# **NEUROPEPTIDE GPCRS IN NEUROENDOCRINOLOGY**

**Topic Editors Hubert Vaudry and Jae Young Seong**

#### *FRONTIERS COPYRIGHT STATEMENT*

© Copyright 2007-2014 Frontiers Media SA. All rights reserved.

All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use.

**ISSN** 1664-8714 **ISBN** 978-2-88919-267-0 **DOI** 10.3389/978-2-88919-267-0

# *ABOUT FRONTIERS*

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

# *FRONTIERS JOURNAL SERIES*

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing.

All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

# *DEDICATION TO QUALITY*

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view.

By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

# *WHAT ARE FRONTIERS RESEARCH TOPICS?*

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area!

Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# **NEUROPEPTIDE GPCRS IN NEUROENDOCRINOLOGY**

Topic Editors: **Hubert Vaudry,** University of Rouen, France **Jae Young Seong,** Korea University, South Korea

The human genome encompasses ≈ 860 G protein-coupled receptors (GPCRs) including 374 non-chemosensory GPCRs. Half of these latter GPCRs recognize (neuro)peptides as natural ligands. GPCRs thus play a pivotal role in neuroendocrine communication. In particular, GPCRs are involved in the neuroendocrine control of feeding behavior, reproduction, growth, hydromineral homeostasis and stress response. GPCRs are also major drug targets and hence possess a

strong potential for the development of innovative pharmaceuticals. The aim of this Research Topic was to assemble a series of review articles and original research papers on neuropeptide GPCRs and their ligands that would illustrate the different facets of the studies currently conducted in this domain.

# Table of Contents

# *Neuropeptide GPCRs in Metabolism and Feeding*


Rim Hassouna, Alexandra Labarthe, Philippe Zizzari, Catherine Videau, Michael Culler, Jacques Epelbaum and Virginie Tolle

*71 Central Urocortin 3 and Type 2 Corticotropin-Releasing Factor Receptor in the Regulation of Energy Homeostasis: Critical Involvement of the Ventromedial Hypothalamus*

Peilin Chen, Christine Van Hover, Daniel Lindberg and Chien Li


Douglas J. MacNeil

*124 Peripheral Injections of Melanin-Concentrating Hormone Receptor 1 Antagonist S38151 Decrease Food Intake and Body Weight in Rodent Obesity Models* Odile Della-Zuana, Valérie Audinot, Viviane Levenez, Alain Ktorza, Françoise Presse,

Jean-Louis Nahon and Jean A. Boutin


João C. R. Cardoso, Rute C. Félix, Vera G. Fonseca and Deborah M. Power

*168 Gonadotropin-Releasing Hormone 2 Suppresses Food Intake in the Zebrafish,*  **Danio rerio**

Ryo Nishiguchi, Morio Azuma, Eri Yokobori, Minoru Uchiyama and Kouhei Matsuda


Fumiko Takenoya, Haruaki Kageyama, Satoshi Hirako, Eiji Ota, Nobuhiro Wada, Tomoo Ryushi and Seiji Shioda

# *Neuropeptide GPCRs in Reproduction*

*186 Decoding High Gonadotropin-Releasing Hormone Pulsatility: A Role for GnRH Receptor Coupling to the cAMP Pathway?*

Joëlle Cohen-Tannoudji, Charlotte Avet, Ghislaine Garrel, Raymond Counis and Violaine Simon


Valérie Simonneaux and Caroline Ancel

*234 Functional Significance of GnRH and Kisspeptin, and Their Cognate Receptors in Teleost Reproduction*

Renjitha Gopurappilly, Satoshi Ogawa and Ishwar S. Parhar

*247 PROK2/PROKR2 Signaling and Kallmann Syndrome* Catherine Dodé and Philippe Rondard

# *Neuropeptide GPCRs in Stress Response*

*255 Is It Really a Matter of Simple Dualism? Corticotropin-Releasing Factor Receptors in Body and Mental Health*

Donny Janssen and Tamás Kozicz


Ludwik K. Malendowicz, Agnieszka Ziolkowska and Marcin Rucinski

# *Neuropeptide GPCRs in Brain Functions*


Magali Basille-Dugay, Hubert Vaudry, Alain Fournier, Bruno Gonzalez and David Vaudry

*321 Brain Neuropeptides in Central Ventilatory and Cardiovascular Regulation in Trout*

Jean-Claude Le Mével, Frédéric Lancien, Nagi Mimassi and J. Michael Conlon

*336 The Stimulatory Effect of the Octadecaneuropeptide ODN on Astroglial Antioxidant Enzyme Systems Is Mediated Through a GPCR*

Yosra Hamdi, Hadhemi Kaddour, David Vaudry, Salma Douiri, Seyma Bahdoudi, Jérôme Leprince, Hélène Castel, Hubert Vaudry, Mohamed Amri, Marie-Christine Tonon and Olfa Masmoudi-Kouki


Miwa Mori, Kenji Mori, Takanori Ida, Takahiro Sato, Masayasu Kojima, Mikiya Miyazato and Kenji Kangawa


Rafael Roesler and Gilberto Schwartsmann

*396 Somatostatinergic Systems: An Update on Brain Functions in Normal and Pathological Aging*

Guillaume Martel, Patrick Dutar, Jacques Epelbaum and Cécile Viollet

*411 The Physiological Role of Orexin/Hypocretin Neurons in the Regulation of Sleep/Wakefulness and Neuroendocrine Functions*

Ayumu Inutsuka and Akihiro Yamanaka

*421 Role of Neuropeptide FF in Central Cardiovascular and Neuroendocrine Regulation*

Jack H. Jhamandas and Valeri Goncharuk


Tomoya Nakamachi, Attila Matkovits, Tamotsu Seki and Seiji Shioda

# *Evolution of Neuropeptides and GPCRs*


Stacia A. Sower, Wayne A. Decatur, Nerine T. Joseph and Mihael Freamat


Takanori Ida, Tomoko Takahashi, Hatsumi Tominaga, Takahiro Sato, Hiroko Sano, Kazuhiko Kume, Mamiko Ozaki, Tetsutaro Hiraguchi, Hajime Shiotani, Saki Terajima, Yuki Nakamura, Kenji Mori, Morikatsu Yoshida, Johji Kato, Noboru Murakami, Mikiya Miyazato, Kenji Kangawa and Masayasu Kojima

*519 More than Two Decades of Research on Insect Neuropeptide GPCRs: An Overview*

Jelle Caers, Heleen Verlinden, Sven Zels, Hans Peter Vandersmissen, Kristel Vuerinckx and Liliane Schoofs

*549 Neuropeptide GPCRs in* **C. elegans** Lotte Frooninckx, Liesbeth Van Rompay, Liesbet Temmerman, Elien Van Sinay, Isabel Beets, Tom Janssen, Steven J. Husson and Liliane Schoofs

# *Molecular Mechanism for GPCR Activation and Signal Transduction*

*567 Neuropeptide Y Receptors: How to Get Subtype Selectivity* Xavier Pedragosa-Badia, Jan Stichel and Annette G. Beck-Sickinger *580 The Melanocortin Receptors and Their Accessory Proteins* Shwetha Ramachandrappa, Rebecca J. Gorrigan, Adrian J. L. Clark and Li F. Chan *588 Neurotensin and Its High Affinity Receptor 1 as a Potential Pharmacological Target in Cancer Therapy* Zherui Wu, Daniel Martinez-Fong, Jean Trédaniel and Patricia Forgez *597 Update on the Urotensinergic System: New Trends in Receptor Localization, Activation, and Drug Design* David Chatenet, Thi-Tuyet M. Nguyen, Myriam Létourneau and Alain Fournier *610 Taltirelin Is a Superagonist at the Human Thyrotropin-Releasing Hormone Receptor* Nanthakumar Thirunarayanan, Bruce M. Raaka and Marvin C. Gershengorn *614 Desensitization, Trafficking, and Resensitization of the Pituitary Thyrotropin-Releasing Hormone Receptor* Patricia M. Hinkle, Austin U. Gehret and Brian W. Jones *628 Relaxin-3/RXFP3 Signaling and Neuroendocrine Function – A Perspective on Extrinsic Hypothalamic Control* Despina E. Ganella, Sherie Ma and Andrew L. Gundlach *639 Galanin Receptors and Ligands* Kristin E. B. Webling, Johan Runesson, Tamas Bartfai and Ülo Langel *653 Mechanisms Underlying the Tissue-Specific and Regulated Activity of the Gnrhr Promoter in Mammals* Anne-Laure Schang, Bruno Quérat, Violaine Simon, Ghislaine Garrel, Christian Bleux, Raymond Counis, Joëlle Cohen-Tannoudji and Jean-Noël Laverrière *672 Sensitivity of Cholecystokinin Receptors to Membrane Cholesterol Content* Aditya J. Desai and Laurence J. Miller *682 Mutation of Phe318 within the NPxxY(x)5,6F Motif in Melanin-Concentrating Hormone Receptor 1 Results in an Efficient Signaling Activity*

Akie Hamamoto, Manabu Horikawa, Tomoko Saho and Yumiko Saito

*693 On the Existence and Function of Galanin Receptor Heteromers in the Central Nervous System*

Kjell Fuxe, Dasiel O. Borroto-Escuela, Wilber Romero-Fernandez, Alexander O. Tarakanov, Feliciano Calvo, Pere Garriga, Mercé Tena, Manuel Narvaez, Carmelo Millón, Concepción Parrado, Francisco Ciruela, Luigi F. Agnati, José A. Narvaez and Zaida Díaz-Cabiale


# *Class B GPCRs in Neuroendocrine Regulation*

*722 Receptor Oligomerization: From Early Evidence to Current Understanding in Class B GPCRs*

Stephanie Y. L. Ng, Leo T. O. Lee and Billy K. C. Chow


Dora Reglodi, Andrea Tamas, Miklos Koppan, Donat Szogyi and Laura Welke

*784 Glucagon-Like Peptide-1 Receptor Overexpression in Cancer and its Impact on Clinical Applications*

Meike Körner, Emanuel Christ, Damian Wild and Jean Claude Reubi


# Neuropeptide GPCRs in neuroendocrinology

# **Hubert Vaudry <sup>1</sup>\* and JaeYoung Seong<sup>2</sup>**

1 INSERM U982, IRIB, University of Rouen, Mont-Saint-Aignan, France

<sup>2</sup> Korea University, Seoul, South Korea

\*Correspondence: hubert.vaudry@univ-rouen.fr

#### **Edited and reviewed by:**

Jeff M. P. Holly, University of Bristol, UK

**Keywords: neuroendocrinology, neuropeptides, biologically active peptides, G protein-coupled receptors, seven-transmembrane domain receptors, heptahelical receptors, signaling mechanisms, transduction pathways**

G protein-coupled receptors (GPCRs) represent the single largest family of plasma membrane receptors, encompassing about 860 members in humans. During the last decades, considerable progress has been made regarding the biochemical identification, signaling mechanisms, allosteric modulation, structural characterization, and pathophysiological implication of GPCRs, which are considered as major targets for the development of new therapeutic agents. Thus, in 2012, Robert J. Lefkowitz and Brian K. Kobila shared the Nobel Prize in Chemistry for their seminal contribution on GPCR structures and functions.

Neuropeptides play a pivotal role in chemical communication in the brain and in peripheral organs. To date, more than 100 neuropeptides have been identified and it is established that a vast majority of them act through GPCRs. Neuropeptide GPCRs are involved in the control of a large array of physiological processes including feeding, reproduction, growth, metabolism, stress response, sleep, and a number of behaviors. The aim of this Research Topic is to illustrate the importance of neuropeptide

GPCRs in neuroendocrine communication by gathering a bouquet of 72 review papers and original articles from leading scientists in this fast-evolving field.

We are deeply indebted to all authors who have contributed to this Research Topic and to the dedicated reviewers who helped us reaching the highest quality standards. We gratefully acknowledge the valuable support of the Frontiers team and Mrs Catherine Beau for her invaluable help in the processing of manuscripts.

*Received: 11 March 2014; accepted: 19 March 2014; published online: 04 April 2014. Citation: Vaudry H and Seong JY (2014) Neuropeptide GPCRs in neuroendocrinology. Front. Endocrinol. 5:41. doi: 10.3389/fendo.2014.00041*

*This article was submitted to Neuroendocrine Science, a section of the journal Frontiers in Endocrinology.*

*Copyright © 2014 Vaudry and Seong . This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Development of the hypothalamic melanocortin system

# **Berengere Coupe1,2 and Sebastien G. Bouret 1,2\***

<sup>1</sup> Neuroscience Program, The Saban Research Institute, Children's Hospital Los Angeles, University of Southern California, Los Angeles, CA, USA <sup>2</sup> U837, Neurobese Lab, INSERM, Jean-Pierre Aubert Research Center, University Lille 2, Lille, France

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Robert Dores, University of Minnesota, USA James A. Carr, Texas Tech University, USA

#### **\*Correspondence:**

Sebastien G. Bouret, Neuroscience Program, The Saban Research Institute, Children's Hospital Los Angeles, University of Southern California, 4650 Sunset Boulevard, MS#135, Los Angeles, CA 90027, USA.

e-mail: sbouret@chla.usc.edu

The melanocortin system is a critical component of the forebrain and hindbrain regulatory systems involved in energy balance. This system is composed of pro-opiomelanocortin (POMC) neurons that act, in part, through the melanocortin-4 receptor (MC4R). Although the importance of the melanocortin system in controlling feeding has been established for two decades, the understanding of the developmental substrates underlying POMC and MC4R neuron development and function has just begun to emerge. The formation of the melanocortin system involves several discrete developmental steps that include the birth and fate specification of POMC- and MC4R-containing neurons and the extension and guidance of POMC axons to their MC4R-expressing target nuclei. Each of these developmental processes appears to require specific sets of genes and developmental cues that include perinatal hormones. Recent evidence has also highlighted the importance of perinatal nutrition in controlling the ultimate architecture of the melanocortin system.

**Keywords: pro-opiomelanocortin,** α**MSH, MC4R, hypothalamus, development, hormones, axon guidance, neurogenesis**

#### **INTRODUCTION**

Pro-opiomelanocortin (POMC) neurons control a variety of physiological functions, the most characterized of which is the regulation of energy balance. They reduce food intake and increase energy expenditure by releasing α-melanocyte-stimulating hormone (αMSH), a product of POMC processing (Cone, 2006; Ellacott and Cone, 2006). More recent targeted deletion studies have specifically shown the importance of POMC neurons in mediating the physiological actions of metabolic hormones, such as leptin and insulin (Belgardt and Brüning, 2010; Williams and Elmquist, 2012). POMC neurons have a limited distribution across the central nervous system (CNS). POMC cell bodies are found only in two CNS nuclei: the arcuate nucleus of the hypothalamus (ARH) in the forebrain and the nucleus of the tractus solitarius (NTS) in the brain stem (Joseph et al., 1983; Cone, 2005). Recent data indicated that while hypothalamic POMC neurons appear to be more involved in integrating long-term adiposity signals, the contribution of hindbrain POMC neurons seem to be more specific to the integration of short-term satiety signals (Zhan et al., 2013). POMC neurons provide extensive projections to various parts of the brain, especially to the paraventricular (PVH) and dorsomedial (DMH) nuclei of the hypothalamus, the lateral hypothalamic area (LHA), and the ventral tegmental area (VTA). Each of these nuclei and areas also plays a major role in the central regulation of feeding behavior and energy balance (King and Hentges, 2011) (**Figure 1**).

Of the five known melanocortin receptors (MCRs), αMSH binds with various affinities to MC1R, MC3R, MC4R, and MC5R. However, it is the MC4R subtype that has been primarily implicated in the regulation of energy balance. The observation that deletion of the MC4R results in a phenotype that is a virtual carbon copy of POMC-deficient mice (including late onset of obesity and diabetes) strongly supports this idea (Huszar et al., 1997; Yaswen et al., 1999; Butler et al., 2001; Challis et al., 2004). Notably, mutations in the MC4R gene are the most common monogenic disorders that cause obesity in humans (Coll et al., 2007). MC4R is a G-protein-coupled receptor that has a widespread distribution throughout the CNS. MC4R mRNA is found in virtually every part of the brain, including the hypothalamus, thalamus, cortex, brainstem, and spinal cord (Mountjoy et al., 1994; Kishi et al., 2003). Soon after the cloning of the MC4R, agouti-related protein (AgRP), which is synthesized in the ARH, has been identified as an endogenous antagonist of the MC4R (Huszar et al., 1997; Ollmann et al., 1997).

This review attempts to summarize our current understanding of the development of the melanocortin system within the context of metabolic programing. Because of the lack of data on the development of POMC neurons in the NTS, this review will largely focus on the development of arcuate POMC neurons and their related hypothalamic MC4R pathways.

#### **DEVELOPMENT OF THE MELANOCORTIN SYSTEM**

The developmental processes that produce the melanocortin system can fall into two broad categories: the birth and determination of the neuronal phenotype of POMC- and MC4R-containing neurons and the formation of functional circuits, which includes POMC axon growth and synaptogenesis onto MC4R-responsive targets.

#### **NEUROGENESIS**

Classic experiments that used birth dating tools, such as the thymidine analog bromodeoxyuridine, revealed that the majority of POMC neurons in the mouse ARH are born primarily on embryonic day (E)11–E12 (Khachaturian et al., 1985; Padilla et al., 2010). However, some POMC neurons, which are located more laterally in the ARH, are generated as late as E13. Gene expression

studies have also shown that neurons in the presumptive ARH first express POMC mRNA on E10–E12 (Padilla et al., 2010) (**Figure 2**), which is consistent with early determination of cell fate. Recent genetic cell lineage tracing studies confirmed this hypothesis and revealed that POMC neurons acquire their terminal phenotype at approximately E15 (Padilla et al., 2010). However, only a portion of embryonic *Pomc*-expressing precursors adopts a POMC fate in adult mice. Half of *Pomc*-expressing precursors acquire a non-POMC fate in adult mice and nearly one quarter of the mature NPY neurons in the ARH share a common progenitor with POMC cells (Padilla et al.,2010). These data show the unique property of *Pomc*-expressing progenitors with respect to giving rise to antagonistic neuronal populations.

#### **DEVELOPMENT OF POMC-CONTAINING AXONAL PROJECTION**

The ontogeny of the POMC neuronal projections remains to be determined specifically. However, axonal tract tracing experiments in mice have revealed that ARH projections are largely immature at birth and develop postnatally during the first 2 weeks of postnatal life. By P6, ARH projections extend through the periventricular zone of the hypothalamus to provide inputs to the DMH first, followed by inputs to the PVH between P8 and P10. Projections from the ARH to the LHA develop significantly later, with the mature pattern of innervation first apparent on P12. Not until P18 does the pattern of ARH axonal projections achieve a distribution that resembles that seen in the adult (Bouret et al., 2004a). These findings suggest that the development of ARH POMC axonal projections toward each of their target nuclei does not occur until the second week of postnatal life, i.e., at a time that is far removed from the birth of these neurons (E11–E12). At the synaptic levels, Colmers and collaborators showed that there is an age-dependent increase in the electrophysiological response of specific sets of PVH neurons to melanocortin, with a maximal response observed at

P28–P35 (Melnick et al., 2007). These results suggest that synapses between POMC axons and PVH MC4R-containing neurons are not structurally and functionally mature until puberty.

# **DEVELOPMENTAL REGULATION OF MC4R**

In spite of the importance of MC4R in energy balance regulation, we still know relatively little about the exact time point at which the assembly of this system is fully established. For example, we still do not know when MC4R neurons are born or when MC4R becomes functional and able to signal. Nevertheless, the fact that peripheral injection of the MCR agonist MTII reduces milk intake and body weight as early as during the first 2 weeks of postnatal life suggests that MC4R receptors might be present and functional in the hypothalamus at this age (Glavas et al., 2007). Consistent with this idea, *in situ* hybridization analysis shows that MC4R mRNA is abundant in the hypothalamus and especially in the PVH at P10 (**Figure 2**). That peripheral injection of MTII induces strong activation of cFos immunoreactivity (a marker of neuronal activation) in the PVH at P5–P15 further supports the functionality of MC4R in the PVH during early postnatal life (Glavas et al., 2007). At the molecular level, it was shown that MC4R mRNA is first expressed at E12 in the proliferative zone surrounding the lower portion of the third ventricle (also known as the neurepithelium) and that this mRNA expression peaks at E16 (**Figure 2**) (Mountjoy and Wild, 1998). These findings are particularly interesting because it

is known that neurons that compose various hypothalamic nuclei in adults are primarily derived from precursors that originate from this proliferative zone, raising the possibility that MC4R could be involved in hypothalamic neurogenesis. Supporting this hypothesis, melanocortins stimulate astrocyte proliferation *in vitro* (Zohar and Salomon, 1992). Other brains sites express MC4R during embryonic development. In addition to being expressed in the diencephalic neurepithelium, MC4R mRNA is also found in the telencephalon and the lamina terminalis at E14 (Mountjoy and Wild, 1998). By E19,MC4R mRNA is widely expressed throughout the brain (Mountjoy and Wild, 1998).

# **HORMONAL REGULATION OF POMC NEURONS DURING EARLY LIFE**

Accumulating evidence suggests that there are physiological differences in the regulation of energy balance between adults and neonates. For example, in sharp contrast to the effects of leptin on adults, several groups have reported that exogenous leptin does not significantly inhibit growth,food intake, or energy expenditure until after weaning (Mistry et al., 1999;Ahima and Hileman, 2000; Schmidt et al.,2001;Proulx et al.,2002). Similarly,in contrast to the well-known orexigenic effects of ghrelin in mature animals, exogenous ghrelin does not significantly promote milk intake in the first 2–3 postnatal weeks (Piao et al., 2008; Steculorum and Bouret, 2011a). The general thinking has been that the neonatal brain is relatively insensitive to metabolic hormones. However, both leptin and ghrelin receptors are expressed in the ARH during early postnatal life (Caron et al., 2010; Steculorum and Bouret, 2011a) and these receptors can initiate cellular responses, particularly in POMC neurons. First, acute peripheral leptin treatment in mice on P10 induces phosphorylation of STAT3 and ERK (two major leptin receptor signaling pathways) in 20–35% of ARH POMC neurons (Caron et al., 2010; Bouret et al., 2012). The same leptin treatment increases *Pomc* mRNA levels in the rat ARH (Proulx et al., 2002). Second, acute peripheral ghrelin treatment in P10 mice causes a reduction in *Pomc* gene expression in the ARH (Steculorum and Bouret, 2011a). Collectively, these findings provide convincing evidence that POMC neurons contain functional receptors for metabolic hormones such as leptin and ghrelin during the postnatal period, and suggest that the "hormone insensitivity" observed during this period could instead be from a failure of these cells to relay hormonal signals to other parts of the hypothalamus.

# **MECHANISMS UNDERLYING POMC NEURON DEVELOPMENT**

The process of developing highly specialized cellular structures, such as POMC neurons require tight temporal and regional regulation of expression for specific sets of genes and developmental cues. A variety of genetic tools are now available to specifically identify the cellular and molecular pathways that are involved in POMC neuronal development.

#### **MECHANISMS UNDERLYING CELL FATE SPECIFICATION: THE ROLE OF TRANSCRIPTION FACTORS**

While the specific programs involved in the determination of POMC terminal fates are unknown, recent data investigating the role of the Mash1-neurogenin 3 (Ngn3) pathway provides some clues as to the role of these transcription factors in the differentiation of the POMC lineage. Using a mouse model of Mash1 deficiency (Mash1−/− mice), McNay et al. (2006) found that this basic helix-loop-helix transcription factor has a pro-neural function and act upstream of Ngn3 to regulate neurogenesis in the ventral hypothalamus. Loss of Mash1 blunts Ngn3 expression in ARH progenitors and is associated with a dramatic reduction in the number of POMC-expressing cells at E12. More recent genetic fate mapping and loss of function studies in mice further demonstrated that the expression of Ngn3 in progenitor cells promotes the development of arcuate POMC neurons (Pelling et al., 2011). Based on the recent observations that a subset of *Pomc*-expressing progenitors in the ARH differentiates into functional mature NPY neurons (Padilla et al., 2010), it would also be interesting to determine the molecular mechanisms that underlie this developmental switch.

#### **CUES INVOLVED IN AXON GROWTH AND AXON GUIDANCE**

The precise molecular mechanisms that are responsible for the formation of POMC circuits are only beginning to be understood. In general, axon development comprises two aspects: the physical act of extension and the molecular mechanisms underlying this process. Axons grow by sending out a highly plastic and sensitive structure called a "growth cone," which travels toward the target and trails behind it the elongating neurite. As noted above, neonatal POMC neurons express LepRb, and the administration of leptin to mouse neonates results in the activation of major LepRb signaling pathways, including pSTAT3 and pERK in POMC neurons, specifically (Caron et al., 2010; Bouret et al., 2012). Recent anatomical data showed that one of the key factors in controlling the development of POMC neural circuits appears to be the expression of LepRb by POMC neurons (**Table 1**). Direct exposure of ARH explants *in vitro* to leptin promotes axon growth (Bouret et al., 2004b). In addition, mice or rats that lack functional LepRb signaling (Leprdb/Leprdb mice and fa/fa rats, respectively) display a reduced density of ARH POMC projections to the PVH (Bouret and Simerly, 2007; Bouret et al., 2012). Similarly, the density of αMSH-immunoreactive fibers is also markedly reduced in the PVH of s/s mice that lack functional LepRb→STAT3 signaling (Bouret et al., 2012). This observation raises the importance of this signaling pathway, specifically, in the development of POMC neural projections. However, not all LepRb signaling pathways play a role in the formation of POMC projections. For example,

#### **Table 1 | List of genetic and pathological conditions that alter development of POMC-derived neural projections.**


mice that lack LepRb→ERK signaling (l/l mice) display comparable densities of αMSH-immunoreactive fibers in the PVH compared to wild-type mice (Bouret et al., 2012). Of particular importance is the fact that leptin appears to exert its developmental action on POMC neural projections during a discrete developmental critical period. POMC neuronal projections are disrupted in leptin-deficient (Lepob/Lepob) mice, and exogenous leptin treatment during the first 2 weeks of postnatal life rescues these projections (Bouret et al., 2004b). In contrast, the treatment of adult Lepob/Lepob mice with leptin is relatively ineffective and does not increase the density of αMSH fibers in the PVH to levels that are characteristic of wild-type mice (Bouret et al., 2004b).

Growing POMC axons must then choose a path to follow and must decide the direction to go on this path to innervate the proper nucleus (e.g., the PVH). The pathways are defined by cell–cell interactions and diffusible chemorepulsive and chemoattractive cues (Tessier-Lavigne and Goodman, 1996). Notably, the diffusible axon guidance cues Netrin, Slit, and Semaphorins are highly expressed in the PVH during development (Xu and Fan, 2008), and POMC terminals express the semaphorin receptor neuropilin 1 (LeGuern and Bouret, unpublished data). Supporting a role for neuropilins/semaphorins in POMC axon guidance, a loss of neuropilin 1 receptors in POMC neurons disrupts the development of POMC axonal projections to the PVH, specifically (LeGuern and Bouret, unpublished data). The formation of POMC neural connections also likely involves cell adhesion molecules. Supporting this idea, mice that are deficient in contactin, a cell adhesion molecule that is involved in the formation of axonal projections, display reduced density of αMSH-immunoreactive fibers in the PVH during postnatal development (Fetissov et al., 2005).

#### **CELLULAR MECHANISMS INVOLVED IN POMC NEURON DEVELOPMENT: ROLE OF AUTOPHAGY**

The development of POMC neurons also requires massive cytoplasmic remodeling. Autophagy is one of the major cellular degradation processes in which parts of the cytoplasm and intracellular organelles are engulfed within double-membraned vesicles, known as autophagosomes (Klionsky, 2007). An important function of autophagy is to promote cell growth, development, and homeostasis, by maintaining a balance between the synthesis, degradation, and subsequent recycling of cellular components (Levine and Klionsky, 2004; Maiuri et al., 2007; Cecconi and Levine, 2008). Recent morphological data revealed that autophagy is constitutively present in the hypothalamus during important periods of development, including in POMC neuron perikarya and processes such as dendrites (Coupe et al., 2012). Consistent with a functional role for autophagy in POMC neurons, conditional deletion of essential autophagy genes, such as the autophagy-related gene (Atg) 7, results in obesity and impaired glucose homeostasis (Coupe et al., 2012;Kaushik et al., 2012; Quan et al., 2012). These metabolic disturbances are associated with neurodevelopmental abnormalities. Neonates lacking ATG7 in POMC neurons display a reduced density of POMC-containing projections to each of their target nuclei, including the PVH, DMH, and LHA (Coupe et al., 2012). These structural abnormalities persist

throughout adult life and appear to be the result of a diminished capacity of POMC neurons for extending axons (Coupe et al., 2012). However, not all of the developmental processes are affected by autophagy deficiency. No changes in POMC cell numbers were reported in mice that lacked autophagy in POMC neurons (Coupe et al., 2012; Kaushik et al., 2012; Quan et al., 2012), which suggests that autophagy does not influence neurogenesis or programed cell death and that instead it has a specific role in axon growth.

# **PATHOLOGICAL CONDITIONS THAT ALTER MELANOCORTIN SYSTEM DEVELOPMENT**

# **MATERNAL OBESITY AND/OR DIABETES**

In the United States, epidemiological studies have estimated that 20% of women are obese when they conceive (Johnson et al.,2006). This disturbing observation highlights the importance of evaluating the outcomes of maternal obesity in the offspring. Maternal high-fat diet (HFD) feeding during pregnancy is most likely the most widely used approach for studying the consequences of maternal obesity. Notably, offspring born to obese females fed an HFD (45–60% of calories from fat) during gestation only or during both gestation and lactation become progressively overweight, hyperphagic, and glucose intolerant, and they display an increase in adiposity (Chen et al., 2009; Kirk et al., 2009). The model of diet-induced obesity (DIO) developed by Levin et al. (1997) also provides a valuable tool for the study of obesity, in part because Levin's DIO rats share severalfeatures with human obesity, including polygenic inheritance. This animal model is therefore particularly well suited for the study of the relative contribution of genetic versus environmental factors in metabolic programing. Impaired organization of POMC neural circuits is a common feature of maternal obesity (**Table 1**). Animals born to either high-fat fed or obese-prone DIO dams display a reduced density of αMSH fibers innervating the PVH (Bouret et al., 2008; Kirk et al., 2009). In DIO rats, the abnormal development of POMC projections appears to be the result of a diminished response to the neurotrophic action of leptin during the early postnatal period (Bouret et al., 2008). In addition, a significant remodeling of synapses onto POMC neurons has been observed in DIO rats, particularly in response to nutritional challenges (Horvath et al., 2010). DIO rats fed a chow diet display increased inhibitory inputs to POMC neurons compared to obesity-resistant DR rats. In addition, DIO rats fed a HFD display a loss of synapses onto POMC neurons, whereas high-fat feeding in control (DR) rats causes an increase in POMC synaptic coverage (Horvath et al., 2010).

However, one caveat to keep in mind is that almost all of the animal models of maternal obesity are also hyperglycemic and insulin-resistant. This complication makes it difficult to differentiate the detrimental effects of maternal obesity *per se* as opposed to maternal diabetes. Nevertheless, the manipulation of glucose and insulin levels without an alteration of the diet can be performed experimentally by injecting streptozotocin, a pancreatic beta-cell toxin. Using this approach, we recently found that maternal diabetes alone (i.e., without maternal obesity) can cause a reduction in the density of αMSH-immunoreactive projections to the PVH (Steculorum and Bouret, 2011b) (**Table 1**). Importantly, an increase in the POMC cell number is observed in the ARH of pups that were born to diabetic dams (Steculorum and Bouret, 2011b), which supports the hypothesis that the low density of POMC-derived fibers is more likely the result of alterations in POMC axon growth as opposed to a reduction in cell numbers.

#### **MATERNAL MALNUTRITION**

Interest in maternal malnutrition is also particularly high given its high prevalence in underdeveloped countries and the evidence regarding its disease-promoting potential in offspring. Animal models commonly used to study the consequences of maternal malnutrition include caloric restriction and a low-protein diet during pregnancy and lactation. Maternal malnutrition (caused either by caloric restriction or a low-protein diet) affects the overall organization of POMC neural circuits by altering the density of αMSH-containing axons (Delahaye et al., 2008; Coupe et al., 2010) (**Table 1**). However, if pups born to malnourished dams do not exhibit a rapid catch-up growth, this reduction of POMC innervation is associated with an increased sensitivity to the anorexigenic effect of the melanocortin agonist MTII (Stocker et al., 2012). Therefore, the timing of catch-up growth appears to be an important determinant for the lifelong regulation of the melanocortin system. Supporting this hypothesis, although early catch-up growth ameliorates the abnormal organization of POMC pathways observed in pups born to protein-restricted dams and is highly beneficial for markers of brain development, including markers of cell adhesion and axon elongation,late catch-up growth causes permanent structural defects of POMC neural projections (Coupe et al., 2010).

#### **POSTNATAL OVERFEEDING/OBESITY**

Because of the importance of hypothalamic development during postnatal life, including the melanocortin system, in rodents, animal models of postnatal metabolic programing are particularly relevant. An animal model that has proven to be extremely fruitful for the study of postnatal overfeeding is the divergent litter size model. In this model, pups are raised in small litters (SL) from birth to weaning to induce accelerated growth during the pre-weaning period. Postnatally overfed animals show increased adult body weight and an accelerated and exacerbated weight gain when fed an HFD (Glavas et al., 2010). Chronic postnatal overfeeding is associated with a reduced activity of the melanocortin system. For example, rats raised in SL display an overall decrease in the expression of *Pomc* gene expression (Srinivasan et al., 2008; Chen et al., 2009). The observed changes appear to reflect an acquired mechanism that originates from a malprograming of the hypothalamic melanocortin system during early life rather than being a consequence of metabolic dysfunctions, such as overweight and hyperphagia. Consistent with this idea, changes in POMC innervation to the PVH are observed as early as during the first postnatal weeks in neonatally overfed mice, i.e., prior to the development of overweight and hyperphagia (Bouret et al., 2007). In addition to its adverse effects on the *Pomc* gene expression, postnatal overfeeding affects the neuronal response to melanocortins. For example, PVH neurons of chronically overfed pups display reduced electrophysiological responses to αMSH (Davidowa et al., 2003).

# **EPIGENETIC CHANGES**

Epigenetic mechanisms of gene regulation, such as DNA methylation and histone modifications (e.g., methylation, acetylation, ubiquitination) can regulate gene expression in response to environmental stimuli. Given that *Pomc* expression can be affected by the prenatal or early postnatal diet (see above) many groups have investigated whether epigenetic modifications of the POMC promoter occur during the neonatal period in response to nutritional insults. Using a bisulfite sequencing approach, Plagemann and colleagues studied the methylation status of CpG dinucleotides of the *Pomc* promoter in rats that were raised in SL and found that postnatal overfeeding causes hypermethylation of the *Pomc* promoter. Interestingly, this hypermethylation occurs within the two Sp1-related binding sequences of the *Pomc* promoter (Sp1, NF-KB), which are essential for mediating the effect of leptin and insulin on *Pomc* gene expression (Plagemann et al., 2009). Similarly, recent human studies showed that childhood obesity is associated with POMC hypermethylation (Kuehnen et al., 2012). In contrast, methylation of the fetal hypothalamic *Pomc* promoter is reduced in underfed sheep, which is associated with reduced DNA methyltransferase activity and altered histone methylation and acetylation (Begum et al., 2012). In addition, rat neonates from protein-restricted dams display a reduction of hypothalamic Dnmt1 and Dmnt3a mRNA expression (Coupe et al., 2010).

#### **CONCLUSION**

At a time when obesity, including in children, is reaching epidemic proportions, it appears crucial to better understand the biological processes that mediate the development of metabolic systems. The melanocortin system plays an important role in this process, and recent studies are providing knowledge on how POMC- and MC4R-containing neurons develop. These studies have also revealed that changes in the hormonal and nutritional milieu during critical periods of life can permanently influence the development and functional activity of the melanocortin system. This finding could represent a key mechanism for effecting long-term changes in weight regulation in response to perinatal insults. It also opens new avenues for understanding congenital eating disorders, such as Prader–Willi syndrome. However, if we want to design therapeutic interventions to reverse the metabolic programing of the fetus and/or neonate, it remains to show the periods of vulnerability for the melanocortin system in humans, which most likely proceeds on a timeline of months, compared to days in rodents.

#### **ACKNOWLEDGMENTS**

Work in the authors' laboratory is supported by grants from the National Institute of Health (Grant DK84142), the Foundation for Prader–Willi Research, the "Fondation pour la Recherche Médicale," the EU FP7 integrated project (grant agreement no. 266408, "Full4Health"), and the "Agence Nationale de la Recherche" (Grants ANR-08-JCJC-0055-01 and ANR-11-BSV1-021-02).

# **REFERENCES**


development: cell makeover rather than cell death. *Dev. Cell* 15, 344–357.


of arcuate nucleus neuropeptidergic development in contactin-deficient mice: comparison with anorexia and food-deprived mice. *Eur. J. Neurosci.* 22, 3217–3228.


distribution of murine hypothalamic proopiomelanocortin neurons innervating distinct target sites. *PLoS ONE* 6:e25864. doi:10.1371/journal.pone.0025864


R. D. (1994). Localization of the melanocortin-4 receptor (MC4-R) in neuroendocrine and autonomic control circuits in the brain. *Mol. Endocrinol.* 8, 1298–1308.


Hypothalamic proopiomelanocortin promoter methylation becomes altered by early overfeeding: an epigenetic model of obesity and the metabolic syndrome. *J. Physiol. (Lond.)* 587, 4963–4976.


compromises the organization of hypothalamic feeding circuits and impairs leptin sensitivity in offspring. *Endocrinology* 152, 4171–4179.


of feeding behavior by POMC neurons in the brainstem and hypothalamus, respectively. *J. Neurosci.* 33, 3624–3632.

Zohar, M., and Salomon, Y. (1992). Melanocortins stimulate proliferation and induce morphological changes in cultured rat astrocytes by distinct transducing mechanisms. *Brain Res.* 576, 49–58.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 February 2013; paper pending published: 04 March 2013; accepted: 12 March 2013; published online: 27 March 2013.*

*Citation: Coupe B and Bouret SG (2013) Development of the hypothalamic melanocortin system. Front. Endocrinol. 4:38. doi: 10.3389/fendo.2013.00038*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Coupe and Bouret. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

#### **Mathieu Méquinion<sup>1</sup> , Fanny Langlet <sup>1</sup> , Sara Zgheib<sup>2</sup> , Suzanne Dickson3,4, Bénédicte Dehouck 1,5 , Christophe Chauveau<sup>2</sup>† and Odile Viltart 1,6\* †**

<sup>1</sup> UMR INSERM 837, Development and Plasticity of Postnatal Brain, Lille, France

<sup>2</sup> Pathophysiology of inflammatory of bone diseases, Université Lille Nord de France-ULCO – Lille 2, Boulogne sur Mer, France

<sup>3</sup> Department of Physiology, Institute of Neuroscience and Physiology, The Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

<sup>4</sup> Department of Endocrinology, Institute of Neuroscience and Physiology, The Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

<sup>5</sup> Université Lille Nord de France – Université d'Artois, Liévin, France

<sup>6</sup> Université Lille Nord de France-USTL (Lille 1), Villeneuve d'Ascq, France

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Masamitsu Nakazato, University of Miyazaki, Japan Paolo Magni, Università degli Studi di Milano, Italy

#### **\*Correspondence:**

Odile Viltart, Development and Plasticity of the Postnatal Brain, Team 2, Jean-Pierre Aubert Research Center, UMR INSERM 837, Bât Biserte, 1 place de Verdun, 59,045 Lille cedex, France. e-mail: odile.viltart@univ-lille1.fr

†Christophe Chauveau and Odile Viltart have contributed equally to this work.

Increasing clinical and therapeutic interest in the neurobiology of eating disorders reflects their dramatic impact on health. Chronic food restriction resulting in severe weight loss is a major symptom described in restrictive anorexia nervosa (AN) patients, and they also suffer from metabolic disturbances, infertility, osteopenia, and osteoporosis. Restrictive AN, mostly observed in young women, is the third largest cause of chronic illness in teenagers of industrialized countries. From a neurobiological perspective, AN-linked behaviors can be considered an adaptation that permits the endurance of reduced energy supply, involving central and/or peripheral reprograming. The severe weight loss observed in AN patients is accompanied by significant changes in hormones involved in energy balance, feeding behavior, and bone formation, all of which can be replicated in animals models. Increasing evidence suggests that AN could be an addictive behavior disorder, potentially linking defects in the reward mechanism with suppressed food intake, heightened physical activity, and mood disorder. Surprisingly, the plasma levels of ghrelin, an orexigenic hormone that drives food-motivated behavior, are increased. This increase in plasma ghrelin levels seems paradoxical in light of the restrained eating adopted by AN patients, and may rather result from an adaptation to the disease.The aim of this review is to describe the role played by ghrelin in AN focusing on its central vs. peripheral actions. In AN patients and in rodent AN models, chronic food restriction induces profound alterations in the « ghrelin »signaling that leads to the development of inappropriate behaviors like hyperactivity or addiction to food starvation and therefore a greater depletion in energy reserves. The question of a transient insensitivity to ghrelin and/or a potential metabolic reprograming is discussed in regard of new clinical treatments currently investigated.

**Keywords: ghrelin, anorexia, food intake, energy balance, central alterations, peripheral alterations, reward, animal models**

#### **INTRODUCTION**

Feeding is a behavior that ensures an adequate and varied supply of nutritional substrates essential to maintain energy levels for basal metabolism, physical activity, growth, and reproduction and hence, for survival of every living organism on Earth. In the case of mammals, that must maintain a stable body temperature, the maintenance of a high metabolic rate requires constant availability of a sufficient amount of energy stores. The tight balance between energy demand and expenditure is fine-tuned by an adapted dialog between homeostatic and hedonic brain systems that are regulated by peripheral signals involved in feeding behavior and energy homeostasis. Mechanisms for feeding control remain a current and crucial scientific subject for understanding the etiology and potential therapeutic approaches for the treatment of food intake disorders that include obesity, on one hand, and severe forms of anorexia nervosa (AN) on the other.

Voluntary anorexia is a disease not unique to man and has even been described in many vertebrate species that favor migration activity (Wang et al., 2006). In this case, surviving food deprivation involves an adaptation of metabolism, such that internal energy stores available at the onset of fasting are used to maintain basal metabolism and physical activity. The biochemical and physiological adaptations that result from a lack of food help to preserve physiological function in order to maintain behaviors like food seeking or predator avoidance and also, to resume all metabolic processes necessary when food becomes available. However, absolute or long term food deprivation observed in nature or in restrictive AN proceeds in stages in which the individual/organism tries to adapt its metabolism to energy costs but that culminates in death, due to exhaustion of energy stores. As clearly described by Wang et al. (2006), the different stages progress from fasting to starvation, but "*The demarcation between these two states*

*is rarely appreciated, perhaps owing to lack of definition. In humans, fasting often refers to abstinence from food, whereas starvation is used for a state of extreme hunger resulting from a prolonged lack of essential nutrients. In other words, starving is a state in which an animal, having depleted energy stores, normally would feed to continue normal physiological processes*." Briefly, three metabolic phases are described during food deprivation (Wang et al., 2006) where energy metabolic adaptations occur to allow supply of fuel in the different parts of the organism, especially the brain (see **Table 1**). In regard to these metabolic stages, the transition from fasting to starvation occurs by the end of phase II or the beginning of phase III. Thus, voluntary anorexia as seen in restrictive AN should correspond to phases I and II.

Restrictive AN is a feeding behavior disorder for which severe chronic food restriction causes dramatic physiological and psychological effects that are detrimental for health. AN is most prevalent in women aged of 25 years old or younger (whose BMI reaches values largely below 18.5 kg/m<sup>2</sup> ) and is currently the third largest cause of chronic illness in teenagers (Lucas et al., 1991). The prevalence of AN has drastically increased within recent decades. It leads to central and/or peripheral reprograming that permits the individual/organism to endure a reduced energy supply. These drastic conditions not only induce severe weight loss and metabolic disturbance, but also infertility, osteopenia, and osteoporosis. Moreover, AN is increasingly recognized as an addictive behavior disorder. <sup>∗</sup> -Indeed, many of its common primary characteristics – food obsession coupled with food restriction, weight loss, heightened physical activity, and the strong association with mood disorder (such as anxiety or depression), strongly suggest a potential alteration of the central (dopaminergic) reward system.

Anorexia nervosa patients exhibit significant changes in the release of key hormones involved in energy balance and feeding control (Hasan and Hasan, 2011). For example, the plasma levels of ghrelin, an orexigenic hormone mostly released from the empty stomach, are increased inAN patients along all the day (Germain et al., 2009, 2010). This hormone acts centrally to increase food intake (Wren et al., 2001a,b) and food-motivated behavior (Skibicka et al., 2012), but has also been suggested to be required for the maintenance of blood glucose homeostasis during severe calorie restriction (Zhao et al., 2010). The increases in plasma ghrelin levels in AN seem paradoxical in light of the restrained eating adopted by these patients and suggest an adaptive response to the disease. In regard to the metabolic deficiencies occurring in restrictive AN (see *infra*), the aim of this review is to highlight the impact of ghrelin in the adaptation of the organism to chronic food restriction until it falls into exhaustion and death. A better understanding of the role of this gastric hormone in dysfunctional AN like feeding behavior is important when evaluating its therapeutic potential for the treatment of AN, envisaged to be used alongside mainstay psychiatric and nutritional therapies.

#### **PHYSIOLOGICAL ALTERATIONS IN ANOREXIA NERVOSA**

#### **TYPES AND SUBTYPES OF ANOREXIA NERVOSA: NEW DSM-V CLASSIFICATION**

Chronic food restriction is linked to several disorders classified in DSM-V (Diagnostic and Statistical Manuel of Mental Disorders). In the provisional version of DSM-V (of spring **Table 1 | Different metabolic phases occurring during food restriction and permitting distinction between fasting and starvation (seeWang et al., 2006).**

All the metabolic changes aim to deliver sufficient amount of glucose for different organs and especially for the brain.

2012; http://www.dsm5.org/meetus/pages/eatingdisorders.aspx), the *Feeding and Eating Disorders* category includes three disorders as manifested by persistent failure to meet appropriate nutritional and/or energy needs and significant weight loss, *AN, avoidant/restrictive food intake disorder (ARFIDO), and atypical AN*.

Diagnostic criteria for AN includes the restriction of energy intake relative to requirements, a drastic significant loss of body weight, an intense fear of gaining weight, body image disturbance, and/or a persistent lack of recognition of the seriousness of the current low body weight. In a recent review (Garcia et al., 2011), the lifetime prevalence of AN was estimated to be 1.9% infemale adults to 2.6% in female adolescents in industrialized countries. In the binge eating/purging subtype, the individual engages in recurrent episodes of binge eating or purging behavior while such episodes do not occur in the restricting subtype. Patients from these two subtypes also exhibit differences in eating disorder symptom indicators (Olatunji et al., 2012). However, the subtype determination at the time of the diagnosis should be considered carefully since, over a 7-year period, the majority of women withAN werefound to cross over to the restricting and binge eating/purging AN subtypes (Eddy et al., 2008). In a 21-year follow-up study, Löwe et al. (2001) showed that 16% of AN patients deceased due to consequences of their illness: about 50% died because of somatic complications leading to heart attack and the remainder committed suicide. Moreover, mortality is significantly more common among inpatients with somatic comorbidity (like renal, cardiac, bone, and digestive pathologies) than among inpatients without a somatic disease (Erdur et al., 2012). Finally, among the psychiatric comorbidities, AN is often associated with depression, anxiety, obsessive compulsive or personality disorders, and drug abuse (Erdur et al., 2012). Whether AN resembles an addiction behavior disorder remains one major question for physicians and researchers alike. The criteria proposed by Goodman (1990) to identify addictive disorder (**Table 2**) are found in AN patients. Indeed, Speranza et al. (2012) showed that 35% of the restrictive AN subtype patients, 48% of the binge eating/purging AN subtype patients and 60% of the patients have substance-use disorders and hence, exhibit an addictive disorder according to Goodman's criteria. From these criteria, an emerging hypothesis of AN implicates neurobiological mechanisms (and hence, investigation strategies for treatments and diagnostic markers) that are based on a reward deficit and on the recognition as an addictive behavior disorder (Alguacil et al., 2011). In fact, among the different theories linking AN and addiction, interestingly, the "auto-addiction opioid model" proposes that this chronic eating disorder could represent an addiction to the body's endogenous opioids, especially β-endorphins (see Davis and Claridge, 1998). Starvation and excessive exercise, that concern a high percentage of AN patients (Davis et al., 1997; Kohl et al., 2004), are associated with increased levels of β-endorphin, known to further stimulate dopamine in the mesolimbic reward centers (Bergh and Södersten, 1996; Casper, 1998). This mesolimbic pathway plays a pivotal role in addictive behaviors related to drugs and dietary behaviors (Avena and Bocarsly, 2012; Perelló and Zigman, 2012). Importantly, this mesolimbic dopamine pathway is activated by ghrelin (Abizaid et al., 2006; Jerlhag et al., 2006; see *infra*) and, since AN patients have high plasma ghrelin levels (Germain et al., 2009, 2010), it follows that there may be a dysfunctional ghrelin-(dopamine) reward signal in these patients. However, as discussed by Barbarich-Marsteller et al. (2011), there are fundamental differences between AN and addiction. Indeed, the main goal of an individual suffering from a substance abuse disorder is to pursue the immediate effects of the drug on mood and/or behavior (alleviation of anxiety, for example), whereas the goals of an AN patient are both immediate and long term. In these patients, dieting and starvation produce immediate feelings of hunger that may induce a sense of control over one's body and thereby a sense of control over one's life while, in the long term, it produces sustained weight loss and thinness that take on an irrationally important value.

In the ARFIDO, insufficient food intake is associated with significant weight loss, nutritional deficiency, dependence on enteral feeding, or nutritional supplements and/or a marked psychosocial dysfunction. In these patients, the eating disturbance does not occur exclusively during the course of AN, and is not associated with body image disturbances. ARFIDO is a new recognized

#### **Table 2 | Addictive disorder criteria according to Goodman (1990).**

	- 1. Frequent preoccupation with the behavior or preparatory activities
	- 2. Frequent engaging in the behavior to a greater extent or over a longer period than intended
	- 3. Repeated efforts to reduce, control, or stop the behavior
	- 4. A great deal of time spent in activities necessary for the behavior, engaging in the behavior, or recovering from its effects
	- 5. Frequent engaging in the behavior when expected to fulfill occupational, academic, domestic, or social obligations
	- 6. Important social, occupational, or recreational activities given up or reduced because of the behavior
	- 7. Continuation of the behavior despite knowledge of having a persistent or recurrent social, financial, psychological, or physical problem that is caused or exacerbated by the behavior
	- 8. Tolerance: need to increase the intensity or frequency of the behavior in order to achieve the desired effect or diminished effect with continued behavior of the same intensity

To reach the categorical diagnosis of addictive disorder according to Goodman (1990), criteria A–D plus criterion E (five among nine symptoms) must be met for at least 1 month.

eating disorder previously classified in the left over category *Eating Disorder Not Otherwise Specified* (EDNOS) corresponding to the majority of in-and-out patients treated for eating disorders. Its prevalence remains to be determined.

The last disorder, *atypical AN*, is in the EDNOS category. It includes all the criteria for AN diagnosis except that, despite significant weight loss, the individual's weight is within or above the normal range. The lifetime prevalence of atypical AN ranges from 2.4% in female adults to 7.7% in female adolescents (Garcia et al., 2011).

Considering this recent classification, in this review we focus on restrictive AN in which the individual is subjected to chronic food restriction that may or may not be associated with intense physical exercise. In fact, the course of AN is extremely variable, with approximately 50–60% of individuals recovering, 20–30% partially recovering, and 10–20% remain chronically ill (Löwe et al., 2001; Fisher, 2003). The unknown etiology of AN renders this complex psychiatric disease difficult to treat and current pharmacological treatments have little efficacy during the acute phase of illness or in preventing relapse (Barbarich-Marsteller, 2007). However, the physiological alterations induced by severe chronic food restriction impact on peripheral compartments (fat, bone, reproductive axis, energy balance) and on central pathways (reward, food intake, mood regulation, etc.) for which the outcome is usually similar whatever the initial cause (personal history, infancy trauma, socio-cultural pressions, personality traits, neurobiological, genetic background, etc.). Some authors even support the view that the physiological mechanisms involved in the regulation of feeding behavior in AN might, in many neurobiological effects, parallel those of obesity. As recently suggested by Jacquemont et al. (2011): "*abnormal eating behaviors, such as hyperphagia and anorexia, could represent opposite pathological manifestations of a common energy balance mechanism, although the precise relationships between these mirror phenotypes remain to be determined*." Thus, the development of effective/new pharmacological treatments for this disease area would be enhanced if the mechanisms maintaining the abnormal behaviors characteristic of AN are better understood.

#### **CENTRAL AND PERIPHERAL ALTERATIONS IN ANOREXIA NERVOSA**

Whatever the initial and causal factors leading to AN, all patients display similar energy metabolic deficits and are unable to adapt their feeding behavior to energy demand and costs. In this state, survival requires the development of physiological changes that drive the individual/animal to adapt itself to these drastic conditions. Among all of the variations induced by chronic food restriction, the endocrine, immune, bone, and metabolic changes first allow adaptations to starvation, and are subsequently often directly involved in the complications of the disease (Estour et al., 2010). In addition, some of the feeding-regulatory factors are also involved directly (or not) in the modulation of reward-related

and motivational processes, as well as in cognition and emotions associated with the disease. It should be noted that some of these endocrine changes persist after recovery and might contribute to susceptibility for AN recurrence (Lawson and Klibanski, 2008).

Among the biological factors whose levels are altered in AN patients, neurotransmitters and neuropeptides regulating appetite and feeding may contribute to some of the occurring central perturbations (**Table 3**). Overall, a high degree of heterogeneity has been observed between studies, for most of the assayed factors (plasma and cerebrospinal fluid samples). This heterogeneity could be explained by differences in the clinical characteristics of the samples (as severity and duration of the illness or subtype) and/or the increasing reliability of the methods used to ascertain factor concentrations over the last two decades. Consequentially, it is impossible to link any tendency in changes of levels and/or sensitivity to (an)orexigenic factors to AN, necessitating further investigation.

Systemic hormones directly regulating food intake have been widely studied in AN patients (**Table 4**). However, some anorexigenic hormones such as leptin decrease while others, such as peptide YY3–36 (PYY3–36), increase. The same pattern is also observed for the orexigenic hormone, ghrelin (see *infra*). There exists sparse and contradictory data about the anorexigenic factors cholecystokinin (CKK) and glucagon-like peptide 1 (GLP1) in relation to this disease area rendering, it difficult to interpret observed

**Table 3 | Compared levels of neuropeptides regulating food intake in AN patients and healthy matched population.**


Adapted from Monteleone (2011).

\*Neuropeptides inhibiting food intake are on a gray background.

#### **Table 4 | Compared levels of hormones regulating food intake in AN patients and healthy matched population.**


<sup>1</sup>Most of the studies found decreased insulin levels.

<sup>2</sup>Only three studies found no significant differences when compared to control group, and one found a decrease, while all the other found increased adiponectin levels.

<sup>3</sup>Most of the studies found increased PYY levels.

\*Hormones inhibiting food intake are on gray background.

variations in the context of AN. Concerning PYY3–36, most studies reported increased levels; although an anorexigenic peptide, this increase is difficult to explain as PYY3–36 is normally released in response to food intake. On the contrary, around 50 studies relate a very low leptin blood level in AN patients compared to a healthy matched control population (**Table 4**). This endogenous signal of energy stores is positively correlated to body mass index. Leptin is considered to be a good predictor of growth hormone (GH) burst, cortisol, estradiol, and thyroid hormone levels and its receptor is widely distributed throughout the body suggesting a

pivotal role in mediating the hormonal adaptation to chronic starvation. Furthermore, AN patients display high plasma levels of adiponectin, another adipose-derived circulating cytokine. This anorexigenic hormone plays an important role in energy homeostasis and insulin sensitivity. The high levels of adiponectin in AN might contribute to the higher insulin sensitivity found in these patients. Indeed, insulin levels are usually strongly decreased that could be related to the hypoglycemia observed in AN patients.

More than 90% of adult women with AN are osteopenic, and almost 40% are osteoporotic at one or more sites (Grinspoon et al., 2000). Osteopenia and osteoporosis are frequent consequences of AN, that very often persist after weight gain. Moreover, as synthesized by Confavreux et al. (2009), "*bone can now be considered as a true endocrine organ secreting osteocalcin, a hormone pharmacologically active on glucose and fat metabolism. Indeed osteocalcin stimulates insulin secretion and* β*-cell proliferation. Simultaneously, osteocalcin acts on adipocytes to induce adiponectin, which secondarily reduce insulin resistance*." For these reasons, studies comparing bone turnover markers in AN patients with healthy control are presented in **Table 5**. Anorectic patients display increased levels of bone resorption markers and decreased bone formation markers. We may conclude that the bone mass alteration in patients with AN is dual: an increase of resorption and a decrease of bone formation. Moreover, the decrease in osteocalcin level could also contribute to the hypoinsulinemia and hypoadiponectinemia usually described.

A number of other endocrine disturbances have also been described in AN patients (**Table 6**). Hypothalamic-pituitaryadrenal axis deregulation is commonly suggested in AN. Indeed, AN is characterized by hypercortisolemia (**Table 6**) and, as mentioned by Miller (2011): "*overnight blood cortisol levels are inversely associated with bone mineral density and positively associated with severity of depression and anxiety symptoms in women with anorexia nervosa (Lawson et al., 2009). Therefore, hypercortisolemia may also contribute to the severe bone loss incurred and the highly prevalent psychiatric comorbidities in women with anorexia nervosa*." Hypothalamic amenorrhea is another characteristic feature of AN, and has been attributed to a state of severe energy deficitfrom restricted energy intake, increased energy expenditure or both. Women and adolescent girls with AN have lower levels of estradiol, luteinizing hormone (LH), and for some of them follicle stimulating

hormone (FSH; **Table 6**). The low levels of insulin-like growth factor-1 (IGF-1) and insulin may also contribute to this hypogonadal state and impact on bone turnover. The GH/IGF-1 axis is also altered in most of the studies (**Table 6**). Indeed, AN is associated with a nutritionally acquired hepatic resistance to GH with decreased production of IGF-1 and increased GH levels. Such an increase is due to a reduced feedback at the level of the pituitary and hypothalamus from low IGF-1 levels, and high levels of ghrelin (**Table 4**, see *infra*). Most studies report low levels of T3 and/or T4 thyroid hormones in patients with AN (**Table 6**). T3 and T4 plasma levels are enhanced by leptin administration in women with AN, and the levels of these three hormones are positively associated (Haas et al., 2005; Misra et al., 2005a). Moreover, ghrelin is known to inhibit the release of pituitary thyroid stimulating hormone (Wren et al., 2000), and studies indicate negative correlations between ghrelin and thyroid hormones plasma levels in AN (Misra et al., 2005c). These data suggest that low leptin and high ghrelin levels may contribute to lower thyroid hormone levels in AN. Finally, inflammatory cytokines were assayed in AN patients and matched healthy controls (Corcos et al., 2003). Data did not show significant variations suggesting that AN might not have an inflammatory component.

#### **ANIMAL MODELS OF CHRONIC FOOD RESTRICTION: A WAY TO DECIPHER THE PHYSIOLOGICAL MECHANISMS OF ANOREXIA NERVOSA**

The use of appropriate animal models mimicking most of the physiological changes occurring in AN might help to determine more precisely the potential mechanisms, central and/or peripheral, involved in the early adaptive state that precedes exhaustion


**Table 5 | Compared levels of bone turnover markers in AN patients and healthy matched population.**

<sup>1</sup>All but three studies found significant or non-significant decreased OC levels.

<sup>2</sup>All but three studies found significant or non-significant decreased BSAP levels

3 Increased CTX found by Caillot-Augusseau et al. (2000), Weinbrenner et al. (2003), Galusca et al. (2006), Ohwada et al. (2007), and Estour et al. (2010).

\*Bone resorption markers are on gray background.

<sup>4</sup>Only Grinspoon found increased urinary NTX/creatinine levels.

#### **Table 6 | Compared levels of other factors altered in AN patients.**


<sup>1</sup>All but two studies found increased GH levels.

<sup>2</sup>All but one study found decreased FSH.

<sup>3</sup>Most of the studies found decreased estrogen levels.

<sup>4</sup>Most of the studies found increased Cortisol levels.

<sup>5</sup>All but two studies found decreased T3 and/or T4 levels.

of the individual/animal. However, developing and using animal models of psychiatric disorders is inherently difficult due to the complex nature of these illnesses. In the literature, numerous models of genetic deficient mice for one or multiple genes involved in the regulation of feeding behavior/reward/energy balance have been developed (for review, see Siegfried et al., 2003; Kim, 2012). These genetic models give essential mechanistic data related to one specific pathway but do not completely mirror the symptoms observed in human disease (Willner, 1984; Smith, 1989). Indeed, the use of more "environmental models" that mimic most of the physiological symptoms of AN are preferred as they provide insight regarding how the disease might progress toward exhaustion. Initially, the most widely used animal model, whatever the species, is the chronic food restriction model (for review, see Kim, 2012). However, it does not take into account several conditions

observed in AN patients that are self-starvation, hyperactivity, and chronic stress. The rat model of self-starvation developed by Routtenberg and Kuznesof (1967) consists in housing one rat in a cage equipped with a running wheel and submitting the animal to a food restriction (1 h-feeding per day). This model later coined the term "the activity-based anorexia (ABA) model." It produces a rapid decrease in body weight and food intake, hyperactivity, hypothermia, loss of estrus, and an increase in HPA axis activity (Hall and Hanford,1954;Routtenberg and Kuznesof,1967;Burden et al., 1993). Moreover, in this model, the rats eat less than inactive rats fed with the same schedule, and can even starve themselves to death. In the ABA model, the long term exposure (few days) to low leptin and high ghrelin levels, induced a tissue-specific expression pattern of ghrelin and leptin receptors (Pardo et al., 2010). Furthermore, Verhagen et al. (2011) found that plasma ghrelin

levels are highly associated with food anticipatory behavior, measured by running wheel activity in rats. This effect is dependent of the central ghrelin signaling system *via* growth hormone secretagogue receptor 1a (GHS-R1A). In many aspects this model mimics numerous physiological alterations observed in AN. However, as specified by Klenotich and Dulawa (2012), the ABA paradigm is strongly dependent upon factors that amplify or reduce some parts of the phenotype, i.e., the choice of rodent strain (more or less resistant to ABA, Gelegen et al., 2007), the sex of animal (female are more vulnerable to ABA), the age (younger animals are more susceptible to ABA), the temperature (increasing the temperature to 32˚C strongly reduces the ABA behavior, Cerrato et al., 2012), the time of the day the animals receive food, and the hydration/dryness of the food. In fact,Boakes and Juraskova, 2001; Boakes, 2007) demonstrated that the "self-starvation" observed in ABA rats might reflect both the reduced palatability of the dry chow for a dehydrated animal and satiety signals from a stomach full of water. Thus, giving hydrated food during the 1 h-feeding schedule completely abolishes the ABA phenotype (rapid weight loss, hyperactivity, etc.). Currently, we are developing an adaptation of the ABA model in female mice that aims to follow the long term (more than 2 weeks) physiological alterations induced by a combination of physical activity and 50% food restriction. Our recent results, involving a 2-week protocol, indicate that our selected ABA model induces a rapid and stable loss of weight, changes the circadian locomotor activity, alters energy balance (corresponding to the passage from phase I to II; **Table 1**), induces hypoglycemia, hypoleptinemia, hyperghrelinemia, and a central alteration in the hypothalamic feeding centers (Méquinion, personal data). Thus, the combination of exercise (in running wheels) and food deprivation following a certain schedule might be considered as two important chronic stress factors that could reinforce the weight loss by modifying feeding behavior, as observed in our study. Other models based on chronic stress that are associated with food deprivation use tail pinching, cold swimming (Shimizu et al., 1989; Wong et al., 1993), or separation. We choose this last model, named "Separation-Based Anorexia" (SBA) to study the impact of chronic stress associated with caloric insufficiency. In our protocol, 8-week old female mice are separated and fed with a time-restricted food access for up to 10 weeks. Our recent results showed a 20–25% body weight loss with a cumulative food intake just under that of the control group. Moreover, SBA mice displayed reduced lean, fat, and bone masses associated with hypoleptinemia and alteration in GH/IGF-1 axis. Finally, an alteration of the estrous cycle was also observed (Zgheib, personal data).

To date, these described "environmental animal models" (ABA and SBA) remain the only models that enable long term studies of how chronic food restriction impacts upon physiology at different levels (energy balance, reproduction, bone/fat regulation, etc.), and of the mechanisms responsible for the sustenance of these alterations on different tissues often not available on patients (brain, bones, fat, muscle, liver, etc.). They will also facilitate determination of whether the dramatic outcome of the patients might be related to a specific dysregulation of one or many biological factors that can be considered as a marker of the disease and potentially also in its evolution. In addition, ABA and SBA models exhibit good face validity for most of the physiological symptoms of AN.

# **GHRELIN A KEY ENERGY BALANCE HORMONE: ROLE IN ANOREXIA NERVOSA**

#### **ORIGIN AND BIOSYNTHESIS OF GHRELIN**

Ghrelin is a 28 amino acid, initially isolated from rat stomach (Kojima et al., 1999). The preproghrelin messenger RNA (mRNA) is mainly expressed in the X/A-like oxyntic gland cells of the gastric fundus mucosa equivalent to P/D1 cells in humans (Bordi et al., 2000). Ghrelin is also produced in other parts of the gastrointestinal tract, and it is expressed at lower levels in pancreas, kidney, testis, placenta, and bone (Gnanapavan et al., 2002; González et al., 2008) and hypothalamic neurons (Cowley et al., 2003). The 117 amino acid preproghrelin is processed by cleavage and results in two peptides (**Figure 1**): obestatin and proghrelin (Jeffery et al., 2005). Des-acyl ghrelin is then cleaved from the 94 amino acid proghrelin precursor by enzymes like the prohormone convertase 1/3 (Zhu et al., 2006). This 28 amino acid peptide is modified post-translationally in the active acylated form of ghrelin, capable to bind to its receptor, the GHS-R1a. The octanoylation at the third N-terminal amino acid, usually serine (Kojima et al., 1999), is catalyzed the enzyme ghrelin octanoyl-acyltransferase (GOAT, Yang et al., 2008),which is expressed predominantly in the stomach, gut, and pancreas, but also at other sites (Kang et al., 2012). Ghrelin concentrations in blood comprise principally des-acyl ghrelin (85– 90%) and in lesser amounts acyl ghrelin (10–15%) and C-terminal proghrelin peptides (Pemberton and Richards, 2007).

#### **GHRELIN RECEPTOR AND DISTRIBUTION**

Ghrelin is the only known ligand to bind to GHS-R1a (Howard et al., 1996; Gutierrez et al., 2008). This receptor belongs to the G-protein coupled receptor family (GPCR; Holst and Schwartz, 2006) and has two variants, GHS-R1a and GHS-R1b, which are splice variants of the same gene. Type 1a is the full length, seventransmembrane domain receptor, and the type 1b isoform is a C-terminally truncated, five-transmembrane domain variant (Kojima et al., 1999). Only the GHS-R1a is fully functional, binding mostly with acylated ghrelin on Gαq-protein, whereas the 1b isoform is thought to be physiologically inactive. It should be noted that des-acyl ghrelin does not compete with acyl ghrelin for GHS-R1a to any significant extent. Indeed, supraphysiological concentrations of des-acyl ghrelin are necessary to allow binding and activation of GHS-R1a (Veldhuis and Bowers, 2010; Delhanty et al., 2012). Although derived from the same precursor, obestatin is a cognate ligand for the orphan receptor GPR39, another member of the ghrelin receptor subfamily (McKee et al., 1997; Holst et al., 2004).

Growth hormone secretagogue receptor 1a is abundantly expressed within the CNS. Notably, a large population of GHS-R1a-expressing neurons are located in the hypothalamic arcuate nucleus (ARC), which has a crucial role in energy balance control. Other hypothalamic areas expressing this receptor of relevance for feeding control include the ventromedial hypothalamus (VMH), paraventricular nucleus (PVN), anteroventral preoptic nucleus, anterior hypothalamic area, lateral hypothalamic area (LHA), suprachiasmatic nucleus, supraoptic nucleus, and the tuberomammillary nuclei (Guan et al., 1997; Gnanapavan et al., 2002; Camiña, 2006; Harrold et al., 2008). Moreover, mRNA studies also demonstrate the presence of this receptor in

limbic and mesolimbic structures known to be involved in motor control, emotional reactivity and reward/motivation systems such as the hippocampus (dentate gyrus, CA2, and CA3 regions), pars compacta of the substantia nigra, ventral tegmental area (VTA), raphe nuclei, laterodorsal tegmental nucleus (Guan et al., 1997; Zigman et al., 2006), and amygdala (Alvarez-Crespo et al., 2012). In the periphery, GHS-R1a is expressed in different tissues and organs implicated in energy balance, e.g.,in the anterior lobe of the pituitary on somatotroph cells (Briggs and Andrews, 2011), pancreas, spleen, stomach, intestine, heart, thyroid, gonads, adrenal, liver, skeletal muscle, and adipose tissues (Papotti et al., 2000). The truncated form is also found in various tissues but its exact role is not well known (Gnanapavan et al., 2002; Camiña, 2006). Interestingly, GHS-R1a is constitutively active even when unstimulated by the afferent ghrelin signal (Petersen et al., 2009). Finally, ghrelin receptors are able to interact with other receptors to form homoor hetero-dimers, like GHS-R1a/GHS-R1a and GHS-R1a/GHS-R1b (Chow et al., 2012). Co-expression of the truncated variant of ghrelin receptor with the full length variant attenuated the

described to be a GH-secretagogue. Beside an obvious role in the regulation

constitutive signaling probably because the translocation of the ghrelin receptor from the plasma membrane to the cell nucleus is decreased (Mokrosinski and Holst, 2010). GHS-R1a also appears to heterodimerize with other GPCRs, at least in *in vitro* test systems, and can explain the differential ghrelin signaling (review in Schellekens et al., 2010). Heterodimerization of GHS-R1A with the dopamine receptor D2 has recently been shown to occur in mouse hypothalamic neurons that regulate appetite (Kern et al., 2012) Since GHS-R1a acts as an allosteric modulator of D1 and D2 signaling, this finding implies a functional role for the expressed GHS-R1a in brain areas that may be less accessible to peripherally produced ghrelin and where there is no local production of ghrelin (Jiang et al., 2006; Kern et al., 2012).

antagonistic effect to ghrelin (see Delhanty et al., 2012).

#### **ROLES OF GHRELIN**

In line with the broad expression of ghrelin and its receptor, this hormone is involved in multiple biological functions, many of which are linked to feeding control. Initially, this gutbrain signal was shown to have direct pituitary GH-releasing

effects, reproducing the known effects of the so-called growth hormone secretagogues (GHS). This term refers a group of synthetic GHS-R1a ligands, the first group of which was derived from metenkephalin (described by Bowers et al., 1977) and included the hexapeptide GHRP-6 (Bowers et al., 1984) that is now recognized as a synthetic ghrelin mimetic. Both ghrelin and its receptor have been strongly conserved during evolution, supporting the notion that GHS-R1a and its natural ligand play a fundamentally important role in biology (Palyha et al., 2000).

Ghrelin's most characterized effects are: (i) its ability to stimulate GH secretion, likely of relevance for glucose homeostasis and energy balance; (ii) its role as an orexigenic hormone acting at key hypothalamic and midbrain circuits involved in feeding control; (iii) its involvement in various other physiological functions like gastrointestinal, cardiovascular, pulmonary and immunefunction, sleep duration, learning, memory, and behavior, cellular proliferation, immunomodulation, reproduction, and bone physiology. Most of these physiological functions are altered in AN indicating a potential role for ghrelin in the pathogenesis of this disease.

#### **Role of ghrelin in the regulation of appetite, food intake, and energy balance**

Ghrelin acts at different levels to stimulate GH secretion and thus modulate hepatic IGF-1 production (Peino et al., 2000). GHS-R1a is expressed by GHRH arcuate neurons but also by GH cells in the anterior pituitary gland (Kojima et al., 1999). Ghrelin and GHS activate ARC cells (Dickson et al., 1993; Hewson and Dickson, 2000) including neuroendocrine cells in this region (Dickson et al., 1996), notably a sub-population of GHRH neurons (Dickson and Luckman, 1997; Osterstock et al., 2010). Indeed, these compounds acts in synergy with GHRH to induce a greater GH release than would be induced by GHRH alone (Bowers et al.,1984; Arvat et al., 2001; Hataya et al., 2001). This ghrelin-stimulated GH release is dose-dependent and could explain why AN patients have elevated circulating GH levels (Kojima and Kangawa, 2005; Miljic et al., 2006; Misra et al., 2006; Germain et al., 2007, 2010; Estour et al., 2010). Chronic starvation is associated with GH resistance and relatively low IGF-1 levels involving a feedback mechanism, rather than body composition parameters or other circulating factors, e.g., free fatty acid or insulin levels (Støving et al., 1999; Brick et al., 2010). Besides its role in growth, GH stimulates lipolysis through a mechanism independent of IGF-1 (Fazeli et al., 2010a), for which the effects are largely anabolic. The increased GH secretion and the reduction of IGF-1 in starvation may be adaptive since they respectively serve the function of mobilizing fat stores in the setting of reduced energy availability and reduce anabolism. However, the reduction of IGF-1 levels may also have deleterious effects, contributing to bone, and muscle loss in AN women. Even if the mechanisms underlying the development of GH resistance in states of chronic undernutrition are not as well established, ghrelin might strongly participate to such endocrine dysregulation. Studies conducted in rodents support a close link between ghrelin signaling and altered GH/IGF-1 status. In rodents, fasting induced higher expression of GHS-R1a and overexpression of GHS-R1a in female mice provoked higher expression of GH and GHRH (Veldhuis and Bowers,2010). Furthermore, suppressed ghrelin signaling (using antisense RNA knockdown against GHS-R1a or a GHS-R1a

antagonist, BIM-28163), caused a decrease in GH peak pulsatility with or without a decrease in plasma levels of IGF-1 (Zizzari et al., 2005; Veldhuis and Bowers, 2010). Finally, plasma ghrelin concentration is negatively correlated with body weight and subcutaneous, visceral, and total adiposity (reviewed in Veldhuis and Bowers, 2010), probably due to a long term effect on ghrelin in driving pulsatile GH secretion, which is strongly lipolytic.

Ghrelin is perhaps best known as a circulating hunger signal necessary for meal initiation and meal anticipation with a secretion occurring in a pulsatile manner starting with a preprandial rise and postprandial fall 1 h after food intake (Ariyasu et al., 2001; Cummings et al., 2001; Tschöp et al., 2001; Zizzari et al., 2011; Merkestein et al., 2012). Moreover, whatever the route of injection, ghrelin increases food intake both in humans and animals (Wren et al., 2001a,b). In addition, prolonged food reduction or severe caloric restriction causes an increase in plasma ghrelin concentration (Wren et al., 2001b; Méquinion, personal data). In AN patients, ghrelin levels are increased up to twofold and return to normal levels after weight restoration (Otto et al., 2001, 2005; Tolle et al., 2003; Germain et al., 2009; Yi et al., 2011). However, it appears that fluctuations in ghrelin are not always influenced by food intake in AN (Germain et al., 2009) suggesting impairment in its regulation, probably due to a chronic adaptation to long term food restriction (Yi et al., 2011). Mice lacking the gene for ghrelin or its receptor have normal food intake when fed chow, probably due to compensatory adaptation during embryonic development, but show a degree of protection from obesity when fed a high-fat diet, especially from an early age (Sun et al., 2003, 2004; Wortley et al., 2004, 2005).

Ghrelin also appears to be of importance in the regulation of lipid and glucose metabolism. It has been attributed a role in the maintenance of normal blood glucose levels (Grove and Cowley, 2005). There are indications that ghrelin's effects on the GH axis may have relevance for glucose homeostasis (and even survival) during chronic food deprivation. Mice lacking acyl ghrelin (due to knockout of GOAT) lose glycemic control and become moribund by 1 week of 60% food deprivation, an effect that can be circumvented by infusion of either acyl ghrelin or GH throughout this 1-week period (Zhao et al., 2010). The work of Sun et al. (2008) on ghrelin homeostasis in ghrelin KO and GHS-R KO mice demonstrate that the ghrelin/GHS-R pathway appears to play an important role in glucose homeostasis by regulating insulin sensitivity and glucose sensing, particularly under conditions of negative energy balance. However, data indicating an action of ghrelin on plasma insulin levels are still controversial (Castañeda et al., 2010; Sangiao-Alvarellos and Cordido, 2010). Intravenous ghrelin injection leads to a decrease in plasma insulin and an increase in blood glucose (Broglio et al., 2001). This result is not found universally and it may be related to the physiological vs. pharmacological doses used (Castañeda et al., 2010). Ghrelin could potentially decrease insulin secretion by altering insulin sensitivity (Castañeda et al., 2010). This is in agreement with the results obtained during insulin and glucose tolerance tests performed in AN patients (Broglio et al., 2004a; Harada et al., 2008). Moreover, under chronic food restriction,fatty acids are mobilized and their oxidation could increase the production of octanoic acid, thereafter used to octanoylate des-acyl ghrelin, leading to a global

increase of plasma ghrelin levels. Thus, the greater levels of glucose observed in ABA mice (Méquinion, personal data) or wild type mice during the early phases of the chronic food restriction might be driven by this increase in ghrelin. This might contribute to the adaptive state in the first stages of AN, before a depletion of the supply of free fatty acids and induction of ketosis. Ghrelin acts directly on the liver, to favor glycogenolysis but also on muscle and adipose tissues. Indeed, subcutaneous injection of ghrelin in rats induces an increase of hepatic triglycerides associated with an increase in the gene expression of enzymes involved in lipogenesis like acetyl-CoA carboxylase (ACC) and fatty acid synthase (FAS, Barazzoni et al., 2005) as well as increased expression of ACC and FAS mRNA in white visceral adipose tissue (Thompson et al., 2004; Barazzoni et al., 2005). By contrast, subcutaneous ghrelin injection induces a decrease of triglyceride content in muscle without modifying ACC expression and AMPK phosphorylation (Barazzoni et al., 2005). These effects are especially observed in the gastrocnemius muscle, a fast-twitch muscle that is predominantly glycolytic. Similar variations are observed in the food-restricted condition (Samec et al., 2002). Finally, Pardo et al. (2010) describe, in the ABA model, a tissue-specific expression pattern of GHS-R1a receptors in visceral and subcutaneous fat and within the muscle. Indeed, the oxidative-soleus type of muscle appears to be more susceptible to circulating ghrelin levels than the glycolytic-gastrocnemius type under exercise and food restriction situations. All of these modifications could provide a defense mechanism to maintain energy homeostasis in the unbalanced energy state that is found in AN patients.

#### **The central orexigenic effect of ghrelin**

Appetite, food intake, and energy balance are finely tuned by a complex intercommunication between neural networks and peripheral tissues (**Figure 2**). Within the CNS, various hypothalamic nuclei containing orexigenic and anorexigenic neurons regulate the different facets of food intake. The ARC cells targeted by ghrelin and its mimetics include the orexigenic neuropeptide Y (NPY) cells (Dickson and Luckman, 1997) that co-express another orexigenic peptide, agouti-related peptide (AgRP). Following administration of ghrelin, these neurons are activated, reflected by the induction of Fos-protein in discrete cell groups (Hewson and Dickson, 2000; Wang et al., 2002), by increased action potential firing (Cowley et al., 2003; Andrews et al., 2008) and by an increased expression of NPY and AgRP mRNA (Kamegai et al., 2000, 2001; Nakazato et al., 2001). Furthermore, the stimulatory effects of ghrelin on NPY/AgRP neurons are complemented by a reduction of the ARC anorexigenic pro-opiomelanocortin (POMC) neuronal activity *via* inhibitory GABA-ergic inputs from NPY/AgRP neurons (Cowley et al., 2003). Interestingly, sensitivity of the ARC cells to ghrelin appears to be nutritionally regulated as the Fos response was increased up to threefold in fasting rats relative to fed animals (Hewson and Dickson, 2000) an effect that was reversed once again upon refeeding (Luckman et al., 1999). Collectively, these data indicate that ghrelin activates a key orexigenic pathway in the hypothalamic ARC, the NPY/AgRP cells and that this response is metabolically regulated. Consistent with this, stimulation of hypothalamic GHS-R1a results in an anabolic response characterized by an increase in food intake (Wren et al., 2000) and

a decrease in energy utilization (Tschöp et al., 2000). The feeding effects ghrelin appear to require normal NPY/AgRP signaling since ablation of NPY or AgRP neurons or the use of NPY receptor antagonists abolish these effects (Chen et al., 2004; Luquet et al., 2007). Conditional deletion of NPY/AgRP co-expressing neurons in the ARC of adult mice (by targeting the human diphtheria toxin to the AgRP locus) caused a rapid starvation to death (Luquet et al., 2005). Moreover, mice homozygous for the anorexia (*anx*) mutation, characterized by poor food intake and death by 3–5 weeks after birth (Maltais et al., 1984) display a lower density of hypothalamic neuropeptides. The data of Nilsson et al. (2011) support the hypothesis of degeneration of hypothalamic ARC neuron populations; the AgRP system appears to be the first system affected and the POMC system being secondary in this process. Finally, in our models of chronic food restriction (SBA and ABA), we found an alteration of the AgRP signal with an accumulation of this peptide in ARC neurons (Méquinion and Nilsson, personal data). Thus, there are numerous lines of evidence supporting the fact that in chronic food restriction (and probably in AN), a dysregulation of the AgRP system occurs that contributes to deficient ghrelin signaling at the level of the ARC. Recent neuroimaging data obtained from AN patients differing in disease duration showed a significant reduction of total white matter volume and focal gray matter atrophy in various brain areas such as the hypothalamus, especially in patients with shorter food restriction. Collectively, these studies highlight the potential role of endocrine and central (hypothalamic) dysfunction in the altered homeostatic metabolic status in AN, as described in animal models (Boghi et al., 2011).

At the hypothalamic level, ghrelin has also been reported to act directly or indirectly on other nuclei linked to feeding control such as theVMH, PVN, and LHA (López et al.,2008;Mano-Otagiri et al., 2009; Lamont et al., 2012). Although not coupled to c-fos expression, VMN cells exhibit a robust electrical response following bath application of a ghrelin agonist (Hewson et al., 1999). The elegant study of López et al. (2008) showed that, in fasted rats, an elevated ghrelin tone was associated with an increased activation of hypothalamic AMPK and a decreased mRNA expression of enzymes (like FAS) involved in the *de novo* fatty acid biosynthesis only in the VMH. They concluded that the energy peripheral signals sensed to regulate fatty acid metabolism in the hypothalamus and consequently the feeding behavior, may not be a nutrient, but ghrelin through an action at the level of the VMH.

Ghrelin's effects in the PVN are also likely linked to feeding control as direct intra-PVN injection of ghrelin induces a robust feeding response that is coupled to neuronal (c-fos) activation (Olszewski et al., 2003). Consistent with this, a reduction of GHS-R1a gene in the PVN using RNA interference in rats, significantly reduced body weight and blood ghrelin levels without affecting food intake (Shrestha et al., 2009). These data reflect a role for ghrelin in the modulation of PVN neuron activity in that is linked to energy homeostasis, but the mechanisms of action remains to be elucidated.

In the LHA, ghrelin is thought to mediate hyperphagia through orexin neurons. Indeed, central administration of ghrelin or a ghrelin mimetic induces Fos expression in orexin-containing, but not melanin-concentrating hormone-containing, neurons

**FIGURE 2 | Action of ghrelin in the brain.** Ghrelin acts at different levels of the brain to stimulate food intake via hypothalamus and meso-cortico-limbic pathway. In the hypothalamus, ghrelin activates orexigenic neurons (AgRP/NPY), which inhibit anorexigenic neurons (POMC/CART) via GABA projections. They are connected to second order neurons like CRH and TRH neurons located in the PVN and/or the orexin neurons found in the LHA. POMC/CART neurons activate MCH neurons. Ghrelin acts also at different levels of meso-cortico-limbic pathway: LDTg, VTA, and Acc. Ghrelin acts directly on VTA to stimulate dopamine release in Acc. dopamine release is controlled by cholinergic LDTg neurons. Ghrelin could also act on NTS to

stimulate the food intake via either vagal nerve or area postrema, see **Figure 4**. Acc, accumbens nucleus; ACh, acétylcholine; AgRP, agouti-related peptide; ARC, arcuate nucleus; CART, cocaine- and amphetamine-regulated transcript; CRH, corticotropin-releasing hormone; DA, dopamine; DYN, dynorphin; ENK, enkephalin; GABA, γ-aminobutyric acid; GHRH growth-hormone-releasing hormone; GLU, glutamate; LDTg, laterodorsal tegmental area; LHA, lateral hypothalamic area; MCH, melanin-concentrating hormone; NPY, neuropeptide Y; NTS, nucleus tractus solitarius; POMC, pro-opiomelanocortin; PVN, paraventricular nucleus; TRH, thyrotropin releasing hormone; VMH, ventromedial nucleus; VTA, ventral tegmental area.

(Lawrence et al., 2002; Olszewski et al., 2003), and activates glucose responding neurons (Chen et al., 2005a) in this area. Furthermore, ghrelin-induced feeding is suppressed in orexin KO mice (Toshinai et al., 2003). The role of orexin neurons to simulate feeding behavior (appetite/metabolism) is now well established although they are also especially important for sleep and wakefulness (España and Scammell, 2011; Gao, 2012) and play important roles in the stress response, in analgesia and reward/addiction (see Kukkonen, 2013). Moreover, ghrelin's effects to increase the reward value of a high-fat diet appear to involve a LHA-VTA orexin pathway (Perello et al., 2010).

The action of ghrelin to increase food intake and associated appetitive behaviors involves an integrated neurobiological response exerted at many levels, not only via the hypothalamus. For example, structures located in the caudal brainstem also express GHS-R1a. In particular, ghrelin receptors are found in all three components of the dorsal vagal complex with a highest expression within the area postrema, a moderately dense signal in the nucleus of the solitary tract and a low density signal in the dorsal motor nucleus of the vagus (Zigman et al., 2006). Peripheral injection of ghrelin and, prior to its discover, ghrelin mimetics, induces c-Fos induction in the nucleus of the solitary tract and area postrema (Bailey et al., 2000; Lawrence et al., 2002). The effects of ghrelin on food intake/behavior are similar when injected into the fourth as for the third ventricle, in terms of the amount of food eaten, the number of meals and meal size during the first few hours after treatment (Faulconbridge et al., 2003). The effects of ghrelin in the dorsal vagal complex might be more related to autonomic effects such as on the cardiovascular system. However, since the hypothalamus is strongly connected with the nucleus solitary tract, we cannot exclude an indirect effect of the ghrelin to hypothalamic structures through an activation of brainstem areas, although appears not to include a noradrenergic pathway (Bailey et al., 2000; see *infra*).

#### **Ghrelin and the reward system**

Besides the homeostatic ghrelin sensitive pathways, ghrelin also appears to target mesolimbic circuits linked to reward. Hedonic (non-homeostatic) brain pathways are also involved in feeding control and are modulated by circulating energy balance signals such as ghrelin, thereby influencing the evaluation of the pleasure derived from the taste, smell, or texture of food. Many regions of the corticolimbic and mesolimbic brain are thus involved in learning,memory, emotion, and reward processing associated with food. Among these complex feeding networks (for review, see Van Vugt, 2010; **Figure 3**), the VTA-Acc (dopaminergic) pathway plays a pivotal role in conferring reward from a wide range of reinforcers, from chemical drugs of abuse to natural rewards such as food (Nestler, 1996; Corwin et al., 2011). Using optogenetic techniques, Adamantidis et al. (2011) demonstrated that phasic activation dopaminergic VTA neurons is associated with reward-predicting cues and facilitates the development of positive reinforcement during reward-seeking and behavioral flexibility. As reviewed in Carr (2011), burst firing of VTA dopaminergic neurons may operate as a "teaching signal." For example, in the case of food intake, when rats are presented with a highly palatable food for the first time, this triggers dopamine release in the Acc (shell), whereas repeated exposure to the same palatable food blunts the dopamine response despite avid consumption. Interestingly, food restriction has been described to sustain the Acc (shell) dopamine release in this model. Moreover, simply delivering cues

linked previously to the food reward can be sufficient to reinstate the Acc (shell) dopamine response, indicating transference of the dopamine response from the reinforcer to the cue (for review, see Volkov et al. (2011). Ghrelin has emerged as one important modulator of the VTA-NAcc (dopaminergic) reward pathway (for review, see Skibicka and Dickson, 2011). More than 50% of the dopaminergic VTA neurons express GHS-R (Zigman et al., 2006), although it is also expressed on other cell types in this area (Abizaid et al., 2006). Whether administered peripherally, into the brain ventricles or into the VTA, ghrelin administration triggers a robust Acc dopamine response (Abizaid et al., 2006; Jerlhag et al., 2006, 2007) that is accompanied by an increased feeding response in rodents (Naleid et al., 2005; Egecioglu et al., 2010) and an increase in food-motivated behavior (Skibicka and Dickson, 2011; Skibicka et al., 2012). Central ghrelin signaling, *via* GHS-R1A, appears to be important not only for food reward (Egecioglu et al., 2010), but also for the reward associated with artificial rewards like alcohol, cocaine, and amphetamine (Wellman et al., 2005; Jerlhag et al., 2009, 2010). As an example, the locomotor stimulating effect of cocaine is decreased in ghrelin KO mice compared to their wild type littermates (Abizaid et al., 2010). Furthermore, in rodents, ghrelin elevates the motivation to obtain high-sugar or high-fat reward (Perello et al., 2010; Skibicka and Dickson, 2011; Skibicka et al., 2012). In particular, both peripheral and central injections of ghrelin augment the food-motivated behavior of a satiated rat to get sugar whereas blockade of ghrelin signaling reduces the operant responding of an hungry rat to the level of a satiated rat (Skibicka et al., 2012). These data strongly support the involvement of ghrelin in behaviors related to food reward. Thereby one can suggest that ghrelin could be considered as a key internal cue, available during period of energy deficit to motivate adequate food intake behavior (Skibicka and Dickson, 2011). These data reinforce the now well-documented role of ghrelin in food reward, considering that the shell region of the Acc is described to process unpredicted rewards and motivational states to reinforce food intake behaviors, but also use of drugs of abuse. However, the action of circulating ghrelin upon GHS-R1A-expressing cells in the VTA, located mostly on dopaminergic neurons, begs the question of how this hormone reaches deep brain structures that are far away from circumventricular organs (see *infra*). Finally, one study shows that presence of food is necessary to induce dopamine release (Kawahara et al., 2009). Chronic ghrelin treatment also modifies the expression of dopaminergic receptors in Acc, more specially D1 and D3 (Skibicka et al., 2012), which are described to be involved in obesity, food reward (D1 receptors) and inhibition of reward behavior (D3 receptors). Similar data were obtained from human imaging studies that emphasize the role of ghrelin in food reward. Indeed, intravenous ghrelin injection to human subjects increases activity in brain areas involved in the evaluation of the reward value attributed to food and food cues including the striatum, amygdala, insula, and orbitofrontal cortex (Malik et al., 2008).

#### **Other functions**

Almost 6000 articles have been published on ghrelin, since its discovery in Kojima et al. (1999) and it is not surprising that its biological effects extend beyond feeding and energy balance,

There is now increasing evidence that ghrelin stimulates motor activity in the gastrointestinal tract (gastric motility and emptying). In fact, ghrelin shares high homology degree with motilin (Kojima and Kangawa, 2005), a hormone released by endocrine cells of duodenum and jejunum duringfasting and which increases gastric motility after feeding (Sanger, 2008). Ghrelin has been described to use both central and local pathways to exert its effects on the gut through receptors located on vagal afferents, in the nodose ganglion and myenteric plexus. In fact in normal rodents, central pathways are operational whereas after vagotomy, ghrelin is able to exert effects *via* the myenteric plexus (see review of Peeters, 2003). Moreover, in mice, central administration of ghrelin accelerates gastric emptying (Asakawa et al., 2001) and changes the excitability of neurons located in the PVN identified as receiving ascending afferent signals from mechanoreceptors in the stomach (Zhao et al., 2003). Clinically, it has been reported that the intravenous administration of ghrelin accelerates the rate of gastric emptying and induces gastrointestinal contraction in healthy volunteers (Fujitsuka et al., 2012). Since the plasma levels of ghrelin are high in AN patients, one can hypothesize an alteration of the signaling both at central and peripheral levels that may worsen the outcome of the patients. These findings suggest that ghrelin could provide a therapeutic target for disorders related to gastrointestinal discomfort.

Surprisingly, ghrelin is also involved in sleep-wakefulness regulation. Indeed, experiments conducted in adult male rats demonstrate that repeated intravenously administrations of ghrelin stimulate wakefulness, decrease slow-wave sleep, and reduce the duration of rapid eye movement sleep (Tolle et al., 2002). This action could involve a reduction in the release of acetylcholine from the dorsal tegmental nucleus (LDTg) on neurons expressing somatotropin release-inhibiting factor known to indirectly regulate the rapid eye movement sleep periods (Tolle et al., 2002). In fact, among the various neurochemical systems involved in wakefulness (acetylcholine, norepinephrine, dopamine, serotonin, histamine), the hypothalamic orexigenic neurons are crucial promoters of wakefulness since deficiency in the orexin system leads to disorders such as narcolepsy (Modirrousta et al., 2005). The status of activity in orexin neurons is closely related with the nutritional and behavioral state of animals. Moreover, Lamont et al. (2012) observed that both GHS-R and ghrelin KO mice had fewer orexinimmunoreactive cells than their wild type littermates. Their data support the synergistic relationship between ghrelin and orexin in the coordination of metabolism, reward and arousal to adopt the adapted behavior for food seeking and restoration of energy deficiency. In humans, AN patients exhibit sleep disorders. As an example, AN adolescents have an increase in wakefulness after sleep onset, a fragmentation of sleep as well as a reduction of slow-wave sleep and slow-wave activity during their total sleep time (Lauer and Krieg, 2004; Nobili et al., 2004). Even if deepening of nocturnal sleep follows a partial weight restoration, the neurobiological mechanisms linking starvation, mood disorders, and sleep disturbance remain to be elucidated.

The impact of ghrelin on anxiety behaviors remains controversial: studies show an anxiolytic effect under caloric restriction or after subcutaneous ghrelin injection (Lutter et al., 2008) while anxiogenic effects are observed in others (acute) studies with intracerebroventricular or intraperitoneal ghrelin injection (Asakawa et al., 2001; see reviewChuang and Zigman, 2010). Interestingly, chronic central ghrelin treatment was found to increase anxiety-like behavior in rats (Hansson et al., 2011). Recently, one study investigating the amygdala as a target for ghrelin found that acute ghrelin injection at this site elicits behaviors consist with a reduction in anxiety-like behavior, but only in rats that were not allowed access to food during the initial hour after injection. It was concluded that ghrelin, acting at the level of the amygdala, may provide an especially important signal to suppress anxiety-like behaviors that would otherwise prohibit the animal from finding food (Alvarez-Crespo et al., 2012). It is not yet known whether the ghrelin system regulates anxiety behavior associated with AN. One study found that a SNP in the preproghrelin gene was associated with panic disorder in a small patient group (Hansson et al., 2011).

Among the disorders described in AN, osteopenia/osteoporosis is also one major problem that cause long term outcomes with in particular a strong increase of the bone fracture incidence (Lucas et al., 1999). Ghrelin has and ghrelin mimetics have been shown to increase bone mineral density (Svensson et al., 2000; Fukushima et al., 2005; Delhanty et al., 2006) by a mechanism that appears to include the promotion of both proliferation and differentiation of osteoblasts (cells involved in bone formation), involving GHS-R1a and GHS-R1b receptors (Fukushima et al., 2005; Delhanty et al., 2006). The etiopathogenesis of bone disease in AN is complex and multifaceted. Indeed, the low bone mineral density (Legroux-Gerot et al., 2005, 2008; Legroux-Gérot et al., 2007; Estour et al., 2010) is usually linked with alteration of multiple factors (**Tables 3**–**6**) that are thought to contribute to the "uncoupling" of bone turnover, leading to increased bone resorption, and decreased bone formation (see Howgate et al., 2013). However, it has been demonstrated that ghrelin affects bone metabolism by operating in an autocrine/paracrine mode, independent of the GH/IGF-1 axis (see Nikolopoulos et al., 2010). Weight recovery is associated with partial recovery of bone mineral density. There is currently no approved effective therapy that completely reverses the bone mineral density deficit. The most convincing results were obtained with a treatment of recombinant human IGF-1 alone or in combination with the oral contraceptive pills (see Misra and Klibanski, 2011). The link between ghrelin and estrogen on bone metabolism is always matter of debate even if it is established that ghrelin suppresses pulsatile LH and FSH pulsatility (Meczekalski et al., 2008; Kluge et al., 2012).

Another criterion used to characterize AN patients is amenorrhea. Indeed, in negative energy balance conditions like in AN, the increase of plasma ghrelin is associated with decrease of LH secretion (**Table 6**). Evidence is mounting that ghrelin may operate as a pleitropic modulator of gonadal function and reproduction (Tena-Sempere, 2008; Muccioli et al., 2011; Repaci et al., 2011). Notably, most of the actions of ghrelin upon the reproductive axis reported to date are inhibitory. Ghrelin can suppress not only LH, but also FSH secretion in male and female rats (Fernández-Fernández et al., 2005; Martini et al., 2006). Such effects are also

described in humans (Kluge et al., 2012). Centrally, ghrelin exerts a predominant action directly at the level of the GnRH pulse generator by inhibiting directly GnRH release (Fernández-Fernández et al., 2005; Muccioli et al., 2011) or by an indirect modulation of other neuronal pathways. For example,Forbes et al. (2009)recently showed the ability of ghrelin to decrease Kiss1 mRNA expression in the medial preoptic area. Given the importance of the kisspeptin system to control the reproductive axis, these data provide new hypothesis for ghrelin-induced suppression of pulsatile LH secretion. Once again,in the AN, ghrelin might dynamically mediate the suppressive effect of energy deficit on the onset of puberty, gonadal function, and fertility. Here, the effects of ghrelin on the gonadal axis might protect females in a condition of strong energy insufficiency to develop a reproductive behavior that can be deleterious for her and her progeny.

Finally, ghrelin is also involved in other physiological functions that are more or less affected in AN like cardiovascular function or immune system. Among the cardiovascular effects, this hormone improves left ventricular contractility and cardiac output in healthy humans (Enomoto et al., 2003; Tesauro et al., 2010) and lowers blood pressure in mice concomitantly with a decrease in sympathetic nerve activity that is not caused by a direct action on blood vessels (Callaghan et al., 2012). In AN, the neuroendocrine alterations are also accompanied by autonomic dysfunctions like lower blood pressure values, lack of circadian variation of blood pressure and bradycardia (Oswiecimska et al., 2007). Thus, ghrelin might participate in AN to the cardiovascular complications observed in AN (Casiero and Frishman, 2006; Jáuregui-Garrido and Jáuregui-Lobera, 2012), but no studies currently display any correlation between these cardiovascular risks and the high plasma levels of ghrelin.

The role of ghrelin on the immune system remains unclear. However, Taub (2008) describes its implication in the regulation of immune factors, by inhibiting inflammatory cytokine production, more specifically in mediating anti-inflammatory effects on IL-1, TNF-α, and IL-6 cytokine expression by T-cells and mononuclear cells *via* GHS-R, and promoting thymic function. In AN, data related to the evaluation of circulating pro-inflammatory or inflammatory cytokines or in adipocytes are still a matter of debate and, as underlined by Nova and Marcos (2006), "controversial findings have been published regarding some aspects of the immune system that are otherwise impaired in more typical types of malnutrition."

**Des-acyl ghrelin and obestatin: a controversial metabolic function?**

Concerning des-acyl ghrelin, its role in food intake has been much debated. The recent paper of Delhanty et al. (2012) gives numerous arguments supporting des-acyl ghrelin as an hormone that can be metabolically active, when co-administrated with acyl ghrelin, by counteracting the effects of acyl ghrelin on insulin secretion and glucose metabolism. Des-acyl ghrelin appears to be increased inAN patients (Harada et al.,2008;Germain et al.,2009). Kojima et al. (1999) showed that des-acyl ghrelin was not able to bind to GHS-R1a. Although the des-acyl ghrelin receptor remains unknown, the increasing data suggesting that des-acyl ghrelin is a biologically active molecular, indicate that a dedicated receptor may exist.

Several studies show controversial effects of des-acyl ghrelin on food intake that are either inhibitory (Asakawa et al., 2005; Chen et al., 2005b) or stimulatory (Toshinai et al., 2006) in rodents. These results can be due to the different methods used like the type of injection, the dose used, the time of injection (light or dark period), the nutritional status, fed, or fasted. However, the overexpression of des-acyl ghrelin in a transgenic mouse model results in a small phenotype, associated with a reduction of food intake and body fat mass, reduced IGF-1 plasma levels without significant changes in circulating GH and also higher des-acyl ghrelin with no change in total ghrelin plasma levels (Ariyasu et al., 2005). In these mice, no significant differences have been noticed for glycemia and insulinemia (Ariyasu et al., 2005; Asakawa et al., 2005) while studies other studies have shown that des-acyl ghrelin inhibits glucose release *in vivo* and *in vitro* (Broglio et al., 2004b; Gauna et al., 2005; Qader et al., 2008). Moreover, it appears that ghrelin and des-acyl ghrelin do not have the same blood concentration in systemic and in portal circulations suggesting that liver could be involved in ghrelin regulation (Goodyear et al., 2010).

Concerning lipid metabolism, *in vivo* studies showed that desacyl ghrelin as well as ghrelin increase bone marrow adipogenesis in rat shinbone (Thompson et al., 2004) and both forms enhance lipid accumulation in visceral tissue in humans (Rodríguez et al., 2009). The mechanisms involved remain unclear. Similarly, acute or chronic des-acyl ghrelin injections in adult male rats cause an inhibition of LH secretion like ghrelin (Martini et al., 2006). By contrast, transgenic mice overexpressing des-acyl ghrelin do not display any changes in LH and FSH levels (Ariyasu et al., 2005).

Des-acyl ghrelin may also have a role in gastric motility. Indeed, intracerebroventricular or intravenous injections alter motor activity in the antrum with a decrease of antrum activity only in fasted rats (Chen et al., 2005b). Mice overexpressing des-acyl ghrelin exhibit a decrease in gastric emptying (Asakawa et al., 2005). Other studies are necessary to understand the mechanisms involved since vagotomy does not disrupt the (intravenous) des-acyl ghrelin effect (Chen et al., 2005b).

Studies about the effects of des-acyl ghrelin on the cardiovascular system are rare. Nevertheless, like acyl ghrelin, it promotes bradycardia and hypotension (Tsubota et al., 2005). Moreover, ghrelin and des-acyl ghrelin display vasodilator effect and no inotropic effects when they are applied on human artery *in vitro* (Kleinz et al., 2006).

Similarly, concerning obestatin, it remains again an open question whether this peptide is a physiologically relevant peptide to regulate energy homeostasis, food intake and gastric motility (Gourcerol et al., 2007). Obestatin binds to GRP 39, a receptor of the same subfamily than ghrelin receptor, to decrease food intake and body weight in an opposite manner to ghrelin (Stengel et al., 2009;Hassouna et al., 2010; Mokrosinski and Holst, 2010;Veldhuis and Bowers, 2010). Subsequent studies failed to show activation of this receptor and only few studies have reproduced the obestatin effects under specific conditions. Such results should be interpreted with caution since variations are observed according to the kits and conditions applied for obestatin assays (Hassouna et al., 2010). Due to their potential functions, it should be interesting to measure ghrelin/obestatin ratio to better understand their roles in the alteration of energy balance. It seems that AN affects obestatin

blood levels with a lower ghrelin/obestatin ratio in AN patients of restrictive type compared to constitutional thin women (Germain et al., 2009, 2010). Moreover, other functions are attributed to this hormone such as the inhibition of thirst, gastric motility, cell survival, pancreatic hormone secretion, sleep, thermoregulation, memory, and anxiety (Szentirmai et al., 2009; Hassouna et al., 2010; Veldhuis and Bowers, 2010).

#### **ACCESS OF GHRELIN TO ITS NEURONAL TARGETS**

To dynamically report energy homeostasis alterations and ensure an appropriate neuronal response, blood-borne ghrelin must rapidly access the central nervous system. Intriguingly, the physiological mechanisms controlling the access of ghrelin to its neuronal target remain currently debated. Indeed, although a central origin of ghrelin has been described (Cowley et al., 2003), it is now recognized that blood-derived ghrelin is able to target neuronal networks within the central nervous system to regulate energy homeostasis. However, it remains unclear how this key energy status-signaling hormone can rapidly access sensory neurons to alter feeding responses. Ghrelin mainly targets neurons located in the ARC where different blood/brain interfaces have been described. The blood–brain barrier is one such interface and one of the best described in the hypothalamic nuclei as in all other regions of the brain. The blood–brain barrier is located on brain capillaries where endothelial cells are tightly apposed by continuous tight junctions that prevent the free passage of molecules through the paracellular pathway. For circulating factors to access to the brain through the blood–brain barrier endothelium requires transcellular transport. Many studies have investigated the transport of circulating ghrelin across the blood–brain barrier. Banks et al. (2002) demonstrated the existence of ghrelin saturable transport system in mice from the brain to the blood but transport into the brain was much less pronounced. Remarkably, human ghrelin, which differs from mouse ghrelin by 2 amino acids, can be transported in both directions in mice. So, although receptor-mediated transport of ghrelin cannot be excluded, uptake mechanisms of this peptide remain unclear. Moreover, the efficiency of this blood– brain barrier remains to be studied in a chronic caloric restriction context. Improved CNS penetration during fating is one possible mechanism to explain the threefold increase in the number of cells expressingfos after peripheral ghrelin injection infasted vs. fed rats (Hewson and Dickson, 2000). The role of another blood/brain interface that is materialized by fenestrated vessels is also to consider. Indeed, these vessels are part of the blood-CSF barrier that is mostly described in the median eminence, the circumventricular organ adjacent to the ARC (Mullier et al., 2010). Median eminence vasculature differs from typical brain vessels as they harbor a fenestrated endothelium that lacks tight junction complexes. These structural characteristics and the presence of various blood-derived molecules in the median eminence and the other circumventricular organs parenchyma suggest high permeability of this specific vasculature (Broadwell et al., 1983; Ciofi, 2011; Morita and Miyata, 2012). Permeable vasculature can be found in the external part of the median eminence forming pituitary portal capillary plexus that displays some long intrainfundibular loops spreading into the median eminence parenchyma. Interestingly, fenestrated vessels are also found within the ARC with a

higher density into the ventromedial ARC where they are bordered by NPY expressing neurons (Ciofi et al., 2009). These data give support to the access of ghrelinfrom the circulation to the ventromedial sensory neurons *via* median eminence/ARC fenestrated vasculature (**Figure 4**).

The action of ghrelin on food intake may not be only due to its action on hypothalamus. Indeed, an indirect role of ghrelin on hypothalamic structures through an activation of brainstem areas has been suggested. Indirect pathway may occur through the binding of ghrelin to gastric vagal afferent neurons (Date et al., 2002). However, the expression of GHS-R1a within the dorsal vagal complex supports a direct action of ghrelin to brain parenchyma. Among this complex, the area postrema is a circumventricular organ that characteristically present fenestrated vasculature. These vessels may be responsible for the diffusion of ghrelin and its delivery to the dorsal vagal complex that communicates with hypothalamic control centers (**Figure 4**).

Fenestrated vasculature could represent a direct vascular route while allowing passive diffusion of peripheral molecules into the hypothalamus and the area postrema. This route may be responsible of acute regulation and complete chronic feedback accomplished by uptake of circulating molecules *via* receptor-mediated transport across the blood–brain barrier.

**FIGURE 4 | Access of ghrelin signal to its neuronal targets.** This schema summarizes the three hypothetic access routes of ghrelin toward its neuronal targets (cf **Figure 2**). First, ghrelin would be able to target neuronal networks thanks to specific transcellular transports at the level of blood–brain barrier (BBB) located on brain capillaries (purple arrows). Most ghrelin sensitive areas present blood–brain barrier vasculature and this route represent the main one described in all regions of the brain. However free-BBB regions, called the median eminence and the area postrema, are recorded in the hypothalamus and the brainstem respectively. These areas contain a rich fenestrated vasculature, which could represent a direct vascular route while allowing passive diffusion of peripheral ghrelin (red arrows). This route may be responsible of acute regulation and complete chronic feedback accomplished by uptake of circulating molecules via

receptor-mediated transport across the blood–brain barrier. Finally, activation of brainstem areas by ghrelin may occur without the entrance of ghrelin in the brain, but through its binding to gastric vagal afferent neurons (orange). Acc, accumbens nucleus; ACh, acétylcholine; AgRP, agouti-related peptide; AP, area postrema; ARC, arcuate nucleus; CART, cocaine- and amphetamine-regulated transcript; CRH, corticotropin-releasing hormone; DA, dopamine; DYN, dynorphin; ENK, enkephalin; GABA, γ-aminobutyric acid; GHRH, growth-hormone-releasing hormone; GLU, glutamate; LDTg, laterodorsal tegmental area; LHA, lateral hypothalamic area; MCH, melanin-concentrating hormone; ME, median eminence; NPY, neuropeptide Y; NTS, nucleus of the solitary tract; POMC, pro-opiomelanocortin; PVN, paraventricular nucleus; TRH, thyrotropin releasing hormone; VMH, ventromedial nucleus; VTA, ventral tegmental area.

The transport of ghrelin through the blood/brain interfaces has been poorly investigated in metabolic disorders excepted in obesity where few data are available. Banks et al. (2008) showed an inverse relation between body weight and ghrelin access to the brain suggesting that physiological states influence the rate at which ghrelin is transported across the blood/brain interfaces. A better understanding of the access of ghrelin to its neuronal target may leads to novel therapeutic interventions.

#### **CONCLUSION: GHRELIN AS A POTENTIAL TREATMENT FOR ANOREXIA NERVOSA**

In restrictive AN, the high plasma levels of ghrelin, even adaptive in view of its main role in meal initiation, let us to hypothesize a potential insensitivity to this endocrine signal both peripherally and centrally in association with this disease. It should be also mentioned thatAN patients often cannot increase theirfood intake not only because of fear of obesity, but also because of chronic or recurrent abdominal discomfort, fullness, and chronic constipation, functions in which efficient ghrelin participation is required. Injections of exogenous ghrelin have been shown to increase the adiposity and to stimulate appetite in healthy individuals and cancer patients (Peino et al., 2000; Wren et al., 2001a; Neary et al., 2004).

For AN patients, only a few preliminary studies have been performed to examine the effects of ghrelin administration (Miljic et al., 2006; Hotta et al., 2009; Ogiso et al., 2011). In these clinical studies, the mode of administration was different leading to different outcomes. In the study of Miljic et al. (2006), a singledose continuous administration of ghrelin for 5 h failed to affect appetite in all AN patients treated. Hotta et al. (2009) report that intravenous administration of ghrelin in anorectic patients twice a day for 14 days improves epigastric discomfort or constipation and increases the hunger score, which is related to gastric emptying. An increase of the body weight was obtained, from 1.5 to 2.4 kg, and daily energy intake during ghrelin infusion increased by 12–36% compared with the pretreatment period. Nutritional parameters such as total protein and triglyceride levels improved. These findings suggest that ghrelin may have therapeutic potentials in AN patients who cannot gain weight because of gastrointestinal dysfunction. Further studies are need to elucidate the potential impact of ghrelin by itself or agonists to ameliorate the outcomes of the AN patients.

In animals, several attempts have been made (see the other chapter submitted by Hassouna et al. "Actions of agonists and antagonists of the ghrelin/GHS-R pathway on GH secretion, appetite, and c-Fos activity"). Recently, Costantini et al. (2011) detailed an unexpected effect of GSK1614343, a novel ghrelin receptor antagonist with no partial agonist properties, that

#### **REFERENCES**


induced both in rats and dogs an increase of food intake, body weight, and reduced the POMC mRNA levels in the hypothalamus of rats chronically treated with the compound. Although it may seem counterintuitive to consider the use of ghrelin antagonists in this disease area, one study in rodents suggested that suppressed ghrelin signaling reduces (food anticipatory) hyperlocomotor activity in the ABA model (Verhagen et al., 2011). Barnett et al. (2010) demonstrated that the administration of a GOAT inhibitor improved glucose tolerance and decreased weight gain in wild type mice but not in ghrelin KO mice. Such treatment also led to decreased serum levels of acyl ghrelin without any effect on serum levels of des-acyl ghrelin. Even if this GOAT inhibitor might be expected to pave the way for clinical targeting of GOAT in metabolic diseases such as obesity and diabetes mellitus, one can suggest that, a GOAT activator, which has not been reported yet, may be a potential new treatment of AN if it improves food intake and body weight gain in patients with AN.

Finally, in the future it will be possible to assess more precisely the exact contribution of ghrelin, des-acyl ghrelin, and obestatin in the evolution of the AN both in humans and adequate animal models (Cardona Cano et al., 2012). The roles played by these peptides in feeding behavior, adaptation to starvation, reward mechanisms, emotional behavior, and stress responses in animals and humans led them to be potential therapeutic targets for AN treatments. The way and the mode of administration remain to be further determined and clarified. Neuroimaging studies have reported reduced brain volumes affecting both ventral and dorsal neural circuit dysfunctions in AN patients, with altered metabolisms of serotonin and dopamine that are closely associated to ghrelin, contributing to their puzzling symptoms (Kaye et al., 2009; Brooks et al., 2011). It will be important to determine whether the ghrelin signal reaches its central targets leading, as it is observed for leptin in obesity, to a « ghrelin-resistance »or a « transient ghrelin-insensitivity ». More investigations are needed to better suppress the neuronal activity of ghrelin signaling and to identify the specific pathways that may underlie the deleterious behaviors in patients suffering from AN. Investigation of the ghrelin peptide system will open up a new window of research for tackling psychosomatic disorders beyond the gastrointestinal tract, particularly restrictive AN and obesity/metabolic syndrome, two disorders at the extreme of the body weight continuum.

#### **AUTHOR NOTE**

Cross reference with another chapter of this issue: *Actions of agonists and antagonists of the ghrelin/GHS-R pathway on GH secretion, appetite and c-Fos activity* by Rim Hassouna, Alexandra Labarthe, Philippe Zizzari, Catherine Videau, Michael Culler, Jacques Epelbaum, and Virginie Tolle (Hassouna et al., 2013).

interrogation of dopaminergic modulation of the multiple phases of reward-seeking behavior.*J. Neurosci.* 31, 10829–10835.

Alguacil, L. F., Salas, E., and González-Martín, C. (2011). Identification of new drug targets and biomarkers related to obesity and eating disorders: an approach based on reward deficit and addiction. *Curr. Pharm. Des.* 17, 462–470.

Alvarez-Crespo, M., Skibicka, K. P., Farkas, I., Molnár, C. S., Egecioglu, E., Hrabovszky, E., et al. (2012). The amygdala as a neurobiological target for ghrelin in rats: neuroanatomical, electrophysiological and behavioral evidence.

*PLoS ONE* 7:e46321. doi:10.1371/ journal.pone.0046321


of circadian rhythm and unresponsiveness to short-term refeeding. *Eur. J. Endocrinol.* 138, 415–420.


psychopathological aspects. *Psychoneuroendocrinology* 31, 131–136.


bone resorption in anorexic patients. *Clin. Endocrinol. (Oxf.)* 52, 113–121.


by CRF type 2 receptors to disrupt fasted stomach motility in conscious rats. *Gastroenterology* 129, 8–25.


Weigle, D. S. (2001). A preprandial rise in plasma ghrelin levels suggests a role in meal initiation in humans. *Diabetes* 50, 1714–1719.


anorexia nervosa. *J. Psychiatry Neurosci.* 34, 323–329.


anorexia nervosa, bulimia nervosa and nonspecific eating disorders correlate with the body mass index but are independent of the respective disease. *Clin. Endocrinol. (Oxf.)* 46, 289–293.


of mice. *Eur. Neuropsychopharmacol.* 17, 199–205.


weight recovery. *Am. J. Clin. Nutr.* 81, 889–896.


short-term weight restoration. *Mol. Psychiatry* 2, 330–334.


of intact and degraded ghrelin and their responses to glucose infusion in anorexia nervosa. *J. Clin. Endocrinol. Metab.* 89, 5707–5712.


cholinergic systems in mice: implications for its involvement in brain reward. *Addict. Biol.* 11, 45–54.


in women with anorexia nervosa. *Metab. Clin. Exp.* 59, 473–477.


mineral density and fat mass. *J. Clin. Psychiatry* 72, 1546–1551.


Minn.: a population-based study. *Am. J. Psychiatry* 148, 917–922.


(GPR38 and GPR39) related to the growth hormone secretagogue and neurotensin receptors. *Genomics* 46, 426–434.


anorexia nervosa and effects on bone metabolism. *J. Clin. Endocrinol. Metab.* 89, 4972–4980.


*Prog. Neuropsychopharmacol. Biol. Psychiatry* 30, 1117–1121.


production of delta sleep in anorectic adolescents. *Sleep* 27, 1553–1559.


with anorexia nervosa. *Endokrynol. Pol.* 61, 117–123.


Casanueva, F. F. (2010). Peripheral leptin and ghrelin receptors are regulated in a tissue-specific manner in activity-based anorexia. *Peptides* 31, 1912–1919.


(2008). Proghrelin-derived peptides influence the secretion of insulin, glucagon, pancreatic polypeptide and somatostatin: a study on isolated islets from mouse and rat pancreas. *Regul. Pept.* 146, 230–237.


gene expression on energy balance. *Peptides* 30, 1336–1341.


growth hormone (GH) secretion and increased process irregularity in women with anorexia nervosa: indirect evidence for disruption of feedback regulation within the GH-insulin-like growth factor I axis. *J. Clin. Endocrinol. Metab.* 84, 2056–2063.


al. (2003b). Fasting plasma ghrelin levels in subtypes of anorexia nervosa. *Psychoneuroendocrinology* 28, 829–835.


anorexia nervosa. *Psychoneuroendocrinology* 30, 364–372.


Van Vugt, D. A. (2010). Brain imaging studies of appetite in the context of obesity and the menstrual cycle. *Hum. Reprod. Update* 16, 276–292.

Veldhuis, J. D., and Bowers, C. Y. (2010). Integrating GHS into the ghrelin system. *Int. J. Pept.* 2010, 1–40.


K. (2006). Expression of ghrelin receptor mRNA in the rat and the mouse brain. *J. Comp. Neurol.* 494, 528–548.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 October 2012; paper pending published: 10 November 2012; accepted: 01 February 2013; published online: 26 February 2013.*

*Citation: Méquinion M, Langlet F, Zgheib S, Dickson S, Dehouck B, Chauveau C and Viltart O (2013) Ghrelin: central and peripheral implications in anorexia nervosa. Front. Endocrinol. 4:15. doi: 10.3389/fendo.2013.00015*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Méquinion, Langlet, Zgheib, Dickson, Dehouck, Chauveau and Viltart. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Activation of somatostatin 2 receptors in the brain and the periphery induces opposite changes in circulating ghrelin levels: functional implications

# *Andreas Stengel 1\* andYvette Taché2\**

<sup>1</sup> Division Psychosomatic Medicine and Psychotherapy, Department of Medicine, Obesity Center Berlin, Charité, Universitätsmedizin Berlin, Berlin, Germany <sup>2</sup> Digestive Diseases Division, CURE: Digestive Diseases Research Center, Center for Neurobiology of Stress andWomen's Health, Department of Medicine, VA Greater Los Angeles Health Care System, University of California at Los Angeles, Los Angeles, CA, USA

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Rafael Vazquez-Martinez, University of Cordoba, Spain Francisco Gracia-Navarro, University of Cordoba, Spain

#### *\*Correspondence:*

Andreas Stengel, Division Psychosomatic Medicine and Psychotherapy, Department of Medicine, Obesity Center Berlin, Charité, Universitätsmedizin Berlin, Luisenstr. 13a, 10117 Berlin, Germany. e-mail: andreas.stengel@charite.de; Yvette Taché, Digestive Diseases Division, CURE: Digestive Diseases Research Center, Center for Neurobiology of Stress and Women's Health, Department of Medicine, VA Greater Los Angeles Health Care System, University of California at Los Angeles, CURE Building 115, Room 117, 11301 Wilshire Boulevard, Los Angeles, CA 90073, USA. e-mail: ytache@mednet.ucla.edu

Somatostatin is an important modulator of neurotransmission in the central nervous system and acts as a potent inhibitor of hormone and exocrine secretion and regulator of cell proliferation in the periphery.These pleiotropic actions occur through interaction with five G protein-coupled somatostatin receptor subtypes (sst1−5) that are widely expressed in the brain and peripheral organs. The characterization of somatostatin's effects can be investigated by pharmacological or genetic approaches using newly developed selective sst agonists and antagonists and mice lacking specific sst subtypes. Recent evidence points toward a divergent action of somatostatin in the brain and in the periphery to regulate circulating levels of ghrelin, an orexigenic hormone produced by the endocrine X/A-like cells in the rat gastric mucosa. Somatostatin interacts with the sst2 in the brain to induce an increase in basal ghrelin plasma levels and counteracts the visceral stress-related decrease in circulating ghrelin. By contrast, stimulation of peripheral somatostatin-sst2 signaling results in the inhibition of basal ghrelin release and mediates the postoperative decrease in circulating ghrelin. The peripheral sst2-mediated reduction of plasma ghrelin is likely to involve a paracrine action of D cell-derived somatostatin acting on sst2 bearing X/Alike ghrelin cells in the gastric mucosa. The other member of the somatostatin family, named cortistatin, in addition to binding to sst1−<sup>5</sup> also directly interacts with the ghrelin receptor and therefore may simultaneously modulate ghrelin release and actions at target sites bearing ghrelin receptors representing a link between the ghrelin and somatostatin systems.

**Keywords: X/A-like cell, somatostatin receptor subtypes, ghrelin cell, somatostatin receptor agonists and antagonist, cortistatin**

"fendo-03-00178" — 2013/1/9 — 15:57 — page 1 — #1

#### **INTRODUCTION**

In 1972, Guillemin's group – while searching for additional releasing factors in hypothalamic extracts after their identification of thyrotropin-releasing hormone (TRH) – identified a novel negative regulator of pituitary somatotropic cells releasing growth hormone (GH; Brazeau et al., 1973). The cyclo peptide was named somatostatin (somatotropin release-inhibiting factor, SRIF), in keeping with its hypophysiotropic action (Guillemin, 2011). Somatostatin was found to be expressed in two biologically active isoforms: the tetradecapeptide somatostatin-14 (Brazeau et al., 1973) and the amino terminally extended octacosapeptide somatostatin-28 generated by differential post-translational processing from a common precursor molecule (Pradayrol et al., 1980). Thereafter, a flow of articles in rodents and humans established the ubiquitous distribution of somatostatin in various brain areas (Finley et al., 1981; Johansson et al., 1984; Uhl et al., 1985) and peripheral organs including the gastrointestinal tract (Costa et al., 1977; Walsh, 1994). This was followed by the identification and characterization of five distinct, high-affinity, specific somatostatin receptors (sst) encoded by five distinct genes (Gahete et al., 2010a). Structurally, these receptors belong to the so-called "superfamily" of membrane G protein-coupled (GPC) receptors. The sst1−<sup>5</sup> have distinct as well as overlapping patterns of distribution in the brain (Fehlmann et al., 2000; Spary et al., 2008) and gut (Schafer and Meyerhof, 1999; Ludvigsen et al., 2004; Corleto et al., 2006) with a prominent expression of sst2 in the gastrointestinal tract (Sternini et al., 1997). Studies using new pharmacological tools, namely selective sst subtype agonists and antagonists (Grace et al., 2006; Cescato et al., 2008; Erchegyi et al., 2008, 2009; **Table 1**) point toward the role of different sst1−<sup>5</sup> in mediating the large spectrum of somatostatin biological actions, mostly inhibitory in nature. Multiple effects of somatostatin can also result from the ability of sst to form both homodimers or heterodimers (sst5 with sst1 or dopamine D2 receptor, sst2a with sst3, D2 or μ opioid receptor subtype 1) resulting in the activation of different intracellular signaling cascades (Rocheville et al., 2000; Baragli et al., 2007; Siehler et al., 2008).

Of interest to neuroendocrinologists, somatostatin's inhibitory action on pituitary GH release was soon extended to a wide range of hypophyseal hormones including prolactin, thyrotropin (thyroid-stimulating hormone, TSH), and adrenocorticotropic hormone (ACTH; Brown et al., 1984; Bertherat et al., 1995;

#### **Table 1 | Structure and receptor binding affinity of somatostatin and somatostatin receptor agonists.**


Receptor affinities were derived from competitive radioligand displacement assays in cells stably expressing the cloned human receptor using <sup>125</sup>I-[Leu8DTrp22Tyr25] SST-28 (Grace et al., 2006, 2008; Cescato et al., 2008; Erchegyi et al., 2008, 2009) except for somatostatin-14 (Viollet et al., 1995), somatostatin-28 (Viollet et al., 1995), and cortistatin (Fukusumi et al., 1997).

"fendo-03-00178" — 2013/1/9 — 15:57 — page 2 — #2

Shimon et al., 1997) as well as a large number of hormones secreted from endocrine cells of the gastrointestinal tract including gastrin (Lloyd et al., 1997), cholecystokinin (CCK; Shiratori et al., 1991), secretin (Shiratori et al., 1991), gastric inhibitory peptide (GIP; Pederson et al., 1975), and neurotensin (Rokaeus, 1984), and from the pancreas glucagon (Marco et al., 1983; Strowski et al., 2000), insulin (Pederson et al., 1975; Marco et al., 1983; Strowski et al., 2000), and pancreatic polypeptide (PP; Marco et al., 1983). Studies to characterize the somatostatin receptor subtypes established that sst2 is primarily involved and, to a smaller extent, sst5 in the broad inhibitory effects of somatostatin on endocrine secretion (Lloyd et al., 1997; Strowski et al., 2000; de Heer et al., 2008). Of clinical relevance was also the compelling evidence that the majority of neuroendocrine tumors (NETs) expressed increased levels of sst2 leading to the development of radiolabeled peptides and peptide receptor radionuclides as diagnostic and therapeutic tools respectively for NETs (van der Hoek et al., 2010; Culler et al., 2011; Giovacchini et al., 2012; Jawiarczyk et al., 2012).

In the past decade, a significant breakthrough came from the identification of a 28-amino-acid octanoylated peptide, ghrelin in the endocrine cells of the gastric mucosa as the cognate ligand for the GH-secretagogue receptor 1a (GHS-R1a, later renamed ghrelin receptor, GRLN-R; Kojima et al., 2001; Davenport et al., 2005). Ghrelin's interaction with the GRLN-R represents a third independent regulatory pathway controlling the pulsatile release of pituitary GH besides the GH-releasing hormone (GHRH) and somatostatin (Kojima et al., 2001; Malagon et al., 2003; Kineman and Luque, 2007).

Consistent with ghrelin, being predominantly produced by the endocrine cells of the stomach (Ariyasu et al., 2001), recent studies also documented that gastrointestinal NETs released ghrelin (Corbetta et al., 2003; Tsolakis et al., 2004). Therefore the regulation of ghrelin release by somatostatin will be of clinical relevance. The present review focuses on recent evidence that somatostatin signaling in the brain and the periphery exerts opposite influence on circulating ghrelin levels. The sst receptor subtype(s) involved and functional relevance of somatostatin-induced altered ghrelin plasma levels in the context of orexigenic and gastric prokinetic actions of the peptide (Kojima and Kangawa, 2005) will also be addressed.

# **SOMATOSTATIN AND ITS RECEPTORS: EXPRESSION AND MODULATION OF FEEDING BEHAVIOR AND GASTROINTESTINAL MOTILITY**

#### **EXPRESSION OF SOMATOSTATIN AND sst**

Somatostatin is expressed throughout the brain except in the cerebellum (Finley et al., 1981; Johansson et al., 1984). Peptide distribution has been investigated by immunohistochemistry indicating the expression in the cortex, limbic system, central nucleus of the amygdala, sensory structures, hypothalamus (including the arcuate, ventromedial, and paraventricular nucleus), and the periaqueductal central gray (Finley et al., 1981; Johansson et al., 1984; Moga and Gray, 1985). Similarly, sst receptor mRNA expression

is widely detected in the rat brain including the deep layers of the cerebral cortex, bed nucleus of the stria terminalis, basolateral amygdaloid nucleus, medial amygdaloid nucleus, paraventricular thalamic nucleus, medial preoptic nucleus, dorsomedial and ventromedial hypothalamic nucleus, arcuate and paraventricular nucleus of the hypothalamus, substantia nigra, dorsal raphe nucleus, granular layer of the cerebellum, locus coeruleus, nucleus of the solitary tract, and the dorsal motor nucleus of the vagus nerve (Fehlmann et al., 2000; Spary et al., 2008). The wide distribution of the ligand along with the receptors is consistent with the pleiotropic action of the sst signaling systems.

Somatostatin is widely expressed in the gastrointestinal tract and the pancreas, namely in endocrine mucosal D cells scattered within the stomach, small and large intestine, and in δ-cells of the pancreatic islets of Langerhans (Walsh, 1994). In addition, somatostatin is also detected in neurons of the gut enteric nervous system within both the submucosal and myenteric plexus (Costa et al., 1977). Similar to the ligand, sst receptor subtypes are widely expressed in the gastrointestinal tract. In rats, mRNA expression of sst1, sst2, sst3, and sst4 is detected in the small and large intestine without any clear predominance to one segment (Schafer and Meyerhof, 1999). The sst1 is the predominant form in gastrointestinal tract of mice, whereas the sst5 was almost undetectable (Schafer and Meyerhof, 1999). In human colon, a differential expression has been reported with sst2 mRNA expression on circular smooth muscle cells and sst1−<sup>3</sup> mRNA on longitudinal muscle layer cells (Corleto et al., 2006). Interestingly, in the pancreas protein expression of all sst1−<sup>5</sup> receptor subtypes has been identified on every major endocrine cell type, namely α, β, δ, and PP cells with a differential expression pattern between species (Ludvigsen et al., 2004).

#### **BRAIN ACTIONS OF SOMATOSTATIN**

The first function assigned to somatostatin was the inhibition of GH release (Brazeau et al., 1973). Now somatostatin is recognized to exert several central extrapituitary actions such as the regulation of other pituitary endocrine hormones especially those responsive to stress, parasympathetic and sympathetic outflow, thermogenesis, visceral functions, and behaviors. Namely, intracerebroventricular (icv) injection of somatostatin-28 inhibited the tail suspension stress-induced rise of circulating ACTH (Brown et al., 1984). This effect was mimicked by the stable pan-somatostatin agonist, ODT8-SST but not by somatostatin-14 (Brown et al., 1984). The sst2 seems to play a pivotal role in these regulatory processes as sst2 receptor knockout mice display an increased ACTH release compared to their wild type littermates (Viollet et al., 2000).

With regard to the central actions of somatostatin to affect visceral functions, recent studies have focused on the gastrointestinal tract. ODT8-SST injected intracisternally (ic) in rats accelerated gastric emptying of a liquid non-nutrient solution. This effect is mimicked by somatostatin-28 and the sst5 preferring agonist, BIM-23052 but not somatostatin-14 or the agonists preferring sst1, CH-275, sst2, DC-32-87, sst3, BIM-23056, and sst4, L-803,087 (Martinez et al., 2000) indicating a prominent role of brain sst5 activation to induce a gastroprokinetic effect. This contention is further corroborated by the prominent mRNA expression of

sst5 in the dorsal motor nucleus of the vagus nerve (Thoss et al., 1995). Since sst receptors are also expressed in hypothalamic brain nuclei regulating colonic functions (Mönnikes et al., 1993; Mönnikes et al., 1994; Tebbe et al., 2005) including the arcuate (sst1−5; Fehlmann et al., 2000; Schulz et al., 2000) and paraventricular nucleus of the hypothalamus (sst2−4; Fehlmann et al.,2000; Schulz et al., 2000) and the locus coeruleus (sst2−4; Fehlmann et al., 2000; Schulz et al., 2000), several studies investigated whether activation of central somatostatin signaling influences colonic motor functions. Stressing mice by short exposure to anesthesia vapor and icv injection of vehicle robustly stimulated propulsive colonic motor function shown by a 99% increase in fecal pellet output (Stengel et al., 2011a). This effect was completely abolished by icv pretreatment with ODT8-SST, somatostatin-28 and the selective sst1 agonist, whereas sst2 or sst4 agonists or octreotide (**Table 1**) had no effect. These data suggest that activation of brain sst1 can prevent the stress-related stimulation of colonic propulsive motor function in mice (Stengel et al., 2011a).

Central somatostatin alters food intake with a divergent action depending on the doses used: an increase is induced by icv injection of lower (picomolar) doses and a decrease following high (nanomolar) in rats (Feifel and Vaccarino, 1990). The stable pan-somatostatin agonist, ODT8-SST (Erchegyi et al., 2008) icv stimulated food intake in rats under basal as well as already stimulated conditions during the dark phase with a rapid onset (during the first hour) and a long duration of action (lasting for 4 h; Stengel et al., 2010a). ODT8-SST's orexigenic action is sst2 mediated as it is reproduced by icv injection of the selective peptide sst2 agonist (Stengel et al., 2010d) and blocked by the selective peptide sst2 antagonist (Stengel et al., 2010a). This represents a central action since injected intraperitoneally at a 30-fold higher dose, the peptide did not influence food intake (Stengel et al., 2010d). The orexigenic effect of central sst2 stimulation observed in rats has been recently expanded to mice (Stengel et al., 2010c). A detailed analysis of the food intake microstructure using an automated food intake monitoring device showed that icv injection of the sst2 agonist increased the number of meals, shortened inter-meal intervals, and induced a higher rate of ingestion, whereas meal sizes were not altered (Stengel et al., 2010c). Collectively, these data indicate that brain activation of sst2 signaling pathways in rodents induces a rapid orexigenic response by decreasing satiety (number of meals), without influencing satiation indicated by normal meal sizes (Stengel et al., 2010c). The physiological role of brain sst2 signaling in modulating food intake is also supported by the decrease of nocturnal food intake induced by the peptide sst2 antagonist injected icv at the beginning of the dark phase (Stengel et al., 2010d). In addition, hypothalamic somatostatin shows a circadian rhythm with a peak in the early dark phase and a nadir in the early light phase (Gardi et al., 1999). Moreover, food restriction increases pituitary somatostatin release (Ishikawa et al., 1997) which could increase the drive to eat under these conditions.

#### **PERIPHERAL ACTIONS OF SOMATOSTATIN**

"fendo-03-00178" — 2013/1/9 — 15:57 — page 3 — #3

Contrasting to the central effects, somatostatin's actions in the periphery are largely inhibitory. In the stomach, somatostatin delays emptying of the food (Smedh et al., 1999) and inhibits gastric acid secretion which is mediated directly by an interaction with acid producing parietal cells but also via the reduced release of histamine from ECL cells and gastrin from G cells (Walsh, 1994). The acid-antisecretory action of somatostatin is no longer observed in sst2 knockout mice, indicative of a primary role of the sst2 (Piqueras et al., 2003). Similarly, in the small intestine somatostatin reduces intestinal peristalsis in cats, rabbits and rats, while stimulating the duodenal, jejunal, and ileal contractile response in dogs (Tansy et al., 1979). In line with the findings in rodents, somatostatin increases gastrointestinal transit time in humans (Gregersen et al., 2011). The effects on colonic motility are likely mediated by the sst1 and sst2 based on *in vitro* studies using circular and longitudinal human colonic smooth muscle cells (Corleto et al., 2006). In addition, somatostatin reduces visceral sensitivity with the sst2 playing a key role as indicated by visceral hypersensitivity to both mechanical and chemical stimulation in the jejunum of sst2 knockout mice (Rong et al., 2007). This finding is likely to be relevant in humans as well. Patients with irritable bowel syndrome injected subcutaneously with octreotide display an anti-hyperalgesic response as shown by the increased threshold of discomfort and pain using rectal barostat manometry compared to injection of placebo (Bradette et al., 1994; Schwetz et al., 2004).

# **GHRELIN AND ITS RECEPTOR: EXPRESSION AND PHYSIOLOGICAL OREXIGENIC AND PROKINETIC ACTIONS EXPRESSION AND REGULATION OF GHRELIN AND GHRELIN RECEPTOR**

Ghrelin bears a unique fatty acid (*n*-octanoyl) residue on the third amino acid which is essential for affinity and binding to the GRLN-R (Kojima et al., 1999; Kojima and Kangawa, 2005). Other dietary fatty acids of medium length can also serve as a direct source for the acylation of ghrelin (Nishi et al., 2005). The enzyme catalyzing the acylation of ghrelin was unknown for several years and just recently identified in mouse and human as a member of the membrane-bound *O*-acyltransferases (MBOATs), namely MBOAT4 which was subsequently renamed ghrelin-*O*acyltransferase (GOAT; Gutierrez et al., 2008; Yang et al., 2008). GOAT mRNA and protein are prominently expressed in rodent and human gastric mucosa in ghrelin expressing cells (Sakata et al., 2009; Stengel et al., 2010f). In addition, GOAT protein has been detected in rodent and human intestine, pancreatic duct, gallbladder, hypothalamus and pituitary gland (Gahete et al., 2010b; Lim et al., 2011; Kang et al., 2012), rodent plasma (Stengel et al., 2010f) and human visceral and subcutaneous adipocytes (Rodriguez et al., 2012) leading to the speculation of additional acylation sites of ghrelin.

Unlike ghrelin, desacyl ghrelin, which does not bear the hydrophobic residue on the third amino acid, is the main circulating form. The acyl:desacyl ghrelin ratio is 1:3 as recently reported using an optimized blood processing protocol to improve the yield of acylated ghrelin in rats (Stengel et al., 2009). Although desacyl ghrelin does not bind to and activate the GRLN-R (Kojima et al., 1999), recent studies indicate that the peptide exerts several biological actions to influence food intake (Stengel et al., 2010e), reduce inflammatory somatic pain (Sibilia et al., 2012), muscle cachexia produced by injury in rats (Sheriff et al.,2012) and basal autophagy

"fendo-03-00178" — 2013/1/9 — 15:57 — page 4 — #4

in human visceral adipocytes (Rodriguez et al., 2012). However, the understanding of the physiological function of this peptide is hampered by the fact that the desacyl ghrelin receptor mediating its effects is still to be identified.

Blood levels of ghrelin vary in relation with the meal pattern with an increase before meals and a decrease thereafter (Cummings et al., 2001; Tschöp et al., 2001a). In addition, fasting also increases ghrelin mRNA expression (Toshinai et al., 2001; Kim et al., 2003; Xu et al., 2009) and reduces ghrelin peptide content in the stomach (Toshinai et al., 2001; Kim et al., 2003) indicative of a stimulated production and release under conditions of food deprivation. Ghrelin levels are not only regulated by shortterm variations in energy status associated with meal patterns but also by long-term changes in body weight. Ghrelin plasma levels are elevated under conditions of reduced body weight such as anorexia nervosa or tumor cachexia and reduced in obesity (Tschöp et al., 2000, 2001b; Cummings et al., 2002). Similar to the ligand, the ghrelin acylating enzyme, GOAT is regulated by the metabolic status with an increased GOAT mRNA and protein expression in rodent gastric mucosa, hypothalamus, and pituitary following a 12- or 24-h fast (Gonzalez et al., 2008; Gahete et al., 2010b; Stengel et al., 2010f). Under conditions of obesity induced by high fat diet or leptin deficiency in *ob/ob* mice, a down-regulation of GOAT mRNA occurred in the mouse pituitary, unlike the stomach or hypothalamus (Gahete et al., 2010b), while patients with obesity-associated type 2 diabetes showed higher levels of GOAT in visceral adipose tissue (Rodriguez et al., 2012). This indicates a tissue specific regulation of GOAT under conditions of obesity.

# **OREXIGENIC AND PROKINETIC EFFECTS OF GHRELIN**

Ghrelin is well established to stimulate food intake in line with its regulation by changes in energy status in many species including humans (Wren et al., 2000; Tang-Christensen et al., 2004; Druce et al., 2005). It is so far the only known peripherally produced and centrally acting orexigenic peptide, contrasting with the numerous anorexigenic peptides in the gut (Suzuki et al., 2011). Ghrelin's action is blocked by pharmacological or genetic approaches using GRLN-R antagonists (Salome et al., 2009) and GRLN-R knockout mice (Sun et al., 2004; Zigman et al., 2005) indicating a key role of ghrelin-GRLN-R interaction in mediating the orexigenic response. The food intake stimulatory action can result from ghrelin crossing the blood–brain barrier and binding to GRLN-R expressed on food intake regulatory brain nuclei (Banks et al., 2002; Pan et al., 2006) or acting directly on vagal afferents which also bear the ghrelin receptor (Date et al., 2002; Sakata et al., 2003). The respective role of these pathways under nutritional changes is still to be delineated. In addition to the stimulation of food intake, ghrelin is also involved in the regulation of body weight inducing an increase of body weight following chronic infusion of the peptide. This occurs through combined actions of stimulating appetite along with increasing fat storage and reducing lipid mobilization (Tschöp et al., 2000; Strassburg et al., 2008; Davies et al., 2009). Further corroborating these findings, ghrelin and GRLN-R double knockout mice display an increased energy expenditure leading to a reduction of body weight (Pfluger et al., 2008) which, however, could not be reproduced with a single genetic deletion of

either ghrelin (Sun et al., 2003; Pfluger et al., 2008) or the GRLN-R (Pfluger et al., 2008). These differential phenotypes may reflect the functional relevance of the high constitutive activity of the GRLN-R (Damian et al., 2012) and also give rise to the speculation that additional ligands for the receptor may exist (Deghenghi et al., 2001a).

# **DIFFERENTIAL MODULATION OF GH RELEASE BY GHRELIN AND SOMATOSTATIN**

Ghrelin exerts endocrine actions opposite to somatostatin by stimulating anterior pituitary release of GH (Kojima et al., 1999; Yamazaki et al., 2002; Kojima and Kangawa, 2011), prolactin, and ACTH (Lanfranco et al., 2010). Somatostatin's GH inhibitory effect is mediated by the sst2 (Briard et al., 1997), sst5 (Saveanu et al., 2001) and also sst1 (Kreienkamp et al., 1999). The GH releasing effect of ghrelin is blunted by intravenous (iv) infusion of somatostatin in healthy volunteers (Di Vito et al., 2002) and was completely blocked in pig pituitary cells *in vitro* (Malagon et al., 2003). The GH releasing action of ghrelin is likely to not only result from inhibiting somatostatin release (Feng et al., 2011) but also from direct activation of GH release (Veldhuis et al., 2006). In addition, ghrelin and somatostatin antagonistically interact on hypothalamic arcuate cells to regulate the release of GHRH with an activation of these neurons following ghrelin and a reduction after application of somatostatin *in vitro* (Mori et al., 2010).

# **ACTIVATION OF BRAIN sst2 SIGNALING INCREASES BASAL AND PREVENTS VISCERAL STRESS-INDUCED SUPPRESSION OF CIRCULATING GHRELIN**

Based on the established centrally sst2-mediated orexigenic action of somatostatin (Stengel et al., 2010a,d), we further investigated whether changes in circulating ghrelin may play a role. We found that the pan-somatostatin peptide, ODT8-SST (**Table 1**) injected icv increased basal plasma acyl ghrelin levels in *ad libitum* fed rats (Stengel et al., 2010a). However, the rise was observed at 3 h postinjection and therefore unlikely to underlie the initial increase in food intake response to central ODT8-SST which occurred within the first hour. However, it may contribute to the sustained significant increase in cumulative food intake still maintained at 4 h after icv injection of ODT8-SST (Stengel et al., 2010a). Other studies showed that activation of brain sst2 receptor prevents the decline in circulating ghrelin induced by visceral stress. Abdominal surgery reproducibly decreased the fasting plasma levels of acyl and desacyl ghrelin with a rapid onset and long lasting effect in rats (Stengel et al., 2010b, 2011b,c). Such a response was completely prevented by the ic injection of ODT8-SST as monitored 50 min postsurgery at a time where the peptide induces a 31 and 46% rise in fasting circulating acyl and desacyl ghrelin, respectively in sham animals (Stengel et al., 2011b). In addition, the sst2 selective peptide agonist injected ic also prevented the postoperative decline in plasma levels of acyl ghrelin, whereas sst1 and sst4 agonists (**Table 1**) did not when tested under the same conditions (Stengel et al., 2011b).

Of interest was the demonstration that ic ODT8-SST or sst2 agonist in addition to preventing the decline in acyl ghrelin induced by abdominal surgery also blunted the postoperative suppression of food intake (Stengel et al., 2011b). However, the

"fendo-03-00178" — 2013/1/9 — 15:57 — page 5 — #5

observed prevention of declining circulating ghrelin levels is not the main factor for the restoration of the orexigenic response postsurgically. The peripheral blockade of the GRLN-R using the GRLN-R antagonist ([D-Lys3]-GHRP-6) injected intraperitoneally did not modify the food intake stimulating effect of ic ODT8-SST (Stengel et al., 2011b). ODT8-SST injected icv also restored gastric emptying inhibited by abdominal surgery to levels observed under basal conditions, an effect mimicked by the sst5 but not the sst1, sst2, and sst4 agonists in rats (Stengel et al., 2011b). Similar to the effect on food intake, this action was not mediated by acyl ghrelin as the intraperitoneal injection of the GRLN-R antagonist did not influence the gastroprokinetic action of ODT8-SST postsurgery (Stengel et al., 2011b).

Taken together, these data indicate that the central activation of sst2 increases basal or prevents the surgical inhibition of circulating levels of ghrelin and food intake while activation of sst5 restores postoperative gastric ileus in rats. It also points to a differential role of brain sst subtypes 2 and 5 in preventing the stress-related suppression of ghrelin release/food intake and gastric emptying, respectively. In addition, the restoration of suppressed circulating acyl ghrelin levels after surgery does not play a major role as underlying mechanisms through which ODT8-SST injected into the brain exerts its prokinetic and orexigenic effects. Other studies showed that activation of sst2 and sst5 receptors in the brain inhibits stimulated CRF release in the hypothalamus and acute stress-related CRF mediatedACTH release (Brown et al.,1984; Tizabi and Calogero, 1992; Saegusa et al., 2011; Tringali et al., 2012). Brain CRF acting on CRF receptors is involved in the stress-related decrease in feeding behavior, ghrelin secretion, and gastric emptying (Hotta et al., 1999; Sekino et al., 2004; Taché and Bonaz, 2007; Yakabi et al., 2011). Therefore, it may be speculated that central activation of sst2 and/or sst5 dampens hypothalamic CRF activated by abdominal surgery (Wang et al., 2011) which may have a bearing with preventing the reduction in feeding, gastric emptying, and circulating ghrelin induced by abdominal surgery (Stengel et al., 2011b).

# **ACTIVATION OF PERIPHERAL sst2 SIGNALING INHIBITS CIRCULATING GHRELIN**

In contrast to the rise in circulating ghrelin induced by central administration of somatostatin agonists, convergent *in vivo* and *in vitro* rodent studies established that somatostatin or the stable agonists octreotide and SOM230 act peripherally through the sst2 to reduce ghrelin release (Seoane et al., 2007; Iwakura et al., 2010; Lu et al., 2012) resulting in lower circulating levels (Shimada et al., 2003; Silva et al., 2005; de la Cour et al., 2007). Such a response is in line with the inhibitory effects of somatostatinsst2 on the endocrine secretion of other intestinal hormones (Pederson et al., 1975; Marco et al., 1983; Rokaeus, 1984; Shiratori et al., 1991; Strowski et al., 2000). Likewise, in humans, peripherally injected somatostatin (Broglio et al., 2002; Norrelund et al., 2002) and somatostatin agonists such as octreotide (Barkan et al., 2003) reduce circulating ghrelin in healthy subjects. Of relevance, chronic subcutaneous infusion of the sst2/sst3/sst5 agonist octreotide-induced suppression of ghrelin plasma levels is not subject to rapid desensitization in rats and is likely to be sst2 mediated based on the prominent sst2 mRNA expression in the rat stomach (Silva et al., 2005). Further support for a role of sst2 came from immunofluorescent double labeling studies detecting the protein expression of sst2 on ghrelin-producing X/A-like cells of the rat stomach (Stengel et al., 2011c) and similarly on human ghrelin-producing gastric mucosal P/D1 cells (Fischer et al., 2008). Moreover, the selective peptide sst2 agonist (**Table 1**) injected intravenously decreased circulating levels of acyl and desacyl ghrelin with a rapid onset (0.5 h) and a long duration of action (still visible at 2 h; Stengel et al., 2011c). Lastly, the selective peptide sst2 antagonist (**Table 1**) injected intravenously prevents the decline in circulating ghrelin induced by the stimulation of endogenous peripheral somatostatin in rats (Stengel et al., 2011c).

The physiological role of peripheral somatostatin-sst2 signaling in the regulation of ghrelin was investigated under conditions of stimulation of endogenous gastric somatostatin. Urethane is well established to stimulate gastric somatostatin mRNA expression and peptide release in rats (Yang et al., 1990). Under these conditions, plasma ghrelin levels are decreased and the selective peptide sst2 antagonist (**Table 1**) injected intravenously prevents the decline in circulating ghrelin induced by urethane (Stengel et al., 2011c). Convergent reports showed that abdominal surgery induces a rapid and sustained inhibition of circulating levels of ghrelin in rats (Stengel et al., 2010b, 2011b,c). Likewise, iv injection of the selective peptide sst2 antagonist (**Table 1**) blocked the abdominal surgery-induced decrease of plasma ghrelin at 0.5 h postsurgery (Stengel et al., 2011c). The peptide is likely to act through paracrine transmission since somatostatin positive D cells directly contact ghrelin immunoreactive X/A-like cells in the rat stomach (Shimada et al., 2003). Interestingly, following abdominal surgery, acyl ghrelin was reduced more rapidly compared to desacyl ghrelin which was associated with a reduction of gastric as well as plasma concentrations of GOAT (Stengel et al., 2011c). Since blockade of peripheral sst2 signaling restores circulating levels of ghrelin (Stengel et al., 2011c), these data collectively suggest that peripheral somatostatin may blunt gastric GOAT mRNA expression and thereby negatively affect the acylation of ghrelin. In primary pituitary cell cultures somatostatin was reported to reduce GOAT mRNA expression and somatostatin knockout mice showed higher GOAT mRNA expression in the pituitary gland than the wild type (Gahete et al., 2010b). Based on these findings, somatostatin may influence ghrelin signaling not only *via* a direct inhibition of secretion but also by modulating the ghrelin activating enzyme GOAT. Further support for a physiological inhibitory action of peripheral somatostatin on ghrelin signaling came from somatostatin knockout mice that displayed an increased gastric ghrelin expression and higher circulating ghrelin levels compared to their wild type littermates (Luque et al., 2006a). These data indicate that endogenous somatostatin exerts a physiological inhibitory tone on gastric ghrelin synthesis and release.

During the past years, several clinical studies described ghrelinproducing NETs (Papotti et al., 2001; Volante et al., 2002; Taal and Visser, 2004) including six insulinomas, gastrinomas, vasoactive intestinal polypeptide (VIP)omas, non-functioning tumors (Volante et al., 2002) and as part of the multiple endocrine neoplasia type 1 (MEN-1; Iwakura et al., 2002; Raffel et al., 2005;

"fendo-03-00178" — 2013/1/9 — 15:57 — page 6 — #6

Ekeblad et al., 2007). Moreover, ghrelinomas have been identified to originate from the stomach or pancreas and were associated with very high ghrelin levels (Corbetta et al., 2003; Tsolakis et al., 2004). Growing evidence from clinical reports indicates that in addition to the surgical, chemotherapeutic, alpha-interferon, and local radiation treatments, the use of somatostatin or somatostatin analog is an established treatment option for gastroenteropancreatic NETs (Pavel and Wiedenmann, 2011). Based on the evidence described above, the determination of the sst subtype expressed on those ghrelinoma tumor cells followed by the use of peripherally acting selective sst compounds may result in an even more targeted approach.

# **INTERACTION OF CORTISTATIN WITH GHRELIN SIGNALING**

In 1996, the discovery of a new peptide sharing 11 of its 14 amino acids with somatostatin-14 was named cortistatin based on its predominant cortical expression and ability to depress cortical activity (de Lecea et al., 1996). Despite the chemical structure homology with somatostatin, both peptides are derived from distinct genes (de Lecea et al., 1997b). Similar to pro-somatostatin, processing of cortistatin precursor generated two mature products, cortistatin-14 and -29 in rodents and cortistatin-17 and -29 in humans (Fukusumi et al., 1997; Spier and de Lecea, 2000). Cortistatin is widely expressed in the brain, namely in the cortex and hippocampus, and although regional overlap exists, the distribution pattern differs from that of somatostatin (de Lecea et al., 1997a). Likewise, the peripheral expression pattern of cortistatin (e.g., in adrenal, thyroid, and parathyroid gland, testis, pancreas, kidney, lung, liver, stomach, ileum, jejunum, colon, endothelial, and immune cells) does not fully match that of somatostatin (Papotti et al., 2003; Dalm et al., 2004; Xidakis et al., 2007). Consistent with being a close somatostatin endogenous analog, cortistatin contains the FWKT tetramer crucial for sst binding, and therefore displays high-affinity (1–2 nM) to all five sst subtypes where the peptide acts as an agonist (Fukusumi et al., 1997; Siehler et al., 2008). However, emerging evidence indicates that cortistatin induces distinct central and peripheral effects that differ from those exerted by the somatostatin-sst interaction such as central acetylcholine release, reduction of locomotor activity, depression of cortical activity, induction of slow wave sleep, anti-inflammatory and immunomodulatory effects, and reduction of vascular calcium deposition (for review, see Spier and de Lecea, 2000; Broglio et al., 2007; Gonzalez-Rey and Delgado, 2008).

The existence of a specific cortistatin receptor has not been identified yet but differential actions between somatostatin and cortistatin may possibly reside in the ability of cortistatin to bind and to activate the GRLN-R, whereas somatostatin does not (Deghenghi et al., 2001b; Muccioli et al., 2001). Divergent from native somatostatin, however, synthetic somatostatin agonists, lanreotide, octreotide, and vapreotide bind to the GRLN-R in human pituitary tissue (Deghenghi et al., 2001c) in addition to their selective affinity for sst2 > sst5 > sst3 (Bauer et al., 1982; Reichlin, 1983; Redding and Schally, 1984). In addition to binding studies, few functional findings also support the possibility that cortistatin may be another endogenous high-affinity ligand of the

GRLN-R. Cortistatin has been reported to inhibit vascular calcification induced experimentally in rats through activation of the GRLN-R receptor rather than sst or Mrg X2 (Liu et al., 2010). Of interest was the demonstration that cortistatin selectively upregulates the GRLN-R mRNA expression in cultured rat vascular smooth muscle cells, further indicative of an interaction between cortistatin and ghrelin signaling (Liu et al., 2010). Other studies in primates and mice demonstrated that endogenous cortistatin, unlike somatostatin, is involved in the stimulation of pituitary prolactin release, an effect that is blocked *in vitro* by the GRLN-R antagonist (Cordoba-Chacon et al., 2011). However, most of the endocrine studies performed *in vivo* or *in vitro* showed parallel inhibitory responses between cortistatin and somatostatin consistent with the activation of classical sst subtypes (Broglio et al., 2008) with some exceptions (Prodam et al., 2008). This is further corroborated by the finding that cortistatin knockout mice display elevated circulating acyl ghrelin levels associated with an upregulated gastric ghrelin and GOAT expression (Cordoba-Chacon et al., 2011) indicating an inhibitory tone of endogenous cortistatin on ghrelin signaling.

To further delineate the actions of cortistatin mediated *via* the GRLN-R another peptide, cortistatin-8, has been shown to bind to the GRLN-R while being devoid of affinity to the sst subtypes (Luque et al., 2006b). However, in one clinical study cortistatin-8 did not influence spontaneous pituitary hormone secretion (GH, prolactin, and ACTH) and did not interfere with ghrelin's endocrine responses when given in equimolar dose ratios in healthy human subjects (Prodam et al., 2008). Therefore,

# **REFERENCES**


"fendo-03-00178" — 2013/1/9 — 15:57 — page 7 — #7

additional specific tools may be needed to characterize a possible direct link between the ghrelin and somatostatin signaling system *via* interaction on the GRLN-R.

# **SUMMARY**

In summary, somatostatin robustly affects circulating levels of ghrelin through interaction with the sst2. However, alterations vary with the site of action. Central somatostatin elevates plasma levels of acyl and desacyl ghrelin *via* interaction with brain sst2 and counteracts the visceral stress-related decrease in circulating ghrelin through pathways still to be elucidated in rodents. By contrast, the activation of peripheral somatostatin-sst2 inhibits circulating ghrelin levels in experimental and clinical studies and mediates the decline in circulating ghrelin induced by abdominal surgery in rodents likely *via* a paracrine action of somatostatin on sst2 bearing ghrelin cells in the stomach. Of interest, cortistatin, the other member of the somatostatin family, in addition to binding to sst1−5, also binds to and activates the GRLN-R. There is evidence that the peptide can exert a dual influence on ghrelin, by inhibiting its release through interaction with sst2 located on gastric ghrelin cells while activating GRLN-R at ghrelin's tissue targets.

# **ACKNOWLEDGMENTS**

This work was supported by NIH R01 DK-33061, NIH Center Grant DK-41301 (Animal Core), VA Research Career Scientist and VA Merit grant (to Yvette Taché), German Research Foundation STE 1765/3-1 (to Andreas Stengel), and Charité University Funding UFF 89-441-176 (to Andreas Stengel).

of cortistatin: in vivo studies. *Mol. Cell. Endocrinol*. 286, 123–127.


of one pancreatic ghrelinoma. *J. Clin. Endocrinol. Metab*. 88, 3117– 3120.


"fendo-03-00178" — 2013/1/9 — 15:57 — page 8 — #8


endocrine neoplasm type I. *J. Clin.*


injection of somatostatin receptor 5 preferring agonists induces a vagal cholinergic stimulation of gastric emptying in rats. *J. Pharmacol. Exp. Ther.* 293, 1099–1105.


"fendo-03-00178" — 2013/1/9 — 15:57 — page 9 — #9

the treatment of gastroenteropancreatic neuroendocrine tumors. *Horm. Metab. Res.* 43, 844–853.


its inhibition by somatostatin. *Acta Physiol. Scand*. 122, 261–267.


(2004). Both corticotropin-releasing factor receptor type 1 and type 2 are involved in stress-induced inhibition of food intake in rats. *Psychopharmacology (Berl.)* 76, 30–38.


localisation of somatostatin receptor subtypes sst1-sst5 in areas of the rat medulla oblongata involved in autonomic regulation. *J. Chem. Neuroanat*. 35, 49–66.


"fendo-03-00178" — 2013/1/9 — 15:57 — page 10 — #10

intake and gastric emptying in rats. *Neurogastroenterol. Motil*. 23, e294– e308.


T. L., and Tschöp, M. (2004). Central administration of ghrelin and agoutirelated protein (83–132) increases food intake and decreases spontaneous locomotor activity in rats. *Endocrinology* 45, 4645–4652.


decrease of circulating human ghrelin levels. *J. Endocrinol. Invest.* 24, RC19–RC21.


receptor by pancreatic islet cells and related endocrine tumors. *J. Clin. Endocrinol. Metab.* 87, 1300–1308.


"fendo-03-00178" — 2013/1/9 — 15:57 — page 11 — #11


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 11 October 2012; paper pending published: 01 November 2012; accepted: 17 December 2012; published online: 11 January 2013.*

*Citation: Stengel A and Taché Y (2013) Activation of somatostatin 2 receptors in the brain and the periphery induces opposite changes in circulating ghrelin levels: functional implications. Front. Endocrin. 3:178. doi: 10.3389/fendo.2012.00178*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Stengel and Taché. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

#### *Yves Mear 1, Alain Enjalbert 1,2 and Sylvie Thirion1 \**

<sup>1</sup> CNRS, CRN2M UMR7286, Aix Marseille University, Marseille, France <sup>2</sup> Molecular Biology Laboratory, Conception Hospital, AP-HM, Marseille, France

#### *Edited by:*

María M. Malagón, University of Cordoba, Spain

#### *Reviewed by:*

Rhonda D. Kineman, University of Illinois at Chicago, USA Jesus P. Camiña, Hospital Clínico Universitario de Santiago, Spain

#### *\*Correspondence:*

Sylvie Thirion, Faculté de Médecine Nord, CRN2M - Neurobiology and Neurophysiological Research Center of Marseille, UMR CNRS 7286, Aix-Marseille University, Bd Pierre Dramard-CS 80011, 13344 Marseille Cedex 15, France e-mail: sylvie.thirion@univ-amu.fr

Abundant evidences have shown that ghrelin, by its binding to GHS-R1a, plays an important role for fundamental physiological functions. Increasing attention is given to the GHS-R1a unusually high constitutive activity and its contribution to downstream signaling and physiological processes. Here, we review recent lines of evidences showing that the interaction between ligand-binding pocket TM domains and the ECL2 could be partially responsible for this high constitutive activity. Interestingly, GHSR-1a constitutive activity activates in turn the downstream PLC, PKC, and CRE signaling pathways and this activation is reversed by the inverse agonist [D-Arg1, D-Phe5, D-Trp7*,*9, Leu11]-substance P (MSP). Noteworthy, GHSR-1a exhibits a C-terminal-dependent constitutive internalization. Non-sense GHS-R1a mutation (Ala204Glu), first discovered in Moroccan patients, supports the role of GHSR-1a constitutive activity in physiological impairments. Ala204Glu-point mutation, altering exclusively the GHSR-1a constitutive activity, was associated with familial short stature syndrome. Altogether, these findings suggest that GHS-R1a constitutive activity could contribute to GH secretion or body weight regulation. Consequently, future research on basic and clinical applications of GHS-R1a inverse agonists will be challenging and potentially rewarding.

**Keywords: ghrelin receptor, GPCR, constitutive activity, signaling pathway, PLC, β-arrestin**

# **INTRODUCTION**

The secretion of growth hormone (GH) by the anterior pituitary is under complex control. Small synthetic molecules termed GH secretagogues (GHS) are synthetic, peptidyl, and nonpeptidyl molecules which possess strong and dose-dependent GH-releasing activity *in vivo* in several species and in humans (Cheng et al., 1989; Bowers et al., 1991; Cheng et al., 1991). The cloning of the GH secretagogue receptor (GHS-R1a, now called ghrelin receptor) in 1996 (Howard et al., 1996; Pong et al., 1996), led to the isolation of the endogenous ligand, ghrelin in 1999 (Kojima et al., 1999).

Ghrelin is a 28 amino acid peptide that differs from all other peptide hormones known by an octanoylation. This fatty acid modification is essential both for the binding to and activation of the receptor and for its pharmacokinetic properties (Kojima et al., 2001). Ghrelin strongly stimulates GH release in humans (Bowers et al., 1991), and it is much more potent than Growth Hormone-Releasing Hormone (GHRH) under similar conditions. Ghrelin potently increase food intake and weight gain, and also regulates energy homeostasis and metabolism following central and systemic administration (Castañeda et al., 2010; Stengel and Taché, 2012).

The GHS-R1a is a G protein-coupled receptor (GPCR) of 366 amino acids with the characteristic seven transmembranespanning domains (7TM receptor; for review Schwartz et al., 2006; Cruz and Smith, 2008). Signal transduction from the extracellular environment via 7TM receptors in general requires a conformational change from an inactive (R) to a active state (R∗). Certain 7TM receptors are stabilized in an active conformation without any ligand present. The ability to propagate the intracellular signal in the absence of agonist is commonly known as constitutive activity.

Constitutive activity is described for mostly all GPCR, however for a large part of them the level of constitutive activity is very low (Arvanitakis et al., 1998; Smit et al., 2007).

The GHS-R1a exhibits unusual high constitutive activity (Holst et al., 2003) as it signal with ∼50% of its maximal capacity in the absence of the agonist, ghrelin. This article aims to review the current knowledge on GHS-R1a ligand-independent constitutive activity and its functions.

# **GHS-R1a AND ITS CONSTITUTIVE ACTIVITY MOLECULAR BASIS**

For several years, it has been noted that mostly all GPCRs exhibit intrinsic constitutive activity (Arvanitakis et al., 1998; Smit et al., 2007). In 1999, studies performed on the β2-adrenergic receptor suggested that this ligand-independent activity could involve an inward movement of the extracellular segments of the transmembrane domains (TMs) VI and VII toward TM III in the ligand-binding pocket (Elling et al., 1999).

Holst and colleagues performed a structural analysis of GHS-R1a and revealed the crucial role of an aromatic cluster formed by three residues (*Phe VI:16, Phe VII:06, and Phe VII:09)* on the inner face of the extracellular ends of GHS-R1a TMs VI and VII. Their close spatial proximity and the formation of this cluster allow GHS-R1a to stabilize in its active conformation in absence of agonist (Holst et al., 2004; Mokrosinski and Holst, 2010).

The residue in position *VI:16* is central for the constitutive activity level that can gradually be increased or decreased depending on the size and hydrophobic properties of the side chain of the amino acid. It has also been suggested that the aromatic residue *VI:16* may work as a tethered agonist located strategically at the interface between TM III TM VI and TM VII, blocking these extracellular TM segments in a conformation promoting GHS-R1a high constitutive activity (Schwartz et al., 2006).

A conserved aromatic lock crucial for GHS-R1a high basal signaling level is formed by the Trp *VI:13* and Phe *V:13* residues (Holst et al., 2004). The Trp *VI:13* (=Trp276) is located in the conserved motif CWxP in the middle of TM VI and is supposed to act as a global toggle switch model allowing the inward movement of this domain, and the GHS-R1a high basal activation level (Schwartz et al., 2006; Floquet et al., 2010).

Specific residues in the vinicity of this cluster have been proposed to orchestrate finely tuned microswitches critical for the activation level in absence of ligand (Holst et al., 2004; Valentin-Hansen et al., 2012).

In order to study the importance of this core peptide in the GHS-R1a constitutive activity, Gozé et al. introduced the mutation Trp276Ala and mutated the two surrounding amino acid residues Val131 and Ile134. Their results revealed that the mutation Trp276A1a dramatically impairs the ligand-independent activity whereas Val131Leu and Ile134Met highly increase GHS-R1a basal activity. According to these results, the three residues Trp276, Val131, and Ile134 could also significantly impact on GHS-R1a constitutive signaling (Gozé et al., 2010).

In 2006, Pantel et al. reported a mutation (Ala204Glu) in the extracellular loop II (ECL2) of the human GHS-R1a affecting selectively the ligand-independent activity (Pantel et al., 2006). The ECL2 structure function analysis revealed that by restricting this segment, and so possibly TM V/TM III, movements either by mutation or by ligand binding, reduces the constitutive signaling level (Mokrosinski et al., 2012), showing that the high GHS-R1a basal signaling level depends on the flexibility in these segments. Other studies showed that a single mutation or space generating a substitution in the GHS-R1a sequence or in the ligand peptide sequence can change the ligand properties from agonist to inverse agonist or from inverse agonist to agonist depending on the residue mutated (Holst et al., 2007; Els et al., 2012).

TM VI and TM VII movements into their inward-bend promoting the ligand-independent active conformation can be stoically blocked, using a modified substance P (MSP) inverse agonist, [D-Arg1, D-Phe5, D-Trp7*,*9, Leu11]-substance P (often referred in the literature as SPA, for substance P analog). The systematic analysis of this peptide structure-function relationship identified the C-terminal heptapeptide (fQwFwLL) as its active core, the D-Phe5 residue being apparently crucial for the inverse agonist property and the binding affinity (Holst et al., 2006). Unlike ghrelin that only interacts with the middle part of the ligand-binding pocket, the inverse agonist binds to an extended-binding pocket comprising all the seven TM domains of the receptor except for the first one. In addition, the spacegenerating mutants located relatively deep in the binding pocket at key positions within the TM III, TM IV, and TM V, upregulate the effects of MSP, suggesting that this molecule could prevent the spontaneous receptor activation across the binding pocket extend (Holst et al., 2006; Mokrosinski and Holst, 2010).

#### **INTRACELLULAR PATHWAYS AND CONSTITUTIVE INTERNALIZATION**

Inositol phosphate (IP) signaling pathway, through Phospholipase C (PLC) activation, was the first specifically associated with the GHS-R1a ligand-dependent activity (Adams et al., 1995; Lei et al., 1995; Chen et al., 1996; Petersenn, 2002) inducing intracellular calcium mobilization (Herrington and Hille, 1994). Consequently, this pathway has been investigated for determining the receptor constitutive signaling. PLC activation was demonstrated comparing heterologous HEK-293 and COS-7 cells overexpressing the GHS-R1a to cells transfected with the motilin receptor (that is the closest GPCR homolog of GHS-R1a without constitutive activity; Holst et al., 2003). The unusualy high GHS-R1a ligand-independent signaling level was similar to that of the most famous highly constitutively active GPCR, the virally encoded ORF74 receptor. To note, this paper allowed characterizing MSP as a full GHS-R1a inverse agonist. Indeed, this compound inhibits the GHS-R1a constitutive ligand-independent IP accumulation and decreases the IP level to that of cells transfected with the empty vector (Holst et al., 2003). Lau et al. reported that GHS-R1a constitutive activity could reduce apoptosis in HEK-293 overexpressing the (seabream) sbGHS-R1a, through PKC-dependent caspase-3 inhibition (Lau et al., 2009).

Gq/11-coupled GHS-R1a constitutive activity also resulted in a dose-dependent but ligand-independent increase in the CRE luciferase reporter while the full inverse agonist MSP partially reversed this effect. To note, heterotrimeric Gq/11 protein-coupled receptors have previously been reported to phosphorylate the cAMP responsive element-binding protein (CREB) by activating the CRE pathway as a result of calcium/calmoduline kinase IV (CaMK IV) and/or protein kinase C (PKC) activation (Matthews et al., 1994; Poulin et al., 2009) (**Figure 1**). Finally, performing serum responsive element (SRE) reporter assay on HEK-293 cells transfected with GHS-R1a revealed a 10-fold increase in the ligand-independent signaling compared to the cells transfected with the empty plasmid.

While many other signaling pathways have been shown to play a role in the GHS-R1a ligand-dependent activation, the GHS-R1a ligand-independent activation has not been investigated. Indeed, Gi/o heterotrimeric pathway was clearly associated with beta-arrestin-mediated ERK1/2 activation following GHS-R1a activation by ghrelin (Camiña et al., 2007). Similarly, c-Src that is involved in the GHS-R1a ghrelin-dependent Akt activation (via Gi/o-protein; Lodeiro et al., 2009) has not been studied for the high basal level associated signaling pathways.

Since GHS-R1a exhibits an unusually high constitutive activity, it could be hypothesized that the downstream signaling level could reflect the membrane expression level. In this context, a better understanding of the mechanisms underlying and modulating its plasma membrane expression was necessary. To this end, GHS-R1a was tagged using M2 anti-FLAG antibody labeling, thus allowing following the intracellular movement of

GHS-R1a. These experiments revealed a GHS-R1a constitutive ligand-independent internalization and the receptor could be trapped at the cell surface by the inverse agonist MSP. The punctiform receptor intracellular labeling co-localizes with clathrincoated vesicles and recycling endosome markers (Holst et al., 2003).

Unlike the GHS-R1a, GPR-39 receptor, a member of the ghrelin receptor family, is not constitutively internalized but it displays a high ligand-independent signaling level (Holst et al., 2003). Based on these observations, Holliday et al. developed an elegant approach by switching the GHS-R1a C tail with that of the GPR-39. The chimera, named GhR-39, was constitutively active through the PLC pathway but its internalization was impaired. Components supporting the constitutive activity could be necessary but not sufficient for GHS-R1a endocytosis and additional regulatory elements in the C-terminal domain may be involved (Holliday et al., 2007).

GHS-R1a constitutive internalization requires the sequential activation of the monomeric G proteins Rab5 and Rab11. Rab proteins control various important cellular processes, such as endocytosis, trafficking, endosome fusion. and exocytosis (Seachrist and Ferguson, 2003). These proteins regulate vesicle transport and fusion with specific target compartments, the early endosomes for Rab5, and the endosomal recycling compartments such as the perinuclear recycling compartment (PNRC) for Rab11. MSP-induced GHS-R1a membrane plasma anchorage is blocked by the constitutive expression of GTP-binding mutants of Rabs (Rab5 Gln79Leu or Rab11 Gln70Leu) confirming that MSP naturally inhibits GHS-R1a internalization rather than activates GHS-R1a neosynthesis and trafficking (Holliday et al., 2007).

β-arrestin recruitment has been obviously investigated as it appears as the most widely standard adaptor for GPCR endocytosis (Lefkowitz, 1998) (**Figure 2**). Only ghrelin stimulation induces GHSR-1a phosphorylation and β-arrestin 2 recruitment. Besides, dominant-negative β-arrestin 2 construct, which competes for clathrin interaction, does not inhibit the constitutive endocytosis, supporting the hypothesis of a GHS-R1a β-arrestinindependent constitutive internalization. These results suggest that the aromatic residue *VI:16*, already mentioned above, could act as "tethered biased agonist" rather than "tethered agonist" because it blocks the receptor in a conformation that only induces the Gq/11 protein activation without β-arrestin 2 recruitment (Shukla et al., 2011; Reiter et al., 2012). To note, β-arrestin 2 recruitment has been reported to active the MAPK pathway. In this context, the absence of β-arrestin 2 recruitment in basal condition has been proposed to explain the absence of MAPK pathway activation (Holliday et al., 2007).

Purified GHS-R1a monomers in a lipid disc showed that the ghrelin receptor *per se* activates Gq/11 in the absence of agonist,

and that GHS-R1a constitutive activity is an intrinsic property of the protein and is not influenced by its cellular environment (Damian et al., 2012). In this context, the receptor isolated in lipid discs recruits arrestin-2 in an agonist-dependent manner (Mary et al., 2012), whereas it interacts with μ-AP2 (plasma membranelocalized clathrin adaptor subunit) in the absence of ligand or in the presence of ghrelin (Damian et al., 2012). Thus, μ-AP2 could be involved in the basal regulation of GHS-R1a trafficking.

# **PHYSIOLOGICAL RELEVANCE**

Many types of GPCR display a high ligand-independent signaling *in vitro* (Seifert and Wenzel-Seifert, 2002). Among them, GHS-R1a has been shown to display both the highest basal activation of Gq/11 (about 50% of its maximal activity) *in vitro* (Holst et al., 2004), and substantial basal signaling for food intake and weight control *in vivo* (Petersen et al., 2009; Els et al., 2012). Pantel et al., who reported naturally occurring mutations in the GHS-R1a sequence, have first shown a putative link with physiological impairments in 2006. They reported a substitution mutation located within the first GHS-R1a exon, predicting the substitution of Alanine by Glutamate (Ala204Glu) in two independent Moroccan families. This missense mutation is located in the GHS-R1a ECL2 and affects a fully conserved amino acid. HEK-293 cells stably transfected with WT and Ala204Glumutant GHS-R1a showed, using POU1F1-luciferase reporter assay that this mutation selectively abolishes the GHS-R1a ligandindependent signaling without altering the Ghrelin-dependent activity. Noteworthy, mutations altering exclusively the constitutive activity are associated with familial short stature syndrome and the latter can be partially reversed with GH treatment (Pantel et al., 2006, 2009; Inoue et al., 2011). Wang reported another uncharacterized GHS-R1a mutation affecting the Phe 279 residue (Phe *VI:16*) recently identified as being of particular interest for GHS-R1a constitutive activity. Noteworthy, the phenotype of patients expressing this mutation is characterized by an increased obesity and short stature (Wang, 2004). Regarding these different but functionally similar mutations, Holst and Schwartz proposed that the absence of GHS-R1a constitutively active signaling results in a syndrome, characterized, not only by a short stature, but also by obesity (Holst and Schwartz, 2006). This suggestion has been confirmed recently using icv MSP administration that significantly decreased the food intake, body weight, and neuropeptide Y (NPY) and uncoupling protein 2 (UCP2) gene expression in the hypothalamus (Petersen et al., 2009).

Ligand-dependent GHS-R1a heterodimerizations have been reported for the subtype 1 dopamine receptor that increases dopamine signaling (Jiang et al., 2006), the somatostatin receptor-5 that regulates insulin release (Park et al., 2012) and the melanocortin-3 receptor that is involved in body weight regulation and energy balance (Rediger et al., 2011, 2012). Also, the dimerization of the GHS-R1a with the dopamine D2 receptor has recently been shown to regulate appetite (Kern et al., 2012).

Its dimerization with the melanocortin-3 receptor, the D1 receptor, and the newly discovered 5-HT2C receptor may be central in modulating and controlling GHS-R1a-mediated downstream signaling and subsequent satiety and appetite signaling, as well as the rewarding and motivational aspects of food intake (Schellekens et al., 2013). The understanding of the underlying mechanisms leading to these activations may ultimately lead to the development of new therapeutic strategies.

On the other hand, ghrelin and GHS-R1a knockout mice explorations may allow the emergence of new regulatory properties of the ghrelin receptor constitutive activity. For example, GHS-R1a constitutive activity increases limbic seizures in rodents, the endogenous ligand being naturally anticonvulsive (Portelli et al., 2012). In the mouse brain, deficits in spontaneous receptor activity cause marked functional impairment in learning and memory (Albarran-Zeckler et al., 2012). GHS-R1a signaling is necessary for hippocampal-dependent learning and habituated feeding responses (Davis et al., 2011). Ghrelin plays also a role in sleep whereas the GHS-R1a will be more implicated in arousal (Esposito et al., 2012).

To move beyond putative redundant compensation mechanisms, further *in vivo* investigations targetting the receptor constitutive activity with appropriate pharmacological tools (Mokrosinski and Holst, 2010; Sivertsen et al., 2011) would be needed to determine the underlying physiological functions linked to this constitutive activity.

#### **CONCLUSION**

Abundant evidence currently indicates that ghrelin, by its binding to GHS-R1a, plays a role in various physiological functions that

# **REFERENCES**


ghrelin. *Eur. J. Intern. Med.* 23, 197–202.


led to the development of clinical trials to translate basic research findings to human disease treatment and diagnosis (for review Akamizu and Kangawa, 2012). Concerning the GHS-R1a constitutive activity, most of the studies performed so far focused on the molecular aspects. Even if some physiological relevance are emerging for food intake regulation (see paragraph above), further studies remain necessary to confirm the role played by the GHS-R1a constitutive activity in the regulation of the GH axis. Pituitary somatotroph adenomas express higher GHS-R1a transcript and protein levels than normal pituitary (Korbonits et al., 1998; Barlier et al., 1999). Moreover, Pantel et al. (2006) have shown that a reduced GHS-R1a constitutive activity impairs the GH secretion. Thus, it should be interesting to test if the GHS-R1a constitutive activity disrupts the GH hypersecretion observed in acromegalic patients with somatotroph tumors (Acunzo et al., 2008; Roche et al., 2012).

The signaling pathways associated with this high constitutive activity should also be addressed. Knowing that the heterogeneity of the coupling on the signaling varies depending on the tissue or cell type, it is crucial to address now these questions in physiological or pathophysiological models. Depending on the model, does the GHS-R1a ligand-independent CRE activation pass through the calcium/calmodulin kinase? And/or via the cAMP/PKA pathways?

These issues are part of the many questions pending in this field. Some of them may support the development of clinical applications of GHS-R1a inverse agonists in physiological disorders in the future, allowing developing novel and unique therapies for various disorders, including intractable and serious diseases. Indeed, research on basic and clinical applications of GHS-R1a inverse agonists will be challenging and potentially rewarding.

# **ACKNOWLEDGMENTS**

We thank very gratefully Prof. Isabelle Limon for helpful criticisms and careful corrections of the manuscript. We acknowledge Dr. Marie-Pierre Blanchard for her manuscript reading. This work was supported by Aix-Marseille University and CNRS.

polymerase chain reaction, triple *in-situ* hybridization and cell culture studies. *J. Neuroendocrinol.* 11, 491–502.


*Front. Neuroendocrinol.* 31, 44–60. doi: 10.1016/j.yfrne.2009.10.008


kinase-C in His-D-Trp-Ala-Trp-D-Phe-Lys-NH2-induced growth hormone release from rat primary pituitary cells. *Endocrinology* 129, 3337–3342.


intracellular trafficking of the ghrelin receptor. *Mol. Endocrinol.* 21, 3100–3112.


growth-hormone-releasing acylated peptide from stomach. *Nature* 402, 656–660.


mutagenesis. *Methods Enzymol.* 484, 53–73.


mechanism of β-arrestin-biased agonism at seven-transmembrane receptors. *Annu. Rev. Pharmacol. Toxicol.* 52, 179–197.


protein-coupled receptor endocytosis and trafficking by Rab GTPases. *Life Sci.* 74, 225–235.


*Annu. Rev. Pharmacol. Toxicol.* 47, 53–87.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 December 2012; accepted: 09 May 2013; published online: 29 May 2013.*

*Citation: Mear Y, Enjalbert A and Thirion S (2013) GHS-R1a constitutive activity and its physiological relevance. Front. Neurosci. 7:87. doi: 10.3389/fnins. 2013.00087*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Neuroscience.*

*Copyright © 2013 Mear, Enjalbert and Thirion. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Actions of agonists and antagonists of the ghrelin/GHS-R pathway on GH secretion, appetite, and cFos activity

#### **Rim Hassouna<sup>1</sup> , Alexandra Labarthe<sup>1</sup> , Philippe Zizzari <sup>1</sup> , Catherine Videau<sup>1</sup> , Michael Culler <sup>2</sup> , Jacques Epelbaum<sup>1</sup> and Virginie Tolle<sup>1</sup>\***

<sup>1</sup> UMR-S 894 INSERM, Centre de Psychiatrie et Neurosciences, Université Paris Descartes, Sorbonne Paris Cité, Paris, France 2 IPSEN, Milford, MA, USA

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Rhonda D. Kineman, University of Illinois at Chicago, USA Vivian J. Costantini, Aptuit s.r.l., Italy

#### **\*Correspondence:**

Virginie Tolle, UMR894 INSERM, Centre de Psychiatrie et Neurosciences, 2 ter rue d'Alésia, 75014 Paris, France. e-mail: virginie.tolle@inserm.fr

The stimulatory effects of ghrelin, a 28-AA acylated peptide originally isolated from stomach, on growth hormone (GH) secretion and feeding are exclusively mediated through the growth hormone secretagogue 1a receptor (GHS-R1a), the only ghrelin receptor described so far. Several GHS-R1a agonists and antagonists have been developed to treat metabolic or nutritional disorders but their mechanisms of action in the central nervous system remain poorly understood. In the present study, we compared the activity of BIM-28163, a GHS-R1a antagonist, and of several agonists, including native ghrelin and the potent synthetic agonist, BIM-28131, to modulate food intake, GH secretion, and cFos activity in arcuate nucleus (ArcN), nucleus tractus solitarius (NTS), and area postrema (AP) in wild-type and NPY-GFP mice. BIM-28131 was as effective as ghrelin in stimulating GH secretion, but more active than ghrelin in inducing feeding. It stimulated cFos activity similarly to ghrelin in the NTS and AP but was more powerful in the ArcN, suggesting that the super-agonist activity of BIM-28131 is mostly mediated in the ArcN. BIM-28163 antagonized ghrelin-induced GH secretion but not ghrelin-induced food consumption and cFos activation, rather it stimulated food intake and cFos activity without affecting GH secretion.The level of cFos activation was dependent on the region considered: BIM-28163 was as active as ghrelin in the NTS, but less active in the ArcN and AP. All compounds also induced cFos immunoreactivity in ArcN NPY neurons but BIM-28131 was the most active. In conclusion, these data demonstrate that two peptide analogs of ghrelin, BIM-28163, and BIM-28131, are powerful stimulators of appetite in mice, acting through pathways and key brain regions involved in the control of appetite that are only partially superimposable from those activated by ghrelin. A better understanding of the molecular pathways activated by these compounds could be useful in devising future therapeutic applications, such as for cachexia and anorexia.

**Keywords: ghrelin, GHS-R, BIM-28163, BIM-28131, antagonist, food intake, GH secretion, cFos**

#### **INTRODUCTION**

Ghrelin is a 28 amino acid acylated peptide originally discovered in stomach tissues by inverse pharmacology (Kojima et al., 1999) as the endogenous ligand for the growth hormone (GH) Secretagogue-1a Receptor (GHS-R1a), an orphan receptor cloned a few years earlier from pig pituitaries (Howard et al., 1996). GH secretagogues are a family of synthetic peptidic or non-peptidic compounds developed from the early 1980s to stimulate GH secretion (Bluet-Pajot et al., 2001). As anticipated, ghrelin was initially described for its ability to stimulate GH secretion in several species including rodents and humans (Kojima et al., 1999; Tolle et al., 2001) but it is still to date the only orexigenic hormone produced in the gastrointestinal tract (Tschop et al., 2000).

The actions of ghrelin on GH secretion and feeding require the addition of an eight-carbon fatty acid that is attached on a serine in position 3 by the enzyme ghrelin-*O*-acyl-transferase (GOAT) (Gutierrez et al., 2008; Yang et al., 2008). They are exclusively mediated through the GHS-R1a (Sun et al., 2004), the only ghrelin receptor described so far. GHS-R1a is highly expressed in the hypothalamus, a region involved in the control of GH secretion and appetite, and also in the brainstem that receives informations from gut vagal afferents (Guan et al., 1997; Katayama et al., 2000). Within the hypothalamus, NPY neurons in the arcuate nucleus (ArcN) express the GHS-R1a (Willesen et al., 1999; Tannenbaum et al., 2003), and are a well-characterized target for ghrelin actions (Tannenbaum et al., 2003; Chen et al., 2004). Nevertheless, several studies have suggested that certain ghrelin actions may be mediated through a receptor that has yet to be identified.

Several GHS-R1a antagonists have been developed to decipher the function of the ghrelin/GHS-R pathway in the regulation of feeding behavior and GH secretion (Asakawa et al., 2003; Okimura et al., 2003; Beck et al., 2004; Halem et al., 2004; Demange et al., 2007; Esler et al., 2007; Petersen et al., 2009; Costantini et al., 2011; Moulin et al., 2013). Although they have all been clearly shown to antagonize exogenous ghrelin actions on GH secretion both in *in vitro* systems and *in vivo*, their biological effects *per se* on several biological parameters are contradictory and their mechanisms of action in the central nervous system remain poorly understood.

Indeed, intracerebroventricular injections of two antagonists [d-Lys3]-GHRP6, an analog of one of the synthetic GHS (Asakawa et al., 2003) or [d-Arg-1, d-Phe-5, d-Trp-7,9,Leu-11]-substance P, an analog of substance P (Petersen et al.,2009) induces suppression of feeding in mice. In contrast [d-Lys3]-GHRP6 has no action on spontaneous ultradian GH secretion (Okimura et al., 2003). But results are biased by the fact that [d-Lys3]-GHRP6 also binds to all melanocortin receptors (Schioth et al., 1997) and [d-Arg-1, d-Phe-5, d-Trp-7,9,Leu-11]-substance P also has full inverse agonist action *in vitro* and *in vivo* (Holst et al., 2003; Petersen et al., 2009). To date, the effects of JMV2810, another recently developed GHS-R1a antagonist, on spontaneous feeding or GH secretion have not been reported (Demange et al., 2007).

Only the full-length ghrelin analog, BIM-28163 (now called RM-28163), has been tested on both spontaneous GH secretion and food intake. Treatment with this selective GHS-R1a antagonist over 48 h in rats reduces pulsatile GH secretion (Zizzari et al., 2005). In contrast, it increases food intake and weight gain as effectively as ghrelin when administered at 5- to 10-fold higher doses than ghrelin (Halem et al., 2004, 2005). Interestingly, BIM-28163 induces cFos activation in the dorsomedial nucleus of the hypothalamus (DMH) while it acts as an antagonist in the ArcN of the hypothalamus. These data are corroborated by a more recent study also describing a stimulatory effect of another GHS-R1a antagonist, GSK1614343, on feeding in rats and dogs (Costantini et al., 2011). Altogether these data are intriguing and suggest the existence of an unknown pathway mediating the effects of these ghrelin antagonists on feeding through either the GHS-R1a or another unknown receptor. Understanding the mechanisms of action of this compound that differentially affect feeding and GH secretion can be of clinical interest.

In the present study, we compared the effect of BIM-28163, BIM-28131(Strassburg et al., 2008; Palus et al., 2011), native ghrelin, and the combination of BIM-28163 + ghrelin in the modulation of food intake, GH secretion, and cFos activity in fed mice. Our aim was to map changes in cFos activation in several key brain regions controlling appetite and/or GH secretion, including the hypothalamic ArcN, ventromedial nucleus of the hypothalamus (VMH), nucleus tractus solitarius (NTS), and area postrema (AP) of the brainstem. In addition, we used NPY-GFP mice to test whether the feeding effects of the analogs were mediated, like ghrelin, through the orexigenic NPY neurons.

# **MATERIALS AND METHODS**

#### **ANIMALS**

About 18–25-week-old C57BL6/J male and female mice, obtained either from Charles River or from our own colony, were used for feeding experiments. About 18–25-week-old C57BL6/J or NPY-Renilla GFP transgenic male and female mice backcrossed on the C57BL6/J background and expressing Renilla GFP under the transcriptional control of the NPY genomic sequence (Van Den Pol et al., 2009) were used for cFos experiments. Mice were housed at constant temperature and humidity, with a fixed 12 h light/dark cycle (lights-on at 7.00 a.m.) and free access to food and water. In addition, the animals were handled weekly to minimize stress. All experiments were carried out in accordance with the European Communities Council Directive (86/609/EEC) and were approved by the animal experimentation committee of Paris Descartes University.

# **PEPTIDES**

Native ghrelin, BIM-28131 (a small peptide ghrelin agonist), and BIM-28163 (a full-length ghrelin analog antagonist) were obtained from IPSEN (Milford, MA, USA). Peptides were dissolved in a vehicle containing 0.9% saline +0.25% of bovin serum albumin (BSA).

# **AUTOMATED FOOD INTAKE MONITORING**

One week prior to the experiments, 18–25-week-old C57BL6/J male mice were individually housed and acclimatized to the automated drinking/feeding stations (TSE Systems, GmbH, Germany). Feeding behavior was recorded continuously by means of high precision sensors, attached to the top of the cage lids. Meal patterns were analyzed using the following definition: a meal consists of the consumption of 0.03 g of food separated from the next feeding episode by at least 10 min as previously described (Yu et al., 2009; Stengel et al., 2010; Wang et al., 2011). For each mouse, the meal number, the total meal size (g), and the total meal duration (min) were measured within 4 h following peptide injections.

Experiments were performed during the light phase (10.00 a.m.–11.00 a.m.) and carried out in a cross-over designed manner so that each mouse received all treatments randomly separated by two washout days. On each experimental day, mice were injected intraperitoneally (ip) with either vehicle (0.9% saline containing 0.25% BSA), native ghrelin (30 nmol), BIM-28131 (30 nmol), BIM-28163 (150 nmol), or native ghrelin (30 nmol) combined with BIM-28163 (150 nmol).

# **cFos IMMUNOHISTOCHEMISTRY, FEEDING, AND GH MEASUREMENTS**

About 18–25-week-old male and female C57BL6/J and NPY-GFP mice were individually housed and had free access to food and water at the time of injections. Vehicle, native ghrelin (30 nmol), BIM-28131 (30 nmol), BIM-28163 (150 nmol), or native ghrelin (30 nmol) combined with BIM-28163 (150 nmol) were injected ip in the early light phase (9.00 a.m.–11.00 a.m.).

A pre-weighed amount of food was distributed in each cage at the time of injections and weighed 90 min later in order to confirm the effects of the treatments on food consumption and to correlate food intake to the number of activated cFos nuclei. 15 min following the injection, 4µl of whole blood was withdrawn from the tail vein, homogenized in 116µl of GH buffer (PBS, 0.05% Tween) for GH measurements. Whole blood GH concentrations were evaluated by EIA as previously described (Steyn et al., 2011).

The number of nuclei immunoreactive for cFos protein were quantified 90 min after ip injection of peptides to determine which brain regions were activated by the compounds. Mice were deeply anesthetized with pentobarbital (5.5 mg/30 g BW) and perfused through the ascending aorta with saline 0.9% for 1 min followed by 4% paraformaldehyde (PFA) in phosphate buffer 0.1 M (PB) for an additional 9 min. The brains were removed, post-fixed for 2 h in 4% PFA and cryoprotected in 30% sucrose for 2 days at 4˚C. Brains were then frozen in 2-methyl-butane and sectioned in the coronal plane at a thickness of 25µm using a freezing microtome (Frigomobile, Leica, Wetzlar, Germany).

For the detection of cFos protein expression, free-floating sections were processed for immunohistochemistry. Sections were incubated in blocking buffer (10% Normal Donkey Serum, 0.3% Triton X-100 in 0.1 M TBS) for 1 h at room temperature then incubated with rabbit cFos antibody (1:20000, Ab-5, Jackson Laboratories,West Grove, PA, USA) in 1% NDS, 0.3% Triton X-100 in 0.1 M TBS overnight at room temperature. Sections were then rinsed 4 × 10 min in 0.1 M TBS and incubated with Cy3 conjugated Donkey Anti-Rabbit antibody (1:800 DAR-Cy3, Jackson Laboratories, West Grove, PA, USA) 1 h at room temperature and then rinsed 4 × 10 min in 0.1 M TBS. Sections were mounted with fluoromount and quantified using a Zeiss Axioplan epifluorescence microscope (Carl Zeiss, Le Pecq, France) under 40× magnification. Quantifications were performed bilaterally every 100µm sections through the ArcN (2.3–1.6 mm anterior to the interaural line) and ventromedial nucleus (VMH, 2.5–2.3 mm anterior to the interaural line) of the hypothalamus and in the NTS (3.7 mm posterior to the interaural line) and AP (3.7 mm posterior to the interaural line) of the brainstem (Franklin and Paxinos, 1997).

NPY neurons were visualized using GFP fluorescence. GFPpositive cell bodies expressing cFos were quantified unilaterally under 40× magnification using a confocal SP5 microscope (Leica, Wetzlar, Germany). Co-localizations were determined with the Image-J software (http://rsbweb.nih.gov/ij/) on series of continuous optical sections with 0.5µm increment along the *z*-axis of the section.

#### **STATISTICAL ANALYSES**

Values are given as mean ± SEM. Statistical analyses were performed using ANOVA, repeated measures ANOVA followed by

treatment on 90 min food intake: \*p < 0.05 BIM-28163 vs. vehicle, \*\*P < 0.01 BIM-28131 vs. vehicle, and native ghrelin, @P < 0.05

Fisher PLSD *post hoc* test using the Statview software (SAS Institute Inc., Cary, NC, USA).

#### **RESULTS**

# **EFFECT OF GHS-R1a AGONISTS AND ANTAGONISTS ON FEEDING BEHAVIOR**

Feeding was monitored after intraperitoneal injections of native ghrelin (30 nmol), BIM-28163 (150 nmol), BIM-28131 (30 nmol), or after co-administration of native ghrelin + BIM-28163 in the early light phase. In a first subset of mice, feeding was monitored manually after injection of the treatments in randomly assigned groups of animals. Ninety minutes after the injections, a significant effect of treatment on food consumption was observed (**Figure 1A**). Although native ghrelin increased food intake by twofold, the effect of this compound was not significant. Only BIM-28131 and BIM-28163 increased food consumption significantly with BIM-28131 being three times more effective and BIM-28163 two times more effective than ghrelin, respectively. Food intake was identical after co-administration of BIM-28163 and native ghrelin or after administration of ghrelin alone. Automated feeding stations were also used to monitor cumulative food intake and meal pattern during 7 h following the injection in a second set of mice: animals received each treatment randomly in a cross-over designed manner (**Figure 2** and **Table 1**). When data were analyzed as repeated measures in the same mouse and over time, there was a significant interaction between time and treatment. BIM-28131 was the only compound to stimulate appetite even in mice that did not respond well to ghrelin. Increased food consumption in this group was associated with a tendency to increased meal number, total meal size, and total meal duration (**Table 1**).

ghrelin + BIM-28163, @@P < 0.01 BIM-28131 vs. native ghrelin + BIM-28163, ##P < 0.01 native ghrelin or BIM-28131 vs. BIM-28163, Fisher PLSD post hoc test.

**Table 1 | Effect of native ghrelin and BIM compounds on meal pattern measured with the automated feeding station in mice.**


Meal number, total meal size, and total meal duration measured (0–7 h) after ip injection of native ghrelin (30 nmol), BIM-28131 (30 nmol), BIM-28163 (150 nmol), and native ghrelin (30 nmol) co-administered with BIM-28163 (150 nmol). Data represent mean ± SEM.

#### **EFFECT OF GHS-R1a AGONISTS AND ANTAGONISTS ON GH SECRETION**

GH plasma levels were monitored 15 min following the injections in the same animals used to monitor food intake manually (**Figure 1B**). Native ghrelin and BIM-28131 equally stimulated GH secretion whereas GH secretion was not increased after injection of BIM-28163 alone. In contrast to feeding data, BIM-28163 antagonized ghrelin-induced GH secretion.

# **EFFECT OF GHS-R1a AGONISTS AND ANTAGONISTS ON cFos ACTIVATION IN THE HYPOTHALAMIC ArcN**

cFos activation was monitored after intraperitoneal injections of native ghrelin (30 nmol), BIM-28163 (150 nmol), BIM-28131 (30 nmol), or after co-administration of native ghrelin + BIM-28163 in the early light phase. Repeated measures ANOVA over the rostro-caudal extent of the ArcN showed an effect of treatment on cFos immunoreactive nuclei (**Figure 3A**). All treatment groups were significantly elevated compared to the vehicle-treated group, except BIM-28163 which had a modest effect (**Figure 3B**). Injection of native ghrelin induced cFos activation and this was not antagonized by co-administration of BIM-28163. Administration of BIM-28131 induced a more pronounced activation than all other treatments. Thus the efficiency of activation in the ArcN was BIM-28131 > native ghrelin > native ghrelin + BIM-28163 > BIM-28163. Differences were greater around 2.1–2.0 mm anterior to the interaural line where most NPY neurons are localized.

**FIGURE 3 | Effect of native ghrelin and BIM compounds on cFos immunoreactivity in mice in the ArcN, AP, NTS, and VMH. (A)** Number of cFos immunoreactive nuclei and **(B)** sum of cFos immunoreactive nuclei along the rostro-caudal extent of the ArcN (2.3–1.6 mm anterior to the interaural line). Data represent mean ± SEM. **(A)** Repeated measures ANOVA over the rostro-caudal extent of the ArcN shows an effect of treatment on the number of cFos-positive nuclei: P < 0.0001 native ghrelin vs. Vehicle, BIM-28163, and BIM-28131, P < 0.0001 BIM-28131 vs. all other treatments, P < 0.0001 BIM-28163 vs. all other treatments except vehicle, P < 0.01 BIM-28163 vs. vehicle, P < 0.0001 native ghrelin + BIM-28163 vs. vehicle, BIM-28163, and BIM-28131, Fisher PLSD post hoc test. **(B)** \*\*\*P < 0.001 vs.

# **EFFECT OF GHS-R1a AGONISTS AND ANTAGONISTS ON cFos ACTIVATION IN THE NTS AND AP OF THE BRAINSTEM**

cFos activation was also observed in the AP and nucleus tractus solitarius (NTS) (**Figures 3C,D**). In the NTS, in contrast to the ArcN, all treatments significantly increased the number of cFos-immunoreactive cells. BIM-28131 stimulated cFos as efficiently as native ghrelin. BIM-28163 had an activity *per se* but was inefficient in antagonizing ghrelin-induced cFos. In the AP, the compounds had different activities. Native ghrelin and BIM-28131 had the same efficiency in activating cFos, but BIM-28163 was ineffective.

vehicle, #P < 0.05 vs. BIM-28163, ##P < 0.01 vs. BIM-28163, ###P < 0.001 vs. BIM-28163, @@P < 0.01 vs. ghrelin + BIM-28163, Fisher PLSD post hoc test. **(C,D)** Number of cFos immunoreactive nuclei in the NTS and AP (3.7 mm posterior to the interaural line). Data represent mean ± SEM. ANOVA shows an effect of treatment on the number of cFos nuclei in the NTS and AP. \*P < 0.05, \*\*P < 0.01 and \*\*\*P < 0.001 vs. vehicle, #P < 0.05 vs. 28163, Fisher PLSD post hoc test. **(E)** Number of cFos immunoreactive nuclei in the VMH (2.5–2.3 mm anterior to the interaural line). Data represent mean ± SEM. No significant effect of treatments is observed in the VMH. **(F)** Summary of the effect of the different BIM compounds on cFos activation in the ArcN, NTS, and AP. Cc, central canal.

# **EFFECT OF GHS-R1a AGONISTS AND ANTAGONISTS ON cFos ACTIVATION IN THE HYPOTHALAMIC VMH**

In a separate group of animals, cFos activation was quantified in the VMH (Interaural line: −2.5, −2.3). Due to the high level of cFos activation and high variability in vehicle-treated animals, statistical differences were not observed between treatments (**Figure 3E**). The number of cFos nuclei in vehicle, ghrelin, and BIM-28131-treated animals were identical; however, in mice treated with BIM-28163, the number of cFos nuclei was twice as elevated as in ghrelin-treated mice.

# **EFFECT OF GHS-R1a AGONISTS AND ANTAGONISTS ON cFos ACTIVATION IN NPY NEURONS**

3V, third ventricle; ME, median eminence.

To measure cFos immunoreactivity in NPY neurons,we used NPY-GFP mice. All treatments induced cFos activation in NPY neurons as compared with vehicle-treated mice (**Figure 4** and **Table 2**). Treatments did not modify the number of GFP-positive neurons. BIM-28131 and native ghrelin co-administered with BIM-28163 induced the greatest activation of NPY neurons with more than 20% of NPY cells activated, whereas native ghrelin and BIM-28163 alone induced cFos in less than 15% of NPY neurons (**Table 2**).

line. Data represent mean ± SEM. Scale bar represents 50µm in the ArcN.

# **DISCUSSION**

The present study demonstrates that two GHS-R1a synthetic ligands, BIM-28163 and BIM-28131, are powerful stimulators of appetite in mice, acting through pathways and brain regions that are distinct from the ones activated by ghrelin (**Figure 3F** and **Table 3**).

BIM-28163 is a full GHS-R1a antagonist as it has no intrinsic activity at the GHS-R1a, and can fully block the ability of ghrelin to activate the GHS-R1a, as well as block ghrelin-induced GH release, both *in vitro* and *in vivo* (Halem et al., 2004). It is thus a pharmacological tool to dissect the role of endogenous ghrelin. However, whereas previous investigations in rats with this GHS-R1a antagonist revealed a role of the endogenous ligand in amplifying GH pulsatile pattern, blocking GHS-R1a fails to inhibit food intake (Zizzari et al., 2005), and BIM-28163 even stimulates appetite and weight gain after daily treatment in rats (Halem et al., 2004, 2005). In the current study performed in mice, we also demonstrate that acute injection of BIM-28163 increases food consumption and show for the first time that the compound induces cFos in two brain regions involved in the control of appetite, the hypothalamus and brainstem. The feeding effect of BIM-28163 is observed as early as 15–30 min after its injection (i.e., same time-course as native ghrelin) (data not shown).

It was previously described in the rat that intraperitoneal (ip) injection of native ghrelin in the early light phase induces feeding and cFos activation in the ArcN, NTS, and AP (Hewson and Dickson, 2000; Lawrence et al., 2002; Takayama et al., 2007). A dose of 30 nmol/30 g body weight was chosen here based on a published study showing that this dose stimulates food intake in mice (Zizzari et al., 2007). Although ghrelin was very potent in activating cFos in the ArcN, NTS, and AP in the present study, its effects on feeding did not reach statistical significance. It may be due to the fact that different sets of animals were used for feeding experiments and cFos experiments. Another possible interpretation is that cFos activation in these nuclei is required but not alone sufficient to induce an effect on appetite. We indeed recently observed that after ip injection of ghrelin, animals can be subdivided into two groups: high and low responders, suggesting an interindividual variability in the feeding effects of ghrelin (Hassouna et al., 2012b).

BIM-28131 is a super-agonist with regard to food intake in rats (Strassburg et al., 2008; Palus et al., 2011). It is also very potent to stimulate both appetite and cFos in the current study even in mice that were not responsive to ghrelin in the cross-over designed study. Differences in stability between native ghrelin and BIM-28131 would partly explain the differential activities of these two compounds. Indeed, in addition to a fivefold higher affinity than native ghrelin in binding to the GHS-R1a and a 10-fold increased potency in activating the receptor, BIM-28131 has a 10-fold greater circulating half-life. However, BIM-28131 is comparable to ghrelin with regard to stimulating GH secretion. The equal effect on GH release may be possible because the action of the compounds may be partly relayed at the pituitary level and the time of blood sampling occurred after a relatively short time (15 min) whereas the food intake was measured over 90 min or greater. By analyzing the feeding response every 15 min following the injection, BIM-28131 did not have a stronger feeding effect over native ghrelin at 15– 30 min post-injection (data not shown) but from 60 min following the injection. Thus, BIM-28131 is more powerful than ghrelin in stimulating appetite at equimolar doses, and is also more powerful in activating cFos in the ArcN, but not in the NTS and AP, suggesting that the activity of this super-agonist is mostly mediated in the ArcN. This is further substantiated by cFos activation


**Table 2 | Effect of native ghrelin and BIM compounds on the number of GFP-positive cells, number of cFos-positive nuclei, number, and percentage of GFP-positive cells expressing cFos protein in the ArcN in NPY-Renilla GFP mice.**

Coronal sections of the ArcN at approximately 2.0–1.9 mm anterior to the interaural line were quantified unilaterally. Data represent mean ± SEM. \*P < 0.05 and \*\*P < 0.01 vs. vehicle.



Food intake and cFos measured 90 min and GH 15 min post-injection.

in orexigenic NPY ArcN neurons in BIM-28131-treated mice as compared with ghrelin-treated ones. NPY neurons are well-known targets for ghrelin actions (Tannenbaum et al., 2003; Chen et al., 2004) and induction of cFos in NPY neurons in rats was previously demonstrated with other ghrelin agonists (Dickson and Luckman, 1997).

In the current study, BIM-28163 activates several brain nuclei. The intensity of activation seems to be different depending on the region and relative to that of native ghrelin. BIM-28163 appears to have greater activity in the NTS than in the ArcN and AP, two structures outside the blood brain barrier (BBB). Indeed, in the NTS, BIM-28163 is as potent as both native ghrelin and BIM-28131 in inducing cFos with more than a twofold increase as compared with saline treated animals although the effect does not reach statistical significance. In the ArcN, however, BIM-28163 is not as effective as ghrelin in stimulating cFos, and is much less active than BIM-28131. This suggests that BIM-28163 uses alternative pathways other than through these BBB free structures to relay its orexigenic actions. These data are consistent with previous reports in rats showing that BIM-28163 selectively activates cFos in the DMH after icv injections, whereas, in the ArcN, it acts exclusively as an antagonist by blocking ghrelin-induced cFos activation without any intrinsic effect (Halem et al., 2004, 2005). Our data here slightly differ from the above studies in rats because BIM-28163 still activates cFos in the ArcN, although to a much lesser extent than ghrelin. In addition, we were not able to observe any antagonistic actions of BIM-28163 either on food intake or cFos activity when co-administered with ghrelin. Differences may be due to the species studied (mouse vs. rat), to the mode of administration (ip vs. icv), as well as the dose injected (5 nmol/g ip vs. 1.5 nmol/rat icv).

Differences in antagonizing ghrelin-induced food consumption or cFos may be due to the fact that BIM-28163 was injected at a fivefold higher dose than ghrelin instead of a 10-fold higher dose (IC50 for BIM-28163 at GHS-R1a is 10-fold higher than for native ghrelin). However, the antagonistic effect of BIM-28163 on ghrelin-induced cFos activity in rats was still observed when BIM-28163 was administered at a fivefold dose (Halem et al., 2005). The recent report by Costantini et al. (2011) shows that another novel and selective GHS-R1a antagonist with no partial agonist activity, GSK1614343 was also not able to antagonize ghrelin-induced food intake at a dose of 10 mg/kg.

Concerning the VMH, a high level of activation and high interindividual variability in vehicle-treated animals are observed. Consequently, ghrelin seems to have no stimulatory action in this nucleus. This is consistent with other data in rats showing that ghrelin does not activate cFos in the VMH (Lawrence et al., 2002). In contrast, the number of cFos-positive nuclei after treatment with BIM-28163 is almost twofold higher than in the ghrelin-treated group.

Within the hypothalamus, the ArcN is one of the main targets of peripheral signals, such as leptin and ghrelin, which relay information about energy stores and/or nutritional status. The efficiency of the compounds used in this study in activating the ArcN is BIM-28131 > native ghrelin > native ghrelin + BIM-28163 > BIM-28163. The GHS-R1a has been shown to be coexpressed with several neuropeptides in the ArcN. It is expressed on the orexigenic NPY and GHRH neurons (Willesen et al., 1999; Tannenbaum et al., 2003) and these populations of neurons relay ghrelin orexigenic and GH-releasing actions in rats (Tannenbaum et al., 2003; Chen et al., 2004). In mice, ghrelin also induces cFos in NPY-expressing neurons (Wang et al., 2002). To determine whether BIM-28163 and BIM-28131 orexigenic actions could be partly mediated through NPY neurons as is the case for ghrelin, we investigated the effect of these compounds in NPY-GFP mice. Indeed,ip administration of 30 nmol ghrelin during the light cycle induced cFos activation in approximately 26% of NPY neurons (Hassouna et al., 2012a,b). Here we observed that about 15% of NPY-positive cells were activated after BIM-28163 or native ghrelin administration whereas about 20% were activated after BIM-28131 treatment. The orexigenic action of BIM-28163 could be partly mediated through an activation of ArcN NPY neurons. Orexigenic actions of ghrelin may also be mediated by a reduced activity in POMC cells (Cowley et al., 2003). Modified activity of the anorexigenic POMC neurons after treatment with GHS-R1a compounds can thus not be excluded. Orexigenic actions of GSK1614343, another GHS-R1a antagonist,was accompanied by a reduced expression of POMC in the ArcN after chronic treatment with the compound (Costantini et al., 2011).

Although the majority of studies demonstrated that acylation is essential for ghrelin feeding activities (Inhoff et al., 2009), one study showed that desacyl ghrelin was able to stimulate food intake after intracerebroventricular administration by a mechanisms independent of the GHS-R1a (Toshinai et al., 2006). Interestingly,feeding effects of desacyl ghrelin was more pronounced in GHS-R1a deficient mice. Thus it can be postulated that blocking the GHS-R1a with the antagonist may allow desacyl to stimulate feeding through a GHS-R1a independent pathway. Whereas a distinct receptor from the known GHS-R1a would possibly mediate the orexigenic actions of BIM-28163, we can not exclude that these effects are also dependent on the GHS-R1a. Indeed, a recent study using the antagonist, GSK1614343, showed that the appetite-mediated action of this compound was abolished in GHS-R null mice (Costantini et al., 2011), suggesting that the orexigenic effects of GSK1614343 is relayed by GHS-R1a or that the ghrelin receptor may be needed. GHS-R1a is associated with multiple signal transduction pathways (Carreira et al., 2004; Holst et al., 2005) and it is possible that BIM-28163 could activate a specific pathway on the GHS-R1a that is independent from the one mediating GH-releasing activities. In addition, the formation of heterodimers between the GHS-R1a and other receptors has

#### **REFERENCES**


been evidenced (Kern et al., 2012), and raises the question as to whether the orexigenic actions of BIM-28163 could be mediated through interaction with a GHS-R1a dimer that differs from the GHS-R1a receptor form that regulates GH secretion. BIM-28163 could possibly interact with a receptor that needs to dimerize with the GHS-R1a.

The present study demonstrates that two GHS-R1a synthetic ligands, BIM-28163, and BIM-28131, are powerful stimulators of appetite in mice, acting through pathways and brain regions that are distinct from those activated by ghrelin.

In conclusion, utilization of synthetic GHS-R1a ligands, such as BIM-28163 and BIM-28131 that are powerful stimulators of appetite and act through pathways that are distinct from those activated by ghrelin, even in situations when ghrelin seems modestly effective, can have important clinical implications, in conditions such as cachexia or anorexia [*see other chapter in the same issue: Ghrelin: Central and Peripheral Implications in Anorexia Nervosa* (Mequinion et al., 2012)]. BIM-28131 was previously demonstrated to be very efficient in a rat heart failure model of cachexia (Strassburg et al., 2008; Palus et al., 2011). In addition, observations that BIM-28163 is able to selectively stimulate feeding and increase weight gain without altering GH secretion may suggest the possibility of treating pathologies in which hyper-activity of the GH/IGF-1 axis may be deleterious. A better understanding of the molecular pathways activated by these compounds will be useful for devising future therapeutic applications.

#### **ACKNOWLEDGMENTS**

This work was supported by an IPSEN grant to Virginie Tolle and Jacques Epelbaum. We are grateful to Alice Cougnon for the care of the animals and to Julie Cognet for technical assistance. The compounds utilized in this study, BIM-28131 and BIM-28163, are now known as RM131 and RM163. RM131 is currently in clinical development by Rhythm, Boston, MA, USA.


with DRD2 in hypothalamic neurons and is essential for anorexigenic effects of DRD2 agonism.*Neuron* 73, 317–332.


growth hormone secretion in mice. *Endocrinology* 152, 3165–3171.


meal structures in mice. *Peptides* 32, 51–59.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 03 October 2012; paper pending published: 26 October 2012; accepted: 22 February 2013; published online: 18 March 2013.*

*Citation: Hassouna R, Labarthe A, Zizzari P, Videau C, Culler M, Epelbaum J and Tolle V (2013) Actions of agonists and antagonists of the ghrelin/GHS-R pathway on GH secretion, appetite, and cFos activity. Front. Endocrinol. 4:25. doi: 10.3389/fendo.2013.00025*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Hassouna, Labarthe, Zizzari, Videau, Culler, Epelbaum and Tolle. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, providedthe original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Central urocortin 3 and type 2 corticotropin-releasing factor receptor in the regulation of energy homeostasis: critical involvement of the ventromedial hypothalamus

# *Peilin Chen1†, Christine Van Hover 2†, Daniel Lindberg1 and Chien Li 1\**

<sup>1</sup> Department of Pharmacology, University of Virginia Health System, Charlottesville, VA, USA <sup>2</sup> Department of Neuroscience, University of Virginia Health System, Charlottesville, VA, USA

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Tamás Kozicz, Radboud University Nijmegen, Netherlands Eric Zorrilla, The Scripps Research Institute, USA

#### *\*Correspondence:*

Chien Li, Department of Pharmacology, University of Virginia Health System, P.O. Box 800735, 1300 Jefferson Park Avenue, Charlottesville, VA 22908, USA. e-mail: cl4xd@virginia.edu

†Peilin Chen and Christine Van Hover have contributed equally to this work. The vital role of the corticotropin-releasing factor (CRF) peptide family in the brain in coordinating response to stress has been extensively documented. The effects of CRF are mediated by two G-protein-coupled receptors, type 1 and type 2 CRF receptors (CRF1 and CRF2). While the functional role of CRF1 in hormonal and behavioral adaptation to stress is well-known, the physiological significance of CRF2 remains to be fully appreciated. Accumulating evidence has indicated that CRF2 and its selective ligands including urocortin 3 (Ucn 3) are important molecular mediators in regulating energy balance. Ucn 3 is the latest addition of the CRF family of peptides and is highly selective for CRF2. Recent studies have shown that central Ucn 3 is important in a number of homeostatic functions including suppression of feeding, regulation of blood glucose levels, and thermoregulation, thus reinforcing the functional role of central CRF2 in metabolic regulation. The brain loci that mediate the central effects of Ucn 3 remain to be fully determined. Anatomical and functional evidence has suggested that the ventromedial hypothalamus (VMH), where CRF2 is prominently expressed, appears to be instrumental in mediating the effects of Ucn 3 on energy balance, permitting Ucn 3-mediated modulation of feeding and glycemic control. Thus, the Ucn 3-VMH CRF2 system is an important neural pathway in the regulation of energy homeostasis and potentially plays a critical role in energy adaptation in response to metabolic perturbations and stress to maintain energy balance.

#### **Keywords: CRF, Ucn 3, VMH, energy balance, feeding, glucose**

**"fendo-03-00180" — 2013/1/3 — 22:54 — page 1 — #1**

# **INTRODUCTION**

Corticotropin-releasing factor (CRF) was discovered in 1981 by Vale and colleagues at the Salk Institute (Vale et al., 1981). Its existence had been hypothesizedfor many years prior to the 1981 paper characterizing the amino acid sequence of the peptide. Since then, the importance of CRF family of peptides and receptors in stress has steadily emerged. Extensive investigation has demonstrated that CRF is critical in regulating the hypothalamic-pituitaryadrenal (HPA) axis and in integrating endocrine, autonomic, and behavioral responses to stressors (Perrin and Vale, 1999).

In addition to CRF, additional members of the CRF peptide family including urocortins (Ucns) 1, 2, and 3 have been identified in mammals, including humans (Vaughan et al., 1995; Hsu and Hsueh, 2001; Lewis et al., 2001; Reyes et al., 2001). Accumulating evidence suggests that the central actions of Ucn peptides may account for some stress-related effects originally attributed to CRF (Bale and Vale, 2004; Hashimoto et al., 2004; Venihaki et al., 2004; Jamieson et al., 2006; Kuperman et al., 2010). For example, Ucn 1 appears to be involved in the later stage of the stress response and adaptation to stress, while Ucns 2 and 3 may be involved in attenuating the stress response (Ryabinin et al., 2012).

Additionally, the Ucns have been shown to be involved in various physiological regulations including energy balance, cardiovascular function, and behavioral modulation. Ucn 1, 2, and 3 all suppress feeding (Spina et al., 1996; Hashimoto et al., 2004; Fekete et al., 2007), and deficiency in Ucn 2 improves glucose and insulin homeostasis (Chen et al., 2006). Ucn 1 and 2 decrease cardiac output and heart rate, and may be protective against ischemia (Latchman, 2002; Bale et al., 2004; Hashimoto et al., 2004). Behaviorally, Ucn 1 plays a critical role in anxiety-like and depressive behavior, and may be involved in the predisposition of alcohol consumption (Vetter et al., 2002; Ryabinin et al., 2012). Ucn 2 may be linked to depression but not anxiety (Ryabinin et al., 2012). Ucn 2 appears to influence social behavior, including aggression (Breu et al., 2012), as mice deficient in Ucn 2 are less aggressive and prefer passive social interaction. Finally, Ucn 1 and CRF receptors have been found in the auditory system (Graham et al., 2010) and the peptide appears to be involved in the development and maintenance of hearing (Vetter et al., 2002).

Urocortin 3 is the latest addition of the CRF family of peptides, initially identified in the brains of humans and rodents (Hsu and Hsueh, 2001; Lewis et al., 2001). Sequence analyses show that Ucn 3 is more closely related to Ucn 2 than Ucn 1 or CRF. Human and mouse Ucn 3 share 40% homology with human and mouse Ucn 2, but only 21 and 18%, respectively, with human and mouse Ucn 1 and 32 and 26% with CRF (Lewis et al., 2001). Accumulating evidence, as discussed below, has suggested that Ucn 3 is a critical regulator in energy homeostasis.

# **CORTICOTROPHIN-RELEASING FACTOR RECEPTORS**

Two receptors have been identified for CRF: type 1 and type 2 CRF receptors (CRF1 and CRF2), and amino acid sequence analysis has shown that the two receptors share approximately 70% homology (Perrin and Vale, 1999; Bale and Vale, 2004). Both CRF1 and CRF2 are G-protein-coupled receptors with seven transmembrane domains. These receptors signal predominantly through increased cAMP production, but additional signaling pathways including Ca2+, mitogen-activated protein kinase (MAPK), phospholipase C, protein kinase B, and ion channels have also been shown to couple to CRFRs (Kiang, 1997; Grammatopoulos, 2000; Brar et al., 2002). The two receptors differ significantly in their binding affinity for CRF peptides and anatomical distribution within the central nervous system (Chalmers et al., 1995; Van Pett et al., 2000; Hsu and Hsueh, 2001; Lewis et al., 2001). Biochemical studies have shown that CRF binds CRF1 with high affinity while showing modest affinity for CRF2 (Bale and Vale, 2004). Ucn 1 binds both CRF1 and CRF2 with equally high affinity while Ucn 2 and Ucn 3 demonstrate preferential specificity for CRF2. However, while Ucn 2 may bind and stimulate CRF1 at high, pharmacological concentrations (Reyes et al., 2001), Ucn 3 is highly selective to CRF2 and displays minimal affinity for CRF1 (Hsu and Hsueh, 2001; Lewis et al., 2001).

A circulating protein has been identified that binds CRF. It has been suggested that the function of CRF binding protein (CRFBP) is mainly to sequester CRF to reduce its HPA axis stimulation. Levels of CRFBP are elevated in pregnancy to dampen the negative effect of stress responses on the developing fetus (Goland, 1986). In addition to CRFBP, a splice variant of CRF2 has been identified that contains only the extracellular domain of the receptor and shown to circulate, bind, and sequester CRF as well (Bon et al., 1997; Chen et al., 2005). This splice variant also binds Ucn 1 but has very low affinity for Ucn 2 and 3 (Chen et al., 2005).

Anatomical mapping studies have demonstrated that CRF1 and CRF2 have distinct distributions in the brain. CRF1 has a wide distribution throughout the brain with high density in the medial septal area, amygdala, and cerebellum (Chalmers et al., 1995; Van Pett et al., 2000). CRF2 has three major variants that differ in their N-terminal domains: CRF2(a), CRF2(b), and CRF2(c). In murine brains, CRF2(a) is predominantly expressed in the hypothalamus, lateral septum (LS), and dorsal raphe (Chalmers et al., 1995; Van Pett et al., 2000). CRF2(b) is found predominantly in the periphery, including in skeletal muscle, the gastrointestinal tract, and the heart (Kanno, 1999; Wiley and Davenport, 2004; Porcher et al., 2005; Tache and Bonaz, 2007). CRF2(c) is found only in the human brain (Kostich et al., 1998).

The function of CRF1 has been closely associated with the stress response, including the release of adrenocorticotropic hormone (ACTH) from the anterior pituitary and behavioral adaptation to stressors (Bale and Vale, 2004). The physiological role of CRF2, however, is less defined. Functional studies have shown this receptor is involved in an array of homeostatic regulations, with most of its actions regulating energy balance by modulating feeding, blood glucose levels, and energy expenditure. This review will focus on recent advances in the understanding of the physiological role of CRF2 and Ucn 3 in the brain, particularly in the hypothalamus, in the regulation of energy balance.

**"fendo-03-00180" — 2013/1/3 — 22:54 — page 2 — #2**

# **CRF2 IN ENERGY HOMEOSTASIS**

Loss-of-function studies with CRF2 null mice and pharmacological studies with CRF2 agonists both suggest that endogenous CRF2 plays a physiological role in energy balance. Evidence from several studies that examined ingestive behavior of CRF2 null mice showed that endogenous CRF2 is required for the anorectic effect of CRF peptide and is involved in control of the meal size during active phase of eating and following acute exposure to the stress (Bale et al., 2000; Coste et al., 2000; Pelleymounter et al., 2000; Tabarin et al., 2007). Similarly, infusion of antisense oligonucleotides to CRF2 mRNA attenuates both CRF and Ucn 1 induced hypophagia and corticosterone secretion (Smagin et al., 1998). Furthermore, CRF2 deletion protects mice from high-fat diet-induced insulin resistance and glucose intolerance (Bale et al., 2003). Taken together, it is clear that CRF2 is responsible for mediating the effect of CRF family peptides on the suppression of feeding and is involved in corticosterone secretion and glucose homeostasis.

In addition to feeding, CRF2 is involved in regulating energy expenditure. CRF2 knockout (KO) mice have more active metabolism and lose heat faster than wildtype (WT) mice (Carlin et al., 2006), suggesting an exaggerated level of sympathetic activity. Moreover, CRF2 KO mice have higher brown adipose tissue (BAT) temperature and, when given a choice between room temperature and warm areas, prefer warmer areas more than WT mice (Carlin et al., 2006). The KO mice also have higher oxygen consumption and carbon dioxide production and reduced respiratory exchange rate (Carlin et al., 2006), indicating a preference in fatty acid oxidation over carbohydrate utilization in the KO mice. It was suggested that a lack of functional CRF2 leads to elevated CRF1 activity, which consequently increases sympathetic nervous system (SNS) activity to promote lipolysis (Carlin et al., 2006). This hypothesis appears to disagree with pharmacological studies, as activation of central CRF2 (discussed below), in most cases, results in increased SNS activity. This apparent discrepancy may result from a number of possibilities including compensatory mechanisms due to total body KO of the receptor as compared to acute, local stimulation of the receptor in the brain. Obviously more studies are needed to further elucidate this issue.

Leptin, a hormone secreted by adipocytes, is a potent appetite suppressant (Uehara et al., 1998; Zigman, 2003). A number of studies suggest that its effect on feeding may involve the CRF receptor system. Though leptin treatment greatly decreases food intake, when co-administered with a non-selective CRF receptor antagonist, food intake remains at a nearly normal level (Gardner et al., 1998), suggesting the CRF system is a downstream target of leptin in the brain. However, this notion was recently challenged by a study (Harris, 2010) demonstrating that CRF2 is not essential for the effects of leptin on energy balance, including feeding and body weight regulation. Again this discrepancy may be due to the nature of global CRF2 KO, which potentially results in functional compensation such as exaggerated CRF1 activity in these mice (Carlin et al., 2006). Therefore, central or specific brain area deletions of CRF2 may be necessary to further evaluate the interaction of leptin and the CRF system in the brain.

# **UROCORTIN 3**

#### **ANATOMICAL LOCATION OF Ucn 3**

Urocortin 3 is found both in the periphery and in the brain. In the periphery, it is expressed in the digestive tract, muscle, thyroid and adrenal glands, pancreas, heart, spleen, and skin (Hsu and Hsueh, 2001; Lewis et al., 2001). In the brain, neurons expressing Ucn 3 are concentrated in the medial amygdala (MeA) and hypothalamus (Lewis et al., 2001; Li et al., 2002). In the hypothalamus, the major Ucn 3 cell population is near the rostral perifornical hypothalamic area (rPFA; Li et al., 2002). Specifically, Ucn 3-positive cells are gathered around the fornix lateral to the paraventricular nucleus of the hypothalamus (PVH). This group extends rostrally and stays close to the fornix into the posterior part of the bed nucleus of the stria terminalis (pBNST) and medially into the anterior parvicellular part of the PVH (PVHap). A recent study has elucidated anatomical heterogeneity within this hypothalamic Ucn 3 cell population, as neurons of the rostral part (PVHap/pBNST) project to the ventromedial hypothalamus (VMH), and those of the caudal part, residing in the rostral perifornical hypothalamus (rPFH), projects to the LS (Chen et al., 2011). A second group of Ucn 3 positive cells is found in the median preoptic nucleus (MnPO; Li et al., 2002). In the forebrain, prominent Ucn 3 nerve fibers and terminals are found in the VMH, LS, MeA, and BNST (Li et al., 2002). These areas also express high levels of CRF2 (Chalmers et al., 1995; Van Pett et al., 2000). This overlap of Ucn 3 and CRF2 distribution and the high affinity of the peptide for the receptor strongly suggest that Ucn 3 is an endogenous ligand for CRF2 in these brain areas.

#### **METABOLIC EFFECTS OF Ucn 3**

All 3 Ucns bind CRF2 and may each be responsible for some of the receptor's energy homeostatic effects. However, Ucn 3 alone has myriad metabolic effects.

# *Feeding*

Central administration, KO, and overexpression studies reveal a role of Ucn 3 in the regulation of feeding. When directly infused into the lateral ventricles, Ucn 3 decreases nocturnal food and water intake in a dose dependent manner, primarily due to decreased meal frequency, and this effect was eliminated with concomitant CRF2 antagonist treatment (Fekete et al., 2007). The anorectic effect of Ucn 3 is not due to distaste for food, as no concurrent taste aversion develops (Fekete et al., 2007). Consistent with pharmacological evidence, genetic Ucn 3 deficiency appears to lead to overeating. Though Ucn 3 KO mice have similar body weights toWT animals, KO mice eat more and have increased accumulated food intake. Similar to CRF2 null mice, Ucn 3 KO mice exhibit elevated nocturnal feeding, when greatest spontaneous food intake naturally occurs (Chao et al., 2012). Taken together, it is clear that Ucn 3 in the brain functions as a potent anorectic agent. However, a study that used genetic overexpression of Ucn 3 challenged this notion. Under regular chow-fed condition, mice with overexpression of Ucn 3 (*Ucn3*+) have higher body mass-adjusted food intake than WT and are heavier than WT controls due to increased lean body mass (Jamieson et al., 2011). On the other hand, *Ucn3*+ mice do not gain as much weight as WT mice when fed with a high-fat diet (Jamieson et al., 2011). This

A temporally and spatially controlled viral approach to overexpress Ucn 3 in the rPFH shows that Ucn 3 in the rPFH does not modulate food intake (Kuperman et al., 2010), as mice with Ucn 3 overexpression in the rPFH ingest consume similar amount of food as control mice (Kuperman et al., 2010). The rPFH-specific Ucn 3-overexpressing mice show a trend toward being heavier, but retain the same fat–lean mass percentages as control mice (Kuperman et al., 2010). As mentioned above, Ucn 3 cells in the rPFH project mainly to the LS with minimal projection to the VMH. Therefore, it is reasonable to assume that CRF2 in the LS will be overstimulated in this mouse model. Interestingly, a number of studies have shown that CRF2 in the LS is involved in suppression of food intake (Wang and Kotz, 2002; Bakshi et al., 2007). Currently, it is unclear as to why overexpression of Ucn 3 in the rPFH fails to suppress feeding. The expression of CRF receptor has been shown to be subject to ligand-induced receptor down-regulation (Rabadan-Diehl et al., 1996; Eghbal-Ahmadi et al., 1997). It is conceivable that chronic elevated Ucn 3 input to the LS in Ucn 3 rPFH overexpression mice may lead to alteration in CRF2 expression and consequently reduced response to Ucn 3 stimulation in the LS. Thus, it is possible that acute stimulation of CRF2 in the LS suppresses feeding, but chronic stimulation of CRF2 in the LS in mice with Ucn 3 overexpression in the rPFH may lead to negative feedback to balance the effect of CRF2 in feeding. Clearly, more studies are needed to elucidate the effect of Ucn 3 overexpression in the regulation of food intake. In addition to Ucn 3, Ucn 1 neurons in the midbrain Edinger–Westphal nucleus have been shown to innervate the LS (Kozicz et al., 1998; Bittencourt et al., 1999). Therefore, both Ucn 1 and 3 may contribute to the effect of CRF2 on feeding in the LS.

The mechanism of Ucn 3-induced anorexia has not been directly studied. Central administration of CRF2 agonists, including Ucn 3, have been shown to inhibit gastric emptying (Martinez et al., 2004; Stengel and Tache, 2009), and elevate blood glucose levels (Jamieson et al., 2006; Chen et al., 2010). Both reduced gut motility and hyperglycemia have been shown to induce satiation and reduce feeding (Ritter, 2004; Cummings and Overduin, 2007; Wolfgang et al., 2007; Cha et al., 2008). Therefore, multiple mechanisms are potentially involved in mediating the anorectic effect of Ucn 3 in the brain.

#### *Energy homeostasis*

"fendo-03-00180" — 2013/1/3 — 22:54 — page 3 — #3

Similarly to CRF2, Ucn 3 is also involved in energy expenditure. Central administration of Ucn 3 increases motor activity (Ohata and Shibasaki, 2004). Similarly, transgenic Ucn 3 overexpression mice (*Ucn3*+) have an increased respiratory exchange ratio and increased motor activity in their home cages (Jamieson et al., 2011). Furthermore, mice with Ucn 3 overexpressed in the rPFH had an increased respiratory exchange ratio and elevated heat production (Kuperman et al., 2010). This is consistent with the notion that Ucn 3 is involved in SNS activity and energy homeostasis. Ucn 3 KO mice, on the other hand, do not show the same pattern; there are no differences in oxygen consumption, heat production, or activity levels (Chao et al., 2012). Therefore, KO of Ucn 3 in specific brain areas may provide better insight into the role of specific populations of Ucn 3 cells in the brain in energy expenditure.

# *Glucose homeostasis*

Several studies have revealed a complex role of Ucn 3 in glucose homeostasis. Though adult Ucn 3 KO mice fed a chow diet show no differences in glucose tolerance and insulin sensitivity compared to WT mice (Li et al., 2007), Ucn 3 KOs have lower basal insulin levels and show a greater rebound in blood glucose levels after the initial hypoglycemia during an insulin tolerance test (Li et al., 2007; Chao et al., 2012). Furthermore, under high-fat diet feeding, adult KO mice are more metabolically resilient. The Ucn 3 KO mice have lower plasma insulin and blood glucose concentrations than WT mice, remain sensitive to insulin and do not develop glucose intolerance and liver steatosis with the same frequency of WT mice (Li et al., 2007). Moreover, aged KO mice show better glucose homeostasis than age-matched WT mice (Li et al., 2007). Overall, Ucn 3 deficiency appears to protect the mice from metabolic disorders caused by high-fat feeding. It is noteworthy that Ucn 3 is expressed in pancreatic β cells and has been shown to play a critical role as a local regulator in insulin secretion (Li et al., 2003). Thus, it is likely that Ucn 3 in both the brain and in the periphery, especially in the pancreas, contribute to the phenotypes observed in Ucn 3 null mice.

On the other hand, *Ucn3*+ transgenic mice also appear to be protected against excessive metabolic challenge, having decreased fed and fasting blood glucose levels and increased tolerance to glucose when challenged in a glucose tolerance test (Jamieson et al., 2011). Fasting insulin levels in *Ucn3*+ mice are also lower than that in WT mice, though an insulin tolerance test shows no differences in insulin sensitivity (Jamieson et al., 2011). When challenged with a high-fat diet, *Ucn3*+ mice fare better than the WT littermates, maintaining normal body weight and low blood glucose levels, but display comparable insulin sensitivity to theWT control mice (Jamieson et al., 2011). Though *Ucn3*+ mice show improved glucose homeostasis and insulin sensitivity, overexpression of Ucn 3 in the rPFH produces the opposite effect; rPFH Ucn 3-overexpressing mice show reduced insulin sensitivity and increases basal insulin levels, however they show no difference in glucose tolerance (Kuperman et al., 2010).

While genetic KO of Ucn 3 appears to have metabolic protective qualities, the effect of Ucn 3 overexpression is unclear. Full body overexpression of Ucn 3 seems protective, while targeted overexpression within the rPFH appears metabolically detrimental. Interestingly, *Ucn3*+ mice have lower fasting blood glucose levels and higher energy intake under basal conditions (Jamieson et al., 2011). As stated earlier, CRF2 is expressed abundantly in a number of peripheral tissues including skeletal muscle (Perrin et al., 1995; Stenzel et al., 1995; Wiley and Davenport, 2004; Porcher et al., 2005; Kuperman et al., 2011). Stimulation of muscle CRF2 has been shown to promote thermogenesis (Solinas et al., 2006). Moreover, muscle CRF2 is involved in regulating skeletal muscle mass (Hinkle et al., 2003) and consistent with this notion, *Ucn3*+ mice have increased muscle mass (Jamieson et al., 2011). Thus, it is conceivable that the improved glucose homeostasis of *Ucn3*+ mice is due, at least in part, to stimulation of muscle CRF2 by ectopic overexpression of Ucn 3 in the periphery.

# *Thermoregulation*

Functional studies have shown that Ucn 3 is involved in thermoregulation, potentially acting on brown fat. It was found that Ucn 3 induced a significant increase in body temperature, from 37.2 to 38.6◦C (99.0 to 101.5◦F), when injected into the lateral ventricles of rats (Telegdy et al., 2006). Temperature gradually decreased after peaking 2 h after Ucn 3 administration, but remained significantly elevated for a total of 4 h (Telegdy et al., 2006). Moreover, pretreating animals with CRF2 but not CRF1 antagonists completely blocked Ucn 3-induced hyperthermia, indicating that the pyrogenic action of Ucn 3 is mediated by CRF2 (Telegdy and Adamik, 2008). Noraminophenazone, a cyclooxygenase inhibitor, simultaneously applied with Ucn 3 also prevented the temperature increase and also attenuated the increase when administered 30 min after Ucn 3 treatment (Telegdy et al., 2006). This indicates that the arachidonic acid cascade forming prostaglandin is a downstream target of central Ucn 3 and CRF2 system in thermoregulation. Currently, the brain loci that may mediate the effect of Ucn 3 in body temperature remain elusive. The MnPO is known to be a center for thermoregulation and expresses high concentrations of the prostaglandin receptor EP3 (Morrison et al., 2008). It has been shown that prostaglandins play an important role in the MnPO through EP3 to regulate body temperature (Morrison et al., 2008). Furthermore, the presence of a group of Ucn 3 cells in the MnPO (Li et al., 2002) suggests that Ucn 3 might be involved in MnPO mediates pyrogenic effects.

#### **REGULATION OF Ucn 3 EXPRESSION IN THE BRAIN**

The expression of Ucn 3 in the brain has been determined in a number of stress paradigms and metabolic challenges. It was found that restraint stress rapidly elevates Ucn 3 gene expression in the MeA and that the elevated Ucn 3 mRNA levels return to basal levels 4 h after the stress (Jamieson et al., 2006). Restraint stress also increases Ucn 3 mRNA levels in the rPFA but with a slower time course compared to that of the MeA (Venihaki et al., 2004; Jamieson et al., 2006). Adrenalectomy greatly elevates Ucn 3 expression in the rPFA,while corticosterone replacement returns the expression to a basal level (Jamieson et al., 2006). This result indicates that corticosterone may be involved in stress-mediated Ucn 3 gene expression in the rPFA. Hemorrhage decreases Ucn 3 expression in the MeA after 30 min, and 48 h of food deprivation also decreases Ucn 3 expression in the MeA (Jamieson et al., 2006).

The expression of Ucn 3 has also been examined in genetically obese rodent models. Food deprivation increases Ucn 3 mRNA expression in the dorsal part of the medial amygdala (MeD) in obese Fa/Fa Zuker rats and has no effect on Ucn 3 expression in the rPFH (Poulin et al., 2012). In contrast, lean Fa/? rats show increased Ucn 3 expression in the rPFH but not the MeD after food deprivation (Poulin et al., 2012). Ucn 3 mRNA expression returns to normal after 24 h of refeeding (Poulin et al., 2012). In ob/ob obese mice, Ucn 3 expression is significantly reduced in the MeA (**Figure 1**; Li and Vale, 2002), and leptin treatment reverses Ucn 3 expression in this area. Interestingly, pair-feeding in ob/ob mice

"fendo-03-00180" — 2013/1/3 — 22:54 — page 4 — #4

failed to modulate Ucn 3 expression in the MeA. These results indicate that Ucn 3 expression in the MeA is regulated by leptin. Taken together, these studies further support the notion that endogenous Ucn 3 in the brain is sensitive to metabolic signals and energy status and potentially plays an important role in regulating energy homeostasis.

#### **THE VENTROMEDIAL HYPOTHALAMUS**

The anatomical distribution of CRF2 in the brain has provided important insight into possible areas that mediate the effects of CRF2 in energy balance. The VMH has received significant attention due to its abundant expression of CRF2 and well-known role in regulating energy homeostasis, feeding, and blood glucose levels.

The VMH is parceled on cytoarchitectonic grounds into three major divisions: dorsomedial (VMHdm), central, and the ventrolateral (VMHvl) parts (Gurdjian, 1927). The VMH volume is larger in males compared with females, and this difference is largely accounted for by the VMHvl, which in females is significantly smaller than that in males (Dugger et al., 2007). Further, VMHvl in female rats express higher levels of estrogen receptors than that in males, and has been shown to play a critical role in regulating lordosis behavior in females (Flanagan-Cato, 2011) and aggressiveness in males (Lin et al., 2011).

The VMH has long been considered a critical brain area in the regulation of energy homeostasis. Lesion of the VMH results in hyperphagia, hyperinsulinemia, reduction of SNS activity, increase of fat mass, and reduction of energy expenditure that ultimately leads to storage of excess of energy and obesity (Bernardis and Frohman, 1971; Inoue et al., 1977; Cox and Powley, 1981; Niijima et al., 1984; Sakaguchi et al., 1988; Ruffin and Nicolaidis, 1999; King, 2006). Conversely, stimulation of the VMH results in predominately opposite phenotypes including induction of satiety, increase in SNS activity, lipolysis, and thermogenesis (King, 2006). In recent years, the importance of the VMH in energy homeostasis has been further ascertained with the aid of improved molecular tools and mouse genetics (Sternson et al., 2005; King, 2006; Chao et al., 2012). For example, mice with VMH-specific deletion of a number of genes including the leptin receptor (Dhillon et al., 2006; Bingham et al., 2008), estrogen receptor α (Musatov et al., 2006, 2007), and vesicular glutamate transporter-2 (VGLUT2; Tong et al., 2007) result in a number of abnormalities such as increased feeding, reduced energy expenditure, impaired glucose homeostatic regulation, and obesity. Moreover, mice bearing a deletion of steroidogenic factor 1 (SF1), a transcription factor involved in steroidogenesis that is highly enriched in the VMH, show similar phenotypes in energy homeostasis and are obese (Luo et al., 1994; Sadovsky et al., 1995; Shinoda et al., 1995).

# **EXPRESSION OF CRF2 IN THE VMH**

"fendo-03-00180" — 2013/1/3 — 22:54 — page 5 — #5

The VMH is one of the brain areas with prominent CRF2 expression (Chalmers et al., 1995;Van Pett et al., 2000). The expression is concentrated in the dorsomedial part of the nucleus with decreasing density toward the ventrolateral division of the VMH. The expression of CRF2 in the VMH has been shown to be sensitive to energy status and stress. For example, leptin injection increases and fasting decreases CRF2 expression in the VMH and ob/ob mice or Fa/Fa Zucker rats have lower CRF2 in the VMH compared to WT controls (Richard et al., 1996; Makino et al., 1998, 1999; Nishiyama et al., 1999). Furthermore, restraint stress and glucocorticoids increase CRF2 levels in the VMH (Makino et al., 1998, 1999). Thus, these data support the notion that CRF2 in the VMH is important in regulating energy balance.

Currently, detailed neurochemical phenotypes of CRF2 positive cells within the VMH remain unclear. Several proteins have been found to be expressed in the VMH including SF1, pituitary adenylate cyclase-activating polypeptide, the leptin receptor, and VGLUT2 (Fei et al., 1997; Elmquist et al., 1998; Ziegler et al., 2002; Segal et al., 2005; Kurrasch et al., 2007). It has been shown that CRF2 extensively colocalizes withVGLUT2 in theVMH (Chen et al., 2010). VGLUT2 mediates glutamate uptake into synaptic vesicles of excitatory neurons (Fremeau et al., 2001; Herzog et al., 2001; Takamori et al., 2001) and has been used extensively as a marker for excitatory glutamatergic neurons. The colocalization of CRF2 and VGLUT2 suggests that CRF2 is expressed predominately in excitatory neurons in theVMH. As mentioned above, SF1 is a nuclear receptor that regulates the transcription of key genes involved in sexual development and reproduction (Parker et al., 2002). In adults, SF1 expression is specifically confined to the VMH (Parker et al., 2002). In SF1 null mice, CRF2 mRNA expression is nearly undetectable in the VMH (Luo et al., 1994; Sadovsky et al., 1995; Shinoda et al., 1995), suggesting CRF2 is expressed in SF1 cells in the VMH. Consistent with this notion, it was found that more than 90% of CRF2 neurons in the dorsomedial part of the nucleus are also SF1-positive (**Figure 2**) with less colocalization of these two materials in the VMHvl (Digruccio et al., 2007).

#### **CRF LIGANDS INPUT INTO THE VMH**

The anatomical distribution of a number of the CRF family peptides has been determined, and it was found that CRF and Ucn 1-expressing neurons provide minor innervation to the VMH (Swanson et al., 1983; Kozicz et al., 1998; Bittencourt et al., 1999). Interestingly, low levels of Ucn 1-immunoreactivity have been detected in cells in the VMH (Kozicz et al., 1998), suggesting Ucn 1 may be a local factor in the nucleus. Although Ucn 2 fiber distribution has not been determined, Ucn 2-positive cells have been found in a number of brain areas including the magnocellular part of the PVH, locus ceruleus, and facial motor nucleus (Reyes et al., 2001), and none of these areas provide extensive projection into the VMH (McBride and Sutin, 1977; Luiten and Room, 1980; Berk and Finkelstein, 1981; Kita and Oomura, 1982; Zaborszky, 1982; Fahrbach et al., 1989; Chen et al., 2011). Compared to other CRF innervations, Ucn 3 neuronal fibers abundantly innervate the VMH (Li et al., 2002). Similar to the expression of CRF2 in the VMH, Ucn 3-positive axonal fibers and terminals concentrate in the dorsomedial part of the VMH with reduced density toward the ventrolateral part of the nucleus (Li et al., 2002). As stated above, Ucn 3 neurons in the PVHap provide the major Ucn 3 afferent input into the VMH and Ucn 3 cells in the pBNST and

**FIGURE 2 | (A)** Darkfield photomicrograph showing CRFR2 mRNA signal revealed by in situ hybridization in the basal hypothalamic area of a transgenic mouse expressing Cre recombinase (Cre) and enhanced yellow fluorescent protein (EYFP) in SF1-positive cells. Note that CRFR2 mRNA hybridization signal (white clusters) was abundant in the dorsomedial part of the VMH (VMHdm). **(B)** Bright field photomicrograph of the same area in **(A)** showing SF1-positive cells, revealed by immunostaining with anti-green fluorescent protein antibody (darkbrown precipitates) in the VMH. **(C)** High magnification of boxed area in **(A)** showing colocalization of CRFR2 (black dot clusters) and SF1 (brown precipitates) in the VMH. ARH, arcuate nucleus of hypothalamus; V3, third ventricle; VMHc, central part of the VMH; VMHdm, dorsomedial part of the VMH; VMHvl, ventrolateral part of the VMH. Scale bar = 50 μm **(B)**, 20 μm **(C)**.

"fendo-03-00180" — 2013/1/3 — 22:54 — page 6 — #6

the MeA comparatively provide moderate input into the nucleus (Chen et al., 2011). Interestingly, Ucn 3 cells in the rPFH, immediately caudal and adjacent to the PVHap, provide minimal Ucn 3 afferent input into the VMH and instead send strong innervation into the LS.

Neural inputs into the PVHap and MeA have been studied (Sawchenko and Swanson, 1983; Swanson and Petrovich, 1998; Uehara et al., 1998; Maras and Petrulis, 2010; Carrillo et al., 2011; Northcutt and Lonstein, 2011; Radley and Sawchenko, 2011; Van Hover et al., 2011). In general, the two brain areas receive similar afferent input from a number of brain regions including the septal nuclei and amygdala. However, a few subtle but significant exceptions should be noted. The PVHap receives prominent inputs from the hypothalamus, cortex, and brainstem, whereas the MeA receives strong inputs from the bed nucleus of the accessory olfactory tract and nucleus of stria medullaris. These findings suggest Ucn 3 cells in the PVHap receive inputs that transmit visceral and autonomic information while the Ucn 3 cell group in the MeA receives afferents that relay olfactory information. Thus, it is conceivable that central Ucn 3 may serve as a common peptide neurotransmitter in the various neural pathways that convey information from assorted neural signals into the VMH to coordinate the regulation of energy homeostasis.

# **FUNCTION OF CRF2 IN THE VMH**

When the function of CRF2 in the VMH was first assessed by injecting Ucn 1 into the VMH, it was found that the peptide potently suppresses food intake (Ohata et al., 2000). However, because CRF1 has been suggested to be expressed in the VMH (Cheng et al., 2007) and Ucn 1 has equally high affinity for both CRF1 and CRF2 (Vaughan et al., 1995), it remains possible that CRF1 may also contribute to the anorectic effect of Ucn 1 in the VMH. More recently, when the CRF2-selective ligand Ucn 3 was identified (Lewis et al., 2001), the function of CRF2 in the VMH was re-examined by site-specific injection of Ucn 3 into the VMH (Fekete et al., 2007; Chen et al., 2010). Consistent with the Ucn 1 injection study, stimulation of CRF2 by Ucn 3 in the VMH significantly suppresses feeding. Moreover, stimulation of CRF2 in other regions including the PVH, amygdala, and the lateral hypothalamus fails to modulate feeding (Ohata et al., 2000; Fekete et al., 2007; Chen et al., 2010), reinforcing the notion that VMH CRF2 plays a critical role in mediating the effect of CRF peptides in suppressing food intake.

In addition to appetite suppression, activation of CRF2 in the VMH results in rapid elevation of blood glucose levels (Chen et al., 2010). This is consistent with the function of VMH neurons in glucose homeostasis, as VMH neurons have been suggested to play an important role in regulating glucose levels via glucose sensing, and modulating glucose production in peripheral tissues (Kang et al., 2004; Levin et al., 2004). On the other hand, it has been shown that stimulation of CRF2 in the VMH suppresses insulininduced release of glucagon and epinephrine (McCrimmon et al., 2006), indicating thatVMH CRF2 exerts a negative control over the counterregulatory response (CRR). Taken together, it is possible that the functional role of CRF2 in the VMH in glucose homeostasis is context dependent. When blood glucose is low as a result of hyperinsulinemia, CRF2 in the VMH negatively regulates the CRR response. On the other hand, under normoglycemia, CRF2 induces acute hyperglycemia to facilitate fuel mobilization perhaps in response to stress. We have also found that CRF2-positive neurons in the VMH are sensitive to glucose, as high glucose inhibits and low glucose stimulates the neuronal activity (Digruccio et al., 2007).

Type 2 CRF receptor has been shown to modulate the HPA axis. For example, CRF2 KO mice display altered HPA hormonal secretion, and central Ucn 3 injection facilitates stress-induced ACTH secretion (Jamieson et al., 2006). On the other hand, activation of CRF2 in the VMH fails to modulate HPA hormone secretion (Chen et al., 2010), indicating that the receptor in the VMH is not essential for central Ucn 3-induced HPA activation. Thus, CRF2 positive brain loci that are important for modulation of the HPA axis remain to be determined.

# **PHYSIOLOGICAL ROLE OF CRF2 IN THE VMH**

To probe the physiological role of endogenous CRF2 in the VMH, VMH-specific CRF2 knockdown mice were generated by injection of a lentiviral vector expressing CRF2 small hairpin RNA (shRNA; Chao et al., 2012). Mice injected with CRF2 shRNA displayed significantly reduced CRF2 mRNA levels and gain more weight, mostly in white fat, than control mice. Furthermore, similar to Ucn 3 null mice, mice with reduced CRFR2 in the VMH exhibited elevated basal food intake and ate more than the control mice after overnight fasting. This result indicates that CRF2 in the VMH serves as a brake to facilitate the cessation of feeding. This study suggests that CRF2 in the VMH plays a critical role in mediating the effect of central Ucn 3 in energy balance.

In addition to elevated feeding, mice with decreased expression of CRF2 in the VMH display reduced lipolysis and increased adiposity in white fat (Chao et al., 2012). This is likely due to reduced SNS activity, as the VMH has been shown to regulate lipolysis via sympathetic outflow (Kumon et al., 1976; Takahashi and Shimazu, 1981; Ruffin and Nicolaidis, 1999). Interestingly, CRF2 knockdown in the VMH has no major impact in heat production or uncoupling protein 1 expression in BAT, suggesting that CRF2 in the VMH does not significantly modulate thermogenesis in BAT. This result appears to disagree with earlier reports that VMH is involved in regulating thermogenesis in BAT (Perkins et al., 1981; Kim et al., 2011). Anatomical studies have demonstrated a compartment-specific organization of innervation of different peripheral organs, as abdominal fat and subcutaneous fat are innervated by different neural pathways (Kreier et al., 2002, 2006). It is possible that CRF2-positive neurons are a subpopulation of cells in the VMH that regulate SNS outflow to white fat without significant functional impact on BAT. Heterogeneity of VMH neurons has been reported as overlapping but distinct subpopulation of neurons in the VMHvl that are critical in regulating fighting and mating (Lin et al., 2011).

Consistent with the function of CRF2 in regulating blood glucose levels, mice with reduced CRF2 expression in the VMH show improved glucose homeostasis compared to control mice (Chao et al.,2012). Moreover, mice with CRF2 knocked down in theVMH exhibit an exaggerated rebound in blood glucose levels compared to control mice after the initial hypoglycemic response to insulin challenge. This result agrees with the study by McCrimmon et al.

"fendo-03-00180" — 2013/1/3 — 22:54 — page 7 — #7

(2006), who found that injection of Ucn 3 into the VMH suppresses the hypoglycemia-induced CRR response. Taken together, these studies strongly argue that CRF2 is a critical molecular mediator in VMH regulation of glucose homeostasis.

#### **NEUROCIRCUITS UNDERLYING THE EFFECT OF VMH**

To understand howVMH CRF2 neurons regulate output functions and to describe an anatomical link between these neurons and the SNS, it is necessary to determine their axonal projections to identify downstream targets in the brain. A number of anterograde tracing studies utilizing different tracers have been performed to evaluate the projections of VMH neurons. Generally, it was found that VMH neurons project extensively to neighboring hypothalamic nuclei including the anterior and paraventricular nucleus, BNST, and periaqueductal gray (PAG; Saper et al., 1976; Canteras et al., 1994). These anatomical studies raise an interesting dilemma. Although it is clear that VMH activity modulates SNS activity, these studies failed to observe directVMH efferents within well-known autonomic centers in the brainstem. Thus, it was concluded that the VMH likely modulates SNS activity indirectly by first projecting to a relay center such as the PAG.

Recently, using a conditional viral tracing approach, we have found that VMH neurons project to a number of important brainstem autonomic centers including the parabrachial nucleus, C1 catecholaminergic cell group in the rostral ventrolateral medulla, and the nucleus of solitary tract (Lindberg et al., 2011). Moreover, we have used the same approach to find that CRF2-positive cells in the VMH show similar axonal projections to these brainstem areas (**Figure 3**; Lindberg et al., 2011). These studies demonstrate that VMH neurons, including cells that express CRF2, can potentially modulate SNS activity by direct projections to brainstem autonomic centers.

# **CONCLUSION**

The function of CRF peptides and their receptors in coordinating hormonal, neuronal, and behavioral responses to stress is well recognized. Pharmacological studies have determined that the CRF2 receptors are involved in the regulation of energy homeostasis. Recent studies using various genetic mouse models and molecular tools have further ascertained the critical role of CRF2 and its selective ligands, including Ucn 3, in feeding, blood glucose regulation, SNS output, and peripheral metabolism. Moreover, CRF2 in the VMH mediates most, if not all, of the effects of central Ucn 3 on energy homeostasis.

It is clear that conflicting results have been observed between whole body and region- or tissue-specific KO or overexpression mouse models. Furthermore, several studies have demonstrated that anatomical or even functional heterogeneity exists within a seemingly single Ucn 3 cell population in the hypothalamus. Thus, a more detailed understanding of the physiological function of CRF2 and its selective ligands in the brain will be aided by brain region-specific transgenic animal models permitting manipulation of ligand or receptor expression. Study of such models will provide insight into the specific roles of CRF2 in modulating metabolic functions.

Regulation of energy balance under diverse challenges including stress, starvation, or high-fat diet requires numerous adaptive

"fendo-03-00180" — 2013/1/3 — 22:54 — page 8 — #8

mechanisms in both central and peripheral tissues. It is now clear that central CRF2 and Ucn 3 are involved in this regulation, as the expression of the receptor and ligands are closely regulated under these challenges. It is thus conceivable that dysregulated CRF2 or ligand function potentially cause or exacerbate metabolic perturbations. This hypothesis can be easily tested with the above mentioned rodent models to determine the functional role of the CRF2 system in metabolic diseases. Moreover, a better understanding of the molecular mechanisms by which various stressors or metabolic signals regulate the expression and/or function of CRF2 and its ligand will provide significant insight into the potential role of the CRF family and its receptors in the pathophysiology of metabolic disorders including obesity and diabetes.

#### **REFERENCES**


CRF receptors. *J. Comp. Neurol.* 415, 285–312.


# **ACKNOWLEDGMENTS**

We thank Dr. Ruth Stornetta for providing comments on the manuscript. This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases Grant R01 DK-078049 (to Chien Li).

ventromedial nucleus of hypothalamus is critical in regulating feeding and lipid metabolism in white adipose tissue. *Endocrinology* 153, 166–176.


"fendo-03-00180" — 2013/1/3 — 22:54 — page 9 — #9

neurons in the VMH, and this action by leptin is required for normal body-weight homeostasis. *Neuron* 49, 191–203.


The expression of vesicular glutamate transporters defines two classes of excitatory synapse. *Neuron* 31, 247–260.


parasympathetic innervation of subcutaneous and intra-abdominal fat – functional implications. *J. Clin. Invest.* 110, 1243–1250.


"fendo-03-00180" — 2013/1/3 — 22:54 — page 10 — #10


of thermogenesis in mammals. *Exp. Physiol.* 93, 773–797.


a cDNA expressed in heart. *Proc. Natl. Acad. Sci. U.S.A.* 92, 2969–2973.


glands and gonads but express P450 side-chain-cleavage enzyme in the placenta and have normal embryonic serum levels of corticosteroids. *Proc. Natl. Acad. Sci. U.S.A.* 92, 10939–10943.


"fendo-03-00180" — 2013/1/3 — 22:54 — page 11 — #11

hormone expressed in the heart. *Mol. Endocrinol.* 9, 637–645.


that stimulates the secretion of corticotropin and β-endorphin. *Science* 213, 1394–1397.


of the corticotropin-releasing hormone family: modulation by stress and attenuation of some anxiety-like behaviours. *J. Neuroendocrinol.* 16, 411–422.


Shulman, G. I., et al. (2007). Regulation of hypothalamic malonyl-CoA by central glucose and leptin. *Proc. Natl. Acad. Sci. U.S.A.* 104, 19285– 19290.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that

"fendo-03-00180" — 2013/1/3 — 22:54 — page 12 — #12

could be construed as a potential conflict of interest.

*Received: 01 October 2012; paper pending published: 22 October 2012; accepted: 18 December 2012; published online: 07 January 2013.*

*Citation: Chen P, Van Hover C, Lindberg D and Li C (2013) Central urocortin 3 and type 2 corticotropin-releasing factor receptor in the regulation of energy homeostasis: critical involvement of the ventromedial hypothalamus. Front. Endocrin. 3:180. doi: 10.3389/fendo.2012.00180*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Chen, Van Hover, Lindberg and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Oxytocin, feeding, and satiety

# **Nancy Sabatier, Gareth Leng and John Menzies\***

Centre for Integrative Physiology, School of Biomedical Sciences, The University of Edinburgh, Edinburgh, UK

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Aldo Lucion, Universidade Federal do Rio Grande do Sul, Brazil Ruud Buijs, Universidad Autónoma de México, Mexico

#### **\*Correspondence:**

John Menzies, Centre for Integrative Physiology, University of Edinburgh, George Square, Edinburgh, EH8 9XD, UK.

e-mail: john.menzies@ed.ac.uk

Oxytocin neurons have a physiological role in food intake and energy balance. Central administration of oxytocin is powerfully anorexigenic, reducing food intake and meal duration. The central mechanisms underlying this effect of oxytocin have become better understood in the past few years. Parvocellular neurons of the paraventricular nucleus project to the caudal brainstem to regulate feeding via autonomic functions including the gastrointestinal vago-vagal reflex. In contrast, magnocellular neurons of the supraoptic and paraventricular nuclei release oxytocin from their dendrites to diffuse to distant hypothalamic targets involved in satiety. The ventromedial hypothalamus, for example, expresses a high density of oxytocin receptors but does not contain detectable oxytocin nerve fibers. Magnocellular neurons represent targets for the anorexigenic neuropeptide α-melanocyte stimulating hormone. In addition to homeostatic control, oxytocin may also have a role in reward-related feeding. Evidence suggests that oxytocin can selectively suppress sugar intake and that it may have a role in limiting the intake of palatable food by inhibiting the reward pathway.

**Keywords: oxytocin, food, appetite, satiety, reward**

# **INTRODUCTION**

Unlike Roald Dahl's "enormously fat" Augustus Gloop – a boy who pursued eating as a hobby – humans usually eat much less than they could. Obesity is typically the result of a modest excess of energy intake over expenditure, but one that is sustained over a prolonged period. In fact, humans, like other animals, are very efficient at balancing energy intake and expenditure. If rats are allowed unlimited access to a high-energy palatable diet, their energy intake diverges quickly from control animals fed a bland diet, but the difference in energy intake stabilizes within days (Archer and Mercer, 2007). Thus rats, which presumably do not feel societal pressure to be slim, will overeat palatable food, but only to a certain extent. When their access to palatable food ends, rats typically *undereat*, failing to defend the extra body weight they have accumulated.

A variety of peripheral signals convey information to control meal size and, in the longer term, these signals are modified according to physiological state and the size of energy reserves. Several "satiety" peptides are secreted from the gastrointestinal tract, including cholecystokinin (CCK), glucagon-like-peptide-1 (GLP-1), and peptide YY (PYY) (Strader and Woods, 2005); some of these act on the brain at sites that lack a blood-brain barrier, others are transported across the blood-brain barrier, and others act via vagal neuronal afferents (Verbalis et al., 1986). Leptin, which is secreted from adipose tissue in proportion to the size of fat stores, is not itself a satiety signal, but by signaling the size of peripheral fat reserves it has a long term anorectic influence, and in part it may act by moderating acute satiety signals arising from the gastrointestinal tract. Whether these satiety signals and leptin all converge at a discrete "satiety center" in the brain is unclear but several neuronal populations have been identified as likely candidates for mediating satiety. In part, these have been identified because they synthesize peptides that have marked anorexic actions when administered centrally.

One of those potently anorectic neuropeptides is oxytocin – a peptide classically thought to be involved mainly in reproductive functions. However, as we review here, there is now considerable evidence that oxytocin also plays an important role in satiety.

# **PHYSIOLOGICAL ROLES OF OXYTOCIN**

Oxytocin is produced in two hypothalamic regions: the supraoptic nucleus (SON) and paraventricular nucleus (PVN). Magnocellular neuroendocrine neurons in these nuclei project to the posterior pituitary gland, from where oxytocin is secreted into the blood. Classically, oxytocin secreted from the pituitary gland is involved in efficient and timely fetal expulsion, and is indispensable for the milk-ejection reflex and successful lactation (Nishimori et al., 1996); in some species, including rodents but not humans, it also regulates sodium excretion (natriuresis) (Verbalis et al., 1991), both by direct actions at the kidneys and indirectly be regulating the secretion of natriuretic peptides from the heart (McCann et al., 2003); in these species it is secreted in response to raised plasma osmotic pressure (Huang et al., 1996). All of the oxytocin neurons in the SON are magnocellular neuroendocrine neurons, but the PVN also contains parvocellular oxytocin neurons that project within the brain – to the spinal cord, caudal brainstem, amygdala, and substantia nigra (Sofroniew, 1980). These neurons are important in sexual behavior in males (Melis et al., 1986), and, as we discuss below, in the regulation of gastric reflexes.

Recently it has become apparent that oxytocin has an important "pro-social" role. In rodents, it has been implicated in recognition and positive social behavior between rodent mothers and their offspring and between adult members of the same social group (Neumann, 2009). However, behaviors linked to oxytocin are not all positive. In the rat, for example, patterns of central oxytocin release correlate with maternal aggression against unfamiliar intruders (Bosch et al., 2005). Various "social" effects of oxytocin in humans have also been reported, but mainly following the intranasal administration of doses of oxytocin so large that it is hard to be confident of where or how they are acting.

It is unclear to what extent these "behavioral" roles of oxytocin are attributable to the parvocellular oxytocin system or the magnocellular system, because many of the sites of action of oxytocin in this regard lack conspicuous innervation by oxytocincontaining fibers. Accordingly, it seems likely that oxytocin reaches these sites by volume transmission following release from possibly quite distant sites. Large amounts of oxytocin are released in some circumstances from the dendrites of magnocellular neurons (Ludwig and Leng, 2006), and interestingly, such dendritic released is regulated independently of axonal secretion – thus magnocellular neurons can release oxytocin either centrally or peripherally depending upon the stimulus (Sabatier et al., 2003).

# **OXYTOCIN AND FOOD INTAKE**

In early studies of hypothalamic function, lesions of oxytocincontaining hypothalamic nuclei were shown to result in an increase in food intake and body weight (Leibowitz et al.,1981; Shor-Posner et al., 1985; Sims and Lorden, 1986; Kirchgessner et al., 1988). Then, in the 1990s, several studies reported anorexigenic effects of central oxytocin: low doses of oxytocin given icv dose-dependently inhibited food intake in rats, increased the latency to begin feeding and reduced meal duration in both hungry and satiated animals, and these actions could be blocked by oxytocin receptor antagonists (Arletti et al., 1990; Olson et al., 1991a). Longer term central infusions of oxytocin were also reported to reduce body weight gain in rats given a high-fat diet, but in contrast to oxytocin's acute effects, chronic oxytocin infusions did not alter total food intake or meal patterning, but instead appeared to stimulate lipid metabolism in adipose tissue (Deblon et al., 2011). It was noted that, in rats, dehydration or sodium loading potently increased oxytocin secretion and at the same time suppressed appetite (Flanagan et al., 1992a); given that (in rats) oxytocin promotes natriuresis (Verbalis et al., 1991), the suppression of appetite by oxytocin appeared to be part of a general homeostatic role in sodium balance. As oxytocin secretion is not stimulated by hyperosmolarity in humans (Williams et al., 1986) it seemed that this role of oxytocin might be one peculiar to rodents. However, recent findings in humans with rare genetic mutations linked to monogenic obesity indicate that oxytocin may also have a role in appetite regulation in humans.

Male (but not female) oxytocin receptor-deficient mice express an obese phenotype in later adulthood despite no difference in food intake or motor activity (Takayanagi et al., 2008). Male and female oxytocin-knockout mice show an elevation in body weight and fat stores in adulthood but as with oxytocin receptor-deficient mice this is not due to an increase in food intake (Nishimori et al., 1996). There have been no published reports of humans completely lacking either oxytocin or its receptor, probably because the absence of oxytocin or its receptor is incompatible with successful reproduction, but a partial deficiency in central oxytocin production has been associated with the development of obesity in humans in two documented conditions.

The transcription factor Single-minded 1 (Sim1) is one of the few genes associated with human monogenic obesity (Holder et al., 2000; Farooqi and O'Rahilly, 2005). In mice, Sim1 is expressed in the SON and PVN and is essential for the development of these nuclei (Michaud et al., 1998). Homozygous Sim1 knockout mice do not survive gestation; but heterozygous mice are viable and are hyperphagic and become obese early in life (Michaud et al., 2001). These mice have much-reduced levels of oxytocin mRNA and immunoreactive oxytocin in both the SON and PVN (Michaud et al., 2001), and their hyperphagia can be reversed by oxytocin given icv (Kublaoui et al., 2008). Conditional deletion of Sim1 after gestational development of the PVN also results in both a reduction in oxytocin mRNA expression, and in hyperphagia and obesity (Tolson et al., 2010). In contrast, mice overexpressing Sim1 do not increase their food intake when given a high-fat diet, and are resistant to diet-induced obesity (Kublaoui et al., 2006b).

Patients affected by Prader–Willi syndrome (PWS) caused by the lack of a segment in the paternal chromosome 15 suffer from morbid obesity due to extreme hyperphagia. The PVN of these patients contains fewer oxytocin neurons than controls (Swaab et al., 1995), leading to speculation that a deficiency in oxytocin may be instrumental in the development of obesity in this condition. Mice in which specific genes associated with the PWS syndrome have been knocked out similarly develop late-onset obesity due to hyperphagia, and this can be partly explained by a deficient production of oxytocin in the hypothalamus (Dombret et al., 2012) or by a reduction in the number of oxytocin neurons (Muscatelli et al., 2000).

# **WHERE DOES APPETITE-INHIBITORY OXYTOCIN COME FROM?**

In the last 20 years it has become apparent that oxytocin neurons in both the SON and PVN are powerfully regulated by appetiterelated signals (Renaud et al., 1987; Olson et al., 1991a,b). The role of central oxytocin in the regulation of energy homeostasis appears to involve both the magnocellular neurons and centrally projecting parvocellular neurons.

#### **PARVOCELLULAR NEURONS OF THE PVN**

The PVN has a major role in the regulation of appetite and metabolism, and is an important direct target of projections from the primary leptin- and ghrelin-receptive neurons of the arcuate nucleus – from orexigenic neurons that co-express neuropeptide Y (NPY) and agouti-related peptide (AgRP) and the inhibitory neurotransmitter GABA, and from pro-opiomelanocortin (POMC) containing neurons, which express the potent satiety peptides α-melanocyte stimulating hormone (α-MSH) and cocaine-and amphetamine regulating transcript (CART) (Valassi et al., 2008). The PVN regulates metabolism via neuroendocrine neurons that release thyrotropin releasing hormone to regulate the thyroid gland (Alkemade, 2010; Nillni, 2010), it regulates glucocorticoid production via its regulation of pituitary adrenocorticotropin secretion (Herman et al., 2003), and it regulates the sympathetic nervous system via a large population of pre-autonomic neurons (Ferguson et al., 2008; Kc and Dick, 2010). However, the oxytocin neurons in the PVN, like those in the SON, are also conspicuous targets for α-MSH (Kim et al., 2000), and are particularly powerfully influenced by food intake and a variety of nutritionally related signals.

In rats, the expression of oxytocin mRNA in the PVN is markedly reduced by fasting; this reduction can be reversed by leptin administration (Kublaoui et al., 2008) and these effects apparently involve both magnocellular neurons and parvocellular neurons. Oxytocin neurons in the PVN are also contacted by fibers arising from the NPY/AgRP/GABA neurons of the arcuate nucleus. The PVN expresses abundant GABA receptors (Kalsbeek et al., 2004) and NPY Y1 receptors (Yokosuka et al., 1999). Optogenetic activation of the GABAergic axons in the PVN that arise from the arcuate nucleus increases food intake. Similarly, direct optogenetic activation of PVN oxytocin neurons increases c-Fos expression in these neurons and suppresses food intake. In the same study, a pharmacogenetic approach showed that while acute silencing of arcuate POMC neurons has surprisingly little effect, silencing Sim1-expressing PVN neurons markedly increases food intake (Atasoy et al., 2012).

Many parvocellular oxytocin neurons project to the nucleus tractus solitarii (NTS) in the caudal brainstem (Rinaman, 1998) where oxytocin modulates vagal efferent pathways that regulate gastric motility (McCann and Rogers, 1990). These neurons are critically involved in a reflex that is triggered by food intake, and mediated in part by gastric distension and in part by the secretion of CCK from the duodenum. Peripheral administration of CCK leads to activation of gastric vagal afferent neurons and thence to activation of brainstem structures, notably the NTS and ventrolateral medulla (Simpson et al., 2012). These in turn project to, and activate, centrally projecting oxytocin neurons of the PVN. As the CCK-stimulated NTS neurons are densely innervated by oxytocin-containing fibers from the PVN (Blevins et al., 2003), it appears that there is a recurrent circuit involving parvocellular oxytocin neurons that modulates the gastrointestinal vago-vagal reflex.

The gastrointestinal vago-vagal reflex involves main threecomponents: the gastrointestinal tract, the NTS, and the dorsal motor nucleus of the vagus (DMV). Visceral afferent fibers carrying digestion-related information ascend the vagus nerve and terminate in the medial NTS. NTS neurons integrate the information and send afferent projections to the DMV, and DMV neurons in turn project back down to the intrinsic ganglia in the gastrointestinal tract. In addition to this main loop, catecholaminergic and peptidergic neurons of the NTS also send ascending projections to structures in the forebrain including the PVN and the SON, and parvocellular PVN neurons project back to both the NTS and the DMN (Saper et al., 1976; Swanson and Kuypers, 1980). There is clear evidence that oxytocin acts in the DMN to inhibit gastric motility: the DMN contains oxytocin receptors (Dubois-Dauphin et al., 1992) and injection of oxytocin into the DMN decreases gastric motility, while oxytocin antagonists have the opposite effect (Rogers and Hermann, 1987). Electrical stimulation of the PVN inhibits gastric motility, and this effect can be attenuated by injection of an oxytocin antagonist in the DMN (Rogers and Hermann, 1987). Similar results were observed in conscious, freely moving animals, in which gastric motility was reduced by icv injection of oxytocin and by stimulation of the PVN, and in both cases the inhibition was prevented by icv injection of an oxytocin antagonist (Flanagan et al., 1992b). Finally, icv administration of an oxytocin antagonist alone increased baseline

gastric motility, suggesting a tonic inhibitory effect of oxytocin on gastric motility. However, it seems unlikely that the inhibitory action of CCK on gastric motility is mediated by oxytocin, as icv injection of an oxytocin antagonist did not prevent CCK-induced inhibition of gastric motility (Flanagan et al., 1992b).

Thus there is a well-established role of parvocellular oxytocin neurons in the regulation of the gastrointestinal tract. Moreover, the magnocellular oxytocin neurons of the SON and PVN also appear to be involved in regulating appetite, possibly by the actions of dendritically released oxytocin on the ventromedial nucleus of the hypothalamus (VMN). This large nucleus is known to have an important role in both energy balance and sexual behavior, and is a site at which oxytocin receptors are expressed at an exceptionally high density (Tribollet et al., 1988) as well as insulin-regulated aminopeptidase (IRAP), an enzyme involved in the inactivation of oxytocin (Fernando et al., 2005).

#### **MAGNOCELLULAR NEURONS OF THE SON AND PVN**

As we discuss further below, there is powerful evidence that magnocellular oxytocin neurons have an important role in the regulation of appetite – but it is important to note that this role is not necessarily mediated by oxytocin alone. In magnocellular neurons of the SON and PVN, oxytocin is co-localized with a number of anorexigenic factors, including the neuropeptides CART (Vrang et al., 1999), pituitary adenylate cyclase activating polypeptide (PACAP) (Koves et al., 1994), corticotropin-releasing factor (CRF) (Sawchenko et al., 1984), CCK (Vanderhaeghen et al., 1981), and nesfatin-1 (Foo et al., 2008). Indeed,it has been suggested that oxytocin actually mediates the inhibitory action of nesfatin-1 on food intake, as nesfatin-1 induces the release of oxytocin in the PVN (Maejima et al., 2009) and as the anorexigenic effect of nesfatin-1 can be blocked by an oxytocin antagonist (Yosten and Samson, 2010). The fat mass and obesity-associated (FTO) gene that has been associated with obesity in humans is also co-localized with oxytocin in both the PVN and SON. This gene encodes a transcription co-factor (Fto) that is believed to regulate the expression of appetite-related genes (Jia et al., 2008), and Fto over-expression increases oxytocin mRNA levels in cell cultures (Olszewski et al., 2011).

Magnocellular oxytocin neurons are activated during feeding: thus, in schedule-fed rats trained to expect to receive food for just 2 h each day, magnocellular oxytocin neurons in both the SON and PVN densely express Fos protein soon after the initiation of food intake (Johnstone et al., 2006). These neurons are also activated by gastric distension, and by systemic application of the satiety peptides CCK (Renaud et al., 1987) and GLP-1 (Bojanowska and Stempniak, 2000). Moreover, the intestinal lipid amide oleoylethanolamide (OEA), which suppresses feeding via activation of the vagus nerve (Lo Verme et al., 2005), stimulates oxytocin mRNA expression in the PVN and SON, this anorexic effect is prevented by blockade of central oxytocin receptors (Gaetani et al., 2010).

The best understood pathway involving feeding-evoked activation of magnocellular oxytocin neurons is that where peripheral injections of CCK leads to secretion of oxytocin from the posterior pituitary gland in rats. In brief, CCK is released from the duodenum in response to food intake and peripheral administration of CCK can inhibit food intake via stimulation of vagal afferent neurons and activation of brainstem structures, notably the NTS and ventrolateral medulla. CCK given intraperitoneally or intravenously increases Fos expression in oxytocin neurons in the PVN and SON (Caquineau et al., 2010), increases the electrical activity of oxytocin neurons in the SON (Renaud et al., 1987; Leng et al., 1991) and increases plasma oxytocin secretion (Kutlu et al., 2010). These actions are mediated by a direct projection from noradrenergic neurons of the A2 cell group,which co-express the potent appetite-inhibiting peptide prolactin-releasing peptide (PrRP). The activation of oxytocin neurons can be blocked by selective lesioning of the noradrenergic afferents or by blocking the actions at the SON of noradrenaline itself or those of PrRP (Onaka et al., 2012). Interestingly, central pretreatment with an oxytocin receptor antagonist reduces the anorexigenic effect of CCK (Olson et al., 1991b; Blevins et al., 2003) suggesting that CCK-evoked satiety may be mediated in part by oxytocin.

#### **INTERACTIONS BETWEEN MAGNOCELLULAR OXYTOCIN NEURONS AND THE MELANOCORTINS**

Like parvocellular oxytocin neurons, magnocellular oxytocin neurons abundantly express leptin receptors and are a target for this important hormone (Hakansson et al., 1998; Yarnell et al., 1998; Brogan et al., 2000). Leptin activates Fos expression in the PVN (Yokosuka et al., 1998; Caquineau et al., 2010; Qi et al., 2010), particularly in parvocellular PVN neurons projecting to CCKsensitive neurons in the NTS. In the SON, however, central administration of leptin does not induce Fos expression (Caquineau et al., 2010) or nuclear translocation of STAT3 (Hakansson and Meister, 1998), but it does increase nuclear STAT5 expression (Mutze et al., 2007).

In addition to this direct modulation by leptin, magnocellular oxytocin neurons are regulated indirectly via the effects of leptin on POMC neurons of the arcuate nucleus. Like the PVN, the SON receives a strong projection from POMC neurons. Oxytocin neurons in the SON, like those in the PVN, densely express α-MSH receptors (MC4) (Garza et al., 2008). α-MSH is a powerful anorexigenic peptide: centrally administered α-MSH reduces food intake and body weight, and mice lacking MC4 receptors are hyperphagic and obese (Adan et al., 2006). As mentioned above, magnocellular oxytocin neurons can secrete a large amount of peptide from their dendrites in response to certain stimuli, and notably they do so in response to α-MSH. In magnocellular oxytocin neurons of the SON, α-MSH acts at MC4 receptors to increase the intracellular calcium concentration; this increase directly evokes oxytocin release from the large dendrites of these neurons and also results in the production of endocannabinoids by the oxytocin neurons. Endocannabinoids produced in response to α-MSH act presynaptically to suppress glutamatergic afferent inputs to the oxytocin neurons. Thus, remarkably, the response of magnocellular oxytocin neurons to α-MSH is an increase in central release of oxytocin but a simultaneous suppression of electrical activity and hence a suppression of secretion into the systemic circulation (Sabatier and Leng, 2006).

Further evidence of an interaction between oxytocin and α-MSH is illustrated in the model of Sim1 heterozygous mice. In wild type mice, an agonist of α-MSH MC4 receptor (MC4R) reduced food intake, and induced Fos expression in PVN neurons, of which some co-express oxytocin and MC4R (Liu et al., 2003), and oxytocin and Sim1 (Kublaoui et al., 2008). However in Sim1 heterozygous mice,MC4R agonist had a much attenuated effect on both food intake and Fos expression in the PVN (Kublaoui et al., 2006a). This suggests that the α-MSH agonist actions on the PVN of Sim1<sup>±</sup> mice were impaired by the lack of oxytocin production in these mice (Michaud et al., 2001).

# **THE VENTROMEDIAL NUCLEUS OF THE HYPOTHALAMUS – A SITE OF ACTION OF OXYTOCIN**

It seems possible that oxytocin released from the dendrites of magnocellular neurons is involved in the regulation of appetite, presumably reaching its targets by volume transmission. We have estimated that a release rate of just one vesicle per oxytocin cell every 10 s would be enough to achieve a mean basal concentration of ∼260 pg/ml throughout the anterior hypothalamus within a minute (Leng and Ludwig, 2008). The half life of oxytocin in CSF is ∼20 min – it is likely to be less than this in the extracellular fluid, but there are no clear data on this point. We have argued elsewhere that the effects of oxytocin depends less on its sites of release but rather on the location of its receptors (Sabatier et al., 2007). Thus it is likely that food intake is inhibited in various physiological conditions in which oxytocin is released from the dendrites of magnocellular neurons. Indeed, appetite is inhibited in rats subjected to dehydration or sodium loading (Flanagan et al., 1992a), two stimuli that result in a hyperosmolar environment,which is known to stimulate the dendritic release of oxytocin from supraoptic neurons (Neumann et al., 1993). The main physiological circumstances in which there is extensive central secretion of oxytocin is in lactation, when suckling-induced dendritic oxytocin release is an essential part of the milk-ejection reflex (Rossoni et al., 2008). Lactation is associated with a marked increase of food intake, rather than a reduction, but despite this, lactation is a time of negative energy balance, because the increased food intake does not adequately compensate for the energy demands of the suckling young. Accordingly, it seems that appetite during lactation is not increased to the degree needed to maintain energy homeostasis – and it may be that suppression of appetite during suckling is necessary to ensure that the mother nurses the young rather than searches for food.

One particularly intriguing potential target for dendritic oxytocin is the VMN. This large hypothalamic nucleus contains a very high density of oxytocin receptors, as shown by intense labeling both for oxytocin receptor binding sites (Tribollet et al., 1988) and oxytocin receptor mRNA expression (Yoshimura et al., 1993), particularly in the ventrolateral region. Oxytocin receptors in the VMN have an established role in sexual behavior in female rats (McCarthy et al., 1994), and we have previously suggested that they are involved in the reciprocal regulation of appetite and sexual behavior (Leng and Ludwig, 2008). Although the VMN has a high density of oxytocin receptors, it contains very few oxytocin fibers, and is therefore a likely target for oxytocin released from the dendrites of magnocellular oxytocin neurons (Leng and Ludwig, 2008).

The VMN is not a functionally homogeneous nucleus so it is not surprising that systemic injections of CCK have diverse effects on VMN neurons, but the most common effect of CCKon the electrical activity of VMN neurons is inhibitory (Sabatier and Leng, 2010); these responses varied particularly between different subpopulations of VMN that displayed contrasting electrophysiological features. To test whether the appetite-inhibiting effects of oxytocin and those of CCK converge at the level of the VMN, we have studied the effects of central icv injection of oxytocin on the electrical activity of VMN neurons *in vivo*, and compared these responses with those of the same neurons to CCK. As with CCK, the responses to oxytocin were diverse. About 30% of neurons responded to oxytocin, 78% with a significant increase in their mean firing rate (**Figure 1**), and again there were differences in the responses of electrophysiologically distinct subpopulations. However,when we compared the responses of the same neurons to CCK and oxytocin, we noted particularly that, while one subpopulation of neurons had clearly divergent responses to CCK and oxytocin, in all other subpopulations the responses were remarkably convergent – for most neuronal types in the VMN, the response to icv oxytocin was a very strong predictor of the response to systemic CCK (**Figure 2**), supporting the hypothesis that VMN neurons are part of a common pathway mediating satiety.

Finally, to test the feasibility of the hypothesis that oxytocin affects VMN neurons via dendritic release of oxytocin from magnocellular neurons, we studied the electrical activity of single VMN neurons while applying α-MSH directly to the ipsilateral SON by microdialysis to evoke dendritic oxytocin release. Like the effects of icv oxytocin, this too had long-lasting, mainly excitatory effects upon a subset of VMN neurons (**Figure 3**).

#### **OXYTOCIN AND REWARD**

In addition to the drive to eat for metabolizable energy and nutritional factors (homeostatic feeding), humans and many other animals attend selectively to palatable foods and are motivated to eat them for pleasure (hedonic feeding). In addition to its effects on homeostatic food intake, oxytocin may have a role in the modulation of hedonic food intake.

Activation of dopaminergic mesolimbic neurons of the ventral tegmental area (VTA) is currently thought to be closely associated with reward and motivation. During palatable food intake, dopamine is released in the VTA's target regions such as the nucleus accumbens (NAcc) and dopamine release in these structures can also be evoked by stimuli associated with or predictive of palatable food. It is likely that activation of this system equates with the motivation to attend to, pursue and consume rewarding stimuli like palatable food. Oxytocin-containing axons from the PVN contact mesolimbic neurons (Sofroniew, 1980; Succu

**FIGURE 1 | Responses to icv oxytocin in VMN neurons in vivo.** Single VMN neurons were recorded extracellularly from the VMN of urethane-anesthetized rats (Sabatier and Leng, 2008). **(A)** Each bar represent the mean change in firing rate (±SD) averaged over the first 10 min after icv injection of oxytocin (1–10 ng) in each VMN cell tested. The cells are ranked by response magnitude and are classed as inhibited, non-responsive, and excited

according to whether the responses were significant or not. **(B)** Representative example of an inhibition (left panel) and activation (right panel) of the mean firing rate in response to icv oxytocin in a single VMN cell recorded extracellularly. **(C)** Mean change in firing rate (±SE) in the VMN cells that were significantly activated ; n = 14), significantly inhibited ; n = 4), and not significantly affected ; n = 43) by icv oxytocin.

et al., 2008) and likely have an inhibitory action as exogenous oxytocin activates NO synthase in mesolimbic dopaminergic neurons *in vivo* (Succu et al., 2008. Exogenous oxytocin also reduces amphetamine-evoked dopamine turnover in the NAcc (Qi et al., 2008) and inhibits the formation of a place-preference to methamphetamine when given into the NAcc core (Baracz et al., 2012). However, administration of oxytocin does not always lead to a reduction in dopamine release, administration into the ventral subiculum for example, a hippocampal region innervated by the PVN and potentially involved in restraining the stress response and reinstating non-food reward responding behavior, results in increased dopamine release in the NAcc (Melis et al., 2009). Inhibition of the PVN by optical stimulation of GABAergic fibers from arcuate NPY/AgRP/GABA neurons increases motivation to obtain sugar in a progressive ratio lever pressing task (Atasoy et al., 2012) suggesting that signals arising from the PVN may suppress the motivation for rewarding foods.

Consumption of readily catabolizable sugars seems to be innate in animals.Without training, rodents given a choice between water and sucrose will consume large volumes of a 10% sucrose solution [over 200 ml a day in adult male Sprague-Dawley rats (unpublished observations)] and voluntarily cease water drinking. Even transgenic mice lacking functional sweet taste receptors prefer sucrose over water (de Araujo et al., 2008) indicating that a postingestive reward is provided by sucrose. Transgenic mice lacking oxytocin show an enhanced preference for and consumption of sweet solutions over water in a two-bottle choice paradigm (Amico et al.,2005;Billings et al.,2006). The increase in intake is driven by a greater number of feeding bouts rather than by an increase in their duration (Sclafani et al., 2007). However, no such effect of oxytocin deficiency is seen on consumption of a palatable high-fat liquid formulation (Miedlar et al., 2007) or of a sucrose-containing solid food (Amico et al., 2005), though scheduled feeding of a highsugar diet to rats increases oxytocin gene expression (Olszewski et al., 2009). Progressive ratio operant conditioning paradigms are often used as a measure of motivation to work for a reward. In contrast to an effect of oxytocin deficiency on freely available sweet solutions, there is less evidence that oxytocin-deficient mice have enhancements (or deficiencies) in motivation to work for food as they are not different to wild type mice in this task (Sclafani et al., 2007). In addition to Fos expression observed at the termination of bland food intake (Johnstone et al., 2006), Fos expression is also increased in PVN oxytocin neurons at the termination of a bout of feeding where only either sucrose or a high-fat food was available (Olszewski et al., 2010). In this study, hypothalamic oxytocin mRNA levels were higher in mice allowed access to a high-sugar diet compared to a high-fat diet despite both being palatable and readily consumed. Furthermore, oxytocin receptor antagonism led to an increase in sucrose consumption but has no effect on fat consumption when presented separately. Regular exposure in rats to a diet high in sugar reduces Fos expression in PVN and SON oxytocin neurons after a high- or low-sugar meal compared to animals regularly receiving a low-sugar diet (Mitra et al., 2010). This suggests that regular sugar consumption might blunt the activity of oxytocin neurons in response to any meal, whether high in sugar or not. Nonetheless, it may be that oxytocin has a satiating effect related to certain components of diet rather than a general effect.

(20µg/kg) in a single VMN neuron.

Thus, the balance of evidence suggests that oxytocin might have a role in limiting the intake of palatable food by suppressing the activation of the reward pathway.

#### **CONCLUSION**

It now seems clear that oxytocin has a physiological role in energy balance through its actions in the caudal brainstem, but probably also through actions within the hypothalamus, including at the VMN, and possibly at other sites in the brain. While the actions in the brainstem appear to be part of a suite of autonomic regulatory functions exercised by parvocellular oxytocin neurons of the PVN, the hypothalamic actions appear to be more associated with motivational drive to eat, and are probably attributable to the magnocellular oxytocin system rather than the parvocellular system. It is possible that the inhibitory effects of oxytocin on appetite

#### **REFERENCES**


hormone signalling in the control of energy metabolism. *J. Neuroendocrinol.* 22, 56–63.

Amico, J. A., Vollmer, R. R., Cai, H. M., Miedlar, J. A., and Rinaman, L. (2005). Enhanced initial and sustained intake of sucrose

are closely linked to the stimulatory effects of oxytocin in sexual behavior and that this link reflects an evolutionarily conserved reciprocal regulation of these two key motivational drives (Caquineau et al., 2012). Whether it is possible to find a way of dissociating the appetite-inhibiting effects of oxytocin from the sexual arousal effects is unclear at present, but this may be the key to success in using the oxytocin system as a target for appetite-reducing drug therapies.

#### **ACKNOWLEDGMENTS**

Supported by the EU Seventh Framework program under grant agreements FP7-KBBE-2009-245009 (NeuroFAST) and FP7- KBBE-2010-266408 (Full4Health). This research was supported in part by a grant award to Dr. Nancy Sabatier from Medical Research Scotland (Grant award 316-FRG-L-0901).

solution in mice with an oxytocin gene deletion. *Am. J. Physiol. Regul. Integr. Comp. Physiol.* 289, R1798–R1806.


Bertrand, M. J., et al. (2012). Loss of Maged1 results in obesity, deficits of social interactions, impaired sexual behavior and severe alteration of mature oxytocin production in the hypothalamus. *Hum. Mol. Genet*. 21, 4703–4717.


defined target neurons of the hypothalamus. *J. Neurosci.* 18, 559–572.


the forebrain. *Acta Biol. Hung.* 45, 297–321.


melanocortin pathway. *Cell Metab.* 10, 355–365.


opioids, orexin, oxytocin and NPY. *Peptides* 30, 226–233.


cases. *J. Clin. Endocrinol. Metab.* 80, 573–579.


chronic osmotic stimuli in man. *J. Endocrinol.* 108, 163–168.


by pretreatment with an oxytocin receptor antagonist. *Am. J. Physiol. Regul. Integr. Comp. Physiol.* 298, R1642–R1647.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 02 October 2012; paper pending published: 22 November 2012; accepted: 05 March 2013; published online: 20 March 2013.*

*Citation: Sabatier N, Leng G and Menzies J (2013) Oxytocin, feeding, and satiety. Front. Endocrinol. 4:35. doi: 10.3389/fendo.2013.00035*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Sabatier, Leng and Menzies. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# A closer look at the role of urotensin II in the metabolic syndrome

# *Pierre-Olivier Barrette and Adel Giaid Schwertani\**

Division of Cardiology, Department of Medicine, McGill University Health Center, Montreal, QC, Canada

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Lan Ma, Fudan University, China Leo T. Lee, The University of Hong Kong, Hong Kong

#### *\*Correspondence:*

Adel Giaid Schwertani, Division of Cardiology, Department of Medicine, McGill University Health Center, 1650 Cedar Avenue, Room C9-166, Montreal, QC, Canada H3G 1A4. e-mail: adel.giaid@mcgill.ca

Urotensin II (UII) is a vasoactive peptide that was first discovered in the teleost fish, and later in mammals and humans. UII binds to the G protein coupled receptor GPR14 (now known as UT). UII mediates important physiological and pathological actions by interacting with its receptor. The metabolic syndrome (MetS) is described as cluster of factors such as obesity, dyslipidemia, hypertension, and insulin resistance (IR), further leading to development of type 2 diabetes mellitus and cardiovascular diseases. UII levels are upregulated in patients with the MetS. Evidence directly implicating UII in every risk factor of the MetS has been accumulated. The mechanism that links the different aspects of the MetS relies primarily on IR and inflammation. By directly modulating both of these factors, UII is thought to play a central role in the pathogenesis of the MetS. Moreover, UII also plays an important role in hypertension and hyperlipidemia thereby contributing to cardiovascular complications associated with the MetS.

**Keywords: metabolic syndrome, insulin resistance, inflammation, obesity, dyslipidemia, hypertension, diabetes**

# **UROTENSIN II**

The peptide Urotensin II (UII) was first identified in the teleost fish *Gillichthys mirabilis* (Pearson et al., 1980). Homologs of this peptide were subsequently found in mammals and humans (Pearson et al., 1980; Conlon et al., 1996; Ames et al., 1999; Coulouarn et al., 1999). Although the amino acid sequence of these homologs changes depending on the species, a cyclic sequence of six amino acids is conserved (Cys-Phe-Trp-Lys-Tyr-Cys). It is believed that this sequence is responsible for the biological activity of the peptide (Conlon et al., 1990). UII acts as a vasoactive peptide by binding to the G protein coupled receptor GPR14, better known as UT (Ames et al., 1999; Liu et al., 1999). A recent study has shown that UII vasoconstrictive effects are mediated by calcium influx via STIM1 and Orai-1 (Domínguez-Rodríguez et al., 2012). Human UII (hUII) and its receptor have been found in cardiac and vascular tissues, spinal cord, central nervous system, kidney, liver, and pancreas (Maguire et al., 2000; Matsushita et al., 2001). UII is also present in the blood plasma in picomolar concentrations (Matsushita et al., 2001). However, it tends to act mainly in an autocrine and paracrine fashion rather than as a hormone (Yoshimoto et al., 2004). Even though the relation between UII and its receptor was made in 1999, the physiological and pathological roles of UII are only beginning to be unraveled. In a healthy state, the binding of UII with UT is known to play an important role in the control of vascular tone, blood pressure, and insulin release (Douglas and Ohlstein, 2000; Douglas et al., 2000; Loirand et al., 2008). UII has been identified as the most potent vasoactive peptide known to date, being more potent vasoconstrictor than endothelin-1 and angiotensin-II (Ames et al., 1999; Douglas and Ohlstein, 2000; Maguire et al., 2000). UII also controls the function of vascular smooth-muscle cells (VSMCs) through the release of endothelial-cell-derived vasodilators such as nitric oxide (Gibson, 1987; Gardiner et al., 2001; Stirrat et al., 2001). The role of UII in pathological states is still debated. Many studies have shown increased levels of UII and its receptor in diverse cardiovascular and metabolic diseases such as type 2 diabetes mellitus, renal dysfunction, atherosclerosis, systematic and essential hypertension, obesity, congestive heart failure, myocardial infarction, cardiac fibrosis, hypertrophy, and remodeling (Reviewed in Hassan et al., 2003; Ross et al., 2010; Gruson et al., 2012). The elevated levels of UII in disease states suggest that UII is expressed either as a protective response to pathologies or as a pathological agent. Some studies claim that temporary elevation of UII levels may lower cardiovascular disease risk factors in end-stage renal disease (Mallamaci et al., 2005; Zoccali et al., 2006), myocardial infarction (Babi´nska et al., 2012) and restore endothelial function (Zoccali and Mallamaci, 2008). On the other hand, other studies have demonstrated that UII acts as a pathological agent inducing cardiac hypertrophy in synergy with angiotensin II by phosphorylation of the Akt kinase (Chanalaris et al., 2005; Gruson et al., 2010b, 2012).

# **METABOLIC SYNDROME**

"fendo-03-00165" — 2012/12/24 — 21:15 — page 1 — #1

As mentioned above, the metabolic syndrome (MetS) consists of a set of risk factors, such as obesity, hyperlipidemia, hypertension, hyperglycemia, and insulin resistance (IR), leading to the development of type 2 diabetes, cardiovascular diseases, non-alcoholic fatty liver disease, and renal impairment (Grundy, 1999; Cho, 2011). The relation between these risk factors and complications seems to be mostly attributed to IR (Eckel et al., 2010), which is why the syndrome was first defined as an IR syndrome (Reaven, 1988). Since then, the National Cholesterol Education Program (NCEP)-Adult Treatment Panel III, the World Health Organization (WHO) and the International Diabetes Federation (IDF) set up some criteria for the clinical definition of MetS.

A recent study using data from the 2003–2006 National Health and Nutrition Examination Survey and the NCEP/IDF MetS criteria, revealed that the prevalence of MetS was 34.3% among all adults in the US (Ford et al., 2010). However, high prevalence of MetS is not only restricted to the Western world. MetS seems to be highly prevalent in sub-Saharan Africa, Middle-East, South America, and South Asian Countries (Misra and Khurana, 2008). Central obesity is considered the main precursor to MetS (Cameron et al., 2008). According to the WHO, the global obese population increased from 200 million in 1995 to 300 million in 2000. The WHO has also determined that the obesity epidemic is also spreading in developing countries with over 115 million people suffering from obesity-related illnesses (Misra and Khurana, 2008; WHO, 2008). Popkin and Doak (1998) have also reported increasing obesity prevalence in the United States, Latin America, Europe, and Asia, which confirms the worldwide epidemic proportions of obesity and overweight. In addition, the Aerobic Center Longitudinal Study, which included 33,230 cancer-free men followed for 14 years, showed 56% enhanced risk of cancer mortality for men affected by MetS (Jaggers et al., 2009). The direct implication of the syndrome in CVD, type 2 diabetes, non-alcoholic fatty liver, and cancer, as well as an increasing prevalence of obesity and decreased physical inactivity in Western societies makes the MetS a major health preoccupation (NCEP, 2002)

Recent researches have shown that UII has an impact on the risk factors as well as the overall pathogenesis of the MetS (Ong et al., 2008). A study using the American Heart Association/National Heart, Lung, and Blood Institute criteria for MetS revealed that MetS patients show higher plasma levels of UII (Gruson et al., 2010a). Our group also demonstrated an increase in UII plasma levels in mice when all risk factors of MetS are present (You et al., 2012). Moreover, we recently found that UII gene deletion in mice (UIIKO) significantly decreased body mass, visceral fat, blood pressure, and increased insulin and glucose tolerance when compared to wild-type mice (You et al., 2012). Even though there is a clear association between MetS and UII, it is still unclear whether the peptide plays a role in the initiation of the disease, or if the elevated plasma levels of UII are a result of the syndrome. This review aims to give a closer look at the implication of UII in the risk factors and pathways involved in the development of the MetS.

# **INSULIN RESISTANCE (IR)**

Insulin plays a major role in glucose and lipid metabolism such as adipose tissue triglyceride lipolysis, lipoprotein lipase activity, muscle and adipose tissue glucose absorption, muscle and liver glycogen synthesis, and endogenous glucose production (Cornier et al., 2008). When cells react poorly to an insulin stimulus, they are defined as IR. The latter is known to be the link between the individual components of MetS (Eckel et al., 2010). For instance, it seems that IR correlates positively with atherosclerosis (Fernández-Real and Ricart, 2003) and coronary artery disease (CAD; Juhan-Vague et al., 1991), and also an indirect cause of hyperinsulinemia (Reaven, 2003), dyslipidemia, hyperglycemia, and hypertension (Sowers, 2004) through an increase in free fatty acid (FFA) synthesis in adipose tissue (Boden and Shulman, 2002) and a chronic inflammatory response (Fernández-Real and Ricart, 2003).

The IR Atherosclerosis Study (IRAS) and the Atherosclerosis Risk in Communities Study showed that insulin sensitivity (measured with an intravenous glucose tolerance test) correlates negatively with intimal-medial thickness of the carotid artery (Howard et al., 1996). Subsequent studies also found a significant correlation between carotid intimal-media thickening and IR and hyperinsulinemia (Golden et al., 2002; Zavaroni et al., 2006). Intimal-media thickness has been shown to be a significant predictor of myocardial infarction and coronary death (Hodis et al., 1998). As a result, IR is a major risk factor in developing cardiovascular complications. Indeed, IR has been associated with increased plasma levels of plasminogen activator inhibitor 1 (PAI-1; Juhan-Vague et al., 1991). Increased levels of PAI-1 are known to impair fibrinolysis, which leads to progression of coronary atherosclerosis in glucose intolerant patients (Bavenholm et al., 1998) and increased risk of myocardial reinfarction within 3 years in less than 45-year-old men (Hamsten et al., 1985, 1987; Gram et al., 1987; Juhan-Vague et al., 1991) as well as increased risk of CAD (Juhan-Vague et al., 1991). UII is known to increase expression of PAI-1 in vascular SMC (Djordjevic et al., 2005), hence UII may contribute to atherosclerosis through PAI-1 inhibition of fibrinolysis.

Free fatty acids play an important role in MetS by having direct effects on dyslipidemia, hypertension (Fagot-Campagna et al., 1998), IR (Boden, 1997), and pancreatic β-cell dysfunction (Sako and Grill, 1990; Unger, 1995; McGarry and Dobbins, 1999; Boden and Shulman, 2002). Impaired insulin action in adipose tissue tends to increase FFAs concentrations in the plasma by reducing the insulin inhibition of adipose tissue triglyceride lipolysis (Randle et al., 1963; Randle, 1998; Kahn and Flier, 2000; Cornier et al., 2008). On the other hand, FFAs induce IR in skeletal muscles (DeFronzo et al., 1985) and liver by inhibiting insulin suppression of glycogenolysis (Boden et al., 2002), which results in a vicious cycle increasing IR, FFA levels, and promotion of hyperglycemia. Furthermore, higher plasma glucose and FFA levels tend to increase insulin secretion in the pancreas (Boden and Shulman, 2002), thus leading to hyperinsulinemia. Such high plasma insulin levels promote hypertension (Welborn et al., 1966; Reaven, 2003), which is another risk factor of MetS. An increase in FFA levels also predicts development of type 2 diabetes (Paolisso et al., 1995; Charles et al., 1997). In the liver, higher FFA levels increase low density lipoprotein (LDL) and triglyceride production, while lowering high density lipoprotein (HDL) production (Brinton et al., 1991; Muramaki et al., 1995).

The direct effects of UII on lipid mobilization and FFA release are somewhat contradictory. In the coho salmon, administration of UII stimulated the activity of triacylglycerol lipase and the release of FFAs in the liver (Sheridan and Bern, 1986; Sheridan et al., 1987). On the contrary, UII injection in dogfish showed no significant increase in FFA release or plasma triglyceride concentrations (Conlon et al., 1994). We recently found that UII gene deletion in mice fed high fat diet significantly reduced serum levels of FFAs in comparison with wild-type (You et al., 2012). In light of our own findings and those of others linking UII to IR and glucose metabolism (Silvestre et al., 2001, 2004; Ong et al., 2006; You et al., 2012), we suggest that UII may mediate lipid mobilization and FFA indirectly through IR. However, the direct influence of UII in lipid metabolism still requires further research.

"fendo-03-00165" — 2012/12/24 — 21:15 — page 2 — #2

The role of UII in the MetS seems to be closely related to insulin activity and the overall glucose metabolism in the pancreas (**Figure 1**). In fact, UII and UT are both expressed in rat pancreatic islets, where their interaction seems to inhibit glucosemediated insulin release (Silvestre et al., 2001, 2004). Also, when high concentrations of UII are administrated to the rat pancreas, glucose and arginine induced insulin response are blocked (Silvestre et al., 2001, 2004). Some UII and UT gene haplotypes are known to be associated with IR (Ong et al., 2006), impaired glucose tolerance (Ong et al., 2006), β-cell pancreatic function, and type 2 diabetes (Sun et al., 2002; Zhu et al., 2002; Wenyi et al., 2003; Suzuki et al., 2004; Tan et al., 2006; Sáez et al., 2011). UII upregulation in plasma and skeletal muscle in type 2 diabetes mellitus is coherent with this association (Totsune et al., 2003; Wang et al., 2009). In addition, UII increases glucose-6-phosphatase activity in salmon liver, which tends to reduce glycogen content and increase glucose levels, leading to hyperglycemia. Our recent study has demonstrated that UII gene knockout in mice reduced serum glucose and insulin, and increased glucose and insulin tolerance in comparison with wild-type mice (You et al., 2012).

# **INFLAMMATORY RESPONSE IN THE MetS**

It is well known that IR and MetS are both associated with an inflammatory response (Sutherland et al., 2004). The relationship between IR and inflammation seems to be bidirectional; IR promotes an inflammatory response and vice versa (Fernández-Real and Ricart, 2003), resulting in a vicious circle increasing risk of MetS incidence. An acute-phase inflammatory response is not only linked with IR (Fernández-Real et al., 2001; Festa et al., 2002; Grimble, 2002; Leinonen et al., 2003), but also with type 2 diabetes (McMillan, 1989; Arnalich et al., 2000; Festa et al., 2002; Leinonen et al., 2003), obesity (Laimer et al., 2002; Leinonen et al., 2003), and the overall MetS (Chan et al., 2002; Das, 2002). The activity of inflammatory markers such as C-reactive protein, tumor necrosis factor-alpha (TNF-α), Interleukin-18 (IL-18) and PAI-1 increases as the presence of MetS risk factors increase (Cornier et al., 2008).

Tumor necrosis factor-alpha prevents the action of insulin in cultured cells and in animal models (Hotamisligil et al., 1993, 1994, 1996) by inducing serine phosphorylation of the insulin receptor substrate-1, which reduces the tyrosine kinase activity

"fendo-03-00165" — 2012/12/24 — 21:15 — page 3 — #3

of the insulin receptor (Hotamisligil et al., 1996), resulting in IR. Furthermore, TNF-α stimulates the production of endothelin-1 (Kahaleh and Fan, 1997) and angiotensin (Brasier et al., 1996) *in vitro*, which are directly involved in vasoconstriction and thus may lead to increased hypertension. Indeed, there is a positive correlation between TNF-α, systolic blood pressure, and IR (Dörffel et al., 1999). TNF-α also stimulates VLDL production (Grunfeld and Feingold, 1992) and decreases HDL cholesterol (Jovinge et al., 1998). Dyslipidemia in mice is significantly related to increased TNF-α levels (Fleet et al., 1992).

Elevated levels of the inflammatory cytokine IL-6 are associated with obesity, IR (Fried et al., 1998; Bastard et al., 2000; Fernández-Real et al., 2001; Kern et al., 2001; Bastard et al., 2002), and type 2 diabetes (Pradhan et al., 2001; Vozarova et al., 2003). IL-6 also plays a role in the pathology of dyslipidemia by decreasing LPL activity in adipocytes (Greenberg et al., 1992) which increases hepatic triglyceride levels (Nonogaki et al., 1995). In man, IL-6 correlates with increased FFA (Stouthard et al., 1995), fasting triglycerides, and VLDL (Fernández-Real et al., 2000), and decreased HDL cholesterol (Zuliani et al., 2007). This cytokine may also have an impact on hypertension by stimulating the sympathetic and central nervous system (Besedovsky and Del Rey, 1996; Papanicolaou et al., 1996), and angiotensin expression (Takano et al., 2000).

The central role of the inflammatory response in MetS is also related to UII and UT expression (**Figure 2**). Inflammatory cells including lymphocytes, macrophages, monocytes, and foam cells express UII and UT mRNA (Bousette et al., 2004). Lymphocytes are by far the largest producers of UII, whereas monocytes and macrophages produce the highest levels of UT. Moreover, inflammatory markers such as TNF-α, lipopolysaccharide, and interferon-γ are all known to induce UT expression *in vitro* (Segain et al., 2007). These findings suggest that an inflammatory response to MetS and its components may increase UII levels by increasing inflammatory cells production of the peptide. On the other hand, UII is known to contribute to the inflammatory response by inducing the release of inflammatory cytokines like IL-6 in cardiomyocytes (Sano et al., 2000; Tzanidis et al., 2003; Johns et al., 2004; Russell and Molenaar, 2004). UII mediates aortic inflammation by stimulating the release of leukotriene C (LTC4), a lipid inflammatory mediator, from aortic adventitial fibroblasts through 5-lipoxygenase pathway (Dong

"fendo-03-00165" — 2012/12/24 — 21:15 — page 4 — #4

et al.,2012). LTC4 plays a key role in the pathogenesis of atherosclerosis as a chemoattractant and activator for monocytes in the vascular endothelium. It is also known to initiate contraction of smooth muscle cells in the adjacent medial tissues of the vasculature (Dong et al., 2012). In addition, we have recently shown that UII gene deletion in mice fed a high fat diet reduces the serum levels of inflammatory cytokines including monocyte chemoattractant protein-1, monokine induced by γ-interferon, and keratinocyte chemoattractant when compared to wild-type mice (You et al., 2012).

# **UII, HYPERTENSION, AND CARDIOVASCULAR DISEASES**

The MetS is identified as a clustering of risk factors that perpetuate cardiovascular diseases. Since UII has been shown to play a role in the pathogenesis of atherosclerosis (Maguire et al., 2004; Hassan et al., 2005; Papadopoulos et al., 2009), congestive heart failure (Hassan et al., 2003), myocardial ischemia (Zhang et al., 2002; Zhou et al., 2003), ventricular hypertrophy, and fibrosis (Sano et al., 2000; Tzanidis et al., 2003; Johns et al., 2004; Russell and Molenaar, 2004), it is therefore reasonable to suggest that induction of UII in the MetS may contribute to the cardiovascular abnormalities associated with this syndrome.

Atherosclerosis is the main cause of morbidity and mortality in cardiovascular diseases and cause of death in the Western world (Libby et al., 2009). Atherosclerosis is known to induce UT and UII overexpression in mice aortic tissues (Douglas et al., 2002; Bousette et al., 2004; Wang et al., 2006; Libby et al., 2009). In mice overexpressing UT receptor (UT+), aortic atherosclerosis lesion formation is significantly increased (Papadopoulos et al., 2009), whereas lesion formation is decreased in UIIKO mice with an atherosclerotic background (APOE;You et al., 2012). UII is known to promote vascular remodeling by mediating VSMC proliferation and migration inside the intima, a process directly involved in atherogenesis (Sauzeau et al., 2001; Watanabe et al., 2001a,b; Tamura et al., 2003; Albertin et al., 2009; Iglewski and Grant, 2010). It has been suggested that vascular endothelial growth factor (VEGF), a major angiogenic protein, could be responsible for UII-mediated vascular remodeling. Even though UII does not affect VEGF expression in human endothelium (Albertin et al., 2009), recent evidence has shown that UII induces secretion of VEGF in the adventitia in synergy with angiotensin II (Song et al., 2012). VEGF stimulates proliferation of endothelial and smooth muscles cells while also stimulating proliferation and migration of adventitial fibroblasts in the intima (Sartore et al., 2001; Stenmark et al., 2006; Zhang et al., 2008). In addition, UII contributes to the inflammatory response in atherosclerosis by acting as a chemoattractant for monocytes (Segain et al., 2007) and increasing IL-6 expression in cardiomyocytes (Sano et al., 2000; Tzanidis et al., 2003; Johns et al., 2004; Russell and Molenaar, 2004). UII interacts with UT receptors on the surface of macrophages to promote expression of ACAT-1 (Watanabe et al., 2005). ACAT-1 is known to accelerate foam cells formation, which has a major impact on atherosclerotic lesion development (Watanabe et al., 2005). Another main effect of UII on atherosclerosis is the increased expression of NADPH oxidase, which is a main source of reactive oxygen species (ROS; Djordjevic et al., 2005; Tsoukas et al., 2011). ROS are central in the early initiation of atherosclerosis by converting LDL into oxidized-LDL (Djordjevic et al., 2005). NADPH oxidase-derived superoxide inactivates nitric oxide, resulting in impaired endothelial dependant vasodilatation (Zou et al., 2004) and hypertension (Lassègue and Clempus, 2003).

Urotensin II also contributes to essential (Matsushita et al., 2001) and secondary (Heller et al., 2002) hypertension by inducing vascular remodeling. Increased blood pressure in cats (Behm et al., 2004), rats (Lin et al., 2003), and sheep (Watson et al., 2003) with UII administration is coherent with these findings. In humans, plasma UII correlates positively with systolic blood pressure, independently of an atherosclerotic background (Cheung et al., 2004). Indeed, the UII gene (UTS2) is associated with essential hypertension (Yi et al., 2006) and myocardial infarction (Nishihama et al., 2007; Oguri et al., 2009). In a chronic heart failure or essential hypertension state, UII loses its dilatory function (Lim et al., 2004; Sondermeijer et al., 2005).

# **UII, OBESITY, AND HYPERLIPIDEMIA**

Obesity seems to be a main cause of IR in both men and women (Carey et al., 1996; Cornier et al., 2008; Stefan et al., 2008). Moreover, it seems that physical activity and a diet rich in monounsaturated or hydrogenated fat leads to an impairment of insulin sensitivity (Mayer-Davis et al., 1998; Vessby et al., 2001; Han et al., 2002). Obesity is also known as a chronic inflammatory state; it induces release of proinflammatory cytokines and adipokines (Yudkin et al., 1999; Festa et al., 2001; Engström et al., 2003; Trayhurn and Wood, 2004). As a result, the pathologic role of inflammation in IR and MetS may be enhanced by obesity. However, some suggest that obesity may be, on the opposite, an outcome of inflammation (Das, 2001). Obesity may contribute to increase local IR in the adipocytes as well as in other tissue such as the liver and skeletal muscles (Cornier et al., 2008; Gregor and Hotamisligil, 2011).

Plasma UII correlates positively with body weight in humans independently of atherosclerotic background (Cheung et al., 2004). Adipokines (dipeptidyl peptidase-4, endocan, insulin-like growth factor-binding proteins), known to play a role in diabetes mellitus and obesity (Janke et al.,2006; Neumiller et al.,2010; Ruan and Lai, 2010) were reduced in UII/ApoE double knockout mice in comparison with ApoEKO (You et al., 2012). Reduction of body mass, visceral fat, visceral adipocytes diameter, serum LDL, triglycerides, as well as increase in HDL is also observed with UII gene deletion in mice (You et al., 2012; **Figure 3**). A reduction in hepatic cholesterol esterification is also observed in UIIKO mice (Kiss et al., 2011). These results are attributed to a reduction in cholesterol and apolipoprotein B production by hepatocytes (Kiss et al., 2011). Furthermore, Jiang et al. (2008) found that the UII gene (UTS2) regulates fat accumulation in humans.

# **CONCLUSION**

"fendo-03-00165" — 2012/12/24 — 21:15 — page 5 — #5

The mechanism that links the various components of MetS relies primarily on IR and inflammation (Sutherland et al., 2004; Eckel et al., 2010). The ability of UII to modulate both of these factors suggests an important role for the peptide in the pathogenesis of the MetS. Other risk factors mediated by UII, such as obesity, may tend to initiate the positive feedback cycle, which contributes to

the development of MetS. In fact, genetic manipulation or the use of UT receptor antagonists seems to have a positive effect on every risk factor of the MetS (You et al., 2012). However, it is still unclear whether UII is important in the initiation or only in the progression and further complications of the disease. The biological activities of UII, and the recent reports of increased serum levels of UII in patients with the MetS suggest that pharmacological

# **REFERENCES**


Bastard, J. P., Maachi, M., Van Nhieu, J. T., Jardel, C., Bruckert, E., Grimaldi, A., et al. (2002). Adipose tissue IL-6 content correlates with resistance to insulin activation of glucose uptake both in vivo and in vitro. *J. Clin. Endocrinol. Metab.* 87, 2084–2089.


manipulation of the UII pathway as a possible future treatment of the syndrome.

# **ACKNOWLEDGMENTS**

We acknowledge the support of the Canadian Institute of Health Research and the Heart and Stroke Foundation of Quebec. We also thank Zhipeng You for his technical help.

(2002). FFA cause hepatic insulin resistance by inhibiting insulin suppression of glycogenolysis. *Am. J. Physiol. Endocrinol. Metab.* 283, E12–E19.


"fendo-03-00165" — 2012/12/24 — 21:15 — page 6 — #6

both low and high risk of NIDDM. *Diabetes* 45, 633–638.


Effects of dogfish urotensin II on lipid mobilization in the fasted dogfish, *Scyliorhinus canicula*. *Gen. Comp. Endocrinol.* 93, 177–180.


hypertension. *Hypertension* 34, 113–117.


atherosclerosis study. *Diabetes* 51, 1131–1137.


"fendo-03-00165" — 2012/12/24 — 21:15 — page 7 — #7


obesity-diabetes. *J. Clin. Invest.* 94, 1543–1549.


human urotensin II: receptor localization in human tissues and comparison of vasoconstrictor responses with endothelin-1. *Br. J. Pharmacol.* 131, 441–446.


"fendo-03-00165" — 2012/12/24 — 21:15 — page 8 — #8

2 diabetes mellitus. *Pharmacotherapy* 30, 463–484.


in pulmonary vascular remodelling. *Physiology (Bethesda)* 21, 134–145.


"fendo-03-00165" — 2012/12/24 — 21:15 — page 9 — #9

of the urotensin II receptor antagonists. *Front. Pharmacol.* 2:38. doi: 10.3389/fphar.2011.00038


T. R. (1966). Serum-insulin in essential hypertension and in peripheral vascular disease. *Lancet* 1, 1336– 1337.


hearts under normal perfusion and ischemia reperfusion. *Sheng Li Xue Bao* 55, 442–448.


"fendo-03-00165" — 2012/12/24 — 21:15 — page 10 — #10

community-dwelling older adults: the In Chianti study. *Atherosclerosis* 192, 384–390.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 29 August 2012; accepted: 29 November 2012; published online: 28 December 2012.*

*Citation: Barrette P-O and Schwertani AG (2012) A closer look at the role of urotensin II in the metabolic syndrome. Front. Endocrin. 3:165. doi: 10.3389/ fendo.2012.00165*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Barrette and Schwertani. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Arcuate AgRP neurons and the regulation of energy balance

# *Céline Cansell 1†, Raphaël G. P. Denis1†, Aurélie Joly-Amado1†, Julien Castel 1,2 and Serge Luquet1,2\**

<sup>1</sup> Unité de Biologie Fonctionnelle et Adaptative, CNRS-EAC 4413, Sorbonne Paris Cité, Université Paris Diderot-Paris 7, Paris, France <sup>2</sup> Centre National de la Recherche Scientifique EAC 4413, Paris, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Chris Scott, Charles Sturt University, Australia Virginie Tolle, Institut National de la Santé et de la Recherche Médicale, France

#### *\*Correspondence:*

Serge Luquet, Unité de Biologie Fonctionnelle et Adaptative, CNRS-EAC 4413, Sorbonne Paris Cité, Université Paris Diderot-Paris 7, 4 rue Marie-Andrée Lagroua Weill-Hallé, Bâtiment Buffon, Case courrier 7126, 75205 Paris Cedex 13, France. e-mail: serge.luquet@univ-parisdiderot.fr

†Céline Cansell, Raphaël G. P. Denis, and Aurélie Joly-Amado have contributed equally to this work.

The arcuate nucleus of the hypothalamus contains at least two populations of neurons that continuously monitor signals reflecting energy status and promote the appropriate behavioral and metabolic responses to changes in energy demand. Activation of neurons making pro-opiomelanocortin (POMC) decreases food intake and increases energy expenditure through activation of G protein-coupled melanocortin receptors via the release of α-melanocyte-stimulating hormone. Until recently, the prevailing idea was that the neighboring neurons [agouti-related protein (AgRP) neurons] co-expressing the orexigenic neuropeptides, AgRP, and neuropeptide Y increase feeding by opposing the anorexigenic actions of the POMC neurons. However, it has now been demonstrated that only AgRP neurons activation – not POMC neurons inhibition – is necessary and sufficient to promote feeding. Projections of AgRP-expressing axons innervate mesolimbic, midbrain, and pontine structures where they regulate feeding and feeding-independent functions such as reward or peripheral nutrient partitioning. AgRP neurons also make gamma aminobutyric acid , which is now thought to mediate many of critical functions of these neurons in a melanocortin-independent manner and on a timescale compatible with neuromodulation.

**Keywords: neuropeptideY, agouti-related protein, GABA, feeding behavior, metabolism, obesity, reward, dopamine**

"fendo-03-00169" — 2012/12/27 — 10:11 — page 1 — #1

During the last several decade, the world has witnessed a pandemic expansion of pathologies related to high-fat and highcarbohydrate diets including obesity, diabetes, dyslipidemia, and cardiovascular diseases – collectively referred to as metabolic syndrome. Obesity is now considered by the World Health Organization (WHO) to be a worldwide epidemic, having more than doubled since 1980. In 2008, there were 1.5 billion overweight adults in both developed and developing countries (http://www.who.int/mediacentre/factsheets/fs311/en/). The WHO believes the fundamental cause of obesity and being overweight is an energy imbalance between calories consumed and calories expended. Appropriate energy balance is reached when energy intake and energy expenditure are adapted to meet energy demands and nutrient availability. It took billions of years for mammalian species to shape a highly responsive homeostatic system in which the multiple aspects of energy expenditure are exquisitely balanced with both hunger and the motivational component of feeding to ensure energy homeostasis. Disruption of this regulation gives rise to life-threatening conditions that include anorexia nervosa at one extreme and metabolic syndrome at the other.

During the last decade, a significant effort has been focused on the identification of neuronal pathways that control food intake and energy expenditure. This review focuses primarily on a tiny neuronal population of about ∼1000 cells located in arcuate nucleus (ARC) of the hypothalamus, namely the neurons that produce agouti-related protein (AgRP), neuropeptide Y (NPY), and gamma aminobutyric acid (GABA; referred to here as AgRP neurons), and the recent conceptual advances that have been made studying their function in energy balance.

# **AgRP AND POMC NEURONS: TWO INTERMINGLED NEURONAL POPULATIONS DEFINING THE MELANOCORTIN SYSTEM**

Agouti-related protein was discovered as an endogenously released neuropeptide that acts as an inverse agonist for the melanocortin receptors, MC3R/MC4R (Fan et al., 1997; Haskell-Luevano et al., 1999; Haskell-Luevano and Monck, 2001; Nijenhuis et al., 2001; Flier, 2006; Tolle and Low, 2008). Shortly after its discovery, Hahn et al. (1998) discovered that AgRP is co-expressed in hungeractivated neurons with NPY, another peptide that stimulates food intake and regulates weight gain (Tatemoto et al., 1982; Clark et al., 1984). The inhibitory nature of the NPY/AgRP neurons was further substantiated through the identification of GABA as their ionotropic neurotransmitter (Horvath et al., 1997). These neurons are now commonly referred to as AgRP neurons because, unlike NPY and GABA which are widely expressed in the nervous system, AgRP is uniquely produced by these neurons. It is a unique molecular signature that has been extensively exploited for the selective manipulation of these neurons. AgRP neurons are located in the ARC subdivision of the hypothalamus at the bottom of the third ventricle close to a circumventricular organ called median eminence (ME). The blood–brain barrier in this region is fenestrated and allows for facilitated blood–brain exchange. As a result, neurons that reside there are referred to as "first order neurons" because they would be the first to respond to the circulating signals of hunger and satiety.

Neurons in the ARC that make pro-opiomelanocortin (POMC) and cocaine- and amphetamine-related transcript (CART) secrete the melanocortin peptides adrenocorticotropic hormone (ACTH) and α, β, and γ-melanocyte-stimulating hormone (MSH), which are derived from post-translational processing of POMC. POMC and AgRP neurons are considered to be two functionally opposed components of the "central melanocortin system," a term that refers to as a set of hormonal, neuropeptidergic, and paracrine signaling pathways that are defined by components that include the five G protein-coupled melanocortin receptors (MCR1 to MCR5; Cone, 2005). These receptors are distributed throughout the body (Mountjoy and Wong, 1997; Liu et al., 2003). In the CNS, MC4R is broadly expressed while MC3R is mainly restricted to POMC and AgRP neurons (Jegou et al., 2000). The integral role of the melanocortin system in body weight homeostasis is supported by the fact that any mutation in the melanocortin signaling pathway including MC3R- or MC4R-null mutants (Huszar et al., 1997) and ectopic expression of MCR3/4 antagonist, agouti, in agouti lethal yellow (Ay ) mutant mice (Miltenberger et al., 1997), results in hyperphagia, hypometabolism, hyperinsulinemia, and hyperglycemia in both rodents and humans (Hinney et al., 1999; Krude et al., 1999). The antagonistic relationship between POMC and AgRP neurons results from a tonic GABAergic inhibition from AgRP neurons onto POMC neurons (Horvath et al., 1992; Broberger and Hokfelt, 2001; Cowley et al., 2001; Williams et al., 2001; Pinto et al., 2004) and the interaction of NPY released by AgRP neurons with the NPY-Y1 receptor expressed on POMC neurons. The two populations project to several nuclei within the hypothalamus, including the paraventricular (PVN), ventromedial, dorsomedial, and lateral hypothalamus, and to the nucleus of tractus solitarii (NTS), the parabrachial nucleus (PBN) and the amygdala and the bed of the stria terminalis (BNST) which lie outside the hypothalamus (Broberger et al., 1998; **Figure 1**). In these regions AgRP- and NPY-containing fibers are found in close apposition to α-MSH-containing fibers and synapse onto secondorder targets (Broberger et al., 1998). The release of α-MSH by POMC neurons and its binding to G-coupled MCR's initiates the central anorexic signaling pathway that results in decreased food intake and increased energy expenditure while AgRP exerts its orexigenic action partly by blocking the binding of α-MSH to its receptor there by preventing the α-MSH-induced anorexic pathway (Michaud et al., 1994; Ollmann et al., 1997; Shutter et al., 1997; Haskell-Luevano et al., 1999; Haskell-Luevano and Monck, 2001; Nijenhuis et al., 2001; Tolle and Low, 2008).

Thyroid-releasing hormone (TRH)-, oxytocin (OT)-, and corticotropin-releasing hormone (CRH)-expressing neurons located in the PVN all express MC4R (Lu et al., 2003). The binding of α-MSH to MC4R on these neurons has a positive action onto the hypothalamic–pituitary–thyroid (HPT) axis and the hypothalamic–corticotropic axis (HPA). During fasting, increased release of AgRP by AgRP neurons has been demonstrated to be a key mechanism for fasting-induced down regulation of the HPT axis and the consequent adaptation during negative energy balance (Fekete et al., 2000; Lechan and Fekete, 2006).

These observations promoted a dominant conceptual framework that, until recently envisioned the orexigenic and anabolic action of AgRP neurons as the result of POMC neuron antagonism (Palmiter, 2012).

However, there are several lines of evidence supporting a melanocortin-independent pathway for AgRP. For exemple, shortand long-term hyperphagic actions of AgRP are still observed in MC4R KO mice and some AgRP fibers—but not α-MSH fibers have been found in close apposition to TRH-synthesizing neurons expressing MC4R (Fekete et al., 2000). Moreover, it has been shown that AgRP can act by a melanocortin-independent pathway that regulates glutamatergic neurons in the ventromedial hypothalamus (Fu and van den Pol, 2008). These data were first to suggest that AgRP, can exert an action as an agonist on unidentified receptors that are independent from the melanocortin signaling pathway.

Recent studies expand further the role of AgRP neurons to include melanocortin-independent mechanisms and nonfeeding-related functions such as goal-directed behavior and peripheral nutrient partitioning.

# **AgRP NEURONS ARE NECESSARY AND SUFFICIENT TO INITIATE THE FULL FEEDING SEQUENCE**

In 2005, several laboratories reported the selective ablation of AgRP neurons. Although the methods and the results differed somewhat, there was agreement that acute depletion of AgRP neurons in the adult mouse leads to life-threatening anorexia (Bewick et al., 2005; Gropp et al., 2005; Luquet et al., 2005; Xu et al., 2005). These experiments demonstrated that ablation of AgRP neurons in adult mice inhibits feeding and can lead to starvation. Ablation of AgRP neurons still caused severe anorexia when performed in the genetic context of A<sup>y</sup> mice (Wu et al., 2008a), a model in which the melanocortin signaling pathway is already tonically inhibited by the ectopic expression of the melanocortin receptor antagonist, agouti (Miltenberger et al., 1997). These data indicated that the anorexia is not the direct consequence of unopposed melanocortin tone. Wu et al. (2008a) went on to show that acute loss of GABA signaling by AgRP neurons was responsible. Although ablation of AgRP neurons in adult mice leads to starvation, mice can adapt to the loss of AgRP neurons and continue to eat adequately. This was first shown by performing the ablation in neonatal mice, before AgRP neurons are mature (Luquet et al., 2005), but it was subsequently shown that this phenomenon can also occur in adult mice (Wu et al., 2009, 2012a,b).

Direct activation of AgRP neurons *in vivo* has been achieved through either forced expression of designer receptors exclusively activated by designer drugs (DREADD) or photoactivated channel rhodopsin allowing for chemical- or light-mediated activation of neurons (Hegemann and Moglich, 2011; Krashes et al., 2011). Using optogenetic techniques,Aponte et al. (2011)found that photoactivation of AgRP neurons promoted feeding in both wild-type and A<sup>y</sup> mice. In a subsequent study, the Sternson group showed that photoactivation of feeding is mediated in part by activation of OT-expressing neurons in the PVN (Atasoy et al., 2012). They also confirmed that inhibition of POMC neurons is neither necessary nor sufficient to trigger feeding since co-stimulation of both POMC and AgRP neurons resulted in rapid feeding response.

"fendo-03-00169" — 2012/12/27 — 10:11 — page 2 — #2

"fendo-03-00169" — 2012/12/27 — 10:11 — page 3 — #3

These results provide an entirely new perspective to the field by showing that extinction of α-MSH signaling cascade was not mandatory for AgRP neurons to initiate feeding.

# **PROCESSING OF TASTE AND VISCEROSENSORY INPUTS IN THE HYPOTHALAMUS–PONS–MEDULLA AXIS**

By using Fos immunostaining to reveal neuron activation,Wu et al. (2008b) found that acute ablation of AgRP neurons in the PBN induces hyperactivity in all known targets of axonal projections from AgRP neurons. Moreover, they showed that GABA replacement in the PBN prevented anorexia and body weight loss after AgRP neuron ablation. This study demonstrated that GABA made by AgRP neurons was critically required to mediate their action in a melanocortin-independent manner (Wu et al., 2009; Wu and Palmiter,2011) and put a new light on the PBN, a pontine structure that links gustatory sensory circuits to the brain center that processes the reward and motivational aspects of feeding (de Araujo, 2009; Suwabe and Bradley, 2009; Tokita et al., 2009; Oliveira-Maia et al., 2011). Gut-initiated viscerosensitive satiety or aversive signals together with food-related cues gathered by sensory neurons innervating the oral cavity are routed to the NTS primarily by the afferent portion of the vagus nerve (Schwartz et al., 1991, 1993; Moran et al., 2001; Schwartz and Moran, 2002). In the rodent, the PBN is a second-order target for NTS taste-related information. It serves as a relay structure for the encoding of the reward and motivational components of food-associated cues through the activation of the mesolimbic dopaminergic system (de Araujo, 2009; Suwabe and Bradley, 2009; Tokita et al., 2009; Oliveira-Maia et al., 2011). Looking for the source of excitatory inputs into the PBN that mediate its hyperactivity once the GABAergic tone from AgRP neurons is removed, Wu et al. (2012a) demonstrated that input to the PBN is glutamatergic and that it arises from a subpopulation of viscerosensitive NTS neurons. They also showed that the latter received tonic activationfrom serotoninergic neurons of the raphe obscurus and the raphe magnus (**Figures 1 and 2**).

The rostral NTS is also a target of descending projections from cognitive and emotional processing centers such as the insular and prefrontal cortex, the central nucleus of the amygdala, the lateral hypothalamus, and the BNST. The study from Wu et al. (2012a) provides an additional dimension to the hypothalamus– pons–medulla axis connection by suggesting a direct role for AgRP neurons in the ability of the medulla to relay rewarding and aversive information and integrate cognitive and emotional feedback from limbic structures (**Figure 1**). This Arc–pons–medulla axis could be instrumental in the gain of body weight that is often

"fendo-03-00169" — 2012/12/27 — 10:11 — page 4 — #4

close to the circumventricular organs, the median eminence (ME). They exert a GABAergic tonic inhibition onto POMC-, OT-, CRH-, and TRH-expressing neurons from the PVN, the PBN and the dopaminergic neurons (DA) of the VTA. The PBN receives glutamatergic input from the NTS which is also a target for serotoninergic neurons of the raphe obscurus (ROb) and the raphe magnus (RMg). In the PVN, the synaptic properties of AgRP axons are such that GABA release could promote post-synaptic inhibition through the long

VTA, the PBN, and preganglionic structure of the PVN. Hence, hunger-activated neurons could have a role that extends beyond the acute regulation of feeding to autonomic control of nutrient partitioning, the modulation of gut-borne signals in the brainstem, and the fine tuning mesolimbic reward and motivational circuitry. (A/P) Antero-posterior stereotaxic coordinates are presented in mm from bregma below each section.

associated with the treatment of depression of bipolar disorders using selective serotonin re-uptake inhibitors.

Based on the circuitry described, one would expect that activation of AgRP neurons stimulate feeding by activating OTexpressing neurons in the PVN, while at the same time dampening activation of the PBN and thereby minimizing the influence of satiety and or visceral malaise. Photoactivation of AgRP axons in the PBN did not promote feeding, whereas activation of the PVN resulted in robust feeding (Atasoy et al., 2012). However, reducing satiety or visceral malaise would not necessarily promote robust feeding. Thus, the discrepancy may be resolved by considering the timescale at which the two events occur as well as their potential contribution to body weight maintenance. AgRP input to the PVN could convey an acute hunger-activated feeding response whereas AgRP input to the PBN could be more tonic in nature and involve

longer-lasting actions that insure the proper excitatory balance of the PBN.

# **CONNECTING METABOLIC NEEDS AND GOAL-DIRECTED BEHAVIOR**

A recent study provides evidence that AgRP neurons could directly participate in the dopaminergic encoding of goal-directed behavior independent from the actual retrieval of food. The reinforcing and motivational aspects of food are closely tied to the release of the neurotransmitter dopamine by midbrain dopamine neurons in the ventral tegmental area (VTA) that project to the nucleus accumbens, and other limbic brain regions. VTAmediated dopamine release is stimulated by high-fat/high-sugar foods as well as by most other objects of desire (e.g., sex, drugs; Wise, 2006). The VTA–striatal network provides a crucial neural

substrate upon which drugs of abuse (e.g., cocaine, nicotine, morphine) exert their effect; thus, this projection is often referred to as the brain "reward circuit" (Kelley et al., 2005).

Using a model in which AgRP neurons are either ablated from birth or rendered hypoactive through selective knockdown of the metabolic sensor of the sirtuin family Sirt1 (silent mating type information regulation 2 homolog), a direct role for AgRP neurons in modulation of dopamine signaling was revealed (Dietrich et al., 2012). They demonstrated a direct projection from AgRP neurons to the VTA and inactivation or ablation of AgRP neurons reduced GABAergic tone to VTA dopamine neurons. This translated to higher excitability of VTA neurons and facilitated the induction of long-term potentiation. Hence, a reduced activity of AgRP neurons resulted in enhanced dopamine-dependent encoding of reward and motivation. At the behavioral level, the two models showed an enhanced response to novelty and a stronger preference for an environment associated with cocaine injection (Dietrich et al., 2012; Palmiter, 2012).

One can therefore envision GABAergic tone from AgRP neurons to the PBN and the VTA as a necessary input to counteract gut-borne aversive input while reducing the threshold at which taste-related information is successfully transferred to cognitive, emotional, and rewarding processing centers in the brain (**Figure 1**). This mechanism could be central to the maintenance of the motivational and rewarding components of feeding when energy stores are low and when the food is deprived of reinforcing properties. The activity of AgRP neurons would be necessary to maintain metabolic need as a significant contributor of goal-directed behavior. Decreased AgRP neuronal input, tonic or phasic, could result in the progressive replacement of hunger by limbic-associated emotional inputs such as stress or anxiety. The overall consequence could be that a drive for reward seeking eventually prevails over metabolic demand in the control of feeding (Dallman et al., 2005; **Figure 2**).

# **ORCHESTRATING NUTRIENT PARTITIONING THROUGH TONIC CONTROL OF PREGANGLIONIC STRUCTURES**

Agouti-related protein neurons project to several preganglionic structures (Broberger et al., 1998) including the regions involved in activating autonomic nervous system (ANS) output such as the PVN (Cowley et al., 1999). ANS innervation of peripheral tissues (pancreas, liver, visceral adipose tissue) has a distinct organization that can be tracked back to pre-autonomic hypothalamic neurons (Kreier et al., 2006). Nutrient partitioning (the integrated processes that control conversion, storage, and utilization of nutrients) relies on the ability of the brain to orchestrate peripheral organ activity through the modulation of the ANS. Using a model of neonatal depletion, we showed that the lack of AgRP

#### **REFERENCES**


neurons affects the relative balance between lipid and carbohydrate metabolism. As a consequence, mice lacking AgRP neurons became obese and hyperinsulinemic on regular chow but displayed reduced body weight gain and paradoxical improvement in glucose tolerance on high-fat diet (Joly-Amado et al., 2012). This action was independent from feeding, and involved GABAergic input to preganglionic structures and the consequent modulation of ANS output onto liver, muscle, and pancreas.

These findings indicate that AgRP neurons play a role that extends beyond the regulation of feeding, to the control of peripheral nutrient partitioning. In good agreement, a recent study showed that Sirt 1 inactivation selectively in AgRP neurons resulted in a shift in the overall metabolic profile, the impairment of metabolic adaptation induced by a fast, and a change in ghrelin-induced excitability of AgRP neurons (Dietrich et al., 2010). Importantly, we found that a GABAA receptor agonist normalized the metabolic profile in mice lacking AgRP neurons (Joly-Amado et al., 2012). These results suggest that, here too, GABA may be a crucial signaling molecule by which AgRP neurons control peripheral nutrient partitioning.

In conclusion, several recent studies have significantly broadened the spectrum of brain structure, mechanism, and timescale that underlie the action of AgRP neurons in the regulation of energy balance. The intrinsic nature of GABA release has received much attention and technological tools and approaches have allowed dissociation of the behavioral and metabolic actions of AgRP from the melanocortin signaling pathway. Long-lasting inhibitory currents that persist long after the action potential indicate a putative function for AgRP neurons in the neuromodulation of post-synaptic targets. A second era is just beginning for this neurocircuitry that will likely provide fundamental insights into the mechanisms that underlie not only food-related but also nonfood-related behavior including cognitive, emotional, and reward processes.

#### **ACKNOWLEDGMENTS**

This work was supported by young investigator ATIP grant from the CNRS, a research fellowship from and a grant by the "Agence Nationale de la Recherche" ANR-09-BLAN-0267-02. Céline Cansell received a PhD fellowship from the Centre National de la Recherche Scientifique (CNRS) and a research grant from the Société Francophone du Diabète-Roche (SFD). Aurélie Joly-Amado received a National Merit Scholarship from the French Department of National Education and Research and a research grant from the SFNEP-ANTADIR. Raphaël G. P. Denis received a research fellowship from the Region Île-de-France. We would like to express our gratitude to Richard D. Palmiter and Diane Durnam for fruitful comments on the manuscript and for editing.


"fendo-03-00169" — 2012/12/27 — 10:11 — page 5 — #5

Y and human pancreatic polypeptide stimulate feeding behavior in rats. *Endocrinology* 115, 427–429.


of NPY, AGRP, and melanocortin signals in the hypothalamic paraventricular nucleus: evidence of a cellular basis for the adipostat. *Neuron* 24, 155–163.


for feeding. *Nat. Neurosci.* 8, 1289– 1291.


hypothalamic-thalamic-striatal axis for the integration of energy balance, arousal, and food reward. *J. Comp. Neurol.* 493, 72–85.


"fendo-03-00169" — 2012/12/27 — 10:11 — page 6 — #6

gut feedback signaling. *Int. J. Obes. Relat. Metab. Disord.* 25(Suppl. 5), S39–S41.


nucleus leads to starvation. *Cell* 137, 1225–1234.


protein activates fos and gliosis in postsynaptic target regions. *J. Neurosci.* 28, 9218–9226.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

"fendo-03-00169" — 2012/12/27 — 10:11 — page 7 — #7

*Received: 03 October 2012; paper pending published: 24 October 2012; accepted: 05 December 2012; published online: 27 December 2012.*

*Citation: Cansell C, Denis RGP, Joly-Amado A, Castel J and Luquet S (2012) Arcuate AgRP neurons and the regulation of energy balance. Front. Endocrin. 3:169. doi: 10.3389/fendo. 2012.00169*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Cansell, Denis, Joly-Amado, Castel and Luquet. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# The role of melanin-concentrating hormone and its receptors in energy homeostasis

# **Douglas J. MacNeil \***

Department of In Vitro Pharmacology, Merck Research Laboratories, Kenilworth, NJ, USA

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Jean Albert Boutin, Institut de Recherches Servier, France Olivier Civelli, University of California Irvine, USA

#### **\*Correspondence:**

Douglas J. MacNeil, Department of In Vitro Pharmacology, Merck Research Laboratories, K15-3-309D, 2015 Galloping Hill Road, Kenilworth, NJ 07033, USA. e-mail: doug.macneil@merck.com

Extensive studies in rodents with melanin-concentrating hormone (MCH) have demonstrated that the neuropeptide hormone is a potent orexigen. Acutely, MCH causes an increase in food intake, while chronically it leads to increased weight gain, primarily as an increase in fat mass. Multiple knockout mice models have confirmed the importance of MCH in modulating energy homeostasis. Animals lacking MCH, MCH-containing neurons, or the MCH receptor all are resistant to diet-induced obesity. These genetic and pharmacologic studies have prompted a large effort to identify potent and selective MCH receptor antagonists, initially as tool compounds to probe pharmacology in models of obesity, with an ultimate goal to identify novel anti-obesity drugs. In animal models, MCH antagonists have consistently shown efficacy in reducing food intake acutely and inhibiting body-weight gain when given chronically. Five compounds have proceeded into clinical testing. Although they were reported as well-tolerated, none has advanced to long-term efficacy and safety studies.

**Keywords: MCH, neuropeptide, MCHR1, orexigenic, obesity, KO mice, antagonist, clinical study**

#### **INTRODUCTION**

The mammalian form of melanin-concentrating hormone (MCH), is a 19-amino acid cyclic peptide encoded within a 165 amino acid preprohormone (**Figure 1**) (Vaughan et al., 1989). MCH has been associated with a wide variety of behaviors (see recent reviews by Saito and Nagasaki, 2008; Antal-Zimanyi and Khawaja, 2009; Chung et al., 2011), but the focus of this review is the role of MCH in energy homeostasis. The amino acid sequence of MCH is identical in all mammals evaluated and alternative processing of the preproMCH peptide can generate two additional putative peptides, designated neuropeptide E-I (NEI) and neuropeptide G-E (NGE) (Nahon et al., 1989). Several *in vivo* studies have shown that MCH plays a role in a variety of physiologic processes mediated within the central nervous system (CNS), including energy homeostasis sleep and arousal, and emotionality (Yumiko and Nagasaki, 2008; Torterolo et al., 2009). Although less studied, MCH may also have a role in peripheral tissues such as in gut and pancreatic islet function (Pissios et al., 2007; Kokkotou et al., 2008).

Two MCH G-protein coupled receptors (GPCRs) have been characterized (Pissios et al., 2006; Chung et al., 2011). MCHR1 is found in all vertebrates, while MCHR2 is found in non-rodent higher species, including primates (Hill et al., 2001; Sailer et al., 2001).

Neurobiology, rodent genetics, and rodent pharmacologic studies all demonstrate that MCH and the MCH receptors are involved in regulating body weight (Gomori et al., 2003, 2007; Hervieu, 2006; Bednarek, 2007;Antal-Zimanyi and Khawaja, 2009; Johansson, 2011; Cheon, 2012). On the basis of this information, many pharmaceutical companies have pursued the development of MCHR1 antagonists for the treatment of obesity (for a recent review, see Johansson, 2011). Unfortunately, although a few MCHR1 antagonists have entered development, no compound has successfully demonstrated anti-obesity efficacy in a clinical trial. It remains unclear if this lack of clinical efficacy is due to a lack of efficacy via the MCH1R pathway or to the inability of teams to identify safe and well-tolerated compounds with sufficient potency and pharmacodynamics properties to test the hypothesis. This review summarizes the evidence for a role of MCH and its receptors in energy homeostasis and the progress made to date toward identifying small-molecule antagonists to treat obesity.

# **MCH ACTS THROUGH TWO G-PROTEIN COUPLED RECEPTORS**

Originally described as the orphan receptor SLC-1/GPR24 (Kolakowski et al., 1996), MCHR1 was later shown by five groups to be activated by MCH (Pissios et al., 2006). The 402-amino acid rodent and human MCHR1 receptors are highly homologous, sharing ∼95% identity (Pissios and Maratos-Flier, 2003), and the highest expression of the receptors is within the brain (Saito et al., 1999; Hill et al., 2001). Like many family A GPCRs, the MCHR1 receptors have consensus *N*-glycosylation sites at the amino terminus and several potential phosphorylation sites in the intracellular loops (Lakaye et al., 1998).

In recombinant cell lines, the natural ligand, MCH, binds to MCHR1 with ∼1 nM affinity, and it couples to Gi, Go, and Gq proteins (Hawes et al., 2000; Pissios et al., 2003). Thus, activation of MCHR1 leads to an increase in intracellular Ca++ accumulation acting through the Gq-coupled pathway and/or to lowered cyclic adenosine monophosphate (cAMP) levels via the Gi/o-coupled pathway. Further analyses of the signaling of MCHR1 in recombinant cell lines and in brain slices demonstrates that activation of MCHR1 also leads to ERK phosphorylation (Pissios et al.,2003). In 3T3-L1 adipocytes, MCH rapidly induced a threefold to fivefold

increase in MAPK pathway activities (Bradley et al., 2002). It is unclear if all, or some, of these signaling pathways contribute to MCH-mediated events *in vivo*.

A second MCHR was later identified and termed MCHR2 by six groups (Antal-Zimanyi and Khawaja, 2009). The functional role of MCHR2 is not well defined, in part because it is not expressed in rodents and related species (hamsters, guinea pigs, or rabbits), but it is expressed in humans, dogs, ferrets, and monkeys (Tan et al., 2002). The amino acid sequence identity between MCHR1 and MCHR2 is low,∼38%, with the highest homology in the seven-transmembrane domains that form the ligand binding pocket (Sailer et al., 2001). Although MCH binds to MCHR1 and MCHR2 with a similar nanomolar affinity, the signal transduction mechanism of MCHR2 is limited to the Gq-mediated increase in intracellular Ca++ levels (Sailer et al., 2001). MCHR2 is largely coexpressed with MCHR1 in the CNS (Sailer et al., 2001), although peripheral expression was also found in adipocytes, pancreas, prostate, and intestine (An et al., 2001). The phylogenetic tree of MCH-related receptors contains opioid, somatostatin, galanin, urotensin 2, and orphan receptors (Sailer et al., 2001). MCH receptors have the highest homology (about 40%) with the somatostatin receptors (Sailer et al., 2001).

#### **NEUROANATOMY OF MCH AND MCH RECEPTORS**

Melanin-concentrating hormone has been implicated in many behaviors. The hypothalamus is one of the primary sites in which MCH-containing nerve fibers and MCH receptors are extensively expressed (Gao, 2009). Although most of the MCH neurons are located within the incerto-hypothalamic and lateral hypothalamic area (LHA), a recent review by Bittencourt details the locations throughout the brain of MCH nerve terminals (Bittencourt, 2011). Neural signaling by MCH via its receptors has been implicated in the control of energy balance, but due to the wide distribution of MCH-containing fibers throughout the brain, the critical sites of action for particular behaviors have not been identified (Zheng et al.,2005). In male rats, neurons expressingMCH arefound in the LHA and medial zona incerta, as well as, sparsely, in the olfactory tubercle and pontine reticular formation. The wide distribution of MCH fibers suggests the involvement of this neuropeptide in a variety of functions, including arousal, neuroendocrine control, and energy homeostasis (Rondini et al., 2007).

Melanin-concentrating hormone-expressing neurons in the LHA play an integrative role between signals from the periphery, acting via first-order neurons in the arcuate nucleus, and then from extra-hypothalamic systems, which modulate regulation of feeding, drinking, and seeking behaviors (Guyon et al., 2009). Factors from the periphery affect brain activity, resulting in changes in food intake and energy expenditure. Neurons from the arcuate nucleus detect changes in homeostatic parameters and transmit information to other brain areas, including the LHA. These secondary area neurons have widespread projections throughout the brain, and their activation leads to coordinated and altered behaviors (Guyon et al., 2009). About 25% of the LHA neurons projecting to the pedunculopontine tegmental (PPT) nucleus are immunoreactive for MCH, and 75% of the LHA neurons projecting to the cerebral motor cortex also contain MCH (Elias et al., 2008). Also, 15% of the incerto-hypothalamic neurons projecting to the PPT express MCH immunoreactivity. The MCH neurons express glutamic acid decarboxylase mRNA, a gammaaminobutyric acid (GABA) synthesizing enzyme, indicating that the MCH/GABA neurons are involved in inhibitory modulation and their activation may lead to decreased motor activity in states of negative energy balance (Elias et al., 2008). Like MCH, vasopressin and oxytocin can influence energy homeostasis and other behaviors. Whole-cell recording in hypothalamic brain slices from the MCH-green fluorescent protein transgenic mouse revealed that both vasopressin and oxytocin evoked a substantial excitatory effect on MCH-expressing cells (Yao et al., 2012). Both neuropeptides reversibly increased spike frequency and depolarized the membrane potential in a concentration-dependent manner, suggesting that vasopressin or oxytocin exerts a robust excitatory effect on presumptive GABA cells that contain MCH (Yao et al., 2012).

Interestingly, projections to ventral medullary sites apparently play a role in the inhibitory effect of MCH on energy expenditure, but not food intake (Zheng et al., 2005). In the rat, a significant proportion (5–15%) of primarily perifornical and far-lateral hypothalamic MCH neurons project to the dorsal vagal complex. Retrograde tracing in the caudal brainstem demonstrated that MCH-immunoreactive axons are distributed densely in the nucleus of the solitary tract, in the dorsal motor nucleus of the vagus, and in sympathetic premotor areas in the ventral medulla (Zheng et al., 2005). In medulla slice preparations, MCH inhibited the amplitude of excitatory postsynaptic currents. Administration of MCH in the fourth ventricle in freely moving rats decreased core body temperature, but it did not change locomotor activity or food and water intake (Zheng et al., 2005).

TheMCH pathways from the lateral hypothalamus to the mammillary nucleus may also enable the animal to look for food during the initial moments of appetite stimulation (Casatti et al., 2002). Injection of the retrograde tracer True Blue in the medial mammillary nucleus led to MCH/True Blue double-labeled neurons in the LHA, the rostromedial zona incerta, and the dorsal tuberomammillary nucleus. The afferents were confirmed using implants of the anterograde tracer *Phaseolus vulgaris* leucoagglutinin. The MCH projections may participate in spatial memory processing mediated by the medial mammillary nucleus (Casatti et al., 2002).

In addition to the LHA, the nucleus accumbens shell (AcSh) is a brain region important for food intake. The AcSh contains high levels of receptor for MCH. MCH receptor activation in the AcSh increases food intake, while AcSh MCH receptor blockade reduces feeding. Moreover, *in vivo* recordings confirm that MCH reduces neuronal activity in the AcSh in freely moving animals, consistent with a model from other pharmacological and electrophysiological studies whereby reduced AcSh neuronal firing leads to food intake (Sears et al., 2010). Since the AcSh mediates reinforcing properties of food, MCH may modulate motivational aspects of feeding. Indeed, chronic loss of rat MCH decreased food intake predominantly via a reduction in meal size during development and reduced high-fat food reinforced operant response in adult rats (Mul et al., 2011). Also, chronic loss of ProMCH in the rat affects the limbic dopamine system, since adult *Pmch*−*/*<sup>−</sup> rats showed increased *ex vivo* electrically evoked dopamine release (Mul et al., 2011). Thus, MCH actions in the AcSh mediate motivational aspects of feeding behavior.

MCHR1 is widely distributed in the brain (Hervieu et al., 2000; Able et al., 2009). Hervieu et al. (2000) used *in situ* hybridization histochemistry and immunohistochemistry to determine that *Mchr1* mRNA and protein were widely expressed throughout the rat brain. Similar to the distribution of MCH, *Mchr1* signals were observed in the cerebral cortex, caudate-putamen, hippocampal formation, amygdala, hypothalamus, and thalamus, as well as in various nuclei of the mesencephalon and rhombencephalon (Hervieu et al., 2000). Able et al. (2009) used an MCHR1-specific radioligand to demonstrate highly MCHR1-specific binding in the rat nucleus accumbens, caudate-putamen, and preform cortex, as well as lower levels of binding in the hippocampus and amygdala. Surprisingly, and in contrast to Hervieu et al. (2000) andAble et al. (2009) did not detect MCHR1 binding in the hypothalamus.

The distribution of MCHR2 in the primate brain nearly overlaps that of MCHR1, but the latter shows much higher relative levels and a wider distribution pattern (Mori et al., 2001). *MCHR2* is expressed in several human brain areas, including the hippocampus and amygdala, although its distribution in the hypothalamus remains controversial. Specifically, *MCHR2* mRNA was reported to be mainly expressed in the arcuate nucleus and ventromedial hypothalamic nucleus in African green monkeys by *in situ* hybridization (Sailer et al., 2001), while three other reports did not detect its expression in the human hypothalamus by RT-PCR (Hill et al., 2001; Mori et al., 2001) or Northern blot analysis (Rodriguez et al., 2001).

#### **HUMAN GENETICS OF MCH AND ENERGY HOMEOSTASIS**

Genetic analysis of obese subjects has identified several variants of MCH and the MCH receptors, but no alterations have been conclusively linked to obesity or leanness. In an association study, among 106 subjects with severe early onset obesity and a history of hyperphagia, two missense variants were found in *MCHR1*: Y181H and R248Q (Gibson et al., 2004). Neither of these was found in 192 normal-weight controls. R248Q co-segregated with obesity across two generations, butfamily data were unavailablefor

Y181H. When tested for functional response, the R248Q variant showed no evidence of constitutive activation, alteration in cAMP signaling, or ligand hypersensitivity (Gibson et al., 2004). Two common single-nucleotide polymorphisms (SNPs) were found to be in linkage disequilibrium, but no association between either of these and obesity-related phenotypes was found among a population cohort of 541 whites. Only two rare, non-coding variants were found in*MCHR2*. However, the relationship of these*MCHR2* variants to metabolic phenotypes has not been clarified (Gibson et al., 2004). Genomic screening of 13.4 kb encompassing the *MCHR1* in extremely obese German children and adolescents identified 11 infrequent variations and two SNPs in the *MCHR1* coding sequence and 18 SNPs (eight were novel) in the flanking sequence. Although an association of an *MCHR1* haplotype (SNPs rs133072 and rs133073) with obesity was observed in two cohorts of German children and adolescents, it was not confirmed in five independent cohorts (Wermter et al., 2005). To investigate the possible polygenic role of *MCHR1*, six common SNPs (minor allele frequency>5%) found in the sequenced regions were screened in 557 morbidly obese adults, 552 obese children, and 1195 non-obese non-diabetic control subjects (Bell et al., 2005). The plausible promoter SNP, rs133068, was found to be associated with protection against obesity in obese children only (Bell et al., 2005).

A functional analysis of 11 MCHR1 variants that had been reported previously in the literature identified two mutant receptors, R210H and P377S, that failed to respond to MCH (Goldstein et al., 2010). Five other variants showed significant alterations in MCH efficacy, ranging from 44 to 142% of the wild-type value. Both inactive receptors had cell surface expression that was comparable to wild-type (Goldstein et al., 2010). It is of note that the two loss-of-function mutants were identified in markedly underweight individuals, raising the possibility that a lean phenotype may be linked to deficient MCHR1 signaling (Goldstein et al.,2010). Additional association studies with larger cohorts are needed to explore the extent to which signaling-deficient MCHR1 variants influence the maintenance of body weight.

The association between *MCHR2* variation and human obesity was investigated in 141 obese children and 24 non-obese adult subjects by DNA sequencing, and by case-control analyses using 628 severely obese children and 1401 controls (Ghoussaini et al., 2007). None of the *MCHR2* variants showed an association with adult severe obesity, but the A76A SNP was associated with severe obesity (*P* = 0.01) and overeating in obese children (*P* = 0.02) (Ghoussaini et al., 2007). Validation of an association of *MCHR2* with obesity requires replication in other cohorts.

The common allele of the *ProMCH* gene, rs7973796, may be associated with a higher body mass index (BMI) in olanzapine-treated patients with schizophrenia. In a subgroup of subjects under 50 years of age among 300 schizophrenia patients, the rs7973796 genotype was associated with an effect on BMI among patients taking olanzapine (interaction *P* = 0.025) (Chagnon et al., 2007). Olanzapine-treated patients with schizophrenia carrying the homozygote genotype showed a higher BMI for rs7973796 (*P* = 0.016 with the least-squares means*t*-test) than the variant homozygotes. The G allele was associated with an increase in the odds of obesity in schizophrenic patients taking olanzapine (Chagnon et al., 2007).

# **RODENT GENETICS INDICATE A ROLE FOR MCH IN ENERGY HOMEOSTASIS**

Multiple mouse knockout (KO) models have been constructed to explore the role of MCH in energy homeostasis and other MCHmediated behaviors. The KO models include multiple constructs that prevent synthesis of MCH and the MCHR1 receptor. Studies of these mice show that loss of MCH function leads to leanness and resistance to obesity.

#### **MCH KO MICE**

Strong evidence that MCH mediates energy homeostasis came from studies of mice in which the *Promch* gene was inactivated. Mice constructed with targeted inactivation of the *Promch* gene in a mixed C57BL/6 × 129SvJ genetic background had reduced body weight and leanness due to hypophagia (reduced feeding) and an increased metabolic rate, despite reduced amounts of both leptin and arcuate nucleus proopiomelanocortin mRNA (Shimada et al., 1998). Evaluation of *Promch* inactivation in pure genetic backgrounds confirmed that *Promch* deficiency increased energy expenditure and promoted increased running-wheel activity (Kokkotou et al., 2005; Zhou et al., 2005). As observed previously on a mixed background, the C57BL/6 *Promch* KO mice were hypophagic; however, the 129SvEv *Promch* KOs were hyperphagic, relative to wild-type. In both C57BL/6 and 129SvEv backgrounds, deletion of *Promch* led to reduced adiposity, attenuated weight gain, and increased locomotor activity, compared with wild-type counterparts. The relative increase in activity was greater on a high-fat diet (HFD) than on regular chow (Kokkotou et al., 2005). The lean phenotype of the *Promch* KO mice persisted as the mice aged. At 19 months, C57BL/6 *Promch*−*/*<sup>−</sup> male and female mice weighed about 25% less than their wild-type counterparts as a result of reducedfat mass in *Promch*−*/*<sup>−</sup> mice. The aged *Promch*−*/*<sup>−</sup> mice exhibited improved glucose tolerance in intraperitoneal glucose tolerance tests, were more insulin sensitive, and were more active compared with wild-type controls (Jeon et al., 2006).

Further confirmation of the role of MCH in energy homeostasis came from studies of *Promch* neuron-ablated mice, generated using toxin (ataxin-3)-mediated ablation strategy in an FVB/n background (Alon and Friedman, 2006). In these mice, the *Promch* gene is present throughout development, but 60–70% of MCHexpressing neurons degenerate in the first few weeks of life (Alon and Friedman, 2006). After 7 weeks of age, the mice developed reduced body weight, body length, fat mass, lean mass, and leptin levels. As observed in the C57BL/6 *Promch*−*/*<sup>−</sup> mice, leanness was characterized by hypophagia and increased energy expenditure. In leptin-deficient *ob/ob* mice, loss of either *Promch* or MCH-containing neurons improved obesity, diabetes, and hepatic steatosis, suggesting that MCH is an important mediator of the response to leptin deficiency (Segal-Lieberman et al., 2003; Alon and Friedman, 2006).

#### **Promch tg MICE**

The phenotype of *Promch* tg mice overexpressing MCH also supports a role for MCH in energy homeostasis. Ludwig et al. (2001) constructed transgenic mice that overexpressed *Promch* in the lateral hypothalamus at ∼twofold higher levels than normal mice. On an FVB background, the homozygous transgenic mice fed a HFD ate 10% more and were 12% heavier than wild-type animals. Blood glucose levels were higher both preprandially and after an intraperitoneal glucose injection, and the transgenic mice were insulin-resistant (Ludwig et al., 2001). *Promch* tg heterozygous mice on a C57Bl/6 background were hyperphagic on regular chow, heavier, and insulin-resistant, but did not have elevated blood glucose (Ludwig et al., 2001).

#### **Mchr1**−**/**<sup>−</sup> **MICE**

As the gene for preproMCH encodes two additional peptides, NEI and NGE of unknown function (Nahon et al., 1989), studies with *Mchr1*KO mice provided clarification and confirmation of the role of MCH in energy homeostasis. Three groups independently produced *Mchr1*−*/*<sup>−</sup> mice. Chen et al. (2002) reported that *Mchr1* KO mice on a C57BL/6 × 129SvJ mixed background were resistant to diet-induced obesity and had fat mass that was significantly lower in both male (4.7 ± 0.6 vs. 9.6 ± 1.2 g) and female (3.9 ± 0.2 vs. 5.8 ± 0.5 g) mice than that of the wild-type control. The mice were hyperphagic on a HFD, but had a 28% higher metabolic rate than wild-type. Both leptin and insulin levels were significantly lower in male *Mchr1*−*/*<sup>−</sup> mice than in the wild-type controls, but there were no detectable differences in glucose levels. No differences were observed between heterozygotes and wild-type mice (Chen et al., 2002). Marsh et al. (2002) observed that *Mchr1*−*/*<sup>−</sup> mice also constructed on a C57BL/6 × 129SvJ mixed background had normal body weights, yet they had reduced fat mass and were hyperphagic when maintained on regular chow. In agreement with Chen et al. (2002) they observed that *Mchr1*−*/*<sup>−</sup> mice were less susceptible to diet-induced obesity, and the KO leanness was a consequence of hyperactivity and altered metabolism (Marsh et al., 2002). A later study by Zhou et al. (2005) showed that the *Mchr1* KO mice had a dramatic 250% increase in running-wheel activity along with hyperphagia (Antal-Zimanyi and Khawaja, 2009). Astrand et al. (2004) also independently generated *Mchr1*−*/*<sup>−</sup> mice on a mixed C57BL/6 × 129SvJ background and observed that the mice had an elevated metabolic rate and were hyperactive, hyperphagic, and lean. A >12% increase in heart rate without any change in blood pressure was noted (Astrand et al., 2004). Two groups studied *Mchr1* KO mice after backcrossing onto a C57BL/6 background (Bjursell et al., 2006;Ahnaou et al., 2011). In leptin-deficient *ob/ob* mice, loss of *Mchr1* reduced adiposity (although body weights were not statistically different), improved the response in an oral glucose tolerance test (OGTT), increased spontaneous movement, and improved thermoregulation upon exposure to cold, suggesting that MCH is an important mediator of the response to leptin deficiency (Bjursell et al., 2006).

The role of MCHR2 has not been investigated in animal models. MCHR2 is absent in rodents but is present in higher species, including primates (Hill et al., 2001; Sailer et al., 2001). The role of MCHR2 might be studied in MCHR2-humanized mice, but no such model has been described. Alternatively, *in vivo* studies with MCHR1- and MCHR2-selective agonists could be used to study the role of MCHR2 in non-rodent species with a functional MCHR2. Indeed,it is surprising that*in vivo* pharmacological studies with selective ligands have not been performed, in light of the availability of MCH peptides that are potent dual MCHR1/R2 agonists, selective MCHR1 agonists,MCHR2-preferring agonists, and



potent MCHR1/R2 antagonists (MacNeil and Bednarek, 2009). At this point, it is unclear if MCHR1 and MCHR2 play redundant or unique roles in MCH signaling in primates, or whether MCHR2 plays any significant role in energy homeostasis in humans.

Multiple mouse models show that disruption of the MCH system, via either the peptide ligand or the receptor, results in altered energy homeostasis (**Table 1**). In general, loss of MCH signaling leads to a lean, diet-induced obese (DIO)-resistant phenotype due primarily to increased energy expenditure, and in some models, lower food intake.

#### **PHARMACOLOGIC STUDIES CONFIRM A ROLE FOR MCH IN ENERGY HOMEOSTASIS**

*In vivo* studies have shown that MCH or MCH analogs increase food intake and body weight, while, conversely, studies with MCHR1 antagonists reduce body weight and associated comorbidities.

#### **IN VIVO EFFECTS OF MCH PEPTIDE AGONISTS**

Melanin-concentrating hormone, and occasionally MCH derivatives, have been used to evaluate the role of MCH in energy homeostasis. Most studies have utilized acute injections of MCH into either specific brain nuclei or, more commonly, the third or fourth ventricle. These injections almost certainly lead to supraphysiologic levels of MCH in at least some of the MCH receptor-containing nuclei. Qu et al. (1996) were the first to show that intracerebroventricular (ICV) MCH increased food intake. In multiple experiments ICV injections of 5 or 30µg into Long Evans rats increased 2-, 4-, and 6-h food intake between 150 and 200% (Qu et al., 1996). ICV injection of MCH into the third ventricle of either Wistar or Sprague-Dawley rats also increased food intake (Qu et al., 1996; Della-Zuana et al., 2002; Shearman et al., 2003). The orexigenic effects of ICV MCH were maximal at 2 h post injection (Della-Zuana et al., 2002). In Wistar rats, low doses (0.1 and 0.5µg/rat) were ineffective,while higher doses (1,5, and 10µg/rat) were equally effective, leading to an approximate doubling of food intake over 2–4 h (Della-Zuana et al., 2002). In Sprague-Dawley rats, only the two highest doses led to significant increases in food intake (Della-Zuana et al., 2002). When presented with sucrose solutions after ICV injection of MCH (2 nmol), Sprague-Dawley rats increased their intake of sucrose solution by increasing the rate of licking (Baird et al., 2006). The role of MCH in energy homeostasis was also confirmed in sheep, which have both MCHR1 and MCHR2; acute ICV doses of MCH increased food intake (Whitlock et al., 2005). Potent MCH analogs (IC<sup>50</sup> < 25 nM), but not the weak analogs (IC<sup>50</sup> > 1000 nM), reduced 2-h food intake after ICV administration of 4.4 nmol to Wistar rats (Suply et al., 2001). In a separate study in Sprague-Dawley rats using a smaller, potent MCH analog, Shearman et al. (2003) observed dose-dependent increases in 6-h food intake after an ICV injection of an MCH agonist (1µg/rat, +68%; 5µg/rat, +76%; 15µg/rat, +122%). Guesdon et al. (2008) confirmed that the smaller MCH analog, when given ICV at 5µg/rat, increased food intake in Wistar rats threefold over a 2-h period.

Intracerebroventricular injection of MCH into the third ventricle of rats significantly increased the ingestion of sucrose and glucose solution, but not of saccharin, indicating that the MCHinduced dipsogenic response is more related to caloric content than to sweet taste *per se* (Sakamaki et al., 2005). Injections of MCH into several brain nuclei led to increases in food intake. MCH (0.6 nmol) elicited a rapid and significant increase in feeding in satiated rats following injection into the arcuate nucleus, the paraventricular nucleus, or the dorsomedial nucleus (Abbott et al., 2003). However, no significant alteration in feeding was observed following injection into other brain regions associated with energy homeostasis, including the supraoptic nucleus, LHA, medial preoptic area, anterior hypothalamic area, or ventromedial nucleus of the hypothalamus (Abbott et al., 2003).

In a side-by-side comparison, MCH was found to be a weaker orexigen than two other hypothalamic neuropeptides. In lean rats, 1 and 3 nmol of the ICV-injected orexigenic peptides, neuropeptide Y (NPY) and agouti-related protein (AGRP), showed robust increases in intake of a sucrose solution, but 3 nmol of MCH mediated only a non-significant trend toward increased feeding (Semjonous et al., 2009). The importance of forebrain hypothalamic regions for MCH action was apparent when injection of 6 nmol of MCH into the fourth ventricle of lean rats or sheep failed to induce increased food intake, while control injections of NPY did increase food intake (Whitlock et al., 2005; Baird et al., 2007). Administration of LiCl, a potent inducer of conditioned taste aversion (CTA), to rats leads to an upregulation of *Mch* and *Mchr1* mRNA (Mitra et al., 2012). However, when MCH was injected prior to the induction of CTA with LiCl, as well as later during the CTA retrieval, MCH treatment did not reduce the magnitude of CTA upon subsequent presentations of the aversive tastant (Mitra et al., 2012). Thus, MCH is not critical to the development of CTA.

Although most ICV studies leading to an increase in food intake utilized injection into the third ventricle, Georgescu et al. (2005) demonstrated that direct injection of 1µg of MCH into the AcSh, a region rich in MCHR1, resulted in a robust increase in food intake by Sprague-Dawley rats lasting at least 4 h. Guesdon et al. (2008) used a potent, truncated MCH analog and also observed that injection of 5µg into the AcSh of Wistar rats increased food intake of regular chow threefold during the 2 h following injection.

Agonist studies in preproMCH-deficient rats confirmed that the orexigenic actions of MCH are independent of two other preproMCH encoded peptides, NGE and NEI. Acute AcSh administration of NGE and NEI, or chronic ICV infusion of NEI, did not affect feeding behavior in adult *Promch*+*/*<sup>+</sup> or *Promch*−*/*<sup>−</sup> rats (Mul et al., 2011). However, acute administration of MCH to the AcSh of adult *Promch*−*/*<sup>−</sup> rats elevated feeding behavior toward wild-type levels (Mul et al., 2011).

In addition to the effects of MCH on food intake, the role of MCH in mediating energy expenditure was also compared with that of two other orexigenic peptides, AGRP and orexin. Both AGRP and orexin, administered ICV (1 nmol/mouse), significantly decreased oxygen consumption compared with artificial cerebrospinal fluid (aCSF) treated controls; in contrast, MCH (1 nmol/mouse) had no significant effect compared with

aCSF-treated controls (Asakawa et al., 2002). However, an effect of MCH on oxygen consumption might not have been detected, since only a relatively low dose of peptide was tested.

Chronic ICV infusions of MCH into rodents were shown to not only increase food intake, but to also cause obesity (Della-Zuana et al., 2002; Gomori et al., 2003; Ito et al., 2003); while MCH-induced insulin resistance in rats was observed acutely in the absence of weight changes (Pereira-da-Silva et al., 2005). Chronic infusions of MCH (8µg/rat/day) over 12 days led to an increase in body weight of about 20 g more than did control aCSF infusions in both Wistar or Sprague-Dawley rats (Della-Zuana et al., 2002). After a 14-day infusion of MCH into the third ventricle of C57BL/6J mice (10µg/day), no significant increase in food intake was observed in mice fed a regular chow, but on a moderately high-fat diet (MHF), the mice ate about 15% more food (Gomori et al., 2003). Mice on both diets were significantly heavier than control mice, with the largest increase in body weight observed in the MCH infused mice on a MHF diet; these mice gained 17% more weight than the control infused mice on an MHF diet (Gomori et al., 2003). Glick et al. (2009) also observed that chronic infusion of 10µg/day of MCH for 14 days into C57BL/6 mice fed regular chow led to a 34% increase in food intake and a 15% increase in body weight after a 14-day infusion. In a separate study, C57BL/6J mice on an MHF diet infused with a lower amount of MCH (3µg/day for 7 days) also were hyperphagic and gained 350% more weight than did the vehicle control mice, with no detectable changes in activity (Ito et al., 2003). In addition, a small, potent, MCH analog given chronically ICV (30µg/day) to Sprague-Dawley rats increased food intake by 23% and body weight by 38% more than in the vehicle controls (Shearman et al., 2003).

Intracerebroventricular-injected MCH has metabolic effects beyond increases in food intake and body weight. The acute effects of single MCH injections probably identify direct effects from increased MCH signaling in the brain, while the chronic effects may be subsequent to increased adiposity associated with body-weight gain. A single ICV injection of MCH into Wistar rats (3 nmol) inhibited the thyroid axis (Kennedy et al., 2001) by suppressing release of thyroid hormone from the hypothalamus, leading to a suppression of plasma thyroid-stimulating hormone (Kennedy et al., 2001). ICV injections of 4 nmol of MCH into Wistar rats for 4 days resulted in an ∼10% increase in fasting plasma glucose (Pereira-da-Silva et al., 2005). Guesdon et al. (2008) used a potent, truncated MCH analog and found that injection of 5µg into the AcSh or the third ventricle of Wistar rats did not significantly increase energy expenditure, but it did increase glucose oxidation and reduce lipid oxidation in the period 3 h after injection.

Chronic ICV infusions of MCH into C57BL/6J mice resulted in increased fat pad weights (∼100%), leptin (∼300%), liver triglycerides (∼120%), fasting plasma glucose (∼10%), and insulin (∼100%) (Gomori et al., 2003). Glick et al. (2009) also observed that chronic infusion of 10µg/day of MCH for 14 days into C57BL/6 mice led to a 4.5-fold increase in adiposity, about a 0.4˚C drop in body temperature during the active dark phase, a 15% decrease in oxygen utilization, and a 26% increase in plasma insulin-like growth factor 1 levels. Chronic MCH agonist infusions into Sprague-Dawley rats also led to increases in insulin (∼200%) and leptin (∼400%) (Shearman et al., 2003).

#### **IN VIVO EFFECTS OF MCH RECEPTOR ANTAGONISTS**

As discussed above, rodent genetic models clearly show a role for MCH in energy homeostasis. However, the applicability of these studies to understanding the physiological role and disease states associated with MCH signaling may be limited due to developmental effects of MCH inactivation or neuronal loss affecting other neuromodulators. Moreover, the relevance of *in vivo* pharmacology studies is complicated, because they utilize MCH and MCH agonists administered ICV at what are probably supraphysiologic levels. To better understand the role of MCH in energy homeostasis, a series of *in vivo* studies utilized MCHR1-specific antagonists. Because many pathways can affect food intake and body weight, *in vivo* effects of MCHR1 antagonists might be due to non-specific effects; however, in a few cases, researchers have demonstrated MCHR1 specificity by showing that an antagonist is inactive in *Mchr1*−*/*<sup>−</sup> mice.

#### **Studies with MCHR1 peptidic antagonists**

Studies on truncated and substituted MCH analogs identified key amino acids for activation of the MCH receptors and led to synthesis of potent antagonist derivatives (Bednarek et al., 2002; Audinot et al., 2009). Several *in vivo* studies employed Ac- (Ava9–10, Ava14–15)-MCH(6–16)-NH2, a truncated, cyclic MCH analog containing γ-aminovaleric acid which is a potent (Kb 4 nM) antagonist (MacNeil and Bednarek, 2009). Shearman et al. (2003) reported that the peptide antagonist given ICV (10µg) did not significantly reduce spontaneous feeding, affect feeding duration or locomotor activity, or alter overnight body-weight gain in lean male Sprague-Dawley rats. However, the peptide antagonist blocked the initial 2- or 3-h hyperphagic activity of an MCH agonist (Shearman et al., 2003; Mashiko et al., 2005). Chronic ICV infusion of the antagonist at 48µg/day for 14 days reduced cumulative food intake by 13% and body-weight gain by 33%, relative to vehicle controls (Shearman et al., 2003). In male C57BL/6J mice fed an MHF diet, 28-day ICV infusion of the peptide antagonist at 7.5µg/day decreased cumulative food intake 15% and the antagonist-treated mice weighed 13% less than the vehicleinfused controls (Mashiko et al., 2005). The antagonist treatment led to improvements in other metabolic parameters, including reductions in plasma glucose, leptin, insulin, and total cholesterol, as well as reductions in fat pad and liver weights (Mashiko et al., 2005). The specificity of the MCHR1 peptide antagonist was confirmed using *Mchr1*−*/*<sup>−</sup> mice, as infusion of the antagonist for 2 weeks had no effect on body weight, fat mass, or cumulative food intake in the KO mice (Mashiko et al., 2005). In contrast to *Mchr1* KO mouse models (**Table 1**), no changes were noted in activity after a 4-week infusion of the MCHR1 antagonist (Mashiko et al., 2005). Obese, aged, male C57BL/6J mice fed a HFD for 1 y (body weight ∼60 g) were subjected to a 4-week ICV infusion with the peptide antagonist at 7.5µg/day (Ito et al., 2008). The antagonist-infused mice lost 21% of their weight, while vehicleinfused mice gained 6% more weight. Not surprisingly, given the large differences in body weight, the MCHR1 antagonist-treated

mice showed significant reductions in serum leptin and insulin (Ito et al., 2008). Also, the liver weights of the antagonist-treated DIO mice decreased by about 50%, while the serum liver markers showed improvements (Ito et al., 2008). In a follow-up study to explore the role of MCH in modulating accumulation of triglycerides in liver, male C57BL/6J mice were fed a diet deficient in methionine and choline to induce steatohepatitis; ICV treatment with the MCHR1 antagonist (7.5µg/day for 10 days) did not result in any body-weight difference vs. vehicle control, but accumulation of triglycerides in liver was reduced 33% (Ito et al., 2008). The ability of MCHR1 antagonist treatment to improve hepatic steatosis was confirmed in a female model of obesity in which ovariectomized mice were fed a regular chow diet. Thirty weeks after the ovariectomy, the female C57BL/6J mice weighed about 25% more than did the sham treated mice (Gomori et al., 2007). When these obese mice were treated ICV with the MCHR1 peptide antagonist for 4 weeks at 7.5µg/day, they lost 13% of their body weight and 32% of their liver triglycerides (Gomori et al., 2007).

Studies with S38151 in multiple rodent models of obesity resulted in significant effects on food intake and body weight. S38151 [*p*-guanidinobenzoyl-[des-Gly10]-MCH(7–17)] is a modified and truncated 11-amino acid MCH analog which is a potent (Kb = 4 nM) MCHR1 antagonist (Audinot et al., 2009). Injection of this peptidic MCHR1 antagonist into the AcSh reduced food intake (Georgescu et al., 2005). Over a 6-h period, in male Wistar rats, S38151 administered ICV dose dependently inhibited the orexigenic effects of previously injected MCH (Audinot et al., 2009). The highest doses of S38151, 30 and 50 nmol per rat, completely blocked MCH-induced food intake for 6 h (Audinot et al., 2009). The orexigenic effect of proMCH (131–165), which is more potent than MCH in stimulating feeding, was blocked for 2 h by 50 nmol/kg of S38151 administered ICV (Maulon-Feraille et al., 2002; Della-Zuana et al., 2012). Once daily ICV injections of 20 nmol/kg of S38151 into male Zucker *fa/fa* rats, reduced food intake, water intake, motility, and body weight (Della-Zuana et al., 2012). A single injection of 20µmol/kg of S38151 intraperitoneally (i.p.), reduced cumulative food intake in Zucker *fa/fa* rats for 24 h (Della-Zuana et al., 2012). S38151 was administered i.p. at 30 mg/kg for 5 days into two mouse models of obesity, female *ob/ob* and female C57BL/6J DIO mice, resulting in reductions in body weight and cumulative food intake. The S38151 effects on energy homeostasis were MCH1R based, since no changes in food intake or body weight were observed after 5 days of i.p. injection into female *Mchr1* KO mice (Della-Zuana et al., 2012).

#### **Studies with non-peptide MCHR1 antagonists**

The overwhelming set of genetic and physiologic data demonstrating that MCHR1 modulates energy homeostasis attracted the interest of a many medicinal chemistry groups who have synthesized numerous structurally distinct MCHR1 antagonists (see a recent review by Cheon, 2012). In all, 23 different companies have published more than 100 medicinal chemistry papers and patents describing attempts to optimize MCHR1 antagonist hit compounds toward drug candidates (see the recent review of the patent literature by Johansson, 2011). Many of the

#### **Table 2 | Acute effects of non-peptide MCH1R antagonists on food intake.**


\*Data shown for the most potent analog in the cited reference at the highest dose tested.

#Measured at 20 min.

resulting optimized lead compounds, representing a diverse set of non-peptide MCHR1 antagonists, have been evaluated *in vivo*. Non-peptide MCHR1 antagonists are effective in different models of acute food intake in a variety of different rodent strains (**Table 2**) (Borowsky et al., 2002; Takekawa et al., 2002; Huang et al., 2005; Palani et al., 2005; McBriar et al., 2006; Sasikumar et al., 2006; Xu et al., 2006; Balavoine et al., 2007; Kowalski and Sasikumar, 2007; Moriya et al., 2009; Nagasaki et al., 2009; Haga et al., 2011; Kamata et al., 2011; Kasai et al., 2011, 2012). In general, MCHR1 antagonists potently block up to 75% of MCH-induced food intake (Borowsky et al., 2002; Takekawa et al., 2002; Moriya et al., 2009; Nagasaki et al., 2009), but they have more modest effects on reducing fasting-induced feeding and spontaneous feeding.

Multiple MCHR1 antagonists have also shown dose-dependent and sustained efficacy in chronic models of obesity (Kym et al., 2005; Souers et al., 2005a,b, 2007; Vasudevan et al., 2005a,b; Carpenter et al., 2006; Hertzog et al., 2006; Tavares et al., 2006a,b; Mendez-Andino and Wos, 2007; Mendez-Andino et al., 2007; Gehlert et al., 2009; Ito et al., 2009; Semple et al., 2009; Suzuki et al., 2009; Hadden et al., 2010; Mihalic et al., 2012; Sasmal et al., 2012a,b). At the highest dose tested in DIO mice, ranging from 10 to 100 mpk, weight loss ranged from 5% at 5 days to 33% at 238 days (Ito et al., 2009; Mihalic et al., 2012). In several cases, weight loss was shown to be primarily due to the loss of fat mass (Souers et al., 2005a,b, 2007; Vasudevan et al., 2005a; Mendez-Andino et al., 2007). The mechanism of weight loss appears to involve a combination of reduced food intake, which was observed

in six studies (**Table 3**), and increased energy expenditure with no increase in activity (Kowalski et al., 2006; Gehlert et al., 2009).

As shown above with peptides, and in **Tables 2** and **3** with nonpeptide compounds, many different MCHR1 antagonists have demonstrated modest to robust efficacy in a variety of rodent obesity models, suggesting that MCHR1 antagonists may have potential for treating human obesity. One caveat for the data is that in most cases, the antagonists have not demonstrated the ability to induce weight loss through an MCHR1-specific mode of action, since efficacy in *Mchr1* KO mice has not been evaluated. In fact, in one report of robust weight loss, the authors caution that some weight loss at the highest dose tested maybe due to non-specific mechanisms, since the compound resulted in high brain levels, 6.46µg/g (Kym et al., 2005). However, in four reports, the MCHR1-specific efficacy of antagonists was confirmed, since the antagonists lacked efficacy in DIO *Mchr1* KO mice, but they showed body-weight loss in DIO wild-type mice (Gehlert et al., 2009; Della-Zuana et al., 2012; Mashiko et al., 2005; Mihalic et al., 2012). Further support for an MCHR1 mechanism based-weight loss has been reported for three compounds, for which efficacy was correlated with brain MCHR1 receptor occupancy (Hervieu et al., 2003; Kowalski et al., 2006; Ito et al., 2009).

#### **Studies with MCHR2 antagonists**

The absence of the *Mchr2* in rodents has limited the interest in developing MCHR2-selective compounds. Only one paper has disclosed a potent MCHR2-selective non-peptide small-molecule

#### **Table 3 | Chronic effects of non-peptide MCH1R antagonists in rodents.**

#### **Rodent model of energy homeostasis\***


\*Data shown for the most potent analog in the cited reference at the highest dose tested.

antagonist (Chen et al., 2012). Compound 38 is a potent (Ki 13 nM) and selective MCHR2 antagonist with good oral bioavailability and pharmacokinetics suitable for *in vivo* studies (Chen et al., 2012). However, no *in vivo* data have been reported with this compound.

#### **CLINICAL STUDIES WITH MCHR1 ANTAGONISTS**

The intense interest in MCHR1 antagonists has resulted in more than 80 publications and 100 patent applications describing various unique antagonists (Johansson, 2011). Five compounds have reached testing in human subjects, but none has proceeded into advanced Phase II studies to rigorously test their efficacy in causing chronic weight loss (**Figure 2**). A major issue with many lead compounds is increased cardiovascular risk due to high-affinity hERG binding and drug-induced QTc prolongation (Mendez-Andino and Wos, 2007).

TheAmgenMCHR1 antagonistAMG 076 entered Phase I safety and tolerability testing in 2004, but there have been no subsequent reports of its status since 2005. GlaxoSmithKline MCHR1 antagonist GW-856464 also entered Phase I studies in 2004 (Cheon,

2012). However, on August 24, 2010, Carmen Drahl reported via Twitter that "low bioavailability precluded further development."

NGD-4715 is a selective MCHR1 antagonist developed by Neurogen. In a May 2, 2007 press release, Neurogen announced that NGD-4715 was safe and well-tolerated in a Phase I clinical trial. Neurogen was acquired by Ligand Pharmaceuticals in 2009; as of February 22, 2013, the Wikipedia entry for NGD-4715 quotes an e-mail from Ligand that there are "no plans for further development on the compound." In a May 31, 2011 press release, AMRI announced that its MCHR1 antagonist, ALB-127158(a), was well-tolerated in a Phase I single ascending-dose study and 14-day multiple ascending-dose safety and tolerability study. An encouraging result reported by some subjects was loss of appetite. Despite the reported tolerability and suggestion of efficacy, in a subsequent press release dated October 4, 2011, AMRI announced that development was terminated before the initiation of Phase II studies.

Bristol–Myers Squibb (BMS) conducted the longest clinical trial with an MCHR1 antagonist. As of February 22, 2013, the BMS website indicates BMS-830216 was evaluated in a 28-day Phase I study to assess the safety, tolerability, and effect on body weight and other obesity-related factors of different doses of BMS-830216. A summary of the clinical results with BMS-830216 is available on the NIH website (http://www.ncats.nih.gov/files/BMS-830216.pdf), which indicates that BMS-830216 is a prodrug of the antagonist BMS-819881 and a potent (*K*<sup>i</sup> = 10 nM) and selective MCHR1 antagonist. BMS-830216 was generally safe and well-tolerated at all doses in the Phase I study for up to 28 days. However, no indications of weight loss or reduced food intake were observed, and the compound did not proceed to Phase II studies.

# **FUTURE PROSPECTS**

A rich literature of rodent genetics and rodent pharmacology demonstrates a significant role for MCH, acting via the MCHR1 within the hypothalamus, in maintaining energy homeostasis. This knowledge has stimulated more than 20 companies to seek MCHR1 selective compounds for the treatment of obesity. Five companies are known to have succeeded in identifying development candidates that proceeded through preclinical safety studies and enabled Phase I clinical safety and tolerability testing. Three of the compounds, NGD-4715, ALB-127158(a), and BMS-830216 (**Figure 1**), were found generally safe and well-tolerated in the Phase I studies. However, BMS-830216, which was tested for 28 days in obese subjects, failed to show any significant weight loss efficacy. No compounds have proceeded into Phase II studies in which chronic efficacy could be evaluated.

After almost 15 years of research, studies have failed to detect anti-obesity efficacy with MCHR1 antagonists in the clinic. There are multiple reasons that might account for this disconnect between rodent and human studies. First, compounds tested in the clinic may not have had the appropriate properties to sufficiently block the MCHR1 receptor and achieve an effect of energy balance. For instance, the MCHR1 antagonists may not have reached the hypothalamic sites of MCH action. Since there have been no reported studies measuring the level of receptor occupancy of the clinical compounds, it is possible that the compounds failed to sufficiently penetrate the brain: blood barrier resulting in low and insufficient receptor occupancy. Second, studies utilizing a positron emission tomography (PET) ligand demonstrated that an NPY5R antagonist must achieve >90% receptor occupancy sustained over 24 h to cause weight loss (Erondu et al., 2006). Similarly, in DIO mouse studies, maximum weight loss was observed only when an MCHR1 antagonist blocked >90% of the MCHR1 receptors for 24 h (unpublished data by the author). Clearly, a MCHR1 PET ligand could guide compound dose selection to assure 24 h of high receptor blockade (Erondu et al., 2006; Philippe et al., 2012). Third, since humans also express MCHR2 in the hypothalamus (Sailer et al., 2001), MCH signaling in humans may involve both MCHR1 and MCHR2, such that blockade of the MCHR1 alone may not be sufficient to achieve efficacy in obesity. Fourth, the role of MCH in human energy homeostasis may not be as significant as its role in rodents, consequently, blockade of MCHR1 would be inherently ineffective as a treatment for obesity.

In the author's opinion, there is, at best, only a modest probability that an MCHR1 antagonist will be developed as a treatment for obesity. Following thefailure of five Phase I compounds to progress to Phase II efficacy studies, the enthusiasm for MCHR1 antagonists has clearly dimmed. Moreover, after more than a decade of drug discovery effort by more than 20 companies, there are currently no known MCHR1 antagonists in the clinic for obesity. Should any company continue to seek MCHR1 antagonists, the author

# suggests that they should proceed only with the aid of a PET ligand, perhaps [11C]SNAP-7941, or another on-target biomarker to assess receptor occupancy (Philippe et al., 2012). Companies might also consider focusing their chemistry efforts on identifying a dual MCHR1 and MCHR2 antagonist, to assure that all MCHmediated signaling involved in energy homeostasis is effectively blocked. Finally, an effective MCH antagonist therapy for obesity must not only achieve meaningful weight loss, but it must also be well-tolerated. Although not discussed in this review, there is significant rodent genetic and pharmacologic data indicating that MCH participates in other CNS behaviors (Saito and Nagasaki, 2008; Yumiko and Nagasaki, 2008; Antal-Zimanyi and Khawaja, 2009; Torterolo et al., 2009; Chung et al., 2011). In particular, studies in multiple rodent models suggest blockade of MCH signaling may be anxiolytic (Antal-Zimanyi and Khawaja,2009;Chung et al., 2011). Tolerability may be a substantial hurdle for an effective MCH receptor antagonist to overcome.

#### **ACKNOWLEDGMENTS**

I thank Tanya MacNeil for reviewing and editing the manuscript and Michele McColgan for preparing the figures. Current information about the structure and function of the MCH receptors can be found at the IUPHAR database (IUPHAR-DB): http://www.iuphar-db.org/DATABASE/ FamilyMenuForward?familyId\$=\$37.

#### **REFERENCES**


*Regul. Integr. Comp. Physiol.* 291, 1265–1274.


1 5' polymorphism with earlyonset extreme obesity. *Diabetes* 54, 3049–3055.


*Physiol. Endocrinol. Metab.* 284, E583–E588.


"The melanin-concentrating hormone," in *Handbook of Chemical Neuroanatomy; Vol. 20 Peptide Receptors*, Part II, eds A. Bjorklund and T. Hokfelt (Amsterdam: Elsevier), 31–101.


hormone receptor 1 antagonists 3-aminomethylquinolines: reducing human ether-a-go-gorelated gene (hERG) associated liabilities. *J. Med. Chem.* 55, 4336–4351.


brain cDNA encoding for the SLC-1 G protein-coupled receptor reveals the presence of an intron in the gene. *Biochim. Biophys. Acta* 1401, 216–220.


(MCHR1) antagonist with reduced hERG inhibition. *Bioorg. Med. Chem. Lett.* 22, 3781–3785.


O. (1999). Molecular characterization of the melanin-concentratinghormone receptor. *Nature* 400, 265–269.


food intake in sheep. *Domest. Anim. Endocrinol.* 28, 224–232.


**Conflict of Interest Statement:** The author is employed by a company with a goal of developing novel therapies.

*Received: 22 February 2013; paper pending published: 12 March 2013; accepted: 09 April 2013; published online: 22 April 2013.*

*Citation: MacNeil DJ (2013) The role of melanin-concentrating hormone and its receptors in energy homeostasis. Front. Endocrinol. 4:49. doi: 10.3389/fendo.2013.00049*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 MacNeil. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Peripheral injections of melanin-concentrating hormone receptor 1 antagonist S38151 decrease food intake and body weight in rodent obesity models

*Odile Della-Zuana1, Valérie Audinot 2, Viviane Levenez 1, Alain Ktorza1, Françoise Presse3,4, Jean-Louis Nahon3,4 and Jean A. Boutin2 \**

<sup>1</sup> Maladies Métaboliques, Institut de Recherches SERVIER, Suresnes, France

<sup>2</sup> Biotechnologie, Pharmacologie Moléculaire et Cellulaire, Institut de Recherches SERVIER, Croissy-sur-Seine, France

<sup>3</sup> Genomics and Evolution in Neuroendocrinology, Institut de Pharmacologie Moléculaire et Cellulaire, UMR7275, Centre National de la Recherche Scientifique, Valbonne, France

<sup>4</sup> Genomics and Evolution in Neuroendocrinology, Université de Nice Sophia Antipolis, Nice, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Berta Levavi-Sivan, The Hebrew University, Israel James A. Carr, Texas Tech University, USA

#### *\*Correspondence:*

Jean A. Boutin, Biotechnologie, Pharmacologie Moléculaire et Cellulaire, Institut de Recherches SERVIER, 125 chemin de Ronde, 78290 Croissy-sur-Seine, France. e-mail: jean.boutin@fr.netgrs.com The compound S38151 is a nanomolar antagonist that acts at the melanin-concentrating hormone receptor 1 (MCH1). S38151 is more stable than its purely peptide counterpart, essentially because of the blockade of its N-terminus. Therefore, its action on various models of obesity was studied. Acute intra-cerebroventricular (i.c.v.) administration of S38151 in wild-type rats counteracted the effect of the stable precursor of melanin-concentrating hormone (MCH), NEI-MCH, in a dose-dependent manner (from 0.5 to 50 nmol/kg). In genetically obese Zucker fa/fa rats, daily i.c.v. administration of S38151 induced dose-dependent (5, 10, and 20 nmol/kg) inhibition of food intake, water intake, and body weight gain, as well as increased motility (maximal effect observed at 20 nmol/kg). In Zucker fa/fa rats, intraperitoneal injection of S38151 (30 mg/kg) induced complete inhibition of food consumption within 1 h. Daily intraperitoneal injection of S38151 (10 and 30 mg/kg) into genetically obese ob/ob mice or diet-induced obese mice is able to limit body weight gain. Furthermore, S38151 administration (10 and 30 mg/kg) does not affect food intake, water intake, or body weight gain in MCHR1-deleted mice, demonstrating that its effects are linked to its interaction with MCH1. These results validate MCH1 as a target of interest in obesity. S38151 cannot progress to the clinical phase because it is still too poorly stable in vivo.

**Keywords: melanin-concentrating hormone, receptors, peptide antagonists, feeding behavior, rat, mice, obesity models**

#### **INTRODUCTION**

Melanin-concentrating hormone (MCH) is a 17-amino-acid, cyclic pseudopeptide that is responsible for the bleaching of skin in teleost fishes (Kawauchi et al., 1983). In mammals, MCH is a 19-amino-acid cyclic peptide (Vaughan et al., 1989). Its actions are not linked to pigmentation, but to a variety of properties, originally centered on appetite regulation and obesity (Qu et al., 1996; Pissios et al., 2006). MCH acts in several processes, including sleep/arousal (Hervieu, 2006; Peyron et al., 2009), emotionality (Hervieu, 2006), and memory (Adamantidis and de Lecea, 2009). MCH is encoded as a pre-pro-hormone, and also gives rise to two other neuropeptides after proteo-cleavage: (1) neuropeptide N-G (NGE), the function of which is elusive, and (2) neuropeptide E-I (NEI) (Nahon et al., 1989), which displays various functions (Maulon-Feraille et al., 2002; Bittencourt and Celis, 2008). NEI-MCH is more potent than MCH in stimulating feeding in rats, for proteolytic-protective reasons (Maulon-Feraille et al., 2002). MCH is mainly synthesized in cell bodies of the lateral hypothalamus and sub-*zona incerta* in the central nervous system (Bittencourt et al., 1992). MCH is also synthesized in peripheral organs, such as the gut, arteries, testes, thymus, and pancreatic β-cells (Hervieu and Nahon, 1995).

Two MCH receptors (MCH1 and MCH2) were identified a decade ago. MCH1 was known for quite some time as SLC-1, an orphan receptor in humans (Lakaye et al., 1998). It was established as a genuine MCH receptor in 1999 by at least three different groups (Bachner et al., 1999; Chambers et al., 1999; Saito et al., 1999). The subsequent search for analogs of the protein led to the cloning of a second receptor, MCH2 (Mori et al., 2001; Rodriguez et al., 2001; Sailer et al., 2001; Wang et al., 2001). This second receptor is expressed in humans, dogs, and ferrets, but not in rodents. In mammalian species, the highest expression levels of both receptors are found in the frontal cortex, amygdala, and nucleus accumbens; however, they are also expressed in hypothalamic areas regulating energy balance, such as the arcuate nucleus and the ventral medial hypothalamus. MHCRs are also moderately expressed in peripheral organs. In rodents, MCH acts through MCH1, which implicates MCH1 in obesity and energy homeostasis.

Several genetically engineered strains of mice have been reported. MCH1-knockout (MCH1-KO) mice (mice do not express a functional MCHR2 protein) exhibit a phenotype of leanness and resistance to diet-induced obesity, characterized by hyperphagia, hyperactivity, and hypermetabolism (Chen et al., 2002; Marsh et al., 2002; Pissios, 2009). This phenotype is attributed to increased energy expenditure, resulting from increased locomotor activity and increased resting energy expenditure. Paradoxically, these mice exhibit significant hyperphagia.

The molecular pharmacology of MCH1 has been studied in some detail. After the centrally important works of teams at Takeda and Synaptic/Lundbeck (Borowsky et al., 2002; Takekawa et al., 2002), many non-peptide compounds were screened and found active at MCHR1. Most of these small molecules inhibit MCH-mediated feeding behavior; however, they may also affect energy expenditure. Interestingly, MCH1 antagonists of different chemical classes exhibit anxiolytic and anti-depressant effects, a finding that would likely contribute to the success of these molecules in obese populations that might also have depression and anxiety. However, cardiovascular risk associated with human ether-a-go-go-related gene (hERG)-binding activity of biaryl-containing compounds has plagued many of these nonpeptide MCH1 antagonist programs (Mendez-Andino and Wos, 2007; Meyers et al., 2007; Johansson, 2011).

Alternatively, several dozen MCH analogs have been synthesized and tested by binding, leading to the discovery of peptide super-agonists and mild antagonists. Some of these have been tested on MCH2, and present major selectivity to MCH1. Structure activity relationship is now well-established concerning the peptide ligands. Although small molecules had been synthesized as potent and moderately selective MCH1antagonists, there was still an interest in finding new peptides with antagonistic activities. Pioneering works by Bednarek's group described such interesting compounds. In agreement with our initial search for better agonists (Audinot et al., 2001a,b), we explored the possibility of designing natural peptides or pseudopeptides [pseudopeptides are peptides including one or more exotic aminoacid(s) in their sequence] with good MCH1 antagonism and fair *in vivo* stability. We succeeded in describing S38151 (Audinot et al., 2009), a potent peptide at MCH1 with antagonistic activity in the nanomolar range. The aim of the present study was to evaluate the possible inhibition of S38151 on food and water intakes, body weight, and motility after one-time or repeated daily peripheral administration in different rodent models of obesity. Confirmation of the MCH1-mediated actions of S38151 was tested in mice with genetic ablation of the receptor. Mice with genetic ablation of MCH1were used to confirm that the transient inhibition of S38151 onto the bodyweight was mediated by MCH1.

# **MATERIALS AND METHODS PEPTIDES**

MCH and S38151 (Audinot et al., 2009) were obtained from Polypeptide Laboratories (Illkirch, France) and/or from Genepep (Saint Jean de Védas, France). They were of purity exceeding 97%. S38151 is a pseudopeptide, the sequence of which is: [p-guanidinobenzoyl-(Des-Gly10)-MCH (7–17)], or pGua-Cys-Met-Leu-Arg-Val-Tyr-Arg-Pro-Cys; both cysteines are linked together with a disulfide bridge. The molecular weight of S38151 is 1485 g/L. All of the various lots were systematically analyzed using mass spectrometry.

#### **S38151 STABILITY STUDIES**

First, S38151 was incubated for 16 h at 4◦C and 37◦C in mouse plasma or in TRIS buffer (pH 7.4). Next, plasma proteins were precipitated with two volumes of acetonitrile. After centrifugation and dilution with a water/acetonitrile mixture (50/50 v/v), the supernatant was injected onto a triple quadripole mass spectrometer (Ultima, Micromass WATERS Corporation, Milford, USA). The liquid chromatographic system consisted of an Agilent 1100 analytic pump (Agilent, Santa Clara, USA) equipped with an Alpha Mos CTC autosampler (Alpha Mos, Toulouse, France). Mass detection was performed using the m/z transition 743.5 *>* 163, in positive electrospray. A second stability study was performed at a lower concentration (1 μM) and for 4 h, with the following sampling times: 0, 15, 45, 90, and 240 min. At each sampling time, 50μL of plasma were precipitated with two volumes of acetonitrile, and after centrifugation were directly injected onto the LC-MS/MS system. The same treatment was used for the buffer samples. Although brains from S38151-treated animals were analyzed to determine the amount of compound detectable in the brain, none was detected (data not shown).

# **ANIMALS**

These experiments were conducted with male Wistar and Zucker *fa/fa* rats (10–13 weeks of age, weighing 325–350 g) together with C57BL/6J mice, *ob/ob* mice, and MCH1-KO mice (14 weeks of age, weighing 25, 45, and 25 g, respectively) (Iffa Credo and Charles River, L'Arbresle, France). The animals were housed individually in a room with a 12-h light/dark cycle (lights on at 07:30 and off at 19:30), at 22 ± 3◦C and 55% relative humidity. Normal food and tap water were available *ad libitum* unless otherwise stated. All animal procedures described in this study comply with French laws regulating animal experimentation (Decree No 87- 848 19th October 1987 and the ministerial Decree of 10 April 1988) and were approved by the animal ethics committee of the Servier Research Institute.

# **INTRA-CEREBROVENTRICULAR CANNULA IMPLANTATION** *Acute i.c.v. injections in rats*

Wistar and Zucker *fa/fa* rats were anesthetized with Forène (Abbott Laboratories, Queenborough, UK), and a stainless steel guide cannula (Plastic Products Co, Roanoke, Va, USA) was stereotaxically implanted into the right lateral ventricle at the following coordinates relative to the bregma: *AP* −0.8 mm, *L* = − 1*.*2 mm, and *V* = −3*.*5 mm. After a 7-day recovery period, during which the animals were handled each day to minimize non-specific stress, they were lightly anesthetized with Forène. Human/mouse/rat MCH (1μg in a volume of 2.5μL, Bachem, Voisins-le-Bretonneux, France) or an equivalent volume of artificial cerebrospinal fluid (CSF) was then injected through the intraventricular cannula. Immediately after injection, the animals quickly recovered and were returned to their home cages. Only those animals responding within 2 h with a robust increase in food intake, indicating correct cannula placement, were used in the experiments described below. Two days after MCH injection, the rats were randomly assigned to different groups and studied in one of the following protocols.

# *Acute central peptide administration and feeding studies in Wistar rats*

Wistar rats were habituated for at least 2 weeks before experiment to a diet of food pellets (6 mm diameter) of the following composition: 67.5% food flour, 26.5% sucrose, 5% gum tragacanth, and 1.25% magnesium stearate (A03 UAR, Orge, France). The food pellets were present in a metal food hopper attached to the inside of each cage. At 09:00 on the day of study, the animals were lightly anesthetized with Forene® (Abbott, France). Thereafter, rats were injected intra-cerebroventricular (i.c.v.) with vehicle (artificial CSF) or with S38151 [0.5, 5, 10, 20, 30, or 50 nmol/kg (0.7, 1.4, 14, 28, 42, or 70μg/kg, respectively)] in a volume of 2.5μL, 20 min prior to the injection of human/rat NEI-MCH [5 nmol/kg (18.7 μg/kg) in a volume of 2.5 μL].

Food intake was measured at different times after peptide injection. Food spillage was carefully quantified during each time period, and intake measurements were corrected for this loss. At the end of the experiments, the animals were euthanized and the positions of the cannulae were assessed by the injection of 100 μL Evans blue dye (2 mg/mL) followed by the visual examination of brain slices. Only data obtained from animals with correctly positioned cannulae were included in the final data analysis.

#### *Daily central peptide administration and feeding studies in Zucker rats*

The TSE Drinking and Feeding Monitor system was used to analyze ingestive behavior in these experiments (TSE Technical & Scientific Equipment GmbH, Bad Homburg, Germany). The rats' motility was also determined in these experiments using the MoTil® system (TSE). Zucker *fa/fa* rats were placed individually into plastic cages to which were attached both a feeding and a drinking sensor. The animals were habituated to normal food (A03 UAR, Orge, France) and water for at least 2 weeks before experimentation. One hour prior to the beginning of the dark phase, Zucker *fa/fa* rats were injected i.c.v. with different doses of S38151 [5, 10, or 20 nmol/kg (3.5, 14, or 28 μg/kg, respectively)] or with an equivalent volume of artificial CSF vehicle (10μL). The injections were repeated daily for a total of 7 days. After 7 days, the injections were stopped and each measured parameter was followed during a 5-day washout period.

# **ACUTE PERIPHERAL PEPTIDE ADMINISTRATION AND FEEDING STUDIES IN ZUCKER RATS**

Zucker fa/fa rats were habituated in the same conditions as Wistar rats (described in section "Acute Central Peptide Administration and Feeding Studies in Wistar Rats"). At 09:00 on the day of study, after fasting for 24 h, either vehicle (normal saline) or S38151 [30 mg/kg (20μmol/kg)] was administered intraperitoneally (i.p.).

#### **CONDITIONED TASTE AVERSION**

These experiments were performed according to previously described methods (Criscione et al., 1998; Welzl et al., 2001). In brief, Wistar rats were individually housed in plastic cages with free access to normal food and water. During a 7-day habituation period, the water bottle was removed from the cages overnight (from 16:00 to 10:00). At 10:00, water was made available to the animals from 250-mL plastic bottles to which a normal drinking spout was attached. The bottles were weighed before and after presentation to the animals to determine the quantity of water ingested during this period. After 4 days of this treatment regime, which served to habituate the animals to the experimental protocol and to randomize the animals, 0.2% saccharine was presented instead of water. Twenty minutes after the test period, the animals were injected i.p. with lithium chloride (75 mg/kg) or i.c.v. with S38151 [20 nmol/kg (28μg/kg)] or vehicle (artificial CSF). Four, 7, 11, and 14 days later, the consumption of saccharine was determined. Considerable care was taken in these experiments to first prevent and then quantify any leakage of saccharine solution during the presentations.

# **PICA**

These experiments were performed with modifications to previously described methods (Madden et al., 1999). Rats were individually housed in wire-bottomed plastic cages with free access to both food pellets (6 mm diameter A03 UAR) and kaolin powder. The food pellets were presented in a small porcelain bowl on the floor of the cage. The kaolin powder was presented in a metal feeding dish attached to the inside of the cage. After a 2-week habituation period to these conditions, the animals were randomized into five treatment groups based on the quantity of food eaten during the stabilization period. In addition, those rats that consistently spilled the kaolin powder were eliminated from the study. One day after randomization, the animals were injected just before the beginning of the dark phase with either cyclophosphamide (100 mg/kg, i.p.), or S38151 [20 nmol/kg (28 μg/kg) i.c.v.] or vehicle [(artificial CSF) i.c.v.]. After injection, the quantity of kaolin and food ingested during the following 24-h period was determined. Food or kaolin spilled by the animals was collected at the end of the experiment, and total food and kaolin intake measurements were corrected for this loss.

# **CONSTRUCTION OF MCHr1−***/***<sup>−</sup> MICE** *Construction of the targeting vector*

The MCHR1−*/*<sup>−</sup> mouse model was custom-constructed by genOway (Lyon, France). The construction of the targeting vector and knock-in strategy were designed and performed by genOway (Lyon, France). A genomic clone containing the murine Mchr1 locus was isolated from a C57BL/6J RPCI-24 BAC genomic library by using a probe corresponding to the murine Mchr1 exon 2: one BAC clone (544N15) containing the Mchr1 locus. The genomic organization of the targeted locus was determined by subcloning the XhoI-SpeI genomic fragment into the pZErO™-2 vector (Invitrogen, Carlsbad, California). The 7.4-kb XhoI-SpeI genomic insert was sequenced and the Mchr1 sequence was generated. The genomic clone (containing the entire gene) was used to construct the targeting vector. Briefly, a 5.1-kb XhoI/ApaI fragment comprising Mchr1 exons 1 and 2 and a 1.1-kb ApaI/BglII fragment located downstream of Mchr1 exon 2 were used to flank a NEO cassette (FRT-PGK promoter-NeoR cDNA-FRT-LoxP) (**Figure 1A**); a negative (DTA) selection cassette was introduced at the 5 end of the long arm of homology.

allele band (6.6 kb).

#### *Screening of Mchr1-recombined ES cell clones*

NotI-linearized targeting vector was transfected into 129SvPas ES cells (genOway, Lyon, France) according to genOway's modified electroporation procedures (i.e., 108 ES cells in the presence of 100μg of linearized plasmid, 800 V, 300μF). Positive selection was started 48 h after electroporation, by adding 200 μg/mL of G418 (150μg/mL of active component, Life Technologies, Inc.). A total of 333 resistant clones were isolated and amplified in 96-well plates, and duplicates were made of the 96-well plates. The set of plates containing ES cell clones amplified on

are shown. The probe used in all Southern blot analyses was a 0.65-kb

gelatin was screened by PCR and further confirmed by Southern blotting.

5 PCR screening conditions were: GW676 primer specific for a region upstream of *Mchr1* exon 1 (5 -GATATCAATTCGGGA CACATGG-3 ) and GW681 primer specific for the NeoR selection cassette (5 -TCTCGGCAGGAGCAAGGTGAGATGACAG-3 ). PCR conditions were 94◦C/2 min, and 35 cycles of (94◦C/30 s, 59◦C/30 s, and 68◦C/6 min) followed by 68◦C/10 min, resulting in a 6509-bp mutated allele. 3 PCR screening conditions were: GW668 primer specific for the Neo® selection cassette (5 -ATCA GGACATAGCGTTGGCTAC-3 ) and GW669 primer to hybridize the region downstream of the *Mchr1* gene (5 -ATGAGAAGTG ACCAGCAAGAGC-3 ). PCR conditions were 94◦C/2 min, and 35 cycles of (94◦C/30 s, 61◦C/30 s, and 68◦C/2 min), followed by 68◦C/10 min, resulting in a 1609-bp recombined allele. Both PCR reactions were performed using Long Expand High Fidelity polymerase (Roche®) and reaction buffer 3. PCR products were then digested with PmeI to confirm the integration of the distal *loxP* site within the recombined allele. Briefly, for Southern blot analysis, genomic DNA was digested with AvrII/SacII and then hybridized with a 0.65-kb probe; *Mchr1* ± clones gave rise to a 4.2-kb wild-type signal and 5.6-kb recombined signal.

*In vitro Cre-mediated deletion of Mchr1-recombined ES cell clones*

Two recombined ES cell clones were used to perform Cremediated deletion of the floxed *Mchr1* region. Supercoiled Creexpressing plasmid was transfected into *Mchr1*-recombined ES cells (genOway, Lyon, France) according to genOway standard procedures. No selection was applied. One hundred ES cell clones were isolated and amplified in 96-well plates, and duplicates were made of the 96-well plates. The set of plates containing ES cell clones amplified on gelatin was screened by PCR and further confirmed by Southern blotting. The PCR screening conditions were: GW787 primer specific for *Mchr1* exon 1 (5 -AGCTCTGAAGGAGAAGGGAATG-3 ) and GW669 primer to hybridize the region downstream of the *Mchr1* gene (5 -ATGAGAAGTGACCAGCAAGAGC-3 ). The PCR conditions were 94◦C/2 min, and 35 cycles of (94◦C/30 s, 59◦C/30 s, and 68◦C/6 min), followed by 68◦C/10 min, resulting in a 4481-bp recombined allele and a 1564-bp deleted allele. The PCR reaction was performed using Long Expand High Fidelity polymerase (Roche®) and reaction buffer 3. Briefly, for Southern blot analysis, genomic DNA was digested with AvrII/SacII and then hybridized with a 0.65-kb probe. The *Mchr1* wild-type allele gave rise to a 4.2-kb signal, the *Mchr1*-recombined allele gave rise to a 5.6-kb signal, and the *Mchr1* deleted allele gave rise to a 6.6-kb signal.

#### *Generation of chimeric mice and breeding scheme*

One floxed mutated *Mchr1* ES cell clone (clone #2A9-2A10) was microinjected into C57BL/6J blastocysts, and gave rise to male chimeras with a significant ES cell contribution (as determined by Agouti coat color). After mating with C57BL/6J females, germline transmission was confirmed by the genotyping of tail DNA from offspring using PCR and Southern blot analysis. Knock-out heterozygous animals were screened as described in section "Screening of Mchr1-Recombined ES Cell Clones." PCR screening conditions were: GW787 primer specific for *Mchr1* exon 1 (5 -AGCTCTGAAGGAGAAGGGAATG-3 ) and GW669 primer to hybridize the region downstream of the *Mchr1* gene (5 -ATGAGAAGTGACCAGCAAGAGC-3 ). PCR conditions were 94◦C/2 min, and 35 cycles of (94◦C/30 s, 59◦C/30 s, and 68◦C/6 min), followed by 68◦C/10 min, resulting in a 3.1-kb signal for the wild-type allele and 1.6-kb signal for the deleted allele. The PCR reaction was performed using Long Expand High Fidelity polymerase (Roche®) and reaction buffer 3. Briefly, for Southern blot analysis, genomic DNA was digested with AvrII/SacII and then hybridized with a 0.65-kb probe. The wildtype *Mchr1* allele gave rise to a 4.2-kb signal and the deleted *Mchr1* allele gave rise to a 6.6-kb signal (**Figure 1B**). F1 male and female heterozygous animals were interbred to obtain *Mchr1-*null mice, and offspring were also screened by PCR and Southern blot analysis as described in this section.

#### **FOOD INTAKE IN C57BL/6J AND ob/ob MICE**

For the MCH daily repeated i.c.v. injection studies, 13-week-old female C57BL/6J and *ob/ob* mice were used (Harlan, Gannat, France).

#### **FOOD INTAKE INDUCED OBESITY IN C57BL/6J, ob/ob, AND MCHr1-KO MICE**

In these studies, 13-week-old female C57BL/6J, ob/ob, and MCH1-KO mice were used. The mice were maintained on a 12-h light/dark cycle (lights on at 07:30 and off at 19:30) at 22 ± 3◦C, and supplied with food pellets (6 mm diameter A03, UAR Laboratory chow, Epinay Villemuisson France) and tap water *ad libitum*. To induce obesity in C57BL/6J (creating diet-induced obese, or DIO, mice), mice were fed a high-fat diet (60 kcal% fat, 20 kcal% protein, 20 kcal% carbohydrate, ref D12492 from Research Diets, New Brunswick, NJ08901, USA) for 8 weeks starting at 4 weeks of age. In these experiments, DIO mice, ob/ob mice, and MCH−*/*<sup>−</sup> <sup>1</sup> mice were individually housed in a modular chamber that was placed on an activity platform to concurrently measure the food intake, water intake, and activity of the mice (ADDENFI, Les Cordeliers, Paris, France). For at least 2 weeks before experimentation, mice were habituated to a diet of 6 mm diameter food pellets of the following composition: 67.5% food flour, 26.5% sucrose, 5% gum tragacanth, 1.25% magnesium stearate (A03 UAR, Orge, France), and to i.p. injection. After this habituation period, food intake, drink intake, and body weight were measured for 3 consecutive days. The mice were randomized into three groups based on their food and water intake over the three preceding 24-h periods, and on their body weight. Next, the quantity of food eaten over each 24-h period during the 5 days of i.p. treatment with S38151 [10 and 30 mg/kg/day (7 and 20μmol/kg/day)] or vehicle was determined and corrected for spillage.

# **DATA AND STATISTICAL ANALYSIS**

Data were grouped together and presented as mean ± standard error (SEM). The individual statistical tests that were used in these studies are described in detail in the figure legends. In each case, *P*-values less than 0.05 were considered statistically significant.

# **RESULTS**

# **S38151 CHARACTERISTICS AND SERUM STABILITY**

S38151 was synthesized during a process that was attempting to find peptides or pseudopeptides with affinity at MCH1 receptor that also had greater stability in biological media. This pseudopeptide is derived from another, very similar agonist, pGua (MCH7-17) (Audinot et al., 2001a, 2009). In Chinese hamster ovary cells over-expressing MCH1, S38151 exhibited a K*<sup>i</sup>* of 80 nM in a [125I]-S36057 binding assay and was a potent and full antagonist (*KB* = 4*.*3 nM in a GTPγS binding assay, and *KB* = 210 nM in a calcium flux assay) (Audinot et al., 2009). This activity was specific to MCH1, since it has no activity with respect to MCHR2 at concentrations of less than 10 μM. This compound was also evaluated on a series of 70 targets, comprising monoaminergic and peptidergic receptors as well as enzymes, ionic channels, and transporters. S38151 did not display any activity regarding these targets at concentrations of less than 10 μM (data not shown). The peptide was incubated either in a buffered solution (pH 7.4) or in mouse plasma for 16 h, at either 4◦C or 37◦C. **Figure 2** clearly shows that at 37◦C, a minimal portion of S38151 is detectable, suggesting that it has limited stability in plasma.

#### **THE EFFECT OF A SINGLE i.c.v. INJECTION OF S38151 ON NEI-MCH-INDUCED FOOD INTAKE IN WISTAR RATS**

The acute effects of a single i.c.v. injection of S38151 on NEI-MCH-induced food intake in Wistar rats are summarized in **Figure 3**. During the first 2 h after its injection, NEI-MCH produced a significant increase in food intake above the corresponding values observed in vehicle-injected control animals (T1h: 3.18 g ± 1.09 vs. 0.45 g ± 0.19; T2h: 2.73 g ± 0.89 vs. 1.25 g ± 0.38). Relative to control values, the increase in food intake induced by NEI-MCH became progressively greater during the following 4-h period (from T1 h to T4 h). During the first and second hours, the MCHR1 antagonist S38151 produced a statistically significant dose-related inhibition of the increase in food intake produced by NEI-MCH. Nearly complete inhibition of NEI-MCH-induced food intake was observed at the dose of 50 nmol/kg S38151. Thereafter, for each dose of S38151, food intake increased essentially in parallel with the food increase in the NEI-MCH-treated group. Food intake during the entire 24-h period in the control group was not significantly altered by NEI-MCH, or by any combination of NEI-MCH and S38151.

#### **THE EFFECTS OF DAILY S38151 i.c.v. INJECTIONS ON FOOD INTAKE, BODY WEIGHT, AND MOTILITY IN ZUCKER fa/fa RATS**

The effects of daily S38151 i.c.v. injection on the measured parameters are depicted in **Figure 4**. During the 7-day experimental

period, injection of 28μg/kg (5 nmol/rat) S38151 1 h before the dark phase resulted in dose-dependent inhibition of food intake. However, the change in food intake only reached statistical significance at the dose of 20 nmol/kg (28μg/kg) S38151. Immediately after the injections were terminated, food intake returned to control levels for the duration of the washout period.

In contrast to food intake, water consumption appeared to be more sensitive to inhibition of MCHR1; water intake was significantly inhibited in response to all injected doses of S38151 except for the 5 nmol/rat dose. Immediately after the injections were terminated, water intake returned to control levels during the washout period. In a manner similar to water intake, motility also appeared to be more sensitive to the effects of MCH1 blockade; motility was significantly increased at both the 10 nmol/kg and 20 nmol/kg doses of S38151. Notably, the increase in motility was not terminated after the S38151 injections were stopped. This finding was most marked during the washout period after administration of the 20 nmol/kg dose of S38151, during which motility remained significantly elevated. During the 7-day experimental period, S38151 produced a dose-dependent reduction in body weight gain compared with control rats at the 10 and 20 nmol/kg doses. After the injections were stopped, body weight gain exhibited a tendency to return to control values. However, body weight gain remained significantly less than control values during the washout period after injection of 10 nmol/kg S38151.

**FIGURE 3 | The effect of S38151 on NEI-MCH-induced food intake in Wistar rats after acute central administration.** Satiated rats were injected (i.c.v.) at the beginning of the light phase with 5 μg NEI-MCH, 5 μg NEI-MCH + differing doses of S38151 (0.5–50 nmol/kg), or an equivalent volume of artificial cereblospinal fluid vehicle (5 μL). Cumulative food intake was measured at 1, 2, 4, 6, and 24 h post-injection. Results are expressed as mean ± SEM (n = 10–11 rats per group). Repeated measures Two-Way ANOVA (treatment × time) was followed by post hoc Student-Neuman-Keuls analysis to compare all groups of animals.

+++P *<* 0*.*001.

# **THE EFFECTS OF S38151 ON CONDITIONED TASTE AVERSION AND PICA**

(i.c.v.) once daily for 7 days, followed by a washout period of 5 days.

**IN ZUCKER fa/fa RATS** The results of the conditioned taste aversion test are shown

in **Figure 5A**. The consumption of 0.2% saccharine during the first exposure (test day 0) before treatment was well matched between the three groups. Subsequent injection of lithium chloride resulted in a significant reduction in 0.2% saccharine consumption during test days 4, 7, 11, and 14 compared with the control group. In contrast, S38151 injection did not affect 0.2% saccharine consumption.

In the Pica test, vehicle-injected control animals ate 27 ± 1 g of normal food and very little kaolin (0.05 ± 0.01 g) during the 24-h experimental period (**Figure 5B**). Cyclophosphamide (i.p. injection) resulted in the nearly complete suppression of spontaneous food intake, which was accompanied by a significant increase in kaolin intake. In contrast, i.c.v. injection of S38151 [20 nmol/kg (28 μg/kg)] significantly reduced food intake but did not significantly alter kaolin intake compared with control values.

# **THE EFFECT OF A SINGLE i.p. INJECTION OF S38151 ON FOOD INTAKE IN FASTED ZUCKER fa/fa RATS**

The acute effect of i.p.-injected S38151 on food intake in Zucker fa/fa rats is summarized in **Figure 6**. After a 24-h period of food restriction, rats ate 3.6 ± 0.4 g during the first hour after the food pellets were put back. S38151 (i.p. injection) at the dose of 30 mg/kg (approximately 60 nmol/rat) 1 h before refeeding induced a dramatic fall in food intake from the first hour (0*.*09 ± 0*.*06 g) to 6 h compared with the control group. This decrease maintained statistical significance up to 24 h after drug administration.

**FIGURE 5 | The effects of S38151 on conditioned taste aversion and pica in Zucker** *fa/fa* **rats. (A)** In the conditioned taste aversion test, after a period of habituation of the animals to the experimental protocol, Zucker fa/fa rats were presented with 0.2% saccharine for 20 min after drinking water was removed from their cages for 18 h (test day 0). Animals were then injected (i.p.) with vehicle or lithium chloride, or with S38151 (20 nmol/kg, i.c.v). On test days 4, 7, 11, and 14, rats were presented with 0.2% saccharine, and the quantity consumed during a 20-min period was determined. Results are expressed as mean ± SEM (n = 10–11 rats per group). Repeated measures

#### **THE EFFECTS OF DAILY i.c.v. INJECTION OF MCH ON FOOD INTAKE AND BODY WEIGHT IN ob/ob MICE** *Food intake*

Food intake in vehicle-infused C57BL/6J mice ranged from 3 to 4 g/day during the 5-day experimental period (**Figure 7A**). Infusion of MCH at 5 μg/mouse/day (100 nmol/kg/day) resulted in a sustained and significant increase in food intake above the values observed in vehicle-infused mice on day 5 (*-* = +1*.*3 g vs. controls). Food intake in ob/ob mice (**Figure 7C**) ranged from 3.5 to 5 g/day during the experimental period, and was significantly greater than that observed in C57BL/6J mice. Infusion of MCH at 5μg/mouse/day in ob/ob mice significantly increased food intake above the values obtained in vehicle-infused control animals. The increase in food intake produced by MCH infusion in ob/ob mice on day 5 was significantly greater than that observed in lean mice (*-*= +3*.*2 g vs. controls).

#### *Body weight*

In C57BL/6J mice, body weight remained stable during the 5-day experimental period (**Figure 7B**). Infusion of MCH resulted in a significant increase in body weight on day 5 (*-* = +1*.*3 g vs. controls). Body weight tended to decrease slightly during the experimental period in vehicle-infused ob/ob mice (**Figure 7D**).

++P *<* 0*.*01, +++P *<* 0*.*001. **(B)** In the pica test, Zucker fa/fa rats were injected (i.c.v.) with 20 nmol/kg S38151, or an equivalent volume of artificial cerebrospinal fluid vehicle. Food and kaolin intake during the following 24-h period were each measured. Results are expressed as mean ± SEM (n = 8 rats per group). One-Way ANOVA was followed by Dunnett's test, which compared the treated group with the control group. +P *<* 0*.*05, +++P *<* 0*.*001.

However, infusion of MCH reversed this trend, and body weight had increased significantly above control levels after 5 days of treatment (*-* = +2*.*1 g vs. controls). The increase in body weight in MCH-infused ob/ob mice on day 5 was significantly greater than that observed in either group of lean mice.

#### **THE EFFECTS OF DAILY S38151 i.p. INJECTION ON FOOD AND WATER INTAKE, BODY WEIGHT, AND MOTILITY IN ob/ob MICE**

The effects of daily i.p. injection of S38151 on the measured parameters are depicted in **Figure 8**. During the 6-day experimental period, i.p. injection of S38151 1 h before the dark phase produced a dose-dependent inhibition of food intake that reached significance starting on day 3 only at the highest dose of 30 mg/kg (20μmol/kg/day). Water intake presented the same profile, but did not reach statistical significance because of the greater variability in water consumption among mice. During the 6-day experimental period, S38151 produced a dose-dependent reduction in body weight compared with control mice, with the 30 mg/kg dose reaching statistical significance at days 1, 4, 5, and 6. In contrast, motility appeared to be similar among all groups. On day 6, an improvement of metabolic parameters (significant reductions in glycemia and insulinemia) was observed at a dose of 30 mg/kg (data not shown).

#### **THE EFFECTS OF DAILY i.p. INJECTION OF S38151 ON FOOD INTAKE, BODY WEIGHT, AND MOTILITY IN DIO MICE**

The effects of daily i.p. injection of S38151 on the measured parameters are depicted in **Figure 9**. During the 5-day experimental period, i.p. injection of S38151 1 h before the dark phase produced a dose-dependent inhibition of food and drink intake that reached significance beginning on day 2 only at the 30 mg/kg dose. A reduction in body weight gain was observed at the 10 and 30 mg/kg doses (7 and 20μmol/kg/day, respectively), which reached statistical significance at day 3 and day 1, respectively, compared with control mice. S38151 induced a similar significant reduction in motility at the two tested doses from day 3 to day 5. At day 6, an improvement of metabolic parameters (significant reductions in glycemia and insulinemia) was observed at a dose of 30 mg/kg (data not shown). In another experiment, daily i.p. administration of 30 mg/kg S38151 for 16 days induced a significant reduction of body weight gain that was accompanied by reduced glycemia and insulinemia, as well as a significant reduction of triglycerides, free fatty acids, and fat mass (data not shown).

#### **THE EFFECTS OF DAILY i.p. INJECTION OF S38151 ON FOOD INTAKE, BODY WEIGHT, AND MOTILITY IN MCH1-KO MICE**

During the 5-day experimental period, i.p. injection of S38151 1 h before the dark phase did not induce any significant effects on the food intake, drink intake, or body weight of MHCR1-KO mice; however, a significant reduction in motility was observed at the highest dose of 30 mg/kg (20 μmol/kg/day) (**Figure 10**).

# **DISCUSSION**

In the present study, we have investigated the effects of a selective MCH1 peptide antagonist, S38151 (Audinot et al., 2001a, 2009), on food intake, drinking, body weight, and/or motility in lean rats and mice, as well as in Zucker *fa/fa* rats, *ob/ob* mice, and DIO mice. Central acute administration of S38151 inhibited proMCH131-165 (MCH-NEI peptide)-induced feeding in lean animals in a dose-dependent manner. This effect lasted for over 6 h, in agreement with the estimated concentration of compound still present in plasma. In several obese rodent models (7-day treatment in Zucker *fa/fa* rats, 6-day treatment in *ob/ob* mice, and 5-day treatment in DIO mice), S38151 produced a robust and dose-dependent reduction in food intake and body weight gain. Drinking behavior was also inhibited in Zucker *fa/fa* rats and DIO mice. However, the cessation of S38151 injections resulted in a rapid resumption of feeding and drinking behaviors in Zucker *fa/fa* rats, a transitory action that greatly contrasted with the persistent weight loss. Repeated daily blockade of MCH1 receptors with S38151 reduced food intake and body weight with attenuated magnitude in leptin-deficient *ob/ob* mice relative to that observed in DIO mice. This observation was also reported with another MCH1 antagonist (Kowalski et al., 2006), presumably owing to the emergence of multiple alterations in leptin-responsive neuronal pathways implicated in the feeding behavior of *ob/ob* mice.

To clearly establish the involvement of MCH1 in the mode of action of S38151 on feeding and weight controls, its effects were evaluated in MCH1-KO mice. Consistent with direct MCH1 antagonism, S38151 did not produce any significant effect on feeding, drinking, or body weight gain in MCH1-KO mice treated for 5 days. Combined with the growing literature on MCH1 antagonists (see reviews Mendez-Andino and Wos, 2007; Johansson, 2011), these data strengthen the validation of MCH1 as a strong candidate for developing anti-obesity drugs.

Intriguingly, repeated daily MCH1 antagonism by S38151 induced persistent hypermobility in Zucker *fa/fa* rats, but did not affect locomotor activity in *ob/ob* mice, and even reduced motility in DIO mice. The differential response may result from differences in species (rat vs. mouse) or genetic background (*ob/ob* vs. C57BL/6J), or may be related to an off-target mechanism. When treated with S38151 for 5 days, MCH1-KO mice exhibited a reduction in motility similar to that noted in DIO mice. Therefore, the observed disparity in motor activity likely arose from off-target (i.e., MCH1-independent) effects of S38151.

In recent years, MCH1 antagonists have been described as central to the regulation of energy expenditure and food intake. However, unlike for neuropeptide Y receptor 5 antagonists [see Block et al. (2002) and references therein], data suggest that small molecule antagonists of MCH1 would have much broader activity, especially regarding mood regulation (and more generally, depression). Vast programs of drug synthesis and testing have been published in the literature. One of the surprising key features of those programs is that it appeared to be difficult to find

compounds with potent antagonists against MCH1 that lacked activity regarding the hERG channel [see Mendez-Andino and Wos for a complete review (Mendez-Andino and Wos, 2007)].

Regarding the onset of obesity, it is well-known that the adipocyte has a central role, so it seemed interesting to find a way to act directly upon this cell type. MCH1 is expressed in adipocytes, suggesting that fat cells may be targets of MCH antagonists under physiological conditions (Bradley et al., 2000). This hypothesis was confirmed by our study using DIO mice treated with daily S38151 injections for 16 days, in which a reduction in fat mass was observed concomitant with improvements in glucose and lipid profiles. Indeed, acting through the central control of appetite has often failed, particularly for two reasons: (1) the number of "rescue" systems in the brain circuits is so important that inhibiting one pathway would lead to the "takeover" of one or several alternate pathways (Nair and Ren, 2009); (2) in some cases acting on a central system, such as the MCH-ergic system, to control appetite has promoted adverse effects, and particularly pro-depressive effects (Roy et al., 2007; Lagos et al., 2011; Garcia-Fuster et al., 2012). Therefore, it was interesting to observe whether a potent (in the nM range) MCH1 antagonist that was theoretically incapable of penetrating the blood/brain barrier and fairly specific (as a derivative of the natural MCH1 agonist MCH) would have an effect *in vivo*, as it was already known to be active i.c.v. (Audinot et al., 2009). S38151 appears to fulfill the task. Indeed, despite a longer stability than unmodified peptides, its stability in blood remains too low to expect further reasonable activity in humans. However, it remains interesting to note that the blockade of the N-terminus of the peptide with a *para*guanidinobenzoyl moiety may enhance its stability in biological media.

Furthermore, we were unable to detect traces of this pseudopeptide in the brains of i.p.-injected animals. Similarly, its effect could not be correlated with aversion owing to a bad test. Finally, S38151 significantly reduced weight gain in rodent obesity models, and (most importantly) was inactive in our MCH1-KO mouse model, strongly suggesting not only the nature of its target but also its potency and selectivity.

The process through which S38151 induces weight loss remains poorly understood at the molecular level. We hypothesize that it acts locally at the adipocyte level. Indeed, in their paper on a small, non-peptide antagonist (SCH-A) of MCH1 activity, Kowalski et al. (2006) performed a careful and complete study to describe the mechanism of action of this compound at the adipocyte level. To our knowledge, few peptide MCH agonists or antagonists have been used *in vivo.* Indeed, as reviewed by

**FIGURE 8 | The effects of repeated daily S38151 injection on food and water intake, body weight, and motility in** *ob/ob* **mice.** Genetically modified ob/ob mice were administered S38151 (10 and 30 mg/kg, i.p.) or vehicle (normal saline i.p.) for 6 consecutive days. Results are expressed as mean ± SEM (n = 8 mice per group). White

**water intake, body weight, and motility in diet-induced obese (DIO) mice.** DIO mice were administered S38151 (10 and 30 mg/kg, i.p.) or vehicle (normal saline i.p.) for 5 consecutive days. Results are expressed as mean ± SEM (n = 8 mice per group). Repeated measures Two-Way ANOVA (treatment × time) was followed by complementary analysis of each treatment's effect at a fixed level of time to compare all groups of mice with the control group. \*P *<* 0*.*05, \*\*P *<* 0*.*01, \*\*\*P *<* 0*.*001.

Bednarek (2007), peptide ligands at MCH1 were used *in vivo* by Mashiko et al. (2005) and Shearman et al. (2003) to demonstrate the effects of peptide MCH1 antagonists on diet-induced obesity. Both compounds were administered through i.c.v. injection. To our knowledge, none of the reported agonists or antagonists was administered through i.p. injection. The validation of the agonistic nature of MCH, MCH derivatives, or peptides derived from them, has been reported on several occasions regarding feeding

#### **REFERENCES**


Audinot, V., Della Zuana, O., Fabry, N., Ouvry, C., Nosjean, O., Henlin, J. M., et al. (2009). S38151 [p-guanidinobenzoyl- [Des-Gly(10)]-MCH(7-17)] is a potent and selective antagonist at the MCH(1) receptor and has anti-feeding properties *in vivo*. *Peptides* 30, 1997–2007.


behavior, and has always involved administration through i.c.v. injection (Suply et al., 2001) or on their other effects (Chung et al., 2009; Lagos et al., 2009). All of these data have been reviewed by MacNeil and Bednarek (2009).

# **ACKNOWLEDGMENTS**

The authors would like to thank Marc Bertrand and his team (TES, Orléans, France) for their work on S38151 stability.

of neuropeptide EI (NEI). *Peptides* 29, 1441–1450.


and anorectic effects of a melaninconcentrating hormone-1 receptor antagonist. *Nat. Med.* 8, 825–830.


Della-Zuana et al. MCHr1 antagonist in vivo activity

diet-induced obesity. *Endocrinology* 143, 2469–2477.


nucleus evaluated in the forced swim test. *Behav. Brain Res.* 218, 259–266.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 20 September 2012; paper pending published: 17 October 2012; accepted: 26 November 2012; published online: 21 December 2012.*

*Citation: Della-Zuana O, Audinot V, Levenez V, Ktorza A, Presse F, Nahon J-L and Boutin JA (2012) Peripheral injections of melanin-concentrating hormone receptor 1 antagonist S38151 decrease food intake and body weight in rodent obesity models. Front. Endocrin. 3:160. doi: 10.3389/fendo.2012.00160*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Della-Zuana, Audinot, Levenez, Ktorza, Presse, Nahon and Boutin. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

REVIEW ARTICLE published: 05 March 2013 doi: 10.3389/fendo.2013.00020

# Physiological roles of GPR10 and PrRP signaling

# **Garron T. Dodd and Simon M. Luckman\***

Faculty of Life Sciences, AV Hill Building, University of Manchester, Manchester, UK

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Günter K. Stalla, Max-Planck-Institute of Psychiatry, Germany Tatsushi Onaka, Jichi Medical University, Japan

#### **\*Correspondence:**

Simon M. Luckman, Faculty of Life Sciences, AV Hill Building, The University of Manchester, Oxford Road, Manchester M13 9PT, UK. e-mail: simon.luckman@ manchester.ac.uk

#### **INTRODUCTION**

Seven-transmembrane-domain receptors (7TMRs) make up a receptor superfamily related by common signaling features and a structure that spans the cell membrane seven times. All 7TMRs are coupled to guanine nucleotide binding proteins (G-proteins) and, as such, are more commonly referred to as G-protein-coupled receptors (GPCRs; Probst et al., 1992). In the human genome, over 800 GPCRs have been annotated (>4% of the genome), many of which since have been implicated in diverse physiological roles from photoreception to olfaction, and from mood to appetite (Fredriksson et al., 2003). This diverse functionality infers immense therapeutic potential for the treatment of disease and, in fact, as many as half of the currently marketed drugs target GPCRs (Flower, 1999). Advances in genomics over the last century, that have allowed genome-wide homology analysis, have facilitated the discovery of so many new GPCRs. Currently the GenBank/EMBL database has over 1000 clones of eukaryotic GPCRs recorded, and many of the predicted receptors have no known ligand. These are termed "orphan" GPCRs. Although many of the GPCR genes probably correspond to homologs of sensory olfactory receptors, which are predicted to exist in considerable number in the genome, the remainder could encode for diverse unknown receptors, which may play important physiological roles

Prolactin-releasing peptide (PrRP) was first isolated from bovine hypothalamus, and was found to act as an endogenous ligand at the G-protein-coupled receptor 10 (GPR10 or hGR3). Although originally named as it can affect the secretion of prolactin from anterior pituitary cells, the potential functions for this peptide have been greatly expanded over the past decade. Anatomical, pharmacological, and physiological studies indicate that PrRP, signaling via the GPR10 receptor, may have a wide range of roles in neuroendocrinology; such as in energy homeostasis, stress responses, cardiovascular regulation, and circadian function. This review will provide the current knowledge of the PrRP and GPR10 signaling system, its putative functions, implications for therapy, and future perspectives.

**Keywords: PrRP, GPR10, energy intake, stress, dorsomedial hypothalamic nucleus, nucleus tractus solitarius, energy metabolism**

> (Buck and Axel,1991). Due to the undoubted therapeutic potential for the treatment of different pathologies, the discovery of ligands by the"de-orphanization"of GPCRs and an understanding of their physiological function is the focus of an intense research effort that has far reaching implications for both frontier and translational science.

> One of the first GPCRs to be de-orphanized was G-proteincoupled receptor 10 (GPR10; also known as hGR3 or UHR-1). GPR10 was originally cloned in hypothalamic tissue using low stringency PCR primers designed against to the highly conserved GPCR transmembrane domains 2 and 6 (Welch et al., 1995). The cloned receptor showed sequence similarity to the neuropeptide Y (NPY) receptor (31% overall and 46% in the transmembrane regions), however, it could not be activated by either NPY or pancreatic polypeptide (Marchese et al., 1995). This presented the scientific community with a novel problem, in that this represented the first GPCR for which its discovery preceded that of its endogenous ligand. Initial GPR10 localization studies indicated high mRNA expression in the anterior pituitary (Fujii et al., 1999). As hypothalamus derived factors frequently play important roles in regulating anterior pituitary function, it seemed intuitive that the natural ligand for GPR10 might exist in the hypothalamus. Using this insight, GPR10 was finally de-orphanized by Hinuma et al. (1998), using a novel reverse pharmacology approach. For reasons described below, the receptor ligand was termed prolactin-releasing peptide (PrRP). Later studies, using other *in vitro* heterologous expression systems, demonstrated that PrRP shows some promiscuous binding to another RFamide peptide family receptor, neuropeptide FF receptor 2 (NPFF-2R) (Engstrom et al., 2003; Ma et al., 2009). However, to date, PrRP is the only ligand known to have significant affinity for GPR10.

> Initial studies showed that PrRP could stimulate prolactin secretion from dispersed anterior pituitary cells; hence, the peptide's name (Hinuma et al., 1998). However since its discovery, the importance of PrRP in the physiological regulation of

**Abbreviations:** 7TMR, seven-transmembrane-domain receptors; ACTH, adrenocorticotropic hormone; AP, area postrema; BL, basolateral amygdaloid nucleus; BNST, bed nucleus of the stria terminalis; CCK, cholecystokinin-8; Ce, central amygdaloid nucleus; DMN, dorsomedial hypothalamic nucleus; GABA, γ-aminobutyric acid; GPCR, G-protein-coupled receptors; GPR10, G-protein-coupled receptor 10; LH, lateral hypothalamic area; MCPO, magnocellular preoptic nucleus; MD, mediodorsal thalamic nucleus; MPO, medial preoptic nucleus; ox, optic chiasm; NPFF-R2, neuropeptide FF receptor 2; NPY, neuropeptide Y; NTS, nucleus of the tractus solitarius; OLETF, Otsuka Long-Evans Tokushima Fatty; Pe, periventricular hypothalamic nucleus; PrRP, prolactin-releasing peptide; PT, paratenial thalamic nucleus; PVN, paraventricular hypothalamic nucleus; Rt, reticular nucleus of the thalamus; SM, nucleus of the stria medullaris; SO, supraoptic hypothalamic nucleus; SpVe, spinal vestibular nucleus; TH, tyrosine hydroxylase; VLH, ventrolateral hypothalamic nucleus; VLM, ventrolateral medulla.

prolactin secretion has been put in doubt (see below). Instead, the PrRP-GPR10 signaling pathway has been implicated in a range of other physiological systems. For example, central administration of PrRP inhibits food intake and increases energy expenditure in rats and mice (Lawrence et al., 2000, 2004), suggesting that PrRP plays roles in the regulation of energy balance. It also elevates circulating plasma levels of adrenocorticotropic hormone (ACTH) level, suggesting an association of PrRP with stress responses (Takayanagi and Onaka, 2010). Moreover, PrRP also can affect the cardiovascular system (Samson et al., 2000) and circadian cyclicity (Zhang et al., 2000, 2001; Lin et al., 2002a). This article aims to review the current understanding of the physiological roles for PrRP and GPR10 signaling in the mammalian system, and to highlight future directions for research.

#### **PrRP AND GPR10 EXPRESSION**

Determining the expression patterns of both receptor and ligand gives key insight into physiological function. *In situ* hybridization histology, RT-PCR, and immunohistochemical studies indicate that PrRP is expressed in neurons of the nucleus tractus solitarius (NTS), the ventrolateral medulla (VLM), and in the caudal portion of the dorsomedial hypothalamic nucleus (DMN) (**Figure 1**) (Chen et al., 1999; Maruyama et al., 1999; Ibata et al., 2000; Lee et al., 2000). PrRP mRNA has also been found in a number of peripheral tissues, including the adrenal gland, pancreas, placenta, and testis (Fujii et al., 1999; Matsumoto et al., 1999a; Kalliomaki et al., 2004).

The co-localization of PrRP with tyrosine hydroxylase (TH) in the caudal NTS and VLM, suggests that these PrRP cells are a subset of A2 and A1 noradrenergic neurons, respectively (Chen et al., 1999). The highest numbers of PrRP cell bodies are found within the NTS, and interestingly as the hypothalamus shows the highest levels of PrRP fiber immunoreactivity, this suggested the possible projection of PrRP from the brainstem to the hypothalamus (Hinuma et al., 1998; Fujii et al., 1999; Matsumoto et al., 1999a). PrRP-immunoreactive fibers are visible in many areas of the brain, such as the DMN, area postrema (AP), pontine parabrachial area, preoptic areas, bed nucleus of the stria terminalis (BNST), amygdala, mediodorsal nucleus of the thalamus, septal nucleus, and ependymal linings of the ventricles and blood vessels (Lin, 2008). One of the major projection sites is the paraventricular hypothalamus (PVN), where PrRP neurons appear to synapse directly on corticotrophin-releasing hormone (CRH) (Matsumoto et al., 1999a) and oxytocin neurons (Maruyama et al., 1999). Cell-specific connections also have been identified on magnocellular oxytocin/vasopressin neurons of the hypothalamic supraoptic nucleus (Maruyama et al., 1999), somatostatin neurons in the hypothalamic periventricular nucleus (Iijima et al., 2001), and on catecholaminergic cells of the adrenal medulla (Fujiwara et al., 2005).

Distribution of the GPR10 receptor has been investigated using autoradiography, *in situ* hybridization, and RT-PCR (Fujii et al., 1999; Roland et al., 1999; Ibata et al., 2000). The relative level of expression is high in the anterior pituitary, reticular nucleus of the thalamus (Rt), periventricular hypothalamus, DMN, AP, and NTS; with moderate expression in the BNST, PVN, medial

preoptic area and nucleus, ventrolateral hypothalamus, stomach, femur, and adrenal gland (Roland et al., 1999).

There is good complementarity in the localization of GPR10 receptor immunoreactive PrRP fiber staining in many brain areas (BNST, supraoptic nucleus, PVN, DMN, and NTS). However, it is interesting to note discrepancies in localization, which might be surprising if GPR10 is the only receptor for PrRP. In fact, many peptide systems have significant mismatches between the distribution of the ligand and their respective cognate receptors. Much of this mismatch might be explained by redundancy in function, that is a receptor will not respond if it is not in contact with the ligand. It may be energetically convenient not to lose the expression of a receptor if there is no evolutionary pressure to do so. Furthermore, peptides often have permissive actions and may not function as classical transmitters at tightly regulated synaptic junctions. For instance, PrRP may be released from neuronal fibers terminating at the ventricular zones, and may enter and diffuse within the cerebral spinal fluid (Iijima et al., 1999); or as seen with substance P, PrRP may diffuse through the neuronal tissue to reach distant receptor sites (Duggan et al., 1990). Although GPR10 is considered to be the cognate receptor for PrRP, others (perhaps currently unknown) may exist. For example, PrRP has significant affinity at neuropeptide FF receptor 2 (NPFF-R2) in *in vitro* studies, and there is potential for overlap between the presence of PrRP and NPFF-R2 particular in the hypothalamus and adrenal gland (Gouarderes et al., 2004). Nevertheless, the diverse distribution profile of receptors and ligand may underlie the diverse physiological roles played by PrRP-GPR10 signaling, and each function needs careful investigation. In the absence of receptor-selective antagonists, this is probably best achieved in receptor knockout mice.

#### **ROLE OF PrRP IN PROLACTIN SECRETION**

As high expression of GPR10 is seen in the anterior pituitary, initial studies investigating the physiological action of PrRP focused on hypophysiotropic secretion (Hinuma et al., 1998; Lin et al., 2002b). Preliminary *in vitro* studies, which gave rise to the name of the peptide, described an action of PrRP on prolactin secretion from anterior pituitary tumor cell lines and primary cell cultures (Hinuma et al., 1998). Subsequent studies investigating the relevance of PrRP *in vivo* as a central mediator of prolactin release were controversial,with positive results being reliant on high intravenous PrRP doses administered during specific phases of female rat estrous cycle (Matsumoto et al., 1999b). Other studies demonstrated no prolactin release following central administration of PrRP (Matsumoto et al., 2000; Seal et al., 2002). Moreover, as no PrRP immunoreactivity is found in the median eminence or in hypophysiotropic cells of the hypothalamus (Matsumoto et al., 1999a;Maruyama et al., 2001), classically associated with the secretion of pituitary hormones, the question remains how does PrRP access the pituitary? PrRP may act upon the anterior pituitary as a hormone secreted from peripheral tissues (adrenal, pancreas, testis, placenta), or by an indirect central mechanism possibly via hypophysiotropic neurons (Morales and Sawchenko, 2003). This is a strong possibility, since central administration of PrRP can affect a number of anterior pituitary hormones (Seal et al., 2002). Interestingly, in fish and amphibians, PrRP fibers project to and

**FIGURE 1 | Schematic drawings showing the neuronal distribution of PrRP and GPR10 receptor in the Paxions andWatson rat brain atlas (Paxinos andWatson, 1998; Sun et al., 2005).** Blue areas represent PrRP-immunopositive nerve fibers; black checkers represent PrRP cells bodies; green areas represent GPR10 expression; and red areas represent overlap of PrRP and GPR10 expression. AP, area postrema; BL, basolateral amygdaloid nucleus; BNST, bed nucleus of the stria terminalis; Ce, central amygdaloid nucleus; DMN, dorsomedial

hypothalamic nucleus; LH, lateral hypothalamic area; MCPO, magnocellular preoptic nucleus; MD, mediodorsal thalamic nucleus; MPO, medial preoptic nucleus; ox, optic chiasm; PVN, paraventricular hypothalamic nucleus; Pe, periventricular hypothalamic nucleus; PT, paratenial thalamic nucleus; Rt, reticular thalamic nucleus; SM, nucleus of the stria medullaris; SO, supraoptic hypothalamic nucleus; NTS, nucleus of the tractus solitarius; SpVe, spinal vestibular nucleus; VLH, ventrolateral hypothalamic nucleus.

terminate on prolactin-producing cells of the pituitary and systemic injection of PrRP into rainbow trout causes a release in prolactin and somatolactin (Moriyama et al., 2002; Seale et al., 2002; Sakamoto et al., 2006). PrRP may, therefore, represent an ancient factor for the direct regulation of prolactin secretion that is now evolutionary redundant in this function in higher mammals. Thus, the name, PrRP, may represent a misnomer, as research over the past decade has implicated this signaling pathway in alternative physiological systems.

# **CONSERVED FUNCTION OF PrRP AND GPR10 SIGNALING IN FEEDING BEHAVIOR**

Prolactin-releasing peptide belongs to the RFamide neuropeptide family (Osugi et al., 2006). Although this family impacts on a diverse range of physiological functions, almost all have been shown to modulate food intake (Bechtold and Luckman, 2007). This involvement of the RFamides in feeding behavior has been demonstrated across most animal taxa, including coelenterates, mollusks, amphibians, birds, and mammals, suggesting an evolutionary conserved role in energy homeostasis (Dockray, 2004).

Numerous studies suggest a pivotal role of PrRP in the homeostatic regulation of feeding and energy balance. Evidence from our group has shown that central administration of PrRP decreases feeding and body weight gain in rats and mice without causing adverse effects (Lawrence et al., 2000, 2002; Bechtold and Luckman, 2006), and that PrRP mRNA in the DMN, NTS, and VLM is downregulated in states of negative energy balance (Lawrence et al., 2000). Importantly, these central anorexic actions of PrRP are not present in mice (Bechtold and Luckman, 2006) or rats (Watanabe et al., 2005) that lack functional expression of GPR10, highlighting the significance of endogenous PrRP-GPR10 signaling in food intake. The significance of this system to energy homeostasis generally is validated further by the obese and hyperphagic phenotypes of both PrRP−/<sup>−</sup> and GPR10−/−null mice (Gu et al., 2004; Takayanagi et al., 2008).

As PrRP induces hypophagia without evoking a conditioned taste aversion or disrupting the normal behavioral satiety sequence (Lawrence et al., 2002), it seems likely that PrRP-GPR10 signaling plays an integral part of the brain's endogenous appetitive neurochemistry. In fact, PrRP induces a significant temporal advancement in the behavioral satiety sequence, an affect associated with natural satiety factors like cholecystokinin-8 (CCK) (Lawrence et al., 2002). Furthermore, experiments with PrRP−/−mice or PrRP-neutralizing antibodies, in the laboratory of Tatsushi Onaka, show that PrRP regulates meal size rather than meal frequency, indicating that PrRP may mediate appetite by direct actions on satiation (Takayanagi et al., 2008). As the brainstem medulla oblongata and, in particular, the NTS receives extensive gastrointestinal vagal inputs, these PrRP neurons are an obvious candidate for a role in gut-brain signaling. CCK is released from enteroendocrine cells in response to a meal, and acts via the CCK<sup>1</sup> receptor on vagal afferent neurons which terminate in the NTS (Saper, 2004). PrRP neurons localized in both the NTS and the VLM show strong functional activation in response to anorexic doses of CCK (Lawrence et al., 2002). Central administration of PrRP elicits a similar pattern of neuronal c-Fos protein expression as that

observed following intraperitoneal administration of CCK (Luckman, 1992; Lawrence et al., 2002; Bechtold and Luckman, 2006), and the anorexic effects of CCK are impaired in both PrRP <sup>−</sup>/<sup>−</sup> and GPR10−/−null mice (Bechtold and Luckman, 2006; Takayanagi et al., 2008).

The downstream actions of PrRP neurons within the brainstem remain to be clarified. However, PrRP receptor is present in the dorsal vagal complex (Roland et al., 1999; Ibata et al., 2000) and PrRP can act pre-synaptically to affect the firing of preganglionic vagal efferents involved in regulating gut function (Morales and Sawchenko, 2003). Thus, although not proven, it is likely that PrRP-GPR10 signaling within the dorsal vagal complex may mediate the effects of CCK on the parasympathetic regulation of gut motility and secretion. The sensation of satiety, and integration with descending motor pathways to regulate feeding, requires integration with higher brain centers. PrRP-immunoreactive fibers and GPR10 mRNA expression have been demonstrated in a number of hypothalamic nuclei (Fujii et al., 1999; Maruyama et al., 1999; Roland et al., 1999; Ibata et al., 2000; Lee et al., 2000). In particular, PrRP-containing neurons in the NTS project directly to the PVN (Onaka, 2004), where neurons containing CRH or oxytocin possess PrRP receptor (Lin et al., 2002a; Takayanagi and Onaka, 2010). Though anorexic doses of PrRP activate neurons expressing CRH or oxytocin in the PVN (Bechtold and Luckman, 2006; Mera et al., 2006), it is difficult to relate this specifically to satiety signaling, as these neurons may equally be involved in responses to stress (see below). However, PrRP-induced anorexia is attenuated by CRH receptor antagonists (Bechtold and Luckman, 2006), while oxytocin receptor antagonists attenuate the anorexic actions of both PrRP and CCK (Olson et al., 1991; Blevins et al., 2003). Further work will be required to dissect the relative importance of ascending PrRP pathways on satiety and stress-related stimuli. Additional consideration for the role of DMN PrRP neurons in the regulation of feeding behavior is needed also. However, our working model is that PrRP-GPR10 signaling mediates the CCKvagal regulation of gut function following a meal at the level of the dorsal vagal complex in the brainstem. Further integration with higher brain centers is achieved through the projection of PrRP neurons to the hypothalamus. This may include CRH and oxytocin neurons of the PVN, the latter, at least, having an accepted role in the descending fine regulation of the dorsal vagal complex and feeding control (Samson et al., 2000; Yamada et al., 2009; Onaka et al., 2010).

# **ENERGY HOMEOSTASIS**

Though the evidence for PrRP-GPR10 functioning in satiation is strong, this does not infer a role in overall energy balance, and an interaction with other metabolic regulators might be expected. We have shown that the expression of PrRP is down regulated in situation where the animal is in real (e.g., fasting or lactation) or in perceived (e.g., Zucker rat) negative energy balance (Ellacott et al., 2002). That is, situations which correlate with reduced leptin signaling. Leptin is an adipose-derived hormone, that signals levels of peripheral fat storage to the brain to regulate long-term metabolism (Denver et al., 2011). Immunohistochemical studies have suggested that PrRP neurons (and TH-positive cells) in the brainstem and hypothalamus of the rat express leptin receptors and, thus, that there is a direct cellular effect of the hormone (Hay-Schmidt et al., 2001; Ellacott et al., 2002). However, a more recent paper failed to co-localize leptin receptor in brainstem PrRP neurons of the mouse (Garfield et al., 2012). Leptin induces the expression of phosphorylated signal transducer and activator of transcription protein 3 (pSTAT3) in PrRP neurons, especially those in the DMH (Takayanagi et al.,2008). Central co-administration of PrRP and leptin results in augmented hypophagia and body weight loss (Ellacott et al., 2002), and the hypophagic effects of leptin are impaired in PrRP−/<sup>−</sup> (Takayanagi et al., 2008) and GPR10−/<sup>−</sup> null mice (our unpublished results). PrRP-GPR10 clearly has a role in the response to leptin, but whether this is due to a direct or indirect effect of leptin on PrRP neurons remains to be determined.

The maintenance of energy homeostasis involves the balance of both energy intake and energy expenditure. Interestingly pairfeeding studies indicate that the reduced weight gain measured in rats treated with PrRP is not accounted for solely by a reduction in food intake, suggesting that PrRP also affects energy expenditure (Lawrence et al., 2000, 2004). PrRP administration acutely increases body temperature, O<sup>2</sup> consumption, and UCP-1 expression of brown adipose tissue in rats (long before any effect on body weight), suggesting direct modulation by PrRP of energy expenditure (Lawrence et al., 2004). Furthermore, GPR10−/<sup>−</sup> knockout mice exhibit a much lower basal metabolic rate, when compared with wild-type mice (our unpublished data), which likely contributes to the obese phenotype of these animals (Gu et al., 2004). Thus, PrRP-GPR10 signaling can induce energy expenditure and thermogenesis, which is interesting considering the known role of the DMN in thermoregulation (Willette et al., 1984; Aicher et al., 1995; Horiuchi et al., 2002). In addition, PrRP may play a role in mediating energy consumption under stressful conditions, as the increase oxygen consumption seen in response to stressful stimuli is attenuated in PrRP−/<sup>−</sup> mice (Onaka et al., 2010).

# **ROLES OF PrRP AND GPR10 SIGNALING IN THE CONTROL OF STRESS RESPONSES**

Brain nuclei expressing PrRP and GPR10, such as in the medulla oblongata and the hypothalamus, have been implicated in mediating stress responses (Onaka, 2004). PrRP neurons within these regions respond to a variety of stressful stimuli including body restraint, fear conditioning (Zhu and Onaka, 2003), footstock, hemorrhage (Uchida et al., 2010), and inflammatory stress (Mera et al., 2006). PrRP neurons may, therefore, play an important role in the neuroendocrine response to stress.

Retro-grade tracing of the PrRP neurons innervating the PVN indicates that the fibers originate within the VLM and NTS, where they co-localizes with noradrenaline in the A1 and A2 neuronal populations, respectively (Chen et al., 1999; Minami et al., 1999; Roland et al., 1999; Morales et al., 2000; Maruyama et al., 2001). These noradrenergic neurons are well known mediators of stress in the central nervous system. Models of emotional stress, including conditioned fear stimulation and water immersion/restraint activate medullary PrRP neurons and increases PrRP mRNA expression (Maruyama et al., 2001; Morales and Sawchenko, 2003; Zhu and Onaka, 2003). Interestingly, PrRP and noradrenaline, which co-localize in A1/A2 cells, act synergistically to induce systemic ACTH release (Maruyama et al., 2001).

One way in which PrRP may influence stress response is by the dense network of PrRP clustered on CRH and oxytocin neurons in the PVN and BNST (Iijima et al.,1999;Maruyama et al.,1999;Ibata et al.,2000). Central administration of PrRP dramatically increases c-Fos expression in CRH neurons in the PVN, an effect that results in the concomitant release of ACTH, oxytocin, and corticosterone into the systemic circulation (Matsumoto et al., 2000; Seal et al., 2002). Importantly, blockade of endogenous PrRP signaling by administration of PrRP neutralizing antibodies attenuates stress induced activation of PVN neurons and reduces systemic oxytocin release (Zhu and Onaka, 2003; Mera et al., 2006).

Although contacts are seen between PrRP fibers and CRH neurons in the PVN (Matsumoto et al., 2000), their relative paucity suggests that these synapses are unlikely to be responsible for the entire modulation of CRH neurons in the PVN. Double *in situ* hybridization shows that the majority of cells expressing GPR10 in the PVN are in fact CRH-negative, whereas GPR10 is co-expressed extensively with CRH in the BNST (Lin et al., 2002b). The BNST not only receives extensive PrRP nerve fibers (Maruyama et al., 1999), it is involved with stress responses via a direct modulation of the PVN (Palkovits et al., 1980; Lin et al., 2002b). It, therefore, seems possible that PrRP may also regulate CRH neurons in the PVN indirectly via the BNST. These results suggest that whether directly or indirectly, PrRP is a potent stimulator of CRH neurons in the PVN, inferring access to the hypothalamic–pituitary–adrenal axial control of stress.

Stressful stimuli affect food intake and energy expenditure, while food intake and energy expenditure affect stress responses (Kawakami et al., 2008). For instance PrRP−/<sup>−</sup> mice show a reduced increase in oxygen consumption following stressful stimuli (Onaka et al., 2010). It seems tempting to suggest that PrRP may modulate food intake in times of stress. Although only speculative further examination of this hypothesis using conditional transgenic mice for PrRP and GPR10 could help shed light on this theory.

#### **EFFECTS OF PrRP AND GPR10 ON BLOOD PRESSURE**

Central injection of PrRP results in a significant increase in blood pressure and cardiovascular output in conscious, unrestrained rats (Samson et al., 2000). Numerous studies describe integral roles played by the NTS, AP, and VLM in mediating cardiovascular function (Yamada et al., 2009). It seems likely that PrRP neurons in these regions may be involved. The NTS and AP receive visceral and hormonal information from peripheral cardiovascular sites (Aicher et al., 1995), so PrRP and GPR10 in these sites may be in a position to modify the ascending and descending efferent connections mediating blood pressure homeostasis (Willette et al., 1984; Aicher et al., 1995). Site specific administration of PrRP directly into the caudal VLM (where a population of PrRP neurons are localized) results in a dose-dependent increase in mean arterial blood pressure, heart rate, and renal sympathetic activity (Horiuchi et al., 2002). Interestingly however, PrRP has no effect when injected directly into the rostral VLM, AP, or the NTS. How PrRP modulates blood pressure homeostasis in an area of minimal GPR10 receptor expression such as the VLM and not in regions of high receptor expression such as the AP and NTS remains enigmatic (Chen et al., 1999; Roland et al., 1999).

Although the mechanisms underlying the pressor effects of PrRP are undefined, a recent study by Yamada et al. (2009) suggests the involvement of CRH neurons in the PVN. PrRP neurons from the VLM project to the PVN where they synapse on CRH positive cells. Central CRH has a known effect of elevating blood pressure in response to stressors (i.e., CRH stimulates sympathetic nerves via the CRH<sup>1</sup> receptor) (Vale et al., 1983; Spina et al., 2000). Yamada et al. (2009) show that pressor- and tachycardiainducing doses of PrRP activate oxytocin-, vasopressin-, and CRHproducing neurons in the PVN. Furthermore, the elevation of blood pressure and heart rate elicited by PrRP administration are completely suppressed by treatment with a CRH antagonist. PrRP neurons in the VLM may, therefore, mediate CRH release to regulate the cardiovascular system via the sympathetic nervous system.

Finally, the receptor mediating PrRP pressor and tachycardia effects remains unclear. Epidemiological human studies show an association of polymorphisms in the GPR10 receptor with blood pressure, thus implying a potential role of the GPR10 receptor in blood pressure regulation (Bhattacharyya et al., 2003). Contrastingly, PrRP can still elicit effects on mean arterial blood pressure and heart rate in Otsuka Long-Evans Tokushima Fatty (OLETF) rat strain, in which the GRP10 receptor gene is naturally mutated (Ma et al., 2009). Instead, PrRP effects were blocked by administration of the NPFF-2R antagonist, RF9, suggesting that PrRP may modulate blood pressure homeostasis via the NPFF-2R (Ma et al., 2009). It will be useful to follow up these studies using other models since neither the OLETF rat (which has at least one other natural mutation, in the CCK<sup>1</sup> receptor gene), nor the R9 antagonist, are the best tools available. Certainly, if unwanted cardiovascular effects of PrRP are mediated solely by the NPFF-R2, there could be therapeutic potential for selective GPR10 agonists as drug targets other metabolic diseases.

## **EFFECTS OF PrRP AND GPR10 CIRCADIAN RHYTHMICITY AND SLEEP REGULATION**

The expression of GPR10 in particular brain regions, including the preoptic area, the histaminergic ventral tuberomammillary nucleus, the noradrenergic locus ceruleus, serotonergic dorsal raphe, and suprachiasmatic nucleus suggested that PrRP-GPR10 signaling may play a part in circadian rhythmicity and/or sleep regulation (Chen et al., 1999; Roland et al., 1999). The relative importance of PrRP-GPR10 signaling in each of these specific nuclei is yet to be investigated, however, a wealth of literature exists implicating an integral role in sleep and arousal (for reviews, see Suntsova et al., 2009; Szymusiak, 2010; Brown et al., 2012; Murillo-Rodriguez et al., 2012). One region of particular interest, which has a high GPR10 receptor expression, is the Rt (Roland et al., 1999). The Rt is predominantly GABAergic and acts as a gateway for ascending inputs into the cortex that regulate the transition into sleep (Steriade, 2005; Timofeev and Chauvette, 2011). Central administration of PrRP is known to modulate sleep oscillation, and promote rapid and prolonged arousal (Zhang et al., 2000; Lin et al., 2002a). Furthermore, electrophysiological experiments on brain slices show that administration of PrRP attenuates oscillatory activity generated in the Rt, a phenomena that could underlie PrRP's modulation of circadian and sleep regulation (Lin et al., 2002a).

# **FUTURE PERSPECTIVES**

Since the de-orphanization of GPR10, research into the physiological roles of PrRP neurotransmission has been varied and exciting. The PrRP peptide is conserved among species (fish, amphibians, birds, and mammals), pointing toward it seems both primitive and important function (Dockray, 2004; Bechtold and Luckman, 2007). Research into the physiological roles of PrRP has evolved from the initial observations as a "PrRP" (perhaps resulting in a misnomer) to a multifunctional protein integral to a number of functions. Given the current understanding of the PrRP-GPR10, it seems likely that this ancient signaling system may act in times of stress to regulate feeding behavior, induce energy expenditure and increased cardiac output, heighten arousal, and allow the systemic release of endocrine factors. It is possible that, in mammalian species, some of these functions have been modified more specifically, for example into a role in satiation and energy regulation.

To further examine the importance of PrRP-GPR10 signaling a number of outstanding questions need to be addressed.

#### **WHAT ARE THE RELATIVE IMPORTANCE OF THE DIFFERENT POPULATIONS OF PrRP-PRODUCING CELLS IN THE BRAINSTEM, HYPOTHALAMUS, AND IN PERIPHERAL TISSUES?**

This review has highlighted potential differential roles for the PrRP-expressing neuronal populations. Although currently only speculative, it seems that the VLM may play a specific role in the pressor effects of PrRP (Horiuchi et al., 2002); whereas, the NTS appears important in mediating CCK's effect on satiation (Bechtold and Luckman, 2006). Interestingly, a recent study has shown that systemic CCK acts to attenuate liver gluconeogenesis independently of insulin production (Cheung et al., 2009). Importantly, the effect requires the integration of a gut–brain–liver axis; effects that could be centrally mediated by CCK responsive PrRP neurons in the NTS.

Further speculation arises over the function of the DMN PrRP population. Takayanagi and Onaka (2010) show significant pSTAT3 co-expression following leptin administration in DMN, suggesting that DMN PrRP neurons are responsive to leptin. Although there is little doubt of the importance of leptin receptor in energy homeostasis, recent research has specifically identified leptin responsive neurons in the DMN as mediators of adaptive thermogenesis (Enriori et al., 2011; Bechtold et al., 2012). As adaptive thermogenesis and the neuronal circuitry innervating brown adipose tissue is currently topical in the domain of anti-obesity therapeutics, it is important to investigate whether the PrRP neurons play a part. Also, both PrRP and GPR10 are expressed in peripheral tissues (Roland et al., 1999). Nothing is known about the importance of peripheral PrRP-GPR10 signaling, and the exploration of these interactions could be vital to the development of viable GPR10 therapeutics.

The recent advancements in the generation of conditional transgenic mice, makes finding the answers to these questions possible. For instance the generation of mice conditionally expressing PrRP under the control of different promoters will allow the genetic dissection of specific populations.

#### **WHAT ARE THE DOWNSTREAM AND UPSTREAM TARGETS OF PrRP – GPR10 SIGNALING?**

Understanding the neurochemical make up of PrRP target neurons will help to further define and dissect out the relative importance of each neuronal population. Recent advancements in anteroand retro-grade labeling using Cre recombinase specific adenoviral vectors could advance our understanding of how specific PrRP neuronal population integrate into both the local and global neuronal circuits (Gautron et al., 2010).

There is as a potential caveat with much of the work already achieved in understanding PrRP-GPR10 signaling. As mentioned above, there is some divergence, at least in the mammalian

#### **REFERENCES**


of the rat medulla. *Brain Res.* 822, 276–279.


system, between PrRP-expressing nerve fibers and the location of GPR10 receptors. Thus, an element of deliberation must be taken to understanding studies whereby PrRP is administered globally into the brain and at non-physiological doses, as this could be activating redundant receptors. Although this problem may be minimal, the use of conditional transgenic animals, conditional viral vectors, and the development of selective GPR10 agonists/antagonists will greatly enhance our understanding of this important neurotransmitter system. Although much of the ground work has been established, there is still much to learn about PrRP-GPR10 signaling and its definitive roles in the nervous system.

receptors in the human genome form five main families. Phylogenetic analysis, paralogon groups, and fingerprints. *Mol. Pharmacol.* 63, 1256–1272.


treatments in rats: evidence for central oxytocin inhibition of food


is responsible for the hyperphagia/dyslipidaemia/obesity locus of Dmo1 in the OLETF rat. *Clin. Exp. Pharmacol. Physiol.* 32, 355–366.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 11 January 2013; accepted: 19 February 2013; published online: 05 March 2013.*

*Citation: Dodd GT and Luckman SM (2013) Physiological roles of GPR10 and PrRP signaling. Front. Endocrinol. 4:20. doi: 10.3389/fendo.2013.00020*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Dodd and Luckman. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Feeding and the rhodopsin family G-protein coupled receptors in nematodes and arthropods

# **João C.R. Cardoso\*, Rute C. Félix,Vera G. Fonseca and Deborah M. Power**

Molecular Comparative Endocrinology, Centre of Marine Sciences, Universidade do Algarve, Faro, Portugal

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Pei-San Tsai, University of Colorado, USA Liliane Schoofs, Catholic University of Leuven, Belgium Lindy Holden-Dye, University of Southampton, UK

#### **\*Correspondence:**

João C.R. Cardoso, Molecular Comparative Endocrinology, Centre of Marine Sciences, Universidade do Algarve, Campus de Gambelas, Faro 8005-139, Portugal. e-mail: jccardo@ualg.pt

In vertebrates, receptors of the rhodopsin G-protein coupled superfamily (GPCRs) play an important role in the regulation of feeding and energy homeostasis and are activated by peptide hormones produced in the brain-gut axis.These peptides regulate appetite and energy expenditure by promoting or inhibiting food intake. Sequence and function homologs of human GPCRs involved in feeding exist in the nematode roundworm, Caenorhabditis elegans (C. elegans), and the arthropod fruit fly, Drosophila melanogaster (D. melanogaster), suggesting that the mechanisms that regulate food intake emerged early and have been conserved during metazoan radiation. Nematodes and arthropods are the most diverse and successful animal phyla on Earth. They can survive in a vast diversity of environments and have acquired distinct life styles and feeding strategies.The aim of the present review is to investigate if this diversity has affected the evolution of invertebrate GPCRs. Homologs of the C. elegans and D. melanogaster rhodopsin receptors were characterized in the genome of other nematodes and arthropods and receptor evolution compared. With the exception of bombesin receptors (BBR) that are absent from nematodes, a similar gene complement was found. In arthropods, rhodopsin GPCR evolution is characterized by species-specific gene duplications and deletions and in nematodes by gene expansions in species with a free-living stage and gene deletions in representatives of obligate parasitic taxa. Based upon variation in GPCR gene number and potentially divergent functions within phyla we hypothesize that life style and feeding diversity practiced by nematodes and arthropods was one factor that contributed to rhodopsin GPCR gene evolution. Understanding how the regulation of food intake has evolved in invertebrates will contribute to the development of novel drugs to control nematodes and arthropods and the pests and diseases that use them as vectors.

**Keywords: rhodopsin GPCR, feeding, conservation, evolution, invertebrates**

#### **INTRODUCTION**

Feeding is the process by which food is obtained to provide energy. It must satisfy growth, survival, and reproductive requirements and has driven the evolution of specialized feeding behaviors and apparatus in metazoan. Regulation of feeding is a complex mechanism, which involves a combination of physical, chemical, and nutritional factors (Neary et al., 2004; Coll et al., 2007; Woods et al., 2008). Food-taking behavior is dependent on environmental signals (odors and taste), hunger signals (metabolic signals), and also endocrine satiety signals that via the blood stream or the vagal afferent terminals act on the hypothalamus, brain stem, or afferent autonomic nerves to modulate feeding response (**Figure 1**; Konturek et al., 2004; Stanley et al., 2005; Chaudhri et al., 2006; Woods et al., 2006, 2008). In mammals, psychological factors such as mood (emotions) and food reward have also been shown to affect eating behavior (Christensen, 1993; Berridge, 1996). In vertebrates, a group of small regulatory peptides that are produced by the brain-gut axis play a major role in the endocrine regulation of feeding and control of energy homeostasis (**Figure 1**; Coll et al., 2007; Chaudhri et al., 2008). These peptide hormones are divided into two groups, those that stimulate appetite (orexigenic

peptides) and induce food intake and those that cause loss of appetite (anorexigenic peptides) and reduce food consumption and increase energy expenditure (Ahima and Osei, 2001; Wilding, 2002; Suzuki et al., 2010). The action of such peptides involves the activation of specific G-protein coupled receptors (GPCRs), which undergo conformational changes and promote the activation of intracellular signaling mechanisms that ultimately lead to a cellular response (**Table 1**; Marinissen and Gutkind, 2001; Xu et al., 2004; Fredriksson and Schioth, 2005).

The involvement of GPCRs in the regulation of vertebrate feeding and appetite is well recognized (Shioda et al., 2008). Much less is known about their homologs and cognate activating peptides in non-vertebrates. However, comparative sequence approaches and functional studies suggest that the involvement of GPCRs in metazoa feeding behavior emerged early and has been maintained during the species radiation (Brody and Cravchik, 2000; Hewes and Taghert, 2001; Fredriksson and Schioth, 2005; Teng et al., 2008). GPCRs have emerged via gene or genome duplication events followed by selection of the gene duplicates. Understanding the origin of GPCRs represents a valuable tool for the characterization of basic physiological functions that have been

maintained during evolution. The present review takes a comparative approach and targets rhodopsin GPCR subfamily members in the model species, *C. elegans* (a nematode) and *D. melanogaster* (a arthropod) that are sequence and function homologs of vertebrate GPCRs implicated in feeding regulation. To enrich the data and provide insight into how divergent life style and feeding strategies may have shaped receptor evolution in invertebrates the sequence of the target GPCRs were identified in other nematodes and arthropods with available genome data.

# **THE VERTEBRATE GPCRs SUPERFAMILY AND THEIR ROLE IN FEEDING**

G-protein coupled receptors are one of the largest groups of receptors present in cells. Based upon their structure and sequence similarity five distinct superfamilies have been defined in human: glutamate (G), Rhodopsin (R), Adhesion (A), Frizzled (F), and Secretin (S) and are collectively known as GRAFS (Fredriksson and Schioth, 2005; **Figure 2**). GPCRs are characterized by a signature motif of seven conserved transmembrane spanning helix domains (TM) in vertebrates and non-vertebrates. Receptor activation is mediated by the extracellular N-terminal domain and also by TM and extracellular loops (receptor core domain) that interact with diverse types of molecules. The cellular response is provoked by the receptor C-terminal domain which activates a series of intracellular signaling cascades via the G-protein coupled pathway complex (Bockaert and Pin, 1999; Marinissen and Gutkind, 2001). Other molecular mechanisms such assembly of receptor heterodimers and allosteric receptor–receptor interactions in the cell membrane are also involved in GPCR regulation, activation and signaling (Prinster et al., 2005; Langmead and Christopoulos, 2006; Fuxe et al., 2012).



Receptor subfamily members, activating peptides and their effect on feed (stimulation or reduction) are indicated. For references please consult the text.

G-protein coupled receptors are ubiquitous and involved in many different physiological functions. The glutamate receptors are involved in synaptic plasticity and participate in numerous functions in the central nervous system (CNS; Niswender and Conn, 2010). Rhodopsin receptors include receptors for hormones, neurotransmitters and photons and they are involved in taste, smell, and also regulate metabolism, reproduction, and neural function (Simoni et al., 1997; Murdoch and Finn, 2000; Gaillard et al., 2004; Waldhoer et al., 2004). Adhesion receptors participate in cell adhesion, signaling, and immune function (Bjarnadottir et al., 2007; Yona et al., 2008). Frizzled receptors are involved in the Wnt signaling pathway and in the control of cell proliferation and embryogenesis (van Amerongen and Nusse, 2009; Schulte, 2010). In contrast to other GPCRs, secretin family members are only activated by peptide hormones and they are implicated in brain-gut functions, calcium homeostasis, and in the stress response (McDermott and Kidd, 1987; Harmar, 2001; Bale and Vale, 2004; Moody et al., 2011). Due to their conserved structure and presence in many phyla, GPCRs are suggested to have a common evolutionary origin and to have arisen via gene/genome duplication early in the species radiation (Krishnan et al., 2012). With the exception of the glutamate family members, they are proposed to share a common ancestor with the cAMP receptors of primitive eukaryote species (Nordstrom et al., 2011).

In humans,more than 700 GPCR genes are predicted and a large proportion are orphans with unknown function (**Figure 2**). The rhodopsin family (a.k.a family A or class 1 GPCRs) comprise the most diverse receptor group and in humans they account for more than 80% of GPCRs and include members that are involved in regulation of feeding (Joost and Methner, 2002; Fredriksson et al., 2003). Rhodopsin family members possess a short N-terminal domain and are characterized by the presence of several conserved amino acid motifs such as N-S-x-x-N-P-x-x-Y within TM7 and the DRY (D(E)-R-Y(F)) motif between TM3 and intracellular loop (IL) 2 (Schioth and Fredriksson, 2005; Suwa et al., 2011). Based upon sequence similarity the human rhodopsin receptors

are sub classified into four main groups (α, β, γ and δ; **Figure 2**; Fredriksson et al., 2003). The α-group contains clusters for the prostaglandin, amine, opsin, melatonin, melanocortin, endothelial, cannabinoid, and adenosine binding receptors. Members of the β-group include a subfamily of receptors for which known ligands are peptides such as orexin (OX), neuropeptide FF (NPFF), neurokinin (NK), gastrin-cholecystokinin (CCK), neuropeptide Y (NPY), endothelin-related (EDN), bombesin and related peptides (BB), neurotensin (NTS), ghrelin and obstatin, neuromedin (NMU), thyrotropin releasing hormone (TRH), arginine vasopressin (AVP), gonadotropin-releasing hormone (GNRH), and oxytocin (OXT). The γ group includes receptors for somatostatin (SST), opioids, galanin (GAL), melanin concentrating hormone (MCH), and chemokine peptides. The δ group contains the olfactory receptors (highly diverse > 400 members) as well as the glycoprotein, purine, and the MAS-related receptor clusters. In humans, twelve members of the rhodopsin family, which are activated by peptide hormones, play an important role in feed intake and stimulate or reduce food consumption (**Table 1**). The majority of these receptors are β group members and their role in the regulation of feed intake in mammals will now be briefly considered.

Receptors for melatonin (MT), gastrin-cholecystokinin (CCK), neurokinin (NK), neuropeptide FF (NPFF), bombesin and related peptides (BB), and neuromedin (NMU) have an inhibitory role in feed intake in vertebrates. Melanocortin receptors (MCR) are activated by melanocortin (ACTH, MSH, and lipotropin) peptides and administration of receptor agonists significantly reduces food consumption in rats (Irani and Haskell-Luevano, 2005). In addition mutant MC3R mice have increased fat mass (Coll et al., 2007) and ablation of the MC4R gene results in severe obesity (Coll et al., 2004; Millington, 2007). In rats, mutations of CCK1R are associated with obesity (Kopin et al., 1999) and peripheral administration of an NK1R antagonist leads to reduced weight gain after a high-fat diet (Karagiannides et al., 2011). Injection of NPFF provokes anorexia in mice and induces satiety (Murase et al., 1996; Bechtold and Luckman, 2006, 2007; Cline et al., 2009).

BB peptides also mediate satiety (Hampton et al., 1998; Yamada et al., 2002; Gonzalez et al., 2008) and knockout BB2R mice have increased body weight (Ladenheim et al., 2002) and BB3Rdeficient mice exhibit a mild obesity phenotype and increased food intake (Ohki-Hamazaki et al., 1997). Mice lacking the NMU gene are hyperphagic and have increased adiposity and obesity and amino acid variants in NMU are associated with human obesity (Brighton et al., 2004; Hainerova et al., 2006).

In contrast, orexin (OXs), neuropeptide Y (NPYs), galanin (GAL), and melanin concentrating hormone (MCH) receptors are activated by orexigenic peptides which stimulate feeding (Sakurai, 1999; Branchek et al., 2000; Chamorro et al., 2002; Lecklin et al., 2002; Lang et al., 2007; Wong et al., 2011). Administration of orexin-A and B stimulates food consumption in a dose-dependent manner (Sakurai et al., 1998; Matteri, 2001). NPY is one of the most potent orexigenic factors and NPYinduced feeding is markedly reduced in Y1-knockout mice and NPY Y1 receptor deficient mice lack appetite (Mercer et al., 2011; Pjetri et al., 2012). GAL1R-KO mice display increased food intake and body weight gain in response to an acute 3 day high-fat challenge (Zorrilla et al., 2007). MCH is a hypothalamic appetite-stimulating peptide that is high in obese mice (Kawauchi, 2006; Coll et al., 2007) and deletions in MCH1R confer resistance to diet-induced obesity (DIO) and MCH1R antagonists are effective in reducing body weight (Chung et al., 2011).

The role of SSTR and their activating peptides in vertebrates is unclear. In rats SSTR can stimulate or inhibit appetite although peptide injections in chickens have an orexigenic effect (Tachibana et al., 2009). In addition receptors for ghrelin-obestatin have opposing effects on feeding and ghrelin is associated with hunger scores and plasma ghrelin levels increase during fasting and decrease after food intake (Rocha-Sousa et al., 2010). Treatment of rats with obestatin suppresses food intake and decreases body weight gain (Zhang et al., 2005).

Other GPCR families activated by peptide hormones may also play a role in food intake and include members of the secretin receptor family: pituitary Adenylate-Cyclase Activating Peptide/Vasoactive Intestinal Peptide (PACR/VIPR; Morley et al., 1992; Chance et al., 1995); Glucagon and related peptide (GCGR/GLPR; McMahon and Wellman, 1997, 1998; Tang-Christensen et al., 2001; Woods et al., 2006); Calcitonin (CTR; Riediger et al., 2004) and Corticotrophin Releasing Factor (CRFR) receptors (Heinrichs and Richard, 1999; Bradbury et al., 2000; Richard et al., 2002). However, the secretin receptor family will not be considered in the present review.

#### **THE INVERTEBRATE GPCRs SUPERFAMILY**

Invertebrates are one of the most diverse animal groups and they represent more than 95% of the species on Earth. Protostomia comprise the majority of the species identified and are of both ecological and economic importance as they are involved in the nutrient cycle, plant fertilization, and include agricultural pests and vectors of human disease, such as malaria and sleeping sickness. The divergence of Protostomes from Deuterostomes occurred more than 700 million years ago (MYA) and their success is associated with adaptations to a variety of ecological niches and modifications in their feeding habits that allow them to live, survive and reproduce in many different environments. Invertebrates can be herbivores (eating plant tissue, nectar, and pollen), carnivores (feeding on other invertebrates as well as larger animals), parasites (living on plant and animals), and detritus feeders (eating dead animal and plants). Surprisingly few studies exist about the regulation of feed intake in invertebrates, despite its importance for their success and this is also a neglected target for alternative control strategies. The genome of several invertebrates has been sequenced and in the metazoan Ensembl genome database (www.ensemblgenomes.org) 48 invertebrate genomes are available. Comparative molecular studies represent an invaluable

mechanism to better understand invertebrate biology and to characterize endocrine factors associated with feeding.

Homologs of the vertebrate GPCR repertoire have been described in many invertebrates and representatives of the five distinct human GRAFS families are proposed to have emerged before the split of nematodes from the chordate lineage (**Table 2**; Fredriksson and Schioth, 2005). The model organisms, the nematode roundworm *C. elegans* and the fruit fly *D. melanogaster* are the most studied Prostostomes. Their genomes have been completely sequenced and are fully annotated and a vast range of functional resources exists and numerous GPCRs have been characterized (Consortium, 1998; Adams et al., 2000; Keating et al., 2003). In the roundworm, GPCRs account for approximately 5% of the genome (there are more than 1000) and the chemoreceptor genes, which are involved in chemoreception of environmental stimuli are unique in nematodes and are also the most abundant and diverse (Schioth and Fredriksson, 2005; Robertson and Thomas, 2006; Nagarathnam et al., 2012). In the fruit fly, approximately 200 GPCRs (1% of the genome) are predicted and the gustatory/taste receptors (Montell, 2009) are specific to insects although a quarter share sequence homology with vertebrate neurohormone receptors (Keating et al., 2003; Fredriksson and Schioth, 2005; Hauser et al., 2006; Nagarathnam et al., 2012). Recently GPCRs were also characterized in the genome of two Platyhelminthes, the blood fluke *Schistosoma mansoni* and the planarian *Schmidtea mediterranea* and a similar gene repertoire to vertebrates has been characterized. A platyhelminth-specific rhodopsin subfamily (PROF1) and a planarian-specific Adhesion-like family (PARF1)

**Table 2 | Gene number and receptor subfamilies of the human rhodopsin GPCRs involved in feeding and the sequence homologs identified in C. elegans and D. melanogaster.**


The total number of receptor genes in human, C. elegans and D. melanogaster is indicated. In C. elegans and D. melanogaster the homologs of the human Neurokinin/Neuropeptide FF/Orexin receptors and Ghrelin-Obestatin/Neuromedin U receptors were grouped due to their high sequence relatedness (Hewes and Taghert, 2001). ni, not identified.

have been identified suggesting lineage specific GPCRs evolved in invertebrates (Suwa et al., 2011; Zamanian et al., 2011).

Comparison of the neuroendocrine GPCR complement in the fruit fly and the honey bee *Apis mellifera* (*A. mellifera*) revealed that a similar gene complement is present (Hauser et al., 2006). In the malaria vector, the mosquito *Anopheles gambiae* (*A. gambiae*) genome, a total of 276 GPCRs are predicted and approximately 30 correspond to putative neuropeptide receptors (Hill et al., 2002). With the exception of *C. elegans*, very little is known about GPCRs in other nematodes despite availability of molecular data in public databases. The activating molecules for the roundworm and fruit fly GPCRs in common with other organisms are in general neurohormones (biogenic amines, protein hormones, and neuropeptides) and they play a central role in the control of behavior, reproduction, development,feeding, and many other physiological

processes. This suggests that GPCR signaling has been conserved during evolution and that neuropeptide signaling plays a key role in both Proto and Deuterostomes (Grimmelikhuijzen and Hauser, 2012).

The present review provides a general overview of the evolution of the rhodopsin GPCR members that are implicated in feeding regulation. It will start by identifying and describing sequence homologs of human rhodopsin GPCRs in the model invertebrate organisms *C. elegans* and *D. melanogaster* followed by the characterization of their homologs in other nematodes and arthropods with distinctive feeding habits and life styles (**Table 4**). The *C. elegans* and *D. melanogaster* rhodopsin GPCR repertoire was obtained from published data and to enrich and confirm the dataset it was complemented with appropriate database searches using the human homologs (**Table 3**). A total of 35 rhodopsin

**Table 3 |The human C. elegans and D. melanogaster rhodopsin GPCRs used for comparative sequence analysis and their accession numbers.**


#### **Table 4 | Nematodes and arthropods used to analyze the rhodopsin GPCRs.**


Information about life style, feeding type and the database interrogated is indicated.

GPCRs are present in*C. elegans* and 22 in*D. melanogaster* genomes (**Table 2**) and a conserved role in feeding regulation has been demonstrated.

#### **FEEDING IN NEMATODES AND ARTHROPODS**

Feeding in invertebrates in common with other animals involves a complex combination of physical, chemical, and nutritional factors (Chapman and De Boer, 1995). Taste and smell are important for feeding behavior and provide the CNS with information on quality and quantity of food and feeding behavior occurs mainly in response to both nutrient and nutritional storage status. Once feeding has been initiated and food ingested, the alimentary canal, and its associated glands triturate, lubricate, store, digest, and absorb the food material and excrete and expel unwanted remains (Audsley and Weaver, 2009).

The Nematoda is a highly diverse, complex, and specialized group of metazoans, about 30,000 species are currently known and many are renowned parasites (15%) and have specialized life cycles that depend on their host to survive and reproduce. Their success is associated with a protective, impermeable cuticle and by the diversity of the pharynx and feeding mechanisms (Coghlan, 2005). The shape and presence or absence of teeth, lancets, stylets, or other structures in the mouth reflects their distinct feeding methods. The majority of nematodes are free-living and inhabit soil and water and feed on microorganisms (bacteria, fungi, algae) and organic debris. The parasites feed on animal and plant tissues and some on vertebrate blood.

The Arthropoda represents the most diverse animal phyla and comprises over 80% of the species identified and the Insecta class is the most specious with approximately 920,000 species. Four main classes of feeding habits are recognized: plant feeders, predators (feed on aphids and mites), scavengers (feeding on dead and decaying organic matter), and parasites (of other insects and vertebrates), some of which are hematophagous. Within each of these classes, various types of feeding can be found such as biting and chewing on leaves or animal tissue and sucking from plant or animal cells or tissues. Despite this unique ability to use almost any organic substrate, most insect species restrict themselves to a particular category of food (Posnien et al., 2010) and feed primarily on a fluid diet (Prakash and Steele, 2010). The variety of feeding habits in arthropods is the result of anatomical and physiological adaptations to distinct food sources (Chapman and De Boer, 1995). The alimentary canal is composed of specialized regions that vary according to feeding habit and life stage.

The organisms selected for analysis of rhodospin GPCRs potentially involved in invertebrate feeding are members of different nematode and arthropod lineages. The specific life style and feeding habits of the invertebrates included in the analysis are indicated in **Table 4**.

#### **HOMOLOGS OF THE VERTEBRATE RHODOPSIN FAMILY GPCRS IMPLICATED IN FEEDING AND APPETITE REGULATION IN NON-VERTEBRATES**

The following section describes the evolution and function of rhodopsin family members in nematodes and arthropods. It will start with an overview of those described in *C. elegans* and *D. melanogaster* involved in or candidates for feed intake regulation (**Tables 2** and **5**). Expression data when available from wormbase and flybase is included to provide insight into receptor function. It is followed by a section in which receptor evolution in invertebrates is discussed including homologs from non-model nematode and arthropod species.

In general, no putative melatonin peptide receptors (MCR) or melanin concentrating hormone receptor (MCHR) homologs have been described or were identified in the present study in any of the selected nematodes or arthropods (**Figure 3**). In addition,


**Table 5 | An overview of the amino acid sequence similarity of the main subfamilies of C. elegans and D. melanogaster rhodopsin GPCRs and their human homologs.**

Percentage of sequence similarity was calculated in the GeneDoc program (http://www.nr.bsc.org/gfx/genedoc/).The maximum and minimum sequence similarity of receptor subgroups between invertebrate and human homologs is indicated.

in nematodes no homolog of the vertebrate and fruit fly bombesin receptors seem to exist (**Table 2**). Duplicates of the human receptor genes were identified in the genomes of nearly all target species and phylogenetic analysis suggests specific gene duplication/deletions occurred within the nematode and arthropod lineages (**Figure 3**).

#### **THE RHODOPSIN GPCRS IN C. ELEGANS AND D. MELANOGASTER GENOMES**

#### **Characterized and functionally assigned subfamily members**

*Gastrin-cholecystokinin receptor subfamily.* In the genomes of *C. elegans* and *D. melanogaster* two putative Gastrin-CCK-like receptor homologs of the human members have been reported (**Figure 3A**; Keating et al., 2003; Janssen et al., 2008). In *C. elegans*, *ckr-1*, and *ckr-2* have been described and functionally characterized. The *ckr-1* is expressed in the nerve ring and functional RNAi knockdown studies reveal that loss of receptor activity provokes fat accumulation (McKay et al., 2007). However, if the receptors are ablated there is no apparent effect on feeding regulation but instead embryonic lethality and reduced brood size is observed (McKay et al., 2007). The neuropeptide *nlp-12* is the ligand of nematode *ckr-2* and the peptide receptor pair shares conserved biological activity with regards to fat storage with the human homolog (Janssen et al., 2008). A cognate peptide for nematode *ckr-1* is yet to be identified.

In *D. melanogaster* the two existent CCK-like receptors were designated CCKL-R17D3 (DSKR1) and CCKL-R17D1 (Kubiak et al., 2002). They are mainly expressed in the CNS and are activated by *Drosophila* sulfakinin (DSK; Nichols et al., 1988), which is a structurally and functionally related peptide to the vertebrate CCK (Audsley and Weaver, 2009). Their role in feeding regulation has not yet been demonstrated in *Drosophila* but in other arthropods the homolog receptor stimulation by SK causes gut emptying and satiety (Nichols, 2007). Injections of SK peptides significantly reduce meal size in locusts (*Schistocerca gregaria*; Wei et al., 2000) and cockroach (*Blattella germanica*; Maestro et al., 2001), carbohydrate feeding in the blowfly (*Phormia regina*), and inhibit female horse flies from blood feeding (Downer et al., 2007).

*Neurokinin/neuropeptide FF/orexin receptor subfamily.* In *C. elegans* two putative neurokinin (a.k.a. tachykinins) receptors *tkr-1* and *tkr-3* have been described (Keating et al., 2003; Greenwood et al., 2005). In *D. melanogaster* three neurokinin-like receptors have been reported: the neurokinin receptor (NKD), the tachykinin receptor (DTKR; Li et al., 1991; Monnier et al., 1992; Rosay et al., 1995; Poels et al., 2009), and the leucokinin receptor (LKR; Radford et al., 2002). Phylogenetic analysis of the invertebrate receptors suggests that they arose from an ancestral Neurokinin/neuropeptide FF/orexin-like receptor gene by speciesspecific duplication events prior to the Proto-Deuterostome divergence (**Figure 3B**; Hewes and Taghert, 2001). Characterization of the *C. elegans tkr-1* revealed expression is restricted to the socket cells (specialized nerve-accessory cells that act as an interface between the sensillum and hypodermis) and RNAi functional screens and the Nile Red fat assay revealed that this gene affects fat metabolism and fat droplet morphology and the pattern of fat deposition (Ashrafi et al., 2003). Knock down nematodes have a substantially lower fat content suggesting that this receptor is a key lipid storage regulator. *Tkr-3* RNAi studies caused mild sluggishness and slowed locomotion in nematodes (Keating et al., 2003), which may be related to modifications in the nervous system. *Tkr-3* is also present in the intestine but no role has yet been assigned in feeding and metabolism.

The *D. melanogaster* NKD and DTK receptors are expressed in the head of both larvae and adults and are activated by *Drosophila* tachykinin (DTK1–6) peptides, which are derived from the *drosotachykinin* (*Dtk*) gene (Birse et al., 2006; Poels et al., 2007) and also by substance P which is involved in the regulation of food intake and energy homeostasis in vertebrates (Birse et al., 2006; Poels et al., 2007). Knock down of DTKR in *D. melanogaster* modulated expression in both fed and starved flies of insulin-like peptides, which play a major role in the regulation of carbohydrates and lipid metabolism (Poels et al., 2009; Birse et al., 2011).

*Neuropeptide Y receptor subfamily.* In *C. elegans* four putative NPY-like receptors (*npr-1*, *npr-2, npr-5*, and *npr-11)* that share conserved sequence with the vertebrate NPYRs have been isolated and function characterized (de Bono and Bargmann,1998;Keating

**FIGURE 3 | Phylogenetic relationship of the Human (Hsa) rhodopsin GPCRs involved in feeding with the nematode C. elegans (Cel) and arthropod D. melanogaster (Dme) sequence homologs.** Trees were constructed using the neighbor joining method with 1000 bootstrap replicates (uniform rate among sites, pairwise deletion using the p-distance substitution model) built in the Mega5.1

et al., 2003; Kubiak et al., 2008; Cohen et al., 2009). Three NPYlike receptors have also been reported in *D. melanogaster*, these are the NepYr receptor and two neuropeptide F (NPF) receptors, the NPFR1 and the short NPFR (SNPFR; **Figure 3C**). The NPF peptide occurs as a long (NPF) and short (sNPF) isoform in arthropods (De Loof et al., 2001) and is the homolog of vertebrate neuropeptide Y (NPY; Li et al., 1992; de Jong-Brink et al., 2001).

In *C. elegans*, the nematode *npr-1*was the first receptor found to influence social feeding behavior and is predominantly expressed in the nervous system (de Bono and Bargmann, 1998). This receptor is activated by *flp-21* peptide (Rogers et al., 2003) and ablation of the peptide does not cause silencing of *npr-1* functions, suggesting that it can be activated by other molecules. In fact, *flp-18* peptide also activates *npr-1* and this peptide is also the ligand of *npr-5*, which is involved, in chemosensory response, foraging behavior, and fat metabolism (Rogers et al., 2003). Nematode *npr-5* is expressed in the head, neck, and body muscles and knock down and gene mutation studies revealed that in common with *npr-2* it is associated with intestinal fat storage regulation (Keating et al., 2003; Cohen et al., 2009), *dauer* formation, and other food-dependent decisions (Cohen et al., 2009). The *npr-11* has a role in reproduction and sensory dynamics of the olfactory system (Chalasani et al., 2010) but no role in feeding has yet been demonstrated (Chalasani et al., 2010).

The fruit fly NepYr and NPF receptors are expressed in the *D. melanogaster* CNS and NepYr is also present in the gut. NepYr is activated by dRYamide-1 and dRYamide-2, which has a C-terminal sequence similar to vertebrate NPY family peptides and in flies dRYamide suppresses feeding motivation (Ida et al., 2011). NPF and its receptors also modulate feeding behavior in *D. melanogaster* (Wu et al., 2003; Garczynski et al., 2005) and they promote feeding in larvae (Wu et al., 2003) and influence the effect of food deprivation in adult flies (Wu et al., 2003; Lingo et al., 2007). In other arthropods their functions have also been described and NPFR is involved in hindgut contraction in the bloodsucking bug (*Rhodnius prolixus*; Gonzalez and Orchard, 2009) and in ovarian maturation in locusts (Schoofs et al., 2001). In *D. melanogaster* sNPF is involved in the control of food intake and in the regulation of body size (Lee et al., 2004). Studies in mutant fruit flies over expressing sNPF peptide exhibit increased food intake and produce bigger and heavier flies, whereas sNPF loss-offunction mutants exhibit suppressed food intake (Lee et al., 2004). Gene expression studies with the red fire ant (*Solenopsis invicta Buren*) revealed SNPFR in brain is down-regulated during starvation (Chen and Pietrantonio, 2006) and expression of long NPF and its receptor in the malaria mosquito (*A. gambiae*) appear to be dependent on the insect nutritional status (Garczynski et al., 2005).

program. Receptors were classified into six distinct subfamilies: **(A)** Gastrin-Cholecystokinin receptors; **(B)** Neurokinin/neuropeptide FF/orexin receptors, **(C)** Neuropeptide Y receptors, **(D)** Bombesin receptors, **(E)** Ghrelin/obstatin and Neuromedin U receptors, and **(F)** Somatostatin and galanin receptors. Accession numbers are described in**Table 3**.

*Bombesin receptor subfamily.* Homologs of the vertebrate bombesin receptors have not been reported in nematodes and were not identified in the present study. Members of this family are only present in *D. melanogaster* and they correspond to the Allatostatin type B receptors (Stay, 2000). In *D. melanogaster*, two bombesin-like receptors have been isolated and function characterized: CCHamide-1r (CCHa1r; Johnson et al., 2003) and CCHamide-2r (CCHa-2r; Johnson et al., 2003; Hauser et al., 2008; **Figure 3D**).

In insects the function of the arthropod bombesin receptor is still poorly explored as a specific ligand has only recently been identified. CCHa-2r expression was detected in *D. melanogaster* brain and in the CNS and midgut of *B. mori* (Roller et al., 2008). Functional analysis reveals the receptors are activated by the peptides CCHamide-1 or CCHamide-2 that have been shown to suppress feeding activity in the cockroach, *Blattella germanica* (Audsley and Weaver, 2009).

*Ghrelin-obestatin/neuromedin U receptor subfamily.* In *C. elegans* four nmur-like receptors: *nmur-1*, *nmur-2*, *nmur-3*, and *nmur-4* have been described. In *D. melanogaster* the capaR and three pyrokinin receptors PK-1R, PK-2-R1, and PK-2-R2 are the homologs of vertebrate NMURs (Iversen et al., 2002; Park et al., 2002; **Figure 3E**). The nematode *nmur-1* is suggested to be involved in the sensory system and with processing information from specific food cues, which enables selection of different food types (Maier et al., 2010). *C. elegans nmur-2* was also shown with its ligand peptide (derived from the *nlp-44* precursor gene) to be involved in the regulation of food intake (Lindemans et al., 2009). To date no functional studies involving *nmur-3* and *nmur-4* have been reported although *nmur-4* is expressed in the pharynx and intestine suggesting it may have a role in feeding.

The *D. melanogaster* capaR is mainly expressed in the Malpighian tubules and it is involved in the increase of fluid transport and diuresis and no direct role in feeding has yet been attributed (Terhzaz et al., 2012). CapaR is activated by two neuropeptides, capa-1 and -2 that are encoded by the *capability* gene and have antidiurectic actions in insects (Pollock et al., 2004;Coast and Garside, 2005; Paluzzi et al., 2010). The *capability* gene also encodes the pyrokinin-1 (PK1) peptide that is a specific activator of PK-1R. PK-2-R1 and PK-2-R2 are activated by pyrokinin-2 (PK2) and Hug-γ that are derived from the hugin (hug) prepropeptide (Cazzamali et al., 2005).

Phylogenetic analysis of the pyrokinin receptors suggests that they share common ancestry and that PK-2-R1 and R2 are the result of a recent duplication in the fly genome. The pyrokinin peptides are involved in rhythmic motor activity in arthropods (Saideman et al., 2007) and receptors are expressed in the abdomen (carcass) and nervous tissue and involvement in modulation of feeding behavior has been suggested. Overexpression of the hugin gene was found to suppress feeding in *Drosophila*, while blockage of the synaptic activity of hugin neurons caused the opposite effect (Meng et al., 2002; Melcher and Pankratz, 2005).

*Somatostatin receptor subfamily.* A homolog of human SSTR in the *C. elegans* genome was predicted in the 1990's (Wilson et al., 1994). Characterization of the deduced protein revealed that the signature motif of the vertebrate SSTR was missing in TM7, suggesting that the receptor is probably activated by other ligands. Since no other homolog of vertebrate SSTR has been reported, the function of the putative SSTR-like receptors in nematodes remains to be explored. In arthropods, Allatostatin type-C receptors are the homologs of the vertebrate somastostatin receptors and in *D. melanogaster*, two receptors star1-RA and AlCR2 were described (Kreienkamp et al., 2002; Mayoral et al., 2010; **Figure 3F**).

The *D. melanogaster* star1-RA and AlCR2 receptors are detected in the CNS and they are activated by allatostatin-C peptides, which are potent modulators of hormone synthesis (Aguilar et al., 2003; Hergarden et al., 2012). These peptides inhibit or stimulate the corpora allata to synthesize juvenile hormone, which is an important regulator of development and reproduction in insects and may indirectly influence feeding behavior (Audsley and Weaver, 2009; Nassel and Winther, 2010).

*Galanin receptor subfamily.* In *C. elegans* and *D. melanogaster* a sequence and function homolog of vertebrate GALR has been described (**Figure 3F**). The *C. elegans* GALR-like receptor, *npr-9* in common with the vertebrate homolog may be involved in food foraging and lipid storage (Bendena et al., 2008). The *npr-9* is expressed in specific neurons around the posterior pharyngeal bulb and*C. elegans* receptor mutants are characterized by impaired food-related roaming behavior and accumulate intestinal fat as a result of fat ingestion and reduced energy expenditure (Lang et al., 2007; Bendena et al., 2008). Peptides involved in the activation of *npr-9* have not been isolated, although *nlp-5* and *nlp-6*, are candidate allatostatin-like peptides that in insects activate the GAL-like receptor (Nathoo et al., 2001).

In arthropods, the Allatostatin type-A receptors are homologs of the vertebrate GALRs (Birgul et al., 1999). Two receptors have been described in *D. melanogaster*, DAR-1 (a.k.a. AlstR) and DAR-2 (Birgul et al., 1999; Lenz et al., 2000; **Figure 3F**). AlstR is expressed in *D. melanogaster* head and CNS while DAR-2 is expressed in the gut suggesting they may have divergent functions. The receptors are activated by FGLamide neuropeptides (Pratt et al., 1991; Woodhead et al., 1994) that in arthropods inhibit food intake (Audsley and Weaver, 2009). Genetic epistasis assays in *D. melanogaster* indicate that FGLamide neuron activation inhibits or limits starvation-induced changes in feeding behavior (Hergarden et al., 2012).

# **Novel subfamily members with an unknown role in feeding regulation**

*Neurokinin/neuropeptide FF/orexin-like receptor subfamily.* In *C. elegans* four additional NKRs members may exist: *npr-14*, *npr-22* and the genes C49A9.7 and C50F7.1 (Keating et al., 2003). In *D. melanogaster* the SIFamide receptor and the gene CG10823 (Hewes and Taghert, 2001) also seem to be novel receptor members (**Table 5**, **Figure 3B**). In the phylogenetic tree, the *C. elegans* gene C49A9.7 clusters with *tkr-1* suggesting they may be duplicates and the nematode *npr-14* and C50F7.1 genes group with the fruit fly CG30340 and SIFamide receptor genes suggesting that they may have emerged from the same gene prior to the nematodearthropod divergence. Functional studies of these receptors are scarce but those that exist indicate that the *C. elegans* MVRFamide neuropeptides but not tachykinin-like peptides activate the *npr-22* receptor (Mertens et al., 2006). The function of *D. melanogaster* CG30340 gene, which is present in low abundance in the digestive and nervous system and of SIFamide receptors are unknown (Jorgensen et al., 2006).

*Neuropeptide Y-like receptor subfamily.* In *C. elegans* at least eight putative novel NPYR gene members are predicted: *npr-3*,*npr-4*, *npr-6*, *npr-7*, *npr-8*, *npr-10*, *npr-12*, and *npr-13* and all remain to be validated and functionally characterized (Keating et al., 2003; **Figure 3C**). The receptors share between 30–40% amino acid sequence similarity with their human counterparts (**Table 5**) and are approximately 20% identical to the *C. elegans* homologs with a characterized function. The high sequence similarity and phylogenetic relationship between *npr-5* and *npr-13* (43%), *npr-4* and *npr-10* (50%) and *npr-11* and *npr-12* (44%) suggests that they may have arisen as a result of a recent duplication event in the nematode genome. These receptors are expressed in nervous tissue and intestine and their function is incompletely described and a specific role in feeding has not been demonstrated (Keating et al., 2003; Styer et al., 2008). In the *D. melanogaster* genome a putative novel insect NPY-like gene of unknown function (CG32547) may also exist (Hewes and Taghert, 2001) and seems to be expressed in the CNS (**Figure 3C**). The CG32547 gene shares less than 14% similarity with the human NPYR members (**Table 5**) and with the other insect family members, although this is probably due to its atypical size of 1008 amino acids, which makes family annotation ambiguous.

*Ghrelin-obestatin/neuromedin U receptor subfamily.* Two putative additional *C. elegans* nmur-like receptor genes the *npr-20* and *npr-21* were retrieved in the present study (**Figure 3E**). They share 30–36% amino acid sequence similarity with human homologs and are probably duplicates (**Table 5**). Expression of *npr-21* in *C. elegans* occurs in nerves of the head, tail, and ventral nerve cord and also in the posterior intestine suggesting that it may have a role in brain-gut function associated with feeding regulation. Similarly in *D. melanogaster* a putative member of this family was also retrieved, the gene CG34381 (**Table 5**) and it clusters with nematode *npr-20* and *npr-21* suggesting that it may have shared common ancestry (Hewes and Taghert, 2001). Expression of the CG34381 gene occurred in the fruit fly head but so far no functional studies have been reported.

*Somatostatin receptor subfamily.* In the *C. elegans* genome at least eight putative SST-like receptor genes are predicted: *npr-15*, *npr-16*, *npr-17*, *npr-18*, *npr-24*, *npr-32*, and the Y54E2A.1 (Vashlishan et al., 2008) and T02E9.1 genes (Keating et al., 2003; **Figure 3F**). No additional putative SST-like receptors were identified or have been reported for *D. melanogaster*. Characterization of the nematode putative SST-like receptors revealed the *C. elegans* members share between 27–35% amino acid sequence similarity with the human SSTRs and that the *npr-24* gene is the most closely related to the insect and human homologs suggesting that they may share a common ancestry (**Table 5**). Comparisons of the putative SSTR in *C. elegans* revealed they are highly divergent suggesting that after their emergence from an ancestral gene they underwent considerable change. Nematode *npr-17* is most similar to *npr-18* and to the T02E9.1 gene with which it shares 23% sequence identity and the three receptors tend to cluster with *npr-16* and *npr-32* suggesting they emerged in the nematode lineage.

The physiological role of the nematode SST-like receptors is poorly characterized but a role in metabolism and feeding behavior is probable. RNAi knockdown studies of *npr-16*, found to be expressed in head/tail neurons and the ventral nerve cord, increased fat deposition (Ashrafi et al., 2003). Ligand binding studies revealed that the peptide *nlp-3* activates the receptor *npr-17,* which seems to be involved in food aversion and has a role in serotonergic modulation via ASH sensory neurons to modulate nematode behavior in response to an external stimuli (Harris et al., 2010). Deletion of the T02E9.1 gene resulted

in an uncoordinated phenotype and nematodes moved slowly and with an increase in circular movement, although feeding was apparently unaffected (Keating et al., 2003). The function of *npr-15*, *npr-18*, *npr-24*, *npr-32*, and Y54E2A.1 remain to be explored.

#### **EVOLUTION OF RHODOPSIN GPCR HOMOLOGS IN INVERTEBRATES**

The evolution of the rhodopsin GPCRs in invertebrates was established (**Figure 4**) by identifying homologs in different nematode and arthropod lineages of the receptors present in *C. elegans* (**Figure 5** and **Table 6**) and *D. melanogaster* (**Figure 6** and **Table 7**). In general, the invertebrate GPCRs with a documented role in feeding or that are sequence homologs of mammalian seem to have evolved differently in nematodes and arthropods. A similar gene complement to that identified in *C. elegans* and *D. melanogaster* was identified in non-model nematodes and arthropods, respectively (**Figure 4**). Nematodes of the superfamily Rhabditoidea generally have more genes than other nematodes (**Table 6**). Gene duplicates in *C. elegans* and *C. brigssae* are more abundant than in arthropods (Lynch and Conery, 2000; Cutter et al., 2009) and a higher number of homologs of the human NPYRs and SSTRs occur

#### **FIGURE 4 | Distribution of rhodopsin subfamily members in nematodes and arthropods.** The phylogenetic relationship of the species analyzed is represented on the right and their feeding habits are indicated. The black circle indicates a putative gene duplication event in the nematode radiation and the black cross potential gene deletion in the T. spiralis genome. Genes that were

identified based upon sequence similarity but that were not considered for phylogenetic analysis are indicated within brackets "()"; ni- GPCR member not identified, and P represent parasitic nematode and arthropod. The evolutionary relationship within nematodes and arthropods was obtained from (Consortium, 2006; Sommer and Streit, 2011).

in nematodes when compared to arthropods (**Figures 5C** and **6C**). In arthropods, species-specific gene duplications exist rather than a conserved gene homolog complement suggesting that,

galanin receptors. The C.elegans (Cel) receptors are annotated in bold. C.

despite their common ancestry, GPCRs have had distinct evolutionary trajectories in the different lineages (**Table 7** and **Figure 6**).

the methodology described in **Figure 3**.


A striking observation is the absence in nematodes of homologs of the arthropod bombesin receptors (BBR; **Figure 4**; **Table 5**). The reason for the loss of BBR in nematodes is unknown and their function and any link to feeding regulation remains to be established. In vertebrates, bombesin and its receptors are involved in smooth muscle contraction, exocrine, and endocrine secretion in the gut, pancreas, and pituitary and they also have a central role in food intake and energy homeostasis (Sano et al., 2004; Gonzalez et al., 2008). Three receptors have been isolated in humans and a similar number exist in arthropods and they share a common ancestry (**Figure 6D**).

A similar number of gastrin-CCK, NKR, NMUR, and GALR subfamily members were characterized in nematodes and arthropods (**Figure 4**). Two putative gastrin-CCK receptors were identified in invertebrates and in humans two gastrin-CCK receptors also exist suggesting that the evolution of the members of this family has been highly conserved. However, phylogenetic analysis suggests that the duplication, which delivered the two gene copies, was not common to all the species and occurred independently within each lineage. The two *ckr* that are present in nematodes resulted from a lineage specific duplication and homologs of the two *C. elegans* genes were identified in most nematode genomes analyzed (**Figure 5A**). In arthropods, a different situation exists and the two *D. melanogaster* genes are very similar and seem to have resulted from a species-specific duplication event (**Figure 6A**). Similarly in the blacklegged tick (*I. scapularis*) three putative gastrin-CCK receptors were also identified. In contrast, no putative homologs were identified in the plant feeding arthropod, the silkworm *B. mori*, even though they had a similar gene complement to other arthropods. It remains to be established if the absence of this receptor in *B. mori* is a consequence of its incomplete genome assembly (Xia et al., 2004) or represents an adaptation relative to feeding regulation.

Members of the NKR, NMUR, and GALR subfamilies have also evolved via lineage specific and species-specific duplication events. In nematodes, a similar number of NKR, NMUR, and GALR receptors exist in *H. contortus* and in the three representatives of the *Caernohabitis* genus analyzed (**Figures 5B,D,E**). In contrast, few genes of these families have been identified in other nematode taxa and a single NKR subfamily member was retrieved from *P. pacificus*, *M. incognita*, *B. malayi*, and *T. spiralis*. In arthropods, gene duplication of the *D. melanogaster* LKR receptor homologs was identified in the mosquito *A. aegypti* and also in *I. scapularies* in which four putative receptors exist (**Figure 6B**). In addition, in the honeybee (*A. mellifera*) three putative homologs of the fruit fly DTKR receptors were also identified. In contrast, no homologs of *D. melanogaster* NKD were detected in the honeybee and *A. aegypti* genomes. Within the NMUR family (**Figure 6E**), the *D. melanogaster* PK2Rs emerged as a consequence of a species-specific duplication event and two putative capaR were also identified in the honeybee, but only a single member was found in *I. scapulars*. In contrast, duplication of GALR occurred in the *I. scapulars* genome and four putative receptors were identified while other arthropods contained a single homolog of *D. melanogaster* DAR-1 and DAR-2 genes (**Figure 6F**).

The complete genome sequence of some of the species used in this study are not yet available, nonetheless gene representatives


#### **Table 7 | Accession numbers of the D. melanogaster homologs in A. gambiae, A. aegypti, A. mellifera, B. mori and I. scapularis.**

\*Indicates sequences not used in the phylogenetic analysis due to poor sequence or non-identification of TM domains. ni: indicates gene not identified.

identified in the selected nematodes and arthropods provides a clear idea of the GPCR evolution in invertebrates. The majority of the *C. elegans* sequence homologs were identified in the target species and an increase in gene number seems to have occurred in Rhabditoidea and Strongyloidea (Abad et al., 2008; Dieterich et al., 2008; Mitreva et al., 2011). The exception was *B. malayi* in which representatives of NMUR and GALR were not identified possibly because of its incomplete genome assembly (Ghedin et al., 2007). The absence of the majority of the *C. elegans* receptor homologs in parasitic nematode genomes and the higher number of genes present in *H. contortus* and in other representatives of the *Caernohabitis* genus is curious. A general comparison of the gene content of *T. spiralis* with *C. elegans* revealed that the parasitic nematode genome contains fewer genes (15,808 compared to 20,060 and 19,507 in *C. elegans* and *C. briggsae*, respectively) and we hypothesize that gene absence is a consequence of the selective pressures provoked by the host on which they live and depend for survival (Mitreva et al., 2011; Sommer and Streit, 2011). The genome of *P. pacificus* is predicted to contain a higher gene number than *C. elegans* and suggests that a specific GPCR gene expansion occurred in the nematode lineage after their divergence (Dieterich et al., 2008; Sommer and Streit, 2011). Comparisons between *T. spiralis* and the other blood feeding parasitic nematode *H. contortus* revealed that the latter has a higher GPCR gene number than *T. spiralis*. One explanation may be related to their life cycles and while both nematodes need blood to survive *T. spiralis* is an obligate parasite, while *H. contortus* has a non-parasitic free-living stage. Intriguingly during the parasitic stage of *H. contortus* significant changes in the active transcriptome occurs when compared to the nematode free-living stage (Hoekstra et al., 2000) and it will be of interest to establish if this affects the diversity of rhodopsin GPCRs expressed.

In arthropods, GPCR gene evolution appears species dependent and specific gene duplications and deletions have occurred despite their common ancestry. The existence of specific gene duplicates in arthropods may indicate that a divergent regulatory system evolved in different species and the origin and maintenance of duplicates in the genome remain to be explored. Gene number in the two mosquito species analyzed are very similar and may reflect their identical life styles (Klowden, 1990). In the tick, which feeds exclusively on blood, a specific expansion of NKR and GALR gene families occurred. Further studies are required to determine the significance of the specific evolution of rhodopsin family GPCRs in arthropods and to consider how life style and feeding activity may have influenced receptor evolution.

#### **FINAL CONSIDERATIONS**

In general, the physiological processes involving GPCRs are conserved and sequence and function homologs of vertebrate rhodopsin GPCRs are present in invertebrates indicating they emerged early in evolution. In Nematoda and Arthropoda the rhodopsin GPCRs have evolved differently. Gene expansion is observed in nematodes with a free-living stage and specific gene deletions seem to have affected parasitic nematode genomes. In arthropods species-specific gene duplications occurred. We hypothesize that the evolving feeding regime and life style of invertebrates was one of the pressure forcing GPCR evolution and that this may explain some of the specific gene family expansions and deletions. Comparative studies of GPCRs gained or lost in the

#### **REFERENCES**


regulation of feeding in insects. *Gen. Comp. Endocrinol.* 162, 93–104.


nematodes and arthropods and their relationship to feeding regulation may provide insights into how GPCRs contributed and shaped adaptation to new ecological niche. Studies of other nematodes and arthropods coupled with experiments to assign function and potential conserved role in feeding will be needed to test this hypothesis.

#### **ACKNOWLEDGMENTS**

This study was funded by the Portuguese Science Foundation PTDC/BIA-BCM/114395/2009 and CCMAR pluriannual grant. Vera G. Fonseca was supported by FCT grant SFRH/BPD/80447/2011.

#### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at http://www.frontiersin.org/Neuroendocrine\_Science/10.3389/ fendo.2012.00157/abstract

**Figure S1 | Sequence of the nematodes GPCR transmembrane (TM) domains from non-model nematodes within each receptor family were extracted by sequence homology using the roundworm C. elegans TM regions.** To facilitate visualization the TM1, 3, 5, and 7 were annotated in gray.

**Figure S2 | Sequence of the Arthropod GPCR transmembrane (TM) domains used in for phylogenetic analysis.** TM domains from non-model arthropods within each receptor family were extracted by sequence homology using the D. melanogaster TMs. To facilitate visualization the TM1, 3, 5 and 7 were annotated in gray.

to the mammalian somatostatin/galanin/opioid receptor family. *EMBO J.* 18, 5892–5900.


C. elegans: a platform for investigating biology. *Science* 282, 2012–2018.


Anopheles gambiae. *Peptides* 26, 99–107.


in Drosophila melanogaster. *Neuroscience* 148, 371–374.


and ecdysis-modifying neuropeptides. *Mech. Dev.* 117, 5–13.


neurokinin-like receptors. *Peptides* 28, 103–108.


regulation of food intake. *Physiol. Rev.* 85, 1131–1158.


Loof, A., and Schoofs, L. (2000). Sulfakinins reduce food intake in the desert locust, Schistocerca gregaria. *J. Insect Physiol.* 46, 1259–1265.


*Trans. R. Soc. Lond. B Biol. Sci.* 361, 1219–1235.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 28 August 2012; accepted: 22 November 2012; published online: 18 December 2012.*

*Citation: Cardoso JCR, Félix RC, Fonseca VG and Power DM (2012) Feeding and the rhodopsin family G-protein coupled receptors in nematodes and arthropods. Front. Endocrin. 3:157. doi: 10.3389/fendo.2012.00157*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Cardoso, Félix, Fonseca and Power. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Gonadotropin-releasing hormone 2 suppresses food intake in the zebrafish, Danio rerio

# *Ryo Nishiguchi1, Morio Azuma1, EriYokobori1, Minoru Uchiyama1 and Kouhei Matsuda1,2\**

<sup>1</sup> Laboratory of Regulatory Biology, Graduate School of Science and Engineering, University of Toyama, Toyama, Japan <sup>2</sup> Laboratory of Regulatory Biology, Graduate School of Innovative Life Science, University of Toyama, Toyama, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Jack Falcon, Université Pierre et Marie Curie and Centre National de la Recherche Scientifique, France John Chang, University of Alberta, Canada

#### *\*Correspondence:*

Kouhei Matsuda, Laboratory of Regulatory Biology, Graduate School of Science and Engineering, University of Toyama, 3190 Gofuku, Toyama 930-8555, Japan. e-mail: kmatsuda@sci.u-toyama.ac.jp Gonadotropin-releasing hormone (GnRH) is an evolutionarily conserved neuropeptide with 10 amino acid residues, of which several structural variants exist. A molecular form known as GnRH2 ([His5 Trp<sup>7</sup> Tyr8]GnRH, also known as chicken GnRH II) is widely distributed in vertebrates except for rodents, and has recently been implicated in the regulation of feeding behavior in goldfish. However, the influence of GnRH2 on feeding behavior in other fish has not yet been studied. In the present study, therefore, we investigated the role of GnRH2 in the regulation of feeding behavior in a zebrafish model, and examined its involvement in food intake after intracerebroventricular (ICV) administration. ICV injection of GnRH2 at 0.1 and 1 pmol/g body weight (BW) induced a marked decrease of food consumption in a dose-dependent manner during 30 min after feeding. Cumulative food intake was significantly decreased by ICV injection of GnRH2 at 1 pmol/g BW during the 30-min posttreatment observation period. The anorexigenic action of GnRH2 was completely blocked by treatment with the GnRH type I receptor antagonist Antide at 25 pmol/g BW. We also examined the effect of feeding condition on the expression level of the GnRH2 transcript in the hypothalamus. Levels of GnRH2 mRNA obtained from fish that had been provided excess food for 7 days were higher than those in fish that had been fed normally. These results suggest that, in zebrafish, GnRH2 acts as an anorexigenic factor, as is the case in goldfish.

**Keywords: zebrafish, GnRH2, food intake, ICV administration, Antide, anorexigenic action**

**"fendo-03-00122" — 2012/10/16 — 12:40 — page 1 — #1**

# **INTRODUCTION**

Gonadotropin-releasing hormone (GnRH) is an evolutionarily conserved decaneuropeptide that plays a crucial role in the regulation of reproduction in vertebrates (Sherwood et al., 1993; Fernald and White, 1999, Millar et al., 2004). The demonstration of GnRH structural variants in vertebrates, and even in invertebrates, has now resulted in the identification of 29 molecules (Guilgur et al., 2006; Kah et al., 2007; Roch et al., 2011). In vertebrates, these peptides are distributed in a wide range of tissues, and have diverse functions as hypophysiotropic hormones, paracrine or autocrine mediators and neuromodulators/neurotransmitters in the central and peripheral nervous systems and tissues (Gore, 2002; Millar, 2005; Kim et al., 2007; Millar et al., 2007). GnRH with substitutions at the N-terminal fifth, seventh, and eighth positions by histidine, tryptophan, and tyrosine residues, respectively, was originally purified and characterized as a second type of GnRH from chicken brain, and was named chicken GnRH II (now called GnRH2; Miyamoto et al., 1984). Subsequently, it has been found that GnRH2 is present throughout the vertebrates from cartilaginous fish to humans, but not in rodents (Conlon et al., 1993; Sherwood et al., 1993; White et al., 1998; Millar, 2003). GnRH2 has been implicated in the central regulation of reproductive behavior as well as neuroendocrine control of the gonads (Maney et al., 1997; Volkoff and Peter, 1999; Schiml and Rissman, 2000; White et al., 2002; Temple et al., 2003; Lethimonier et al., 2004; Hofmann, 2006). Recently, it has been reported that in an insectivore, the musk shrew, GnRH2 also influences feeding behavior, and that intracerebroventricular (ICV) administration of GnRH2 induces a marked decrease in food consumption, suggesting that GnRH2 controls reproduction and energy balance (Temple et al., 2003; Kauffman, 2004; Kauffman and Rissman, 2004a,b; Kauffman et al., 2005b, 2006). However, the involvement of GnRH in the regulation of feeding behavior had not been studied in other animal models.

Previous studies have indicated that ICV injection of GnRH2 also induces an anorexigenic effect in a goldfish model (Hoskins et al., 2008; Matsuda et al., 2008; Kang et al., 2011). In addition to goldfish, the zebrafish has now been widely used as an excellent animal model to investigate the effects of neuropeptides on feeding behavior (Yokobori et al., 2011, 2012; Matsuda et al., 2012). As in rodents and goldfish, it has been found that, in zebrafish, neuropeptide Y (NPY) and orexin A stimulate food intake (Yokobori et al., 2011, 2012; Matsuda et al., 2012). However, the exact role of GnRH2 is unclear, and there is no information about the effect of GnRH2 on feeding behavior in this species.

Therefore, the aim of the present study was to investigate the effect of GnRH2 on food intake in the zebrafish model, and the effect of ICV injection of Antide, a GnRH type I receptor antagonist, on the action of ICV-administered GnRH2. We also examined the effect of feeding condition on the expression level of the GnRH2 transcript in the hypothalamus.

# **MATERIALS AND METHODS**

#### **ANIMALS**

Adult zebrafish (*Danio rerio*, 0.5–1.0 g body weight, BW) of both sexes were obtained commercially, and kept for 2 weeks under controlled light/dark conditions (12 h light/12 h dark) in a water-temperature-regulated fish tank (20–24◦C) before use in experiments, since prevention of gonadal development. The fish were fed a commercially available granule diet (containing 32% protein, 4% dietary fat, 3% dietary fiber, 9% mineral, 8% water, and 44% other components; Hikari MariGold, Kyorin, Kobe, Japan) every day at noon. For 1 week before the experiments each fish was kept in a small experimental tank (24 cm in diameter) with 3.5 l of tap water. All animal experiments were conducted in accordance with the University of Toyama guidelines and the Declaration of Helsinki for the care and use of animals. Every effort was made to minimize the number of animals used and their suffering.

#### **CHEMICALS**

The zebrafish possesses two molecular forms of GnRH: GnRH2 (pGlu-His-Trp-Ser-His-Gly-Trp-Tyr-Pro-Gly-NH2), and GnRH3 (pGlu-His-Trp-Ser-Tyr-Gly-Trp-Leu-Pro-Gly-NH2; Powell et al., 1996; Steven et al., 2003). Therefore, in order to examine the effect of ICV injection of GnRH2 on food intake, GnRH2 was purchased from Bachem AG (Bubendorf, Switzerland) and used. Zebrafish possesses four kinds of GnRH receptors (GnRH R1–R4), GnRH R1 and R3 being of the GnRH type I receptor type whereas GnRH R2 and R4 are GnRH type III receptors (Tello et al., 2008). In the present study, we used the GnRH type I receptor antagonist, Antide (acetyl-D-Ala(2-naphthyl)-D-Phe(4-Cl)-D-Ala(3-pyridyl)- Ser-Lys(Nε-nicotinoyl)-D-Lys(Nε-nicotinoyl)-Leu-Lys(Nε-isopropyl)-Pro-D-Ala-NH2), obtained from Sigma-Aldrich Co. (St. Louis, MO, USA). Antide was dissolved in 0.1% acetic acid and diluted with 0.6% NaCl and 0.02% Na2CO3 solution (saline) before use.

#### **EFFECT OF ICV ADMINISTRATION OF GnRH2 ON FOOD INTAKE**

Details of the methods used for evaluating feeding behavior in zebrafish have been reported elsewhere (Yokobori et al., 2011, 2012). Each fish was normally fed before the experiments began at noon, and placed in a wet sponge under anesthesia with MS-222 (3-aminobenzoic acid ethyl ester; Sigma-Aldrich). A small part of the parietal bone was carefully removed using a surgical blade (No. 19, Futaba, Tokyo, Japan), and then 0.5 μl/g BW of GnRH2 at doses of 0.1 and 1 pmol/g BW was injected into the third ventricle of the brain using a small Hamilton syringe. The gap in the bone was then filled with a surgical agent (Aron Alpha, Sankyo, Japan). The accuracy of the injection site was confirmed after the experiment by examining whether Evans blue dye, injected at the same time, was present in the ventricle (**Figure 1A**). Control fish in each experiment were injected with the same volume of vehicle (less than 0.01% acetic acid diluted with saline) in the same way as for the experimental group. Each fish that had received an injection was individually placed in a small experimental tank (24 cm in diameter) containing 3.5 l of tap water, and supplied with food equivalent to 3% of its BW. Food intake was measured by directly observing and recording the number of diet pellets

eaten by individual fish over 15 and 30 min of commencement of feeding.

# **EFFECT OF ICV INJECTION OF ANTIDE ON GnRH2-INDUCED ANOREXIGENIC ACTION**

Because pilot experiments had shown that ICV injection of cGnRH2 at a dose of 1 pmol/g BW induced a marked decrease of food intake, Antide at 25 pmol/g BW, a dose previously determined to be sufficient to suppress the action of GnRH2 in goldfish (Matsuda et al., 2008), was delivered by ICV injection in addition to GnRH2 at 1 pmol/g BW. Control fish were injected with the same volume of vehicle (less than 0.01% acetic acid diluted with saline) in the same way as for the experimental group. Food intake was then measured over the first 15 and 30 min of commencement of feeding, as described above.

#### **EFFECT OF FEEDING CONDITION ON GnRH2 mRNA EXPRESSION IN THE HYPOTHALAMUS**

"fendo-03-00122" — 2012/10/16 — 12:40 — page 2 — #2

Every day for 7 days, fish were supplied an excessive amount of food corresponding to 9% of their BW. Other fish were fed an amount of food corresponding to 3% of their BW for the same period. On day 7, the fish were anesthetized with MS-222 and decapitated. Because the zebrafish brain is very small, each brain was dissected out, and the hypothalamus was collected, weighed, and immersed immediately in liquid nitrogen, before being stored at −80◦C until use. Total RNA was extracted from each part of the brain with Isogen (a solution containing phenol and guanidinium isothiocyanate; Nippon Gene, Tokyo, Japan). For amplification and quantitation of the cDNA fragments encoding GnRH2 and β-actin, the one-step reverse transcription polymerase chain reaction (RT-PCR) method (SYBR Green RT-PCR Reagents Kit, Applied Biosystems, Foster City, CA, USA) was used. Reactions (including 5 μM primers, 2× SYBR Green PCR master mix, 6.25 U MultiScribe reverse transcriptase, 10 U RNase inhibitor, RNA template, and water) were set up in a 96 well reaction plate and placed in a sequence detection system for cycling (TP 800, Takara, Tokyo, Japan). Reverse transcription was carried out at 48◦C for 30 min and the resulting cDNA was subsequently amplified using 40 cycles of 95◦C for 15 s followed by 60◦C for 60 s. The PCR products from each cycle were monitored using SYBR Green I fluorescent dye (Applied Biosystems). Gene-specific primers for amplification of the GnRH2 cDNA fragment were based on the nucleotide sequence of zebrafish GnRH2 (GenBank ID, BC162945.1, NM\_181439; Ensembl ID, ENSDARG00000044754). PCR with the sense primer (5 -CAA AAT ATT AGA CTG AAG TGA TGG T-3 ) and the antisense primer (5 -GGT CTA TCT CTC TCT TTC CTC CA-3 ) yielded a 86-bp product encoding zebrafish GnRH2 cDNA. Zebrafish βactin-specific primers were used as the internal control for PCR amplification (GenBank accession number, NM\_181601; Ensembl ID, ENSDART00000055194). Using these primers (sense primer, 5 -GTG ATG GAC TCT GGT GAT GGT GT-3 ; antisense primer, 5 -TGA AGC TGT AGC CTC TCT CGG TC-3 ), a 148-bp product corresponding to a region in the central part of the β-actin cDNA sequence was obtained. The expression levels of GnRH2 mRNA were calculated quantitatively as a ratio relative to the expression of β-actin mRNA.

#### **DATA ANALYSIS**

All the results are expressed as mean ± SEM. Statistical analysis was performed by one- and two-way ANOVA with Bonferroni's method or Student's *t*-test. Statistical significance was determined at the 5% level.

#### **RESULTS**

#### **EFFECT OF ICV ADMINISTRATION OF GnRH2 ON FOOD INTAKE**

Intracerebroventricular injection of GnRH2 (at 0.1 and 1 pmol/g BW) inhibited food intake over a 30-min feeding period. A significant reduction in cumulative food consumption was observed at a dose of 1 pmol/g BW at both 15 and 30 min after commencement of feeding (**Figure 1B**). The *df*, *F*, and *P* values between treatments with saline and GnRH2 were: 2, 3.08, and 0.06 at 15 min; 2, 3.51, and 0.04 at 30 min.

#### **EFFECT OF ICV INJECTION OF ANTIDE ON ANOREXIGENIC ACTION OF GnRH2**

Intracerebroventricular administration of GnRH2 alone at 1 pmol/g BW suppressed food intake over a 30-min feeding period, and ICV-injected Antide alone at 25 pmol/g BW did not affect food intake. On the other hand, the same dose of Antide completely

abolished the anorexigenic action of ICV-injected GnRH2 at a dose of 1 pmol/g BW, and the efficacy of the antagonist was shown to be significant by two-way ANOVA with Bonferroni's method (*df*, *F*, and *P* values, 1, 4.39, and 0.04, respectively; **Figure 2**).

#### **EFFECT OF FEEDING CONDITION ON GnRH2 mRNA EXPRESSION IN THE HYPOTHALAMUS**

**Figure 3** shows the expression levels of GnRH2 mRNA in the hypothalamus of zebrafish supplied an excessive amount of food corresponding to 9% of their BW, and normal amount of food corresponding to 3% of their BW. Expression of GnRH2 mRNA was estimated quantitatively as a ratio relative to the expression of β-actin mRNA. In the hypothalamus, excessive feeding for 7 days induced a significant increase (approximately four times higher) in the level of GnRH2 mRNA compared with that in fish that had been fed normally (*t* and *P* values, 2.82 and 0.012, respectively; **Figure 3**).

# **DISCUSSION**

"fendo-03-00122" — 2012/10/16 — 12:40 — page 3 — #3

We have developed methods for administering ICV test substances and for measuring food consumption in a small fish, the zebrafish, and our previous studies have demonstrated that, in this species, NPY and orexin A act as orexigenic neuropeptides (Yokobori et al., 2011, 2012). In the present study, we investigated the effect of central administration of GnRH2 on food

intake, and demonstrated for the first time that GnRH2 strongly suppresses food consumption in the zebrafish. In matured female musk shrew, an insectivore, ICV injection of GnRH2 inhibits food intake, and feeding status influences the levels of both GnRH2 mRNA expression and immunoassayable GnRH2 in the brain, which are decreased by food restriction (Kauffman and Rissman, 2004a; Kauffman et al., 2006). The present study indicates the anorexigenic action of GnRH2 in the zebrafish, as is the case in female musk shrew and goldfish (Hoskins et al., 2008; Matsuda et al., 2008; Kang et al., 2011), indicating that it may be involved generally in the regulation of feeding behavior in vertebrates. GnRH3 is widely distributed in several regions of the zebrafish brain, including the olfactory bulb, the area of the terminal nerve, and telencephalon, and GnRH3 is implicated in pituitary control (Torgersen et al., 2002; Steven et al., 2003; Palevitch et al., 2007). On the other hand, GnRH2-expressing neurons are localized mainly in the midbrain tegmentum (Palevitch et al., 2007). However, the exact role of GnRH2 in the zebrafish has been unclear. Because GnRH2 is implicated in the regulation of reproductive behavior and energy balance in the female musk shrew, sparrow, and goldfish (Maney et al., 1997; Temple et al., 2003; Kauffman, 2004; Kauffman and Rissman, 2004a; Kauffman et al., 2005a,b; Hofmann, 2006; Hoskins et al., 2008; Matsuda et al., 2008; Kang et al., 2011), it is likely that, in zebrafish, GnRH2 is involved in both feeding control and reproductive behavior. GnRH systems have been well studied in teleost fish (Lethimonier et al., 2004). The zebrafish possesses two GnRHs – GnRH2 and GnRH3 – which are encoded by two distinct genes (Steven et al., 2003), and four GnRH receptors, GnRH R1–R4 – which are members of the G protein-coupled receptor family (Tello et al., 2008). There are marked differences in structure between these two groups of GnRHRs. GnRH R1 and R3 are evolutionarily derived from the common ancestor of the GnRH type I receptor, and GnRH R2 and R4 belong to the GnRH type III receptor group (Tello et al., 2008). In the present study, ICV administration of GnRH2 at 1 pmol/g BW induced a significant decrease of food intake, and this effect was completely blocked by treatment with Antide. Antide is the GnRH type I receptor antagonist (Kauffman et al., 2005b; Matsuda et al., 2008). These results suggest that the anorexigenic action of GnRH2 is mediated by the Antide-sensitive receptor system, perhaps involving GnRH R1 and/or R3. The knowledge about the expression and the distribution of GnRH receptors in the zebrafish brain should help us to speculate on the mechanisms underlying GnRH2-induced anorexigenic effect. However, there has been no information about distribution of GnRH receptors in the hypothalamus. It is unclear which receptor type mediates the anorexigenic action of GnRH2 in zebrafish.

In mammals, several neuropeptides including CRH, galaninlike peptide, LHRH (GnRH1), kisspeptin, α-MSH, NPY, orexin, and 26RFa are implicated in the regulation of nutrition and reproduction, suggesting that feeding and reproductive functions are closely linked (Catzeflis et al., 1993; Iqubal et al., 2001; Chartrel et al., 2003; Kauffman et al., 2005a; Crown et al., 2006; Martynska et al., 2006; Navarro et al., 2006; Maeda et al., 2007). Orexin, which has crucial role in the sleep–wakefulness cycle and appetite control, affects GnRH1 release directly or via the NPY-, CRH-, and βendorphin-signaling pathways (Li et al., 1999; Tamura et al., 1999; Irahara et al., 2001; Yang et al., 2005; Iwasa et al., 2007). In goldfish, ICV administration of orexinA suppresses spawning behavior, and ICV administration of GnRH2 reduces the level of orexin precursor mRNA in the brain (Hoskins et al., 2008). These data suggest that, in goldfish, GnRH2 and orexin have opposite roles in appetite and satiety regulation. Our previous studies have revealed that mRNA expression levels for NPY and orexin in the hypothalamus obtained from zebrafish fasted for 7 days are higher than those in zebrafish that had been fed normally, suggesting NPY and orexin in the hypothalamus act as orexigenic factors in this species. In the present study of zebrafish, we focused on the GnRH2 neuronal system, which has been implicated in the regulation of food intake in the goldfish (Matsuda et al., 2008), and demonstrated for the first time that the expression of GnRH2 mRNA in the hypothalamus is affected by feeding status. The results support the idea that GnRH2 in the hypothalamus acts as an anorexigenic neuropeptide in this species. Further investigations to clarify the regulatory mechanism of food intake by GnRH2 and other neuropeptides and factors are warranted.

In conclusion, the present study has demonstrated for the first time that ICV administration of GnRH2 suppresses food intake in the zebrafish. These results suggest that GnRH2 induces behavioral changes, and in particular acts as an anorexigenic factor in this species. The present findings also indicate that evolutionary pressure has acted to preserve the function of GnRH2 as a feeding regulator across the vertebrates.

#### **ACKNOWLEDGMENTS**

"fendo-03-00122" — 2012/10/16 — 12:40 — page 4 — #4

This work was supported by a Grant-in-Aid from the Japan Society for Promotion of Science (21370025 to Kouhei Matsuda), and by a research grant from the University of Toyama (to Kouhei Matsuda).

# **REFERENCES**


*Biophys. Res. Commun.* 281, 232–236.


nonhypothalamic tissues. *Semin. Reprod. Med.* 25, 326–336.


"fendo-03-00122" — 2012/10/16 — 12:40 — page 5 — #5

evidence that gonadotropin secretion is probably controlled by two distinct gonadotropin-releasing hormones in avian species. *Proc. Natl. Acad. Sci. U.S.A.* 81, 3874–3878.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 August 2012; paper pending published: 24 September 2012; accepted: 30 September 2012; published online: 17 October 2012.*

*Citation: Nishiguchi R, Azuma M, Yokobori E, Uchiyama M and Matsuda K (2012) Gonadotropin-releasing hormone 2 suppresses food intake in the zebrafish, Danio rerio. Front. Endocrin. 3:122. doi: 10.3389/fendo.2012.00122*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Nishiguchi, Azuma, Yokobori, Uchiyama and Matsuda. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

"fendo-03-00122" — 2012/10/16 — 12:40 — page 6 — #6

# Neurotensin and its receptors in the control of glucose homeostasis

# *Jean Mazella\*, Sophie Béraud-Dufour, Christelle Devader, Fabienne Massa and Thierry Coppola\**

Institut de Pharmacologie Moléculaire et Cellulaire, UMR 7275, Centre National de la Recherche Scientifique, Université de Nice-Sophia Antipolis, Valbonne, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Isabelle Dubuc, Université de Rouen, France Gina Leinninger, Michigan State University, USA

#### *\*Correspondence:*

Jean Mazella and Thierry Coppola, Institut de Pharmacologie Moléculaire et Cellulaire, UMR 7275, Centre National de la Recherche Scientifique, Université de Nice-Sophia Antipolis, 660 route des Lucioles, Sophia Antipolis, 06560 Valbonne, France. e-mail: mazella@ipmc.cnrs.fr; coppola@ipmc.cnrs.fr

The pharmacological roles of the neuropeptide neurotensin through its three known receptors are various and complex. Neurotensin is involved in several important biological functions including analgesia and hypothermia in the central nervous system and also food intake and glucose homeostasis in the periphery.This review focuses on recent works dealing with molecular mechanisms regulating blood glucose level and insulin secretion upon neurotensin action. Investigations on crucial cellular components involved in the protective effect of the peptide on beta cells are also detailed. The role of xenin, a neurotensin-related peptide, on the regulation of insulin release by glucose-dependent insulinotropic polypeptide is summarized. The last section comments on the future research areas which should be developed to address the function of new effectors of the neurotensinergic system in the endocrine pancreas.

**Keywords: neurotensin, G protein-coupled receptor, pancreas, beta cell, sortilin**

**"fendo-03-00143" — 2012/11/23 — 18:50 — page 1 — #1**

# **INTRODUCTION**

The endogenous peptide neurotensin (NT) was discovered in 1973 in bovine hypothalami by Carraway and Leeman (1973) for its ability to induce vasodilatation. This property added to its cerebral expression indicated that the peptide could fulfill a dual function, as a neurotransmitter/neuromodulator in the central nervous system, and as a hormone in the periphery. NT is expressed in the central nervous system where it is located in neuronal synaptic vesicles (Uhl and Snyder, 1976; Bayer et al., 1991) and in the gastrointestinal tract (Goedert and Emson, 1983) in neuroendocrine cells (Atoji et al., 1995). NT is synthesized from a precursor following excision by prohormone convertases (Kitabgi, 2006).

The pharmacological and physiological effects of NT are triggered following its interaction, depending on the tissue or the cell type, by three known receptors (NTSRs). Two of them, NTSR1 and NTSR2, are classical neuropeptide receptors coupled to Gproteins (GPCR) bearing seven transmembrane domains (Vincent et al., 1999; Myers et al., 2009). The third one, NTSR3 also called sortilin (Petersen et al., 1997; Mazella et al., 1998), is a type I receptor with a single transmembrane domain, non-coupled to G-proteins, which belongs to the Vps10p containing domain receptor family (Mazella, 2001; Hermey, 2009). The heterogeneity in the structure of NTSRs, added to their ability to form homo and/or heterodimers (Martin et al., 2002; Beraud-Dufour et al., 2009; Hwang et al., 2010), underlines the complexity to study the neurotensinergic system both in the brain and in peripheral tissues.

In the central nervous system, the effects of NT include the interaction of the peptide with the dopaminergic system (Studler et al., 1988) and its ability to induce opioid-independent analgesia (Dubuc et al., 1999) and hypothermia (Popp et al., 2007). The latter property could allow the use of bioavailable NT analogs to reduce the risk of brain damage following hypoxia (Choi et al., 2012). The anti-psychotic and hypothermia effects of NT are mediated through NTSR1 (Dobner, 2005; Hadden et al., 2005; Choi et al., 2012), whereas NTinduced analgesia involves both NTSR1 and NTSR2 (Dubuc et al., 1999; Smith et al., 2012). NTSR3/sortilin is responsible for the migration and the release of cytokines and chemokines from microglial cells induced by NT (Martin et al., 2003; Dicou et al., 2004).

At the neuroendocrine point of view, the actions of NT as a regulator of anterior pituitary secretions and food intake has been recently well reviewed (Rostene and Alexander, 1997; Stolakis et al., 2010; Kalafatakis and Triantafyllou, 2011). For this reason, these sections will be only summarized at the beginning of this review focusing essentially on the role of NT in the control of glucose homeostasis. The roles of NT in insulin regulation and pancreatic cell growth have been initiated by several works (Kaneto et al., 1978; Dolais-Kitabgi et al., 1979; Wood et al., 1988). However, molecular characterization of the signaling pathways and the receptors involved in both pancreatic cell proliferation and beta cell secretion have been only recently investigated (Guha et al., 2003; Friess et al., 2003; Kisfalvi et al., 2005; Elghazi et al., 2006; Beraud-Dufour et al., 2010).

In addition, we will present the role of a NT-related peptide, named xenin, in the regulation of glucose homeostasis. Xenin is a peptide of 25 amino acids discovered in 1992 (Feurle et al., 1992), and which is synthesized from a precursor of 35 amino acids (Hamscher et al., 1996). Intriguingly, the sequence of this precursor is 100% identical to the N-terminal sequence of the mammalian coat protein α (α-COP; Chow and Quek, 1996), a cytoplasmic protein which cannot be released.

# **NT AND METABOLISM**

The energetic balance is regulated by the control of satiety, a domain which has evoluted following the discovery of leptin. This peptide is secreted by adipocytes and acts as a hormone directly on specific hypothalamic areas involved in the control of food intake (Jequier, 2002). Interestingly, several works reported that the effects of leptin are controlled by NT expressing neurons (Cui et al., 2006). Indeed, the anorectic effect of leptin is impaired in NTSR1-deficient mice (Kim et al., 2008), suggesting that the complex NT-NTSR1 is crucial for the action of leptin. Moreover, Leinninger et al. (2011) published an elegant demonstration showing that the action of leptin via NT neurons controls orexin release, the mesolimbic dopamine system and energy balance. In conclusion, the control of food intake is mediated by leptin on NT expressing neurons through NTSR1.

The other important aspect of the neurotensinergic system in the periphery concerns the control of nutrient absorption. Rapidly following its discovery, it was observed that the level of circulating NT increased several minutes after a meal, and this increase was more important when the food was enriched with fatty acid (Leeman and Carraway, 1982). Recently, Gui and Carraway (2001) completed this observation by demonstrating that NT acts as a hormone released from intestine following ingestion of fat, and facilitates lipid digestion by stimulating pancreatic secretion. It was also demonstrated that NT enhanced taurocholate absorption from proximal rat small intestine indicating a role in the regulation of enterohepatic circulation. This effect is largely mediated by the release of mast cell mediators, and is regulated by NO (Gui and Carraway, 2004). In conclusion, NT acting as a neurotransmitter or as a hormone, regulates food intake (satiety) and lipids absorption indicating a general role for NT in the regulation of energy balance and in the control of homeostasis.

# **NT AND PANCREAS**

Immunoreactive-NT (IR-NT) has been detected in plasma extracts and plasma IR-NT elevation occurs in response to nutrient stimuli (Rosell and Rokaeus, 1979). This suggested that NT may influence pancreatic regulation by a hormonal mechanism. IR-NT is present in the pancreas (Fernstrom et al., 1981) and a direct (paracrine) influence on islet hormone secretion was suggested. The presence of NT in the circulation and in the pancreas can be correlated to its effects on insulin and glucagon release (Dolais-Kitabgi et al., 1979) and on the growth of ductal adenocarcinoma of the pancreas (Reubi et al., 1998). Actually, we know that the action of NT on islet secretion is the consequence of the expression of the three NTSRs in normal endocrine pancreas (Beraud-Dufour et al., 2010). By contrast, NTSRs are not expressed in normal exocrine pancreas, their expression being linked with the development of tumors (Kitabgi, 2002; Myers et al., 2009).

Administration of NT increased pancreatic weight, DNA, RNA, and protein contents as well as lipase concentration (Feurle et al., 1987). The proliferative effect of NT on the pancreas has been also demonstrated by Wood et al. (1988). The role of NT and other gastrointestinal hormones like cholecystokinin or gastrin releasing peptide in the growth of normal and neoplastic tissues, including pancreas, has been well documented in a review article (Thomas et al., 2003).

"fendo-03-00143" — 2012/11/23 — 18:50 — page 2 — #2

## **NT AND PANCREATIC CANCER GROWTH**

It is now well established that NT receptors are expressed in exocrine pancreatic tumors and chronic pancreatitis whereas these receptors were not found in normal exocrine pancreatic tissues (Reubi et al., 1998; Wang et al., 2000). Numerous studies have demonstrated that NT stimulates mitogenic signaling pathways and DNA synthesis in human pancreatic cancer cell lines including PANC-1 and MIA PaCa-2 cells (Herzig et al., 1999; Ehlers et al., 2000; Ryder et al., 2001; Guha et al., 2002; Falciani et al., 2010). These growth effects are mediated through NTSR1 stimulation (Iwase et al., 1997). However, a recent study concluded that NT-induced migration of pancreatic ductal adenocarcinoma cells *in vitro* occurs via NTSR3/sortilin pathways (Mijatovic et al., 2007).

#### **NT AND ENDOCRINE PANCREAS**

The first *in vivo* observation that NT displayed a role in glucose homeostasis was performed in the rat where the peptide produced hypoinsulinemia and consequently hyperglycemia (Brown and Vale, 1976). Then after, a more detailed work performed on rat islets of Langerhans demonstrated that NT regulates endocrine pancreatic hormones release (Dolais-Kitabgi et al., 1979). In this study, NT was shown to stimulate insulin and glucagon release at low glucose concentration whereas at high glucose or arginine levels, NT inhibits the release of both peptides. However, in another study performed on isolated neonatal rat islets, NT was unable to alter insulin secretion under high glucose concentration (Fujimoto, 1981). This result may be due to the absence of some NTSRs at this stage of development (Zsurger et al.,1992). The involvement of NT in the regulation of glucose homeostasis has been confirmed in a clinical study dealing with healthy elderly and young subjects. It was demonstrated that in addition to hyperinsulinemia and hypergastrinemia, the postprandial responses for NT were significantly higher in the aged subjects (Arnalich et al., 1990). However, no abnormality in the content of NT was detected in human diabetes, as demonstrated in insulin-dependent diabetic patients and in lean or obese non-insulin-dependent diabetic patients (Service et al., 1986).

# **MOLECULAR MECHANISMS OF NT ACTION ON BETA CELLS PROTECTION OF BETA CELLS**

Although there is a lack of evidence for a role of NT in diabetes, there are convincing data for its implication in glucose homeostasis. Surprisingly, no more important studies have been carried out to identify the NT receptor(s) involved in the effects observed. Only very recently our studies investigated the molecular mechanisms involved in the activation of the signaling pathways responsible for NT functions in cultured beta cells. We first demonstrated that all the identified NTSRs are expressed in murine Langerhans islets and in beta cell lines (Coppola et al., 2008; Beraud-Dufour et al., 2009). We demonstrated that NT efficiently protects beta cells from external cytotoxic agents (staurosporine, IL-1beta) through the PI3 kinase pathway (Coppola et al., 2008; **Figure 1**). The NTSR2 partial agonist levocabastine exerts a protective effect similar to that of NT whereas the NTSR1 antagonist SR48692 is unable to block the effect of NT suggesting the predominant involvement of the NTSR2 in the protective action of NT on beta cells. Moreover, we showed that this effect is mediated by NTSR2 via the protein complex formed between the GPCR NTSR2 and the type I receptor NTSR3/sortilin (Beraud-Dufour et al., 2009; **Figure 1**). In this case, the role of NTSR3/sortilin has been postulated to direct NTSR2 to its functional plasma membrane compartment as shown for NTSR1 in HT29 cells (Martin et al., 2002) and for the potassium channel TREK-1 in neurons (Mazella et al., 2010).

The protective action of NT on beta cells is of importance since in diabetes beta cell death is generally the consequence of prolonged hyperglycemia and/or hyperlipidemia. Therefore taking into account that NT is released in the circulation after a meal, in particular after lipid absorption, the neurotensinergic system may save endocrine pancreas as previously demonstrated for glucagon-like peptide 1 (for review Desgraz et al., 2011).

# **REGULATION OF INSULIN SECRETION**

From experiments carried out on rat beta cell lines, we confirmed the dual action of NT, which is able to increase insulin secretion at low glucose concentration and also to decrease the glucose-induced insulin release (Dolais-Kitabgi et al., 1979; Beraud-Dufour et al., 2010). At the cellular level NT, as well as the NTSR2 selective ligand levocabastine, rapidly and transiently increases the intracellular concentration of calcium in Ins1-E cell line. NT-evoked calcium regulation involves PKC and the translocation of PKCα and PKCε to the plasma membrane. A similar response is obtained with levocabastine, indicating that NTSR2 triggers the effect of NT both on insulin secretion and calcium concentration (**Figure 1**). This result is in total agreement with a study from 1978 in which the hyperglycemia action of NT was shown to be mediated through histamine since blockers of both H1 and H2 histamine receptors were able to inhibit the effect of NT (Nagai and Frohman, 1978). Indeed, it is important to remember that levocabastine, originally developed against H1 histamine receptors (Kitabgi et al., 1987), is a selective ligand of NTSR2 (Chalon et al., 1996; Mazella et al., 1996). Moreover, we demonstrated that the propeptide issued from the maturation of the precursor form of NTSR3/sortilin (Munck Petersen et al., 1999) triggers an increase of intracellular calcium and insulin secretion as observed for NT and levocabastine (Beraud-Dufour et al., 2010). The propeptide acts as an antagonist of NTSR3/sortilin (Martin et al., 2003), however it does not block the NT effect. This

"fendo-03-00143" — 2012/11/23 — 18:50 — page 3 — #3

#### **FIGURE 1 | Neurotensin receptors signaling in beta cells.**

Two G protein-coupled receptors, NTSR1 and NTSR2, and one type I receptor, NTSR3, are expressed in beta cells. In one hand, the binding of NT to the complex NTSR2/NTSR3 leads to the stimulation of phospholipase C to enhance the intracellular concentration of calcium responsible for the secretion of insulin. In a second hand, the interaction of NT with the same receptor complex activates the PI3 kinase, resulting in the phosphorylation of Akt to protect beta cells from the apoptosis induced by cytotoxic external agents. Although NTSR1 is expressed in beta cells, an unusual positioning or the absence of the receptor at the plasma membrane may explain the absence of interaction with NT. The propeptide (PE) which is released from the precursor form of NTSR3 displays agonist activity to increase intracellular calcium concentration.

suggests that NTSR3/sortilin may be also involved, in combination with NTSR2, in the action of NT. We know that NTSR2 and NTSR3/sortilin form heterodimers in these beta cells (Beraud-Dufour et al., 2009). However, we do not know the exact role of NTSR3 in this complex. NTSR3 may regulate the membrane expression of NTSR2, as reported for NTSR1 (Martin et al., 2002) and the two pore potassium channel TREK-1 (Mazella et al., 2010), and so only NTSR2 selective ligand may activate the receptor complex to trigger the effect. NTSR3/sortilin selective ligands (i.e., the propeptide) may also contribute to the final activation. Further studies carried out on NTSR KO animal models would be crucial for a better understanding of the role of each receptor on insulin secretion.

Interestingly, xenin, a peptide related to NT (Feurle et al., 1992) and co-secreted with glucose-dependent insulinotropic polypeptide (GIP) from intestinal K-cells (Anlauf et al., 2000), regulates glucose homeostasis and potentiates the action of GIP on glucosemediated insulin release (Taylor et al., 2010; Wice et al., 2010). Xenin enhances GIP-mediated insulin release by a mechanism which does not involve direct action on beta cells. Rather, the effect of xenin appears to be mediated by activating non-ganglionic cholinergic neurons that innervate islets (Wice et al., 2010). The effect of xenin in combination with molecules regulating insulin secretion rates is observed only in humans with normal or impaired glucose tolerance but not with type 2 diabetes (Wice et al., 2012). However, although xenin has been suggested to interact with NTSR1 in guinea pig enteral smooth muscles (Feurle et al., 2002) and in the effect of the peptide in food intake (Kim and Mizuno, 2010), evidence for non-neurotensin receptor-mediated effects of xenin has also been documented in rat intestine (Heuser et al., 2002). In the absence of demonstration that xenin interacts directly with one of the three known NTSRs, the effects of this peptide could be mediated by a system which is, at least partly, independent from the neurotensinergic system for the regulation of insulin secretion.

# **CONCLUSION AND FUTURE**

From the overall data obtained on pancreatic beta cells and islets, it is clear that although the three NT receptors are expressed, the effects of NT involve both NTSR2 and NTSR3/sortilin but not NTSR1 (**Figure 1**). This is intriguing since the majority of the actions of NT both in the brain and in the periphery involves, at least partly, NTSR1, a receptor which was always shown to be expressed and functional at the plasma membrane of NT target cells. One possible explanation of its lack of implication in beta cells could be that although NTSR1 is detected by PCR and Western blot, the protein is absent or not correctly sorted at the plasma membrane. One of the arguments in favor of this hypothesis is that the binding of iodinated NT measured on membrane homogenate from beta cells is totally displaced by levocabastine (personal observation). Unfortunately, in the absence of efficient antibodies directed against NTSR1 for immunocytochemistry, sub-cellular localization of protein expression appears difficult to be correctly investigated.

On another hand, the presence of NTSR3/sortilin, as well as the effect of the propeptide on the intracellular concentration of calcium in beta cells, may be also of importance. Indeed, the role of NTSR3/sortilin as a sorting protein and its ability to be translocated to the plasma membrane upon activation by NT (Chabry et al., 1993; Navarro et al., 2001) where the propeptide can be released could serve as a new system of regulation of endocrine hormone release. Moreover, NTSR3/sortilin is also expressed on adipocytes and skeletal muscles, targets of insulin, in which its translocation to the cell surface has been also demonstrated (Lin et al., 1997; Shi and Kandror, 2007). Here again, the propeptide can be released in the circulation and then can

"fendo-03-00143" — 2012/11/23 — 18:50 — page 4 — #4

regulate by feedback the secretion of insulin by interacting with NTSR3/sortilin and TREK-1 (**Figure 2**). Taken together, these observations refine the regulation scheme of glucose homeostasis through NT and its receptors and will lead to novel research areas with new components and new functions like the propeptide release and action.

#### **REFERENCES**


a levocabastine-sensitive neurotensin


# **ACKNOWLEDGMENTS**

This work was supported by the Centre National de la Recherche Scientifique and by grants from the Plan National de la Recherche sur le Diabète, INSERM (grant no. PNRD0701) and Association pour la Recherche sur le Diabète (ARD). We thank Franck Aguila for artwork.


"fendo-03-00143" — 2012/11/23 — 18:50 — page 5 — #5


"fendo-03-00143" — 2012/11/23 — 18:50 — page 6 — #6


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 24 September 2012; paper pending published: 05 October 2012; accepted: 05 November 2012; published online: 26 November 2012.*

*Citation: Mazella J, Béraud-Dufour S, Devader C, Massa F and Coppola T* *(2012) Neurotensin and its receptors in the control of glucose homeostasis. Front. Endocrin. 3:143. doi: 10.3389/ fendo.2012.00143*

*This article was submitted to Frontiers in Neuroendocrine Science, a* *specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Mazella, Béraud-Dufour, Devader, Massa and Coppola. This is an open-access article distributed under the terms of the Creative Commons* *Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

"fendo-03-00143" — 2012/11/23 — 18:50 — page 7 — #7

# Neuropeptide W

#### *Fumiko Takenoya1,2, Haruaki Kageyama1,3, Satoshi Hirako1, Eiji Ota1, Nobuhiro Wada1, Tomoo Ryushi 1,4 and Seiji Shioda1 \**

<sup>1</sup> Department of Anatomy, Showa University School of Medicine, Tokyo, Japan

<sup>2</sup> Department of Exercise and Sports Physiology, Hoshi University School of Pharmacy and Pharmaceutical Science, Tokyo, Japan

<sup>3</sup> Faculty of Health Care, Kiryu University, Gunma, Japan

<sup>4</sup> Department of Sports and Health Science, Daito Bunka University, Saitama, Japan

#### *Edited by:*

Australia

Jae Y. Seong, Korea University, South Korea

*Reviewed by:* Willis Samson, Saint Louis University, USA Herbert Herzog, Garvan Institute,

#### *\*Correspondence:*

Seiji Shioda, Department of Anatomy, Showa University School of Medicine, 1-5-8 Hatanodai, Shinagawa-ku, Tokyo 142-8555, Japan. e-mail: shioda@med.showa-u.ac.jp

Neuropeptide W (NPW), which was first isolated from the porcine hypothalamus, exists in two forms, consisting of 23 (NPW23) or 30 (NPW30) amino acids. These neuropeptides bind to one of two NPW receptors, either NPBWR1 (otherwise known as GPR7) or NPBWR2 (GPR8), which belong to the G protein-coupled receptor family. GPR7 is expressed in the brain and peripheral organs of both humans and rodents, whereas GPR8 is not found in rodents. GPR7 mRNA in rodents is widely expressed in several hypothalamic regions, including the paraventricular, supraoptic, ventromedial, dorsomedial, suprachiasmatic, and arcuate nuclei. These observations suggest that GPR7 plays a crucial role in the modulation of neuroendocrine function. The intracerebroventricular infusion of NPW has been shown to suppress food intake and body weight and to increase both heat production and body temperature, suggesting that NPW functions as an endogenous catabolic signaling molecule. Here we summarize our current understanding of the distribution and function of NPW in the brain.

**Keywords: food intake, hypothalamus, brain mapping, peptides, CPCRs**

# **INTRODUCTION**

In 1995, O'Dowd et al. (1995) used oligonucleotides based on the opioid receptor as well as the structurally related somatostatin receptor to identify two genes, GPR7 (NPBWR1) and GPR8 (NPBWR2), which were predicted to encode two G proteincoupled receptors (GPCRs) in the human brain. NPBWR1 mRNA expression is demonstrated in the human and rodent brain, whereas the NPBWR2 gene is detected in the human and rabbit, but not the rodent, brain (Lee et al., 1999). In 2002, Shimomura et al. (2002) identified the endogenous ligand for NPBWR1- 2 by exposing Chinese Hamster Ovary (CHO) cells to porcine hypothalamic extracts with monitoring changes in the level of cAMP. Moreover, when cell lines that expressed either NPBWR1 or NPBWR2 were incubated with the hypothalamic extracts, forskolin-induced cAMP production was inhibited. These receptors were coupled to the heterotrimeric Gi protein-mediated receptors (Shimomura et al., 2002). Further structural analysis of the ligands responsible for the inhibition of cAMP production led to the identification of a novel peptide, neuropeptide W (NPW). Shimomura et al. identified mature peptide sequences of both 23 and 30 residues from porcine, rat, and human prepro NPW (**Figure 1**). NPW is named after the tryptophan residues appearing at both its N- and C-terminals in its two mature forms: NPW30 (the 30 amino acid form) and NPW23 (consisting of 23 amino acids, which are identical to the N-terminal 23 residues of NPW30) (**Figure 1**). At the same time, the human gene for NPW was identified by Tanaka et al. (2003), who also cloned the mouse gene and reported restricted expression in specific neurons in the mid-brain and brainstem. Similarly, Brezillon et al. (2003) identified the shorter 23 amino acid form of the peptide, which

they termed L8 that was derived by proteolytic processing from the longer peptide. The prepropeptide was designated prepro-G protein-coupled receptor 8 ligand.

# **DISTRIBUTION OF NPW**

Based on RT-PCR analysis, Brezillon et al. (2003) have reported that NPW mRNA is highly expressed in the substantia nigra and spinal cord, and moderately expressed in the hippocampus, amygdala, hypothalamus corpus callosum, cerebellum, and dorsal root ganglia in the human central nervous system. In rodents, *in situ* hybridization histochemistry has revealed that NPW mRNA is expressed in a few restricted brain regions, including the rat periaqueductal gray (PAG), Edinger-Westphal nucleus (EW), and dorsal raphe nucleus (Baker et al., 2003; Tanaka et al., 2003; Kitamura et al., 2006), while Kitamura et al. (2006) have reported that it is confined to specific nuclei in the rat midbrain and brainstem. However, based on RT-PCR analysis, we have reported that NPW mRNA is expressed in the rat hypothalamic paraventricular nucleus (PVN), ventromedial nucleus (VMH), arcuate nucleus (ARC), and lateral hypothalamus (LH) (Takenoya et al., 2010), with another study also reporting the expression of NPW in various areas of the rat brain (Dun et al., 2003). Immunohistochemical studies have shown that NPW-like immunoreactive (NPW-LI) neuronal cell bodies mainly observed in the hypothalamic areas, ARC and posterior pituitary gland, with a lower level in the PVN. Interestingly, NPW-LI cells appear to be more numerous in the male than the female hypothalamus (Dun et al., 2003). In another study, Kitamura et al. (2006) reported a heavy localization of NPW-LI cell bodies in the midbrain, including the PAG and EW. Furthermore, we first

identified the presence of NPW-LI cell bodies and their processes in the PVN, VMH, and amygdala at the electron microscopic level (Takenoya et al., 2009). Moreover, NPW-LI nerve fibers were abundantly distributed in the midbrain and limbic system, including the CeA and BST, suggests that NPW may play a role in the regulation of fear and anxiety as well as in feeding behavior (Dun et al., 2003; Hondo et al., 2008; Takenoya et al., 2009, 2010).

In the peripheral tissues, NPW is expressed in the trachea, as well as in lymphoblastic leukemia in the fetal kidney and colorectal adenocarcinoma (Brezillon et al., 2003). Rat adrenocortical cells are also shown to produce NPW (Hondo et al., 2008), as have noradrenalin-containing cells in the rat adrenal medulla (Seki et al., 2008) and gastric antral G cells in rats and mice (Mondal et al., 2006), with expression of NPW in the rat stomach mucosa being regulated by nutritional status, glucocorticoids, and thyroid hormones (Caminos et al., 2008). Hochol et al. (2006) have reported expression of NPW in the thyroid and parathyroid glands, pancreatic islets, adrenal glands, ovary, and testis of the rat, while Rucinski et al. (2007) have demonstrated NPW immunoreactivity in all of the cells of the pancreatic islets, including the A, B, and D cells, and PP cells. In contrast, Dezaki et al. (2008) found NPW immunoreactivity in the B cells, but not in the A or D cells. In addition, NPW mRNA is expressed in the urogenital system, including the kidney, testis, uterus, ovary, and placenta (Fujii et al., 2002). Based on RT-PCR analysis, we have confirmed the presence of NPW mRNA in the pituitary gland, adrenal gland, and stomach (Seki et al., 2008). These observations suggest that NPW may play an important role in the regulation of the endocrine system in response to stress, as well as in the activation of the hypothalamus-pituitary-adrenal (HPA) axis (Niimi and Murao, 2005; Seki et al., 2008).

#### **DISTRIBUTION OF NPBWR1-2**

In humans, RT-PCR analysis has demonstrated that NPBWR1 mRNA is highly expressed in the amygdala, hippocampus, neocortex, and hypothalamus (Lee et al., 1999). *In situ* hybridization histochemical studies have demonstrated that NPBWR1 mRNA is present in the rat hypothalamus, including the ARC, VMH, PVN, and DMH, with Ishii et al. (2003) reported that NPBW1 knockout mice exhibit hyperphagia and develop adult-onset obesity. Singh et al.(2004) used [125I] -NPW receptor autoradiography and demonstrated a significant expression of NPBW1 in the rat amygdala and hypothalamus, as well as in the BST, medial preoptic area (MPA), PAG, superficial gray layer of the superior colliculus, and subfornical organ. In general, NPBWR1 is most commonly expressed at high levels in the amygdala (Singh et al., 2004; Kitamura et al., 2006; Skrzypski et al., 2012). Although the BST shows the highest level of NPBWR1 expression in small mammals, this phenomenon has not been demonstrated in humans. Kitamura et al. (2006) have reported that NPBWR1 is most abundantly expressed in the rat CeA and BST, which may indicate that NPBWR1 is involved in the regulation of stress, emotion, fear, and anxiety. On the other hand, NPBWR1-2 mRNAs are also localized in the pituitary and adrenal glands (in both the adrenal cortex and adrenal medulla) (Mazzocchi et al., 2005). These observations suggest that NPBWR1-2 may be involved in responding to stress via the HPA axis (Mazzocchi et al., 2005; Niimi and Murao, 2005). Ziolkowska et al. (2009) recently examined the expression and function of the NPW, NPB, and NPBWR1 system in cultured rat calvaria osteoblast-like cells, with their results suggesting a direct effect on this cell proliferation. NPB has been identified in larger mammals, as well as in rabbits, but has not been described in either rats or mice. RT-PCR analysis has shown that NPBWR2 mRNA is highly expressed in the human amygdala, hippocampus, pituitary gland, adrenal gland, and testis, as well as in the cortical cells in the adrenal gland (Mazzocchi et al., 2005).

# **REGULATION OF FEEDING AND ENERGY METABOLISM BY NPW**

NPBWR1 knockout mice are hyperphagic and show decreased energy expenditure, suggesting that NPW may act as a modulator of feeding. icv infusion of NPW in male rats has been shown to increase food intake during the first 2 h in the light phase (Shimomura et al., 2002). Similarly, Levine et al. (2005) have reported that injection of NPW into the PVN increases food intake. These results suggest that NPW acts as an acute orexigenic peptide. However, Mondal et al. (2003) have reported that both forms of NPW suppress dark-phase and fasting-induced food intake, indicating that the effect of NPW on feeding differs depending on whether animals are maintained in a light or dark phase.

We have carried out a series of neuroanatomical studies to examine the neural relationship between NPW and other neuropeptides involved in the regulation of feeding. Very close neuronal interactions were observed between NPW-containing nerve fibers and orexin- or melanin-concentrating hormonecontaining neuronal cell bodies and nerve fibers in the rat brain (Takenoya et al., 2005), while Levine et al. (2005) demonstrated that c-Fos expression was induced in orexin-containing neurons in the perifornical region of the LH after intracerebroventricular (icv) infusion of NPW. Interestingly, we also identified NPW-LI cell bodies in the VMH, which is known as a center for satiety (Takenoya et al., 2010). Leptin acts on neurons in the VMH, thereby reducing food intake, and Date et al. (2010) have recently reported that NPW-LI neurons and leptin receptors are colocalized in this region of the brain. NPW expression is also significantly up-regulated in *ob/ob* and *db/db* mice. Therefore, NPW may play important roles in feeding and energy metabolism, functioning as a substitute for leptin (Date et al., 2010) (**Figure 2**). Furthermore, NPW reduces food intake via the melanocortin-4-receptor signaling pathway, suggesting that it may activate POMCcontaining neurons and inhibit NPY-containing neurons to control feeding regulation in the ARC (Date et al., 2010) (**Figure 2**).

Very recently, Skrzypski et al. (2012) have demonstrated that NPB and NPW regulate the expression and secretion of leptin and resistin, and increase lipolysis in isolated rat adipocytes. When NPW was administered to rats, the authors could not detect enhanced locomotor activity, but did observe increased O2 consumption and increased CO2 production, as well as an increase in body temperature (Mondal et al., 2003). Interestingly, Mondal et al. (2006) have reported that the levels of NPW isolated from rat stomach antral cells are lower in fasted animals, increasing once the animals have been re-fed. In contrast, female NPBWR1 −*/*− mice do not display hyperphagic activity compared with wild-type mice (Ishii et al., 2003). In addition, Dun et al. (2003) reported the existence of differences between male and female rats with respect to the distribution of NPW.

# **NEUROENDOCRINE FUNCTION OF NPW**

Immunohistochemical studies have reported that NPBWR1 is expressed in the PVN, pituitary gland, and adrenal medulla in the human, mouse and rat (O'Dowd et al., 1995; Lee et al., 1999; Brezillon et al., 2003; Tanaka et al., 2003), particularly in the parvocellular subdivisions of the PVN and the posterior pituitary. However, NPW has not been reported to influence the release of other anterior pituitary hormones. These neuroendocrine effects of NPW are not directly mediated through NPWBR1 on the pituitary gland cells, but may occur indirectly through controlling the release of the hypothalamic hormone,

**Table 1 | Physiological effect of NPW in Rat.**


The plus or minus indicates stimulatory (↑) or inhibitory (↓) effects, respectively.

corticotrophin-releasing factor (CRF) (Brezillon et al., 2003; Samson et al., 2004). Colocalization of NPW with urocortin, a stress-related neuropeptide of the CRF family, has also been demonstrated in the EW.

On the other hand, icv infusion of NPW23 stimulates prolactin release in the rat (Shimomura et al., 2002), and an *in vitro* study has reported that log molar NPW23 concentrations, significantly alter prolactin, growth hormone and ACTH release from dispersed rat anterior pituitary cells (Baker et al., 2003). In addition, *in vivo* studies have revealed that icv infusion of NPW23 stimulates an increase in plasma prolactin levels (Baker et al., 2003). icv infusion of NPW23 also results in a significant, elevation in plasma corticosterone levels, suggesting that NPW plays a role in the hypothalamic response to stress. However, growth hormone levels in plasma are inhibited by icv infusion of this peptide. These findings suggest that NPW is the endogenous ligand for GPR7 and/or GPR8 and acts as a mediator of neuroendocrine function (Shimomura et al., 2002; Baker et al., 2003). Moreover, Taylor et al. (2005) have reported that the infusion of NPW activates the HPA axis, as demonstrated by changes in plasma corticosterone levels in conscious rats; NPW increases plasma corticosterone levels but does not stimulate the release of oxytocin or vasopressin in the peripheral circulation, or alter blood pressure or heart rate. Furthermore,

#### **REFERENCES**


Caminos, J. E., Bravo, S. B., Garcia-Rendueles, M. E., Ruth Gonzalez, C., Garces, M. F., Cepeda, L. A., et al. (2008). Expression of neuropeptide W in rat stomach mucosa: regulation by nutritional status, glucocorticoids and thyroid hormones. *Regul. Pept.* 146, 106–111.

Date, Y., Mondal, M. S., Kageyama, H., Ghamari-Langroudi, M., Takenoya, F., Yamaguchi, H., et al. (2010). Neuropeptide W: an anorectic peptide regulated by leptin and metabolic state. *Endocrinology* 151, 2200–2210.

icv infusion of a CRF antagonist does not significantly reduce corticosterone levels, although CRF antagonist pretreatment significantly reduces the capacity of centrally administered NPW to increase corticosterone levels (Taylor et al., 2005). Using electrophysiological studies with whole-cell patch recording of hypothalamic slice preparations, Taylor et al. (2005) have also shown that NPW depolarizes and increases the spike frequency of the majority of putative neuroendocrine PVN neurons. Furthermore, the response of these cells in the presence of tetrodotoxin confirmed that NPW was acting post-synaptically (Taylor et al., 2005).

Niimi and Murao (2005) have reported that double immunostaining for both NPW and c-Fos is significantly increased in response to stress. Immobilization stress increases parasympathetic outflow, whereas cold stress is primarily considered to increase sympathetic outflow, suggesting that NPW may participate in the regulation of both the sympathetic and parasympathetic branches of the autonomic nervous system. In the PVN, vasopressin- expressing neurons are known to be involved in the production of hormonal outputs in response to endocrine and autonomic stress (Ulrich-Lai and Herman, 2009). Kawasaki et al. (2006) have reported that centrally administered NPW activates magnocellular neurosecretory neurons in the SO and PVN, and significantly increases plasma arginine-vasopressin and plasma oxytocin levels. Price et al. (2008) have reported that the inhibition of growth hormone release due to the central effects of NPW results from activation of arcuate somatostatin neurons, which could produce inhibition of neurons expressing growth hormone releasing hormone. These findings suggest that endogenous NPW may play a physiologically relevant role in the neuroendocrine response to stress in the brain (**Table 1**).

# **ACKNOWLEDGMENTS**

This study was supported in part by the Ministry of Education, Culture, Sports, Science and Technology of Japan via a Grant-in-Aid for Scientific Research (C) to Fumiko Takenoya (23500863) and to Haruaki Kageyama (21590222). Support was also received from the Ministry of Education, Culture, Sports, Science and Technology of Japan via a Grant-in-Aid for Exploratory Research to Seiji Shioda (21659059) and by a High-Technology Research Center project grant from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to Seiji Shioda).


Y., et al. (2002). Identification of a neuropeptide modified with bromine as an endogenous ligand for GPR7. *J. Biol. Chem.* 277, 34010–34016.


Effects of neuropeptides B and W on the rat pituitary-adrenocortical axis: *in vivo* and *in vitro* studies. *Int. J. Mol. Med.* 19, 207–211.


et al. (2012). Neuropeptide W stimulates adrenocorticotrophic hormone release via corticotrophinreleasing factor but not via arginine vasopressin. *Endocr. J.* 59, 547–554.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 October 2012; paper pending published: 13 November 2012; accepted: 06 December 2012; published online: 21 December 2012. Citation: Takenoya F, Kageyama H,*

*Hirako S, Ota E, Wada N, Ryushi T and Shioda S (2012) Neuropeptide W. Front. Endocrin. 3:171. doi: 10.3389/ fendo.2012.00171*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Takenoya, Kageyama, Hirako, Ota, Wada, Ryushi and Shioda. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *Joëlle Cohen-Tannoudji\*, Charlotte Avet, Ghislaine Garrel, Raymond Counis and Violaine Simon*

Equipe Physiologie de l'Axe Gonadotrope, Unité de Biologie Fonctionnelle et Adaptative, CNRS-EAC 4413, Sorbonne Paris Cité, Université Paris Diderot-Paris 7, Paris, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Jong-Ik Hwang, Korea University, South Korea Yves Combarnous, Centre National de la Recherche Scientifique, France

#### *\*Correspondence:*

Joëlle Cohen-Tannoudji, Equipe Physiologie de l'Axe Gonadotrope, Unité de Biologie Fonctionnelle et Adaptative, CNRS-EAC 4413, Sorbonne Paris Cité, Université Paris Diderot-Paris 7, Case courrier 7007, 4 rue Marie-Andrée Lagroua-Weill-Hallé, 75013 Paris, France. e-mail: joelle.cohen-tannoudji@univparis-diderot.fr

The gonadotropin-releasing hormone (GnRH) pulsatile pattern is critical for appropriate regulation of gonadotrope activity but only little is known about the signaling mechanisms by which gonadotrope cells decode such pulsatile pattern. Here, we review recent lines of evidence showing that the GnRH receptor (GnRH-R) activates the cyclic AMP (cAMP) pathway in gonadotrope cells, thus ending a long-lasting controversy. Interestingly, coupling of GnRH-R to the cAMP pathway as well as induction of nitric oxide synthase 1 (NOS1) or follistatin through this signaling pathway take place preferentially under high GnRH pulsatility. The preovulatory surge of GnRH in vivo is indeed associated with an important increase of pituitary cAMP and NOS1 expression levels, both being markedly inhibited by treatment with a GnRH antagonist. Altogether, this suggests that due to its atypical structure and desensitization properties, the GnRH-R may continue to signal through the cAMP pathway under conditions inducing desensitization for most other receptors. Such a mechanism may contribute to decode high GnRH pulsatile pattern and enable gonadotrope cell plasticity during the estrus cycle.

**Keywords: pituitary, GnRH pulsatile pattern, GnRH receptor, gonadotrope cell signaling, cAMP pathway**

"fendo-03-00107" — 2012/8/30 — 13:35 — page 1 — #1

#### **INTRODUCTION**

Reproduction success depends on the coordinated synthesis and release of several hormones from the hypothalamus– pituitary–gonadal axis. Mammalian gonadotropin-releasing hormone (known as GnRH 1) mediates brain control of reproductive activity and, as a major actor of gonadotrope axis, has received considerable attention. The neurohormone GnRH is a decapeptide produced by a small number of scattered hypothalamic neurons whose cell bodies are primarily located in the preoptic area and basal hypothalamus. GnRH neurons project to the median eminence and secrete GnRH in a pulsatile fashion into the hypophysial portal vessels through which GnRH is transported to the anterior pituitary gland. The release of GnRH by hypothalamic neurons is influenced by numerous external and internal factors acting through central nervous pathways. The pattern of pulsatile GnRH secretion thus varies widely in both males and females depending on the reproductive status. GnRH acts *via* its receptor specifically expressed in gonadotrope cells to stimulate both synthesis and exocytosis of the two gonadotropins, luteinizing hormone (LH) and follicle-stimulating hormone (FSH). LH and FSH will in turn act on the gonads in a coordinated manner to initiate sexual maturation and regulate gonadal steroidogenesis and gametogenesis in both sexes. Gonadotrope hormones are complex endocrine signals constituted of non-covalently associated glycoprotein dimers. Each gonadotropin is composed of an alpha glycoprotein subunit common to LH, FSH, thyrotropin (TSH) and placental choriogonadotropin (for a few species) and a unique beta subunit.

The GnRH pulsatile pattern is critical for appropriate regulation of LH and FSH synthesis and secretion. Indeed, intermittent stimulation *in vivo* or *in vitro* that mimics the physiological pulsatile release of GnRH efficiently stimulates the secretion of gonadotropins. In contrast, a continuous pattern leads to desensitization of gonadotrope cells and this has been exploited by clinicians to suppress gonadotropin secretion (Lahlou et al., 1987). Furthermore, pulsatility of GnRH varies throughout the ovarian cycle and accounts for the differential secretion of LH and FSH. At mid-cycle, during proestrus, an abrupt and massive increase in GnRH pulsatility is responsible for gonadotropin surge and ovulation. Only little is known about the signaling mechanisms by which the pituitary gonadotrope cells decode GnRH pulse pattern. The aim of this article is to review the current knowledge on GnRH receptor (GnRH-R) coupling to the cyclic AMP (cAMP) signaling pathway in order to highlight its potential role in decoding high GnRH pulsatility.

#### **COUPLING OF THE GnRH RECEPTOR TO THE cAMP SIGNALING PATHWAY**

GnRH binds to a receptor belonging to the G protein-coupled receptor (GPCR) family with seven transmembrane domains connected by extracellular and intracellular loops. Agonist binding is mainly associated with a rapid Gq/11-mediated increase in phospholipase Cβ (PLCβ) activity, which will in turn initiate a wide array of signaling events. Hydrolysis of phosphatidylinositol 4,5 bisphosphate (PIP2) results in the formation of diacylglycerol

(DAG) and inositol trisphosphate (IP3). Rapid formation of IP3 induces calcium release from intracellular stores and, together with GnRH-stimulated calcium influx, accounts for calcium oscillations that trigger gonadotropin exocytosis. Elevation of calcium also activates the nitric oxide synthase (NOS) cascade (NOS1/NO/soluble guanylate cyclase), resulting in a rapid increase of cyclic GMP (cGMP) levels (Naor et al.,1980; Lozach et al.,1998). GnRH-induced DAG formation activates protein kinase C (PKC) isoforms belonging to the three known families of PKC (conventional, novel, and atypical), which mediate notably activation of mitogen-activated protein kinases (MAPK) cascades. PKC and MAPK signaling are crucial for the regulation of gonadotropin subunit gene expression (Thackray et al., 2009). Following a short time lag, GnRH also activates phospholipase D (PLD) and phospholipase A2 (PLA2). PKC favors the coupling of GnRH-R to PLD leading to a sustained second wave of DAG that may contribute to maintain PKC activation during prolonged stimulation by GnRH (Zheng et al., 1994). GnRH-mediated PLA2 activation generates arachidonic acid and its lipoxygenase products that have been implicated in GnRH-induced gonadotropin synthesis and release (Naor, 2009). The GnRH-R thus activates a wide array of signaling entities to regulate gonadotropin synthesis and release.

It has been clearly established that the cAMP/protein kinase A (PKA) pathway is essential for gonadotrope function. Indeed, cAMP analogs mimic most of the effects of GnRH as they enhance the release of newly synthesized LH and the expression of several key genes including those encoding LHβ and α subunits as well as GnRH-R and NOS1 (Starzec et al., 1989; Garrel et al., 2002; Horton and Halvorson, 2004). However, the ability of GnRH to induce cAMP production in gonadotrope cells as well as the involvement of cAMP in GnRH action has long remained debated. Early observation of Borgeat et al. (1975) showed that a prolonged exposure (3 h) of rat hemipituitaries to GnRH stimulates cAMP accumulation and this observation was confirmed soon after by others (Naor et al., 1975). Since then, studies performed on dispersed rat pituitary cell cultures did not evidence any cAMP production or stimulation of adenylylcyclase (AC) activity in response to GnRH (Theoleyre et al., 1976; Conn et al., 1979). In 1989, the pituitary AC-activating polypeptide (PACAP) was discovered based on its ability to strongly activate AC in rat pituitary cells (Miyata et al., 1989). This probably contributed to lower the interest paid to GnRH-R signaling through the cAMP pathway. Furthermore, no GnRHinduced cAMP production could be substantiated on the first gonadotrope cell line that was established in the same period of time, the αT3-1 cell line (Horn et al., 1991). Accordingly, using photolabeling experiments, Grosse et al. (2000) argued for the exclusive coupling of GnRH-R to Gαq/11 in αT3-1 cells. Up to the early 2000s, the coupling of GnRH-R to the cAMP pathway in gonadotrope cells thus remained elusive. The ability of the mammalian GnRH-R to stimulate the cAMP pathway was however reported in several heterologous systems such as GH3, COS-7, or HeLa cells (Arora et al., 1998; Lin et al., 1998; Oh et al., 2005), thus reinforcing the interest for such coupling. The key to moving forward was the development of novel models and technologies. The ability of GnRH to induce cAMP accumulation in gonadotrope cells was demonstrated by two different groups including ours (Liu et al., 2002; Lariviere et al., 2007) using the gonadotrope cell line LβT2, which is more fully differentiated than the αT3-1 cell line. cAMP accumulation was evidenced by biochemical and enzyme-linked immunosorbent assays (ELISAs) and was specifically induced by GnRH-R activation as shown by competition with a GnRH antagonist. Further analysis revealed that different mechanisms contribute to GnRH-R-induced cAMP accumulation. Indeed, based on photolabeling experiments and on cell-permeable peptides that uncouple the receptor from Gαs, it was shown that GnRH-R activates cAMP production *via* Gs recruitment (Liu et al., 2002). Additionally, we demonstrated, using selective pharmacological inhibitors as well as dominant negative mutants of PKC isoforms, that novel PKCδ and ε mediate GnRH-R activation of the cAMP pathway (Lariviere et al., 2007). In both studies, GnRH-induced increase in cAMP levels exhibited atypical features. Indeed, GnRH-induced maximal accumulation of cAMP levels was observed only after a lag of time of approximately 30 min and was delayed as compared to response to PACAP (Lariviere et al., 2007) or to the beta-adrenergic receptor agonist, isoproterenol (J. Cohen-Tannoudji, personal communication). Furthermore, maximal cAMP accumulation was weaker and this may explain why GnRH-R coupling to this signaling pathway has long remained controversial. This prompted us to reevaluate the coupling of GnRH-R to the cAMP signaling pathway in αT3-1 cells using a recently developed technology. As reported previously by Horn et al. (1991), we did not detect any significant increase in cAMP levels by ELISAs when cells were challenged with a GnRH agonist in contrast to response to PACAP or forskolin stimulation. We then took advantage of a sensitive technology based on the use of a cAMP biosensor (pGloSensorTM-22F cAMP Plasmid). αT3-1 cells were transfected with a plasmid encoding an engineered cAMP sensitive luciferase, which becomes active upon cAMP binding. Transfected cells were challenged with a GnRH agonist and the luminescence intensity was measured in real time in living cells. This strategy allowed us to demonstrate for the first time that GnRH-R couples to the cAMP pathway in αT3-1 cells (Avet et al., 2012). Indeed, GnRH dose-dependently increased luciferase activity reflecting cAMP production with an EC50 in the range of 1 nM and the increase was significantly inhibited by co-incubation with the GnRH antagonist, antide. Once the ability of GnRH to stimulate cAMP production was established in both gonadotrope cell lines, our efforts turned toward trying to understand whether this could also occurred in native gonadotrope cells. A massive pituitary cAMP level increase has been reported long ago to occur during the proestrus afternoon of the rat estrus cycle, coincidently with the GnRH preovulatory surge (Kimura et al., 1980). However, despite such temporal coincidence, the precise contribution of GnRH has never been elucidated. We reevaluated this contribution *in vivo* by administrating a potent GnRH antagonist, the ganirelix, to females at the evening of the diestrus II (Garrel et al., 2010). The marked decrease in pituitary cAMP levels provided evidence for a predominant role of GnRH in mediating the proestrus-specific rise of cAMP (**Figure 1**).

**"fendo-03-00107" — 2012/8/30 — 13:35 — page 2 — #2**

**FIGURE 1 | Profiles of serum LH and pituitary cAMP and NOS1 levels during the proestrus and following administration of a GnRH antagonist. (A)** Serum LH, pituitary cAMP, and pituitary NOS1 levels were determined by radioimmunoassay, ELISA, and western blot analysis, respectively.

#### **THE ATYPICAL STRUCTURE OF GnRH RECEPTOR IS ASSOCIATED WITH RESISTANCE TO RAPID DESENSITIZATION**

All mammalian pituitary GnRH-R (named type I GnRH-R) exhibit a striking structural feature, which is the lack of a cytoplasmic carboxyl-terminal tail. This carboxyl-terminal tail is present in all other GPCR including GnRH-R from non-mammalian vertebrates and is a target for kinases such as GRK (GPCR kinase). GRK-mediated phosphorylation promotes the binding of β-arrestins, which not only uncouple receptors from G proteins resulting in desensitization but also target many GPCR for internalization in clathrin-coated vesicles (Ferguson, 2001). Given the unique absence of cytoplasmic carboxyl-terminal tail, the initial rate of inositol phosphate accumulation is linear upon 10 min of GnRH stimulation showing that GnRH-R is resistance to rapid desensitization (Davidson et al., 1994). Accordingly, mammalian GnRH-R is not phosphorylated upon agonist stimulation and does not recruit β-arrestins, in contrast to nonmammalian GnRH-R (Willars et al., 1999; McArdle et al., 2002). Extensive analysis of GnRH-R internalization has been performed using biochemical approaches and fluorescent or radiolabeled GnRH in gonadotrope and heterologous cells. These studies indicated that mammalian GnRH-R is poorly internalized as compared to non-mammalian GnRH-R and that addition of

a functional intracellular carboxyl-terminal tail to the receptor significantly increased internalization rates and rapid desensitization. Whereas these studies were rather measuring ligand internalization, investigations conducted by Millar's group, using an epitope-tagged receptor allowed to trace the receptor itself. These studies established that GnRH-R exhibits a low level of constitutive internalization and does not undergo rapid agonistdependent internalization as compared to non-mammalian GnRH or the TSH-releasing hormone receptors (Pawson et al., 2008). The refractory state of gonadotrope cells under a sustained GnRH challenge is thus believed to occur through desensitization mechanism affecting downstream signaling entities such as Gαq/11, PKC isoforms, or IP3 receptors (Willars et al., 2001; Liu et al., 2003) rather than the receptor itself. Consequently, mammalian GnRH-R has the unique property amongst GPCR to face prolonged activation by its ligand.

#### **MECHANISMS CONTRIBUTING TO GnRH-R COUPLING TO THE cAMP PATHWAY**

One intriguing property of GnRH-R coupling to the cAMP pathway in LβT2 cells is that it appears to be dependent on GnRH stimulation pattern. We, indeed, observed that an intermittent stimulation of cells with GnRH, known to stimulate calcium release or MAPK activation, was ineffective in inducing cAMP

"fendo-03-00107" — 2012/8/30 — 13:35 — page 3 — #3

accumulation (Lariviere et al., 2007). Levels of cAMP were significantly increased only under sustained stimulation of GnRH-R suggesting that this pathway may be preferentially recruited under high frequency of GnRH pulses, in agreement with the proestrusspecific elevation of pituitary cAMP levels during rat ovarian cycle (Kimura et al., 1980; Garrel et al., 2010). This shows that GnRH-R maintains cAMP pathway activation under stimulations inducing desensitization of most GPCR. This atypical feature was recently confirmed using fluorescence resonance energy transfer reporters in LβT2 cells in which high pulse frequency of GnRH desensitized DAG and calcium but not cAMP activation (Tsutsumi et al., 2010).

Only few genes regulated by GnRH through the cAMP/PKA signaling pathway have been identified so far. Among them are those encoding follistatin, Nur77, PACAP, and NOS1 (Winters et al., 2007; Hamid et al., 2008; Grafer et al., 2009; Mutiara et al., 2009; Garrel et al., 2010). Interestingly, analysis of how GnRH regulates two of these genes has revealed that they are preferentially induced under high GnRH pulse frequency. Follistatin transcript and protein are increased by GnRH administered continuously or at fast frequency pulses either *in vivo* in rat or in a perifused pituitary cells (Besecke et al., 1996) and not at slow GnRH pulse frequencies. Because follistatin binds to and neutralizes activin, stimulation of FSHβ by GnRH, which requires activin is blocked. This is believed to explain why expression of FSHβ is preferentially stimulated under low frequency of GnRH pulses. In rat anterior pituitary gland, NOS1 is expressed specifically in gonadotrope and folliculostellate cells. Our group has demonstrated from experiments performed *in vivo* in rats and *in vitro* that NOS1 expression is regulated by GnRH through the cAMP pathway (Garrel et al., 1998; Lozach et al., 1998). This was shown in particular by taking advantage of a cell-permeant PKA inhibitor (PKI) peptide delivery in primary cultures of rat pituitary cells (Garrel et al., 2010). Interestingly, we observed that NOS1 is preferentially induced by prolonged GnRH treatment. Indeed, NOS1 expression remained at high levels for at least 48 h after treatment of rats with a long-lasting GnRH agonist, whereas, at the same time, LH and FSH secretion was desensitized. Accordingly, we demonstrated in perifused rat pituitary cells that NOS1 is not induced by GnRH delivered at a frequency of one pulse per hour, although this frequency is able to trigger a massive LH release. In contrast, NOS1 was maximally induced by a continuous GnRH stimulation, which suppressed gonadotropin release. Such a result is in agreement with *in vivo* data showing a marked increase of NOS1 expression at proestrus during the preovulatory GnRH surge, which is characterized by a very high GnRH pulse frequency (Lozach et al., 1998; **Figure 1**). Interestingly, treatment with the GnRH antagonist ganirelix not only blocked the proestrus-specific cAMP increase (see above) but also significantly reduced the rise of NOS1 expression levels (Garrel et al., 2010; **Figure 1**) further supporting the physiological relevance of GnRH-R coupling to the cAMP pathway. Altogether, experiments reported here, suggest that under conditions of high GnRH pulsatility, GnRH-R, which is atypically maintained at the cell surface, still interacts with its intracellular machinery and activates the cAMP/PKA pathway. This initiates the expression of a new set of target proteins among which are NOS1 and follistatin. The role of GnRH-dependent NOS1 induction at proestrus remains to be determined. The specific increase of NOS1 expression leads to a concomitant huge increase of pituitary cGMP levels at proestrus (Lozach et al., 1998). cGMP may regulate some cyclic nucleotidegated channels that we have characterized in pituitary gonadotrope cells (J. Cohen-Tannoudji, personal communication) and this is under investigation.

How is the preferential coupling of GnRH-R to the cAMP pathway selected? This question is still unresolved. GnRH-R is constitutively localized in low-density membrane microdomains such as rafts (Navratil et al., 2003). Intense stimulation by GnRH may alter the signaling platform associated with GnRH-R in rafts and hence favor receptor coupling to the cAMP pathway. Another potential mechanism may be related to GPCR-interacting proteins (GIP) that interact through receptor intracellular domains to regulate signaling efficacy and specificity (Bockaert et al., 2004). The idea that GIP regulate GnRH-R signaling is suggested by the fact that introduction in αT3-1 cells of synthetic peptides corresponding to intracellular domains of mammalian GnRH-R increases GnRH-R coupling to the inositol phosphate pathway (Shacham et al., 2005). We very recently identified a protein, the protein SET, as the first direct interacting partner of GnRH-R. Using cAMP and calcium biosensors in αT3-1 living cells combined with small interfering RNA directed against SET, we showed that SET significantly enhances GnRH-R coupling to the cAMP pathway. The mechanisms contributing to SET recruitment and the potential regulatory roles of SET on gonadotrope function are under current investigation.

# **CONCLUSION**

An important question in the field of reproductive endocrinology is to understand how the GnRH pulsatile pattern is decoded by pituitary gonadotrope cells. Due to its atypical structure and desensitization properties, GnRH-R and some of its signaling mechanisms are continuing to respond under massive GnRH stimulations. Such stimulations of gonadotrope cells by GnRH thus do not lead to general cell desensitization since some genes are induced in response to GnRH at high pulse frequencies. Variation in GnRH pulse profiles may lead to preferential activation of different signaling networks and there is now evidence that the cAMP/PKA pathway contributes to the decoding of high GnRH pulsatility. This emphasizes the unique feature of gonadotrope cells, which maintain part of their functional response under conditions of stimulation inducing desensitization of most GPCR. During the ovarian cycle, the GnRH preovulatory surge lasts much longer than the LH surge (Caraty et al., 2002). Maintenance of some gonadotrope responsiveness while gonadotropins are no longer secreted may contribute to gonadotrope cell plasticity during the estrus cycle. Identifying the mechanisms directing the preferential coupling of GnRH-R to the cAMP pathway is an exciting challenge. Insight into GnRH-R interacting partners and into the dynamics of their interactions with the receptor will undoubtedly help to better understand the plasticity of gonadotrope cell signaling.

# **ACKNOWLEDGMENT**

"fendo-03-00107" — 2012/8/30 — 13:35 — page 4 — #4

Authors are grateful to Richard Wargnier for his contribution to the illustration of the manuscript.

# **REFERENCES**


gonadotrophs, a process altered by desensitization and, indirectly, by gonadal steroids. *Endocrinology* 139, 2163–2170.


pituitary and hypothalamus during the rat estrous cycle and effects of administration of sodium pentobarbital in proestrus. *Endocrinology* 106, 631–635.


"fendo-03-00107" — 2012/8/30 — 13:35 — page 5 — #5

pituitary cells. *Biochem. Biophys. Res. Commun.* 164, 567–574.


stimulate the biosynthesis of luteinizing hormone polypeptide chains in a nonadditive manner. *Mol. Endocrinol.* 3, 618–624.


GnRH in mouse gonadotroph cell lines: evidence for a role for cAMP signaling. *Mol. Cell. Endocrinol.* 271, 45–54.

Zheng, L., Stojilkovic, S. S., Hunyady, L., Krsmanovic, L. Z., and Catt, K. J. (1994). Sequential activation of phospholipase-C and -D in agonist-stimulated gonadotrophs. *Endocrinology* 134, 1446–1454.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 20 July 2012; paper pending published: 02 August 2012; accepted: 15*

"fendo-03-00107" — 2012/8/30 — 13:35 — page 6 — #6

*August 2012; published online: 31 August 2012.*

*Citation: Cohen-Tannoudji J, Avet C, Garrel G, Counis R and Simon V (2012) Decoding high gonadotropin-releasing hormone pulsatility: a role for GnRH receptor coupling to the cAMP pathway? Front. Endocrin. 3:107. doi: 10.3389/ fendo.2012.00107*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Cohen-Tannoudji, Avet, Garrel, Counis and Simon. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Development and aging of the kisspeptin–GPR54 system in the mammalian brain: what are the impacts on female reproductive function?

# **Isabelle Franceschini 1,2,3,4\* and Elodie Desroziers 1,2,3,4†**

<sup>1</sup> UMR85 Physiologie de la Reproduction et des Comportements, Institut National de Recherche Agronomique, Nouzilly, France

<sup>2</sup> UMR7247, Centre National de la Recherche Scientifique, Nouzilly, France

<sup>3</sup> Université François Rabelais de Tours, Tours, France

4 Institut Français du Cheval et de l'Equitation, Nouzilly, France

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Manuel Tena-Sempere, University of Cordoba, Spain Vance Trudeau, University of Ottawa, Canada

#### **\*Correspondence:**

Isabelle Franceschini, Centre INRA de Tours, Unité de Physiologie de la Reproduction et des Comportements, UMR 7247 INRA/CNRS/Univ. Tours/IFCE, 37380 Nouzilly, France. e-mail: isabelle.franceschini@ tours.inra.fr

#### **†Present address:**

Elodie Desroziers, Groupe Interdisciplinaire de Génoprotéomique Appliquée – Neurosciences, University of Liege, Liege, Belgium.

#### **INTRODUCTION**

The G protein coupled receptor GPR54 was initially cloned based on its high homology with the galanin receptor (Lee et al., 1999). Its peptide ligands were purified from human placenta (Ohtaki et al., 2001) and identified as the proteolysis products of a 145 amino acid protein encoded by the tumor suppressor gene *Kiss1* (Lee et al., 1996). These ligands have been termed kisspeptins and share a 10 amino acid long sequence at their amidated C terminal, a sequence that is sufficient to bind and activate GPR54 (Ohtaki et al., 2001). The importance of the kisspeptin–GPR54 system in reproductive function came to light in 2003 when it was discovered that some human families displaying hypogonadotropic hypogonadism and absence or delay in puberty bore mutations in the *GPR54* gene (de Roux et al., 2003; Seminara et al., 2003).

*Kiss1* and *GPR54* have since been cloned from many mammalian species. Kisspeptin–GPR54 signaling predominantly acts at the level of the brain to control reproductive function (Oakley et al., 2009 for review). Kisspeptins are in fact the most potent gonadotropin-releasing hormone (GnRH) peptide secretagogue discovered to date (Messager et al.,2005). Kisspeptins are produced by two main populations of neurons, one within the preoptic area (POA) and one within the arcuate nucleus (ARC, or infundibulum in primates), that have been implicated in the regulations

The prominent role of the G protein coupled receptor GPR54 and its peptide ligand kisspeptin in the progression of puberty has been extensively documented in many mammalian species including humans. Kisspeptins are very potent gonadotropin-releasing hormone secretagogues produced by two main populations of neurons located in two ventral forebrain regions, the preoptic area and the arcuate nucleus.Within the last 2 years a substantial amount of data has accumulated concerning the development of these neuronal populations and their timely regulation by central and peripheral factors during fetal, neonatal, and peripubertal stages of development.This review focuses on the development of the kisspeptin–GPR54 system in the brain of female mice, rats, sheep, monkeys, and humans.We will also discuss the notion that this system represents a major target through which signals from the environment early in life can reprogram reproductive function.

**Keywords: kisspeptin, GPR54, ontogenesis, neuron, differentiation, regulation, reproduction, environment**

of the preovulatory GnRH surge and GnRH pulsatile release, respectively (**Figure 1**; Lehman et al., 2010 for review). Importantly, kisspeptin neurons express a large variety of hormonal receptors not expressed by GnRH neurons and may integrate and convey to GnRH neurons a large panel of information about the body, enabling adaptive outcomes on reproduction.

Within the last 2 years, a substantial amount of data has accumulated on the physiological regulation and function of the kisspeptin–GPR54 system in the developing brain and we consider it as timely and worthwhile to review this topic. In addition, we review recent evidence indicating that the kisspeptin–GPR54 system may represent a major target through which signals from the environment early in life can produce long-lasting defects on reproductive function. This review focuses on females and discusses data from mice, rats, sheep,monkeys, and humans in separate sections since marked differences exist between species and sexes relative to the developmental regulation of the kisspeptin– GPR54 system. For more detailed specific information on the neuroanatomy and regulation of the system in developing males or in adulthood, please refer to recent reviews on these subjects (Lehman et al., 2010; Clarke, 2011; Poling and Kauffman, 2013). Barely a decade has passed since the discovery of the importance of GPR54 in the development of reproductive function and, despite

the relative youth of the field, some consensus ideas and matters of controversies are starting to emerge with implications for important future directions.

GnRH (red) secretion: kisspeptin neurons in the POA drive GnRH surges and

# **A ROLE FOR THE KISSPEPTIN–GPR54 SYSTEM IN PUBERTY AND SEXUAL DIFFERENTIATION**

The prominent involvement of central nervous system kisspeptin signaling in the maturation of reproductive function is strongly suggested from genetic association studies of developmental reproductive disorders in humans and from experimental data on a variety of animal models. These studies are detailed below for each species and summarized in **Table 1**.

#### **MOUSE**

In the mouse, targeted genetic disruption of *GPR54* (de Roux et al., 2003; Funes et al., 2003; Seminara et al., 2003) or of *Kiss1* (d'Anglemont de Tassigny et al., 2007; Lapatto et al., 2007) produce a similar hypogonadotropic hypogonadism phenotype, suggesting that kisspeptins represent the main ligands for this receptor. This major effect on the development of reproductive function does not appear to involve a reduction of GnRH peptide levels in the brain nor a reduction in GnRH neuronal numbers (Seminara et al., 2003; d'Anglemont de Tassigny et al., 2007; Lapatto et al., 2007) but involves an impairment in tonic GnRH release (Dungan et al., 2007). *In vitro* studies further suggest that central kisspeptins can already signal developing GnRH neurons well before birth. Nasal placodes from embryonic day (E) 11.5 mouse embryos can be explanted *in vitro* to produce a GnRH neuronal network that releases GnRH in a pulsatile manner (Constantin et al., 2009a). If kisspeptins are applied to these cultures, it increases both GnRH pulse frequency and pulse amplitude (Constantin et al., 2009a). Kisspeptins have also been shown to promote neurite outgrowth of GnRH neurons from embryonic POA explants, suggesting their

participation in morphogenetic events (Fiorini and Jasoni, 2010). Notably, developmental strategies leading to the maturation of the GnRH system through puberty appear remarkably plastic at early stages of mouse development. For instance, if kisspeptin neurons are conditionally ablated with a toxin during the juvenile stage, mice display hypogonadism, persistent diestrus, and infertility but this is not the case if these cells are congenitally ablated indicating that strong compensatory mechanisms can occur earlier in development (Mayer and Boehm, 2011). *GPR54* deletion has also been found to inhibit defeminization/masculinization of brain circuits and behavior in males (Kauffman et al., 2007a) without interfering with the perinatal testosterone surge (Poling and Kauffman, 2012). In particular, *GPR54* knock-out males display a pattern of kisspeptin expression in the POA and an olfactory-mediated partner preference behavior characteristic of females (Kauffman et al., 2007a). By contrast, no behavioral deficits have yet been reported in female *GPR54* knock-out mice (Kauffman et al., 2007a). Nevertheless, a major neuroendocrine effect has been found since these females become unable to perform a surge of luteinizing hormone (LH) under estrogen positive feedback conditions in adulthood (Clarkson et al., 2008), raising the possibility that the development of female-specific brain circuits may be impacted by this genetic deletion. Further investigations on the impact of the kisspeptin–GPR54 system on the sexual differentiation of the brain and behaviors in females remain an important future direction.

mammillary bodies; APit, anterior pituitary; PPit, posterior pituitary.

# **RAT**

The involvement of the kisspeptin–GPR54 system in the central maturation of rat reproductive function has been explored with great precision by the group of Tena-Sempere since 2004 (Navarro et al., 2004a,b; Castellano et al., 2005, 2006). These studies used various modes of administration of synthetic forms of kisspeptins at different stages of development, followed by the



weeks.

monitoring of endocrine and physiological changes. In prepubertal female rats, kisspeptins, whether administered centrally or peripherally, induce an immediate elevation of circulating LH levels and this effect can be abolished by central administration of a GnRH receptor antagonist (Matsui et al., 2004; Navarro et al., 2004a). The GnRH-releasing activity of kisspeptins was directly tested on hypothalamic explants derived from female rats of various postnatal stages and found similar between the neonatal, infantile, and juvenile stages (Castellano et al., 2006). Repeated intracerebroventricular injections during the late juvenile period are able to increase uterus weight, circulating levels of LH and estradiol, and to advance the age of vaginal opening, a peripheral landmark of puberty onset in rodents (Navarro et al., 2004a). Conversely, central infusion of a GPR54 antagonist to peripubertal female rats decreases uterus and ovary weights without affecting total body weights and delays puberty onset (Pineda et al., 2010). Taken together, these observations provide convincing evidence that prepubertal kisspeptins are both necessary and sufficient for triggering various indices of female puberty in this species.

#### **SHEEP**

The functional relevance of the kisspeptin–GPR54 system in the maturation of female reproductive function has been well characterized during the last decade in mice and rats, as detailed in the preceding sections; however it was not until 2011 that this aspect has been explored in species other than rodents such as sheep and monkeys. In ewes, chronic hourly intravenous injections of kisspeptins during the prepubertal period stimulated a pulsatile release of LH within 15 min following injections, increasing both pulse frequencies and amplitudes (Redmond et al., 2011a). Mean circulating levels of LH and estradiol were increased and a surge like release of LH developed in some lambs 17 h post treatment. These animals were however unable to develop a long-lasting luteal phase as attested by circulating progesterone levels and did not achieve regular estrus cyclicity, suggesting that the reproductive neuroendocrine axis was not yet fully mature (Redmond et al., 2011a).

#### **MONKEY**

In female monkeys, it has recently become possible to directly monitor the GnRH-releasing activity of kisspeptins *in vivo* (Guerriero et al., 2012a). The GnRH-releasing response to kisspeptin infusion directly within the medial basal hypothalamus and stalk median eminence is dose dependent and greater in pubertal than prepubertal females (Guerriero et al., 2012a). Conversely, mean GnRH levels are diminished following central infusion with a GPR54 antagonist at both developmental stages (Roseweir et al., 2009; Guerriero et al., 2012a). This provides convincing evidence that a GPR54-mediated mechanism is required for the reactivation of GnRH release at puberty in the female monkey.

#### **HUMAN**

The fundamental role of the kisspeptin–GPR54 system in Pubertal development was initially discovered by two independent groups that identified loss of function mutations in the *GPR54* gene within human families with idiopathic hypogonadotropic hypogonadism (IHH) (de Roux et al., 2003; Seminara et al., 2003). Mutations of the *GPR54* gene have since been found in other families with IHH associated with various levels of GnRH deficiency (Nimri et al., 2011; Wahab et al., 2011). A very recent report also discovered an inactivating mutation of the *Kiss1* gene associated with IHH (Topaloglu et al., 2012). Idiopathic central precocious puberty (ICPP) is another developmental reproductive disorder that has been associated with either*GPR54* mutations (Teles et al., 2008), *Kiss1* mutations (Silveira et al., 2010), or polymorphisms (Luan et al., 2007a,b). Taken together, these pathologies attest for the critical role played by kisspeptin signaling in the development of the human hypothalamic–pituitary–gonadal axis.

# **DEVELOPMENTAL CHANGES OF KISSPEPTIN–GPR54 COINCIDE WITH CHANGES IN LH SECRETION THROUGHOUT LIFE**

**MOUSE** Recent evidence points to an early onset of *Kiss1* and *GPR54* expression in the mouse nervous system. For instance,*Kiss1* mRNA has been detected in the mediobasal hypothalamus of the mouse as early as embryonic day (E)13, by reverse transcription polymerase chain reaction (Fiorini and Jasoni, 2010). At this early developmental stage, *in situ* hybridization further identified *GPR54* mRNA in some GnRH neurons along the nasal portion of their migratory route (Constantin et al., 2009b). Furthermore, *in vitro* studies suggest that GnRH neurons are already able to respond to kisspeptins by enhanced secretion during prenatal life (Constantin et al., 2009a,b). However, the presence of the kisspeptin protein remains to be shown in mouse embryos.

Postnatally, *Kiss1*-expressing cells can already be detected in the ARC a few hours only after birth, using *in situ* hybridization (Poling and Kauffman, 2012). *Kiss1* mRNA levels increase in this region at the time of puberty but only in hypogonadal hpg mice, which are deficient in GnRH receptor signaling and hence display very low levels of sex steroids (Gill et al., 2010, 2012). In wild type mice, it appears that *Kiss1* expression in the ARC is strongly repressed postnatally by circulating estradiol and no developmental changes have yet been detected in this region at the *Kiss1* transcript or kisspeptin protein level (Clarkson and Herbison, 2006; Gill et al., 2010, 2012). Of interest, however, is the significant increase in neurokinin B (NKB; another GnRH secretagogue expressed by kisspeptin neurons) transcript levels that has been observed in this nucleus prior to puberty onset (Gill et al., 2012). In the POA, expression of *Kiss1* appears to develop later than in the ARC, between postnatal day 8 and 10 (Semaan et al., 2010) and increases thereafter until puberty is reached (Gill et al., 2010). Similarly, a peripubertal increase in the number of POA kisspeptin-immunoreactive neurons has been shown (Clarkson and Herbison, 2006; Clarkson et al., 2009; Gill et al., 2010; Mayer et al., 2010). This increase has been correlated with the development of the capacity for GnRH/LH surges (Clarkson et al., 2009). Another interesting feature observed in the development of this system is a peripubertal increase in the proportion of GnRH neurons closely apposed to kisspeptin-immunoreactive fibers, suggesting a postnatal morphological maturation of kisspeptin circuitry associated with puberty (Clarkson and Herbison, 2006). Furthermore, a longitudinal analysis of transgenic mice in which

*LacZ* has been introduced in the *GPR54* locus identified a prepubertal rise in the percentage of GnRH neurons expressing *GPR54* (Herbison et al., 2010). These different changes observed in the female POA at the time of puberty may be related to the increase in circulating LH levels that precedes puberty onset in the mouse (Michael et al., 1980; Gill et al., 2012). In any case, debate still persists about the relative importance of ARC versus POA kisspeptin neurons in triggering the onset of puberty in the female mouse.

#### **RAT**

A recent study in the rat has provided strong foundation for our understanding of the embryonic stages of kisspeptin neuron development. Using a BrdU pulse chasing approach, we have established that the neurogenesis period of ARC kisspeptin cells in the female rat hypothalamus begins at about E12.5, peaks around E15.5, and is not yet over at E17.5 (Desroziers et al., 2012a). Using immunohistochemistry, we further show that some cells in the developing ARC already synthesize kisspeptins from about E14.5 (Desroziers et al., 2012a; **Figure 2**). The number of kisspeptinimmunoreactive cells in the fetal ARC increases between E14.5 and E18.5 which coincides with the time when GnRH fibers reach the portal vessels of the median eminence and circulating LH levels reach their maximum in the rat fetus (Ugrumov et al., 1985; Huhtaniemi, 1995). At E18.5, close appositions between kisspeptinand GnRH-immunoreactive fibers can be detected in the median eminence (**Figure 2D**), consistent with the hypothesis that kisspeptins already control GnRH release prenatally. The number of kisspeptin-immunoreactive cells, as well as hypothalamic *Kiss1* mRNA levels, decrease at the end of gestation (Desroziers et al., 2012a). The mechanism of action and physiological relevance of this decrease have yet to be fully elucidated.

During postnatal development, *Kiss1* expression continues to be tightly regulated as shown in the pioneering study by Navarro et al. (2004b). In particular, real-time quantitative RT-PCR of *Kiss1* and *GPR54* content in the hypothalamus revealed a transient decline in the expression levels of both genes during the infantile period, followed by an increase around the time of puberty (Navarro et al., 2004b). These changes have since been spatially refined by studying the respective expression of these genes either by RT-PCR on tissue punches (Knox et al., 2009; Takase et al., 2009; Lederman et al., 2010; Li et al., 2012) or by *in situ* hybridization (Cao and Patisaul, 2011; Takumi et al., 2011; Patisaul et al., 2012). In the ARC, *Kiss1* expression is relatively high during the neonatal and early infantile period but decreases during the third postnatal week (Cao and Patisaul, 2011; Takumi et al., 2011). A decline in the density of kisspeptin-immunoreactive fibers has also been measured during that period of time (Desroziers et al., 2012b). These observations may be correlated to the reported transient activation of LH secretion during early postnatal life (Döhler and Wuttke, 1975). *Kiss1* expression in the ARC increases again at puberty (Knox et al., 2009; Takase et al., 2009; Takumi et al., 2011). This increase has been associated with an increase in basal

kisspeptin-immunoreactive cells were detected (green dots). Numerous kisspeptin-immunoreactive cells were detected in the ARC at embryonic day (E) 18.5 as shown on this immunoperoxidase-labeled brain section with anti-kisspeptin AC067 **(C)** [corresponding to the box in **(B)**]. In the median eminence, close appositions (white arrows) between kisspeptin fibers (green) and GnRH fibers (red) were detected as shown in this

anterior commissure; PaP, paraventricular nucleus; 3V, third ventricle; OC, optic chiasma; POA, preoptic area; SO, supraoptic nucleus; ME, median eminence; VMH, ventro-medial hypothalamus; MB, mammillary bodies; Pit, pituitary; Apit, anterior pituitary; Ppit, posterior pituitary; BSph, sphenoid bones; MRe, mammillary recess. Scale bars: 100µm **(C)** and 20µm **(D)**.

circulating LH levels (Takase et al., 2009). Similarly, kisspeptinimmunoreactive cell numbers and fiber densities increase in the ARC during the peripubertal period (Desroziers et al., 2012b). As in mice, *Kiss 1* expression in the POA of rats is detected later than in the ARC, starting during the second postnatal week (Cao and Patisaul, 2011), and continues to increase until the fifth postnatal week (Desroziers et al., 2010; Takumi et al., 2011). By contrast to mice, however, no kisspeptin-immunoreactive cells has been detected so far in the POA throughout rat postnatal life, unless colchicine was administered (Takase et al., 2009; Desroziers et al., 2010,2012b;Lederman et al.,2010).A very recent study on Sprague Dawley rats has enabled the peripubertal changes in *Kiss1* and *GPR54* expression in the ARC and POA to be assessed with greater time resolution and in relation to changes in the frequency of LH pulses (Li et al., 2012). No changes in *GPR54* expression are observed during this developmental time window but*Kiss1* mRNA levels significantly increase prior to puberty onset, first in the ARC and later in the POA. Notably, these changes are accompanied by an increase in LH pulse frequencies (Li et al., 2012). Our recent analysis of kisspeptin-immunoreactive fiber density across each of these two regions similarly suggest a sequential activation of kisspeptin-immunoreactivity at the time of puberty, first in the ARC, and second in the POA (Desroziers et al., 2012b). Collectively, these findings provide strong support for an important role of kisspeptins produced by ARC neurons in triggering puberty in female rats. During pubertal progression, this would be followed by an estrogen-dependent amplification system of GnRH release mediated by kisspeptins neurons of the POA.

In adult cycling females, the number of *Kiss1*-expressing cells varies across the estrous cycle, in opposite phases between the ARC and the POA (Smith et al., 2006a). Changes in *Kiss1* expression have also been evaluated across aging. In the rat, unlike in humans, reproductive senescence is associated with a decrease in the frequency and amplitude of GnRH/LH pulses and a progressive disappearance of GnRH/LH surges (Scarbrough and Wise, 1990). Consistently, a significantly lower number of kisspeptinimmunoreactive cells are detected in the POA of middle-aged rats compared to young rats, under estradiol positive feedback conditions (Lederman et al., 2010).

Thus, in the rat, *Kiss1* expression appears tightly regulated throughout the reproductive life cycle, not only in the POA, but also in the ARC where it is already detected prenatally. The developmental profile of *Kiss1* expression in the ARC and POA differ significantly. This may be related to different roles played by the two kisspeptin cell populations in the control of GnRH pulse amplitude and frequency.

#### **SHEEP**

Studies of the kisspeptin–GPR54 system in sheep have focused on two critical periods of development, the prenatal period and the peripubertal period. *Kiss1* expression has been detected by RT-PCR in the hypothalamus of 110 day old sheep fetuses (birth occurring around gestational day 145 in this species) both in rostral and caudal hypothalamic slices including the POA and ARC, respectively (Bellingham et al., 2009). This is a time when LH is already secreted in a GnRH-dependent, pulsatile manner (Matwijiw and Faiman, 1987). Thus, it is tempting to speculate that kisspeptins already provide tonic drive to GnRH neuronal activity *in utero*. *Kiss1* expression has further been studied by *in situ* hybridization in an estradiol-supplemented ovariectomized model of puberty (Redmond et al., 2011b). In this study, puberty was preceded by an elevation in the number of *Kiss1*-expressing neurons both in the POA and ARC. However, it is the increase in *Kiss1* expression levels in the ARC more specifically that could be correlated with an increase in LH pulse frequency (Redmond et al., 2011b). In intact ewes, the number of kisspeptin-immunoreactive cells in the ARC was shown to increase at puberty, concomitantly to an increase in LH pulse frequency but not pulse amplitude (Nestor et al., 2012). Over that same period of time, the proportion of GnRH neurons with close appositions of kisspeptinimmunoreactive fibers increase (Nestor et al., 2012), as previously reported in mice (Clarkson and Herbison, 2006). Whether this reflects a postnatal morphological plasticity of the system, or is related to enhanced *Kiss1* expression remains to be investigated. Taken together, these findings point to a potential relationship between postnatal changes in the kisspeptin–GPR54 system and the increase of LH secretion accompanying ovine puberty.

# **MONKEY**

In the female monkey, hypothalamic *Kiss1* and *GPR54* expression levels have been monitored across puberty and aging. Realtime quantitative RT-PCR of *Kiss1* and *GPR54* mRNA levels in the mediobasal hypothalamus revealed that both of these transcripts significantly increase at puberty (Shahab et al., 2005). More recently, using an *in vivo* microdialysis method, the team of Ei Terasawa was able to show an increase in kisspeptin pulsatile release within the stalk median eminence during puberty (Guerriero et al., 2012b). During late puberty, *in vivo* secretion of kisspeptin is pulsatile and in almost perfect synchrony with GnRH pulses, a finding consistent with a possible role of kisspeptins in the tuning of GnRH pulsatility at puberty (Keen et al., 2008). Notably, no changes in the POA have yet been reported across monkey physiology. With aging, *Kiss1* and *GPR54* mRNA levels increase in the mediobasal hypothalamus but remain unchanged in the POA (Kim et al., 2009; Eghlidi et al., 2010). By analogy, LH pulsatile secretion increases in aging monkeys but these are still able to generate GnRH/LH surges (Gore et al., 2004). Collectively, these observations suggest that GnRH release in the monkey continues to be under the tight regulatory control of ARC and POA kisspeptins late in postnatal life.

#### **HUMAN**

A very recent study detected kisspeptin and GPR54 immunoreactivities in the hypothalamus of second trimester human fetuses (Guimiot et al., 2012). Kisspeptin-immunoreactivity declined at the end of gestation, as previously observed in rats (Desroziers et al., 2012a). A causal link between this decline and the decline in circulating levels of gonadotropins that was monitored in cord blood between the 30th week of prenatal life and birth was suggested (Guimiot et al., 2012). Postnatal analysis of *Kiss1* expression or kisspeptin-immunoreactivity in human brains has only been reported in post-pubertal subjects (Rometo et al., 2007; Hrabovszky et al., 2010, 2011). *Kiss1* expression in the infundibular nucleus has been shown to increase after menopause in women (Rometo et al., 2007). This increase may be related to the increase in GnRH/LH secretion characterizing human reproductive senescence (Kermath and Gore, 2012 for review). Again, these findings are consistent with the hypothesis that GnRH secretion in women continues to be regulated by kisspeptins late in life.

In conclusion, analogies can be found in each species between fluctuations in GnRH/LH levels and fluctuations in some aspects of the kisspeptin–GPR54 system, supporting the view that kisspeptin signaling in the brain controls GnRH secretion throughout life, including prenatally (**Figure 3**). The regulatory mechanism by which these changes in the kisspeptin–GPR54 system operate will be the focus of the following section.

#### **ENDOGENOUS REGULATORS OF KISSPEPTIN–GPR54 DURING DEVELOPMENT**

#### **DEVELOPMENTAL REGULATION OF THE KISSPEPTIN–GPR54 SYSTEM BY GONADAL STEROIDS**

Ever since the discovery that *Kiss1* expression is under tight regulation of estrogen receptor signaling (Smith et al., 2005) and that kisspeptin neurons express numerous sex steroid receptors in adult mice, rats, and sheep (Smith et al., 2005, 2006a, 2007; Franceschini et al., 2006; Adachi et al., 2007; Clarkson et al., 2008; Cheng et al., 2010), numerous studies have investigated the regulation of the kisspeptin–GPR54 system by gonadal steroids across development (**Table 2**).

#### **Mouse**

Female mice display a much greater number of kisspeptinimmunoreactive cells in the POA than male mice (Clarkson and Herbison, 2006) even after gonadectomy (González-Martínez et al., 2008), suggestive of organizational effects of sex steroids on this cell population during development. This sexual dimorphism appears to develop between postnatal days 10 and 12, that is shortly after the first *Kiss1*-expressing cells can be detected in this region (Semaan et al., 2010). It does not appear to involve bax-mediated apoptosis (Semaan et al., 2010) but differential transcription patterns of the *Kiss1* gene governed by sex-specific epigenetic mechanisms during development (Semaan et al., 2012). For instance, the methylation pattern of the *Kiss1* gene in the POA (but not the ARC) significantly differs at several CpG sites between males and females (Semaan et al., 2012). The contribution of estrogen signaling to the sexual differentiation of the kisspeptin cell population of the POA has mostly been explored by the means of genetic manipulations. Mice genetically deficient in alpha-fetoprotein (AFP-knock-out) have excess estradiol signaling in the brain during the perinatal period. Female AFPknock-out mice develop as few kisspeptin cells in the POA as males and their circulating LH levels are not increased by estradiol and progesterone-induced positive feedback conditions (González-Martínez et al., 2008). Thus, it may be that estradiol signaling in the brain at the time of the perinatal testosterone surge may be in part responsible for the defeminization/masculinization of the POA kisspeptin cell population in males. Interestingly, transgenic reduction of developmental sex steroid hormone signaling in a variety of mouse models also results in a marked decrease in the adult number of kisspeptin-immunoreactive neurons in the POA (Clarkson et al., 2009; Bakker et al., 2010; Gill et al., 2010; Mayer et al., 2010). For example, in the adult hypogonadal hpg mice, the number of kisspeptin-immunoreactive cells in the POA is reduced by half and it is not possible to increase this number by estradiol administration as it is for wild type mice (Gill et al., 2010). In aromatase knock-out (ArKO) mice which completely lack the capacity to synthesize estrogens, the number of kisspeptin-immunoreactive cells in the POA is also significantly reduced compared to wild type mice (Clarkson et al., 2009), even under identical positive feedback conditions (Bakker et al., 2010). In the same vein, ArKO mice, whether ovariectomized or treated with estradiol and/or progesterone to mimic positive feedback conditions, display fewer *Kiss1*-expressing cells in the POA compared to wild type mice (Szymanski and Bakker, 2012), consistent with the hypothesis that postnatal estrogens may reprogram *Kiss1*



embryonic

 day; P, postnatal day; W, weeks.

transcription in the POA. Furthermore, while odors from male urine are able to induce c-fos expression in a good proportion of POA kisspeptin cells in wild type female mice, this proportion is markedly reduced in the ArKO mice (Bakker et al., 2010). Data from experimental gonadectomy and estradiol replacement during development suggest that estrogens stimulate kisspeptin expression in the POA before puberty (Clarkson et al., 2009). These authors proposed that this estrogen-dependent increase of kisspeptin expression in the POA may facilitate the emergence of pulsatile gonadotropin secretion necessary for puberty onset, as well as the preovulatory estrogen-dependent surge (Clarkson et al., 2009). However, it was recently shown that ArKO mice maintain the ability to mount an LH surge when treated with hormones mimicking positive feedback conditions in adulthood, questioning the absolute requirement of the full female-typical (estrogen induced) complement of POA kisspeptin neurons in generating GnRH/LH surge (Szymanski and Bakker, 2012).

In the mouse ARC, the sexual dimorphism of the kisspeptin cell population is not as obvious as in the POA. In this nucleus, female mice display a greater number of *Kiss1*-expressing cells than males on the day of birth but this difference is no longer visible during infancy or in adulthood (Kauffman et al., 2009; Poling and Kauffman, 2012). By gonadectomizing the mice at different stages of development, Kauffman et al. (2009) revealed an important gonadal hormone-independent sex difference in the number of ARC *Kiss1* cells during the infantile period. Prepubertal gonadal hormones appear to exert a greater repressive action on ARC *Kiss*1 expression in females than in males (Kauffman et al., 2009). Of interest, this strong downregulation of *Kiss1* expression in the prepubertal female ARC most likely occurs through Erα signaling within kisspeptin cells themselves and may be of physiological relevance in the timing of puberty (Mayer et al., 2010). Indeed, female mice with a targeted deletion of *Esr1* within kisspeptin cells not only display enhanced *Kiss1* expression in the ARC at the juvenile stage but also a dramatic advance in the timing of vaginal opening. Moreover, circulating LH levels in these genetically modified mice are higher than in control mice (Mayer et al., 2010). These authors proposed that the prepubertal repressive action of estrogen signaling on *Kiss1* expression in the ARC represents an essential break to the central activation of the gonadotropic axis, preventing premature puberty onset (Mayer et al., 2010).

Collectively, these findings suggest that in the mouse, estrogen signaling may exert organizational effects on the kisspeptin cell population of the POA during the perinatal period in males and during the prepubertal period in females. Further studies are needed to elucidate the developmental origin of the sexual dimorphism observed at the level of ARC kisspeptin cell population. In particular, it is still unclear whether the sex-specific pattern of *Kiss1* expression detected in the absence of gonadal hormones during infancy in this nucleus is conditioned by a sexspecific and gonad-independent developmental program or by an organizational effect of the perinatal testosterone surge.

#### **Rat**

The number of POA *Kiss1*-expressing cells is greater in female rats than in male rats under identical estrogen positive feedback conditions, suggesting that, like in mice, developmental sex steroids may have important organizational effects on this cell population (Kauffman et al., 2007b; Homma et al., 2009). The importance of the perinatal period in this process is suggested from several studies using gain or loss of function approaches. Male pups neonatally orchidectomized develop a female-specific pattern of *Kiss1* expression in the POA (Homma et al., 2009; Takumi et al., 2012a). Conversely, female pups exposed to synthetic estrogens or aromatizable androgens such as estradiol benzoate or testosterone propionate during the perinatal period display a male-specific pattern of *Kiss1* expression or kisspeptin-immunoreactivity in adulthood (Navarro et al., 2004b; Kauffman et al., 2007b; Homma et al., 2009; Dickerson et al., 2011a).

By contrast to the POA, similar numbers of *Kiss1*-expressing cells and similar levels of *Kiss1* transcripts have been detected in the ARC of gonadectomized male and female rats (Adachi et al., 2007; Kauffman et al., 2007b; Homma et al., 2009). Furthermore, the downregulating effect of estradiol on *Kiss1* transcription in this nucleus appears similar in both sexes (Adachi et al., 2007; Kauffman et al., 2007b; Homma et al., 2009). In the ARC of intact adult rats, however, several studies have shown that females display more *Kiss1*-expressing cells or kisspeptin-immunoreactivity than males (Iijima et al., 2011; Takumi et al., 2011; Desroziers et al., 2012b). A real-time quantitative RT-PCR analysis of *Kiss1* transcript content in the ARC also points to the fact that the sex difference in *Kiss1* transcript levels is more or less visible, depending on the stage of the estrus cycle (Adachi et al., 2007). This sex difference develops during the neonatal period (Cao and Patisaul, 2011; Desroziers et al., 2012b) and its amplitude appears to fluctuate across postnatal development (Cao and Patisaul, 2011; Takumi et al., 2011; Desroziers et al., 2012b), presumably reflecting different developmental fluctuations in circulating gonadal hormones between males and females, as well as sex-specific developmental changes in the sensitivity of ARC kisspeptin cells to these hormones. For instance, data from experimental ovariectomy and estradiol replacement across different developmental time windows suggest that the regulatory action of estrogen signaling on *Kiss1* expression may change at the time of female puberty, leading to a developmental increase of *Kiss1* mRNA levels in both regions, and to an increase in circulating LH levels (Takase et al., 2009). Accordingly, we recently measured a peripubertal increase in kisspeptin-immunoreactivity within both regions at a time when peripheral estradiol levels were relatively constant (Desroziers et al., 2012b).

Another period of life when the sensitivity of kisspeptin cells to the feedback action of ovarian hormones appears to change is during aging. Middle-aged female rats that are gonadectomized and supplemented with estradiol benzoate and progesterone to mimic positive feedback conditions display an attenuated rise in the number of kisspeptin-immunoreactive cells in the POA as compared to young females under the same conditions (Lederman et al., 2010). By contrast, the repressive action of these gonadal hormones on ARC *Kiss1* mRNA and kisspeptin levels appears unaffected by aging (Lederman et al., 2010).

Collectively, these results suggest that estrogen signaling during the neonatal period organizes sex differences in *Kiss1* expression in the POA and that the sensitivity of both kisspeptin cell populations to estrogen signaling is dynamic across the lifespan of female rats.

#### **Sheep**

Female sheep display a greater number of kisspeptinimmunoreactive cells than male sheep not only in the POA but also in the ARC (Cheng et al., 2010). This sex difference in the ARC is also observed in pubertal sheep after gonadectomy, suggesting that it may result from organizational effects of gonadal steroids (Nestor et al., 2012). In this precocious species, sexual differentiation of the neuroendocrine circuits controlling reproductive function is mainly under the organizational influence of prenatal androgens during fetal life. Accordingly, female sheep treated prenatally with androgens display a lower responsiveness of the GnRH system to the negative feedback influence of progesterone. This translates into increased LH pulse frequencies and mean concentrations in peripheral blood (Cheng et al., 2010). However, this prenatal androgenization does not alter the number of kisspeptin-immunoreactive cells in the ARC nor in the POA. Rather, it is the number of cells immunoreactive for dynorphin or NKB, two other neuropeptides co-expressed with kisspeptins in the ARC, that is reduced compared to control females (Cheng et al., 2010). It may be that a greater exposure to androgens is required (dose and length) or that the influence of sex-specific genetic factors prevail in the sexual differentiation of the ovine kisspeptin system. Around the time of puberty, the negative feedback action of gonadal hormones on ARC kisspeptin-immunoreactivity diminishes parallel to an increase in LH pulse frequencies, suggesting, as in the rat, that a change in the sensitivity of ARC kisspeptin cells to the negative feedback action of estradiol signaling may initiate puberty onset (Nestor et al., 2012).

#### **Monkey**

There is to our knowledge no published information of a clear sexual dimorphism of the kisspeptin system in the monkey. In the female monkey, the positive and negative feedback action of estradiol on the kisspeptin–GPR54 system have been studied at the time of puberty (Guerriero et al., 2012b) and during aging (Eghlidi et al., 2010). Prepubertal ovariectomy does not prevent the peripubertal rise in kisspeptin pulsatile release within the median eminence. In fact, the repressive action of estradiol on this release appears to develop later, once puberty has already started (Guerriero et al., 2012b). Ovariectomy and short-term estradiol replacement appears to have little effect on ARC *Kiss1* expression levels in young monkeys (Eghlidi et al., 2010). By contrast, long-term ovariectomy (mimicking the loss of ovarian steroid negative feedback that occurs during perimenopause) significantly increases ARC *Kiss1* expression levels. *Kiss1* expression levels can be restored to normality by short-term estradiol supplementation (Eghlidi et al., 2010). Collectively, these results minimize the role played by estradiol signaling in the upregulation of kisspeptin release at puberty in the female monkey but suggest a role of this signaling in the upregulation of *Kiss1* expression in the ARC with aging.

#### **Human**

Little is known in humans about the impact of sex steroids on the developing kisspeptin–GPR54 system. In adult women, the number of kisspeptin-immunoreactive neurons is greater than in men both in the infundibular nucleus and in the POA (Hrabovszky et al., 2010). However, the involvement of developmental sex steroids in this dimorphism is yet to be defined.

#### **DEVELOPMENTAL REGULATION OF KISSPEPTIN NEURONS BY OTHER FACTORS**

In the adult, there exists neuroanatomical, electrophysiological, and/or pharmacological evidence that the kisspeptin–GPR54 system may be the target of a variety of other hormones, neurotransmitters, and neuropeptides (**Figure 4**, squared by dotted lines). These include leptin (Smith et al., 2006b; Backholer et al., 2010), corticosterone (Kinsey-Jones et al., 2009), prolactin (Li et al., 2011), melanocortin hormone (Cravo et al., 2011), pheromones (Bakker et al., 2010), melatonin (Revel et al., 2006), ghrelin (Forbes et al., 2009),melanin-concentrating hormone (Wu et al., 2009; Cravo et al., 2011), corticotropin-releasing hormone (Takumi et al., 2012b), vasopressin (Vida et al., 2010), glutamate (Ducret et al., 2010), GABA (García-Galiano et al., 2012a), dopamine (Goodman et al., 2012), NKB (Navarro et al., 2011), and RFRP3 (Rizwan et al., 2012). However, only a few of these molecular factors have yet been identified as potential actors in the developmental regulation of the kisspeptin–GPR54 system (**Figure 4**, squared by full lines).

#### **Leptin**

Among hormonal factors, leptin, a well known permissive factor for pubertal maturation, is a potential upstream regulator of the kisspeptin–GPR54 system. Leptin deficient ob/ob mice display a marked reduction of *Kiss1* expression in the ARC and in the number of kisspeptin-immunoreactive cells in the POA but it is not yet known when exactly during development this effect starts (Quennell et al., 2011). The gonadotropin response of prepubertal rats to acute central administration of kisspeptin is preserved in different models of leptin deficiencies such as after central immunoneutralization of leptin or food restriction and in leptin resistant Zucker rats, indicating that leptin must act upstream of kisspeptin signaling (Navarro et al., 2004a; Castellano et al., 2011). In prepubertal rats under different food regimens, a positive correlation has been found between circulating leptin levels and hypothalamic *Kiss1* and *GPR54* mRNA levels (Iwasa et al., 2010a; Castellano et al., 2011). Pharmacological manipulations of mTOR, a transducer of leptin's effect on energy homeostasis, suggested that this signaling pathway is essential for activating *Kiss1* transcription in the ARC at puberty onset (Roa et al., 2009). However, administration of leptin was able to restore hypothalamic levels of *GPR54* but not *Kiss1* mRNA in prepubertal rats displaying reduced levels of both *GPR54* and *Kiss1* mRNA after 24 h food deprivation (Iwasa et al., 2010a). The precise site of leptin's action in the brain during the pubertal transition period may vary between species and remains a matter of debate (Elias and Purohit, 2013).

#### **Neuropeptides**

Among neuropeptides potentially regulating the kisspeptin– GPR54 system during development, a particular attention has been paid to NKB. Indeed, human genetic studies have associated loss of function mutations on either TAC3 or TACR3 genes (encoding NKB and its receptor NK3R respectively) with hypogonadism and infertility, similar to the phenotypes of *Kiss1* or *GPR54* mutants

**FIGURE 4 | Neural, hormonal, and environmental factors regulating the kisspeptin–GPR54 system.** Scheme summarizing the different factors that have been shown to regulate the kisspeptin–GPR54 system only during adulthood (squared by dashed lines) or also during development (squared by full lines and colored). Hormonal factors are codified by an arrow and central factors by a triangle. Molecular factors have been included whose receptors have been found on some kisspeptin neurons, factors found within fibers in close apposition to kisspeptin neurons, factors eliciting c-fos expression, or an electrophysiological response within kisspeptin neurons or changing Kiss1 or GPR54 mRNA levels, kisspeptin or GPR54 immunoreactivities, or the number of Kiss1/kisspeptin expressing cells when exogenously administered. Of note this synthetic scheme combines data from mice, rats, sheep, and monkeys and therefore occults potential species differences that may exist in these regulations. It is hypothesized that the developmental pattern of GnRH release (red graph below the tap) is shaped by interactions of these different neural and hormonal factors with an intrinsic differentiation program of the

(Topaloglu et al., 2009). Interestingly, NKB receptors have been found on kisspeptin neurons of the ARC, which also co-express NKB, but not on GnRH neurons (Navarro et al., 2012). In prepubertal hpg mice, administration of a specific NK3R antagonist showed that NKB does not control *Kiss1* expression (Gill et al., system (central clock). The developing kisspeptin–GPR54 system is particularly vulnerable to some environmental factors like endocrine disruptors, diet, and stress which can alter GnRH secretion and reproductive function on the long-term. POA, preoptic area; ARC, arcuate nucleus; E2, estradiol; T, testosterone; P4, progesterone; ER, estrogen receptor; AR, androgen receptor; PR, progestin receptor; IGF, insulin-like growth factor; IGF-R, insulin-like growth factor receptor; FGF, fibroblast growth factor; FGF-R, fibroblast growth factor receptor; GABA-R, GABA receptor; RFRP3, RF-amides related peptide-3; RFRP3-R, RFRP3 receptor; LepR, leptin receptor; Prl-R, Prolactin receptor; NKB, neurokinin B; NK3R, NKB receptor; Glut-R, glutamate receptor; VP, vasopressin; VP-R, vasopressin receptor; MCH, melanocortin; MCH-R, MCH receptor; Dyn, dynorphin; KOR, kappa-opioid receptor (Dyn-receptor); GR, Glucocorticoid receptor; CRH, corticotrophin-releasing hormone; CRH R, corticotrophin-releasing hormone receptor; D2-R, dopamine-receptor. The illustrations in the arrows were obtained from Clipart Microsoft Word® .

2012). Instead, there is increasing evidence that this neuropeptide stimulates kisspeptin release. For instance, the LH releasing activity of the NKB agonist senktide is abolished in *GPR54* knock-out mice (García-Galiano et al., 2012b), in the presence of a GPR54 antagonist in prepubertal rats (Grachev et al., 2012) or after GPR54 desensitizing in agonadal juvenile monkeys (Ramaswamy et al., 2011). A model has recently been proposed where NKB would participate in the initiation of pulses of kisspeptin release by synchronizing ARC kisspeptin neuronal activity through an autofeedback loop (Navarro, 2012).

Another neuropeptide that may cross-talk with the kisspeptin– GPR54 system during development is RFRP3. This mammalian ortholog of GnIH is already found closely apposed to a large proportion of GnRH neurons in prepubertal rats (Losa-Ward et al., 2012) where it may act by decreasing the GnRH neuronal response to kisspeptins as previously shown by electrophysiological recordings on brain slices from adult mice (Wu et al., 2009). In addition, RFRP3 may act upstream of kisspeptin neurons in light of a recent study in hamster where this peptide has been proposed to convey a melatoninergic signal to GnRH neurons through regulation of *Kiss1* transcription (Ancel et al., 2012).

#### **Growth factors**

The developing kisspeptin–GPR54 system may also be the target of growth factors: mice harboring deficiencies in FGF8 and/or FGFR-1 display a greater number of kisspeptin-immunoreactive cells in the POA at some stages of peripubertal development specifically, suggesting that FGF signaling pathways may control kisspeptin cell numbers or act on the *Kiss1* gene at transcriptional or posttranscriptional levels (Tata et al., 2012). Another growth factor that may be implicated during development of kisspeptin neurons is IGF1. Female rats that receive an intracerebroventricular administration of IGF1 during the prepubertal period display increased *Kiss1* mRNA levels in the POA specifically. This effect can be abolished by the administration of an IGF1 receptor antagonist and appears dependent on the presence of gonadal estrogens (Hiney et al., 2009). This contrasts with results obtained in adulthood where the same antagonist produces no effect on *Kiss1* expression, suggesting that the IGF1 receptor response of kisspeptin neurons may change as development proceeds (Todd et al., 2007).

# **GABA**

In the female rhesus monkey, a very recent study demonstrated the fundamental role of GABA signaling in restraining kisspeptin release prior to puberty. GABA<sup>A</sup> receptor antagonist administration during the prepubertal period but not during the pubertal period stimulates kisspeptin release in the medial basal hypothalamus (Kurian et al., 2012). In the same study, the use of a GPR54 antagonist suggested that kisspeptin neurons may relay inhibitory GABA signals to GnRH neurons prior to puberty. In addition, the response of GnRH neurons to kisspeptins can be modulated by GABA signaling in adult mice and rats and it will be interesting to further explore when during development this cross-talk is established (Pielecka-Fortuna and Moenter, 2010; García-Galiano et al., 2012a).

# **Transcription factors**

The concept has recently been put forward that puberty is controlled by regulatory gene networks composed of multiple functional modules operating with overlaps of partially redundant pathways (Ojeda et al., 2010). In this context, there has been a great interest in positioning the *Kiss1* gene within a framework of puberty-associated identified transcription factors. *In vitro* promoter assays in human cell lines suggest that the *Kiss1* gene is regulated by a set trans-activators and repressors involved in the system-wide control of mammalian puberty, among which TTF1, CUX1-p200, EAP1, YY1, and CUX1-p110 (Mueller et al., 2011). It will be of great interest to confirm the relevance of these findings *in vivo*.

# **THE DEVELOPING KISSPEPTIN–GPR54 SYSTEM AS TARGET OF ENVIRONMENTAL DISRUPTORS OF REPRODUCTION ENDOCRINE DISRUPTORS**

A variety of endocrine disrupting chemicals (EDCs) have recently been shown to disrupt the orderly progression of the female reproductive life cycle in association with changes in the development of the kisspeptin–GPR54 system. Most studies reporting alterations of the kisspeptin–GPR54 system by EDCs have been performed in rats. For example, estradiol benzoate, genistein, and polychlorinated bisphenyls (PCBs), if administered during rat perinatal development, have each been shown to advance vaginal openings and accelerate reproductive senescence, associated with a reduction in hypothalamic kisspeptin-immunoreactivity in adulthood (both at the level of the POA and ARC). This reduction in kisspeptin-immunoreactivity is observed after normalization of sex steroid circulating levels and is associated with a decline in the proportion of GnRH neurons being activatable by hormonal stimuli mimicking estradiol positive feedback conditions (Bateman and Patisaul, 2008; Dickerson et al., 2011a; Patisaul et al., 2012). Long Evans rats administered with estradiol benzoate or genistein during the first 4 days of life have further been analyzed for kisspeptin-immunoreactivity around puberty (Losa et al., 2011). A significant reduction in kisspeptin-immunoreactivity (both at the level of the POA and ARC) was detected in these EDC-treated animals compared to control rats. A recent *in situ* hybridization analysis shows that female Long Evans rats neonatally administered with estradiol benzoate display lower *Kiss1* hybridization signals than controls in both brain regions during the pubertal transition period (Patisaul et al., 2012). This reduction has been evidenced as early as postnatal day 4 in the ARC and postnatal day 10 in the POA (Cao et al., 2012), suggesting that the *Kiss1* gene may represent an early target of this endocrine disruptor in the hypothalamus. By postnatal day 10, sex differences in *Kiss1* mRNA signal in the ARC are no longer observed following neonatal administration of estradiol benzoate (Cao et al., 2012). Interestingly, this study also suggests that a neonatal exposure to estradiol benzoate can induce a rapid downregulation of *Esr1* and *Esr2* transcript levels in various hypothalamic nuclei,including the POA and ARC (Cao et al., 2012). Of note, other hypothalamic targets of estradiol benzoate have recently been evidenced in the POA of newborn Sprague Dawley rats that had been exposed *in utero* to this endocrine disruptor, using a 48 gene TaqMan PCR-based array (Dickerson et al., 2011b). For instance, on the first day of postnatal life, the gene encoding prodynorphin was shown to be upregulated and genes encoding subtypes of NMDA and GABA receptors downregulated in the POA (Dickerson et al., 2011b). A similar result was obtained following prenatal PCB exposure (Dickerson et al., 2011b). At the age of 2 months, 4 POA genes out of 48 appear significantly reduced to male levels in the EDCexposedfemales,including the androgen receptor,NMDA receptor 2b, IGF1, and TGFβ1. It is noteworthy that all four of these genes play roles in hypothalamic development, including in the regulation of GnRH and kisspeptin neurons (Hiney et al., 2009; Oakley et al., 2009; Kurian et al., 2012). However, no changes in *Kiss1* expression levels could be detected in this study, despite the reduction in kisspeptin-immunoreactivity observed (Dickerson et al., 2011a). This may be due to a methodological limitation or to a different regulation between the mRNA and the protein. The same approach has been used to examine the molecular consequences in the POA of the aged female progeny, of a perinatal exposure to estradiol benzoate, a treatment that results in an acceleration of reproductive aging in Fisher rats (Gore et al., 2011). A complex regulatory neural/glial network of 17 genes controlling reproductivefunction appears upregulated in the POA by this perinatal estradiol benzoate treatment, including sex steroid hormone receptors, GABA and glutamate receptor subunits, growth factors, neuropeptide receptors, and a transcription factor. The *Kiss1* gene appears significantly downregulated and *Esr1* upregulated, relative to controls. Interestingly, an increase in methylation at some CpG sites in the *Esr1*gene is observed, suggesting that early exposure to this endocrine disruptor can induce lifelong epigenetic changes in this gene (Gore et al., 2011). It will be of great interest to further assess whether endocrine disruptors can also interfere with epigenetic marks on the *Kiss1* gene (Semaan et al., 2012) and to position *Kiss1* within this identified network of EDC-sensitive hypothalamic genes.

Bisphenol A (BPA), another endocrine disruptor with both estrogenic and anti-androgenic activities, may also target the kisspeptin–GPR54 system during development. However, studies are sparser and often incomplete, making it difficult to draw a clear picture of its mechanism of action. Administration of high (but not low) doses of BPA to neonatal female Long Evans rat results in adulthood in decreased kisspeptin-immunoreactivity in the ARC, independently of the steroidogenic milieu (Patisaul et al., 2009). If administered at high doses to neonatal Wistar rats, BPA induces a decrease in *Kiss1* hypothalamic expression levels, as well as a decrease in kisspeptin-immunoreactivity in the ARC at puberty (Navarro et al., 2009; Losa-Ward et al., 2012). If administered at low doses to neonatal Wistar rats, it advances the time of vaginal opening without any changes in kisspeptin-immunoreactivity being detectable in the ARC or POA at puberty. Finally, a low dose neonatal administration to Long Evans rats has recently been shown to decrease *Kiss1* hybridization signals in the POA as early as P10 (Cao et al., 2012). In CD1 mice, a lifelong-exposure to high doses of BPA results in an increase rather than a decrease in hypothalamic *Kiss1* mRNA levels (Xi et al., 2011). Furthermore, BPA-exposed mice display higher levels of circulating estradiol than control mice (Xi et al., 2011). Similarly, a perinatal exposure of female CD1 mice to extremely low levels of BPA results in an increase in the number of kisspeptin-immunoreactive cells in the POA (Panzica et al., 2011). These opposite effects of BPA between mice and rats illustrate the species-specificity of the effect of endocrine disruptors in general.

The consequences of real life exposure to EDCs was furthermore assessed in a farm animal species, the sheep. The fetuses (110 day old) of pregnant ewes exposed from the first day of conception to sewage sludge containing common endocrine disruptors display lower *Kiss1* mRNA levels in the ARC and in the POA relative to control animals maintained on pasture treated with conventional inorganic fertilizers (Bellingham et al., 2009). The physiological and behavioral outcomes of this exposure remains to be assessed.

Whether there is a causal link between disruption of the kisspeptin–GPR54 system and the different reproductive defects engendered by these EDCs remains to be fully investigated. Many recent studies on neuroendocrine disruption of reproductivefunction have focused on analysis of the kisspeptin–GPR54 system as potential target but it is clear that developmental exposure to some endocrine disruptors can advance the time of vaginal opening independently of kisspeptin signaling (Witham et al., 2012). This may in some cases derive from direct peripheral effects of the EDC or through interferences with other steroid-sensitive neural circuits regulating GnRH secretion. For example, neonatal exposure of female rats to low levels of BPA was recently found to advance the time of vaginal opening and has been associated to a decrease in the number of RFRP3-immunoreactive neurons and in the proportion of GnRH neurons displaying RFRP3-immunoreactive fiber appositions (Losa-Ward et al., 2012). On the other hand, no changes in kisspeptin-immunoreactivity could be detected (Losa-Ward et al., 2012).

#### **DIET**

It has been known for a long time that diet can have a strong impact on reproductive function, including on the timing of puberty onset. This observation has led several laboratories to investigate a potential regulatory role of diet on the development of the kisspeptin–GPR54 system. In mice, a high fat diet given from the time of weaning can induce infertility in the DJA strain but not in the C57/Bl6 strain. Accordingly, this food regimen results in adulthood in a decrease of ARC and POA *Kiss1* expression specifically in the DJA strain (Quennell et al., 2011). Mouse nutrition has also been manipulated during early postnatal development by varying litter sizes during lactation from postnatal day 4 until weaning and this was shown to produce long-lasting changes in kisspeptinimmunoreactivity and in physiological parameters (Caron et al., 2012). Under-nourished pups raised in large litters during lactation display delayed puberty onset and reduced fertility index and this has been associated with a decrease in the density of fibers double labeled for kisspeptin and NKB in medial preoptic nuclei (Caron et al., 2012). DiI anterograde tracing studies suggest that this results from a neonatal impairment of kisspeptin neural projections from the ARC toward the medial preoptic nucleus (Caron et al., 2012).

In rats, chronic undernutrition from the time of weaning onward (Castellano et al., 2005; Navarro et al., 2012) diminishes hypothalamic *Kiss1* mRNA levels around puberty. This is associated with delayed onset of vaginal opening, and decreased circulating levels of LH, effects that can be rescued by prepubertal chronic daily central administration of either kisspeptin (Castellano et al., 2005) or senktide, a NKB agonist (Navarro et al., 2012). Chronic undernutrition exclusively during the fetal period (Iwasa et al., 2010b) or during lactation (Castellano et al., 2011) is sufficient to induce long-term changes, at least until the pubertal period: puberty onset is again delayed and hypothalamic *Kiss1* mRNA levels are decreased. Neuroanatomical analysis further showed a

decrease in the number of kisspeptin immunoreactive cells in the ARC following undernutrition during lactation (Castellano et al., 2011). Conversely, overnutrition by litter size manipulation during lactation increases hypothalamic *Kiss1* mRNA levels at the pubertal transition period, advances puberty onset, and increases kisspeptin-immunoreactive fiber density in the POA (Castellano et al., 2011). High fat diet from the time of weaning can also advance puberty onset and results in an increase in *Kiss1* mRNA levels in the ARC prior to vaginal opening, followed by a decrease in the POA (Li et al., 2012).

#### **STRESS**

The functioning of the hypothalamic–pituitary–gonadal axis can also be altered by stressful experiences early in life. Female rats that are exposed to an immunological stress before 7 days of postnatal life exhibit a significant delay in puberty-associated with decreased *Kiss1* but not *GPR54* expression in the POA (Knox et al., 2009). Thus, kisspeptin neurons may represent important cellular relays through which stress-related factors impact the reactivation of GnRH pulsatile release at puberty. Notably, corticotropin-releasing factor receptors and glucocorticoid receptors have recently been detected by immunohistochemistry in kisspeptin neurons of the ARC in adulthood (Takumi et al., 2012b), implying a possible direct effect of stress-related factors on kisspeptin neurons. It will be interesting to determine when during development expression of these receptors start.

# **CONCLUSION**

Loss and gain of function studies during development have now provided compelling evidence that kisspeptin signaling in the brain is essential for the maturation of reproductive function through puberty in several mammalian species including humans. Upregulation of *Kiss1* transcription both in the ARC and POA appears to play a major role in the onset and progression through puberty in several mammalian species. Numerous potential regulators of *Kiss1* transcription during development have been identified and light has also recently been shed on developmental regulators of kisspeptin release, including NKB and GABA. Nevertheless, a lively debate still persists on the respective roles played by the ARC and POA populations of kisspeptin neurons in puberty onset, progression, and completion. For example,*Kiss1* expression or kisspeptin-immunoreactivity does not appear to increase in the ARC at the time of puberty in mice, as opposed to rats, sheep, and monkeys for which positive correlations have been found between *Kiss1* mRNA levels in the ARC and LH pulse frequencies. On the other hand, the mouse is the only speciesfor which an upregulation of *GPR54* expression has been described in GnRH neurons during female postnatal development. Since studies using different species often rely on different approaches with different sensitivities, it may be hazardous at this stage to put these different observations at the account of true species differences. Clearly,further investigations using a variety of complementary approaches on each animal model are needed in order to identify and ascertain species-specific processes in the developmental regulation and function of the kisspeptin–GPR54 system.

Numerous studies have shown that the kisspeptin–GPR54 system is particularly sensitive to gonadal steroids. In fact, all species studied develop a clear sexual dimorphism in the pattern of *Kiss1* expression in the POA (with greater expression in females) but the precise roles of developmental sex steroids in this process have so far only been studied in mice and rats using different yet complementary approaches. Estrogen receptor signaling appears to exert important organizational effects during the perinatal and peripubertal periods that may involve epigenetic regulations of the *Kiss1* and/or *Esr1* genes. In mice, rats, sheep, and humans, sex differences in the amounts of *Kiss1* expression and/or kisspeptinimmunoreactivity have also been detected in the ARC at some points of development (different depending on species) but the respective roles played by organizational and activational effects of sex steroids in these sex differences remain to be fully analyzed and documented. In mice, rats, and sheep, different yet dynamic sensitivities of male and female ARC kisspeptin cells to circulating sex steroids have been highlighted at the time of puberty. It seems clear that estrogen signaling can exert both organizational and activational effects on the kisspeptin–GPR54 system albeit at times of development and at cellular levels that can greatly vary between species and strains. More detailed studies on the mechanism of action of sex steroids on the development of the kisspeptin–GPR54 system with its neuroendocrine and behavioral consequences should be conducted in the future.

Moreover, studies in rodents suggest that kisspeptin cells may be reprogrammed during development by environmental threats including endocrine disruptors, diet, and stress, with long-term often deleterious effects on reproductive function. Therefore, it remains important to decipher for each species the critical periods of plasticity of the kisspeptin–GPR54 system and to better understand the molecular and cellular mechanisms involved in its developmental programing. Most recently, it has been shown in rats and humans that kisspeptins are already synthesized in some ARC cells well before birth. The physiological significance of these observations have yet to be revealed but one likely hypothesis is that kisspeptins already regulate tonic GnRH release prenatally, hence contributing to early maturation processes of the gonads and possibly to sexual differentiation of some brain circuits. The embryonic period of progenitor cell proliferation and neurogenesis has recently been identified for the ARC kisspeptin neurons of rat. This represents an important first step in the exploration of the morphogenetic processes shaping this neuronal system during early development. In a translational perspective, these studies should help the development of predictive cellular models for assessing the danger of environmental chemicals on reproductive function.

# **ACKNOWLEDGMENTS**

Authors are very grateful to Drs. Alain Caraty, Yves Tillet, and Anne Duittoz for valuable discussions and assistance in performing the research presented in this review, to the anonymous referees that have stimulated improvements of this review article and to Laura Szymanski for carefully checking the English of this manuscript. The work presented herein was supported by the Institut National de Recherche Agronomique (INRA), Centre National de la Recherche Scientifique (CNRS), Université of Tours, and by grants from the French National Research Agency (ANR) and Région Centre. Elodie Desroziers was recipient of a Ph.D. fellowship from the INRA and Région Centre.

# **REFERENCES**


female rats. *J. Comp. Neurol.* 519, 2954–2977.


neurons. *Endocrinology* 147, 5817–5825.


serum gonadotropins, prolactin and gonadal steroids in prepubertal male and female rats. *Endocrinology* 97, 898–907.


*J. Physiol. Endocrinol. Metab.* 303, E1252–E1263.


Developmental changes in GnRH release in response to kisspeptin agonist and antagonist in female rhesus monkeys (*Macaca mulatta*): implication for the mechanism of puberty. *Endocrinology* 153, 825–836.


immunohistochemical study on the expressional dynamics of kisspeptin neurons relevant to GnRH neurons using a newly developed anti-kisspeptin antibody. *J. Mol. Neurosci.* 43, 146–154.


L., Chaudhary, A. A., et al. (2009). Down-regulation of hypothalamic kisspeptin and its receptor, Kiss1r, mRNA expression is associated with stress-induced suppression of luteinising hormone secretion in the female rat. *J. Neuroendocrinol.* 21, 20–29.


Patisaul, H. B. (2011). Neonatal exposure to genistein adversely impacts the ontogeny of hypothalamic kisspeptin signaling pathways and ovarian development in the peripubertal female rat. *Reprod. Toxicol.* 31, 280–289.


(2011). Transcriptional regulation of the human KiSS1 gene. *Mol. Cell. Endocrinol.* 342, 8–19.


the brain. *Endocr. Rev.* 30, 713–743.


mechanisms of gonadotropin regu-


female reproduction and kisspeptin neurons in mice. *Biol. Reprod.* 86, 119.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 October 2012; paper pending published: 24 October 2012; accepted: 22 February 2013; published online: 28 March 2013.*

*Citation: Franceschini I and Desroziers E (2013) Development and aging of the kisspeptin–GPR54 system in the mammalian brain: what are the impacts on female reproductive function? Front. Endocrinol. 4:22. doi: 10.3389/fendo.2013.00022*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Franceschini and Desroziers. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Gonadotropin-inhibitory hormone action in the brain and pituitary

# *Takayoshi Ubuka,You Lee Son,Yasuko Tobari and Kazuyoshi Tsutsui\**

Laboratory of Integrative Brain Sciences, Department of Biology, Center for Medical Life Science, Waseda University, Tokyo, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Lance Kriegsfeld, University of California, USA José A. Muñoz-Cueto, University of Cadiz, Spain

#### *\*Correspondence:*

Kazuyoshi Tsutsui, Laboratory of Integrative Brain Sciences, Department of Biology, Center for Medical Life Science, Waseda University, 2-2 Wakamatsu-cho, Shinjuku-ku, Tokyo 162-8480, Japan. e-mail: k-tsutsui@waseda.jp

Gonadotropin-inhibitory hormone (GnIH) was first identified in the Japanese quail as a hypothalamic neuropeptide inhibitor of gonadotropin secretion. Subsequent studies have shown that GnIH is present in the brains of birds including songbirds, and mammals including humans. The identified avian and mammalian GnIH peptides universally possess an LPXRFamide (X = L or Q) motif at their C-termini. Mammalian GnIH peptides are also designated as RFamide-related peptides from their structures. The receptor for GnIH is the G protein-coupled receptor 147 (GPR147), which is thought to be coupled to Gα<sup>i</sup> protein. Cell bodies of GnIH neurons are located in the paraventricular nucleus (PVN) in birds and the dorsomedial hypothalamic area (DMH) in mammals. GnIH neurons in the PVN or DMH project to the median eminence to control anterior pituitary function. GPR147 is expressed in the gonadotropes and GnIH suppresses synthesis and release of gonadotropins. It was further shown in immortalized mouse gonadotrope cell line (LβT2 cells) that GnIH inhibits gonadotropin-releasing hormone (GnRH) induced gonadotropin subunit gene transcriptions by inhibiting adenylate cyclase/cAMP/PKA-dependent ERK pathway. GnIH neurons also project to GnRH neurons in the preoptic area, and GnRH neurons express GPR147 in birds and mammals. Accordingly, GnIH may inhibit gonadotropin synthesis and release by decreasing the activity of GnRH neurons as well as directly acting on the gonadotropes. GnIH also inhibits reproductive behavior possibly by acting within the brain. GnIH expression is regulated by a nocturnal hormone melatonin and stress in birds and mammals. Accordingly, GnIH may play a role in translating environmental information to inhibit reproductive physiology and behavior of birds and mammals. Finally, GnIH has therapeutic potential in the treatment of reproductive cycle and hormone-dependent diseases, such as precocious puberty, endometriosis, uterine fibroids, and prostatic and breast cancers.

**Keywords: gonadotropin-inhibitory hormone, gonadotropin-releasing hormone, gonadotropins, reproductive behavior, melatonin, stress, GPR147, RFamide-related peptide**

# **INTRODUCTION**

The decapeptide gonadotropin-releasing hormone (GnRH) is the primary factor responsible for the hypothalamic control of gonadotropin secretion. GnRH was first isolated from mammals (Matsuo et al., 1971; Burgus et al., 1972) and subsequently from birds (King and Millar, 1982; Miyamoto et al., 1982, 1984) and other vertebrates (for reviews, see Millar, 2003, 2005). Gonadal sex steroids and inhibin can modulate gonadotropin secretion. However, no hypothalamic neuropeptide inhibitor of gonadotropin secretion was known in vertebrates, although dopamine has been reported as an inhibitor of gonadotropin secretion in several fish groups. In 2000, a previously unidentified hypothalamic neuropeptide was shown to inhibit gonadotropin release from the cultured quail anterior pituitary gland and it was named gonadotropin-inhibitory hormone (GnIH; Tsutsui et al., 2000). Although it is now known that GnIH and its receptor are also expressed in the gonads of birds (Bentley et al., 2008; Maddineni et al., 2008; McGuire and Bentley, 2010; McGuire et al., 2011) and mammals (Zhao et al., 2010; Singh et al., 2011a,b; Li et al., 2012) including humans (Oishi et al., 2012), this review highlights the discovery of GnIH in the quail brain and the progress of GnIH research investigating its function in the brain and pituitary of birds and mammals. We also briefly review recent progresses in the study of GnIH peptides in fish and humans.

# **DISCOVERY OF GnIH IN BIRDS**

"fendo-03-00148" — 2012/11/26 — 21:47 — page 1 — #1

Gonadotropin-inhibitory hormone was first discovered in the brain of Japanese quail, *Coturnix japonica*, while searching a novel RFamide peptide in birds. RFamide peptides, which have an Arg-Phe-NH2 motif at its C-terminus, were first isolated in invertebrate species in the late 1970s. The first RFamide peptide, Phe-Met-Arg-Phe-NH2 (FMRFamide), was a cardioexcitatory molecule isolated from the ganglia of the venus clam *Macrocallista nimbosa* (Price and Greenberg, 1977). After the discovery of FMRFamide peptide, numerous RFamide peptides that act as neurotransmitters, neuromodulators, and peripheral hormones have been identified in various invertebrates, including cnidarians, nematodes, annelids, molluscs, and arthropods. Subsequent immunohistochemical studies in vertebrates suggested the presence of RFamide peptides in the central nervous system. It was revealed that some of the FMRFamide-immunoreactive (-ir) neurons projected close to the pituitary gland, suggesting a role in the regulation of pituitary function in vertebrates.

Tsutsui et al. (2000) have isolated a novel RFamide peptide from 500 brains of the Japanese quail by high-performance liquid chromatography (HPLC) and a competitive enzyme-linked immunosorbent assay using an antibody raised against the dipeptide Arg-Phe-NH2. The isolated peptide had a novel dodecapeptide structure, SIKPSAYLPLRFamide. Its C-terminus was identical to chicken LPLRFamide that was reported to be the first RFamide peptide isolated in vertebrates (Dockray et al.,1983), which is likely to be a degraded fragment of the dodecapeptide. The isolated peptide was localized in the hypothalamo-hypophyseal system, and shown to decrease gonadotropin release from cultured quail anterior pituitary glands (Tsutsui et al., 2000). This RFamide peptide was therefore named GnIH (Tsutsui et al., 2000).

Following the isolation of GnIH in the quail brain, the precursor polypeptide for GnIH was determined (Satake et al., 2001). A cDNA that encoded GnIH precursor polypeptide was identified by a combination of 3 and 5 rapid amplification of cDNA ends (3- /5- RACE; Satake et al., 2001). The GnIH precursor consisted of 173 amino acid residues that encoded one GnIH and two GnIH-related peptides (GnIH-RP-1 and GnIH-RP-2) possessing an LPXRFamide (X = L or Q) sequence at their C-termini (**Figure 1**). These peptide sequences were flanked by a glycine C-terminal amidation signal and a single basic amino acid on each end as an endoproteolytic site (**Figure 1**). GnIH-RP-2 was also identified as a mature peptide by mass spectrometry in quail (**Figure 1**; Satake et al., 2001). GnIH was further isolated as mature peptides in European starlings (Ubuka et al., 2008a) and zebra finch (Tobari et al., 2010) within the class of birds (**Figure 1**; for reviews, see Tsutsui and Ukena, 2006; Tsutsui et al., 2007, 2010a,b, 2012; Tsutsui, 2009; Tsutsui and Ubuka, 2012).

# **IDENTIFICATION OF GnIH IN MAMMALS**

In mammals, cDNAs that encode GnIH orthologs, LPXRFamide peptides, have been investigated by a gene database search (Hinuma et al., 2000; for a review, see Tsutsui and Ubuka, 2012). Mammalian LPXRFamide peptides are also designated as RFamide-related peptides (RFRP) from its structure. Although human, macaque, and bovine LPXRFamide precursor cDNAs encoded three RFRPs (RFRP-1, -2, and -3), only RFRP-1 and RFRP-3 possessed a C-terminal LPXRFamide (X = L or Q) motif, and RFRP-2 had C-terminal RSamide or RLamide sequences (**Figure 1**). On the other hand, rodents do not have RFRP-2 sequence in their precursors (see the precursor sequences of rat and hamster in **Figure 1**). Although the positions that encode GnIH-RP-1/RFRP-1 or GnIH/RFRP-2 in the precursor polypeptides were conserved between birds and mammals (only RFRP-1 in rodents), the positions of GnIH-RP-2 and RFRP-3 in their precursor polypeptides were different between birds and mammals (**Figure 1**).

The LPXRFamide motif at the C-terminus is followed by glycine as an amidation signal and arginine or lysine as endoproteolytic basic amino acids in mammals as well as birds (**Figure 1**). Endogenous LPXRFamide peptides can also be cleaved at basic amino acids at their N-termini. However, there were some exceptions in the cleavage site at the N-terminal, such as that of

Siberian hamster RFRP-1 (**Figure 1**). Up until now, bovine RFRP-1 (Fukusumi et al., 2001) and -3 (Yoshida et al., 2003), rat RFRP-3 (Ukena et al., 2002), Siberian hamster RFRP-1 and -3 (Ubuka et al., 2012a), macaque RFRP-3 (Ubuka et al., 2009b), and human RFRP-1 and -3 (Ubuka et al., 2009c) are identified as mature peptides in mammals (**Figure 1**; for reviews, see Tsutsui and Ukena, 2006; Tsutsui et al., 2007, 2010a,b, 2012; Tsutsui, 2009; Tsutsui and Ubuka, 2012).

# **GnIH RECEPTOR**

Bonini et al. (2000) have identified two G protein-coupled receptor (GPCR) for neuropeptide FF (NPFF), and designated them as NPFF1 (identical to GPR147) and NPFF2 (identical to GPR74). Hinuma et al. (2000) have also reported a specific receptor for RFRP and named it OT7T022, which was identical to GPR147. The binding affinities for GPR147 and GPR74 and the efficacies on signal transduction pathway were examined, using various analogs of RFRPs and NPFF. RFRPs showed a higher affinity for GPR147, whereas NPFF had potent agonistic activity for GPR74 (Bonini et al., 2000; Liu et al., 2001). Taken together, GPR147 (NPFF1, OT7T022) was suggested to be the receptor for RFRP (mammalian GnIH). **Figure 2** shows the predicted two-dimensional structure of human GPR147 from its nucleotide sequence (AB040104; Ubuka et al.,2008b, 2009c). It was shown that RFRPs suppress the production of cAMP in ovarian cells of Chinese hamster transfected with GPR147 cDNA, suggesting that GPR147 couples to Gα<sup>i</sup> protein (Hinuma et al., 2000).

To elucidate the mode of action of GnIH in birds, Yin et al. (2005) have identified GnIH receptor (GnIH-R) in the quail diencephalon and characterized its expression and binding activity. First, a cDNA encoding a putative GnIH-R was cloned by a combination of 3- /5- RACE using PCR primers designed from the sequence of the receptor for RFRPs (GPR147). The crude membranefraction of COS-7 cells transfected with the putative GnIH-R cDNA specifically bound GnIH and GnIH-RPs in a concentrationdependent manner, indicating that GPR147 is GnIH-R (Yin et al., 2005). GnIH-R also bound with high affinities to GnIH, GnIH-RPs and RFRPs, which have LPXRFamide (X = L or Q) motif at their C-termini. In contrast, C-terminal non-amidated GnIH failed to bind the receptor. Accordingly, the C-terminal LPXR-Famide (X = L or Q) motif seems to be critical for its binding to GnIH-R (Yin et al., 2005; for reviews, see Tsutsui and Ukena, 2006; Tsutsui et al., 2007, 2010a,b, 2012; Tsutsui, 2009; Tsutsui and Ubuka, 2012). It was suggested that there is no functional difference among GnIH and GnIH-RPs because GnIH-R bound GnIH and GnIH-RPs with similar affinities (Yin et al., 2005). However, further studies are required to investigate if GnIH and GnIH-RPs work additively or synergistically to achieve their effects on the cells that express GnIH-R.

# **DISTRIBUTION OF GnIH CELLS AND FIBERS IN THE BRAIN LOCATION OF GnIH NEUNONS IN THE BRAIN**

The location of GnIH precursor mRNA was first investigated by Southern blot analysis of the RT-PCR products of GnIH precursor cDNA. Within the samples from telencephalon, diencephalon, mesencephalon, and cerebellum, GnIH precursor mRNA was only expressed in the quail diencephalon (Satake et al., 2001). *In situ*

"fendo-03-00148" — 2012/11/26 — 21:47 — page 2 — #2

"fendo-03-00148" — 2012/11/26 — 21:47 — page 3 — #3

GnIH-RP-2, RFRP-1, RFRP-3 with Gly (G) as an amidation signal and Arg (R) or Lys (K) as an endoproteolytic basic amino acid at the Ctermini are shown in bold. Identified mature peptides are underlined. Endoproteolytic basic amino acid (R or K) at the N-termini are also shown hamster (JF727837), white-crowned sparrow (AB128164), zebra finch (AB522971), European starling (EF486798) and Japanese quail (AB039815).

hybridization for GnIH precursor mRNA further showed that cells expressing GnIH mRNA were clustered in the paraventricular nucleus (PVN) in the hypothalamus (Ukena et al., 2003). Immunohistochemistry using GnIH antibody has revealed that clusters of GnIH-ir neurons were expressed in the PVN in quail

(Tsutsui et al., 2000; Ubuka et al., 2003). GnIH expressing cell bodies were also clustered in the PVN in other birds (Bentley et al., 2003; Osugi et al., 2004; Ubuka et al., 2008a).

In mammals, expression of precursor mRNA of RFRP (mammalian GnIH) was only detected in the dorsomedial hypothalamic

from Ubuka et al. (2008b).

area (DMH) in the mouse and hamster brains by *in situ* hybridization (Kriegsfeld et al., 2006; Ubuka et al., 2012a). In the rat brain, RFRP precursor mRNA was expressed in the periventricular nucleus (PerVN), and the portion between the dorsomedial nucleus (DMN) and the ventromedial nucleus (VMN) of the hypothalamus (Hinuma et al., 2000; Legagneux et al., 2009). The majority of RFRP mRNA expressing neuronal cell bodies were localized in the intermediate periventricular nucleus (IPe) of the hypothalamus in the macaque (Ubuka et al., 2009b), and in the DMN and PVN in the sheep (Clarke et al., 2008).

# **GnIH INNERVATION IN THE BRAIN**

"fendo-03-00148" — 2012/11/26 — 21:47 — page 4 — #4

Although a dense population of GnIH neuronal cell bodies was only found in the PVN in quail, GnIH-ir neuronal fibers were widely distributed in the diencephalic and mesencephalic regions (Ukena et al., 2003). Dense networks of GnIH-ir fibers were found in the ventral paleostriatum, septal area, preoptic area (POA), median eminence, optic tectum, and the dorsal motor nucleus of the vagus. GnIH-ir neuronal fibers were also widely distributed in the diencephalic and mesencephalic regions in European starlings (Ubuka et al., 2008a) and white-crowned sparrows (Ubuka et al., 2012b). Thus, it was hypothesized that GnIH may participate not only in the regulation of pituitary function, but also in behavioral and autonomic mechanisms in birds.

GnIH-ir neuronal fibers were also widely distributed in the diencephalic and mesencephalic regions in rodents. Dense GnIHir fibers were observed in the lateral septal nucleus, medial POA, amygdala, arcuate nucleus (ARC); moderate GnIH-ir fibers were observed in the paraventricular thalamic nucleus, the paraventricular hypothalamic nucleus, and the central gray in Siberian hamsters (Ubuka et al., 2012a). Dense GnIH-ir fibers were also observed in limbic and hypothalamic structures in rats (Johnson et al., 2007). Accordingly, it was hypothesized that GnIH may also participate in behavioral and autonomic mechanisms in mammals.

Innervation of GnIH neuronal fibers was intensively investigated in the rhesus macaque brain (Ubuka et al.,2009b). Abundant GnIH-ir fibers were observed in the nucleus of the stria terminalis in the telencephalon; habenular nucleus, PVN of the thalamus, POA, PVN of the hypothalamus, IPe, ARC of hypothalamus, median eminence and dorsal hypothalamic area in the diencephalon; medial region of the superior colliculus, central gray substance of the midbrain and dorsal raphe nucleus in the midbrain; and parabrachial nucleus in the pons. GnIH-ir fibers were observed in close proximity to GnRH-I, dopamine, proopiomelanocortin, and GnRH-II neurons in the POA, IPe, ARC of hypothalamus, and central gray substance of midbrain, respectively. Qi et al. (2009) have shown that RFRP (mammalian GnIH) cells project to neuropeptideY and pro-opiomelanocortin neurons in the ARC, orexin and melanin-concentrating hormone neurons in the lateral hypothalamic area, as well as orexin cells in the DMN and corticotrophin-releasing hormone and oxytocin cells in the PVN, GnRH neurons in the POA in sheep. GnIH neurons might thus regulate these important neural systems in addition to directly regulating pituitary gonadotropin release (Ubuka et al., 2009b; for reviews, see Tsutsui, 2009; Tsutsui et al., 2010a,b, 2012; Tsutsui and Ubuka, 2012; **Figure 3**).

#### **GnIH ACTION IN THE BRAIN**

#### **MODULATION OF THE ACTIVITY OF GnRH AND KISSPEPTIN NEURONS BY GnIH**

Immunohistochemical studies using light and confocal microscopy indicated that GnIH (RFRP)-ir axon terminals are in probable contact with GnRH neurons in birds (Bentley et al., 2003), rodents (Kriegsfeld et al., 2006; Ubuka et al., 2012a), monkeys (Ubuka et al., 2009b), and humans (Ubuka et al., 2009c). Thus, there is potential for the direct regulation of the activity of GnRH neurons by GnIH (RFRP) neurons (**Figure 3**).

Ubuka et al. (2008a) investigated the interaction of GnIH neuronal fibers and GnRH neurons in the European starling brain. It is generally accepted that birds possess at least two forms of GnRH in their brains. One form is GnRH-I which is thought to be released at the median eminence to stimulate the secretion of gonadotropins from the anterior pituitary (King and Millar, 1982; Miyamoto et al., 1982; Sharp et al., 1990; Ubuka and Bentley, 2009, 2011; Ubuka et al., 2009a). The second form of GnRH is GnRH-II (Miyamoto et al., 1984; Millar, 2003), which is thought to influence reproductive behaviors in birds (Maney et al., 1997) and mammals (Temple et al., 2003; Barnett et al., 2006). Double-label immunocytochemistry showed GnIH axon terminals on GnRH-I and GnRH-II neurons in the songbird brain (Bentley et al., 2003; Ubuka et al., 2008a). *In situ* hybridization of starling GnIH-R mRNA combined with GnRH immunocytochemistry further showed the expression of GnIH-R mRNA in GnRH-I and GnRH-II neurons (Ubuka et al., 2008a; **Figure 3**).

Double-label immunocytochemistry also showed GnIH axon terminals on GnRH neurons in the Siberian hamster brain (Ubuka et al., 2012a) and GPR147 was expressed in GnRH neurons (Ubuka et al., 2012a). Central administration of GnIH inhibits the release of gonadotropin in white-crowned sparrows (Bentley et al., 2006), Syrian hamsters (Kriegsfeld et al., 2006), rats (Johnson et al., 2007), and Siberian hamsters (Ubuka et al., 2012a) in a manner similar to peripheral administration of GnIH (Osugi et al., 2004; Kriegsfeld et al., 2006; Ubuka et al., 2006). Accordingly, GnIH may inhibit the secretion of gonadotropins by decreasing the activity of GnRH neurons in addition to directly regulating pituitary gonadotropin secretion (**Figure 3**).

Direct application of RFRP-3 to GnRH cells in cultured mouse brain slices decreased firing rate in a subpopulation of cells, further indicating a direct action of RFRP-3 on GnRH neurons (Ducret et al., 2009). In addition, RFRP-3 inhibited firing of kisspeptin-activated vGluT2 (vesicular glutamate transporter 2)-GnRH neurons as well as of kisspeptin-insensitive GnRH neurons (Wu et al., 2009). More recent data have confirmed a role for RFRP-3 using an antagonist, RF9, against GnIH-R. Central administration of RF9 to rats and mice led to marked increases in gonadotropin concentrations, providing a pronounced role of RFRP-3 as a key regulator of the reproductive axis in mammals (Pineda et al., 2010).

Recently, Rizwan et al. (2012) have investigated (1) whether RFRP-3 can directly inhibit LH secretion without inhibiting GnRH neurons; (2) whether RFRP-3 neurons project to GnRH neurons and rostral periventricular kisspeptin neurons in mice, and (3) whether GPR147 and GPR74 are expressed by these neurons. Intravenous treatment with the GPR147 antagonist RF9 increased plasma LH concentrations in castrated male rats but was unable to do so in the presence of the GnRH antagonist cetrorelix. Approximately 26% of GnRH neurons of male and diestrous female mice were apposed by RFRP-3 fibers, and 19% of kisspeptin neurons of proestrous female mice were apposed by RFRP-3 fibers. They further showed that 33% of GnRH neurons and 9–16% of rostral periventricular kisspeptin neurons expressed GPR147, whereas GPR74 was not expressed in either population. These data show that RFRP-3 can act on GnRH neurons as well as kisspeptin neurons to modulate reproduction in rodents (**Figure 3**).

#### **EFFECT OF CENTRAL ADMINISTRATION OF GnIH ON BEHAVIORS OF BIRDS AND MAMMALS**

Central administration of GnIH or RFRP-3 to the third ventricle of the brain inhibited reproductive behavior of females in white-crowned sparrows (Bentley et al., 2006) or of males in rats (Johnson et al., 2007). It was known that GnRH-II enhances copulation solicitation in estrogen-primed female white-crowned sparrows exposed to the song of males (Maney et al., 1997).

"fendo-03-00148" — 2012/11/26 — 21:47 — page 5 — #5

"fendo-03-00148" — 2012/11/26 — 21:47 — page 6 — #6

activity of GnRH-I neurons as well as directly inhibiting the pituitary gonadotrope. GnIH (RFRP) neurons may also regulate GnRH-I neurons by regulating the activity of kisspeptin (Kiss) neurons that project to GnRH-I neurons. There are also reports showing that GnIH (RFRP) neurons project their axons to GnRH-II, dopamine, pro-opiomelanocortin (POMC),

receptor (Mel-R), glucocorticoid receptor (GC-R), or estrogen receptor α (ERα) in GnIH (RFRP) neurons were shown in several species. These mechanisms of action of GnIH (RFRP) on gonadotropin secretion or regulatory mechanism of GnIH (RFRP) expression may vary between species, sexes, and developmental stages.

Because of the putative contact of GnIH neurons with GnRH-II neurons in white-crowned sparrows (Bentley et al., 2003), Bentley et al. (2006) investigated the effect of GnIH administration on copulation solicitation in females of this species. A centrallyadministered physiological dose of GnIH inhibited copulation solicitation in estrogen-primed female white-crowned sparrows exposed to the song of males. Johnson et al. (2007) investigated the effect of central administration of RFRP-3 on reproductive behaviors of male rats. Behavioral tests indicated that RFRP-3 dose-dependently suppressed all facets of male sexual behavior. In contrast, immunoneutralization of RFRP in the rat brain increased male sexual behaviors. These results suggest that GnIH and RFRP inhibit reproductive behavior by inhibiting GnRH neuronal activities or by acting directly within the brain (**Figure 3**).

To identify the mechanism of GnIH neurons in the regulation of behavior, Ubuka et al. (2012b) investigated the effect of

RNA interference (RNAi) of the GnIH gene on the behavior of white-crowned sparrows, a highly social songbird species. Administration of small interfering RNA against GnIH precursor mRNA into the third ventricle of male and female birds reduced resting time, spontaneous production of complex vocalizations, and stimulated brief agonistic vocalizations. GnIH RNAi further enhanced song production of short duration in male birds when they were challenged by playbacks of novel male songs. These behaviors resembled those of breeding birds during territorial defense. The overall results suggest that GnIH gene silencing induces arousal. In addition, the activities of male and female birds were negatively correlated with GnIH mRNA expression in the PVN. Density of GnIH neuronal fibers in the ventral tegmental area was decreased by GnIH RNAi treatment in female birds, and the number of GnRH neurons that received close appositions of GnIH neuronal fiber terminals was negatively correlated with the activity of

male birds. In summary, GnIH may decrease arousal level resulting in the inhibition of specific motivated behavior, such as in reproductive contexts (Ubuka et al., 2012b).

Central administrations of GnIH or RFRP-3 can also stimulate feeding behavior in chicken or rats (Tachibana et al., 2005; Johnson et al., 2007; Murakami et al., 2008). There is a recent report showing that central administration of RFRP-3 can stimulate adrenocorticotropic hormone and oxytocin release, and induce anxiety behavior in rats (Kaewwongse et al., 2011). The fact that RFRP-ir fibers project to various neurons in the brain, such as dopamine and/or pro-opiomelanocortin neurons in the rat, sheep, and macaque (Qi et al., 2009; Ubuka et al., 2009b; **Figure 3**) suggests multiple functions of GnIH or RFRP in the brain.

# **GnIH ACTION IN THE PITUITARY**

Dense population of GnIH-ir fibers at the median eminence (ME) in quail (Tsutsui et al., 2000; Ukena et al., 2003) as well as in other birds (Bentley et al., 2003; Osugi et al., 2004; Ubuka et al., 2008a), suggested a direct action of GnIH in the regulation of pituitary function in birds (**Figure 3**). The fact that GnIH inhibits synthesis and/or release of gonadotropins from cultured quail and chicken anterior pituitary gland provides strong support for this function (Tsutsui et al., 2000; Ciccone et al., 2004). Peripheral administration of GnIH also inhibits gonadotropin synthesis and/or release in birds (Osugi et al., 2004; Ubuka et al., 2006). In mammals, abundant RFRP-ir fibers were observed in the ME of sheep (Clarke et al., 2008), macaque (Ubuka et al., 2009b), and humans (Ubuka et al., 2009c). As GnIH in birds, RFRP-3 inhibits gonadotropin synthesis and/or release from cultured pituitaries in sheep (Sari et al., 2009) and cattle (Kadokawa et al., 2009). Peripheral administration of RFRP-3 also inhibits gonadotropin release in sheep (Clarke et al., 2008), rats (Murakami et al., 2008), and cattle (Kadokawa et al., 2009). It was further shown that GnIH-R (GPR147) mRNA is expressed in gonadotropes in the human pituitary (Ubuka et al., 2009c). Taken together, it is likely that GnIH and RFRP-3 directly act on the pituitary to inhibit gonadotropin secretion from the pituitary at least in these avian and mammalian species (**Figure 3**).

Recently, Smith et al. (2012) measured the concentration of RFRP-3 in hypophyseal portal blood in ewes during the nonbreeding (anestrous) season and during the luteal and follicular phases of the estrous cycle in the breeding season. Pulsatile RFRP-3 secretion was observed in the portal blood of all animals. Mean RFRP-3 pulse amplitude and pulse frequency were higher during the non-breeding season. RFRP-3 was virtually undetectable in peripheral blood plasma. To determine the role of secreted RFRP-3, Smith et al. (2012)further examined its effects on GnRHstimulated LH secretion in hypothalamo–pituitary-disconnected ewes, and a significant reduction in the LH response to GnRH was observed by RFRP-3 administration. These data show that RFRP-3 is secreted into portal blood to act on pituitary gonadotropes, reducing the action of GnRH in sheep (Smith et al., 2012).

On the contrary it was suggested that GnIH or RFRP-3 may not act directly on the pituitary in some birds and rodents, because there are relatively few or no GnIH (RFRP)-ir fibers in the ME of Rufous-winged sparrows (Small et al., 2007), hamsters (Kriegsfeld et al., 2006; Ubuka et al., 2012a), and rats (Rizwan et al., 2009). Rizwan et al. (2009) have injected a retrograde tracer Fluoro-Gold intraperitoneally in rats. The majority of GnRH neurons were labeled but essentially no RFRP neurons were labeled. In contrast, intracerebral injections of Fluoro-Gold into the rostral POA resulted in the labeling of 75 ± 5% of RFRP cell bodies. These observations suggested that RFRP-3 is not a hypophysiotropic neuroendocrine hormone in rats (Rizwan et al., 2009). However, there are also studies indicating that RFRP-3 can act directly to inhibit gonadotropin release from the pituitary of rats (Murakami et al., 2008). More extensive studies, analyzing peptide release into the hypophyseal portal blood, fiber projections by retrograde labeling, etc., are needed to elucidate the hypophysiotropic action of GnIH in various animals (for reviews, see Tsutsui et al., 2007, 2010a,b, 2012; Tsutsui, 2009; Tsutsui and Ubuka, 2012).

Since GPR147 couples to Gα<sup>i</sup> to inhibit adenylyl cyclase (AC; Hinuma et al., 2000), GPR147 (GnIH-R) activation may reduce intracellular cAMP levels, and reduce the activities of cAMPdependent protein kinase (PKA) and mitogen-activated protein kinase (MAPK) signaling cascade. On the other hand, GnRH receptor (GnRH-R) couples with Gαq/<sup>11</sup> to activate phospholipase C (PLC), which results in the production of inositol tri-phosphate (IP3) and diacylglycerol (DAG). In turn, IP3 and DAG increase intracellular Ca2<sup>+</sup> and activate the protein kinase C (PKC) pathway and MAPK signaling. GnRH-R was also reported to be coupled to Gα<sup>s</sup> to stimulate AC/cAMP/PKA pathway. A recent study using immortalized mouse gonadotrope cell line (LβT2 cells) has demonstrated that the inhibitory action of mouse RFRPs on gonadotropin gene expression is mediated by an inhibition of AC/cAMP/PKA-dependent extracellular signal-regulated kinase (ERK) pathway (Son et al., 2012). In the sheep, RFRP-3 can inhibit both GnRH-induced intracellular Ca2<sup>+</sup> increase and ERK phosphorylation, impacting GnRH-induced gonadotropin release and synthesis (Sari et al., 2009). In the chicken, activation of GnRH-R can activate AC and stimulate cAMP responsive element (CRE) binding protein. GnIH may partly inhibit this GnRH-induced CRE activation, thus impacting gene transcription (Shimizu and Bédécarrats, 2010).

# **REGULATION OF GnIH EXPRESSION EFFECT OF MELATONIN**

"fendo-03-00148" — 2012/11/26 — 21:47 — page 7 — #7

Identification of the regulatory mechanisms governing GnIH expression is important in understanding the physiological role of the GnIH system. Photoperiodic mammals rely on the annual cycle of changes in nocturnal secretion of melatonin to drive their reproductive responses (Bronson, 1990). In contrast, a dogma has existed that birds do not use seasonal changes in melatonin secretion to time their reproductive effort, and a role for melatonin in birds has remained enigmatic (Wilson, 1991; Juss et al., 1993). Despite the accepted dogma, there is strong evidence that melatonin is involved in the regulation of several seasonal processes, including gonadal activity and gonadotropin secretion (Ohta et al., 1989; Guyomarc'h et al., 2001; Rozenboim et al., 2002). Ubuka et al. (2005) hypothesized that melatonin may be involved in the induction of GnIH expression, thus influencing gonadal activity. The action of melatonin on the expression of GnIH was studied in quail, a highly photoperiodic bird species. Because the pineal gland and eyes are the major sources of melatonin in the quail (Underwood et al., 1984), Ubuka et al. (2005) analyzed the effects of pinealectomy (Px) combined with orbital enucleation (Ex; Px plus Ex) on the expression of GnIH precursor mRNA and GnIH peptide. Subsequently, melatonin was administered to Px plus Ex birds. Px plus Ex decreased the expression of GnIH precursor mRNA and the content of mature GnIH peptide in the hypothalamus. Melatonin administration to Px plus Ex birds caused a dose-dependent increase in the expression of GnIH precursor mRNA and the production of mature peptide. They also investigated the expression of melatonin receptor in GnIH neurons. *In situ* hybridization combined with immunocytochemistry for GnIH revealed that the mRNA of Mel1c, a melatonin receptor subtype, was expressed in GnIH-ir neurons in the PVN. Autoradiography of melatonin receptors further revealed specific binding of melatonin in the PVN. Accordingly, melatonin appears to act directly on GnIH neurons through its receptor to induce expression of GnIH (Ubuka et al., 2005; **Figure 3**).

Chowdhury et al. (2010) further investigated the role of melatonin in the regulation of GnIH release and the negative correlation of GnIH release with LH release in quail. Melatonin administration dose-dependently increased GnIH release from hypothalamic explants *in vitro*. A clear diurnal change in GnIH release was observed in quail, and this change was negatively correlated with changes in plasma LH concentrations. GnIH release during the dark period was greater than that during the light period in explants from quail exposed to long-day (LD) photoperiods. Conversely, plasma LH concentrations decreased during the dark period. In contrast to LD, GnIH release increased under short-day (SD) photoperiods, when the duration of nocturnal secretion of melatonin increases. These results indicate that melatonin may play a role in stimulating not only GnIH expression but also GnIH release, thus inhibiting plasma LH concentrations in quail (**Figure 3**).

A similar, but opposite, action of melatonin on the inhibition of the expression of RFRP was shown in Syrian and Siberian hamsters, both photoperiodic mammals (Revel et al., 2008; Mason et al., 2010; Ubuka et al., 2012a). The level of RFRP mRNA and the number of RFRP-ir cell bodies were reduced in sexually quiescent Syrian and Siberian hamsters acclimated to SD photoperiod, compared to sexually active animals maintained under LD photoperiod. The photoperiodic variation of RFRP expression was abolished in Px hamsters and injections of LD hamsters with melatonin reduced the expression of RFRP down to SD levels, indicating a dependence upon melatonin (Revel et al.,2008; Ubuka et al., 2012a). There are also reports showing that the expression of RFRP precursor mRNA is regulated by melatonin in sheep (Dardente et al., 2008) and rats (Gingerich et al., 2009). These results demonstrate that the expression of GnIH and RFRP is photoperiodically modulated via a melatonin-dependent process (**Figure 3**).

Quail is a LD breeder, which activates its reproductive activity in LD and suppresses its reproductive activity in SD. It is understandable that the expression of GnIH is stimulated by a nocturnal hormone melatonin and SD when the duration of melatonin secretion is long. Accordingly, it was hypothesized that the increase of GnIH expression may inhibit reproductive activity in SD in quail (Ubuka et al., 2005). The opposite but similar thoughts can be applied to sheep. Sheep is a SD breeder that activates its reproductive activity in SD and suppresses its reproductive activity in LD. It is also understandable that the expression of RFRP is inhibited by a nocturnal hormone melatonin and SD when the duration of melatonin secretion is long. Accordingly, it was hypothesized that the increase of RFRP expression may inhibit reproductive activity in LD in sheep (Dardente et al., 2008). On the other hand, it was difficult to understand the inhibitory effect of melatonin or SD on RFRP expression in hamsters, because hamsters are LD breeders (Revel et al., 2008; Mason et al., 2010). However, recent report has shown that RFRP may have a stimulatory effect on gonadotropin secretion in SD in Siberian hamsters (Ubuka et al., 2012a). Long duration of melatonin secretion in SD may need to decrease RFRP expression to inhibit reproductive activities of Siberian hamsters in SD. Another recent work in male Syrian hamsters suggested that GnIH might be excitatory in LD as well (Ancel et al., 2012). Long duration of melatonin secretion in SD may decrease RFRP expression to inhibit reproductive activities and short duration of melatonin secretion in LD may increase RFRP expression to stimulate reproductive activities of Syrian hamsters in LD (Ancel et al., 2012). Further studies are required to understand the role of melatonin controlling GnIH and RFRP expression in seasonal breeders.

# **EFFECT OF STRESS**

"fendo-03-00148" — 2012/11/26 — 21:47 — page 8 — #8

Stress leads to reproductive dysfunction in many species, including rodents and humans. Calisi et al. (2008) hypothesized that stress effects upon reproduction are mediated via the hypothalamic GnIH system in birds. They examined the effects of capture-handling stress in the hypothalamus of male and female adult house sparrows. There were more GnIH-positive neurons in fall birds versus those sampled in spring, and a significant increase in GnIH-positive neurons was seen in stressed birds in spring. These data imply an influence of stress upon the GnIH system that changes over the annual cycle of reproduction (**Figure 3**).

Kirby et al. (2009)showed that both acute and chronic immobilization stress lead to an up-regulation of the expression of RFRP in the DMH of adult male rats and that this increase in RFRP is associated with inhibition of downstream HPG activity. They also showed that adrenalectomy blocks the stress-induced increase in RFRP expression. Immunohistochemistry revealed that 53% of RFRP cells express receptors for glucocorticoids (GCs), suggesting that adrenal GCs can mediate the stress effect through direct action on RFRP cells. These data show that stress-induced increases in adrenal GCs cause an increase in RFRP that contributes to hypothalamic suppression of reproductive function (**Figure 3**).

Papargiris et al. (2011) tested if GnIH (RFRP) mediates the inhibitory effect of stress on LH secretion in ovariectomized ewes using a psychosocial stressor, isolation/restraint. Isolation/restraint stress increased plasma cortisol concentrations and decreased plasma LH concentrations. However, there was no significant effect of stress on RFRP peptide or mRNA levels, with no difference between control or stressed ewes. Furthermore, there was no difference in the number of RFRP-ir cells doublelabeled for Fos between control and stressed ewes and there was

no difference in the cellular expression of RFRP mRNA between groups. Accordingly, GnIH (RFRP) may not mediate the effects of stress on LH secretion in ewes or the effect of stress may depend on the presence of gonadal sex steroids.

#### **EFFECT OF SEX STEROIDS**

Estrogen secreted by the ovary feedbacks to the brain and pituitary to regulate gonadotropin secretion (Herbison, 1998; Petersen et al., 2003). Wintermantel et al. (2006) found that estrogen positive feedback to generate the preovulatory gonadotropin surge was normal in estrogen receptor (ER) β knockout mice, but absent in ERα mutant mice. Because GnRH neurons do not express ERα, estrogen positive feedback upon GnRH neurons must be indirect. RFRP (mammalian GnIH) neuronal system may be involved in estrogen feedback signaling to GnRH neurons because RFRP neurons in rodents express ERα and respond with c-Fos expression to an acute administration of estradiol-17β (E2; Kriegsfeld et al., 2006; **Figure 3**).

The role of RFRP-3 in regulating ovulatory function was investigated in female hamsters (Gibson et al., 2008). The cellular activity of RFRP neurons was suppressed at the time of the LH surge, suggesting removal of negative feedback by RFRP-3 at this time. The SCN, the master circadian clock triggering ovulation in rodents, projects to a large proportion of RFRP neurons, providing a mechanism for timing removal of negative drive on the GnRH system. Activities of the SCN, GnRH, and RFRP neurons were coordinated with ovulation (Gibson et al., 2008). Li et al. (2012) investigated the expression patterns in the reproductive axis of the female pig across the estrous cycle. The hypothalamic levels of both RFRP and its receptor mRNA were lowest in estrus and peaked in the proestrus and diestrus phases. Smith et al. (2010) examined Kiss1 and RFRP mRNA throughout the menstrual cycle of a female primate, rhesus macaque. Kiss1-expressing cells werefound in the POA and ARC, and RFRP-expressing cells were located in the PVN/DMN. Kiss1 expression in the caudal ARC and POA was higher in the late follicular phase of the cycle (just before the GnRH/LH surge) than in the luteal phase. RFRP expression was also higher in the late follicular phase suggesting that RFRP fine tunes GnRH/LH surge in primates. There are also reports showing the correlation of RFRP expression and testicular activities of male mice (Sethi et al., 2010a,b).

Molnár et al. (2011) investigated the possibility that RFRP neurons are involved in estrogen feedback signaling to the reproductive axis in mice. They compared the expression of RFRP mRNA of ovariectomized mice, with and without E2 replacement. Subcutaneous administration of E2 via silastic capsules for 4 days significantly down-regulated RFRP mRNA expression. In ovariectomized mice, low levels of ERα immunoreactivity were detectable in 18.7 ± 3.8% of RFRP neurons, whereas RFRP neurons did not exhibit ERβ immunoreactivity. The estrogenic down-regulation of RFRP expression may contribute to estrogen positive feedback to the reproductive axis. However, whether E2 regulates RFRP neurons directly or indirectly remains an open question because ERα immunoreactivity is present only in a subset of RFRP cells (**Figure 3**).

Poling et al. (2012) examined changes in RFRP neurons in mice of both sexes during development and under different

"fendo-03-00148" — 2012/11/26 — 21:47 — page 9 — #9

adulthood hormonal milieus. They identified two interspersed subpopulations of RFRP cells (high RFRP-expressing, HE; low RFRP-expressing, LE), which have unique developmental and steroidal regulation characteristics. The number of LE cells robustly decreased during postnatal development, whereas HE cell number increased significantly before puberty. In adults, they found that E2 and testosterone moderately repress RFRP expression in both HE and LE cells, whereas the non-aromatizable androgen dihydrotestosterone has no effect. They determined that approximately 25% of RFRP neurons coexpress ERα in each sex, whereas RFRP cells do not express androgen receptor in either sex, regardless of hormonal milieu. They detected coexpression of GPR147 but no coexpression of GPR74 in GnRH neurons of both intact and gonadectomized males and females. RFRP-3 may thus exert its effects on reproduction either directly, via GPR147 in a subset of GnRH neurons, and/or indirectly, via upstream regulators of GnRH (**Figure 3**).

# **RECENT PROGRESS IN GnIH STUDIES ON FISH AND HUMANS**

We have described the progress of GnIH research investigating its function in the brain and pituitary of birds and mammals. Recently, there were important progresses in the study of GnIH peptides in fish and humans as we briefly summarize them below.

A cDNA encoding three GnIH orthologs, LPXRFamide peptides, was cloned from the goldfish brain by a combination of 3- /5- RACE (Sawada et al., 2002). Mass spectrometric analyses revealed that a tridecapeptide (SGTGLSATLPQRFamide) is expressed in the brain as an endogenous ligand. Immunoreactive cell bodies were restricted to the nucleus posterioris periventricularis and the nervus terminalis and immunoreactive fibers were distributed in several brain regions including the nucleus lateralis tuberis pars posterioris and pituitary (Sawada et al., 2002). Amano et al. (2006) analyzed the hypophysiotropic activity of the three goldfish LPXRFamide peptides (gfLPXRFa-1, -2, and -3) in sockeye salmon. gfLPXRFa-ir cell bodies were detected in the nucleus posterioris periventricularis of the hypothalamus and immunoreactive fibers were distributed in various brain regions and the pituitary in sockeye salmon. gfLPXRFamide peptides stimulated the release of FSH, LH, and GH from cultured pituitary cells (Amano et al., 2006). In contrast, Zhang et al. (2010) have identified the orthologous GnIH genes in zebrafish, stickleback, medaka, and Takifugu. The zebrafish GnIH precursor contained three putative LPXRFamide peptides. Intraperitoneal administration of the mature zebrafish LPXRFa-3 (zfLPXRFa-3) significantly reduced the basal serum LH level in goldfish (Zhang et al., 2010).

Moussavi et al. (2012) examined the effects of synthetic gfLPXRFamide peptides on pituitary LHβ and FSHβ subunit, and gfLPXRFamide peptide receptor (gfLPXRFa-R) mRNA levels and LH secretion in goldfish. Intraperitoneal injections of gfLPXRFa-3 increased pituitary LHβ and FSHβ mRNA levels at early to late gonadal recrudescence, but reduced serum LH and pituitary gfLPXRFa-R mRNA levels, respectively, at early to midrecrudescence and later stages of recrudescence. Static incubation with gfLPXRFa-3 elevated LH secretion from dispersed pituitary cell cultures from prespawning fish, but not at other recrudescent stages. gfLPXRFa-3 suppressed LHβ mRNA levels at early recrudescence and prespawning but elevated LHβ at mid-recrudescence, and consistently attenuated FSHβ mRNA in a dose-specific manner *in vitro*. These results indicate that the effect of gfLPXRFa-3 depends on maturational status and administration route in goldfish (Moussavi et al., 2012).

Recently, the mature peptide structures of human GnIH peptides (human RFRP-1 and RFRP-3) were identified by mass spectrometry (Ubuka et al., 2009c). Because the structure of human RFRP-3 was identical to the structure of sheep RFRP-3, physiological functions of human RFRP-3 were studied in the sheep. It was shown that sheep/human RFRP-3 inhibits gonadotropin synthesis and release *in vitro* (Sari et al., 2009) and gonadotropin

# **REFERENCES**


neuropeptide FF. *J. Biol. Chem.* 275, 39324–39331.


release *in vivo* (Clarke et al., 2008; Smith et al., 2012). In view of its potent inhibition of gonadotropin secretion, human RFRP-3 (human GnIH) has the potential of an alternative or adjunct therapeutic agent to inhibit endogenous levels of gonadotropins and steroid hormones in humans. Thus, RFRP has therapeutic potential in the treatment of hormone-dependent diseases, such as precocious puberty, endometriosis, uterine fibroids, benign prostatic hyperplasia, and prostatic and breast cancers.

# **ACKOWLEDGMENTS**

This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science and Culture, Japan (22132004 and 22227002 to Kazuyoshi Tsutsui).

photoperiod. *J. Neuroendocrinol.* 20, 1252–1259.


"fendo-03-00148" — 2012/11/26 — 21:47 — page 10 — #10

Hosoya, M., et al. (2000). New neuropeptides containing carboxyterminal RFamide and their receptor in mammals. *Nat. Cell Biol.* 2, 703–708.


D. (2009). Stress increases putative gonadotropin inhibitory hormone and decreases luteinizing hormone in male rats. *Proc. Natl. Acad. Sci. U.S.A.* 106, 11324–11329.


"fendo-03-00148" — 2012/11/26 — 21:47 — page 11 — #11


"fendo-03-00148" — 2012/11/26 — 21:47 — page 12 — #12

the quail brain. *Cell Tissue Res.* 312, 73–79.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 28 August 2012; paper pending published: 24 September 2012; accepted:*

*11 November 2012; published online: 28 November 2012.*

*Citation: Ubuka T, Son YL, Tobari Y and Tsutsui K (2012) Gonadotropininhibitory hormone action in the brain* *and pituitary. Front. Endocrin. 3:148. doi: 10.3389/fendo.2012.00148 This article was submitted to Frontiers in Neuroendocrine Science, a specialty of*

*Frontiers in Endocrinology.*

*Copyright © 2012 Ubuka, Son, Tobari and Tsutsui. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and* *reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

"fendo-03-00148" — 2012/11/26 — 21:47 — page 13 — #13

# RFRP neurons are critical gatekeepers for the photoperiodic control of reproduction

# *Valérie Simonneaux\* and Caroline Ancel*

Neurobiologie des Rythmes, Institut des Neurosciences Cellulaires et Intégratives, UPR CNRS 3212, Strasbourg, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Lance Kriegsfeld, University of California, USA Iain J. Clarke, Monash University, Australia

#### *\*Correspondence:*

Valérie Simonneaux, Neurobiologie des Rythmes, Institut des Neurosciences Cellulaires et Intégratives, UPR CNRS 3212, 5, rue Blaise Pascal, 67084 Strasbourg, France. e-mail: simonneaux@inci-cnrs. unistra.fr

Seasonally breeding mammals rely on the photoperiodic signal to restrict their fertility to a certain time of the year.The photoperiodic information is translated in the brain via the pineal hormone melatonin, and it is now well-established that it is the variation in the duration of the nocturnal peak of melatonin which synchronizes reproduction with the seasons. The Syrian hamster is a long day breeder, and sexual activity is therefore promoted by exposure to a long day photoperiod and inhibited by exposure to a short day photoperiod. Interestingly, in this species electrolytic lesion of the mediobasal hypothalamus abolishes the short day-induced gonadal regression.We have shown that the expression of a recently discovered neuronal population, namely RFamide-related peptide (rfrp) neurons, present in the mediobasal hypothalamus, is strongly down-regulated by melatonin in short day conditions, but not altered by circulating levels of sex steroids. The role of rfrp and its product RFRP-3 in the regulation of reproductive activity has been extensively studied in mammals, and our recent findings indicate that this peptide is a potent stimulator of the reproductive axis in the Syrian hamster. It induces a marked increase in GnRH neuron activity and gonadotropin secretion, and it is able to rescue reproductive activity in short day sexually inactive hamsters. Little is known about the localization of the RFRP-3 receptor, GPR147, in the rodent brain. Accumulating evidence suggests that RFRP-3 could be acting via two intermediates, the GnRH neurons in the preoptic area and the Kiss1 neurons in the arcuate nucleus, but future studies should aim at describing the localization of Gpr147 in the Syrian hamster brain. Altogether our data indicate that the rfrp neuronal population within the mediobasal hypothalamus might be a serious candidate in mediating the photoperiodic effects of melatonin on the regulation of the reproductive axis.

**Keywords: RFamide peptide, RFRP-3, GPR147, kisspeptin, seasonal reproduction, melatonin**

"fendo-03-00168" — 2012/12/14 — 19:08 — page 1 — #1

# **MAMMALS USE THE RHYTHMIC SECRETION OF THE PINEAL HORMONE MELATONIN TO SYNCHRONIZE REPRODUCTION WITH THE SEASONS**

A large number of species restrict their fertility to a particular time of the year to ensure that the birth of the offspring occurs during the most favorable season. To determine the time of the year and synchronize their reproductive activity accordingly, mammals use the highly reproducible annual variations in light duration (or photoperiod). Photoperiod is transduced by a photoneuroendocrine system composed of the retina, the suprachiasmatic nucleus (seat of the master circadian clock) and the pineal gland which releases the hormone melatonin exclusively at night, so that the duration of the secretion varies according to night length. Therefore, the photoperiodic variations in circulating levels of melatonin throughout the year provide the body with a robust and reproducible representation of the seasons (Simonneaux and Ribelayga, 2003). It has long been established that photoperiodic variations in the duration of the nocturnal peak of melatonin synchronize reproduction in seasonal species like sheep or hamsters (Hoffman and Reiter, 1965; Carter and Goldman, 1983; Pevet, 1988; Bartness et al., 1993; Malpaux et al., 2001).

Syrian hamsters are long day breeders, meaning that they are sexually active in long day (LD: 14 light/10 h dark) conditions. Upon exposure to short day (SD: 10 light/14 h dark) conditions, they undergo a dramatic inhibition of reproductive activity within 8–10 weeks manifested by a marked atrophy of the gonads and accessory organs resulting in low levels of circulating sex steroids. Surgical removal of the pineal gland before exposure to SD conditions prevents hamsters from undergoing sexual inactivation. Conversely, exogenous melatonin injections mimicking SD conditions in hamsters raised in LD conditions induce sexual inactivation. In contrast to small rodents, large mammals with a longer gestation time like sheep are SD-breeders; they are sexually active in SD conditions and become quiescent after transfer to LD conditions. Although the reproductive timing is opposite in hamsters and sheep, in both cases the circulating levels of melatonin synchronize reproduction with photoperiod. However, why the reproductive systems of LD- and SD-breeders respond in opposite ways to the same melatonin signal is not known.

#### **MELATONIN ACTS ON THE PARS TUBERALIS TO TRANSMIT PHOTOPERIODIC INFORMATION**

It is clear that melatonin does not act directly on GnRH neurons and responsiveness to GnRH does not change with photoperiod (Urbanski et al., 1991). Melatonin binding sites are found in a number of brain structures but with considerable species differences (Masson-Pevet et al., 1994). Besides, a high density of melatonin receptors has been identified in the pars tuberalis of the adenohypophysis in a large number of mammalian species (Masson-Pevet and Gauer, 1994). Notably, the pars tuberalis cells expressing melatonin receptors synthesize thyroid-stimulating hormone (TSH) in a photoperiod/melatonin-dependent manner, with a higher level of expression in LD conditions (Klosen et al., 2002; Dardente et al., 2003, 2010). TSH produced by the pars tuberalis has been recently recognized as a key messenger through which melatonin acts on the gonadotropic axis for the seasonal control of reproduction. TSH acts on a specialized glial cell type of the hypothalamic ependymal wall, the tanycytes, to induce a marked up-regulation of the thyroid hormone-activating enzyme deiodinase 2 (Dio2) which in turn increases local concentrations of the bioactive T3 thyroid hormone (Yoshimura et al., 2003; Hanon et al., 2008, 2010; Nakao et al., 2008). In quail (Yoshimura et al., 2003) and Siberian hamsters (Barrett et al., 2007) local T3 administration was reported to increase reproductive activity although through unknown mechanisms.

Altogether these observations point to the pars tuberalis as a key site for the integration of the endocrine melatoninergic message for the seasonal regulation of reproductive activity. However, the central reproductive site(s) actually controlled by the melatonin/TSH system is(are) still unknown. Various structures in the mediobasal hypothalamus have been proposed to be direct or indirect sites of action for melatonin, particularly in the sheep (Malpaux et al., 1998) and hamster (Maywood and Hastings, 1995).

# **RFRP-3 NEURONS LOCATED IN THE MEDIOBASAL HYPOTHALAMUS ARE STRONGLY REGULATED BY MELATONIN**

In the Syrian hamster, the dorsal part of the mediobasal hypothalamus appears as a key structure for the photoperiodic regulation of reproductive activity since it contains melatonin binding sites and its ablation by electrolytic lesion prevents the inhibitory effect of melatonin on reproductive activity (Maywood et al., 1996). Recently, we reported that neurons located in this hypothalamic area express the *RFamide-related peptide* (*rfrp)* gene in a photoperiodic-dependent manner in Siberian and Syrian hamsters (Revel et al., 2008).

The *rfrp* gene was discovered in 2000 in mammals (Hinuma et al., 2000), concurrently with the discovery of its avian ortholog *gonadotropin-inhibitory hormone* (*gnih*; Tsutsui et al., 2000). The *rfrp* and *gnih* genes were found to produce new peptides of the RFamide family of peptides, which share a common C-terminal LPXRFamide (X = L or Q) motif. In the quail, GnIH was shown to act directly at the level of the pituitary to inhibit gonadotropin release (Tsutsui et al.,2000). In mammals, the *rfrp* gene is expressed in neurons located in the mediobasal hypothalamus and encodes a precursor that produces two peptides, RFRP-1 and RFRP-3 (Ukena and Tsutsui, 2001; Kriegsfeld et al., 2006; Clarke et al., 2008; Dardente et al., 2008; Revel et al., 2008; Smith et al., 2008; Rizwan et al., 2009). The demonstration that GnIH is a potent inhibitor of gonadotropin release in birds spurred great interest in the roles of RFRP-1 and particularly RFRP-3 in the regulation of endocrine functions in mammals (Bentley et al., 2010; Kriegsfeld et al., 2010; Smith and Clarke, 2010; Tsutsui et al., 2010 for reviews).

In Syrian and Siberian hamsters, we observed that the level of *rfrp* mRNA is strongly down-regulated in sexually inactive SDadapted animals (Revel et al., 2008). This variation is solely photoperiodic as there are no daily changes in *rfrp* mRNA levels either in LD or SD conditions. In both species, the SD-induced decrease in *rfrp* gene expression is associated with a similar decrease in peptide immunoreactivity in perikarya and fibers (Revel et al., 2008; Mason et al., 2010; Ubuka et al., 2012; **Figure 1**). We have recently found a similar SD-induced inhibition of *rfrp* expression in other LD-breeders, notably the European hamster (**Figure 1**) and the jerboa (Janati et al., 2012). Strikingly, in sheep, a SD-breeder, two studies reported that hypothalamic *rfrp* mRNA levels and RFRP immunoreactivity are also reduced in SD conditions while animals are sexually active (Dardente et al., 2008; Smith et al., 2008). In contrast, in the non-photoperiodic rat *rfrp* mRNA levels are not modified by photoperiodic conditions (Revel et al., 2008).

In the Syrian hamster, we demonstrated that the SD downregulation of *rfrp* expression is not due to the lower levels of

by Dr. Greg Anderson; Syrian and Siberian hamster pictures were kindly provided by Julien Bartzen and European hamster pictures by Cristina Saenz de Miera.

"fendo-03-00168" — 2012/12/14 — 19:08 — page 2 — #2

circulating sex steroids since neither testosterone implants in sexually inactive SD hamsters, nor testis ablation in LD-adapted hamsters altered the levels of *rfrp* mRNA. This lack of major sex steroid feedback on *rfrp* expression is in agreement with other studies conducted in rats, mice, Siberian hamsters, and sheep (Smith et al., 2008; Quennell et al., 2010; Poling et al., 2012; Ubuka et al., 2012). Of note however, studies in female Syrian hamsters reported that RFRP neurons contain Er-α and respond to estrogen administration (Kriegsfeld et al., 2006; Gibson et al., 2008) and RFRP expression in ewe is reduced during the preovulatory period (Clarke et al., 2012).

Importantly, we found that pineal gland ablation before transferring hamsters to SD conditions, a protocol which prevents the SD-induced inhibition of reproductive activity, prevented the decrease in *rfrp* mRNA levels (Revel et al., 2008). Conversely, repeated melatonin injections during the late afternoon to LD-adapted hamsters, a protocol known to inhibit reproductive activity, also induced a marked decreased in *rfrp* mRNA levels (Revel et al., 2008). In the Siberian hamster as well, the down-regulation of *rfrp* expression in SD conditions is induced by melatonin (Ubuka et al., 2012). Remarkably, in the quail, melatonin also regulates GnIH expression but in an opposite manner compared to mammals. Melatonin binds to Mel1c receptors located on GnIH neurons to increase GnIH synthesis and release, and as a consequence, expression of this inhibitory peptide is increased under SD conditions (Ubuka et al., 2005).

Altogether these observations indicate that in a number of seasonal mammalian species, *rfrp* expression is decreased in SD conditions regardless of whether the species is a LD- or a SDbreeder. Experiments with melatonin manipulation carried out in Syrian and Siberian hamsters demonstrate that *rfrp* downregulation results from the larger production of melatonin in SD conditions. It is tempting to speculate that melatonin may act primarily on RFRP neurons in the mediobasal hypothalamus to control seasonal reproduction. Several observations, however, indicate that this central effect of melatonin is probably indirect. Melatonin binding sites are found in the area where RFRP neurons are located in the Syrian hamster, but this is not the case in the other seasonal species like the Siberian and European hamsters or the sheep. Furthermore, in the Syrian hamster, we found that at least 3 weeks of daily melatonin administration is required to induce a significant reduction in the level of *rfrp* mRNA (Revel et al., 2008) whereas the pineal hormone is much faster to control the expression of other photoperiodically regulated genes like *tsh* in the pars tuberalis or *deiodinase 2* in the tanycytes (Revel et al., 2006a; Yasuo et al., 2007; Dardente, 2012). Therefore, it appears likely that there is an intermediate between the endocrine melatoninergic message and the photoperiodic regulation of RFRP expression, and it might be interesting to investigate whether it is the melatonin-driven TSH/T3 signal.

# **RFRP-3 STIMULATES THE GONADOTROPIC AXIS AND RESCUES REPRODUCTIVE ACTIVITY IN PHOTO-INHIBITED HAMSTERS**

An increasing number of studies now indicate that RFRP-3 is implicated in the regulation of mammalian reproductive function (Bentley et al., 2010; Tsutsui et al., 2010 for reviews).

"fendo-03-00168" — 2012/12/14 — 19:08 — page 3 — #3

In mice RFRP-3 was found to exhibit rapid and repeatable inhibitory effects on the firing rate of a subpopulation of GnRH neurons (Ducret et al., 2009). In male rats, intracerebroventricular (icv) RFRP-3 significantly suppresses all facets of sex behavior and also significantly reduces plasma levels of luteinizing hormone (LH; Johnson et al., 2007; Pineda et al., 2010). In female rats, chronic icv infusion of RFRP-3 causes a dosedependent inhibition of GnRH neuronal activation at the LH surge peak and also suppresses neuronal activation in the anteroventral periventricular region, which provides stimulatory input to GnRH neurons (Anderson et al., 2009). In ovariectomized mature rats, intravenous administration of RFRP-3 significantly reduces plasma LH concentrations (Murakami et al., 2008). Finally, in the ovine and bovine species, RFRP-3 administration inhibits gonadotropin release (Clarke et al., 2008; Kadokawa et al., 2009; Sari et al., 2009) although this is still controversial (Caraty et al., 2012).

Until recently, and based on the plethora of publications supporting this hypothesis, it was assumed that RFRP-3 functioned in mammals as GnIH functioned in birds and served as an inhibitory component regulating the hypothalamic–pituitary– gonadal axis. However, this statement was somehow contradictory with our observation of an increased synthesis of RFRP in LDadapted sexually active hamsters. We have recently reported novel findings in the male Syrian hamster (Ancel et al., 2012) which have led to call this assumption into question, concurrently with another group working on the male Siberian hamster (Ubuka et al., 2012). In the male Syrian hamster, we reported that acute icv administration of RFRP-3 stimulates GnRH cell activity, gonadotropin release, and testosterone production under LD conditions (**Figure 2**; Ancel et al., 2012). In the same manner, in SD-adapted male Syrian hamsters a single central injection of RFRP-3 increases gonadotropin release [LH (ng/ml) vehicle 1.72 ± 0.31 vs. RFRP-3 4.36 ± 0.82; *n* = 6; *p* < 0.05]. Furthermore, under the same photo-inhibitory conditions, 5 weeks of continuous central administration of RFRP-3 to male Syrian hamsters produces a complete reactivation of the reproductive axis, manifested by increased testis weight and circulating levels of testosterone, similar to those observed in LD conditions (**Figure 3A**; Ancel et al., 2012). In the Siberian hamster, while administration of RFRP-3 in LD conditions inhibits gonadotropin release, the same protocol stimulates gonadotropin secretion in SD conditions (Ubuka et al., 2012). Remarkably, these findings of a stimulatory action of RFRP-3 on the male hamster reproductive axis are in sharp contrast with a previous study reporting an inhibitory effect of icv GnIH in ovariectomized female Syrian hamsters (Kriegsfeld et al., 2006). In LD conditions, reproductive activity of female rodents displays a well-described estrous cycle, characterized by varying levels of circulating gonadotropins and sex steroids. It has been hypothesized that RFRP-3 might be an inhibitory component of the negative feedback loop which regulates the estrous cycle, since RFRP cellular activity is decreased at the time of the LH surge in the Syrian hamster (Gibson et al., 2008). In this context it would be interesting to determine whether the effect of RFRP-3 on the female reproductive axis depends on the stage of the estrous cycle at which it is administered.

Taken together, these data indicate that there are certainly species- and gender-dependent differences in the involvement of RFRP-3 in the regulation of reproductive activity. As a consequence one might be cautious when calling the mammalian peptide "GnIH" based on its effect in birds, and in the light of the recent reports on work carried out in male hamsters the peptide should be termed "RFRP-3." From a seasonal point of view, when considering LD- (hamster) and SD- (sheep) breeders, it is remarkable to find that while the expression of the peptide in increased in LD conditions in both kinds of breeders, the effect of the peptide on the gonadotropic axis is opposite, as it is stimulatory in hamsters and inhibitory in sheep. This suggests a non-conserved role and/or site of action for RFRP-3 across seasonally breeding species. It is tempting to speculate that RFRP neurons may play a key role in discriminating between long and short day breeders because RFRP expression is down-regulated by a short day profile of melatonin in both kinds of seasonal breeders but the peptide appears to

"fendo-03-00168" — 2012/12/14 — 19:08 — page 4 — #4

have an opposite effect on the reproductive axis, being stimulatory in long day and inhibitory in short day breeders. Complementary experiments will have to be carried out in other species to test this hypothesis and fully understand the role of RFRP-3 in the seasonal control of reproduction.

Interestingly, the RFRP peptides were shown to have a modulatory action on feeding behavior (Bechtold and Luckman, 2007; Johnson et al., 2007; Clarke et al., 2012) thus it might be worth investigating whether the photoperiodic variation in RFRP expression might also impact on food intake and body weight regulation in seasonal species.

# **RFRP-3 MODES AND SITES OF ACTION**

The complex involvement of RFRP peptides in the regulation of the hypothalamic–pituitary–gonadal axis has raised a number of questions regarding the sites of action of the peptides. In various mammalian species including humans, RFRP fiber networks are found in multiple brain regions including the preoptic area, the arcuate nucleus, the lateral septum, the anterior hypothalamus, and the bed nucleus of the stria terminalis (Ukena and Tsutsui, 2001; Kriegsfeld et al., 2006; Johnson et al., 2007; Mason et al., 2010). Notably, RFRP-immunoreactive fibers make apparent contact with a subpopulation of GnRH neurons in rodents and sheep (Kriegsfeld et al., 2006; Smith et al., 2008; Poling et al., 2012; Rizwan et al., 2012; Ubuka et al., 2012) suggesting that RFRP-3 acts centrally to control the reproductive axis.

There is still uncertainty as to whether RFRP-3 also exerts a hypophysiotropic effect in mammals as reported in birds. A large body of evidence now reports the absence of fibers in the median eminence of rodents (Ukena and Tsutsui, 2001; Yano et al., 2003; Kriegsfeld et al.,2006; Rizwan et al.,2009; Smith et al.,2010; Ubuka et al., 2012). However, there are controversial data as to whether RFRP-3 acts (Kriegsfeld et al., 2006; Murakami et al., 2008; Pineda et al., 2010) or not (Anderson et al., 2009; Rizwan et al., 2009; Ubuka et al., 2012) on the rodent pituitary to regulate LH secretion. In the male hamster we reported no effect of the peptide on LH secretion when injected peripherally, nor on the basal and GnRH-stimulated production of LHfrom isolated pituitary glands (Ancel et al., 2012). In contrast in the sheep, RFRP fibers terminating in the median eminence have been identified and the peptide is released into the portal blood and appears to induce a marked inhibition of gonadotropin secretion (Clarke et al., 2008; Sari et al., 2009; Smith et al., 2012).

The RFRP peptides bind with high affinity to GPR147 (also known as NPFF1R) and with a lower affinity to GPR74 (also known as NPFF2R), which were first identified as neuropeptide FF receptors (Hinuma et al., 2000; Liu et al., 2001; Engstrom et al., 2003). The GPR147 receptor couples with Gαi3 and Gα*<sup>s</sup>* proteins (Gouarderes et al., 2007) suggesting that GPR147 can have both inhibitory and stimulatory downstream effects on cellular activity. However, in CHO cells, activation of the receptor inhibits forskolin-stimulated cAMP accumulation (Mollereau et al., 2002).

NPFF receptors have been detected in rodent, lagomorph, and monkey brains suggesting that they are phylogenetically conserved (Gouarderes et al., 2004a). Importantly, however, remarkable variations in GPR147 and GPR74 receptor contents exist from one

**"fendo-03-00168" — 2012/12/14 — 19:08 — page 5 — #5**

species to another and from one strain to another among the same species (Gouarderes et al., 2004a,b). Early studies describing the autoradiographic distribution of GPR147 in mice and rats indicated that the receptor was present throughout the hypothalamus (Gouarderes et al., 2002, 2004a,b). More recently, the use of *in situ* hybridization facilitated precise localization and made it possible to show that about 25% of GnRH neurons express *Gpr147*, but not *Gpr74*, in various rodent species (Poling et al., 2012; Rizwan et al., 2012; Ubuka et al., 2012). This is in agreement with the observation that RFRP-3 fibers are in contact with 20–40% of GnRH neurons (Rizwan et al., 2012; Ubuka et al., 2012). Furthermore, in mice expressing GnRH-green fluorescent protein-tagged neurons, RFRP-3 was found to exert a direct inhibitory effect on the firing rate of 41% of GnRH neurons, while 12% increased their firing rate, and the remainder were unaffected (Ducret et al., 2009). These observations support the hypothesis that RFRP-3 may exert its effects on reproduction directly via GnRH neurons. However, the peptide may also act indirectly, via upstream regulators of GnRH. This hypothesis is supported by data in rats indicating that RFRP-3 fibers are in contact with kisspeptin neurons, a subpopulation (20%) of which expresses the *Gpr147* gene (Rizwan et al., 2012).

In the Syrian hamster, RFRP-ir fibers project throughout much of the brain, including into the preoptic area and the arcuate nucleus (Kriegsfeld et al., 2006). We demonstrated that central injection of RFRP-3 to Syrian hamsters induces c-Fos expression in 30% of the GnRH neurons (**Figure 2A**) suggesting that the effects observed on the reproductive axis are mediated via these neurons (Ancel et al., 2012). Whether this effect is due to a direct action of RFRP-3 on GnRH neurons or whether it is linked to an effect on upstream regulators of the reproductive axis remains to be determined. Indeed, in the same study, although c-Fos expression was not observed in kisspeptin neurons following acute administration, the continuous central administration of RFRP-3 led to an increase in *Kiss1* expression in the arcuate nucleus together with an increase of testicular activity (**Figure 3**; Ancel et al., 2012). It is therefore possible that the RFRP-3 neuronal system regulates reproductive activity by acting at two levels of the reproductive axis: the GnRH and the kisspeptin neurons. In order to answer this question, it seems essential to carry out a detailed mapping of the *gpr147* in the Syrian hamster.

Altogether, a large amount of evidence now indicates that in various mammalian species RFRP-3 regulates reproductive activity by acting via its receptor located on GnRH neurons. This is supported by results showing that RFRP-3 fibers are in contact with a subpopulation of GnRH neurons and that *Gpr147* is expressed in GnRH neurons in rodents. However, another line of evidence points to*Kiss1* neurons as possible intermediates between RFRP peptides and the regulation of the reproductive function. Indeed, in rats RFRP-3 fibers are in contact with kisspeptin neurons which express *Gpr147* and in Syrian hamsters the reactivation of the reproductive function following continuous RFRP-3 administration goes alongside with an increase in *Kiss1* expression. Future studies using Kiss1R and GnRHR antagonists could help to understand the role of each one of these neuronal populations in mediating the effects of RFRP peptides on the reproductive axis.

# **RFRP-3 AND KISSPEPTIN ACT IN CONCERT TO SYNCHRONIZE RODENT REPRODUCTION WITH THE SEASONS**

There is strong evidence that RFRP neurons are regulated by photoperiod/melatonin to adapt reproductive activity to the seasons. A few years ago, the same supposition was made for kisspeptin, another member of the large RFamide family of peptides. In 2003, milestone studies reported that loss-of-function mutations of the kisspeptin receptor (KiSS1R/GPR54) in humans and rodents (de Roux et al., 2003; Seminara et al., 2003) prevented pubertal development and caused infertility, leading to a large number of studies aiming at investigating the role of kisspeptins in Vertebrate reproduction (Pinilla et al., 2012 for review). The *Kiss1* gene is mainly expressed in the arcuate nucleus and the anteroventro-periventricular nucleus of the hypothalamus, and kisspeptin neurons project specifically onto the GnRH cell bodies in the preoptic area and nerve terminals in the median eminence. In all mammalian species studied to date, kisspeptin appears as a powerful stimulator of the gonadotropic axis, acting primarily on GnRH neurons.

In the male Syrian hamster, we demonstrated that kisspeptin expression in the arcuate nucleus is down-regulated by melatonin in SD conditions, despite a negative feedback effect of testosterone on these neurons (Revel et al., 2006b). In the anteroventroperiventricular nucleus, kisspeptin expression is also decrease in SD conditions, but as a result of the absence of the positive feedback of testosterone consecutive to testicular regression (Ansel et al., 2010). Importantly, we demonstrated that chronic infusion of kisspeptin in SD-adapted sexually inactive male hamsters rescues reproductive activity to levels comparable to animals kept in photo-stimulatory LD conditions (Revel et al., 2006b; Ansel et al., 2011; Simonneaux et al., 2012). In the Siberian hamster as well, kisspeptin expression displays photoperiod variations and the peptide stimulates LH release (Greives et al., 2008a,b). Interestingly in the sheep, kisspeptin expression is also regulated by photoperiod, with a higher level of expression in SD conditions when animals are sexually active, and kisspeptin infusion in LD-adapted anestrous ewes induces ovulation in a majority of treated animals (Franceschini et al., 2006; Caraty et al., 2007; Smith et al., 2007, 2009). These observations indicate that kisspeptin expression, like RFRP, is regulated by photoperiod in seasonal species but, unlike RFRP, the direction of the regulation is different according to whether animals are LD- or SD-breeders.

To make things more complicated, other parameters also influenced by seasons regulate kisspeptin expression. This is particularly true for sex steroids which inhibit kisspeptin expression in the arcuate nucleus and increase it in the anteroventroperiventricular nucleus in various species including the Syrian (Revel et al., 2006b; Ansel et al., 2010) and the Siberian (Mason et al., 2007) hamsters. Additionally, metabolic factors that are also under the influence of seasonal changes were shown to impact on kisspeptin expression (Castellano et al., 2010). Therefore, although kisspeptin has been identified as an essential component of the photoperiodic regulation of reproductive activity in seasonal breeders (Revel et al., 2007; Simonneaux et al., 2009, 2012 for reviews) recent observations indicate that kisspeptin neurons are not the primary target of melatonin action but are controlled upstream by seasonally regulated intermediates.

Our current findings in the Syrian, Siberian, and European hamsters (Revel et al., 2008; **Figure 1**), in the jerboa (Janati et al., 2012), and other reports in the sheep (Dardente et al., 2008; Smith et al., 2008) suggest that RFRP-3 expression undergoes a conserved down-regulation by the SD melatonin signal irrespective of the reproductive response to seasons. In contrast, kisspeptin expression is increased when animals become sexually active, irrespective of the photoperiod. On the other hand, kisspeptin is always stimulatory of reproductive activity whereas RFRP-3 displays species-specific effects, being stimulatory in the Syrian hamster and inhibitory in the sheep.

These observations have led us to propose a working model for the seasonal control of reproduction in rodents (**Figure 4**). We propose that in LD conditions, RFRP-3 would activate GnRH neuronal activity directly and/or indirectly via the kisspeptinergic neurons. The former pathway is supported by the report of RFRP-3 fibers apposed to subpopulations of GnRH neurons (Kriegsfeld et al., 2006) whereas the latter

**Syrian hamster.** In long day conditions, RFRP-3 activates GnRH neuron activity directly in the preoptic area (POA) and/or indirectly via the kisspeptinergic neurons in the arcuate nucleus (ARC). In short day conditions, the large production of melatonin inhibits RFRP expression in the dorsomedial hypothalamus, which in turn decreases gonadotropic activity. Kiss1 expression in the ARC is reduced in short day conditions, but it is also inhibited by the testosterone feedback and may be regulated by metabolic factors.

"fendo-03-00168" — 2012/12/14 — 19:08 — page 6 — #6

is supported both by a report in rats indicating that RFRP-3 fibers are in contact with subpopulations of kisspeptin neurons (Rizwan et al., 2012) and by our observation that chronic central infusion of RFRP-3 increases arcuate *Kiss1* expression together with a reactivation of testicular function in the Syrian hamster (Ancel et al., 2012). In SD conditions, the melatoninergic signal would be primarily integrated by the hypothalamic RFRP neurons leading to a decreased expression of the peptide, but whether melatonin acts directly or not on RFRP neurons has yet to be determined. Current studies do not support a major feedback effect of sex steroids on RFRP neurons. In contrast, kisspeptinergic neurons integrate other factors related to the sexual and metabolic status of the animal in order to finely tune reproductive activity with the seasons. From studies in sheep, a SD-breeder, it appears that an increased production of RFRP-3 in LD conditions would inhibit reproductive activity by acting directly on the GnRH neurons and/or on the pituitary (Smith et al., 2008; Sari et al., 2009).

### **REFERENCES**


**CONCLUSION**

In regard to the seasonal regulation of reproductive activity, recent data have shed light on the involvement of two hypothalamic peptides of the RFamide family, RFRP-3 and kisspeptin. Clearly both peptides are regulated by the photoperiodic melatoninergic signal but may also be sensitive to other seasonally regulated factors. Our current hypothesis is that RFRP expression undergoes a conserved inhibition in short photoperiod but the peptide may be stimulatory or inhibitory according to the reproductive physiology of the species; on the other hand, kisspeptin has a conserved stimulatory action on the gonadotropic axis but its seasonal regulation shows species differences. Whether RFRP-3 and kisspeptin act on each other's expression or independently to regulate reproductive activity has yet to be clarified in different species. In the years to come, we believe that thorough comparative analyses on the effects and sites of action of both peptides between LD- and SD-breeders should help to resolve the yet unanswered question of why the same photoperiodic cue induces opposite behavioral responses.

A., Millar, R. P., et al. (2012). Gonadotropin-inhibitory hormone is a hypothalamic peptide that provides a molecular switch between reproduction and feeding. *Neuroendocrinology* 95, 305–316.


"fendo-03-00168" — 2012/12/14 — 19:08 — page 7 — #7

in the mouse. *Endocrinology* 150, 2799–2804.


peptide-3 suppresses luteinizing hormone (LH) secretion from the pituitary as well as pulsatile LH secretion in bovines. *Domest. Anim. Endocrinol.* 36, 219–224.


"fendo-03-00168" — 2012/12/14 — 19:08 — page 8 — #8

regulatory mechanisms. *Physiol. Rev.* 92, 1235–1316.


GPR54 gene as a regulator of puberty. *N. Engl. J. Med.* 349, 1614–1627.


"fendo-03-00168" — 2012/12/14 — 19:08 — page 9 — #9

Yoshimura, T., Yasuo, S., Watanabe, M., Iigo, M., Yamamura, T., Hirunagi, K., et al. (2003). Lightinduced hormone conversion of T4 to T3 regulates photoperiodic response of gonads in birds. *Nature* 426, 178–181.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 October 2012; paper pending published: 11 November 2012; accepted: 04 December 2012; published online: 18 December 2012.*

*Citation: Simonneaux V and Ancel C (2012) RFRP neurons are critical gatekeepers for the photoperiodic control of reproduction. Front. Endocrin. 3:168. doi: 10.3389/fendo.2012.00168*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Simonneaux and Ancel. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

REVIEW ARTICLE published: 08 March 2013 doi: 10.3389/fendo.2013.00024

# Functional significance of GnRH and kisspeptin, and their cognate receptors in teleost reproduction

#### **Renjitha Gopurappilly† , Satoshi Ogawa and Ishwar S. Parhar \***

Brain Research Institute, School of Medicine and Health Sciences, Monash University Sunway Campus, Selangor, Malaysia

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Vincent Prevot, INSERM, France Jae Young Seong, Korea University, South Korea

#### **\*Correspondence:**

Ishwar S. Parhar, Brain Research Institute, School of Medicine and Health Sciences, Monash University Sunway Campus, Petaling Jaya 46150, Selangor, Malaysia. e-mail: ishwar@monash.edu

#### **†Present address:**

Renjitha Gopurappilly, Manipal Institute of Regenerative Medicine, Manipal University, GKVK Post, Bellary Road, Allalasandra, Yelahanka, Bangalore 560 065, India.

Guanine nucleotide binding protein (G-protein)-coupled receptors (GPCRs) are eukaryotic transmembrane proteins found in all living organisms. Their versatility and roles in several physiological processes make them the single largest family of drug targets. Comparative genomic studies using various model organisms have provided useful information about target receptors. The similarity of the genetic makeup of teleosts to that of humans and other vertebrates aligns with the study of GPCRs. Gonadotropin-releasing hormone (GnRH) represents a critical step in the reproductive process through its cognate GnRH receptors (GnRHRs). Kisspeptin (Kiss1) and its cognate GPCR, GPR54 (=kisspeptin receptor, Kiss-R), have recently been identified as a critical signaling system in the control of reproduction. The Kiss1/Kiss-R system regulates GnRH release, which is vital to pubertal development and vertebrate reproduction.This review highlights the physiological role of kisspeptin-Kiss-R signaling in the reproductive neuroendocrine axis in teleosts through the modulation of GnRH release. Moreover, we also review the recent developments in GnRHR and Kiss-R with respect to their structural variants, signaling mechanisms, ligand interactions, and functional significance. Finally, we discuss the recent progress in identifying many teleost GnRH-GnRHR and kisspeptin-Kiss-R systems and consider their physiological significance in the control of reproduction.

**Keywords: GnRH receptor, kisspeptin receptor, GPR54, reproduction, teleost fish**

# **INTRODUCTION**

Guanine nucleotide binding protein (G-protein)-coupled receptors (GPCRs) play a pivotal role in various physiological processes (Marinissen and Gutkind, 2001), and GPCRs are now recognized as major pharmaceutical drug targets (Shoichet and Kobilka, 2012). GPCR sequencing in model organisms has proven vital for identifying novel GPCRs and their ligands with potential therapeutic value (Metpally and Sowdhamini, 2005). Evolutionary comparisons of GPCR sequences may allow for the identification of conserved motifs and the recognition of key functional residues (Attwood, 2001; Bjarnadottir et al., 2005). As research models, fish have been used largely to exploit their aquaculture potential. However, in recent years, there has been a trend toward using them in biomedical research as models of human disease. Because fish are phylogenetically diverse, they can be used to understand the fundamental principles of vertebrate evolution and disease processes. Therefore, the teleost genome provides an additional model to study the evolution and function of GPCRs (Metpally and Sowdhamini, 2005).

The hypothalamic-pituitary-gonadal (HPG) axis regulates puberty in vertebrates, primarily through the hypothalamic secretion of gonadotropin-releasing hormone (GnRH). This decapeptide hormone stimulates the release of the following gonadotropins from the anterior pituitary: follicle-stimulating hormone (FSH) for gamete growth and luteinizing hormone (LH) for gamete maturation and release (Ankley and Johnson, 2004; Weltzien et al., 2004). Recently, kisspeptin, a novel neuropeptide encoded by the metastasis suppressor gene, Kiss1/KISS1

(rodents/human), and its cognate GPCR, kisspeptin receptor (Kiss-R) (=GPR54), have been identified as potent regulators of reproduction, particularly for the onset of puberty (de Roux et al., 2003; Seminara et al., 2003). The kisspeptin-GPR54 system is vital to the onset of puberty, as it is critical to the activation of GnRH neurons. In 2004, we were the first to report the expression of GPR54 mRNA in GnRH neurons in a cichlid fish, tilapia, which strongly supports this concept (Parhar et al., 2004). Following this discovery, the presence of GPR54 in GnRH neurons has been demonstrated in several mammalian and non-mammalian species (Irwig et al., 2004; Han et al., 2005; Messager et al., 2005; Grone et al., 2010). Moreover, the innervation of kisspeptin fibers to GnRH neurons has been illustrated in several vertebrates (Clarkson and Herbison, 2006; Ramaswamy et al., 2008; Servili et al., 2011). The HPG axis is regulated by a number of GPCRs that play important roles in reproduction and sex hormone-dependent diseases (Heitman and Ijzerman, 2008). These receptors are therefore referred to as "reproductive" GPCRs. Numerous GPCRs with roles in reproduction have been discovered in recent years (Millar and Newton, 2010). In this review, we focus on the GnRH receptor (GnRHR), the recently deorphanized Kiss-R, and the regulation of GnRH secretion by an intricate interplay between the two in the teleosts.

#### **GnRH RECEPTOR TYPES STRUCTURAL VARIANTS**

Gonadotropin-releasing hormone is the hypothalamic decapeptide, which is mainly responsible for reproductive function. It is secreted from the hypothalamus into the hypophyseal portal circulation via the median eminence, which binds and activates the GnRHR that is expressed on the surface of the pituitary gonadotrope cells in mammals (Neill, 2002). In teleosts, the equivalent of the median eminence is incorporated into the rostral neurohypophysis of the pituitary (Batten and Ingleton, 1987). GnRH fibers have been observed to directly innervate the neurohypophysis in the pituitary (Kah et al., 1986; Parhar et al., 2002). In the pituitary, GnRH interacts with high-affinity GnRHRs in cell membranes of gonadotrophs, leading to biosynthesis and secretion of LH and FSH (Karges et al., 2003). To date, more than 20 forms of GnRH and the corresponding genes have been identified (Tsai, 2006; Chen and Fernald, 2008; Okubo and Nagahama, 2008; Roch et al., 2011; Tostivint, 2011; Kochman, 2012). These have a broad range of functions,including neuroendocrine, neurotransmitter/neuromodulatory, paracrine and autocrine functions (Millar et al., 2004), and each GnRH type is capable of serving all these roles. Structural and phylogenetic analysis of the GnRH genes classifies GnRH types into three forms (White and Fernald, 1998). Most vertebrate species possess at least two or three GnRH forms (Sherwood, 1987; Sherwood et al., 1993; Sealfon et al., 1997; Millar, 2003). GnRH-II (=GnRH2) is ubiquitously found in the most vertebrates, which was first isolated from the chicken (cGnRH-II) (Miyamoto et al., 1984). Because GnRH2 structure is highly conserved in the vertebrate species, it most likely serves critical functions and evolved the earliest (Millar et al., 2004). The hypothalamic form is designated GnRH type I (GnRH1) (Troskie et al., 1998), which is the species-specific form and regulates pituitary LH release (Parhar, 2002), and most of which have been identified in fish (Okubo and Nagahama, 2008). A third GnRH type (salmon GnRH-III = GnRH3) is only exhibited in the forebrain of teleost fish. A recent gene synteny analysis of the genomic regions comprising fish GnRH3 has found similar arrangement of GnRH3 gene cluster in the tetrapod genomes (Roch et al., 2011), which indicates that tetrapod GnRH3 was lost after divergence of fish and tetrapod lineages (Kim et al., 2011; Tostivint, 2011). Lamprey GnRH-I and -III, have recently been categorized as GnRH type IV (GnRH4), exclusive to the jawless fish (Roch et al., 2011).

GnRH receptor gene sequence was first identified from the mouse αT3 gonadotrope cell line (Tsutsumi et al., 1992; Millar et al., 2004). The ligand binding pocket of the GnRHR is formed by residues in the extracellular loops and the interior of transmembrane helices, indicating that GnRH partially enters the transmembrane core (Forfar and Lu, 2011). Mammalian GnRHR homologs have been isolated from non-mammalian species such as reptile, amphibians, birds, teleosts, and invertebrate species (Millar, 2003). The first teleost GnRHR was isolated from the African catfish (Tensen et al., 1997), and piscine GnRHR homologous sequences have been isolated from various teleosts such as the goldfish (Illing et al., 1999), Japanese eel (Okubo et al., 2000), trout (Madigou et al., 2000), striped bass (Alok et al., 2000), cichlid (Robison et al., 2001; Parhar et al., 2005a), medaka (Okubo et al., 2001), salmon (Jodo et al., 2003), fugu (Moncaut et al., 2005), and cobia (Mohamed et al., 2007). In some teleosts, there are four or five GnRHR isoforms (Jodo et al., 2003; Moncaut et al., 2005). Recent phylogenetic analyses based on amino acid sequence identity have classified teleosts GnRHR isoforms into three major

groups (Mollusk et al., 2011; Roch et al., 2011). Furthermore, a recent classification based on genome synteny analysis has also revealed three major lineages of fish GnRHR types that further subdivide into five classes: non-mammalian type I (GnRHRn1 and GnRHRn1b), non-mammalian type II (GnRHRn2), and nonmammalian type III (GnRHRn3 and GnRHRn3b) (Kim et al., 2011) (**Table 1**). The genome synteny-based classification has clearly demonstrated high conservation of neighboring genes of GnRHR types, which are also found in tetrapod GnRHR containing genome fragments (Kim et al., 2011). The neighboring gene sets of fish GnRHRn1, found in GnRHRn1b are probably generated by the teleost-specific third round genome duplication (Kim et al., 2011).

#### **LOCALIZATION AND FUNCTION OF GnRHR TYPES**

Localization of the various GnRHRs may provide some insight into their role in the physiology of reproduction. Five teleost GnRHR types (GnRHRn1, GnRHRn1b, GnRHRn2, GnRHRn3, and GnRHRn3b) have been localized in various reproductive organs including the gonads, brain, and pituitary (Lethimonier et al., 2004). In the European sea bass, five GnRHR genes show differential expression in various tissues, in which GnRHR1A (GnRHRn3b) and GnRHR (GnRHRn2) are widely distributed in reproductive and non-reproductive tissues including the eye, kidney, gills, gut, and liver, whereas GnRHRII (GnRHRn3), GnRHR2B (GnRHRn1), and GnRHR2C (GnRHRn1b) are more restricted to the central nervous system (Moncaut et al., 2005). This finding suggests multiple neuroendocrine and neuromodulatory roles of GnRH types throughout the body of teleosts (Jodo et al., 2003).

In the brain, the distribution of GnRHR types has been studied by RT-PCR, *in situ* hybridization and immunohistochemical approaches (**Table 1**). *In situ* hybridization studies in the European seabass have demonstrated the expression of GnRHR (GnRHRn2) in the forebrain, and the midbrain (Gonzalez-Martinez et al., 2004). In *Astatotilapia burtoni*, GnRHR1 (classified as GnRHRn1 based on sequence homology) is expressed in restricted brain regions including the telencephalon, preoptic area, ventral hypothalamus, and thalamus, whereas GnRH-R2 (GnRHRn3) is expressed in many more brain areas (Chen and Fernald, 2006). In the tilapia, GnRHR1 (GenBank accession number: XM\_003437572) and/or GnRHR3 (XM\_003437677) (both are classified as GnRHRn1) immunoreactive cells have been shown in the forebrain and midbrain (Soga et al., 2005). Most GnRHR types are found in brain areas involved in reproductive functions (Volkoff and Peter, 1999), and several GnRHR types are also found in the brain region that is known to be involved in appetite control, feeding, and stress responses (Chandroo et al., 2004; Volkoff et al., 2005).

*In situ* hybridization studies have demonstrated the expression offish GnRHR in the pituitary: GnRHRn1 [*A. burtoni* GnRHR1, striped seabass GnRHR (AF218841)], GnRHRn1b (European seabass GnRHR2C), GnRHRn2 (European seabass GnRHR, African catfish GnRHR1), GnRHRn3 (*A. burtoni* GnRH-R2, European seabass GnRHRII), and GnRHRn3b [European seabass GnRHR1A, Rainbow Trout GnRHR (NP\_001117823), African catfish GnRHR1 (X97497), Goldfish GnRHRA, and Goldfish

#### **Table 1 | GnRH receptors in teleosts.**


ND, not determined.

GnRHRB] (Illing et al., 1999; Alok et al., 2000; Madigou et al., 2000; Gonzalez-Martinez et al., 2004; Moncaut et al., 2005; Chen and Fernald, 2006) (**Figure 1A**). Within the pituitary cells, most GnRHR types have mainly been localized in the proximal pars

distal is of the pituitary where the gonadotrophs (LH and FSH) are exist, which includes GnRHRn1: Tilapia GnRHR1/R3, *A. burtoni* GnRHR1; GnRHRn2: European seabass GnRHR; GnRHRn3/3b: *A. burtoni* GnRH-R2, Rainbow trout GnRHR, Goldfish GnRHRA

and GnRHRB (Madigou et al., 2000; Parhar et al., 2002; Gonzalez-Martinez et al., 2004). In some teleosts, the presence of multiple GnRHR types have also been demonstrated in other pituitary cell types such as lactotropes, somatotropes, thyrotropes, melanotropes, corticotropes, and somatolactin cells (Illing et al., 1999; Parhar et al., 2002).

# **GnRHR SIGNALING, CYCLING, AND DESENSITIZATION**

A recent review by Levavi-Sivan and Avitan (2005) elegantly describes the GnRHR signaling pathway in gonadotrophs (**Figure 2A**). Mammalian and non-mammalian GnRHRs promote a conformational change in the GnRH peptide structure, which is essential for G-protein coupling and signal transduction

Rønnekleiv et al. (2010).

(Illing et al., 1999; Cheung and Hearn, 2002; Levavi-Sivan and Avitan, 2005), suggesting similar receptor activation mechanisms. Ligand binding to GnRHR activates phospholipase C to generate inositol (Levavi-Sivan and Avitan, 2005), which mobilizes intracellular Ca2<sup>+</sup> stores and activates protein kinase C. In teleosts, the cAMP signaling pathway is involved in GnRH release (Yaron et al., 2003; Levavi-Sivan and Avitan, 2005). In the tilapia, GnRH regulates glycoprotein hormone α and LHβ transcription through the PKC and PKA pathways and regulates FSHβ transcription through the PKC and PAPK-independent

the calmodulin (CaM). It has also been proposed that GnRH caused GTP

pathways (Gur et al., 2002). GnRHRs and GPCRs typically undergo desensitization and internalization including receptor phosphorylation, after being activated (McArdle et al., 2002). This phosphorylation stabilizes the association of GPCRs with β-arrestin, which prevents effector activation and acts as an adapter, targeting desensitized GPCRs for internalization (Ferguson, 2001) by endocytosis. Different GnRHR type may have different rates of desensitization and internalization and may also have different repertoire of signaling possibilities (McArdle et al., 2002).

# **REGULATION OF FISH GnRHR**

GnRH receptor is known to be regulated by several factors. In the tilapia, GnRHR3 has been shown to be up-regulated by its own ligand in the pituitary (Levavi-Sivan et al., 2004). A dopamineagonist suppressed tilapia GnRHR3 mRNA levels, which indicates an inhibitory effect of dopamine on GnRHR3 synthesis (Levavi-Sivan et al., 2004). Furthermore, tilapia GnRHR3 mRNA levels are higher in vitellogenic females than in maturing males, which could be due to the effect of higher estradiol-17β levels in females on tilapia GnRHR3 mRNA levels (Levavi-Sivan et al., 2004). In female *A. burtoni*, GnRH-R2 mRNA levels are elevated in mouthbrooding female brain compared to fed condition, which could be due to low androgens and estrogens levels (Grone et al., 2012). In rare minnow*Gobiocypris rarus*, bisphenol A, an endocrine disrupting chemical has been recently reported to affect gene expression of GnRHR1A gene in the brains of females (Qin et al., 2013). Although not many studies on promoter analysis of a fish GnRHR have been done, fish GnRHR gene seems to be sensitive to sex steroids as reported in mammals (Hapgood et al., 2005). In the Atlantic cod, pituitary GnRH-R2a but not GnRH-R1b gene expression increases in late-vitellogenic and running females. Continuous light inhibits the increase of pituitary GnRH-R2a expression seen during the normal spawning period (Hildahl et al., 2013). These results suggest that GnRHR is also influenced by nutrition and environmental factors most probably via steroid hormone levels.

# **GPR54**

Kisspeptin, encoded by the KISS1/Kiss1 gene, is an endogenous ligand for GPR54 (thus called Kiss-R) and promotes GnRH secretion (Kotani et al., 2001; Kaiser and Kuohung, 2005; Dungan et al., 2006). Kiss-R is widely expressed in many reproductive tissues. Mounting evidence suggests the critical role of kisspeptin in the modulation of GnRH secretion in the central nervous system (Irwig et al., 2004; Han et al., 2005; Messager et al., 2005). The kisspeptin-Kiss-R signaling system in various vertebrate species has been described in terms of distribution and physiology (Oakley et al., 2009). This signaling pathway is significant in the reproductive axis in vertebrates, although the location, developmental timing, and expression patterns differ. The anatomy and physiology of piscine kisspeptin-Kiss-R signaling has been extensively reviewed (Elizur, 2009; Zohar et al., 2010; Ogawa and Parhar, 2013). Current knowledge on the gene expression, distribution, and physiological function of piscine kisspeptin and Kiss-R is as follows.

# **KISSPEPTIN RECEPTOR TYPES IN TELEOSTS**

Kisspeptin receptor was first identified as an orphan GPCR (Lee et al., 1999). Mammalian Kiss-R is weakly homologous to galanin receptors (44–45%) but does not bind to either galanin or galaninlike peptides (Lee et al., 1999). We have previously reported a non-mammalian GPR54 for the first time in the tilapia (Parhar et al., 2004). Since then, many groups have identified Kiss-R in several non-mammalian vertebrates. In fish, there are two Kiss-R types (Kiss1Ra/GPR54-1 and Kiss1Rb/GPR54-2 in zebrafish, GPR54a and GPR54b in goldfish) (Lee et al., 2009; Li et al., 2009), while *Xenopus* have three Kiss-R types (GPR54-1a, GPR54-1b, and GPR54-2) (Lee et al., 2009). Genome and cDNA analyses has revealed that the Kiss-R genes contain five exons, although medaka Kiss-R1 and sea bass Kiss-R2 have six exons (Tena-Sempere et al., 2012). Our recent *in silico* study has identified four Kiss-R homologous sequences in an early sarcopterygian, Coelacanth *Latimeria chalumnae* genome, which leads to further clarification of molecular evolution of Kiss-R in vertebrates (describe in below).

Two Kiss-R types have been identified in the Senegalese sole (Mechaly et al., 2009), goldfish (Li et al., 2009), zebrafish (Biran et al., 2008), medaka (Lee et al., 2009), striped bass (Zmora et al., 2012), and sea bass (Tena-Sempere et al., 2012). In Perciform teleosts, the Southern Bluefin Tuna, and the Yellowtail Kingfish, two mRNA transcript variants of Kiss-R2 have been identified (Nocillado et al., 2012). With two different Kiss-R genes, two kisspeptin types (Kiss1 and Kiss2) have been identified in several teleosts such as zebrafish, medaka (Kanda et al., 2008; van Aerle et al., 2008; Kitahashi et al., 2009), goldfish (Li et al., 2009), sea bass (Felip et al., 2009), chub mackerel (Selvaraj et al., 2010), and striped bass (Zmora et al., 2012). Two kisspeptin types have also been identified in *Xenopus* and elephant shark (Lee et al., 2009), which suggests that the kisspeptin-Kiss-R systems function independently, especially in controlling teleost reproduction. Some teleosts possess only one kisspeptin type (Kiss2) or/and its receptor (Kiss-R2) (Tena-Sempere et al., 2012). Previous genome synteny analyses (Um et al., 2010; Tena-Sempere et al., 2012) have suggested that the Kiss-R genes previously identified from tilapia (Parhar et al., 2004), a cichlid *A. burtoni* (Grone et al., 2010), gray mullet (Nocillado et al., 2007), cobia (Mohamed et al., 2007), fathead minnow (Filby et al., 2008), Atlantic croaker (Mechaly et al., 2009), Senegalese sole (Mechaly et al., 2011), Atlantic halibut (Mechaly et al., 2010), and European eel (Pasquier et al., 2011) belong to the Kiss-R2 subfamily. Therefore, in teleosts, Kiss2-Kiss-R2 is evolutionarily highly conserved and may be functionally equivalent to mammalian Kiss1-Kiss-R.

The nomenclature for two Kiss-R types has been classified based on phylogenetic analysis (Um et al., 2010). Recent genome synteny-based classification studies have clearly demonstrated high conservation of neighboring genes of four fish Kiss-R types (Lee et al., 2009). The fish Kiss1Ra-containing genome fragments have a large array of common neighboring genes, which include ATP6V0B, RABL5, ODF2L, CYR61, C1QL3, GPT, FUZ, CCDC24, and B4GALT2, while the fish Kiss1Rb-containing genome fragments have conserved PSAT1, ZCCHC6, and ISCA1. Ligandreceptor binding assays have shown that while Kiss1 peptide (Kiss1–10) activates Kiss1Rb (GPR54-2) more efficiently than Kiss2 peptide (Kiss2-10) in zebrafish (Lee et al., 2009) and sea bass (Tena-Sempere et al., 2012), Kiss2–10 performs more efficiently in goldfish (Li et al., 2009). Kiss1Ra (GPR54-1) is activated by Kiss1–10 in goldfish (Li et al., 2009) and activated by Kiss1–10 and Kiss2–10 in zebrafish and sea bass (Lee et al., 2009; Tena-Sempere et al., 2012). Interestingly, human Kiss1–10 was an effective homolog in activating Kiss-R2 in the orange spotted grouper (which ordinarily has only Kiss2/Kiss-R2 pair) (Shi et al., 2010). Therefore, in the following sections, fish Kiss1Ra and Kiss1Rb are designated as Kiss-R2 and Kiss-R1, respectively.

# **DISTRIBUTION**

Two Kiss-R genes are highly expressed in various reproductive tissues including the brain, pituitary, and gonads and partially in other peripheral tissues (Nocillado et al., 2007; Biran et al., 2008; Filby et al., 2008; van Aerle et al., 2008; Li et al., 2009; Mechaly et al., 2009; Tena-Sempere et al., 2012) (**Table 2**). Kiss-R gene expression has been described in specific parts and throughout the teleost brain (Biran et al., 2008; Filby et al., 2008; Martinez-Chavez et al., 2008; van Aerle et al., 2008; Shahjahan et al., 2010). In the Senegalese sole, the two Kiss-R isoforms (Kiss1r\_v1 and Kiss1r\_v2) exhibit differential patterns in the brain (Mechaly et al., 2009). In the brain of fathead minnow, *Pimephales promelas*, *kiss1r* (homologous to Kiss-R2) gene expression is largely seen in various brain regions including the olfactory bulbs, the dorsal and ventral telencephalon, the hypothalamic nuclei, the midbrain, and the hindbrain (Filby et al., 2008). In the zebrafish, *in situ* hybridization has shown that most *kiss-r1* (GPR54-2) mRNA is expressed in the habenula (Ogawa et al., 2010, 2012a; Servili et al., 2011), whereas *kiss-r2* (GPR54-1) mRNA is widely expressed in the fish brain, especially in the olfactory bulb, telencephalon, preoptic area, midbrain, hypothalamic nuclei, cerebellum, and spinal cord (Grone et al., 2010; Servili et al., 2011; Ogawa et al., 2012a) (**Figure 1B**).

The distinct expression patterns of two Kiss-R types indicate their specific roles in reproductive and non-reproductive functions in fish. The habenula, a conserved structure in the brain of vertebrates has been shown to play important roles in various brain functions and behaviors, which include circadian rhythmicity, feeding, stress, sleep, affective states, and maternal and sexual behaviors (Teitelbaum and Epstein, 1962; Modianos et al., 1974; Corodimas et al., 1992; Matthews-Felton et al., 1995; Klemm, 2004; Zhao and Rusak, 2005). In mammals, the habenula expresses neuropeptides such as substance P (SP) and neuropeptide Y (NPY) (Neckers et al., 1979; Smith et al., 1985). In rats, GnRH-immunoreactive non-neuronal mast cells have been observed in the habenula (Khalil et al., 2003), which have also been noted in GFP-GnRH transgenic rats (Parhar et al., 2005b). Similarly in the habenula of teleosts, there are groups of cells immunoreactive to neuropeptides such as SP and corticotropinreleasing factor,(Villani et al., 1991; Mousa and Mousa, 2006), and expression of *tac2a*/*tac3a* gene (encoding neurokinin B) has been recently identified in the habenula of zebrafish (Biran et al., 2012; Ogawa et al., 2012b). These neuropeptide containing cells in the habenula may co-express Kiss-R to regulate a variety of neuroendocrine functions. As Kiss-R2 are widely expressed in the preoptic-hypothalamic regions in fish, Kiss-R2 could also be expressed in other neuronal populations in addition to GnRH neurons, which could play a key role in the control of reproduction, as has been suggested in mice (Herbison et al., 2010; Hanchate et al., 2012).

In some teleosts, Kiss-R is also expressed in the pituitary (Filby et al., 2008; Martinez-Chavez et al., 2008; Shahjahan et al.,


ND, not determined. SRE, binding response shown with serum response element; CRE, binding response shown with cAMP response element.

2010) (**Table 2**). In the pituitary of goldfish, Kiss-R2 (GPR54a) is expressed in the gonadotrophs, somatotrophs, and lactotrophs (Yang et al., 2010), which corresponds with the existence of fiber innervation of Kiss2-immunoreactive fibers in the pituitary of zebrafish (Servili et al., 2011). In mammals, kisspeptin appears to directly stimulate the secretion of LH, growth hormone, and prolactin secretion (Gutiérrez-Pascual et al., 2007; Kadokawa et al., 2008a,b). These results suggest that the hypothalamic kisspeptin-Kiss-R system regulates the reproductive functions at the level of the brain as well as at the level of the pituitary.

#### **EXPRESSION OF KISS-R IN GnRH NEURONS**

Our group (Parhar et al., 2004) was the first to report Kiss-R (Kiss-R2) and GnRH co-localization in tilapia using single-cell gene profiling coupled with laser-captured microdissection. This finding has established the concept that kisspeptin directly regulates GnRH neurons. We demonstrated expression of Kiss-R2 mRNA transcripts in all three GnRH neuronal types in tilapia (Parhar et al., 2004), which has also been confirmed in another cichlid by *in situ* hybridization (Grone et al., 2010). A close association between Kiss2 fibers and GnRH neurons have been established by a recent immunohistochemical study in zebrafish (Servili et al., 2011). In the mullet, Kiss-R2 gene expression positively correlates with GnRH2 and GnRH3 (Nocillado et al., 2007). In the early larval and juvenile cobia brain, Kiss-R (homologous to Kiss-R2) expression and all three GnRH mRNAs remains remarkably similar (Mohamed et al., 2007). These results suggest that Kiss-R2 and GnRH have coordinated roles at early pubertal stages in fish. In zebrafish, *kiss1*, *kiss2*, GnRH2, and GnRH3 mRNA levels are increased at the start of the pubertal phase (Kitahashi et al., 2009), indicating its role in controlling the onset of puberty. A very recent study in the *Morone* species has demonstrated that Kiss-R2 is co-localized in GnRH1 neurons in the preoptic area, while Kiss-R1 is expressed in cells attached to GnRH1 fibers, indicating two different GnRH1 regulatory methods (Zmora et al., 2012). These findings also indicate potential relationships between Kiss-Rs and multiple GnRHs, implicating kisspeptin-Kiss-R in the development and maturation of the piscine reproductive system.

#### **KISS-R SIGNALING IN TELEOSTS**

The ligand specificity of the piscine Kiss/Kiss-R system has previously been demonstrated by analyzing PKC-MAPK activation in several teleosts species (Biran et al., 2008; Lee et al., 2009; Tena-Sempere et al., 2012) (**Figure 2B**). All piscine Kiss-Rs tested successfully activated luciferase expression with the help of an SRE promoter, indicating the significance of PKC-MAPKs as a signaling pathway (similar to mammals) (Tena-Sempere et al., 2012). However, variances in kisspeptin length and receptor-ligand combinations have resulted in observable differentiation. Mammalian Kiss-Rs require a core sequence of kisspeptin (Kiss-10) to optimally activate Kiss-R (Kotani et al., 2001). All available fish Kiss2 sequences show a conserved Arg (position 13), indicating a putative mature Kiss2–12 peptide, while Kiss1 sequencing revealed a mature Kiss1–15 peptide due to the presence of a conserved N-terminal Gln (Tena-Sempere et al., 2012), which has further been supported by more active kisspeptin results

from pyroglutamylation of the N-terminal Gln (Lee et al., 2009). Only zebrafish and sea bass have served in the testing of longer kisspeptins (Kiss1–15 and Kiss2–12) (Lee et al., 2009; Zmora et al., 2012). In all cases, the longer peptides have activated Kiss-Rs more effectively than their respective shorter peptide form with some exceptions (Lee et al., 2009; Tena-Sempere et al., 2012).

The cAMP/PKA pathway is activated more potently by Kiss-R1 than Kiss-R2 in zebrafish, goldfish, and sea bass (Tena-Sempere et al.,2012).Maximum activation of Kiss-R1 in goldfish is achieved with Kiss2–10 (Li et al., 2009), while it is most potently activated by Kiss1–15 in sea bass (Tena-Sempere et al., 2012). The singleligand test using Kiss1–10 in zebrafish found that the cAMP/PKA pathway could be activated through Kiss-R1 (but not Kiss-R2) activation (Biran et al., 2008). In the orange spotted grouper,where the Kiss-R2 was solely studied, there was no observable PKA signaling pathway (Shi et al., 2010),which is consistent with previous piscine observations that Kiss-R2s have compromised signaling through this pathway. In some species,GPR54 duplicates have exhibited differentiation in signaling. However, further analysis of fish species is required to specify defined signaling patterns (Tena-Sempere et al., 2012).

#### **INDEPENDENT FUNCTION OF TWO KISSPEPTIN-KISS-R SYSTEMS**

In several teleost species, the onset of puberty marks a significant increase in Kiss-R mRNA expression (Biran et al., 2008; Filby et al., 2008; Martinez-Chavez et al., 2008). GPR54 is only induced in the mullet brain during the pubertal stages, leading to speculation that it also plays a role in reproductive development (Nocillado et al., 2007). In the brain of the fathead minnow, the highest expression of Kiss-R1 mRNA coincides with higher levels of GnRH gene expression in sexually mature females compared to prepubertal females (Filby et al., 2008). Although kisspeptin is considered to be a potent regulator of reproductive function, the specific function of two kisspeptin-Kiss-R systems has not been well understood due to poor ligand selectivity of two Kiss-R types for Kiss1 and Kiss2 peptides (Um et al., 2010). Physiological studies in some teleosts have suggested that Kiss2 exhibits higher potency in reproductive regulation compared to Kiss1 (Felip et al., 2009; Kitahashi et al., 2009; Li et al., 2009; Akazome et al., 2010; Shahjahan et al., 2010). Conversely, in the medaka, Kiss1 but not Kiss2 neurons in the hypothalamus exhibit estrogen sensitivity (Mitani et al., 2010), which suggests that Kiss1 but not Kiss2 plays a role in central reproductive regulation in the medaka (Oka, 2009). In the *Morone* species, Kiss1 is more potent in inducing LH release, with Kiss2 downregulating GnRH1 and Kiss-R2 gene expression during recrudescence (Zmora et al.,2012). These highly diverse observations suggest that the kisspeptin-Kiss-R pathway plays important roles in piscine reproduction; however, given the diversity of reproduction strategies, environmental niches and the timing of sexual maturation, and inconsistency in experimental approaches, it is difficult to establish a unifying theme (Oakley et al., 2009).

A recent study in zebrafish has managed to demonstrate an independent function of two kisspeptin-Kiss-R systems (Ogawa et al., 2012a). In zebrafish, Kiss1 neurons are solely present in the habenula, whereas Kiss2 neurons are only expressed in the preoptic-hypothalamic nuclei (Kitahashi et al., 2009; Servili et al., 2011). Similarly in zebrafish, Kiss-R1 mRNA expression occurs mainly in the habenula, whereas Kiss-R2 mRNA expression occurs widely around the brain (Kitahashi et al., 2009; Servili et al., 2011). This anatomical correspondence between kisspeptin and Kiss-R types clearly indicates that their ligand-receptor pairs functionally represent the key-keyhole interaction in zebrafish. Therefore, zebrafish is an attractive model system to study independent functions of two kisspeptin-Kiss-R systems in vertebrates. We have recently revealed that the Kiss1 neurons in the habenula that project into the interpeduncular nucleus (IPN) could modulate the serotonergic system through an autocrine mechanism (Ogawa et al., 2012a). However, the precise mechanism that underlies this finding and its role in serotonin regulation remains unknown.

#### **EVOLUTIONARY SIGNIFICANCE OF THE KISSPEPTIN-KISS-R SYSTEM**

The fishes provide excellent animal models to study the principles that underlie the vertebrate kisspeptin and Kiss-R systems from an evolutionary viewpoint (Akazome et al., 2010). The conserved genomic organization and gene synteny of the *kiss* and *kiss-r* genes indicates that they originated from a common ancestral gene (Akazome et al., 2010; Um et al., 2010). The vertebrate lineage exhibits this conserved genome organization, spanning mammalian, and non-mammalian species (Tena-Sempere et al., 2012). Our genome synteny analysis found four predicted Kiss-R homologous types in the Coelacanth genome (Kiss-R1a, Kiss-R1b, Kiss-R2a, and Kiss-R2b) (**Table 2**). Genes surrounding Coelacanth Kiss-R1a gene are also found in the genomic region of *Xenopus* GPR54-1a and mammalian Kiss-R genes but absent in fish Kiss-R types. Coelacanth Kiss-R1b is close to fish Kiss1Rb and *Xenopus* GPR54-1b. Neighboring genes of Coelacanth Kiss-R2a were found in a region surrounding *Xenopus* GPR54-2 and fish Kiss1Ra but were absent in the human Kiss-R. Coelacanth Kiss-R2b is close to Platypus GPR54b and Green anole (*Anolis carolinensis*) GPR54-like sequence. Based on the phylogenetic analysis, it can be hypothesized that the two ancestral Kiss-R lineages (Kiss-R1 and Kiss-R2) were first duplicated into four Kiss-R lineages during chromosome duplication in ancestral fish (**Figure 3**). Among the four lineages, Kiss-R1a and Kiss-R2b lineages have been lost in bony fish, Kiss-R2b lineage could have been lost in amphibian but may exist in reptile and platypus, and only Kiss-R1a lineage is preserved in human. However, in the Coelacanth genome, all four Kiss-R lineages are still conserved. These results correspond with the concept of molecular evolutional history of vertebrate Kiss-R genes that has been previously proposed (Lee et al., 2009; Zohar et al., 2010).

The alternating actions and importance of both Kiss1 and Kiss2 were very recently demonstrated in the *Morone* species (Zmora et al., 2012). They concluded that the organization of the kisspeptin system suggests a transitional evolutionary state between early to late evolving vertebrates. The evolutionary transition between multiple forms of kisspeptin, present in evolutionarily older vertebrates such as frogs and some fish, to a single form, as evident in higher vertebrates, is exemplified in the kisspeptin systems of the various fish species studied thus far (Um et al., 2010; Zmora et al., 2012). In the Coelacanth genome, only one kisspeptin-homologous sequence (ENSLACP00000010201,

FNFNPFGLRF) was identified, which is close to fish Kiss2. Genes surrounding the Coelacanth Kiss1 (LDHB, GYS2, SLC25A3, STRAP, and GOLT1B) were found in the Stickleback Kiss2, *Xenopus* Kiss2, and Zebrafish Kiss2, but not in the human Kiss1. Therefore, in the Coelacanth, Kiss1 lineage could have been lost, although genes rounding the human Kiss1 such as REN, ETNK2, and SNRPE are still conserved in the Coelacanth genome. The complete disappearance of Kiss1 and its functional relocation in different fish species clearly shows its evolutionary transition (Zmora et al., 2012). However, the physiological significance of two or loss of one Kiss system in fish species still remains unknown. The evolution of kisspeptin-Kiss-R system may be closely related to the evolution of reproductive traits. It is hence crucial to examine kisspeptin-Kiss-R system in fish species with one-Kiss and those with two-Kiss systems from the viewpoint of diversity of reproductive physiology, i.e., single- or multiple-spawner, seasonal breeder, social or non-social species, viviparous or non-viviparous fish, lifespan, and sex changing fish.

#### **CONCLUDING REMARKS**

The hypothalamic GnRH system has been well studied, and the specific functional roles of the various receptor-ligand pairs have been delineated for both mammalian and non-mammalian vertebrates. However, research on the kisspeptin/GnRH relationship in non-mammalian vertebrates, including fish, is still in its infancy. Studies have so far shown in fish that kisspeptins directly regulate GnRH neurons and GnRH release via interactions with Kiss-Rs. However, there are numerous unknown or unconfirmed (confirmed in mammals but not in fish) matters regarding the GPCRs in fish reproduction. Understanding the precise mechanism of endocrine regulation of fish reproduction based on GPCRs is necessary to determine the precise physiological roles of kisspeptin-GnRH pathways.

#### **REFERENCES**


#### **ACKNOWLEDGMENTS**

We thank Dr. Takashi Kitahashi for his contractive comments. This work was supported by grants from Malaysian Ministry of Higher Education, FRGS/2/2010/ST/MUSM/03/2 (to Satoshi Ogawa and Ishwar S. Parhar), Malaysian Ministry of Science, Technology, and Innovation, 02-02-10-SF0044 (to Ishwar S. Parhar and Satoshi Ogawa), and Monash University Sunway Campus, M-2-2-06 and M-2-07 (to Satoshi Ogawa), MM-2-5-06 and MM-7-07 (to Ishwar S. Parhar) and Neuroscience Research Strength grant (to Ishwar S. Parhar).


in an African cichlid fish. *Horm. Behav.* 62, 18–26.


steroid-sensitive sexually dimorphic kisspeptin neurons in medaka (*Oryzias latipes*). *Endocrinology* 149, 2467–2476.


isoforms of the kisspeptin-1 receptor (kiss1r) generated by alternative splicing in a modern teleost, the Senegalese sole (*Solea senegalensis*). *Biol. Reprod.* 80, 60–69.


fish cobia. *J. Mol. Endocrinol.* 38, 235–244.


of puberty. *N. Engl. J. Med.* 349, 1614–1627.


two endogenous ligands on a single cognate gonadoliberin receptor. *Eur. J. Biochem.* 243, 134–140.


sites of three gonadotropin-releasing hormone genes in one species. *Gen. Comp. Endocrinol.* 112, 17–25.


Neuroendocrinology of reproduction in teleost fish. *Gen. Comp. Endocrinol.* 165, 438–455.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 October 2012; paper pending published: 24 October 2012; accepted: 22 February 2013; published online: 08 March 2013.*

*Citation: Gopurappilly R, Ogawa S and Parhar IS (2013) Functional significance of GnRH and kisspeptin, and their cognate receptors in teleost reproduction. Front. Endocrinol. 4:24. doi: 10.3389/fendo.2013.00024*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Gopurappilly, Ogawa and Parhar. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# **Catherine Dodé<sup>1</sup>\* and Philippe Rondard<sup>2</sup>**

1 INSERM U1016, Institut Cochin, Université Paris-Descartes, Paris, France

<sup>2</sup> CNRS UMR5203, INSERM U661, Institut de Génomique Fonctionnelle, Université Montpellier 1, 2, Montpellier, France

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Ishwar S. Parhar, Monash University, Malaysia Nicolas De Roux, INSERM U676, France

#### **\*Correspondence:**

Catherine Dodé, INSERM U1016, Institut Cochin, Département de génétique et développement, 24 rue du Faubourg Saint-Jacques, 75014 Paris, France. e-mail: catherine.dode@inserm.fr

Kallmann syndrome (KS) is a developmental disease that associates hypogonadism and a deficiency of the sense of smell. The reproductive phenotype of KS results from the primary interruption of the olfactory, vomeronasal, and terminal nerve fibers in the frontonasal region, which in turn disrupts the embryonic migration of neuroendocrine gonadotropinreleasing hormone (GnRH) synthesizing cells from the nose to the brain. This is a highly heterogeneous genetic disease, and mutations in any of the nine genes identified so far have been found in approximately 30% of the KS patients. PROKR2 and PROK2, which encode the G protein-coupled prokineticin receptor-2 and its ligand prokineticin-2, respectively, are two of these genes. Homozygous knockout mice for the orthologous genes exhibit a phenotype reminiscent of the KS features, but biallelic mutations in PROKR2 or PROK2 (autosomal recessive mode of disease transmission) have been found only in a minority of the patients, whereas most patients carrying mutations in these genes are heterozygotes. The mutations, mainly missense mutations, have deleterious effects on PROKR2 signaling in transfected cells, ranging from defective cell surface-targeting of the receptor to defective coupling to G proteins or impaired receptor-ligand interaction, but the same mutations have also been found in apparently unaffected individuals, which suggests a digenic/oligogenic mode of inheritance of the disease in heterozygous patients. This non-Mendelian mode of inheritance has so far been confirmed only in a few patients. However, it may account for the unusually high proportion of KS sporadic cases compared to familial cases.

**Keywords: PROK2, PROKR2, Kallmann syndrome, hypogonadotropic hypogonadism, anosmia, digenic/oligogenic mode of inheritance**

#### **INTRODUCTION**

Kallmann syndrome (KS) is a developmental disease that associates hypogonadotropic hypogonadism, due to gonadotropinreleasing hormone (GnRH) deficiency, and anosmia, related to the absence or hypoplasia of the olfactory bulbs (Kallmann, 1944; deMorsier, 1954; Naftolin et al., 1971). The degree of the hypogonadism and that of the smell deficiency can vary significantly, not only between unrelated KS patients, but also between patients from the same family. The prevalence of KS has been estimated at one out of 8000 in boys. In girls, the prevalence is thought to be five times lower, but it is probably underestimated because some affected females only have mild hypogonadism, and also because primary amenorrhea in females often remains unexplored (Jones and Kemmann, 1976).

Pathohistological studies of fetuses with olfactory bulb agenesis have shown that the reproductive phenotype of KS results from a pathological sequence in embryonic life,whereby premature interruption of the olfactory, vomeronasal, and terminal nerve fibers in the frontonasal region disrupts the migration of neuroendocrine GnRH cells, which normally migrate from the nose to the brain along these nerve fibers (Schwanzel-Fukuda and Pfaff, 1989; Teixeira et al., 2010). What causes the primary failure of these fibers to establish proper contact with the forebrain is, however, still unknown.

Kallmann syndrome is genetically heterogeneous and involves various modes of transmission, specifically, X-chromosome linked, autosomal recessive, autosomal dominant with incomplete penetrance, and also digenic/oligogenic inheritance (Dodé and Hardelin, 2009; Sykiotis et al., 2010). Because the common infertility in affected individuals and, most importantly, the incomplete penetrance of the disease impede genetic linkage analysis, researchers have used various strategies to identify genes involved in KS, including mutation screening in genes that are disrupted by deletion or translocation breakpoints in chromosomal rearrangements associated with the disease phenotype, and candidate gene approaches. Nine causal genes have been reported to date, namely, by chronological order of discovery, *KAL1* (Franco et al., 1991; Legouis et al., 1991; Hardelin et al., 1992), *FGFR1* (Dodé et al., 2003), *PROKR2* and *PROK2* (Dodé et al., 2006), *FGF8* (Falardeau et al., 2008), *CHD7* (Kim et al., 2008; Jongmans et al., 2009), *WDR11* (Kim et al., 2010), *HS6ST1* (Tornberg et al., 2011), and *SEMA3A* (Hanchate et al., 2012; Young et al., 2012) (**Table 1**). Various loss-of-function mutations in *KAL1*, encoding the extracellular matrix glycoprotein anosmin-1, and in *FGFR1* or *FGF8*, encoding fibroblast growth factor receptor-1 and fibroblast growth factor-8, underlie the X-chromosome linked form and an autosomal dominant form of KS, respectively. Mutations in *KAL1* and *FGFR1/FGF8* account for roughly 8 and 10% of


#### **Table 1 | KS genes**.

?, not known.

all KS cases, respectively. The *KAL1* gene product, anosmin-1, binds to heparan-sulfate glycosaminoglycans, and may act as a coreceptor for FGF signaling through FGFR1, which also requires interaction with heparan-sulfate glycosaminoglycans for receptor activation. Mutations in the genes encoding heparan-sulfate 6-*O*-sulfotransferase 1, an enzyme involved in glycosaminoglycan modifications, WDR11, an intracellular protein that interacts with the transcriptionfactor EMX1, and semaphorin 3A, a secreted protein involved in axonal pathfinding, have also been found in some KS patients. Mutations in the chromodomain helicase DNAbinding protein 7 gene (*CHD7*) are present in approximately 70% of the patients affected by the CHARGE syndrome, which in most patients includes KS (Pinto et al., 2005), and mutations in this gene have been found in some patients who initially presented with KS. Finally, mutations in *PROKR2* and *PROK2*, encoding prokineticin receptor-2 and prokineticin-2, respectively, have been identified in approximately 9% of the KS patients, both in heterozygous and in homozygous or compound heterozygous states.

#### **PROKINETICINS AND THEIR RECEPTORS**

Prokineticins are secreted cysteine-rich proteins that possess diverse biological activities. The first identified member of this family was isolated from the venom of the black mamba snake (Joubert and Strydom, 1980), and was named mamba intestinal toxin 1 (MIT1) owing to its ability to induce intestinal contraction (Schweitz et al., 1999). Then, a small protein of similar size (77 amino acid residues, 8 kDa), with 58% sequence identity with MIT1, was isolated from skin secretions in the amphibian *Bombina variegata,* and called Bv8 (Mollay et al., 1999). Soon after, two mammalian proteins of this family were identified and named prokineticin-1 and -2 (PROK1 and PROK2) (Li et al., 2001; Kaser et al., 2003). PROK1 has 80% sequence identity with MIT1 and 58% identity with PROK2 (Li et al., 2001). The amino-terminal domain of prokineticins contains a sequence of six amino acid residues (AVITGA), which is conserved

in all mammalian and non-mammalian orthologs. Substitutions, deletions, or insertions to this hexapeptide result in the loss of agonist activity on prokineticin receptors (Kaser et al., 2003; Bullock et al., 2004). The carboxy-terminal region of prokineticins contains 10 cysteine residues forming five disulfide bonds. Apart from its potent effect on gastrointestinal smooth muscle contraction, PROK1 was also characterized as an angiogenic factor with specific effects on steroidogenic glands, thus earning its initial name EG-VEGF (endocrine gland vascular endothelial growth factor) (LeCouter et al., 2003). Mouse and human orthologs of Bv8, also known as PROK2, have been involved in a variety of biological activities, including effects on neuronal survival, gastrointestinal smooth muscle contraction (Li et al., 2001), circadian locomotor rhythm (Cheng et al., 2002), survival and migration of adrenal cortical capillary endothelial cells (LeCouter et al., 2003). PROK2 also has a role in appetite regulation and its anorectic effect is mediated partly by the melanocortin system (Gardiner et al., 2006).

Prokineticins can bind to two different G protein-coupled receptors, prokineticin receptor-1 and -2 (PROKR1 and PROKR2), which have about 85% sequence identity. These receptors were characterized simultaneously by three different groups (Lin et al., 2002; Masuda et al., 2002; Soga et al., 2002). Both are able to bind to PROK1 and PROK2 with similar nanomolar range affinities. They have a central core formed by seven transmembrane α-helical segments (TM1–TM7) connected by intracellular (i1–i3) and extracellular (e1–e3) loops, an extracellular aminoterminal end, and an intracellular carboxy-terminal end. These plasma membrane receptors operate as molecular switches to relay extracellular ligand-activation to intracellular heterotrimeric G proteins. PROKR1 is mainly expressed in peripheral tissues, including endocrine glands and organs of the reproductive system, the gastrointestinal tract, lungs, and the circulatory system (Soga et al., 2002; Battersby et al., 2004), whereas PROKR2 shows relatively localized distribution in the central nervous system (Cheng et al., 2002; Lin et al., 2002).

The olfactory bulb is one of the few areas in the mammalian brain that produce neurons throughout life. New interneurons originating from progenitors in the subventricular zone (SVZ) are continually added to the olfactory bulbs. The mRNAs of both receptors (PROKR1 and PROKR2) are expressed in the SVZ and the olfactory bulbs. The expression of PROKR1 have been detected in these areas although less abundantly than PROKR2. Whereas PROK1 mRNAs were not detected in any of these brain region (Ng et al., 2005). It was first shown that PROK2 functions as a chemoattractant for these neuronal progenitors, which follow a rostral migratory stream. Accordingly, PROK2 deficiency in mice leads to a loss of normal olfactory bulb architecture, and accumulation of neuronal progenitors in the rostral migratory stream (Ng et al., 2005). Soon after,it was found that *Prokr2*−*/*<sup>−</sup> knockout mice exhibit early hypoplasia of the olfactory bulbs and severe atrophy of the reproductive organs in both sexes, a phenotype reminiscent of the KS features. In addition, immunohistochemical analysis of these mice revealed that the neuroendocrine GnRH cells were absent from the hypothalamus (Matsumoto et al., 2006).

*PROK2* is a clock-controlled gene: the level of its messenger RNA shows a circadian oscillation profile in the suprachiasmatic nuclei (Cheng et al., 2002; Li et al., 2006). It has been postulated that PROK2 signaling through PROKR2 is a suprachiasmatic nuclei clock output signal that regulates circadian rhythms (Prosser et al., 2007; Li et al., 2012). PROK2-null mice show accelerated acquisition of food anticipatory activity during a daytime food restriction (Li et al., 2006), exhibit reduced total sleep time predominantly during the light period, and also have an impaired response to sleep disturbance (Hu et al., 2007).

PROK2 is a functional target gene of proneural basic helixloop-helix (bHLH) factors. Neurogenin-1 (NGN1) and MASH1 activate *PROK2* transcription by binding to E-box motifs on the *PROK2* promoter with the same set of E-boxes critical for another pair of bHLH factors, CLOCK and BMAL1, in the regulation of circadian clock (Cheng et al., 2002; Zhang et al., 2007).

#### **COMPLEX GENETICS OF KALLMANN SYNDROME CAUSED BY MUTATIONS IN PROKR2 OR PROK2**

We first considered that *PROKR2* was a relevant KS candidate because of the KS-like phenotype of PROKR2-null mice (see above). We thus sequenced the two coding exons of *PROKR2* in a cohort of patients affected by KS, and identified 10 different mutations (one frame-shifting and nine missense mutations) in 14 patients, either in heterozygous state (10 cases) or in homozygous or compound heterozygous state (4 cases) (Dodé et al., 2006) (**Table 2**). Notably, most of these mutations were missense mutations, and many were also found in apparently unaffected individuals, thus initially raising some questions regarding their pathogenic role. A deleterious effect on the signaling activity of PROKR2 was, however, confirmed in transfected HEK-293 cells for most of the mutations (Cole et al., 2008; Monnier et al., 2009).

Then, we considered the possibility that mutations in *PROK2* also account for some KS cases, especially since mutant mice defective in PROK2 showed a marked reduction in the size of their olfactory bulbs. *PROK2* contains four coding exons (Bullock et al., 2004), of which exon 3, encoding an arginine- and lysine-rich peptide of 21 amino acid residues, may or may not be included in the mature transcript due to alternative splicing (Wechselberger et al., 1999) (**Figure 1**). We sequenced the entire coding sequence of *PROK2* in the patients, and identified four different point mutations (two missense mutations, one frame-shifting mutation, and one single nucleotide substitution in the translation initiation sequence), all in the heterozygous state (Dodé et al., 2006). *PROK2* mutations in homozygous state were subsequently found in a few patients, and a KS-like phenotype was concomitantly reported in PROK2-null mutant mice (Pitteloud et al., 2007; Leroy et al., 2008). Since then, additional mutations in *PROKR2* and *PROK2* have been reported in KS patients. A list of the mutations, together with the corresponding references, is provided in **Table 2**.

The finding, for given *PROKR2* and *PROK2* mutations, of both heterozygous and homozygous (or compound heterozygous) unrelated patients is quite remarkable, and argues in favor of a digenic or oligogenic mode of inheritance in heterozygous patients. To date, digenic inheritance of KS has been shown in few patients who had monoallelic missense mutations both in *PROKR2* or *PROK2*, and in other KS genes (*KAL1*, *FGFR1*) or genes underlying normosmic congenital hypogonadotropic hypogonadism (*GNRHR*, *KISS1R*) (Dodé et al., 2006; Cole et al., 2008; Raivio et al., 2009; Martin et al., 2010; Sarfati et al., 2010). The other patients harboring monoallelic mutations in *PROKR2* or *PROK2* are expected to carry at least one additional pathogenic mutation in as yet uncharacterized genes. Indeed, mutations in any of the currently known KS genes have been identified in only 30% of all KS patients, thus indicating that other disease genes remain to be discovered. Notably, it has been found that patients carrying biallelic mutations in *PROKR2* or *PROK2* consistently have a severe, complete KS phenotype, whereas the phenotype of patients carrying monoallelic mutations in these genes is more variable, and likely depends on the additional genetic hits in these patients (Martin et al., 2010; Sarfati et al., 2010).

Kallmann syndrome patients who carry biallelic mutations in *PROK2* or *PROKR2* do not seem to have any of the nonolfactory, non-reproductive occasional anomalies that have been reported in the previously characterized genetic forms of the disease (i.e., X-linked KAL1 form and autosomal dominant KAL2 form), specifically, bimanual synkinesis, renal agenesis, dental agenesis, and cleft lip or palate, and even though sleep disorders or increased body mass index have been reported in some of these patients, there is so far no evidence that these clinical features can be ascribed to the *PROKR2* and *PROK2* mutations, despite potential roles of these genes in sleep-wake regulation and ingestive behavior (see above). In addition, plasma cortisol levels were measured during 24 h in five patients mutated in *PROK2* or *PROKR2*, including one patient with biallelic *PROKR2* mutations, and normal circadian variation was observed in all cases (Sarfati et al., 2010), which argues against a major contribution of PROK2/PROKR2 signaling to physiological circadian variation of plasma cortisol levels in humans. Of course, this result does not exclude the presence of more subtle defects in the patients.

# **CONSEQUENCES OF THE PROKR2 AND PROK2 MISSENSE MUTATIONS ON RECEPTOR SIGNALING ACTIVITY**

A total of 24 *PROKR2* missense mutations have been identified (Dodé et al., 2006; Cole et al., 2008; Sinisi et al., 2008; Chan et al., 2009;Abreu et al., 2010; Sarfati et al., 2010) in KS patients (**Table 2**;

#### **Table 2 | PROKR2 and PROK2 mutations in Kallmann syndrome.**




Mutations reported in PROKR2 and PROK2 are mainly missense mutations. In most patients, the mutations have been found in heterozygous state.The R85C, R85H, R164Q, L173R, and P290S PROKR2 mutations, as well as R73C, c.163del, and c.297\_298insT PROK2 mutations have, however, been found both in heterozygous and homozygous (or compound heterozygous) states, which suggests that patients heterozygous for PROKR2 or PROK2 mutations carry additional mutations, presumably in other, as yet unidentified Kallmann syndrome genes in most cases. Notably, two such patients have the L173R mutation in PROKR2 together with S396L or R423X mutations in KAL1 (Dodé et al., 2006; Sarfati et al., 2010), another patient has the V115M mutation in PROKR2 together with the A24P mutation in PROK2 (Cole et al., 2008), and still another patient has the R85L mutation in PROKR2 together with a A604T mutation in FGFR1 (Sarfati et al., 2010). In addition, the patient who has the S202G mutation in PROKR2 also has I239T and R31C monoallelic mutations in FGFR1 and GNRH1, respectively (Chan et al., 2009). Finally, two patients who carry R268C and V331M mutations in PROKR2 also carry A189T and R240Q monoallelic mutations in KISS1R and GNRHR, respectively (Sarfati et al., 2010). Abbreviation: NMD, nonsense-mediated mRNA decay.

?, not known.

peptide, respectively. The AVITGA motif is shown in red, and the 10 cysteinyl residues (forming five disulfide bonds) are in blue. Vertical arrows indicate the positions of the mutations identified in the patients.

**Figure 2**). Most of the mutant receptors have been characterized regarding their ability to induce intracellular release of calcium upon PROK2-stimulation, their cell surface expression, and their PROK2-binding, in mammalian cell lines (Cole et al., 2008; Monnier et al., 2009; Abreu et al., 2012; Raivio et al., 2012; Sbai et al., in preparation). Although PROKR2 can activate different

**the structural models of the ligand and the receptor.** The mutations are classified in different categories according to their effects on PROKR2 signaling activity: similar to wild-type (green), absence of the receptor at the cell surface (brown), absence of ligand-binding (blue), and strong or mild effect on signaling (red and orange, respectively). The mutations for which functional data are not available are denoted in black. The colored balls indicate the atom of alpha carbon of the polypeptidic chain of the mutated residue. Note that residue R357 is located in the proximal part of the C-terminal region of the receptor.

G protein pathways (Lin et al., 2002; Soga et al., 2002; Chen et al., 2005), its coupling to Gq, leading to intracellular release of calcium, represents the best characterized transduction mechanism. Only five mutants (A51T, R80C, R248Q, M321I, R357W) have properties similar to the wild-type PROKR2, thus calling into question the pathogenic effect of these missense variants. Notably, the R268C mutation has also been found in heterozygous state in a relatively large proportion (174 out of 2203, i.e., 7.9%) of individuals from the African American general population, and in homozygous state in six individuals from the same population (0.3%; see Exome Variant Server website URL: http://evs.gs.washington.edu/EVS/). Moreover, a clear deleterious effect of the R268C mutation on PROKR2 signaling through Gq protein activation could not befound in transfected HEK-293 cells, thus calling into question the pathogenic effect of this missense variant too. Most of the *PROKR2* mutations, however, impair cell surface expression, ligand-binding, or G protein-binding of the receptor (**Figure 2**). Three mutations affecting conserved residues located in the middle of the transmembrane helices, W178S, G234D, and P290S, impede targeting of the receptor to the cell surface. The Q210R mutant receptor is present at the cell surface, but is not able to bind to the ligand. The other mutants

impair, either mildly or strongly, intracellular release of calcium in the Gq signaling pathway, and for some of them this effect might be due to the low expression of the mutant at the cell surface. Interestingly, the mutations that mostly impair the signaling are located nearby the extracellular side or the intracellular side of the receptor, and for these mutants, the loss-of-function is associated to a loss of cell surface expression. The mutations that result in a mildly impaired signaling activity are located in the intracellular loops, in agreement with the important role played by these loops in G protein-coupling. In addition, when wild-type and mutant receptors were coexpressed in HEK-293 cells, none of the mutant receptors that were retained within the cells did affect cell surface-targeting of the wild-type receptor, and none of the mutant receptors properly addressed at the plasma membrane did affect wild-type receptor signaling activity. This argues against a dominant negative effect of the mutations *in vivo*.

A total of 10 missense mutations have been identified in PROK2 (**Table 2**, **Figure 2**) (Dodé et al., 2006; Cole et al., 2008; Leroy et al., 2008; Sarfati et al., 2010). A mutation located at the translation initiation site (−4C >A) likely reduced the protein synthesis. Mutations of conserved residues in prokineticins are expected to strongly impair PROK2 activity such as those in the N-terminal conserved region AVITGA important for prokineticin activity (G32R), and those of the conserved cysteines (C34Y and C46Y). In addition, mutations that introduce an additional cysteine (R73C) in the cysteine-rich region of PROK2 may affect folding of the hormone then leading to a loss-of-function. This is consistent with the functional analysis of Cole et al. (2008) who have examined the signaling properties of three PROK2 mutants: the C34Y and R73C mutations totally abolished and strongly impaired the intracellular calcium release activity of PROKR2, respectively, whereas the I50M mutation had an activity similar to that of the wild-type PROK2. The functional consequence of the other mutations (A24P, S54N, R101Q, R101W, and H104Y) are more difficult to predict in the absence of model of the complex between PROK2 and PROKR2. Further studies will be necessary to better analyze whether the mutations in PROK2 affect the process of binding and/or activation of PROKR2 by the hormone.

#### **CONCLUSION**

The characterization of new genes involved in KS is a difficult goal. Because of the hypogonadotropic hypogonadism, sizes of the KS families are small and the mode of inheritance is very often difficult to establish. Animal models in which a gene has been inactivated and involving a KS phenotype are an alternative approach for the identification of new KS genes. The inactivation of *PROK2* or *PROKR2* lead to defective olfactory morphogenesis and hypogonadism in mice and humans. Curiously, *PROK2* and *PROKR2* mutations in homozygous state were found in a few patients and the main part of the KS patients carried only heterozygous mutations. In all of the functional studies done so far, only Gq signaling pathways have been investigated to characterize the PROK2 and PROKR2 mutants and further studies will be necessary to analyze other signaling pathways. Interestingly, few patients who had monoallelic missense mutations both in *PROK2* or *PROKR2*, and in other KS or normosmic congenital hypogonadotropic hypogonadism genes, raising the idea of oligogenism. So far, this hypothesis has been validated only in a few patients suggesting that the second mutation in the other heterozygous patients resides in unexplored regions of the genes or as yet undiscovered KS genes.

### **REFERENCES**


I. Agénésie des lobes olfactifs (telencéphaloschizis latéral) et des commissures calleuse et antérieure (telencéphaloschizis médian). La dysplasie olfacto-génitale. *Schweiz. Arch. Neurol. Psychiatr.* 74, 309–361.


# **ACKNOWLEDGMENTS**

We thank Jean-Pierre Hardelin for critical reading of the manuscript, Oualid Sbai and Carine Monnier for sharing unpublished data, and Guillaume Lebon for his help in preparing the 3D models of PROKR2 and PROK2. This work was supported by ANR-09-GENO-017-01 and GIS maladies rares (Project A09051KS).

(2007). Altered circadian and homeostatic sleep regulation in prokineticin 2-deficient mice. *Sleep* 30, 247–256.


related to adhesion molecules. *Cell* 67, 423–435.


the regulation of circadian behaviour by suprachiasmatic nuclei. *Proc. Natl. Acad. Sci. U.S.A.* 104, 648–653.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 17 September 2012; paper pending published: 25 October 2012; accepted: 14 February 2013; published online: 12 April 2013.*

*Citation: Dodé C and Rondard P (2013) PROK2/PROKR2 signaling and Kallmann syndrome. Front. Endocrinol. 4:19. doi: 10.3389/fendo.2013.00019*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Dodé and Rondard. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Is it really a matter of simple dualism? Corticotropin-releasing factor receptors in body and mental health

# **Donny Janssen<sup>1</sup> and Tamás Kozicz 1,2,3\***

<sup>1</sup> Department of Cellular Animal Physiology, Donders Institute for Brain, Cognition and Behavior, Nijmegen, Netherlands

<sup>2</sup> Department of Anatomy, Donders Institute for Brain, Cognition and Behavior, Nijmegen, Netherlands

<sup>3</sup> Human Genetics Center, Tulane University, New Orleans, LA, USA

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

James A. Carr, Texas Tech University, USA Rafael Vazquez-Martinez, University

#### of Cordoba, Spain **\*Correspondence:**

Tamás Kozicz, Department of Anatomy, Radboud University Nijmegen Medical Centre, P.O. Box 9101, 6500 HB Nijmegen, Netherlands. e-mail: t.kozicz@anat.umcn.nl

Physiological responses to stress coordinated by the hypothalamo-pituitary-adrenal axis are concerned with maintaining homeostasis in the presence of real or perceived challenges. Regulators of this axis are corticotrophin releasing factor (CRF) and CRF related neuropeptides, including urocortins 1, 2, and 3. They mediate their actions by binding to CRF receptors (CRFR) 1 and 2, which are located in several stress-related brain regions. The prevailing theory has been that the initiation of and the recovery from an elicited stress response is coordinated by two elements, viz. the (mainly) opposing, but well balanced actions of CRFR1 and CRFR2. Such a dualistic view suggests that CRF/CRFR1 controls the initiation of, and urocortins/CRFR2 mediate the recovery from stress to maintain body and mental health. Consequently, failed adaptation to stress can lead to neuropathology, including anxiety and depression. Recent literature, however, challenges such dualistic and complementary actions of CRFR1 and CRFR2, and suggests that stress recruits CRF system components in a brain area and neuron specific manner to promote adaptation as conditions dictate.

**Keywords: CRFR1, CRFR2, CRF, Urocortins, stress, anxiety, depression, HPA-axis**

*"The idea would be not to abolish the CRFR1 receptor's response to the brain's stress hormone but to bring it into the normal range so that it would have appropriate levels of anxiety and stress as conditions dictate."*

*Wylie W. Vale (1941–2012)*

# **INTRODUCTION**

The concept of stress and adaptation was first observed in 1936 and further defined in 1951 (Selye, 1936, 1951). After exposure to a stressor, corticotrophin releasing factor (CRF) is released by endocrine cells of the paraventricular nucleus (PVN) of the hypothalamus into the portal vessels thereby stimulating the release of adrenocorticotropic hormone (ACTH) by the anterior pituitary. Subsequently, circulating ACTH stimulates secretion of glucocorticoids from the adrenal gland (Vale et al., 1981). In the 1980s, the group of Vale isolated the 41-residue CRF peptide from the ovine hypothalamus (Vale et al.,1981) and throughout the decades, many members of the CRF family have been identified, including urocortin 1 (Ucn1), urocortin 2 (Ucn2), and urocortin 3 (Ucn3). The urocortins differ in their tissue distribution and receptor pharmacology (Vaughan et al., 1995; Hsu and Hsueh, 2001; Lewis et al., 2001; Reyes et al., 2001). First shown to be important regulators of the endocrine stress response, the CRF family of neuropeptides is now known to play a role in diverse roles of homeostatic balance, important in mobilization of resources and behaviors during stress (Bale and Vale, 2004). In addition, members of the CRF peptide family play a role in regulation of food intake and satiety, as well as gastrointestinal tract motility, vascular tone, and development, and also acoustic and cardiac function (Heinrichs et al., 1992; Spina et al., 1996; Parkes et al., 1997; Okosi et al., 1998; Koob and Heinrichs, 1999; Maillot et al., 2000; Terui et al., 2001;Vetter et al., 2002; Inoue et al., 2003).

The actions of these peptides are initiated by binding and activating G-protein coupled receptors, CRFR1 and CRFR2, which display distinct affinity for members of the CRF peptide family, with CRF and Ucn1 binding to both receptors, while Ucn2 and 3 selectively bind to CRFR2 (Vaughan et al., 1995; Lewis et al., 2001). Besides CRF receptors (CRFRs) and their cognate ligands, another peptide that plays an important role in the neurobiology of the CRF family of neuropeptides is CRF binding protein (CRF-BP), a 37 kDa N-linked glycoprotein which binds both CRF and Ucn1 with high affinity (Orth and Mount, 1987; Boorse et al., 2005).

In this review, we highlight evidence supporting the notion of a dualistic action of CRFR1 and CRFR2 in mediating an adequate physiological, endocrine, and behavioral response to stress, followed by recent studies challenging this view. As this function is largely dependent upon structure, we will start with a short summary on the functional neuroanatomy of CRF system components followed by data on the animals' stress (mal)adaptation response.

# **DISTRIBUTION OF CRF SYSTEM COMPONENTS CRF**

The distribution of CRF is consistent with its hypothesized functions of controlling the endocrine, physiological, and behavioral response to stress, and it is abundantly expressed in the mammalian brain, with especially high amounts of the peptide concentrated in the parvocellular division of the hypothalamic PVN, bed nucleus of the stria terminalis, central amygdala, lateral hypothalamus, and locus coeruleus (Merchenthaler et al., 1982; Morin et al., 1999). Dense CRF fibers have been located in the lateral septum, bed nucleus of the stria terminalis, central nucleus of the amygdala, median eminence, the raphe nuclei, and the spinal cord (Merchenthaler et al., 1982; Morin et al., 1999; Korosi et al., 2007).

#### **UROCORTIN 1**

Urocortin 1 is related to CRF, with a sequence identify of 45% (Vaughan et al.,1995). It binds CRFR2 with 100-fold higher affinity than CRF does, indicating that this peptide might be an endogenous ligand for CRFR2 (Vaughan et al., 1995; Chalmers et al., 1996; Perrin and Vale, 1999). The most dominant sites of Ucn1 expression in the mammalian brain includes the centrally projecting Edinger–Westphal nucleus (EWcp) in the rostroventral midbrain, the supraoptic nucleus in the hypothalamus and the superior lateral olive, which have been confirmed in mammalian and non-mammalian species (Iino et al., 1997; Takahashi et al., 1998; Bittencourt et al., 1999; Kozicz et al., 2002, 2008, 2011a; Ryabinin et al., 2005). Networks of Ucn1 fibers have been identified in the lateral septum, internal layer of the median eminence, dorsal raphe nucleus, and the spinal cord, with scattered fibers located in the hypothalamus, hippocampus, cortex, and posterior pituitary (Kozicz et al., 1998; Bittencourt et al., 1999; Iino et al., 1999; Morin et al., 1999; Korosi et al., 2007).

#### **UROCORTIN 2**

Urocortin 2 is a 38 amino acid peptide, which selectively binds CRFR2 (Reyes et al., 2001). Ucn2 mRNA shows a distinct subcortical distribution, including regions known to be involved in physiological and behavioral responses to stress, such as the PVN, locus coeruleus, and the arcuate nucleus (Reyes et al., 2001). Ucn2 mRNA partially overlaps with the expression of CRF in the PVN and Ucn1 in the brainstem and spinal motor nuclei (Swanson et al., 1983; Bittencourt et al., 1999; Reyes et al., 2001). To date no reliable Ucn2 immunohistochemistry has been performed (due to lack of reliable and specific antibody against Ucn2), hence the distribution of Ucn2 immunoreactive fiber terminals remains to be mapped.

#### **UROCORTIN 3**

Urocortin 3 is just like Ucn2 a 38 amino acid peptide which selectively binds CRFR2 (Lewis et al., 2001). Ucn3 expressing neurons have a limited distribution compared to CRF and Ucn1, and the peptide is mostly located in forebrain regions, including the preoptic region, hypothalamus, and amygdala. Two areas of the hypothalamus express Ucn3, the first population being neurons in the preoptic nucleus and the second group of neurons are associated with the fornix ("perifornical"Ucn3 neuron population) near the PVN of the hypothalamus (Lewis et al., 2001; Li et al., 2002). More Ucn3 expressing neurons are located in the medial amygdala and paraolivary nucleus (Lewis et al., 2001; Li et al., 2002). Densest innervation of Ucn3 includes the ventromedial hypothalamus, lateral septum, posterior division of the bed nucleus of the stria terminalis and the medial amygdala, which are areas known to express CRFR2 (Lewis et al., 2001; Li et al., 2002).

#### **CRF BINDING PROTEIN**

CRF binding protein is relatively widely expressed in the mammalian brain, including the cerebral cortex, subcortical limbic system, various sensory relays, raphe nuclei, hypothalamus, and pituitary (Orth and Mount, 1987; Behan et al., 1989; Potter et al., 1992). In humans, the CRF-BP has been found in the liver and in the circulation, where it inactivates CRF and/or Ucn1. The exact function of the protein is still unknown, but it has been proposed that CRF-BP prevents inappropriate pituitary-adrenal activation, e.g., during pregnancy (Potter et al., 1991). Upon binding to CRF-BP, CRF forms a dimer complex and is thought to modulate the endocrine activity of CRF (Lowry et al., 1996). In rat pituitary cells, recombinant CRF-BP blocks secretion of ACTH (Potter et al., 1991). The binding proteins have also been detected in brain regions not associated with CRF actions, suggesting it may have other functions independent from CRF (Potter et al., 1992; Cortright et al., 1995; Bale and Vale, 2004).

#### **CRFR1**

CRFR1 is a 415 amino acid peptide in mammals and has a widespread expression in stress-related areas located in the central nervous system (CNS). CRFR1 mRNA has been located in the cortex, cerebellum, hippocampus, amygdala, olfactory bulb, lateral septum, thalamus, basal ganglia, the raphe nuclei, pituitary, brain stem, and spinal cord (Van Pett et al., 2000; Korosi et al., 2006, 2007; Justice et al., 2008; Kuhne et al., 2012). Outside the CNS, the CRFR1 is expressed in the thymus, spleen, skin, ovary, testis, gastrointestinal tissue, and adrenal gland (Dufau et al., 1993; Nappi and Rivest, 1995; Slominski et al., 1995; Baigent and Lowry, 2000; Muller et al., 2001; Chatzaki et al., 2004).

A recent study, which aimed at accurately determining the presence of CRFR1 in the brain, revealed that this receptor is present in glutamatergic neurons of the cortex and hippocampus, in gammaaminobutyric acid containing neurons of the reticular thalamic nucleus, globus pallidus, and septum, and in dopaminergic neurons of the substantia nigra pars compacta and ventral tegmental area. It is also expressed in few serotonergic neurons of the dorsal and media raphe nuclei (Refojo et al., 2011).

#### **CRF RECEPTOR 2**

CRF receptor 2 is a 397–437 amino acid protein in mammals and it is abundantly expressed in both the CNS and in the periphery (Kishimoto et al., 1995; Lovenberg et al., 1995b; Kostich et al., 1998; Palchaudhuri et al., 1999; Van Pett et al., 2000; Korosi et al., 2007; Justice et al., 2008; Kuhne et al., 2012). In the CNS, CRFR2 is expressed in the olfactory bulb, hippocampus, amygdala, septum, the dorsal and median raphe nuclei, cortex, pituitary, and spinal cord (Palchaudhuri et al., 1999; Bittencourt and Sawchenko, 2000; Van Pett et al., 2000; Korosi et al., 2006, 2007; Lukkes et al., 2011). In the periphery, the receptor has been identified in the retina, heart, skeletal muscle, vasculature, adrenal gland, and gastrointestinal tissue (Lovenberg et al., 1995a; Palchaudhuri et al., 1999; Muller et al., 2001; Chatzaki et al., 2004).

Despite the fact that the distribution of CRFR1 and CRFR2 overlaps in the brain, clear distinctions can be made among brain subregions. For example, the CRFR1 receptor is expressed in the basolateral nucleus of the amygdala, while both CRFR1 and CRFR2 are present in the medial nucleus of the amygdala (Bittencourt and Sawchenko, 2000; Van Pett et al., 2000). Interestingly, mRNA of CRFR2 has been discovered in non-neuronal structures as well, like the choroid plexus and cerebral arterioles (Lovenberg et al., 1995a). In the cerebral arteries and arterioles, stimulation of CRFR2 leads to an increased cerebral blood flow (De Michele et al., 2005).

#### **EVIDENCE SUPPORTING A DUALISTIC ACTION OF CRFR1 AND CRFR2 DURING STRESS (MAL)ADAPTATION STRESS**

Stress has been defined as various physiologic alterations, including misbalance of homeostasis and activation of the hypothalamopituitary-adrenal (HPA)-axis (Selye, 1936, 1951; Dallman et al., 1987). Upon exposure to a stressor, the HPA-axis is activated by the release of CRF into portal vessels and subsequently acts upon the CRFR1 located in the anterior pituitary (Turnbull and Rivier, 1997;Vale et al., 1997). The initial understanding on the biological significance of CRF system components governing the endocrine stress response has come from various genetically modified animal models. As follows we will overview these animal models.

#### **CRF knockout**

In order to assess the central physiological role of CRF a knockout model was created (Muglia et al., 1995). Mice lacking CRF show decreased glucocorticoid levels after exposure to stress, showing the importance of CRF in regulating the HPA-axis (Muglia et al., 1995; Venihaki and Majzoub, 1999). However, interpretation of data from this animal model is significantly hampered/influenced by compensatory changes in other members of the CRF system. For example, in the absence of CRF, Ucn1 levels in the EWcp are elevated, while distribution remains unchanged (Weninger et al., 1999, 2000).

#### **CRF overexpression**

In order to study chronic activation of the HPA-axis, two independent CRF overexpressing mouse lines were created (Stenzel-Poore et al., 1992; Groenink et al., 2002). Both lines revealed elevated basal plasma corticosterone levels in response to stress. Delayed and attenuated HPA-axis hormone responses to stress, which might be the result of HPA-axis desensitization, were reported in the mice created by Stenzel-Poore et al. (Coste et al., 2001).

#### **Ucn1 knockout**

Two independent Ucn1 knockout models were generated by two groups (Vetter et al., 2002; Wang et al., 2002). Both lines have normal endocrine stress responses, supporting the notion that Ucn1 has a minor or no role in stress induced activation of the HPAaxis. It has been proposed that Ucn1 plays a role in adaptation to stress, rather than initiating the stress response. The fact that Ucn1 KO mice have impaired adaptation to repeated stress supports this notion (Zalutskaya et al., 2007).

#### **Ucn2 knockout**

Ucn2 knockout model showed that HPA-axis activity was normal in male and female mice deficient for Ucn2 (Breu et al., 2012). Stress induced release of corticosterone was equal between knockout and wild type animals. Interestingly, a gender specific phenotype was detected in Ucn2 deficient mice. Females, but not males, lacking Ucn2 show a significant increase in basal daily rhythms of ACTH and corticosterone (Chen et al., 2006). It is known that CRFR2 can modulate the HPA-axis, however, Ucn2 deficiency does not seem to have an impact on the stress induced activation of the axis (Breu et al., 2012).

#### **Ucn3 knockout**

Ucn3 knockout mouse model recently showed that absence of Ucn3 does not have an impact on basal activity of the HPA-axis, as corticosterone levels remained unchanged when compared to wild type littermates (Deussing et al., 2010). After exposure to stress, the corticosterone levels in Ucn3 knockout and wild type mice were equal, which resulted in an operational negative feedback loop of corticosterone on the HPA-axis (Deussing et al., 2010).

#### **Multiple Ucn knockouts**

The importance of the urocortins in the stress system was confirmed by multiple urocortin knockout studies.A double knockout of Ucn1 and 2 in mice showed equal basal corticosterone, but elevated levels after exposure to stress (Neufeld-Cohen et al., 2010a). However, the elevated corticosterone levels after exposure to stress were not observed in an urocortin 1, 2, and 3 triple knockout (Neufeld-Cohen et al., 2010b). These mice deficient for the urocortins were unable to recover properly and this was paired with dysregulated serotonergic function in stress-related neuronal circuits (Neufeld-Cohen et al., 2010b).

#### **CRFR1 knockout**

Two independent lines of CRFR1 knockout lines have demonstrated the importance of CRFR1 in regulation of the HPA-axis in response to stress. KO mice in both lines show an attenuated response to restraint stress by a minimal increase in ACTH and corticosterone. However, basal levels in the CRFR1 null mice are equal between KO and wild type littermates (Smith et al., 1998; Timpl et al., 1998).

#### **CRFR2 knockout**

CRFR2 knockout mice reveal a role for CRFR2 in regulating HPAaxis activation in response to stress, but it appears initiation of the response seems to be normal. These null mice show an early termination of ACTH release, suggesting a role for CRFR2 in maintaining activation of the HPA-axis (Coste et al., 2000). Furthermore, coping with stressors seems to be reduced in CRFR2 KO mice (Coste et al., 2000). Mice deficient for CRFR2 are also hypersensitive to stress, which leads to increased anxiety-like behaviors (Bale et al., 2000; Gammie et al., 2005). Mutant CRFR2 mice have increased CRF mRNA in the central amygdala and increased Ucn-1 and -3 in the EWcp and lateral perifornical region, respectively suggesting a compensatory activation of extrahypothalamic CRF and Ucn systems (Bale et al., 2000, 2002a; Kozicz, 2009). Chronic activation of CRFR2 also promotes an anxiety-like

state, with attenuated behavioral and HPA-axis responses to stress (Neufeld-Cohen et al., 2012).

#### **CRFR1 and 2 double knockout**

CRFR1 and CRFR2 double knockout mice show the central role of these receptors in the stress response system. In the absence of both receptors, the double knockout mice show an impaired HPAaxis activation in response to stress (Preil et al., 2001; Bale et al., 2002b). Further, ACTH and corticosterone levels after exposure to stress are lowered in the double knockout mice compared to CRFR1 deficient mice, suggesting a role for CRFR2 in mediating HPA-axis sensitivity (Coste et al., 2000; Bale et al., 2002b).

#### **ANXIETY**

The CRF system has been proposed to be involved in the development of anxiety-related disorders. Cumulative evidence relates dysregulation of CRF systems to the etiology and pathobiology of stress-associated diseases, including anxiety. The following studies which are related to anxiety- and depressive-like behaviors are summarized in **Table 1**.

# **CRF knockout**

Surprisingly, no behavioral abnormalities have been reported with the CRF knockouts generated by Muglia et al. (1995). Similarly, anxiety-like behavior is the same between knockout and wild type animals under normal and stressful conditions in another CRF deficient mouse created by Weninger et al. (1999). Interestingly, CRFR antagonists have an anxiolytic effect in CRF knockout animals, suggesting that CRFR1 activation is crucial to induce anxiety, while CRF itself may not. These data point toward the significance of another CRFR ligand, like Ucn1, or a yet unidentified neuropeptide, compensatingfor CRF deficiency in the brain (Weninger et al., 1999; Kozicz, 2007; Kozicz et al., 2011b).

#### **CRF overexpression**

In the model created by Stenzel-Poore et al. (1992), increased anxiety-like and decreased exploratory behavior was observed (Heinrichs et al., 1997; van Gaalen et al., 2002). This increase in anxiety-like behavior could be reversed by administration of CRF antagonist alpha-helical CRF (Stenzel-Poore et al., 1994). Removing the adrenal gland in these animals did not inhibit the anxiogenic effects of CRF overexpression, although corticosterone levels were normalized. This suggests that behavioral effects caused by CRF overexpression are mediated by CRF and CRFRs, rather than being affected by corticosterone (Heinrichs et al., 1997). Along these lines, Refojo et al. (2011) demonstrated that overexpression of limbic CRF in CRF-COECamk2aCre mice (Camk2aCre, Cre driven by the calcium/calmodulin-dependent protein kinase type II alpha chain promoter) resulted in increased anxiety-like behavior too, suggesting that limbic CRF in particular would be instrumental in mediating anxiety-like behavior. To date, no data is available on anxiety-like behavior of the CRF overexpressing mice created by Groenink et al. (2002).

#### **Ucn1 knockout**

The behavioral phenotype of Ucn1 KO mice is controversial. The group of Wang et al. (2002) have shown no differences in anxiety-like behavior, whereas Vetter et al. (2002) have demonstrated that Ucn1 deficient animals display increased anxiety-like behavior.

#### **Ucn2 knockout**

Ucn2 knockout mice have revealed no significant changes in anxiety-like behavior. However, when investigating social activities, it was seen that male Ucn2 deficient mice had more passive social interactions and reduced aggressiveness compared to wild type littermates, suggesting that Ucn2 may rather modulate aggressive behavior in male mice (Breu et al., 2012).

#### **Ucn3 knockout**

Ucn3 knockout mice created by Deussing et al. (2010) showed unchanged anxiety-related behaviors compared to wild type mice. However, based on the prevalent expression of Ucn3 throughout the accessory olfactory bulb and altered social discrimination abilities of male and female Ucn3 knockout mice, it has been suggested that Ucn3 plays a role in processing of social cues and establishment of social memories (Deussing et al., 2010).

#### **Multiple Ucn knockout**

Urocortin 1 and 2 double knockout mouse model has demonstrated an anxiolytic phenotype (Neufeld-Cohen et al., 2010a). A further reduction in anxiety-like behavior was observed after double Ucn1/Ucn2 deficient mice were exposed to acute stress, and this reduction in anxiety was correlated with levels of serotonin in anxiety-related brain regions (Neufeld-Cohen et al., 2010a). Remarkably, the triple urocortin knockout mouse showed an increase in anxiety-like behavior 24 h post-stress. This increase in anxiety in triple urocortin knockout mice was also associated with serotonergic function in stress-linked neurocircuits. It has been suggested that Ucn3 is pivotal to the observed phenotype of the triple knockout model (Neufeld-Cohen et al., 2010b).

#### **CRFR1 knockout**

Three studies reported that mice deficient for CRFR1 exhibited, what seemed to be, reduced anxiety-like behavior. In different tests of spontaneous anxiety, open field (a measure of exploratory behavior and general activity), light-dark box (based on the aversion of mice to well illuminated areas and on exploratory behavior in response to stressors), defensive withdrawal (a measure of conflict between exploratory behavior and retreat), and elevated plus maze (based on the aversion for open and elevated areas and on exploratory behavior in novel areas), which normally inhibit behavioral activity, CRFR1 deficient mice showed heightened levels of locomotion consistent with anxiety-like behavior (Smith et al., 1998; Timpl et al., 1998; Contarino et al., 1999). The anxiolytic effect of CRFR1 deficient mice seems to be mediated by reduced CRFR1 expression in the basolateral amygdala (Sztainberg et al., 2010). Furthermore, stress induced hormone secretion showed that CRFR1 knockout mice had reduced levels of ACTH and corticosterone, providing evidence that CRFR1 mediates anxiety and stress induced-hormone activation (Smith et al., 1998; Timpl et al., 1998). Opposite effects were found in CRFR1Camk2aCre mice generated by Muller et al. (2003)in which Cre mediated deletion of CRFR1 starts in the second week of postnatal life. After

#### **Table 1 | Summary of animal models targeting CRF system components.**



↑, indicates an increase; ↓, indicates a decrease; =, indicates no difference in anxiety- or depressive-like behavior as comparing mutant and wildtype animals.

exposure to restraint stress, these mice exhibited elevated levels of ACTH and corticosterone, while showing reduced anxiety-like behavior.

#### **CRFR2 knockout**

In contrast to CRFR1 deficient mice, CRFR2 deficient mice do not show a consistent change in anxiety-like behavior. One study has shown that CRFR2 knockouts produced no significant effects on anxiety responses in the elevated plus maze or in an open field test (Coste et al., 2000). Another study has demonstrated no behavioral changes in the light-dark test, but appealingly, the CRFR2 knockout mice have increased anxiety in the elevated plus maze and open field test (Kishimoto et al., 2000). These anxiogeniclike effects may be influenced by the reported increase in CRF mRNA in the central amygdala of the CRFR2 knockout mice (Bale et al., 2000). This brain region is associated in the activation of diverse responses induced by stress (Davis, 1992). However, some data suggests a sex effect of emotional behavior mediated by CRFR2. Female knockout mice seem to have behavior comparable to wild type littermates, while the male knockouts exhibit more anxiogenic-like behavior (Kishimoto et al., 2000).

#### **CRFR 1 and 2 double knockout**

CRFR 1 and 2 double knockout mice, in terms on anxiety-like behavior, displayed a sexually dichotomous phenotype (Preil et al., 2001; Bale et al., 2002b). Female double mutant mice showed reduced levels of anxiety-like behavior, while male double knockouts did not, compared to their wild type littermates (Bale et al., 2002b). The behavioral phenotype of the double knockout model generated by Preil et al. (2001) has not been researched yet.

#### **DEPRESSION**

Strong evidence links the stress response, and the sensitivity to stressful encounters, to the development of depression (e.g., Nestler et al., 2002; de Kloet et al., 2005; McEwen et al., 2012). While the stress response is essential for successful adaptation, chronic stress can accelerate disease processes, and lead to depression or other mood disorders (Nestler et al., 2002). As follows we will highlight some of the available large body of evidence linking CRF family of neuropeptides and their receptors to the development of depression.

#### **CRF**

Elevated CRF levels and decreased receptor expression have been found in post-mortem examination of suicide victims. In addition, disproportionate activation of the HPA-axis has been reported in more than one-half of patients diagnosed with depression, and these symptoms can be corrected by treatment with antidepressants (Holsboer, 1999). CRF was also found to be elevated in cerebrospinal fluid of depressed patients, which was reversed in patients treated with antidepressants (De Bellis et al., 1993; Heuser et al., 1998). Reduced CRF binding sites in the frontal cortex of suicide victims have also been identified (Nemeroff et al., 1984, 1988). This was interpreted as central CRF overabundance leading to CRFR desensitization, a common phenomenon for G-protein coupled receptors (Holsboer and Ising, 2010). Furthermore, hypercortisolemia and impairment of negative feedback by cortisol on the HPA-axis have also been attributed to elevated CRF levels (Reul and Holsboer, 2002; Keck, 2006). In support for a role for CRF overexpression in the pathobiology of depression, a recent study performed by Keen-Rhinehart et al. (2009), in which CRF was chronically overexpressed in the central nucleus of the amygdala in female rats revealed an amplified CRF concentration in the PVN and a decreased glucocorticoid negative feedback, both markers which have previously been associated with the pathophysiology of depression (Nemeroff et al., 1984, 1988). Depressive-like behavior was further confirmed in the

forced swim test (FST; measures escape behavior and behavioral despair), as mutant rats overexpressing CRF showed increased floating times and reduced escape behavior (Keen-Rhinehart et al., 2009).

No behavioral tests were performed on the CRF null mice created by Muglia et al. (1995) and Weninger et al. (1999) to assess depressive-like behavior.

#### **Urocortins**

A study where urocortin peptides have been injected i.c.v. into mice (Tanaka and Telegdy, 2008) revealed that while Ucn1 had no effect on the animal's behavior, both Ucn2 and 3 displayed strong antidepressant-like activity by decreasing immobility time and increased climbing and swimming behavior in a FST (Tanaka and Telegdy, 2008). Moreover, single nucleotide polymorphisms (SNP) located in the gene for Ucn3 have also been associated with the antidepressant response (Wong et al., 2008). While a study assessing levels of Ucn1 and 2 has not found any significantly changes in patients suffering from major depressive disorder vs. controls (Kang et al., 2007), Ucn1 mRNA is markedly upregulated in the EWcp in depressed male suicide victims, compared to healthy males (Kozicz et al., 2008). This upregulation was not observed in female suicide victims, suggesting a gender specific neuropathology (Kozicz et al., 2008).

In animal models for depression, Ucn2 knockout mice showed a significant decrease in depressive-like behavior as assessed by the FST. These effects were only evident in the female null mice, suggesting a role for Ucn2 in mediating the sex differences observed in the stress response (Wang et al., 2002;Chen et al., 2006) albeit in an opposite direction as Ucn1 may (Kozicz et al., 2008). In contrast, mice deficient for Ucn3 have unchanged depressive-like behavior compared to wild type mice (Deussing et al., 2010).

#### **CRFR1**

CRFR1 activation of CRFR1 has been associated with anxiety or depressive-like behaviors, which could be treated by administrating CRFR antagonists in patients suffering from major depression (Kehne and Cain, 2010). Furthermore, quantitative PCR analyses have shown that mRNA for CRFR1, but not CRFR2, is reduced in the frontopolar cortex in suicide brains, which might be secondary to high CRF or Ucn1 levels (Merali et al., 2004; Kozicz et al., 2008). These data suggests that at least for CRFR1, dysregulation in the forebrain may contribute to the neuropathology of depression. To date,major pharmaceutical companies are focusing on CRFR1 as the primary target for antidepressant development. However, limited phase 2/3 clinical trial results with two CRFR1 antagonists, suggest a lack of efficacy in patients suffering from major depressive disorder (Kehne and Cain, 2010). This suggests that an overall inhibition of CRFR1 does not inhibit depressivelike behavior, which could mean a more complex role for CRFR1 in mediating mood disorders (see later in text).

#### **CRFR2**

The role of CRFR2 is more complex, as the receptor has been associated with enhancement as well as inhibition of stress responsivity (Kehne and Cain, 2010). More specifically, CRFR2 deficient mice tested in the FST displayed increased immobility which indicates depressive-like behavior (Bale and Vale, 2003; Todorovic et al., 2005) which could be effectively reversed by the CRFR1 antagonist antalarmin (Bale and Vale, 2003).

Taken together, the findings listed above have led to the notion that activation of CRFR1 is responsible for the initiation of the stress response and mediates a rather pro anxiety/depression behavior (**Table 1**). In contrast, CRFR2 is involved in the recovery phase of the stress response, and exhibit a more anxiolytic anti-depression function (**Table 1**). Although a great deal of experimental evidence supported this dualistic action of CRFR1 and CRFR2 in health and disease, not all data are in favor of this notion. Consequently, the fundamental question has recently been raised; are the endocrine, physiological and behavioral changes controlled/mediated by CRFR1 and CRFR2 the consequence of a balanced, dualistic function of these receptors or in a broader sense that of CRF system components? Or can we possibly identify specific brain regions and/or neuron populations responsible for the endocrine, physiological and behavioral stress response? Recent technical developments have allowed us to directly address this issue by creating unique animal models.

## **FINDINGS THAT CHALLENGE THE DUALISTIC ACTION OF CRFR1 AND CRFR2 IN STRESS (MAL)ADAPTATION**

Conditional mutagenesis of CRFR1 using a *Cre-Lox* system has led to the development of various conditional knockouts were CRFR1 has been deleted in specific neuron populations previously implicated in stress and stress-associated anxiety and depression: (a) CRFR1GLU-CKO, where CRFR1 is deleted in forebrain glutamatergic neurons; (b) CRFR1GABA-CKO, deleting the receptor in forebrain GABAergic neurons; (c) CRFR1DA-CKO, carrying CRFR1 deletion in midbrain dopaminergic neurons; and (d) CRFR15HT-CKO, deleting CRFR1 in brainstem serotonergic neurons (Refojo et al., 2011) (**Figure 2**). The roles of the different neuronal populations expressing CRFR1 on emotional behavior, these knockouts were subjected to a series of tests. CRFR1GLU-CKO mice showed reduced anxiety-like behavior pointing toward a central role of glutamate neurotransmission in stress induced anxiety. These mice lack CRFR1 expression in glutamatergic neurons in the hippocampus and amygdala, two important limbic regions in the neuropathology of mood disorders (Refojo et al., 2005; Sanacora et al., 2008). In line with these data, CRFR1Glu-CKO animals also show impairments in CRF-induced changes on excitatory neurotransmission in these limbic regions (Refojo et al., 2011). It was also shown that CRFR1 facilitates neuronal activity propagation from the classical hippocampal input region (dentate gyrus) to the CA1 output area. In conclusion, activation of CRFR1 in limbic glutamatergic neurons, as it would in response to a stressor, modulates glutamatergic neurotransmission, giving rise to neuronal excitation in the hippocampal network, and consequently anxiogenesis (Refojo et al., 2011).

Unexpectedly, for CRFR1DA-CKO, an increase in anxietylike behavior was found, whereas for CRFR1GABA-CKO and CRFR15HT-CKO no anxiety-related behavioral changes were observed (Refojo et al., 2011). As to the CRFR1DA-CKO, the deletion of CRFR1 in dopaminergic neurons in the midbrain ventral tegmental area and substantia nigra pars compacta (VTA/SNpc), together with the fact that CRF increases the action potential

**FIGURE 1 | Schematic cross-section of the mouse brain showing site specific overexpression of CRF and the resulting change on anxiety- or depressive-like behavior.** Areas of interest are the Bed nucleus of the Stria Terminalis (BST) (Regev et al., 2011), central nucleus of the Amygdala (CeA) (Keen-Rhinehart et al., 2009; Regev et al., 2011), and the paraventricular nucleus (PVN) (Elliott et al., 2010). \*Change in anxiety- or depressive-like behavior is dependent on the conditions the mice were subjected to. AON, anterior olfactory nucleus; Apit, anterior pituitary; ARC, arcuate nucleus; BLA, basolateral amygdala; BST, bed nucleus of the stria terminalis; CA1-3, fields CA1-3 of Ammon's horn; CC, corpus callosum;

firing of dopamine neurons in the VTA (Wanat et al., 2008) are important to control the animal's mood. It was also shown that CRFR1DA-CKO displayed a decreased response to stress induced dopamine release in the prefrontal cortex (PFC), indicating that CRFR1 targets dopamine cells to control dopamine release into the PFC under stressful conditions (Refojo et al., 2011). The findings mentioned above strongly imply that stress-associated anxietylike behavior might be a consequence of an imbalance between CRFR1 controlled glutamatergic and dopaminergic neuronal populations involved in mediating emotional behavior (Refojo et al., 2011).

In addition, the contrasting roles for CRFR1 expressed by dopaminergic vs. glutamatergic neurons suggest that under physiological conditions, CRF and CRFR1 controlled glutamatergic and dopaminergic systems might function in an antagonistic manner to maintain adaptive anxiety responses during periods of stress (Refojo et al., 2011). The fact that simultaneous deletion of CRFR1 in both glutamatergic and dopaminergic neurons do not show alterations in anxiety-like behavior, supports this notion.

A neuron population specific role for CRF system components in mediating stress-associated emotional response is further supported by a study by Sztainberg et al. (2011). These authors have demonstrated that a viral-mediated specific knock-down (KD) of CRFR1 in the globus pallidus results in increased anxietylike behavior in mice (Sztainberg et al., 2011) (**Figure 2**). Earlier studies have demonstrated that this region is a key mediator of anxiety behavior, as it is associated with motor and associative functions (Baumann et al., 1999; Critchley et al., 2001;

CeA, central nucleus of the amygdala; Cereb, cerebellum; CingCx, cingulate cortex; DBB, diagonal band of Broca; DG, dentate gyrus; EWcp, centrally projecting Edinger–Westphal nucleus; FrCx, frontal cortex; IC, inferior colliculus; IPit, intermediate pituitary; LC, locus coeruleus; LS, lateral septum; MS, medial septum; NTS, nucleus tractus solitarii; OB, olfactory bulb; OccCx, occipital cortex; PAG, periaqueductal gray; ParCx, parietal cortex; PPit, posterior pituitary; PVN, paraventricular nucleus; RN, raphe nuclei; SC, superior colliculus; SN, substantia nigra; SON, supraoptic nucleus; VLM, ventrolateral medulla; VMH, ventromedial hypothalamus; VTA, ventral tegmental area.

Kita, 2007; Sztainberg et al., 2011). In contrast CRFR1 expressed in the basolateral amygdala mediates anxiogenic behaviors in mice, further substantiating the view that the very same receptor can mediate opposing behavioral responses depending on the brain site of action (Sztainberg et al., 2010) (**Figure 2**).

To date there is only one study examining a site specific action of CRFR2. A lentiviral KD of CRFR2 specifically in the bed nucleus of the stria terminalis reduces susceptibility to stress induced anxiety in an animal model for post-traumatic stress disorder, suggesting an important role for CRFR2 in stress coping (Lebow et al., 2012) (**Figure 2**).

With regard to CRF, a site specific action can also been proposed. More specifically, prolonged and site specific overexpression of CRF in the central amygdala attenuates stress induced anxiety-like behaviors without effecting depression-like behavior, whereas CRF overexpression in the bed nucleus of the stria terminalis increased depressive-like behavior, without affecting anxiety levels in male mice (Regev et al., 2011) (**Figure 1**). Somewhat conflicting data were reported in a mouse model, in which unrestrained CRF synthesis in CeA produced a dysregulation of the HPA-axis, depression-like behavior, as well as physiological and reproductive consequences associated with stress-related disorders (Keen-Rhinehart et al., 2009) (**Figure 1**).

Using two lentiviral-based approaches to specifically KD or conditionally overexpress (OE) CRF in the CeA of adult mice anxiety- and depression-like behaviors were evaluated under basal and stressful conditions. Intriguingly, changing CeA-CRF levels mildly affected anxiety-like behaviors under basal conditions. In contrast, following exposure to an acute stressor, CeA-CRF-KD

change on anxiety-like behavior in mice which lack CRFR1 in specific brain areas (Refojo et al., 2011). \*In one study, CRFR1 was deleted in only the GP (Sztainberg et al., 2011), while in another study CRFR1 was deleted in all areas expressing GABA (Refojo et al., 2011). Other areas of interest are the basolateral Amygdala (BLA) (Sztainberg et al., 2010) and Bed nucleus of the Stria Terminalis (BST) (Lebow et al., 2012). AON, anterior olfactory nucleus; Apit, anterior pituitary; ARC, arcuate nucleus; BLA, basolateral amygdala; BST, bed nucleus of the stria terminalis; CA1-3, fields CA1-3 of

strongly attenuated stress induced anxiety-like behaviors, whereas a short-term CeA-CRF-OE enhanced the stress induced effects on these behaviors (Regev et al., 2012). This study suggests that an adequate behavioral response to stress is not only determined by the level and site of expression of CRF system components, but depends also on the condition an animal is exposed to.

Based on the pioneering studies discussed above (for overview see **Table 1** as well as **Figures 1** and **2**), a picture is emerging that the roles of CRF system components in the animal's stress response and mood cannot be simplified to a dualistic model of action, but are rather linked to the recruitment of specific brain areas and neuron populations. Therefore, researches in the field must embark on a long journey to systematically and comprehensively exploring the significance of the many brain areas and neuron populations expressing one or several of the CRF system components, before

#### **REFERENCES**


lateral septum; MS, medial septum; NTS, nucleus tractus solitarii; OB, olfactory bulb; OccCx, occipital cortex; PAG, periaqueductal gray; ParCx, parietal cortex; PPit, posterior pituitary; RN, raphe nuclei; RTN, reticular thalamic nucleus; SC, superior colliculus; SN, substantia nigra; SON, supraoptic nucleus; VLM, ventrolateral medulla; VMH, ventromedial hypothalamus; VTA, ventral tegmental area.

we can conclude on the role(s) of CRFR1 and CRFR2 in health and disease.

### **CONCLUSION**

Taken together, recent studies using conditional mutagenesis or viral knock-down or overexpression of CRF system components strongly indicate that the involvement of CRF system components in stress-associated anxiety and depression-like behavior cannot be explained by a universal and brain wide mechanisms, and therefore the earlier proposed dualistic action of CRFR1 and CRFR2 in stress may not hold. Rather, CRF system components would be specifically recruited in particular functional neuronal circuits, which in turn would allow the expression of the necessary and sufficient endocrine, physiological, and behavioral responses to stress as conditions dictate.


ganglia in mood disorders: preliminary data from a postmortem study. *J. Neuropsychiatry Clin. Neurosci.* 11, 71–78.


R. (1999). Differential distribution of urocortin- and corticotropinreleasing factor-like immunoreactivities in the rat brain. *Neuroscience* 92, 281–291.


*Proc. Natl. Acad. Sci. U.S.A.* 102, 6183–6188.


Gold, L. H., et al. (1998). Corticotropin releasing factor receptor 1-deficient mice display decreased anxiety, impaired stress response, and aberrant neuroendocrine development. *Neuron* 20, 1093–1102.

Spina, M., Merlo-Pich, E., Chan, R. K., Basso, A. M., Rivier, J., Vale, W., et al. (1996). Appetite-suppressing effects of urocortin, a CRF-related neuropeptide. *Science* 273, 1561–1564.


al. (1998). Impaired stress response and reduced anxiety in mice lacking a functional corticotropin-releasing hormone receptor 1. *Nat. Genet.* 19, 162–166.


al. (2002). Urocortin-deficient mice display normal stress-induced anxiety behavior and autonomic control but an impaired acoustic startle response. *Mol. Cell. Biol.* 22, 6605–6610.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 18 September 2012; accepted: 22 February 2013; published online: 12 March 2013.*

*Citation: Janssen D and Kozicz T (2013) Is it really a matter of simple dualism? Corticotropin-releasing factor receptors in body and mental health. Front. Endocrinol. 4:28. doi: 10.3389/fendo.2013.00028*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Janssen and Kozicz. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Stress responsiveness of the hypothalamic–pituitary– adrenal axis: age-related features of the vasopressinergic regulation

# **Nadezhda D. Goncharova1,2\***

<sup>1</sup> Research Institute of Medical Primatology of Russian Academy of Medical Sciences, Sochi, Russia <sup>2</sup> Sochi State University, Sochi, Russia

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

James A. Carr, Texas Tech University, USA Dóra Zelena, Institute of Experimental

Medicine, Hungary

#### **\*Correspondence:**

Nadezhda D. Goncharova, Research Institute of Medical Primatology of Russian Academy of Medical Sciences, Veseloye 1, Adler, Sochi 354376, Krasnodarskii Krai, Russia. e-mail: ndgoncharova@mail.ru

The hypothalamic–pituitary–adrenal (HPA) axis plays a key role in adaptation to environmental stresses. Parvicellular neurons of the hypothalamic paraventricular nucleus secrete corticotrophin releasing hormone (CRH) and arginine vasopressin (AVP) into pituitary portal system; CRH and AVP stimulate adrenocorticotropic hormone (ACTH) release through specific G-protein-coupled membrane receptors on pituitary corticotrophs, CRHR1 for CRH and V1b for AVP; the adrenal gland cortex secretes glucocorticoids in response to ACTH.The glucocorticoids activate specific receptors in brain and peripheral tissues thereby triggering the necessary metabolic, immune, neuromodulatory, and behavioral changes to resist stress. While importance of CRH, as a key hypothalamic factor of HPA axis regulation in basal and stress conditions in most species, is generally recognized, role of AVP remains to be clarified. This review focuses on the role of AVP in the regulation of stress responsiveness of the HPA axis with emphasis on the effects of aging on vasopressinergic regulation of HPA axis stress responsiveness. Under most of the known stressors, AVP is necessary for acute ACTH secretion but in a context-specific manner. The current data on the AVP role in regulation of HPA responsiveness to chronic stress in adulthood are rather contradictory.The importance of the vasopressinergic regulation of the HPA stress responsiveness is greatest during fetal development, in neonatal period, and in the lactating adult. Aging associated with increased variability in several parameters of HPA function including basal state, responsiveness to stressors, and special testing. Reports on the possible role of the AVP/V1b receptor system in the increase of HPA axis hyperactivity with aging are contradictory and requires further research. Many contradictory results may be due to age and species differences in the HPA function of rodents and primates.

**Keywords: hypothalamic–pituitary–adrenal axis, vasopressin, stress, aging, V1b receptors**

# **INTRODUCTION**

The hypothalamic–pituitary–adrenal (HPA) axis is a key adaptive neuroendocrine system. Regulation of glucocorticoid secretion through adrenocorticotropic hormone (ACTH) is critical to life and essential to maintain the mammalian response to stressor (Pedersen et al.,2001;McEwen,2007; Lupien et al.,2009). The HPA axis consists of the nucleus paraventricularis hypothalami (PVN), which secretes corticotrophin releasing hormone (CRH) and arginine vasopressin (AVP); the pituitary gland that is sensitive to CRH and AVP, which stimulate ACTH release; the adrenal gland cortex, which secretes the glucocorticoids, mainly cortisol in humans and non-human primates, and corticosterone in rodents, and dehydroepiandrosterone (DHEA) in humans and non-human primates in response to interaction with ACTH. DHEA and its sulfate DHEAS are important regulators of glucocorticoid activity (Yen and Laughlin, 1998; Maninger et al., 2009). The glucocorticoids act on specific receptors present in most peripheral tissues and the brain and trigger the metabolic, immune, neuromodulatory, and behavioral changes needed to cope with the impact of the stressors

(Tuckermann et al., 2005; McEwen, 2007;Vegiopoulos and Herzig, 2007). In turn, the glucocorticoids act through the pituitary and limbic structures, especially the hippocampus, in a negative feedback loop to regulate the activity of CRH-producing neurons in the PVN and, thus the entire the HPA axis. The adaptive influence of the HPA axis under stress is realized not only through ACTH and the corticosteroids but also through the hypothalamic neuropeptides CRH and AVP, which are released in the brain, where they are responsible for behavioral and autonomic responses to stress (Bale and Vale, 2004; Roper et al., 2011).

In resting conditions, activity of the HPA axis shows circadian and ultradian changes with pulsatile glucocorticoid secretion that is greater in amplitude during the phase of wakefulness. This leads to higher average levels of glucocorticoids during the day in humans and most other primates and to higher activity of the HPA axis during the night in rodents (Goncharova et al., 2002; Kalsbeek et al., 2010; Russell et al., 2010; Walker et al., 2010). Stress responsiveness of the pituitary– adrenal axis in humans and non-human primates also shows the

circadian rhythms with higher stress response in the afternoon and evening, and lower in the morning (Goncharova, 2006; Goncharova et al., 2006a; De Weerth et al., 2007). Rhythmicity in the HPA axis is essential for the normal functioning of the brain and other glucocorticoid responsive organs (Van Cauter et al., 1997; Lupien et al., 2002; Yoshimura et al., 2007; Goncharova et al., 2008a,b).

While the acute stress activation of the HPA axis is critical for life, chronic exposure to stressors leads to its excessive stimulation and hypercortisolemia. Hypercortisolemia plays a pathophysiological role in the development of a variety of stress-related diseases: psychiatric, reproductive, immune, metabolic, and others. It is a major factor in aging and age-related pathology (Chrousos, 2000; de Kloet et al., 2005, 2008; Perez-Neri et al., 2008). While CRH is widely recognized as a major hypothalamic factor controlling the HPA axis activity in resting conditions and under stress, the role of AVP in regulation of the HPA axis remains to be studied. Almost all current data on the role of AVP in regulation of the HPA axis were obtained in experiments on adult rodents. In the available literature, there are no review articles devoted to the role of AVP in the regulation of HPA axis stress responsiveness in aged subjects. In any event, aging is generally characterized by hyperactivation of the HPA axis, breakdown of the circadian rhythm, progressive decline in DHEA and DHEAS production, and high incidence of stress-dependent diseases, including depression, in which AVP production and the corticosteroids are of great importance (Yen and Laughlin, 1998; de Winter et al., 2003; Dinan and Scott, 2005; Kondratova and Kondratov, 2012). This review will focus on the evaluation of the role of AVP in the regulation of stress responsiveness of the HPA axis with emphasis on HPA axis vasopressinergic regulation during healthy aging in rodents as well as in humans and non-human primates.

#### **MECHANISMS OF HPA AXIS REGULATION CORTICOTROPHIN RELEASING HORMONE**

Corticotrophin releasing hormone is the main physiological regulator of the HPA axis in basal conditions and in response to most acute stressors (Jacobson et al., 2000; Bale and Vale, 2004; Aguilera, 2011). CRH is a peptide consisting of 41 amino acid residues, which was discovered and sequenced by Vale et al. (1983). It is produced in the medial parvicellular neurons of the PVN, which project to the outer zone of the median eminence. There it is released into the portal system of the pituitary, which reaches the anterior lobe. CRH stimulates ACTH secretion through activation of type 1 CRH receptors (CRHR1) in the pituitary corticotrophs. CRHR1 belongs to the CRH family of G-protein-coupled membrane receptors (GPCRs) (Bale and Vale, 2004; Zorrilla and Koob, 2010).

The collection of CRH-producing neurons (CRH neurons) of the PVN is a key center of the central nervous system, integrating the neuroendocrine effects of stress, and a key part of the HPA axis. On the one hand, the CRH neuron is under the regulatory influence of numerous afferent nerve pathways that carry information about the stressor. In addition, it is regulated by glucocorticoids and it is the central link of the autoregulation mechanism in the HPA axis.

#### **Afferent regulation**

The CRH neuron receives projections from ascending catecholaminergic pathways including noradrenergic projectionsfrom the A2 noradrenergic cell group within the nucleus of the solitary tract and the locus ceruleus. It also receives input from areas of the limbic system, notably the bed nucleus of the stria terminalis, the hippocampus, and the amygdala (Herman et al., 1996; Dinan and Scott, 2005). Indirect pathways from the sensory systems in the forebrain through the amygdala and bed nucleus of the stria terminalis may stimulate CRH neurons by activating stimulatory (glutamatergic) or repressing inhibitory (γ-aminobutyric acid – GABA-ergic) interneurons (Herman et al., 1996;Arvat et al., 2002; Aguilera, 2011). There is also significant evidence in relation to the stimulatory effect of the cholinergic system on CRH secretion implemented through direct and indirect projections from both the basal forebrain and the mesopontine tegmentum (Rhodes and Rubin, 1999; Rubin and Rhodes, 2002).

#### **Glucocorticoid feedback mechanism**

Feedback mechanism plays an essential role in the regulation of CRH production and in limiting the stress response. The central sensors of feedback to the HPA axis are two corticosteroid receptors, the high-affinity mineralocorticoid receptor (MR) and the low-affinity glucocorticoid receptors (GR), which are expressed in the brain and on the corticotrophs of the pituitary (Ratka et al., 1989). GRs are distributed throughout the brain, but mostly in hypothalamic neurons and corticotrophs (Jacobson and Sapolsky, 1991; Funder, 1997), whereas MRs are present in the hypothalamus and in the greatest numbers in the hippocampus (Ratka et al., 1989; Funder, 1997). A significant role of direct inhibition of transcription of pro-opiomelanocortin (POMC) and ACTH secretion by glucocorticoids in the regulation of the HPA axis by a feedback mechanism has been previously demonstrated (Autelitano et al., 1989; Dallman et al., 1994). Development of MR antagonists has enabled demonstration of the important role of glucocorticoid feedback at the suprapituitary level, probably from the hippocampus in regulation of the HPA axis both in animals and humans (Wellhoener et al., 2004; Giordano et al., 2005). The presence of GRs in CRH neurons (Liposits et al., 1987), as well as a demonstrated inhibitory effect of intra-PVN glucocorticoid injection on CRH mRNA, and the reduction in CRH promoter activity in reporter gene assays following incubation with glucocorticoids, give reason to believe that a direct inhibitory effect of glucocorticoids on CRH transcription underlies the central mechanisms of feedback regulation of CRH transcription and desensitization of the HPA axis to stress responses (Kovacs et al., 1986; Sawchenko, 1987; Harbuz and Lightman, 1989; Makino et al., 1995; Fenoglio et al., 2004; Aguilera, 2011).

# **ARGININE VASOPRESSIN Distribution of AVP**

While it is generally accepted that CRH is a critical coordinator of HPA axis function in resting conditions and in response to stress (Aguilera, 1998, 2011; Jacobson et al., 2000; Bale and Vale, 2004), the role of AVP in the regulation of the HPA axis remains a subject of active research. Following its identification in 1954, AVP, a nonapeptide, was considered the principal factor in the regulation

of ACTH secretion until the subsequent identification of CRH, which led to the replacement of AVP with CRH as the principle regulator of the HPA axis. Nevertheless, AVP is still considered as an important regulator of the HPA axis (Antoni, 1993; Herman, 1995; Volpi et al., 2004; Aguilera, 2011). In addition, AVP is seen as a hormonal regulator of water homeostasis and has significant effects on vascular tone (Knepper, 1997; Koshimizu et al., 2006). AVP is also believed to regulate circadian rhythms (Kalsbeek et al., 2010; Kondratova and Kondratov, 2012) and numerous central functions such as synaptic transmission, control of body temperature, memory, and behavior (Barberis and Tribollet, 1996; Wersinger et al., 2002).

Knowledge of the functional anatomy, physiology, and pathophysiology of AVP in the regulation of the HPA axis is mainly based on studies carried out on rodents. There are two major vasopressinergic systems in the brain. The first system consists of hypothalamic magnocellular neurons in the PVN and supraoptic nucleus, which project to the neurohypophysis and deliver AVP and oxytocin to the peripheral circulation; it is largely responsible for the peripheral actions of AVP that maintain water homeostasis and blood pressure (Knepper, 1997; Koshimizu et al., 2006). The second system is represented by the CRH-producing parvicellular neurons of the PVN, which also synthesizes AVP. AVP produced by the latter neuron is secreted into the pituitary portal circulation from axon terminals in the external zone of the median eminence; they are involved in regulation of the HPA axis (Antoni, 1993). It has been shown in non-stressed rats that 50% of CRH neurosecretory cells coexpress AVP (Whitnall, 1987). In humans, there is evidence that all CRH neurons in the hypothalamus may contain AVP (Mouri et al., 1993; Dinan and Scott, 2005). In primates, including humans, large numbers of immunoreactive AVP neurons have been demonstrated in the limbic system, notably the bed nucleus of the stria terminalis, the hippocampus and the amygdala, and also in the pituitary intermediate lobe, the granular layers of cerebellar cortex, and the dentate gyrus. Lower levels are found in the medial habenula, adenohypophysis, area postrema, pineal body, subfornical and subcommissural organs (see review Dinan and Scott, 2005). It is thought that the AVP levels in the pituitary portal vascular system is derived from the PVN, but morphological and neurochemical studies suggest that AVP from supraoptic magnocellular AVP-secreting cells also accesses the hypophyseal portal blood (Antoni et al., 1990; Dinan and Scott, 2005). AVP neurons in the medial amygdala and the bed nucleus of the stria terminalis project to the lateral septum and ventral hippocampal sites affecting memory and behavior (Caff et al., 1987; Aguilera, 2011). Finally, AVP is expressed in the suprachiasmatic nucleus (SCN), which is involved in the regulation of circadian rhythms (Rhodes and Rubin, 1999; Arima et al., 2002; Kalsbeek et al., 2010; Kondratova and Kondratov, 2012).

#### **AVP receptors**

The actions of AVP are mediated through interaction with specific plasma membrane receptors of target cells that have been identified and cloned (Jard et al., 1987; Birnbaumer, 2000; Derick et al., 2004). AVP receptors belong to the family of GPCRs,i.e., they transmit a signal to the cell through G-protein (guanyl-nucleotidebinding protein) (Jard et al., 1987; Birnbaumer, 2000; Thibonnier et al., 2002; Derick et al., 2004). AVP receptors have been divided into three major types:V1a,V1b (or V3), and V2 according to their pharmacological and G-protein-coupled properties. The V1a and V1b subtypes are both coupled to Gq and signal via phospholipase C (Jard et al., 1987; Thibonnier et al., 2002; Roper et al., 2011). The V2 receptor subtype is coupled to Gs that signal via adenylate cyclase (Thibonnier et al., 2002;Derick et al., 2004). TheV1a receptor is predominantly found in vascular smooth muscle where it is involved in control of vasoconstrictor effects and blood pressure (Koshimizu et al., 2006;Vantyghem et al., 2011). The renal V2 AVP receptor is responsible for water resorption in collecting ducts of the kidney by promoting the translocation of aquaporin-2 channels to the plasma membrane (Knepper, 1997). TheV1b receptor is primarily located in corticotrophs of the anterior pituitary where it stimulates ACTH release (Antoni, 1993; Derick et al., 2004; Roper et al., 2011). A number of studies utilizing *in situ* hybridization histochemistry and reverse transcription-polymerase chain reaction in rodents have shown thatV1b receptor mRNA is also expressed at a lower level in some brain regions (hippocampus, PVN, olfactory bulb, and amygdala) and a number of peripheral tissues (kidney, pancreas, heart, lung, breast, adrenal medulla, and others) (Grazzini et al., 1996; Roper et al., 2011), where its function is unknown (Roper et al., 2011).

#### **Effects of AVP on the HPA axis function and interaction with CRH**

The ability of both neuropeptides CRH and AVP to stimulate the secretion of ACTH has been demonstrated in humans (DeBold et al., 1984; Salata et al., 1988; Favrod-Coune et al., 1993) and various other species, including non-human primates (Goncharova and Lapin, 2002; Goncharova, 2009a), rats (Vale et al., 1983; Antoni, 1993), mice (Muller et al., 2000; Lolait et al., 2007), horses (Evans et al., 1993), and sheep (McFarlane et al., 1995;Hassan et al., 2003). The relative importance of each seems to vary depending on the species. For example, in humans (DeBold et al., 1984; Favrod-Coune et al., 1993), non-human primates (Goncharova, 2009a), rats (Vale et al., 1983; Antoni, 1993; Serradeil-Le Gal et al., 2005), and horses (Evans et al., 1993), CRH is a much more potent secretagogue than AVP. At the same time, some groups have reported that in the sheep AVP is the more potent secretagogue (Familari et al., 1989; Hassan et al., 2003). In all cases, a marked synergism between CRH and AVP has been observed (Antoni, 1993; Favrod-Coune et al., 1993; Aguilera et al., 1994; Aguilera, 1998; Serradeil-Le Gal et al., 2005). The impact of AVP on ACTH secretion is often regarded as ancillary to CRH as far as AVP alone is a weak ACTH secretagogue but acts synergistically with CRH to facilitate ACTH release both in humans (DeBold et al., 1984; Salata et al., 1988; Favrod-Coune et al., 1993) and rodents (Rivier and Vale, 1983; Antoni, 1993). It has been shown that the corticotroph AVP/V1b and the CRH/CRHR1 signaling pathways converge to increase ACTH secretion (Abou-Samra et al., 1986; Roper et al., 2011). In particular, it has been suggested that AVP plays a role in stimulating the primary nuclear transcripts induced by CRH in pituitary corticotrophs (Antoni, 1993). Studies in recent years suggest that the CRH–AVP synergism may also be conditioned by change in the physical state of corticotroph V1b and CRHR1 receptors (Young et al., 2007; Roper et al., 2011). The two receptors may physically heterodimerize (Young et al., 2007).

# **VASOPRESSINERGIC REGULATION OF STRESS RESPONSIVENESS OF THE HPA AXIS**

#### **AVP IN REGULATION OF HPA AXIS REACTIVITY IN RESPONSE TO ACUTE STRESS**

#### **Acute stress**

Acute immobilization stress has been observed to lead to upregulation of AVP mRNA along with up-regulation of CRH mRNA expression in parvicellular neurons of the PVN (Bartanusz et al., 1993). Moreover, compensatory increase in basal vasopressinergic system in mice lacking the CRHR1 gene was revealed (Muller et al., 2000). All this allows us to consider AVP as an important modulator of the ACTH response to stress, as a compensatory mechanism to maintain the activity of the HPA axis when the CRH/CRHR1 signaling pathway is damaged (Muller et al., 2000). The importance of AVP along with CRH as physiological modulators of ACTH secretion in response to insulin-induced hypoglycemia has been demonstrated in humans (DeBold et al., 1984; Ellis et al., 1990).

The precise role of AVP production in response to stress remains controversial because most studies have involved indirect or correlative measurements of activation of AVP-producing neurons after exposure to stress (DeBold et al., 1984; Ellis et al., 1990; Makara et al., 2004; Lolait et al., 2007). Some of the best evidence for AVP involvement in the response of the HPA axis response to stress is derived from studies that have shown that immunoneutralization of AVP diminish the increase in plasma ACTH produced by various stressors such as restraint, insulininduced hypoglycemia, or injection of bacterial lipopolysaccharide (LPS) (Turnbull et al., 1998; see introduction in Lolait et al., 2007). In addition, studies in which AVP levels have been sampled in pituitary portal blood indicate that AVP may be secreted preferentially over CRH in response to acute stressors such as insulin-induced hypoglycemia (Plotsky et al., 1985; Engler et al., 1989).

Nevertheless, several studies do not support the conclusion of an important role for AVP in the regulation of HPA axis stress responsiveness. Animals in some studies using modern genetic and pharmacological models to create an AVP deficit have, nevertheless shown a normal response of the HPA axis in acute stress, specifically to restraint, resident-intruder, and LPS stressors (Lolait et al., 2007; Chen et al., 2008a; Zelena et al., 2011). For example, mice lacking AVP V1b receptors exhibited normal corticosterone responses to acute physical–psychological stress in the resident-intruder paradigm (Wersinger et al., 2002).

In other studies using genetic and pharmacological models of AVP deficit the pituitary–adrenal responses to acute stress were substantially reduced. Thus, recent studies with the V1b receptor knockout mice demonstrated reduced plasma levels of ACTH and corticosterone or only ACTH in response to insulin-induced hypoglycemia (Lolait et al.,2007),forced-swimming (Tanoue et al., 2004; Stewart et al.,2008), novel environment (Stewart et al.,2008), mild restraint (Stewart et al., 2008), and in response to some of the other stressors (Roper et al., 2011) compared to wild-type mice. The use of SSR149415, the first selective V1b receptor antagonist, at 2 h after treatment significantly suppressed the increase in ACTH after a 5 min forced-swimming stress and decreased

the dramatic elevation in plasma corticosterone concentrations in mice (Serradeil-Le Gal et al., 2005). Acute administration of another antagonist of V1b receptor (Org) also reduced ACTH secretion following both restraint and LPS but did not antagonize the ACTH response to noise (Spiga et al., 2009). Interestingly, the OrgV1b antagonist,while reducing the stress-induced increase in ACTH secretion, had no effect on plasma corticosterone levels (Spiga et al., 2009). Similar results were obtained when another V1b receptor antagonist (SSR149415) was used: the ACTH plasma level was reduced in rats in response to restraint compared to controls, but showed no inhibitory effect on the secretion of corticosterone (Serradeil-Le Gal et al., 2003).

Findings from studies of the role of AVP in responsiveness of the HPA axis in acute stress using V1b receptor antagonists in rodents are consistent with studies of naturally AVP-deficient Brattleboro rats (Zelena et al., 2009; Makara et al., 2012). The ACTH response of homozygous animals to severe stressors, such as those mentioned above, as well as to anaphylactoid stress (hypertonic saline, egg white injection), elevated plus-maze, and novelty, was sharply decreased compared to that of heterozygous littermates. The ACTH response also was decreased in response to volume load, restraint, or aggressive attack in AVP-deficient rats. At the same time, there were no significant differences in the ACTH reaction in response to such stressors as social avoidance, footshock, and ether inhalation. Differences were also observed in the corticosterone response to the three groups of stressors. It did not change in response to the first group of stressors, decreased in parallel with ACTH in AVP-deficient rats in response to the second group of stressors and, like ACTH, did not change, in response to the third group of stressors (Zelena et al., 2009; Makara et al., 2012). Herman et al. (1996) and Zelena et al. (2009) suggest that the brain categorizes stressors and utilizes neural response pathways that vary in accord with their assigned category as well as that AVP increase the ACTH response to stress in a context-specific manner (Zelena et al., 2009).

Age can also influence the importance of AVP for the HPA axis response to acute stress. Reduced ACTH and corticosterone release was observed in AVP-deficient rats in the neonatal period, while adult rats of the same strain demonstrated a normal response of the pituitary–adrenal axis to LPS injection (Zelena et al., 2011).

#### **Basal conditions**

There is a wide range of studies on the role of AVP in the HPA axis regulation not only in response to acute stressors but during basal conditions. Studies using AVP immunoneutralization have not identified a role of AVP in basal ACTH regulation (Ono et al., 1985; Tilders et al., 1985). Studies using genetic models of AVP deficit based on knocking out the corticotroph V1b receptor resulted either in reduced (Tanoue et al., 2004) or normal (Lolait et al., 2007) resting ACTH and corticosterone levels. A selectiveV1b receptor antagonist (Org),which decreased restraintand LPS-induced ACTH secretion, did not affect the resting levels or stress-induced corticosterone responses (Spiga et al., 2009). The resting pituitary ACTH concentration of homozygous Brattleboro rats appeared to be normal (Lolait et al., 1986; Zelena et al., 2009), although there was a trend toward elevated corticosterone

at decapitation (Zelena et al., 2009) and also there are publications of impaired basal ACTH and corticosterone plasma levels (Burgess and Balment, 1992).

#### **AVP REGULATION OF HPA AXIS RESPONSIVITY TO CHRONIC STRESS**

Regulation of V1b receptors in the pituitary gland appears to play a major role in corticotroph responses to chronic stressors, as a good correlation has been established between the concentration of the receptor and the secretion of ACTH by the pituitary gland (Aguilera, 1994; Aguilera et al., 1994). Furthermore, increased expression of AVP mRNA in parvicellular neurons of the PVN has been demonstrated along with a positive correlation between the density of corticotroph V1b receptors and ACTH release in response to chronic stress (Aguilera, 1994).

Numerous studies performed in rats have revealed a shift of hypothalamic CRH/AVP signal in favor of AVP in some chronic stress states. This is manifested by enhanced AVP synthesis in CRH-producing cells in parvicellular neurons of the PVN after repeated restraint (Whitnall, 1989; De Goeij et al., 1992; Bartanusz et al., 1993; Aguilera, 1994), foot shock (Sawchenko et al., 1993), intraperitoneal injection of hypertonic saline (Ma and Aguilera, 1999), and AVP storage in the CRH-containing nerve terminals in the external zone of the median eminence (De Goeij et al., 1992). An increased proportion of hypothalamic neurons coexpressing AVP has also been demonstrated with such stressors (De Goeij et al., 1992). At the same time, CRH production remained unchanged or decreased (De Goeij et al., 1992). CRH expression increased only in cases of chronic stress when circulating ACTH and glucocorticoid levels increased, for example, in response to stress caused by repeated injections of intraperitoneal hypertonic saline or footshock (Imaki et al., 1991; Ma and Aguilera, 1999).

The results of direct manipulation of the HPA axis reinforce the idea that chronic stress induces a shift of the hypothalamic CRH/AVP signal in favor of AVP. Prolonged administration of CRH by osmotic minipumps led to a reduction in CRH receptor number in the pituitary corticotrophs of control rats (Tizabi and Aguilera, 1992), which was enhanced by the simultaneous infusion of AVP. At the same time adrenalectomized Brattleboro rats lacking hypothalamic AVP, showed only a minimal reduction in CRH receptor number, which was increased by AVP infusion in the period after adrenalectomy (Tizabi and Aguilera, 1992).

Repeated restraint stress or repeated hypertonic saline injections were found to produce sustained increases in the expression of V1b receptor mRNA in the pituitary (Rabadan-Diehl et al., 1995). This suggests an up-regulation of AVP receptors in situations of chronic stress, which may explain the preserved enhanced response to novel superimposed acute stress exposure (so-called heterotypic stressor) in these animal models. Thus, the study using V1b receptor knockout mice demonstrated the importance of AVP and V1b receptor for a normal response of the HPA axis to heterotypic stressor (hypoglycemia) in chronic stress induced by repeated restraint (Tanoue et al., 2004; Lolait et al., 2007). In addition, the plasma ACTH response to repeated mild restraint,forcedswimming, and novel environments in V1b receptor knockout mice was decreased in comparison with wild-type mice (Stewart et al., 2008; Roper et al., 2011).

The sensitivity of CRH and AVP transcription to glucocorticoid feedback is markedly different (Bilezijian et al., 1987). CRH mRNA and CRH receptor mRNA levels are reduced by elevated glucocorticoids, whereas V1b receptor mRNA levels and coupling of the receptor to phospholipase C are stimulated by glucocorticoids, effects which may contribute to the refractoriness of AVPstimulated ACTH secretion to glucocorticoid feedback (Rabadan-Diehl and Aguilera, 1998; Aguilera and Rabadan-Diehl, 2000). This suggest that vasopressinergic regulation of the HPA axis is critical for sustained corticotroph responsiveness in the presence of high circulating glucocorticoid levels during chronic stress (Rabadan-Diehl and Aguilera, 1998; Aguilera and Rabadan-Diehl, 2000).

All these findings support the proposal that CRH plays a predominantly permissive role in the HPA axis regulation in conditions of chronic stress, but AVP is a dynamic modulator of ACTH release (Plotsky, 1991; Dinan and Scott, 2005).

Some recent studies of chronic stress in naturally AVP-deficient Brattleboro rats, suggest that the role of V1b and AVP in adaptation of the HPA axis to chronic stress may not be as convincing as first thought. Experiments on Brattleboro rats and control heterozygous littermates utilized three different chronic stress models: repeated restraint to produce physical–psychological stress (Zelena et al., 2004; Makara et al., 2012); repeated short periods of morphine withdrawal (Domokos et al., 2008; Makara et al., 2012); streptozotocin induced diabetes (Zelena et al., 2006; Makara et al., 2012). The changes characteristic to chronic stress, including body weight reduction, involution of the thymus, adrenal gland hypertrophy, and increases in basal POMC mRNA and plasma corticosterone, developed in all three models, but no difference was observed between the genotypes. Both male AVP-deficient Brattleboro rats and their heterozygous littermates exposed to chronic unpredictable stress, exhibited elevated levels of POMC mRNA in the anterior lobe of the pituitary and plasma ACTH along with a less significant increase in plasma corticosterone (Varga et al., 2011). In addition, the ACTH and corticosterone response to repeated shaker stress in V1b receptor knockout mice is not different from that seen in wild-type mice (Roper et al., 2011). There are also studies using pharmacological models of AVP-deficiency that have shown adequate HPA axis responses to chronic stress in the absence of AVP (Chen et al., 2008b). These data suggest that AVP is not so important for the development of the chronic stress response.

Perhaps the contradictory results can be explained in terms of age and species differences in the HPA function of rodents and primates, including humans. Some evidence suggests that AVP is the main regulator of the HPA axis during the perinatal period (Zelena et al., 2008; Makara et al., 2012). And in some key aspects of HPA function non-human primates are a more appropriate animal model for human than rodents. For example, the adrenal cortex of monkeys, in contrast to rodents, produces DHEA and DHEAS (Goncharova, 1997; Kemnitz et al., 2000; Conley et al., 2004), which are involved in stress response of the HPA axis and may regulate the release of AVP (Deuster et al., 2005). It is apparent that we need more in-depth research on the role of AVP in the HPA axis regulation in response to different types of chronic stress on various rodent models, including both wild-type and

homozygous AVP-deficient Brattleboro mutant strains, as well as V1b receptor knockout rodents. It should also investigate in detail the function of the HPA axis in AVP-deficient lines in basal conditions using various functional tests to get a solid baseline data that are comprehensively characterize the artificial models. Extensive basic research is required on the effects of V1b receptors antagonists and agonists on the HPA axis function in resting and stress conditions in primates.

## **DEVELOPMENTAL SPECIALTIES OF VASOPRESSINERGIC HPA AXIS REGULATION**

#### **FETAL DEVELOPMENT**

Extensive clinical and experimental data indicate variation in the importance of AVP in regulation of the HPA axis, including its stress responsiveness, at different stages of ontogeny and under different physiological conditions. The role of AVP in regulation of the HPA axis in the fetus has been described only in a few studies (Carey et al., 2007, 2009). In fetal sheep late in gestation, secretion of ACTH by the pituitary gland increases in response to stimulation with AVP. This effect is associated with an increase in the generation of inositol 1,4,5-trisphosphate, a second messenger of activated V1b receptors (Carey et al., 2007). The same study also found that hypothalamo-pituitary disconnection significantly impaired this ontogenetic development and prevents the developmental increase in fetal plasma cortisol concentrations in late gestation (Carey et al., 2007). Another study confirmed the hypothesis that cortisol is an important modulator of pituitary responsiveness to AVP in the late gestation sheep fetus (Carey et al., 2009).

#### **NEONATAL PERIOD**

Developmental studies have shown that neonatal rodents (Yi and Baram, 1994; Dent et al., 1999; Schmidt et al., 2009; Zelena et al., 2011) and humans (Gunnar and Donzella, 2002) show a reduced adrenocortical response to stress. Thus, this period is called the stress-hyporesponsive period (Yi and Baram, 1994; Dent et al., 1999; Schmidt et al.,2009). In some studies the reduced response of the pituitary–adrenal axis to stressors was not associated with significant change in CRH gene transcription (Yi and Baram, 1994) or even mentioned its decrease in response to certain stressors, such as endotoxin (Dent et al., 1999). The use of conditional knockout mice with a deletion of the GR at the pituitary (GRPOMCCre) showed that during the neonatal period expression of CRH mRNA in parvicellular neurons of the PVN is reduced while expression of AVP mRNA is increased (Schmidt et al., 2009). Similar data were obtained in 10-day-old pups of AVP-deficient Brattleboro rats, which showed significantly lower resting CRH mRNA levels in the PVN and have revealed that AVP-deficiency abolishes the ACTH stress response in perinatal rats compared to control littermates (Zelena et al., 2008). Recent studies by this research group using the V1b antagonist, SSR149415, confirmed a primary role for AVP in ACTH stress responsiveness of control rats during the neonatal period (Zelena et al., 2011; Makara et al., 2012). Apparently, the shift in the ratio CRH/AVP in favor AVP (Whitnall, 1989; Dinan and Scott, 2005) can take place not only in adult individuals under chronic stress or acute stress induced by a severe stressor, as noted in Section "Vasopressinergic Regulation of Stress

Responsiveness of the HPA Axis" above but also in the neonatal period.

#### **LATE PREGNANCY, LACTATION**

Such important physiological conditions of the organism as pregnancy and lactation are of considerable interest in terms of vasopressinergic regulation of stress reactivity of the HPA axis. It is well known that the HPA axis response to stressors is markedly attenuated in late pregnancy and during lactation period in women (Altemus et al., 1995; de Weerth and Buitelaar, 2005) and female rodents (Fischer et al., 1995; Neumann et al., 1998; Walker et al., 2001; Ma et al., 2005; Brunton et al., 2008). The precise mechanisms of maternal HPA axis hyporesponsiveness in these two states are different (see review Brunton et al., 2008), but both states are characterized by marked changes in AVP secretion from parvicellular neurons of the PVN and subsequent change in the secretion of ACTH in response to stress. The rat's response to stress exposure late in pregnancy is associated with reduced AVP expression (Ma et al., 2005) or reduced expression of both AVP and CRH (da Costa et al., 1996; Douglas et al., 2003). On the contrary, AVP mRNA expression in parvicellular neurons of the PVN in postpartum females exposed to restraint stress during lactation was greater than that of virgin females (Fischer et al., 1995; Walker et al., 2001). At the same time, the expression of CRH mRNA in the PVN significantly decreased during lactation (Fischer et al., 1995; Shanks et al., 1999; Walker et al., 2001). It suggests that AVP may play an important role in the stimulation of ACTH secretion during lactation in rats (Walker et al., 2001). Thus, as in the fetus and neonate, AVP performs an important role in regulation of the HPA axis stress reactivity with a shift in the CRH/AVP ratio in favor of AVP in pregnant and lactating females.

Gestation and early postnatal periods, as we know, are the most vulnerable periods of ontogenesis in terms of programing disruptions of the HPA axis for adulthood (Schmidt et al., 2009; Long et al., 2010; Renard et al., 2010) and aging (Goncharova, 2009b; Solas et al., 2010). Reduced HPA axis response to stressors during late pregnancy, lactation, and the neonatal period, apparently, is an adaptive response of the body that provides protection of the fetus and newborn from an excess of toxic glucocorticoids and that can also have a programing effect on the development of the HPA axis and other physiological systems in later life.

# **AVP IN REGULATION OF HPA AXIS STRESS RESPONSIVENESS DURING AGING**

#### **ACTIVITY OF THE HPA AXIS DURING AGING**

It is generally accepted that mainly hyperactivation of the HPA axis occurs during aging. In small laboratory animals hyperactivation of the HPA axis in the aging process develops at all levels of the system. Thus, aged rodents demonstrated elevated resting ACTH and corticosterone plasma levels (Sapolsky et al., 1986; Tizabi et al., 1992; Hauger et al., 1994; Sapolsky, 1999; Meijer et al., 2005; Lo et al., 2006), an increased release of resting hypothalamic CRH (Scaccianoce et al., 1990; Hauger et al., 1994; Sapolsky, 1999), a decreased sensitivity of the adrenal glands to ACTH (Tang and Phillips, 1978; Sapolsky et al., 1986), of the anterior pituitary to CRH (Hylka et al., 1984; Sapolsky et al., 1986; Hauger et al., 1994; Sapolsky, 1999), of the pituitary, hippocampus, and

hypothalamus to the level of circulating glucocorticoids (Sapolsky et al., 1986; Tizabi et al., 1992; Hatzinger et al., 2000; Revskoy and Redei, 2000). In old rodents, hyperactivity of the HPA axis is likely to reflect loss of resiliency and reduced sensitivity to negative glucocorticoid feedback, which mainly reflects hippocampal receptor damage (Sapolsky et al., 1986; Seeman and Robbins, 1994; Pedersen et al., 2001; Giordano et al., 2005). As a result, older animals tend to show a more prolonged elevation of corticosterone in response to stress exposure (Sapolsky et al., 1986; Seeman and Robbins, 1994; Sapolsky, 1999). Marked disturbances have been shown in the reaction of aged rat adrenals to stress.

These have been expressed mainly as increases in the maximum peak values of plasma corticosterone rise and the prolongation of elevated corticosterone secretion (Sapolsky et al., 1986; Seeman and Robbins, 1994; Sapolsky, 1999; Lo et al., 2006). Some studies, however, have shown individual and species differences ranging from unchanged and reduced basal and stress responses in aged rats compared to young rats (Cizza et al., 1994; Hauger et al., 1994; Lightman et al., 2000).

Hyperactivation of the HPA axis during aging in humans and non-human primates is generally associated with elevated plasma levels of ACTH and cortisol in basal conditions (Sapolsky and Altman, 1991; Guazzo et al., 1996; Lupien et al., 1999, 2007). Decrease in HPA axis sensitivity to glucocorticoid regulation by a feedback mechanism with age has also been observed in humans (Born et al., 1995; Heuser et al., 2000; Giordano et al., 2005) and nonhuman primates (Sapolsky and Altman, 1991; Brooke et al., 1994; Goncharova et al., 2000; Gust et al., 2000; Goncharova and Lapin, 2002, 2004). However, in contrast to rodents there was an increase in response of the pituitary–adrenal axis to CRH and ACTH injection in humans (Born et al., 1995; Kudielka et al., 1999; Ferrari et al., 2001) and monkeys (Goncharova and Lapin, 2002, 2004; Goncharova et al., 2002), as was a higher responsiveness of the pituitary–adrenal axis to combined administration of CRH and AVP (Born et al., 1995). With healthy aging basal activity of CRHproducing neurons in the hypothalamic PVN slightly increased in the human (see review Swaab et al., 2005).

It should be noted that aging is associated with an increase in the variability of disturbances in the HPA axis both in primates and rodents but to a greater extent in humans and non-human primates than in rodents. Many authors have failed to observe significant age-related changes in basal plasma levels of ACTH and cortisol in humans (Ohashi et al., 1986; Waltman et al., 1991; Goncharova, 1997; Goncharova et al., 2002) and monkeys (Moore et al., 1979; Goncharova and Lapin, 2000, 2002; Goncharova et al., 2000; Kemnitz et al., 2000). Some even have observed decreased basal cortisol levels in humans (Blichert-Toft, 1978; Sherman et al., 1985) and monkeys (Gust et al., 2000). Other researchers have found no marked age-related changes in sensitivity of the HPA axis to inhibition by glucocorticoid feedback (Blichert-Toft, 1978; Waltman et al., 1991; Huizenga et al., 1998). There was no change in the reaction of the adrenal cortex to injection of CRH, ACTH, or insulin in humans (Blichert-Toft, 1978; Ohashi et al., 1986) or non-human primates (Goncharova and Lapin, 2000; Goncharova et al., 2000).

Some researchers have found higher responses to acute psycho-emotional stress in physically untrained aged individuals (Traustadóttir et al., 2005). Older men responded to psychological stress with increased cortisol levels and more intense cardiovascular responses compared to women of similar age (Traustadottir et al., 2003). Comparative analysis of cortisol response to various pharmacological stimuli including dexamethasone, CRH, naloxone,ACTH,insulin, epinephrine,metapyrone, physostigmine, and hypertonic saline, as well as to psychological stressors in healthy volunteers, young and old age through the analysis of the data published by different authors in 45 articles, allowed the authors (Otte et al., 2005) to come to a conclusion about a higher HPA axis response in older subjects compared to younger persons. The effect of age on the release of cortisol was significantly greater in women.

At the same time, no age-related differences in ACTH and cortisol levels in response to acute psychosocial stress have been reported in postmenopausal women compared to young women (Kudielka et al., 1999). The response to acute psychological stress in the elderly who were aerobically fit did not differ from that of young adults (Traustadóttir et al., 2005). Marked difference in function of the HPA axis in response to surgical stress is not seen between young persons and centenarians (Kudoh et al., 2001), even though clinical observations indicate that the mortality rate of elderly patients, undergoing surgery is much higher than that of young patients (Blichert-Toft, 1978). The higher rate may be due to reduced adaptive ability of the aging body and deterioration of HPA axis function.

Experiments on healthy non-human primates (*Macaca mulatta*) revealed that age-related dysfunctions of the HPA axis are associated with adaptive behavior of animals (Goncharova et al., 2010). Monkeys with depression-like behavior show agerelated changes in the HPA axis function that are accompanied by maximal absolute and relative hypercortisolemia (high cortisol/DHEAS molar ratio) under resting conditions, as well as by a significantly greater increase in plasma cortisol levels with acute psycho-emotional stress. Young aggressive monkeys, in comparison with young monkeys of other behavior groups, demonstrated the highest plasma levels of DHEAS and the lowest molar ratios of cortisol to DHEAS. Such differences were not exhibited by old monkeys with aggressive behavior. Minimal age-related changes in the HPA axis have been observed in monkeys with average (standard) behavior.

It should be noted that a striking difference in physiology of the HPA axis between primates and rodents is the secretion of large amounts of DHEA and DHEAS by the primate adrenal cortex, which undergoes a decline with age (Orentreich et al., 1992; Goncharova, 1997; Yen and Laughlin, 1998; Ferrari et al., 2001; Goncharova and Lapin, 2002; Conley et al., 2004; Heaney et al., 2012). DHEA and DHEAS serve as precursors to androgens and estrogens (Yen and Laughlin, 1998; Labrie et al., 2001; Mellon and Vaudry, 2001). In addition, they may act as functional antagonists to the effects of glucocorticoids (Yen and Laughlin, 1998; Maninger et al., 2009). In the brain, they are neuromodulators of neuronal receptors, such as GABA-A and *N*-methyl-d-aspartate (NMDA) receptors (Baulieu, 1998; Yen and Laughlin, 1998; Mellon and Vaudry, 2001), which are part of the neurons of afferents nerves regulating – among others – the PVN activity (Arvat et al., 2002; Deuster et al., 2005).

#### **CIRCADIAN RHYTHMS**

Significant variability has been found with respect to age-related changes in circadian rhythms of ACTH and glucocorticoid secretion both in humans and animals. While a number of studies have failed to show marked circadian disorders of cortisol and ACTH in humans and non-human primates (Chambers et al., 1982; Waltman et al., 1991; Ceresini et al., 2000) and some strains of rats (Honma et al., 1996), an increasing number of studies have shown disruption of circadian rhythms in HPA axis activity with aging in both humans and animals. Many authors have reported ageassociated increases in basal ACTH and cortisol concentrations in terms of basal evening levels and nadirs in humans (Ferrari et al., 2001; Giordano et al., 2005; Hofman and Swaab, 2006) and non-human primates (Gust et al., 2000; Khavinson et al., 2001; Goncharova et al., 2002, 2006b; Zhdanova et al., 2011), and fall of the levels of ACTH and corticosterone in the evening and increase during the circadian trough in rodents (Hauger et al., 1994; Lightman et al., 2000; Froy, 2011). A damaged daily rhythm may involve the SCN, as fetal transplants containing SCN can restore the circadian rhythm in old Sprague-Dawley rats (Cai et al., 1997).

The master circadian clock is located in the SCN of the hypothalamus. Its activity is synchronized with the natural day–night cycle, and it coordinates the circadian rhythms of the body, including the HPA axis rhythm (Reppert and Weaver, 2002; Kalsbeek et al., 2010; Walker et al., 2010; Kondratova and Kondratov, 2012). A characteristic feature of the SCN in many species, including humans and other primates is a subpopulation of AVP-containing neurons (Kalsbeek et al., 2010). The SCN regulates HPA axis activity through the PVN via AVP, which serves as a neurotransmitter in the SCN afferents to the hypothalamic PVN (Kalsbeek et al., 2010). The regulatory function of these afferents is confirmed by experiments with microinjections of AVP-cytotoxic monoclonal antibodies into the SCN, which produce a significant reduction of immunoreactive AVP and AVP mRNA expression there as well as in parvicellular neurons of the PVN (Gomez et al., 1997). Study of post-mortem human brains revealed a pronounced circadian variation in the activity of AVP-containing neurons in the SCN (Kalsbeek et al., 2010). Animal experiments have shown an important role for SCN-derived AVP in the control of neuroendocrine day/night rhythms of the HPA axis (Kalsbeek et al., 2010). The number of AVP-expressing neurons in the SCN declines with age in humans (Swaab et al., 1985), non-human primates (Cayetanot et al., 2005), and rats (Roozendaal et al., 1987). Decreased AVPproducing neurons in the SCN in aged humans and animals are accompanied by deterioration of the circadian rhythms of HPA axis function (Kalsbeek et al., 2010).

#### **AVP REGULATION OF STRESS RESPONSIVENESS OF THE HPA AXIS WITH AGING**

Age-related changes in AVP production vary depending on location in the brain. While secretion of AVP decreases with aging in the SCN of animals and humans (Cai et al., 1997; Cayetanot et al., 2005; Hofman and Swaab, 2006;Kalsbeek et al., 2010), its secretion in the PVN seems to increase. This is indicated by immunocytochemical studies of human post-mortem brains that have revealed increased amounts of AVP in AVP-expressing cells and in the size of AVP-containing neurons in the PVN of aged humans (Lucassen

et al., 1993; Prelevic and Jacobs, 1997; Ishunina and Swaab, 1999). In addition, an age-dependent change in colocalization of CRH and AVP in favor of AVP in the human PVN has been identified (Raadsheer et al., 1993). Moreover, old rats showed signs of hyperactive AVP-expressing neurons (Terwell et al., 1992; Bazhanova et al., 2000). Other studies, however, provide indications of a modest activation of CRH neurons with aging in healthy persons (see review Swaab et al., 2005).

In the literature there are several studies that investigated the role of the AVP/V1b receptor system in the regulation of stress responsiveness of the HPA axis with aging. Most of these studies have been performed in experiments on rodents. Features of vasopressinergic regulation of stress responsiveness of the HPA axis during aging has been studied significantly less in humans and non-human primates.

#### **Rodents**

On the other hand, the idea that secretion of AVP in the PVN can increase with aging is evidenced by the results of studies on the role of AVP in the regulation of stress reactivity of the HPA axis, performed mainly on rodents. Thus, some reports in the literature describe work with rats of the Fischer-344/N strain, in which aging is associated with a progressive decline in hypothalamic CRH production associated with increased production of AVP (Cizza et al., 1994). In this case, basal plasma ACTH levels were similar across age groups. Injection of exogenous rat CRH elicited significantly greater ACTH and corticosterone responses in aged rats, which was consistent with the observation of hypothalamic CRH deficiency. CRH mRNA levels in the PVN, CRH content, and *in vitro* secretion of CRH by whole explanted hypothalamus showed progressive and significant reductions with age, whereas the steady state levels of AVP mRNA significantly increased with age (Cizza et al., 1994). Perhaps the importance of PVN-derived AVP for HPA axis regulation increases with advancing age, while the importance of CRH is reduced.

Subsequent studies have confirmed the important role of AVP expressed in parvicellular neurons of the PVN in the regulation of ACTH secretion in response to stimulus. Application of the combined DEX/CRH test to old male Wistar rats resulted in excessive release of ACTH and corticosterone as compared to young rats. Administration of a V1b receptor antagonist between the dexamethasone and CRH injections blocked the effect (Hatzinger et al., 2000). In DEX-pretreated aged rats the authors detected increased numbers of AVP mRNA-expressing neurons in the PVN; this was in combination with increased numbers and activity of CRH mRNA-expressing neurons. This study provided the first direct evidence of the involvement of excessive activity of AVPexpressing neurons in age-dependent hyperactivity of the HPA axis. The results of this study suggest an important role of AVP in the mechanism of increased HPA axis activity in aging rats, possibly secondary to defective function of corticosteroid receptors (Hatzinger et al., 2000).

Another study has demonstrated an increase in the content of AVP within the PVN in basal conditions. The approximately twofold increase correlated with an increase of ACTH and corticosterone basal plasma levels in old rats (Keck et al., 2000). However, in response to acute stressor, forced-swimming, the responses of AVP, ACTH, and cortisol were significantly lower than in young animals (Keck et al., 2000). Some authors report that aging in rats is likely to be associated with reduced production of AVP and hypersecretion of CRH in the hypothalamus and subsequent down-regulation of corticotroph CRH receptor (Hauger et al., 1994).

Some of the contradictory results reported in the above papers may by attributable to individual differences in HPA axis function both in resting and stress conditions. Thus, in the study of Meijer et al. (2005) young and aged rats had previously been classified as inferior and superior learners on the basis of their performance in a water maze test. The authors tested for a correlation between the production of the hypothalamic neuropeptides, CRH and AVP, and their status as inferior and superior learners. They found that AVP expression in parvicellular neurons of the PVN was virtually unchanged with aging in animals of either group. CRH expression, however, decreased significantly in old animals compared to young animals in the superior learning group, which produced a change in the ratio of CRH/AVP expression in favor of AVP. By contrast, the ratio of CRH/AVP expression in inferior learners was virtually unchanged with aging. Apparently, there are marked individual differences in rats of the same strain, as well as differences between strains in aging of the brain in general and aging of the HPA axis in particular. This conclusion is supported by the results of studies that demonstrate differences in the functioning of the HPA axis in adult rats of the same species with high and low trait anxiety (Keck et al., 2002). Using surgical microdialysis *in vivo*, the investigators showed that the content of AVP in the PVN is significantly higher in rats with increased anxiety (Keck et al., 2002).

Thus, reports on the possible role of the AVP/V1b receptor system in the increase of HPA axis hyperactivity with aging in rodents are contradictory, suggesting (a) the possible increase of the AVP production in the PVN for aged rodents, (b) the possible lack of any age changes in the AVP secretion in basal and stress conditions, and (c) the possible reduced basal AVP production along with the CRH hypersecretion.

#### **Humans and non-human primates**

Only a few studies have addressed age-related changes in vasopressinergic regulation of HPA axis stress reactivity in humans and non-human primates (Raskind et al., 1995; Deuster et al., 2005; Goncharova, 2009a). Most studies have focused on age differences in response of the pituitary–adrenal axis to stress. The cross-species findings are quite variable with regard to responsiveness of the HPA axis in old age. They depend not only on the nature, intensity and duration of exposure of the stressor, the genetic characteristics of the individual, and the environment in old age, but also on the influence of genetic and environmental factors in the early stages of development, particularly the fetal and neonatal periods. Environmental stress in early life may program the future development, including disruptions of HPA axis stress responsiveness in old age (Pivina et al., 2007; Goncharova, 2009b; Solas et al., 2010).

It has been noted that the healthy elderly exhibit a more pronounced adrenocortical response to hypertonic saline infusion than young people (Raskind et al., 1995). The increase in levels of AVP and plasma osmolarity was similar in representatives of both age groups, but no rise in the levels of ACTH was observed in either older or young subjects. The authors, therefore, suggested that the increased cortisol secretion in response to stimulation with the hypertonic solution was due to age-related changes in the adrenocortical level of the HPA axis (Raskind et al., 1995). Perhaps, stimulation of cortisol secretion was influenced by increased levels of AVP in the general circulation. This effect could be a direct AVP effect on the adrenal gland as the pituitary contains predominantly V1b receptor, but the adrenal cortex has V1a receptor (Grazzini et al., 1996; Vezzosi et al., 2007; Roper et al., 2011). Thus, there is evidence that AVP potentiates cortisol secretion *in vitro* by normal human adrenal cortex and some cortisol-producing adrenocortical tumors or hyperplasia through the activation of V1a receptors (Perraudin et al., 1993, 1995).

Interestingly, chronic administration of DHEA to healthy young men significantly increased exercise-induced release of AVP along with ACTH and cortisol secretion (Deuster et al., 2005). Perhaps, the stimulating effect of DHEA on pituitary–adrenal axis stress responsiveness is due to its action as antagonist of GABA-A receptors restrained the HPA axis and the stimulating effect of DHEA on NMDA receptors, involved in AVP release (Rossi and Chen, 2002; Deuster et al., 2005). In this regard, DHEA deficiency during aging in humans and non-human primates may play an important role in development of the age-related disturbances of the HPA axis and, in particular, of its stress responsiveness (Kudielka et al., 1998; Goncharova et al., 2000).

Recent experiments on non-human primates (Goncharova, 2006, 2009a; Goncharova et al., 2006a, 2008b), had shown that stress responsiveness of the HPA axis depends to a large extent on the time of day when stimulus is applied as well as on initial sensitivity of the HPA axis. It was shown that the circadian rhythm of ACTH and cortisol level is evident not only in basal conditions but also in response to acute psycho-emotional stress. Young (6–8 years old) female rhesus monkeys showed a much higher increase in ACTH and cortisol levels in response to a 2 h restraint stress imposed at 15:00 h than to the same stress imposed at 09:00 h. This difference attenuates with aging. Old (20–27 years old) female animals showed much lower responsiveness of the HPA axis to the afternoon stress and a tendency toward higher responsiveness to the morning stress. A similar circadian rhythm in responsiveness of the HPA axis to mild acute stress was observed in young pregnant women (De Weerth et al., 2007). The timedependent responsiveness of the HPA axis to acute immobilization stress was observed in young male *Papio hamadryas* (Chirkov et al., 1987) and young female Wistar rats (Pivina et al., 2007). Moreover, senescent rats demonstrated flattening of circadian rhythm in stress reactivity of the HPA axis (Pivina et al., 2007).

The attempt to understand the mechanism of decreasing stress reactivity of the pituitary–adrenal axis with aging led to a series of experiments with the administration of a standard dose of CRH and AVP (1µg/kg b.w., intravenously) to young and old female rhesus monkeys at different times of day (09:00 and 15:00 h). The response of the pituitary–adrenal axis to the injection of CRH revealed a circadian rhythm with a more pronounced response in the afternoon; this rhythm was not affected by aging (Goncharova, 2009a). Apparently, the circadian rhythm in the secretion of CRH underlies the circadian rhythm in the stress reactivity of the pituitary–adrenal axis of the HPA axis. At the same time, the rise of ACTH and cortisol levels in response to AVP injection did not show a dependence on the time of the day in either young or old animals. A significant age-related difference in the rise of ACTH and cortisol in response to AVP injection was, however, found in the response to AVP administered in the afternoon; lower values were seen in older animals (Goncharova, 2009a). It appears that age differences in response of the anterior pituitary to AVP may underlie age-related differences in stress responsiveness of the HPA axis in primates in the afternoon. Perhaps age differences in response of the anterior pituitary to AVP underlie age-related differences in stress responsiveness of the HPA axis in primates in the afternoon.

#### **CONCLUSION**


#### **REFERENCES**


of the hypothalamic-pituitaryadrenal axis: implications for stress adaptation. *Regul. Pept.* 96, 23–29.

	- (b) For the most experiments on rodents, the hyperactivation of the HPA axis was observed at all levels of the HPA organization with aging under the basal and stress conditions.
	- (c) With healthy aging of non-human primates and humans the HPA hyperactivation is usually associated with the increase of the cortisol level in the evening time, the slight changes in the regulation of the HPA axis by glucocorticoid feedback with relative hypercortisolemia due to the decline in secretion of the adrenal antagonists of cortisol – DHEA and DHEAS.
	- (a) the possible increase of the AVP production in the PVN for aged rodents and humans,
	- (b) the possible lack of any age changes in the AVP secretion in basal and stress conditions,
	- (c) the possible reduced basal AVP production along with the CRH hypersecretion.

nucleus in vitro. *Endocrinology* 143, 4165–4171.


responsiveness in humans. *Eur. J. Endocrinol.* 132, 705–711.


of diurnal changes in hormones with sexual behavior and age in male rhesus macaques. *Neurobiol. Aging* 3, 37–42.


to subordinate status. *Endocrinology* 131, 847–853.


relevance. *Annu. Rev. Med.* 48, 231–240.


and Lapin, B. A. (2008b). Circadian and age-related changes in stress responsiveness of the adrenal cortex and erythrocyte antioxidant enzymes in female rhesus monkeys. *J. Med. Primatol.* 37, 229–238.


rats to release ACTH and corticosterone in response to CRF administration. *Proc. Soc. Exp. Biol. Med.* 175, 1–3.


mechanism: regulation of aquaporin water channels by vasopressin.*Am. J. Physiol.* 272, F3–F12.


(CRF). *Histochem. Cell Biol.* 87, 407–412.


vasopressin in the paraventricular nucleus of human hypothalamus. *Neuroendocrinology* 57, 34–39.


Vaudry, H. (1993). Vasopressin stimulates cortisol secretion from human adrenocortical tissue through activation of V1 receptors. *J. Clin. Endocrinol. Metab.* 76, 1522–1528.


with age among wild baboons. *Biol. Psychiatry* 30, 1008–1013.


in man. *J. Clin. Endocrinol. Metab.* 61, 439–443.


behavior in male mice. *Mol. Psychiatry* 7, 975–984.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 30 September 2012; accepted: 22 February 2013; published online: 12 March 2013.*

*Citation: Goncharova ND (2013) Stress responsiveness of the hypothalamic–pituitary– adrenal axis: age-related features of the vasopressinergic regulation. Front. Endocrinol. 4:26. doi: 10.3389/fendo.2013.00026*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Goncharova. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *Kazunori Kageyama\**

Department of Endocrinology and Metabolism, Hirosaki University Graduate School of Medicine, Hirosaki, Aomori, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

David Lovejoy, University of Toronto, Canada Jae Young Seong, Korea University, South Korea

#### *\*Correspondence:*

Kazunori Kageyama, Department of Endocrinology and Metabolism, Hirosaki University Graduate School of Medicine, 5 Zaifu-cho, Hirosaki, Aomori 036-8562, Japan. e-mail: kkageyama@hkg.odn.ne.jp

While stress activates the hypothalamic–pituitary–adrenal (HPA) axis, it suppresses the hypothalamic–pituitary–gonadal (HPG) axis. Corticotropin-releasing factor (CRF) is a major regulatory peptide in the HPA axis during stress. Urocortin 1 (Ucn1), a member of the CRF family of peptides, has a variety of physiological functions and both CRF and Ucn1 contribute to the stress response via G protein-coupled seven transmembrane receptors. Ucn2 and Ucn3, which belong to a separate paralogous lineage from CRF, are highly selective for the CRF type 2 receptor (CRF2 receptor). The HPA and HPG axes interact with each other, and gonadal function and reproduction are suppressed in response to various stressors. In this review, we focus on the regulation of gonadotropins by CRF and Ucn2 in pituitary gonadotrophs and of gonadotropin-releasing hormone (GnRH) via CRF receptors in the hypothalamus. In corticotrophs, stress-induced increases in CRF stimulate Ucn2 production, which leads to the inhibition of gonadotropin secretion via the CRF2 receptor in the pituitary. GnRH in the hypothalamus is regulated by a variety of stress conditions. CRF is also involved in the suppression of the HPG axis, especially the GnRH pulse generator, via CRF receptors in the hypothalamus. Thus, complicated regulation of GnRH in the hypothalamus and gonadotropins in the pituitary via CRF receptors contributes to stress responses and adaptation of gonadal functions.

**Keywords: corticotropin-releasing factor, urocortin, stress, gonadotropin**

"fendo-04-00012" — 2013/2/18 — 18:53 — page 1 — #1

# **INTRODUCTION**

A variety of stressors have been shown to suppress gonadal function (Chand and Lovejoy, 2011). Proteins that play key roles in vertebrate reproduction include the neuropeptides gonadotropinreleasing hormone (GnRH) and kisspeptin and their receptors (Kim et al., 2012): kisspeptin stimulates GnRH release from hypothalamic GnRH neurons via Gpr54, a G protein-coupled receptor (Messager et al., 2005), while the gonadal steroid estrogen mediates its inhibitory effect on GnRH secretion by acting on kisspeptin-expressing neurons of the arcuate nucleus (Oakley et al., 2009; Ohkura et al., 2009). The expression of kisspeptin and kisspeptin receptor mRNA is downregulated by stressors including restraint, hypoglycemia, and lipopolysaccharide, which suggests that kisspeptin/kisspeptin receptor signaling plays a critical role in the transduction of stress-induced suppression of reproduction (Kinsey-Jones et al., 2009). In fact, kisspeptin–GPR54 signaling in the arcuate nucleus of the mediobasal hypothalamus is a critical neural component of the hypothalamic GnRH pulse generator (Li et al., 2009).

Gonadotropin-inhibitory hormone (GnIH), an RFamiderelated peptide, can also modulate the reproduction of vertebrates (Ubuka et al., 2008). GnIH neurons interact directly with GnRH neurons, and the action of GnIH is mediated by a novel G protein-coupled receptor, Gpr147 (Ubuka et al., 2008). In mice, higher concentrations of GnIH-like substances are expressed in the hypothalamus and GnIH reduces GnRH release from the mouse hypothalamus (Bentley et al., 2010). The glucocorticoid and corticotropin-releasing factor (CRF) receptors are expressed in a large population of GnIH/RFamide-related peptide-expressing cells (Kirby et al., 2009). Glucocorticoids increase the inhibitory actions of GnIH on GnRH secretion (Kirby et al., 2009), while the regulation of GnIH via the CRF receptor remains to be determined.

Corticotropin-releasing factor activates and regulates the hypothalamic–pituitary–adrenal (HPA) axis during stress (Vale et al., 1981, 1997). Stress-induced CRF synthesis and secretion from the hypothalamic paraventricular nucleus (PVN) stimulates adrenocorticotropic hormone (ACTH) release from pituitary corticotrophs (Gillies et al., 1982; Mouri et al., 1993), which, in turn, stimulates the release of glucocorticoids from the adrenal glands (Whitnall, 1993). These glucocorticoids then moderate the stress response by inhibiting hypothalamic PVN production of CRF and pituitary production of ACTH (Whitnall, 1993). Urocortin 1 (Ucn1), a 40-amino acid peptide originally cloned from the Edinger–Westphal nucleus, is a member of the CRF family of peptides (Vaughan et al., 1995). Both CRF and Ucn1 contribute to stress responses and cardiovascular and gonadal functions via G protein-coupled seven transmembrane receptors (Vale et al., 1997; Kageyama et al., 1999a; Suda et al., 2004). CRF exhibits high affinity for CRF type 1 receptor (CRF1 receptor; IC50 = 1.6 nM) but not for CRF type 2b receptor (CRF2b receptor; IC50 = 42 nM), while Ucn1 exhibits similar affinity for CRF1 receptor (IC50 = 0.16 nM) and CRF2b receptor (IC50 = 0.86 nM; Jahn et al., 2004). CRF1 receptor is predominately expressed in the brain and pituitary gland (Chang et al., 1993; Chen et al., 1993; Vita et al., 1993; Potter et al., 1994). In the pituitary, the CRF1 receptor is mainly expressed by corticotrophs and is responsible for mediating the effects of

hypothalamic CRF on ACTH secretion in response to stress (Wynn et al., 1985; Antoni, 1986).

Ucn2 and Ucn3 prohormones were identified in the human genome database and in mouse genomic DNA, respectively (Hsu and Hsueh, 2001; Lewis et al., 2001; Reyes et al., 2001), from which the identity and existence of endogenous peptides were predicted (Fekete and Zorrilla, 2007). Ucn2 and Ucn3 are more similar to each other than to CRF with regard to receptor binding (Kishimoto et al., 1995; Lovenberg et al., 1995a; Perrin et al., 1995; Stenzel et al., 1995). Ucn2 exhibits high affinity for CRF2b receptor (IC50 = 0.25 nM) but low affinity for CRF1 receptor (IC50 > 350 nM; Jahn et al., 2004). Similarly, Ucn3 binds with moderate affinity to CRF2b receptor (IC50 = 14 nM), but its specific binding to CRF1 receptor is not detectable (IC50 > 2000 nM; Jahn et al., 2004). It is hypothesized that an ancient gene duplication event is behind why Ucn1 belongs to the CRF lineage and why Ucn2 and Ucn3 represent a separate paralogous lineage (Fekete and Zorrilla, 2007).

The CRF1 receptor is primarily involved in stress responses and depression, while the CRF2 receptor is believed to mediate "stresscoping" responses in the brain, such as anxiolysis (Suda et al., 2004), because mice deficient in the CRF2 receptor or treated with a CRF2 receptor antagonist display increased anxiety-like behaviors and hypersensitive stress responses (Bale et al., 2000). Furthermore, both Ucn2 and Ucn3 act as anorexigenic neuropeptides via the CRF2 receptor (Fekete et al., 2011; Chao et al., 2012) and Ucn3 regulates glucose-stimulated insulin secretion and energy homeostasis (Li et al., 2007). Ucn3 signaling through the CRF2 receptor is also a critical molecular mediator in the ventromedial nucleus of the hypothalamus in regulating feeding and peripheral energy metabolism (Chao et al., 2012).

Corticotropin-releasing factor is involved in the suppression of the hypothalamic–pituitary–gonadal (HPG) axis (Rivier et al., 1986), especially the GnRH pulse generator in the hypothalamus (Knobil, 1992). Stress profoundly inhibits the reproductive function by suppressing the pulsatile release of GnRH and consequently luteinizing hormone (LH), at least in part via the CRF system as well as through the GABAergic system (Lin et al., 2012). Although CRF and Ucn clearly have potent effects on the HPG system, their possible roles and how they are regulated have yet to be fully determined. In this review, we focus on the regulation and the roles of Ucn2 in pituitary gonadotrophs and discuss the regulation of GnRH via CRF receptors in the hypothalamus.

#### **REGULATION OF GONADOTROPINS BY CRF AND Ucn2 IN THE PITUITARY**

Changes in CRF1 receptor expression and desensitization of the receptor in pituitary corticotrophs play a major role in modulating adaptive responses to stressors (Kageyama et al., 2006). CRF, vasopressin, lipopolysaccharides, cytokines, and glucocorticoids can negatively modulate the levels of pituitary CRF1 receptor mRNA (Pozzoli et al., 1996; Sakai et al., 1996;Aubry et al., 1997). However, CRF2 receptor mRNA is also found in the anterior pituitary and combined immunohistochemistry and *in situ* hybridization have demonstrated that CRF2 receptor mRNA colocalizes mainly with gonadotrophs, not corticotrophs (**Figure 1**).

RNase protection assays of anterior pituitary mRNA show that the dominant receptor type is the CRF type 2a receptor (CRF2a) receptor and not the CRF2b receptor (Lovenberg et al., 1995a; Kageyama et al., 2003). Rat CRF2a receptor, linked to various roles in the brain, is expressed primarily in several discrete brain regions, including the hypothalamus, lateral septum, and raphe nuclei (Lovenberg et al., 1995b), whereas the CRF2b receptor is found predominately in peripheral tissues such as the heart, gastrointestinal tract, arterioles, and muscles (Kageyama et al., 1999b). These data suggest that the CRF2a receptor in pituitary gonadotrophs is involved in the modulation of gonadotropin secretion and/or gonadal function.

Activation of the stress system could potentially influence reproduction at any level of the HPG axis (Tilbrook et al., 2002). The stress-induced decreases in LH/follicle-stimulating hormone (FSH) secretion influence gonadal functions such as sex steroidogenesis and sperm production (Demura et al., 1989; Tilbrook et al., 2002). Ucn2 is expressed mainly in corticotrophs of rat pituitary (Yamauchi et al., 2005), and its secretion and expression levels are increased by CRF and suppressed by glucocorticoids (Nemoto et al., 2007).

The CRF2 receptor-selective ligand Ucn2 suppresses both expression and secretion of gonadotropins in rats, while a CRF2 receptor antagonist increases the secretion of gonadotropins (Nemoto et al., 2009). In addition, an anti-CRF antibody blocks stress-induced increases in plasma ACTH and corticosterone, and an anti-Ucn2 antibody blocks stress-induced suppression of LH secretion without affecting stress-induced ACTH and corticosterone release (Nemoto et al., 2010). Stress-induced increases in microRNA-325-3p also suppress gonadotropin secretion (Nemoto et al., 2012). Although the presence and/or secretion of mature Ucn2 has not been determined in the pituitary or other tissues, it is possible that stress-induced increases in CRF stimulate Ucn2 in corticotrophs, which inhibits gonadotropin secretion via CRF2 receptors in the pituitary.

# **REGULATION OF GnRH BY CRF AND Ucn VIA CRF RECEPTORS IN THE HYPOTHALAMUS**

Although peripheral administration of CRFfails to affect LH secretion (D'Agata et al., 1984; Rivier and Vale, 1984), central injection of CRF inhibits secretion of gonadotropins (Rivier et al., 1986). These effects of CRF probably reflect a central mechanism that involves modulation of the activity of GnRH neurons in the hypothalamus (Petraglia et al., 1987; Li et al., 2010). Indeed, in monkeys, a CRF antagonist attenuates suppression of the GnRH pulse generator in response to hypoglycemic stress (Chen et al., 1996). Furthermore, a recent *in vivo* rat study indicated that CRF innervation of the dorsolateral bed nucleus of the stria terminalis plays a central role in stress-induced suppression of the GnRH pulse generator (Li et al., 2011).

Corticotropin-releasing factor also suppresses GnRH gene expression levels in murine GnRH GT1-7 cells (Kinsey-Jones et al., 2006). In fact, GT1-7 GnRH-producing cells have been used extensively in studies of the basic control mechanisms involved in GnRH neuronal function. Belsham and colleagues have managed to develop cell lines that are representative of the enormous range of cell types of the hypothalamus (Dalvi et al., 2011). N39,

"fendo-04-00012" — 2013/2/18 — 18:53 — page 2 — #2

"fendo-04-00012" — 2013/2/18 — 18:53 — page 3 — #3

developed from primary mouse fetal hypothalamic culture, is one of these homologous neuronal cell lines. To further understand the possible function of Ucn and the regulation of GnRH by CRF receptors in the hypothalamus, hypothalamic N39 cells have been studied because they express both CRF1 and CRF2 receptor mRNA and protein (Kageyama et al., 2012). It has been shown in these cells that a CRF1 receptor antagonist, antalarmin, inhibits CRF-induced decreases in GnRH mRNA levels, which suggests that CRF decreases GnRH mRNA levels via the CRF1 receptor (**Figure 2**).

The CRF2 receptor may also be involved in the regulation of GnRH gene expression. It has been reported that CRF regulates

GnRH mRNA levels via, at least in part, the CRF2 receptor in GT1-7 cells (Kinsey-Jones et al., 2006). In N39 cells, Ucn2 increases GnRH mRNA levels, and these Ucn2-induced increases in GnRH mRNA levels are blocked by the CRF2 receptor antagonist antisauvagine-30 (**Figure 2**). These results suggest that Ucn2 stimulates GnRH mRNA levels via the CRF2 receptor in hypothalamic cells. In an *in vivo* study, hypoglycemia- and lipopolysaccharideinduced suppression of LH involves activation of CRF2 receptor while restraint stress-induced inhibition of LH pulses involves both CRF1 and CRF2 receptors (Li et al., 2006). On the other hand, a more recent *in vivo* study showed that a CRF1 receptor antagonist blocks the acute stress-induced increases in gonadotropin secretion on the morning of proestrus while a CRF2 receptor antagonist weakly blocks the increase in FSH secretion (Traslaviña and Franci, 2012). Although GnRH production and secretion may be differentially modulated via CRF receptors under different stressors, further study will be required to elucidate the involvement of CRF receptors.

Glucocorticoids were recently shown to increase CRF2a receptor expression while simultaneously inhibiting CRF1 receptor expression in pancreatic β cell-derived insulinoma MIN6 cells expressing glucocorticoid receptors (Huising et al., 2011). The differential effects of the glucocorticoids on the expression of these receptors in the endocrine pancreas represent a mechanism of shifting sensitivity from CRF1 to CRF2 receptor ligands (Huising et al., 2011). In the hypothalamus, glucocorticoids, released in response to stress, inhibit GnRH and gonadotropins through activation of GnIH (Kirby et al., 2009). It has yet to be determined whether glucocorticoid-induced changes in CRF and Ucn are involved in the regulation of GnRH and gonadotropins.

### **RELATION BETWEEN SEXUAL DIFFERENCES AND THE CRF SYSTEM IN THE HYPOTHALAMUS**

Sexual dimorphism is associated with stress sensitivity and interaction of the HPA and HPG axes (Chand and Lovejoy, 2011). Estrogens are implicated in the differing stress responses between the sexes and modulate activation of the HPA axis; females, but not males, generally have slight hypercortisolism (Magiakou et al., 1997). Estrogen replacement increases the basal levels of ACTH in ovariectomized rats (Ochedalski et al., 2007) and in postmenopausal women (Fonseca et al., 2001). Moreover, women in the midluteal phase, when both progesterone and estrogen levels are relatively high, show enhanced ACTH levels in response to a stressor (Altemus et al., 2001).

Estrogens acting centrally, including in the pituitary corticotrophs and the hypothalamus, are able to modulate the stress responses (Nakano et al., 1991), and direct estrogenic regulation of CRF gene expression has also been demonstrated in various tissues (Vamvakopoulos and Chrousos, 1993; Dibbs et al., 1997). As high levels of estrogen replacement increase the basal levels of CRF mRNA in the PVN of ovariectomized rats (Ochedalski et al., 2007), estrogen would regulate the HPA axis *in vivo* by stimulating CRF gene expression in the hypothalamus. CRF mRNA levels in the PVN are not affected by estrogen treatment in either gonadectomized estrogen receptor (ER) type β (ERβ) knockout mice or wild-type male mice (Nomura et al., 2002). Therefore, it is likely that estrogen modulates CRF gene expression in a sex-dependent manner.

Hypothalamic 4B cells show characteristics of the parvocellular neurons of the PVN because these cells express CRF, vasopressin, CRF1 receptor, and glucocorticoid receptors. Estrogen directly stimulates CRF gene expression in hypothalamic 4B cells (Ogura et al., 2008), suggesting that estrogen is involved in the positive regulation of CRF gene expression in the parvocellular region of the PVN *in vitro*. Neurons expressing both CRF and ERβ are found in the medial parvocellular division (Miller et al., 2004) and project to the median eminence, and CRF in parvocellular PVN neurons exerts effects on corticotroph ACTH secretion (Gillies et al., 1982; Mouri et al., 1993). Therefore, estrogen and ERβ would contribute to the enhancement of stress responses through stimulation of CRF neurons of the hypothalamus, and may constitute the basis of sexual dimorphism in the regulation of the CRF gene (Straub, 2007). In addition, estrogen also enhances CRF- and stress-induced suppression of pulsatile LH secretion (Cates et al., 2004), and upregulation of the CRF2 receptor may contribute to the sensitizing influence of estradiol on the CRF- and stressinduced suppression of the GnRH pulse generator (Kinsey-Jones et al., 2006).

Meanwhile, Ucn1 in the non-preganglionic Edinger–Westphal nucleus plays an important role in stress adaptation. Estrogens exert a differential transcriptional regulation of the Ucn1 gene through either ER type α (ERα) or ERβ receptors (Haeger et al., 2006). Ucn1 mRNA levels in the non-preganglionic Edinger–Westphal nucleus of male rats are much higher than those of females (Derks et al., 2010), and estrogens may

"fendo-04-00012" — 2013/2/18 — 18:53 — page 4 — #4

contribute to stress adaptation through modulation of Ucn1 production.

#### **CONCLUSION**

In summary, Ucn2, mainly produced in corticotrophs in response to CRF, acts on gonadotrophs expressing the CRF2 receptor and inhibits the production of gonadotropins in the pituitary (**Figure 3**). CRF is involved in the suppression of the HPG axis, especially the GnRH pulse generator in the hypothalamus, and also decreases GnRH mRNA levels via the CRF1 receptor (**Figure 3**). The CRF2 receptor may be involved in the regulation

# **REFERENCES**


deficient for corticotropin-releasing hormone receptor-2 display anxietylike behaviour and are hypersensitive to stress. *Nat. Genet.* 24, 410–414.


of GnRH production and secretion. GnRH production and secretion may be differentially modulated via CRF receptors in response to different stressors. Thus, complicated regulation of GnRH and gonadotropins via the CRF receptors contributes to stress responses and adaptation in gonadal functions.

## **ACKNOWLEDGMENT**

This work was supported in part by Health and Labour Sciences Research Grants (Research on Measures for Intractable Diseases) from the Ministry of Health, Labour, and Welfare of Japan.

controversies and challenges. *Gen. Comp. Endocrinol.* 171, 253–257.


"fendo-04-00012" — 2013/2/18 — 18:53 — page 5 — #5

activity in the rhesus monkey: roles of vasopressin and corticotropinreleasing factor. *Endocrinology* 137, 2012–2021.


male rats. *Proc. Natl. Acad. Sci. U.S.A.* 106, 11324–11329.


"fendo-04-00012" — 2013/2/18 — 18:53 — page 6 — #6

2 α and CRF2 β receptor mRNAs are differentially distributed between the rat central nervous system and peripheral tissues. *Endocrinology* 136, 4139–4142.


of pituitary urocortin 2 in the regulation of expression and secretion of gonadotropins. *J. Endocrinol.* 201, 105–114.


roles of urocortins, human homologues of fish urotensin 1, and their receptors. *Peptides* 25, 1689–1701.


"fendo-04-00012" — 2013/2/18 — 18:53 — page 7 — #7

and functional expression of mouse pituitary and human brain corticotrophin releasing factor receptors. *FEBS Lett.* 335, 1–5.


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 09 May 2012; accepted: 30 January 2013; published online: 20 February 2013.*

*Citation: Kageyama K (2013) Regulation of gonadotropins by corticotropinreleasing factor and urocortin. Front. Endocrin. 4:12. doi: 10.3389/fendo.2013. 00012*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Kageyama. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Neuromedins U and S involvement in the regulation of the hypothalamo–pituitary–adrenal axis

# *Ludwik K. Malendowicz\*, Agnieszka Ziolkowska and Marcin Rucinski*

Department of Histology and Embryology, Poznan University of Medical Sciences, Poznan, Poland

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

James A. Carr, Texas Tech University, USA Gábor B. Makara, Hungarian Academy of Sciences, Hungary

#### *\*Correspondence:*

Ludwik K. Malendowicz, Department of Histology and Embryology, Poznan University of Medical Sciences, 6 Swie¸ cicki St., 60-781 Poznan, Poland. e-mail: lkm@amp.edu.pl

#### **INTRODUCTION**

In search for new biologically active peptides, the group of Minamino, Kangawa, and Matsuo in the 1980s isolated numerous small neuropeptides from porcine spinal cord. All of them exerted potent smooth-muscle stimulating activity. These short peptides have been named neuromedins (Minamino et al., 1985). Their sequences and biological activities are similar to some known neuropeptides and therefore they are commonly divided into four groups (classes):

Bombesin-like neuromedins—neuromedin B (NMB) and neuromedin C (NMC) (Minamino et al., 1983, 1984b).

Kassinin-like neuromedins—neuromedin K (NMK) and neuromedin L (NML) (Kangawa et al., 1983; Minamino et al., 1984a). Neurotensin-like neuromedins—neuromedin N (NMN) (Minamino et al., 1984c).

Neuromedins U (NMU), for which no substantial homology with other known neuropeptides was found (Minamino et al., 1985). This group, however, was expanded in 2005, when Mori et al. (2005, 2008) isolated neuromedin S (NMS) from rat brain. NMS is composed of 36 amino acid residues and both peptides share the same amidated C-terminal heptapeptide. Furthermore, both NMU and NMS appeared to be endogenous ligands for the orphan G protein-coupled receptors FM-3/GPR66 and FM-4/TGR-1, identified earlier as type-1 and type-2 NMU receptors (NMUR1 and NMUR2), respectively (Tan et al., 1998; Howard et al., 2000; Raddatz et al., 2000; Mori et al., 2005).

**Abbreviations:** CRH, corticotropin releasing hormone; GPCR, G protein coupled receptor; HPA, hypothalamo–pituitary–adrenal; icv, intracerebroventrical; iPVN, intraparaventricular nucleus; KO, knock out; NMS, neuromedin S; NMU, neuromedin U; NMUR1, neuromedin U receptor 1; NMUR2, neuromedin U receptor 2; PCR, polymerase chain reaction; PVN, paraventricular nucleus; QPCR, quantitative real time polymerase chain reaction; RT-PCR, reverse transcription PCR; sc, subcutaneously; WT, wild type.

We reviewed neuromedin U (NMU) and neuromedin S (NMS) involvement in the regulation of the hypothalamo–pituitary–adrenal (HPA) axis function. NMU and NMS are structurally related and highly conserved neuropeptides. They exert biological effects via two GPCR receptors designated as NMUR1 and NMUR2 which show differential expression. NMUR1 is expressed predominantly at the periphery, while NMUR2 in the central nervous system. Elements of the NMU/NMS and their receptors network are also expressed in the HPA axis and progress in molecular biology techniques provided new information on their actions within this system. Several lines of evidence suggest that within the HPA axis NMU and NMS act at both hypothalamic and adrenal levels. Moreover, new data suggest that NMU and NMS are involved in central and peripheral control of the stress response.

**Keywords: neuromedin U, neuromedin S, hypothalamus, pituitary, adrenal**

Identification of specific NMU receptors (NMUR1 and NMUR2) and its anorexigenic action have enhanced interest in physiological role of NMU and NMS (Howard et al., 2000; Ida et al., 2005). Advances in these studies were recently reviewed (Brighton et al., 2004; Mori et al., 2008; Mitchell et al., 2009a; Budhiraja and Chugh, 2009). Present review, on the other hand, will focus on expression and of role of NMU/NMS system in hypothalamo–pituitary–adrenal (HPA) axis functioning, updating thus our earlier reviews (Malendowicz and Markowska, 1994; Malendowicz, 1998; Malendowicz et al., 2013).

# **ISOLATION, STRUCTURE, AND SYNTHESIS OF NMU AND NMS**

Originally NMU was isolated from porcine spinal cord in two molecular forms, one containing 25 (NMU25) and the other 8 (NMU8) amino acid residues (Minamino et al., 1985). Subsequently NMU was isolated from other vertebrates, among them humans (Austin et al., 1995), rat (Conlon et al., 1988), guinea pig (Murphy et al., 1990), dog (O'Harte et al., 1991a), rabbit (Kage et al., 1991), chicken (O'Harte et al., 1991b; Domin et al., 1992), frogs—*Rana temporaria*, *Litoria caerulea*, and *Bombina maxima* (Domin et al., 1989b; Salmon et al., 2000; Lee et al., 2005), and goldfish (Maruyama et al., 2008). Amino acid sequences of NMU from different species are shown in **Figure 1**.

In the mammalian NMUs a common C-terminal sequence— Phe-Leu-Phe-Arg-Pro-Arg-Asn-NH2—contains the active site of the neuropeptide, which is formed by the amino acid residues between positions 2 and 8 (Hashimoto et al., 1991; Sakura et al., 1991). Recent data indicate that in most species studied, the five amino acids at the C-terminus of the NMUs are totally conserved, suggesting that this region is of major importance for biological activity (Lo et al., 1992; Brighton et al., 2004; Mitchell et al., 2009b). Amidation of the C-terminus of NMU is required for receptor activation (for review see Brighton et al., 2004).


**FIGURE 1 | Amino acid sequences of neuromedin U from some mammalian, avian, piscine, and amphibian species.** The box, highlighting the C-terminal pentapeptide, shows conservation of this sequence in vertebrates, except of goldfish. From Mitchell et al. (2009a), modified. Amino acid sequences were acquired from NCBI. Numbers after NMU denote peptide length.

Unexpectedly, in 2005, Mori et al. (2005) isolated from the rat brain a new 36 amino acid peptide related to NMU, which appeared to be another endogenous ligand of FM-4/TGR-1 receptor. This neuropeptide is highly expressed in the suprachiasmiatic nucleus of the hypothalamus and therefore was designed as NMS. Human NMS, on the other hand, is composed of 33 amino acid residues. All NMS share a C-terminal core structure with NMU. NMU and NMS share the same amidated C-terminal heptapeptide and bind to the same receptors NMUR1 and NMUR2. Amino acid sequences of NMS from different species are shown in **Figure 2**.

It should be emphasized that NMU and NMS genes are located on different chromosomes (NMU on 4q12 and NMS on 2q11.2) (Mori et al., 2005). Moreover, evidences indicate that during the

**FIGURE 2 | Amino acid sequences of neuromedin S from some mammalian and amphibian species.** The box, highlighting the C-terminal decapeptide, shows conservation of this sequence in vertebrates. NMU and NMS share the same amidated C-terminal heptapeptide. Amino acid sequences were acquired from NCBI. Numbers after NMS denote peptide length.

process of evolution NMU and NMS genes had already diverged at the level of the Amphibia (Chen et al., 2006).

In humans both NMU and NMS genes are composed of 10 exons and 9 introns. The mRNA lengths encoded by these genes are 816 and 485 bp, respectively. The exon–intron boundaries in the NMU and NMS prepro-proteins are comparably conserved (Mori et al., 2008). General structures of encoded prepro-NMU and prepro-NMS are shown in **Figure 3**.

# **NMU AND NMS RECEPTORS**

Early studies revealed the presence of highly specific 125I-NMU binding sites on membranes prepared from the rat uterus. The binding was saturable and specific and Scatchard analysis suggested a single class of binding site with a Kd of 0.35 nM (Nandha et al., 1993).

By means of modern molecular biology techniques two receptors for NMU were identified. In 1998, Tan et al. (1998) cloned a GPCR (FM-3) from human and murine cDNA libraries. This DNA segment has homology to the growth hormone secretagogue receptor and the neurotensin receptor. Subsequently, by means of a "reverse pharmacological" method NMU was identified as an endogenous ligand for the orphan human GPCR, FM-3 (or GPR66) (Fujii et al., 2000; Hedrick et al., 2000; Hosoya et al., 2000; Howard et al., 2000; Kojima et al., 2000; Raddatz et al., 2000; Shan et al., 2000; Szekeres et al., 2000). After identification of the second NMU receptor, the first one had been named NMUR1.

The NMUR2 (FM-4, TGR1) gene, on the other hand, was identified based on its sequence similarity with NMUR1 (Hosoya et al., 2000; Howard et al., 2000; Raddatz et al., 2000; Shan et al., 2000). NMUR1 gene is located on human chromosome 2—position q37.1 and NMUR2 gene on chromosome 5—position q33.1 (Mitchell et al., 2009a).

In humans NMUR1 gene consists of 3 exons and 2 introns, the size of encoded mRNA is 3274 bp and the receptor is composed of 426 aa residues. NMUR2 gene, on the other hand, consists of 4 exons and 3 introns. The size of its mRNA is 2067 bp and the receptor is composed of 415 aa residues (**Figure 4**).

NMS has also been identified as an endogenous ligand of NMUR1 and NMUR2 receptors and some data indicate that

**humans.** Data from Protein Knowledgebase (UniProtKB) P48645 and Q5H8A3, respectively. Schematic structure of prepro-NMU is modified from Austin et al. (1995). Numbers refer to residues and cleavage sites are given in red. Asterisk marks amidated asparagine.

NMUR2 has greater affinity to NMS than NMU (Mori et al., 2005).

Interaction of NMU and NMS with their receptors results in intracellular calcium mobilization and subsequent stimulation of inositol phosphates. These effects are mediated by both Gq*/*<sup>11</sup> and Gi*/*<sup>0</sup> proteins (Raddatz et al., 2000; Shan et al., 2000; Szekeres et al., 2000; Funes et al., 2002; Mori et al., 2005; Maruyama et al., 2011). Moreover, both neuromedins are also able to activate the mitogen activated protein kinase ERK1 and ERK2 (Brighton et al., 2004).

# **EXPRESSION OF NMU AND NMS AND THEIR RECEPTORS IN THE HYPOTHALAMO–PITUITARY–ADRENAL AXIS HYPOTHALAMUS** *NMU and NMS*

Soon after NMU identification, high concentrations of NMUlike immunoreactivity were found in extracts of the rat, mouse, and human hypothalamus (Domin et al., 1986, 1988). In rat hypothalamus concentration of NMU-like immunoreactivity was reported as 31*.*2 ± 5*.*6 pmol/g (nearly 6 times higher than in the anterior pituitary lobe). Subsequent immunohistochemical studies demonstrated the presence of NMU-like substances in the hypothalamic paraventricular (PVN) and arcuate nucleus of the rat (Honzawa et al., 1987; Ballesta et al., 1988; Steel et al., 1988). On the other hand, molecular biology techniques demonstrated low to moderate levels of NMU mRNA in rat (RT-PCR) and human (QPCR) hypothalamus (Fujii et al., 2000; Szekeres et al., 2000). Single-cell reverse transcription-multiplex polymerase chain reaction (single-cell RT-mPCR) technique revealed that 14.7% parvocellular neurons of the rat PVN expressed NMU mRNA (Chu et al., 2012).

NMU mRNA is present in hypothalamus of WT mice and, in contrast, NMU mRNA could not be detected in NMU KO mice (Fukue et al., 2006).

In the frog (*Rana esculenta*) NMU-like immunoreactivity was observed in perikaria of the dorsal nucleus of the hypothalamus and the caudal part of the infundibulum (Maderdrut et al., 1996). Although authors emphasize that this pattern of immunoreactivity distribution differs notably from that seen in mammals, it is necessary to emphasize that the ventral infundibular nucleus of the frog is homologous to the mammalian arcuate nucleus (Neary and Northcutt, 1983; Ten Donkelaar, 1998).

The highest expression of NMS mRNA was found by RT-PCR in rat hypothalamus (Mori et al., 2005). Subsequently, *in situ* hybridization demonstrated that this neuropeptide was specifically expressed in the hypothalamic suprachiasmatic nuclei (SCN). In the rat hypothalamus expression of NMS gene was nearly 3-fold higher than that of NMU gene (Rucinski et al., 2007). During postnatal development levels of NMS mRNA attained maximum at prepubertal stage and adulthood (Vigo et al., 2007).

# *NMU receptors*

Earliest studies revealed that NMUR1 is expressed predominantly in periphery while NMUR2 in the central nervous system (for review see Brighton et al., 2004; Mitchell et al., 2009a). In agreement with this principle in the rat hypothalamus, by means

of QPCR Raddatz et al. (2000) demonstrated high expression of NMUR2 and a notably lower one of NMUR1. This finding was confirmed by numerous groups (Fujii et al., 2000; Howard et al., 2000; Qiu et al., 2005; Rucinski et al., 2007; Vigo et al., 2007). NMUR2 gene is also expressed in the PVN. By means of the single-cell RT-mPCR analysis (multiplex) Qiu et al. (2005) demonstrated that NMU-sensitive PVN neurons abundantly expressed NMUR2 mRNA but expressed NMUR1 mRNA to a lesser extent or not at all.

Detailed mapping of NMUR2 mRNA expression in the rat brain by *in situ* hybridization revealed the most intense signal in ependymal cells of the third ventricle and moderate signal in the PVN (Guan et al., 2001; Graham et al., 2003). Presence of NMUR2 receptors in these regions of the brain had also been confirmed by receptor autoradiography (Mangold et al., 2008).

Similar pattern of NMU receptor expressions is observed in human hypothalamus, with high expression of NMUR2 and a negligible one of NMUR1 (Szekeres et al., 2000; Gartlon et al., 2004).

# **PITUITARY GLAND**

#### *NMU and NMS*

High concentrations of NMU-like immunoreactivity were found in the pituitary gland of the rat as early as in 1987 (Domin et al., 1987, 1988, 1989a). In rat anterior lobe concentration of NMUlike immunoreactive substances was 6.3 pmol/gland (mean) while in posterior lobe 0.3 pmol/gland only. Furthermore, Northern blot analysis using total RNA extracted from rat anterior pituitary demonstrated high levels of NMU mRNA in the gland (Lo et al., 1992).

High concentrations of NMU protein in rat pituitary gland are accompanied by high expression of NMU gene (Fujii et al., 2000). High expression of this gene in anterior pituitary lobe of the rat was confirmed by QPCR (Rucinski et al., 2007; Shimizu et al., 2008). It should be emphasized that NMU mRNA is also expressed in cultured pituitary cells of the rat. Using quantitative *in situ* hybridization, high concentration of NMU transcripts was also found in part tuberalis of the rat pituitary (Ivanov et al., 2002; Nogueiras et al., 2006).

Immunohistochemistry demonstrated the presence of NMUimmunoreactive substances in the intermediate and the anterior pituitary gland lobes of mouse, rat, and human (Ballesta et al., 1988; Steel et al., 1988). Electron microscopy studies revealed NMU-like immunoreactivity in some thyrotropes and most corticotropes of the rat pituitary gland. NMU is colocalized with galanin and ACTH in the same secretory granules (Cimini et al., 1993; for review see Malendowicz, 1998; Cimini, 2003).

In developing rat NMU-immunopositive cells appear in anterior pituitary at day E15 (Cimini, 2003). Their number increases at day E20 and then decreases. More than 60% of NMUimmunoreactive cells contain ACTH at E15, and, after falling to about 40% at E16, this value is more or less constant until E21.

NMU mRNA is expressed in pituitary gland of WT mice, but in contrast, NMU mRNA could not be detected in NMU KO mice (Fukue et al., 2006).

Expression of NMS gene in rat pituitary gland was found by Mori et al. (2005). In rat anterior pituitary expression level of NMS gene however, was notably lower than that of NMU gene (Rucinski et al., 2007).

# *NMU receptors*

Conflicting data were reported on expression of NMUR1 and NMUR2 in pituitary gland. By means of QPCR low expression of both receptors in human pituitary gland was reported by Raddatz et al. (2000). These data were confirmed by other groups (Shan et al., 2000; Gartlon et al., 2004).

The earliest studies did not reveal NMUR1 gene expression in the rat pituitary gland while that of NMUR2 was very low (Fujii et al., 2000; Gartlon et al., 2004). On the other hand, in rat adenohypophysis expression of NMUR1 gene, but not of NMUR2 was observed by our group (Rucinski et al., 2007).

Expression of both NMUR1 and NMUR2 genes was observed in mouse pituitary gland of both WT and NMU KO mice (Fukue et al., 2006).

#### **ADRENAL GLAND**

Only scanty data are available on expression of NMU–NMS and their receptors in adrenal glands. Very low levels of NMU mRNA in the rat adrenal were reported by Fujii et al. (2000). Detailed studies on rat adrenal gland revealed very low expression of NMU and NMS genes at mRNA levels (Rucinski et al., 2007).

In human adrenal gland both NMUR2 mRNA (PCR) and protein (dot blot method) were identified in 2000 (Raddatz et al., 2000; Shan et al., 2000).

In the rat adrenal gland NMUR2 mRNA could not be demonstrated (Hosoya et al., 2000; Fujii et al., 2000). In immature rat adrenal NMUR1 was expressed at both mRNA and protein (immunohistochemistry) levels while the signal from NMUR2 was absent. NMUR1 mRNA was detected in all adrenocortical zones and in medulla of the gland (Ziolkowska et al., 2008). Moreover, the presence of NMUR1 mRNA in isolated zona glomerulosa and fasciculata/reticularis cells rules out the possibility that the expression was due to the presence in the specimens assayed of the non-parenchymal components of the gland. Expression of NMUR1 as mRNA and protein was demonstrated in adrenal gland of intact rat, in enucleation-induced regenerating gland, in hemiadrenalectomized animals (compensatory adrenal growth) as well as in ACTH-stimulated one (Trejter et al., 2008, 2009; Malendowicz et al., 2009).

# **NMU AND NMS IN THE**

# **HYPOTHALAMO–PITUITARY–ADRENAL AXIS FUNCTIONING HYPOTHALAMUS**

The above described localization of elements of NMU/NMS and NMUR2 system in hypothalamus forms a base of regulation by NMU and NMS of HPA axis functioning.

The earliest experiments with intracerebroventricular (icv) injection of NMU demonstrated a strong increase in Fosimmunoreactive nuclei in the PVN and supraoptic nucleus (SON) of the rat hypothalamus (Niimi et al., 2001; Ozaki et al., 2002). Almost all CRH-containing neurons in the parvocellular divisions of the PVN expressed Fos-like immunoreactivity 90 min after icv administration of NMU (Yokota et al., 2004).

Subsequent studies demonstrated direct NMU effects on CRH and arginine vasopressin (AVP) release by rat hypothalamic explants *in vitro* (Wren et al., 2002). At the 100 nM NMU concentration CRH and AVP release was nearly doubled when compared to controls. Moreover, results of whole cell patch-clamp recordings revealed that NMU directly depolarized the subpopulation of PVN parvocellular, but not magnocellular neurons via enhancement of the hyperpolarization-activated inward current (Qiu et al., 2003). Direct effects of both neuromedins on PVN was also suggested by *in vitro* electrophysiological studies which showed

that NMU and NMS increased the neuronal firing rates in both arcuate and PVN nuclei slices (Nakahara et al., 2010).

NMU also regulates HPA axis in birds. Icv administration of NMU in chicks significantly upregulated mRNA expression of CRH in the hypothalamus (Kamisoyama et al., 2007).

NMS likewise affects CRH neurons in PVN. In the rat icv administration of this neuromedin increased POMC mRNA expression in the arcuate nucleus and CRH mRNA in the PVN (Ida et al., 2005; Nakahara et al., 2010). NMS notably stimulated Fos-immunoreactive cells in both hypothalamic nuclei. NMS also significantly increased firing rate of PVN cells. Stimulating effects of NMS on Fos-immunoreactive cells of the PVN was confirmed by another group (Sakamoto et al., 2007, 2008).

#### **PITUITARY**

Expression of NMUR1 in pituitary gland and colocalization of NMU and ACTH in pituitary corticotropes suggest NMU and NMS involvement in regulation of ACTH secretion. Regarding this, there is a growing body of evidence that in the rat NMU administered icv, into PVN or subcutaneously increases blood ACTH concentrations, via stimulation of CRH release.

First reports demonstrated that a single sc injection of NMU8 resulted in a transient increase in ACTH blood concentration while after 2–6-day treatment (low NMU8 dose) blood ACTH level remained unchanged (Malendowicz et al., 1993, 1994b; Trejter et al., 2009). Only the higher dose of NMU8 (6 µg/ 100 g/day for 6 days) increased the level of circulating ACTH.

In the rat unilateral adrenalectomy notably increased plasma ACTH concentrations and NMU administration (sc) into hemiadrenalectomized rats did not significantly change corticotropin levels (Malendowicz et al., 2009).

In the rat icv administration of NMU (0.1, 1, and 3 nmol/rat) resulted in a dose-dependent increase of plasma ACTH concentrations, an effect significantly reduced by pretreatment with anti-NMU IgG (Ozaki et al., 2002). Similar dose-dependent effects of NMU on plasma ACTH concentrations were found after iPVN neuropeptide administration (Wren et al., 2002). Again, chronic iPVN NMU administration (twice-daily of 0.3 nmol NMU for 7 days) did not change plasma ACTH concentrations (Thompson et al., 2004). Only one group was unable to demonstrate stimulating effect of icv administered NMU on plasma ACTH concentrations in the rat (Rokkaku et al., 2003).

Unfortunately, no studies have as yet investigated direct effects of NMU and NMS on pituitary ACTH secretion. Our preliminary data indicate that neither NMU nor NMS affect ACTH release by quarters of the rat adenohypophysis, while the response to CRH was normal. This observation may suggest that observed *in vivo* stimulating effect of NMU/NMS on ACTH secretion is mediated via hypothalamus.

#### **ADRENAL**

Potent stimulating effects of exogenous NMU on adrenocortical steroid secretion in the rat have been described as early as in 1993. A single sc injection of NMU resulted in a transient increase in ACTH blood concentration (between 3 and 12 h) and a sustained (24 h) elevation of plasma corticosterone concentration (Malendowicz et al., 1993, 1994b). These data demonstrated stimulating effect of neuropeptide on adrenal cortex, possibly partially due to the direct effect of NMU on the gland.

In subsequent searches for mode of NMU action on corticosteroid secretion our group found that NMU had no effect on basal and ACTH-stimulated corticosterone secretion by freshly isolated or cultured inner zone adrenocortical cells, nor did it change their cytosolic Ca2<sup>+</sup> concentration (Malendowicz et al., 1994b; confirmed by Ziolkowska et al., 2008). However, this neuropeptide stimulated corticosterone output by adrenal slices, but not by fragments of adrenocortical autotransplants lacking medullary chromaffin cells (Malendowicz et al., 1994a). Detailed analysis revealed that at all concentrations tested NMU increased basal pregnenolone and total post-pregnenolone steroid output by gland slices containing both cortex and medulla. The increase in total post-pregnenolone steroid output induced by low concentrations of NMU8 was due to similar rises in the production of non-18-hydroxylated steroids; conversely, that provoked by higher concentrations of the peptide was almost exclusively caused by the rise in the yield of 18-hydroxylated steroids. The stimulating effect of NMU8 on pregnenolone output was blocked by both alpha-helical-CRH and corticotropin-inhibiting peptide, which are competitive inhibitors of CRH and ACTH, respectively. These data suggest that NMU affects corticosteroid secretion indirectly by acting on the medullary chromaffin cells, which in turn may paracrinally stimulate cortex of the gland. We also suggested that the medullary mediator of NMU action on adrenal cortex may activate 18-hydroxylation and aldosterone synthase activity. On the other hand, we cannot rule out the possibility of NMU action on adrenal cortex via stimulation of adrenaline secretion by adrenal medulla. Recently it has been demonstrated that centrally administered NMU evokes secretion of adrenaline from the adrenal medulla (Sasaki et al., 2008). Medullary adrenaline may in turn modulate secretion of corticosteroids (for review see Nussdorfer, 1996; Bornstein et al., 1997; Ehrhart-Bornstein et al., 1998).

Stimulating effects of NMU on corticosteroid secretion also were observed after icv or iPVN neuropeptide administration. In the rat acute iPVN administration of NMU dose-dependently increased plasma corticosterone concentrations (Wren et al., 2002). Similar effect was observed after icv NMU administration (Ozaki et al., 2002). In contrast, another group reported lack of icv NMU administration effects on plasma corticosterone levels (Gartlon et al., 2004).

Stimulating effects of icv administered NMU on plasma corticosterone levels were confirmed by experiments with anti-NMU IgG (Jethwa et al., 2006). In rats administration of anti-NMU IgG significantly attenuated the dark phase rise in plasma corticosterone concentrations.

In contrast to acute administration, chronic iPVN administration of NMU produced an elevation of plasma corticosterone levels while plasma ACTH concentrations remained unchanged (Thompson et al., 2004). Thus, in chronically NMU administered rats elevated plasma corticosterone levels, not accompanied by increased ACTH concentrations, were independent of the mode of NMU administration (sc vs iPVN) (Malendowicz et al., 1993, 1994b; Thompson et al., 2004).

The stimulating effect of sc administered NMU on plasma corticosterone concentrations also was found in rats during enucleation-induced regeneration, as well as in rats treated with low ACTH doses (Trejter et al., 2008, 2009). On the other hand, NMU8 administration to immature rats was found to raise aldosterone, but not corticosterone concentrations (Ziolkowska et al., 2008).

NMS, like NMU also stimulates corticosteroid secretion. In rats icv NMS administration resulted in nearly 5-fold increase in plasma corticosterone concentrations and the effect was dependent on neuropeptide dose (Jászberényi et al., 2007). Likewise, in castrated Holstein steers both NMS and NMU evoked a doserelated increase in plasma cortisol concentrations (Yayou et al., 2009). Of interest is that also in chicks icv NMS administration stimulated corticosterone release (Tachibana et al., 2010).

In contrast to the above described NMS effects on corticosterone/cortisol secretion, an opposite effect was seen in Rhesus monkeys (Jahan et al., 2011). In their experiments NMS was infused through a teflon cannula implanted into saphenous vein. Blood samples were collected 45 min before and 120 min after NMS administration at 15 min intervals. Under these conditions NMS resulted in a significant decrease in plasma cortisol levels at 40 and 60 nmol doses while a non-significant difference was observed at 20 nmol. This unexpected finding may suggest that in non-human primates NMS exerts inhibitory effect on HPA axis functioning. However this finding remains to be confirmed.

Recent data also demonstrated direct stimulating effect of NMU on proliferative activity of immature rat inner adrenocortical cells in primary culture (Ziolkowska et al., 2008). Moreover, in enucleation-induced adrenal regeneration NMU notably enhanced proliferative activity of adrenocortical cells (Trejter et al., 2008). A similar stimulating effect of sc administered NMU on proliferative activity (metaphase index) was seen in ACTH treated rats (Trejter et al., 2009). In contrast, in hemiadrenalectomized rats NMU notably inhibited adrenocortical cell proliferation in both zona glomerulosa and zona fasciculata, as assessed by the metaphase index (Malendowicz et al., 2009). Since these effects were independent of changes in blood ACTH, they could reflect an interaction of NMU with the neural system innervating the adrenal gland, which is responsible for compensatory adrenal growth.

## **INVOLVEMENT OF NMU AND NMS IN THE STRESS RESPONSE**

As it follows from the above presented data, NMU and NMS are linked to the HPA axis functioning. Independently of mode of administration (icv, iPVN, or sc) NMU activates CRH containing neurons and stimulates CRH secretion, which in turn triggers pituitary ACTH and adrenal corticosterone/cortisol secretion. In this regard it is not astonishing that NMU and NMS are involved in central and peripheral control of the stress response. These studies were initiated by Hanada et al. (2001) who observed stress-related behavior (gross locomotor activity, face washing and grooming) in icv NMU administered rats. This response was attenuated by pretreatment with alpha-helical CRH (antagonist of CRH) or anti-CRH IgG. Furthermore, NMU-induced increases in oxygen consumption and body temperature were attenuated in CRH KO mice. Subsequent studies demonstrated that blood corticosterone levels were significantly increased after 10 min of immobilization stress in wild-type mice, but not in NMU KO mice (Nakahara et al., 2004). Excessive grooming induced by icv NMU administration were abolished in the NMUR2 KO mice, an observation suggesting that NMUR2 plays a decisive role in stress/anxiety induction (Zeng et al., 2006).

Stress related behavior induced by icv NMU or NMS administration was also observed in cattle (Yayou et al., 2009). Administration of both neuropeptides tended to shorten the duration of lying and increase the number of head shaking in studied cattle. These behavioral changes were accompanied by increased levels of plasma cortisol.

#### **CONCLUDING REMARKS**

The above reviewed data clearly demonstrate NMU and NMS involvement in regulation of HPA axis growth and functions. However, their mechanism of action is far from being completely understood. The available experimental data suggest that within the HPA axis these neuromedins may exert both endocrine and/or paracrine/autocrine effects on target cells.

Endocrine effects of discussed neuropeptides require their release into the bloodstream. Only scanty data are available on NMU and NMS presence in general circulation. Manufacturers' manuals for various kits contain data on NMU and NMS concentration in plasma and comparable figures are shown in clinical studies (Ketterer et al., 2009). However, experimental data suggest that neither endogenous NMU nor NMS are circulating, at least in the rat (Peier et al., 2009). Furthermore, in general circulation the biological half-life of NMU is shorter than 5 min.

At the hypothalamus level the actions of NMU and NMS are rather well documented. Stimulation of CRH, argininevasopresssin and oxytocin release both *in vivo* and *in vitro* activates HPA axis and apart from elevation of ACTH and corticosterone/cortisol secretion, triggers behavioral parameters typical for stress reaction.

More puzzling is the effect of NMU on pituitary gland. As discussed earlier, in pituitary gland high levels of NMU are present, and the gland contains high concentration of NMU-like immunoreactive substances and is provided with NMUR1 receptor. NMU is colocalized with ACTH and numerous NMU-like immunoreactive cells are present in human extrapituitary corticotropinomas (Steel et al., 1988). Denef (2008) suggests that in the pituitary NMU may paracrinally regulate function of adjacent cells and a similar mode of action may be present in extrapituitary corticotropinomas. So far, direct effects of NMU/NMS on pituitary ACTH secretion have not been reported. Our preliminary data indicate that neither NMU nor NMS affect ACTH release by quarters of the rat adenohypophysis. It remains to be established whether inability of NMU and NMS to directly stimulate ACTH release from pituitary gland may be caused by possible high receptor occupancy by endogenous NMU.

In view of these findings, questions arise concerning the possible role of NMU in pituitary tumor formation. In their review on pituitary tumorogenesis Korbonits et al. (2004) mention NMU only as a potential messenger in the anterior pituitary. Thus, the possible role of NMU in development of pituitary and extrapituitary corticotropinomas requires further investigation. As far as endocrine system is concerned, an elevated expression of NMU gene was found in human ovarian cancer cell lines (Euer et al., 2005).

The available experimental data suggest that at the adrenal level NMU affects the steroid secretion indirectly by acting on the medullary chromaffin cells. On the other hand some data suggest direct effect of neuromedin on proliferation and growth of rat adrenocortical cells.

Thus, although great progress has been made in understanding NMU and NMS action within the HPA axis during the past years, much remains to be learned about their mechanisms of action within this system. Recent development of a metabolically stable analog of NMU, based on derivatization of the native peptide with high molecular weight poly(ethylene) glycol (PEG) ("PEGylation") may be helpful in these attempts (Ingallinella et al., 2012).

#### **ACKNOWLEDGMENTS**

Review was prepared using funds from the grant no N N401 227839 from the Ministry of Science and Education in Poland.

#### **REFERENCES**


Morphological and functional studies of the paracrine interaction between cortex and medulla in the adrenal gland. *Microsc. Res. Tech.* 36, 520–533.


mammalian neuropeptide, neuromedin S (NmS), in the dermal venoms of Eurasian bombinid toads. *Biochem. Biophys. Res. Commun.* 345, 377–384.


neuromedin U-like immunoreactivity in rat corticotropes after alteration of endocrine status. *Cell Tissue Res.* 272, 137–146.


and rat central nervous system and effects of neuromedin-U following central administration in rats. *Psychopharmacology (Berl.)* 177, 1–14.


obesity and diabetes. *Bioorg. Med. Chem.* 20, 4751–4759.


U analog from the skin secretions of toad *Bombina maxima*. *Regul. Pept.* 129, 43–47.


et al. (2009). The antiobesity effects of centrally administered Neuromedin U and Neuromedin S are mediated predominantly by the Neuromedin U receptor 2 (NMUR2). *Endocrinology* 150, 3101–3109.


U analog from the defensive skin secretion of the Australasian tree frog, *Litoria caerulea*. *J. Biol. Chem.* 275, 4549–4554.


*Biochem. Biophys. Res. Commun.* 323, 65–71.


secretion. *Endocrinology* 148, 813–823.


*Received: 30 August 2012; accepted: 20 November 2012; published online: 05 December 2012.*

*Citation: Malendowicz LK, Ziolkowska A and Rucinski M (2012) Neuromedins U and S involvement in the regulation of the hypothalamo–pituitary– adrenal axis. Front. Endocrin. 3:156. doi: 10.3389/fendo.2012.00156*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Malendowicz, Ziolkowska and Rucinski. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *Marie-Odile Guimond and Nicole Gallo-Payet\**

Division of Endocrinology, Department of Medicine, Faculté de Médecine et des Sciences de la Santé, Université de Sherbrooke, Sherbrooke, QC, Canada

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Lie Gao, University of Nebraska Medical Center, USA Thomas Unger, Maastricht University, Netherlands

#### *\*Correspondence:*

Nicole Gallo-Payet, Service d'Endocrinologie, Département de Médecine, Faculté de Médecine et des Sciences de la Santé, Université de Sherbrooke, 3001, 12e Avenue Nord, Sherbrooke, QC, Canada J1H 5N4. e-mail: nicole.gallo-payet@ usherbrooke.ca

The angiotensin type 2 (AT2) receptor of angiotensin II has long been thought to be limited to few tissues, with the primary effect of counteracting the angiotensin type 1 (AT1)receptor. Functional studies in neuronal cells have demonstrated AT2 receptor capability to modulate neuronal excitability, neurite elongation, and neuronal migration, suggesting that it may be an important regulator of brain functions. The observation that the AT2 receptor was expressed in brain areas implicated in learning and memory led to the hypothesis that it may also be implicated in cognitive functions. However, linking signaling pathways to physiological effects has always proven challenging since information relative to its physiological functions has mainly emerged from indirect observations, either from the blockade of the AT1receptor or through the use of transgenic animals. From a mechanistic standpoint, the main intracellular pathways linked to AT2 receptor stimulation include modulation of phosphorylation by activation of kinases and phosphatases or the production of nitric oxide and cGMP, some of which are associated with the Gi-coupling protein. The receptor can also interact with other receptors, either G protein-coupled such as bradykinin, or growth factor receptors such as nerve growth factor or platelet-derived growth factor receptors. More recently, new advances have also led to identification of various partner proteins, thus providing new insights into this receptor's mechanism of action. This review summarizes the recent advances regarding the signaling pathways induced by the AT2 receptor in neuronal cells, and discussed the potential therapeutic relevance of central actions of this enigmatic receptor. In particular, we highlight the possibility that selective AT2 receptor activation by non-peptide and selective agonists could represent new pharmacological tools that may help to improve impaired cognitive performance in Alzheimer's disease and other neurological cognitive disorders.

**Keywords: AT2 receptor, angiotensin, brain, differentiation, regeneration, neurodegenerative disorders, signaling, cognitive functions**

"fendo-03-00164" — 2012/12/17 — 21:16 — page 1 — #1

#### **INTRODUCTION**

It is now well accepted that the effects of the various components of the renin-angiotensin system (RAS) range in various aspects of peripheral and brain functions well beyond those of regulating blood pressure and hydro-mineral balance. In particular, the existence of a complete RAS in the brain is fully acknowledged. Its activation leads to angiotensin II (Ang II) production, which is usually viewed as the end-product of this system (de Gasparo et al., 2000). Ang II binds two receptors from the G protein-coupled receptor family (GPCR), namely the angiotensin type 1 (AT1) and angiotensin type 2 (AT2) receptor. Although physiological functions of the AT1 receptor are relatively well-established, ranging from vasoconstriction and aldosterone release to cell growth, the effects associated with the AT2 receptor are surrounded by controversy. Both AT1 and AT2 receptors are expressed in various brain areas involved in the regulation of fluid and electrolyte balance and in the regulation of arterial pressure, as well as in structures involved in cognition, behavior, and locomotion (Phillips and de Oliveira, 2008; Horiuchi et al., 2010; Horiuchi and Mogi, 2011; Wright and Harding, 2011, 2012; Mogi and Horiuchi, 2012).

One of the biggest challenges in studying the AT2 receptor is to apply what has been observed using cell lines to *in vivo* models. Indeed, studies using cell lines expressing the AT2 receptor either endogenously or by transfection, have provided paramount information regarding its intracellular mechanisms of action, although associating these mechanisms with biological functions has proven to be much more difficult. Indeed, most of the relevant information regarding AT2 receptor functions in the brain has emerged from indirect observations, either by use of AT1 receptor blockers (ARB) or *via* transgenic "knock-down" animals for AT2 receptor expression. The present review summarizes recent advances in AT2 receptor signaling pathways, and discusses how they could be related to the neuroprotective functions of the receptor.

# **BRAIN EXPRESSION AND ROLE OF THE AT2 RECEPTOR**

As summarized in several reviews (de Gasparo et al., 2000; Porrello et al., 2009; Gallo-Payet et al., 2011; Wright and Harding, 2011; Mogi and Horiuchi, 2012), the AT2 receptor is widely expressed during fetal life, which decreases rapidly after birth (Grady et al., 1991; Breault et al., 1996; Schutz et al., 1996; Nuyt et al., 1999), although a recent study has reported opposite results (Yu et al., 2010). This study is indeed in sharp contrast with previous reports using more specific methods, like autoradiography or *in situ* hybridization. In the adult, AT2 receptor expression is limited to a few tissues and cell types, such as vascular endothelial cells, adrenal gland, kidney, heart, myometrial cells, and ovaries (review in Porrello et al., 2009; Gallo-Payet et al., 2011, 2012;Verdonk et al., 2012). In the adult central nervous system (CNS), the AT2 receptor is observed in certain specific brain areas involved in the control and learning of motor activity, control of autonomous functions, sensory areas, and selected limbic system structures (Lenkei et al., 1996, 1997). In particular, it is the major Ang II receptor in the medulla oblongata (control of autonomous functions), septum and amygdala (associated with anxiety-like behavior), thalamus (sensory perception), superior colliculus (control of eye movements in response to visual information) as well as subthalamic nucleus and cerebellum (areas associated with learning of motor functions). On the other hand, certain areas involved in cardiovascular functions, learning, behavior, and stress reactions (cingulate cortex, molecular layer of the cerebellar cortex, superior colliculus, and paraventricular nuclei) contain both AT1 and AT2 receptors (Millan et al., 1991; Tsutsumi and Saavedra, 1991; Lenkei et al., 1996, 1997). More recently, expression of the AT2 receptor was also detected in the substantia nigra pars compacta, an area involved in dopaminergic signals and associated with Parkinson's disease (Grammatopoulos et al., 2007), and in the hippocampus (Arganaraz et al., 2008; AbdAlla et al., 2009). At the cellular level, the AT2 receptor is expressed in neurons, but not in astrocytes (Bottari et al., 1992a; Lenkei et al., 1996; Gendron et al., 2003). Evidence also suggests that the AT2 receptor is expressed in the vasculature wall, where it acts on cerebral blood flow (review in Horiuchi and Mogi, 2011; Horiuchi et al., 2012). It should also be noted that existence of a non-AT1/non-AT2 receptor in the CNS has been suggested, which displays high affinity for Ang I, II, and III (Karamyan and Speth, 2007).

# **ROLE OF THE AT2 RECEPTOR IN NEURONAL EXCITABILITY**

One of the first roles of the AT2 receptor to be identified was the modulation of neuronal excitability, which plays a crucial role not only in neuronal differentiation, but also in neuronal functions (review in Gendron et al., 2003; Gao et al., 2011). In particular, in cells of neuronal origin, activation of the AT2 receptor decreases activity of T-type calcium channels (Buisson et al., 1992, 1995). On the other hand, in rat brain neuronal culture, Kang et al. (1994) showed that the AT2 receptor stimulates a delayed rectifier K<sup>+</sup> current (IK) and a transient K<sup>+</sup> current (IA), an effect dependent on the G-protein Gi and the serine/threonine phosphatase PP2A. Consistent with these observations, a recent study showed that AT2 receptor induces a hyperpolarization and a decrease in firing rate in rostral ventrolateral medulla (RVLM) neurons suggesting that central activation of the AT2 receptor in this region decreases excitability (Matsuura et al., 2005). More recently, another study using C21/M024 demonstrated that selective stimulation of AT2 receptor in the neuronal cell line (called

"fendo-03-00164" — 2012/12/17 — 21:16 — page 2 — #2

CATH.a neurons) increases the potassium current activity (*I*Kv) in a nitric oxide (NO)-dependant pathway (Gao et al., 2011). Moreover, intracerebroventricular infusion of C21/M024 was associated with a decrease in norepinephrine excretion and in blood pressure. Indeed, the modulation of the receptor on neuronal excitability in this region could be one of the mechanism associated with its effect on blood pressure, since RVLM is often considered as the main regulator of vascular tone (review in Dupont and Brouwers, 2010). An inhibitory effect of the AT2 receptor on neuronal excitability has also been observed in the locus coeruleus from brain slice preparations (Xiong and Marshall, 1994) and in the superior colliculus (Merabet et al., 1997). Finally, using the selective agonist C21/M024, Jing et al. (2012) recently demonstrated that direct stimulation of cerebral AT2 receptor increases postsynaptic potential, thus corroborating previous*in vitro* observations. Interestingly, AT2 receptor-induced neuronal activation of delayed rectifier potassium channels has also been demonstrated to have a neuroprotective effect (Grammatopoulos et al., 2004a). In fact, these AT2 receptor effects on ionic channel activity suggest that it may be implicated in synaptic plasticity, an important process involved in learning and memory.

# **ROLE OF THE AT2 RECEPTOR IN NEURONAL DIFFERENTIATION**

One of the best recognized effects of AT2 receptor stimulation in neuronal cells is the induction of neurite outgrowth (review in Gallo-Payet et al., 2011). In the early 1990s, our group observed that stimulation of the AT2 receptor with its selective agonist CGP42112A induces neurite outgrowth in the neuronal NG108-15 cell line (Laflamme et al., 1996), results that were further confirmed using the recently developed non-peptide selective AT2 receptor agonist C21/M024 (Wan et al., 2004). This effect was associated with an increase in mature neural cell markers, such as βIII-tubulin, and microtubule-associated proteins (MAPs) such as MAP2c (Laflamme et al., 1996), both known to stabilize tubulin in a polymerized state, thus participating actively in differentiation (Sanchez et al., 2000). Similar results have also been reported in the pheochromocytoma-derived cell line PC12W, where Ang II was found to promoted neuronal differentiation characterized by an increase in neurite elongation (Meffert et al., 1996) and enhanced levels of polymerized βIII-tubulin and MAP2 associated with microtubules (Stroth et al., 1998). However, neurite outgrowth in PC12W cells has also been associated with a reduced expression of MAP1B (Stroth et al., 1998) and neurofilament M (Gallinat et al., 1997), two proteins specifically associated with axon elongation (Gordon-Weeks, 1991). These results were further confirmed in primary neuronal cultures, including retinal explants (Lucius et al., 1998), microexplant cultures of the cerebellum (Coté et al., 1999), in neurospheres from mouse fetal brain (Mogi et al., 2006) as well as primary cultures of newborn brain cortex neurons (Li et al., 2007) and hippocampal neurons (Jing et al., 2012). Some studies also showed that this neurite elongation was associated with an increase in the repair of damaged DNA by induction of methyl methanesulfonate sensitive-2 (MMS2), a neural-differentiating factor (Mogi et al., 2006; Jing et al., 2012). Altogether, these results suggest that activation of the AT2 receptor is associated with important rearrangements of the cytoskeleton necessary for induction of neurite elongation.

# **ROLE OF THE AT2 RECEPTOR IN NEURONAL MIGRATION**

In cerebellar microexplants, where both neuronal and glial cells are present, AT2 receptor activation induces not only neurite outgrowth, but cell migration as well (Coté et al., 1999). Indeed, application of Ang II in this model induced cell migration of neurons from the center toward the periphery of the microexplant (Coté et al., 1999). These effects were more pronounced in cells treated with Ang II and DUP 753 (known as the ARB losartan) or in cells treated with 10 nM of CGP42112A an AT2 receptor agonist, and conversely blocked with the AT2 receptor antagonist PD123,319. Similar cell migration has also been observed during AT2 receptor-induced regeneration of post-natal retinal microexplants (Lucius et al., 1998). During migration and neurite outgrowth, cells are characterized by a myriad of advancing, retracting, turning, and branching behavioral patterns. Such dynamics and plasticity are driven by the reorganization of actin and the microtubular cytoskeleton. In particular, during the process of migration, actin filaments play a major role and are putatively considered as the primary target of guidance cues, due to their localization at the cell periphery, and in filopodium in the growth cone, where they are considered to be the driving force for the forward extension of the cell membrane (Gallo and Letourneau, 2004; Kalil and Dent, 2005). Our results on NG108- 15 cells have shown that the underlying mechanism involves an Ang II-induced decrease in the amount of F-actin in filopodium and an increase in the pool of unpolymerized actin, through a pertussis toxin (PTX)-sensitive increase in ADF/cofilin activity. These latter effects were found to be AT2 receptor-dependent, since the increase in the rate of migration was abolished by the selective antagonist PD123,319, but not by the selective AT1 receptor antagonist losartan. Interestingly, some co-localization of F-actin with microtubules was also observed in control conditions, but which disappeared during Ang II-induced migration (Kilian et al., 2008). Among the candidate molecules that possibly crosslink actin filaments and microtubules are MAP2c and MAP1B (Dehmelt et al., 2003; Dehmelt and Halpain, 2004), proteins previously shown by our group to be affected during the process of AT2 receptor-stimulated neurite outgrowth, both in NG108- 15 cells and in cerebellar granule cells (Laflamme et al., 1996; Coté et al., 1999).

# **MAIN SIGNALING PATHWAYS OF THE AT2 RECEPTOR**

Although the AT2 receptor displays most of the classical features of a GPCR, it is usually considered as an atypical member of this family, since it fails to induce all of the classical signaling pathways such as cAMP, production of inositol triphosphate (IP3) or intracellular calcium release. Signaling pathways associated with the AT2 receptor mainly involve a balance between phosphatase and kinase activities and according to whether the cell is undifferentiated or differentiated and whether it expresses angiotensin AT1 receptors or not. Thus, there is still much controversy surrounding this receptor, and its effects, either protective or deleterious, remain a subject of debate (Widdop et al., 2003; Steckelings et al., 2005, 2010; Porrello et al., 2009; Horiuchi et al., 2012;Verdonk et al., 2012). In our endeavor to elucidate the mechanisms associated with AT2 receptor-induced neurite outgrowth, we and others have investigated signaling pathways activated by this receptor, including G-protein coupling, regulation of kinase activity, interaction with growth factor receptors, and production of NO. Moreover, recent observations have also delineated new partners for the AT2 receptor which play key functions in its regulation (**Figure 1**).

# **G-PROTEIN COUPLING**

While coupling of G-protein to AT1 receptors is well described (de Gasparo et al., 2000; Hunyady and Catt, 2006), such coupling is not the rule for the AT2 receptor. Former studies have described a coupling to subunit Gαi2 and Gαi3 in rat fetus (Zhang and Pratt, 1996). In some models (rat hippocampal neurons and other selected cell types), blocking Gα<sup>i</sup> with PTX or antibodies directed against Gα<sup>i</sup> inhibited the AT2 receptor effects on actin depolymerization, activation of endothelial NO synthase (NOS), stimulation of neuronal K+ current and on anti-proliferative activity (Kang et al., 1994; Ozawa et al., 1996; Li et al., 2004; Olson et al., 2004; Kilian et al., 2008), indicating that coupling of the AT2receptor to Gα<sup>i</sup> is at least implicated in these pathways. However, aside from a few exceptions (Kang et al., 1994), PTX failed to inhibit either p42/p44mapk activation in the neuronal cell line NG108-15 (Gendron et al., 2002) or phosphatase activity in several models (for review see Nouet and Nahmias, 2000; Gendron et al., 2003).

#### **REGULATION OF KINASE ACTIVITY**

"fendo-03-00164" — 2012/12/17 — 21:16 — page 3 — #3

# *AT2 Receptor-induced phosphatase activation*

Phosphatase activation has been one of the first signals associated with AT2 receptor activation. After the earlier studies in PC12W cells (Bottari et al., 1992b; Brechler et al., 1994), results have been confirmed in other cell lines, including N1E-115 cells (Nahmias et al., 1995), NG108-15 cells (Buisson et al., 1995), and R3T3 fibroblasts (Tsuzuki et al., 1996a,b). This phosphatase activation by the AT2 receptor is essential for its anti-proliferative and pro-apoptotic effects (for reviews, see Nouet and Nahmias, 2000; Steckelings et al., 2005; Porrello et al., 2009; Verdonk et al., 2012). Currently, three main phosphatases have been implicated in AT2 receptor signaling, namely SH2-domain-containing phosphatase 1 (SHP-1), mitogen-activated protein kinase phosphatase 1 (MKP-1), and the serine–threonine phosphatase PP2A.

SHP-1 is a cytosolic phosphatase rapidly activated by the AT2 receptor following Ang II binding. Activation of SHP-1 is associated with AT2-induced growth inhibition in various cells, including neuronal cells (Bedecs et al., 1997; Elbaz et al., 2000; Feng et al., 2002; Li et al., 2007), vascular smooth muscle cells (Cui et al., 2001; Matsubara et al., 2001), CHO, and COS-7 cells transfected with the AT2 receptor (Elbaz et al., 2000; Feng et al., 2002). Activation of SHP-1 is associated with inhibitory effects of the AT2 receptor on the AT1 receptor, including transactivation of the epidermal growth factor (EGF) receptor and activation of c-Jun N-terminal kinase (JNK) (Matsubara et al., 2001; Shibasaki et al., 2001), but also on insulin-induced activation of the phosphatidylinositol 3-kinase (PI3K), its association with the insulin receptor substrate IRS-2 and phosphorylation of Akt (Cui et al., 2001). This inhibition of insulin signaling by AT2 receptor-induced SHP-1 activation has also been associated with an increase in PC12W

"fendo-03-00164" — 2012/12/17 — 21:16 — page 4 — #4

cell apoptosis (Cui et al., 2002). More recently, Li et al. (2007) have shown that induction of neurite outgrowth in fetal rat neurons by AT2 receptor involves the association of SHP-1 with the newly identified AT2-receptor interacting protein (ATIP; see section AT2 Receptor Interacting Proteins) and an increase in MMS2 protein (Li et al., 2007). Finally, although the mechanisms associated with AT2 receptor-induced activation of SHP-1 have yet to be fully elucidated, implication of G-protein coupling (Bedecs et al., 1997; Feng et al., 2002) as well as activation of Src kinase (Alvarez et al., 2008) have been reported; other studies have also implicated a constitutive association between AT2 receptor and SHP-1 in overexpressing models (Feng et al., 2002; Miura et al., 2005). Another phosphatase associated with AT2 receptor activation is MKP-1, which is a key regulator of p42/p44mapk activity. AT2 receptor-activated MKP-1 has been observed in various cell types, including PC12W cells (Yamada et al., 1996), fibroblasts (Horiuchi et al., 1997; Calo et al., 2010), and cardiac myocytes (Fischer et al., 1998; Hiroi et al., 2001). Activation of MKP-1 by AT2 leads to a decrease in p42/p44mapk activity, and is associated to growth inhibition induced by the AT2 receptor. Moreover, Horiuchi et al. (1997) demonstrated that AT2 receptor-induced MKP-1 activation is implicated in apoptotic effects of the AT2 receptor, leading to Bcl-2 dephosphorylation and an increase in Bax, resulting in cell death. Finally, the serine–threonine phosphatase PP2A is also activated by the AT2 receptor following Ang II binding and may be associated with AT2 receptor regulation of p42/p44mapk. Indeed, in primary neuronal cultures, AT2 receptor-induced activation of PP2A is associated with inhibition of AT1 receptor-induced p42/p44mapk phosphorylation (Huang et al., 1995, 1996a,b) and is implicated in AT2-induced modulation of potassium currents (Huang et al., 1995, 1996a; Caballero et al., 2004). More recently, we have also shown an implication of PP2A activation in actin depolymerization and an increase in neuronal migration (Kilian et al., 2008; **Figure 1**).

#### *Mitogen-activated protein kinase p42/p44*

Among all signaling pathways associated with AT2 receptor activation, regulation of p42/p44mapk is probably the one where variability is the most important. The effect of AT2 receptor stimulation on activation or inhibition of p42/p44mapk activity is dependent on the models studied, on whether they express AT1 receptors or not and whether cells are under physiological or pathological conditions. Thus, AT2 receptor effects on p42/p44mapk remain controversial. Many studies have shown that the AT2 receptor leads to dephosphorylation of p42/p44mapk *via* one the phosphatases associated with AT2 receptor signaling (see above). This decrease in p42/p44mapk activity is associated with inhibition of growth and pro-apoptotic effects of the AT2 receptor (review in Nouet et al., 2004; Porrello et al., 2009). In addition to activation of phosphatase, AT2 receptor-induced inhibition of p42/p44mapk can be mediated by inhibition of growth factor receptors. Indeed, in vascular smooth muscle cells overexpressing the AT2 receptor, stimulation with Ang II decreases EGF receptor phosphorylation and inhibits p42/p44mapk activation (Shibasaki et al., 2001). Similar observations have also been reported in CHO cells overexpressing the AT2 receptor (Elbaz et al., 2000). Worthy of note is the fact that inhibition of p42/p44mapk induced by the AT2 receptor is observed only in certain conditions, such as in cells overexpressing the AT2 receptor or already exhibiting pathological conditions such as serum-starving (Bedecs et al., 1997; Horiuchi et al.,1997; Elbaz et al.,2000;Cui et al.,2001; Shibasaki et al.,2001).

By contrast, in neuronal cells such as NG108-15 and PC12W cells, theAT2 receptor leads to sustained activation of p42/p44mapk. In these cells, activation of p42/p44mapk is essential to AT2 receptor-induced neurite elongation (Gendron et al., 1999; Stroth et al., 2000). In NG108-15 cells, we observed that this increase in p42/p44mapk activity was associated with the Rap1/B-Raf pathway. However, this Rap1 activation appears to be dependent of nerve growth factor receptor TrkA activation (see latter; Plouffe et al., 2006) rather than through cAMP and protein kinase A (PKA), as usually observed with other GPCR (**Figure 1**). This activation of p42/p44mapk by the AT2 receptor has also been observed in non-neuronal COS-7 and NIH3T3 cells overexpressing the AT2 receptor (Hansen et al., 2000; De Paolis et al., 2002).

# *Src family kinase*

There are few studies showing an implication of Src family members in AT2 receptor signaling. However, Src family kinases (SFKs) are key regulators in cell growth and differentiation and are implicated in most growth factor signaling pathways. In the CNS, five members of SFK are expressed, namely Src, Fyn, Lyn, Lck, and Yes, where they act as modulators of neurotransmitter receptors as well as in the regulation of excitatory transmission (review in Kalia et al., 2004; Theus et al., 2006; Ohnishi et al., 2011). Recently, we have shown that stimulation of the AT2 receptor in NG108-15 cells leads to rapid but transient activation of SFK and that expression of inactive Fyn abolished AT2 receptor-induced neurite outgrowth in these cells (Guimond et al., 2010). However, inhibition of Fyn had no effect on other signaling pathways induced by the AT2 receptor, including p42/p44mapk and Rap1 activation, suggesting that it may be involved either downstream of these proteins, or in a parallel pathway. Of note, among the five SFKs expressed in the brain, only a deficiency in Fyn-induced neurological deficits, including impairment in spatial learning and in hippocampal development (Grant et al., 1992; Kojima et al., 1997). Interestingly, similar physiological perturbations were also observed in mice lacking the AT2 receptor (Hein et al., 1995; Ichiki et al., 1995; Okuyama et al., 1999; Maul et al., 2008). Therefore, regulation of Fyn activity could be considered as a new player implicated in the protective effect of this receptor in cognitive disorders. Indeed, Fyn has been shown to be involved in tau phosphorylation, thus regulating its affinity for tubulin and stability of microtubules, two parameters implicated in the development of Alzheimer's disease (AD) and other neurodegenerative diseases (Lee et al., 1998, 2004). Thus, it appears that Fyn is involved in the final steps of induction of elongation, but not in the initial events of AT2 receptor activation. This implication of Fyn in AT2 receptor signaling is further strengthened by the fact that activation of SFKs, as the AT2 receptor, was shown to be important for the induction of long-term potentiation, a key element in learning and memory, in CA1 pyramidal neurons of hippocampal slices (Yu et al., 1997).

To the best of our knowledge, only one other group has demonstrated the implication of a Src family member in AT2 receptor signaling (Alvarez et al., 2008). In this latter study, it was shown that activation of c-Src was present in an immunocomplex including the tyrosine phosphatase SHP-1 and theAT2 receptor following Ang II stimulation in rat fetal membranes. Pre-incubation of membranes with the non-selective inhibitor PP2 inhibited SHP-1 activation and c-Src association. These results indicate that c-Src may represent an important step leading to AT2 receptor-induced SHP-1 activation. More recently, the same group demonstrated that this association also occurred in hindbrain membranes from post-natal day 15 rats, and was associated with focal adhesion kinase (p85FAK) (Seguin et al., 2012). These observations strongly suggest that c-Src may also be implicated in cytoskeleton remodeling associated with neurite elongation and neuronal migration induced by the AT2 receptor.

# **LINKING THE AT<sup>2</sup> RECEPTOR WITH THE GROWTH FACTOR RECEPTORS**

Recently, we demonstrated that activation of Rap1/B-Raf/ p42/p44mapk pathway by the AT2 receptor was dependent on the nerve growth factor receptor TrkA, although the mechanism involved remains unknown (Plouffe et al., 2006). In addition, we further showed that a SFK member was essential for the initial activation of TrkA by the AT2 receptor, since pre-incubation of NG108-15 cells with the non-selective inhibitor PP1 disrupted this effect (Guimond et al., 2010). However, although Fyn was essential for neurite outgrowth induced by the AT2 receptor, it did not appear to be implicated in TrkA activation, since expression of a dominant negative form did not impede AT2-induced TrkA activation (Guimond et al., 2010). In light of recent data obtained by Ciuffo's group regarding the involvement of c-Src and other SFK members with AT2 receptors (Alvarez et al., 2008; Seguin et al., 2012), it would be of interest to see whether the association of the AT2 receptor with SHP-1 and c-Src is implicated in this transactivation, and whether TrkA could be involved in

"fendo-03-00164" — 2012/12/17 — 21:16 — page 5 — #5

FAK activation. Interestingly, transactivation of the TrkA receptor in neurons has also been observed for the pituitary adenylyl cyclase-activating polypeptide receptor (PACAP; Rajagopal et al., 2004), which is also associated with neuronal development in the cerebellum (Basille et al., 2006).

Curiously, although the expression of inactive Fyn is known to disrupt AT2 receptor-induced neurite elongation, non-selective inhibition of SFK in NG108-15 cells with the inhibitor PP1 is sufficient to increase neurite elongation to levels similar to those observed with AT2 receptor stimulation (Guimond et al., 2010), which could be a consequence of a decrease in proliferative signal. Indeed, our group showed that induction of neurite outgrowth was associated with a decrease in cell proliferation through inhibition of PKCα and p21Ras (Gendron et al., 1999; Beaudry et al., 2006). Moreover, as in the case of SFK, inhibition of the plateletderived growth factor (PDGF) receptor was sufficient to induce neurite outgrowth and to increase microtubule polymerization more extensively than Ang II alone (Plouffe et al., 2006). These findings are in agreement with a previous report demonstrating that expression of an inactive form of the PDGF receptor in PC12 cells was sufficient to increase neurite elongation (Vetter and Bishop, 1995). However, whether AT2 receptor directly inhibits PDGF receptor or inhibits its signaling pathway is still unknown.

#### **NITRIC OXIDE AND cGMP PRODUCTION – A ROLE FOR BRADYKININ**

Nitric oxide has been shown to regulate several types of K+ channels, including ATP-dependent K<sup>+</sup> channels and Ca2+-activated K+ channels (review in Prast and Philippu, 2001). Indeed, in neuronal cell lines, observations with the selective AT2 receptor agonist C21/M024 revealed that this production of NO induced by AT2 was necessary for AT2-induced hyperpolarization of potassium channel function (Gao and Zucker, 2011). Production of NO following AT2 receptor stimulation has been observed in various cell types, such as neuronal cells (Chaki and Inagami, 1993; Coté et al., 1998; Gendron et al., 2002; Zhao et al., 2003; Muller et al., 2010), vascular endothelial cells (Wiemer et al., 1993; Seyedi et al., 1995; Saito et al., 1996; Thorup et al., 1998; Baranov and Armstead, 2005) as well as in smooth muscle cells (de Godoy et al., 2004). It is already well accepted that AT2 receptor activation plays an important role in the control of renal function particularly in chronic kidney diseases. The AT2 receptor is believed to counterbalance the effects of the AT1 receptor at least by influencing vasodilation through NO production and natriuresis (Carey and Padia, 2008; Siragy, 2010; Siragy and Carey, 2010). This promoter effect of AT2 on natriuresis in pathological conditions (obese Zucker rats) was also recently confirmed using C21/M024 (Ali and Hussain, 2012). Activation of NOS by the AT2 receptor can occur by direct signaling such as in neuronal cells, or indirectly *via* stimulation of bradykinin production and subsequent activation of its receptor B2. Indeed, heterodimerization between the AT2 receptor and bradykinin has also been described in PC12W cells (Abadir et al., 2006). Moreover, it is already known that bradykinin can modulate AT2 receptor-induced NO production (Siragy and Carey, 1996; Gohlke et al., 1998; Searles and Harrison, 1999). Such involvement of B2 receptors in AT2 receptor-induced production of NO is of prime importance in the modulation of cerebral blood flow. Indeed, an AT2-induced increase in spatial learning was recently observed to be associated with an increase in cerebral blood flow, an effect reduced by co-administration of the B2 receptor antagonist icatibant. This observation strongly suggests that the beneficial effect of the AT2 receptor in cognitive function is partly dependent on bradykinin (Jing et al., 2012). In addition, Abadir et al. (2003) demonstrated in conscious bradykinin B2-null and wild-type mice that the AT2 receptor can induce production of NO in both null and wild-type models, indicating that the B2 receptor may participate in this process, although is not the only means for the AT2 receptor to induce NO production.

# **AT<sup>2</sup> RECEPTOR ASSOCIATED PROTEINS** *ATIP*

Recently, using a yeast two-hybrid system, the ATIP was cloned and identified as a protein interacting with the C-terminal tail of the AT2 receptor (Nouet et al., 2004). This protein is expressed as five different transcripts, namely ATIP1, ATIP2, ATIP3a, ATIP3b, and ATIP4 (review in Rodrigues-Ferreira and Nahmias, 2010; Horiuchi et al., 2012). While ATIP3 appears to be the major transcript in tissues, ATIP1 and ATIP4 are mainly expressed in the brain, indicating that they may play biological roles in brain functions. ATIP2, on the other hand, is almost undetectable by real-time PCR (Di Benedetto et al., 2006). In CHO cells expressing the AT2 receptor, ATIP is known to decrease growth factor-induced p42/p44mapk activation and DNA synthesis, therefore decreasing cell proliferation, as well as decrease insulin receptor autophosphorylation, similarly to the AT2 receptor. Of particular interest is the fact that, although expression of the AT2 receptor was essential in this instance, stimulation by Ang II was not necessary, and that ATIP was able to exert its effect by its sole expression. Implication of ATIP in AT2 receptor-induced neurite outgrowth has also been reported. In this context, Ang II stimulation of the AT2 receptor induces translocation of ATIP with SHP-1 into the nucleus, resulting in the transactivation of MMS2 (Li et al., 2007). Moreover, ATIP, also known as ATBP50 (AT2 receptor binding protein of 50 kDa), has been reported as a membrane-associated Golgi protein implicated in intracellular localization of the AT2 receptor and necessary for its membrane expression (Wruck et al., 2005). ATIP3, which is also expressed in the CNS, has been shown to strongly interact with stabilized microtubules in a model of breast cancer, suggesting an implication on cell division, where it induces a delayed metaphase, thus decreasing tumor progression (Rodrigues-Ferreira et al.,2009). The brain-specific isoformATIP4 is highly expressed in the cerebellum and fetal brain, two sites where the AT2 receptor is also highly expressed. Therefore considering (i) the previously described function of the AT2 receptor in preservation of cognitive function, (ii) the role of ATIP protein in AT2 receptor function, and (iii) the link between ATIP protein and microtubule cytoskeleton, it could be suggested that regulation of ATIP expression and regulation of its association with the AT2 receptor could be an important element to consider with regard to the development of neurological disorders, such as AD.

# *PLZF*

"fendo-03-00164" — 2012/12/17 — 21:16 — page 6 — #6

Association between the AT2 receptor and the promyelocytic leukemia zinc finger (PLZF) protein has been observed using a yeast two-hybrid system (Senbonmatsu et al., 2003). In CHO cells expressing both PLZF and AT2 receptors, Ang II stimulation induces co-localization of PLZF with the AT2 receptor, followed by internalization of the complex. This observation is in contrast with other studies observing no internalization of the AT2 receptor following Ang II stimulation (Hunyady et al., 1994; Hein et al., 1997). Since internalization of the receptor was observed only in cells expressing PLZF, this could represent a new regulatory pathway of AT2 receptor function, specific only to selected cell types. However, beside internalization of AT2 receptor, a recent study showed that PLZF was implicated in neuroprotection in a stroke model (Seidel et al., 2011). In this study, the authors showed that PLZF exerts neuroprotective effect in a model of *in vitro* glutamate toxicity. They also showed that overexpression of PLZF in neuronal cells in culture induced a significant increase in AT2 receptor expression, suggesting that PLZF could also be implicated in the regulation of AT2 receptor expression.

# *PPAR***γ**

A new partner for the AT2 receptor has recently emerged from the study of Zhao et al. (2005) who observed that neurite outgrowth induced by AT2 receptor stimulation in PC12W cells was dependent on the activation of peroxisome proliferator-activated receptor gamma (PPARγ). This observation is in keeping with the implication of PPARγ in NGF-induced neurite outgrowth in the same cell type (Fuenzalida et al., 2005), clearly suggesting a possible crosstalk between the AT2 receptor and NGF pathways. This hypothesis is further reinforced by the observation that inhibition of the NGF receptor TrkA significantly decreases AT2 receptorinduced neurite outgrowth (Plouffe et al., 2006). Moreover, Iwai et al. (2009), using atherosclerotic ApoE-KO mice with an AT2 receptor deficiency (AT2R/ApoE double knockout mice), observed that the lack of AT2 receptor expression decreased the expression of PPARγ in adipocytes cells. These observations strongly suggest a link between the AT2 receptor and PPARγ functions. PPARγ is a transcriptional factor regulating the expression of multiple genes, hence promoting the differentiation and development of various tissues, specifically in adipose tissue, brain, placenta, and skin. Interestingly, neuroprotective effects of PPARγ agonist have also been observed (review in Gillespie et al., 2011). However, a major component of the hypothesis regarding the possible implication of PPARγ in AT2 receptor function is the PPARγ-like activity associated with certain ARBs, including telmisartan, irbesartan, and candesartan (Benson et al., 2004; Schupp et al., 2004; review in Horiuchi et al., 2012). Indeed, there is some evidence suggesting that this PPARγ activation following blockade of the AT1 receptor could be part of its anti-inflammatory and anti-oxidative effects, leading to neuroprotection against ischemia and amyloid β (Aβ) accumulation (Tsukuda et al., 2009; Iwanami et al., 2010; Washida et al.,2010). PPARγ has also been implicated in neural cell differentiation and death, as well as inflammatory and neurodegenerative conditions (review in Gillespie et al., 2011).

# **LESSONS FROM NEURONAL DIFFERENTIATION: HOW CAN THE AT2 RECEPTOR IMPROVE BRAIN FUNCTION? ROLE OF THE AT<sup>2</sup> RECEPTOR IN NEURONAL REGENERATION**

The capacity for nerve regeneration in lower vertebrates has been mostly lost in higher vertebrates and regeneration within the

"fendo-03-00164" — 2012/12/17 — 21:16 — page 7 — #7

CNS in mammals is essentially inexistent. However, after injury in the peripheral nervous system, regeneration can be achieved successfully. Observations that AT2 receptor stimulation induces neurite elongation associated with modulation of MAP expression strongly suggested that this effect could also be observed following nerve injury. In 1998, two studies demonstrated that theAT2 receptor improved nerve recovery in both optic (Lucius et al., 1998) and sciatic (Gallinat et al., 1998) nerve following nerve crush or in perivascular nerves implicated in vasodilation (Hobara et al., 2007). This effect was accompanied by an increase in AT2 receptor expression, the activation of NFκB and induction of growthassociated protein (GAP-43) leading to a reduction in lesion size. Moreover, Reinecke et al. (2003) demonstrated that activation of NFκB by the AT2 receptor was an essential step to recovery following sciatic nerve crush. This implication of AT2 receptor in neuronal regeneration has even led to the suggestion that Ang II, *via* the AT2 receptor, could act as a neurotrophic factor.

# **AT<sup>2</sup> RECEPTOR IN COGNITIVE FUNCTION**

There is increasing evidence suggesting that the AT2 receptor could be associated with improvement of cognitive function following cerebral ischemia-induced neuronal injury (Iwai et al., 2004; Li et al., 2005; Mogi et al., 2006; McCarthy et al., 2009). Indeed, it has been shown that central administration of CGP42112A increases neuronal survival and minimizes experimental post-stroke injury (McCarthy et al., 2009), indicating that activation of brain AT2 receptors exhibits a neuroprotective effect. More recently, stimulation of the AT2 receptor with the selective agonist C21/M024 was observed to prevent cognitive decline in an AD mouse model with intracerebroventricular injection of Aβ(1-40) (Jing et al., 2012). Indeed, some of the signaling pathways described above may be linked to improvement in impaired signalingfunctions as observed in AD. One of the major hallmarks of AD is Aβ deposition in senile plaques and the presence of neurofibrillary tangles (NFTs). Formation of NFTs is a consequence of protein tau accumulation, due to its hyperphosphorylation, and the dissociation of microtubules. Thus, regulation of tau phosphorylation is of paramount importance with regard to AD progression. On the other hand, several studies have reported that the AT2 receptor activates PP2A phosphatase (Huang et al., 1995, 1996a; Kilian et al., 2008), which is markedly deficient in AD (Gong et al., 1993, 2000; Wang et al., 2007) and implicated in glycogen synthase kinase-3 (GSK-3) inactivation *via* a sustained increase in p42/p44mapk. Since tau is a substrate for PP2A phosphatase, GSK-3 and Fyn, the latter of which is also implicated in the AT2 receptor effect on neurite outgrowth (Guimond et al., 2010), AT2 receptor activation could participate in controlling the equilibrium between tau phosphorylation and dephosphorylation (Hernandez and Avila, 2008; Hanger et al., 2009; Hernandez et al., 2009). In addition to acting on tau regulation, the AT2 receptor may also improve neurite architecture, through effects on MAPs, as observed in neuronal cell lines (Laflamme et al., 1996; Meffert et al., 1996; Coté et al., 1999; Li et al., 2007). The observation that central AT2 receptor activation using its selective agonist C21/M024 decreases cognitive loss induced by Aβ intracerebroventricular injection lends further support to this hypothesis (Jing et al., 2012). Although the mechanisms underlying these neuroprotective effects of the AT2

receptor remain to be fully elucidated, they may include PPARγ and the protein MMS2 (Mogi et al., 2006, 2008; for recent reviews see Gallo-Payet et al., 2011, 2012).

Moreover, as indicated earlier, another important feature of AT2 receptor signaling is induction of NO and cGMP production. Recently, Jing et al. (2012) observed that direct stimulation of central AT2 receptors increases NO *via* a bradykinin-dependent pathway, an effect which leads to an increase in cerebral blood flow and enhanced spatial memory. A further study also showed that administration of C21/M024 reduced early renal inflammatory response with production of NO and cGMP (Matavelli et al., 2011). This increase in NO-cGMP production has also been shown to lead to a decrease in nicotinamide adenine dinucleotide phosphate-oxidase (NADPH) superoxide production (Volpe et al., 2003; Widdop et al., 2003; de la Torre, 2004; Steckelings et al., 2005; Iadecola et al., 2009), thus reducing oxidative stress and potentially associated neuronal apoptosis. This hypothesis is coherent with the observation that the AT2 receptor attenuates chemical hypoxia-induced caspase-3 activation in primary cortical neuronal cultures (Grammatopoulos et al., 2004b). Finally, inflammation is also a common feature of neurodegenerative diseases. In this regard, a recent study conducted in primary cultures of human and murine dermal fibroblasts, has shown that C21/M024 has anti-inflammatory effects, inhibiting tumor necrosis factor (TNF)-α-induced interleukin-6 levels and NFκB activity. This effect was notably initiated through increased activation of protein phosphatases and increased synthesis of epoxyeicosatrienoic acid (Rompe et al., 2010).

# **CONCLUSION**

Since its identification in the early 90s, the AT2 receptor has been and still is shrouded by controversy, its low expression in the adult and its atypical signaling pathways adding to the challenge of studying this receptor. Thanks to the major advances achieved in the past few years, several studies have confirmed

#### **REFERENCES**


(2008). Involvement of c-Src tyrosine kinase in SHP-1 phosphatase activation by Ang II AT2 receptors in rat fetal tissues. *J. Cell. Biochem.* 105, 703–711.


that stimulation of the AT2 receptor activates multiple signaling pathways which are linked to beneficial effects on neuronal functions (including excitability, differentiation, and regeneration), inflammation, oxidative stress, and cerebral blood flow (**Figure 1**). Several neurodegenerative diseases (including cognitive deficits and dementia) are closely associated with these neuronal and synaptic dysfunctions (Iadecola, 2004; Zlokovic, 2005; LaFerla et al., 2007; Boissonneault et al., 2009; Mucke, 2009; Nelson et al., 2009). Moreover, an increasing number of studies suggest that the protective effects of ARBs on brain damage and cognition may result not only from the inhibition of AT1 receptor effects, but also from the beneficial effect due to unopposed activation of the AT2 receptor. Thus, if further research confirms the promising early results obtained with the recently developed selective non-peptide AT2 receptor agonist C21/M024, the latter may represent a new pharmacological tool in the fight against neurological cognitive disorders. In addition, unraveling the underlying effects of the AT2 receptor on neuronal plasticity may lead to the development of even more potent and selective therapies.

#### **ACKNOWLEDGMENTS**

The authors are grateful to Pierre Pothier for critical reading of the manuscript and editorial assistance (Les Services PM-SYS Enr., Sherbrooke). This work presented in this review was supported by grants from the Canadian Institutes of Health Research (MOP-82819 to Nicole Gallo-Payet) and from the Alzheimer's Society of Canada to Nicole Gallo-Payet with Louis Gendron (Université de Sherbrooke) and Thomas Stroh (McGill University) and by the Canada Research Chair program to Nicole Gallo-Payet. Nicole Gallo-Payet is a past holder of the Canada Research Chair in Endocrinology of the Adrenal Gland. Marie-Odile Guimond is a postdoctoral fellowship in the laboratory of Nicole Gallo-Payet. Nicole Gallo-Payet and Marie-Odile Guimond are both members of the FRSQ-funded Centre de recherche clinique Étienne-Le Bel.

development. *J. Comp. Neurol.* 496, 468–478.


"fendo-03-00164" — 2012/12/17 — 21:16 — page 8 — #8

receptor antagonist with selective PPARgamma-modulating activity. *Hypertension* 43, 993–1002.


et al. (2001). Pivotal role of tyrosine phosphatase SHP-1 in AT2 receptormediated apoptosis in rat fetal vascular smooth muscle cell. *Cardiovasc. Res.* 49, 863–871.


"fendo-03-00164" — 2012/12/17 — 21:16 — page 9 — #9

oxide and cyclic GMP are involved in angiotensin II AT(2) receptor effects on neurite outgrowth in NG108- 15 cells. *Neuroendocrinology* 75, 70–81.


activation in primary cortical neuronal cultures. *Brain Res. Bull.* 62, 297–303.


amyloid, and hypertension. *Stroke* 40, S40–S44.


"fendo-03-00164" — 2012/12/17 — 21:16 — page 10 — #10

Fyn transgene. *Proc. Natl. Acad. Sci. U.S.A.* 94, 4761–4765.


doi: 10.1111/j.1447 [Epub ahead of print].


"fendo-03-00164" — 2012/12/17 — 21:16 — page 11 — #11


of peroxisome proliferator-activated receptor-gamma activation. *Hypertension* 54, 782–787.


receptor-gamma activation in mice with chronic cerebral hypoperfusion. *Stroke* 41, 1798–1806.


"fendo-03-00164" — 2012/12/17 — 21:16 — page 12 — #12


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 November 2012; paper pending published: 26 November 2012; accepted: 29 November 2012; published online: 19 December 2012.*

*Citation: Guimond M-O and Gallo-Payet N (2012) How does angiotensin AT2 receptor activation help neuronal differentiation and improve neuronal pathological situations? Front. Endocrin. 3:164. doi: 10.3389/fendo.2012.00164*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Guimond and Gallo-Payet. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Activation of PAC1 receptors in rat cerebellar granule cells stimulates both calcium mobilization from intracellular stores and calcium influx through N-type calcium channels

**Magali Basille-Dugay 1,2,3,4, Hubert Vaudry 1,2,3,4, Alain Fournier 4,5, Bruno Gonzalez 2,6 and David Vaudry 1,2,3,4\***

1 INSERM U982, Laboratory of Neuronal and Neuroendocrine Differentiation and Communication, University of Rouen, Mont-Saint-Aignan, France

<sup>3</sup> PRIMACEN, University of Rouen, Mont-Saint-Aignan, France

4 International Associated Laboratory Samuel de Champlain, University of Rouen, Mont-Saint-Aignan, France

5 Institut National de la Recherche Scientifique-Institut Armand Frappier, University of Québec, Laval, QC, Canada

<sup>6</sup> Région INSERM ERI28, Laboratory of Microvascular Endothelium and Neonate Lesions, University of Rouen, Rouen, France

#### **Edited by:**

Jae Young Seong, Korea University, South Korea

#### **Reviewed by:**

Hitoshi Hashimoto, Osaka University, Japan Dora Reglodi, University of Pecs, Hungary

#### **\*Correspondence:**

David Vaudry, INSERM U982, Laboratory of Neuronal and Neuroendocrine Differentiation and Communication, International Associated Laboratory Samuel de Champlain, University of Rouen, 76821 Mont-Saint-Aignan, Rouen, France.

e-mail: david.vaudry@univ-rouen.fr

High concentrations of pituitary adenylate cyclase-activating polypeptide (PACAP) and a high density of PACAP binding sites have been detected in the developing rat cerebellum. In particular, PACAP receptors are actively expressed in immature granule cells, where they activate both adenylyl cyclase and phospholipase C. The aim of the present study was to investigate the ability of PACAP to induce calcium mobilization in cerebellar granule neurons. Administration of PACAP-induced a transient, rapid, and monophasic rise of the cytosolic calcium concentration ([Ca2+]<sup>i</sup> ), while vasoactive intestinal peptide was devoid of effect, indicating the involvement of the PAC1 receptor in the Ca2<sup>+</sup> response. Preincubation of granule cells with the Ca2<sup>+</sup> ATPase inhibitor, thapsigargin, or the D-myo-inositol 1,4,5 trisphosphate (IP3) receptor antagonist, 2-aminoethoxydiphenyl borate, markedly reduced the stimulatory effect of PACAP on [Ca2+]<sup>i</sup> . Furthermore, addition of the calcium chelator, EGTA, or exposure of cells to the non-selective Ca2<sup>+</sup> channel blocker, NiCl2, significantly attenuated the PACAP-evoked [Ca2+]<sup>i</sup> increase. Preincubation of granule neurons with the N-type Ca2<sup>+</sup> channel blocker, ω-conotoxin GVIA, decreased the PACAP-induced [Ca2+]<sup>i</sup> response, whereas the L-type Ca2<sup>+</sup> channel blocker, nifedipine, and the P- and Q-type Ca2<sup>+</sup> channel blocker, ω-conotoxin MVIIC, had no effect. Altogether, these findings indicate that PACAP, acting through PAC1 receptors, provokes an increase in [Ca2+]<sup>i</sup> in granule neurons, which is mediated by both mobilization of calcium from IP3-sensitive intracellular stores and activation of N-type Ca2<sup>+</sup> channel. Some of the activities of PACAP on proliferation, survival, migration, and differentiation of cerebellar granule cells could thus be mediated, at least in part, through these intracellular and/or extracellular calcium fluxes.

**Keywords: pituitary adenylate cyclase-activating polypeptide, cerebellum, granule cells, calcium, cytoplasmic calcium stores, calcium channels, autoradiography**

#### **INTRODUCTION**

Pituitary adenylate cyclase-activating polypeptide (PACAP) is a 38- or 27-amino acid peptide that was isolated from hypothalamic extracts for its ability to stimulate cAMP formation in anterior pituitary cells (Miyata et al., 1989). PACAP belongs to a superfamily of peptides that originate from a common ancestral sequence and have evolved through exon and gene duplications. In particular, PACAP27 exhibits 68% identity with vasoactive intestinal peptide (VIP). The sequence of PACAP has been very well conserved during evolution, suggesting that it may exert vital physiological activities (Vaudry et al., 2009). Indeed, numerous studies have shown that PACAP is widely expressed throughout the body and regulates a large array of biological functions, both in the central nervous system and in peripheral organs (Reglodi et al., 2012).

Three PACAP/VIP receptors, which belong to the class B family of seven-transmembrane G protein-coupled receptors, have been cloned and termed respectively PAC1,VPAC1, and VPAC2, according to their relative affinity for PACAP and VIP (Harmar et al., 2012). PAC1 recognizes PACAP27 and PACAP38 with 1000-fold higher affinity than VIP while, VPAC1 and VPAC2 exhibit similar affinity for both PACAP and VIP (Cauvin et al., 1990). All three receptors can activate a variety of second messengers including cAMP, cGMP, IP3, or calcium, depending on the receptor types, splice variants, G proteins, and other intracellular components expressed by each cell type (Vaudry et al., 2009). This cellular context plays a crucial role in determining the effects of PACAP, as differential expression of PAC1 splice variants is sufficient to

<sup>2</sup> Institute for Research and Innovation in Biomedicine, University of Rouen, Mont-Saint-Aignan, France

**Abbreviations:** 2APB, 2-aminoethoxydiphenyl borate; Ca2+, calcium; [Ca2+]<sup>i</sup> , intracellular calcium concentration; EGTA, ethylene glycol-bis(β-aminoethylether)- N,N,N<sup>0</sup> ,N0 -tetraacetic acid; P, postnatal day; PACAP, pituitary adenylate cyclaseactivating polypeptide; PACAP27, 27-amino acid form of PACAP; PACAP38, 38-amino acid form of PACAP; VIP, vasoactive intestinal peptide.

trigger opposite activities on neuronal precursor proliferation (Nicot and DiCicco-Bloom, 2001).

It was initially shown that PACAP can increase calcium ion concentration in pituitary gonadotrope, somatotrope, and somatolactotrope cells (Canny et al., 1992; Gracia-Navarro et al., 1992; Matsuda et al., 2008). PACAP also induces calcium mobilization in chromaffin cells (Watanabe et al., 1992), hippocampal neurons (Tatsuno et al., 1992), folliculo-stellate cells (Yada et al., 1993), and type-2 astrocytes (Tatsuno and Arimura, 1994). Several transduction pathways are clearly involved in the effect of PACAP on calcium fluxes. For instance, in somatotrope cells, PACAP activates calcium mobilization in a cAMP- and protein kinase-A-dependent manner (Rawlings et al., 1993) while in gonadotrope cells, PACAP increases calcium through an inositol trisphosphate-dependent mechanism (Rawlings et al., 1994). Calcium mobilization induced by PACAP is essential for stimulation of acetylcholine (Masuo et al., 1993), catecholamine (Isobe et al., 1993), and insulin (Yada et al., 1994) release. Besides triggering neurosecretion, calcium influx also activates a variety of transcription factors that control gene expression involved in long-lasting changes of neuronal function. For instance, cAMP-dependent calcium mobilization is required for the ability of PACAP to regulate astrocyte differentiation (Cebolla et al., 2008).

In suprachiasmatic neurons, PACAP activates L-type calcium channel conductance by activating the MAPK pathway (Dziema and Obrietan, 2002). In this model, PACAP can either induce transient calcium mobilization (Dziema and Obrietan, 2002) or increase the amplitude and/or frequency of spontaneous calcium spikes (Michel et al., 2006). In cultured granule cells, some studies report that PACAP fails to increase calcium levels (Aoyagi and Takahashi, 2001) while others indicate that PACAP induces calcium mobilization (Mei, 1999). In cerebellar tissue slices, PACAP has no effect on the frequency of calcium transients but decreases the amplitude of calcium oscillations in granule cells (Cameron et al., 2007). Since PACAP exerts important neurodevelopmental and neuroprotective activities in the cerebellum (Botia et al., 2007), and since calcium plays a pivotal role in the control of cell proliferation, migration, and differentiation of granule cells (Sato et al., 2005), the aim of the present study was to investigate the ability of PACAP to induce calcium mobilization in cerebellar granule neurons and to characterize the channels involved.

# **MATERIALS AND METHODS**

#### **ANIMALS**

Wistar rats were obtained from a colony raised at the University of Rouen in an accredited animal facility (approval B.76-451-04), according to the French recommendations for the care and use of laboratory animals. All experiments were performed in accordance with the French Ministry of Agriculture and the European Communities Council Directive 2010/63/UE of September 22, 2010 (approval number N/01-12-11/24/12-14) under the supervision of authorized investigators (David Vaudry, Magali Basille-Dugay). Animals were housed in a temperature-controlled room (22 ± 1˚C) under a 12-h light/dark cycle and with *ad libitum* access to food and water. Pups, of both sexes, were killed by decapitation at postnatal day 8 (P8) and the cerebella were quickly dissected out.

#### **REAGENTS**

The 38-amino acid form of PACAP (PACAP38) was synthesized by solid phase methodology as previously described (Jolivel et al., 2009). PACAP27 and VIP were obtained from PolyPeptide Laboratories (Strasbourg, France). Fluo-3 acetoxymethylester (Fluo-3/AM) was from Molecular Probes (Invitrogen, Cergy Pontoise, France). Thapsigargin, ethylene glycol-bis(β-aminoethylether)-N,N,N<sup>0</sup> ,N0 -tetraacetic acid (EGTA), nifedipine, ω-conotoxin GVIA, and ω-conotoxin MVIIC were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France). 2-aminoethoxydiphenyl borate (2APB) was from Fisher Scientific (Illkirch, France).

# **CELL CULTURE**

Granule cell suspensions were prepared from cerebella of P8 rats, as previously described (Kaddour et al., 2013). For calcium experiments, dispersed cells were seeded on poly-l-lysine-coated glass coverslips in Falcon 3001 dishes at a density of 1 × 10<sup>6</sup> cells/mL in a cultured medium consisting of 75% DMEM and 25% Ham's F12 supplemented with 10% fetal bovine serum, 25 mM KCl, and 1% antibiotic antimycotic solution. Cells were grown at 37˚C in a humidified incubator with an atmosphere of 5% CO2/95% air. After 20 h, cytosine arabinoside, a mitotic inhibitor, was added at a final concentration of 5µM to avoid replication of non-neuronal cells. In these conditions, the culture mainly contains cerebellar granule cells and contaminant cells are mostly astrocytes, which can be recognized based on their morphology (Levi et al., 1989).

#### **CALCIUM MEASUREMENT**

Forty-eight hours after plating, granule cells were loaded at 37˚C for 45 min with 5µM fluo-3/AM diluted in culture medium. Then, the calcium-dye-probe was washed off and replaced with a Ringer's solution containing 10 mM Hepes, 120 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, and 6 mM glucose (pH 7.4). Under certain conditions, pharmacological agents were added in the Ringer's solution 30 min before mounting the glass coverslip on the stage of a Nikon inverted microscope (Eclipse TE200) equipped with a Nikon 60x, 1.4 NA, oil-immersion objective. Fluo-3/AM was excited with the 488-nm laser line, and the emitted fluorescence was recorded with a 500-nm long-pass filter on a Noran OZ confocal laser-scanning microscope (Noran Instruments, Middleton, WI, USA), equipped with a standard argon/kripton laser for illumination. Images were acquired as a time series of scan of the same focal plane (512 × 480 pixels at 7.5 images/s) and data processing was carried out using the Intervision software (Noran Instruments). The cells were continuously perifused with Ringer's solution at constant flow rate (2 mL/min) and temperature (37˚C). The perifusion system was also used to deliver test substances at the vicinity of the cultured neuroblasts. Any modification of fluorescence directly reflects a change of intracellular calcium concentration ([Ca2+]i). To determine the increase of [Ca2+]<sup>i</sup> , the value of the peak amplitude as well as the area under the curve (AUC) of each response profile were calculated by using the PRISM Software (GraphPad Software, San Diego, CA, USA). Results are expressed as histograms showing (i) the distribution of cells according to the amplitude of their response to PACAP and (ii) the mean AUCs of their [Ca2+]<sup>i</sup> increase.

#### **BINDING STUDIES**

PACAP27 was radioiodinated by means of the lactoperoxidase technique as previously described (Basille et al., 1994). The radioligand was purified by reverse-phase HPLC on a Vydac C<sup>18</sup> column (25 × 0.46 cm; Sigma-Aldrich), using a gradient of acetonitrile/water containing 0.1% TFA. The specific radioactivity of the tracer was approximately 800 Ci/mmol.

Twenty-µm-thick sections or granule cells from P8 rat cerebella cultured for 1 day were preincubated at 20˚C in 50 mM Tris buffer (pH 7.4) supplemented with 1% bovine serum albumin (BSA), 32 mM saccharose, 5 mM MgCl2, and 0.5µg/mL bacitracin, for 30 or 10 min, respectively. Sections or cells were incubated with [ <sup>125</sup>I]PACAP27 (40 or 400 pM, respectively) at 20˚C for 1 h in the same buffer, supplemented with 2% BSA. To visualize non-specific

**(D)** on [Ca<sup>2</sup><sup>+</sup>

]i

peptide application.


Ca<sup>2</sup><sup>+</sup>

in cultured granule cells. The arrows indicate the onset of

binding, slices, or cells were incubated with the radioligand in the presence of 10−<sup>6</sup> M PACAP38. At the end of the incubation, sections, or cells were washed with cold Tris buffer and dried under an air stream. Finally, slices were apposed onto Hyperfilm-3H (GE Healthcare, Les Ulis, France) for 6 days and the radioactivity associated with the cells was counted in a gamma-counter (LKB, Wallac, Rockville, MI, USA). Tissue slices were photographed by means of a computer-assisted image-analysis station (Samba, Grenoble, Lyon).

#### **STATISTICAL ANALYSIS**

Data are expressed as mean ± SEM values from at least three independent experiments. Statistical analyses were conducted with the PRISM software.

#### **RESULTS**

#### **EFFECT OF PACAP38, PACAP27, AND VIP ON [Ca2**+**]i**

Perifusion of cultured cerebellar granule cells with 10−<sup>6</sup> M PACAP38 provoked a marked increase of [Ca2+]<sup>i</sup> in more than 90% of the cells (**Figure 1A**). In all responding cells [Ca2+]<sup>i</sup> increased rapidly, i.e., within less than 5 s following PACAP infusion, and reached a maximum after 15–20 s, but the amplitude of the response was variable depending on the cell (**Figure 1A**). The response profile determined as the ratio between the fluo-3 fluorescence intensity under resting conditions and after exposure to test substances, revealed that both PACAP38 and PACAP27 (10−<sup>6</sup> M) induced a transient, rapid, and monophasic increase in [Ca2+]<sup>i</sup> followed by gradual return to baseline (**Figures 1B,C**) while application of vehicle had no effect. Conversely, administration of 10−<sup>6</sup> M VIP did not induce any modification of the [Ca2+]<sup>i</sup> level (**Figure 1D**). When the same cells were exposed to a second pulse of PACAP, the stimulatory effect of the peptide on [Ca2+]<sup>i</sup> was totally abolished (data not shown).

#### **CONTRIBUTION OF INTRACELLULAR Ca2**<sup>+</sup> **POOLS IN THE PACAP38-EVOKED INCREASE OF [Ca2**+**]i**

Preincubation of granule cells with the Ca2<sup>+</sup> ATPase inhibitor thapsigargin (10−<sup>6</sup> M; 15 min) induced a substantial reduction of the amplitude of the [Ca2+]<sup>i</sup> response to 10−<sup>6</sup> M PACAP38 in most cells (*p* < 0.001; **Figure 2A**). Indeed, in the absence of thapsigargin, more than 80% of the responding cells exhibited at least a 50% increase of their fluorescence intensity whereas in the presence of the inhibitor, 80% of the cells exhibited an increase of their fluorescence intensity that was lower than 30% (**Figure 2A**). As shown in **Figure 2B**,

thapsigargin reduced by 86% the mean AUC of the PACAP38 evoked [Ca2+]<sup>i</sup> response (*p* < 0.001). Preincubation of granule cells with the permeable d-myo-inositol 1,4,5-trisphosphate receptor antagonist 2APB (10−<sup>5</sup> M) markedly reduced the [Ca2+]<sup>i</sup> response to 10−<sup>6</sup> M PACAP38. In the presence of 2APB, more than 90% of responding cells exhibited an increase of their fluorescence intensity that was lower than 30% (*p* < 0.001; **Figure 2C**). As shown in **Figure 2D**, 2APB reduced by 89% the mean AUC of the PACAP38-evoked [Ca2+]<sup>i</sup> response (*p* < 0.001).

#### **CONTRIBUTION OF EXTRACELLULAR Ca2**<sup>+</sup> **IN THE PACAP38-EVOKED STIMULATION OF [Ca2**+**]i**

Preincubation of granule cells with the calcium chelator EGTA (6 mM; 10 min), significantly attenuated the amplitude of the [Ca2+]<sup>i</sup> response to 10−<sup>6</sup> M PACAP38 (*p* < 0.01; **Figure 3A**) and diminished by 91% the AUC of the PACAP38-evoked [Ca2+]<sup>i</sup> increase (*p* < 0.001; **Figure 3B**). Similarly, a 10-min preincubation with 3 mM NiCl2, a blocker of voltage-operated calcium channels (VOCCs), significantly reduced the amplitude of the [Ca2+]<sup>i</sup> peak in response to 10−<sup>6</sup> M PACAP38 (*p* < 0.001; **Figure 3C**). Indeed, while 90% of the responding cells had a ratio of fluorescence intensity greater than 130 in the absence of NiCl2, 79% of them had an amplitude lower than 130 in the presence of the VOCC blocker (**Figure 3C**). Furthermore, addition of NiCl<sup>2</sup> to the Ringer's solution reduced by 62% the AUC of the PACAP38-evoked [Ca2+]<sup>i</sup> response (*p* < 0.001; **Figure 3D**).

#### **ABSENCE OF EFFECT OF EGTA ON [125I]PACAP27 BINDING**

To verify that EGTA did not influence PACAP binding on its recognition sites, autoradiography experiments were carried out using [125I]PACAP27 as a radioligand. As previously reported [ <sup>125</sup>I]PACAP27 binding was observed in the external granule cell layer of P8 rat cerebella (**Figure 4A**). Addition of 10−<sup>6</sup> M PACAP38 to the incubation medium completely displaced [125I]PACAP27 binding (**Figure 4B**). In the presence of 6 mM EGTA, specific binding of [125I]PACAP27 on cerebellar tissue slices was not impaired (**Figure 4C**) and displacement of [125I]PACAP27 binding by PACAP38 was not affected (**Figure 4D**). Binding experiments were also performed on 1-day-old cultured granule cells with [125I]PACAP27 as a radioligand and it appeared that 6 mM EGTA did not modify the specific binding of [125I]PACAP27 on cultured cells (**Figure 4E**).

Autoradiographic visualization of [<sup>125</sup>I]PACAP27 binding sites in the absence **(A,C)** or presence of 10<sup>−</sup><sup>6</sup> M PACAP38 **(B,D)** in consecutive sections of P8 rat cerebellum. Scale bar = 1.7 mm. **(E)** Histograms showing specific binding of [<sup>125</sup>I]PACAP27 in the absence (black bars) or presence (white bars) of 6 mM EGTA on cultured cerebellar granule cells. EGTA, ethylene glycol-bis(β-aminoethylether)-N,N,N<sup>0</sup> ,N<sup>0</sup> -tetraacetic acid. NS, not significantly different from Hepes buffer.

#### **INVOLVEMENT OF N-TYPE CALCIUM CHANNELS IN THE PACAP38-EVOKED STIMULATION OF [Ca2**+**]i**

To determine which type of Ca2<sup>+</sup> channel is responsible for the stimulatory effect of PACAP38 on calcium influx, selective blockers of VOCCs were used. A 30-min incubation of cultured granule cells with the L-type Ca2<sup>+</sup> channel blocker nifedipine (10−<sup>5</sup> M; **Figures 5A,B**) or the P- and Q-type Ca2<sup>+</sup> channel blocker ωconotoxin MVIIC (10−<sup>6</sup> M; **Figures 5C,D**) did not significantly modify the amplitude of the [Ca2+]<sup>i</sup> response to 10−<sup>6</sup> M PACAP38 (*p* > 0.05; **Figures 5A,C**). Consistent with these observations, nifedipine, and ω-conotoxin MVIIC did not impair the AUCs of the PACAP38-induced [Ca2+]<sup>i</sup> rise (*p* > 0.05; **Figures 5B,D**). In contrast, preincubation of cells with the N-type Ca2<sup>+</sup> channel blocker ω-conotoxin GVIA (10−<sup>6</sup> M; 30 min; **Figures 5E,F**)

provoked a significant decrease of the amplitude of the [Ca2+]<sup>i</sup> increase induced by 10−<sup>6</sup> M PACAP38, with 76 and 52% of responding cells exhibiting a ratio of fluorescence intensity greater than 130 in the absence or presence of ω-conotoxin GVIA, respectively (*p* < 0.001; **Figure 5E**). Furthermore, ω-conotoxin GVIA reduced by 46% the AUC of the PACAP38-evoked [Ca2+]<sup>i</sup> response (*p* < 0.001; **Figure 5F**).

# **DISCUSSION**

The respective contribution of extracellular and intracellular Ca2<sup>+</sup> pools in PACAP-induced [Ca2+]<sup>i</sup> increase in cerebellar neuroblasts has not been previously investigated. Because primary cultures of rat cerebellar granule cells are mainly composed of a single population of cells, the effect of PACAP on the [Ca2+]<sup>i</sup> response could be monitored on a large number of neurons. Thus, we found that, in 90% of granule cells, infusion of 10−<sup>6</sup> M PACAP38 or PACAP27 induced a transient, rapid, and monophasic [Ca2+]<sup>i</sup> rise, similar to the PACAP-induced Ca2<sup>+</sup> response observed in neural NG108-15 cells (Holighaus et al., 2011) but different from that observed in rat pancreatic acinar AR42J cells (Barnhart et al., 1997) or human neuroblastoma NB-OK-1 cells (Delporte et al., 1993) in which the initial peak is followed by a plateau phase that lasts for approximately 2–3 min before the return of [Ca2+]<sup>i</sup> to baseline level. These differences could be explained by the diversity of the PACAP receptor subtype repertoire expressed in each cell type and by the different intracellular signaling systems involved. For instance, in rat gonadotrophs, PACAP stimulates Ca2<sup>+</sup> oscillations through activation of VPAC1 coupled to a PTX-insensitive G protein and phospholipase C-β (Hezareh et al., 1996) whereas, in cerebellar neurons, in which VIP failed to affect [Ca2+]<sup>i</sup> , the ability of PACAP to induce [Ca2+]<sup>i</sup> rise can be ascribed to activation of the PAC1 receptor, as previously reported in primary cultures of rat cortical neurons and astrocytes (Grimaldi and Cavallaro, 1999). In neural NG108-15 cells, the PACAP38-evoked [Ca2+]<sup>i</sup> response is mainly mediated by the PAC1 hop1 splice variant (Mustafa et al., 2007; Holighaus et al., 2011), an isoform also expressed in granule neurons (Kienlen-Campard et al., 1997). Interestingly, this hop isoform has been reported to be mandatory for calcium influx in cortical precursors (Yan et al., 2013). In β-islet cells, the PAC1 TM4 splice variant increases [Ca2+]<sup>i</sup> by stimulating Ca2<sup>+</sup> influx *via* a L-type Ca2<sup>+</sup> channel without activating adenylyl cyclase or phospholipase C in response to PACAP (Chatterjee et al., 1996). In cerebellar granule cells, the TM4 splice variant is expressed at a very low level (Chatterjee et al., 1996) and is thus probably not functionally relevant considering the strong activation of the cAMP pathway evoked by PACAP (Basille et al., 1993, 1995).

It has previously been demonstrated that, in cerebellar granule neurons, PACAP activates the phospholipase C pathway (Basille et al., 1995). We have thus explored the possible contribution of intracellular Ca2<sup>+</sup> stores in the stimulatory effect of PACAP on [Ca2+]<sup>i</sup> . The results showed that depletion of the endoplasmic reticulum Ca2<sup>+</sup> store with the Ca2<sup>+</sup> ATPase inhibitor thapsigargin reduced the amplitude and the AUC of the PACAP-evoked [Ca2+]<sup>i</sup> response in cerebellar neurons. Consistent with the notion that intracellular Ca2<sup>+</sup> from IP3-sensitive Ca2<sup>+</sup> pools could play an important role in the mechanism of action of PACAP, incubation of the cells with 2APB, a cell permeable d-myo-inositol

1,4,5-trisphosphate receptor antagonist, decreased the [Ca2+]<sup>i</sup> response to PACAP. Such a contribution of intracellular Ca2<sup>+</sup> stores to the PACAP-evoked [Ca2+]<sup>i</sup> response has already been reported in rat gonadotrophs (Rawlings et al., 1994), in rat acinar AR42J cells (Barnhart et al., 1997), and in human neutrophils (Harfi and Sariban, 2006).

Alongside, suppression of extracellular Ca2<sup>+</sup> by EGTA or exposure of cells to the non-selective Ca2<sup>+</sup> channel blocker NiCl2, also attenuated the stimulatory effect of PACAP38 on [Ca2+]<sup>i</sup> , indicating that Ca2<sup>+</sup> influx is also required for the transient phase of the [Ca2+]<sup>i</sup> increase. Recruitment of both intracellular and extracellular sources of Ca2<sup>+</sup> after activation of PAC1 receptors has already been reported in several models including the rat acinar cell line AR42J (Barnhart et al., 1997), the human neuroblastoma cell line NB-OK-1 (Delporte et al., 1993), and primary cultures of rat cortical neurons (Grimaldi and Cavallaro, 1999). To determine which type of Ca2<sup>+</sup> channel was responsible for the stimulatory effect of PACAP on calcium influx, selective blockers of VOCCs were used. Preincubation of granule cells with the N-type Ca2<sup>+</sup> channel blocker ω-conotoxin GVIA decreased the PACAP-evoked [Ca2+]<sup>i</sup> response, whereas the L-type Ca2<sup>+</sup> channel blocker nifedipine and the P- and Q-type Ca2<sup>+</sup> channel blocker ω-conotoxin MVIIC had no effect. These findings contrast with previous data showing the involvement of L-type Ca2<sup>+</sup> channels in the stimulatory effect of PACAP on human neutrophils (Harfi et al., 2005), on porcine somatotrope cells (Martinez-Fuentes et al., 1998) or on bovine adrenal chromaffin cells (Tanaka et al., 1996) and the implication of T-type Ca2<sup>+</sup> channels in mouse adrenal chromaffin cells (Hill et al., 2011). Nevertheless, other factors acting on cerebellar granule neurons, such as IGF-1, have already been shown to regulate N-type Ca2+-channels (Blair and Marshall, 1997).

Ca2<sup>+</sup> is an essential intracellular messenger required for numerous cellular functions (Carafoli et al., 2001). For instance, during development, modifications of intracellular Ca2<sup>+</sup> concentrations have been shown to regulate proliferation (Owens et al., 2000), migration (Komuro and Rakic, 1992), differentiation (Benquet et al., 2002; Ronn et al., 2002), and apoptosis (Turner et al., 2002) of immature neurons. Various studies suggest that, depending on the pool involved, Ca2<sup>+</sup> can exert different effects on immature neurons. For instance, in rat cortical neurons, Ca2<sup>+</sup> release from intracellular stores is implicated in neurite elongation, while Ca2<sup>+</sup> influx regulates dendritic branching (Ramakers et al., 2000). In the immature cerebellum, a transient elevation of intracellular Ca2<sup>+</sup> levels increases the migration rate of granule neuroblasts (Komuro and Rakic, 1996) and the thapsigargin-sensitive Ca2<sup>+</sup> store plays an essential role in growth and maturation of cerebellar granule cells (Yao et al., 1999). Even though the implication of Ca2<sup>+</sup> on granule cell development is now well established, the neurotrophic factors able to control Ca2<sup>+</sup> levels in cerebellar neuroblasts are not yet clearly identified. Nevertheless, peptides acting on G protein-coupled receptors are thought to be important mediators to control neurite outgrowth and growth cone guidance in cerebellar granule cells through a Ca2+-dependent pathway (Xiang et al., 2002).

Granule cells are, by far, the major population of interneurons in the cerebellum and they represent the main source of glutamate, the second most abundant population being GABAergic neurons (Voogd and Glickstein, 1998). Previous studies have demonstrated that plasticity of granule cells can be modulated by neuropeptides (Cote et al., 1999; Yacubova and Komuro, 2002). In particular, in cerebellar granule cells, PACAP has been shown to inhibit proliferation (Nicot et al., 2002), stop migration (Cameron et al., 2007), protect from apoptosis (Vaudry et al., 2000), and

#### **REFERENCES**


PACAP-38 causes phospholipase Cdependent calcium signaling in rat acinar cell line. *Surgery* 122, 465–474.

Basille, M., Gonzalez, B. J., Desrues, L., Demas, M., Fournier, A., and Vaudry, H. (1995). Pituitary adenylate cyclase-activating polypeptide (PACAP) stimulates adenylyl cyclase and phospholipase C activity in rat cerebellar neuroblasts. *J. Neurochem.* 65, 1318–1324.

promote differentiation (Gonzalez et al., 1997). *In vivo*, PACAP receptors are expressed by precursors of cerebellar granule cells in the external granule cell layer (Basille et al., 1993) and PACAP administration increases the number of mature neurons in the post-migratory internal granule cell layer (Vaudry et al., 1999). The implication of adenylyl cyclase and MAP kinases in the effects of PACAP have been extensively investigated (Villalba et al., 1997; Vaudry et al., 1998; Nicot et al., 2002) but the contribution of PACAP-induced Ca2<sup>+</sup> increase in maturation of cerebellar neuroblasts is still poorly understood. So far, it has only been shown that calcium is involved in the inhibitory effect of PACAP on granule cell migration (Cameron et al., 2007). However, it is well established that potassium depolarization-induced Ca2<sup>+</sup> entry is essential for granule cell survival (Gallo et al., 1987), suggesting that the antiapoptotic effect of PACAP may involve calcium mobilization. As reported with cortical neurons, PACAP may induce BDNF expression in a Ca2+-dependent manner to indirectly promote granule cell survival (Shintani et al., 2005; Kokubo et al., 2009). The stimulation of granule cell proliferation by calcium influx is suppressed by nifedipine but not by ω-conotoxin GVIA (Borodinsky and Fiszman, 1998), suggesting that the ability of PACAP to block granule cell division does not depend on calcium regulation. Consistent with this latter hypothesis, it has been shown that, in cortical neurons, PACAP inhibits cell proliferation through activation of the cAMP signaling pathway (Nicot and DiCicco-Bloom, 2001). Finally, potassium depolarization and NMDA treatment require calcium influx to induce neurofilament assembly in cerebellar granule cells (Bui et al., 2003), supporting the idea that PACAP-induced neurite outgrowth also requires calcium mobilization.

Altogether, these data indicate that some of the activities of PACAP on cerebellar granule cell proliferation, survival, migration, and differentiation involve, at least in part, intracellular and/or extracellular calcium mobilization, but further investigations are needed to decipher the precise role of calcium in each process.

#### **ACKNOWLEDGMENTS**

The authors want to thank Dr. Arnaud Arabo, Mrs. Huguette Lemonnier, and Mr. Donovan Liot for skillful technical assistance. David Vaudry and Hubert Vaudry are Affiliated Professors at the Institut National de la Recherche Scientifique – Institut Armand Frappier. This study was supported by grants from INSERM (U982), the Interreg TC2N project, the LARC-Neuroscience network, and the Région Haute-Normandie.


evidence for a stimulatory effect of PACAP on immature cerebellar granule cells. *Neuroscience* 57, 329–338.


channels in a PI 3-kinase-dependent manner. *Neuron* 19, 421–429.


J. (1993). Contrasting effects of PACAP and carbachol on [Ca2+]i and inositol phosphates in human neuroblastoma NB-OK-1 cells. *Peptides* 14, 1111–1118.


cyclase-activating peptide (PACAP) recruits low voltage-activated T-type calcium influx under acute sympathetic stimulation in mouse adrenal chromaffin cells. *J. Biol. Chem.* 286, 42459–42469.


hedgehog interact to control cerebellar granule precursor cell proliferation. *J. Neurosci.* 22, 9244–9254.


et al. (2005). Neuroprotective action of endogenous PACAP in cultured rat cortical neurons. *Regul. Pept.* 126, 123–128.


H. (1999). Neurotrophic activity of pituitary adenylate cyclaseactivating polypeptide on rat cerebellar cortex during development. *Proc. Natl. Acad. Sci. U.S.A.* 96, 9415–9420.


Voogd, J., and Glickstein, M. (1998). The anatomy of the cerebellum. *Trends Cogn. Sci. (Regul. Ed.)* 2, 307–313.


activating polypeptide (PACAP) increases cytosolic-free calcium concentration in folliculo-stellate cells and somatotropes of rat pituitary. *Peptides* 14, 235–239.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 03 April 2013; accepted: 25 April 2013; published online: 10 May 2013.*

*Citation: Basille-Dugay M, Vaudry H, Fournier A, Gonzalez B and Vaudry D (2013) Activation of PAC1 receptors in rat cerebellar granule cells stimulates both calcium mobilization from intracellular stores and calcium influx through Ntype calcium channels. Front. Endocrinol. 4:56. doi: 10.3389/fendo.2013.00056*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Basille-Dugay, Vaudry, Fournier, Gonzalez and Vaudry. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Brain neuropeptides in central ventilatory and cardiovascular regulation in trout

#### *Jean-Claude Le Mével <sup>1</sup> \*, Frédéric Lancien1, Nagi Mimassi <sup>1</sup> and J. Michael Conlon2*

<sup>1</sup> INSERM UMR 1101, Laboratoire de Traitement de l'Information Médicale, Laboratoire de Neurophysiologie, SFR ScInBioS, Faculté de Médecine et des Sciences de la Santé, Université Européenne de Bretagne, Université de Brest, CHU de Brest, Brest, France

<sup>2</sup> Department of Biochemistry, Faculty of Medicine and Health Sciences, United Arab Emirates University, Al Ain, United Arab Emirates

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Tobias Wang, University of Aarhus, Denmark Pei-San Tsai, University of Colorado, USA

#### *\*Correspondence:*

Jean-Claude Le Mével, INSERM UMR 1101, Laboratoire de Traitement de l'Information Médicale, Laboratoire de Neurophysiologie, SFR ScInBioS, Faculté de Médecine et des Sciences de la Santé, Université Européenne de Bretagne, Université de Brest, CHU de Brest, 22 avenue Camille Desmoulins, CS 93837, 29238 Brest Cedex 3, France. e-mail: jean-claude.lemevel@ univ-brest.fr

#### **INTRODUCTION**

In addition to classical neurotransmitters, numerous brain neuropeptides and their G-protein coupled receptors (GPCRs) have been identified in several cardiovascular and ventilatory nuclei (Fuxe et al., 1986). Considerable data have accumulated in the literature concerning the central cardiovascular actions of these neuropeptides in mammals but much less is known about the roles of central neuropeptides on ventilatory regulation (Niewoehner et al., 1983; Doi and Ramirez, 2008; Pilowsky et al., 2009). Since the central ventilatory system and the cardiovascular system share some nuclei and mutually interact (Niewoehner et al., 1983; Taylor et al., 1999, 2009; Mandel and Schreihofer, 2006; Dampney et al., 2008), it is crucial to determine the integrative role of neuropeptides on these two vital regulatory mechanisms. Fish are aquatic vertebrates that use their gills to breathe, and mammals are vertebrates that breathe using their lungs. Nevertheless, there are important similarities between fish and mammals in the neuroanatomical networks and nervous mechanisms that control the ventilatory and cardiovascular systems (Taylor et al., 1999, 2010a,b; Bolis et al., 2001). In addition, neuropeptides appeared very early during evolution and the primary structures of these peptides and their receptors have been

Many neuropeptides and their G-protein coupled receptors (GPCRs) are present within the brain area involved in ventilatory and cardiovascular regulation but only a few mammalian studies have focused on the integrative physiological actions of neuropeptides on these vital cardio-respiratory regulations. Because both the central neuroanatomical substrates that govern motor ventilatory and cardiovascular output and the primary sequence of regulatory peptides and their receptors have been mostly conserved through evolution, we have developed a trout model to study the central action of native neuropeptides on cardio-ventilatory regulation. In the present review, we summarize the most recent results obtained using this non-mammalian model with a focus on PACAP, VIP, tachykinins, CRF, urotensin-1, CGRP, angiotensin-related peptides, urotensin-II, NPY, and PYY. We propose hypotheses regarding the physiological relevance of the results obtained.

**Keywords: neuropeptides, brain, ventilatory variables, heart rate, blood pressure, evolution, fish**

generally well conserved from fish to mammals (Holmgren and Jensen, 2001). Furthermore, in fish as in mammals, the neuropeptidergic systems are frequently present within brain areas involved in cardiovascular and ventilatory functions, including the hypothalamus and the brainstem autonomic nuclei (Batten et al., 1990; Dampney et al., 2005). Consequently, we have developed a trout model to gain insight into the effects of exogenously administered synthetic replicates of endogenous neuropeptides on ventilatory and cardiovascular functions in trout.

In this review, we summarize the neuroanatomical and functional pathways involved in cardio-respiratory control in fish. We describe the trout model and report methods to study the ventilatory and cardiovascular responses to centrally administered neuropeptides. We briefly summarize the available information regarding the primary structures of the fish neuropeptides and the similarities with their mammalian counterparts, the neuroanatomical location of the neuropeptides and their receptors in the fish brains. The neuropeptides investigated in this programme are those whose primary structures are known in the trout and whose neuroananatomical distribution is well characterized. We describe the ventilatory and cardiovascular actions of these neuropeptides following their intracerebroventricular (ICV) injection and we briefly contrast these central effects with their actions following intra-arterial (IA) injection. Finally, we propose hypotheses relating to the potential mechanisms of actions and physiological significance of central neuropeptides in the brain of the trout.

**Abbreviations:** BRS, baroreflex sensitivity; CNS, central nervous system; CPG, central pattern generator; DVN, dorsal motor nucleus of the vagus; HR, heart rate; HRV, heart rate variability; IA, intra-arterial; ICV, intracerebroventricular; NPO, preoptic nucleus; PDA, dorsal aortic blood pressure; VF, ventilation frequency; VA, ventilation amplitude; VTOT, total ventilation.

# **NEURAL PATHWAYS REGULATING CARDIO-RESPIRATORY FUNCTIONS IN FISH**

The central control of cardiorespiratory functions in fish has been previously described (Taylor et al., 1999). In fish, the visceral sensory area in the medulla oblongata to which the chemoreceptor and baroreceptor afferent fibers project is homologous to the nucleus tractus solitarius (NTS) of higher vertebrates (Nieuwenhuys and Powels, 1983; Sundin et al., 2003a). The NTS is the site where the first synapse on the chemo- and baroreflexes takes place. Rhythmic ventilatory movements in fish are generated by a diffuse central pattern generator (CPG) whose activity is modulated by inputs from the peripheral chemoreceptors and also from higher brain centers, including the mesencephalon and the forebrain (Taylor et al., 1999). The CPG controls the activity of trigeminal Vth, facial VIIth, glossopharyngeal IXth, and vagal Xth motor nuclei all of which drive the breathing muscles (Taylor et al., 1999). There is a close association between the neural mechanisms controlling the ventilatory and the cardiovascular systems at the level of the medulla oblongata (Taylor et al., 1999). Furthermore, anatomical and functional links between the hypothalamus and the medullary cardio-respiratory centers in teleosts have been described (Ariëns-Kappers et al., 1936; Hornby and Demski, 1988). Electrical stimulation of hypothalamic sites in the goldfish *Carassius auratus* induces concomitant changes in ventilatory variables and heart rate (HR) (Hornby and Demski, 1988). Within the brainstem, the cardiac vagal pre-ganglionic neurons are located within the dorsal motor nucleus of the vagus (DVN). Some cardiac vagal pre-ganglionic neurons are also present in a more lateral position, probably constituting a primitive nucleus ambiguus. However, little is known regarding the neurotransmitters and/or neuropeptides and their receptors that permit integration of the various inputs at the level of the brainstem to control the final output motor impulses that ultimately govern the ventilatory and cardiovascular variables (Gilmour and Perry, 2007). In the brainstem of the dogfish *Squalus acanthias*, catecholamines regulate the electrical activity of respiratory neurons (Randall and Taylor, 1991). In the channel catfish, *Ictalurus punctatus*, glutamatergic pathways within the caudal part of the NTS are essential for the control of ventilation and studies in the shorthorn sculpin *Myoxocephalus scorpius* reveal that *N*-methyl-D-aspartate (NMDA) receptors mediate ventilatory frequency (VF), while kainate receptors mediate ventilatory amplitude (VA) (Sundin et al., 2003b; Turesson and Sundin, 2003; Turesson et al., 2010). In addition, it was shown that α-amino-3-OH-5-methyl-4 isooxazole-propionic-acid (AMPA) receptors located within the NTS control the parasympathetic activity to the heart and that NMDA and non-NMDA receptors are involved in the hypoxia activated sympathetic hypertension (Turesson et al., 2010). The hearts of teleost fish receive both a cholinergic vagal innervation and an adrenergic sympathetic supply (Taylor et al., 1999). Although humoral catecholamines increase HR after binding to β-adrenoreceptors (Wood and Shelton, 1980), the functional role of the nervous sympathetic system in teleost cardiac control is not clearly recognized (Burnstock, 1969; Taylor et al., 1999). At rest, the teleost heart is under strong inhibitory control mediated by the vagus nerve acting on muscarinic cholinergic receptors on the pacemaker cells (Laurent et al., 1983; Taylor, 1992; Taylor et al., 1999).

# **THE TROUT MODEL**

For the *in vivo* experiments, we use rainbow trout *Oncorhynchus mykiss* (body wt 240–270 g) of both sexes. The experiments were made on unanesthetized trout under controlled and constant levels of oxygen partial pressure in water (*P*wO2), pH and temperature, maintained at constant levels (*P*wO2 = 20 kPa; pH = 7*.*4 − 7*.*6; T = 10 − 11◦C). Experimental protocols were approved by the Regional Ethics Committee in Animal Experiments of Brittany, France (registration number: 07).

An overview of trout equipped with arterial and buccal catheters, electrocardiographic (ECG) leads, and the ICV guide is presented **Figure 1**. Examples of the recorded signals are also plotted on the figure. The ventilatory and cardiovascular signals are processed off-line with custom-made programs written in LabView 6.1 (Laboratory Virtual Instrument Engineering Workbench, National Instruments).

#### **THE VENTILATORY SIGNAL (FIGURE 1A)**

A flared cannula is inserted into a hole drilled between the nares such that its flared end is resting against the roof of the mouth. This cannula is used to record any changes in buccal ventilatory pressure (Holeton and Randall, 1967). Segments free of any movement artifacts on the ventilatory signal are selected and the VF and VA are determined. VF is calculated from the first harmonic of the power spectrum of the ventilatory signal using the fast Fourier transform (FFT) algorithm. VA is calculated from the difference between the maximal abduction phase and the maximal adduction phase for each of the ventilatory movements. The net effect of the changes in VF and VA on ventilation is determined according to the formula VTOT = VF × VA where VTOT is total ventilation. Thus, the overall ventilatory response is determined by the combined output of the VF and ventilatory stroke volume (by approximation VA).

# **THE BLOOD PRESSURE SIGNAL (FIGURE 1B)**

Catheterization of the dorsal aorta (Soivio et al., 1972) permits the recording of the dorsal aortic blood pressure (PDA) and injection of various compounds, including the neuropeptides. Blood is collected via this catheter for routine hematocrit determination and for measurement of the concentration of hormones in plasma. The pulsatile PDA enables the measurement of systolic blood pressure (SBP) and diastolic blood pressure (DBP). Mean PDA is calculated as the arithmetic mean between the SBP and the DBP. The detection of the SBP along the recordings together with the R-R tachogram permits the determination of the cardiac baroreflex response. The baroreflex has been evolutionary conserved from Agnatha (lamprey) to humans (Bagshaw, 1985). The baroreflex in fish, as in humans, is working spontaneously under baseline conditions and also responds to adverse blood pressure changes (Lancien and Le Mével, 2007; Karemaker and Wesseling, 2008). In fish, the baroreflex response is probably used as a mechanism to protect the delicate vasculature of the fish gill against high blood pressure (Sundin and Nilsson, 2002). We evaluate the cardiac baroreflex sensitivity (BRS) using both a time domain

method, the sequence method (Bertinieri et al., 1988; Lancien and Le Mével, 2007) and a frequency domain method, the cross spectral analysis technique (Parati et al., 2000; Lancien et al., 2011). In teleost fish (Lancien and Le Mével, 2007; Sandblom and Axelsson, 2011), as in mammals (Bertinieri et al., 1988), the parasympathetic nervous system plays a crucial role in the short term cardiac baroreflex response.

# **THE ELECTROCARDIOGRAPHIC (ECG) SIGNAL (FIGURE 1C)**

Two ECG AgCl electrodes are subcutaneously implanted ventrally and longitudinally at the level of the pectoral fins. After amplification, the ECG signal, which is very similar to the human ECG, displays its different waves (P, Q, R, S, T) (Satchell, 1991). The QRS complex is the largest deflection and the R waves are routinely measured to determine HR. The R-R interval of the ECG can be used to plot the tachogram and to quantify the heart rate variability (HRV) using either the FFT algorithm (for review see Task Force of the European Society of Cardiology, and the North American Society of Pacing and Electrophysiology, 1996) or the Poincaré plot (Brennan et al., 2001). HRV reflects modulation on a beat to beat basis of the cardiac sinus node activity by both limbs of the autonomic nervous system. The high frequency component of the HRV in humans reflects respiratory sinus arrhythmia and provides information primarily on the degree of vagal tone on the heart (Médigue et al., 2001). Interestingly, studies of HRV in teleost fish demonstrate that the parasympathetic nervous system is the main, or even the only, contributor to HRV (Altimiras et al., 1995; Le Mével et al., 2002; Grossman and Taylor, 2007). Nevertheless, the physiological significance of HRV in teleost fishes is poorly understood.

#### **THE ICV GUIDE**

Fish do not possess large and expanded cerebral hemispheres with a developed neocortex (Nieuwenhuys et al., 1998). Consequently, these animals offer the opportunity to insert directly, under stereomicroscopic guidance, a 25-gauge needle fitted with a PE-10 polyethylene catheter between the two habenular ganglia toward the third ventricle until its tip lies between the two preoptic nuclei (NPO) (Le Mével et al., 2009a). The method is rapid and accurate since no stereotaxic placement is needed. In addition, no post-injection confirmation of the injected site is required. The rationale for this ICV implantation between the two NPO is that neuropeptides, directly or after diffusion through the cerebrospinal fluid, can access sites which are known to be critical to ventilatory and cardiovascular control. In teleost fish, as in mammals, these are the hypothalamus and the brainstem (Hornby and Demski, 1988; Taylor et al., 1999; Dampney et al., 2008).

# **EFFECTS OF INTRACEREBROVENTRICULAR INJECTIONS OF NEUROPEPTIDES**

#### **PITUITARY ADENYLATE CYCLASE-ACTIVATING POLYPEPTIDE (PACAP) AND VASOACTIVE INTESTINAL PEPTIDE (VIP)**

PACAP and VIP belong to the secretin-glucagon superfamily of peptides (Sherwood et al., 2000). PACAP is found in two forms, a 38 amino-acid peptide (PACAP-38) and the C-terminally truncated 27 amino-acid peptide (PACAP-27). PACAP and VIP share sequence similarity and, in mammals, these peptides exert their actions by binding to three receptors, PAC1, VPAC1, and VPAC2 (Laburthe et al., 2007). Within the brains of mammals, PACAP and VIP are known to control multiple physiological functions including some cardiovascular and ventilatory processes (Wilson and Cumming, 2008; Vaudry et al., 2009).

PACAP and VIP appeared very early during evolution and the primary structure of these peptides and their receptors have been remarkably well conserved from fish to mammals (Wong et al., 1998; Sherwood et al., 2000; Montpetit et al., 2003). Within the central nervous system (CNS) of teleosts, PACAPand VIP-like immunoreactivities are localized mainly in neuronal perikarya of the diencephalon at the level of the NPO. Their fibers project not only into the adenohypophysis (Matsuda et al., 1997; Montero et al., 1998; Wong et al., 1998) but also toward many extra hypothalamo-hypophysial areas such as the mesencephalon and the medulla oblongata (Montero et al., 1998). These observations suggest that PACAP and VIP act not only as hypophysiotropic hormones (Montero et al., 1998; Wong et al., 1998) but also as neurotransmitters, and/or neuromodulators. In the goldfish, peripheral PACAP reduces food intake (Matsuda et al., 2006). We also demonstrated that trout PACAP-27 and trout VIP act on the CNS to increase ventilation (Le Mével et al., 2009b) and to reduce the cardiac BRS (Lancien et al., 2011).

After ICV injection, PACAP (25–100 pmol) evokes a dose- and time-dependent elevation of VF and VA. Consequently, the net effect of the peptide is a hyperventilatory response involving a gradual and significant dose-dependent increase in VTOT. The threshold dose for an effect of PACAP on VF is 100 pmol, but a significant effect on VA and VTOT is seen at 50 pmol and this latter effect is observed 15 min after the injection of the peptide. The actions of PACAP on the ventilatory variables are long-lasting since values have not returned to baseline levels by the end of the post-injection period of 25 min. The most pronounced action of PACAP is evoking hyperventilation through an increase in VA rather than VF. For instance, after 50 and 100 pmol PACAP this maximal change in VA, expressed as a percentage of the preinjection value, reaches about 100 and 200%, respectively, while the elevation of VF is only about 10 and 35% (Le Mével et al., 2009b).

Upon ICV injection, the effects of trout VIP on the ventilatory variables are quite different from those following ICV injection of PACAP. VIP does not produce a significant increase in VF and VA but nonetheless the resultant action of this peptide is a small, transient but significant elevation of VTOT at the highest dose tested. Moreover, statistical analysis of the results obtained following ICV injection indicates that the maximum increase in VF, VA, and VTOT after ICV injection of 100 pmol PACAP relative to the pre-injection values is about 2.5-fold higher than the maximum ventilatory effects of the same dose of VIP.

After ICV injection, only the highest dose of PACAP produces a weak, but significant, sustained increase in PDA. However, there is no significant change in HR. ICV injections of VIP do not cause any change in either PDA or HR. The greater action of PACAP on ventilation and blood pressure compared with VIP suggests that PACAP may bind preferentially to PAC1 receptors rather than to VPAC receptors. The lack of HR response to elevation of blood pressure suggests that the cardiac BRS is depressed following central PACAP. Compared with vehicle, ICV injections of PACAP and VIP (25–100 pmol) dose-dependently reduce the cardiac BRS to the same extent with a threshold dose of 50 pmol for a significant effect (Lancien et al., 2011).

In contrast to their ICV effects, IA injections of PACAP and VIP at doses of 25–100 pmol produce no change in the ventilatory variables. Peripherally injected PACAP does not cause any significant change either in PDA or in HR, but bolus peripheral injection of VIP produces a robust dose-dependent and sustained hypertensive response without any change in HR.

#### **NEUROPEPTIDE GAMMA (NPγ), NEUROKININ A (NKA) AND SUBSTANCE P (SP)**

The tachykinins are a family of biologically active peptides that are characterized structurally by the common carboxy-terminal pentapeptide sequence Phe-Xaa-Gly-Leu-Met-NH2. This Cterminally amidated sequence is of primary importance for the interaction with the tachykinin receptors (Conlon, 2004). In mammals, SP, NKA, NPγ, and neuropeptide K (NPK) are encoded by the single copy preprotachykinin A gene. Neurokinin B is derived from the preprotachykinin B gene while the preprotachykinin C gene encodes three peptides (hemokinin 1, endokinin C, and endokinin D) with limited structural similarity with SP (for references, see Conlon, 2004). The tachykinins exert their actions by binding to GPCRs that are widely distributed within vascular, endocrine and nervous tissues. SP is the preferential agonist of the NK-1 receptor, NKA along with NPγ and NPK are regarded as endogenous ligands of the NK-2 receptor, and NKB is the preferred agonist of the NK-3 receptor (Patacchini and Maggi, 2004). In mammals, there is strong evidence for the importance of CNS tachykinins in the control of respiration (Gray et al., 1999). In addition, central tachykinins are involved in cardiovascular regulation, neuroendocrine secretion, pain transmission, and in certain behavioral responses (Satake and Kawada, 2006).

Orthologs of the mammalian tachykinins have been isolated and structurally characterized in a wide range of tetrapod and non-tetrapod species (for references, see Conlon, 2004). In particular, SP (Jensen and Conlon, 1992), NKA (Jensen and Conlon, 1992), and NPγ (Jensen et al., 1993) have been purified from tissues of the rainbow trout *O. mykiss*. Neuroanatomical studies have revealed the presence of tachykinin-like immunoreactivity in neuronal cell bodies and fibers throughout the brains of several teleost fish, including the trout (Vecino et al., 1989; Batten et al., 1990; Holmqvist and Ekstrom, 1991; Moons et al., 1992) together with high density of tachykinin binding sites from the hypothalamus to the medulla oblongata (Moons et al., 1992). We recently demonstrated that, after ICV injection, exogenously administered trout tachykinins are differentially implicated in the neuroregulatory control of ventilation in trout (Le Mével et al., 2007).

Compared with ICV injection of vehicle, NPγ (25–100 pmol) evokes a gradual elevation of VF but a progressive dosedependent reduction of VA. Therefore, the net effect of the peptide is a hypoventilatory response involving a significant decrease in VTOT. The threshold dose for an effect of NPγ on VF, VA, and VTOT is 50 pmol and this is observed 15 min after the injection of the peptide (**Figure 2A**). Interestingly, in some trout, the ICV injection of 100 pmol NPγ was followed by a dramatic reduction in VA to near the noise level of the recording system for periods of 10–20 s, giving the appearance of an apneic response. All actions of NPγ on the ventilatory variables are of long duration, since parameters do not return to baseline values by the end of the recording period.

In contrast to the action of NPγ, the effects of SP(50–250 pmol) on ventilation are not dose dependent and only the highest dose of SP (250 pmol) produces a significant elevation of VF, a significant reduction of VA, and a resultant

significant decrease of VTOT. The changes in these parameters reach significance 10–15 min after ICV injection.

As with SP, the effect of NKA (50–250 pmol) on the ventilatory variables are relatively minor, with only the highest dose (250 pmol) producing a significant decrease in VA and an overall significant fall in VTOT. This action of NKA is of short duration with VA returning rapidly to baseline values.

None of the three tachykinin peptides produce significant changes in mean PDA or HR following ICV injection. Further studies are required to determine whether the central action of NPγ on ventilatory variables in trout involves interaction with a receptor that resembles the mammalian NK-2 receptor more closely than the NK1-receptor.

Because centrally controlled cardiorespiratory coupling contributes to HRV in teleost fish (Grossman and Taylor, 2007), we made the assumption that changes in the VF after central injection of NPγ and SP, but not NKA, also produce changes in HRV (Lancien et al., 2009). Compared to vehicle-injected trout, Poincaré plot analysis of HRV demonstrates that ICV injection of NPγ dose-dependently increases HRV. SP evokes a significant elevation of HRV but only at the highest dose (250 pmol). In contrast, NKA is without any effect on HRV. The physiological significance of HRV in teleost fish is poorly understood. Recent studies favor the hypothesis that HRV may be an important component of the mechanisms optimizing the efficiency of respiratory gas exchange over the counter-current at the gill lamellae (Grossman and Taylor, 2007). Taken together, our data are consistent with a possible selective central action of NPγ on neuronal

networks implicated in the control of cardiorespiratory coupling in teleost fish.

IA injections of NPγ, SP or NKA at doses of 50–250 pmol produce no change in any of the ventilatory variables. However, all three tachykinins at their highest dose of 250 pmol cause a significant increase in mean PDA and, except for NPγ, a concomitant and significant fall in HR.

#### **CORTICOTROPIN-RELEASING FACTOR (CRF) AND UROTENSIN-I (U-I)**

CRF, a 41-amino acid peptide originally isolated from ovine hypothalamus (Vale et al., 1981), plays a key role in regulating the release of adrenocorticotropic hormone from the pituitary during stress. In mammals, CRF and the CRF-related peptide urocortin 1, an ortholog of the fish U-I (Vaughan et al., 1995; Barsyte et al., 1999), are also known to play a crucial neurotropic role in the CNS in coordinating the autonomic and behavioral responses to stressful situations (Koob and Heinrichs, 1999). In mammals, the actions of the CRF-family peptides are mediated by two types of G-protein-coupled receptors: CRF type 1 receptor (CRF-R1) and CRF type 2 receptor (CRF-R2) (Bale and Vale, 2004). CRF, urocortin-1, and the non-mammalian CRF-related peptides U-I, and sauvagine (SVG) bind with similar affinity to CRF-R1 while CRF has only a low affinity for CRF-R2. U-I, SVG, and the urocortins bind with high affinity to CRF-R2. In mammals including humans, CRF immunoreactivity (Swanson et al., 1983) and CRF receptors (Bale and Vale, 2004) are widely distributed in brain areas involved in the control of cardiovascular regulation and breathing movements. After ICV administration, CRF and urocortin 1 induce marked changes in cardiovascular variables (Parkes et al., 2001) and CRF acts centrally to produce a strong stimulatory effect on ventilatory movements (Bennet et al., 1990).

The CRF family of peptides and their receptors are of ancient origin (Chang and Hsu, 2004). In teleost fish, CRF, U-I, CRF receptors, and CRF binding protein are present not only in neurons of the preoptic region and hypothalamus (Olivereau and Olivereau, 1990) but also in extra-hypothalamic brain regions including the telencephalon and the posterior brain (Batten et al., 1990; Bernier et al., 1999a; Lovejoy and Balment, 1999; Alderman et al., 2008). Taken together, these neuroanatomical findings raise the possibility that CRF and U-I in teleosts also exert extrahypothalamo-hypophyseal actions and mediate some autonomic and/or behavioral effects within the brain. In fact, physiological data have indicated that, after ICV injection, CRF and U-I are implicated in the autonomic regulation of the cardiovascular system (for review see Le Mével et al., 2006), the control of locomotor activity (Clements et al., 2002) and in the regulation of food intake (for review see Bernier, 2006). Our results demonstrate that CRF and U-I also produce a potent hyperventilatory response when injected centrally in trout (Le Mével et al., 2009a).

After ICV injection, trout CRF (1–10 pmol) evokes both gradual and dose-dependent elevations of VF and VA. The net effect of the peptide is, therefore, a hyperventilatory response involving a significant dose-dependent elevation in VTOT. The minimum dose to elicit a statistically significant response in both ventilatory variables is 5 pmol and this is observed 15 min after the injection of the peptide. In contrast to the sustained action of ICV injection of CRF on VF and VA, a significant stimulatory action of trout U-I (1–10 pmol) on these two variables appears only after ICV injection of the highest dose of peptide tested. VTOT also indicates that the significant hyperventilatory action of U-I is delayed by 10 min (U-I, 5 pmol) and 5 min (U-I, 10 pmol) compared to corresponding doses of CRF. Moreover, the maximum increase in VTOT after ICV injection of CRF relative to the preinjection value is 2-fold higher than the hyperventilatory effect of U-I during the 25–30 min post-injection period.

At the dose of 5 pmol, only CRF transiently increases PDA, but a clear sustained hypertension is observed for the highest dose of 10 pmol of CRF and U-I. ICV injection of either CRF or U-I has no significant effect on HR for all doses tested.

ICV administration of alpha helical CRF9−<sup>41</sup> (ahCRF9−41) alone (50 pmol) does not affect the baseline ventilatory and cardiovascular variables. However, pre-treatment of the trout with this CRF antagonist at a dose ratio of ahCRF9−41: CRF of 10:1 delays and significantly reduces (by at least 3-fold) the CRFinduced increase in VF, VA, and VTOT and inhibits CRF-induced elevation in PDA. In fish, the pharmacological characteristics of the CRF receptors are quite different from their mammalian counterparts (see above). In catfish, where a third CRF receptor (CRF-R3) has been identified (Arai et al., 2001), CRF-R1 binds CRF, U-I, and SVG with similar affinity, while CRF-R2 preferentially binds SVG. CRF-R3 binds CRF with a 5-fold higher affinity than U-I and SVG (Arai et al., 2001). Pohl et al. (2001) concluded that, in Chum salmon, neither CRF-R1 nor CRF-R2 could discriminate between CRF and U-I. The lack of an intrinsic effect of ahCRF9−<sup>41</sup> when injected centrally suggests that endogenous CRF and U-I are not involved in the regulation of VA and VF in baseline situations. The fact that this antagonist significantly reduces the central hyperventilatory effects of exogenous CRF is indicative of a selective receptor-mediated hyperventilatory action of CRF in the brain of the trout. However, the type of CRF receptor involved cannot be determined at this time.

After IA injection, CRF and U-I are devoid of any ventilatory or cardiovascular activities except a transient increase in blood pressure at the highest dose of U-I (50 pmol).

#### **CALCITONIN GENE-RELATED PEPTIDE (CGRP)**

The 37-amino- acid peptide CGRP is derived from the tissuespecific splicing of the primary transcript of the calcitonin gene (Amara et al., 1982). CGRP is thus a member of the calcitonin/CGRP peptide family that includes adrenomedullin (AM), adrenomedullin-2 (or intermedin), amylin, and calcitonin receptor-stimulating peptide (Ogoshi et al., 2006; Sawada et al., 2006). CGRP binds to a seven transmembrane G-protein-coupled calcitonin receptor-like receptor that is complexed with one of three receptor activity-modifying proteins (Tam and Brain, 2006). In mammals, CGRP and its receptors are widely distributed throughout the peripheral and central CNS. In the CNS, CGRP acts as a neurotransmitter and/or neuromodulator involved in multiple physiological and behavioral processes including the hypothalamic regulation of feeding (Krahn et al., 1984). In addition, CGRP regulates the local vasodilation of cerebral vessels contributing to the pathophysiology of migraine headache and the peptide modulates pain responses at the level of the spinal cord (Tam and Brain, 2006). Central CGRP also plays a role in the autonomic regulation of the cardiovascular system. In contrast to its hypotensive effect in the periphery, ICV injection of CGRP produces a hypertensive response by activating the sympathetic nerves in rats (Fisher et al., 1983) and CGRP augments the baroreflex controls of renal sympathetic nerve activity and HR in the unanesthetized rabbit (Matsumura et al., 1999).

CGRP has an ancient evolutionary history. In fish, cDNAs encoding for CGRP have been isolated from a number of species (Jansz and Zandberg, 1992; Clark et al., 2002; Ogoshi et al., 2006; Martinez-Alvarez et al., 2008) and CGRP mRNA is expressed in peripheral and central tissues (Clark et al., 2002; Martinez-Alvarez et al., 2008). Moreover, the primary sequence of the peptide has been highly conserved from fish to humans (Shahbazi et al., 1998). As in many cerebral regions, the hypothalamus expresses CGRP mRNA (Martinez-Alvarez et al., 2008) and some CGRP-like immunoreactive fibers represent ascending projections from brainstem areas involved in autonomic functions (Batten and Cambre, 1989; Batten et al., 1990). Interestingly, in the goldfish *Carassius auratus* and in the puffer fish *Fugu rubripes* (Clark et al., 2002) the strongest expression of calcitonin/CGRP transcripts was observed in the posterior brain at the level of autonomic nuclei and spinal cord. In addition, CGRP receptors are present within the brain and heart of the flounder *Paralichthys olivaceus* (Suzuki and Kurokawa, 2000). Collectively these neuroanatomical data support a role for CGRP not only in neuroendocrine function and behavior but also in autonomic and cardiovascular regulation in fish. The anorexigenic action of centrally administered CGRP in the goldfish *Carassius auratus* has been previously described (Martinez-Alvarez et al., 2009). The cardio-ventilatory actions of centrally administered trout CGRP in trout has been recently described (Le Mével et al., 2012).

ICV administration of CGRP (1–50 pmol) evokes a dose- and time-dependent elevation of VF and VA. As a result, the net effect of the peptide is a hyperventilatory response involving a gradual and significant dose-dependent increase in VTOT. The threshold dose for an effect of CGRP on VF is 50 pmol but only 5 pmol for VA (**Figure 2B**). As for many neuropeptides, the actions of CGRP on these ventilatory variables are long-lasting since values had not returned to baseline levels by the end of the post-injection period of 25 min. This observation suggests that CGRP may act as a long-term hyperventilatory peptide *in vivo*. The most pronounced action of CGRP is evoking hyperventilation through an increase in VA instead of VF. For instance at a dose of 50 pmol, during the 15–20 min post-injection period when VTOT is maximal and increased by 300% from baseline value, the change in VA, expressed as a percentage of pre-injection value, is more than 200% while the elevation of VF is only about 30%.

After ICV injection, CGRP produces a significant dosedependent and sustained increase in PDA but the increase in HR does not reach the level of statistical significance. The receptor(s) mediating the ventilatory and cardiovascular action of CGRP in trout have not been determined. In eel, the paralogs AM2 and AM5 exhibit different central cardiovascular responses suggesting that they may act through different receptors (Ogoshi et al., 2008).

In contrast to its ICV effects, IA injections of CGRP at doses of 5–50 pmol produce no change in VF, VA, or VTOT. Nonetheless, peripherally injected CGRP causes an overall robust, dose-dependent and sustained hypertensive response without any change in HR. IA injection of the highest dose of CGRP causes at first a rapid but transient decrease in PDA followed by a hypertensive phase that does not return to the pre-injection level until 60 min.

# **ANGIOTENSIN PEPTIDES**

Data from mammalian studies have demonstrated that angiotensin II (Ang II) and angiotensin III (Ang III) are the two main effector peptides of the brain renin-angiotensin system (RAS). However, angiotensin IV (Ang IV) and to a lesser extend angiotensin 1–7 (Ang 1–7) are also implicated in various physiological functions, particularly body fluid homeostasis and cardiovascular regulation (Paul et al., 2006; Fyhrquist and Saijonmaa, 2008). Ang II and Ang III bind to angiotensin receptor type 1 (AT1) and type 2 (AT2). Ang IV binds exclusively to angiotensin receptor type 4 (AT4). The type of receptor that mediates the actions of Ang 1–7 is somewhat controversial. Studies on the effects of the RAS on ventilation are limited and only the action of Ang II has been explored in mammals. In both anaesthetized and unanaesthetized dogs (Potter and McCloskey, 1979; Ohtake and Jennings, 1993) and in unanaesthetized rabbits (Potter and McCloskey, 1979) but not in unanaesthetized Sprague-Dawley rats (Walker and Jennings, 1996), Ang II stimulates ventilation through a central mechanism that is independent of baroreceptor or chemoreceptor stimulation (Potter and McCloskey, 1979). In spontaneous hypertensive rats (SHR), but not in normotensive control Wistar-Kyoto rats, intravenous injection of the Ang II receptor antagonist, saralasin, has a depressant action upon ventilation (O'Connor and Jennings, 2001). Because SHR rats exhibit high brain RAS activity compared with normotensive control Wistar-Kyoto rats, the authors speculated that central Ang II is involved in the control of respiration only in SHR rats. However, in anaesthetized Sprague-Dawley rats, intracisternal ANG II provokes a decrease that becomes less when the doses of Ang II are increased (Aguirre et al., 1991) and ICV injection of saralasin reduces respiratory rate and respiratory rate variability in Wistar rats (Olsson et al., 2004). These data indicate that the brain RAS plays a role in the control of ventilation in mammalian species. In humans, Ang II may be implicated in the regulation of the respiratory sensitivity during pregnancy but the mechanism involved in this effect has not been elucidated (Wolfe et al., 1998).

The RAS has an ancient evolutionary history and most of its components are present in lampreys, elasmobranchs, and teleosts (Olson, 1992; Takei et al., 1993; Nishimura, 2001; Rankin et al., 2004; Wong and Takei, 2011). In contrast to the well known peripheral cardiovascular and osmoregulatory hormonal actions of Ang II (Olson, 1992; Le Mével et al., 1993; Bernier et al., 1999b; Takei and Balment, 2009), studies in fish on the central action of Ang II are sparse. Furthermore, two Ang II isoforms [Asn1] and [Asp1]-Ang are present in plasma and tissues (Conlon et al., 1996; Wong and Takei, 2012) but the physiological roles of the latter form have only been recently explored (Lancien et al., 2012). Central administration of [Asn1]-Ang II into the third or fourth ventricle of the eel *Anguilla japonica* induces drinking (Kozaka et al., 2003) and increases HR and blood pressure (Nobata et al., 2011). This procedure elevates HR and blood pressure but reduces both HRV and the cardiac BRS sensitivity in the trout (Le Mével et al., 1994, 2002, 2008b; Lancien et al., 2004b; Lancien and Le Mével, 2007) In addition, local injection of [Asn1]-Ang II within the DVN of the trout potently enhances HR but only weakly increases blood pressure (Pamantung et al., 1997). Taken together, these results demonstrate that in the brains of teleosts, as in mammals, Ang II may act as a neuromodulator or a neurotransmitter involved in key osmoregulatory and cardiovascular regulations. Recently, the cardio-ventilatory actions of exogenously administered [Asn1]-Ang II, [Asp1]-Ang II, Ang III, Ang IV, and Ang 1–7 within the third ventricle of the trout brain have been described (Lancien et al., 2012). In addition, the angiotensin peptides produced in the brain and circulating in plasma of trout were characterized using a high performance liquid chromatography (HPLC) system that can separate these peptides (Lancien et al., 2012; Wong and Takei, 2012).

After ICV injection (5–50 pmol), [Asn1]-Ang II and [Asp1]- Ang II gradually elevate VTOT through a selective stimulatory action on VA. However, the hyperventilatory effect of [Asn1]-Ang II is 3-fold higher than the effect of [Asp1]-Ang II at the 50 pmol dose. Ang III, Ang IV, and Ang 1–7 (25–100 pmol) are without effect on the ventilatory variables. In addition, both Ang II peptides and Ang III dose-dependently increase PDA and HR. These results suggest that the N- and C-terminal amino acid residues of Ang II are important for full effect on the central receptor(s) that mediate(s) hyperventilation and cardiovascular actions. It was previously proposed that in trout, [Asn1]-Ang II was the product of angiotensinogen cleavage in plasma but that this peptide is converted to [Asp1]-Ang II by plasma asparaginase (Conlon et al., 1996). In eel plasma, asparaginase activity is low and the conversion seems to occur in the tissues such as liver and kidney with angiotensinogen, not Ang II, as substrate (Wong and Takei, 2012). In brain tissue, comparable amounts of [Asn1]-Ang II and [Asp1]-Ang II were detected (ca. 40 fmol/mg brain tissue) but Ang III was not present, and the amount of Ang IV was about 8-fold lower than the content of the Ang II peptides. In plasma, Ang II peptides were also the major angiotensins (ca. 110 fmol/ml plasma), while significant but lower amounts of Ang III and Ang IV were present. These results demonstrate that the two Ang II peptides are present in trout plasma and brain tissue and suggest that the conversion Asn<sup>1</sup> <sup>→</sup> Asp1occurs not only in plasma but also in brain. It has been proposed that the teleost AT receptor is an AT1-like receptor (Russell et al., 2001). The demonstration that both Ang II peptides and Ang III elevate PDA and HR while the other angiotensins were without action supports the idea that an AT1-like receptor might also be involved in the central cardiovascular actions of Ang II and Ang III. A novel receptor that binds specifically or with a higher affinity to the Ang II peptides, but not the truncated forms, might mediate the ventilatory effect of the brain RAS in trout.

Within the brain of the trout, [Asn1]-Ang II affects not only the mean HR but also the beat to beat change in R-R intervals of the ECG since ICV injection of the peptide reduces HRV (Le Mével et al., 2002) and the cardiac BRS (Lancien and Le Mével, 2007).

None of the angiotensin peptides injected peripherally alter any of the ventilatory variables but the two Ang II isoforms and to a lesser extent Ang III provoke a pressor response. The concomitant decrease in HR following the IA injections of these angiotensins is not significant. In addition, Ang IV and Ang 1–7 are without effect on the cardiovascular variables. Collectively, these results support the view that the N- and C-terminal residues of the Ang II peptides play a role in optimal interaction with the putative cardiovascular angiotensin receptor in trout vascular tissue (Nishimura, 2001).

# **UROTENSIN II (U-II)**

U-II is a cyclic neuropeptide that was originally isolated from the caudal neurosecretory system of the teleost fish *Gillichthys mirabilis* on the basis of its smooth muscle-stimulating activity (Pearson et al., 1980; Bern et al., 1985). U-II is widely expressed in peripheral and nervous structures of species from lampreys to mammals including humans (Vaudry et al., 2010). It has now been demonstrated that U-II belongs to a family of structurally related peptides that include U-II and the UII-related peptides (URPs), URP, URP-1, and URP-2. In the teleost lineage, four U-II/URP paralogs are present but only two of these ancestral genes, U-II and URP, are found in tetrapods (Quan et al., 2012). U-II, URP, and URP isoforms exhibit the same cyclic hexapeptide core sequence (Cys-Phe-Trp-Lys-Tyr-Cys) while the N- and C-terminal regions are highly variable (Lihrmann et al., 2006; Conlon, 2008). Studies on UII/URP/URP-2 gene expression in teleosts and tetrapods suggest that U-II, URP, and URP-2 exert different functions (Parmentier et al., 2011). In teleost fish URP, URP-1, and URP-2 mRNA occur both in brain and spinal cord (Parmentier et al., 2008; Nobata et al., 2011) but in the eel, *Anguilla japonica* the U-II gene is exclusively expressed in the urophysis (Nobata et al., 2011). In tetrapods, the U-II gene is expressed primarily in motoneurons of the brainstem and spinal cord (Vaudry et al., 2010). U-II has been identified as a specific natural ligand of the orphan, G-protein-coupled receptor GPR14 (now renamed the UT receptor) in mammals (Vaudry et al., 2010) and in teleost fish (Lu et al., 2006). U-II and URP both activate the UT receptor with the same potency. The cardiovascular effects of centrally administered U-II and URP in trout and eel have been analyzed. In trout and eel, only a relatively high dose of U-II (500 pmol) evokes an increase in PDA with variable action on HR (Le Mével et al., 1996; Nobata et al., 2011). In addition, the central vasopressor action of URP in the eel is equally efficacious but less potent than the action of U-II (Nobata et al., 2011). The brain structures controlling the ventilatory system in trout seem to be more sensitive to the central action of U-II as only a 50 pmol dose of the peptide produces an hyperventilatory response through a significant increase in VF and VA (Lancien et al., 2004a). At this dose, U-II produces a long-lasting increase in locomotor activity (Lancien et al., 2004a). The effects of central URP and URP isoforms on the ventilatory and cardiovascular systems in trout have not yet been determined.

IA injection of U-II and URP in trout and eel evokes an elevation in PDA. In both species, the hypertensive effect of U-II is longer-lasting than that of URP (Le Mével et al., 2008a; Nobata et al., 2011). In trout, U-II only provokes a dose-dependent bradycardia (Le Mével et al., 1996, 2008a), while in the eel, U-II and URP significantly increase HR (Nobata et al., 2011). U-II is devoid of ventilatory actions following systemic injection in trout (unpublished observations).

# **NEUROPEPTIDE TYROSINE (NPY) AND PEPTIDE TYROSINE TYROSINE (PYY)**

NPY and PYY are two members of the pancreatic polypeptide family of regulatory peptides. These two 36-amino acid peptides contain a tyrosine residue at their N- and C-termini. NPY is the most abundant peptide within the CNS of mammals. The peptide and its GPCRs, designated Y1, Y2, Y4, Y5, and Y6, are widely distributed in nerve terminals throughout the brain (Dumont and Quirion, 2006). NPY-like immunoreactivity has been demonstrated in many noradrenergic and adrenergic neurons of the medulla oblongata (Fuxe et al., 1986). NPY-containing cell bodies are found in the lateral hypothalamus and NPY innervation of the paraventricular nucleus (PVN) of the hypothalamus arises from the medulla oblongata (Dumont and Quirion, 2006). In mammals, including rat, mouse, sheep, pig, rabbit, and pigeon, NPY has many neuroendocrine regulatory effects within the brain including an orexigenic action and regulation of food intake, anxiety, circadian rhythms, and memory. The central actions of NPY on cardiorespiratory functions remain unclear due to the fact that activation of different NPY receptors have opposite effects on cardiorespiratory variables (Fuxe et al., 1983; Scott et al., 1989; Morton et al., 1999). NPY is also present in peripheral organs, notably in blood vessels and heart. PYY is primarily located in endocrine cells of the lower intestine. In mammals, these two peptides are also involved in peripheral vasoregulation (Zukowska-Grojec et al., 1987; Playford et al., 1992).

NPY and PYY are present in both peripheral and brain tissues in fish (Jensen and Conlon, 1992; Danger et al., 1991). Seven NPY receptors subtypes (Y1, Y2, Y4–Y8) bind both NPY and PYY in fish (Salaneck et al., 2008). ICV injection of NPY in goldfish increases feeding (Volkoff et al., 2009). In trout, ICV administration of trout NPY and PYY at doses up to 100 pmol does not have any effect on ventilatory and cardiovascular variables (unpublished data). These cardiovascular results are consistent with a previous study demonstrating that ICV injection of human NPY (0.6–0.8 nmol) in trout exerts only a weak hypertensive action without any change in HR (Le Mével et al., 1991). These results obtained in trout suggest that, contrary to the actions of the other peptides mentioned in this review, NPY and PYY do not appear to have an important role in the central cardiorespiratory regulation in trout. In contrast, cod NPY causes vasodilation in the cod celiac artery (Shahbazi et al., 2002). In the elasmobranchs, the unanesthetized *Scyliorinus canicula* (Conlon et al., 1991) and the anesthetized *Heterodontus portjacksoni* (Preston et al., 1998), IA or intravenous injection of relatively high doses of dogfish NPY or PYY significantly increase blood pressure. However in trout, IA injection of trout NPY and PYY at a 100 pmol dose is devoid of significant cardiovascular effects (unpublished data). These differences between the effects of NPY and PYY in elasmobranchs and teleosts can possibly be explained by the different experimental protocols used or that the location of NPY/PYY cardiovascular receptors in the cardiovascular systems of elasmobranchs differs from that in teleost fishes.

# **POSSIBLE MECHANISMS OF ACTION AND PHYSIOLOGICAL SIGNIFICANCE**

In order to produce changes in ventilatory and cardiovascular variables, ICV injections of neuropeptides must access receptors critical for the control of cardio-ventilatory motor neurons. However, the receptor site(s) initiating cellular transduction mechanisms cannot be deduced from the experiments in which the peptides are injected into the third cerebral ventricle. Nevertheless, neuroanatomical prerequisites and some neurophysiological data exist that might explain the ventilatory and cardiovascular responses to ICV neuropeptides. Since the neuropeptides are injected in close proximity to a major neuroendocrine hypothalamic nucleus, the NPO, it is reasonable to assume that these exogenous neuropeptides may mimic the action of the endogenous peptides after release from neurons belonging to this nucleus. These neuropeptides can then activate arginine vasotocin (AVT) and isotocin (IT) preoptic neurons. AVT and IT preoptic neurons project to the neurohypophysis where the two nonapeptides are released into the general circulation. AVT is well known to increase vascular tone and elevate blood pressure *in vivo* (Le Mével et al., 1996; Conklin et al., 1997). In addition to this neuroendocrine pathway, projection from the preoptic neurons could influence brainstem respiratory and cardiovascular neurons including the NTS and the DVN through the neurogenic route by the release of AVT, IT or other neuropeptides or classical neurotransmitters (Batten et al., 1990; Saito et al., 2004). In the goldfish, functional-anatomical studies have demonstrated the existence of a neural pathway from the preoptic area to the DNV controlling concomitantly ventilation and HR (Hornby and Demski, 1988). In mammals, stimulation of the

**FIGURE 3 | A model based on a parasagital view of the CNS of the trout depicting the potential central sites and pathways for the effects of intracerebroventricular administered neuropeptides (NPs) on central ventilatory and cardiovascular functions.** Target ventilatory and cardiovascular tissues are also shown. Projecting fibers from preoptic nucleus (NPO) neurons to brainstem ventilatory and cardiovascular nuclei and to spinal sympathetic neurons (sn) are shown with a bold hatched line. Motor outputs from ventilatory and cardiovascular central nuclei to peripheral effectors are shown in continuous line. Feedback information from peripheral tissues to CNS nuclei is shown using thin hatched lines.

The sites and pathways described are highly schematic and speculative (see also text in section 2 and 5 for further explanations). Other abbreviations: AVT, arginine-vasotocin; Cb, cerebellum; CPG, central pattern generator; CNSS, caudal neurosecretory system; De, diencephalon; DVN, dorsal motor nucleus of the vagus; Hy, hypothalamus; MC, massa caudalis; Me, mesencephalon; NTS, nucleus tractus solitarius; OB, olfactory bulb; ON, optic nerve; OT, optic tectum; Pit, pituitary gland (hypophysis); rmn, respiratory motor nuclei; Te, telencephalon; Ur, urophysis; VC, valvula cerebelli. V, trigeminal; VII, facial; IX, glossopharyngeal; X, vagal cranial nerves.

Le Mével et al. Central neuropeptides on cardio-respiratory functions

PVN, a nucleus homologous to the teleostean NPO, can influence brainstem and spinal cord respiratory related mechanisms. Vasopressin and oxytocin parvocellular neurons of the PVN project to important respiratory-related regions of the medulla and spinal cord, including the pre-Bõtzinger complex and the phrenic motor nuclei (Mack et al., 2002). The PVN is also part of the central cardiovascular network that controls the rostral ventrolateral medulla (RVLM) (Nunn et al., 2011). Neurons of the RVLM send excitatory projections to the sympathetic preganglionic neurons in the intermediolateral cell column of the spinal cord to increase HR and blood pressure (Dampney et al., 2005; Pilowsky et al., 2009; Nunn et al., 2011). Although in fish the locations of sympathetic pre-vasomotor nuclei within the medulla are unknown, neuropeptides may also act at the medulla oblongata to influence sympathetic outflow to vascular tissue and chromaffin cells increasing blood pressure. In addition, we can speculate about a possible diffusion of the injected neuropeptides within the cerebrospinal fluid toward critical ventilatory and cardiovascular brainstem nuclei. Consistent with this, receptors for some of the aforementioned neuropeptides are also expressed within the hindbrain (Cobb and Brown, 1992; Moons et al., 1992; Lovejoy and Balment, 1999; Lu et al., 2006). However, the pharmacological characterization of these receptors using specific antagonists/agonists is difficult in fish due to the fact that only drugs designed for mammalian receptors are available. We have noted consistently that the effects exerted by the neuropeptides are usually long lasting, a characteristic that is probably related to their slow rate of metabolism. Alternatively, this long lasting effect of neuropeptides may be due to complex intracellular signalling pathways after binding to their metabotropic GPCRs.

We cannot excluded that some neuropeptides when injected at high doses within the periphery may act at central target sites to increase PDA, through leakage of the blood brain barrier. However, a direct action of neuropeptides on vasculature is probably the physiological mechanism involved after peripheral injection.

**Figure 3** gives a summary of the proposed mechanisms for central neuropeptidergic cardio-respiratory regulation in trout.

**Table 1** provides a summary of the central ventilatory and cardiovascular actions of neuropeptides in our trout model. However, it remains to be determined whether the observed actions of exogenously administered neuropeptides can be translated into evidence for endogenous regulation of physiological functions. It is probable that a cocktail of neuropeptides within the trout brain is involved in fine control of ventilation, each peptide having a selective action either on VA or VF or on both ventilatory variables. Neuropeptides may be part of the neurochemical systems that are involved in the hypoxic ventilatory response in fish (Porteus et al., 2011). However, as previously stated, their precise implication in the CNS pathways that control the VA and VF during intermittent, repeated or chronic hypoxia is unknown. A balance between the action of hyperventilatory and hypoventilatory peptides may permit the fine control of ventilation so as to maintain homeostasis. In addition, endogenous neuropeptides may regulate cardiovascular function.

**Table 1 | Summary of the effects of intracerebroventricular injection of neuropeptides on ventilatory and cardiovascular variables in the unanesthetized trout.**


PACAP, pituitary adenylate cyclase-activating polypeptide; VIP, vasoactive intestinal peptide; NPγ, neuropeptide gamma; SP, substance P; NKA, neurokinin A; CRF, corticotropin-releasing factor; U-I, urotensin-I; CGRP, calcitonin generelated peptide; [Asn1]-Ang II, [Asn1]-angiotensin II; [Asp1]-Ang II, [Asp1] angiotensin II; Ang III, angiotensin III; Ang IV, angiotensin IV; Ang 1–7, angiotensin 1–7; U-II, urotensin-II; NPY, neuropeptide Y; PYY, peptide YY. *–*, no effect; , increase; , decrease; symbol in bold reflects a higher effect.

Besides conservation of the amino acid sequence of neuropeptides during evolution, our physiological results obtained with unanesthetized trout also give support for a strong conservation of cardiovascular and ventilatory functions throughout the vertebrate classes. Determination of the central ventilatory and cardiovascular actions of these neuropeptides in our trout model suggests that these neuropeptides act as neuromodulators and/or neurotransmitters. In addition, neuropeptides acting as local peptides or hormones may be involved in peripheral cardiovascular regulation. We hope that our comparative physiological studies provide new insights into evolution of the basic neuroregulatory mechanisms that operate in the CNS of vertebrates, including humans,

The circumstances leading to the release of endogenous neuropeptides into the synaptic cleft to control the ventilatory and cardiovascular autonomic nuclei during adverse metabolic or environmental situations remain to be delineated. It may be hypothesized that the central neuropeptidergic regulation of cardio-respiratory functions may be critical for proper uptake of oxygen from the aquatic environment and distribution of oxygen to tissues during hypoxic stress for example. Hypoxic stress is known to induce an hyperventilatory response in the rainbow trout through a selective action on VA (Gilmour and Perry, 2007) and environmental hypoxia increases the expression of CRF, UI, and CRFbinding protein genes within the NPO (Bernier and Craig, 2005).

# **REFERENCES**


*Annu. Rev. Pharmacol. Toxicol.* 44, 525–557.


of corticotropin-releasing factor (CRF) and urotensin I precursor genes, and evidence of CRF gene expression regulated by cortisol in goldfish brain. *Gen. Comp. Endocrinol.* 116, 461–477.

to control these vital cardio-respiratory functions.

**CONCLUSION AND PERSPECTIVES**


(2002). Central administration of corticotropin-releasing hormone stimulates locomotor activity in juvenile chinook salmon (*Oncorhynchus tshawytscha*). *Gen. Comp. Endocrinol.* 125, 319–327.


protein-coupled receptors for VIP and PACAP: structure, models of activation and pharmacology. *Peptides* 28, 1631–1639.


*mykiss*. *Am. J. Physiol.* 271, R1335–R1343.


polypeptide (PACAP)-like peptide from the brain of a teleost, stargazer, *Uranoscopus japonicus*. *Peptides* 18, 723–727.


*Physiol. A Mol. Integr. Physiol.* 128, 11–30.


Le Mével et al. Central neuropeptides on cardio-respiratory functions

teleost fish. *Eur. J. Pharmacol.* 430, 193–202.


Cambridge, MA: Cambridge University Press.


immunoreactive cells and fibers in the rat brain: an immunohistochemical study. *Neuroendocrinology* 36, 165–186.


demonstration of PACAP in the pituitary, PACAP stimulation of growth hormone release from pituitary cells, and molecular cloning of pituitary type I PACAP receptor. *Endocrinology* 139, 3465–3479.


subtypes in Japenese eel acclimated to various salinities from deionized water to double-stength seawater. *Gen. Comp. Endocrinol.* 178, 250–258.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 24 August 2012; paper pending published: 11 September 2012; accepted: 01 October 2012; published online: 30 October 2012.*

*Citation: Le Mével J-C, Lancien F, Mimassi N and Conlon JM (2012) Brain neuropeptides in central ventilatory and cardiovascular regulation in trout. Front.* *Endocrin. 3:124. doi: 10.3389/fendo. 2012.00124*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Le Mével, Lancien, Mimassi and Conlon. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# The stimulatory effect of the octadecaneuropeptide ODN on astroglial antioxidant enzyme systems is mediated through a GPCR

*Yosra Hamdi 1, Hadhemi Kaddour1, David Vaudry2,3,4, Salma Douiri 1, Seyma Bahdoudi 1, Jérôme Leprince2,3,4, Hélène Castel 2,4, Hubert Vaudry2,3,4\*, Mohamed Amri 1\*, Marie-Christine Tonon2,4 and Olfa Masmoudi-Kouki <sup>1</sup>*

<sup>1</sup> Laboratory of Functional Neurophysiology and Pathology, Research Unit UR/11ES09, Department of Biological Sciences, Faculty of Science of Tunis, University Tunis El Manar, Tunis, Tunisia

<sup>2</sup> Inserm U982, Laboratory of Neuronal and Neuroendocrine Communication and Differentiation, University of Rouen, Mont-Saint-Aignan, France

<sup>3</sup> International Associated Laboratory Samuel de Champlain, Mont-Saint-Aignan, France

<sup>4</sup> Regional Platform for Cell Imaging of Haute-Normandie, Institute for Medical Research and Innovation, University of Rouen, Mont-Saint-Aignan, France

#### *Edited by:*

Jae Young Seong, Korea University, South Korea

#### *Reviewed by:*

María M. Malagón, University of Cordoba, Spain Kouhei Matsuda, University of Toyama, Japan

#### *\*Correspondence:*

Mohamed Amri, Laboratory of Functional Neurophysiology and Pathology, Research Unit UR/11ES09, Department of Biological Sciences, Faculty of Science of Tunis, University Tunis El Manar, 2092 Tunis, Tunisia. e-mail: mohamed.amri@fst.rnu.tn; Hubert Vaudry, Inserm U982, Laboratory of Neuronal and Neuroendocrine Communication and Differentiation, International Associated Laboratory Samuel de Champlain, Regional Platform for Cell Imaging of Haute-Normandie, Institute for Medical Research and Innovation, University of Rouen, 76821 Mont-Saint-Aignan, France. e-mail: hubert.vaudry@univ-rouen.fr

Astroglial cells possess an array of cellular defense systems, including superoxide dismutase (SOD) and catalase antioxidant enzymes, to prevent damage caused by oxidative stress on the central nervous system. Astrocytes specifically synthesize and release endozepines, a family of regulatory peptides including the octadecaneuropeptide (ODN). ODN is the ligand of both central-type benzodiazepine receptors (CBR), and an adenylyl cyclase- and phospholipase C-coupled receptor. We have recently shown that ODN is a potent protective agent that prevents hydrogen peroxide (H2O2)-induced inhibition of SOD and catalase activities and stimulation of cell apoptosis in astrocytes. The purpose of the present study was to investigate the type of receptor involved in ODN-induced inhibition of SOD and catalase in cultured rat astrocytes.We found that ODN induced a rapid stimulation of SOD and catalase gene transcription in a concentration-dependent manner. In addition, 0.1 nM ODN blocked H2O2-evoked reduction of both mRNA levels and activities of SOD and catalase. Furthermore, the inhibitory actions of ODN on the deleterious effects of H2O2 on SOD and catalase were abrogated by the metabotropic ODN receptor antagonist cyclo1 8[Dleu<sup>5</sup> <sup>−</sup> ]OP, but not by the CBR antagonist flumazenil. Finally, the protective action of ODN against H2O2-evoked inhibition of endogenous antioxidant systems in astrocytes was protein kinase A (PKA)-dependent, but protein kinase C-independent. Taken together, these data demonstrate for the first time that ODN, acting through its metabotropic receptor coupled to the PKA pathway, prevents oxidative stress-induced alteration of antioxidant enzyme expression and activities. The peptide ODN is thus a potential candidate for the development of specific agonists that would selectively mimic its protective activity.

**Keywords: astrocyte, catalase, ODN, ODN metabotropic receptor, oxidative stress, SOD, protein kinase A**

"fendo-03-00138" — 2012/11/21 — 10:23 — page 1 — #1

#### **INTRODUCTION**

The octadecaneuropeptide (ODN) is a peptide generated through the proteolytic cleavage of the 86-amino acid precursor diazepambinding inhibitor (DBI; Guidotti et al., 1983) which is exclusively expressed in astroglial cells in the central nervous system (CNS) of mammals (Malagon et al., 1993; Burgi et al., 1999; Tonon et al., 2006). DBI and its derived peptides are collectively designated by the term endozepines (Tonon et al., 2006). It was initially reported that ODN acts as an inverse agonist of central-type benzodiazepine receptors (CBR) that are intrinsic components of the GABAA receptor-chloride channel complex (Ferrero et al., 1986). It has been subsequently shown that ODN can also activate a Gi/<sup>0</sup> protein-coupled receptor leading to the activation of phospholipase C (PLC) in astrocytes (Patte et al., 1995; Leprince et al., 2001). In addition, recent data indicate that the ODN G protein-coupled receptor can also activate adenylyl cyclase (AC; Hamdi et al., 2012). ODN exerts a wide range of biological activities which are mediated either through CBR, i.e., increase of aggressiveness and anxiety (Kavaliers and Hirst, 1986; De Mateos-Verchere et al., 1998), reduction of pentobarbital-induced sleeping time and drinking (Dong et al., 1999; Manabe et al., 2001), or through a metabotropic receptor, i.e., inhibition of food intake (Do Rego et al., 2007). Similarly, at the cellular level, the diverse effects of ODN are mediated either through CBR, i.e., stimulation of glial cell and neuroblast proliferation (Gandolfo et al., 1999; Alfonso et al., 2012) and activation of neurosteroid biosynthesis (Do Rego et al., 2001), or through a metabotropic receptor, i.e., increase of intracellular calcium concentration in astrocytes

**Abbreviations:** AC, adenylyl cyclase; CBR, central-type benzodiazepine receptors; CREB, cAMP-responsive element-binding protein; DBI, diazepam-binding inhibitor; GADPH, glyceraldehyde-3-phosphate dehydrogenase; H2O2, hydrogen peroxide; ODN, octadecaneuropeptide; PBS, phosphate buffered saline; PKA, protein kinase A; PKC, protein kinase C; PLC, phospholipase C; ROS, reactive oxygen species; SOD, superoxide dismutase.

(Leprince et al., 2001) and modulation of neuropeptide expression in neurons (Compère et al., 2003, 2004).

Oxidative stress, resulting from excessive production of reactive oxygen species (ROS), such as hydrogen peroxide (H2O2), is implicated in the pathology of several neurological disorders including cerebral ischemia and neurodegenerative diseases (Garcia et al., 2012; Hayashi et al., 2012). An excess of H2O2 induces imbalance in ROS generation, impairs cellular antioxidant defences and finally triggers cell death by apoptosis (Emerit et al., 2004; Shibata and Kobayashi, 2008). It is well known that astroglial cells contain high levels of ROS scavenger molecules such as glutathione (Dringen et al., 1999) and the antioxidant enzymes Mn- and Cu,Zn-superoxide dismutases (Mn- and Cu,Zn-SOD), catalase and glutathione peroxidase (Lindenau et al., 2000; Sokolova et al., 2001; Saha and Pahan, 2007). Nonetheless, astroglial cells can be affected, in terms of viability and functionality, by an insurmountable oxidative stress (Ferrero-Gutierrez et al., 2008; Park et al.,2009). In particular, it has been shown that inhibition of SOD and/or catalase activities in cultured astrocytes is associated with an exacerbation of oxidative damages induced by H2O2 or hypoxia (Desagher et al., 1996; Bi et al., 2008; Li et al., 2008). Reciprocally, cultured astrocytes derivedfrom Cu,Zn-SOD-overexpressing transgenic mice exhibit increased resistance to oxidative stress (Chen et al., 2001; Wang et al., 2005). However, little is known regarding the endogenous factors that modulate glial antioxidant systems. In this context, we have previously reported that, in cultured astrocytes, ODN exerts a potent protective effect against oxidative stress-induced apoptosis, and attenuates H2O2 evoked inhibition of SOD and catalase activities (Hamdi et al., 2011). More recently, we have shown that the anti-apoptotic activity of ODN is mediated through the metabotropic endozepine receptor (Hamdi et al., 2012). In contrast, regarding the effects of ODN on endogenous antioxidant systems, the receptor and the signaling mechanism are currently unknown. The purpose of the present study was thus to examine the effects of ODN on SOD and catalase gene expression and to determine the type of receptor involved in the antioxidant action of ODN on astroglial cells.

# **MATERIALS AND METHODS**

#### **ANIMALS**

All experiments were performed in accordance with AmericanVeterinary Medical Association. Approval for these experiments was obtained from the Medical Ethical Committee for the Care and Use of Laboratory Animals of Pasteur Institute of Tunis (approval number: FST/LNFP/Pro 152012).

#### **REAGENTS**

Dulbecco's modified Eagle's medium (DMEM), F12 culture medium, D(+)-glucose, L-glutamine, *N*-2-hydroxyethylpiperazine-*N*-2-ethane sulfonic acid (HEPES), fetal bovine serum (FBS), the antibiotic-antimycotic solution, and trypsin-EDTA were obtained from Gibco (Invitrogen, Grand Island NY, USA). Bovine liver catalase, chelerythrine, DL-epinephrine, H89, Triton X-100, and insulin were purchased from Sigma Aldrich (St. Louis, MO, USA). Flumazenil was a generous gift from Hoffmann-La Roche (Basel, Switzerland). Rat ODN and the G protein-coupled receptor antagonist cyclo1−8[DLeu5]OP were synthesized by using the standard Fmoc procedure, as previously described (Leprince et al., 2001). All other reagents were of A grade purity.

# **SECONDARY CULTURES OF RAT CORTICAL ASTROCYTES**

Secondary cultures of rat cortical astrocytes were prepared as previously described (Castel et al.,2006). Briefly, cerebral hemispheres from newborn Wistar rats were collected in DMEM/F12 (2:1; v/v) culture medium supplemented with 2 mM glutamine, 1% insulin, 5 mM HEPES, 0.4% glucose, and 1% of the antibiotic-antimycotic solution. Dissociated cells were resuspended in culture medium supplemented with 10% FBS, plated in 175-cm2 flasks (Greiner Bio-one GmbH, Frickenhausen, Germany), and incubated at 37◦C in a 5% CO2/95% O2 atmosphere. When cultures were confluent, astrocytes were isolated by shaking overnight the flasks with an orbital agitator and plated on 35-mm Petri dishes at a density of 0.3 <sup>×</sup> <sup>10</sup><sup>6</sup> cells/ml. All experiments were performed on 5- to 7-day-old secondary cultures.

#### **QUANTITATIVE RT-PCR ANALYSIS**

Cultured cells were incubated at 37◦C with fresh serum-free medium. At the end of the incubation, the culture medium was removed and the cells were washed twice with phosphate buffered saline (PBS; 0.1 M, pH 7.4). Total RNA was extracted by using Tri reagent (Sigma, St Quentin Fallavier, France) and purified using the NucleoSpin kit (Macherey-Nagel, Hoerd, France). cDNA was synthetized from 3–4 μg of total RNA with ImProm II Promega kit (Promega). Quantitative RT-PCR was performed on cDNA in the presence of a 1× Fast SYBR Green universal PCR Master mix (Applied Biosystems, Courtaboeuf, France) containing concentrations of dNTPs, MgCl2, SYBR green reporter dye, AmpliTaq Gold DNA polymerase, and forward (5 -CCTTCTTGTTCTGCAACC-TGCTA-3 ) and reverse (5 -CCGGACTCTCCGGTATCTGA-3 ) SOD (GenBank accession no. NM\_012880) primers, or forward (5 -CCACAGTCGCTGGAGAGTCA-3 ) and reverse (5 -GTTTC-CCACAAGGTCCCAGTT-3 ) catalase (GenBank accession no. NM\_012520) primers, or forward (5 -CAGCCTCGTCTCATAGA-CAAGATG-3 ) and reverse (5 - CAATGTCCACTTTGTCACAAG-AGAA-3 ) glyceraldehyde-3-phosphate dehydrogenase (GADPH, GenBank accession no. NM\_017008) primers (300 nM, each; Proligo, Paris, France), using the ABI Prism 7000 sequence detection system (Applied Biosystems). The amount of SOD and catalase cDNA in each sample was calculated by the comparative threshold cycle (Ct) method and expressed as 2−ΔΔCt using GADPH as an internal control.

#### **MEASUREMENT OF ANTIOXIDANT ENZYME ACTIVITIES**

"fendo-03-00138" — 2012/11/21 — 10:23 — page 2 — #2

Cultured cells were incubated at 37◦C with fresh serum-free medium. At the end of the incubation, cells were washed twice with PBS and total cellular proteins were extracted by using the lysis buffer containing 50 mM Tris–HCl (pH 8.0), 10 mM EDTA, 100μM phenylmethylsulfonyl fluoride, and 1% Triton X-100. The homogenate was centrifuged (16,000 *g*, 4◦C, 20 min) and the cellular extract contained in the supernatant was stored at −20◦C until enzyme activity determinations.

The activity of SOD was measured using a spectrophotometric assay, which consists in measuring epinephrine autoxidation induced by superoxide anion. Samples, prepared as described above, were incubated for 3 min with a mixture containing bovine catalase (0.4 U/μl), DL-epinephrine (5 mg/ml), and Na2CO3/NaHCO3 buffer (62.5 mM, pH 10.2). The oxidation of epinephrine was measured at 480 nm with a Bio-Rad spectrophotometer (Bio-Rad Laboratories, Philadelphia, PA, USA).

The activity of catalase was determined on the basis of the decrease of H2O2. Samples, prepared as described above, were mixed with 30 mM H2O2 in PBS. The disappearance of H2O2 was measured at 240 nm for 180 s at 30 s intervals. Catalase activity was calculated using the extinction coefficient of 40/mM/cm for H2O2.

#### **STATISTICAL ANALYSIS**

Data are presented as the mean ± SEM from three independent experiments performed in quadruplicate. Statistical analysis of the data was performed by using Student's *t*-test, ANOVA, followed by Bonferroni's test, and two-way ANOVA test. A *p*-value of 0.05 or less was considered as statistically significant.

#### **RESULTS**

#### **ODN INCREASES SOD AND CATALASE mRNA LEVELS IN CULTURED ASTROCYTES**

We have previously shown that picomolar concentrations of ODN suppress the inhibitory effects of 300 μM H2O2 on SOD and catalase activities in cultured rat astrocytes (Hamdi et al., 2011). To explore the mechanism involved in the effect of ODN on antioxidant enzyme systems we monitored SOD and catalase gene expression by quantitative PCR. Time-course experiments revealed that ODN (0.1 nM) significantly enhanced SOD and catalase mRNA levels within 2 min with a maximum effect after 10 min and 5 min of incubation, respectively (**Figure 1A**). Thereafter, the stimulatory effect of ODN on SOD and catalase expression gradually declined and vanished 60 and 30 min after the onset of ODN administration, respectively. Exposure of astrocytes to increasing concentrations of ODN (0.01 pM to 0.1 nM) induced a concentration-dependent increase of SOD and catalase mRNA levels (**Figure 1A**, inset). In contrast, incubation of astrocytes with graded concentrations of H2O2 (100–500 μM) dose-dependently decreased both SOD and catalase mRNA levels (**Figure 1B**). We next examined the effects of ODN/H2O2 co-incubation on enzyme expression. For moderate concentrations of H2O2 (100–300 μM), ODN (0.1 nM) restored SOD and catalase mRNA levels above control, whereas for higher concentrations of H2O2 (400 and 500 μM), ODN only partially prevented the decrease of SOD and catalase gene expression (**Figure 1B**).

#### **ODN BLOCKS H2O2-EVOKED INHIBITION OF SOD AND CATALASE mRNA LEVELS AND ACTIVITIES THROUGH ACTIVATION OF A METABOTROPIC RECEPTOR COUPLED TO THE PKA PATHWAY**

We next examined the type of receptor of ODN involved in the stimulatory effects of ODN on endogenous antioxidant systems. Administration of the selective metabotropic receptor antagonist

cyclo1−8[DLeu5]OP (1 <sup>μ</sup>M) to cultured astrocytes did not induce any modification of SOD and catalase mRNA levels and activities, but totally abolished the effects of 0.1 nM ODN on H2O2-evoked inhibition of antioxidant enzyme gene transcription and activities. In contrast, the CBR antagonist flumazenil (1 μM) did not affect the protective action of ODN against the deleterious effect of H2O2 on endogenous antioxidant systems (**Figures 2A,B**).

Incubation of astrocytes with the selective protein kinase A (PKA) inhibitor H89 (20 μM) abrogated the effect of ODN on the inhibitory action of H2O2 on SOD and catalase mRNA levels and activities. In contrast, administration of the protein kinase C (PKC) inhibitor chelerythrine (0.1 μM) did not modify the effects of ODN (**Figures 3A,B**), indicating that only the PKA pathway is involved in the protective activity of ODN.

# **DISCUSSION**

"fendo-03-00138" — 2012/11/21 — 10:23 — page 3 — #3

Alteration of endogenous antioxidant systems, especially decrease of SOD and catalase activities, causes exacerbation of oxidative damages leading to apoptosis in various cell types, including astroglial cells (Giffard and Swanson, 2005; Lopez et al., 2007; Bi et al., 2008). Here, we demonstrate that ODN prevents the decrease of SOD and catalase mRNA levels and activities induced by H2O2 in cultured astrocytes, through activation of a metabotropic receptor positively coupled to the AC/PKA signaling pathway.

We have previously reported that, in cultured astrocytes, ODN at very low concentrations (in the picomolar range) stimulates SOD and catalase activities (Hamdi et al., 2011). The present study shows that, in the same range of concentrations, ODN induces a dose-dependent increase of SOD and catalase mRNA levels in cultured astroglial cells, indicating that ODN regulates not only enzyme activity but also gene transcription. Although SOD and catalase genes exhibit hallmarks of typical housekeeping genes, it has been shown that their promoters encompass consensus sequences for regulatory elements such as metal-responsive element, antioxidant responsive element, glucocorticoid-response element, and nuclear factor-κB (Nenoi et al., 2001; Zhu et al., 2001; Zelko et al., 2002), suggesting that these genes are actually regulated in the CNS. As a matter of fact, SOD and catalase gene expression is selectively increased by inflammatory mediators such as interleukin-1β, interferon (IFN)-γ, IFN-β, or lipopolysaccharides in astrocytes (Mokuno et al., 1994; Kifle et al., 1996; Vergara et al., 2010). Previous data have shown that ODN is specifically produced by astroglial cells (Tonon et al., 1990; Malagon et al., 1993; Compère et al., 2010) and that its release is regulated by various factors including agonists of formyl peptide receptors (Tokay et al., 2008) which are involved in inflammation. These observations suggest that ODN may act as an autocrine factor to finely regulate SOD and catalase gene expression in the brain.

Kinetic experiments indicate that the action of ODN on antioxidant enzyme gene transcription is very rapid but transient. Nevertheless, ODN exerts a protective effect against H2O2-reduced SOD and catalase mRNA levels. Similar time-response curves have already been observed on SOD and catalase activities, in cultured astrocytes (Hamdi et al., 2011). These data suggest that

"fendo-03-00138" — 2012/11/21 — 10:23 — page 4 — #4

**FIGURE 1 | Protective effects of ODN against H2O2-evoked inhibition of SOD and catalase mRNA levels in cultured rat astrocytes. (A)** Effects of ODN on SOD and catalase mRNA levels. Astrocytes were incubated in the absence or presence of ODN (0.1 nM) for the times indicated. Inset, cells were treated for 10 min with increasing concentrations of ODN (0.01 pM to 0.1 nM). SOD and catalase mRNA levels were measured by quantitative RT-PCR. Data were corrected using the GAPDH signal as an internal control and the results are expressed as a percentage of controls. Each value is the mean (±SEM) of at least four different wells from three independent experiments. ANOVA followed by the Bonferroni's test. \*p < 0.05;

\*\*p < 0.01; \*\*\*p < 0.001; NS, not statistically different vs. control. **(B)** Effects of ODN on H2O2-evoked inhibition of SOD and catalase mRNA levels. Cells were pre-incubated for 10 min in the absence or presence of 0.1 nM ODN, and then incubated for 1 h with medium alone or with graded concentrations of H2O2 (100–500 μM) in the absence or presence of ODN. The results are expressed as a percentage of control. Each value is the mean (±SEM) of at least four different wells from three independent experiments. Analyses similar to those in **(A)** were performed and symbols show the significance vs. H2O2-treated cells: #<sup>p</sup> <sup>&</sup>lt; 0.05; ##<sup>p</sup> <sup>&</sup>lt; 0.01; ###<sup>p</sup> <sup>&</sup>lt; 0.001; ns, not statistically different.

"fendo-03-00138" — 2012/11/21 — 10:23 — page 5 — #5

**in its protective effects against the deleterious action of H2O2 on SOD and catalase in cultured rat astrocytes.** Astrocytes were pre-incubated for 30 min in the absence or presence of the metabotropic receptor antagonist cyclo1−<sup>8</sup> [Dleu5] OP (1 <sup>μ</sup>M) or the CBR antagonist flumazenil (1 μM) and then incubated for 1 h with medium alone or with 300 μM H2O2 without or with ODN (0.1 nM). **(A)** SOD and catalase mRNA levels were quantified as described in Figure 1. The results are expressed as a percentage of control. Each value is the mean (±SEM) of at least three different wells from three independent experiments. **(B)** The activity of SOD was measured using a spectrophotometric assay which consists in measuring epinephrine autoxidation induced by superoxide anion, and

of H2O2. The results are expressed as a percentage of SOD or catalase activity with respect to control. Each value is the mean (±SEM) of at least four different dishes from three independent experiments. ANOVA followed by the Bonferroni's test: [(**A**, SOD) F = 7.69, df = 40; (**A**, catalase) F = 8.58, df = 39; (**B**, SOD) F = 13.41, df = 44; (**B**, catalase) F = 7.58, df = 44]; \*p < 0.05; \*\*\*p < 0.001; NS, not statistically different vs. control. ###<sup>p</sup> <sup>&</sup>lt; 0.001; ns, not statistically different vs. H2O2-treated cells. Two-way ANOVA test: [(**A**, SOD) F = 5, df = 17; (**A**, catalase) F = 8.13, df = 17; (**B**, SOD) <sup>F</sup> <sup>=</sup> 5.87, df <sup>=</sup> 20; (**B**, catalase) <sup>F</sup> <sup>=</sup> 2.79, df <sup>=</sup> 19]; §<sup>p</sup> <sup>&</sup>lt; 0.05; §§<sup>p</sup> <sup>&</sup>lt; 0.01; Ns, not statistically different vs. ODN <sup>+</sup> H2O2-cotreated cells.

"fendo-03-00138" — 2012/11/21 — 10:23 — page 6 — #6

**FIGURE 3 | Protein kinase A-dependence of the protective effects of ODN on the expression and activities of SOD and catalase in cultured rat astrocytes.** Astrocytes were pre-treated for 30 min in the absence or presence of the PKA inhibitor H89 (20 μM) or the PKC inhibitor chelerythrine (Chel; 0.1 μM) and then incubated for 1 h with medium alone or with 300 μM H2O2 without or with ODN (0.1 nM). **(A)** SOD and catalase mRNA levels were quantified as described in Figure 1. The results are expressed as a percentage of controls. Each value is the mean (±SEM) of at least three different wells from three independent experiments. **(B)** The activity of SOD and catalase were quantified as described in Figure 2. The results are

expressed as a percentage of SOD or catalase activity with respect to control. Each value is the mean (±SEM) of at least four different dishes from three independent experiments. ANOVA followed by the Bonferroni's test: [(**A**, SOD) F = 7.49, df = 65; (**A**, catalase) F = 5.08, df = 53; (**B**, SOD) F = 12.37, df = 55; (**B**, catalase) F = 9.06, df = 60]; \*p < 0.05; \*\*\*p < 0.001; NS, not statistically different vs. control. ###p < 0.001; ns, not statistically different vs. H2O2-treated cells. Two-way ANOVA test: [(**A**, SOD) F = 2.40, df = 20; (**A**, catalase) F = 2.63, df = 17; (**B**, SOD) F = 6.97, df = 18; (**B**, catalase) <sup>F</sup> <sup>=</sup> 4.68, df <sup>=</sup> 24]; §<sup>p</sup> <sup>&</sup>lt; 0.05; §§<sup>p</sup> <sup>&</sup>lt; 0.01; Ns, not statistically different vs. ODN + H2O2-cotreated cells.

ODN-induced rapid activation of antioxidant systems is required for the long-lasting inhibition of the deleterious effect of H2O2. That ODN-induced increase of transcription and activity of antioxidant enzymes is responsible, at least in part, for inhibition of cell death is consistent with previous data showing that SOD and catalase blockers suppress the protective effect of ODN against H2O2-induced astrocyte apoptosis (Hamdi et al., 2011). Furthermore, it has been reported that, in cultured astrocytes, overexpression of SOD is able to prevent ROS-induced alteration of mitochondrial integrity, caspase-3 activation and thus cells apoptosis (Yang et al., 2008).

Previous studies have shown that, in cultured astrocytes, ODN can interact with either CBR associated with the GABAA receptor (Gandolfo et al., 1999) or with a G protein-coupled receptor positively coupled to PLC (Patte et al., 1995; Leprince et al., 2001). Here, we found that the inhibitory effects of ODN on H2O2 evoked reduction of SOD and catalase mRNA levels and activities were suppressed by the ODN analog cyclo1−8[DLeu5]OP but were not affected by the specific CBR antagonist flumazenil. It has been reported that the cyclic analog of ODN exerts potent antagonistic activities on ODN-induced polyphosphoinositide turnover increase and intracellular calcium mobilization in rat astrocytes (Leprince et al., 2001). Thus, these data indicate that the antioxidant action of ODN is mediated through the activation of the G protein-coupled receptor.

We next investigated the signaling cascade involved in the effect of ODN on endogenous antioxidant systems. ODN blockage of H2O2-evoked inhibition of SOD and catalase gene transcription and enzyme activities was totally abrogated by the PKA inhibitor H89, while the PKC inhibitor chelerythrine had no effect. That, the ODN G protein-coupled receptor could stimulate the AC/PKA transduction cascade is in agreement with recent data indicating that ODN increases the production of cAMP in astrocytes (Hamdi et al., 2012). Altogether, these observations indicate that the antioxidant action of ODN against H2O2-induced oxidative stress can be specifically ascribed to the activation of the AC/PKA signaling pathway. Consistent with this notion, it has been shown that the SOD and catalase promoters contain a cAMPresponsive element-like sequence (Das et al., 1995; Zelko et al., 2002; Colombo and Moncada, 2009) and that siRNA knockdown of cAMP-responsive element-binding protein (CREB) or inhibition of CREB phosphorylation blocks the expression of SOD in the rat hypothalamus (Hsieh et al., 2008) and the expression of catalase in human vascular endothelial cells (Colombo and Moncada, 2009), respectively. The fact that ODN provokes ERK phosphorylation via a cAMP-dependent pathway in astrocytes (Hamdi et al., 2012), strongly suggests that the stimulatory

#### **REFERENCES**


H2O2-induced oxidative stress in astrocytes primary cultures. *Neurosci. Lett.* 442, 224–227.

Burgi, B., Lichtensteiger, W., Lauber, M. E., and Schlumpf, M. (1999). Ontogeny of diazepam binding inhibitor/ acyl-CoA binding protein mRNA and peripheral benzodiazepine receptor mRNA expression in the rat. *J. Neuroendocrinol.* 11, 85–100.

effect of ODN on SOD and catalase expression can also be ascribed to activation of the ERK-type MAP kinase transduction pathway.

The protective effect of ODN against H2O2-reduced antioxidant enzyme expression and activities might have a physiopathological significance in neurodegenerative diseases and stroke. CNS is sensitive to oxidative stress due to its high metabolic rate and high levels of unsaturated lipids so that up-regulation of antioxidant enzyme systems in astroglial cells could be beneficial against cell death observed during and after ischemia and neurodegenerative diseases. In agreement with this hypothesis, we have recently shown that the endozepine ODN exerts a potent protective action against apoptosis induced by oxidative stress in astrocytes (Hamdi et al., 2011, 2012) and that the anti-apoptotic effect of ODN is attributable to activation of the antioxidant enzymes that act as scavengers of H2O2 and ROS (Hamdi et al., 2011). The fact that the glioprotective action of ODN is likely mediated through the metabotropic receptor is of particular interest. Previous data indicate that ODN induces a wide range of activities through activation of CBR (Tonon et al., 2006). In particular, ODN has been initially described as an anxiogenic peptide (De Mateos-Verchere et al., 1998). Since cyclic analog of ODN do not recognize CBR (Leprince et al., 2001), the development of specific cyclic agonists that would selectively mimic the glioprotective effect of ODN might prove useful for the treatment of ischemia and neurodegenerative diseases.

In conclusion, the present study has demonstrated that the endozepine ODN, acting through a metabotropic receptor sensitive to the cyclo1−8[DLeu5]OP antagonist, exerts a potent antioxidant action against H2O2-induced oxidative stress in astrocytes. This antioxidant effect of ODN is attributable to activation of both gene expression and activities of enzymatic antioxidant systems and can be ascribed to the stimulation of the AC/PKA transduction pathway.

#### **ACKNOWLEDGMENTS**

Yosra Hamdi and Hadhemi Kaddour were recipients of fellowships from the University of Tunis El Manar and a France-Tunisia exchange program CMCU-Utique. Seyma Bahdoudi and Salma Douiri were recipients of fellowships from the University of Tunis El Manar and a France-Tunisia exchange program Inserm-DGRS. This study was supported by the Research Unit UR/11ES09, a CMCU-Utique program (to Mohamed Amri and Marie-Christine Tonon; grant number 07G0822), an Inserm-DGRS program (to Mohamed Amri and Marie-Christine Tonon; grant number M 10/M), Inserm (U982), the Institute for Medical Research and Innovation (IRIB) and the Région Haute-Normandie.


"fendo-03-00138" — 2012/11/21 — 10:23 — page 7 — #7

have increased resistance to oxidative injury. *Glia* 33, 343–347.


diazepam-binding inhibitor, the precursor of the anorexigenic octadecaneuropeptide ODN, in mouse glial cells. *J. Mol. Endocrinol*. 44, 295–299.


binding protein modulates superoxide dismutase and neuropeptide Ymediated feeding behavior in freely moving rats. *J. Neurochem.* 105, 1438–1449.


"fendo-03-00138" — 2012/11/21 — 10:23 — page 8 — #8

gene promoter by Sp1, CCAATrecognizing factors, and a WT1/Egrrelated factor in hydrogen peroxideresistant HP100 cells. *Cancer Res.* 61, 5885–5894.


decreases ischemia-like astrocyte injury. *Free Radic. Biol. Med.* 38, 1112–1118.


(SOD2), and EC-SOD (SOD3) gene structures, evolution, and expression. *Free Radic. Biol. Med.* 33, 337–349.

Zhu, C. H., Huang, Y., Oberley, L. W., and Domann, F. E. (2001). A family of AP-2 proteins downregulate manganese superoxide dismutase expression. *J. Biol. Chem.* 276, 14407–14413.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 05 September 2012; accepted: 26 October 2012; published online: 21 November 2012.*

*Citation: Hamdi Y, Kaddour H, Vaudry D, Douiri S, Bahdoudi S, Leprince J, Castel H, Vaudry H, Amri M, Tonon M-C and Masmoudi-Kouki O (2012) The stimulatory effect of the octadecaneuropeptide ODN on astroglial antioxidant enzyme systems is mediated through a GPCR. Front. Endocrin. 3:138. doi: 10.3389/fendo.2012.00138*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Hamdi, Kaddour, Vaudry, Douiri, Bahdoudi, Leprince, Castel, Vaudry, Amri, Tonon and Masmoudi-Kouki. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

"fendo-03-00138" — 2012/11/21 — 10:23 — page 9 — #9

# Role of calcitonin gene-related peptide in cerebral vasospasm, and as a therapeutic approach to subarachnoid hemorrhage

# *Stelios Kokkoris1, Peter Andrews 2\* and David J.Webb3*

<sup>1</sup> Intensive Care Unit, Western General Hospital, Edinburgh, UK

<sup>2</sup> Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, UK

<sup>3</sup> Clinical Pharmacology Unit, British Heart Foundation Centre for Cardiovascular Science, Queen's Medical Research Institute, University of Edinburgh, Edinburgh, UK

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Akiyoshi Takahashi, Kitasato University, Japan Jo G. De Mey, Maastricht University, Netherlands

#### *\*Correspondence:*

Peter Andrews, Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, UK. e-mail: p.andrews@ed.ac.uk

Calcitonin gene-related peptide (CGRP) is one of the most potent microvascular vasodilators identified to date. Vascular relaxation and vasodilation is mediated via activation of the CGRP receptor. This atypical receptor is made up of a G protein-coupled receptor called calcitonin receptor-like receptor (CLR), a single transmembrane protein called receptor activity-modifying protein (RAMP), and an additional protein that is required for Gas coupling, known as receptor component protein (RCP). Several mechanisms involved in CGRP-mediated relaxation have been identified. These include nitric oxide (NO)-dependent endothelium-dependent mechanisms or cAMP-mediated endothelium-independent pathways; the latter being more common. Subarachnoid hemorrhage (SAH) is associated with cerebral vasoconstriction that occurs several days after the hemorrhage and is often fatal. The vasospasm occurs in 30–40% of patients and is the major cause of death from this condition. The vasoconstriction is associated with a decrease in CGRP levels in nerves and an increase in CGRP levels in draining blood, suggesting that CGRP is released from nerves to oppose the vasoconstriction. This evidence has led to the concept that exogenous CGRP may be beneficial in a condition that has proven hard to treat. The present article reviews: (a) the pathophysiology of delayed ischemic neurologic deficit after SAH (b) the basics of the CGRP receptor structure, signal transduction, and vasodilatation mechanisms and (c) the studies that have been conducted so far using CGRP in both animals and humans with SAH.

**Keywords: GPCR, CGRP, subarachnoid hemorrhage, cerebral vasospasm, G proteins**

**"fendo-03-00135" — 2012/11/9 — 22:03 — page 1 — #1**

#### **INTRODUCTION**

In the US, over 30,000 persons each year experience a subarachnoid hemorrhage (SAH). Whereas intracranial aneurysms are found in 2−5% of all autopsies, the incidence of rupture is only 2−20/100,000 individuals/year (Ingall et al., 2000). SAH is more frequent in women than men (3:2 ratio) over the age of 40, but the reverse is the case in those younger than 40 (The ACROSS Group, 2000; Ohkuma et al., 2002). Peak rupture rates occur between the ages of 50 and 60 years (The ACROSS Group, 2000; Ohkuma et al., 2002). Intracranial aneurysms account for approximately 85% of cases of non-traumatic SAH, whereas 10% have the pattern of non-aneurysmal perimesencephalic hemorrhage, a relatively harmless condition (van Gijn et al., 2007). The other causes include bleeding from other vascular malformations, moyamoya syndrome, coagulopathy, and, rarely, extension of an intracerebral hematoma (van Gijn et al., 2007). In up to 15%, no source of bleeding is identified (Kim et al., 2012). Approximately 10−15% of patients die before receiving medical treatment from the initial bleed or its immediate complications (Huang and van Gelder, 2002) and over 40% of hospitalized patients die within 1 month of the event (Ingall et al., 2000). Those that survive the initial bleed are at risk for a number of secondary insults including rebleeding (Winn et al., 1977; Ohkuma et al., 2001), hydrocephalus, and cerebral vasospasm (van Gijn et al., 2007).

Calcitonin gene-related peptide (CGRP) is one of the most potent microvascular vasodilator peptides identified to date. In the cerebral circulation, CGRP is released from sensory fibers originating in the trigeminal ganglia and acts to dilate cerebral vessels (McCulloch et al., 1986). CGRP has been found to be at least 1,000 times more potent than acetylcholine, substance P, ATP, adenosine, and 5-hydroxytriptamine, and 10−100 times more potent than the β-adrenergic agonist isoprenaline. Consequently, a dose of 15 pmol injected into human skin produces an erythema that lasts for 5−6h(Brain et al., 1985). As we discuss later, CGRP has a particularly potent vasodilator activity in the cerebral circulation, rendering it a promising agent for the treatment of SAH-triggered cerebral vasospasm.

In the present review, we summarize the etiology and therapy of cerebral vasospasm, the biology of CGRP and its receptors, and review the role of CGRP as a treatment in SAH-associated vasospasm in both animals and humans.

#### **CEREBRAL VASOSPASM AFTER SAH**

#### **DEFINITIONS**

Throughout the literature, authors have used various means of defining vasospasm including terms like angiographic vasospasm, symptomatic vasospasm, and delayed cerebral ischemia (DCI). Angiographic vasospasm is a narrowing of the lumen of the major cerebral arteries, which is usually focal but may be diffuse. Vasospasm has its onset usually on day 3 after SAH, is maximal at days 6–8, and usually lasts for 2–3 weeks (Wilkins, 1990). Symptomatic vasospasm is characterized by the insidious onset of confusion and decreased level of consciousness, followed by focal motor and/or speech impairments. It is mainly a diagnosis of exclusion, when clinical deterioration occurs and hydrocephalus, rebleeding, hypoxia, and metabolic abnormalities have been ruled out. DCI is defined as symptomatic vasospasm, infarction attributable to vasospasm, or both (Frontera et al., 2009). Although about 70% of patients may develop arterial narrowing, only 30% will manifest neurological deficits. The outcome of DCI itself is death in about one-third and permanent deficit in another third (Dorsch, 1995). In the present review the term vasospasm is defined as arterial vessel narrowing.

# **VASOSPASM PATHOPHYSIOLOGY**

#### *Nitric oxide*

Loss of the biological effect of nitric oxide (NO) is considered to play a pivotal permissive role in the development of cerebral vasospasm. The principal effect of NO on cerebral vessels is the relaxation of vascular smooth muscle cells, with decreased bioavailability of NO being implicated in the formation of SAHinduced vasospasm. The depletion of NO has been assumed to occur via several mechanisms in the setting of SAH. First, due to its high affinity for hemoglobin (Hb), NO is scavenged by Hb released during the breakdown of subarachnoid blood (Goretski and Hollocher, 1988; Ignarro, 1990). Second, it is possible that the production of NO is decreased in SAH, as a result of the down-regulation of endothelial NO synthase (eNOS) and neuronal NOS (nNOS; Pluta, 2005). This is supported by studies that revealed the down-regulation/dysfunction of eNOS, and loss of nNOS in spastic arteries after SAH (Hino et al., 1996; Pluta et al., 1996), as well as the finding that levels of asymmetric dimethylarginine (ADMA), an endogenous inhibitor of eNOS, are elevated in the setting of cerebral vasospasm (Jung et al., 2004). Third, NO may reverse the effects of the potent vasoconstrictor endothelin-1 (ET-1; Thomas et al., 1997). Therefore, in the setting of decreased NO levels, the balance of vasodilator and vasoconstrictor influences is altered, and the relatively increased actions of ET-1 can potentiate cerebral vasospasm.

#### *Endothelin-1*

ET-1 is an extremely potent vasoconstrictor. In the brain, it is primarily produced by endothelial cells in response to ischemia, though it can also be produced by neurons, astrocytes, and activated leukocytes (Fassbender et al., 2000; Chow et al., 2002; Dumont et al., 2003). Levels of ET-1 are high in the plasma and cerebrospinal fluid (CSF) of SAH patients, correlate with the persistence of cerebral vasospasm (Seifert et al., 1995; Juvela, 2000), and decline in the absence of vasospasm (Seifert et al., 1995). Conversely, the administration of ET-1 antagonists or endothelin converting enzyme inhibitors prevents vasospasm (Kwan et al., 2002; Macdonald et al., 2008). Lastly, ET-1 induces NADPH oxidase expression and oxidative stress in human endothelial cells (Duerrschmidt et al., 2000).

#### *Inflammation*

Expression of adhesion molecules facilitates leukocyte adherence to the endothelium. Adhesion molecules, such as ICAM-1, VCAM-1, and E-selectin, have been found to be elevated in the CSF of patients with SAH and in blood vessel walls exposed to clot (Polin et al., 1998; Dumont et al., 2003). Leukocytes can contribute to vasospasm by promoting free radical formation that may evoke endothelial dysfunction (Grisham et al., 1998; Sullivan et al., 2000), and by producing a variety of vasoactive substances, including ET-1 and cytokines (Fassbender et al., 2000). Several cytokines have been found to be up-regulated in cerebral vasospasm, including TNF-alpha, IL-1, IL-6, and IL-8 (Hirashima et al., 1997; Fassbender et al., 2001; Takizawa et al., 2001).

# *Oxidative stress*

Oxyhemoglobin (OxyHb) may catalyze generation of reactive oxygen species (ROS). Free radicals are considered to play a pivotal role in cerebral vasospasm through various mechanisms. First, they can initiate lipid peroxidation, whose products, lipid peroxides, are capable of producing vasospasm and damaging the structure of arteries (Lin et al., 2006). Second, it has been hypothesized that ROS can activate the protein kinase C (PKC) pathway directly and indirectly, through enhancement of the metabolism of membrane phospholipids resulting from peroxidative damage. This, in turn, can lead to vasospasm (Asano and Matsui, 1999). Other possible vasoactive compounds are bilirubin oxidation products (BOXes). Once bilirubin is formed, it is subsequently oxidized into BOXes, reaching maximum concentrations during the peak vasospasm period of 4–11 days. They are thought to be potentiators of cerebral vasospasm once it has been initiated, rather than primary initiators (Clark and Sharp, 2006).

#### *Hemoglobin*

A large body of evidence suggests that OxyHb, the ferrous form of hemoglobin, released from lysed erythrocytes, is a mediator of vasospasm. More specifically, OxyHb causes prolonged contraction of isolated cerebral arteries (Toda et al., 1991), and intracisternal injections of this agent result in cerebral vasospasm (Macdonald et al., 1991). Indeed, the presence of OxyHb in the CSF of patients after SAH and the extent of hemorrhage are correlated with the distribution, severity, and time course of vasospasm (Mayberg et al., 1990). Ferrous hemoglobin released from subarachnoid clot could lead to delayed arterial narrowing by a number of mechanisms, such as scavenging or decreased production of NO (Pluta, 2005), free radical production, modification of K<sup>+</sup> and Ca2<sup>+</sup> channels (Ishiguro et al., 2008), differential upregulation of genes (Vikman et al., 2006), and activation of the Rho/Rho kinase and PKC pathways (Wickman et al., 2003).

# *Intracellular Ca***2<sup>+</sup>**

"fendo-03-00135" — 2012/11/9 — 22:03 — page 2 — #2

Vasospasm can be regarded as an abnormal and prolonged contraction of vascular smooth muscle. The intracellular free Ca2<sup>+</sup>

level plays a pivotal role in the regulation of smooth muscle contractility (Horowitz et al., 1996). Following SAH, changes have been reported in the electrical properties of smooth muscle cells of small diameter cerebral arteries leading to enhanced Ca2<sup>+</sup> influx, vasoconstriction, and decreased cerebral blood flow (Koide et al., 2011). Cerebral arteries from healthy animals express only L-type voltage-dependent Ca2<sup>+</sup> channels. Expression of an additional type of voltage-dependent Ca2<sup>+</sup> channels (R-type) occurs after SAH, leading to increased Ca2<sup>+</sup> channel density, increased Ca2<sup>+</sup> influx, and vasoconstriction (Ishiguro et al., 2005).

#### *Cortical spreading depolarization*

This is a pathogenetic process that has attracted much attention lately. The term "cortical spreading depolarization" describes the wave of near-complete neuronal depolarization and neuronal swelling in the brain that is ignited when passive cation influx across the cellular membranes exceeds ATP-dependent Na<sup>+</sup> and Ca2<sup>+</sup> pump activity. The cation influx is followed by water influx and shrinkage of the extracellular space by ∼70% (Dreier et al., 2009). Although the ignition of cortical spreading depolarization occurs passively, driven by electrical and diffusion forces, energy consumption paradoxically increases since Na<sup>+</sup> and Ca2<sup>+</sup> pumps are immediately activated to correct the intracellular Na<sup>+</sup> and Ca2<sup>+</sup> surge. As a consequence, regional cerebral blood flow increases during the neuronal depolarization phase. The opposite of this physiological hemodynamic response to cortical spreading depolarization is termed "the inverse hemodynamic response," and occurs when there is local dysfunction of the microvasculature. With the inverse response, severe microvascular spasm instead of vasodilatation is coupled to the neuronal depolarization phase, and the term "cortical spreading ischemia" describes the cortical spreading depolarization-induced perfusion deficit (Dreier et al., 2009).

## *Neurogenic factors*

The cerebral arteries have sympathetic, parasympathetic, and sensory innervation. It has been postulated that SAH causes a derangement of neuronal regulatory mechanisms, which in turn leads to vascular smooth muscle contraction. The vasoconstriction is associated with a decrease in CGRP levels in cerebral perivascular nerves (Edvinsson et al., 1991) and an increase in CGRP levels in blood draining from the external jugular vein (Juul et al., 1990), suggesting that CGRP is released antidromically from trigeminal sensory perivascular nerves to oppose the vasoconstriction. This evidence has led to the concept that administration of CGRP may be beneficial in SAH-associated vasospasm. The molecular characteristics of CGRP and its use as a treatment option in SAH are reviewed in Sections "Calcitonin Gene-related Peptide Biology" and "Calcitonin Gene-related Peptide and SAH," respectively, of the present article.

#### **TREATMENT OF VASOSPASM**

The management of vasospasm involves routine "prophylactic" measures as well as more aggressive interventions, reserved for situations where there are signs or symptoms of DCI.

#### *Hemodynamic therapy*

The use of triple-H therapy (hypervolemia, hypertension, and hemodilution) stems from numerous clinical observations noting improvement in patients' clinical symptoms following induced hypertension and volume expansion (Kosnik and Hunt, 1976; Kassell et al., 1982). The relative contribution of each component is debated. However, there are many uncertainties for the use of prophylactic hemodynamic therapy following SAH. Two studies randomly assigned normovolemic or hypervolemic therapy to patients and reported no difference in the incidence of DCI between groups (Lennihan et al., 2000; Egge et al., 2001).

### *Nimodipine*

Nimodipine is safe, cost-effective, and reduces the risk of poor outcome and secondary ischemia (Neil-Dwyer et al., 1987; Welty, 1987; Kostron et al., 1988; Mee et al., 1988), but has very modest effects. It is used prophylactically in all patients with SAH. Its precise mechanism of action remains unclear. Despite being shown to reduce the incidence of DCI and cerebral infarction in clinical trials, it has negligible effects on angiographic vasospasm; nimodipine may be neuroprotective by blocking Ca2<sup>+</sup> influx at a neuronal level (Al-Tamimi et al., 2010).

#### *Intracisternal thrombolysis*

A meta-analysis looking at a total of 652 patients who were treated with intracisternal thrombolytics concluded that thrombolytic therapy had a statistically significant beneficial effect. However, the authors acknowledged the lack of large, randomized prospective trials (Amin-Hanjani et al., 2004).

#### *Endovascular techniques*

Endovascular techniques frequently play a role in the aggressive treatment of vasospasm. They include transluminal angioplasty and intra-arterial infusion of vasodilators (papaverin, nicardipine, verapamil, etc.; Brisman et al., 2006). Transluminal balloon angioplasty is very effective at reversing angiographic spasm of large proximal vessels and produces a sustained reversal of arterial narrowing (Brisman et al., 2006; Jestaedt et al., 2008). The optimal timing of angioplasty in relation to medical therapy is uncertain. Major complications occur in ∼5% of procedures and include vessel rupture, occlusion, dissection, hemorrhagic infarction, and hemorrhage from unsecured aneurysms (Zwienenberg-Lee et al., 2006).

#### *Statins*

Statins have been shown to possess cholesterol-loweringindependent pleiotropic effects in different clinical settings, including a decrease in the incidence and duration of severe vasospasm as well as a reduction in the mortality rate after SAH (Lynch et al., 2005; Tseng et al., 2005, 2007). Statins are thought to be beneficial in the prevention of cerebral vasospasm by down-regulating inflammation and up-regulating the expression of eNOS and therefore NO (Sugawara et al., 2011).

#### *Other treatments*

"fendo-03-00135" — 2012/11/9 — 22:03 — page 3 — #3

Clazosentan, an endothelin receptor A (ETA) antagonist decreased the incidence of severe vasospasm, DCI and new infarcts seen on CT scans in a dose-dependent fashion. However, CONSCIOUS 1 study (a phase 2 trial) did not show a reduction in patient mortality, though the study was underpowered for this endpoint (the primary end point of this study was moderate or severe vasospasm within 14 days; Macdonald et al., 2008). CONSCIOUS 2 study (a phase 3 trial) included 1157 patients and its primary composite end point comprised all-cause mortality and vasospasm related morbidity. This study showed that clazosentan at 5 mg/h had no significant effect on mortality and vasospasm-related morbidity or functional outcome (Macdonald et al., 2011).

Erythropoietin (EPO) has also been examined in the setting of cerebral vasospasm. Apart from being potentially neuroprotective, EPO may play a role in preventing vasospasm by increasing the phosphorylation of eNOS (Santhanam et al., 2005), a potentially important mechanism for increasing NO production.

A recent randomized controlled trial (MASH 2) including 1204 patients did not show any benefit from intravenous (i.v.) magnesium sulfate administration in clinical outcome after aneurysmal SAH (Dorhout Mees et al., 2012).

Other drugs under investigation are tirilazad, a free radical scavenger (Haley et al., 1997), fasudil, a Rho-kinase inhibitor that inhibits vascular smooth muscle contraction (Shibuya et al., 1992), sodium nitrite, an NO donor (Pluta et al., 2005) and cisternal placement of prolonged-release nicardipine-loaded polymers (Kasuya et al., 2005).

# **CALCITONIN GENE-RELATED PEPTIDE BIOLOGY CALCITONIN GENE-RELATED PEPTIDE**

Calcitonin gene-related peptide is expressed in a subgroup of small neurons in the dorsal root, trigeminal, and vagal ganglia, which respond to noxious, thermal, or visceral input. These peptidergic neurons use L-glutamate as their primary neurotransmitter and project to the dorsal horn, trigeminal nucleus caudalis, or nucleus of the solitary tract. CGRP increases neurotransmitter release and neuronal responsiveness to noxious stimulation at all these levels, which leads to central sensitization underlying chronic pain states (Benarroch, 2011). CGRP can also be released antidromically in the periphery, eliciting vasodilatation as a component of neurogenic inflammation. CGRP may be involved in the pathophysiology of inflammatory and neuropathic pain. Involvement of CGRP in migraine headache has led to the development of CGRP antagonists for treatment of this disorder (Benarroch, 2011).

Calcitonin gene-related peptide is a 37-amino acid neuropeptide that was identified in 1982 by molecular biological techniques in the thyroid of aging rats and medullary thyroid carcinomas in humans, which were found to contain an alternative peptide product from the calcitonin gene (Amara et al., 1982). CGRP, in common with other members of this peptide family, is derived from the calcitonin gene. Other members of this family include adrenomedulin (AM), which is a potent vasodilator, amylin (AMY), which is important for maintaining glycemic control, and calcitonin, which contributes to calcium metabolism (Hay, 2007). CGRP exists in two forms, named αCGRP and βCGRP. While these two isoforms share the same biological activities, and differ by only three amino acids in the human (Steenbergh et al., 1985, 1986), they are formed from two distinct genes, which share >90% homology, at different sites on chromosome 11. CALC I gene forms calcitonin and αCGRP, whereas CALC II forms βCGRP (Alevizaki et al., 1986). αCGRP synthesis is caused by alternative splicing of the calcitonin gene (Amara et al., 1982; **Figure 1**). βCGRP is known to be transcribed from its own distinct gene (Steenbergh et al., 1985, 1986). The majority of CGRP within the body is αCGRP and primarily expressed in the peripheral and central nervous system. βCGRP is mainly expressed in the gut (Mulderry et al., 1988). However, it has also been identified in the central nervous system, pituitary, thyroid, and in medullary thyroid carcinoma as a major CGRP form together with αCGRP (Petermann et al., 1987).

Data from NMR studies suggest that CGRP consists of a characteristic N-terminal disulfide bridge-linked loop between cysteines Cys2 and Cys7, followed by an alpha-helix in amino acids Val8- Arg18 (Breeze et al., 1991). The next domain at residues 19–27 forms a hinge region (Conner et al., 2002). The C-terminus lies at residues 28–37, and contains two turn regions which form a putative binding epitope (Carpenter et al., 2001). It appears that the N-terminal cyclic portion of the CGRP molecule, containing a ring structure with a disulfide bond, is essential for agonistic activity (Maggi et al., 1990). It is interesting to note that the C-terminal fragment, CGRP8−37, is devoid of any agonist activity at CGRP receptors, although it behaves as a competitive antagonist against the intact peptide (Chiba et al., 1989).

Calcitonin gene-related peptide is widely distributed in the central and peripheral nervous systems, primarily in sensory fibers that are closely associated with blood vessels (Uddman et al.,1986). CGRP is often co-localized with other peptides in these fibers, especially the tachykinin substance P (Uddman et al., 1986). In the cerebral circulation, CGRP is released from sensory fibers originating in the trigeminal ganglia and acts to dilate cerebral vessels (McCulloch et al., 1986). In the gut, CGRP is also released from spinal afferents, where it dilates mucosal blood vessels and may protect against the acidic environment (Holzer, 2000). CGRPcontaining fibers also innervate coronary arteries of the heart (Gulbenkian et al., 1993).

The regulation of CGRP production is poorly understood. At a cellular level, nerve growth factor (NGF) up-regulates CGRP via the Ras/Raf/mitogen-activated protein kinase kinase-1 (MEK-1)/p42/p44 pathway (Freeland et al., 2000).

In the human circulation, CGRP has a half-life of approximately 7–10 min (Kraenzlin et al., 1985; Struthers et al., 1986). Regarding its metabolism, it seems that there is not an obvious mechanism, and it is probably broken down via a number of routes. First, mast cell tryptase has a potent effect in cleaving CGRP into inactive fragments, both *in vivo* and *in vitro*. More specifically, if both CGRP and substance P are released simultaneously, then CGRP could be inactivated by enzymes (tryptases), released by mast cells in response to substance P. This mechanism has been demonstrated in skin (Brain and Williams, 1988, 1989). Second, a matrix metalloproteinase II has the ability to metabolize CGRP and remove its vasodilator activity (Fernandez-Patron et al., 2000). Third, Sams-Nielsen et al. (2001) have provided evidence that CGRP is taken back up into sensory nerve terminals after repolarization *in vitro*. Finally, in the CSF, αCGRP is degraded by an endopeptidase that cleaves the peptide at the Leu*l*6-Ser17 bond (Le Greves et al., 1989).

"fendo-03-00135" — 2012/11/9 — 22:03 — page 4 — #4

"fendo-03-00135" — 2012/11/9 — 22:03 — page 5 — #5

#### **STRUCTURE OF CGRP RECEPTORS**

Many peptides, including the CGRP family, mediate their actions via G protein-coupled receptors (GPCR). The GPCRs form the largest family of cell-surface proteins that are capable of interacting with an extracellular stimulus and transducing that stimulus to produce a reaction inside a cell (Pierce et al., 2002). All GPCRs have seven transmembrane spanning domains, an extracellular N-terminus and an intracellular C-terminus and can be divided into three families based on signature amino acid sequences. Family A is the largest and generally binds small molecules and short peptides. Receptors in this class have been studied extensively, including photoreceptor rhodopsin, as well as adrenergic and olfactory receptors. Family B receptors bind larger peptides in the range of 27 to approximately 50 amino acids (secretin, glucagons, VIP, etc.). These receptors mediate the actions of CGRP and related peptides (Poyner et al., 2002; Hoare, 2005). Family C receptors include glutamate and GABAB receptors (Pierce et al., 2002).

Calcitonin receptor-like receptor (CLR), which belongs to family B of the GPCRs, comprises the main functional unit of the CGRP receptor (**Figure 2**). It was not until McLatchie's work (McLatchie et al., 1998) was published that it was recognized that a novel family of single transmembrane domain proteins, called receptor activity-membrane proteins (RAMP), were required to allow CLR to bind peptide and transduce signal. Three RAMPs have been identified so far (RAMP1, RAMP2, and RAMP3). Each RAMP has a single transmembrane-spanning domain, a short intracellular C-terminal tail (∼9 amino acids) and a long extracellular-terminus (∼100 amino acids; McLatchie et al., 1998). As a result of CLR and calcitonin receptor (CTR) interactions with RAMP, the International Union of Pharmacology (IUPHAR) nomenclature recognizes that CGRP interacts with CLR/RAMP1 (CGRP1) receptors, whereas AM interacts with CLR/RAMP2 (AM1) or CLR/RAMP3 (AM2) receptors. The CTR without RAMP is sufficient for calcitonin binding, but CTR with RAMP 1, 2, or 3 are AMY1, AMY2, and AMY3 receptors, respectively (Poyner et al., 2002). The discovery of RAMPs has led to evolution of our understanding of how receptor diversity is implemented, providing a novel mechanism for generating receptor subtypes within a subset of family B GPCRs (Sexton et al., 2006).

The primary function of CLR is thought to be related to ligand binding, whereas the RAMP molecule plays a crucial role in receptor trafficking to the membrane and determination of receptor pharmacology. The RAMP family regulate the glycosylation and transport of the CLR. However, they are not CGRP receptors by themselves (McLatchie et al., 1998; Sexton et al., 2009). Terminal glycosylation of the receptor and transit from the endoplasmic reticulum/Golgi apparatus to the cell surface require interaction of CLR with RAMP (Sexton et al., 2009).

Calcitonin gene-related peptide receptor activation is known to involve several crucial elements, in common with other GPCRs, such as the presence of a proline "kink" in transmembrane helix (TM)6 (Conner et al., 2005), and a putative'DRY'motif equivalent (Conner et al., 2007), similar to family A GPCRs. There is also evidence suggesting stabilization of the CLR interaction with G "alpha" s (Gas) by another 17kDa intracellular membrane protein, called RCP (Evans et al., 2000).

The existence of two receptors, CGRP1 and CGRP2, was originally proposed in the late 1980s, with the CGRP1 receptor being the predominant mediator of cardiovascular effects. This receptor classification was developed as a consequence of pharmacological studies carried out with different agonists and antagonists in a range of tissue preparations, especially the positive inotropic effect in the guinea pig or rat atrium for determination of CGRP1 receptor activity, and the inhibition of electrically evoked twitch responses in the rat vas deferens for determination of CGRP2 receptor activity (Dennis et al., 1989, 1990; Dumont et al., 1997). In general, receptors that can be antagonized by the 30-amino acid fragment of CGRP, CGRP8−37, with an approximate pA2 value of 7.0 are designated as CGRP1 receptors, while those that CGRP8−<sup>37</sup> block with a pA2 of 6.0 or less are classified as CGRP2 receptors (Quirion et al., 1992; Poyner, 1995). However, it is questionable whether the CGRP2 receptor is a single receptor type or whether it is, in fact, explained by multiple molecular entities (Hay, 2007).

"fendo-03-00135" — 2012/11/9 — 22:03 — page 6 — #6

**FIGURE 2 | Structure of CGRP receptor.** CGRP receptor components and important residues for receptor signaling and internalization. The CGRP receptor is formed by CLR (blue), RAMP1 (yellow), and RCP (orange). Functionally important residues are shown as single letter abbreviations.

CGRP, calcitonin gene-related peptide; CLR, calcitonin receptor-like receptor; RAMP, receptor activity-modifying protein; RCP, receptor component protein; C , C-terminal; EC, extracellular loop; ICL, intracellular loop; N , N-terminal; TM, transmembrane. From Walker et al. (2010), with permission.

In contrast, CGRP1 is a well-defined receptor type consisting of CLR and RAMP1.

#### **SIGNAL TRANSDUCTION OF CGRP RECEPTOR**

Several mechanisms involved in CGRP-mediated vasorelaxation have been identified. These mechanisms include either NO-dependent endothelium-dependent mechanisms or cAMPmediated endothelium-independent pathways. The most common pathway is NO- and endothelium-independent. Activation of the CGRP receptor is generally accepted to result in Gas-mediated activation of adenylate cyclase, with a subsequent increase in cAMP and activation of protein kinase A (PKA). In the absence of endothelium, CGRP is able to cause relaxation, suggesting it must directly act on the smooth muscle cells to stimulate adenylate cyclase (Edvinsson et al., 1985, 1998; Crossman et al., 1990). The resulting rise in cAMP then activates PKA, which phosphorylates and opens up ATP-sensitive K+ channels, thus leading to relaxation (**Figure 3A**; Nelson et al., 1990).

Endothelium-independent relaxation to CGRP occurs in the majority of tissues examined to date. Exceptions include the rat aorta, where the relaxation to CGRP occurs only in the presence of an intact endothelium and is attenuated by inhibitors of NO synthase, implying an NO-dependent mechanism (Brain et al., 1985; Gray and Marshall, 1992a,b). A significant increase in both cAMP and cGMP occurs and is also dependent on the presence of endothelium (Gray and Marshall, 1992a). This implicates the release of NO from the endothelium, which then relaxes the smooth muscle cells through activation of guanylate cyclase and accumulation of cGMP. Moreover, it has been shown that cAMP is able to stimulate eNOS activity, leading to increased synthesis and release of NO (Ferro et al., 1999; Queen et al., 2000). The activation of eNOS via cAMP is probably mediated via PKA, as a study demonstrated that various protein kinases can phosphorylate and activate eNOS (Butt et al., 2000). It is a possibility that CGRP causes an increase in cAMP in endothelial cells, which leads to PKA activation. PKA, in turn, activates eNOS, which results in NO release, and thus relaxation of the smooth muscle (**Figure 3A**).

There is some evidence for Gai/<sup>o</sup> signaling by the CGRP receptor, which is traditionally identified by sensitivity to pertussis toxin (PTX; **Figure 3B**). The CGRP-mediated stimulation of Ca2<sup>+</sup> transients in rat nodose neurons and the activation of c-Jun N-terminal kinase (JNK) in SK-N-MC cells (which express endogenous CGRP receptors) both displayed PTX sensitivity (Wiley et al., 1992; Disa et al., 2000).

"fendo-03-00135" — 2012/11/9 — 22:03 — page 7 — #7

**FIGURE 3 | CGRP receptor-mediated intracellular signaling. (A)** Gas signaling increases AC (green) activity, elevating intracellular cAMP, activating PKA and subsequently many potential downstream effectors. **(B)** The CGRP receptor might also couple to Gai/o, reducing AC (red) activity, decreasing intracellular cAMP and reducing PKA activity. **(C)** CGRP signaling via Gaq activates PLC-b, which cleaves PIP2 into IP3 and DAG, resulting in elevated intracellular Ca2<sup>+</sup> and PKC activation. **(D)** The CGRP receptor might also utilize Ga-independent signaling, and Gβγ- or b-arrestin-mediated signaling

pathways. Arrows represent reported pathways; broken arrows represent potential or inferred pathways. CGRP, calcitonin gene-related peptide; CLR, calcitonin receptor-like receptor; Gα, α subunit of the G protein; NO, nitric oxide; NOS, nitric oxide synthase; AC, adenylyl cyclase; cAMP, cyclic adenosine monophosphate; PKA, protein kinase A; PKC, protein kinase C; RCP, receptor component protein; AC, adenylate cyclase; ER, endoplasmic reticulum; PIP2, phosphatidylinositol 4,5-bisphosphate; DAG, diacylglycerol. From Walker et al. (2010), with permission.

The CGRP receptor may also be able to stimulate intracellular activity through a different G protein. Aiyar et al. (1999) reported that CGRP was able to activate phospholipase C (PLC) in HEK293 cells, leading to an increase in intracellular Ca2<sup>+</sup> via inositol trisphosphate (IP3) activity. This increase in Ca2<sup>+</sup> occurred concurrently with the stimulation of adenylyl cyclase and accumulation of cAMP. Activation of PLC is considered to occur through Gq/11α, rather than through Gαs, suggesting that the activated CGRP receptor is able to interact with both types of G protein. If this mechanism is present in endothelial cells, it provides an alternative explanation for CGRP activation of eNOS (which is traditionally considered to be dependent on Ca2+/calmodulin for activation), independently of cAMP accumulation. The possibility that CGRP receptors may be coupled to phosphatidylinositol turnover is supported by another study that found this secondary messenger pathway in skeletal muscle (Laufer and Changeux,1989; **Figure 3C**).

Recently, Meens et al. (2012) reported that activated CGRP receptors induce cyclic nucleotide-independent relaxation of vascular smooth muscle cells in mesenteric resistance arteries and terminate arterial effects of ET-1 via Gβγ. More specifically, CGRP receptor activation causes cAMP production but the relaxation of rat mesenteric resistance arteries induced by activation of this receptor involves Gβγ and is not dependent on cAMP (**Figure 3D**).

Another study by Meens et al. (2010) discovered that CGRP released from peri-arterial sensory motor nerves terminates longlasting vasoconstrictor effects of ET-1 by promoting dissociation of ET-1/ETA-receptor complexes.

The CGRP receptor can also potentially activate other downstream signaling molecules, such as PKC and mitogen-activated protein kinase (MAPK) cascades, such as p38, JNK, and extracellular receptor activated kinase 1/2 (ERK ½; Walker et al., 2010). CGRP receptor signaling is regulated by desensitization, internalization, and trafficking, which, as with other GPCRs, involves GPCR kinases (GRK), β arrestin, and clathrin- and dynamindependent endocytosis (Walker et al., 2010). Padilla et al. (2007) proposed a mechanism by which endosomal endothelin converting enzyme-1 (ECE-1) degrades CGRP in endosomes to disrupt the peptide/receptor/β-arrestin complex, freeing internalized receptors from β-arrestins and promoting recycling and resensitization, resulting in long-lasting vascular relaxing response to CGRP.

#### **CALCITONIN GENE-RELATED PEPTIDE AND SAH**

#### **PRELIMINARY OBSERVATIONAL STUDIES**

An animal study of experimental SAH in rats revealed that the sensory innervation of the cerebral circulation by CGRP-containing fibers appeared to be reduced after SAH (estimated by the number of fibers present), and there was also a larger vasodilating response to CGRP in basilar arteries after SAH as compared to vessels from control animals. The reduction in CGRP could be due to release of the transmitter from the perivascular nerve terminals caused by blood in the subarachnoid space (Edvinsson et al., 1990).

In another study (Edvinsson et al., 1991), the proximal parts of the middle cerebral artery (MCA) were collected within 24 h after death from five humans suffering SAH (5−10 days beforehand) and from six subjects dying from myocardial infarction. In

"fendo-03-00135" — 2012/11/9 — 22:03 — page 8 — #8

humans who had died from SAH the level of CGRP was nearly not detectable, being in contrast to that seen in age and sex matched subjects who had died of myocardial infarction. The trigeminocerebrovascular system was suggested by the authors to act as an anti-vasoconstrictor system by releasing stored peptides, CGRP being the most likely candidate.

Juul et al. (1995) measured CGRP levels with specific radioimmunoassays (RIA) in patients with SAH, after operation with aneurysm clipping and nimodipine treatment. They used samples taken either from the external jugular vein (*n* = 20) or from the CSF (*n* = 14) during the postoperative course. They also used samples from healthy volunteers. The degree of vasoconstriction in the patients was monitored with Doppler ultrasound recordings. CGRP concentrations from the external jugular vein were significantly higher than from controls. Also, the CGRP level was measurable in SAH CSF but not in CSF of controls.

Others (Tran Dinh et al., 1994) showed that the basal level of endogenous CGRP in CSF was 0.77 nmol/L in rabbits. The CGRP concentration peaked at 14 nmol/L within 30 min, and at 8 nmol/L within 24 h, after SAH. They further showed that 3 days after SAH the CGRP concentration in CSF declined to 3.5 nmol/L.

Nozaki et al. (1989a) produced a model of SAH by a single injection of fresh autologous arterial blood into the cisterna magna of dogs. Then, they examined changes of CGRP immunoreactivity immunohistochemically in perivascular nerve fibers of the large pial arteries. CGRP in cerebrovascular nerve fibers was suppressed after SAH. The suppression was first detected on the third day after SAH, and was most marked during the 7th to 14th day. CGRP, however, recovered to a normal level by the 42nd day after SAH.

Arienta et al. (1991) isolated the basilar artery from five rabbits subjected to SAH and five control animals. A mild or severe vasospasm was observed in the basilar artery about 15 min after injection of blood in the cisterna magna, while fluorescence immunohistochemistry revealed a marked decrease of the perivascular nerves containing CGRP in the animals of the experimental group, as compared to the control group.

#### **EFFECTS OF CGRP ADMINISTRATION ON CEREBRAL VASOSPASM AFTER EXPERIMENTAL SAH IN ANIMALS (Table 1)**

Nozaki et al. (1989b) produced experimental SAH in 30 dogs by injecting autologous arterial blood into the cisterna magna. They used two models of injection: in the first, single-injection model, 1 ml/kg of blood was injected on day 0, while 0.5 ml/kg of blood was injected successively 48 h apart in the second, double-injection model, on day 0 and day 2. The diameter of the basilar artery was measured by angiography. The most marked constriction of the basilar artery was seen on day 3 after SAH in the singleinjection model and on day 7 in the double-injection model. When 10−<sup>10</sup> mol/kg of CGRP was administered intracisternally (i.c.) on day 3 in the single-injection model, cerebral vasospasm reversed completely. The effect began to appear 5 min after CGRP administration, continued for 4 h, and disappeared by 24 h after the administration. When CGRP was administered at doses of 10−<sup>11</sup> to 2 <sup>×</sup> <sup>10</sup>−<sup>10</sup> mol/kg on day 7 after SAH in the double-injection model, the cerebral vasospasm was reversed in a dose-dependent manner: 2 <sup>×</sup> <sup>10</sup>−<sup>10</sup> mol/kg of CGRP reversed the vasospasm completely. The effect began to appear 5 min after the CGRP


"fendo-03-00135" — 2012/11/9 — 22:03 — page 9 — #9

**Table 1 | Studies of CGRP administration after experimental SAH in animals.**

i.c., intra-cisternal; i.v., intravenous; AP, arterial pressure; HR, heart rate; SAH, subarachnoid hemorrhage; NA, non-applicable.

administration, continued for 4 h, and disappeared by 24 h. Of note, when the amounts of CGRP mentioned above were administered i.c., both mean arterial blood pressure and heart rate were only slightly increased and returned to the previous levels within several minutes.

In a similar study by Imaizumi et al. (1996), experimental SAH was produced by i.c. injection of arterial blood in rabbits. The animals were treated with intrathecal administration of CGRP 3 days after SAH. The degree of vasospasm and the effect of CGRP were evaluated angiographically by measuring the basilar artery diameter. The basilar artery constricted to 73% of the pre-SAH values 3 days after SAH. Fifteen minutes after 10−<sup>10</sup> mol/kg CGRP injection, the basilar artery dilated from 73 to 117% (*n* = 8), which was significantly larger than 67.1% in the vehicle group (*<sup>n</sup>* <sup>=</sup> 8; *<sup>p</sup>* <sup>&</sup>lt; 0.01). At 6 h after 10−<sup>10</sup> mol/kg CGRP injection, the basilar artery was still dilated to 90% (*p* < 0.05). In the 10−<sup>11</sup> mol/kg CGRP group, the basilar artery was dilated to 87% (*p* < 0.05) 15 min after the injection. The injection of 10−<sup>12</sup> mol/kg CGRP had no significant effect. The dilatory effect in the 10−<sup>10</sup> mol/kg CGRP group was demonstrated up to 6 h after injection. Arterial blood pressure was stable after injection of CGRP.

Toshima et al. (1992) produced SAH in 41 rabbits by injecting i.c. autologous blood. The animals were randomly assigned to five groups and were sacrificed on day 2 post-SAH. Group 1 was the control group. Immediately prior to sacrifice, group 2 and 3 animals received a 2-h i.c. injection of vehicle or CGRP (100 ng/kg/min), respectively. Group 4 and 5 animals received a 2-h i.v. injection of vehicle or CGRP (100 ng/kg/min), respectively. The diameter of basilar artery in group 3 (i.c. CGRP) was significantly larger than that in group 2 (i.c. vehicle, *p* < 0.001). Similarly, the diameter of basilar artery in group 5 (i.v. CGRP) was significantly greater than that in group 4 (i.v. vehicle, *p* < 0.01). Although no significant difference was observed in mean arterial blood pressure between groups 2 and 3 (i.c. groups), there was a significant difference between i.v. groups 4 and 5 (lower in group 5, *p* < 0.01).

Ahmad et al. (1996) implanted a CGRP slow-release tablet i.c., containing either 24 or 153 μg of human αCGRP, 24 h after experimental SAH was induced in rabbits. Following implantation, the CGRP level in the CSF remained elevated for 5 days. The implantation of the tablet almost completely ameliorated angiographic vasospasm. Moreover, no significant systemic hypotension or neurological adverse event was associated with the treatment.

In a similar approach, Inoue et al. (1996) investigated the efficacy of a CGRP slow-release tablet for the prevention of cerebral vasospasm after SAH in monkeys. Experimental SAH was produced by the method of Espinosa et al. (1984). The animal underwent a right frontotemporal craniectomy under sterile conditions. The dura mater was opened, and the arachnoid membrane was microsurgically incised until the ipsilateral internal carotid


"fendo-03-00135" — 2012/11/9 — 22:03 — page 10 — #10

**Table 2 | Studies of CGRP** 

**administration**

 **after** 

**aneurysmal**

 **SAH in humans.** artery (ICA) and proximal portions of the MCA and anterior cerebral artery (ACA) were exposed. An autologous blood clot (1 ml/kg) was then placed around the exposed arteries to produce experimental SAH. For animals in the CGRP (*n* = 5) and placebo (*n* = 5) groups, a total of three tablets (total drug 1200 μg) were ipsilaterally placed under the frontal and temporal lobes at the time of SAH production. In the control group, cerebral vasospasm developed on day 7 (56% as an average of the ICA, MCA, and ACA). In the CGRP group, vasospasm was significantly ameliorated on average (75%, *p* < 0.02). The CGRP concentration in CSF was measurable only on day 7 for the CGRP group (6.5 nmol/L). No significant untoward reactions were recorded.

Toyoda et al. (2000) sought to determine whether adenovirusmediated gene transfer *in vivo* of CGRP, ameliorates cerebral vasoconstriction after experimental SAH. Arterial blood was injected into the cisterna magna of rabbits to mimic SAH 5 days after injection of adenovirus or vehicle. After injection of adenovirus (*n* = 8), there was a 400-fold increase in CGRP in CSF. In rabbits treated with vehicle (controls, *n* = 8), basilar artery diameter after SAH was 25% smaller than before SAH (*p* < 0.0005). In rabbits treated with adenovirus, arterial diameter was similar before and after SAH. Furthermore, treatment of rabbits with adenovirus after experimental SAH prevented spasm of the basilar artery 2 days after SAH.

Likewise, Satoh et al. (2002) investigated whether a delayed treatment with adenovirus encoding CGRP gene, 2 days after experimental SAH, reduces cerebral vasospasm in a doublehemorrhage model (on days 0 and 2) of severe vasospasm in dogs. Severe vasospasm was observed in control SAH dogs (*n* = 12) on day 7, and the mean basilar artery diameter was 53% of baseline. In the group treated with adenovirus (*n* = 8), vasospasm was significantly reduced (the basilar artery diameter was 78% of baseline, *p* < 0.05 compared with the control SAH group). High levels of CGRP were measured in CSF from dogs that received adenovirus (115-fold greater than baseline levels).

Intracisternal gene transfer of CGRP was initially thought to be more useful than i.v. infusion, because the local gene transfer might avoid systemic effects of CGRP and achieve its sustained release into the central nervous system. However, there are several concerns, such as the inflammatory process induced by adenovirus, the difficulty in approaching the target cells in the presence of a large subarachnoid blood clot, and its potential ability for cancerous transformation of the affected cells.

#### **EFFECTS OF CGRP ADMINISTRATION ON CEREBRAL VASOSPASM AFTER SAH IN HUMANS (Table 2)**

Juul et al. (1994) investigated the effect of i.v. CGRP infusion at a rate of 0.6 μg/min in five patients with vasoconstriction in the postoperative course after SAH, where the hemodynamic index (ratio between middle cerebral and ICA mean velocities) was used as an indicator of vasoconstriction. A significant reduction was found in the hemodynamic index during the CGRP infusion as compared to that before infusion (4.3 vs. 6.2, *p* < 0.05). However, no significant change was observed in pulsatility index (another indicator of vasospasm, equal to the difference between the systolic and diastolic flow velocities divided by the mean flow velocity), blood pressure, or consciousness during CGRP infusion. A

Johnston et al. (1990) undertook a multicenter, randomized, placebo-controlled trial to study the safety and efficacy of i.v. CGRP treatment to reverse neurological deficits after surgical clipping of a ruptured intracranial aneurysm. Patients were enrolled if they had postoperative neurological deficit. Patients received CGRP or placebo in random order, 24 h apart. Fifteen patients were eventually included in the study. Infusion started at a rate sufficient to deliver 0.035 μg/min CGRP, and was doubled every 10 min until either a clinical response was obtained or a maximum dose of 1.15 μg/min was reached at 1 h. If the neurological deficit had not deteriorated and the patient had no side-effects by that time, the maximum infusion rate was continued for another 20 min. Regarding neurological changes according to the modified Glasgow Coma Scale, five patients did not improve on either treatment, one improved on both, eight improved on CGRP but not on placebo, and one improved on placebo but not on CGRP. Of the nine patients who showed a treatment preference, eight (88.9%) favored CGRP (*p* < 0.05). The mean duration of neurological improvement was 25 min, after which patients returned to their previous neurological status. There was a significant decrease in both systolic and diastolic blood pressures during the infusion of CGRP.

A larger, multicenter, randomized controlled trial (European CGRP in SAH study, 1992) investigated the effect of a postoperative infusion of CGRP on outcome at 3 months. Patients with aneurysmal SAH who underwent surgery entered the trial if an ischemic neurological deficit developed after the operation. A total of 117 patients entered the study (62 patients received CGRP and 55 standard management). The CGRP-treated patients received the drug by i.v. infusion at a rate of 0.6 μg/min. If systemic hypotension developed, the infusion rate was reduced to 0.45 μg/min, then to 0.3 μg/min, if the hypotension was still apparent. CGRP treatment was given for at least 4 h; patients who showed a satisfactory neurological response continued to receive treatment for up to 10 days (minimum of 4 days). The percentage of patients with a good outcome was slightly but not significantly higher in the CGRP than in the control group. The relative risk of a bad outcome in CGRP-treated compared with control patients was 0.88 (95% CI: 0.60−1.28). Interestingly, only a third of patients randomized to receive CGRP completed treatment, so two-thirds included in the treatment group for the analyses had limited exposure to CGRP, mainly due to arterial hypotension.

# **CONCLUSION**

"fendo-03-00135" — 2012/11/9 — 22:03 — page 11 — #11

The pathogenesis of vasospasm after SAH is complex, multifactorial, and incompletely understood. CGRP has shown promising results both *in vitro* and *in vivo*, mainly in animal models of experimental SAH. However, there is a lack of studies in humans. Systemic hypotension induced by the i.v. administration of the drug seems to be a serious problem. The encouraging results from the i.c. application of CGRP in animals could warrant large studies in humans with CGRP instillation into the subarachnoid space, in order to avoid hypotension and achieve even more efficient dilatation of the cerebral arteries.

# **REFERENCES**


flow in rat skin. *Br. J. Pharmacol.* 97, 77–82.


(2007). Ligand binding and activation of the CGRP receptor. *Biochem. Soc. Trans.* 35, 729–732.


"fendo-03-00135" — 2012/11/9 — 22:03 — page 12 — #12

CGRP2 in vitro bioassays. *Can. J. Physiol. Pharmacol.* 75, 671–676.


et al. (2001). Inflammatory cytokines in subarachnoid haemorrhage: association with abnormal blood flow velocities in basal cerebral arteries. *J. Neurol. Neurosurg. Psychiatry* 70, 534–537.


"fendo-03-00135" — 2012/11/9 — 22:03 — page 13 — #13

cerebrospinal fluid following subarachnoid haemorrhage in man. *Acta Neurochir. (Wien)* 132, 32–41.


regulation. *Proc. Natl. Acad. Sci. U.S.A.* 83, 5731–5735.


"fendo-03-00135" — 2012/11/9 — 22:03 — page 14 — #14

and CGRP2 receptor subtypes. *Ann. N. Y. Acad. Sci.* 657, 88–105.


Human calcitonin gene related peptide: a potent endogenous vasodilator in man. *Clin. Sci. (Lond.)* 70, 389–393.


of intracisternal and intravenous calcitonin gene-related peptide on experimental cerebral vasospasm in rabbits. *Acta Neurochir. (Wien)* 119, 134–138.


(1986). Calcitonin gene-related peptide (CGRP): perivascular distribution and vasodilatory effects. *Regul. Pept.* 15, 1–23.


"fendo-03-00135" — 2012/11/9 — 22:03 — page 15 — #15


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 23 July 2012; accepted: 24 October 2012; published online: 15 November 2012.*

*Citation: Kokkoris S, Andrews P and Webb DJ (2012) Role of calcitonin generelated peptide in cerebral vasospasm, and as a therapeutic approach to subarachnoid hemorrhage. Front. Endocrin. 3:135. doi: 10.3389/fendo.2012.00135*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Kokkoris, Andrews and Webb. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Different distribution of neuromedin S and its mRNA in the rat brain: NMS peptide is present not only in the hypothalamus as the mRNA, but also in the brainstem

# *Miwa Mori 1†, Kenji Mori 1†,Takanori Ida2,Takahiro Sato3, Masayasu Kojima3, Mikiya Miyazato1\* and Kenji Kangawa<sup>1</sup>*

<sup>1</sup> Department of Biochemistry, National Cerebral and Cardiovascular Center Research Institute, Osaka, Japan

<sup>2</sup> Interdisciplinary Research Organization, University of Miyazaki, Miyazaki, Japan

<sup>3</sup> Molecular Genetics, Institute of Life Sciences, Kurume University, Fukuoka, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Etienne Challet, Centre National de la Recherche Scientifique, France Manuel Tena-Sempere, University of Cordoba, Spain

#### *\*Correspondence:*

Mikiya Miyazato, Department of Biochemistry, National Cerebral and Cardiovascular Center Research Institute, 5-7-1 Fujishirodai, Suita, Osaka 565-8565, Japan. e-mail: miyazato@ri.ncvc.go.jp

†Miwa Mori and Kenji Mori have contributed equally to this work.

Neuromedin S (NMS) is a neuropeptide identified as another endogenous ligand for two orphan G protein-coupled receptors, FM-3/GPR66 and FM-4/TGR-1, which have also been identified as types 1 and 2 receptors for neuromedin U structurally related to NMS. Although expression of NMS mRNA is found mainly in the brain, spleen, and testis, the distribution of its peptide has not yet been investigated. Using a newly prepared antiserum, we developed a highly sensitive radioimmunoassay for rat NMS. NMS peptide was clearly detected in the rat brain at a concentration of 68.3 ± 3.4 fmol/g wet weight, but it was hardly detected in the spleen and testis. A high content of NMS peptide was found in the hypothalamus, midbrain, and pons–medulla oblongata, whereas abundant expression of NMS mRNA was detected only in the hypothalamus. These differing distributions of the mRNA and peptide suggest that nerve fibers originating from hypothalamic NMS neurons project into the midbrain, pons, or medulla oblongata. In addition, abundant expression of type 2 receptor mRNA was detected not only in the hypothalamus, but also in the midbrain and pons–medulla oblongata. These results suggest novel, unknown physiological roles of NMS within the brainstem.

**Keywords: neuromedin S, neuropeptide, radioimmunoassay, peptide distribution, brain, brainstem**

"fendo-03-00152" — 2012/11/30 — 10:12 — page 1 — #1

# **INTRODUCTION**

Neuromedin S (NMS) is a neuropeptide originally isolated by our group from the rat brain as an additional endogenous ligand for neuromedin U (NMU) receptors types 1 (NMUR1) and 2 (NMUR2), which were previously identified as the orphan G protein-coupled receptors FM-3/GPR66 and FM-4/TGR-1, respectively (Mori et al., 2005). In rats, NMS and NMU are structurally related peptides composed respectively of 36 and 23 amino-acid residues (Brighton et al., 2004; Mori et al., 2005). They have an identical C-terminal amidated seven-residue sequence that is essential for their agonist activity. NMS and NMU precursors are encoded by separate genes mapped to human chromosomes 2q11.2 and 4q12, respectively (Mori et al., 2005). Because these genes are under different transcriptional control, the mRNAs of NMS and NMU have different patterns of distribution (Brighton et al., 2004; Mori et al., 2005), indicating that they have different physiological roles despite having the same agonist effect on their receptors. Expression of NMS mRNA is observed mainly in the brain, spleen, and testis, and the highest level of expression is found in the hypothalamus (Mori et al., 2005). In the rat brain, NMS mRNA is expressed predominantly in the suprachiasmatic nucleus (SCN) of the hypothalamus, although low-level expression of NMS mRNA is also found in other hypothalamic nuclei, such as the arcuate nucleus (ARC), paraventricular nucleus (PVN), and supraoptic nucleus (SON; Mori et al., 2005). In contrast to the mRNA distribution, the distribution of the NMS peptide in the brain has not yet been elucidated.

Because of the high level expression of NMS mRNA in the hypothalamus, there has been a focus on the role of NMS in the brain. The functions of NMS in the brain have been validated by intracerebroventricular (ICV) administration of the peptide to rats: ICV-administered NMS causes a phase shift in circadian rhythm (Mori et al., 2005), suppresses food intake (Ida et al., 2005), reduces urine volume (Sakamoto et al., 2007), increases milk secretion (Sakamoto et al., 2008), and elevates plasma levels of luteinizing hormone (Vigo et al., 2007). The activity of NMS is approximately 10 times as great as that of NMU in the brain, although ICV-administered NMU induces the same responses (Nakahara et al., 2004; Ida et al., 2005; Mori et al., 2005; Sakamoto et al., 2007, 2008). These findings suggest that NMS plays important roles in the brain. However, it is not certain whether these actions described above are the physiological roles of NMS. Interesting results have been obtained by ICV administration of anti-NMS antibody to rats to neutralize the activity of endogenous NMS. In contrast to the actions induced by NMS administration, ICV pretreatment with anti-NMS IgG, but not anti-NMU IgG, blocks the suppression of food intake induced by both ICV and intraperitoneally administered leptin in rats (Nakahara et al., 2010), and ICV administration of anti-NMS antiserum suppresses suckling-induced milk ejection in lactating rats (Sakamoto et al., 2008), suggesting that endogenous NMS plays physiological roles in the regulation of food intake and milk ejection. However, there is no anatomical evidence supporting these roles in the brain: NMS mRNA expression levels are low in the ARC, PVN, and SON in the hypothalamus (Mori et al., 2005), even though these nuclei are involved in the control of feeding behavior and release of oxytocin, a hormone important for milk ejection. Therefore, to understand completely the physiological role of NMS, it is necessary to elucidate the neural circuits formed by NMS-producing neurons in the brain.

Here, to clarify the physiological significance of NMS, we analyzed the regional distribution of NMS in the rat brain at the peptide level. We established a highly sensitive radioimmunoassay (RIA) specific for rat NMS by using a newly prepared antiserum. We then characterized NMS-immunoreactive molecules separated by reversed-phase high performance liquid chromatography (RP-HPLC) and determined the NMS content of the brain and other tissues. An examination of both NMS content and NMS mRNA expression in the various regions of the brain suggested that NMS neurons project directly from the hypothalamus to the brainstem.

# **MATERIALS AND METHODS**

#### **PREPARATION OF ANTI-RAT NMS ANTISERA**

[Cys21]-rat NMS[1-20], a C-terminally Cys-extended rat NMS fragment (positions 1–20), was synthesized and then conjugated to maleimide-activated mariculture keyhole limpet hemocyanin. Anti-rat-NMS serum (batch OB1321-1) was obtained from a Japanese white rabbit after 10 weekly subcutaneous injections of conjugate with adjuvant; a second batch (OB1321-2) was obtained from another rabbit in the same way. The titers of the antisera were determined by RIA.

# **RADIOIODINATION OF TRACER LIGAND**

A tracer ligand, [Tyr21]-rat NMS[1-20], was synthesized and then radioiodinated by using the lactoperoxidase method (Washimine et al., 1994). After radioiodination, monoiodinated tracer was purified by RP-HPLC on a COSMOSIL 5C18-AR-II column (4.6 × 150 mm; Nacalai Tesque, Japan). The purified tracer was stored at −30◦C after the addition of bovine serum albumin at a concentration of 0.1%.

#### **RIA FOR RAT NMS**

The RIA for rat NMS was performed by using the method used for ghrelin, with minor modifications (Hosoda et al., 2000). Each RIA incubation mixture was composed of 100 μl of standard rat NMS or unknown sample in RIA buffer (50 mM Na2HPO4, 80 mM NaCl, 25 mM EDTA, 0.5% Triton X-100, 0.5% bovine serum albumin, 0.05% NaN3, pH 7.4) and 100 μl antiserum diluted with RIA buffer containing 3.1% dextran T40. OB1321-2 anti-rat-NMS antiserum was used at a final dilution of 1/295,000. After a 24-h incubation, 18,000 cpm of [125I-Tyr21]-rat NMS[1-20] in 100 μl of RIA buffer containing 3.1% dextran T40 was added. Bound and free tracers were separated by precipitation with polyethylene glycol (PEG). After an additional 24-h incubation, 100 μl of 1% γ-globulin in precipitation buffer [50 mM phosphate buffer (pH 7.4) containing 80 mM NaCl and 0.05% NaN3] and 500 μl of

16% PEG-6000 in precipitation buffer were added. The solution was then mixed vigorously and incubated for 15 min. The bound tracer was precipitated by centrifugation at 1500 *g* for 20 min, and then the free tracer was removed by aspiration. All procedures described above were performed at 4◦C. The radioactivity in the precipitation was counted with a γ-counter ARC-600 (Aloka, Japan).

#### **PREPARATION OF TISSUE EXTRACTS FOR RIA**

All procedures using animal experiments were performed in accordance with the Japanese Physiological Society's guidelines for animal care and the National Cerebral and Cardiovascular Center Research Institute Guide for the Care and Use of Experimental Animals. Ten-week-old male Wistar rats were maintained under 12 h light/dark cycles (light on from 7:00 to 19:00). Food and water were provided *ad libitum* except during the fasting experiments. The brains were sampled during the light period at ZT6 (Zeitgeber time, ZT; ZT0 is lights on and ZT12 is lights off) from free-fed and 48 h-fasted rats, and during the dark period at ZT18 from freefed rats. Immediately after decapitation of rats, the brain, spleen, and testis were resected. The brain was dissected into the cerebral cortex, cerebellum, hippocampus–thalamus–striatum, hypothalamus, midbrain, and pons–medulla oblongata by using the method of Glowinski and Iversen (1966). After being weighed, each tissue was minced and then boiled for 5 min in 5× volumes of water. After being cooled, acetic acid was added to make a final concentration of 1 M, and then the boiled tissue was homogenized with a Polytron homogenizer (Kinematica, Switzerland). The supernatant of the extract, obtained after a 10-min centrifugation for at 14,000 *g*, was concentrated by evaporation and then subjected to precipitation with acetone at a concentration of 66% acetone. After removal of precipitates, the supernatant was evaporated to remove the acetone, and then loaded onto a Sep-Pak C18 cartridge (Waters, USA) equilibrated with 0.1% trifluoroacetic acid (TFA). After the cartridge had been washed with 10% acetonitrile (CH3CN) in 0.1% TFA, the absorbed material was eluted with 60% CH3CN in 0.1% TFA, lyophilized, and dissolved in RIA buffer. Portions of samples, equivalent to 20–300 mg wet weight, were subjected to RIA.

The cerebrospinal fluid was collected from three anesthetized 12-week-old male Wistar rats, and then subjected to Sep-Pak purification.

# **CHROMATOGRAPHIC CHARACTERIZATION OF IMMUNOREACTIVE NMS IN RAT BRAINS**

Immediately after peptide extraction, a portion of sample equivalent to 1.2 g wet weight of whole brain was separated by RP-HPLC on a Symmetry 300 C18 column (3.9 × 150 mm; Waters, USA), with a linear gradient from 10 to 60% of CH3CN in 0.1% TFA for 40 min at a flow rate 1 ml/min. The purified cerebrospinal fluid was also subjected to RP-HPLC under the same conditions. The eluates were collected every minute, evaporated, and then lyophilized. The portions of eluate equivalent to 0.5 g wet weight of tissue and 130 μl of cerebrospinal fluid were subjected to RIA for rat NMS. Synthetic rat NMS was also applied to the same RP-HPLC system to compare the retention time with that of immunoreactive (ir)-NMS.

"fendo-03-00152" — 2012/11/30 — 10:12 — page 2 — #2

#### **QUANTITATIVE REVERSE TRANSCRIPTION–POLYMERASE CHAIN REACTION**

The brains were sampled from 10-week-old male Wistar rats and various brain regions were dissected as described above. RNA was isolated by using TRIzol reagent (Ambion, USA). Reverse transcription was performed with a QuantiTect reverse transcription kit (Qiagen, USA). Quantitative PCR was conducted with a LightCycler system (Roche, USA) using a SYBR Premix Ex Taq kit (TaKaRa, Japan). The primer set used for rat NMS was 5 -CTCATCTGTGGTCTGCAAAGAG-3 and 5 -GCATACAGAA-GCAGTAGATGAC-3 , and for rat NMUR1 was 5 -CTCAGTCCA-CTCTATGTACC-3 and 5 -AGAAGAGCAGTATGGTCGTC-3 , and for rat NMUR2 was 5 -ACTTGAACAGCACAGAGGAG-3 and 5 -TTCAAAGTCTGATGTCGGAC-3 . Known amounts of rat NMS, NMUR1, and NMUR2 cDNA were used to obtain a standard curve. The data were normalized against total RNA quantity, wet weight of tissues, or 36B4 level.

#### **RESULTS**

#### **RIA FOR RAT NMS**

To generate specific antibodies, the N-terminal fragment (positions 1–20) of rat NMS was used as antigen, because this region has no sequence homology to other known peptides including rat NMU, a structurally related peptide (**Figure 1A**). Two types of antisera were obtained; antiserum OB1321-2 exhibited a titer against [125I-Tyr21]-rat NMS[1-20] approximately 17 times higher than OB1321-1 (**Figure 1B**). Therefore, the OB1321-2 antiserum was used in the subsequent experiments. On the standard RIA curve, the half-maximum inhibition of [125I-Tyr21]-rat NMS[1-20] binding to antibody by rat NMS was 6.2 fmol/tube, and the peptide was detectable at a low level of 0.25 fmol/tube (**Figure 1C**). The intra- and inter-assay variations were 3.1 and 3.8%, respectively. This antiserum had no cross-reactivity with rat NMU (**Figure 1C**) as well as neuropeptide Y, somatostatin, pituitary adenylate cyclase activating polypeptide, vasoactive intestinal polypeptide, calcitonin gene-related peptide, adrenocorticotropic hormone, α-melanocyte-stimulating hormone, and orexin-A, which are abundantly present in the hypothalamus (data not shown).

In rats, apparent expression of NMS mRNA is found in the brain, spleen, and testis (Mori et al., 2005). Therefore, we quantified NMS peptide in these tissues. The dilution curve generated from whole-brain extract paralleled the standard curve (**Figure 1C**), and the concentration of NMS in rat whole brain was 68.3 ± 3.4 fmol/g wet weight (mean ± SEM, *n* = 3). A small amount of ir-NMS was detected in the cerebrospinal fluid of rats. However, its concentration was not determined, because the value was outside of the range of the standard curve. On the other hand, specific ir-NMS was hard to detect in the spleen and testis (data not shown).

#### **CHROMATOGRAPHIC CHARACTERIZATION OF ir-NMS IN RAT WHOLE BRAIN**

The molecular forms of ir-NMS found in the extract of rat whole brain were characterized by RIA coupled with RP-HPLC. More than 90% of ir-NMS was eluted as a single peak, with a retention time (20–21 min) identical to that of synthetic rat NMS (**Figure 2**).

(open circles) antisera, and then co-precipitated with antibody in a concentration-dependent manner. **(C)** Standard curve of radioimmunoassay for rat NMS using antiserum OB1321-2. [125I-Tyr21]-rat NMS[1-20] binding to antibody was displaced by increasing concentrations of rat NMS (open circles) but not rat NMU (closed circles). B/B0 means tracer bound/tracer bound in zero standards. Inset shows the inhibition of tracer ligand binding to antibody by diluted rat brain extract (open squares).

A very minor ir-NMS was also detected at a retention time of 18– 20 min; this corresponded to that observed for oxidized NMS (data not shown). Non-specific immunoreactivity was not observed in any other eluted fractions.

On the other hand, ir-NMS in the cerebrospinal fluid was also eluted at a retention time identical to that of synthetic rat NMS (data not shown).

#### **REGIONAL DISTRIBUTION OF ir-NMS AND NMS mRNA IN RAT BRAIN**

"fendo-03-00152" — 2012/11/30 — 10:12 — page 3 — #3

The rat brain was dissected into six parts, and both the NMS contents and the NMS mRNA expression levels in these regions were then determined. The highest concentration of ir-NMS was observed in the midbrain (286.5 ± 26.6 fmol/g wet weight, mean ± SEM), followed by the pons–medulla oblongata

(208.8 ± 18.7 fmol/g wet weight) and hypothalamus (196.2 ± 15.2 fmol/g wet weight; **Figure 3**). Non-specific immunoreactivity was not found in these brain regions by RIA coupled with RP-HPLC (data not shown). Low levels of ir-NMS were detected in the other brain regions.

Abundant expression of NMS mRNA was found only in the hypothalamus (4.69 ± 0.52 copies/ng total RNA, mean ± SEM); low levels of mRNA expression were observed in the other regions (**Figure 4**). Normalization of the mRNA levels against the wet weight of tissues or 36B4 expression did not change this mRNA expression profile in the rat brain (data not shown).

#### **REGIONAL DISTRIBUTION OF NMUR1 AND NMUR2 mRNA IN RAT BRAIN**

To elucidate the regional distribution of NMS receptor in rat brain, the mRNA expression level of NMUR1 and NMUR2 was analyzed by quantitative PCR (**Figure 5**). NMUR1 mRNA was widely distributed throughout the brain with low level of expression (approximately 0.8 copies/ng total RNA). On the other hand, abundant expression of NMUR2 mRNA was observed not only in the hypothalamus (7.27 ± 0.81 copies/ng total RNA, mean ± SEM), but also in the thalamus–striatum– hippocampus (3.72 ± 1.11 copies/ng total RNA), midbrain

(4.20 ± 0.16 copies/ng total RNA), and pons–medulla oblongata (4.76 ± 0.40 copies/ng total RNA).

#### **LEVELS OF NMS PEPTIDE IN THE BRAIN OF RATS MAINTAINED UNDER LIGHT/DARK CYCLING, AND OF FASTED RATS**

Most abundant expression of NMS mRNA is found in the SCN within the brain, and its expression level fluctuates under light/dark cycling (Mori et al., 2005). Therefore, we measured the levels of NMS peptide in the hypothalamus, midbrain, and pons– medulla oblongata of rats during both light and dark periods. The brains were sampled during the light period at ZT6 and the dark period at ZT18. In these brain regions, there is no significant difference in peptide levels between groups of rats at ZT6 and ZT18 (**Figures 6A–C**).

On the other hand, because ICV-administered NMS suppresses the food intake (Ida et al., 2005), we examined the effect of fasting on the levels of NMS peptide in the hypothalamus, midbrain, and pons–medulla oblongata. The peptide levels in the brain regions tested did not significantly differ between groups of free-fed and 48-h fasted rats (**Figures 6D–F**).

# **DISCUSSION**

"fendo-03-00152" — 2012/11/30 — 10:12 — page 4 — #4

We developed a highly sensitive RIA for rat NMS; the lowest detection limit was 0.25 fmol/tube (equivalent to 10.6 pg/ml). We evaluated the specificity of the assay by using a peptide extract of rat whole brain, because NMS was originally isolated from this tissue (Mori et al., 2005). The dilution curve for rat brain extract paralleled the standard curve. Moreover, no immunoreactivity other than that originating from native (major) and oxidized (minor) NMS was detected in any of the fractions of extracts

separated by RP-HPLC, suggesting that the OB1321-2 antiserum is not cross-reactive to other neuropeptides that are abundantly expressed in the brain. This is no doubt because NMS shows no sequence homology with any other known neuropeptide except for NMU, a structurally related peptide. In addition, this RIA is unable to detect rat NMU. These data indicate that the assay was highly specific.

Within the rat brain, the distribution profile of ir-NMS was different from that of NMS mRNA. It was interpreted that we clearly detected NMS peptide in the hypothalamus, because its mRNA was abundantly expressed in this region. However, we also detected NMS peptide in the midbrain and pons–medulla oblongata at levels comparable to those in the hypothalamus, whereas the mRNA levels in these regions were much lower than in the hypothalamus. Differences in the distributions of the orexin peptide and its mRNA have also been observed. Expression of orexin mRNA is restricted to the lateral hypothalamic area (LHA) and posterior hypothalamus (PH; Sakurai et al., 1998), whereas orexin peptides are distributed in various areas of the brain (Mondal et al., 1999). Furthermore, immunohistochemical mapping has shown that orexin-containing nerve fibers are abundant in various brain areas, including the olfactory bulb, cerebral cortex, thalamus, hypothalamus, and brainstem (Date et al., 1999). These data indicate that orexin neurons in the LHA or PH send projections widely across the brain (Date et al., 1999). The predominant expression of NMS mRNA only in the hypothalamus thus indicates that the cell bodies of the NMS neurons are located mainly in this region. In addition, NMS-containing nerve fibers are likely to be present in the midbrain and pons–medulla oblongata, because NMS peptide was abundant in these regions despite the low levels of expression

of NMS mRNA. On the other hand, it is likely that there are fewer NMS-producing cells in the spinal cord than in the hypothalamus, because NMS mRNA expression levels are low in the spinal cord, as they are in the brainstem (Mori et al., 2005). Therefore, it is likely that nerve fibers originating from hypothalamic NMS neurons project to the brainstem, which consists of midbrain, pons, and medulla oblongata.

In our previous study, quantitative RT-PCR and *in situ* hybridization analyses showed that NMS mRNA was predominantly expressed in the SCN of the hypothalamus (Mori et al., 2005), indicating that, within the hypothalamus, most NMS neurons are located in this nucleus. Together with this situation, in view of the larger distribution in the brain of NMS peptide compared to NMS mRNA, we speculated the existence of neural circuits from NMS neurons located in the SCN to the brainstem. Little is known that the nerve fibers originating from SCN neurons project directly to the brainstem, although it is reported that the SCN indirectly projects to the locus coeruleus of pons (Aston-Jones et al., 2001) and the ventral tegmental area of midbrain (Luo and Aston-Jones, 2009). The results we described here may lead to the discovery of novel neural circuits. On the other hand, another possibility is that NMS peptide in the brainstem comes from hypothalamic nuclei other than the SCN, such as the PVN that are known to project to various targets in the brainstem (Geerling et al., 2010), whereas the levels of NMS mRNA in the PVN and other hypothalamic nuclei are extremely low compared with that in the SCN (Mori et al., 2005). To clarify these possibilities, immunohistochemical analysis of NMS in the brain is needed, and further investigation is continued in this regard.

"fendo-03-00152" — 2012/11/30 — 10:12 — page 5 — #5

"fendo-03-00152" — 2012/11/30 — 10:12 — page 6 — #6

There are two kinds of receptors for NMS, NMUR1 and NMUR2 (Brighton et al., 2004; Mori et al., 2005), and their expression patterns are unique. Expression of NMUR1 is widely distributed throughout various tissues, and high levels of expression are found in peripheral tissues (Fujii et al., 2000). In contrast, NMUR2 mRNA is mainly expressed in the central nervous system (Hosoya et al., 2000), and its expression is clearly detected in the PVN, the wall of the third ventricle in the hypothalamus, and the CA1 region of the hippocampus (Howard et al., 2000). In addition, regional distribution of these receptors in rat brain was investigated in this study. Quantitative RT-PCR showed the apparent expression of NMUR2 mRNA in the midbrain and pons–medulla oblongata as well as in the hypothalamus. These data suggest the functional possibility of NMS within the brainstem.

We and another group have demonstrated that ICV administration of NMS induces a phase shift in the circadian rhythm (Mori et al., 2005), suppression of food intake (Ida et al., 2005), reduction of urine volume (Sakamoto et al., 2007), increased in milk ejection (Sakamoto et al., 2008), and release of luteinizing hormone (Vigo et al., 2007). All of actions are induced as a result of the direct effects of NMS on hypothalamic nuclei such as the SCN, ARC, PVN, and SON (Ida et al., 2005; Mori et al., 2005; Sakamoto et al., 2007, 2008; Vigo et al., 2007). On the other hand, NMS peptide and NMUR2 mRNA are present in the brainstem, suggesting the physiological role of NMS system in this region. However, the function of NMS by direct action on the brainstem remains to be clarified. The brainstem is important for basic bodily functions (especially regulation of the cardiovascular and respiratory systems). Therefore, regulation of these systems may be a major role of NMS within the brainstem, although it is not yet certain whether this action is a result of a direct effect of NMS in this region. On the other hand, our group recently reported that, in mice, NMS regulates the heart rate through the sympathetic nervous system (Sakamoto et al., 2011).

In the rat brain, NMS mRNA is predominantly expressed in the SCN, with weak expression in other nuclei of hypothalamus, and its expression fluctuates within the SCN under light/dark cycling: the largest decrease occurs during the dark period, followed by a gradual increase until the late light period (Mori et al., 2005). In contrast, there is no difference in hypothalamic content of NMS peptide between groups of rats during the light and dark periods. In general, the neuropeptide is stored in the secretory vesicles of neuron, and released in response to relevant stimuli. In view of this mechanism, discrepancy between mRNA and peptide levels of NMS may be caused by mechanism of storage and release of the neuropeptide as well as differences in mRNA and peptide turnover rates, because tissue content of peptide is modulated by the presence or absence of releasing stimuli, even if the expression level of mRNA is constant. On the other hand, hypothalamic content of NMS peptide was not influenced by fasting, although NMS is an anorexigenic peptide (Ida et al., 2005). To consider the relationship between hypothalamic content of peptide and physiology of NMS, further investigation is needed.

The negligible detection of NMS peptide in the spleen and testis was unexpected, because NMS gene expression levels in these tissues are higher than that in the whole brain (Mori et al., 2005). Several neuropeptides are produced in the spleen and testis. For example, somatostatin and its mRNA are detected in the spleen and pituitary adenylate cyclase-activating polypeptide and its mRNA are detected in the testis, indicating that both neuropeptides are synthesized and stored until release in these tissues (Aquila et al., 1991; Shioda et al., 1994). In contrast, because only NMS mRNA – not its peptide – was detected in the spleen and testis, NMS peptide accumulation is likely to be difficult in these tissues because NMS may be secreted *via* constitutive secretory pathways in the spleen and testis, similar to the way in which adrenomedullin is secreted from both endothelial and vascular smooth muscle cells (Sugo et al., 1994a,b). This may occur because the sequences of the cleavage sites involved in post-translational modification

#### **REFERENCES**


of the NMS precursor coincide with those of the consensus motifs (Arg-X-Lys/Arg-Arg and Arg-X-Arg-X-X-Arg) for proteolytic processing by furin and/or furin-like proteases (proprotein convertases functioning within constitutive secretory pathways; Seidah and Chrétien, 1999; Mori et al., 2005). On the other hand, there is another possible explanation as follows: if NMS mRNA is highly expressed in a discrete cell population within the spleen and testis, the peptide gets massively diluted when whole organ is homogenized. To verify these hypotheses, it is important to identify the NMS-producing cells in the spleen and testis by *in situ* hybridization analysis and immunohistochemistry.

In conclusion, we developed a highly sensitive and highly specific RIA for rat NMS. The different distribution of NMS peptide and its mRNA within the brain suggested the direct projection of nerve fibers originating from hypothalamic NMS neurons to the brainstem. In addition, the gene expression of NMS receptor, NMUR2, was clearly detected in the brainstem. Because the functions of NMS within the brainstem have not yet been identified, our findings provide a foundation on which further research can build to provide novel insights into the physiological roles of NMS in the brain.

#### **ACKNOWLEDGMENTS**

We thank Dr. Kazuhiko Harada for helpful support. This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture of Japan, the Improvement of Research Environment for Young Researchers Program of the Ministry of Education, Science, Sports, and Culture of Japan, a grant for Scientific Research on Priority Areas from the University of Miyazaki, and the Takeda Science Foundation.


"fendo-03-00152" — 2012/11/30 — 10:12 — page 7 — #7


neuropeptides and G proteincoupled receptors that regulate feeding behavior. *Cell* 92, 573–585.


Sakata, J., et al. (1994a). Endothelial cells actively synthesize and secrete adrenomedullin. *Biochem. Biophys. Res. Commun.* 201, 1160– 1166.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 August 2012; accepted: 16 November 2012; published online: 03 December 2012.*

*Citation: Mori M, Mori K, Ida T, Sato T, Kojima M, Miyazato M and Kangawa*

"fendo-03-00152" — 2012/11/30 — 10:12 — page 8 — #8

*K (2012) Different distribution of neuromedin S and its mRNA in the rat brain: NMS peptide is present not only in the hypothalamus as the mRNA, but also in the brainstem. Front. Endocrin. 3:152. doi: 10.3389/fendo.2012.00152*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Mori, Mori, Ida, Sato, Kojima, Miyazato and Kangawa. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Diverse roles of neurotensin agonists in the central nervous system

# **Mona Boules\*, Zhimin Li, Kristin Smith, Paul Fredrickson and Elliott Richelson**

Neuropsychopharmacology Laboratory, Department of Neuroscience, Mayo Clinic Florida, Jacksonville, FL, USA

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Sarah J. Spencer, RMIT University, Australia Jean H. Costentin, University of Rouen, France

#### **\*Correspondence:**

Mona Boules, Neuropsychopharmacology Laboratory, Department of Neuroscience, Mayo Clinic, 4500 San Pablo Road, Jacksonville, FL 32224, USA. e-mail: boules.mona@mayo.edu

**INTRODUCTION**

Neurotensin (NT) is a 13 amino acid neuropeptide (*p*Glu-Leu-Tyr-Glu-Asn-Lys-Pro-Arg-Arg-Pro-Tyr-Ile-Leu) that was first isolated from bovine hypothalamus by Carraway and Leeman (1973). It is abundant in gastrointestinal tract where it plays a role in gut motility. NT has also been detected in peripheral organs including: heart, liver, lung, pancreas, spleen, and small intestine. It is a paracrine and endocrine modulator of the cardiovascular system and of the digestive tract and acts as a growth factor on a variety of normal and cancerous cells. In addition, NT is widely distributed in the central nervous system (CNS), with the highest concentration in the amygdala, lateral septum, ventral tegmental area (VTA), and substantia nigra (SN).

In brain, NT plays a role in naloxone-independent antinociception (Clineschmidt et al., 1979), hypothermia (Bissette et al., 1976), control of anterior pituitary hormone secretion, and muscle relaxation (Kitabgi et al., 1992). NT also controls central blood pressure, and inflammation (St-Gelais et al., 2006 for review). The physiological functions of NT have been recently reviewed (Mustain et al., 2011).

Many literature reviews have presented the basic role of brain NT through its relationship with dopaminergic mesotelencephalic projection system and its modulating glutamatergic transmission and other signals (Seutin, 2005; Boules et al., 2006; Antonelli et al., 2007; Ferraro et al., 2008). The current review discusses the potential clinical function of NT and the potential clinical use of its analogs as novel therapy for neuropsychiatric disorders such as schizophrenia, addiction, Parkinson's disease (PD), and pain.

# **NEUROTENSIN RECEPTORS**

Neurotensin mediates its effects through three NT receptors (NTRs), the high affinity NTS1, the low affinity NTS2, and NTS3. Both NTS1 and NTS2 are G-protein coupled receptors, with seven-transmembrane domain. NTS1, the most studied NTR, is expressed in both neurons and glial cells and is broadly

Neurotensin (NT) is a tridecapeptide that is found in the central nervous system (CNS) and the gastrointestinal tract. NT behaves as a neurotransmitter in the brain and as a hormone in the gut. Additionally, NT acts as a neuromodulator to several neurotransmitter systems including dopaminergic, sertonergic, GABAergic, glutamatergic, and cholinergic systems. Due to its association with such a wide variety of neurotransmitters, NT has been implicated in the pathophysiology of several CNS disorders such as schizophrenia, drug abuse, Parkinson's disease (PD), pain, central control of blood pressure, eating disorders, as well as, cancer and inflammation. The present review will focus on the role that NT and its analogs play in schizophrenia, endocrine function, pain, psychostimulant abuse, and PD.

**Keywords: neurotensin, endocrinology, schizophrenia, pain, psychostimulant abuse, central nervous system**

distributed in the CNS including medial septal nucleus, nucleus basalis magnocellularis, suprachiasmatic nucleus, SN, and VTA (Elde et al., 1990) as well as small dorsal root ganglion neurons of the spinal cord (Zhang et al., 1995). Activation of NTS1 induces excitatory effect through G-proteins, resulting in intracellular calcium influx. In turn, there is increased intracellular levels of cAMP, cGMP, and IP3; and increased activities of PLC, PKC, and Na+/K+-ATPase (Gilbert and Richelson, 1984; Watson et al., 1992; Yamada and Richelson, 1993; Hermans et al., 1994; Slusher et al., 1994; Poinot-Chazel et al., 1996; Lopez Ordieres and Rodriguez de Lores Arnaiz, 2000; Trudeau, 2000). NTS1 has a close association with dopaminergic and glutamatergic systems.

A high density of NTS1 receptors is co-localized with dopamine (DA) neurons in the ventral mesencephalon (Brouard et al., 1992; Nicot et al., 1995; Boudin et al., 1998). NT antagonizes DA effects at D2 receptors via NTS1/D2 receptor–receptor interaction (Jiang et al., 1994; Farkas et al., 1996). Several studies suggested that NTS1 receptors on mesencephalic DA neurons play an important role in sensitization to psychostimulant drugs and drug addiction (Binder et al., 2001a; Berod and Rostene, 2002; Panayi et al., 2005). Our recent work shows that (1) NTS1 knockout mice (NTS1−/−) mice are spontaneously hyperactive and more sensitive to d-amphetamine-induced hyperactivity (Liang et al., 2010); (2) NTS1−/<sup>−</sup> mice have higher basal levels of DA and higher levels of amphetamine-induced DA release in striatum as compared with results for WT mice (Liang et al., 2010); and (3) NTS1−/<sup>−</sup> have abnormal D2/D1 ratio in striatum (Liang et al., 2010), leading to a decrease of d-amphetamine-induced glutamate and GABA release in striatum (Li et al., 2010b). The results suggest that NTS1 through possible modulation of DA receptor plays an important role in the dysregulation of striatal DA function which is thought to be secondary to a glutamate deficiency in schizophrenia through possible modulation of DA receptor. Additionally, a possible interaction between NTS1 and D1 receptor in medial prefrontal cortex (mPFC) has been proposed in one of our recent studies (Li et al., 2010c). The same work also shows that there are significantly lower basal glutamate levels and lower density of *N*-methyl-d-aspartate (NMDA) receptor 2A subunit in the mPFC of NTS1 null mice as compared with results for WT mice (Li et al., 2010c). These data are consistent with the hypothesis that NTS1 is involved with the pathophysiology of the hypofunction of the glutamatergic system in schizophrenia.

The relationship between NTS1 and glutamatergic system has been investigated by Antonelli's group for years. Activation of NTS1 promotes and reinforces endogenous glutamate signaling in discrete brain regions by increasing the activation of PKC and leading to phosphorylation of the NMDA receptor. NTS1 receptors are highly expressed in nigro-striatal dopaminergic neurons, which degenerate in PD. It has been hypothesized (Antonelli et al., 2007) that in Parkinson's patients, possible elevated NT levels in the basal ganglia cause the NTS1 enhancement of excitotoxic glutamate signaling contributing to the DA neuron neurodegeneration in SN. However, paradoxically, NT or an NTS1 agonist has antiparkinson-like effect (Jolicoeur et al., 1991; Boules et al., 2001b). The discrepancy is likely due to different routes and time course of administration: local infusion in the striatum or SN for 60 min versus acute systemic i.p. administration. We have discussed previously that the effects of a NTR agonist given outside the brain cannot be predicted from studies in which a NTR agonist is injected into discrete areas of the brain (Richelson et al., 2003). Both groups agree that there is a critical role for NTS1 in PD.

NTS2 is localized mainly in the olfactory system, the cerebral and cerebellar cortices, the hippocampal formation, and selective hypothalamic nuclei, VTA, and SN. Importantly, strong expression of NTS2 receptors is in the areas related to pain, namely, the periaqueductal gray (PAG) and the rostral ventrolateral medulla (RVLM). Local administration of NT into the PAG and RVLM induces opioid-independent analgesia (Behbehani, 1992) through non-NTS1 dependent mechanisms (Dubuc et al., 1994). Antisense oligonucleotide knockdown of NTS2 receptor significantly decreases NT-induced analgesia, while oligodeoxynucleotides against NTS1 had no effect in this regard. These results are also supported by recent data from NT (Remaury et al., 2002) and the NTS2 receptor-selective agonist NT79 (Boules et al., 2010). Taken together it suggests that NTS2 plays an important role in pain.

NTS3/sortilin is a single transmembrane amino acid receptor, structurally unrelated to either NTS1 or NTS2. Its mRNA is expressed throughout the brain with high levels of expression in the SN, hippocampal formation, supraoptic nucleus, piriform and cerebral cortices, and medial and lateral septal nuclei (Mazella, 2001; Sarret et al., 2003). NTS3/sortilin participates in the modulation of NT intracellular sorting (hence the name "sortilin") and signaling processes (Sarret et al., 2003) and has been associated with the phosphatidylinositol 3-kinase and mitogenactivated protein kinase pathways in glial cells (Martin et al., 2003). NTS3/sortilin has been implicated in cell death in responsive cells including neurons. This is thought to be due to the ability of NTS3/sortilin to bind the unprocessed form of nerve growth factor (proNGF) (Nykjaer et al., 2004). Additionally, SorLA/LR11, a mosaic protein of the vacuolar protein sorting 10 protein (Vps10p) domain receptor family and the low density lipoprotein receptor (LDLR) family, represents the fourth NTR (NTS4). It has a similar structure as NTS3/sortilin, with a wide distribution in the brain, especially in the hippocampus, cerebellum, cingulate gyrus, entorhinal cortex, red nucleus, and oculomotor nucleus (Motoi et al., 1999). NTS4 may be involved in the intracellular trafficking and in termination of NT signaling (Jacobsen et al., 2001).

# **NT AND NEUROTRANSMITTER SYSTEMS**

Neurotensin acts as a primary neurotransmitter as well as a modulator of other neurotransmitter systems such as the dopaminergic, glutamatergic, GABAergic, cholinergic, and serotonergic systems (Rakovska et al., 1998; Ferraro et al., 2008; Petkova-Kirova et al., 2008). The functional and anatomical NT/DA interactions have been the most extensively studied and reviewed (Binder et al., 2001a; Jomphe et al., 2006; Fawaz et al., 2009; Thibault et al., 2011; Tanganelli et al., 2012).

#### **NEUROTENSIN AND DOPAMINE**

As mentioned in the previous section, NT and DA are co-localized in a subpopulation of mesencephalic neuron within the VTA (Hokfelt et al., 1984; Seroogy et al., 1988; Bean and Roth, 1992). Additionally, NT-like immunoreactivity is highly expressed in areas enriched with DA cell bodies and nerve terminals such as SN, VTA, neostriatum, and nucleus accumbens (NA) (Quirion et al., 1982). NT also forms synaptic contacts with a subpopulation of tyrosine hydroxylase (TH) immunoreactive neurons (Woulfe and Beaudet, 1989). These data strongly indicate a modulatory function of NT on DA neurotransmission.

Neurotensin directly or indirectly (e.g., through glutamate release) modulates DA neurotransmission. NT can modulate DA through: (1) up-regulating TH gene expression (Burgevin et al., 1992a,b) or (2) decreasing the DA binding affinity for the DA D<sup>2</sup> receptors (Fuxe et al.,1992a; Li et al.,1995). NT also opposes DA D<sup>2</sup> receptor agonist-induced auto-inhibition of DA cell firing (Shi and Bunney, 1992). Allosteric receptor–receptor interactions between NT and DA D<sup>2</sup> receptors, as well as second messenger-dependent receptor alterations, such as phosphorylation and dephosphorylation, have also been implicated (see Fuxe et al., 1992b for review). It is important to mention that within the terminal fields, NT opposes the effects of DA both pre- and post-synaptically, leading either to an increase or to a decrease in DA transmission, depending on the intrinsic anatomical location of NTRs (for a more comprehensive review of the modulatory effects of NT on DA neurotransmission, we refer the reader to Binder et al. (2001a) and Kinkead and Nemeroff (2004).

#### **NT AND SEROTONIN**

Neurotensin receptors are present on serotonergic neurons in the nucleus raphe magnus and dorsal raphe, where NT causes an increase in their firing rates (Jolas and Aghajanian, 1996; Li et al., 2001). Therefore, NT has been proposed to play a role in functions known to be affected by the serotonergic system including antinociception (Buhler et al.,2005;Boules et al.,2010), sleep-wake cycle (Jolas and Aghajanian, 1997), and stress (Corley et al., 2002).

#### **NT AND GLUTAMATE**

There has been controversy regarding the effect of NT on glutamate release and modification of the glutamatergic and NMDA receptors. NT increases glutamate release in the striatum, globus pallidus, frontal cortex, and SN (Ferraro et al., 2011, 2012) implicating NT in conditions such as stroke, Alzheimer's disease, and PD.

Others have reported that systemic administration of NT analogs blocks ethanol-induced increases in glutamate levels in the striatum (Li et al., 2011), and decreases phencyclidine (PCP) induced increases in glutamate in the prefrontal cortex (Li et al., 2010a). These data support the idea that the antipsychotic-like effects of NT may be mediated in part by its modulatory effect on glutamate.

#### **NT AND HYPOTHALAMIC-PITUITARY AXIS**

#### **EFFECT OF NT ON HORMONE RELEASE**

The presence of NT in the anterior pituitary gland and hypothalamus and its storage and release at the median eminence implicate NT and its receptors in neuroendocrine regulation (Rostene and Alexander, 1997). NT is thought to possess a paracrine or an autocrine role within the hypothalamus and the anterior pituitary (Bachelet et al.,1997;Bello et al.,2004). NT stimulates both directly and indirectly, the synthesis of corticotrophin releasing hormone (CRH), gonadotropin-releasing hormone (GnRH) (Cooke et al., 2009), growth hormone-releasing hormone (GHRH) (Blackburn et al., 1980), and prolactin secretion at anterior pituitary and median eminence (Memo et al., 1986).

#### **CRH-ACTH**

Neurotensin stimulates the activity of the hypothalamic-pituitary CRH-adrenocorticotropin hormone (ACTH) system. Central administration of NT increases ACTH and corticosterone in the presence of CRH receptor activation in the paraventricular nucleus (PVN) (Nussdorfer et al., 1992; Rowe et al., 1995). The modulatory action of NT on the pituitary-adrenocortical function in rats has been reported to be biphasic. The lower dose of NT exerts a stimulatory effect while the higher dose appears to have an inhibitory effect on both the pituitary-ACTH release and adrenocortical secretion (Malendowicz and Nussdorfer, 1994). NT also enhances the release of both CRH-ir and ACTH-ir in rat adrenal medulla (Mazzocchi et al., 1997).

#### **Gonadotropic hormones**

Many neurons in the anteroventral periventricular (AVPV) and medial preoptic nuclei (MPN) express estrogen receptors and project to GnRH neurons (Smith and Wise, 2001). Surprisingly, GnRH neurons do not seem to possess intracellular estrogen receptors (Herbison and Theodosis, 1993) and evidence suggests that other neurons mediate the stimulatory effects of estrogen on GnRH secretion (Shivers et al.,1983). Interestingly,GnRH neurons were found to express NTS1 receptors (Herbison and Theodosis, 1992) and GABA receptors (Herbison et al., 1993). The number of GnRH neurons expressing NTS1-mRNA peaks during proestrus suggesting that NT directly stimulates GnRH neurons contributing to luteinizing hormone (LH) surge (Smith and Wise, 2001). There has been some controversy regarding the role of NT on LH surge. While some report that the administration of NT directly into the medial preoptic area (where GnRH cell bodies reside) evokes pre-ovulatory GnRH/LH surge (Alexander et al., 1989a), others indicate that NT has no effect on GnRH-stimulated LH release (Leiva and Croxatto, 1994). The former group showed that immunoneutralization of NT in the preoptic region attenuates the LH surge induced by estrogen and progesterone treatment in ovariectomized rats (Alexander et al., 1989b). The effect of NT on LH secretion requires intact dopaminergic and alpha adrenergic systems (Akema and Kimura, 1989).

#### **Growth hormone**

Neurotensin is co-expressed with growth hormone (GH) releasing factor in the arcuate nucleus (Niimi et al., 1991) and modulates GH release. NT regulates GH secretion, in prepubertal children and adults (Bozzola et al., 1998a), as well as in neonates (Bozzola et al., 1998b), through the modulation of somatostatin release from the median eminence. In rats, estrogen plays a facilitatory role on NTinduced GH release that is independent of hypothalamic GHRH or somatostatin release (Ibanez et al., 1993).

#### **Prolactin**

With respect to prolactin release, NT has opposite actions, an inhibitory effect at the hypothalamic site and an excitatory effect at the pituitary. NT elevates plasma prolactin and GH levels in both normal and estrogen-progesterone pretreated male rats (Rivier et al., 1977). The inhibitory action of NT on prolactin release is mediated by the release of DA into the hypophyseal portal vein, which delivers the neurotransmitter to the anterior pituitary gland causing inhibition of prolactin release (Vijayan et al., 1988; McCann and Vijayan, 1992; Pan et al., 1992).

#### **Thyroid hormones**

Neurotensin also participates in the neuroendocrine control of the thyroid hormones by regulating thyroid releasing hormone (TRH) function and thyroid stimulating hormone (TSH) secretion. These data indicate that NT is involved in the metabolic actions of these hormones. The role of NT in the hypothalamic-anterior pituitarythyroid axis and the functional cooperation between NT and thyroid hormones are excellently reviewed by Stolakis et al. (2010).

#### **Effect of NT on feeding**

In addition to its presence in the CNS NT is also found in neuroendocrine cells of the small bowel. There, NT participates in enteric digestive processes, gut motility, and intestinal inflammatory mechanisms. It also plays an important role in intestinal lipid metabolism, thus controlling appetite, weight status, and food intake (Kalafatakis and Triantafyllou, 2011). Additionally, NT exerts central control on blood glucose, feeding patterns, and body weight.

Centrally administered NT reduces appetite (Luttinger et al., 1982; Hawkins, 1986; Cooke et al., 2009), an effect that is mainly mediated by NTS1 (Remaury et al., 2002; Kim et al., 2008; Kim and Mizuno, 2010).

The anorectic effect of centrally administered NT is potentiated by peripheral injection of DA agonists, l-DOPA and bromocriptine, results suggesting that the effect of NT may be mediated by its ability to increase the activity of dopaminergic neurotransmission in the CNS (Hawkins et al., 1986).

Similar effects were shownfor the NT agonist,PD149163,in rats and ob/ob mice (Feifel et al., 2010a) as well as the non-selective NT agonist NT69L in Sprague-Dawley and in obese Zucker rats (Boules et al., 2000). The later study also showed that the effect of NT69L on food intake and body weight is due to functional interactions of NT with brain amines, and metabolic and endocrinological systems. NT69L transiently increases blood glucose and corticosterone levels and decreases TSH and T4 in Sprague-Dawley and in Zucker rats. NT69L also decreases norepinephrine in both the hypothalamus and NA, and increases DA, its metabolite 3,4-dihydroxyphenylacetic acid (DOPAC), and serotonin. These results indicate that feeding and energy expenditures are modulated by the interplay of hormones and neurotransmitters in the CNS (Brunetti et al., 2005).

Additionally, the effect of NT on feeding and body weight is also mediated, in part, by its action on the anorexigenic hormone leptin, which is secreted by adipocytes and regulates food intake by acting on hypothalamic neurons including NT-producing neurons (Beck et al., 1998; Sahu et al., 2001). Leptin together with insulin and α-melanocyte-stimulating hormones increase NT expression in the hypothalamus (Sahu, 1998; Cui et al., 2005). Similarly, the anorectic effect of leptin is at least partly mediated through central NTS1 and the leptin-NTS1 signaling pathway is involved in the regulation of food intake and the in the control of energy balance (Leinninger et al., 2011) since the lack of NTS1 reduces sensitivity to the anorectic action of leptin, causing hyperphagia and abnormal weight gain (Kim et al., 2008).

#### **EFFECT OF HORMONES ON NT SYNTHESIS AND RELEASE IN HYPOTHALAMIC-PITUITARY AXIS**

As NT modulates hypothalamic-pituitary axis (HPA) hormones, circulating hormones influence NT synthesis in the hypothalamus and anterior pituitary, results that suggest that NT mediates feedback effects of the hormones on neuroendocrine cells (Rostene and Alexander, 1997). Estrogen and progesterone regulate the activity of NT-synthesizing neurosecretory cells located in the hypothalamic arcuate nucleus (Alexander, 1999). Estrogen increases the secretion of NT at the median eminence and alters NT binding and NTS1-mRNA expression in the rostral preoptic nucleus (Moyse et al., 1988; Alexander, 1993; Watanobe and Takebe, 1993; Alexander and Leeman, 1994). GHRH neurosecretory cells synthesize NT under basal conditions (Niimi et al., 1991) and GH injection increases NT plasma levels in human (Schimpff et al., 1994).

Neurotensin's endocrine activity, and its modulatory effects on several neurotransmitter systems, kindled many studies suggesting that NT plays a role in the pathophysiology of several CNS disorders including PD, schizophrenia, and psychostimulant and nicotine addiction as well as pain and eating disorders (Caceda et al., 2006).

#### **NT AND NEUROPSYCHIATRIC DISORDERS SCHIZOPHRENIA**

Schizophrenia is a devastating psychotic disorder that affects approximately 1% of the population worldwide. The onset of illness usually occurs relatively early in life with most patients experiencing long-lasting adverse effects accompanied by severe impairment (reviewed in Carpenter and Buchanan (1994). The disease is manifested by positive symptoms (delusions, hallucinations, and an altered perception of reality), and negative symptoms (apathy, cognitive blunting, and social withdrawal), as well as a disorganization of thought and behavior. Many schizophrenic patients have difficulty holding a job or caring for themselves, placing a significant burden on their families and society. Additionally, increased mortality is associated with schizophrenia as patients experience a 20% shorter lifespan than the general population and are at an increased risk for committing suicide (Newman and Bland, 1991; Goff et al., 2001).

The exact cause(s) for the pathophysiology and progression of schizophrenia are not known; however it is generally accepted that associated symptoms are syndromal in nature, rather than manifestations of a single disease (Carpenter et al., 1999). These symptoms are likely caused by several compounding biochemical abnormalities and influenced by both genetics and environment. Theories as to the origin of these abnormalities focus on the dopaminergic system, but have more recently been expanded to include the serotonergic, γ-aminobutyric acid-ergic, and glutamatergic systems, as well as the NT system.

One of the earliest and most studied theories is the DA hypothesis of schizophrenia (Carlsson, 1988; Toda and Abi-Dargham, 2007; Howes and Kapur, 2009). While this theory originally emphasized general hyperdopaminergia as a causative factor in schizophrenia (Snyder, 1976), more recent versions focus on the balance of DA within discreet regions of the brain. The current theory predicts that hyperactive subcortical mesolimbic DA projections (resulting in hyperstimulation of D<sup>2</sup> receptors) are the main cause for positive symptoms, while hypoactive mesocortical DA projections to the PFC (resulting in hypostimulation of D<sup>1</sup> receptors) are the main cause for negative symptoms and cognitive impairment (Weinberger, 1987; Toda and Abi-Dargham, 2007). There is strong evidence for a close relationship between the dopaminergic system and NT (Nemeroff, 1986; Kitabgi et al., 1989), suggesting a role for NT in the pathophysiology of schizophrenia.

# **Neurotensin and schizophrenia**

A rolefor NT in schizophrenia has been hypothesizedfor over three decades (Nemeroff, 1980). Specifically, the NT system is closely associated with the dopaminergic system, and deficits in NT neurotransmission have been implicated in the pathophysiology of schizophrenia. Schizophrenics have a 40% decrease in NTRs in the entorhinal cortex (Wolf et al., 1995), and schizophrenics with decreased NT concentrations in their cerebral spinal fluid have significantly higher levels of pretreatment psychopathology (Sharma et al., 1997).

NT receptors have been detected on DA cell bodies in the SN and VTA (Palacios and Kuhar, 1981; Szigethy and Beaudet, 1989) and NTR activation has an excitatory effect on midbrain DA neurons (Pinnock, 1985; Seutin et al., 1989; Mercuri et al., 1993). NT can inhibit DA D<sup>2</sup> autoreceptor function, thus relieving auto-inhibition of DA transmission (Jomphe et al., 2006) thereby increasing DA release, firing rate, and the synthesis of the rate limiting enzyme for DA synthesis, TH (Binder et al., 2001b). Because of this close association of NT with dopaminergic neurons and its neuromodulatory effects on the dopaminergic system, NT is hypothesized to be therapeutic in the treatment of schizophrenia. Central administration of NT does in fact cause antipsychotic-like

effects (Jolicoeur et al., 1993), and therefore a great interest exists for developing NT-mimetics for the treatment of neuropsychiatric diseases, such as schizophrenia.

#### **The potential role of NT as an antipsychotic drug**

Several lines of evidence suggest that antipsychotics, the traditional treatment for schizophrenia, act through the induction of endogenous NT (Kinkead et al., 1999): (1) all clinically effective antipsychotic drugs affect in some way the NT system of rats. The therapeutic potency of typical APDs is derived mainly from their ability to antagonize DA D<sup>2</sup> receptors with high affinity (Creese et al., 1976). Typical antipsychotics are associated with extrapyramidal side effects (ESP) characterized by movement disorders including tardive dyskinesia, dystonic reactions, and Parkinsonism. Alternatively, it is suggested that atypical APDs have a higher affinity for 5-HT2A relative to DA D<sup>2</sup> receptors (Meltzer et al., 1989; Nordstrom et al., 1995; Kapur et al., 1998; Gefvert et al., 2001). They are characterized by enhanced antipsychotic efficacy and lower risk of EPS (Meltzer et al., 1989). Both typical and atypical antipsychotics, those associated with little or no EPS, increase NT mRNA in specific regions of the brain after both acute and chronic treatment (Govoni et al., 1980; Merchant et al., 1991, 1992; Merchant and Dorsa, 1993). (2) These effects are selective for drugs with antipsychotic efficacy, and are not seen with other classes of psychotropic drugs. (3) Typical and atypical antipsychotic drugs differentially affect the NT system. Typical antipsychotics act on both the mesolimbic and nigro-striatal NT systems. By contrast, the actions of atypical antipsychotics on NT are thought to be restricted to the mesolimbic system, the region thought to be the site of therapeutic effect in schizophrenia. (4) Centrally administered NT elicits similar behavioral effects as systemically administered atypical antipsychotic drugs. It is hypothesized that the increase in NT neurotransmission in the mesolimbic system mediates the therapeutic effects of APDs, while the increase in NT neurotransmission in the nigrostriatum underlies the propensity to cause ESP (Binder et al., 2001b; Caceda et al., 2006). Interestingly, the pattern of increase in NT mRNA expression, NT tissue content and release (Kinkead et al., 1999) is similar to that of the immediate early genes, c-fos, FosB, and ∆Fos B, after acute and chronic administration of typical and atypical APDs (Binder et al., 2001b).

The ability of a compound to inhibit climbing induced by apomorphine (a DA agonist), hyperactivity induced by amphetamine (indirect DA agonist), and to prevent the disruption of prepulse inhibition (PPI) induced by amphetamine [1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane] (DOI) (5-HT2A agonist), and dizocilpine (non-competitive NMDA antagonist) are tests used to predict antipsychotic-like activity of a drug. These tests also distinguish between typical and atypical antipsychotic-like effects (Merchant and Dorsa, 1993; Arnt et al., 1995) (see review by Geyer and Ellenbroek, 2003). The ability of a compound to block apomorphine-induced climbing over oro-facial stereotypies (sniffing and licking) induced by high doses of apomorphine, suggests atypicality of that compound (Gerhardt et al., 1985).

Neurotensin has been shown to prevent the disruption of amphetamine- and apomorphine-induced PPI (Feifel et al., 1997), as well as attenuate apomorphine-induced climbing (Jolicoeur et al., 1993), and amphetamine- and apomorphine-induced

hyperactivity without affecting stereotypy (Jolicoeur et al., 1983). For these reasons it is hypothesized that NT transmission is integral to the mechanism of action of antipsychotic drugs and that NT can serve as an endogenous antipsychotic-like compound (reviewed in Kinkead et al., 1999; Binder et al., 2001b; Kinkead and Nemeroff, 2002).

#### **The behavioral effects of neurotensin analogs in animal models of schizophrenia**

As discussed above, there is strong evidence to support NT as an endogenous antipsychotic drug. However, NT is easily degraded by peptidases and cannot cross the blood-brain barrier. In an effort to study more easily the effects of NT, our group has developed an extensive series of NT(8–13) analogs that can be delivered peripherally, and that elicit effects similar to centrally administered NT. These NT analogs have demonstrated antipsychotic-like activity similar to endogenous NT as determined by behavioral tests in rats. Most of our work has been with NT69L and NT79. NT69L binds human NTS1 and human NTS2 with high affinity. By contrast, NT79 preferentially binds to NTS2, as its binding affinity for the human NTS2 is approximately 25-fold higher than that for human NTS1. NT69L (Cusack et al., 2000), NT79 (Boules et al., 2010), and another analog NT77L (Boules et al., 2001c) are each modified from NT at amino acid positions located in the C-terminal 8–13 sequence. Additionally, PD149163, which is a reduced amide bond NT(8–13) mimetic shows antipsychotic-like activity.

PD149163 has strong affinity for NTRs (*K*<sup>i</sup> = 31.2 nM in newborn mouse brain membranes) and improved metabolic stability (Wustrow et al., 1995), a factor that promotes central activity after systemic administration (Banks et al., 1995).

**Table 1** shows the structures of some NT analogs.

NT69L elicits similar neurochemical and behavioral effects as endogenous NT (for review, see Boules et al., 2003a). Intraperitoneal delivery of NT69L selectively inhibits stereotyped apomorphine-induced climbing at an ED<sup>50</sup> of 16µg/kg, without affecting licking or sniffing at any dose given (Cusack et al., 2000). Additionally, NT69L does not cause catalepsy (muscle rigidity) at any dose given, but reverses haloperidol-induced catalepsy with an ED<sup>50</sup> of 0.2 mg/kg (Cusack et al., 2000). Catalepsy in rats has been used as a predictor of EPS potential of APDs in humans.

NT69L significantly increases both amphetamine- and dizocilpine-induced decreases in PPI with acute administration with an ED<sup>50</sup> of 0.08 and 0.05 mg/kg respectively (Shilling et al., 2003; Boules et al., 2010), as well as with chronic administration (Briody et al., 2010). Additionally, NT69L blocks PCP-induced hyperactivity (Li et al., 2010a), a test that may reflect both positive and negative symptoms of schizophrenia (Snyder, 1980; Javitt and Zukin, 1991). These results demonstrate that NT69L has properties similar to those of atypical APDs.

NT77 is also thought to have antipsychotic-like properties, although less potently than NT69L, based on similar rat behavioral studies (Boules et al., 2001c). NT77L selectively blocks apomorphine-induced climbing with an ED<sup>50</sup> of 5.6 mg/kg without affecting sniffing or licking behavior at any dose. NT77L moderately prevents/reverses the cataleptic effect of haloperidol (ED<sup>50</sup> 5.6 mg/kg) while it does not cause catalepsy itself at any dose. The effects of NT77L on PPI have yet to be tested.

Like NT69L and NT77L, NT79 also selectively blocks apomorphine-induced climbing without affecting sniffing or licking behavior at any dose. Similarly, NT79 significantly blocks amphetamine-induced hyperactivity as well as prevents the disruption of PPI induced by both amphetamine and DOI (Boules et al., 2010). PD149163 antagonizes the reduction in PPI of the rat startle reflex produced by amphetamine and dizocilpine (Feifel et al., 1999), blocks the disruption in PPI induced by the 5-HT2A agonist DOI (Feifel et al., 2003), and inhibits conditioned avoidance responding,which is a highly validated test for screening APD in rats,without causing catalepsy (Holly et al.,2011). These analogs show promise for use in the treatment of schizophrenic symptoms with a lower potential for adverse side effects. The antipsychotic effects of NT analogs are summarized in **Table 2**.

Studies with the use of knockout mice lacking NT or NTR subtypes suggest a role for NTS1 in the antipsychotic-like effect of NT. NT knockout mice have reduced PPI and are not sensitive to the PPI-disrupting effects of amphetamine as compared to wild type mice. These data indicate the importance of endogenous NT in the effects of amphetamine on PPI (Mechanic et al., 2009). Additionally, the use of NTS1- and NTS2-knockout (NTS1−/<sup>−</sup> and NTS2−/−) mice revealed hyper-dopaminergic state in the NTS1−/<sup>−</sup> mice similar to the excessive striatal DA activity reported in schizophrenia (Liang et al., 2010). Furthermore, the NTS1−/<sup>−</sup> show changes in behavior, prefrontal cortex neurotransmitters, and protein expression that are similar to wild type mice treated with the psychomimetic PCP, an animal model for schizophrenia (Li et al., 2010c). The involvement of NTS1 in the antipsychoticlike effect of NT has been further demonstrated by the use of NT agonists. Administration of the NT agonists, NT69L and NT-2, reversed apomorphine-induced climbing in wild type but not in NTS1−/<sup>−</sup> mice (Mechanic et al., 2009). Likewise, PD149163

#### **Table 1 | Structures of NT analogs.**

significantly facilitates PPI and decreases the acoustic startle response in wild type but not in NTS1−/<sup>−</sup> mice (Feifel et al., 2010b).

#### **PAIN**

#### **NT and pain**

Neurotensin and NTRs modulate nociception at several different levels (Dobner, 2005). In fact, on a molar basis, NT is more potent than is morphine as an antinociceptive agent (Nemeroff et al., 1979; al-Rodhan et al., 1991). Central administration of NT induces an analgesic response in both the hot plate (HP) and acetic acid-induced writhing tests in rats (Clineschmidt and McGuffin, 1977; Clineschmidt et al., 1979, 1982). NT also produces a longlasting antinociceptive effect in the tail-flick (TF) assay following intra rostroventral medulla (RVM) administration (Fang et al., 1987). These results indicate that NT can influence nociceptive transmission at several different points in the descending pain modulatory circuitry in the brain. There is also limited evidence that NT may affect nociception transmission directly in the spinal cord (Dobner, 2006). Intrathecal NT administration increased both HP and aversive (hypertonic saline) response latencies, but had no significant analgesic effect in the TF (Hylden and Wilcox, 1983). In addition, intrathecal injection of either NT or one of our novel NT(8–13) into the spinal cord of rats modifies pain perception in a rodent model of persistent (Roussy et al., 2006) and of neuropathic pain (Dansereau et al., 2006).

#### **NT receptor subtypes and pain**

The effect of NT on pain modulation is dose-dependent and receptor-selective (Urban and Smith, 1993). NT not only inhibits but also facilitates pain transmission in a dose-dependent manner. High doses (nanomolar range) of the peptide injected into the


#### **Table 2 | Antipsychotic effects of NT analogs.**


#### Y = has an effect.

Cusack et al. (2000), Mechanic et al. (2009); Cusack et al. (2000); Boules et al. (2001a), Shilling et al. (2003), Briody et al. (2010); Shilling et al. (2003), Secchi et al. (2009); Li et al. (2010a); Boules et al. (2001c); Boules et al. (2010); Feifel et al. (2008); Feifel et al. (1999), Shilling et al. (2004); Feifel et al. (1999), Feifel et al. (2003), Shilling et al. (2004).

RVM have an antinociceptive effect as measured by the TF latency (TFL) in response to heat stimulus. In contrast, lower doses (picomolar range) have been shown to reduce latencies in the HP and TF tests, facilitate spinal nociception response, and increase the visceromotor response to noxious heat and visceral stimulation respectively (Urban and Smith, 1993, 1994).

The basis of the opposing actions of NT and its dose-dependent modulation is most probably due to separate and distinct NTR subtype with varying affinity for the peptide (Smith et al., 1997), as well as the involvement of separate and distinct neuronal pathways that modulate pain at the spinal level (Smith et al., 1997). Evidence suggests that both NTS1 and NTS2 mediate the antinociceptive effects of NT, depending upon the antinociceptive test and possibly, the species of rodent used. NTS1 and NTS2 modulate pain-induced behavioral responses by acting on distinct spinal and/or supra-spinal neural circuits (Roussy et al., 2008). Contradictory results have been reported regarding the NTR subtype mediating the antinociceptive effects of NT.

*NTS1 and pain.* Reports on mice lacking the NTS1 gene reveal that NT and NT analogs fail to induce antinociception in the HP test (Pettibone et al., 2002). Consistent with the knockout mice studies, our group showed that the inhibition of NTS1 synthesis with the use of antisense peptide nucleic acids (PNAs) targeting the NTS1 gene also results in loss of the analgesic properties of NT in the HP test (Tyler et al., 1998b), results that implicate NTS1 in thermal analgesia. Conversely, other evidence that argues against the involvement of NTS1 in the analgesic effect of NT has been reported and summarized by Dubuc et al. (1999a). Thus, as summarized by this group, the analgesic effects of NT are not antagonized by the NTS1-selective antagonist SR48692 (Dubuc et al., 1994), the binding affinity of NT analogs to NTS1 does not correlate with their analgesic effects (Labbe-Jullie et al., 1994), and the administration of antisense oligonucleotides targeted to NTS1 does not reduce NT-induced analgesia in the writhing test in mice (Dubuc et al., 1999b). With respect to formalin-induced pain, Roussy et al. (2010) concluded that NTS1 is important for modulation of persistent pain following systemic administration of morphine in NTS1−/<sup>−</sup> mice.

*NTS2 and pain.* NTS2 has also been implicated in the analgesic effects of NT (Dubuc et al., 1999b; Remaury et al., 2002; Yamauchi et al., 2003; Maeno et al., 2004; Sarret et al., 2005; Bredeloux et al., 2006). Dubuc et al. showed, with the use of NTS2 oligonucleotides, that NTS2 plays an important role in the writhing test. These data were supported by a close correlation between NT analog binding affinity at NTS2 and potency of the analog-induced analgesia in mice (Dubuc et al., 1999b). Maeno et al. (2004) working with mice, suggested that NTS2 plays an important role in thermal nociception. The role of NTS2 in mediating NT-induced analgesia has been based on observations using NTS2 antisense oligodeoxynucleotides and the NTS2 ligand levocabastine (Bredeloux et al., 2006). NTS2 has also been implicated in mediating visceral antinociception (Dubuc et al., 1999a).

#### **NT analogs and pain**

As mentioned, NT must be administered directly into the brain to exert its effect since it gets degraded by peptidases (Tyler-McMahon et al., 2000a; Boules et al., 2005). Our laboratory has been testing our brain-penetrating stable analogs of NT(8–13) in several animal models of pain.

Studies by our group and others show that NT analogs are effective in treating thermal, visceral (acetic acid-induced writhing), and persistent inflammatory (formalin-induced) pain (Tyler-McMahon et al., 2000b; Bredeloux et al., 2008; Boules et al., 2009, 2010; Mechanic et al., 2009). However, as previously stated, evidence suggests that the analgesic efficacy of NT analogs varies with their selectivity for NTS1 and NTS2, the pain model, and, probably, animal species.

Intrathecal injection of the NTS1-selective agonist, PD149163, and the non-selective NT agonist NT69L significantly reduced pain-evoked responses during the inflammatory phase of the formalin test (Roussy et al., 2008). The same analogs also produced potent antiallodynic and antihyperalgesic effects in nerve injured rats, a model of neuropathic pain (Guillemette et al., 2012). Systemic injection of NT analogs NT69L, NT72 (NTS1-selective), and NT77 causes analgesia in the HP test (Tyler et al., 1998a, 1999; Boules et al., 2001c; Smith et al., 2011) with synergy to morphine (Boules et al., 2009). Additionally, NT69L and NT72, significantly reduce acetic acid-induced writhing (Smith et al., 2012).

Supporting the role of NTS2 in antinociception, intrathecal injection of NTS2 agonists levocabastine and JMV431 significantly inhibit the aversive behavior induced by formalin (Roussy et al., 2009) and induced dose-dependent antinociceptive response in the TF test (Sarret et al., 2005).

Interestingly, the NTS2-selective analog, NT79, is ineffective in reducing thermal pain, but blocks acetic acid-induced writhing in rats (Boules et al., 2010), without causing tolerance to its analgesic effects (Smith et al., 2012). It is also interesting that NT79 does not cause hypothermia. Since the NTS1-selective (NT72), and nonselective (NT69L) NT agonists attenuate visceral nociception, it is suggested that both NTR subtypes are involved in mediating visceral analgesia and their roles appear to be NT analog dependent (Smith et al., 2012). Additionally, the NTS2-selective analog, NT79, reduces formalin-induced pain and does so in synergy with morphine (Boules et al., 2011a). Further evidence for the involvement of NTS2 in reducing persistent pain comes with the use of knockout mice. Lafrance et al. (2010) demonstrated that mice lacking NTS2 exhibit significantly lower stress-induced analgesia following cold-water swim stress as compared to their wild type littermates.**Table 3** summarizes the analgesic effects of NT analogs.

#### **Analgesic synergy between NT and morphine**

Morphine is a µ-opioid receptor agonist that is widely used for the treatment of many types of chronic pain. However, morphine and other opioids are usually associated with some serious side effects. Additionally, tolerance develops to their analgesic effects, but not to all the side effects, and this tolerance requires dosage increases over time, to attain a consistent level of analgesia. Increasing morphine dosage increases the potential for serious undesired side effects, such as respiratory depression.

Neurotensin exerts a potent µ-opioid-independent, antinociceptive effects in a variety of analgesic screening tests, including tail-flick, HP, and writhing induced by acetic acid (Clineschmidt et al., 1979, 1982; al-Rodhan et al., 1991; Wustrow et al., 1995;



Y = has an effect; N = has no effect.

Tyler-McMahon et al. (2000b), Boules et al. (2009), Smith et al. (2011); <sup>2</sup>Smith et al. (2012); Roussy et al. (2008), Roussy et al. (2009); Guillemette et al. (2012); Boules et al. (2001c); Boules et al. (2010); Boules et al. (2011a); Boules et al. (2012); Roussy et al. (2008), Roussy et al. (2009).

Sarhan et al., 1997). Both NTS2 and µ-opioid receptors have been found in the same brain structures involved in pain perception (Basbaum and Fields, 1984; Asselin et al., 2001) and neurotensinergic system seems to play an important role in the non-opioid form of stress-induced analgesia (Seta et al., 2001; Gui et al., 2004; Lafrance et al., 2010).

The functional analgesic interaction between endogenous NT and the opioid system has been further illustrated by Tershner and Helmstetter. These authors show that the antinociception induced by the µ-opioid receptor activation in the amygdala is partly dependent on the recruitment of NTRs in the ventral PAG (Tershner and Helmstetter, 2000). Additionally, mice rendered tolerant to morphine showed a reduced analgesic effect to NT (Luttinger et al., 1983) and morphine analgesia was considerably reduced in NTS1-deficient mice. These data indicate that the NTS1 actively participates inµ-opioid analgesia (Roussy et al., 2010). Conversely, opioid receptor antagonists do not block NT-mediated antinociception (Clineschmidt et al., 1982; al-Rodhan et al., 1991), but the perfusion of morphine in the PAG increases the extracellular levels of NT-ir in a naloxone-dependent manner (Stiller et al., 1997).

The use of adjuncts such as the NMDA receptor antagonist ketamine in combination with opioids to enhance their efficacy at low doses, and attenuate the occurrence of tolerance has been reported (Lutfy et al., 1996; Nishiyama, 2000). However, serious motor impairment is observed at doses of ketamine that are antinociceptive in the rat and in human, and ketamine can be psychotomimetic. Additionally, long-term exposure of ketamine on spinal tissue leads to tissue necrosis (Vranken et al., 2005). Studies in our laboratory show analgesic synergism between NT analogs and morphine in the use of HP (Boules et al., 2009), in acetic acidinduced writhing, and in formalin-induced pain (Boules et al., 2011a). These studies provide novel potential use for NT analogs in combination with morphine (and perhaps other opioids) to improve the pharmacological treatment of pain while minimizing specific adverse effects of each of the drugs at a higher dose.

#### **NT AND PSYCHOSTIMULANT ABUSE**

Psychostimulants, including nicotine, amphetamine, and cocaine, are drugs that may be employed to improve cognitive and motor function, but can be highly addictive. The mesocorticolimbic system, to which NT colocalizes with DA and other neurotransmitters, provides the anatomic substrate for psychostimulant dependence and craving (McBride et al., 1999; Di Chiara, 2000).

NT acts as a neurotransmitter as well as a modulator of DA and other monoamine neurotransmitter systems and has been linked through several lines of evidence to psychostimulant effects (Richelson et al., 2003).

Behavioral sensitization to psychostimulants, an animal model of addiction, is a process by which repeated administration of the same dose of a drug produces increasing degrees of locomotor effects (Domino, 2001). This process is thought to be due to changes in the NA. Initiation of sensitization is associated with changes in the NA shell, and maintenance of sensitization with changes in the NA core (Iyaniwura et al., 2001; Balfour, 2004). Behavioral sensitization models human acquisition of psychostimulant addiction and risk of relapse (Miller et al., 2001; De Vries et al., 2002).

Locomotor sensitization depends on activation of the mesolimbic DA system with additional long-term influences on glutamate, GABA, κ-opioid, and other neurotransmitter systems (Pierce and Kalivas, 1997; Hahn et al., 2000). DA neurons originating in the VTA and projecting to the shell and core of the NA lead to the initiation and expression of sensitization, respectively (Pierce and Kalivas, 1997). NT colocalizes with DA neurons and acts as a modulator of DA effects (Binder et al., 2001a). When injected into the VTA, NT causes hyperactivity and DA release in the NA, similar to the effects of psychostimulants (Kalivas and Duffy, 1990; Kalivas, 1994). However, NT injected into the NA reduces the response to psychostimulants, an effect similar to that of brain-penetrating NT analogs given extracranially (Ervin et al., 1981; Robledo et al., 1993; Richelson et al., 2003).

The NT agonist NT69L, given intraperitoneally, blocks the acute locomotor effects of cocaine and d-amphetamine (Boules et al., 2001a), and blocks both the initiation and expression of sensitization to nicotine after subcutaneous administration (Fredrickson et al., 2003a,b). This is one line of evidence suggesting NT agonists may be effective for treatment of nicotine, amphetamine, and cocaine addiction. Curiously, the NT antagonist SR48692 when administered chronically decreased locomotor sensitization to cocaine (Felszeghy et al., 2007).

How then can a NTR agonist and a NTR antagonist have similar effects in this particular animal model of addiction? Although further research is needed, several lines of evidence bear on this question. Repeated doses of the NT antagonist were required to produce the intended effect on sensitization. Classically, chronic receptor blockade can lead to up-regulation of the receptor, thus making the agonist more effective. Chronic blockade of the receptors may also lead to increased NT synthesis and release. Another approach to this question is with animals lacking NTRs. Studies with SR48692 would predict that mice lacking NTS1 would not sensitize to psychostimulants. However, such null mice are more sensitive to an acute injection of d-amphetamine and had an enhanced response to the sensitizing effects of this psychostimulant as compared to wild type mice (Boules et al., 2006).

The NT agonist NT69L attenuates intravenous (IV) selfadministration of nicotine in rats (Boules et al., 2011b). Once animals acquired stable responding to nicotine they were pretreated with either NT69L or saline. Pretreatment with NT69L attenuated nicotine self-infusion under FR1 (fixed ratio of 1) and FR5 schedules of reinforcement. Control rats that were response-independent "yoked" as well as rats that self-infused saline showed minimal responses, indicating that nicotine served as a reinforcer. The stimulant and reinforcing effects of nicotine are attributed to stimulation of the mesolimbic DA system (Stein et al., 1998; Di Chiara, 2000; George et al., 2000). In this study, nicotine self-infusion increased TH, DA D1, and DA D<sup>2</sup> receptor mRNA in the VTA, consistent with increased burst firing. NT69L antagonized these effects. In the PFC, an area implicated with learning, NT69L increased TH and DA D<sup>1</sup> mRNA. TH and DA D<sup>1</sup> mRNA levels were also increased in the striatum in response to the NT agonist. Taken together these results show that the NT agonist NT69L attenuated nicotine self-administration and suggest the drug would reduce craving and withdrawal symptoms (for example, cognitive complaints that are common during nicotine withdrawal).

Is there abuse potential for NT agonists? Although NT mimics the effects of psychostimulants when injected directly into the VTA, it blocks certain stimulant effects when injected into the NA. NT analogs given extracranially (native NT is degraded by peptidases when given peripherally) also block locomotor effects of stimulants, and attenuate nicotine self-administration in rats. Rats do not self administer NT. Therefore, the preponderance of evidence argues against an abuse or addiction potential for NT analogs. This conclusion is supported by data in rhesus monkeys; the animals did not self-infuse NT69L (Fantegrossi et al., 2005).

Although questions remain, a growing body of data supports a key role for NT in regulation of responses to psychostimulants. To the extent animal models are predictive, and NT agonists show promise for the treatment of human addiction. The relative contributions of NTS1 and NTS2 receptors in these models have not been extensively explored, and further research may lead to development of agonists which retain efficacy with fewer side effects such as hypothermia and hypotension.

#### **NT AND PARKINSON'S DISEASE**

Studies show that plasma NT concentrations are consistently higher in PD patients as compared to controls and untreated as compared to treated patients (Schimpff et al., 2001). PD patients also have a twofold increase in NT content in both zona compacta and zona reticulate of the SN compared to that for controls (Fernandez et al., 1995). Brains from PD patients have fewer dopaminergic neurons and very low expression of NTS1 (Yamada and Richelson, 1995). Similar results were reported in an animal model for PD, where MPTP treated mice had significantly lower [ <sup>3</sup>H] NT and [3H] mazindol binding in the striatum and SN, results indicating severe reduction in NTRs and DA uptake sites respectively. These data suggest that the dysfunction in NTRs may be involved in the degradation processes causing PD. Caceda et al. (2006) suggested that the increase in striatal and nigral NT tissue concentrations, as well as in CSF and plasma levels may be due either to a compensatory mechanism for the loss of DA neurons to preserve motor function and/or to a dysregulation of NT neurotransmission on striatal output favoring the striatopallidal pathway (Caceda et al., 2006). Injection of NT resulted in a dose related attenuation of muscular rigidity and tremors caused by bilateral injection of the neurotoxin 6-hydroxydopamine in the medial forebrain bundle (Jolicoeur et al., 1991). In addition, we

showed that injection of the NT agonist, NT69L, attenuated the amphetamine and apomorphine rotatory behavior caused by the unilateral injection of the same drug in the nigro-striatal pathway (Boules et al., 2001b).

# **POTENTIAL SIDE EFFECTS OF NT AGONISTS HYPOTHERMIA**

Neurotensin (Bissette et al., 1976) and NT agonists (Tyler-McMahon et al., 2000b; Boules et al., 2001c, 2003a; Feifel et al., 2010a) elicit hypothermia in rodents. Hypothermia is mediated by NTS1. This has been established with the use of NTS1−/<sup>−</sup> and NTS2−/<sup>−</sup> mice (Remaury et al., 2002; Mechanic et al., 2009). Additionally, NTS2-selective analogs do not induce hypothermia (Boules et al., 2010), results that provide further proof for the involvement of NTS1 receptor subtype in mediating hypothermia. Interestingly, studies show potential therapeutic use for the NT agonist-induced hypothermia. The administration of the NT agonist ABS-201 immediately or up to 60 min after stroke attack significantly reduced infarct formation and brain cell death in an animal model of focal ischemia (Choi et al., 2012). In addition, it was effective in promoting long-term functional recovery in poststroke animals (Choi et al., 2012). Similar studies on regulated hypothermia induced by the NT agonists reduce oxidative stress in the brain during reperfusion from asphyxial cardiac arrest (Katz et al., 2004a). Also, lowering body temperature with a NT agonist provided a better neurologic outcome than brief external cooling in a rat model of near drowning (Katz et al., 2004b). Consistent with these results, hypothermia induced by the NT analog JMV-449 had a neuroprotective effect in a mouse model of permanent distal middle cerebral artery occlusion (Torup et al., 2003).

#### **HYPOTENSION**

Intracerebroventricular (ICV) or IV injections of NT induces a dose-dependent drop in arterial blood pressure in anesthetized rats. The drop in blood pressure is of rapid onset (30–60 s) and of short duration (1–4 min) (Rioux et al., 1981). Interestingly, the hypotensive effect of NT is not accompanied by any alteration in cardiac actions (Rosell et al., 1976; Rioux et al., 1981).

#### **DEVELOPMENT OF TOLERANCE**

Animal studies show that tolerance develops to some but not all the effects of NT analogs. Tolerance develops to the hypothermia, thermal analgesia, and anticataleptic effect of NTS1 agonists (Boules et al., 2003b). Conversely, tolerance does not occur to the NT agonist's effects on the reversal of amphetamine- and cocaineinduced locomotor activity (Boules et al., 2003b; Hadden et al., 2005; Feifel et al., 2008), the reversal of apomorphine-induced climbing (Boules et al., 2003b), and the reversal of amphetamine or DOI-induced disruption of PPI (Feifel et al., 2007; Briody et al., 2010).

Interestingly, NTS2 seem to play an important role in the development of tolerance of NT69L-mediated hypothermia and thermal analgesia (Smith et al., 2011), while tolerance to its analgesic and anticataleptic effects the NTS2 analog NT79 does not occur (Boules et al., 2010).

Thus, the development of tolerance depends on the NT analog, their selectivity to the NTR subtypes, the paradigm being tested and the dosing regimen (Wang et al., 2005; Feifel et al., 2008).

# **CONCLUSION**

The association of NT with several neurotransmitter and endocrine systems suggests its involvement in a wide range of physiologic processes throughout the body and implicates it in the pathology of many neuropsychiatric diseases. Furthermore, the development of NT analogs that can be administered systemically may provide therapy for the treatment of several disorders including schizophrenia, pain, and psychostimulant abuse. Additionally, the use of NT analogs and genetically modified animals will allow

# **REFERENCES**


the brain: relevance for neurodegenerative diseases and their treatment. *Prog. Neurobiol.* 83, 92–109.


further understanding of the molecular role of NT in health and disease.

# **ACKNOWLEDGMENTS**

This work was supported by Mayo Foundation for Medical Education and Research, the Siragusa Foundation Career Development Award in Neuroscience Research, and grant number 2KF01 from the Florida Department of Health to Mona Boules.


to the hypothermic and anticataleptic effects of a neurotensin analog that crosses the blood-brain barrier. *Brain Res.* 987, 39–48.


neuropathic pain management. *Biol. Psychiatry* 71, 302S.


of neurotensin increases tyrosine hydroxylase mRNA in substantia nigra. *Ann. N. Y. Acad. Sci.* 668, 311–313.


alpha-melanocyte-stimulating hormone directly induce neurotensin (NT) gene expression in novel NT-expressing cell models. *J. Neurosci.* 25, 9497–9506.


not exhibit tolerance. *Psychopharmacology (Berl.)* 200, 197–203.


interactions. Relevance for Parkinson's disease and schizophrenia. *Neurochem. Int.* 20(Suppl.), 215S– 224S.


the antipsychotic potential of orally bioavailable neurotensin (8-13) analogues containing non-natural arginine and lysine residues. *Neuropharmacology* 49, 1149–1159.


schizophrenia: version III – the final common pathway. *Schizophr. Bull.* 35, 549–562.


in the immune and neuroendocrine modulation of normal and abnormal enteric function. *Regul. Pept.* 170, 7–17.


K. (1999). Neuronal localization of a novel mosaic apolipoprotein E receptor, LR11, in rat and human brain. *Brain Res.* 833, 209–215.


plasma adrenocorticotropin concentration by stimulating corticotropin-releasing hormone secretion. *Life Sci.* 50, 639–643.


Pepeu, G. (1998). Neurotensin modulation of acetylcholine and GABA release from the rat hippocampus: an in vivo microdialysis study. *Neurochem. Int.* 33, 335–340.


N., Richelson, E., et al. (2008). Spinal NTS1 receptors regulate nociceptive signaling in a rat formalin tonic pain model. *J. Neurochem.* 105, 1100–1114.


et al. (1988). A subpopulation of dopaminergic neurons in rat ventral mesencephalon contains both neurotensin and cholecystokinin. *Brain Res.* 455, 88–98.


the antinociceptive effect of morphine. *J. Pharmacol. Exp. Ther.* 282, 899–908.


dopamine D(2)/neurotensin NTS1 and NMDA/neurotensin NTS1 receptor interaction in psychiatric and neurodegenerative disorders. *Curr. Med. Chem.* 19, 304–316.


ventral midbrain tegmentum. *Brain Res.* 479, 402–406.


expressing neurotensin receptor messenger RNA in the substantia nigra and the nucleus paranigralis of control and Parkinson's disease brain. *Neuroscience* 64, 405–417.


rat spinal cord – II. Co-existence of galanin with other peptides in local neurons. *Neuroscience* 64, 875–891.

**Conflict of Interest Statement:** Drs Richelson, Fredrickson, and Boules have a U.S. Patent # 7,087,575, August 8, 2006. "Treating the Effect of Nicotine: Elliott Richelson, Paul Fredrickson, Mona Boules."

*Received: 31 August 2012; accepted: 06 March 2013; published online: 22 March 2013.*

*Citation: Boules M, Li Z, Smith K, Fredrickson P and Richelson E* *(2013) Diverse roles of neurotensin agonists in the central nervous system. Front. Endocrinol. 4:36. doi: 10.3389/fendo.2013.00036*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Boules, Li, Smith, Fredrickson and Richelson. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Gastrin-releasing peptide receptors in the central nervous system: role in brain function and as a drug target

# *Rafael Roesler1,2,3\* and Gilberto Schwartsmann2,3,4*

<sup>1</sup> Laboratory of Neuropharmacology and Neural Tumor Biology, Department of Pharmacology, Institute for Basic Health Sciences, Federal University of Rio Grande do Sul, Porto Alegre, Brazil

<sup>2</sup> Cancer Research Laboratory, University Hospital Research Center (CPE-HCPA), Federal University of Rio Grande do Sul, Porto Alegre, Brazil

<sup>3</sup> National Institute for Translational Medicine, Porto Alegre, Brazil

<sup>4</sup> Department of Internal Medicine, School of Medicine, Federal University of Rio Grande do Sul, Porto Alegre, Brazil

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Hélène Volkoff, Memorial University of Newfoundland, Canada Mitsuhiro Kawata, Kyoto Prefectural University of Medicine, Japan

#### *\*Correspondence:*

Rafael Roesler, Laboratory of Neuropharmacology and Neural Tumor Biology, Department of Pharmacology, Institute for Basic Health Sciences, Federal University of Rio Grande do Sul, 90050-170 Porto Alegre, Rio Grande do Sul, Brazil. e-mail: rafael.roesler@pq.cnpq.br

Neuropeptides acting on specific cell membrane receptors of the G protein-coupled receptor (GPCR) superfamily regulate a range of important aspects of nervous and neuroendocrine function. Gastrin-releasing peptide (GRP) is a mammalian neuropeptide that binds to the GRP receptor (GRPR, BB2). Increasing evidence indicates that GRPR-mediated signaling in the central nervous system (CNS) plays an important role in regulating brain function, including aspects related to emotional responses, social interaction, memory, and feeding behavior. In addition, some alterations in GRP or GRPR expression or function have been described in patients with neurodegenerative, neurodevelopmental, and psychiatric disorders, as well as in brain tumors. Findings from preclinical models are consistent with the view that the GRPR might play a role in brain disorders, and raise the possibility that GRPR agonists might ameliorate cognitive and social deficits associated with neurological diseases, while antagonists may reduce anxiety and inhibit the growth of some types of brain cancer. Further preclinical and translational studies evaluating the potential therapeutic effects of GRPR ligands are warranted.

**Keywords: gastrin-releasing peptide, gastrin-releasing peptide receptor, bombesin receptors, neuropeptide signaling, brain disorders**

# **INTRODUCTION**

Neuropeptide signaling regulates a variety of aspects of nervous and neuroendocrine function (Hökfelt et al., 2003; Salio et al., 2006). Neuropeptides act by activating specific cell membrane receptors that are members of the G protein-coupled receptor (GPCR) superfamily, leading to stimulation of downstream protein kinase signaling pathways and ultimately altering gene expression (Oh et al., 2006).

Gastrin-releasing peptide (GRP), a neuropeptide originally isolated from the porcine stomach, is a 27-amino acid peptide synthesized as a 148-amino acid precursor (PreproGRP) and subsequently metabolized posttranslationally (Spindel et al., 1984, 1990; Lebacq-Verheyden et al., 1988). GRP is the mammalian homolog of the amphibian 14-amino acid peptide bombesin, isolated from the skin of the European frog *Bombina bombina* in 1970 (Erspamer et al., 1970). GRP and bombesin display similar biological activities and share the same seven C-terminal amino acid sequence. Early experiments examining the effects of bombesin when administered in the brain showed that intracerebroventricular (i.c.v.) infusions of bombesin induced hypothermia and hyperglycemia in rats (Brown et al., 1977a,b). In peripheral tissues, the physiological functions of GRP include regulating gastrin and somatostatin release, gastric acid secretion, pancreatic secretion, gastrointestinal motility, lung development, and chemoattraction in immune system cells (Ruff et al., 1985; Schubert et al., 1991; Del Rio and De la Fuente, 1994; Niebergall-Roth and Singer, 2001; Ohki-Hamazaki et al., 2005; Gonzalez et al., 2008; Jensen et al., 2008b; Czepielewski et al., 2012). Another member of the bombesin-like peptide (BLP) family found in mammals is neuromedin B (NMB), the mammalian equivalent of ranatensin, which acts on the NMB receptor (NMBR; Minamino et al., 1983). An additional peptide originally named neuromedin C (NMC) is in fact a decapeptide of GRP (GRP-10, GRP18−27; Minamino et al., 1984). Thus, BLPs in mammalian tissues have been increasingly shown to constitute a class of signaling peptides regulating a large range of physiological functions.

Gastrin-releasing peptide acts by binding to the GRP receptor (GRPR, also called BB2), a GPCR that binds preferentially to GRP and bombesin, with much lower affinity for NMB (Jensen and Gardner, 1981; Moody et al., 1988, 1992; von Schrenck et al., 1989, 1990; Ladenheim et al., 1990, 1992; Wang et al., 1992). Increasing evidence indicates that GRPR-mediated signal transduction in the central nervous system (CNS) plays an important role in regulating behavior, especially aspects related to emotional responses, social interaction, memory, and feeding. In addition, we have proposed that dysfunctions in GRPR expression and signaling might play a role in CNS disorders including anxiety, autism, memory dysfunction associated with neurodegenerative disorders, and brain tumors. Here we review the role of GRPRs in regulating brain function, and its potential as a drug target for CNS disorders.

# **MOLECULAR ORGANIZATION OF THE GRPR**

"fendo-03-00159" — 2012/12/13 — 20:10 — page 1 — #1

All mammalian bombesin receptors (GRPR, NMBR, and the orphan receptor BRS-3 or BB3) exhibit the characteristic seven

transmembrane domain structure of GPCRs. This review will focus solely on the GRPR. For a comprehensive review of the classification, nomenclature, structure, expression, signaling, and functions of the different types of bombesin receptors, see Jensen et al. (2008b).

The GRPR, clonedfrom murine Swiss 3T3 cells in 1990 (Spindel et al., 1990; Battey et al., 1991), is a 384-amino acid protein in humans, mice, and rats. The chromosomal location for the GRPR gene (named *GRPR* in humans and *Grpr* in mice and rats) is at chromosome Xp22.2-p22.13 (human), X F4 (mouse), and Xq21 (rat; Jensen et al., 2008a; **Table 1**).

#### **GRPR SIGNALING**

Experiments using different types of normal and tumor cells from humans and rodents have provided consistent evidence that the GRPR is directly coupled to the Gq type of G protein, and GRPR activation leads to an increase in cellular [Ca2+] and stimulation of the phospholipase C (PLC)/protein kinase C (PKC) and extracellular signal-regulated protein kinase (ERK)/mitogen-activated protein kinase (MAPK) pathways (Hellmich et al., 1999; Chen and Kroog, 2004; Stangelberger et al., 2005). GRPR signaling also interacts with a range of other enzymes (e.g., phospholipases A2 and D, tyrosine kinases, phosphatidylinositol 3-kinase – PI3K, and ciclooxigenase-2), growth factor receptor systems (including epidermal growth factor receptor, EGFR, and TrkB), and

**Table 1 | Molecular structure of the gastrin-releasing peptide receptor (GRPR).**


Aminoacid sequence (Homo sapiens)


Structural data are from Spindel et al. (1990), Battey et al. (1991), Wada et al. (1991), and Jensen et al. (2008a). Modified from Roesler et al. (2012), with permission).

immediate-early genes (c-fos and c-jun; Szepeshazi et al., 1997; Chatzistamou et al., 2000; Thomas et al., 2005; Hohla et al., 2007; Ishola et al., 2007; Liu et al., 2007; Flores et al., 2008; de Farias et al., 2010; Czepielewski et al., 2012; Petronilho et al., 2012). Data on signaling mechanisms mediating GRPR actions specifically in the CNS will be discussed below.

# **GRPR EXPRESSION IN THE CNS**

Early studies investigating the presence of bombesin receptors binding sites in the mammalian CNS showed that bombesin could bind with high affinity to rat brain membranes (Moody et al., 1978). Subsequently, autoradiographic studies indicated that areas containing high densities of GRPRs include the olfactory bulb, nucleus accumbens, caudate putamen, central amygdala, dorsal hippocampus, as well as the paraventricular, central medial, and paracentral thalamic nuclei (Wolf et al., 1983; Wolf and Moody, 1985; Zarbin et al., 1985). A detailed immunohistochemical characterization of GRPR expression in the mouse brain showed high GRPR immunoreactivity in the basolateral and central nuclei of the amygdala (BLA and CeA, respectively), hippocampus, hypothalamus, brain stem, nucleus tractus solitarius (NTS), and several cortical areas. Importantly, GRPR expression was restricted to neuronal cell bodies and dendrites, and was not present in axons or glial cells (Kamichi et al., 2005). Thus, the pattern of GRPR location in the brain suggests that it is specifically involved in regulating synaptic transmission. In some rat brain areas, GRPR expression shows marked changes during development – specifically between postnatal (PN) days 1 and 16 – with its expression increasing in the dentate gyrus and decreasing in the caudate putamen and lateral cerebellar nucleus (Wada et al., 1992).

Regarding receptor ligands, the use of radioimmunoassay techniques allowed demonstrating the presence of endogenous BLPs in the rat brain, with high concentrations in brain areas including the NTS, amygdala, and hypothalamus (Moody and Pert,1979; Moody et al., 1981). GRP mRNA has the highest density in forebrain areas and hypothalamus (Wada et al., 1990; Battey and Wada, 1991; for reviews, see Moody and Merali, 2004; Roesler et al., 2006a; Jensen et al., 2008b).

In the rodent spinal cord, GRPR expression is restricted to lamina I of the dorsal spinal cord, and GRP is expressed in a subset of dorsal root ganglion neurons including lumbar spinothalamic neurons (Sun and Chen, 2007; Fleming et al., 2012; Kozyrev et al., 2012). Importantly, the GRP system in the spinal cord is sexually dimorphic. In male rats, neurons in the L3 and L4 levels of the lumbar spinal cord project to the lower lumbar spinal cord (L5–L6 level) and release GRP onto somatic and autonomic centers containing GRPRs, whereas this system is vestigial in females (Sakamoto et al., 2008; Sakamoto, 2011). This has important implications for the control of male sexual reflexes by GRPR signaling (see below).

#### **GRPR REGULATION OF CNS FUNCTION**

"fendo-03-00159" — 2012/12/13 — 20:10 — page 2 — #2

Evidence that GRPRs in the brain and spinal cord regulate several physiological functions has come mostly from *in vivo* studies using pharmacological or genetic manipulation of the GRPR in rats or mice. Below, we summarize relevant findings of selected studies focusing on GRPR regulation of memory, stress and anxiety responses, feeding, itching, and sexual behavior.

# **SYNAPTIC PLASTICITY AND MEMORY**

In the late 1980s, Flood and Morley (1988) demonstrated that systemic or i.c.v. injections of GRP or bombesin after learning modulated memory retention for a T-maze footshock avoidance task in mice. When i.c.v. infusions were used, both peptides facilitated memory consolidation, whereas systemic injections produced memory enhancement or impairment depending on the drug dose and training conditions. Consistently with these findings, bombesin given after training through systemic injections (Rashidy-Pour and Razvani, 1998) or infusions directly into the NTS (Williams and McGaugh, 1994) enhanced memory retention in rats.

Memory modulation by GRPRs seems to be particularly important for memories involving emotional arousal and fear. Thus, pretraining injections of the GRPR antagonist [Leu13-(psi-CH(2)NH)-Leu14]BN impaired memory for inhibitory avoidance conditioning in mice (Santo-Yamada et al., 2003), and injection of another selective GRPR antagonist, RC-3095, in rats impaired memory for inhibitory avoidance but not for a task with less emotional content, novel object recognition (Roesler et al., 2004b). Similar impairing effects of RC-3095 on inhibitory avoidance memory were obtained with systemic posttraining injections (Roesler et al., 2004c), pre- or posttraining intrahippocampal microinfusions (Roesler et al., 2003; Venturella et al., 2005; Dantas et al., 2006; Preissler et al., 2007), or posttraining infusions into the BLA (Roesler et al., 2004c). The effects of the GRPR antagonist followed a typical inverted U-shaped dose–response pattern, in which intermediate doses resulted in memory impairment, whereas higher doses had no effect or produced memory enhancement (Roesler et al., 2003, 2004b; Dantas et al., 2006). Conversely, intrahippocampal infusion of bombesin resulted in enhancement of inhibitory avoidance memory at intermediate doses and impairment at higher doses (Roesler et al., 2006b). In addition to influencing memory formation, pharmacological manipulation GRPRs in specific brain areas has been shown to regulate fear memory expression, extinction, and reconsolidation-like processes (Luft et al., 2006, 2008; Mountney et al., 2006, 2008; Merali et al., 2011). For example, infusion of the GRPR antagonist RC-3095 into the rat dorsal hippocampus after memory reactivation blocks the extinction and reconsolidation of fear memory (Luft et al., 2006, 2008; for a review, see Roesler et al., 2012).

The role of GRPRs in regulating fear memory and synaptic plasticity has also been revealed by genetic studies using GRPR knockout mice. Contextual and cued fear conditioning were enhanced by the genetic deletion of GRPR, whereas spatial in the Morris water maze was unaffected. The enhancement of fear memory in GRPR knockout mice was accompanied by enhanced synaptic plasticity measured by long-term potentiation (LTP) in the amygdala. In wild-type mice, GRPR was preferentially expressed in amygdalar inhibitory interneurons releasing gamma-aminobutyric acid (GABA), and GRP might be released as a co-transmitter from glutamatergic neurons to activate preferentially GRPRs located on GABAergic interneurons to stimulate inhibitory transmission within the amygdala and function as an inhibitory constraint for the formation of fear-motivated memories (Shumyatsky et al., 2002).

Additional studies recently found enhanced retention and impaired extinction of cued fear conditioning, associated with an increase in c-fos activity in the BLA and reduced c-fos in the prefrontal cortex, in GRPR knockout mice. However, these mice showed unaltered contextual fear conditioning, multiple-trial cued fear conditioning, and conditioned taste aversion (Chaperon et al., 2012; Martel et al., 2012). Together, these findings indicate that the GRPR acts as a negative regulator of synaptic plasticity in the BLA and specific types of fear conditioning. However, the use of first generation knockout mouse models might confound the interpretation of the results, given that they do not allow the investigation of separate phases of memory (encoding, consolidation, and expression), and knockout mice might have up-regulation of compensatory pathways and non-specific alterations in CNS development in response to gene ablation (reviewed in Roesler et al., 2012).

We have shown that a number of signal transduction mechanisms downstream of receptor activation are involved in mediating memory regulation by the GRPR. In the CA1 area of the dorsal hippocampus, memory enhancement induced by bombesin was prevented by inhibitors of PKC, MAPK, PKA, and PI3K (Roesler et al.,2006b,2009,2012), and potentiated by coinfusion of stimulators of the dopamine D1/D5 receptor (D1R)/cAMP/PKA pathway, namely the D1R agonist SKF 38393, the adenylyl cyclase activator forskolin, and the cAMP analog 8-Br-cAMP (Roesler et al., 2006b). These findings indicate that the PKC, MAPK, PI3K, and PKA pathways are critical in mediating memory modulation by hippocampal GRPRs, and that GRPR activation can interact with cAMP/PKA signaling in enhancing hippocampal memory formation (**Figure 1**). GRPRs in the rat brain also show functional interactions with other growth factor systems including basic fibroblast growth factor (bFGF/FGF-2), nerve growth factor (NGF), and brain-derived neurotrophic factor (BDNF; Kauer-Sant'Anna et al., 2007; Preissler et al., 2007).

#### **EMOTIONAL BEHAVIOR**

"fendo-03-00159" — 2012/12/13 — 20:10 — page 3 — #3

Gastrin-releasing peptide and GRPR are highly expressed in brain regions, such as the amygdala, activated by stressful stimuli, and, as discussed above, GRPR signaling is likely to be a major regulator of memory associated with fear and emotional arousal. Merali and colleagues have shown that chronic stressor exposure leads to an elevation of GRP levels in the anterior pituitary in rats, and GRP release in the rat amygdala is increased in response to exposure to a shock. GRP may stimulate the release of adrenocorticotropic hormone (ACTH), playing a role in mediating the corticotropinreleasing hormone (CRH) stress response, and increasing the activity of the hypothalamic–pituitary–adrenal (HPA) axis. In addition, bombesin administration induces endocrine, autonomic, and behavioral effects associated with stress, and bombesin receptor antagonists attenuate the behavioral and neurochemical effects of stressors (Merali et al., 2002, 2009; Moody and Merali, 2004; Mountney et al., 2011). Moreover, we have shown that systemic administration of a GRPR antagonist can induce an anxiogenic-like effect in the elevated plus maze test in rats

(Martins et al., 2005). Together, these data suggest that brain GRPRs might regulate emotional behavioral and responses to stress.

#### **FEEDING BEHAVIOR**

It has been known for over 30 years that systemic or i.c.v. administration of bombesin or GRP in rats reduces the intake of liquid and solid food in rats (Gibbs et al., 1981). Similar effects on meal size are observed after systemic bombesin injections in mice and intravenous (i.v.) injections in baboons and humans (Gibbs, 1985). In addition, brief vena caval infusions of GRP and NMB in rats, given alone or together at the onset of the first nocturnal meal, significantly reduced meal size and duration (Rushing et al., 1996), and bombesin or GRP given systemically extended the duration of the intermeal interval (Thaw et al., 1998). The suppression of glucose intake induced by systemic administration of GRP or bombesin was blocked by infusion of a GRPR antagonist into the fourth ventricle in rats (Ladenheim et al., 1996), and was absent in GRPR knockout mice (Hampton et al., 1998; Ladenheim et al., 2002), indicating that central GRPRs are critical in mediating the effects of peripheral bombesin and GRP on feeding. In addition, GRPR knockout mice ate significantly more at each meal than wild-type controls (although total 24 h food consumption was equivalent), and showed elevated body weight compared with wild-type littermates beginning at 45 weeks of age (Ladenheim et al., 2002). The finding that systemic GRP potently reduced independent intake of both sucrose and milk from a bottle but did not affect intraoral intake of either solution indicated that the GRPR regulates the appetitive-related aspects of the feeding process (Rushing and Houpt, 1999). The amygdala is likely a key brain area involved in mediating the regulatory action of GRPRs on feeding: bilateral infusion of GRP into the central amygdala produced a transient inhibition of food intake, an effect that was prevented by the GRPR antagonist [Leu(13)-psi(CH(2)NH)-Leu(14)]BN (Fekete et al., 2002).

These findings provide strong support for a role of GRP/GRPR signaling in regulating feeding. It has been proposed that BLPs may also be released from the gastrointestinal tract in response to food ingestion, acting to bridge the gut and brain to inhibit further food intake. Conversely, the suppression of release of BLPs in the brain may trigger the initiation of a feeding episode (reviewed in Merali et al., 1999).

#### **SEXUAL BEHAVIOR**

"fendo-03-00159" — 2012/12/13 — 20:10 — page 4 — #4

One of the most exciting recent developments in GRPR research was the identification by Sakamoto et al. (2008) of a sexually dimorphic GRPR system in the spinal cord that is crucial in regulating male sexual function. In male rats, but not in females or males with a dysfunctional androgen receptor gene, GRP-containing neurons in the upper lumbar spinal cord innervate lower lumbar regions controlling erection and ejaculation. Pharmacological stimulation of spinal GRP receptors restores penile reflexes and ejaculation after castration, whereas intrathecal administration of the GRPR antagonist RC-3095 inhibits penile reflexes and ejaculations. The inhibitory effect of castration on GRP expression in this spinal center suggests that androgen signaling plays a major role in regulating GRP expression in the male spinal cord (Sakamoto et al., 2009b). Moreover, exposure to traumatic stress decreases the local GRP content and reduces penile reflexes in male rats (Sakamoto et al., 2009a; Sakamoto, 2010). Thus, GRP/GRPR signaling has emerged as a new target for the understanding of psychogenic erectile dysfunction and the development of potential therapeutic approaches to masculine reproductive dysfunction (Sakamoto et al., 2008, 2009a,b; Sakamoto and Kawata, 2009; Sakamoto, 2010, 2011).

#### **ITCHING**

Another function in which GRPRs in the spinal cord have been shown to play a major role is itching. GRPR knockout mice show normal thermal, mechanical, inflammatory, and pain responses, but reduced responses to pruritogenic stimuli, and GRP-induced pruritus in wild-type mice is blocked by intrathecal administration of a GRPR antagonist (Sun and Chen, 2007). The selective ablation of GRPR-expressing lamina I neurons in the mouse spinal cord of mice results in scratching deficits in response to itching stimuli, but does not affect pain behaviors (Sun et al., 2009). These findings allowed the identification of GRPR as a central molecular mediator of the itch sensation in the spinal cord (Sun and Chen, 2007; Sun et al., 2009).

A recent seminal study showed that the μ-opioid receptor (MOR) isoform MOR1D heterodimerizes with GRPR in the spinal cord to relay itch information. Blocking the association between MOR1D and GRPR attenuates morphine-induced scratching. Morphine triggers internalization of both GRPR and MOR1D, whereas GRP specifically triggers both GRPR internalization and morphine-independent scratching. These data suggest that opioid-induced itch is independent of opioid analgesia and occurs via cross-activation of GRPR signaling by MOR1D heterodimerization (Liu et al., 2011).

# **POSSIBLE ROLE OF ALTERATIONS IN GRPR EXPRESSION AND SIGNALING IN THE PATHOGENESIS OF BRAIN DISORDERS**

Since GRPRs are highly expressed in neurons in brain areas including the hippocampus and BLA, and regulate crucial aspects of behavior that can be altered in patients with CNS disorders, it is possible that deregulated GRPR signaling contribute to the pathogenesis of neurological and psychiatric diseases. Although a causative role of GRPR dysfunction in CNS disorders has not been directly established, some alterations in the levels of BLPs peptides or GRPR density or function have been observed in patients with psychiatric, neurodegenerative, and neurodevelopmental disorders. In addition, the use of preclinical models has provided further evidence indicating a role for the GRPR in some CNS pathologies. Based on these findings, we have put forward that the GRPR may be a novel molecular target for the development of therapeutic strategies for patients with neurological and psychiatric disorders (Roesler et al., 2004a, 2006a). **Table 2** summarizes the findings from studies examining possible alterations in GRP and GRPR content or signaling found in patients with brain disorders.

### **NEURODEGENERATIVE DISORDERS**

The concentration of BLPs was found to be significantly reduced in the caudate nucleus and globus pallidus of patients with Parkinson's disease (PD; Bissette et al., 1985). However, Stoddard et al. (1991)found no alterations in bombesin-like immunoreactivity in the adrenal medullary tissue of patients with PD, although the concentration of several other neuropeptides was reduced. A reduction in bombesin receptor density and altered bombesin-induced calcium signaling have been reported in fibroblasts from patients with Alzheimer's disease (AD; Ito et al., 1994; Gibson et al., 1997).

**Table 2 | Findings from selected studies examining possible alterations in the GRPR system in patients with CNS disorders. Modified from Roesler et al. (2006a), with permission.**


"fendo-03-00159" — 2012/12/13 — 20:10 — page 5 — #5

**induced by beta-amyloid peptide in the rat hippocampus.** Data are mean ± SEM retention test step-down latencies (s), in an inhibitory avoidance conditioning, of rats given a bilateral infusion of the GRPR agonist bombesin (BB; 0.002 μg) or saline (SAL; control group) 10 min before being trained in IA, and beta-amyloid peptide (Abeta; 25–35) or distilled water (DW; controls) immediately after IA training. The number of animals was 8–14 per group. \*\*P < 0.01 compared to the control group treated with SAL and DW. Reproduced from Roesler et al. (2006b), with permission.

For example, in fibroblasts from patients with familial AD presenting the Swedish APP670/671 mutation, elevations in calcium induced by bombesin were reduced by 40% (Gibson et al., 1997).

Using the memory impairment produced by a microinfusion of a low dose of beta-amyloid peptide (25–35; Abeta) into the rat CA1 area of the dorsal hippocampus as a model of memory dysfunction associated with AD, we showed that an intrahippocampal infusion of bombesin completely prevented the Abeta-induced impairment in inhibitory avoidance memory (Roesler et al., 2006b; **Figure 2**). This finding provided preliminary preclinical evidence suggesting that pharmacological stimulation of the GRPR might rescue memory deficits associated with AD.

#### **NEURODEVELOPMENTAL DISORDERS**

The first evidence suggesting that the GRPR might be a candidate gene in autism spectrum disorders (ASD) was the finding of a translocation breakpoint on the X chromosome in the first intron of the *GRPR* gene in a patient with autism accompanied by mental retardation and epilepsy (Ishikawa-Brush et al., 1997). Although a subsequent study investigating two polymorphic sites in the second exon of the *GRPR* gene in patients did not support the *GRPR* as a candidate locus for autism (Marui et al., 2004), more recently a possible role of C6S and L181F mutations of the *GRPR* gene in GRPR function and ASD was found in two patients (Seidita et al., 2008).

In order to examine the role of GRPR in CNS development and its possible involvement in ASD, we submitted rat pups to pharmacological GRPR blockade by systemic administration of RC-3095 from PN days 1–10, and examined long-lasting behavioral and molecular alterations produced by this treatment. Rats given neonatal RC-3095 showed pronounced deficits in social interaction (a hallmark of rodent models of ASD) when tested at PN days 30–35 (Presti-Torres et al., 2007; **Figure 3**) or PN day 60 (Presti-Torres et al., 2012). In addition, RC-3095-treated rats showed impaired 24-h retention of memory for inhibitory avoidance and novel object recognition, whereas body weight during development, open field behavior, and short-term memory were not affected (Presti-Torres et al., 2007, 2012). Neonatal GRPR blockade also reduced maternal odor preference, a behavioral measure of attachment behavior (Garcia et al., 2010). The impairment in social behavior induced by GRPR blockade was rescued by treatment with the atypical antipsychotic clozapine (Presti-Torres et al., 2012). Together, these findings suggest that GRPR blockade during CNS development can lead to specific behavioral alterations that are consistent with ASD, and support the possibility that abnormal GRPR expression or function during development might play a role in disease pathogenesis. Also, we have proposed that neonatal GRPR blockade in rats may serve as a novel animal model of ASD (Presti-Torres et al., 2007, 2012).

# **OTHER NEUROPSYCHIATRIC DISORDERS**

The findings from rodent studies discussed above, indicating that normal GRPR function during development might be important for behaviors related to social interaction, attachment, and cognition, and that clozapine rescues social behavior deficits produced by GRPR blockade, are also consistent with the possibility that GRPR signaling is altered in schizophrenia. In addition, we found that GRPR blockade by systemic injections of RC-3095 prevent apomorphine-induced stereotypy in mice and amphetamine-induced hyperlocomotion in rats, which are models of schizophrenic psychosis and mania (Meller et al., 2004; Kauer-Sant'Anna et al., 2007). In patients with schizophrenia, a reduction in the levels of radioimmunoassay-detectable bombesin in the cerebrospinal fluid (CSF; Gerner et al., 1985), and reduced urinary levels of BLPs (Olincy et al., 1999) have been found. Further studies using samples from patients and animal models are required to examine whether GRPR signaling is involved in schizophrenia.

As reviewed above, data from animal studies also consistently show that GRPRs in brain areas including the amygdala regulate memory related to fear and anxiety responses, raising the possibility that GRPR signaling plays a role in anxiety disorders (Moody and Merali, 2004; Roesler et al., 2012). For example, pharmacological manipulation of the GRPR in the hippocampus can affect extinction and reconsolidation of fear memory, which are preclinical models used in the investigation and screening of potential therapeutic strategies for post-traumatic stress disorder (PTSD) and other fear-related disorders (Luft et al., 2006, 2008). In postmortem analyses of brains from suicides compared to control subjects, Merali et al. (2006) reported discrete alterations in the levels of GRP and NMB. More recently, however, the possibility that GRP and GRPR are candidate genes in panic disorders was not confirmed in an association and linkage analysis (Hodges et al., 2009).

Anxiety disorders may show comorbidity with eating disorders, anorexia and bulimia nervosa. Given the important role of GRPR in regulating feeding behavior (see above), it is possible that it contributes to eating disorders. One study found significantly reduced GRP levels in the CSF of women who were recovered from bulimia nervosa, compared to women recovered from anorexia or healthy control women. The authors suggested that persistent alterations in GRP levels after recovery indicate that this alteration might be

"fendo-03-00159" — 2012/12/13 — 20:10 — page 6 — #6

"fendo-03-00159" — 2012/12/13 — 20:10 — page 7 — #7

permission.

1 to 10. A social behavior test was carried out at PN 30. **(A)** Representative

trait-related and contribute to episodic hyperphagia in patients with bulimia nervosa (Frank et al., 2001).

#### **BRAIN TUMORS**

Gastrin-releasing peptide receptor overexpression has been demonstrated in many types of cancer (Cornelio et al., 2007), and we have recently shown widespread expression and a high content of GRPR in human glioma, the most common and lethal type of neurological cancer (Flores et al., 2010; **Figure 4**). GRPR activation by GRP or bombesin stimulates the growth of glioma cell lines (Moody et al., 1989; Pinski et al., 1994; Sharif et al., 1997; de Farias et al., 2008; Flores et al., 2008). We have recently shown that the stimulatory effect of GRPR activation on proliferation of glioma cells depends on PI3K signaling (Flores et al., 2008) and is potentiated by co-activation of the cAMP/PKA pathway (de Farias et al., 2008; reviewed in Roesler et al., 2010).

Gastrin-releasing peptide receptor antagonists inhibit the growth of human U-87MG and U-373MG gliomas xenografted into nude mice (Pinski et al., 1994; Kiaris et al., 1999). In addition, GRPR antagonism by RC-3095, alone or combined with temozolomide, significantly reduced the growth of C6 gliomas both *in vitro* and *in vivo*, with the combined administration of TMZ and RC-3095 being the most effective treatment (**Figure 5**; de Oliveira et al., 2009). These findings strongly suggest that targeting GRPR may be a promising strategy for the development of novel therapies

**FIGURE 4 | GRPR content in human normal brain tissue and brain tumors.** Representative sections of **(A)** normal brain and **(B)** astrocytoma grade IV from an immunohistochemical study of GRPR content from samples of patients with gliomas and normal brain samples. GRPR staining is shown in the right column (brown, ×400) and hematoxylin–eosin (HE) in the left column (×400). GRPR staining in the normal brain tissue is restricted to neuronal bodies and dendrites, whereas its presence in astrocytoma samples is widespread. Sections were incubated with anti-GRPR antibody, sequentially treated with biotinylated anti-rabbit IgG and streptavidin-biotin peroxidase solution, and then developed with diaminobenzidine as chromogen. Modified from Flores et al. (2010), with permission.

against glioma. The GRPR might also regulate the growth of neuroblastoma (Kim et al., 2002; Qiao et al., 2008; Abujamra et al., 2009), although, in contrast, we could not find a role for GRPR in regulating the *in vitro* growth of medulloblastoma, the most common brain cancer of childhood (Schmidt et al., 2009).

# **GRPR LIGANDS AS CANDIDATE THERAPEUTIC DRUGS IN BRAIN DISORDERS**

The evidence reviewed above indicates that the GRPR might be considered a novel molecular target in different types of CNS disorders, and raise the possibility that GRPR agonists might ameliorate cognitive and social deficits associated with neurological diseases, while antagonists may, for example, reduce anxiety and inhibit the growth of some types of brain cancer. Studies examining the effects of GRP administration on satiety and eating behavior in humans (Gutzwiller et al., 1994), as well as a phase I trial of the GRPR antagonist RC-3095 in patients with solid tumors (Schwartsmann et al., 2006) have suggested that GRP and peptidergic GRPR antagonists can be safely administered intravenously in human subjects. Thus, the potential therapeutic effect of GRPR ligands in preclinical models as well as in patients with CNS disorders warrants further investigation.

#### **ACKNOWLEDGMENTS**

"fendo-03-00159" — 2012/12/13 — 20:10 — page 8 — #8

This work was supported by the National Council for Scientific and Technological Development (CNPq; grant number 303703/2009-1 to Rafael Roesler); the National Institute for Translational Medicine (INCT-TM); and the South American Office for Anticancer Drug Development.

# **REFERENCES**


in experimental anticancer therapy. *Ann. Oncol.* 18, 1457–1466.


"fendo-03-00159" — 2012/12/13 — 20:10 — page 9 — #9

suppression in gastrin-releasing peptide receptor-deficient mice. *Proc. Natl. Acad. Sci. U.S.A.* 95, 3188– 3192.


receptors for secretagogues on pancreatic acinar cells. *Fed. Proc.* 40, 2486–2496.


peptides in the mediation or integration of the stress response. *Cell. Mol. Life Sci.* 59, 272–287.


"fendo-03-00159" — 2012/12/13 — 20:10 — page 10 — #10

elevates cytosolic calcium and stimulates the growth of small cell lung cancer cell lines. *J. Pharmacol. Exp. Ther.* 263, 311–317.


the memory impairment induced by gastrin-releasing peptide receptor antagonism in area CA1 of the rat hippocampus. *Neurochem. Res.* 32, 1381–1386.


molecular target for psychiatric and neurological disorders. *CNS Neurol. Disord. Drug Targets* 5, 197–204.


reproductive functions. *Nat. Neurosci.* 11, 634–636.


"fendo-03-00159" — 2012/12/13 — 20:10 — page 11 — #11

M., Hampton, L., et al. (2002). Identification of a signaling network in lateral nucleus of amygdala important for inhibiting memory specifically related to learned fear. *Cell* 111, 905–918.


**Conflict of Interest Statement:** The authors declare that the research was

"fendo-03-00159" — 2012/12/13 — 20:10 — page 12 — #12

conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 17 October 2012; paper pending published: 04 November 2012; accepted: 23 November 2012; published online: 17 December 2012.*

*Citation: Roesler R and Schwartsmann G (2012) Gastrin-releasing peptide receptors in the central nervous system: role in brain function and as a drug target. Front. Endocrin. 3:159. doi: 10.3389/ fendo.2012.00159*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Roesler and Schwartsmann. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Somatostatinergic systems: an update on brain functions in normal and pathological aging

# *Guillaume Martel, Patrick Dutar, Jacques Epelbaum and Cécile Viollet\**

Inserm UMR894 - Center for Psychiatry and Neuroscience, Université Paris Descartes, Sorbonne Paris Cité, Paris, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Paola Bagnoli, University of Pisa, Italy Kyriaki Thermos, University of Crete, Greece

*\*Correspondence:*

Cécile Viollet, Inserm UMR894 - Center for Psychiatry and Neuroscience, Université Paris Descartes, Sorbonne Paris Cité, 2 ter rue d'Alésia, 75014 Paris, France. e-mail: cecile.viollet@inserm.fr

Somatostatin is highly expressed in mammalian brain and is involved in many brain functions such as motor activity, sleep, sensory, and cognitive processes. Five somatostatin receptors have been described: sst1, sst2 (A and B), sst3, sst4, and sst5, all belonging to the G-protein-coupled receptor family. During the recent years, numerous studies contributed to clarify the role of somatostatin systems, especially long-range somatostatinergic interneurons, in several functions they have been previously involved in. New advances have also been made on the alterations of somatostatinergic systems in several brain diseases and on the potential therapeutic target they represent in these pathologies.

**Keywords: interneurons, SRIF, GPCR, sst, cortistatin, long-range, Alzheimer's disease**

"fendo-03-00154" — 2012/12/4 — 23:44 — page 1 — #1

# **NEUROANATOMY OF SOMATOSTATINERGIC SYSTEM**

**SOMATOSTATIN PEPTIDES IN THE BRAIN** Somatostatin14 (SRIF, somatotropin release inhibiting factor) was serendipitously discovered in 1972 by Roger Guillemin and colleagues who were aiming to purify and characterize growth hormone (GH)-releasing hormone from sheep hypothalamus (Brazeau et al., 1973). Soon thereafter, an N-terminally extended peptide, SRIF28, was purified from the gut. Both peptides, arising from a common propeptide encoded from a single gene, were found in the mammalian nervous system where SRIF14 is the

predominant form (for review, see Epelbaum, 1986). Two brain SRIF-related bioactive peptides have been discovered later. Cortistatin (CST) has been cloned in 1996 (De Lecea et al., 1996) and shares 11 amino acids with SRIF (**Figure 1**). CST peptides are predicted to occur as 14-AA or 17-AA short forms in rodents and humans, respectively, and a 29-AA extended form in both species. CST is mainly restricted to the cerebral cortex and the hippocampus in the central nervous system (CNS). CST has been implicated in several brain functions such as learning and memory, regulation of sleep/wakefulness rhythms and it is suspected to have an anticonvulsant activity (for review, see de Lecea, 2008). Recently, bioinformatics analyses of evolutionary conserved sequences identified neuronostatin, a 13-AA amidated peptide also encoded by the somatostatin gene. Mostly found in pancreas, spleen, and brain, it is involved in metabolic, cardiovascular, and neuronal functions (Samson et al., 2008).

Somatostatin induces many transduction mechanisms in transfected systems (for review, see Lahlou et al., 2004; Olias et al., 2004), but deciphering the physiological actions of the native receptors *in situ* remains an intense field of study. The last decade showed increasing progress in understanding the role of SRIF in brain functions using molecular, pharmacological, and behavioral approaches. The development of innovative molecular, genetic, and imaging tools now allows to go a step further and to assess the cellular contribution of SRIF-expressing cells in neuronal networks *ex vivo* and soon *in vivo*. In this review we will give an overview of the latest findings concerning SRIF systems in brain, report some recent data concerning their synaptic actions and their physiological roles within the brain, in normal or pathological conditions.

#### **SOMATOSTATINERGIC NETWORKS IN THE BRAIN**

Somatostatin is ubiquitously expressed in mammalian brain, including humans (**Figure 2A**). SRIF-immunoreactivity is found at high level in the mediobasal hypothalamus and median eminence, amygdala, preoptic area, hippocampus, striatum, cerebral cortex, sensory regions, and the brainstem (for review, see Epelbaum, 1986; Viollet et al., 2008).

Somatostatin peptide colocalizes with gamma-aminobutyric acid (GABA), a major inhibitory neurotransmitter and labels mostly non-glutamatergic cells in the brain. In order to target SRIF interneurons *in situ*, most recent studies took advantage of rodent models expressing the green fluorescent protein (GFP) under the control of the GAD67 promoter to visualize GABAergic populations more easily. In the GIN (GFP-expressing inhibitory neurons) strain (Oliva et al., 2000) nearly all GFP cells stained for SRIF while in the GAD67-GFP strain, SRIF immunohistochemistry labels 37% of total GFP cells (Ma et al., 2006). The recent development of specific Cre recombinase and knock-in inducible driver lines for SRIF (Taniguchi et al., 2011) opens promising avenues to study SRIF functions at the cellular level combined to optogenetic and imaging tools.

Previous immunohistochemical and tracing studies have identified two main categories of SRIF neurons: those acting locally in a given structure within microcircuits (interneurons) and those projecting to a distant structure (long-projecting neurons). Nevertheless, recent data using GFP transgenic mice revisit previous anatomical records by demonstrating that some formerly called interneurons also belonged to the projecting neurons category. The different kinds of GABAergic interneurons are classified

"fendo-03-00154" — 2012/12/4 — 23:44 — page 2 — #2

according to their molecular, physiological, and morphological properties. Immunohistochemical characterization of neuronal populations in rat cortex initially stated, based on calcium-binding proteins and peptides expressions that parvalbumin (PV), SRIF, calretinin, and cholecystokinin labeledfour main non-overlapping chemical classes of interneurons (Xu et al., 2006). However, several studies later reported a significant colocalization of SRIF and calretinin in mouse brain (Xu et al., 2006; Kosaka and Kosaka, 2007; Lepousez et al., 2010a), pointing out species-dependent variations in the repertoire of calcium-binding proteins and neuropeptides. It seems that neuronal populations immunoreactive for calbindin or the neuropeptide Y strongly overlap with the somatostatinergic population in rats whereas calretinin is preferentially coexpressed with SRIF in the mouse.

Recent morphological and electrophysiological studies using GIN mice focused on SRIF-expressing populations in cortical circuits. In mouse cortex, parvalbumin- and SRIF-expressing neurons respectively constitute 40 and 30% of the total GABAergic neurons, calretinin being expressed in 50% of the somatostatinergic population (Rudy et al., 2011). The remaining cortical inhibitory interneurons, expressing ionotropic serotonergic receptor 5HT3a, include VIP- and NPY-positive subpopulations whose partial colocalization with SRIF has been reported (Gonchar et al., 2007; Xu et al., 2010; Rudy et al., 2011). SRIF-positive interneurons are homogeneously distributed in all cortical layers (2–6), as compared to PV-positive inhibitory interneurons that are concentrated in the upper part of the layer (Perrenoud et al., 2012).

The major class of SRIF interneurons, the Martinotti cells, have ascending axons that arborize and spread horizontally in layer 1, targeting the distal dendritic parts of excitatory pyramidal neurons (for review, see Viollet et al., 2008). Excitatory inputs onto Martinotti cells are generally strongly facilitating, allowing feedback inhibition of the excited pyramidal cell that increases as function of the rate and the duration of the presynaptic discharge (Kapfer et al., 2007; Silberberg and Markram, 2007). The relative distance between excitatory and interneurons inputs may also impact feedback selectivity and grade, inhibition being stronger for closer inputs. A recent study using a two-photon microscopy approach coupled to uncaged glutamate in cortical slices of GIN mouse mapped the inhibitory network between SRIF-positive interneurons and pyramidal cells at the single-cell resolution (Fino and Yuste, 2011). Whatever the pyramidal cell stimulated, it led to a dense innervation of the surrounding somatostatinergic interneurons, with activity related to the proximity of the cells. Notably, this inhibitory connectivity looked unspecific as all inhibitory interneurons were locally connected to every sampled pyramidal cells regardless whether these were connected among themselves or not. This dense circuit and the fact that somatostatinergic neurons electrically communicate via gap junctions (Ma et al., 2006; Hu et al., 2011) favors the hypothesis that the entire somatostatinergic population belongs to a same inhibitory cortical circuit, contradicting the hypothesis of specific inhibitory cortical subnetworks.

Additional classes of cortical SRIF inhibitory interneurons have been recently described according to their localization, intrinsic firing properties, expression of molecular markers, and connectivity (Gonchar et al., 2007; McGarry et al., 2010). On one hand, calretinin expression was proposed as a distinctive marker (Rudy et al., 2011), since its expression is associated to distinct neuronal morphology and connectivity in populations with distinct ontogenic origin (Sousa et al., 2009; Xu et al., 2010). On the other hand, two novel SRIF-positive subtypes were identified after cluster and principal component analysis of a whole range of morphological or electrophysiological parameters (McGarry et al., 2010). These cell types have some similarities to neurons labeled in a GABAergic-GFP strain distinct from GIN (X94 strain; Ma et al., 2006), such as the lack of expression in the layer 1, but they target different cortical layers. Future identification of their respective calcium-binding proteins and neuropeptides repertoire as well as their molecular phenotype will help to conciliate these independent classifications based on morphological and electrophysiological properties.

"fendo-03-00154" — 2012/12/4 — 23:44 — page 3 — #3

Somatostatin is found in most sensory systems, i.e., retina (Thermos, 2003; Cervia and Bagnoli, 2007). In the olfactory system, SRIF expression has also been described in sparse shortaxon cells scattered in the deep part of the granule cell layer (the main site of intrinsic inhibitory neurons; Shipley and Ennis, 1996; Eyre et al., 2009) and in the peripheral glomerular layer (which receives sensory inputs) in some species (Hwang et al., 2004). Recently, a novel type of somatostatinergic interneurons has been described as predominant in the murine olfactory bulb and specific to this species (Lepousez et al., 2010a). SRIF-positive somata and dendritic fields are restricted to the layer of the olfactory bulb where intrinsic GABAergic interneurons and bulbar principal cells interact through dendrodendritic reciprocal

synapses to initiate local gamma oscillations responsible for odor processing. Electron microscopy evidences suggest that SRIFpositive interneurons also establish reciprocal dendrodendritic synapses with the bulbar principal cells (mitral cells). SRIFpositive neurons have also been described downstream in the olfactory pathway; SRIF interneurons constitute a major GABAergic population in the pars principalis of the anterior olfactory nucleus and in the olfactory tubercle (Brunjes et al., 2011). In both piriform and entorhinal cortices, two cortical structures involved in the processing of odor coding, multipolar SRIFpositive interneurons displaying Martinotti-like morphological and electrical properties are found in the deep (Young and Sun, 2009; Saiz-Sanchez et al., 2010; Suzuki and Bekkers, 2010) and superficial (Saiz-Sanchez et al., 2010; Tahvildari et al., 2012) layers respectively.

As mentioned before, in addition to the somatostatinergic interneurons acting within microcircuits, long-projecting somatostatinergic cells have been described in several regions (Viollet et al., 2008). As glutamatergic pyramidal cells projections do, longrange inhibitory connections mediate communication between multiple brain areas. Long-range inhibitory terminals have a larger diameter and a thicker myelin layer than excitatory projection neurons, suggesting that inhibitory signal may precede the arrival of excitation in co-innervated cortical areas (Jinno et al., 2007). Long-range projecting SRIF-containing neurons are encountered in numerous brain areas (for review, see Viollet et al., 2008) such as the hippocampus (Jinno et al., 2007), the cerebral cortex (Tomioka et al., 2005), and the amygdala (McDonald et al., 2012). For instance, virtually all non-pyramidal neurons in the amygdala that have long-range projecting axons to the basal forebrain in the rat express SRIF (McDonald et al., 2012). **Figure 3** represents the projections of all long-range somatostatinergic interneurons known to date in the brain.

#### **SOMATOSTATIN RECEPTORS IN THE CENTRAL NERVOUS SYSTEM**

Autoradiographic studies characterized initially two SRIF binding site according to their affinity for the synthetic agonist octreotide and their pattern of expression. In the early 1990s, five receptors (sst1−5) belonging to the G-protein-coupled receptors (GPCRs) family were cloned and characterized from various species. Sequence homology is 39–57% among the five subtypes, each being highly conserved across species. They activate multiple intracellular targets (Olias et al., 2004) and display distinct internalization and dimerization properties (Csaba et al., 2012). Based on structural, pharmacological, and operational features, they are now divided into two groups displaying nanomolar affinity for both SRIF and CST: SRIF-1 (sst2, sst3, and sst5 receptors) and SRIF-2 (sst1 and sst4 receptors). **Figure 2B** represents the wide expression of SRIF receptors in the CNS. In contrast to most GPCRs, sst1−<sup>5</sup> are unique because their gene coding sequence is devoid of introns. However, this does not preclude the generation of spliced variants such as a shorter isoform of mouse sst2, named sst2B, originating by the excision of a cryptic intronic sequence (Vanetti et al., 1992), and spliced variants of sst5 in human and rodents (Córdoba-Chacón et al., 2011). While some data suggested that CST also acts through the proadrenomedullin receptorMgrX2 or the ghrelin orexigenic peptide receptors, the existence of specific CST receptors has not been demonstrated (Siehler et al., 2008). Neuronostatin does not bind to SRIF receptors, but some of its effects seem mediated through the central melanocortin system (Yosten et al., 2011). Recent findings have shown that neuronostatin is involved in regulating depressive behavior and nociception (Yang et al., 2011a,b, 2012).

The extended distribution of sst2 receptors in the CNS together with studies using subtype selective SRIF analogs in both *in vivo* and *in vitro* experiments, suggested that these subtypes are the major players in the SRIF receptor family. They have broad inhibitory effects in many neuronal networks including cortex, hippocampus, limbic regions, and sensory systems (retina and

"fendo-03-00154" — 2012/12/4 — 23:44 — page 4 — #4

olfactory system;Viollet et al.,2008; Lepousez et al.,2010a; Radojevic et al., 2011). The sst1 receptor may function as an autoreceptor in basal ganglia, hypothalamus, and sensory systems (Thermos et al., 2006), and in the hippocampus (de Bundel et al., 2010). sst3 receptors are localized to mature neuronal cilia in most brain regions (Stani´c et al., 2008), and pharmacological or genetic blockade of sst3 have marked behavioral effects (Einstein et al., 2010). sst4 receptors are highly expressed in the olfactory bulb, cortex, and hippocampus, where their role remains to be clarified. In the mouse they modulate epileptic activity, whereas in the rat it seems that this effect is largely related to sst2 receptors. Hippocampal sst4 have also been involved in cognitive processes (Gastambide et al., 2009; Sandoval et al., 2011), functionally interacting with sst2 (Dutar et al., 2002; Gastambide et al., 2010). sst5 receptors mediate regulation of GH release and inhibit cell proliferation by SRIF/CST, mainly through sst2/sst5 receptors interaction. The detection of functional truncated forms of sst5 suggests that they could interfere in and modulate those interactions (Córdoba-Chacón et al., 2011).

# **SOMATOSTATINERGIC FUNCTIONS IN THE BRAIN NEURONAL ACTIONS OF SRIF** *Presynaptic mechanisms*

Somatostatin, like other neuropeptides, can modulate CNS excitability via presynaptic mechanisms (Baraban and Tallent, 2004). In rat hippocampus and cortex, SRIF induces a presynaptic inhibition of excitatory neurotransmission leading to a decrease in glutamate release and in the amplitude of evoked synaptic responses (Ishibashi and Akaike, 1995; Boehm and Betz, 1997; Tallent and Siggins, 1997; Grilli et al., 2004). The SRIF-induced decrease in glutamate release is explained by an inhibition of excitatory transmission via a G-protein of the Gi/Go family and modulation of calcium channels. Indeed, SRIF selectively inhibits N-type Ca2<sup>+</sup> channel via the picrotoxin-sensitive G(i)/G(o) protein. Somatostatin can also inhibit N-type Ca2<sup>+</sup> channels in the dentate gyrus (Baratta et al., 2002). By these inhibitory effects on excitatory synaptic transmission, SRIF, co-released with GABA on dendritic shafts of principal neurons, increases and prolongs GABA effect. This presynaptic action on Ca2<sup>+</sup> conductance could explain, at least in part, the inhibitory effect of SRIF on longterm potentiation in the mouse dentate gyrus (Baratta et al., 2002). Other studies suggest that presynaptic K+ channels modulation may also be involved in the SRIF inhibition of excitatory transmission (Tallent and Siggins, 1997). More precisions on the mechanisms have been given by Grilli et al. (2004), demonstrating on synaptosomal preparations from mouse cerebral cortex that activation of sst2 presynaptic receptors may inhibit the cAMP/PKA pathway stimulated by high potassium concentration, leading to a decrease of the evoked glutamate release. If in the hippocampus, cortex and also hypothalamus, the presynaptic effects of SRIF concern almost exclusively the excitatory transmission (Peineau et al., 2003), SRIF is also able to decrease GABA release in different brain structures, such as the rat basal forebrain (Momiyama and Zaborszky,2006), the neostriatum (Lopez-Huerta et al.,2008), and the thalamus (Leresche et al., 2000). In the basal forebrain, SRIF presynaptically inhibits both GABA and glutamate release onto cholinergic neurons in a Ca2+-dependent way.

"fendo-03-00154" — 2012/12/4 — 23:44 — page 5 — #5

**FIGURE 3 | Schematic representations of long-range somatostatinergic interneurons in the central nervous system. (A)** Telencephalic efferent projections to the rest of the brain. **(B)** Efferent projections arising from the diencephalon and projecting to the telencephalon and the pons. **(C)** Efferent projections arising from the mesencephalon, the medulla oblongata, the pons and the spinal cord. Enteped, entopeduncular; BLA, basolateral amygdala; BNST, bed nucleus of the stria terminalis; CEA, central amygdala; DG, dentate gyrus; DRG, dorsal root ganglia; MPOA, medial preoptic area; NTS, nucleus of the solitary tract; OB, olfactory bulb; PAG, periaqueductal gray; POA, preoptic area; PVN, paraventricular nucleus.

#### *Postsynaptic mechanisms*

Effects of SRIF on intrinsic neuronal membrane properties are well documented. Somatostatin induces a membrane hyperpolarization resulting from the activation of two distinct K+ current, the voltage-sensitive K<sup>+</sup> current or M-current (*I*M; Moore et al., 1988; Jiang et al., 2003), and a voltage-insensitive leak current (Schweitzer et al., 1998). In hippocampal CA1 pyramidal neurons, sst4 seems to be the receptor subtype that couples to *I*<sup>M</sup> (Qiu et al., 2008).

In medium spiny neostriatal neurons, SRIF produces a qualitative change in the firing pattern from a tonic regular to an interrupted"stuttering"-like pattern (Galarraga et al., 2007). These authors demonstrated that SRIF changes the firing pattern via sst2-subtype activation, which reduces the small conductance Ca2+-activated K<sup>+</sup> currents (SK-channels) and activates large conductance g(K)Ca2<sup>+</sup> (GK channels). These results highlight the fact that SRIF is a regulator of cellular function in the striatum. The numerous effects of SRIF on Ca2<sup>+</sup> and K<sup>+</sup> channel conductance in different structures are reviewed by Cervia and Bagnoli (2007).

A huge amount of literature has tried to define the pharmacological nature of SRIF effects, using agonists and antagonists of SRIF receptors or mice invalidated for receptor subtypes. Results are often controversial and are different in mice and rats (Aourz et al., 2011). Therefore, the classification of SRIF effects is complex and it is accentuated by the description of functional cooperation between different receptor subtypes sst2/sst3, sst2/sst4, sst3/sst4 (Moneta et al., 2002; Gastambide et al., 2010; Aourz et al., 2011). Recent publications suggest that sst3 and sst4 (but not sst1; de Bundel et al., 2010) have potent anticonvulsive properties (Aourz et al., 2011), and that sst2, the major receptor subtype involved in the anticonvulsant effect of SRIF in the hippocampus exerts a functional cooperation with sst3/sst4. In hippocampus, sst1 activation inhibits both NMDA- and AMPA-mediated responses but did not affect the inhibitory transmission (Cammalleri et al., 2009).

# **SRIF-CONTAINING NEURONS ARE INVOLVED IN PHYSIOLOGICAL FUNCTIONS**

#### *Interneurons*

A large diversity of inhibitory interneurons is able to exert inhibition on specific compartments of principal cells. Among these populations is the dendrite-targeting SRIF-expressing interneuron located in oriens-lacunosum moleculare of the hippocampus. These SRIF-containing neurons are the only subtype of interneuron that reliably follows synaptic stimulation of the alveus in the theta frequency range via activation of their kainate receptors, suggesting that they play an important role in theta band frequency oscillations (Goldin et al., 2007). Spontaneous activities of inhibitory interneurons have been characterized and SRIFcontaining neurons are described in the cortex and piriform cortex as regular-spiking (Kawaguchi and Kubota, 1998; Suzuki and Bekkers, 2010) or low-threshold spiking neurons (Goldberg et al., 2004), often opposed to the fast spiking PV-containing neurons. In the hippocampus, SRIF neurons are locked to the ascending phase of the theta cycle. However, using an optogenetic inhibition of different populations of interneurons, it was recently demonstrated that silencing SRIF interneurons increases burst firing of

pyramidal cells without altering the theta phase of spikes (Royer et al., 2012). Applying optogenetic technique to animals trained to run head-fixed on a treadmill belt rich with visual-tactile stimuli, these authors provided evidence that the dendritic (but not somatic) inhibition of pyramidal neurons by SRIF interneurons is critical for controlling spike burst firing during active exploration. They concluded that perisomatic PV-targeting interneurons control the spikes' theta phase while the dendrite-targeting SRIF interneurons control the rate of discharge. This is in agreement with the fact that dendritic but not somatic GABAergic inhibition is decreased in experimental epilepsy (Cossart et al., 2001). Combining optogenetic stimulation with *in vivo* two-photon imaging in the mouse visual cortex, Wilson et al. (2012) demonstrate that soma-targeting PV neurons regulate the gain of cortical response, while dendritic-targeting SRIF neurons shift response level and alter stimulus selectivity, leaving response gain unaffected.

Another demonstration of the role of SRIF interneurons in cellular function has been given recently (Gentet et al., 2012). In this study, SRIF neurons recorded in the barrel cortex of awake mice were tonically active during quiet wakefulness but they decreased their firing during whisker sensorimotor processing. This decrease in firing relieves the dendrites of excitatory pyramidal neurons from inhibition.

It is known that inhibitory neurons have diverse roles in physiological and synaptic function, based on their connectivity patterns and intrinsic properties. All the experiments described above demonstrated that SRIF interneurons have a prominent role in the regulation of distal dendrites excitability.

#### *Long-range projecting neurons*

The long-range projecting somatostatinergic non-pyramidal cells found in the hippocampus target the medial septum and the medial entorhinal cortex (Viollet et al., 2008; Melzer et al., 2012) and more specifically form inhibitory synapses on GABAergic interneurons of these areas. They coordinate activity between distant brain regions, contributing to the generation and the synchronization of rhythmic oscillatory activity in the hippocampus and entorhinal cortex (Melzer et al., 2012). They are therefore involved in spatial and temporal coding. Interestingly, early-generated GABA-containing hub neurons, dendrite-targeting interneurons, express preferentially SRIF and give long-range projecting neurons (Picardo et al., 2011). These superconnected hub cells are present early in the developing hippocampus. They develop a widespread axonal arborization and remain into adulthood. They play a key role in the control of the hippocampal giant depolarizing potentials as well as in the modulation of network dynamics. In the other brain areas, the precise contribution of these long-projecting SRIF neurons in the oscillatory activity still needs to be addressed.

#### *Hypophysiotropic neurons*

"fendo-03-00154" — 2012/12/4 — 23:44 — page 6 — #6

Somatotropin release inhibiting factor was initially discovered as a neurohormone that inhibits GH secretion from anterior pituitary somatotroph cells. This function is exerted by hypophysiotropic neurons, located in the anterior periventricular hypothalamic nucleus, which project to the median eminence and release the peptide in the fenestrated capillaries of the hypothalamo– hypophyseal portal vessels; thus directly connecting the brain to

the anterior pituitary. SRIF is also a potent inhibitor of many hormonal and exocrine secretions as well as an antiproliferative agent in normal and tumoral tissue (Epelbaum, 1986). SRIF analogs (octreotide and lanreotide) have potent inhibitory effects on hypersecretion, thereby alleviating the symptoms associated with neuroendocrine tumors. Furthermore, the antitumor potential of octreotide is now well documented. Pasireotide, a long-acting SRIF analog, has the advantage of targeting a wider range of SRIF receptors (subtypes 1, 2, 3, and 5) than the analogs previously used in clinical practice (which preferentially target subtype 2) and has a broader spectrum of activity (for review, see Bousquet et al., 2012).

#### **INVOLVEMENT OF SRIF SYSTEMS IN SENSORY, MOTOR, AND COGNITIVE FUNCTIONS**

Since SRIF systems are widely expressed in CNS, they are involved in numerous functions including nociceptive and vasoconstrictor properties. Here, we will present recent advances about the role of SRIF systems in autonomic responses (digestion, cardiac rate, and respiration) and motor functions as well as cognitive functions such as learning and memory and emotion (for review, see Viollet et al., 2008).

#### *Somatostatinergic involvement in sensory functions*

*Somatostatin and visual information.* Somatostatinergic system is expressed in mammalian retina (for review, see Thermos, 2003; Casini et al., 2005; Cervia et al., 2008), where it is suspected to exert multiple actions on neurons and on retinal physiology. SRIF acts as a positive factor in the retina by regulating homeostasis and protecting neurons against damage. Both sst2 and sst5 somatostatinergic receptors are involved. Indeed, activation of sst2 protects the retina from ischemic insults *ex vivo* (Mastrodimou et al., 2005) and sst2 as well as sst5 receptor activation protect from excitotoxicity *in vivo* (Kiagiadaki and Thermos, 2008; Kiagiadaki et al., 2010; Kokona et al., 2012). The severity of angiogenic responses to hypoxia is correlated to the sst2 expression level in the retina (Dal Monte et al., 2007). Moreover, the sst2-preferring agonist octreotide prevents hypoxia-induced VEGF up-regulation (Dal Monte et al., 2009).

*Somatostatinergic modulation of olfactory discrimination.* Recent studies have shown that SRIF modulates olfactory processing in mice (Lepousez et al., 2010a,b). In mouse main olfactory bulb, SRIF is mainly concentrated in local GABAergic interneurons synaptically connected to the mitral cells by reciprocal dendrodendritic synapses. When activated by an odor, mitral cells synchronize and generate gamma oscillations of the local field potentials that are involved in olfactory processing. Pharmacological or genetic blockade of sst2 transmission in the olfactory bulb of awake animal selectively decreased the gamma oscillations power while pharmacological activation of sst2 had opposite effects. These treatments were respectively correlated to either impairment or improvement of odor discrimination performances of the pharmacologically injected animals. Thus, bulbar endogenous SRIF, presumably released from external plexiform layer interneurons, affects gamma oscillations through the dendrodendritic reciprocal synapse and contributes to olfactory processing.

#### *Involvement of SRIF in learning and memory*

It has been reported for decades that SRIF plays a role in learning and memory at different stages of information processing. The first studies investigating its role in cognition showed that intracerebroventricular administrations of SRIF improved learning in active avoidance tasks (Bollok et al., 1983; Vecsei et al., 1983; Vecsei and Widerlov, 1988) and prevented electroshockinduced amnesia in passive avoidance paradigms (Vecsei et al., 1983, 1984). Conversely, the depletion of SRIF in the brain by cysteamine (which depletes SRIF levels; Szabo and Reichlin, 1981) produced major memory deficits in passive avoidance (Bakhit and Swerdlow, 1986; Schettini et al., 1988; DeNoble et al., 1989). These studies revealed that SRIF is involved in the acquisition of information but other studies showed that cysteamine produced memory deficits not only when given before the training session but also within a critical time window (0–4 h) after acquisition, suggesting that SRIF plays a critical role in memory consolidation processing (Haroutunian et al., 1987, 1989; Schettini et al., 1988; Vecsei et al., 1990).

The hippocampus is an essential structure in learning and memory (Jeneson and Squire, 2012), and is also a chosen site to study the effects of SRIF on learning and memory since injection of cysteamine impairs tasks requiring its integrity (DeNoble et al., 1989; Guillou et al., 1998). Surprisingly in the rodent hippocampus, both activation of SRIF receptors as well as depletion of SRIF contents generate hippocampal memory impairments. Indeed, microinjections of cysteamine, SRIF or CST directly into the hippocampus impaired hippocampal-dependent spatial learning (Guillou et al., 1993; Sanchez-Alavez et al., 2000; Lamirault et al., 2001; Mendez-Diaz et al., 2005; Gastambide et al., 2009). Consistent with these pharmacological results, transgenic mice overexpressing CST display a profound impairment of spatial learning (Tallent et al., 2005). Studies that investigated which SRIF receptor mediates SRIF memory effect showed that intrahippocampal injections of the sst4 agonist, but not sst1, sst2, or sst3 agonists, dramatically impaired spatial memory formation (Gastambide et al., 2009). Importantly, these authors found that concomitantly to the impairment of spatial memory, an sst4 agonist also enhanced the use of striatum-dependent memory. Therefore, it was hypothesized that hippocampal sst4 controls the use of cognitive strategies by switching from hippocampusbased multiple associations to simple striatum-based behavioral response through a functional interaction with sst2 receptor (Gastambide et al., 2010). The precise cellular and molecular mechanisms involved in this functional interaction between sst2 and sst4 are not fully understood but some studies showed that sst4 mediates increases in glutamatergic excitability and bursting frequency, which were blocked by sst2 agonists or antagonists and were lacking in sst2 knockout (KO) mice (Moneta et al., 2002; Cammalleri et al., 2006). Therefore, sst4 is not the unique SRIF receptor in the hippocampus mediating SRIF memory effects as sst2 also modulates memory as previously suggested by Dutar et al. (2002).

#### *Involvement of SRIF in the control of emotion*

"fendo-03-00154" — 2012/12/4 — 23:44 — page 7 — #7

Somatostatin and its receptors are strongly expressed in the different nuclei of the amygdala (Hannon et al., 2002), a key brain

structure involved in the emotional assessment of the environment (Schumann et al., 2011). Despite the extensive expression of SRIF systems in this area, the effects of SRIF on emotions have not yet been studied extensively. Nevertheless, some studies reported an involvement of SRIF systems in the control of emotion and anxiety. Indeed, a very recent work revealed that the pattern of activation of SRIF-positive interneurons was specific to the nuclei of the amygdala considered and also to the kind of stressor used (Butler et al., 2012). Moreover, SRIF has anxiolytic- and antidepressant-like effects (Engin et al.,2008b) that are associated with the suppression of the frequency of hippocampal theta activity, a neurophysiological signature common to most classes of anxiolytic drugs (i.e., benzodiazepines, selective 5-HT reuptake inhibitors, 5-HT1A agonists). These effects seem to be mediated by sst2 receptor since both intra-septal and intra-amygdala SRIF microinfusions induced anxiolytic effects that were completely reversed by selective sst2 receptor antagonist injection in these brain areas (Yeung and Treit, 2012). Additional evidence for a specific role of sst2 receptor came from the observation that a stressful experience is associated with an increase of sst2 mRNA levels within the amygdala (Nanda et al., 2008) and that mice lacking sst2 receptor display increased anxiety-like behaviors associated with increased pituitary ACTH levels, a main regulator of the stress response (Viollet et al., 2000).

#### *Involvement of SRIF in locomotion*

An involvement for SRIF was also reported in motor functions. Increased motor activity was shown in rats receiving intracerebroventricular administration of SRIF (Havlicek et al., 1976) as well as in mice receiving unilateral striatal infusions of the peptide by retrodialysis (Hathway et al., 2004) and in animals receiving direct injections of SRIF in the nucleus accumbens (Raynor et al., 1993). Tashev et al. (2001) showed that SRIF modulated locomotor activity in biphasic manner. Indeed, shortly after SRIF striatal injection a decrease of locomotor activity is observed whereas later, the locomotor behavior is increased. Similar effects have been found after striatal injection of sst2 and sst4 agonists. On the other hand, genetic invalidation of sst2 receptor in two different strains of mice as well as SRIF null mice showed an impairment of motor functions (Viollet et al., 2000; Zeyda et al., 2001; Allen et al., 2003). But the role of SRIF in locomotion seems to be limited to fine motor control since these different lines of transgenic mice only develop impaired motor coordination in tasks that require a fine motor control and display normal levels of motor activity and coordination in undemanding tasks (Viollet et al., 2000; Zeyda et al., 2001; Allen et al., 2003).

#### *Autonomic responses*

Somatotropin release inhibiting factor and its receptors are found in several medulla oblongata nuclei that control autonomic functions such as digestion, cardiac rate, and respiration (Llona and Eugenín, 2005; Spary et al., 2008; Viollet et al., 2008). In the preBötzinger complex (preBötC), a critical component of the respiratory rhythm generator that underlies mammalian breathing, SRIF is expressed in a subpopulation of glutamatergic neurokinin 1 receptor-positive neurons, a kind of neuron rhythmically active (Stornetta et al., 2003). Originating from the homeogene Dbx1 lineage, these cells are mandatory for breathing, since invalidation of the Dbx1 gene impaired their differentiation and disrupted respiratory rhythm generation in the preBötC (Bouvier et al., 2010; Gray et al., 2010). Acute silencing of somatostatinergic preBötC neurons increased respiratory rhythm, leading to persistent apnea (Tan et al., 2008). Similar effects were found *in vitro* after pharmacological blockade of sst2 transmission, while exogenous SRIF application decreased rhythms generation (Pantaleo et al., 2011; Ramírez-Jarquín et al., 2012). This demonstrated that the peptide exerts a tonic inhibitory control on the rythmogenic neurons in order to avoid deleterious overactivity, probably through cellular subdomain-specific inhibitory and excitatory synaptic contacts (Wei et al., 2012). The existence of long-range somatostatinergic projections to either contralateral PreBötC (Stornetta et al., 2003) or downstream premotor neurons (Tan et al., 2010) favors a neuromodulatory role for Pre-BötC SRIF (Llona and Eugenín, 2005), whose developmental impairment may be involved in human pathologies (Schwarzacher et al., 2011) such as the sudden infant death syndrome (Lavezzi and Matturri, 2008).

# **SOMATOSTATINERGIC NETWORKS IN PATHOLOGICAL CONDITIONS**

In animals, an alteration of SRIF systems is observed during normal aging (Stanley et al., 2012) and pathological models of aging. In human a similar specific dysregulation is observed in normal pathological disorders such as some neurodegenerative and psychiatric diseases (Glorioso et al., 2011; Gleichmann et al., 2012).

#### **ALZHEIMER'S DISEASE**

"fendo-03-00154" — 2012/12/4 — 23:44 — page 8 — #8

Somatostatin has been involved in Alzheimer's disease (AD) pathology for a number of years. Indeed, since the early 1980s, it is known that SRIF levels in cortex and hippocampus are decreased in AD patients (Davies et al., 1980). Later, it was demonstrated that the decline in SRIF concentrations in the CSF (Tamminga et al., 1987) or in the middle frontal gyrus (Dournaud et al., 1995) correlates with cognitive deficits. Using quantitative real-time PCR, a recent study confirmed this decrease of SRIF in the inferior, medial, and superior temporal lobe of AD patients (Gahete et al., 2010). Interestingly, SRIF concentrations were reported to be significantly lower in Alzheimer patients carrying the epsilon 4 allele of APOE (Grouselle et al., 1998), the main genetic risk factor described to date for late-onset AD (Genin et al., 2011). In addition, two different studies found in Finnish and Chinese patients that polymorphisms in the SST gene are associated with the risk of developing AD (Vepsalainen et al., 2007; Xue et al., 2009).

Regarding SRIF receptors, data are limited and controversial. Although all studies agreed on a decrease of SRIF receptors in AD, controversies appeared about the proportion, the localization, and receptor subtype specificity of this decrease. SRIF receptor quantification using quantitative real-time PCR in AD temporal lobe showed a decrease of sst1, sst3, and sst4 receptors whereas sst2 and sst5 receptors were unchanged (Gahete et al., 2010). Previously, an immunohistochemistry study reported a similar decrease of sst4 but showed a reduction in neuronal sst5 – and a modest decrease in sst2 –like immunoreactivity without any changes in sst1 immunoreactive neurons (Kumar, 2005). Surprisingly, in the same study, an increase of sst3 subtype was observed in AD cortex. A radioligand binding and functional study showed a general receptor decrease in AD brain (Beal et al., 1985). More specifically, receptors levels in the frontal and temporal cortex were reduced by approximately 50% of control values in AD patients while a 40% reduction was reported in the hippocampus and no significant changes were found in the cingulate cortex, postcentral gyrus, temporal pole, and superior temporal gyrus. Another radioligand binding study revealed that while the maximal binding capacity of the SRIF-1 receptor subtype (primarily sst2, and possibly sst5) is altered in frontal and temporal cortices, other putative cortical SRIF receptor classes (SRIF-2 sites, i.e., sst1 and sst4) are not as broadly affected (Krantic et al., 1992). Finally, a last study showed a significant decrease only in the frontal cortex, but not in other brain regions (Bergstrom et al., 1991). Because of the cholinergic hypothesis regarding AD etiology, it was concluded that the pattern of change of SRIF binding in AD cortex might be secondary to the degeneration of SRIF receptor-bearing cholinergic afferents arising from the nucleus basalis. In line with this idea, experiments in the literature demonstrate that the selective destruction of cholinergic neurons of the basal forebrain with intracerebroventricular injection of 192-IgG saporin produces an irreversible loss of SRIF-immunoreactive neurons in the hilus of the hippocampus (Jolkkonen et al., 1997) and in the cortex (Zhang et al., 1998). This last study shows a correlation between the intensity of acetylcholinesterase in the cortex and the number of remaining SRIF cells. These data highlight a trophic dependence of SRIF neurons on cholinergic inputs and are consistent with observations in AD and aging.

Although SRIF deficit is not correlated with the amyloid load in AD brain patients (Dournaud et al., 1995), SRIF was identified as a modulator that increases brain neprilysin activity, one of the main enzymes involved in Aβ degradation (Saito et al., 2005). Recently, it has been shown that neuropeptide pituitary adenylate cyclase-activating polypeptide slows down AD-like pathology and improves cognition in a transgenic mouse model of AD through the activation of SRIF-neprilysin cascade (Rat et al., 2011). In mouse primary embryonic neurons, SRIF concomitantly increased neprilysin activity and decreased Aβ42 in the culture medium and these effects were blocked by an sst5 antagonist (but also an agonist at sst1 and sst3 receptors; for review, see Epelbaum et al., 2009). Moreover, neprilysin activity was decreased by 50% and Aβ42 increased by a similar extent in SRIF KO mice (Saito et al., 2005). Such findings may have important implications for understanding the cellular mechanisms leading to AD and suggest that SRIF and its receptors are potential pharmacological targets for AD. Indeed, FK962, which promotes SRIF production in the brain, co-administrated with donepezil, an acetylcholinesterase inhibitor widely used to treat patients, enhances cognition in rat and has been proposed as an add-on therapy for AD (McCarthy et al., 2011). In addition, Rubio et al. (2012) recently suggested that SRIF and CST act as a protective agent against Aβ toxicity. However, in APP transgenic mouse models, data concerning SRIF-containing interneurons are contradictory. In the triple-transgenic model of AD, 3×Tg-AD, inhibitory neurotransmission is unchanged in the cerebral cortex and hippocampus (Gleichmann et al., 2012). In a APP/PS1 mouse model of AD, as soon as 6 months of age, a decrease in the number of oriens-lacunosum moleculare hilar perforant path-associated SRIF-positive interneurons was evidenced in the hippocampus, when no change was demonstrated for 21 additional mRNA markers tested (Ramos et al., 2006). In the APPswe/PS1dE9 mouse model,Aβ deposition disrupted cognitive circuits when the cholinergic and somatostatinergic systems remained relatively intact (Savonenko et al., 2005). Another study on this last model even found that, in most brain regions tested, SRIF concentrations were increased rather than decreased relative to controls (Horgan et al., 2007). Thus, the validity of a direct and major role for SRIF in the regulation of Aβ42 degradation remains to be further confirmed (Iwata et al., 2005). More recent studies, focusing on olfaction, an early-altered function in AD (Wilson et al., 2009), account for evidence of a relationship between Aβ pathology and SRIF alterations in the disease. Indeed, SRIF interneurons and receptors are selectively reduced by approximately 50% in the anterior olfactory nucleus of AD patients (Saiz-Sanchez et al., 2010). These authors suggested that SRIF decreases in AD might be linked with Aβ. Moreover, an increase in the levels of aggregated Aβ peptide is observed with aging in olfactory cortices of APP/PS1 transgenic mouse model of AD, and it is accompanied by a fall in numbers of SRIF-positive interneurons (Saiz-Sanchez et al., 2012).

Experiments from our laboratory demonstrated that intrahippocampal injections of Aβ in rats induced aberrant inhibitory septo-hippocampal network activity associated with an impairment of hippocampal memory processes (Villette et al., 2010). This effect can be explained by the selective loss of long-range hippocampo-septal projecting neurons population containing calbindin and SRIF (Villette et al., 2012). This population of SRIF neurons could be a favored target for Aβ, explaining the early decrease of SRIF observed in AD.

Somatotropin release inhibiting factor is not only interacting with Aβ42 in AD, it has also an effect on Tau phosphorylation. Rubio et al. (2008) indicated that in mouse cortex SRIF and CST induce Tau phosphorylation at Ser262, a site modified in AD (Wang et al., 2007), although with different kinetics. An sst2/sst4 interaction seems implicated in this process but the types of phosphatases that are involved remain to be determined. Moreover, in human apoE4 knock-in mice where Tau phosphorylation and intracellular neurofibrillary tangle-like deposits are detected (Huang et al., 2001; Harris et al., 2003; Brecht et al., 2004), Huang's group showed that the number of SRIF-positive interneurons correlated inversely with the performance of these mice in a spatial memory task (Andrews-Zwilling et al., 2010).

#### **PARKINSON'S DISEASE**

"fendo-03-00154" — 2012/12/4 — 23:44 — page 9 — #9

Alteration of SRIF levels is also observed in other neurodegenerative diseases. Indeed, decrease in SRIF levels has been described in demented Parkinson's disease patients (Epelbaum et al., 1989) as well as in a unilateral 6-OHDA experimental mouse model of Parkinson's disease (Nilsson et al., 2009). Recent data obtained in a rat model of Parkinsonism showed that an alteration of presynaptic modulation by SRIF after dopamine deprivation. This

observation may underlie a homeostatic mechanism trying to compensate for the excitability imbalance between direct and indirect basal ganglia pathways found during Parkinson's disease (Lopez-Huerta et al., 2012).

#### **MAJOR DEPRESSIVE DISORDER**

Evidence in major depressive disorder (MDD) suggests an impaired excitation/inhibition balance that is potentially mediated by decreased GABA content (Levinson et al., 2010). More specifically, Sibille et al. (2011)reported a down-regulation of SRIF in the dorsolateral prefrontal cortex (PFC), the subgenual cingulate cortices (Tripp et al., 2011), and the amygdala (Guilloux et al., 2011) of MDD patients. Engin et al. (2008a) and Engin and Treit (2009) revealed an antidepressant effect of SRIF mediated by either sst2 or sst3 receptor and suggested that while SRIF itself is not appropriate for clinical use because of its short half-life and diverse range of effects (Pinter et al., 2006), a closely related SRIF derivative may have some potential for the pharmacological treatment of depression.

#### **SCHIZOPHRENIA**

One of the most consistent findings in schizophrenia neuropathology is deficits in cortical inhibitory interneurons across multiple cortical regions (Hashimoto et al., 2008). It has been known for years that cerebral cortical concentrations of SRIF are reduced in schizophrenics (Roberts et al., 1983) as well as hippocampal concentration (Ferrier et al., 1983; Konradi et al., 2011). Moreover, Hashimoto et al. (2008) found that subjects with schizophrenia exhibited deficits in SRIF expression in the PFC, and this was further confirmed after global analysis from six previously published microarray studies (Perez-Santiago et al., 2012). A recent study suggested that this decrease of SRIF-positive inhibitory interneurons in the PFC may be related to changes in an inflammatory response pathway that are often observed in schizophrenics (Fillman et al., 2012). In addition, Beneyto et al. (2012) showed that SRIF neurotransmission in the PFC of subjects with schizophrenia is also altered at the postsynaptic level in a receptor subtype-, layer-, and cell type-specific manner. The expression of sst2, but not sst1, mRNA is preferentially lower in layers 5–6, and in larger, putative pyramidal neurons in those layers. These authors suggested converging pre- and postsynaptic mechanisms to reduce inhibitory neurotransmission in pyramidal neurons in the PFC, which could alter the synchronization of low frequency oscillations and disturb working memory performance in subjects with schizophrenia.

#### **EPILEPSY**

Somatostatin is highly expressed in brain regions associated with seizures and has been implicated as playing a prominent role

#### **REFERENCES**

Allen, J. P., Hathway, G. J., Clarke, N. J., Jowett, M. I., Topps, S., Kendrick, K. M., et al. (2003). Somatostatin receptor 2 knockout/lacZ knockin mice show impaired motor coordination and reveal sites of somatostatin action within the striatum. *Eur. J. Neurosci.* 17, 1881–1895.

Andrews-Zwilling, Y., Bien-Ly, N., Xu, Q., Li, G., Bernardo, A., Yoon, S. Y., et al. (2010). Apolipoprotein E4 causes ageand Tau-dependent impairment in epilepsy (Vezzani and Hoyer, 1999) based on the observation of an activity-dependent release of SRIF during seizures, the modulation of SRIF mRNA expression, peptide and receptors levels by seizures and the effect of SRIF and its analogs on seizures (Tallent and Qiu, 2008; Zafar et al., 2012). Temporal lobe epilepsy (TLE) is characterized by hippocampal sclerosis together with profound phenotypic changes of different classes of interneurons. Hilar SRIF interneurons undergo extensive degeneration in patients with hippocampal sclerosis (de Lanerolle et al., 1989; Robbins et al., 1991). Recently, this selective neurodegeneration has been linked to the specific enrichment of somatostatinergic neurons in striatum-enriched phosphatase, an enzyme that counteracts the MAPK neuroprotective pathway (Choi et al., 2007; Florio et al., 2008). SRIF receptors may represent potential therapeutic targets for TLE. Indeed, SRIF is released in characteristic conditions of seizures and SRIF and its analogs affect seizures (Vezzani and Hoyer, 1999; Buckmaster et al., 2002). However, information on the precise contribution of each SRIF receptor on the SRIF-induced inhibition of epileptiform activity is still limited. Although the sst2 receptor is likely to mediate the anticonvulsant effects of SRIF in rat hippocampus (Vezzani and Hoyer, 1999), observations in the mouse support a central role for sst4 (Moneta et al., 2002) and/or sst1 receptors (Cammalleri et al., 2004, 2006) in mediating SRIF inhibition of epileptiform activity. In a rodent model of cortical focal ischemia, sst2 is also activated while the infarct size is significantly reduced in sst2 KO mice (Stumm et al., 2004). However, recent data in rats showed that sst1 receptors do not appear to mediate the *in vivo* anticonvulsive effect of SRIF (de Bundel et al., 2010), whereas sst3 and sst4 mediate this effect through a functional interaction with sst2 receptor (Aourz et al., 2011).

#### **CONCLUSION**

Somatostatin systems are widely expressed in the different brain regions and are involved in numerous processes from sensory to cognitive functions, suggesting that they play major roles in brain functioning. These key roles are illustrated by the decrease of SRIF concentrations observed in neurodegenerative diseases such as AD and Parkinson's disease but also in psychiatric diseases such as schizophrenia and MDD. From this perspective, SRIF systems represent a potential and challenging therapeutic target. Further studies need to be carried on to unravel the role of SRIF systems in all functions they have been implicated in.

#### **ACKNOWLEDGMENTS**

Supported by NeRF post-doctoral fellowship (to Guillaume Martel), ANR-10-MALZ-003-01 SOMADOLF and Fondation Recherche Plan Alzheimer.


"fendo-03-00154" — 2012/12/4 — 23:44 — page 10 — #10

somatostatin sst3 and sst4 receptors mediate anticonvulsive effects*in vivo*: indications of functional interactions with sst2 receptors. *Neuropharmacology* 61, 1327–1333.

Bakhit, C., and Swerdlow, N. (1986). Behavioral changes following central

injection of cysteamine in rats. *Brain Res.* 365, 159–163.


apolipoprotein e4 proteolysis is associated with increased tau phosphorylation in brains of transgenic mice. *J. Neurosci.* 24, 2527–2534.


protein tyrosine phosphatase. *J. Neurosci.* 27, 2999–3009.


"fendo-03-00154" — 2012/12/4 — 23:44 — page 11 — #11


markers identified in the dorsolateral prefrontal cortex of individuals with schizophrenia. *Mol. Psychiatry.* doi: 10.1038/mp.2012.110 [Epub ahead of print].


"fendo-03-00154" — 2012/12/4 — 23:44 — page 12 — #12

neurons. *Proc. Natl. Acad. Sci. U.S.A.* 98, 8838–8843.


Schmid, H. A., and Thermos, K. (2012). Pasireotide (SOM230) protects the retina in animal models of ischemia induced retinopathies. *Exp. Eye Res.* 103, 90–98.


depressive disorder. *Biol. Psychiatry* 67, 458–464.


et al. (2005). Cortistatin modulates memory evocation in rats. *Eur. J. Pharmacol.* 507, 21–28.


"fendo-03-00154" — 2012/12/4 — 23:44 — page 13 — #13

genes implicated in schizophrenia. *J. Psychiatr. Res.* 46, 1464–1474.


26-9100 enhances learning and memory in SAMP8 mice. *Eur. J. Pharmacol.* 654, 53–59.


"fendo-03-00154" — 2012/12/4 — 23:44 — page 14 — #14

preBötzinger Complex somatostatinexpressing neurons induces persistent apnea in awake rat. *Nat. Neurosci.* 11, 538–540.


and open-field activity in rats. *Neuropeptides* 12, 237–242.


in Alzheimer neurofibrillary degeneration. *Eur. J. Neurosci.* 25, 59–68.


"fendo-03-00154" — 2012/12/4 — 23:44 — page 15 — #15

motor learning of somatostatin null mutant mice. *Brain Res.* 906, 107–114.

Zhang, Z. J., Lappi, D. A., Wrenn, C. C., Milner, T. A., and Wiley, R. G. (1998). Selective lesion of the cholinergic basal forebrain causes a loss of cortical neuropeptide Y and somatostatin neurons. *Brain Res.* 800, 198–206.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 11 October 2012; paper pending published: 06 November 2012; accepted: 20 November 2012; published online: 06 December 2012.*

*Citation: Martel G, Dutar P, Epelbaum J and Viollet C (2012) Somatostatinergic systems: an update on brain functions in normal and pathological aging. Front. Endocrin. 3:154. doi: 10.3389/ fendo.2012.00154*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Martel, Dutar, Epelbaum and Viollet. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# The physiological role of orexin/hypocretin neurons in the regulation of sleep/wakefulness and neuroendocrine functions

# *Ayumu Inutsuka and AkihiroYamanaka\**

Department of Neuroscience II, Research Institute of Environmental Medicine, Nagoya University, Nagoya, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Luis De Lecea, Stanford University, USA Christelle Peyron, CNRS UMR5292, INSERM U1028, France

#### *\*Correspondence:*

Akihiro Yamanaka, Department of Neuroscience II, Research Institute of Environmental Medicine, Nagoya University, Furo, Chikusa, Nagoya 464-8601, Japan. e-mail: yamank@riem.nagoya-u.ac.jp

The hypothalamus monitors body homeostasis and regulates various behaviors such as feeding, thermogenesis, and sleeping. Orexins (also known as hypocretins) were identified as endogenous ligands for two orphan G-protein-coupled receptors in the lateral hypothalamic area. They were initially recognized as regulators of feeding behavior, but they are mainly regarded as key modulators of the sleep/wakefulness cycle. Orexins activate orexin neurons, monoaminergic and cholinergic neurons in the hypothalamus/brainstem regions, to maintain a long, consolidated awake period. Anatomical studies of neural projections from/to orexin neurons and phenotypic characterization of transgenic mice revealed various roles for orexin neurons in the coordination of emotion, energy homeostasis, reward system, and arousal. For example, orexin neurons are regulated by peripheral metabolic cues, including ghrelin, leptin, and glucose concentration. This suggests that they may provide a link between energy homeostasis and arousal states. A link between the limbic system and orexin neurons might be important for increasing vigilance during emotional stimuli. Orexins are also involved in reward systems and the mechanisms of drug addiction. These findings suggest that orexin neurons sense the outer and inner environment of the body and maintain the proper wakefulness level of animals for survival. This review discusses the mechanism by which orexins maintain sleep/wakefulness states and how this mechanism relates to other systems that regulate emotion, reward, and energy homeostasis.

**Keywords: orexin, hypocretin, sleep, hypothalamus, optogenetics, neuropeptide**

"fendo-04-00018" — 2013/3/4 — 15:44 — page 1 — #1

#### **INTRODUCTION**

The hypothalamus plays a critical role in maintaining energy homeostasis by coordinating behavioral, metabolic, and neuroendocrine responses (Bernardis and Bellinger, 1996). Within this region, the lateral hypothalamic area (LHA) has been regarded as an important center for feeding and arousal because animal models with LHA lesions exhibit hypophagia and decreased arousal that frequently leads to death. Orexin A and orexin B (also known as hypocretin 1 and hypocretin 2) are neuropeptides expressed exclusively by LHA neurons. Orexin-producing neurons (orexin neurons) project their axons throughout the brain (Peyron et al., 1998; Nambu et al., 1999), which suggests that their functions are varied. Remarkably, dense projections of orexin neurons are observed in the serotonergic dorsal raphe nucleus (DR), noradrenergic locus coeruleus (LC), and histaminergic tuberomammillary nucleus (TMN); and all of these nuclei are involved in promoting arousal (Saper et al., 2005). Prepro-orexin knockout mice, orexin receptor knockout mice, and orexin neuron-ablated transgenic mice all show severely defective sleep/wakefulness cycles (Chemelli et al., 1999; Hara et al., 2001; Willie et al., 2003). Consistently, deficiencies of orexin function were found in human narcolepsy (Nishino et al., 2000; Peyron et al., 2000; Thannickal et al., 2000). These findings clearly show the importance of the orexin system in the regulation of sleep/wakefulness. Past studies also revealed roles for orexin neurons beyond feeding and arousal, including autonomic nervous system control (Sellayah et al., 2011; Tupone et al., 2011) and in reward and stress systems (Boutrel et al., 2005; Harris et al., 2005).

In this review, we first discuss the basic biological features of orexins and their receptors, and we then describe the neuronal inputs and outputs of the orexin neurons. Finally, we discuss the various physiological roles of the orexin system, focusing on the regulation of sleep and wakefulness.

#### **OREXIN AND OREXIN RECEPTORS**

In 1998, two groups independently found the same new peptides by using different strategies. Sakurai et al. (1998) used reverse pharmacology to identify ligands of orphan G-protein-coupled

**Abbreviations:** AgRP, agouti-related peptide; Arc, arcuate nucleus; BAT, brown adipose tissue; BST, bed nucleus of the stria terminalis; CTB, cholera toxin B subunit; DR, dorsal raphe nucleus; DREADD, designer receptors exclusively activated by designer drugs; FEO, food-entrainable oscillator; GPCRs, G-protein-coupled receptors; ICV, intracerebroventricular; LC, locus coeruleus; LDT, laterodorsal tegmental nucleus; LHA, lateral hypothalamic area; NMDAR, *N*-methyl-D-aspartate receptor; NPY, neuropeptide Y; OX1R, orexin receptor 1; OX2R, orexin receptor 2; PPT, pedunculopontine tegmental nucleus; PTX, pertussis toxin; PVN, paraventricular thalamic nucleus; SCN, suprachiasmatic nucleus; TMN, tuberomammillary nucleus; TTC, C-terminal fragment of tetanus toxin; VLPO, ventrolateral preoptic nucleus; VTA, ventral tegmental area.

receptors (GPCRs). They found a novel family of neuropeptides that binds to two closely related orphan GPCRs. Because the injection of the ligands induced feeding behavior, they named the ligands "orexin" after the Greek word orexis, which means appetite (Sakurai et al., 1998). At the same time, de Lecea et al. (1998) isolated cDNAs selectively expressed within the hypothalamus. Two peptides of the cDNAs showed substantial amino acid sequence homology with the gut peptide hormone secretin, so they named these peptides "hypocretin." They suggested that hypocretins function within the central nervous system as neurotransmitters.

Prepro-orexin polypeptide is proteolysed to produce two orexins, orexin A and orexin B. Orexin A is a 33-amino acid peptide of 3.5 kDa, with an N-terminal pyroglutamyl residue and C-terminal amidation. The four Cys residues of orexin A form two sets of intrachain disulfide bonds. This structure is completely conserved among several mammalian species (human, rat, mouse, cow, sheep, dog, and pig). On the other hand, rat orexin B is a 28-amino acid, C-terminally amidated, linear peptide of 2.9 kDa, which is 46% identical in sequence to rat orexin A. The 3.2 kb fragment of the 5- -upstream region of the human prepro-orexin gene is reported to be sufficient to express genes in orexin-containing neurons (Sakurai et al., 1999; Moriguchi et al., 2002).

*In situ* hybridization of prepro-orexin shows orexin-containing neurons are located in the LHA. Prepro-orexin mRNA was shown to be upregulated under fasting conditions, indicating that these neurons somehow sense the animal's energy balance (Sakurai et al., 1998). Recently, the forkhead box transcription factor Foxa2, a downstream target of insulin signaling, was reported to be involved in this transcriptional regulation (Silva et al., 2009).

Orexin A acts on both orexin receptor 1 (OX1R) and 2 (OX2R), while orexin B selectively acts on OX2R (Sakurai et al., 1998). While orexin neurons are localized within the LHA, they have widespread projections throughout the brain (Peyron et al., 1998; Nambu et al., 1999; **Figure 1**). Therefore, it is important to know the distribution pattern of orexin receptors to identify the functional neuronal network. Marcus et al. (2001) used *in situ* hybridization to demonstrate that OX1R and OX2R differ in distribution. OX1R mRNA was observed in many brain regions including hippocampus, paraventricular thalamic nucleus (PVN), ventromedial hypothalamic nucleus, DR, and LC. OX2R mRNA was prominent in a complementary distribution including the cerebral cortex, hippocampus, DR, and many hypothalamic nuclei including PVN, TMN, and the ventral premammillary nucleus. Among these regions, DR, LC, and TMN are well known to be involved in maintenance of the awake state. Consistently, orexindeficient mice display a narcolepsy-like phenotype (Chemelli et al., 1999), as do dogs with a mutation preventing the expression of OX2R (Lin et al., 1999). Note that the regions expressing orexin receptors contain several areas of the hypothalamus, including LHA, PVN, and the arcuate nucleus (Arc), which are all strongly implicated in the modulation of feeding.

#### **SIGNAL TRANSDUCTION SYSTEM OF OREXIN NEURONS**

OX1R and OX2R are seven-transmembrane GPCRs, which transmit information into the cell by activating heterotrimeric G proteins. The signal transduction pathways of orexin receptors were examined in cells transfected with OX1R or OX2R. The inhibitory effect of orexin on forskolin-stimulated cyclic adenosine monophosphate (cAMP) accumulation was not observed in

"fendo-04-00018" — 2013/3/4 — 15:44 — page 2 — #2

**neurons.** Orexin neurons are found only in the lateral hypothalamic area but project throughout the entire central nervous system. Red arrows show excitatory projections, while blue lines show inhibitory projections.

coeruleus; LDT, laterodorsal tegmental nucleus; PPT, pedunculopontine tegmental nucleus; TMN, tuberomammillary nucleus; VTA, ventral tegmental area; VLPO, ventrolateral preoptic nucleus.

OX1R-expressing cells. In addition, orexin-stimulated elevation in [Ca2+]i in OX1R- or OX2R-expressing cells was not affected by pertussis toxin (PTX) pretreatment. These results suggest that OX1R does not couple to Gi proteins. On the other hand, forskolin-stimulated cAMP accumulation in OX2R-expressing cells was inhibited by orexin in a dose-dependent manner, and this effect was abolished by pretreatment with PTX. These results indicate that OX2R couples to both PTX-sensitive and PTX-insensitive proteins (Zhu et al., 2003). Note that orexin has two independent actions on neuronal activity: activation of noisy cation channels that generate depolarization and activation of a protein kinase C (PKC)-dependent enhancement of Ca2<sup>+</sup> transients mediated by L-type Ca2<sup>+</sup> channels (Kohlmeier et al., 2008).

Orexin neurons innervate monoaminergic neurons. In particular, noradrenergic neurons of the LC, dopaminergic neurons of the ventral tegmental area (VTA), and histaminergic neurons of the TMN are activated by orexins (Hagan et al., 1999; Horvath et al., 1999b; Nakamura et al., 2000; Yamanaka et al., 2002). LC neurons exclusively express OX1R, while TMN neurons express OX2R, suggesting that both OX1R and OX2R signaling are excitatory on neurons. Orexin colocalizes with dynorphin (Chou et al., 2001) and glutamate (Abrahamson et al., 2001). It has also been demonstrated that orexin increases local glutamate signaling by facilitation of glutamate release from presynaptic terminals (Li et al., 2002).

# **INPUT TO OREXIN NEURONS**

#### **ANATOMICAL ANALYSIS OF NEURONAL INPUT TO OREXIN NEURONS**

It has been challenging to study the neuronal afferents to orexin neurons because they are scattered sparsely within the LHA. To address this point, retrograde tracing studies were performed. The non-toxic C-terminal fragment of tetanus toxin (TTC) can be utilized to retrogradely transfer the fused protein to interconnected neurons and transport toward the cell bodies of higher-order neurons (Maskos et al., 2002). Sakurai et al. (2005) generated transgenic mouse lines expressing a fused protein of TTC and green fluorescent protein (GFP) exclusively in orexin neurons by using the promoter of human prepro-orexin. They identified several brain regions including the basal forebrain cholinergic neurons, gamma-aminobutyric acid (GABA)ergic neurons in the ventrolateral preoptic nucleus (VLPO), and serotonergic neurons in the median raphe and paramedian raphe nucleus. Moreover, regions associated with emotion including the amygdala, infralimbic cortex, shell region of the nucleus accumbens, and the bed nucleus of the stria terminalis (BST) were found to innervate orexin neurons.

In addition to TTC, the cholera toxin B subunit (CTB) is also used to retrogradely trace neuronal projections. Yoshida et al. (2006) injected CTB into the LHA and counted every labeled cell in rats. Interestingly, they found strong projections from the lateral septum, preoptic area, BST, and posterior hypothalamus. In addition, they also found that hypothalamic regions preferentially innervate orexin neurons in the medial and perifornical parts of the field, but most projections from the brainstem target the lateral part of the field.

The results of these two papers present slight distinctions. TTC::GFP sometimes labeled regions with no known projections to the orexin field such as the medial septum possibly because of transport to second-order neurons or ectopic expression of the transgene. In addition, the TTC::GFP technique also appears to be less sensitive than conventional retrograde tracers, as it failed to label neurons in the lateral septum or VTA – regions that probably innervate orexin neurons as indicated by anterograde tracing and other retrograde tracing studies (Yoshida et al., 2006; Richardson and Aston-Jones, 2012).

Given that inputs to orexin neurons are so anatomically varied and associated with multiple functions, it might be reasonable to hypothesize the existence of subgroups of orexin neurons. Indeed, anterograde tracers injected into the DR marked the lateral LHA preferentially, while injections into the VMH preferentially stained neurons in the medial LHA (Yoshida et al., 2006). With current technology we cannot only trace neuronal projections but also analyze functional connectivity by utilizing optogenetic and pharmacogenetic tools (Lammel et al., 2012).

#### **INPUT FROM HYPOTHALAMUS**

Previous studies indicated that the LHA is innervated by several hypothalamic regions, and some of these innervations project toward orexin neurons in the LHA. The LHA has long been considered a brain region regulating food intake and body weight. The localization of orexin neurons to the LHA and the functions ascribed to orexin neurons suggest that they may constitute components of a central circuitry controlling energy metabolism. Therefore, it is reasonable to assess their connectivity to other neuronal populations involved in ingestive behaviors. Neuropeptide Y (NPY), produced by specific neurons in the hypothalamic Arc (Allen et al., 1983; Chronwall et al., 1985), was demonstrated to be a prominent inducer of food intake upon central administration (Clark et al., 1984; Stanley and Leibowitz, 1985). Projections to orexin neurons from NPY/agouti-related peptide (AgRP) neurons in the Arc were identified by anatomical studies (Broberger et al., 1998; Elias et al., 1998). Furthermore, orexin neurons express NPY receptors, and direct administration of NPY agonists into the LHA increases Fos-like immunoreactivity in orexin neurons (Campbell et al., 2003). These findings suggest NPY excites orexin neurons; however, electrophysiological analyses showed that direct application of NPY instead reduces spike frequency and hyperpolarizes the membrane potential of orexin neurons (Fu et al., 2004).

#### **INPUT FROM LIMBIC SYSTEM**

"fendo-04-00018" — 2013/3/4 — 15:44 — page 3 — #3

Given that depletion of orexin neurons induces the sleep disorder narcolepsy, the limbic system might also provide important projections to orexin neurons. Narcolepsy patients often suffer from an attack called "cataplexy," which is characterized by sudden weakening of postural muscle tone. Cataplexy is often triggered by strong, generally positive emotion while consciousness is preserved during the attack (Honda et al., 1986). This fact implies that orexin neurons may play a role in the physiological responses associated with emotions. Consistently, local injection of orexin into the pedunculopontine tegmental nucleus (PPT) strongly inhibited rapid eye movement (REM)-related atonia in the cat (Takakusaki et al., 2005). Therefore, it is hypothesized that emotional stimuli

increase orexin release in the PPT to prevent muscle atonia in wild-type animals.

The innervations from limbic system may mediate emotional arousal and fear-related responses. Prepro-orexin knockout mice showed weaker cardiovascular and locomotor responses to emotional stress in an awake and freely moving condition (Kayaba et al., 2003). Consistently, air jet stress-induced elevations of blood pressure and heart rate were attenuated in conscious orexin/ataxin-3 mice, in which orexin neurons were specifically ablated by expressing neurotoxic protein (Zhang et al., 2006).

#### **INPUT FROM PREOPTIC AREAS**

The preoptic area, especially the VLPO, plays a critical role in non-REM sleep initiation and maintenance (Lu et al., 2002). The VLPO has multiple inhibitory projections to neurons that release wakepromoting neurotransmitters, including histamine neurons in the TMN, noradrenergic neurons in the LC, serotonergic neurons in the DR, and acetylcholinergic neurons (Sherin et al., 1996, 1998; Steininger et al., 2001; Lu et al., 2002).

Neurons in the VLPO fire at a rapid rate during sleep, with attenuation of firing during the awake period. Likewise, neurons in wake-promoting centers fire rapidly during wakefulness and are relatively quiescent during sleep, with the exception of cholinergic neurons, which are divided into two classes of neurons: one is active in both the awake and REM sleep period, and the other is active only in the REM sleep period.

Orexin neurons are strongly inhibited by both a GABAA agonist, muscimol (Yamanaka et al., 2003b), and a GABAB receptor agonist, baclofen (Xie et al., 2006). Orexin neurons are also innervated by cells in the VLPO that also contain GABA (Sakurai et al., 2005; Yoshida et al., 2006). These observations suggest that VLPO neurons send GABAergic inhibitory projections to wakepromoting neurons including orexin neurons. This pathway might be important to initiate and maintain sleep.

#### **INPUT FROM SUPRACHIASMATIC NUCLEUS**

Given that sleep/wakefulness is a circadian phenomenon, it is reasonable to consider that orexin neurons receive information from the suprachiasmatic nucleus (SCN), which is the center of the circadian rhythm according the environmental light–dark information. Indeed, the circadian fluctuation of orexin levels in the cerebrospinal fluid (CSF) disappears when the SCN is removed (Deboer et al., 2004). Although direct input to orexin neurons from the SCN appears to be sparse, orexin neurons receive abundant innervations from the BST, supraventricular zone, and dorsomedial hypothalamus (DMH; Sakurai et al., 2005; Yoshida et al., 2006), which receive input from the SCN (Leak and Moore, 2001). This suggests the possibility that orexin neurons receive circadian influences from the SCN via these regions. Note that excitotoxic lesions of the DMH reduce the circadian rhythmicity of wakefulness. The DMH projects to orexin neurons (Chou et al., 2003), although the DMH also projects to multiple brain areas such as the LC and the VLPO that are involved in sleep and wakefulness. In addition, considering that the orexin system is involved in food-entrainable oscillator (FEO; Mieda et al., 2004), the circadian change in the activity of orexin neurons might be regulated by other elements, such as energy balance.

# **FACTORS THAT INFLUENCE ACTIVITY OF OREXIN NEURONS**

Electrophysiological studies have identified several modulators that regulate activity of orexin neurons. Recordings from transgenic mice expressing GFP in orexin neurons demonstrated that agonists of ionotropic glutamate receptors activated orexin neurons, while glutamate antagonists reduced their activity (Li et al., 2002; Yamanaka et al., 2003b). These results indicate that orexin neurons are tonically activated by glutamate.

Dopamine, noradrenaline, and serotonin (5-HT) hyperpolarize and inhibit orexin neurons via alpha2 and 5-HT1A receptors, respectively (Yamanaka et al., 2003b; Muraki et al., 2004; Li and van den Pol, 2005). Dopamine-induced hyperpolarization is most likely mediated by alpha2-adrenergic receptors since a very high concentration of dopamine is necessary to induce hyperpolarization and also because dopamine-induced hyperpolarization is inhibited by the alpha2-adrenergic receptor antagonist, idazoxan (Yamanaka et al., 2006). However, it is noteworthy that dopamine potentially affects both dopamine receptors and adrenergic receptors, while the dopamine D2 receptor antagonist eticlopride blocks the actions of dopamine on spike frequency and membrane potential (Li and van den Pol, 2005). Thus, dopamine might act through both alpha2-adrenergic receptor and dopamine D2 receptor.

Recently, it was found that orexin itself excites orexin neurons via OX2R (Yamanaka et al., 2010). This suggests that orexin neurons form a positive-feedback circuit through indirect and direct pathways, which results in the preservation of the orexin neuron network at a high activity level and/or for a longer period.

Calcium imaging using transgenic mice in which orexin neurons specifically express yellow cameleon 2.1 showed that neurotensin, sulfated octapeptide form of cholecystokinin, oxytocin, and vasopressin activate orexin neurons, while 5-HT, noradrenaline, dopamine, and muscimol, a GABAA receptor agonist inhibit these cells (Tsujino et al., 2005). Recently, it was also reported that orexin neurons express glycine receptors throughout adulthood and that glycine inhibits the electric activity of orexin neurons directly and indirectly (Hondo et al., 2011; Karnani et al., 2011b).

Other factors that reportedly influence the activity of orexin neurons include corticotrophin-releasing factor (Winsky-Sommerer et al., 2004), ATP (Wollmann et al., 2005), NPY (Fu et al., 2004), and physiological fluctuations in acid and CO2 levels (Williams et al., 2007). It is noteworthy that the factors that are supposed to be influenced by feeding (such as glucose, ghrelin, and leptin) inhibit the activity of orexin neurons (Yamanaka et al., 2003a). The large variety of factors regulating orexin neuronal activities demonstrates the integral role of orexin neurons in monitoring circadian rhythms, energy balance, and vigilance level.

#### **REGULATION OF OREXIN NEURONS BY HUMORAL FACTORS**

Motivated behaviors such as food-seeking are deeply involved in maintenance of arousal. Orexin neurons are thought to function as the sensor of energy balance. Electrophysiological studies revealed that increasing extracellular glucose concentrations induce striking hyperpolarizations, while decreasing the glucose concentration induces depolarization and increases the frequency of action

"fendo-04-00018" — 2013/3/4 — 15:44 — page 4 — #4

potentials of orexin neurons (Yamanaka et al., 2003a; Burdakov et al., 2005). Importantly, this mechanism is sufficiently sensitive to respond to physiological fluctuations of glucose concentration induced by normal feeding. Note that other dietary nutrients, amino acids, also activate orexin neurons, and they can suppress the glucose response of orexin neurons at physiological concentration (Karnani et al., 2011a).

In addition, the orexigenic peptide ghrelin activated isolated orexin neurons with depolarization and an increase in action potential frequency (Yamanaka et al., 2003a). In contrast, the strong anorectic factor leptin robustly inhibited orexin neurons, causing hyperpolarization and decreasing the firing rate (Yamanaka et al., 2003a). Notably, insulin exerted no direct effects on orexin neurons. These findings are consistent with the idea that orexin neurons act as a sensor of nutritional status (Sakurai et al., 1998). Indeed, transgenic mice without orexin neurons fail to show fasting-induced arousal (Yamanaka et al., 2003a).

# **PHYSIOLOGICAL FUNCTIONS OF OREXIN NEURONS FUNCTIONS IN FEEDING BEHAVIORS AND ENERGY HOMEOSTASIS**

Orexin neuron-ablated transgenic mice show hypophagia and lateonset obesity (Hara et al., 2001), although the severity of the obese phenotype critically depends on genetic background (Hara et al., 2005). These findings imply a role for orexin in the regulation of energy homeostasis. Although orexins stimulate feeding behavior, they do not slow metabolic rate, which might be expected in a system geared for weight gain. Instead, orexins increase both food intake and metabolic rate (Lubkin and Stricker-Krongrad, 1998). Because animals must be aware and active when they seek and eat food, this function might be important for feeding behavior.

The Arc is attributed to the regulation of feeding behaviors. Orexin neurons densely project to this region (Date et al., 1999; Horvath et al., 1999a). Principal components of the regulating system of feeding behaviors include antagonistic and complementary appetite-stimulating (orexigenic) and appetite-suppressing (anorectic) pathways: NPY/AgRP neurons and proopiomelanocortin (POMC) neurons. It was reported that intracerebroventricular (ICV) injection of orexin induced c-Fos expression in NPY neurons of the Arc (Yamanaka et al., 2000). Therefore, orexin-stimulated feeding may occur at least partly through NPY pathways. However, because NPY antagonist (which completely abolished NPY-induced feeding) only partially abolished orexin-induced feeding in rats, other pathways by which orexin induces feeding might exist. POMC neurons of the Arc are known to suppress appetite, and lack of POMCderived peptides or electrical silencing of POMC neurons causes obesity. Orexin neurons might affect feeding behavior by inhibiting POMC-expressing neurons (Muroya et al., 2004). Indeed, orexin suppresses action potential firing and hyperpolarizes the membrane potential of POMC neurons in theArc (Ma et al.,2007).

Intracerebroventricular injection of orexin induces water intake as well as food intake (Kunii et al., 1999). Additionally argininevasopressin, also known as antidiuretic hormone, activates orexin neurons via the V1a receptor (Tsunematsu et al., 2008). These results suggest a role for orexin neurons in fluid homeostasis too.

## **FUNCTIONS IN SLEEP AND WAKEFULNESS**

The roles of orexin neurons in the regulation of sleep and arousal have been reported repeatedly. ICV injection of orexin A or orexin B during the light period increased awake time and reciprocally decreased REM and non-REM sleep time (Hagan et al., 1999; Bourgin et al., 2000; Piper et al., 2000). Sleep fragmentation observed in orexin knockout mice (Chemelli et al., 1999), orexin receptor knockout mice (Willie et al., 2003), and orexin neuron-ablated transgenic mice (Hara et al., 2001) shows us the importance of their physiological functions. Narcolepsy is a sleep disorder characterized by primary disorganization of sleep/wakefulness cycles. It has also been reported that the number of orexin neurons is greatly reduced, and orexin peptide levels in the cerebrospinal fluid are decreased to undetectable levels in narcoleptic patients (Nishino et al., 2000; Peyron et al., 2000; Thannickal et al., 2000). Orexin-ataxin-3 mice are a well known mouse model of narcolepsy. However, in orexin-ataxin-3 mice, orexin neurons are absent from birth, and therefore other neuronal mechanisms might compensate for the function of orexin neurons during development. Indeed, the frequency of cataplexy is not high in these mice. Timing-controlled neuronal ablation models using the tTA-TetO system might overcome this problem.

The activities of monoaminergic neurons in the brainstem and hypothalamus are known to be associated with sleep and awake states. Furthermore, the DR, LC, and TMN monoaminergic neurons express orexin receptors and are densely innervated by orexin neurons. These findings suggest that these regions mediate the effects of orexins. Consistently, noradrenergic neurons of the LC (Hagan et al., 1999), serotonergic neurons of the DR (Brown et al., 2002; Liu et al., 2002), and histaminergic neurons of the TMN (Huang et al., 2001; Yamanaka et al., 2002) have been shown to be activated by orexins. These observations suggest that the activity of these monoaminergic neurons is at least partly regulated by orexins. Orexins also have a strong direct excitatory effect on cholinergic neurons of the basal forebrain (Eggermann et al., 2001), which is hypothesized to play an important role in arousal.

The PPT and the laterodorsal tegmental nucleus (LDT) provide cholinergic afferents to several brain regions and play a pivotal role in the regulation of REM sleep and wakefulness. These regions are also strongly innervated by orexin neurons (Peyron et al., 1998; Nambu et al., 1999). Electrophysiological experiments revealed that the firing rate of cholinergic neurons is increased by orexin A (Burlet et al., 2002). Microinjection of orexin A into the LDT increases awake time and decreases REM sleep time in cats (Xi et al., 2001), and when orexin A is injected into the PPT in cats, an increased stimulus at the PPT is required to induce muscle atonia (Takakusaki et al., 2005). However, it is noteworthy that the LDT/PPT contains other neuronal types beside cholinergic neurons that show activity associated with sleep/wake cycles (Sakai, 2012).

Emerging new studies using optogenetics have revealed the physiological roles of orexin neurons *in vivo*. Direct selective photostimulation of orexin neurons expressing channelrhodopsin2 increases the probability of transition from non-REM or REM sleep to wakefulness (Adamantidis et al., 2007) and activates downstream wake-promoting nuclei such as LC and TMN (Carter et al., 2009). Consistently, it was also shown that direct selective

"fendo-04-00018" — 2013/3/4 — 15:44 — page 5 — #5

inhibition of orexin neurons expressing halorhodopsin induces non-REM sleep (Tsunematsu et al., 2011). Furthermore, optogenetic stimulation of the LC produces immediate sleep-to-wake transitions, whereas the inhibition causes a decrease in wakefulness (Carter et al., 2010). Recently it was also reported that photoinhibition of LC neurons during the photostimulation of orexin neurons cancels these sleep-to-wake transitions (Carter et al., 2012). These findings indicate that the LC is a major effector of orexin neurons in the regulation of sleep and wakefulness. However, it is noteworthy that LC noradrenergic neurons express only OX1R, and that OX1R knockout mice show only weak fragmentation of sleep and no cataplexy (Mieda et al., 2011).

Recently, not only photo-activated ion channels or ion pumps but also other natural and modified proteins have been used to regulate the activity of specific neuronal circuits *in vivo*. One interesting example is "Designer Receptors Exclusively Activated by Designer Drugs (DREADD)." This method employs modified muscarinic receptors (hM3Dq for excitation and hM4Di for inhibition) that have lost their affinity for endogenous acetylcholine but can be activated by a synthetic ligand, clozapine-N-oxide, which can cross the blood–brain barrier (Armbruster et al., 2007; Alexander et al., 2009). Because stimulation of GPCRs with a specific ligand has a longer effect on cellular signaling than optical stimulation, the DREADD system can facilitate the examination of the chronic effects of modulating the activity of specific neurons. Using this technique, it was reported that the excitation of orexin neurons significantly increased the amount of time spent in wakefulness and decreased both non-REM and REM sleep times and that inhibition of orexin neurons decreased wakefulness time and increased non-REM sleep time (Sasaki et al., 2011). Additionally, melanopsin, a photosensitive G-protein-coupled photopigment, makes it possible to control wakefulness by blue light in a way similar to channelrhodopsin (Tsunematsu et al., 2012).

# **FUNCTIONS IN AUTONOMIC NERVOUS SYSTEM**

Orexin-deficient mice show lower blood pressure than wild-type littermates (Kayaba et al., 2003; Zhang et al., 2006). Consistently, ICV injection of orexins increases blood pressure and heart rate (Shirasaka et al., 1999), and these effects are abolished by administration of alpha1-adrenergic receptor antagonist, prazosin, or beta-adrenergic receptor antagonist, propranolol. These results suggest that orexins physiologically stimulate the sympathetic nervous system and regulate energy expenditure.

Heat production in brown adipose tissue (BAT) also contributes to body weight regulation through the maintenance of body temperature. Recently Tupone et al. (2011) reported that orexinergic projections to raphe pallidus increase BAT thermogenesis in rat. This finding provides a new mechanism contributing to the disrupted regulation of body temperature and energy metabolism in the absence of orexin. Orexin neuron-ablated transgenic mice show late-onset obesity, although they also show hypophagia (Hara et al., 2001). The regulation of BAT thermogenesis by orexin neurons might account for this phenotype of energy metabolism (Sellayah et al., 2011).

### **FUNCTIONS IN REWARD AND STRESS SYSTEMS**

To attenuate the symptoms of the sleep disorder, psychostimulants such as amphetamine or methylphenidate are often given to

"fendo-04-00018" — 2013/3/4 — 15:44 — page 6 — #6

neuropeptide Y; POA, preoptic area; POMC, proopiomelanocortin; PPT, pedunculopontine tegmental nucleus; SCN, suprachiasmatic nucleus; TMN, tuberomammillary nucleus; VTA, ventral tegmental area.

system may be important to regulate the activity of orexin neurons to evoke emotional arousal or fear-related responses. Abbreviations: 5-HT, serotonin; ACh, acetylcholine; Arc, arcuate nucleus; BAT, brown adipose narcolepsy patients. Interestingly, drug addiction hardly occurs in these patients. This finding suggests that the orexin system mediates the establishment of drug addiction. The LHA, where orexin neurons exist, is a brain region historically implicated in reward and motivation, and orexin neurons project to many brain areas including the LC, nucleus accumbens, and VTA (Fadel and Deutch, 2002) that are implicated in behavioral responses to drugs of abuse. Orexin directly activates VTA dopaminergic neurons (Nakamura et al., 2000; Korotkova et al., 2003). ICV or local VTA infusion of orexin drives behavior motivated by either food or drug rewards (Sakurai et al., 1998; Boutrel et al., 2005; Harris et al., 2005).

It was demonstrated that orexin A input to the VTA potentiates *N*-methyl-D-aspartate receptor (NMDAR)-mediated neurotransmission via a PLC/PKC-dependent insertion of NMDARs in VTA dopamine neuron synapses (Borgland et al., 2006). Furthermore, intra-VTA microinjection of an OX1R antagonist abolished a conditioned place preference for morphine (Narita et al., 2006) and locomotor sensitization to cocaine (Borgland et al., 2006). These data indicate that orexin signaling plays an important role in neural plasticity relevant to addiction in the VTA.

# **CONCLUDING REMARKS**

Although the name orexin is derived from the word orexigenic after its function in feeding, mounting evidence has revealed various physiological roles for orexin other than feeding, such as maintenance of sleep, autonomous regulation, and reward processing. Orexin neurons in the LHA are anatomically well placed to provide a link between the limbic system, energy homeostasis, and brain stem monoaminergic or cholinergic neurons. Like the hypothalamus where orexin neurons

#### **REFERENCES**


exist, orexin neurons themselves monitor various physiological conditions and coordinate various behaviors to respond to environmental change adequately (**Figure 2**). For example, feeding behaviors affect the activity of orexin neurons through changes in concentration of glucose or amino acids, and these changes modulate the vigilance state, regulating aspects of the autonomic nervous system such as blood pressure, heart rate, and thermogenesis at the same time. These findings indicate a critical role for orexin neurons in the regulation of vigilance states, according to internal and external environments, for survival.

By combining viral-mediated tracing, electrophysiology, and optogenetic manipulations, it might be determined that there are several subpopulations of orexin neurons that project to different target areas. For example, the distribution pattern of orexin neurons appears to be divided into two groups: medial and lateral. Some of the input projections to orexin neurons demonstrate a preference between these two areas as well. With new tools to manipulate specific neuronal projections, we can now study physiological differences within the orexin system. These upcoming findings may reveal that discrete functional units underlie the integral role of the orexin system.

# **ACKNOWLEDGMENTS**

This study was supported by JST PRESTO program and Grant-in-Aid for Scientific Research (B) (23300142), Grant-in-Aid for Scientific Research on Innovative Area "Mesoscopic Neurocircuitry" (23115103) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (Akihiro Yamanaka), and Grant-in-Aid for Young Scientists (B) (24790192) (Ayumu Inutsuka).

the rodent lateral hypothalamus: relationship to the neuropeptide Y and agouti gene-related protein systems. *J. Comp. Neurol.* 402, 460–474.


"fendo-04-00018" — 2013/3/4 — 15:44 — page 7 — #7

neurons express a functional pancreatic polypeptide Y4 receptor. *J. Neurosci.* 23, 1487–1497.


contain dynorphin. *J. Neurosci.* 21, RC168.


"fendo-04-00018" — 2013/3/4 — 15:44 — page 8 — #8


Inutsuka and Yamanaka Orexin neurons and sleep/wakefulness

mediated by the dopaminergic system. *Brain Res.* 873, 181–187.


by a genetically encoded tracer in mice. *Neuron* 46, 297–308.


"fendo-04-00018" — 2013/3/4 — 15:44 — page 9 — #9


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 07 September 2012; accepted: 12 February 2013; published online: 06 March 2013.*

*Citation: Inutsuka A and Yamanaka A (2013) The physiological role of orexin/hypocretin neurons in the regulation of sleep/wakefulness and neuroendocrine functions. Front. Endocrinol. 4:18. doi: 10.3389/fendo.2013.00018*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Inutsuka and Yamanaka. This is an open-access article distributed under the terms of the* *Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

"fendo-04-00018" — 2013/3/4 — 15:44 — page 10 — #10

# Role of neuropeptide FF in central cardiovascular and neuroendocrine regulation

# *Jack H. Jhamandas1\* and Valeri Goncharuk1,2*

<sup>1</sup> Division of Neurology, Department of Medicine, Centre for Neuroscience, University of Alberta, Edmonton, AB, Canada <sup>2</sup> Russian Cardiology Research Center, Moscow, Russia

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Michiru Hirasawa, Memorial University, Canada Jean-Marie Zajac, Centre National de la Recherche Scientifique, France Guy Simonnet, Université Bordeaux Ségalen et Centre National de la Recherche Scientifique, France

#### *\*Correspondence:*

Jack H. Jhamandas, Department of Medicine (Neurology), 530 Heritage Medical Research Centre, University of Alberta, Edmonton, AB, Canada T6G 2S2. e-mail: jack.jhamandas@ualberta.ca

Neuropeptide FF (NPFF) is an octapeptide belonging to the RFamide family of peptides that have been implicated in a wide variety of physiological functions in the brain including central cardiovascular and neuroendocrine regulation. The effects of these peptides are mediated via NPFF1 and NPFF2 receptors that are abundantly expressed in the rat and human brain. Herein, we review evidence for the role of NPFF in central regulation of blood pressure particularly within the brainstem and the hypothalamic paraventricular nucleus (PVN). At a cellular level, NPFF demonstrates distinct responses in magnocellular and parvocellular neurons of the PVN, which regulate the secretion of neurohypophyseal hormones and sympathetic outflow, respectively. Finally, the presence of NPFF system in the human brain and its alterations within the hypertensive brain are discussed.

**Keywords: RFamide, FMRFamide, NPFF1, NPFF2, hypothalamus, paraventricular nucleus, blood pressure, hypertension**

"fendo-04-00008" — 2013/2/5 — 17:45 — page 1 — #1

# **INTRODUCTION**

An understanding of the mechanisms that regulate arterial blood pressure under physiological conditions and in the context of pathophysiological situations such as hypertension represents a major challenge. Essential hypertension is the most common form of hypertension in humans, although its cause is poorly understood. However, there is substantial evidence to indicate that essential hypertension may be related to elevated levels of sympathetic nervous activity, which originate within the central nervous system (CNS).Within the CNS, neural networks governing arterial blood pressure are contained within topographically segregated but interactive cell groups represented at all levels of the neuraxis. Amongst the many neurotransmitters and neuropeptides present in these autonomic regions, emerging evidence indicates that a group of RFamide peptides play an important role in CNS regulation of cardiovascular function.

# **WHAT ARE RFamide PEPTIDES?**

Historically, the cardioexcitatory peptide FMRFamide from the bivalve mollusc *Macrocallista nimbosa* was the first peptide isolated and identified with an Arg-Phe-amide C-terminus (Price and Greenberg, 1977). Since then many bioactive peptides have been isolated from invertebrates and vertebrates, and the extended family of peptides terminating in a penultimate Arg and an amidated Phe residue at the C-terminus (RFamide) exists in all phyla (Yang et al., 1985). These peptides are designated as FMRFamide related peptides (FaRPs) and collectively referred to as RFamide peptides. RFamide peptides have been identified to have diverse biological functions that include pain modulation, inhibition of food intake, regulation of water balance, and potent cardiovascular actions that are mediated through the peripheral and CNSs (Raffa, 1988; Murase et al., 1996; Panula et al., 1996; Hinuma et al., 2000; Zajac and Gouarderes, 2000; Sunter et al., 2001; Fukusumi et al., 2003; Samson et al., 2003; Dockray, 2004).

The recent rapid accumulation of cDNA and genomic DNA sequence data and the development of bioinformatics have had a profound impact on the field of RFamide peptide research, especially on gene identification and analyses of RFamide peptides and their receptors. While some confusion exists on the precise nomenclature used in the literature, five genes encoding five prepropeptide precursors that yield five groups of RFamide peptides have been described in mammals (**Figure 1**). These include the prolactin-releasing peptide (PrRP) family (Hinuma et al., 1998), the family of neuropeptide FF (NPFF; and related peptides neuropeptide AF (NPAF), neuropeptide SF (NPSF), and neuropeptide VF (NPVF); Perry et al., 1997; Vilim et al., 1999; Bonini et al., 2000; Liu et al., 2001), human RFamide related peptides (hRFRPs; Hinuma et al., 2000; Fukusumi et al., 2001), metastin/kisspeptins (Ohtaki et al., 2001), and pyroglutamylated RFamide peptide (QRFP) (26RFa) family (Chartrel et al., 2003; Fukusumi et al., 2003). Of these, NPFF peptides and PrRP have been identified by our laboratory and others to play an important role in CNS regulation of cardiovascular function (Thiemermann et al., 1991; Allard et al., 1995; Jhamandas et al., 1998; Samson et al., 2000; Jhamandas and MacTavish, 2002). hRFRPs, which are encoded by a human gene, have a significant homology to the NPFF family of peptides and their receptors (Fukusumi et al., 2006). Members of the RFRP family (RFRP-1 and RFRP-3) have been recently identified as mammalian orthologs of the avian gonadotropin inhibitory hormone and administration of the selective NPFF receptor antagonist results in potent secretion of gonadotropins that is presumed to be mediated via the NPFF1

receptor (Pineda et al., 2010), Metastatin/kisspeptins have been shown to have anti-migratory effects*in vitro*, metastasis-inhibiting effects *in vivo* (Muir et al., 2001; Ohtaki et al., 2001), and identified to play an important role in regulation of puberty and reproduction via gonadotropin release (Richard et al., 2009). QRFPs are the most recently discovered members of the RFamide family and postulated to play a role in food intake and increased locomotor activity (Chartrel et al., 2003; Fukusumi et al., 2006; Bruzzone et al., 2007).

# **NEUROPEPTIDE FF**

#### **CHARACTERISTICS AND TISSUE DISTRIBUTION**

Neuropeptide FF (Phe-Leu-Phe-Gln-Pro-Gln-Arg-Phe-NH2) is an important member of the RFamide peptide family that is present in the CNS and in the periphery of several mammalian species including humans (for review see Panula et al., 1996). Initial interest in NPFF stemmed from its ability to modulate the antinociceptive effects of opioids (Yang et al., 1985; Chen et al., 2006; Mouledous et al., 2010), however, emerging studies have shown that the neuropeptide may play an equally important role in the central processing of visceral autonomic signals related to feeding, generation of central cardiovascular responses, stress, and neuroendocrine regulation (Panula et al., 1996; Jhamandas and MacTavish, 2003; Simonin et al., 2006). NPFF was the first RFamide peptide to be identified in mammals (Yang et al., 1985). The gene for NPFF has been cloned from human, bovine, rat, and mouse tissue and is highly conserved amongst these species (Yang et al., 1985; Vilim et al., 1999; Hinuma et al., 2000). The precursor mRNA encodes for NPFF and other related peptides (NPAF, NPSF, and NPVF) and distribution of the NPFF mRNA in the brain matches that of NPFF immunoreactivity (Kivipelto et al., 1989; Vilim et al., 1999; Liu et al., 2001). Immunocytochemical and receptor autoradiographic studies reveal that brain regions involved in pain transmission, autonomic and endocrine regulation are enriched in NPFF and its binding sites (Kivipelto et al., 1989; Allard et al., 1992). Concentrations of NPFF and its receptors in the hypothalamus are amongst the highest in the brain (Bonini et al., 2000; Zajac and Gouarderes, 2000; Liu et al., 2001).

### **LIGANDS AND RECEPTORS**

"fendo-04-00008" — 2013/2/5 — 17:45 — page 2 — #2

Two NPFF receptors, NPFF1 (also referred to as OT7TO22) and NPFF2 (also known as HLWAR77), have been cloned and characterized (Bonini et al., 2000; Elshourbagy et al., 2000; Zajac and Gouarderes, 2000). Both receptors are Gi/<sup>o</sup> protein-coupled when expressed in Chinese hamster ovary cells or human embryonic kidney 293 cells (Kotani et al., 2001). These receptors demonstrate some of the highest levels of expression within the rat and human brain and spinal cord but data on distribution of specific NPFF receptor subtypes in these regions is controversial (Bonini et al., 2000; Zajac and Gouarderes, 2000). The emerging picture, based on autoradiographic binding, immunohistochemical and *in situ* hybridization studies, is that NPFF1 receptors are predominantly localized in the hypothalamus and the forebrain, whereas NPFF2 receptors are mainly within the spinal cord, the brainstem visceral autonomic sensory nuclei, and the hypothalamus (Bonini et al., 2000; Gouarderes et al., 2002; Zeng et al., 2003; Goncharuk and Jhamandas, 2004).

NPFF1 receptor has a high affinity for the avian peptide LPLR-Famide and is the candidate receptor for NPVF, NPSF as well as hRPRF1 and hRPRF2 peptides (Bonini et al., 2000; Elshourbagy et al.,2000; Liu et al.,2001). NPFF2, on the other hand, binds NPFF and NPAF (Liu et al., 2001; Fukusumi et al., 2006). Interestingly, PrRP, for which GPR10 has been identified as the endogenous receptor (**Figure 1**), has a relatively high affinity for the NPFF2 receptor and may even have a higher efficacy at the NPFF2 receptor than NPFF (Engström et al., 2003). When administered

intracerebroventricular (icv), PrRP has been observed to produce elevations in arterial blood pressure and heart rate that are strikingly similar to those evoked by icv NPFF and infact can be blocked with the selective NPFF antagonist, RF9 (Ma et al., 2009). Thus, many of the postulated physiological functions of the newer members of the RFamide family, PrRP and hRFRPs, may in fact be mediated via NPFF1 and NPFF2 receptors.

In the past few years, with the synthesis of peptide analogs of NPFF and related peptides, the essential requirements for ligand recognition at the NPFF receptor have emerged (Vyas et al., 2006). However, a major impediment to delineating a physiological role for NPFF peptides has been a lack of suitable antagonists that show selectivity for each of the receptors. DesaminoYLFQPQRa was the first analog to attenuate morphine abstinence signs induced by NPFF (Malin et al., 1995; Prokai et al., 2001) but suffers from poor bioavailability and/or low affinity for NPFF receptors (Fang et al., 2005). PFR(Tic)amide has been shown to demonstrate antagonist activity toward NPFF effect *in vitro*, but behaves as an agonist *in vivo* (Chen et al., 2006). Neuropeptide Y (NPY) ligands such as BIBP 3226 have been reported to interact with NPFF receptors, likely on the basis of structural similarities between these receptors and the C-terminal end of NPY peptides (Mollereau et al., 2002; Fang et al., 2006). Unfortunately, BIBP 3226 and its derivatives that were most potent at the NPFF1 receptor were also able to displace NPY Y1 binding (Fang et al., 2005). The discovery of RF9, a *selective* antagonist at the NPFF receptor (Simonin et al., 2006) has represented a significant advance in dissecting the role of NPFF in a variety of physiological functions. This compound potently and selectively binds to NPFF receptors and indeed blocks the acute cardiovascular effects induced by icv NPFF. In addition, its chronic administration blocks delayed and long-lasting opioidinduced hyperalgesia.

# **NPFF AND CENTRAL CARDIOVASCULAR AND NEUROENDOCRINE REGULATION**

Experimental evidence supporting a key role for NPFF in cardiovascular regulation first became apparent in the mid-1980s when Roth et al. (1987) reported that two NPFF analogs could produce significant pressor effects when administered systemically. Subsequently, focal injections of NPFF into the brainstem nucleus of tractus solitarius, which is the first terminus for cardiovascular inputs originating from the periphery, resulted in an increase in blood pressure and bradycardia that could be attenuated with adrenergic antagonists (Laguzzi et al., 1996). NPFF-synthesizing neurons in the same brainstem nucleus were shown to be activated in response to hemorrhage and to a lesser extent acute drug-induced hypertension (Jhamandas et al., 1998). These NPFF neurons in turn project to more rostral brainstem and hypothalamic cardiovascular centers. Intrathecal and icv administration of NPFF has been demonstrated to evoke dose-dependent elevations in arterial blood pressure and heart rate (Jhamandas and Mac-Tavish, 2002, 2003; Fang et al., 2010). Identity of neural circuits that participate in centrally generated NPFF responses have been best studied in the hypothalamus, where icv NPFF evokes activation of specific sets of chemically defined paraventricular nucleus (PVN) neurons, that control CNS humoral and autonomic outflow to the periphery (Jhamandas and MacTavish, 2003).

# **WHOLE ANIMAL OBSERVATIONS**

The PVN is viewed as a key site for homeostasis and a model nucleus for understanding the central regulation of autonomic and neuroendocrine function in the brain (Cunningham and Sawchenko, 1991). The magnocellular neurosecretory cells of the PVN synthesize either vasopressin or oxytocin and following stimulation, release these hormones from their axonal projections to the posterior pituitary into the systemic circulation (Sawchenko and Swanson, 1982). On the other hand, the parvocellular component of the PVN is more complex and consists of two broad categories of cells, neurosecretory cells and non-neurosecretory (autonomic) cells. The neurosecretory parvocellular neurons are located within the dorsal medial and periventricular PVN and their axons terminate on median eminence portal capillaries to facilitate the release of "factors" regulating anterior pituitary secretion. Neurons of this type for example express corticotrophin-releasing hormone or thyrotrophin-releasing hormone. Parvocellular nonneurosecretory (autonomic) neurons are located within the dorsal cap and ventral medial PVN and project their axons to the brainstem and the spinal cord. Some of the chemical messengers expressed in these types of cells include tyrosine hydroxylase, oxytocin, and somatostatin (Roland and Sawchenko, 1993; Dawson et al., 1998). Central administration of NPFF results in a preferential activation of oxytocin-synthesizing parvocellular PVN neurons that project to the brainstem. Oxytocinergic projections to the solitary-vagal complex have previously been shown to modulate baroreflex control of heart rate and other aspects of circulatory control (Higa et al., 2002; Vela et al., 2010). On the other hand, icv NPFF does not activate the magnocellular vasopressin-secreting PVN neurons as measured by Fos immunohistochemistry suggesting that the effects of NPFF on these subset of PVN neurons are inhibitory (Jhamandas et al., 2006). The latter posit being supported by observation from electrophysiological studies where NPFF inhibits activity of magnocellular vasopressin neurons (see below).

#### **CELLULAR ACTIONS OF NPFF**

"fendo-04-00008" — 2013/2/5 — 17:45 — page 3 — #3

Work from whole animal experiments described above suggests that the differential activation of subsets of hypothalamic PVN neurons may result from distinct effects of NPFF on synaptic activation of magno- and parvocellular neurons of this nucleus. Indeed, cellular electrophysiological recordings from hypothalamic brain slices reveal that NPFF increases the inhibitory synaptic drive to magnocellular PVN neurons through a GABAsynthesizing network of interneurons located within the sub-PVN region (Roland and Sawchenko, 1993; Jhamandas et al., 2006; **Figure 2**). This observation of NPFF augmenting an inhibition of magnocellular vasopressin-secreting PVN cells fits well with the *in vivo* hormone release data, which shows that hypovolemia-induced vasopressin release from the pituitary is blunted by centrally administered NPFF (Arima et al., 1996). On the other hand, NPFF presynaptically *disinhibits* the GABAergic input to the parvocellular PVN, thereby increasing the net excitability of these neurons (**Figure 2**). NPFF also exerts a distinct depolarizing [tetrodotoxin (TTX)-independent] postsynaptic effect on parvocellular PVN neurons (Jhamandas et al., 2007). NPFF-induced excitation of parvocellular PVN neurons

"fendo-04-00008" — 2013/2/5 — 17:45 — page 4 — #4

would be expected to result in an increase in autonomic outflow and sympathetic activation which is precisely what is observed with acute infusions of icv NPFF in the conscious animal. At present the lack of selective antagonists acting at NPFF1 versus NPFF2 receptors makes it difficult to determine whether the differential effects of NPFF are in fact mediated via distinct NPFF receptor subtypes. However, PrRP, which binds to NPFF2 receptors, shows a similar profile of electrophysiological effects on parvocellular PVN neurons as NPFF. Moreover RF9, a selective NPFF receptor antagonist, blocks PrRP actions on these neurons suggesting that both the PrRP and the NPFF effects in the parvocellular PVN are likely NPFF2 receptor mediated (Ma et al., 2009).

#### **NPFF AND HUMAN HYPERTENSION**

There is currently a paucity of knowledge on the role of NPFF, and RFamide peptides in general, in human pathophysiological states such as essential hypertension. All of our current knowledge on the role of NPFF in the regulation of arterial blood pressure is derived from paradigms that rely on studying acute effects of this peptide in experimental animal models, which do not recapitulate the chronic human hypertensive condition. Nonetheless, anatomical relationships between NPFF and its receptors within autonomic centers in the human brain may provide important clues as to the role of this peptide in diverse biological functions. In this regard, immunohistochemical data from our laboratory over several years has identified striking similarities in the distribution of NPFF and its receptors, NPFF1 and NPFF2, in the normal human brain compared to the rat, a species in which much of the behavioral and physiological studies have been done to date(Goncharuk and Jhamandas, 2004, 2008; Goncharuk et al., 2006). In these studies, we have identified significant numbers of NPFF fibers, NPFF1, and NPFF2 receptors in the human parvocellular PVN. The relative preponderance of NPFF (and its receptors) and its intimate anatomical relationship to important cardiovascular regulatory peptides such as corticotropin releasing hormone (CRH) in human hypothalamus suggests an important role for this peptide in hypertension. Interestingly, an up-regulation of CRH-secreting cells in the human hypothalamic PVN of patients who suffered from essential hypertension has been reported (Goncharuk et al., 2001, 2002). Recent immunohistochemical observations from post-mortem brain tissue of hypertensive individuals and age-matched controls indicate a marked reduction of NPFF in discrete cardiovascular brainstem and hypothalamic nuclei of hypertensives (Goncharuk et al., 2011, 2012). In these studies, NPFF immunoreactivity was severely reduced in a subnuclear zone adjacent to the hypothalamic PVN and supraoptic nucleus, a site where dense networks of GABAergic neurons reside. These GABAergic neurons have been identified to mediate arterial baroreceptor inputs that control the release of the pressor hormone vasopressin from the neurohypophysis (Jhamandas et al., 1989). Thus loss of NPFF input to GABaergic cells has the potential to dysregulate cardiovascular reflexes and control of arterial blood pressure.

# **CONCLUSION**

Emerging evidence indicates that structure of RFamide peptides including NPFF is remarkably conserved during evolution. What makes these peptides attractive as therapeutic targets is that they are involved in essential functions such as pain, appetite and feeding, stress, and cardiovascular regulation. Anatomical, molecular, and physiological studies indicate that NPFF plays an important role in brain control of neurohormones and sympathetic outflow. Advances in identification and pharmacology of NPFF receptors and the availability of new and specific antagonists such as RF9 provide a unique opportunity to identify the specific role and

#### **REFERENCES**


in rat hypothalamic paraventricular nucleus. *J. Neuroendocrinol.* 10, 453–459.


relevance of these receptors in physiologicalfunction and in pathophysiological states such as hypertension. NPFF receptor based compounds could serve as potential therapeutic agents in the treatment of hypertension and other autonomic disorders.

#### **ACKNOWLEDGMENTS**

Research contributions of the author were supported by the Canadian Institutes of Health Research and the Institute of Diabetes, Nutrition and Metabolism (MOP 111426). We thank Mr. David Mactavishfor assistance with graphics andMs. C. Krysfor editorial assistance.

function in rats. *J. Biol. Chem.* 278, 46387–46395.


"fendo-04-00008" — 2013/2/5 — 17:45 — page 5 — #5

and [125I]EYF as selective radioligands. *Neuroscience* 115, 349–361.


properties of the neuropeptide FF system. *Biofactors* 36, 423–429.


"fendo-04-00008" — 2013/2/5 — 17:45 — page 6 — #6

for pain modulatory neuropeptide NPFF: induction in spinal cord by noxious stimuli. *Mol. Pharmacol.* 55, 804–811.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 30 October 2012; paper pending published: 14 November 2012; accepted: 22 January 2013; published online: 07 February 2013.*

*Citation: Jhamandas JH and Goncharuk V (2013) Role of neuropeptide FF in central cardiovascular and neuroendocrine regulation. Front. Endocrin. 4:8. doi: 10.3389/fendo.2013.00008*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Jhamandas and Goncharuk. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Neuropeptide GPCRs in neuroendocrinology: the case of activity-dependent neuroprotective protein (ADNP)

# *Illana Gozes\**

The Lily and Avraham Gildor Chair for the Investigation of Growth Factors, Department of Human Molecular Genetics and Biochemistry, Sagol School of Neuroscience, Adams Super Center for Brain Studies, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel \*Correspondence: igozes@post.tau.ac.il

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Dora Reglodi, University of Pecs, Hungary Terry Moody, National Cancer Institute, USA Mario Delgado, Consejo Superior de Investigaciones Científicas, Spain

#### **VASOACTIVE INTESTINAL PEPTIDE (VIP): NEUROPROTECTION AND SYNAPSE FORMATION**

This review is a selected overview of the chosen subject. As, I have recently summarized my work on VIP (Gozes, 2008), I will describe here only selected topics focusing on the GPCR connection of the VIP-regulated protein, activity-dependent neuroprotective protein (ADNP) (Gozes, 2007).

In the mid 1980s we were the first to clone the gene encoding the 28 amino acid neuropeptide, vasoactive intestinal peptide (VIP) (Bodner et al., 1985). This allowed us to follow up VIP gene expression in the brain showing developmental increases at the time of postnatal synapse formation and glial expansion (Gozes et al., 1987). Given the increased expression of VIP at the time of synapse formation and glial expansion, we hypothesized a developmental role for the peptide—associated with synaptogenesis and neuroprotection. Followup studies by the laboratory of Douglas Brenneman indeed showed that VIP provided neuroprotection through glial cells (Brenneman et al., 1987). We teamed up to show together with the group of Ronald McKay that VIP enhances synapse formation through glial cell activation (Blondel et al., 2000).

Together with the late Frank Baldino, we showed extensive VIP mRNA expression in the brain, for example, in the suprachiasmatic nucleus (SCN) (Card et al., 1988), an area regulating diurnal rhythms. Our further studies utilizing our VIP hybrid antagonist showed that VIP function during development

was required for maintenance of diurnal rhythmicity (Gozes et al., 1995). These findings were later confirmed by knocking out either VIP (Loh et al., 2011), or the VIP receptor VPAC2 (Harmar et al., 2002), which brings us to part of the subject of this review, GPCRs.

# **VIP RECEPTORS**

We have recently reviewed the current status of research on VIP receptors (Harmar et al., 2012). In short, VIP and pituitary adenylate cyclase-activating polypeptide (PACAP) are members of a superfamily of structurally related peptide hormones that includes glucagon, glucagon-like peptides, secretin, gastric inhibitory peptide (GIP) and growth hormone-releasing hormone (GHRH). VIP and PACAP exert their actions through three GPCRs— PAC1, VPAC1 and VPAC2—belonging to class B (also referred to as class II, or secretin receptor-like GPCRs). PAC1 receptors are selective for PACAP, whereas VPAC1 and VPAC2 respond to both VIP and PACAP with high affinity (Harmar et al., 2012). As indicated above, VPAC2 was associated with diurnal rhythmicity (Harmar et al., 2002) as well as astrogenesis (Zupan et al., 1998). Other functions, including association with cancer propagation and immunomodulation have been extensively reviewed (as cited above).

#### **VIP RECEPTORS AND NEUROPROTECTION, OUR POINT OF VIEW**

Prior to the molecular cloning of VIP and PACAP receptors, binding and displacement assays coupled to functional assays suggested high affinity binding for VIP on glial cells (astrocytes)—associated with the release of neuroprotective proteins and low affinity binding to astrocytes, associated with cAMP formation (Gozes et al., 1991). With the available VPAC1, VPAC2, and PAC1 clones (including the alternatively spliced PAC1 receptors), we have used antisense oligodeoxynucleotides specific for each of the then known receptor subtypes to identify receptors involved in neuroprotection. Our data showed that one of the PAC1 splice variants, the hop2-like receptor is a receptor that mediates neuroprotection (Ashur-Fabian et al., 1997). The PAC1 cDNA was cloned from rat cerebral astrocytes. Using genetic manipulation we obtained the hop2 splice variant and expressed it in COS-7 cells. Results showed that VIP bound the cloned hop2 splice variant (Pilzer and Gozes, 2006). Stearyl-neurotensin (6–11) VIP (7– 28) (SNH), an antagonist for VIP, was also found to bind hop2. Other studies have shown that this particular antagonist that we have developed (Gozes et al., 1995) binds to all know VIP receptors (Moody et al., 2002). In addition, VIP protected COS-7 cells expressing hop2 from oxidative stress. Parallel assays demonstrated that VIP increased cAMP accumulation in COS-7 cells expressing hop2. These results support the hypothesis that hop2 mediates some of the cytoprotective effects attributed to VIP (Pilzer and Gozes, 2006).

However, the study cited above did not address the requirement for astrocytes for VIP neuroprotection, suggesting additional players.

#### **VIP STIMULATE ASTROCYTE SECRETION OF GROWTH FACTORS**

We teamed up together with Douglas Brenneman to isolate novel growth factors from VIP-stimulated astrocyte conditioned media. Our first attempt included sequential chromatography coupled to functional assays of neuroprotection, yielding activity-dependent neurotrophic factor (ADNF) (Brenneman and Gozes, 1996). This study took us to three directions: (1) identification of an active peptide site within ADNF—potential future therapeutic; (2) production of antibodies against ADNF as tools for functional studies and (3) use of the antibodies for expression cloning toward the identification of novel ADNF-like molecules, thus cloning ADNP (Bassan et al., 1999). We have further identified ADNP immunoreactivity in astrocyte conditioned medium, which was increased upon exposure to VIP (Furman et al., 2004).

# **GPCR INVOLVEMENT IN ADNP EXPRESSION**

Using VIP analogues specific for the VPAC1 and the VPAC2 receptors, we discovered that VIP-induced changes in ADNP expression in astrocytes via the VPAC2 receptor. The constitutive synthesis of ADNP and VPAC2 was shown to be age-dependent and increased as the astrocyte culture developed. The VIP-related peptide, PACAP also induced changes in ADNP expression. The apparent changeinduced by VIP and PACAP on ADNP expression were developmentally dependent, and while stimulating expression in young astrocytes (Bassan et al., 1999), an inhibition was demonstrated in older cultures suggesting that VIP, PACAP, and the VPAC2 receptor may all contribute to the regulation of ADNP gene expression in the developing astrocyte (Zusev and Gozes, 2004).

Parallel studies by the discoverer of PACAP, the late Akira Arimura, showed that when PACAP38 was added to mouse neuroglial cultures, it induced ADNP mRNA expression in a bimodal fashion at subpico- and nanomolar concentrations with greater response at subpicomolar level. The response was attenuated by a PAC1 receptor antagonist at both concentrations and by a VPAC1 receptor antagonist at nanomolar concentration only. An IP3/PLC inhibitor attenuated the response at both concentrations of PACAP38, but a MAPK inhibitor had no effect. A PKA inhibitor suppressed the response at nanomolar concentration only. The findings suggest that ADNP expression is mediated through multiple receptors and signaling pathways that are regulated by different concentrations of PACAP (Nakamachi et al., 2006) as well as VIP. A further study showed that ADNP-immunoreactive cells in the cerebral cortex were multi-polar-shaped and co-immunostained with the astrocyte marker, glial fibrillary acidic protein (GFAP). ADNP-immunoreactive cells in the cerebellum were found to surround Purkinje cells and showed GFAP immunoreactivity. In contrast, ADNPimmunoreactive cells in the hippocampus and septum were round in shape and co-immunostained with neuron-specific enolase (NSE). Importantly, all of the ADNP-immunopositive cells co-localized with PAC1 immunoreactivity. The observations suggest that ADNP is expressed in both astrocytes and neurons, and that ADNP expression may be regulated in part by PACAP (Nakamachi et al., 2008).

Using bioinformatics analysis, we have identified an ADNP family member which we have named, ADNP2 (Zamostiano et al., 2001; Kushnir et al., 2008).

To evaluate the impact of VIP expression *in vivo* on ADNP and the related protein ADNP2, we examined gene expression in adult wild-type (VIP+/+) and VIP null (VIP−/−) offspring of VIP deficient mothers (VIP+/−) comparing them to wild-type offspring of wild-type mothers. Quantitative real time polymerase chain reaction (PCR) revealed regionally specific reductions of ADNP mRNA in the brains of VIP−/− mice compared with the brains of wild-type offspring of a wild-type mother. ADNP was significantly reduced in the cortex and hypothalamus of VIP−/− mice, but not in the hippocampus or thalamus. ADNP2 exhibited a similar pattern but reached a statistically significant reduction only in the hypothalamus. The RNA transcripts for ADNP and ADNP2 also tended to be reduced in the cortex and hippocampus of the wild-type littermates of the VIP−/− mice, indicating that the VIP genotype of the mother may have had an impact on the ADNP expression of her offspring, regardless of their own VIP genotype. Thus, VIP regulates brain ADNP expression in a regionally specific manner and both maternal and offspring VIP genotype may influence ADNP expression in the brain (Giladi et al., 2007).

# **CLINICAL IMPLICATIONS**

Recent genetic studies titled: duplications of the neuropeptide receptor gene VPAC2 confer significant risk for schizophrenia implicated the VPAC2 receptor in susceptibility to schizophrenia. Further studies implicated the PAC1 receptor and PACAP in post-traumatic stress disorder as recently reviewed (Harmar et al., 2012).

In this respect, we found deregulation of ADNP/ADNP2 in the postmortem hippocampus of schizophrenia patients (Dresner et al., 2011) and correlated with disease duration.

Other findings associated reduced ADNP expression to the progression of multiple sclerosis and increased proinflammatory markers (Braitch et al., 2009), and potential modulation by the ADNP derived drug candidate davunetide (NAP).

While VIP may present a dual function in the regulation of multiple sclerosis (Abad et al., 2010; Loh et al., 2011), it should be borne in mind that ADNP regulation, while in part associated with VIP and PACAP, is regulated by other control mechanisms. While VIP knockout exhibited a subtle phenotype (e.g., Abad et al., 2010; Loh et al., 2011) and VIP over expression, eventually led to downregulation of activity with learning deficits (Gozes et al., 1993), ADNP knockout is lethal at embryonic time of brain formation, implicating ADNP as crucial for brain formation (Pinhasov et al., 2003; Mandel et al., 2007). Interestingly, VIP is suggested as a growth factor to the developing embryo (Gressens et al., 1993, 1994). Furthermore, partial ADNP deficiency leads to cognitive and social dysfunction (Vulih-Shultzman et al., 2007), similar to partial deficiency in VIP functions (Glowa et al., 1992; Gozes et al., 1993; Hill et al., 2007). This deficiency can be also associated with deficiency at the cellular level, with ADNP knockdown (or partial deficiency) associated with blocked neurite outgrowth (Mandel et al., 2008) and reduced glial neurotrophic milieu (Pascual and Guerri, 2007; Vulih-Shultzman et al., 2007) and with VIP blockade resulting in neuronal damage (Hill et al., 1994) and inhibition of synaptogenesis (Blondel et al., 2000).

In the case of ADNP, davunetide, an eight amino acid peptide snippet of ADNP is in phase 2/3 clinical trials in a severe neurodegeneration, progressive supranuclear palsy (PSP) (Gold et al., 2012).

# **FUTURE PERSPECTIVE**

While this review focused on VIP and ADNP, from my point of view, it should be taken into consideration that we have developed a family of peptide hybrids/fragments and lipophilic VIP analogs, agonists and antagonists (Gozes et al., 1995), with activities ranging from neurtrophism/neurodevelopment neuroprotection (Gozes et al., 1999) to stimulation of sexual/social function and cancer growth inhibition. The precise GPCR involvement in these analogues activities remains to be elucidated (Gourlet et al., 1998; Moody et al., 2002). Interestingly, ADNP expression was shown to be reduced as a consequence of ischemia in tissue culture, and this was inhibited by treatment with the neuroprotective lypophilic VIP analog Stearyl-Norleucine17 VIP(7–28) (SNV) (Sigalov et al., 2000), awaiting future development.

# **ACKNOWLEDGMENTS**

Current support is provided by AMN Foundation, Canadian Friends of Tel Aviv University—Montreal Circle of Friends, Joe and Grace Alter, Barbara and Don Seal, the Oberfeld family, the Adams family and Allon Therapeutics Inc. Illana Gozes is the incumbent of the Professorial Lily and Avraham Gildor Chair for the Investigation of Growth Factors, and the Director of the Adams Super Center for Brain Studies, and the Elton Laboratory at Tel Aviv University.

# **REFERENCES**

Abad, C., Tan, Y. V., Lopez, R., Nobuta, H., Dong, H., Phan, P., et al. (2010). Vasoactive intestinal peptide loss leads to impaired CNS parenchymal T-cell infiltration and resistance to experimental autoimmune encephalomyelitis. *Proc. Natl. Acad. Sci. U.S.A.* 107, 19555–19560.


VIP2/PACAP receptors. *Eur. J. Pharmacol.* 354, 105–111.


Effects of vasoactive intestinal peptide genotype on circadian gene expression in the suprachiasmatic nucleus and peripheral organs. *J. Biol. Rhythms* 26, 200–209.


differentiation through extracellular signalregulated protein kinase and Akt pathways, and protects neurons co-cultured with astrocytes damaged by ethanol. *J. Neurochem.* 103, 557–568.


of the human activity-dependent neuroprotective protein. *J. Biol. Chem.* 276, 708–714.


*Received: 28 August 2012; accepted: 23 October 2012; published online: November 2012. 16*

*Citation: Gozes I (2012) Neuropeptide GPCRs in neuroendocrinology: the case of activity-dependent neuroprotective protein (ADNP). Front. Endocrin. 3:134. doi: 10.3389/fendo.2012.00134*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Gozes. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# NPBWR1 and NPBWR2: implications in energy homeostasis, pain, and emotion

# **Takeshi Sakurai \***

Kanazawa University, Kanazawa, Japan

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Irvine, USA

Kazuhiro Nakamura, Kyoto University, Japan Olivier Civelli, University of California,

**\*Correspondence:**

Takeshi Sakurai, Kanazawa University, 13-1 Takara-machi, Kanazawa 920-8640, Japan. e-mail: takeshi.sakurai@gmail.com

Neuropeptide B/W receptor-1 (NPBWR1) and NPBWR2 had been known as orphan receptors GPR7 and 8, respectively. Endogenous peptide ligands of these receptors, neuropeptide B (NPB) and neuropeptide W (NPW), were identified in 2002 and 2003 (Fujii et al., 2002; Brezillon et al., 2003; Tanaka et al., 2003). These peptides have been implicated in regulation of feeding behavior, energy homeostasis, neuroendocrine function, and modulating inflammatory pain. In addition, strong and discrete expression of their receptors in the extended amygdala and bed nucleus of the stria terminalis suggests a potential role in regulating stress responses, emotion, anxiety, and fear. Recent studies of NPB/NPW using both pharmacological and phenotypic analyses of genetically engineered mice as well as a human study support this hypothesis.

**Keywords: neuropeptide B, neuropeptideW, hypothalamus, limbic system, amygdala, pain, emotions**

# **INTRODUCTION**

Receptors for neuropeptide B (NPB)/neuropeptide W (NPW) were originally identified as orphan receptors GPR7 and GPR8, which share 70% nucleotide identity and 64% amino acid identity with each other. They have relatively high similarities with opioid and somatostatin receptors. GPR8 was not found in the rodent genomes, while GPR7 was highly conserved in both human and rodents (O'Dowd et al., 1995). This suggests the GPR8 gene was relatively recently generated through gene duplication.

In 2002–2003, two endogenous peptide ligands for these receptors were identified and named NPB/NPW (Fujii et al., 2002; Shimomura et al., 2002; Brezillon et al., 2003; Tanaka et al., 2003). Following the deorphaning of these receptors, GPR7 and GPR8 were reclassified by IUPHAR as NPB/W receptor-1 (NPBWR1) and NPB/W receptor-2 (NPBWR2), respectively (Davenport and Singh, 2005).

In rodents, *Npbwr1* mRNA is localized in some discrete brain regions, including the hypothalamus (dorsomedial hypothalamus and suprachiasmatic nucleus), hippocampus, ventral tegmental area (VTA), and extended amygdala (CeA and bed nucleus of the stria terminalis; BNST) (Lee et al., 1999; Tanaka et al., 2003), suggesting this receptor is involved in regulatory mechanisms of homeostasis, reward system, and emotion.

This review discusses recent findings concerning the pharmacology, histology, and phenotypic analysis of genetically engineered mice of the NPB/NPW system.

### **IDENTIFICATION OF NPB AND NPW**

In 2002–2003, three groups independently identified endogenous peptide ligands for GPR7 and GPR8 by so-called "reverse pharmacology" in combination with bioinformatics approaches. To identify the cognate endogenous ligands for GPR7 (NPBWR1) and GPR8 (NPBWR2), Shimomura et al. (2002) expressed these receptors in Chinese Hamster Ovary (CHO) cells and measured the decrease in forskolin-induced cAMP production in these cells as the read-out for receptor activation. While screening the bovine hypothalamic extract fractions, they detected activities that specifically inhibit cAMP production in those cells, and purified the activities. During their purification process, they identified two forms of NPW with different peptide lengths of 23 and 30 amino acid residues, and named them NPW23 and NPW30, respectively.

In a similar manner, three groups (Fujii et al., 2002; Brezillon et al., 2003; Tanaka et al., 2003) independently purified and identified an additional endogenous ligand for NPBW1 and NPBW2. Fujii et al. first screened the Celera database to identify novel secretory peptides and then expressed the cDNAs of the putative secretory peptides to find novel peptide ligands. Subsequent pharmacological studies and purification of the peptide from bovine hypothalamic extracts led to identification of the second ligand for GPR7 and GPR8, which was named NPB due to the unique modification, namely, bromination of the first tryptophan residue. Brezillon et al. (2003) also identified human NPB mRNA using a bioinformatics approach by searching the EST database using the NPW sequence as a query.

Tanaka et al. used a unique melanin-pigment aggregation assay in *Xenopus* melanophore cells expressing GPR7 as the assay system to purify NPB from bovine hypothalamus. Through EST database searches they also identified NPW as a putative paralogous peptide (Tanaka et al., 2003).

#### **STRUCTURES OF NPW AND NPB**

Neuropeptide W and NPB do not display any significant sequence similarity to members of other known peptide families, while sharing a high degree of sequence similarity with each other, constituting a distinct family of peptides (**Figure 1A**) (Fujii et al., 2002; Shimomura et al., 2002; Brezillon et al., 2003; Tanaka et al., 2003).

As mentioned earlier, sequence analysis of purified peptides showed that NPW has two isoforms with lengths of 23 and 30 amino acid residues, i.e., neuropeptide W23 (NPW23) and

neuropeptide W30 (NPW30), respectively. NPW23 is produced as a result of proteolytic processing at a pair of arginine residues in the 24th and 25th positions in NPW30 (Fujii et al., 2002; Brezillon et al., 2003; Tanaka et al., 2003). *In vitro* experiments using cells expressing these receptors showed that synthetic NPW23 activates and binds to both NPBWR1 and NPBWR2 at similar effective doses, while NPW30 shows slightly lower affinity to both receptors as compared with NPW23 (**Figure 1B**) (Brezillon et al., 2003; Tanaka et al., 2003).

As mentioned earlier, NPB has a unique modification at the N-terminus tryptophan residue, C-6-bromination (Fujii et al., 2002; Tanaka et al., 2003). While this represents the first evidence of bromination of the protein in mammals, the biological significance of this modification is unclear, as it has been demonstrated that des-bromo-NPB is equipotent to brominated NPB in *in vitro* cAMP inhibition assays (Tanaka et al., 2003). By analogy with NPW, Brezillon et al. (2003) predicted that two isoforms of NPB, NPB23 and NPB29, could be produced from the processing of a 125-amino acid human precursor through the alternative usage of a dibasic amino acid pair. However, the dibasic motif, Arg24–Arg25, which is seen in human NPB, does not exist in NPB of other mammalian species including bovine, rat, and mouse. Therefore, it is unlikely that the NPB23 isoform exists as a mature peptide in non-human species. Consistently, both

Fujii et al. and Tanaka et al. were only able to isolate NPB29 from bovine hypothalamus extracts during their purification procedures.

In *X. laevis* melanophore pigment aggregation assay, NPB29 binds and activates human NPBWR1 or NPBWR2 with median effective concentrations (EC50) of 0.23 and 15.8 nM, respectively (Tanaka et al., 2003), suggesting that NPB is a relatively selective agonist for NPBWR1 (**Figure 1B**).

The preferred conformations of des-bromo-NPB and -NPW have been determined by a combination of <sup>1</sup>H NMR, CD, and molecular modeling (Lucyk and Miskolzie, 2005).

#### **STRUCTURE-ACTIVITY RELATIONSHIPS OF NPW AND NPB**

The rank order of potency of NPB and NPW isoforms has been determined in cell lines expressing NPBWR1 and NPBWR2. Both NPW and NPB bind and activate NPBWR1 and NPBWR2 receptors with varying degrees of affinity. NPBWR1 has slightly higher affinity for NPB as compared with both forms of NPW, whereas NPBWR2 shows a potency rank order of NPW23 > NPW30 > NPB (**Figure 1B**) (Fujii et al., 2002; Shimomura et al., 2002; Brezillon et al., 2003; Tanaka et al., 2003). NPB23 and NPB29 have similar affinity for both receptors. NPW23 and NPW30 bind to both receptors almost equally, while NPB acts as a relatively selective agonist for NPBWR1.

Tanaka et al. (2003) showed that deletion of Trp-1 from NPB or NPW markedly decreased their activity, suggesting that the Nterminus is involved in receptor binding. This is consistent with the fact that NPB and NPW have similarity in their sequences in the N-terminal region.

#### **STRUCTURES AND FUNCTIONS OF NPBWR1 AND NPBWR2**

The human NPBWR1 and NPBWR2 genes are localized on chromosome 10q11.2–121.1 and 20q13.3, respectively. Human NPBWR1 and NPBWR2 have 328 and 333 amino acid residues, respectively, and share 64% sequence homology with each other (**Figure 2**). Among other family members of GPCRs, NPBW1 and NPBW2 are most closely related to opioid and somatostatin receptors (**Figure 2**) (O'Dowd et al., 1995). Amino acid analysis of NPBW1 orthologs in other mammalian species has revealed a high degree of conservation throughout evolution (Lee et al., 1999). In contrast,while the gene encoding NPBWR2 has been discovered in several mammalian species such as monkey, lemur, bat, shrew, and rabbit, it has not been found in rodent genomes (Lee et al., 1999), suggesting that these two receptors were produced by relatively phylogenetically recent gene duplication (Lee et al., 1999).

As suggested by the assay systems used during their purification process and structural similarities to somatostatin/opioid receptors, both NPBWR1 and NPBWR2 couple to the Gi-class of G-proteins (Tanaka et al., 2003). This suggests that these neuropeptides have inhibitory properties on neurons via activation of GIRK (Kir3) channels. NPB and NPW were also shown to stimulate Erk p42/p44 activity in human adrenocortical carcinomaderived NCI-H295 cells (Andreis et al., 2005). This activation is probably mediated by beta/gamma subunits released from Giproteins (Tim van et al., 1995). Recently, a synthetic low molecular weight antagonist for NPBWR1 has been reported (Anthony Romero et al., 2012).

# **DISTRIBUTION OF NPB/NPW AND NPBWR1/NPBWR2 IN BRAIN AND OTHER TISSUE**

#### **NEUROPEPTIDE B**

*Npb* mRNA was found in several specific regions in the mouse brain such as the paraventricular hypothalamic nucleus (PVN), CA1–CA3 fields of the hippocampus, and several nuclei in the midbrain and brainstem, including the Edinger–Westphal (EW) nucleus, the sensory and motor nuclei of the trigeminal nerve, locus coeruleus (LC), inferior olive, and lateral parabrachial nucleus (**Figure 3**) (Tanaka et al., 2003; Jackson et al., 2006).

Schulz et al. reported that NPB-immunoreactive cell bodies were observed in many regions within the hypothalamus. High levels of *Npbwr1* and *Npb* mRNA expression were also observed in the hypothalamus, including the ventromedial hypothalamic


**FIGURE 3 | Schematic representation of distribution of NPB/NPW and Npbwr1 mRNA in mouse brain coronal sections.** Receptor distribution is shown in the right hemisphere, while ligand distribution is shown in the left. Npb mRNA (red regions) was observed in the hippocampus (CA1, CA2, CA3), lateral habenular nucleus (LHb), paraventricular hypothalamic nucleus, medial parvicellular part (PaMP), Edinger–Westphal (EW) nucleus, motor root of the trigeminal nerve (m5), sensory root of the trigeminal nerve (s5), lateral parabrachial nucleus

alpha part (Sub CA), locus coeruleus (LC), noradrenergic cell group A5 (A5), and inferior olive subnucleus B (IOB) (Tanaka et al., 2003). Npw mRNA (blue regions) was observed in the periaqueductal gray (PAG)

matter, EW nucleus (EW), ventral tegmental area (VTA), and dorsal raphe nucleus (DR). NPBWR1 mRNA (yellow regions) was observed in the claustrum (Cl), dorsal endopiriform nucleus (DEn), bed nucleus of the stria terminals, laterodorsal part (BSTLD), bed nucleus of the stria terminalis, medioventral part (BSTMV), suprachiasmatic nucleus (Sch), magnocellular preoptic nucleus (MCPO), paraventricular hypothalamic nucleus, posterior part (PaPo), dorsomedial hypothalamic nucleus (DM), central amygdala (CeA), CA1 field, hippocampus (CA1), ventral tegmental area (VTA), sensory root trigeminal nerve (Su5), subiculum (S), anterior hypothalamic area, posterior part (AHP), and arcuate hypothalamic nucleus (Arc). Modified from Hondo et al. (2008).

nucleus, dorsomedial hypothalamic nucleus, arcuate nucleus, supraoptic retrochiasmatic nucleus, and in the area ventral to the zona incerta (Schulz et al., 2007). Although *Npb* mRNA was detected in several regions outside the hypothalamus such as the hippocampus and brain stem (Tanaka et al., 2003; Jackson et al., 2006; Schulz et al., 2007), the existence of NPBimmunoreactive neurons in these regions in the brain has not been clearly demonstrated, since no good antibody for NPB is available.

In peripheral tissues, expression of human *Npb* mRNA was detected by RT-PCR in the kidney, uterus, ovary, testis, and placenta, while murine *Npb* mRNA was detected by Northern blot at high levels in the stomach, spinal cord, and testis and at lower levels in the liver and kidney (Brezillon et al., 2003). The biological significance of this expression remains unknown.

#### **NEUROPEPTIDE W**

The existence of NPW in the mouse brain was more confined to several nuclei in the midbrain and brainstem including the EW, VTA, periaqueductal gray (PAG), and dorsal raphe nucleus (DR) (**Figure 3**) (Tanaka et al., 2003; Kitamura et al., 2006). Most of these neurons expressed tyrosine hydroxylase (TH), suggesting that these cells are also catecholaminergic (our unpublished observation). In humans, high levels of *Npw* mRNA were detected in the substantia nigra, and moderate expression levels were detected in the amygdala and hippocampus (Fujii et al., 2002). In peripheral tissues, expression of human *Npw* mRNA was confirmed by RT-PCR in the progenital system, comprising the kidney, testis, uterus, ovary, and placenta, and also in the stomach and respiratory system, while murine *Npw* mRNA was detected by Northern blot at

high levels in the lung and lower levels in the stomach (Brezillon et al., 2003).

Consistent with the mRNA distribution,NPW-immunoreactive (ir) cells were also exclusively detected in the EW, VTA, PAG, and DR in rats (Kitamura et al., 2006). NPW-ir fibers were observed in several brain regions in rats including the lateral septum, bed nucleus of the stria terminalis (BNST), dorsomedial and posterior hypothalamus, CeA, CA1 field of hippocampus, interpeduncular nucleus, inferior colliculus, lateral parabrachial nucleus, facial nucleus, and hypoglossal nucleus. Among these regions, NPW-ir fibers were most abundantly observed in the CeA and BNST, the output nuclei of the extended amygdala. These observations suggest that NPW-producing neurons are exclusively localized to the mid brain, and they project mainly to the limbic system, especially the CeA and BNST (**Figure 3**).

Some reports showed the existence of NPW-ir cell bodies in the PVN in rats and mice (Dun et al., 2003). However, the staining of NPW-like immunoreactivity-positive cells in the PVN is likely to be non-specific staining, for two main reasons (Kitamura et al., 2006). First, PVN immunoreactivity is also observed in *NPW* <sup>−</sup>/<sup>−</sup> mice using many of the commercially available antibodies. Second, *Npw* mRNA is not expressed in the PVN in either mice or rats (Kitamura et al., 2006).

#### **NPBWR1 (GPR7)**

*In situ* hybridization study showed that the CeA and BNST express the highest levels of NPBWR1 mRNA expression in the mouse brain (**Figure 3**) (Tanaka et al., 2003; Singh et al., 2004; Jackson et al., 2006). Other nuclei with high levels of *Npbwr1* are the suprachiasmatic (SCN) and ventral tuberomammillary nuclei of the hypothalamus. Moderate levels of expression are seen in the CA1–CA3 regions of the hippocampus, dorsal endopiriform, dorsal tenia tecta, bed nucleus, and red nucleus. Low level expression of *Npbwr1* is seen in the olfactory bulb, parastrial nucleus, hypothalamus, laterodorsal tegmentum, superior colliculus, LC, and nucleus of the solitary tract.

#### **NPBWR2 (GPR8)**

Because *Npbwr* 2 does not exist in rodent genomes, very limited information about the tissue distribution of NPBWR2 has been available thus far. RT-PCR analysis showed that *Npbwr2* mRNA is strongly expressed in the human amygdala and hippocampus. Lower levels of expression were also detected in the corpus callosum, cerebellum, substantia nigra, and caudate nucleus (Brezillon et al., 2003).

#### **BIOLOGICAL ACTIVITIES OF NPB AND NPW**

Several papers reported pharmacological actions of NPB and NPW *in vivo* (**Table 1**). In this section, I discuss biological activities of these peptides, referring the pharmacological actions and phenotypes of genetically-modified mice on NPB/W systems (**Table 2**).

#### **FEEDING AND ENERGY HOMEOSTASIS**

The NPB/W system was initially thought to be involved in regulation of feeding behavior (Shimomura et al., 2002; Tanaka et al., 2003), and therefore many of the currently available reports have

focused on the roles of NPB/W in this function. The first physiological study on the action of NPW showed acute hyperphagia in male rats when NPW was administered intracerebroventricularly (i.c.v.) (Shimomura et al., 2002; Tanaka et al., 2003). However, Tanaka et al. (2003) showed that the effect of NPB on feeding behavior in mice is very complex. When NPB was i.c.v. injected during the light period, no significant effect of NPB on feeding was observed (Tanaka et al., 2003). In contrast, in the dark period, i.c.v. administration of 3 nmol NPB increased feeding only within the first 2 h. A higher dose of NPB suppressed food intake in this period. After 2 h, both doses of NPB decreased food intake. Because rodents only have NPBWR1 as a receptor for NPB, the biphasic, i.e., earlier orexigenic and later anorexic, effect cannot stem from the different potency rank orders of the two peptides on NPBWR1 and NPBWR2. In fact, we also observed a similar biphasic action of NPW on feeding behavior when administered i.c.v. in mice or rats (our unpublished observation). Mondal et al. (2003) also reported anorexic effects of NPW. These findings suggest a complex role of NPB and NPW in the regulation of food intake.

Interestingly, the anorexic effect of NPB was markedly enhanced when corticotrophin-releasing factor (CRF), a known anorexic peptide, was co-administered (Tanaka et al., 2003). The i.c.v. administration of these two peptides almost completely suppressed food intake over 4 h. The biphasic effects of NPB/W on feeding behavior and the synergistic anorexic effects of NPB and CRF suggest complex roles of these peptides in the regulation of feeding behavior in relation to the stress response. The synergic effect of NPB with CRF in suppression of feeding suggests that this neuropeptide is likely to be implicated in inhibition of feeding under stressful conditions.

Continuous i.c.v. infusion of NPW was reported to suppress feeding and body weight gain over the infusion period (Mondal et al., 2003). Conversely, i.c.v. administration of anti-NPW IgG stimulated feeding, suggesting that endogenous NPW plays an inhibitory role in feeding behavior (Mondal et al., 2003). However, unlike the results of continuous i.c.v. infusion of NPW, bolus intra-PVN microinjection of NPW23 at doses ranging from 0.1 to 3 nmol increased feeding for up to 4 h, and a bolus dose ranging from 0.3 to 3 nmol was reported to increase feeding for up to 24 h (Levine et al., 2005). These observations suggest that the orexigenic versus anorectic effects of NPB/W could result from different sites of action. When these peptides are injected into the lateral ventricle, they might initially act on the PVN, followed by diffusion to other regions implicated in the suppression of feeding. Alternatively, delayed inhibition of feeding by NPB/W might result from the production of other anorectic factors that are stimulated by NPB or NPW.

The i.c.v. administration of NPW also increased body temperature and heat production (Mondal et al., 2003). These effects suggest that endogenous NPB/W might increase energy expenditure, which is consistent with the late-onset obesity seen in male *NPBWR1*−/<sup>−</sup> mice and *NPB*−/<sup>−</sup> mice (Ishii et al., 2003; Kelly et al., 2005).

Male *NPBWR1*−/<sup>−</sup> mice have been shown to develop adultonset obesity that progressively worsens with age and is greatly exacerbated when animals are fed a high-fat diet (Ishii et al., 2003). The obesity is caused by hyperphagia along with decreased


**Table 2 | Summary of behavioral phenotypes of NPBWR1 knockout mice (modified from Nagata-Kuroiwa et al., 2011).**


energy expenditure. *Npbwr1*−/<sup>−</sup> male mice showed decreased hypothalamic *neuropeptide Y* mRNA level and increased *proopiomelanocortin* mRNA level, a set of effects opposite to those evident in *ob/ob* mice. Furthermore, *ob/ob*;*NPBWR1*−/<sup>−</sup> and *Ay/a*;*NPBWR1*−/<sup>−</sup> double mutant male mice had an increased body weight compared with normal *ob/ob* or *Ay/a* male mice, suggesting that the obesity of *NPBWR1*−/<sup>−</sup> mice is independent of leptin and melanocortin signaling. Female mice did not show any significant weight increase or associated metabolic defects. These data suggest a potential role for NPBWR1 in regulating energy homeostasis independent of leptin and melanocortin signaling, in a sexually dimorphic manner (Ishii et al., 2003). Consistent with the phenotype of *Npbwr1*−/<sup>−</sup> mice, *Npb*−/<sup>−</sup> mice were also reported to exhibit mild adult-onset obesity (Kelly et al., 2005).

#### **PAIN REGULATION**

Intracerebroventricular administration of NPB produces analgesia to pain induced by subcutaneous formalin injection in rats (Tanaka et al., 2003). Intrathecal (i.t.) injection of either NPW23 or NPB also decreased agitation behaviors induced by formalin injection into the paw and attenuated the level of mechanical allodynia (Yamamoto et al., 2005). The effects were not antagonized by naloxone, suggesting that this effect is not mediated through the opioid receptor system. While i.t. injection of either NPW23 or NPB did not show any effect in the hot plate test or mechanical nociceptive test, i.t. injection of either NPW23 or NPB significantly suppressed the expression of Fos-like immunoreactivity of the L4–5 spinal dorsal horn induced by formalin injection into the paw. These data suggest that spinally applied NPW/B suppressed the input of nociceptive information to the spinal dorsal horn, producing an analgesic effect on inflammatory pain, but not mechanical or thermal pain (Yamamoto et al., 2005).

Consistent with the pharmacological studies suggesting an analgesic action of NPB on inflammatory pain, *NPB*−/<sup>−</sup> mice exhibited hyperalgesia to inflammatory pain, while they showed normal responses to mechanical or thermal pain (Kelly et al., 2005). These observations suggest that NPB in the brain and/or spinal cord modulates pain in a modality-specific manner. These observations suggest that NPB might play a role in the pathophysiology of allodynia. Unfortunately, however there are no available data about expression of NPW, NPW, or NPBWR1 in the spinal cord or dorsal root ganglia.

A low level of NPBWR1 receptor expression was observed in Schwann cells in both normal human and rat nerves as well as in primary rat Schwann cell cultures. Peripheral nerve samples taken from patients exhibiting inflammatory/immune-mediated neuropathy showed a marked increase of NPBWR1 receptor expression restricted to myelin-forming Schwann cells. Complementary animal models of immune-inflammatory and ligation-induced nerve injury and neuropathic pain similarly exhibited increased myelin-associated expression of NPBWR1 (Zaratin et al., 2005). These observations suggest that NPBWR1 is involved in regulation of inflammatory pain at least in part by modulating Schwann cell function.

#### **NEUROENDOCRINE REGULATION**

Central administration of NPB or NPW elevated the plasma corticosterone level in rats (Samson et al., 2004; Taylor et al., 2005). NPB also increased prolactin and decreased growth hormone levels (Samson et al., 2004). Pretreatment with a polyclonal anti-CRF antiserum or CRF antagonist completely blocked the effect of NPB or NPW to stimulate ACTH release, and significantly inhibited the effect of NPB/W on plasma corticosterone level (Samson et al., 2004; Taylor et al., 2005). These findings suggest that NPW and NPB play a role in the neuroendocrine response to stress via activation of the hypothalamus-pituitary-adrenal (HPA) axis. Consistent with these observations, whole cell patch-clamp recording from hypothalamic slice preparations showed that bath application of NPW depolarized and directly increased the spike frequency of neuroendocrine PVN neurons (Taylor et al., 2005).

A whole cell patch-clamp study showed that NPW23 exhibited effects on oxytocin, vasopressin, and thyrotrophin-releasing hormone neurons in the PVN, although both depolarizing and hyperpolarizing effects were observed in each of these cell groups (Price et al., 2009). Corticotrophin-releasing hormone cells were unaffected. Further subdivision of chemically phenotyped cell groups into magnocellular, neuroendocrine, or pre-autonomic neurons, using their electrophysiological fingerprints, revealed that neurons projecting to medullary and spinal targets were predominantly inhibited by NPW, whereas those that projected to the median eminence or neural lobe showed almost equivalent numbers of depolarizing and hyperpolarizing cells (Price et al., 2009).

The intracellular mechanisms of these electrophysiological effects of NPW have remained unclear. Since NPBWR1 primarily couples to Gi subclass of G-proteins, activation of this receptor should lead to inhibition of neurons. Depolarizing effect of NPW described in above mentioned report might be indirect effect due to inhibition of inhibitory interneurons of recorded cells.

#### **AUTONOMIC REGULATION**

Intracerebroventricular administration of NPW to rats was reported to increase arterial blood pressure (ABP), heart rate (HR), and plasma catecholamine concentration (Yu et al., 2007). The same report showed that most of the PVN neurons were excited, while a subset of a smaller population of PVN neurons was inhibited by NPW30, although the chemical identities of these neurons were not determined. These observations suggest that NPB/W modulate PVN neuronal activities, which play an important role in regulation of the autonomic nervous system as well as the HPA axis. The expression of NPB mRNA in the PVN suggests that NPB is more likely to be involved in HPA axis regulation. However, *NPBWR1*−/<sup>−</sup> mice have normal blood pressure and HR in a basal state (Nagata-Kuroiwa et al., 2011). The physiological relevance of the NPB/W system in regulation of the autonomic nervous system remains unclear.

#### **SLEEP AND WAKEFULNESS**

Neuropeptide B is reported to induce slow wave sleep in mice when injected intracerebroventricularly (Hirashima et al., 2011). However, since *NPBWR1*−*/*− mice did not show any sleep abnormality, the biological relevance of this finding remains unclear.

#### **EMOTION AND BEHAVIOR**

*Npbwr1* mRNA is abundantly expressed in discrete brain regions in rodents, including the hypothalamus (dorsomedial hypothalamus and suprachiasmatic nucleus), hippocampus, VTA, and extended amygdala (the CeA and bed nucleus of the stria terminalis; BNST) (Lee et al., 1999; Tanaka et al., 2003). These regions are implicated in mood and emotion. Especially, the particularly strong expression of *Npbwr1* in the CeA and BNST, together with the robust projections of NPW-containing axons to these regions (Kitamura et al., 2006), suggest that this receptor might be an important modulator or regulator of the output signal from the extended amygdala, which is a well-defined subcortical nuclear group that is the center of emotion including fear (Phelps and LeDoux, 2005).

A sensory stimulus associated with an aversive outcome will change neural transmission in the amygdala to produce somatic, autonomic, and endocrine signs of fear, as well as increased attention to that stimulus. Fear learning involves the lateral and basolateral amygdala (BLA), where the association between incoming sensory stimuli leads to potentiation of synaptic transmission. The BLA receives sensory information from the thalamus, hippocampus, and cortex, and then activates or modulates synaptic transmission in target areas appropriate for the reinforcement signal with which the sensory information has been associated. The BLA projects to the CeA and BSNT, whose efferents to the hypothalamus and brainstem trigger the expression of fear. Histological and electrophysiological studies have shown that NPBWR1 acts as an inhibitory regulator on a subpopulation of GABAergic neurons in the lateral division of the CeA and terminates stress responses (Nagata-Kuroiwa et al., 2011). These anatomical and functional findings suggest that the NPB/W systems might have important roles in the modulation of output from the extended amygdala.

The role of NPBWR1 in social behavior was recently investigated using *Npbwr1*−/<sup>−</sup> mice (Nagata-Kuroiwa et al., 2011). When presented with an intruder mouse,*Npbwr1*−/<sup>−</sup> mice showed impulsive contact with the strange mice, produced more intense approaches toward them, and had longer contact and chasing time along with greater and sustained elevation of HR and blood pressure compared to wild type mice. *Npbwr1*−/<sup>−</sup> mice also showed increased autonomic and neuroendocrine responses to various physical stresses, suggesting that NPBWR1 might play a role in coping with stress.

Another interesting phenotype of *NPBWR1*−/<sup>−</sup> mice is the impairment of contextual fear conditioning and inversion of "safety conditioning." In the safety conditioning paradigm, mice received unpaired presentations of a CS (tone) and a US (electrical shock). Because the shock never occurs during the CS, mice typically learn to treat the CS as a safety signal, so that fear-related behavior is inhibited during presentation of the CS. *Npbwr1*−/<sup>−</sup> mice show an inversion of this learning in that the safety-trained CS elicits fear rather than a reduction in fear. This observation suggests that *Npbwr1*−/<sup>−</sup> mice undergo trace conditioning rather than safety conditioning in this procedure, in that they associate the CS and US across a long trace interval, while the control mice treat CS and US as unpaired. These results suggest that *Npbwr1*−/<sup>−</sup> mice are more stimulus-bound, meaning that they preferentially attend to discrete stimuli, to the exclusion of more complex, conjunctive stimuli such as context. This hypothesis would also explain the deficit in contextual fear conditioning. To determine the neural mechanisms and the possible developmental and/or extra-amygdalar origins of the phenotype, further investigation using spatially restricted knockout mice and/or genetic rescue of the phenotype of these mice by expressing NPBWR1 in a region-specific manner is needed.

The human *NPBWR1* gene has a frequent single nucleotide polymorphism at nucleotide 404 (SNP rs33977775) in the coding region (404A >T). Importantly, this polymorphism causes an amino acid substitution (Y135F) within the highly conserved DRY motif of G-protein-coupled receptors at the junction of the third transmembrane domain and second intracellular loop, which is known to play an important role in G-protein coupling. Recently, Watanabe et al. (2012) demonstrated that this change partially impairs receptor function as measured by inhibition of cAMP production. Furthermore, this variation affects the emotional response to stimuli showing human faces with four categories of emotional expression (anger, fear, happiness, and neutral). The subject's emotional levels at seeing these faces were rated on scales of hedonic valence, emotional arousal, and dominance (V–A–D). A significant genotype difference was observed in valence evaluation; the 404AT group (position 404 in one allele is replaced by T) perceived the face as more pleasant than did the 404AA group (position 404 in both alleles of *NPBWR1* is A), regardless of the category of facial expression. The 404AT group tended to feel less submissive to an angry face than did the 404AA group. These results suggest that a single nucleotide polymorphism of NPBWR1 seems to affect human behavior in a social context. Future studies using functional brain imaging of human subjects are needed to clarify the mechanisms of the effects of the polymorphisms.

#### **EFFECTS ON CIRCADIAN RHYTHM**

The abundant expression of NPBWR1 in the suprachiasmatic nucleus suggests the possibility that this neuropeptides/receptor system has a role in regulating circadian rhythm (Lee et al., 1999; Tanaka et al., 2003; Singh et al., 2004). However,we did not observe any effects of NPB/W on circadian activity in rats or mice when administered by i.c.v. injection. *Npbwr1*−/<sup>−</sup> mice displayed a normal circadian pattern of behavior in both light-dark and constant dark conditions (Uchio et al., 2009). Both light-entrainable and food-entrainable oscillation were also normal in these mice (our unpublished observations). The function of NPBWR1 in the SCN remains unclear thus far.

#### **PERIPHERAL ACTIONS**

Expression of *Npb*, *Npw*, and *Npbwr1* mRNA in both the adrenal cortex and adrenal medulla has been reported (Andreis et al., 2005; Mazzocchi et al., 2005; Hochol et al., 2007). NPB and NPW were shown to pharmacologically stimulate adrenal glucocorticoid secretion by an ACTH-independent mechanism. It was also reported that NPW stimulates *in vitro* aldosterone secretion by enhancing the release of medullary catecholamines, which activate beta-adrenoceptors located on zona glomerulosa cells (Hochol et al., 2007).

Bolus intraperitoneal (i.p.) injection of NPB or NPW increased the plasma levels of parathyroid hormone, corticosterone, and testosterone. NPB was also reported to increase the blood concentration of thyroxine, and NPW was shown to increase ACTH and estradiol levels. These findings suggest that NPB and NPW play a role in regulation of the endocrine system (Hochol et al., 2006).

The existence of NPW in rat gastric antral cells was reported. The level of NPW in the stomach was decreased in fasted animals, while it was increased by re-feeding (Mondal et al., 2003), which is consistent with the notion that NPW may act as a suppressant of feeding. However, we did not observe any effects on feeding in mice when NPB or NPW was intravenously administered, suggesting that peripheral NPB/W has limited, if any, significant role in modulating feeding behavior (our unpublished observations).

Neuropeptide B and NPW are reported to inhibit proliferative activity of rat calvarial osteoblast-like (ROB) cells (Ziolkowska et al., 2009). However, the physiological relevance of this finding is unknown, and requires further investigation.

#### **CONCLUSION**

Neuropeptide B and NPW are likely to be multi-tasking factors. Both *Npb*−/<sup>−</sup> and *Npbwr1*−/<sup>−</sup> mice show late-onset obesity and hyperphagia, suggesting that the endogenous NPB-NPBWR1 pathway negatively regulates feeding behavior and positively regulates energy expenditure (Ishii et al., 2003; Kelly et al., 2005). This notion is further supported by several pharmacological studies that showed that i.c.v. NPB/W increased heat production and sympathetic outflow (Yu et al., 2007). However, the biphasic (early orexigenic followed by anorexic) effects of NPB/NPW suggest complex actions of NPB and NPW in the regulation of food intake.

Many studies have also shown that the NPB/W system is involved in the modulation of inflammatory pain. Consistent with these results, *Npb*−/<sup>−</sup> mice are hypersensitive to inflammatory pain but display no significant difference in chemical or thermal pain. These data together strongly support a physiological role of central NPB in pain regulation, and agonists for NPBWR1 or NPBWR2 might be good candidatesfor analgesic drugsfor chronic inflammatory pain.

#### **REFERENCES**


with bromine as an endogenous ligand for GPR7. *J. Biol. Chem.* 277, 34010–34016.


Finally, the strong, discrete expression of NPBWR1 in the extended amygdala and abundant projections of NPW fibers in these regions suggest that this neuropeptide system has a role in the regulation of fear and anxiety. The CeA and BNST have been implicated in a variety of emotional functions including expression of fear, modulation of memory, and mediation of social communication (Davis and Shi, 1999) (**Figure 3**). Therefore, the expression of NPBWR1 in the CeA and BNST suggests modulatory roles of this receptor in these functions. Recent studies in *Npbwr1*−/<sup>−</sup> mice support this concept (**Table 2**). Furthermore, a recent report suggested that a human polymorphism in the *NPBWR1* gene might affect emotional responses to facial expressions (Watanabe et al., 2012). In daily life, people show various emotional reactions to others depending on their personalities, even in the same situation. NPBWR1 may provide part of the cause for such differences in reactions in social interactions.

#### **ACKNOWLEDGMENTS**

This study was supported in part by a Technology (MEXT) of Japan, and the Cabinet Office, Government of Japan through its "Funding Program for Next Generation World-Leading Researchers."


in human adrenocortical cells, and their endogenous ligands neuropeptides B and w enhance cortisol secretion by activating adenylate cyclaseand phospholipase C-dependent signaling cascades. *J. Clin. Endocrinol. Metab.* 90, 3466–3471.


of a family of endogenous neuropeptide ligands for the G proteincoupled receptors GPR7 and GPR8. *Proc. Natl. Acad. Sci. U.S.A.* 100, 6251–6256.


expressions. *PLoS ONE* 7:e35390. doi:10.1371/journal.pone.0035390


activity of ROB cells. *Int. J. Mol. Med.* 24, 781–787.

**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 30 November 2012; paper pending published: 17 December 2012; accepted: 22 February 2013; published online: 18 March 2013.*

*Citation: Sakurai T (2013) NPBWR1 and NPBWR2: implications in energy homeostasis, pain, and emotion. Front. Endocrinol. 4:23. doi: 10.3389/fendo.2013.00023*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Sakurai. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

**REVIEW ARTICLE** published: 23 November 2012 doi: 10.3389/fendo.2012.00145

# Distribution and protective function of pituitary adenylate cyclase-activating polypeptide in the retina

# *Tomoya Nakamachi 1,2, Attila Matkovits1,3,Tamotsu Seki 1,4 and Seiji Shioda1\**

<sup>1</sup> Department of Anatomy, Showa University School of Medicine, Tokyo, Japan


<sup>4</sup> Department of Ophthalmology, Showa University School of Medicine, Tokyo, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Dora Reglodi, University of Pécs, Hungary

Illana Gozes, Tel Aviv University, Israel *\*Correspondence:*

Seiji Shioda, Department of Anatomy, Showa University School of Medicine, 1-5-8 Hatanodai, Shinagawa-ku, Tokyo 142-8555, Japan. e-mail: shioda@med.showa-u.ac.jp

Pituitary adenylate cyclase-activating polypeptide (PACAP), which is found in 27- or 38 amino acid forms, belongs to theVIP/glucagon/secretin family. PACAP and its three receptor subtypes are expressed in neural tissues, with PACAP known to exert a protective effect against several types of neural damage. The retina is considered to be part of the central nervous system, and retinopathy is a common cause of profound and intractable loss of vision.This review will examine the expression and morphological distribution of PACAP and its receptors in the retina, and will summarize the current state of knowledge regarding the protective effect of PACAP against different kinds of retinal damage, such as that identified in association with diabetes, ultraviolet light, hypoxia, optic nerve transection, and toxins. This article will also address PACAP-mediated protective pathways involving retinal glial cells.

**Keywords: PACAP, PACAP receptor, retina, distribution, protection, rodent, review, knockout mouse**

"fendo-03-00145" — 2012/11/21 — 22:56 — page 1 — #1

# **INTRODUCTION**

Pituitary adenylate cyclase-activating polypeptide (PACAP) is a neuropeptide which was originally isolated from ovine hypothalamus on the basis of its capacity to stimulate adenylate cyclase activity in pituitary cells (Miyata et al., 1989). PACAP exists in two forms containing 27 and 38 amino acid residues, respectively, with PACAP27 sharing 68% sequence homology with that of vasoactive intestinal polypeptide (VIP). PACAP belongs to the VIP/secretin/glucagon family of peptides (Arimura and Shioda, 1995). PACAP38 is the predominantly expressed form in mammals (Arimura et al., 1991). The amino acid sequence of PACAP has been well conserved in vertebrates, implying that PACAP may act as an important neuropeptide (Sherwood et al., 2000). The receptors for PACAP, of which there are three main types, belong to the G protein-coupled receptor family with seven transmembrane domains. The PAC1 receptor (PAC1R) binds PACAP with high affinity and VIP with a much lower affinity, while the VPAC1 and VPAC2 receptors (VPAC1R, VPAC2R) bind VIP and PACAP with similar affinities (Harmar et al., 2012). PACAP is a pleiotropic biological peptide which regulates vasodilation, activates intestinal motility, increases insulin and histamine secretion, and modulates immune responses in peripheral tissues (Vaudry et al., 2009). In the central nervous system (CNS), PACAP acts as a neurotransmitter, neuromodulator, and/or neurotrophic factor (Arimura, 1998; Nakamachi et al., 2011). One of the most important functions of PACAP is that of neuroprotection. In this way, PACAP suppresses neuronal damage against acute brain injuries such as brain ischemia, traumatic brain injury, and spinal cord injury (Farkas et al., 2004; Ohtaki et al., 2008; Tsuchikawa et al., 2012). Moreover, PACAP protects the brain against neurodegenerative disease (Reglodi et al., 2011), reducing the level of neuronal damage. The retina is considered as a specialized neuronal tissue containing different types of neurons and glial cells. Therefore, to reveal any functions of PACAP in this tissue, it is necessary to understand which cell types in the retina express PACAP and PACAP receptors. The purpose of this review is to provide an overview of the distribution of PACAP and its receptors in the retina, and to summarize current knowledge of the protective functions of PACAP in animal models of retinopathy.

#### **EXPRESSION OF PACAP AND ITS RECEPTORS IN RETINAL TISSUE PACAP DISTRIBUTION**

From a morphological perspective, light microscopy immunohistochemistry studies have shown PACAP-like immunoreactivity (-LI) exists in a population of sensory neurons in the rat uvea (Moller et al., 1993; Mulder et al., 1994) and in rabbit trigeminal ganglion cells (Onali and Olianas, 1994). Nerve fibers with PACAP-LI have also been found in the uvea of the rat eye (Moller et al., 1993; Mulder et al., 1994). PACAP-LI in the eye was studied by radio-immunoassay and the highest concentrations were found in the iris sphincter and ciliary body (Onali and Olianas, 1994; Wang et al., 1995). Furthermore, immunohistochemical studies revealed that PACAP-positive nerve fibers were present in the nerve fiber layer (NFL), the ganglion cell layer (GCL), and the inner plexiform layer (IPL). PACAP-positive neuronal cell bodies were also found in amacrine and horizontal cells in the inner nuclear layer (INL). No PACAP-LI was found in photoreceptors in the outer nuclear layer (ONL) or retinal pigmented epithelium (Seki et al., 1997, 1998; Izumi et al., 2000).

At the ultrastructural level, PACAP-LI was found in some amacrine and horizontal cells in the INL. PACAP-LI was detected in the plasma membrane and rough endoplasmic reticulum, and diffusible immunoreactive products were detected in the cytoplasmic matrix of amacrine cells and horizontal cells. Intense PACAP-LI was detected in cell processes of the IPL, GCL, and NFL. In the IPL, PACAP-positive amacrine cell processes make synaptic contacts with retinal ganglion cell (RGC) terminals, as well as amacrine and bipolar cell processes. PACAP-positive amacrine cell processes have also been identified to make synaptic contacts with each other. In the IPL, PACAP-positive presynaptic axon terminals were found to contain several dense granular vesicles (80–100 nm in diameter) and many small clear synaptic vesicles (30–50 nm in diameter). However, precise ultrastructural localization of PACAP-LI in the axon terminals is very difficult to determine with the pre-embedding immunostaining method. Electron microscopy observations of the IPL revealed non-specific immunoreactive products associated with post-synaptic membranes, the outer mitochondrial membrane, and the cytoplasmic matrix of axon terminals (Seki et al., 2000b). PACAP-LI is also expressed in the cell bodies of some amacrine and horizontal cells in the INL and their processes in the IPL, GCL, and NFL in the rat retina, as shown by both light and electron microscopic immunocytochemistry. PACAP-positive axon terminals make synaptic contact with RGC, bipolar cells, amacrine cells, and horizontal cells in the GCL, NFL, and IPL. On the other hand, VIP-positive cells have also been found in the GCL and INL, and their fibers have been found in the IPL (Loren et al., 1980). Both PACAP- and VIP-positive cells and fibers are found in the rat retina but their distributions are quite different from each other. These studies strongly suggest that PACAP and VIP function as neurotransmitters and/or neuromodulators, but the functions of PACAP may be different from those of VIP.

# **PACAP RECEPTOR DISTRIBUTION**

Receptor binding sites for PACAP and VIP, positively coupled to adenylate cyclase, have been previously described in the retina of different mammalian species (D'Agata and Cavallaro, 1998). As to the localization of PAC1R in the rat retina (Nilsson et al., 1994), we have described the distribution and localization of PAC1R and its mRNA in the rat retina by immunohistochemistry and *in situ* hybridization histochemistry (Seki et al., 1997, 2000a). PAC1R-LI was found in the cell bodies and processes of RGC and amacrine cells. No PAC1R-LI was observed in photoreceptors. At the ultrastructural level, PAC1R-LI was detected in the plasma membrane, rough endoplasmic reticulum, and the cytoplasmic matrix of RGCs and amacrine cells in the INL. There are certain areas in which the localization of PACAP does not match that of PAC1R. For example, in the rat brain, PAC1R has been found at very high levels in the olfactory bulb, hippocampus, and cerebellar cortex, where few PACAP-containing neurons are identified (Seki et al., 1997). Reports also suggest that PACAP is a transmitter and/or modulator which regulates RGCs and amacrine cells in the rat retina. Müller cells were difficult to identify in histological observations, but PAC1R-LI was observed in rat primary cultures of Müller cells (Seki et al., 2006a). PACAP and PAC1R distributions in the rodent retina are summarized in **Figure 1**.

# **PACAP AND NEUROPROTECTION**

Protective effect of PACAP on retina and retinal cells against various types of retinopathy animal models and toxic reagent showing below was summarized in **Table 1**.

#### **PACAP AND DIABETES (DIABETIC RETINOPATHY)**

It is well known that diabetes causes numerous health complications in the human body, with one of the most serious consequences of this disease being diabetic retinopathy. The reason for the retinal degeneration is that the retina is unable to adapt to the metabolic changes caused by hyperglycemia. This situation leads to chronic inflammation and microvascular angiopathy (Liu et al., 2008). One of the most important consequences of diabetes is the changes to enzyme activities and altered expression patterns of growth and transcription factors (Seki et al., 2004). Another reason for the degeneration of the retina is related to oxidative stress (Kowluru and Abbas, 2003). Increased production of reactive oxygen species (ROS) may play an important role in the development of diabetic complications as it has been shown that ROS levels are elevated in the diabetic rat retina and in retinal cells incubated in high glucose media (Kowluru and Abbas, 2003; Cui et al., 2006) due to the hyperglycemiainduced impairment of antioxidant defense systems (Kowluru et al., 1996). The progression of diabetic retinopathy is slow, and leads to a decrease in a number of cell types in the retina such as amacrine cells, RGCs, and both types of photoreceptors (rod and cone cells; Holopigian et al., 1997; Gastinger et al., 2006). Müller cells and retinal astrocytes are also affected, becoming reactive within 3 months of the onset of diabetes (Lieth et al., 1998). Increased glial fibrillary acidic protein (GFAP) expression has been reported, particularly in the early phase of the disease (Lieth et al., 1998), which is closely associated with the activation of Müller cells.

Some studies have shown that PACAP treatment could protect the retina against the harmful effects of diabetes. In streptozotocintreated rats, PACAP (100 pmol in 5 μl saline) administered into the vitreous body three times over the course of a week was able to attenuate the decrease of the cell number in the GCL (Szabadfi et al., 2012b). The PACAP treatment was also effective with respect to Müller cells in that the streptozotocin injection increased GFAP immunoreactivity in the retina due to Müller cell activation, but the PACAP treatment was able to significantly decrease the number of GFAP-positive Müller cells compared with untreated animals (Szabadfi et al., 2012b).

The PACAP treatment also increased the expression pattern of PAC1R in the diabetic retina, particularly in tyrosine-hydroxylase (TH)-positive cells (Szabadfi et al., 2012b). The most important TH-positive cells in the retina are the amacrine cells. The morphological or functional degeneration of dopaminergic amacrine cells was observed during the early stage of STZ-induced diabetes (Seki et al., 2004), and it was recently confirmed immunohistochemically that intravitreally injected PACAP is able to protect the amacrine cells from such degeneration (Atlasz et al., 2010b). Diabetes also causes changes in the expression of several apoptotic factors. Three weeks after streptozotocin injection, B-cell lymphoma 2 (Bcl-2) expression was decreased and p53 expression was increased, both of which indicate an elevated level of apoptosis

"fendo-03-00145" — 2012/11/21 — 22:56 — page 2 — #2

of the retinal cells. The intravitreal injection of PACAP (100 μmol in 5 μl of saline) was able to block these changes and restore the Bcl-2 and p53 levels to near control (Giunta et al., 2012). These results suggest that PACAP is able to up-regulate anti-apoptotic pathways and down-regulate pro-apoptotic pathways.

#### **PACAP AND ULTRAVIOLET-A LIGHT-INDUCED RETINOPATHY**

Only one paper has been published concerning the capacity of PACAP to protect against ultraviolet (UV) light-induced retinopathy. Exposure of animals for 45 min to diffuse UV-A irradiation caused cell death especially in the ONL and INL of the retina, and the thickness of the retina decreased in line with the duration of the UV-A treatment. One day after the UV-A treatment the GCL was not affected, but following a second day after illumination a significant decrease of the number of RGCs was observed. The intravitreal administration of PACAP (100 pmol in 5 μl saline) immediately after the irradiation was able to protect the ONL and INL of the retina which are seriously affected by the UV-A light, and the number of the cells in INL,ONL, and GCL was significantly increased as a consequence of the PACAP treatment (Atlasz et al., 2011). This effect of PACAP may have been via a reduction in the level and toxicity of free radicals generated as a consequence of the exposure to the UV-A. PACAP may thus reduce UV-A radiationinduced retinal damage and edema in a manner similar to that in which it reduces ischemia-induced cerebral damage and edema (Ohtaki et al., 2008; Nakamachi et al., 2010).

# **PACAP AND HYPOXIA**

"fendo-03-00145" — 2012/11/21 — 22:56 — page 3 — #3

The retina has an elevated oxygen uptake compared to other tissues, making it one of the most sensitive tissues in the human body to hypoxia. Several diseases can cause retinal ischemia, the most common of which are cardiovascular disorders such as carotid artery stenosis, retinal artery occlusion, diabetic retinopathy, or high intraocular pressure, which can compress the blood vessels of the retina and cause hypoperfusion (Seki et al., 2011). An optimal way to model retinal ischemia in rats is the bilateral common carotid artery occlusion (BCCAO) technique; however the outcome of the operation may depend on the rat strain and technique used (Block et al., 1992; Szabadfi et al., 2010). The permanent ligation of both common carotid arteries leads to a reduction


"fendo-03-00145" — 2012/11/21 — 22:56 — page 4 — #4

 **cells.**

**Table 1** NMDA,

N-methyl-D-aspartate;

 ONL, outer nuclear layer; OPL, outer plexiform layer; PD, postnatal day; SC,

subcutaneous

 injection; UV, ultraviolet.

in the cerebral blood flow in rats (Farkas et al., 2005; Lavinsky et al., 2006), which can cause severe retinal damage. All layers of the retina can be affected, with the reduction of the blood flow usually leading to a decrease of the thickness of the retina, especially with respect to the inner and outer plexiform layers (Atlasz et al., 2007; Atlasz et al., 2010b). The photoreceptors can also be damaged, with their outer segments becoming shorter, and alterations to their structure appearing after the BCCAO (Atlasz et al., 2010b).

In recent years, several studies have been published concerning retinoprotective strategies and agents developed to combat retinal ischemia. Because decreased perfusion leads to a range of changes in the retina, such as an altered metabolism of glutamate, increased levels of ROS, mitochondrial failure, and activation of inflammatory mediators, several possibilities to reduce the harmful effects of ischemia have been proposed (Szabadfi et al., 2010). For example, numerous retinoprotective agents or methods have been tested in the last few years against the harmful effects of retinal ischemia; these include VIP, poly ADP-ribose polymerase (PARP) inhibitors, brain-derived neurotrophic factor (BDNF), antioxidants, flavonoids, pre- or post-conditioning, etc. (Atlasz et al., 2010a; Szabadfi et al., 2010). Both *in vitro* and *in vivo* studies have shown PACAP to be one of the best candidates to protect retinal cells and to reduce the effects of ischemia. In an early report, turtle retina fragments were maintained in non-oxygenated Ringer solution for 46 h, with added PACAP38 (0.165 μM) able to protect the horizontal cells against ischemia; after 42 and 46 h, the light response of the cells was significantly higher than responses obtained from control group fragments (Rabl et al., 2002).

Several papers were published thereafter concerning the retinoprotective action of PACAP against the effects of hypoxia. PACAP possibly acts via PAC1R, which is detectable in all layers of the retina and is strongly expressed in the GCL, INL, NFL, and more weakly in the IPL, OPL, and ONL (Seki et al., 1997, 2000a). PACAP (10 pmol in 5 μl saline) intravitreally administered immediately after the BCCAO operation significantly reduced the harmful effects of ischemia compared to sham-operated animals. This protective effect was significantly attenuated by the PACAP38 antagonist, PACAP6-38 (Atlasz et al., 2007). Several cell types in the retina can be damaged by ischemia. A decrease of vesicular glutamate transporter 1 (VGLUT1) transporters causes damage to photoreceptors, bipolar cells, and calcium binding proteins, giving rise to the degeneration of different types of neurons. Moreover, increased GFAP expression is a sign of Müller cell and astrocyte activation. These effects were attenuated by PACAP treatment after the BCCAO (Atlasz et al., 2010b), suggesting that PACAP has a general cytoprotective effect in the retina against hypoxic conditions.

This effect of PACAP38 was confirmed in another study on wild-type and PACAP-null CD1 mice exposed to transient (10 min) BCCAO. Directly after the operation PACAP38 (100 pmol in 3 μl saline) was administered into the vitreous body. The results of the operation and treatment were tested 2 weeks later. The 10-min BCCAO resulted in a thinner retina, with significantly greater damage evident in the PACAP-null animals. In this group all the retinal layers were affected, while in the wild-type

"fendo-03-00145" — 2012/11/21 — 22:56 — page 5 — #5

animal abnormalities were only evident in the INL. Intravitreal PACAP38 treatment significantly attenuated the deleterious effects of BCCAO in both groups. These results suggest that the retina in PACAP-null animals is more sensitive to ischemia compared to that in wild-type mice, and that PACAP treatment is effective against retinal ischemia in both wild-type and PACAP-null animals (Szabadfi et al., 2012a).

Another technique to transiently decrease retinal blood flow is to artificially elicit intraocular hypertension. This is a glaucoma model, where a thin needle is inserted into the anterior chamber of the eye of adult rats and 0.9% saline is injected to temporarily increase the intraocular pressure up to 100 mmHg for 60 min. The result of this procedure is similar to that achieved with BCCAO, in that under these experimental conditions the intraocular pressure is greater than the blood pressure in the vessels of the retina, thereby causing a hypoxic state in the retina. This treatment leads to a decrease in the number of RGCs and in the thickness of the retina, especially the IPL (Seki et al., 2011). PACAP38 treatment was also effective in combating this situation given that, compared with untreated eyeballs; the number of the RGCs was higher in animals treated with PACAP38. The effect of the PACAP was significantly stronger compared to the vehicle-treated animals when 3 μl of PACAP38 solution at concentrations of 10 fM, 10 or 100 pM was administered. This result suggests that the PACAP has a bimodal effect with peaks at 10 fM and 10–100 pM, even though the mechanism accounting for these effects was not the same. In the case of the 10fM PACAP administration, the mitogenactivated protein kinase (MAPK) inhibitor PD-98059 significantly reduced the protective effect of PACAP, but there was no significant difference observed at the other PACAP concentrations tested. Meanwhile, the cAMP antagonist Rp-cAMP significantly decreased the effect of PACAP at all the concentrations tested (Seki et al., 2011). These results suggest that pathways involving MAPK and cAMP play key roles in the protective effect of PACAP against hypoxia. A recent study revealed additional information concerning this phenomenon. Adult rat retinas were examined 5, 30, and 60 min after BCCAO accompanied by injections of 100 nmol/3 μl of PACAP. The results showed that in the absence of any harmful stimulus, the PACAP did not cause any changes compared with the control group. Ischemia itself caused several changes, such as increases in the phosphorylation of Akt, ERK 1/2, JNK, or p38MAPK, particularly at the 30 and 60 min time points following the operation. PACAP treatment caused a significant increase in the phosphorylation of Akt and ERK 1/2 at all time points, and decreased the activity of JNK and p38MAPK (Szabo et al., 2012). Several results were published previously concerning the effect of PACAP on these signal transduction pathways in the retina and other tissues (Dohi et al., 2002; Ohtaki et al., 2006; Racz et al., 2006; Shioda et al., 2006). PACAP has important, but variable, effects on the expression of interleukins and cytokines. The expression of several interleukins such as inter interleukin (IL)-1, intercellular adhesion molecule (ICAM), L-selectin, regulated and normal T cell expressed and secreted (RANTES) etc, was decreased by PACAP, while levels of others such as IL-2, IL-6, IL-10, and tumor necrosis factor (TNF)-α were not changed. In contrast, the levels of vascular endothelial growth factor (VEGF) and thymus chemokine were increased (Szabo et al., 2012).

Not only ischemia, but also hyperoxia cause oxidative stress which could be harmful for cells in the retina. Oxidative stress is one of the most important apoptosis-inducing factors in the human body, especially in the CNS and sensory organs (Vaudry et al., 2002; Racz et al., 2010). Retinal pigment epithelial cells are sensitive to oxidative stress, with hydrogen peroxide- or aldehydeinduced oxidative stress leading to apoptosis of these cells *in vitro* (Kalariya et al., 2008; Kook et al., 2008). Given that retinal pigment epithelial cells express PAC1R and VPAC receptors (Zhang et al., 2005), PACAP could have a cytoprotective effect on these cells. When retinal pigment epithelial cells were treated with 0.25 mM H2O2 and 10 nM PACAP38 for 3 h, the survival of these cells was significantly ameliorated compared with control cells not treated with PACAP. An MAPK inhibitor in this case had no influence on the effect of PACAP38, but inhibition of the phosphoinositide 3 kinase (PI3K)/Akt pathway and PACAP6-38 treatment were able to antagonize this cytoprotective effect (Mester et al., 2011). The p38MAPK, c-Jun N-terminal kinase (JNK), extracellular signalregulated kinase (ERK) 1/2, and Akt pathways were also activated in response to H2O2 treatment, while PACAP38 decreased the level of p38MAPK and pJNK, and increased the activity of the ERK 1/2 and Akt pathways. Other cytokine and signal transduction pathways are modified by H2O2 and PACAP38 treatment. Oxidative stress induced the expression of several apoptosis-inducing factors such as Bad, Bax, Trail, Fas-associated protein with death domain (FADD), Fas, second mitochondrial-derived activator of caspase (SMAC), and several heat-shock proteins (HSP32, HMOX2, and HSP27), as well as p53. Co-treatment with both 100 and 10 nM PACAP38 decreased the activation all of these factors (Fabian et al., 2012).

#### **PACAP AND OPTIC NERVE TRANSECTION**

Optic nerve injury caused by trauma, glaucoma, or neurodegenerative disease can cause apoptotic RGCs death (Quigley et al., 1995), for which, based on the above findings, PACAP could be a good candidate to protect injured RGCs against apoptosis. In one recent study, different concentrations of PACAP38 were administered into the vitreous body of adult rats, and immediately after the injection, the optic nerve was transected (Seki et al., 2008). Fourteen days later the number of the RGCs was found to be significantly decreased in the control (untreated) and PACAP-treated groups due to apoptosis. The PACAP38 treated groups, however, showed increased RGC survival compared with control, particularly with regard to the groups receiving injections of 3 μl of saline containing 10 and 100 pM PACAP.

#### **PACAP AND TOXINS**

Many toxic agents have been identified which can cause severe retinal injury. Some of them, like glutamate, are found in the retinal cells and are necessary in small concentrations for the normal functioning of the retina. Others, like anisomycin and thapsigargin, are not endogenous to the retina and are commonly used in *in vivo* studies to induce retinal injury by direct administration into the vitreous body.

#### *Anisomycin*

The capacity of PACAP to prevent the deleterious effects of this drug was examined on *in vitro* preparations of the retinal neuroblastic layer from newborn rats. This drug inhibits protein synthesis and causes cell death in the neuroblastic layer. It was previously found that increased cAMP levels protect retinal cells against retinal damage induced by protein synthesis inhibition (Rehen et al., 1996). In this way, 1 nM PACAP38 or PACAP27 administered in parallel with anisomycin had a protective effect on the neuroblastic layer (Silveira et al., 2002). PACAP38 exerts its action via the PAC1 receptor, which is expressed in all layers of the neonatal retina, and activates the cAMP/protein kinase A (PKA) pathway, which is essential for this effect. The activation of other PAC1R-activated signal transduction pathways, such as the phospholipase C (PLC) or PI3K pathways, is not necessary for the protective effect of PACAP against anisomycin (Silveira et al., 2002). On the other hand, PACAP6-38 and Maxadilan (a specific PAC1R antagonist) inhibit this neuroprotective effect of PACAP38 (Silveira et al., 2002).

#### *Thapsigargin*

Thapsigargin is a non-competitive inhibitor of the endoplasmic reticulum Ca2+-ATPase. This drug inhibits autophagia, which leads to cell death. In the case of retinal explants of newborn rats, thapsigargin causes apoptosis of the photoreceptors in the ONL (Chiarini et al., 2000). PACAP38 treatment was also effective against thapsigargin-induced damage, with 1 nM PACAP38 administered simultaneously with 10 nM thapsigargin effectively preventing damage to the photoreceptors of the ONL (Silveira et al., 2002).

#### *Kainic acid*

Kainic acid is a glutamate receptor agonist which is able to cause dose-dependent excitotoxicity-related injury to the retina. The intraocular injection of a low dose (6–20 nmol/retina) of this drug causes damage to the amacrine cells, with a 60 nmol/retina dose sufficient to cause degeneration of the bipolar and horizontal cells. Higher doses of kainic acid lead to the disappearance of the inner and outer plexiform layers in the chicken retina, while the photoreceptors and RGCs survived the treatment across the range of doses employed (Morgan and Ingham, 1981). The intravitreal administration of this drug in rats causes excitotoxic injury and cell death in the retina, especially in the INL, IPL, and CGL. PACAP (10 pmol) administered into the vitreous body 2 days before the kainic acid treatment resulted in a significantly lower incidence of cell death in the mentioned layers; however the co-administration of PACAP and kainic acid did not provide any protective effect in the retina (Seki et al., 2006b).

#### *N-methyl-D-aspartate*

"fendo-03-00145" — 2012/11/21 — 22:56 — page 6 — #6

*N*-methyl-D-aspartate (NMDA) is a selective glutamate receptor agonist, which is able to mimic the effect of glutamate. It is however specific to its NMDA receptor only, and it has no effect on other glutamate receptors such as the AMPA or kainite receptors. NMDA treatment induces an excitatory effect on several cell types in the CNS (Parsons et al., 1998; Watkins and Jane, 2006). In the retina, NMDA receptors are expressed by RGCs and amacrine cells (Fletcher et al., 2000), with the overstimulation of NMDA receptors causing retinal excitotoxicity which leads to neuronal cell death in the retina (Sucher et al., 1991), particularly in the

GCL. NMDA is a drug commonly used to artificially create normotensive glaucoma models, and for testing anti-glaucoma drugs (Yamashita et al., 2011). PACAP could be a good candidate to combat NMDA-induced retinal injury because it is able to reduce the harmful effects associated with the intravitreal injection of NMDA. The effect of NMDA treatment was more severe in PACAP-null mice than in wild-type animals, with the decrease in the number of RGCs significantly greater on the first, third, and seventh days after the injection of 40 nmol NMDA in 2 μl saline (Endo et al., 2011). In this way, endogenous PACAP production can be seen to play an important role in the protection of retinal cells against NMDA-induced damage (Endo et al., 2011). PACAP38 co-injected with NMDA significantly increased the survival of RGCs both in wild-type and PACAP-null animals. The most effective concentration of PACAP38 in the case of treatment of the retina of wild-type mice was 100 pM (Endo et al., 2011).

#### *Monosodium glutamate*

Under normal conditions, the amino acid derivate glutamate is one of the most important excitatory neurotransmitters in the CNS. This molecule and its receptors are also present in the retina where they have an essential function. However, glutamate also plays a key role in neurological and retinal diseases, as well as in pathological conditions of the eye (Sucher et al., 1991; Danysz and Parsons, 2002). Monosodium glutamate (MSG) is one of the most commonly used drugs to induce retinal injury in different animal models or *in vitro* studies. In recent years, many papers have been published concerning MSG-induced retinal cytotoxicity, and treatments that are able to protect retinal cells against the harmful effects of increased glutamate levels. Several studies suggest that drugs such as glutamate receptor blockers, hormones, neuropeptides, or pre- and post-conditioning could be effective in treating pathological conditions where the glutamate level is increased (Lombardi and Moroni, 1994; Russo et al., 2008; Fernandez et al., 2009). PACAP could also be an interesting candidate for this given that it has been shown to be retinoprotective in several retinal degeneration models, as mentioned above, and is known to exert a protective effect against glutamate toxicity.

Newborn rat retinal cell primary cultures were exposed for 10 min to 1 mM glutamate which caused a significant decrease in their viability (Shoge et al., 1999). Treatment of the cells with either PACAP38 or PACAP27 was protective in a dose-dependent manner (1 nM to 1 μM) against this glutamate-induced damage, with a maximum protective effect observed at a concentration of 100 nM. After a 10-min treatment with PACAP, the level of PKA and the MAPK activity of the cells were elevated. PACAP6-38 and H-89 (a selective PKA inhibitor) were able to attenuate the positive effects of PACAP27 and PACAP38.

In addition to these in vitro results, some *in vivo* studies have suggested that PACAP is effective against glutamate toxicity. The intravitreal treatment with MSG of newborn rats causes severe degeneration in many layers of the retina. The average thickness of the retina was decreased, the IPL almost disappeared, and the INL and CGL layers seemed to fuse with each other (Tamas et al., 2004; Babai et al., 2005;Atlasz et al., 2009). Simultaneously administered PACAP38 and PACAP27 (100 pmol in 5 μl saline) were able to significantly attenuate the MSG-induced damage to the retina. Although the retina was thinner than that in untreated controls, damage to the affected layers of the retina was not so severe (Tamas et al., 2004; Babai et al., 2006; Atlasz et al., 2007). However, if more than one MSG injection was given into the vitreous body, one PACAP treatment was not enough to provide protection against the repetitive excitotoxic stimuli. The PACAP treatment was successful only in the case where PACAP (100 pmol) was administered at least two times in parallel with the MSG injection (Babai et al., 2005).

An important question concerns the combined and additive effects of different treatments against the excitotoxic effect of the glutamate. An interesting study published in 2010 addressed the effect of enriched environments to combat the effects of MSG. MSG was injected into the vitreous body of newborn rat pups and it was found that animals maintained in a bigger cage with colorful objects exhibited reduced glutamate damage compared with controls maintained under standard conditions. A similar or enhanced result was seen in cases where pups maintained under standard conditions were treated with PACAP (100 pmol diluted in 5 μl saline). However, the protective effect of the enriched environment and PACAP co-injection was not additive (Kiss et al., 2011).

Most studies published in the last few years examined the cytoprotective effect of PACAP from a morphological perspective, but no information is usually given about improvements in retinal function following the treatment of glutamate-induced damage. Until now, only one paper has been published in which the functional effects of MSG and PACAP treatments were measured by electroretinogram (ERG). Newborn rats were treated with MSG (2 mg/g bodyweight) administered subcutaneously on the first, fifth, and ninth days postnatal. PACAP (100 pmol diluted in 5 μl saline) was administered into the vitreous body in half of the pups on the same days. The ERG examination was performed 2 months later, with results showing that the subcutaneous MSG injection attenuated the ERG wave amplitude, and that PACAP treatment was able to significantly improve the functional performance of the damaged retina (Varga et al., 2011).

#### **DIRECT AND INDIRECT PATHWAYS**

"fendo-03-00145" — 2012/11/21 — 22:56 — page 7 — #7

Pituitary adenylate cyclase-activating polypeptide suppresses neuronal activity in the CNS via direct and indirect pathways (Shioda et al., 2006). The direct pathway infers that PACAP affects target neurons expressing the PACAP receptor. As shown above, PAC1R was detected in GCLs and amacrine cells in the retina (**Figure 1**), suggesting that PACAP affects these kinds of neurons directly. On the other hand, PACAP is also expressed by RGCs and amacrine cells (**Figure 1**). It is known that endogenous PACAP has neuroprotective functions in the CNS (Ohtaki et al., 2006; Nakamachi et al., 2010; Reglodi et al., 2012). Indeed, PACAP knockout mouse deteriorate retinal damage in some retinopathy animal models (Endo et al., 2011; Szabadfi et al., 2012a), which implies that endogenous PACAP protects retinal neurons via auto- or paracrine mechanisms. Signaling pathways involving the protective function of PACAP on neurons have been elegantly summarized by Atlasz et al. (2010a).

Nakamachi et al. PACAP in the retina

In the case of indirect pathways, PACAP is known to stimulate the secretion of neuroprotective factors in the CNS (Somogyvari-Vigh and Reglodi, 2004), although little is known about PACAP's indirect mode of action in the retina. IL-6 is recognized as a proinflammatory cytokine, but it acts as a neuroprotectant in the CNS (Loddick et al., 1998; Moidunny et al., 2010), and has been considered as a possible player in PACAP's indirect neuroprotective pathway. PACAP administration significantly increases IL-6 mRNA and protein expression levels in the murine brain, with neurons and astrocytes identified as the source of PACAPinduced IL-6 secretion (Tatsuno et al., 1996; Ohtaki et al., 2006; Nakamachi et al., 2012). In the mouse brain, PACAP-induced neuroprotection is absent in the IL-6 knockout mouse, suggesting that PACAP suppresses neuronal damage via an IL-6-mediated pathway (Ohtaki et al., 2006). PACAP also has the potential to stimulate IL-6 secretion in the retina. The addition of PACAP to primary cultures of rat Müller cells significantly augmented IL-6 levels in the culture medium in a manner that was inhibited by PACAP6-38 treatment (Nakatani et al., 2006; Seki et al., 2006a). Indeed, PAC1R protein was detected in rat primary cultures of Müller cells (Kubrusly et al., 2005). These data suggest that IL-6 released from Müller cells may mediate PACAP-induced retinal protection. Recent reports suggest that Müller cells secrete many types of neurotrophic factors, growth factors, and cytokines, such as BDNF and neurotrophin-3, glial cell-line derived neurotrophic factor (GDNF), neurturin, ciliary neurotrophic factor (CNTF), endothelin-2, leukemia inhibitory factor (LIF), basic fibroblast growth factor (bFGF), and prostaglandin E2 (Bringmann et al., 2009). These factors may be related to the neuroprotective effect of PACAP as well as IL-6. Furthermore, microglial cells exist in the retina, and infiltrating macrophages were identified in the

#### **REFERENCES**


degeneration models in rats. *J. Mol. Neurosci.* 43, 51–57.


retina after injury (Chen et al., 2002; Davies et al., 2006). This microglia/macrophage system has been considered as a regulator of immunity and inflammation after retinal damage (Wraith and Nicholson,2012). Further studyfocusing on retinal glial cells in the mechanism of PACAP neuroprotection will provide new insights of the protective network involving PACAP.

# **CONCLUSION AND FUTURE PERSPECTIVES**

In the clinical setting, retinal neuropathies are a major cause of visual impairment that are yet to have a definitive cure. Alternative therapeutic strategies for addressing optic neuropathies are therefore required. Neurotrophic factors such as BDNF, ciliary neurotrophic factor, and glial cell line-derived neurotrophic factor have been shown to ameliorate RGC damage in an animal model of retinopathy (Johnson et al., 2011). However, the actions of these growth factors have not yet been well elucidated to enable them to be considered as target candidates of neuroprotective drugs to treat retinal neuropathies. In this review, we have shown that PACAP consistently exerts a potent protective effect against neural damage in a broad range of retinal diseases. As future studies involving PACAP will likely shift to obtaining a fuller understanding of the mechanisms underlying such protective functions, an important aspect of this will be to examine pathways involving glial cells. These insights will help in the development of new neuroprotective strategies to treat retinal neuropathies.

# **ACKNOWLEDGMENTS**

This work was supported by Grants-in Aid for Scientific Research (KAKENHI: 23249079, 24592681, 24592680), and by the MEXT-Support Program for the Strategic Research Foundation at Showa University (2008–2012, 2012–16).


"fendo-03-00145" — 2012/11/21 — 22:56 — page 8 — #8

retina. *Cell Death Differ.* 7, 272–281.


neurons from apoptosis by inhibiting JNK/SAPK and p38 signal transduction pathways. *Regul. Pept.* 109, 83–88.


E. (1997). Evidence for photoreceptor changes in patients with diabetic retinopathy. *Invest. Ophthalmol. Vis. Sci.* 38, 2355–2365.


"fendo-03-00145" — 2012/11/21 — 22:56 — page 9 — #9

astrocyte - from neural development to neural repair. *Curr. Pharm. Des.* 17, 973–984.


neurotoxicity of retinal ganglion cells in culture. *Brain Res.* 551, 297–302.


"fendo-03-00145" — 2012/11/21 — 22:56 — page 10 — #10

(2002). Pituitary adenylate cyclaseactivating polypeptide protects rat cerebellar granule neurons against ethanol-induced apoptotic cell death. *Proc. Natl. Acad. Sci. U.S.A.* 99, 6398–6403.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 November 2012; paper pending published: 05 November 2012; accepted: 07 November 2012; published online: 23 November 2012. Citation: Nakamachi T, Matkovits A,*

*Seki T and Shioda S (2012) Distribution and protective function of pituitary adenylate cyclase-activating polypeptide in the retina. Front. Endocrin. 3:145. doi: 10.3389/fendo.2012.00145*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Nakamachi, Matkovits, Seki and Shioda. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

REVIEW ARTICLE published: 10 April 2013 doi: 10.3389/fnins.2013.00028

# Observations on the evolution of the melanocortin receptor gene family: distinctive features of the melanocortin-2 receptor

# **Robert M. Dores\***

Department of Biological Sciences, University of Denver, Denver, CO, USA

#### **Edited by:**

Eric W. Roubos, Radboud University Nijmegen, Netherlands

#### **Reviewed by:**

Dan Larhammar, Uppsala University, Sweden Li Chan, Queen Mary University of London, UK

#### **\*Correspondence:**

Robert M. Dores, Department of Biological Sciences, University of Denver, 2190 E. Iliff, Olin Hall 102, Denver, CO 80210, USA. e-mail: rdores@du.edu

The melanocortin receptors (MCRs) are a gene family in the rhodopsin class of G proteincoupled receptors. Based on the analysis of several metazoan genome databases it appears that the MCRs are only found in chordates. The presence of five genes in the family (i.e., mc1r, mc2r, mc3r, mc4r, mc5r) in representatives of the tetrapods indicates that the gene family is the result of two genome duplication events and one local gene duplication event during the evolution of the chordates. The MCRs are activated by melanocortin ligands (i.e., ACTH, α-MSH, β-MSH, γ-MSH, δ-MSH) which are all derived from the polypeptide hormone/neuropeptide precursor, POMC, and as a result the functional evolution of the MCRs is intimately associated with the co-evolution of POMC endocrine and neuronal circuits. This review will consider the origin of the MCRs, and discuss the evolutionary relationship between MC2R, MC5R, and MC4R. In addition, this review will analyze the functional evolution of the mc2r gene in light of the co-evolution of the MRAP (Melanocortin-2 Receptor Accessory Protein) gene family.

**Keywords: melanocortin receptors,ACTH,** α**-MSH, MRAP, MC2R, MC5R, constructive neutral evolution, evolutionary cell biology**

# **INTRODUCTION**

An analysis of tetrapod (amphibians, reptiles, birds, and mammals) genomes indicates that the melanocortin receptors (MCRs) are a family of five G protein-coupled receptors (GPCRs) genes (i.e., *mc1r, mc2r, mc3r, mc4r, mc5r*) that have been implicated in the mediation of integument pigmentation, appetite regulation, glucocorticoid synthesis, and exocrine gland secretion (Gantz and Fong, 2003; Cone, 2006). A unifying feature of this gene family is that all of the receptors can be activated by one or more of the melanocortin peptides (i.e., ACTH, α-MSH, β-MSH, γ-MSH; Cone,2006) with varying degrees of efficacy. The melanocortin ligands are derived from the precursor protein proopiomelanocortin (POMC; Nakanishi et al., 1979), a member of the opioid/orphanin gene family (Dores et al., 2002). As a result, the functional evolution of MCRs co-evolved with the *POMC* gene. However, the functional evolution of at least some of the MCRs is also tied to the co-evolution of two other gene families; the *Melanocortin-2 Receptor Accessory Protein* (*MRAP*) gene family (Metherell et al., 2005; Hinkle and Sebag, 2009; Webb and Clark, 2010; Liang et al., 2011; Vastermark and Schiöth, 2011), and the *AGRP/ASIP* gene family (Vastermark and Schiöth, 2011). The later polypeptides function as antagonists or "inverse agonists" for several MCRs. The evolution of the *AGRP/ASIP* gene family has recently been reviewed (Vastermark and Schiöth, 2011) and will not be discussed in this review. Instead, this review will consider the origins of the MCRs and POMC, the origin of the melanocortin-2 receptor (MC2R) and the melanocortin-5 receptor (MC5R), and the co-evolution of MC2R and MRAP.

## **THE PHYLOGENETIC DISTRIBUTION OF MELANOCORTIN RECEPTORS AND POMC**

The MCRs are placed in the A-13 family within the rhodopsin class of GPCRs (Horn et al., 2003; Vassilatis et al., 2003). In terms of origin, the MCR gene family appears to be a relatively "recent" addition as compared to other hormone/neuropeptideactivated GPCR gene families such as the vasopressin/oxytocin receptor gene family (Mohr et al., 1996), the CRH receptor gene family (Denver, 2009), or the GnRH receptor gene family (Kah et al., 2006). A search for MCRs in the genome databases of protostomes has not revealed any orthologous genes in these phyla (Vastermark and Schiöth, 2011). In addition among the deuterostomes, it appears that MCR genes also are not present in the genomes of echinoderms, cephalochordates, or urochordates (Vastermark and Schiöth, 2011). However, the presence of *MCR*-related genes in hagfish, lamprey, cartilaginous fish, teleost, and tetrapod genomes (Vastermark and Schiöth, 2011) provides support for the assumption that the MCRs are a chordate gene family.

Coincidentally, orthologous *POMC* genes have been detected in lamprey, cartilaginous fish, teleost, and tetrapod genomes as well (Dores and Baron, 2011). As a result, the proliferation of the paralogous *mcr*-coding genes and the radiation of the paralogous genes in the opioid/orphanin gene family have been influenced by the genome duplication events which have played a critical role in the proliferation of gene families within the various classes of vertebrates (Ohno et al., 1968; Lundin, 1993; Holland et al., 1994).

#### **GENOME DUPLICATION EVENTS AND THE EVOLUTION OF POMC AND MELANOCORTIN RECEPTORS**

The chordates can be divided into three major lineages, the protochordates represented today by the arrow worm, Amphioxus, and tunicates (superclass Cephalochordata), vertebrates lacking a true jaw such as the lampreys and hagfishes (superclass Agnatha) and the jawed vertebrates such as the cartilaginous fishes, the ray-finned fishes, lobe-finned fishes, and the tetrapods (superclass Gnathostoma). These three major lineages in chordate evolution emerged sequentially; that is the ancestral protochordate lineages are most ancient and the ancestral gnathostome lineages are most recent (Carroll, 1988). There is general agreement that during the radiation of the ancestral agnathans two genome duplication events occurred in a lineage which ultimately gave rise to the ancestral gnathostomes (Ohno et al., 1968; Lundin, 1993; Holland et al., 1994). As a result where there may have been a single copy of a particular gene in the ancestral protochordates, there was now the potential for four paralogous copies of this gene in the gnathostomes. To add to the proliferation of paralogous members within a gene family the modern ray-finned fishes (teleosts) have undergone an additional genome duplication event (Meyer and Van de Peer, 2005), and there is evidence for local gene duplication events in many gnathostome gene families. It should also be noted that gene loss has occurred in several of the gnathostome gene families.

An operating assumption of the chordate genome duplication process has been that extant agnathans are 1R, where "R" indicates replication of the entire genome, extant gnathostomes are 2R, and the teleosts are 3R. Schemes based on this operating assumption for the opioid/orphanin gene family and the melanocortin receptor gene family is presented in **Figure 1**. To date neither opioid/orphanin-related genes nor melanocortin receptor-related genes have been detected in the genome of an extant cephalochordate. However, it would be reasonable to propose that the ancestral gene for each gene family emerged in some now extinct protochordate lineage. That said both *POMC*related genes (Heinig et al., 1995; Takahashi et al., 1995) and *Melanocortin Receptor-*related genes (Haitina et al.,2007), and have been characterized from the lamprey genome.

As shown in **Figure 1A**, there are two distinct paralogs of the *POMC* gene in the lamprey genome (*POM* and *POC* genes). These genes encode overlapping yet distinct melanocortin and opioid peptides sequences and are expressed in different regions of the lamprey pituitary (Heinig et al., 1995; Takahashi et al., 1995). Enkephalin-like peptides have also been identified in the CNS of the marine lamprey, *Petromyzon marinus* (Dores and Gorbman, 1990) which would suggest that another opioid precursor is present in the lamprey genome. In this scenario, the *Proenkephalin* and *Prodynorphin* genes are the result of the 2 R event, and these genes are found in all the extant groups of gnathostomes; whereas, the *Proorphanin* gene is viewed as the result of a local gene duplication of the *pomc* gene (Sundstrom et al., 2010) which is predicted to have occurred after the 2R event.

The characterization of two melanocortin receptor genes in the lamprey genome (**Figure 1B**) that are the orthologs of the *MC1R* gene (MCaR) and the *MC4R* gene (MCbR), respectively would be consistent with the assumption that lampreys are 1R organisms

(Haitina et al., 2007). In this scenario the second genome duplication event would result in the MC1R, MC2R, MC3R, and MC4R paralogs in the ancestral gnathostomes. At some later point it is assumed the MC5R paralog emerged as a result of a localized gene duplication of one of the otherMCR paralogs. The origin of MC5R will be discussed in Section "Origin of MC5R and the Speculations on the Relationship between MC5R and MC4R." However, there are some aspects of the schemes presented in **Figure 1** which challenge the status of the lamprey as a 1R organism. For example, while plausible explanations have been made to explain the presence of three opioid coding genes in the lamprey genome, assuming that lampreys are 1R organisms (Dores et al., 2002), perhaps the status of the extant agnathan genomes needs to be reevaluated.

Although the agnathan vertebrates emerged at least 450 million years ago and at their zenith were represented by at least three subclasses and numerous orders (Carroll, 1988), today this superclass has been reduced to two extant subclasses: [Myxini (hagfishes) and Cephalaspidomorphi (lampreys; Nelson, 1994). While the lampreys have been considered a 1R group, recent analyzes of the lamprey genome database (McEwen et al., 2009) have found more members within gene families than would be predicted for a 1R organism. Collectively, these observations have led to the premise that the 2R genome duplication event may have occurred in a group of agnathans that were ancestral to both the lamprey lineage and the ancestral gnathostome lineage (Kuraku et al., 2009; Smith et al., 2013). The ramifications of this hypothesis are reflected in the revised evolutionary trees for opioid/orphanin precursors and for MCRs shown in **Figure 2**.

When considering the radiation of the opioid/orphanin gene family (**Figure 2A**), the assumption that the lampreys are 2R organisms provides a more satisfying explanation for the presence of the two *pomc* paralogous genes in the lamprey genome (i.e., *POM* and *POC*). The presence of distinct POM and POC precursor proteins synthesized in the anterior pituitary and intermediate pituitary, respectively, would appear to be the result of the second genome duplication event followed by divergence of the regulatory regions of the *POM* and *POC* genes (Takahashi and Kawauchi, 2006). This scenario also predicts that *Proenkephalinlike* and *Prodynorphin-like* genes may also be present in the lamprey genome. Since the lamprey genome project is only half completed there is likelihood that these other opioid genes may be present. This scenario also is consistent with the current view of the radiation of the opioid/orphanin genes in the gnathostomes (**Figure 2A**; Sundstrom et al., 2010).

However, regardless of whether the lamprey genome is 1R or 2R, the general organization of POMC has not been radically altered either for the lamprey or the gnathostomes (Vallarino et al., 2012). POMC encodes one copy of a core opioid sequence (YGGF; βendorphin) and at least one copy of a core melanocortin sequence (HFRW). In the lamprey POC sequence the melanocortin ligand is a highly derived form of ACTH (Heinig et al., 1995). In the lamprey POM sequence there are two melanocortin ligands, melanotropin A and melanotropin B, that correspond to β-MSH and α-MSH, respectively (Takahashi et al., 1995). In the POMC sequences of the cartilaginous fishes there are five melanocortin sequences, ACTH, α-MSH, β-MSH, γ-MSH, and a melanocortin sequence unique

to the cartilaginous fishes, δ-MSH (Amemiya et al., 1999). In teleost POMC sequences both the γ-MSH and δ-MSH are absent; whereas, among the tetrapods the γ-MSH sequence is present and there is no equivalent to a δ-MSH sequence (Dores and Lecaude, 2005; Takahashi and Kawauchi, 2006).

It should be noted that the sequence of α-MSH comprises the first 13 amino acids within the ACTH(1–39) sequence. Hence, another feature of the melanocortin network that has been rigorously retained is the differential posttranslational processing of the POMC precursor by the endoproteolytic cleavage enzymes, prohormone convertase 1 (anterior pituitary) and prohormone hormone convertase 2 (intermediate pituitary; Seidah and Chrétien, 1999).

Applying this same scenario to the evolution of the melanocortin receptor genes (**Figure 2B**), the lamprey genome may contain two additional MCRs. Hence the unique sequence motifs in the MC1R ortholog (MCaR) and the MC4R ortholog (MCbR) may be more derived features than ancestral features. Lamprey MCaR has been expressed in heterologous mammalian cells, and an unexpected observation was that this receptor is selective for ACTH-related analogs, and is much less reactive to MSHrelated ligands (Haitina et al., 2007). At present the pharmacology of the lamprey MCbR has not been investigated.

Regardless of whether the lamprey genome is 1R or 2R, the genome of the ancestral gnathostomes should have had at least four paralogous melanocortin receptor genes (i.e., *MC1R*, *MC2R*, *MC3R*, and *MC4R*; **Figure 2B**), and this feature should be evident in the extant members of this subclass (i.e., the cartilaginous fishes, the bony fishes, and the lobe-finned fishes and tetrapods). To date, all five *MCR* paralogs have been found in the several tetrapod genomes that have been analyzed. However, in teleost genomes some deviations from this scheme have been observed. For example the fugu genomes (*Takifugu rubripes* and *Tetraodon nigroviridis*) lack a *MC3R* gene (Klovins et al., 2004a), and the zebrafish genome has an additional *MC5R* paralog (Ringholm et al., 2002). Finally, with respect to the cartilaginous fishes, while three MCR paralogs (*MC1R,MC2R,MC3r)* have been found in the genome of the holocephalan,*Callorhynchus milii* (Vastermark and Schiöth, 2011), and three MCR paralogs (*MC3R, MC4R, MC5R*) have been cloned from the genome of the elasmobranch, *Squalus acanthias*; Ringholm et al., 2003; Klovins et al., 2004b). However, to date all five paralogs have not been characterized from a single cartilaginous fish species. While gene loss could account for the later observation, it should be noted that the *C. milii* genome project has not been completed, and the apparent absence of *mc1r* and *mc2r* from the *S. acanthias* genome may only require a new cloning strategy that takes advantage of the sequence data on the MCR paralogs from the *C. milii* genome project. At this stage it would be reasonable to propose that the ancestral gnathostomes had a minimum of four MCR paralogs (i.e.,*MC1R, MC2R, MC3R, MC4R*), which then begs the question of the origin of the *mc5r* gene.

#### **ORIGIN OF MC5R AND THE SPECULATIONS ON THE RELATIONSHIP BETWEEN MC5R AND MC4R**

When genomes duplicate, paralogous genes will initially be located on distinct homologous chromosomes (Holland et al., 1994).

However, it is appreciated that one or both of the paralogs could be subsequently lost, or that non-homologous chromosomes could fuse resulting in two paralogs on the same chromosome. However, the distribution of paralogous genes on distinct chromosomes has been considered a clear indication that a genome duplication event has occurred (Ohno et al., 1968; Lundin, 1993). For example, an analysis of the human, mouse, chicken, fugu (*Takifugu rubripes*), and zebrafish (*Danio rerio*) genomes revealed that the *MC1R* gene, the *MC2R* gene, the *MC3R* gene, and the *MC4R* gene are all located on different chromosomes (Schiöth et al., 2003; Klovins et al., 2004a). In addition, in all six genomes the *MC5R* gene was located on the same chromosome as the *MC2R* in relatively close proximity. These observations provide support for the hypothesis that the *MC5R* gene was the result of a local duplication of the *MC2R* gene (Klovins et al., 2004a). In this scenario ancestral gnathostomes are viewed as having a *MC5R/MC2R* proto-gene, which gave rise to a distinct *mc2r* gene and a distinct *mc5r* gene on the same chromosome as a result of the local gene duplication event (Baron et al., 2009).

Recently these conclusions on the relationship between the *MC2R* and the *MC5R* gene have been called into question (Vastermark and Schiöth, 2011). The issue is that a comparison of the amino acid sequences of MC2R and MC5R indicate that these two MCRs vary considerably in amino acid identity. However, when the amino acid sequences of MC4R and MC5R are compared, these two receptors share a number of identical residues. Furthermore, in a phylogenetic analysis of human and cartilaginous fish MCRs sequences, the MC4R and the MC5R sequences formed a clade (Vastermark and Schiöth, 2011). Based on these observations, Vastermark and Schiöth (2011) predicted that it is more likely that the *MC5R* gene was the result of a local duplication of the *MC4R* gene.

Are these two interpretations for the origin of the *MC5R* gene mutually exclusive? The *MC4R/MC5R* duplication could have occurred at an ancestral gnathostome locus prior to the divergence of the ancestral cartilaginous fishes and the ancestral bony fishes. In this scenario the *MC5R* locus could have moved to the chromosome carrying the *MC2R* locus in a common ancestor prior to the divergence of the ancestral cartilaginous fishes and the ancestral bony fishes. However, when genes duplicate, either as a result of a local duplication event or as a result of a genome duplication event, the new copies of the ancestral gene will accumulate mutations independent of each other. Furthermore based on selection pressures, these independently evolving genes could retain separate functions of the ancestral gene (subfunctionalization) or become adapted for some new function (neofunctionalization; Force et al., 1998).

**Figure 3** provides an alternative interpretation for the origin of *MC5R* gene that combines the major aspects of the two primary studies (Schiöth et al., 2003; Vastermark and Schiöth, 2011). **Figure 3** is based on the assumption that the *MC4R* gene was the ancestral melanocortin gene. A corollary to this assumption is that MCR paralogs would contain a "MC4R" signature; that is, sets of amino acid motifs derived from the proposed ancestral *MC4R* gene. In this scenario a *MC2R/MC5R* gene in

the ancestral gnathostomes could have undergone a local gene duplication event. This assumption would be consistent with the synteny studies (Schiöth et al., 2003). Following the duplication event, selection pressures may have favored the *MC5R* duplicate gene maintaining the sequence features found in MC4R, while the *MC2R* duplicate gene apparently accumulated mutations and as a result evolved new functional properties not found in any of the other MCR paralogs.

A quick test of the preceding hypothesis would be to take a set of MCR sequences from the same group of organisms, and identify the consensus sequence common to these receptor. In **Figure A1A** in Appendix, the amino acid sequences of five cartilaginous fish MCRs were aligned, and common residues that are found at each position are identified in red. The consensus residues at 287 positions were identified. Interesting, 88% of these residues were present in the MC4R sequence. A pair-wise comparison (**Figure A1B** in Appendix) indicated that the MC4R sequence had the highest sequence identity for the MC5R sequence and the MC3R sequence, respectively. A maximum parsimony analysis of the sequences in **Figure A1** in Appendix indicated that the MC4R, the consensus sequence, and the MC3R and MC5R sequences formed a clade (**Figure A2** in Appendix). While this correlation analysis is suggestive, an analysis of the hagfish genome may be more useful for testing the validity of the hypothesis presented in **Figure 3**.

#### **Table 1 | Human melanocortin ligands.**


The human melanocortin ligand sequences were derived from the sequence of human POMC (accession # CAG46625.1). The HFRW motif highlighted in red is required for the activation of all melanocortin receptor. The KKRRP motif highlighted in blue is required for the activation of the melanocortin-2 receptor (Schwyzer, 1977).

# **EVOLUTION OF MC2R LIGAND SELECTIVITY**

As noted in the Introduction, the ligands for the MCRs are the melanocortins, ACTH, α-MSH, β-MSH, γ-MSH (Gantz and Fong, 2003), and for cartilaginous fishes, δ-MSH (Takahashi and Kawauchi, 2006). The sequences of the melanocortins derived from human POMC are presented in **Table 1**. The proposed origin and the primary sequence variability of vertebrate melanocortins have been reviewed recently (Dores and Baron, 2011). In brief, α-MSH is derived from the first 13 amino acids in the sequence of ACTH via posttranslational processing mechanisms (Eipper and Mains, 1980). It would appear that the γ-MSH and β-MSH are the result of duplications and reinsertions of the α-MSH sequence within the *POMC* gene (Dores et al., 2003); whereas, the δ-MSH sequence appears to be derived from a duplication and reinsertion of the β-MSH sequence in the *POMC* gene of cartilaginous fishes (Amemiya et al., 1999).

Several studies on mammalian MCRs [reviewed by Gantz and Fong (2003)], as well as a study on bird MCRs (Ling et al., 2004), and studies on teleosts (Ringholm et al., 2002;Klovins et al., 2004a) and cartilaginous fish MCRs (Ringholm et al., 2003; Klovins et al., 2004b; Reinick et al., 2012a) indicate that MC1R, MC3R, MC4R, and MC5R can be activated by ACTH or any of the MSH-sized ligands with varying degrees of efficacy. As indicated in **Table 1**, all of the melanocortin ligands have the HFRW motif which is required for activation of all MCRs (Schwyzer, 1977; Mountjoy et al., 1992; Gantz and Fong, 2003). From the perspective of the receptors, Pogozheva et al. (2005) identified critical amino acid positions in transmembrane regions 2, 3, 6, and 7 of human MC4R which are required for activation of that receptor by α-MSH. These residues are conserved in other mammalian MCRs (Pogozheva et al., 2005) and have been found in the sequences of MCRs of non-mammalian tetrapods, amphibian MCRs, teleost MCRs, and lamprey MCRs (Baron et al., 2009; Dores, 2009).

That said, none of the MSH-sized ligands in **Table 1** can activate either teleost or tetrapod MC2R (Schwyzer, 1977; Mountjoy et al., 1992; Gantz and Fong, 2003; Agulleiro et al., 2010; Liang et al., 2011). Since teleost and tetrapod MC2Rs have retained many of the residues associated with the HFRW binding site in MC1R, MC3R, MC4R, and MC5R, it would appear that the HFRW binding site in teleost and tetrapod MC2Rs is masked in some manner. The apparent key to unmasking the HFRW binding site appears to reside in the KKRRP motif in the sequence of ACTH (**Table 1**; Schwyzer, 1977; Costa et al., 2004; Liang et al., 2013a). The KKRRP motif is not present in any of the MSH-sized ligands. In addition, either deletions (Schwyzer, 1977) of this motif, or alanine substitutions (Liang et al., 2013a) within this motif will greatly decrease the potency of the ligand. All of these observations point to a KKRRP binding site in teleost and tetrapod MC2Rs, and raise the question of when MC2R orthologs became exclusively selective for ACTH.

Studies on a MC2R ortholog in the genome of the holocephalan cartilaginous fish, *Callorhynchus milii*, have provided an opportunity to address the latter question (Reinick et al., 2012b). Expression of the *C. milii MC2R* ortholog in CHO cells indicated that this receptor could be activated by either human ACTH(1–24) or NDP-MSH. In addition, stimulation of *C. milii MC2R* transiently transfected CHO cells with spiny dogfish (*Squalus acanthias*) ACTH(1–25), α-MSH, β-MSH, γ-MSH or δ-MSH yielded dose response curves with varying degrees of efficacy (Reinick et al., 2012b). Although the sample size is small, it is possible that other cartilaginous fish MC2R orthologs have similar ligand selectivity properties. If so, then the summary presented in **Figure 4A** would indicate a dichotomy in MC2R ligand selectivity between the cartilaginous fishes and the teleosts and tetrapods. In this scenario it is assumed that the MC2R ortholog in the ancestral gnathostomes could also be activated by ACTH or the MSH-related peptides. Hence, the exclusive selectivity for ACTH would appear to have evolved after the divergence of the ancestral cartilaginous fishes and the ancestral bony fishes.

#### **CO-EVOLUTION OF MC2R AND MRAP; AN EXAMPLE OF CONSTRUCTIVE NEUTRAL EVOLUTION**

Another feature of teleost and tetrapod MC1Rs, MC3Rs, MC4Rs, and MC5Rs is that these receptors can be functionally expressed in heterologous mammalian cell lines such has HEK-293 cells, CHO cells, or COS cells (Rachel et al., 2005; Schiöth et al., 2005). These observations are in sharp contrast to teleost and tetrapod MC2Rs which cannot be functionally expressed in those cells lines unless the cells are co-transfected with accessory protein *MRAP1* cDNA (Hinkle and Sebag, 2009; Agulleiro et al., 2010; Webb and Clark, 2010; Liang et al., 2011). Melanocortin-2 Receptor Accessory Protein 1 (MRAP1) is a transmembrane protein with a single transmembrane domain (Metherell et al., 2005). The features of this accessory protein are discussed in another chapter in this book (Clark and Chan, 2013). For the purposes of this review, the salient features of MRAPs include: (a) there are two *MRAP* paralogous genes (*MRAP1* and *MRAP2*) in the vertebrate genome; (b) MRAP1 and MRAP2 form antiparallel homodimers; and (c) MRAP1 is required for the trafficking of MC2R from the ER to the plasma membrane, and for the functional activation of MC2R at the plasma membrane following the binding of ACTH; and (d) MRAP2 can only facilitate the trafficking of MC2R to the plasma membrane, but has a very weak effect on the functional activation of the receptor at the plasma membrane (Hinkle and Sebag, 2009; Webb and Clark, 2010; Gorrigan et al., 2011). In terms of the phylogeny of the *MRAP* genes, it appears that these genes may be restricted to the lamprey genome and the genomes of gnathostomes (Vastermark and Schiöth, 2011). In addition, the two *MRAP* paralogs are not uniformly distributed in these organisms. As summarized in **Figure 5**, to date only an *MRAP2* gene has been detected in the genome databases for the marine lamprey (*Petromzyon marinus*) and the cartilaginous fish, *Callorhynchus*

**(B)** This summary of the melanocortin-2 receptor interaction with MRAP1 indicates that teleost and tetrapod MC2R orthologs are dependent on MRAP1

> MC2Rs of teleosts and tetrapods are MRAP1 dependent; whereas, the MC2R of at least one species of cartilaginous fish is MRAP independent. This conclusion would suggest that the MC2R

expression of MC2R is dependent on forming a complex with MRAP1.

ortholog of ancestral gnathostomes was also MRAP1 independent. Collectively, these observations point to a number of changes and serendipitous events that have occurred during the evolution of the *MC2R* gene. To understand the functional evolution of the *MC2R* gene it may be easiest to start with the current status of this gene in mammals. In all mammals, MC2R serves as the "ACTH" receptor on cells of the adrenal cortex, and is a critical component of the hypothalamus/pituitary/adrenal axis (HPA;Clark and Cammas, 1996). Activation of MC2R results in the synthesis and release of the glucocorticoid, cortisol, a steroid that influences the normal function of many cell types, and facilitates the body's response to chronic stressors (Engelmann et al., 2004). In humans, mutations in the *MC2R* gene that either effect the trafficking of MC2R from the ER to the plasma membrane or inhibit residues on the receptor responsible for binding ACTH will result in Type 1 Familial Glucocorticoid Deficiency (FGD; Chung et al., 2008). However, the functionality of mammalian MC2Rs is totally dependent on the interaction with MRAPα, one of two splice variants of the

*milii*; whereas, *MRAP1* and *MRAP2* genes have been detected in the genomes of several species of teleost fishes, the chicken (*Gallus gallus*) genome, and the genomes of several mammals (Agulleiro et al., 2010; Liang et al., 2011; Vastermark and Schiöth, 2011). It would appear that a duplication of the *MRAP* gene occurred during the radiation of the ancestral bony fishes (**Figure 5**) resulting in distinct *MRAP1* and *MRAP2* genes. In addition, it is now clear that the functional expression of teleost and tetrapod MC2Rs is dependent on interaction with MRAP1 beginning right after synthesis of MC2R at the rough endoplasmic reticulum (Sebag and Hinkle, 2007).

A *Mc2r* ortholog has been detected in genome of the cartilaginous fish, *C. milli* (Vastermark and Schiöth, 2011). When this *MC2R* ortholog was transiently transfected in CHO cells the receptor could be activated by either ACTH or MSH-sized ligand. Hence, the functional expression of the *C. milli* MC2R ortholog is MRAP1 independent (Reinick et al., 2012b). In addition, the functional expression of the *C. milii* MC2R was not affected, either in a positive or negative manner, following co-expression with either mouse MRAP1, zebrafish MRAP1, or *C. milii* MRAP2. Once again the sample size is small, but as summarized in **Figure 4B**, the

human *MRAP1* gene (Metherell et al., 2005). Hence, mutations to critical regions in the human *MRAP1* gene will result in Type 2 Familial Glucocorticoid Deficiency (Metherell et al., 2005). In the absence of a functional MRAP1, *in vitro* experiments indicate that the MC2R is misfolded (Sebag and Hinkle, 2007) and is tagged for degradation by the ER protein quality control system. For either Type 1 or Type 2 FGD, the congenital defect is potentially life threatening if not treated.

Projecting the preceding observations to the functional activation of non-mammalian tetrapod and teleost MC2Rs, *in vitro* experiments have demonstrated that an amphibian MC2R ortholog (Liang et al., 2011) and teleost MC2R orthologs (Klovins et al., 2004a; Agulleiro et al., 2010; Liang et al., 2011) cannot be functionally expressed unless the MC2R ortholog is expressed in cells derived from a mammalian adrenal cell line (that presumably is expressing an endogenous *MRAP1* gene), or in the case of HEK-293 or CHO cells, the *MC2R* ortholog is co-expressed with a *MRAP1* cDNA. Given these observations, it would be reasonable to predict that either functional mutations in the *MC2R* gene, or mutations to the *MRAP1* gene will have a negative effect on the fitness of non-mammalian tetrapods and teleosts.

Hence, the detection of an MRAP1 independent *MC2R* ortholog in the *C. milii* genome that can be activated by either ACTH or MSH-sized ligands would suggest that the *MC2R* gene present in the ancestral gnathostomes after the 2R genome duplication event (**Figure 3**) was also MRAP independent and perhaps capable of being activated by either ACTH or MSH-sized ligands. This ancestral gene could have been the MC2R/MC5R proto-gene proposed in **Figure 3**. In any event, as shown in **Figure 4C**, following the divergence of the ancestral cartilaginous fishes and the ancestral bony fishes, an interaction between MRAP1 and MC2R in the ancestral bony fish linage must have occurred. Initially this interaction could have been neutral (no apparent advantage for the function of either transmembrane protein). However, as mutations altered the trafficking features of MC2R and the ligand selectivity of MC2R, the pre-adaptation for MC2R and MRAP1 to form a complex at the ER rescued this GPCR that if expressed alone could not function properly. Since both teleost and tetrapod MC2Rs are dependent on the interaction with MRAP1 for functional expression, the interaction must have developed in a common ancestor to both the teleosts and the tetrapods. In this scenario, the interaction between MC2R and MRAP1 would have transitioned over time from a neutral interaction to a functionally dependent interaction with respect to MC2R functionality, and serves as an example of constructive neutral evolution (Stolzfus, 1999).

# **CONCLUSION**

The evolution of the MCRs is intertwined with the co-evolution of the ligand-encoding *POMC* gene, the accessory protein *MRAP* genes, and the inverse agonist *AGRP/ASIP* genes. The presence of five *Melanocortin Receptor* genes in the genomes of tetrapods indicates that the gene family has been shaped by two genome duplication events and one local gene duplication event. Based on these observations, the origin of this gene family may have occurred over 500 MYA prior to the emergence of jawless vertebrates. Synteny studies provide support for the conclusions that the local gene duplication involved the *MC2R* gene and the *MC5R* gene (Schiöth et al., 2003; Klovins et al., 2004a).

Studies on the functional activation of cartilaginous fish MCRs may provide some insights into the properties of the melanocortin receptor genes in the ancestral gnathostomes (Ringholm et al., 2003; Klovins et al., 2004b; Reinick et al., 2012a,b; Liang et al., 2013a). Current studies indicate that orthologs of MC2R, MC3R, MC4R, and MC5R can all be stimulated by ACTH or MSH-sized ligands with varying degrees of efficacy, and none of these receptors apparently requires interaction with an accessory protein to facilitate trafficking to the plasma membrane or activation once at the plasma membrane following a ligand binding event.

Among the descendents of the ancestral bony vertebrates (e.g., teleost and tetrapods) MC1R, MC3R, MC4R, and MC5R have retained the proclivity for stimulation by ACTH or the MSHsized ligands, and none of these receptors requires an interaction with an accessory protein to facilitate trafficking to the plasma membrane. The exception to this generalization is MC2R. These features evolved in this receptor which made the receptor exclusively selective for ACTH, but also dependent on MRAP1 not only for trafficking to the plasma membrane but also for functional activation following an ACTH binding event.

The interaction between MC2R/MRAP1 in teleosts and tetrapods insures the strict signaling selectivity of the hypothalamus/pituitary/adrenal (HPA) axis and the hypothalamus/

#### **REFERENCES**


melanocortin system. *Endocr. Rev.* 27, 736–749.


pituitary/interrenal (HPI) axis. As noted in the Introduction, MCRs are also involved in integument pigmentation, appetite regulation, glucocorticoid synthesis, and exocrine gland secretion (Gantz and Fong, 2003; Cone, 2006). The role of MCRs in these physiological processes have been extensively analyzed in mammals. For non-mammalian vertebrates it is now time to rectify the pharmacology on MCRs with the physiology of these processes in non-mammalian vertebrates. As just one example, do the cartilaginous fishes have a true HPI axis if all cartilaginous fish MCRs can be activated by either ACTH or MSH-sized ligands (Liang et al., 2013b)? What role does receptor dimerization, homo-, or hetero-play in the functionality of MCRs? Is MRAP2 an evolutionary anachronism, or does this accessory protein have a role to play in some melanocortin physiological processes? Although MCRs were characterized nearly 20 years ago, there are still many questions about this gene family that are waiting to be resolved.

#### **ACKNOWLEDGMENTS**

This work was supported by funds provided by the University of Denver.

opioid/orphanin gene family. *Mass. Spectrom. Rev.* 21, 220–243.


(1995). The appearance of proopiomelanocortin early in vertebrate evolution: cloning and sequencing of POMC from a lamprey pituitary cDNA library. *Gen. Comp. Endocrinol.* 99, 137–144.


molecular biology and evolution of the peptide hormones and their receptors. *Vitam. Horm.* 51, 235–266.


Kukkonen, J. P., et al. (2003). Presence of melanocortin (MC4) receptor in spiny dogfish suggests an ancient vertebrate origin of central melanocortin system. *Eur. J. Biochem.* 270, 213–221.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 January 2013; accepted: 19 February 2013; published online: 10 April 2013.*

*Citation: Dores RM (2013) Observations on the evolution of the melanocortin receptor gene family: distinctive features of the melanocortin-2 receptor. Front. Neurosci. 7:28. doi: 10.3389/fnins.2013.00028*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Neuroscience.*

*Copyright © 2013 Dores. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*


values for 100 repetitions are included. The outgroup for the rooted tree was the lamprey MCb receptor (accession number: DQ213060) sequence. eMC1R, elephant shark MC1R; eMC2R, elephant shark MC2R; sMC3R, dogfish MC3R; sMC4R, dogfish MC4R; sMC5R, dogfish MC5R.

# Evolution of vertebrate GnRH receptors from the perspective of a basal vertebrate

# **Stacia A. Sower \*,Wayne A. Decatur, Nerine T. Joseph and Mihael Freamat**

Department of Molecular, Cellular and Biomedical Sciences, Center for Molecular and Comparative Endocrinology, University of New Hampshire, Durham, NH, USA

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Olivier Kah, CNRS UMR 6026, France Jae Young Seong, Korea University, Korea

#### **\*Correspondence:**

Stacia A. Sower, Department of Molecular, Cellular and Biomedical Sciences, Center for Molecular and Comparative Endocrinology, University of New Hampshire, 46 College Road, Durham, NH 03824-3544, USA. e-mail: sasower@unh.edu

This minireview provides the current status on gonadotropin-releasing hormone receptors (GnRH-R) in vertebrates, from the perspective of a basal vertebrate, the sea lamprey, and provides an evolutionary scheme based on the recent advance of whole genome sequencing. In addition, we provide a perspective on the functional divergence and evolution of the receptors. In this review we use the phylogenetic classification of vertebrate GnRH receptors that groups them into three clusters: type I (mammalian and non-mammalian), type II, and type III GnRH receptors. New findings show that the sea lamprey has two type III-like GnRH receptors and an ancestral type GnRH receptor that is more closely related to the type II-like receptors. These two novel GnRH receptors along with lGnRH-R-1 share similar structural features and amino acid motifs common to other known gnathostome type II/III receptors. Recent data analyses of the lamprey genome provide strong evidence that two whole rounds of genome duplication (2R) occurred prior to the gnathostomeagnathan split. Based on our current knowledge, it is proposed that lGnRH-R-1 evolved from an ancestor of the type II receptor following a vertebrate-shared genome duplication and that the two type III receptors resulted from a duplication within lamprey of a gene derived from a lineage shared by many vertebrates.

**Keywords: gonadotropin-releasing hormone receptors, G protein-coupled receptors, evolution, lamprey, basal vertebrate, receptor, hormone, pituitary**

#### **INTRODUCTION**

The study of gonadotropin-releasing hormone (GnRH) receptors in basal and later-evolved vertebrates can provide insight into the evolution and molecular mechanisms of signaling of this receptor family. Numerous full-length GnRH receptor (GnRH-R) sequences have been identified, with more than one receptor isoform identified within a single species. In most vertebrates there are usually two to three forms of GnRH-Rs present (Millar, 2005), although there are fewer GnRH-R genes in mammals compared to protochordates, fish, and amphibians (Morgan and Millar, 2004). To date, there is only a partial understanding of the physiological significance of each receptor type in regards to the spatial expression of GnRH-Rs since more than one receptor type can be expressed in the same tissue.

Of the numerous available forms of GnRH-R nomenclature that have been proposed by several investigators, in this review, we use the phylogenetic classification in which three distinct classes of GnRH-Rs from the vertebrate lineage group into separate clusters: type I (mammalian and non-mammalian), type II, and type III GnRH-Rs (Millar, 2005), with a new slight modification based on partial synteny analysis. Vertebrate type I GnRH-Rs are represented in teleosts, amphibians, reptiles, and avian species and in mammals. The type II GnRH-Rs include receptors from amphibians, reptiles, and mammals; however, type II GnRH-Rs are inactivated in most mammals studied to date (Stewart et al., 2009). Unlike the type II GnRH-Rs, the type III GnRH-Rs include sequences from teleost fish, amphibians, reptiles, and avian species

but do not occur in mammals. New findings show that an ancestral extant vertebrate, the lamprey, has two type III-like GnRH-Rs and an ancestral type GnRH-R that is more closely related to the type II-like GnRH-Rs (Joseph et al., 2012). Based on our current knowledge, it is proposed that lGnRH-R-1 evolved from an ancestor of the type II receptor following a vertebrate-shared genome duplication and that the two type III receptors resulted from a duplication within lamprey of a gene derived from a lineage shared by many vertebrates.

Gonadotropin-releasing hormone action is mediated through high affinity binding with the GnRH receptor (GnRH-R), a rhodopsin-like seven transmembrane G protein-coupled receptor (GPCR). Pituitary GnRH receptors are thought to signal primarily through Gαq/11, resulting in the stimulation of the inositol phosphate (IP) second messenger system; however, Gα<sup>s</sup> activation and cAMP signaling have been reported as well (Arora et al., 1995; Stanislaus et al., 1998; Grosse et al., 2000; Liu et al., 2002; Oh et al., 2005).

Lampreys along with hagfish are the only living representatives of the agnathans, the most ancient class of vertebrates, whose lineage dates back over 550 million years (Sower et al., 2009). Lampreys, which express three hypothalamic peptides of GnRH, lamprey GnRH-I, -II, and -III are important to our understanding of the reproductive endocrinology of the first vertebrates and are likely to have retained key characteristics of the ancestral GnRH and GnRH receptor from which modern GnRH isoforms and GnRH receptors arose, as reviewed (Sower et al., 2009).

# **MOLECULAR EVOLUTION**

As a basal vertebrate, the sea lamprey is well positioned to give us insight into the evolution of the GnRH receptors, particularly in reference to the seminal event of the acquisition of the hypothalamic-pituitary-gonadal axis in vertebrates (Sower et al., 2009). In framing this review, it is important to recognize that there are substantial factors acting in this immensely broad and profound transition that started approximately 550 million years ago with the beginnings of the vertebrate lineages. This shift was facilitated by two rounds of whole-genomic duplication (WGD) between the divergence of the protochordates and the lineage leading to modern vertebrates (Putnam et al., 2008; Van de Peer et al., 2010). Up until very recently, the placement of these two rounds of WGD relative to the split of the agnathans and gnathostomes was in question with the prevailing view placing the second round after the branching of the lamprey from the vertebrate lineage (see for example, Andreakis et al., 2011; Dores, 2011; Parmentier et al., 2011). Recent analysis of the numbers of paralogous gene regions and the pattern of shared synteny represented in the current sea lamprey genome assembly provides new evidence supporting two rounds of WGD occurring prior to the split of the agnathans and gnathostomes (Smith et al., submitted).

Recent identification and phylogenetic analysis of the three GnRH receptors in the sea lamprey (Joseph et al., 2012) in tandem with a reasonable-quality genome assembly (Smith et al., submitted) followed by more in-depth synteny analysis will help to elucidate the overall molecular evolution of the GnRH receptor family. Indeed, a similar approach has been most illuminating in the case of the family of vertebrate GnRHs. Shared synteny analysis of the genome (Decatur et al., 2011; Smith et al., submitted) clarified the ancestry of the genes for the lamprey GnRH peptides

compared to the previous extensive phylogenetic studies that had been done (Guilgur et al., 2006; Kah et al., 2007; Kavanaugh et al., 2008; Okubo and Nagahama, 2008; Tsai and Zhang, 2008; Zhang et al., 2008). The synteny data provide evidence for an alternate view of the evolution of the GnRH peptide family and suggest that all duplication events that generated the different fish and tetrapod GnRH groups likely took place before the split of the ancestral lamprey and gnathostome lineages (Smith et al., submitted). The lamprey GnRH-I and -III, formerly referred to as group IV (Kavanaugh et al., 2008; Zhang et al., 2008) share a more recent common ancestry with GnRH2 and three paralogs (Decatur et al., 2011).

However, in the case of the GnRH receptors themselves, the state of the lamprey genome assembly precludes full analyses of the shared synteny, and genomic structure (**Figure 1**). Although half of the assembly is in scaffolds of 173 kb or longer (Smith et al., submitted), significant portions of the genome remain incompletely assembled and prevents investigations of many of the neuroendocrine genes in lamprey (Decatur et al., 2011). A similar issue was faced by Meyer and colleagues in their work resolving orthology of vertebrate genes (Qiu et al., 2011). These authors concluded that single exon genes are the best candidates to use given the current state of the assembly (Qiu et al., 2011). The genes for the lamprey GnRH peptides possess introns in the coding region except one (Kavanaugh et al., 2008) and the genes themselves encode small precursors (ca. 90 amino acids) that serendipitously fall on scaffolds of much larger size (95 and 302 kb) than those of the GnRH receptors. The open reading frame for the GnRH receptors in lamprey are substantially larger than those of the GnRH precursor and span three exons and two introns, similar to GnRH receptor genes of gnathostomes and even some genes of the protochordate

amphioxus (Tello and Sherwood, 2009). A number of factors, including the high GC-content, vast numbers of repetitive elements, and dramatic genetic rearrangements that occur during somatic cell development, have been suggested as contributing to the challenge of sequencing and assembling a complete version of the lamprey genome (Smith et al., 2009, submitted).

Accepting these limitations, we can use the initial characterization of the lamprey GnRH receptor evolution proposed by Kim et al. (2011) that was based on shared synteny and provide a new current model on GnRH receptor evolution based on the latest, although limited,synteny from the lamprey genome and our recent phylogenetic analysis (**Figure 2**). Maximum likelihood-based phylogenetic analysis showed a substantial degree of similarity (ca. 55% identity) between lamprey GnRH receptor 2 and 3 suggesting that these genes shared a recent ancestor. We further propose that they arose by local tandem or segmental duplication of one of the receptors within the lamprey lineage (Joseph et al., 2012). Based on this information, we expect lamprey GnRH receptors 2 and 3 are on the lineage shared by the zebrafish type III GnRH receptor, whereas lamprey GnRH receptor 1 is most similar to the lineage on which the mammalian type II receptor occurs. Teleost fish have undergone a third round (3R) of WGD. Following 3R, there was retention of some of these paralogs resulting in a greater number of GnRH receptor genes in teleosts (Kim et al., 2011).

#### **FUNCTIONAL DIVERGENCE OF GNRH RECEPTORS**

As a measure of function, we looked at the binding characteristics of the GnRH receptors using GnRH ligands, and as a measure of receptor activation we looked at the signal transduction (Joseph et al., 2012). In an attempt to decipher the evolutionary lineage of specific motifs in terms of binding ability and signal transduction, we compared the three identified lamprey GnRH receptors (lGnRH-R-1, lGnRH-R-2, and lGnRH-R-3) with GnRH receptors of later vertebrates. Given the limited synteny data, we will describe this functional divergence of the GnRH receptors based solely in phylogenetic analysis, and not incorporate the emerging new current model on GnRH receptor evolution (described above). On the premise that lGnRH-R-1 evolved from a common ancestor of the type II GnRH receptor, we propose that identification of key motifs can assist in the elucidation of these motifs/residues in terms of evolutionary stringency. lGnRH-R-2 and lGnRH-R-3 which likely occurred due to a local gene duplication in the lamprey lineage may provide evidence of plasticity in amino acid residue functionality. The table below summarizes the key motifs of binding kinetics and signal transduction through studies on vertebrates in relation to those of an extant agnathan, the lamprey (**Table 1**).

All three GnRH receptors in the lamprey have similar nonmammalian GnRH receptor characteristics (Zhou et al., 1994; Flanagan et al., 1999; Millar et al., 2004). However in terms of ligand binding, there is less conservation of key residues in lGnRH-R-1 compared to the two type III lGnRH receptors. Of the four ligand binding residues considered, two His residues are substituted in lGnRH-R-1 for Asn (Davidson et al., 1996) and Glu (Flanagan et al., 1994), suggesting there is less conservation of key residues in lGnRH-R-1 when compared to the type III lamprey GnRH receptors. However, lGnRH-R-1 remains as the only lamprey receptor displaying an affinity to lGnRH-I, despite all the lamprey receptors coding for Asp in the helix2/7 microdomain,

**FIGURE 2 |Working Hypothesis on the evolution of the GnRH receptor gene family in vertebrates with emphasis on placement of the lamprey genes.** "D" represents duplication events for each of the 3 rounds of whole genome duplication in vertebrate evolution (1R, 2R, and 3R), the third being specific to teleosts. Open rectangles with red X's indicate lost loci. Zebrafish are used here as a representative for teleosts. The half-shaded box for mammalian Type2 indicates that there is no

functional gene product produced in several mammals, including humans. The cloud suggests the ambiguity concerning the relative time between early events, in particular the duration available for resolution of duplicated paralogs between the last common whole genome duplication event and the split of the agnathans from the gnathostome lineage. The dashed lines specify a proposed gene translocation, see Kim et al. (2011) for details and more vertebrates.


**Table 1 | Comparison of characteristics contributing to non-mammalian receptors and motifs pertaining to ligand binding and receptor activation of GnRH-R types.**

Representative sequences of non-mammalian type I GnRH-Rs (mouse and human) and amphibian type I, II, and III GnRH-Rs (bullfrog) were compared with the three identified GnRH receptors in the lamprey (Flanagan et al., 1994, 1999; Zhou et al., 1994, 1995; Arora et al., 1995, 1997; Davidson et al., 1996; Ballesteros et al., 1998; Myburgh et al., 1998; Chung et al., 1999; Fromme et al., 2001; Kitanovic et al., 2001; Millar et al., 2004; Wang et al., 2004; Li et al., 2005; Oh et al., 2005).

enabling binding of configured and non-configured ligands (Zhou et al., 1994; Flanagan et al., 1999). Perhaps there are motifs retained within the ancestral type II-like lamprey GnRH receptor (lGnRH-R-1) that were stringently selected for in terms of binding affinity which would need to be determined by mutation analysis.

In relation to the examined signal transduction residues and motifs known to be responsible for Gαq/11 coupling and cAMP accumulation there are no striking substitutions between the three lamprey GnRH receptors except for the HFRK motif recognized for its attribution for cAMP signal transduction (Oh et al., 2005). lGnRH-R-1 has less substitutions at this motif when compared to the type III lGnRH-Rs and is the only lamprey GnRH receptor able to result in cAMP accumulation when activated by endogenous ligands (Kavanaugh et al., 2008; Joseph et al., 2012). lGnRH-R-1, which is proposed to have arisen from a common ancestor of type II receptors, retained this feature compared to the type III lamprey GnRH receptors that do not have this HRFK-like motif. It is difficult to attribute receptor types to specific signal transduction pathways, although it will be more feasible, when more representative receptors from each vertebrate are examined. However, it may remain that the plasticity in formation of signal transduction complexes drives the retention of specific GnRH receptor types.

Interestingly, given lGnRH-R-1 remains as the only receptor to bind to lGnRH-I (Joseph et al., 2012), investigation of evolutionary stringency of receptor residues/motifs in comparing lGnRH-R-1 (type II-like GnRH receptor) and other type II GnRH receptors suggests a selective co-evolution of cognate ligand/receptor pairing of lGnRH-I and lGnRH-R-1. However, in terms of plasticity in amino acid residue functionality, we surmise that the examined residues of the type III lGnRH receptors are pertinent for receptor functionality.

#### **PHYSIOLOGICAL ROLES OF GnRH RECEPTORS**

Physiological roles of GnRH receptor isoforms are the result of the combined effects of binding and activation of the receptor by one or multiple forms of GnRH at the molecular structural level and the spatio-temporal regulation of its expression in one or multiple tissues of an organism (True and Carroll, 2002). Major mutational events like whole genome duplications as well as local gene duplications generate new input for the evolutionary processes that result in creation of new morphological or functional characteristics. Functional divergence of protein sequences through neo- or sub-functionalization of protein sequences is one of the main mechanisms of the generation of evolutionary novelty. Consequently, the last decade has seen an increase in the number of research studies dedicated to detection of the traces of purifying or adaptive changes in the coding regions of various proteins as well as to understanding of co-evolutionary mechanisms that link receptors to their ligands. We briefly reviewed here the main sequence determinants of ligand binding and signal transduction in GnRH receptors and their conservation from the perspective of sea lamprey isoforms.

In contrast, much less attention has been given to the other source of adaptive potential generated by duplication events, i.e., the non-coding genomic regions. Investigation of GnRH receptors, of their expression pattern in relation to their role in reproductive or non-reproductive physiological processes has generated a wealth of experimental data regarding these aspects in numerous vertebrate taxa. An overview of these data suggests that the GnRH/GnRH receptor system is a very interesting system to be approached from the perspective of the role of gene expression regulation in its evolution, given the relative promiscuity of GnRH receptors and their wide spectrum of tissue expression (Chen and Fernald, 2008). This is however a considerably more difficult endeavor than modeling the coding sequence evolution. At the DNA level, the regions of particular interest may include not only the *cis*-acting control elements of the target gene but also the *cis* elements of the network of *trans*-acting factors that ultimately determine the expression not only of the receptor but also of its GnRH ligands (Fraser, 2011).

The strength of lamprey as an evolutionary model is that no other organism is better positioned to offer clues in respect to the earliest events in the evolution of the neuroendocrine mechanisms in vertebrates. It has evolved directly from the jawless ancestor of all vertebrates, independently from the gnathostome radiation. Its pituitary (hypophysis) is probably a plesiomorphic character in vertebrates and given the surprising overall morphological similarity of this animal with lamprey-like Devonian fossils (Janvier, 2006), it may also reflect the ancestral organization of this gland. Moreover, the experimental evidence accumulated so far on the function of lamprey pituitary is consistent with its involvement in the endocrine control of reproduction (Sower and Kawauchi, 2001; Sower et al., 2009).

As mentioned before, lamprey receptor isoforms are hypothesized to belong to two paralogous lineages shared with the gnathostomes (type II and III). A likely local duplication event within the type III lineage resulted in subsequent divergence of the lamprey-specific GnRH-R 2 and 3 pair of paralogs. Similarly with their orthologs in a majority of the gnathostome lineages, tissue expression of the lamprey receptors show a diverse, frequently overlapping pattern, connected to their three main functional roles in brain, pituitary, and gonadal physiology. The recently found lGnRH-R-2 precursor transcript was detected in the pituitary as well as in a wide variety of non-reproductive tissues. Interestingly, the expression control of the lGnRH-R-3 precursor transcript seems to have sharply diverged from its more recent paralog, its expression was detected only in the brain and eye of male and female lampreys (Joseph et al.,2012). In addition it seems to exhibit a sexually dimorphic expression pattern, being detected in the ovary but not in the testes. The previously described lGnRH-R-1

showed limited expression, its transcripts being detected only in the pituitary and testes of sea lamprey (Silver et al., 2005). A diverse and often overlapping functionality appears to characterize their interaction with lamprey GnRH isoforms as well: the widely expressed lGnRH-R-2 is activated by hypothalamic lGnRH-III and in a lesser extent by ubiquitous lGnRH-II. lGnRH-R-3 from brain and eyes has lGnRH-II as putative main activator and in a lesser extent lGnRH-III. lGnRH-R-1 in pituitary and gonads is promiscuously activated by GnRH-I, -II and -III but the interaction with hypothalamic GnRH-I seems to be highly selective. An interesting aspect of the evolutionary history of the lamprey isoforms is that their functional diversity was initiated by duplication events that are estimated to have taken place on different scales in the genome, i.e., local gene duplication (2 and 3) versus whole genome duplication (2, 3, and 1). Since the local duplications are more likely to change the regulatory environment of a gene and therefore to induce an immediate major shift in gene expression, this might offer the opportunity to compare the noncoding versus coding sequence roles in influencing the fate of paralogs.

# **SUMMARY**

The accumulation of more data regarding the GnRH and GnRH receptors both in mammals as well as in more basal groups of vertebrates has started to unravel a more complex picture of their specificity of expression and ligand-receptor interaction. In turn, this leads to a more nuanced understanding of their physiological roles, both in relation to their central role in reproduction as well as outside the hypothalamic-pituitary system. Increase in the number of isoforms of vertebrate GnRH and GnRH receptors raises the question of their origin and of the evolutionary events and mechanisms that contributed to the situation seen in today's vertebrates.

The constant in the convoluted history of GnRH receptor in vertebrates is the organization of the hypothalamo-pituitary axis. Different vertebrate lineages have solved the question of neural control of pituitary gonadotropin secretion using different pieces of the material generated by successive genome and/or gene duplications.

Understanding of the evolutionary mechanisms that molded the GnRH regulation of reproduction in vertebrates by acting at the level of protein expression requires further knowledge on genetic and epigenetic mechanisms and understanding the change in the endocrine, paracrine, and neural pathways that converge into regulating the release of the ligand from the GnRH neurons and expression of receptors in gonadotropes.

#### **ACKNOWLEDGMENTS**

This research was supported by NSF IOS-0849569. Partial funding was provided by the New Hampshire Agricultural Experiment Station. This is scientific contribution number 2489. We thank members of the Sea Lamprey Genome Consortium especially Drs. Weiming Li and Jeramiah Smith for their contributions in the annotation of the sea lamprey genome that provided key information for our synteny data.

# **REFERENCES**


(GnRH) in vertebrates: identification of a novel GnRH in a basal vertebrate, the sea lamprey. *Endocrinology* 149, 3860–3869.


al. (2004). Position of Pro and Ser near Glu7.32 in the extracellular loop 3 of mammalian and nonmammalian gonadotropin-releasing hormone (GnRH) receptors is a critical determinant for differential ligand selectivity for mammalian GnRH and chicken GnRH-II. *Mol. Endocrinol.* 18, 105–116.


mutation supports helix 2 and helix 7 proximity in the gonadotropinreleasing hormone receptor. *Mol. Pharmacol.* 45, 165–170.

Zhou, W., Rodic, V., Kitanovic, S., Flanagan, C. A., Chi, L., Weinstein, H., et al. (1995). A locus of the gonadotropin-releasing hormone receptor that differentiates agonist and antagonist binding sites. *J. Biol. Chem.* 270, 18853–18857.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 30 August 2012; paper pending published: 24 September 2012; accepted:* *26 October 2012; published online: 19 November 2012.*

*Citation: Sower SA, Decatur WA, Joseph NT and Freamat M (2012) Evolution of vertebrate GnRH receptors from the perspective of a basal vertebrate. Front. Endocrin. 3:140. doi: 10.3389/fendo.2012.00140*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Sower, Decatur, Joseph and Freamat. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Comparative evolutionary histories of kisspeptins and kisspeptin receptors in vertebrates reveal both parallel and divergent features

#### **Jérémy Pasquier <sup>1</sup> , Anne-Gaëlle Lafont <sup>1</sup> , Hervé Tostivint <sup>2</sup> , Hubert Vaudry <sup>3</sup> , Karine Rousseau<sup>1</sup> and Sylvie Dufour <sup>1</sup>\***

<sup>1</sup> Research Unit BOREA, Biology of Aquatic Organisms and Ecosystems, Centre National de la Recherche Scientifique 7208, Institut de Recherche pour le Développement 207, Université Pierre et Marie Curie, Muséum National d'Histoire Naturelle, Paris, France

<sup>2</sup> UMR 7221 CNRS/MNHN Evolution des Régulations Endocriniennes, Muséum National d'Histoire Naturelle, Paris, France

<sup>3</sup> Laboratory of Cellular and Molecular Neuroendocrinology, INSERM U982, European Institute for Peptide Research (IFRMP 23), University of Rouen, Mont-Saint-Aignan, France

#### **Edited by:**

Jae Young Seong, Korea University, South Korea

#### **Reviewed by:**

Kazuyoshi Tsutsui, Waseda University, Japan Robert Dores, University of Minnesota, USA

#### **\*Correspondence:**

Sylvie Dufour, Muséum National d'Histoire Naturelle – UMR BOREA, 7 rue Cuvier, CP32, 75231 Paris Cedex 05, France. e-mail: sylvie.dufour@mnhn.fr

During the past decade, the kisspeptin system has been identified in various vertebrates, leading to the discovery of multiple genes encoding both peptides (Kiss) and receptors (Kissr).The investigation of recently published genomes from species of phylogenetic interest, such as a chondrichthyan, the elephant shark, an early sarcopterygian, the coelacanth, a non-teleost actinopterygian, the spotted gar, and an early teleost, the European eel, allowed us to get new insights into the molecular diversity and evolution of both Kiss and Kissr families. We identified four Kissr in the spotted gar and coelacanth genomes, providing the first evidence of four Kissr genes in vertebrates. We also found three Kiss in the coelacanth and elephant shark genomes revealing two new species, in addition to Xenopus, presenting three Kiss genes. Considering the increasing diversity of kisspeptin system, phylogenetic, and synteny analyses enabled us to clarify both Kiss and Kissr classifications. We also could trace back the evolution of both gene families from the early steps of vertebrate history. Four Kissr and four Kiss paralogs may have arisen via the two whole genome duplication rounds (1R and 2R) in early vertebrates.This would have been followed by multiple independent Kiss and Kissr gene losses in the sarcopterygian and actinopterygian lineages. In particular, no impact of the teleost-specific 3R could be recorded on the numbers of teleost Kissr or Kiss paralogs. The origin of their diversity via 1R and 2R, as well as the subsequent occurrence of multiple gene losses, represent common features of the evolutionary histories of Kiss and Kissr families in vertebrates. In contrast, comparisons also revealed un-matching numbers of Kiss and Kissr genes in some species, as well as a large variability of Kiss/Kissr couples according to species. These discrepancies support independent features of the Kiss and Kissr evolutionary histories across vertebrate radiation.

**Keywords: kisspeptin, kisspeptin receptor, phylogeny, synteny, evolutionary history, spotted gar, coelacanth, European eel**

#### **INTRODUCTION**

As increasing vertebrate genomes have been explored, the understanding of their structure and evolution has progressed in parallel. Indeed, the comparison of their gene organization shed light on various large-scale genomic events that occurred along vertebrate radiation. Among those events, in the early stages of their history, vertebrates experienced two rounds of whole genome duplication (1R and 2R), resulting in fourfold-replicated genomes (Dehal and Boore, 2005; Van de Peer et al., 2010). These two events can be traced through the study of gene families currently presenting up to four paralogs. In addition, the comparison of teleost genomes with other vertebrate genomes revealed a teleost-specific third round of whole genome duplication (3R), resulting in up

to eight paralogs in the same gene family in this lineage (Amores et al., 1998; Meyer and Van de Peer, 2005; Kasahara et al., 2007).

In 1996, kisspeptin was first discovered as an anti-metastatic peptide in human carcinoma (Lee et al., 1996). In 2001, the orphan receptor GPR54 was identified as the cognate receptor of kisspeptin (Kotani et al., 2001; Muir et al., 2001). Two years later, both kisspeptin (Kiss) and its receptor (Kissr) were demonstrated as key players of the reproductive function in mammals (de Roux et al., 2003; Funes et al., 2003; Seminara et al., 2003). They act upstream in the gonadotropic axis mediating gonadotropin releasing hormone (GnRH) and steroid effects on gonadotropin secretion, and are considered as major puberty gatekeepers and reproduction regulators (Pinilla et al., 2012). To date, the kisspeptin system

has been identified in various vertebrate species, leading to the discovery of multiple genes encoding Kiss as well as multiple genes encoding Kissr (Biran et al., 2008; Felip et al., 2009; Kitahashi et al., 2009; Lee et al., 2009).

Concerning *Kiss* and *Kissr* diversity, contrasting situations are found in the different vertebrate phyla. Indeed, in eutherian species, one single gene, named *Kiss1r*, encodes the kisspeptin receptor and one single gene, named *Kiss1*, encodes kisspeptin. In prototherians, such as platypus (*Ornithorhynchus anatinus*), two *Kiss*, and two *Kissr* are present (Lee et al., 2009). To date, in teleosts, two situations have been reported. One *Kiss* gene and one *Kissr* gene are present in some species such as fugu (*Takifugu niphobles*), tetraodon (*Tetraodon nigroviridis*), and stickleback (*Gasterosteus aculeatus*). In contrast, a second *Kiss* as well as a second *Kissr* genes have been characterized in some species including zebrafish (*Danio rerio*; Biran et al., 2008), goldfish (*Carassius auratus*; Li et al., 2009), medaka (*Oryzias latipes*; Lee et al., 2009), and striped bass (*Morone saxatilis*; Zmora et al., 2012). Until recently the maximum number of *Kiss* and *Kissr* genes was found in an amphibian species, the *Xenopus* (*Xenopus tropicalis*), with three paralogs of each gene. On the opposite, in birds (chicken, *Gallus gallus*, zebra finch,*Taeniopygia guttata*, and turkey, *Meleagris gallopavo*) neither *Kiss* nor *Kissr* have been found. So far, in all these cases, a matching number of *Kiss* and *Kissr* genes had been reported, leading to the suggestion of the occurrence of "paired Kiss/Kissr" systems in vertebrates (Kim et al., 2012).

The recent publications of genomes from representative species of key phylogenetic positions makes it possible to revisit the diversity, the origin and the evolution of *Kiss* and *Kissr* in vertebrates. These genomes include a chondrichthyan, the elephant shark (*Callorhinchus milii*; Venkatesh et al., 2007), a representative of early sarcopterygian, the coelacanth (*Latimeria chalumnae*, coelacanth genome project, Broad Institute), a non-teleost actinopterygian, the spotted gar (*Lepisosteus oculatus*; Amores et al., 2011), and an early teleost (elopomorphe), the European eel (*Anguilla anguilla*; Henkel et al., 2012). Gene characterization, phylogenetic, and syntenic analyses allowed us to provide new insights on the respective evolutionary histories of *Kiss* and *Kissr* families. Furthermore, the comparison of proposed *Kiss* and *Kissr* phylogenetic histories highlighted common processes as well as divergent events leading to discuss the existence of conserved Kiss/Kissr couples among the various vertebrate lineages.

# **MATERIALS AND METHODS**

#### **GENOMIC DATABASES**

The following genomic databases were investigated:

,


#### **TBLASTN SEARCH**

The TBLASTN algorithm of the CLC DNA Workbench software (CLC bio, Aarhus, Denmark) was used on the European eel genome database and the elephant shark genome database. The TBLASTN algorithm (search sensitivity: near exact matches) of the *e*!ENSEMBL website<sup>13</sup> was used on the coelacanth and spotted gar genomic databases.

.

#### **GENE PREDICTIONS**

Considering that *Kissr* gene structure as well as coding sequences (CDS) are well conserved among vertebrate species, it was possible to predict the exon and intron sequences for new *Kissr* genes (Pasquier et al., 2012). The splicing junctions were predicted using the empirical nucleotidic splicing signatures, i.e., intron begins with "GT" and ends with "AG." Concerning *Kiss* structures, the fact that their CDS are split on two exons (**Figure 1**), appear to be conserved across vertebrates (for review: Tena-Sempere et al., 2012). However, they are highly variable among species, except for the sequence encoding the Kp(10) localized on the final exon (**Figure 1**). Therefore, only this Kp(10) conserved sequence can be predicted when investigating new genomes. This small sequence was used to identify the open reading frame (ORF) encompassing a part of the putative *Kiss* final exon and a part of the putative intron sequence (**Figure 1**). We therefore focused on this ORF encompassing the sequence encoding Kp(10), in the various genomes. The ORFs of the European eel, the coelacanth, the spotted gar, and the elephant shark, were determined using ORF finder tool of the CLC DNA Workbench software.

#### **SYNTENIC ANALYSES**

The synteny analyses of the eel genomic regions were manually performed using CLC DNA Workbench 6 software and the European eel genome database. The analyses of the spotted gar genomic regions were performed using the preliminary gene annotation of the genome assembly LepOcu1 generated by Ensembl release 67. Synteny maps of the conserved genomic regions in human, platypus, lizard (*Anolis carolinensis*), *Xenopus*, zebrafish, medaka, stickleback, tetraodon, and coelacanth, as well as of the corresponding region in chicken, *G. gallus*, were performed using the PhyloView of Genomicus v67.01 web site<sup>14</sup> (Muffato et al., 2010).

<sup>9</sup>http://pre.ensembl.org/Lepisosteus\_oculatus/Info/Index

<sup>1</sup>http://www.ensembl.org/Gallus\_gallus/Info/Index/

<sup>2</sup>http://www.ensembl.org/Latimeria\_chalumnae/Info/Index/

<sup>3</sup>http://esharkgenome.imcb.a-star.edu.sg/resources.html

<sup>4</sup>http://www.zfgenomics.org/sub/eel

<sup>5</sup>http://www.ensembl.org/Homo\_sapiens/Info/Index/

<sup>6</sup>http://www.ensembl.org/Anolis\_carolinensis/Info/Index/

<sup>7</sup>http://www.ensembl.org/Ornithorhynchus\_anatinus/Info/Index/

<sup>8</sup>http://www.ensembl.org/Petromyzon\_marinus/Info/Index/

<sup>10</sup>http://www.ensembl.org/Gasterosteus\_aculeatus/Info/Index/

<sup>11</sup>http://www.ensembl.org/Xenopus\_tropicalis/Info/Index/

<sup>12</sup>http://www.ensembl.org/Danio\_rerio/Info/Index/

<sup>13</sup>http://www.ensembl.org/index.html

<sup>14</sup>http://www.dyogen.ens.fr/genomicus-67.01/cgi-bin/search.pl

#### **RESULTS AND DISCUSSION**

As one of the aims of this study was to compare the *Kiss* and *Kissr* histories, we first propose to make a short overview of our recent findings concerning the diversity, classification, and origin of *Kissr* gene family. Then, we will expose and discuss our new findings concerning *Kiss* family. Finally, we will compare and discuss the *Kiss* and *Kissr* evolutionary histories in order to get a better understanding of the kisspeptin system evolution.

#### **DIVERSITY AND EVOLUTIONARY HISTORY OF Kissr IN VERTEBRATES Diversity and classification of Kissr**

*New advances in Kissr gene characterization.* Recently, we described three *Kissr* genes in the genome of a basal teleost, the European eel, providing the first evidence of the existence of three *Kissr* genes in a teleost species (Pasquier et al., 2012). Furthermore, we described four *Kissr* in the genome of a non-teleost actinopterygian, the spotted gar, as well as in the genome of a basal sarcopterygian, the coelacanth (Pasquier et al., 2012). This provided the first evidence for four*Kissr* genes in vertebrate species and revealed a larger diversity of *Kissr* than previously described.

So far, no *Kissr* sequence has been reported in chondrichthyans. Our search in the elephant shark genome has only led to the identification of multiple partial sequences, corresponding at least to two *Kissr* (unpublished data). Ongoing sequencing of other chondrichthyan genomes, such as dogfish (*Scyliorhinus canicula*), may provide more insights into the *Kissr* diversity in the sister group of osteichthyans.

#### **Phylogeny, synteny, and classification of Kissr**

Phylogenetic analysis of 51 peptidic Kissr sequences clustered the osteichthyan Kissr into four clades, each one encompassing a coelacanth and spotted gar Kissr (Pasquier et al., 2012). Clade-1 mainly encompassed mammalian Kissr including human Kiss1r, as well as *Xenopus* Kissr-1a, European eel Kissr-1, spotted gar Kissr-1, and coelacanth Kissr-1. Clade-2 mainly encompassed teleost Kissr including European eel Kissr-2, as well as amphibian, spotted gar, and coelacanth Kissr-2. Clade-3 encompassed a few teleost Kissr including European eel Kissr-3 as well as *Xenopus* Kissr1b, spotted gar, and coelacanth Kissr-3. Clade-4 encompassed two early osteichthyan (spotted gar and coelacanth) Kissr-4 and two tetrapod Kissr (lizard and platypus; Pasquier et al., 2012).

Synteny analysis of *Kissr* neighboring genes, performed on 11 representative vertebrate species including the European eel, coelacanth,and spotted gar,fully supported the phylogenetic repartition of Kissr in four clades. Based on this classification, we proposed a new nomenclature of the *Kissr* family (*Kissr-1*,*Kissr-2*,*Kissr-3*, and *Kissr-4*; Pasquier et al., 2012).

#### **EVOLUTIONARY HISTORY OF Kissr Origin of Kissr diversity via 1R and 2R**

Synteny analysis revealed that the four *Kissr* neighboring genomic regions were highly conserved, each presenting paralogs from eight gene families, i.e., *PALM*, *PTBP*, *GRIN3*, *GADD45*, *DIRAS*, *ZCCHC*,*LPAR*,*ZNF644/WIZ* (Pasquier et al., 2012). The hypothesis of the potential existence of four *Kissr* paralogons in vertebrates had been previously raised by Lee et al. (2009) and Kim et al. (2012), although only a maximum of three *Kissr* genes had been discovered at that time. Our finding of four *Kissr* genes, located on four paralogous genomic regions, in coelacanth and spotted gar, provides direct evidence validating this former hypothesis. These four *Kissr* paralogons likely resulted from the two successive genomic duplications (1R and 2R) of a single ancestral genomic region (**Figure 2**).

The currently available data led to a polytomy of the four Kissr clades. This polytomy did not allow to fully solve the homology relationships between the four *Kissr* resulting from the 2R (Pasquier et al., 2012). A recent study proposed the phylogenetic reconstruction of the *PALM* family (Hultqvist et al., 2012). The study of these genes, located in the vicinity of *Kissr* genes, allows us to infer further relationships between the four *Kissr*. We can

thus hypothesize that *Kissr-1* and *Kissr-3*, on one side, and *Kissr-2* and *Kissr-4*, on the other side, could be sister genes resulting from the 2R.

Recently, one study proposed the reconstruction of 10 proto-chromosomes of the ancestral vertebrate karyotype and their linkage to the corresponding tetra-paralogons in the human genome (Nakatani et al., 2007). Considering our localization of the four *Kissr* syntenic regions in the human genome, we can hypothesize that the corresponding tetra-paralogons resulted from the duplications of one single region localized on the proto-chromosome-A of the vertebrate ancestor (Pasquier et al., 2012).

#### **A subsequent history of Kissr losses**

Since the spotted gar and the coelacanth are the only two vertebrate species in which we discovered four paralogous *Kissr*, we can hypothesize multiple *Kissr* loss events to explain the status of this receptor in current vertebrates (**Figure 3A**). In the sarcopterygian lineage, *Kissr-4* may have been lost in amphibians, while *Kissr-1* and *Kissr-2* would have been lost in early amniotes. Subsequent additional losses may have led to the presence of only *Kissr-1* in eutherian mammals and to the absence of any *Kissr* in birds.

Considering the presence of four *Kissr* in a non-teleost actinopterygian, the spotted gar, the teleost-specific 3R could have resulted in the potential existence of up to eight *Kissr* genes. However, we only found three *Kissr* in the European eel, representing the current maximum number of this gene in teleosts. Furthermore, each eel *Kissr* is orthologous to a different tetrapod *Kissr*, supporting the absence of any teleost-specific *Kissr*. Synteny analysis demonstrated that each of the four *Kissr* paralogous genomic regions, present in the spotted gar, was duplicated in zebrafish, in agreement with the 3R. This analysis also indicated that all 3R-copies of *Kissr* were lacking in the corresponding duplicated regions (**Figure 2**). This suggests an early loss of duplicated *Kissr* genes, which would have suppressed the impact of the 3R on the number of *Kissr* in teleosts (**Figure 3A**). Additional successive deletions may have led to the presence of three *Kissr* in a basal teleost (the eel), two *Kissr* in a cypriniform

(zebrafish),and only one Kissr in a more recent teleost (stickleback; **Figure 3A**).

names of the current representative species of each phylum are given at the

#### **DIVERSITY AND EVOLUTIONARY HISTORY OF Kiss IN VERTEBRATES**

In contrast to the receptor proteins which present several conserved domains, the *Kiss* genes encode precursors which are highly variable among vertebrates, except for the short sequence of the mature decapeptide [Kp(10)]. This makes it difficult to obtain *Kiss* mRNA sequences by classical molecular strategies. Genomic database analyses thus represent the best approach to characterize the *Kiss* set for a given species. However, the small characteristic sequence of *Kiss*, encoding Kp(10), could be missing in genomic databases due to sequencing or assembly limitations.

All previously investigated osteichthyan species possessed the same number of *Kiss* and *Kissr* genes: one in eutherian mammals, two in prototherians, three in *Xenopus*, none in birds, and one or two in teleosts (Lee et al., 2009). In cyclostomes, two *Kiss* genes have been reported (Lee et al., 2009), while only one *Kissr* could be predicted until now (Pasquier et al., 2011).

#### **Diversity and classification of Kiss**

*New advances in Kiss gene characterization.* To further assess the *Kiss* diversity in vertebrates, we re-investigated the presence of these genes in the genome of the elephant shark, the coelacanth, the spotted gar, and the European eel, representative species from four groups of relevant phylogenetical positions. Most of the vertebrate *Kiss* genes are made of two exons except for some mammalian *Kiss1*, including human *Kiss1*, pig (*Sus scrofa*) *Kiss1*, and mouse *Kiss1*, that are made of three exons (**Figure 1**). However, the CDS of all the *Kiss* described so far are split on two exons (**Figure 1**). In fact the first of those two exons encodes the signal

peptide while the final exon encodes mainly the mature peptides including the conserved Kp(10) (**Figures 1** and **4**; Tomikawa et al., 2010, 2012; Cartwright and Williams, 2012; Tena-Sempere et al., 2012). Considering that the *Kiss* gene sequences are highly variable among species except for the sequence encoding the Kp(10), we focused our prediction on the ORF containing this sequence. We performed TBLASTN in the four genomes, resulting in the identification of several ORF containing conserved sequences encoding for Kp(10).

are indicated in the various lineages.

*Two Kiss genes in the European eel genome*. The two ORFs containing the sequences encoding Kp(10) are 296 and 327 bp long, respectively (**Figure A1** in Appendix). Once translated, each of them leads to a peptidic sequence encompassing a putative Kp(10): YNWNSFGLRY [European eel Kp1(10)] and FNRNPFGLRF [European eel Kp2(10)], respectively (**Figure 4**). The C-terminal end of the Kp1(10) sequence is followed by a GK-Stop motif, while the Kp2(10) sequence is followed by a GKR motif (**Figure A1** in Appendix). The sequences "X-G-Basic-Basic" or "X-G-Basic" are characteristic of the conserved proteolytic cleavage and alpha-amidation sites of neuropeptides (Eipper et al., 1992).

*Two Kiss genes in the spotted gar genome*. The two ORFs containing the sequences encoding Kp(10) are 348 and 300 bp long, respectively (**Figure A2** in Appendix). Once translated, each of them leads to a peptidic sequence presenting a putative Kp(10): YNWNSFGLRY [spotted gar Kp1(10)] and FNFNPFGLRF [spotted gar Kp2(10)], respectively (**Figure 4**). The C-terminal ends of these two sequences are followed by a GKR motif (**Figure A2** in Appendix).

identical amino-acids are shaded in dark gray and similar amino-acids in light gray. Newly identified sequences are underlined and unique sequences are marked by an asterisk.

*Three Kiss genes in the coelacanth genome*. The three ORFs containing the sequences encoding Kp(10) are 363, 396, and 81 bp long, respectively (**Figure A3** in Appendix). Once translated, each of them leads to a peptidic sequence encompassing a putative Kp(10): YNWNTFGLRY [coelacanth Kp1(10)], FNFNPFGLRF [coelacanth Kp2(10)], and FNWNSFGLRF [coelacanth Kp3(10)], respectively (**Figure 4**). The C-terminal ends of the Kp1(10) and the Kp2(10) sequences are followed by a GKR motif, while the Kp3(10) is followed by a GKK motif (**Figure A3** in Appendix)*.* Seven amino-acids up-stream the sequence of the Kp3(10), a stop codon appears (**Figure A3** in Appendix) suggesting that coelacanth *Kiss3* gene could have a different intro-exon structure compared to what has been described so far or it can suggest that this gene is no longer expressed.

*A third Kiss gene in the elephant shark genome*. While two *Kiss* (*Kiss1* and *Kiss2*) were previously identified in the elephant shark genome (Lee et al., 2009), we were able to localize a new ORF of 315 bp containing a third sequence encoding a Kp(10) (**Figure A4** in Appendix). Once translated, it leads to a peptidic sequence encompassing a putative Kp(10): YNLNSFGLKF [elephant shark Kp3(10)] (**Figure 4**). The C-terminal end of this peptide is followed by a GKR motif (**Figure A4** in Appendix).

*Kiss sequence alignment and comparisons.* The alignment of 56 kisspeptin precursors revealed a high variability of their amino-acid sequences except for the sequences corresponding to Kp(10) and its few surrounding amino-acids which, in contrast, are highly conserved (data not shown). Such a pattern, which is representative of many other neuropeptide precursors, provides poor phylogenetic information within alignment matrix.

This lack of information makes the use of phylogenetic analysis inappropriate to establish homology relationships within this kind of peptide precursor family.

*Novel Kp(10).* Among the new Kiss genes predicted in the present study, four of them encode novel Kp(10) (**Figure 4**). The singularity of the elephant shark Kp3(10) is the presence of a lysine (K) instead of an arginine (R) at the ninth position. The coelacanth Kp1(10) provides the first case of a threonine (T) at the fifth position. The coelacanth Kp3(10) is the only one to present both a phenylalanine (F) at the first position and a serine (S) at the fifth position. The European eel Kp2(10) presents at its third position an arginine (R), which possesses different physical and chemical properties from amino-acids commonly present at this position. Up to now, only the musk shrew (*Suncus murinus*) Kp1(10) presented an arginine at the third position and it was demonstrated that its kisspeptin system was involved in the reproductive function as in other mammals (Inoue et al., 2011). Since the impacted positions by the amino-acid substitutions have not been characterized as highly critical for Kp(10) functional properties (Gutiérrez-Pascual et al., 2009; Curtis et al., 2010), those novel Kp(10) may have conserved their functionality. Since Kp(10) is considered as the smallest required sequence to specifically bind to the receptor (Kotani et al., 2001), it could be of interest to test all those peptides in future pharmacological studies in order to assess their structure/function relationships.

*Syntenic analysis and classification of Kiss genes.* In order to classify the different *Kiss* paralogs, we performed a syntenic analysis of the *Kiss* neighboring genes. We considered the following vertebrate representatives: mammals (human), birds (chicken), squamates (lizard), amphibians (*Xenopus*), basal sarcopterygian (coelacanth), non-teleost actinopterygians (spotted gar), and teleosts (zebrafish, stickleback, and European eel). Our syntenic analysis demonstrated that the *Kiss* genes are localized in three different genomic regions.

The human *Kiss1*, *Xenopus Kiss1a*, coelacanth *Kiss1*, spotted gar *Kiss1*, and zebrafish *Kiss1* are positioned in genomic regions containing common loci, including *TEAD3*, *NAV1*, *PPP1R12B*, *PPFIA4*, *MYBPH*, *KCNC4*, *REN*, *GOLT1A*, *PLEKHA6*, *PPP1R15B*, *PIK3C2B*, and *SYT6*, thus exhibiting well conserved synteny (**Figure 5A**). This supports the orthology of these *Kiss* genes, all considered as *Kiss1* genes. Syntenic analysis suggests that the stickleback, lizard, and chicken genomes do not contain any *Kiss1* gene, although the above-mentioned neighboring genes are present in the corresponding genomic regions (**Figure 5A**). The peptidic sequence of eel Kiss1 presents many similarities to the other Kiss1, but the eel *Kiss1* gene is located on too small scaffolds to contain any other gene, preventing from any syntenic analysis.

The lizard *Kiss2*, coelacanth *Kiss2*, spotted gar *Kiss2*, zebrafish *Kiss2*, stickleback *Kiss2*, and European eel *Kiss2* genes are positioned in genomic regions containing common loci including *STRAP*, *PLEKHA5*, *GOLT1B*, *C12orf39*, *GYS2*, *LDHB*, *KCNJ8*, *ABCC9*, *CMAS*, *SYT10*, *NAV3*, *PPFIA2*, and *KCNC2*, thus exhibiting well conserved synteny (**Figure 5B**). This supports the orthology of these *Kiss* genes, all considered as *Kiss2* genes. Syntenic analysis suggests that human and chicken genomes do not contain

any *Kiss2* gene, although the above-mentioned neighboring genes are present in the corresponding genomic region (**Figure 5B**).

European eel (Anguilla anguilla). This map was established using the PhyloView of Genomicus v67.01 web site, manual annotation of the European

The coelacanth *Kiss3* and the *Xenopus Kiss1b* genes are positioned in genomic regions containing common loci, including *TEAD2*, *PIH1D1*, and *ALDH16A1*, thus exhibiting well conserved synteny (**Figure 5C**). This supports the orthology of these two *Kiss* genes, both considered here as *Kiss3* genes. Syntenic analysis suggests that human, lizard, spotted gar, and teleost genomes do not contain any *Kiss3* gene, although the above-mentioned neighboring genes are present in the corresponding genomic regions (**Figure 5C**). Syntenic analysis also suggests that the whole considered region is absent from the chicken genome.

#### **Evolutionary history of Kiss**

*Origin of Kiss diversity via 1R and 2R.* The syntenic analysis also allowed us to investigate the origin of the multiple*Kiss* genes found in vertebrates. The three conserved genomic regions, presenting *Kiss* genes, also comprise other paralogs from 11 gene families: *TEAD* (4 paralogs), *NAV* (3 paralogs), *PPFIA* (4 paralogs), *KCNC* (4 paralogs),*GOLT1* (2 paralogs), *PLEKHA* (4 paralogs), *PPP1R15* (2 paralogs), *PIK3C2* (3 paralogs), *SYT* (4 paralogs), *GYS* (2 paralogs), and *PTH* (2 paralogs) (**Figures 5A–C**). The members of those families are present among the three *Kiss* syntenic regions and they also delineate a fourth conserved region (**Figure 5D**), which does not present any *Kiss* gene in the osteichthyan representative species studied so far. The four conserved regions delineated by the 11 gene families can be considered as paralogous (tetra-paralogon).

genomic locations are given in Table S1 in Supplementary Material. Chr, chromosome; LG, linkage group; Gr, group; Sc, scaffold; Cg, contig.

Considering the reconstruction of the ancestral vertebrate chromosomes, their linkage to the tetra-paralogons in the human genome (Nakatani et al., 2007) and our localization of the four *Kiss* syntenic regions in the human genome (on Chromosomes 1, 11, 12, and 19), we can hypothesize that the *Kiss* tetra-paralogons resulted from the duplications of one single genomic region

localized on the proto-chromosome-D of the vertebrate ancestor. Therefore, we can infer that the current three *Kiss* genes may have resulted from a single ancestral gene duplicated through 1R and 2R that occurred in early steps of vertebrate evolution (**Figure 6**).

#### *A subsequent history of Kiss losses.*

*Multiple loss events in sarcopterygians and actinopterygians*. The 1R and 2R events should have resulted in four different *Kiss* genes in vertebrates. Since the fourth *Kiss* gene (referred to as *Kiss4* in this study) has not been observed in any species studied so far, we can hypothesize an early loss of this gene after the 2R. As only a chondrichthyan, the elephant shark, and two sarcopterygians, the coelacanth and *Xenopus*, still present three *Kiss* genes, whereas all other species possess less than three *Kiss*, we can hypothesize multiple additional events of *Kiss* losses in vertebrates (**Figure 3B**).

Among the sarcopterygian lineage, in tetrapods, *Kiss3* would have been lost in amniotes. Further alternative losses may have occurred in this lineage, with only *Kiss1* remaining in eutherian mammals and only *Kiss2* in squamates (lizard). Finally, additional losses would have led to the complete absence of *Kiss* in birds (**Figure 3B**). Among the actinopterygian lineage, an early loss of *Kiss3* would have occurred since it is lacking in the actinopterygian species investigated so far (**Figure 3B**).

*No impact of the teleost-specific 3R on Kiss number in current species*. In the actinopterygian lineage, the teleost-specific 3R and the presence of two *Kiss* in a non-teleost actinopterygian, the spotted gar, implied the potential existence of at least four *Kiss* genes in the early teleost history. However, our study showed that so far the largest number of *Kiss* exhibited by current teleosts, including the eel, is two. Furthermore, each teleost *Kiss* is orthologous to a different tetrapod *Kiss,* indicating that no teleost-specific *Kiss* exists. Synteny analysis revealed that each of the four *Kiss* genomic regions present in the spotted gar is duplicated in teleosts in agreement with the 3R event but that duplicated *Kiss* genes are lacking (**Figures 5** and **6**). This suggests an early loss of duplicated *Kiss* genes suppressing the impact of the 3R on the number of *Kiss* in teleosts (**Figure 3B**). Additional deletion may have led to the presence of only *Kiss2* in gasterosteiforms (the stickleback; **Figure 3B**). *Kiss* evolutionary history was punctuated by numerous loss events through vertebrate radiation (**Figure 3B**).

#### **COMPARISON OF THE EVOLUTIONARY HISTORIES OF Kiss AND Kissr IN VERTEBRATES**

These new data concerning *Kiss* and *Kissr* diversities enabled us to improve their respective classifications and evolutionary histories. A remaining challenge was to elucidate whether *Kiss* and *Kissr* families have experienced parallel histories during vertebrate radiation. The comparative study of the current status of both families in vertebrates allows a better understanding of the whole kisspeptin system.

#### **Features in agreement with parallel histories**

*Origin of the Kiss and Kissr multiplicity via 1R and 2R.* Our syntenic studies suggest that the vertebrate *Kiss* and *Kissr* families both resulted from the successive duplications of a single ancestral gene through the 1R and 2R (**Figure 3**). Thus, *Kiss* and *Kissr* experienced, in parallel, the two first genome duplication rounds resulting in four copies of each gene in the early steps of the vertebrate evolutionary history (**Figure 3**). While *Kissr* gene homologs were characterized in non-vertebrate species (*Strongylocentrotus purpuratus*, GenBank accession numbers: XP\_793873.1 and XP\_796286.1; *Saccoglossus kowalevskii*, GenBank accession numbers: NP\_001161573.1 and NP\_001161574.1), tracing back the presence of an ancestral*Kissr* in early deuterostomes,*Kiss* genes have not yet been discovered in non-vertebrate species.

*Subsequent history of gene losses.* Both *Kiss* and *Kissr* families were composed of four genes in the early steps of the vertebrate history. However, most of the current vertebrate species investigated so far present less than four copies of *Kiss* and *Kissr* genes. The current numbers of both *Kiss* and *Kissr* genes suggest that both families underwent numerous independent loss events across vertebrate history (**Figure 3**).

*No impact of the teleost-specific 3R.* The teleost lineage, which has experienced a third whole genome duplication round (3R), could have been expected to possess up to eight *Kiss* and *Kissr* genes. However, the analyses of the *Kiss* and *Kissr* within teleost genomes revealed a maximum of three *Kissr* and two *Kiss* genes and did not reveal any 3R-specific copies of *Kiss* or *Kissr* genes*.* This suggests that the teleost-specific 3R did not impact the current number of *Kiss* or *Kissr* genes, reflecting massive losses of the 3R-copies of both *Kiss* and *Kissr* genes in early teleosts (**Figure 3**).

**Features in opposition to parallel histories: independent loss events** *Un-matching number of Kiss and Kissr in some species.* In the current gnathostomes, we observed a maximum of four *Kissr* but only three *Kiss* paralogs. This difference suggests that this lineage inherited the four *Kissr* copies resulting from the 1R and 2R, whereas the fourth *Kiss* may have been lost before or at an early stage of the gnathostome emergence (**Figure 3**). This situation was observed in an early sarcopterygian, the coelacanth, while an even larger difference in *Kissr* (four) and *Kiss* (two) numbers was found in the spotted gar, reflecting an additional independent loss of *Kiss* in the actinopterygian lineage. An un-matching number of *Kissr* (three) and *Kiss* (two) was also observed in an early teleost, the European eel, while additional losses may have led to equal numbers of *Kissr* and *Kiss* in more recent teleosts (two or one according to the species). These variations in *Kissr/Kiss* numbers reflect different timing of *Kiss* and *Kissr* loss events. Those hypotheses suggest that *Kiss* losses occurred independently among the different gnathostome lineages and also independently from the *Kissr* losses.

*Various Kiss/Kissr combinations across vertebrates.* The hypothesis of independent *Kiss* and *Kissr* evolutionary histories is also strengthened by the comparison of the gene sets present in species with even matching numbers of *Kiss* and *Kissr*. For example, lizard, and stickleback both present the *Kiss2* gene, whereas they possess different *Kissr*, i.e., *Kissr-4* in the lizard and *Kissr-2* in the stickleback (**Figure 3**). The same observation can be done comparing the kisspeptin systems of platypus and zebrafish, which both present the *Kiss1* and *Kiss2* genes, whereas their sets of receptors are completely different, i.e., *Kissr-1* and *Kissr-4* in platypus versus *Kissr-2*

and *Kissr-3* in zebrafish (**Figure 3**). These observations strongly suggest a large variety of *Kiss* and *Kissr* combinations, resulting from independent loss events. These data shed new lights on the evolution of the kisspeptin system in vertebrates and challenge the former hypothesis of a conservation of *Kiss/Kissr* couples across vertebrate evolution. This diversity among vertebrates opens new research avenues for comparative physiology and endocrinology of kisspeptin system.

*What could have favored the independent evolutions of Kiss and Kissr?* A few *in vitro* studies using recombinant receptors have showed cross-reactivities between various kisspeptins and kisspeptin receptors (Biran et al., 2008;Lee et al., 2009; Li et al., 2009). For instance, Lee et al. (2009) tested the specificity of recombinant human GPR54 (Kissr-1 according to our nomenclature), zebrafish GPR54-1, and -2 (Kissr-3 and Kissr-2 according

to our nomenclature), and *Xenopus* GPR54-1a, -1b, and -2 (Kissr-1, Kissr-4, and Kissr-2 according to our nomenclature) toward various kisspeptins. They showed that human, zebrafish, and *Xenopus* kisspeptins were able to activate all the receptors with differential intra and inter-specific ligand selectivity. Such crossreactivity could have promoted the independence of the *Kiss* and *Kissr* evolutionary histories. This could explain the situation of species presenting un-matching numbers of *Kiss* and *Kissr* genes, as well as the high variability of *Kiss/Kissr* gene combinations across vertebrate species. Another study, using goldfish recombinant GPR54a and GPR54b (Kissr-3 and Kissr-2 according to our nomenclature), revealed that the ligand potency strikingly differed depending on the responsive element used in the reporter gene construction (Li et al., 2009). These data showed the difficulty to define specific Kiss/Kissr couples based only on pharmacological properties.

In the case of the presence of multiple *Kiss/Kissr* genes in a given species, anatomical relationships between projections of *Kiss* neurons and target cells expressing *Kissr* may provide further cues for determining Kiss/Kissr functional couples. Thus, in zebrafish which possess two*Kiss* genes and two*Kissr* genes,*in situ* hybridization and immunocytochemical studies localized the *Kiss1* neurons in different nuclei from *Kiss2* neurons (Servili et al., 2011). Moreover, *Kiss1* neurons are projecting to *Kiss1r* (*Kissr-3* according to our nomenclature) expressing cells, while *Kiss2* neurons are projecting to *Kiss2r* (*Kissr-2* according to our nomenclature) expressing cells (Servili et al., 2011). This reveals anatomically separated kisspeptin systems and distinct specific Kiss/Kissr functional couples in zebrafish (Servili et al.,2011). In striped bass,another teleost possessing two *Kiss* genes and two *Kissr* genes, *in situ* hybridization and laser capture microscopy coupled to quantitative PCR showed, in contrast, that *Kiss1* and *Kiss2* were co-expressed in neurons of the hypothalamus, indicating promiscuous Kiss synthesis sites (Zmora et al., 2012). However, the two *Kissr* of the striped bass were expressed in different brain cells, indicating that the kisspeptin systems are not fully redundant (Zmora et al., 2012). All these data underline the importance of investigating the gene diversity, the anatomical organization and the functional properties of the kisspeptin system in various species, regarding the potential high variability of this system among vertebrates.

# **CONCLUSION**

Kisspeptin system is known to play a role in many physiological processes such as antimetastasis, energy metabolism homeostasis, pregnancy, and puberty onset. Even though this system has been widely studied in the last few years, its diversity and evolutionary history remained unclear. Thanks to the newly published genomes of osteichthyans of key phylogenetical positions, we were able to provide new data on the diversity of *Kiss* and *Kissr* genes, to clarify the classification of these genes and to bring new insights on the evolutionary history of these gene families. Four *Kissr* and four *Kiss* genes may have arisen via the 1R and 2R in early vertebrates. This would have been followed by multiple independent *Kiss*

#### **REFERENCES**


pre-eclampsia. *J. Endocrinol.* 214, 79–85.


and *Kissr* gene loss events in the sarcopterygian and actinopterygian lineages. In particular, due to massive *Kiss* and *Kissr* gene losses, no impact of the teleost-specific 3R can be recorded on the number of *Kissr* or *Kiss* paralogs in current teleost species. The comparison of both *Kiss* and *Kissr* gene status, in the current vertebrates, supports both parallel and independent evolutionary histories of the *Kiss* and *Kissr* families across vertebrate radiation. It also underlines a large diversity of *Kiss*/*Kissr* possible combinations that needs to be taken into account in future comparative studies.

#### **ACKNOWLEDGMENTS**

We thank Dr. B. Quérat (CNRS, Paris, France) for his helpful advices and discussions concerning phylogeny and synteny. We also thank all the different consortiums at the initiative of sequencing, assembly, annotation, and publication of genomes, in particular the consortiums of the European eel genome, the spotted gar genome, the coelacanth genome, the elephant shark genome, and the sea lamprey genome. Our thanks also include the authors of the Ensembl genome browser web site and all web sites providing free access to their genomic databases. We thank the authors of the Genomicus web site providing a very useful and friendly tool for synteny analysis. Jérémy Pasquier is a recipient of a Ph.D. fellowship from the Ministry of Research and Education. This work was supported by grants from the National Research Agency, PUBERTEEL N ANR-08-BLAN-0173 to Karine Rousseau and Sylvie Dufour, and from the European Community, Seventh Framework Program, PRO-EEL N No. 245257 to Anne-Gaëlle Lafont and Sylvie Dufour.

#### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at http://www.frontiersin.org/Neuroendocrine\_Science/10.3389/ fendo.2012.00173/abstract

#### **Table S1 | Names, references, and locations of the genes used in the Kiss synteny analysis.**


Structural and functional multiplicity of the kisspeptin/GPR54 system in goldfish (*Carassius auratus*). *J. Endocrinol.* 201, 407–418.


question anymore. *Nat. Rev. Genet.* 11, 166.


Thillart, G., et al. (2012). Multiple kisspeptin receptors in early osteichthyans provide new insights into the evolution of this receptor family. *PLoS ONE* 7:e48931. doi:10.1371/journal.pone.0048931

Pasquier, J., Lafont, A. G., Leprince, J., Vaudry, H., Rousseau, K., and Dufour, S. (2011). First evidence for a direct inhibitory effect of kisspeptins on LH expression in the eel, Anguilla anguilla. *Gen. Comp. Endocrinol.* 173, 216–225.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 05 October 2012; accepted: 07 December 2012; published online: 26 December 2012.*

*Citation: Pasquier J, Lafont A-G, Tostivint H, Vaudry H, Rousseau K and Dufour S (2012) Comparative evolutionary histories of kisspeptins and kisspeptin receptors in vertebrates reveal both parallel and divergent features. Front. Endocrin. 3:173. doi: 10.3389/fendo.2012.00173*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Pasquier, Lafont , Tostivint ,Vaudry, Rousseau and Dufour. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# **APPENDIX**


**genome.** Nucleotide and deduced amino-acid sequences of the ORF encoding the European eel Kp1(10) **(A)** and Kp2(10) **(B)**. Nucleotides (top) are numbered from 5<sup>0</sup> to 3<sup>0</sup> . The amino-acid residues (bottom) are

numbered beginning with the first amino-acid residue encoded by the ORF. The asterisk (\*) indicates the stop codon. The predicted Kp(10) are underlined. The amino-acids of the C-terminal α-amidation and cleavage site are shaded in gray.


**genome.** Nucleotide and deduced amino-acid sequences of the ORF encoding the spotted gar Kp1(10) **(A)** and Kp2(10) **(B)**. Nucleotides (top) are numbered from 5<sup>0</sup> to 3<sup>0</sup> . The amino-acid residues (bottom) are

underlined. The amino-acids of the C-terminal α-amidation and cleavage site are shaded in gray.



# Ghrelin receptors in non-mammalian vertebrates

# **Hiroyuki Kaiya<sup>1</sup>\*, Kenji Kangawa<sup>2</sup> and Mikiya Miyazato<sup>1</sup>**

<sup>1</sup> Department of Biochemistry, National Cerebral and Cardiovascular Center Research Institute, Osaka, Japan

<sup>2</sup> National Cerebral and Cardiovascular Center Research Institute, Osaka, Japan

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Hélène Volkoff, Memorial University of Newfoundland, Canada Catharina Olsson, University of Gothenburg, Sweden

#### **\*Correspondence:**

Hiroyuki Kaiya, Department of Biochemistry, National Cerebral and Cardiovascular Center Research Institute, 5-7-1 Fujishirodai, Suita, Osaka 565-8565, Japan e-mail: kaiya@ncvc.go.jp

# **GENERAL INTRODUCTION**

As implied by their name, growth hormone secretagogues (GHSs), which are artificial derivatives of enkephalin, exhibit growth hormone (GH)-releasing activity (1). Some of these GHSs also stimulate appetite in mammals (2). In 1996, Howard et al. (3) discovered a G-protein-coupled receptor (GPCR) with seven transmembrane domains (TMDs) in humans and pigs, and found that GHSs bound to this receptor and elicited an increase in the intracellular Ca2<sup>+</sup> concentration of cells in which it was stably expressed. They named this receptor the GHS-receptor type-1a (GHS-R1a); in addition, they found an alternative splice variant of the receptor that lacked the Ca2<sup>+</sup> signaling capacity and named it GHS-R type-1b (GHS-R1b). The mammalian GHS-R gene (*ghsr*) comprises two exons separated by one intron (4, 5). GHS-R1a comprises 366 amino acids (AAs), where the first exon (exon 1) encodes the first 265 AAs from TMD 1–5, and the second exon (exon 2) encodes the remaining 101 AAs from TMD 6 and 7. In contrast, the alternative splice variant of *ghsr*, GHS-R1b, is formed from the first exon and part of the intron. Thus, the protein sequence of the entire 289- AA GHS-R1b is identical to GHS-R1a from the N-terminal end to TMD 5.

Extensive investigations were performed to identify the endogenous ligand for the orphan GHS-R1a following discovery of the receptor, and reverse pharmacology facilitated the identification of a natural ligand in 1999 by Kojima et al. (6). The peptide ligand, which contains 28 AAs, was isolated from stomach extracts of rats and named "ghrelin." Ghrelin has a unique fatty acid modification on its N-terminal third serine (Ser3), with an *n*-octanoyl group linked to the hydroxyl group of Ser3. This modification is essential for the binding of ghrelin to the receptor (7) and for eliciting various physiological actions. After the discovery of its endogenous ligand, GHS-R1a was found to mediate various physiological functions of ghrelin: neuroendocrine function; appetite regulation; cardiovascular function; gastro-entero-pancreatic function; glucose metabolism; and cell

The growth hormone secretagogue-receptor (GHS-R) was discovered in humans and pigs in 1996. The endogenous ligand, ghrelin, was discovered 3 years later, in 1999, and our understanding of the physiological significance of the ghrelin system in vertebrates has grown steadily since then. Although the ghrelin system in non-mammalian vertebrates is a subject of great interest, protein sequence data for the receptor in non-mammalian vertebrates has been limited until recently, and related biological information has not been well organized. In this review, we summarize current information related to the ghrelin receptor in non-mammalian vertebrates.

**Keywords: ghrelin, ghrelin receptor, GHS-R, GHS-R-like receptor, fishes, amphibians, reptiles, birds**

functions including apoptosis, proliferation, and differentiation (8–10).

In non-mammalian vertebrates, GHSs affect the regulation of GH release and of appetite in fish and birds (11–14), suggesting the presence of an endogenous ghrelin-like substance and a corresponding receptor system. We first isolated ghrelin from a non-mammalian vertebrate, the bullfrog (15). Subsequently, ghrelin was determined to be present in various non-mammalian vertebrates, and its physiological effects were gradually revealed [for reviews, see Ref. (16, 17)]. However, investigations of nonmammalian ghrelin receptors still lag behind those on mammalian ghrelin receptors. In this review, we summarize our recent work and those of others on ghrelin receptors in non-mammalian vertebrates and provide a comprehensive discussion of their general features.

# **CLASSIFICATION AND NOMENCLATURE OF GHRELIN RECEPTORS**

We begin by describing the nomenclature for the ghrelin receptors in mammals, because the nomenclature for the receptors in non-mammalian vertebrates is more complicated and various names have been used based on the presence of splice variants, paralogs, and different AA lengths. In the first description provided by Howard et al. (3), GHS-R1a was defined as a functional receptor induced by agonist-dependent intracellular Ca2+, and GHS-R1b as a splice variant of unknown function. They classified them simply as "a" and "b" because their sequences and functions differed. Thus the names are based on the sequence and structure: "GHS-R1" refers to the receptor with a "type-1" AA sequence, "a" signifies "activated by ghrelin or GHSs," and "b" indicates "a splice variant of *ghsr*" which contains the first exon and an unspliced intron that continues the coding sequence in the mRNA and terminates at a stop codon within the intron. The International Union of Pharmacology Committee on Receptor Nomenclature and Drug Classification has accepted "GHS-R1a" as the name for

the functional ghrelin receptor (18). Hence, two GHS-Rs exist in mammals: GHS-R1a, which is derived from regular splicing of the gene; and GHS-R1b, which originates from alternative splicing of the gene (**Figure 1**). On the basis of these names, we describe the naming of the receptors in non-mammalian vertebrates as follows.

The non-mammalian GHS-Rs are also roughly divided into two types: (i) an isoform that arises from regular splicing of the gene and (ii) an isoform derived from alternative splicing of the gene (**Figure 1**). The former is further classified into two isoforms (**Figure 1**): one denotes an isoform that we designated "GHS-Ra," which has structural properties similar to those of the mammalian GHS-R1a and is activated by ghrelin and GHSs. GHS-Ra is further divided into two paralogs "1a" and "2a," where "GHS-R2a" refers to the receptor with a "type-2" AA sequence distinct from that of GHS-R1a and whose existence is confirmed only in specific fish. The other denotes another isoform that we designated "GHS-R1a-like receptor (GHS-R1a-LR)," which has structural features that differ from those of GHS-Ra and for which intracellular Ca2<sup>+</sup> increase in response to ghrelin or GHS treatment is either small or not confirmed. This distinction between GHS-Ra and GHS-R1a-LR is evident in the phylogenetic analysis based on the AA sequences of ghrelin receptors (**Figure 2**).

The isoforms derived from alternative splicing of the gene are divided into five types: 1b, 1aV (1c), 1bV, tv, and tv-like receptors. These receptors are formed by different modes of alternative splicing and have distinct structures.

#### **NON-MAMMALIAN VERTEBRATE SPECIES WITH SEQUENCED GHRELIN RECEPTORS**

We have summarized the non-mammalian vertebrates for which the cDNA or genes of GHS-R have been identified and made available in public databases in **Table 1** (fish) and **Table 2** (reptiles, amphibians, and birds). The AA sequences of GHS-R1a, 2a; GHS-R1a-LR; and their multiple alignments are shown in **Figure 3**.

Many GHS-Rs have been identified in non-mammalian vertebrates, and the most of the GHS-R types that have been found are present in fish (19 species). With the recent identification of a GHS-R in bullfrog and Japanese tree frog (19), we now know the GHS-Rs for three kinds of frogs, including African clawed frogs. In reptiles,there are no reports about GHS-Rs at present,although the Ensembl genome database search (http://www.ensembl.org/index. html) yields the GHS-R1a gene for the green anole (*Anolis carolinensis*) and painted turtle (*Chrysemys picta bellii*). Very recently, massive numbers of partial nucleotide sequences (approximately 450-bp encoding a 150-AA protein) of GHS-R have been registered for 124 species of *Squamata*, including snakes and Iguanidae, by Wiens et al. (98) at Stony Brook University in the NCBI database. In birds, GHS-Rs have been found in five species.

# **STRUCTURAL FEATURES OF THE GHRELIN RECEPTOR IN NON-MAMMALIAN VERTEBRATES**

Three features are prominent in non-mammalian GHS-Rs: (1) the presence of paralogs in a few species of teleosts; (2) two isoforms, GHS-Ra and GHS-R1a-LR; and (3) avian-specific alternative splice forms of GHS-R (**Figure 1**). Further details are provided below (see also Classification and Nomenclature of Ghrelin Receptors).

#### **PRESENCE OF PARALOGS IN ONLY A FEW SPECIES OF TELEOSTS**

The GHS-Ra paralog GHS-R2a is found only in a limited number of teleosts, and little is known about the presence of GHS-R paralogs in other vertebrates. GHS-R2a has an AA sequence that is approximately 70% identical to that of GHS-R1a. At present, this receptor has been identified in Cypriniformes such as goldfish, zebrafish, and carp, and in channel catfish in the order Siluriformes

#### **FIGURE 2 | Phylogenetic tree of GHS-Ra and GHS-R1a-LR in**

**non-mammalian vertebrates**. The phylogenetic tree was constructed by using the neighbor-joining method with MEGA4 (http://www.megasoftware.net/). The numbers on the branch points are the bootstrap values (as percentages based on 2000 replicates). The scale bar indicates the average number of substitutions per position (a relative measure of evolutionary distance). Receptors for human motilin (MTLR), neuromedin-U (NMUR1), and neurotensin (NTSR1) were used as the outgroup.

(**Figures 2** and **3**). The two isoforms are encoded by different genes (i.e., the zebrafish GHS-R1a and 2a genes are located separately on chromosomes 4 and 24, respectively), which are considered to have diverged via the third round of whole-genome duplication (3R-WGD) that occurred in the ray-finned fish lineage (20, 21).

In addition, isoforms with approximately 95% identity have been found in goldfish (Cypriniformes) and rainbow trout (Salmoniformes). In goldfish, there are two paralogs each for GHS-R1a and 2a: GHS-R1a-1, 1a-2, 2a-1, and 2a-2 (**Figures 2**, **3**, and **5**). Each receptor originated from a separate gene demonstrated to have a different intron sequence (22). In the rainbow trout, two paralogous sequences, namely the DQTA/LN-type and ERAT/IStype, have been identified (23) (**Figure 3**). Their names indicate AA substitutions at D20E, Q32R, T54A, A62T, L168I, and N264S. These two receptor sequences are known to be derived from at least three distinct genes (the DQTA/LN-type derives from two genes and the ERAT/IS-type originates from one gene), on the basis of analyses of an intron sequence of each receptor (23). These paralogs of goldfish and rainbow trout are considered to have originated from polyploidization events that occurred after 3R-WGD (24) and tandem duplication of the genes, which also affected the opsin gene in these species (25). The presence of multiple paralogs may be a peculiar characteristic of *Ostariophysi* and *Protacanthopterygii* in euteleosts (20, 21).

#### **TWO GHRELIN RECEPTOR ISOFORMS: GHS-Ra AND GHS-R1a-LR**

As shown in **Figure 1**, there are two isoforms in non-mammalian vertebrates: GHS-Ra and GHS-R1a-LR. GHS-Ra includes GHS-R1a and 2a. Tetrapods including mammals, birds, reptiles, and amphibians have GHS-R1a, whereas some bony fish such as Coelacanthiformes, Cypriniformes (e.g., goldfish, carp, and zebrafish), and Siluriformes (e.g., channel catfish) have both GHS-R1a and 2a. GHS-R1a-LRs show considerable AA identity to GHS-R1a, but have a unique structural feature not found in any tetrapod: the second extracellular loop (ECL2) that connects TMD 4 and 5 is notably longer than that of GHS-R1a (**Figure 4**). In addition, GHS-R1a-LRs have the characteristic that ghrelin or GHS treatment either does not increase intracellular Ca2<sup>+</sup> (23, 26) or requires pharmacological doses to activate the receptor (27, 28). This type of receptor is seen in a limited number of fish classified as Percomorpha within the superorder *Acanthopterygii*, which is the most evolutionally advanced group of teleosts, including Perciformes such as black porgy and tilapia, Gasterosteiformes such as stickleback and medaka, Tetraodontiformes such as pufferfish,and Salmoniformes such as rainbow trout (**Figure 3**). An exception is the orange-spotted grouper, which belongs to Perciformes but has an ECL2 that is not long (**Figure 3**). These species have some morphological characteristics such as a highly mobilized upper jaw, a respiratory tract not linked to the swim bladder, and a splinter article in their fins. Salmoniformes belong to *Protacanthopterygii*, which contains a number of moderately advanced teleosts. This evolutionary background may be reflected in the molecular evolution and structure of the ghrelin receptor.

A partial sequence similar to that of the ghrelin receptor was found in a database for the sea lamprey (*Petromyzon marinus*). This receptor could not be placed at the branch of GHS-Ra or GHS-R1a-LR in the phylogenetic analysis (**Figure 2**). The sea lamprey belongs to the group Cyclostomata in the class Agnatha, which is a class of fish with the characteristics of ancient basal vertebrates. Therefore, the receptor in the sea lamprey may contain ancestral characteristics of the ghrelin receptor.

#### **AVIAN-SPECIFIC GHS-Rs**

Birds have specific alternative spliced forms of GHS-R other than GHS-R1b, i.e., 1aV (or 1c), 1bV, tv, and tv-like receptor (29– 32), which are generated by differential modes of splicing from GHS-R1b. GHS-R1aV (30) and GHS-R1c (29) are identical receptors found in chickens. Here, "V" is considered to mean "variant" (30), whereas Geelissen et al. (29) used the designation "c" to indicate an isoform different from "a" or "b." We proposed that GHS-R1c should be referred to as GHS-R1aV because the receptor is identical to GHS-R1a with the exception that it lacks

#### **Table 1 | Ghrelin receptor and ghrelin receptor-like receptor in fish.**



NCBI (http:// www.ncbi.nlm.nih.gov/ ); NCBI genome database (http:// www.ncbi.nlm.nih.gov/ genome/ ); Ensembl Genome Browser (http:// www.ensembl.org/ index. html).

16 AAs (46 bp) in TMD 6 (16). GHS-R1bV is found in quail. Its C-terminal part differs from that of GHS-R1b, and an AA sequence that differs from 1b is translated from the intermediate intron by a frame-shift due to an 8-bp deletion of the intermediate intron of *ghsr*. GHS-Rtv is found in chickens (31). The signature "tv" was first used by Sirotkin et al. (31), although its meaning is unclear. The composition of GHS-Rtv is complex: two distinct parts of the intermediate intron sequence of *ghsr* lie between the exon 1 and exon 2 sequences of GHS-R1a [see Ref. (33)]. Kitazawa et al. (32) reported a receptor similar to chicken GHS-Rtv in the Japanese quail. Because the composition was different from that of GHS-Rtv, it was designated as a GHS-Rtv-like receptor and considered to be a possible ortholog of GHS-Rtv. The functions of these avian variants are completely unknown.

Kitazawa et al. (32) reported five isoforms of GHS-Rs in the Japanese quail: GHS-R1a-L, 1a-S, 1aV-L, 1b-L, and 1bV-L. The "L" and "S" appended to GHS-R1a signify the long-type (354 AAs) and short-type (347 AAs) receptors for GHS-R1a, respectively. GHS-R1a-S is a receptor that lacks 7 AAs at the N-terminus of GHS-R1a-L. Two ATG initiation codons are present in the cDNA and the functional codon is unknown.

# **TISSUE EXPRESSION OF GHRELIN RECEPTOR mRNAs AND THEIR ISOFORMS**

#### **EXPRESSION OF GHS-Ra AND GHS-R1a-LR**

In agreement with a wide range of physiological functions of ghrelin, GHS-R1a transcripts have been detected in human tissues such as the brain, heart, lung, liver, kidney, pancreas, stomach, intestines, and adipose tissue (34, 35). In particular, high expression levels have been detected in the pituitary gland (36), which is consistent with the role of ghrelin in regulating GH release. In the brain, where expression levels are relatively high, GHS-R1a mRNA is widely distributed in regions linked to energy homeostasis such as the arcuate nuclei of the hypothalamus; area postrema; nucleus of the solitary tract; the dorsal motor nucleus of the vagus; hippocampus; dopaminergic neurons in the ventral tegmental area and substantia nigra; parasympathetic preganglionic neurons; the dorsal and medial raphe nuclei; and the dentate gyrus (9, 34, 37, 38).

In non-mammalian vertebrates, GHS-R1a or GHS-R1a-LR transcripts have been found in the central nervous system and various peripheral organs. As in humans, predominant expression occurs in the pituitary in channel catfish (39), chickens (29, 30, 40–43), and ducks (44) for GHS-R1a, as well as in the black porgy (28), orange-spotted grouper (45), and rainbow trout (23)


**2|Ghrelinreceptorandghrelinreceptor-likereceptorinreptiles,amphibians,and**


**FIGURE 3 | Continued**


motifs of the G-protein-coupled receptor transmembrane domains 5 and 7. Sequences were aligned using GENETYX-Mac version 15.0.1.

for GHS-R1a-LR. However, expression in the pituitary gland is not dominant in all species. In Mozambique tilapia, GHS-R1a-LR mRNA is mainly detected in the brain. The distribution of the ghrelin receptor in other tissues also differs among animal species.

In fish, GHS-R transcripts have been detected in most organs. The genes are expressed in all regions of the brain, including the olfactory bulbs and tracts, telencephalon, diencephalon, optic tectum, vagal lobe, hypothalamus, cerebellum and medulla, and spinal cord. Gene expression has also been detected in the eyes, heart, thymus, liver, stomach, intestine, spleen, gill, gall bladder, muscle, kidney, head kidney, Brockmann bodies, skin, muscle, and gonads (23, 26, 28, 39, 45, 46). In rainbow trout, GHS-R1a-LR mRNA expression has been detected in blood leukocytes and head kidney leukocytes (47). Cypriniformes such as goldfish and zebrafish, as well as Siluriformes such as channel catfish, possess paralogs of GHS-Ra, each of which has different levels and patterns of expression (22, 39, 46).

In amphibians, strong GHS-R1a mRNA expression has been found in brain regions such as the diencephalon and mesencephalon; the stomach and testis; and to a lesser extent in the small and large intestines, adrenal gland, and kidney in the bullfrog (19). In the Japanese tree frog, GHS-R1a transcripts have been detected in almost all tissues examined, although relatively high expression was detected in the duodenum, small and large intestines, and ovary. However, unlike in other animals, pituitary expression was absent in both species (19).

In birds, GHS-R1a mRNA has also been detected in almost all tissues examined. GHS-R1a mRNA is expressed in chicken tissues such as the hypothalamus, telencephalon, cerebrum, cerebellum, optic lobes, brainstem, heart, lung, thymus, liver, spleen, pancreas, proventriculus, gizzard, duodenum, adrenal gland, kidney, gonads, breast muscle, subcutaneous fat, leg muscle, abdominal fat, and uropygial gland (29, 30, 40, 41, 44). Comparing the relative expression levels in these tissues is difficult; nonetheless,

the expression levels in the brain, gastrointestinal tract, liver, and spleen appear to be relatively high compared with other tissues, although strain differences may exist (29, 30, 33). In ducks, mRNA expression has been detected in the subcutaneous fat, hypothalamus, small intestine, testis, cerebellum, and cerebrum (44). In the Japanese quail, GHS-R1a mRNA expression was examined only in the gastrointestinal tract (32), where region-specific expression was detected at relatively high levels in the upper and lower intestines such as the esophagus, crop, and colon, but weak levels in the middle portions of the gastrointestinal tract (e.g., the proventriculus, duodenum, gizzard, jejunum, and ileum).

#### **EXPRESSION OF GHRELIN RECEPTOR ISOFORMS OTHER THAN GHS-Ra AND GHS-R1a-LR**

Growth hormone secretagogue-receptor type-1b is a splice variant of the mammalian GHS-R. In humans, its mRNA distribution is more widespread than that of GHS-R1a, and varies spatially and quantitatively from that of GHS-R1a (34). This suggests the possibility that GHS-R1b is involved in specific GHS-R1a-independent physiological activities, although these remain unknown.

In non-mammalian vertebrates, there are a few reports on the mRNA distribution of GHS-R1b. First, GHS-R1b mRNA has been detected in the brain of fish. In the black porgy, the level of expression was highest in the telencephalon, followed by the hypothalamus, pituitary, optic tectum, thalamus, and spinal cord, whereas little was detected in peripheral tissues (28). In Mozambique tilapia, the brain is the site with the highest expression of GHS-R1b mRNA, although transcripts were also detected in the stomach, adipose tissue, gill, liver, intestine, spleen, kidney, and muscle (26). In orange-spotted grouper, the expression levels of GHS-R1b mRNA were high in the pituitary, hypothalamus, cerebellum, medulla, spinal cord, gill filament, spleen, liver, stomach, head kidney, kidney, gonad, red muscle, skin, and fat body (45). In rainbow trout, GHS-R1b mRNA was strongly expressed in the pituitary, whereas weak expression was observed in the hypothalamus, pyloric appendage, middle intestine, spleen, and head kidney (23). In channel catfish, the expression level of GHS-R1b mRNA was highest in the pituitary, but it was approximately 400 times lower in most peripheral tissues compared with the expression level of GHS-R1a (39).

In birds, GHS-R1aV or GHS-Rtv mRNA expression was detected in almost all tissues examined, a pattern almost identical to that of GHS-R1a mRNA expression, although expression levels of each isoform differed (29, 30, 33). GHS-Rtv transcripts were first detected in chicken ovaries (31). In Japanese quail, the expression of the GHS-Rtv-like receptor was detected in the gastrointestinal tract but only in the proventriculus and gizzard (32). The function of these avian variants is entirely unknown.

# **REGULATION OF GHRELIN RECEPTOR EXPRESSION**

Satiation and hunger signals regulate *ghsr* expression. A condition of negative energy balance such as fasting increases GHS-R1a mRNA expression in the hypothalamus and pituitary of rats, while re-feeding restores the increased expression level to a normal level (48, 49). The gene expression of *ghsr* is affected by various hormonal factors, it is stimulated by ghrelin (5, 49–51), GH-releasing hormone (GHRH) (52), thyroid hormone (53), and glucocorticoid (dexamethasone) (54, 55). In contrast, it is inhibited by GH (56–58), leptin (49), glucocorticoid (50), and insulin-like growth factor-I (IGF-I) (59). These are summarized in **Table 3**.

Acute or chronic changes in the energy status or environmental conditions appear to have varying effects on *ghsr* expression in non-mammalian vertebrates (**Table 3**). In Mozambique tilapia, GHS-R1a-LR mRNA levels in the brain are unaffected by fasting, whereas GHS-R1b mRNA expression is increased (60). Peddu et al. (61) reported acute pre- and post-prandial changes in GHS-R1a-LR and GHS-R1b mRNA expression, whereas pre-GHS-R mRNA levels (immature mRNA, hetero-nuclear RNA) did not reflect changes in feeding status. Riley et al. (62) showed that acute increased blood glucose reduced GHS-R1a-LR mRNA levels in the brain and increased gastric ghrelin mRNA expression as well as plasma ghrelin levels. This change in plasma ghrelin levels is

#### **Table 3 | Regulation of ghrelin receptor expression.**


the opposite of that seen in humans or goldfish, where a glucose load decreases plasma ghrelin levels (63, 64). In conditions of chronic negative energy balance, there was no change in the GHS-R1a-LR expression levels in the brains of Atlantic salmon fasted for 14 days (65). In contrast, goldfish GHS-R1a-1 mRNA levels decreased in the vagal lobe and GHS-R1a-2 mRNA levels increased in the liver after 7 days of fasting (22). In bullfrogs, GHS-R1a mRNA expression was up-regulated in the stomach and ventral skin, whereas that in the brain did not change after 10 days of starvation (19). These results suggest that the nutritional condition of the body affects ghrelin receptor expression. Furthermore, GHS-R1a mRNA expression was up-regulated in the brain, stomach, and ventral skin after 10 days of dehydration of tree frog (19). This result may support the view that ghrelin is involved in the regulation of water balance in frogs, as seen in rats (66) and chicks (67).

Hormonal control of *ghsr* expression has been reported. Ghrelin appears to have a stimulatory effect on *ghsr* expression in non-mammalian vertebrates, as it does in mammals. However, the effects differ depending on the ghrelin form, receptor isoform, and target tissue. In channel catfish, the C-terminal structure of ghrelin affects *ghsr* expression (39). In the pituitary, catfish ghrelin-Gly (this is naturally occurring 23-AA ghrelin where Gly is extended at the C-terminus) increased the levels of GHS-R1a mRNA but not of GHS-R2a mRNA. In contrast, catfish ghrelin-amide (22- AA ghrelin with an amide structure at the C-terminus) had no effect on either receptor. In the Brockmann bodies, catfish ghrelinamide or ghrelin-Gly dramatically increased the GHS-R2a mRNA expression levels with different time courses. In zebrafish, goldfish ghrelin12-amide stimulated the mRNA expression of both GHS-R1a and 2a in the brain, but with different time courses (46). In orange-spotted grouper, rat ghrelin (10−<sup>5</sup> M) inhibited the expression of GHS-R1a-LR and GHS-R1b mRNA in the hypothalamus and pituitary (45). In chickens, Geelissen et al. (29) reported that ghrelin down-regulated GHS-R1a and GHS-R1aV mRNA expression in the pituitary *in vitro*. In another *in vitro* study, GHRP-6 stimulated the promoter activity of black porgy GHS-R1a-LR expressed in HEK293 cells (68).

The effects of GH or glucocorticoids on non-mammalian *ghsr* expression also vary depending on the GH species used, target tissue, and GHS-R isoform. In orange-spotted grouper, sea bream GH (10−<sup>7</sup> M) did not affect GHS-R1a-LR levels in the hypothalamus but reduced them in the pituitary, whereas it decreased GHS-R1b mRNA levels in both the hypothalamus and pituitary (45). In chickens, bovine GH and corticosterone decreased mRNA expression of both GHS-R1a and GHS-R1aV, but human GHRH1- 29 reduced only GHS-R1a mRNA expression in the pituitary *in vitro* (29).

Yeung et al. (68) analyzed the 5<sup>0</sup> -flanking region of *ghsr* in black porgy and identified a number of putative binding sites for transcription factors such as AP1, NF-1, Oct-1, and USF. Changes in *ghsr* expression during embryogenesis have been reported in orange-spotted grouper (45) and channel catfish (39). In both species, *ghsr* expression fluctuates depending on the embryonic stage, and the expression levels of GHS-R isoforms are separately regulated.

#### **SIGNALING PATHWAYS OF THE GHRELIN RECEPTOR**

Howard et al. (3) observed increases in intracellular Ca2<sup>+</sup> levels in cells transfected with GHS-R1a. The intracellular signaling of GHS-R1a is mediated by the activation of a G-protein subtype, Gaq/11, which induces the production of inositol triphosphate (IP3), release of Ca2+, and activation of protein kinase C (PKC)

(69). These events are seen in cells transfected with GHS-R1a as well as in somatotrophs (70–74).

In addition, GHS-R1a functions in an agonist-independent manner and causes high basal IP3 production in the absence of agonists, indicating that GHS-R1a is a constitutively active receptor (71, 74, 75). This activity in turn triggers phospholipase C (PLC)–PKC-dependent Ca2<sup>+</sup> mobilization, which is associated with the L-type voltage-gated calcium channel via PKC. Furthermore, extracellular signal-regulated kinase 1 and 2 (ERK1/2) are activated by GHRP-6. A GHS-R antagonist (d-Lys3)-GHRP-6, was shown to inhibit basal PLC and ERK1/2 activity (76).

When a non-mammalian ghrelin receptor was expressed in mammalian cells, a rise in intracellular Ca2<sup>+</sup> was observed with ghrelin or GHSs (19, 22, 27, 28, 32, 77, 78). A similar Ca2<sup>+</sup> mobilization was also induced by ghrelin in the primary culture of goldfish pituitary cells (79, 80), which was important for inducing the release of GH and luteinizing hormone (LH) from goldfish somatotrophs (79) and gonadotrophs (80), respectively. Little is known about the intracellular signaling pathways involved.

In addition to binding ghrelin, non-mammalian ghrelin receptors are capable of binding GHSs such as GHRP-2 and GHRP-6; ipamorelin; and L163,255, L692,585, and L163,540, although the agonistic activity varies according to the receptor present in each animal (19, 22, 27, 28, 32, 77). In addition, a GHS-R1a antagonist (d-Lys3)-GHRP-6, is also capable of inhibiting ghrelin binding to the receptor (22). These results indicate that the structural interactions between the ligand and the AAs of the receptor essential for ligand binding and receptor activation are conserved among vertebrates. However, ligand selectivity has been found in the case of GHRP-6 and hexarelin for goldfish GHS-R1a-1, 1a-2, and 2a-2 (**Figure 5**) (22).

In fish-specific GHS-R1a-LRs, particularly of the pufferfish and black porgy, pharmacological doses of receptor agonists are required in some cases to activate the receptors (27, 28), whereas no reaction was found at all in the receptors in tilapia and rainbow trout, even with homologous ghrelin (23, 26). The reason behind this phenomenon remains to be elucidated.

Receptor functionality has not been examined in the African clawed frog or teleosts such as channel catfish, zebrafish, and Jian carp where GHS-Ra has been identified. We expect that these receptors will be responsive to ghrelin or GHS because of their structural properties, such as the short ECL2 loop (**Figure 4**). However, confirmation of these receptor activities will be required to test this hypothesis in the future.

#### **KEY AMINO ACIDS RELATED TO LIGAND SELECTIVITY AND RECEPTOR FUNCTIONALITY IN THE GHRELIN RECEPTOR STRUCTURE**

Feighner et al. (81) reported key AAs that play essential roles in GHS-R1a activation on the basis of the structure of human GHS-R1a and three types of GHSs with different structures, i.e., MK-0677, GHRP-6, and L692,585. Their results showed that D99, C116, E124, M213, S217, and H280 in human GHS-R1a have crucial roles in receptor activation. In particular, M213 is required for the binding of GHRP-6 and L692,585. S217 and H280 are specifically involved with the binding of GHRP-6. In ghrelin receptors identified in non-mammalian vertebrates, all of the AAs listed

**FIGURE 5 | Ligand selectivity and intracellular Ca<sup>2</sup>**<sup>+</sup> **signaling in four goldfish ghrelin receptors**. Four goldfish ghrelin receptors exhibited different ligand selectivity. The schematic figures above show the strength of the ligand-receptor affinity based on the thickness of the arrow, while the bar graphs below show the maximum value of the stimulated increase in the intracellular Ca<sup>2</sup><sup>+</sup> signal. Goldfish ghrelin (gfGHRL) 12-C8 (octanoylated ghrelin with 12 amino acids, AAs), 17-C8 (octanoylated ghrelin with 17 AAs), and 17-C10 (decanoylated ghrelin with 17 AAs); rat ghrelin (rGHRL); and two

GHSs, GHRP-6 and hexarelin, were used in the experiment. For example, the arrows indicate that the intracellular Ca<sup>2</sup><sup>+</sup> increased in cells expressing GHS-R1a-1 after exposure to gfGHRL12-C8, 17-C8, and 17-C10; rat ghrelin; and hexarelin, but not after exposure to GHRP-6 at a similar dose. The corresponding bar graph shows that gfGHRL17-C10 increased Ca<sup>2</sup><sup>+</sup> much more strongly than the other agonists. Furthermore, although GHS-R2a-2 was capable of binding all of the agonists examined at a low dose, none of the agonists increased the intracellular Ca<sup>2</sup><sup>+</sup> level.

above are conserved, with the exception of an AA that is equivalent to S217 in the stickleback receptor (**Figure 3**). This may suggest that the GHS-Ra and GHS-R1a-LR identified in nonmammalian vertebrates have the ability to bind GHSs. However, as described earlier, goldfish GHS-Ra has ligand selectivity (22). In addition, the GHS-R1a-LR in rainbow trout and tilapia shows no Ca2<sup>+</sup> response in receptor-expressing mammalian cells (23, 26). Although AAs equivalent to M213, S217, and H280, which are essential for binding of GHRP-6 to the receptor, are all conserved in goldfish GHS-Ra, GHRP-6 does not increase the intracellular Ca2<sup>+</sup> in HEK 293 cells expressing goldfish GHS-R1a-1 and 1a-2. Thus, the interaction between the ligand and key AAs in the receptor related to ligand binding may be more complicated than anticipated.

Holst et al. (82) found that the ghrelin receptor elicited strong, ligand-independent signaling in transfected COS-7 or HEK293 cells. Independent of ligand selectivity, the relationship between constitutive receptor activity and the AA composition of the receptor has also been examined (83–85). These studies suggest that of the AAs in human GHS-R1a, V160, F279, A204, I134, and A204 are important for controlling constitutive receptor activity (**Figure 3**). These AAs are conserved in the GHS-Ra and GHS-R1a-LR identified in non-mammalian vertebrates (**Figure 3**); therefore, all of them may be constitutively active receptors, although their activity has been confirmed only in the black porgy receptor (86).

# **PHYSIOLOGICAL FUNCTION OF GHS-Rs**

GHS-R1a mediates the information conveyed by ghrelin and elicits various physiological functions. In addition to its hypophysiotropic effects and regulation of appetite, ghrelin affects many physiological functions, including gastrointestinal motility, cardiovascular performance, cell proliferation, immune function, bone metabolism, sleep, and the promotion of learning and memory (9, 87, 88). Recent evidence suggests that ghrelin functions as a blood glucose regulator (89).

#### **ROLES OF GHS-R1a AND 2a**

Growth hormone secretagogue-receptor type-1a or 2a is thought to mediate various physiological functions of ghrelin, although direct evidence in non-mammalian vertebrates remains sparse. Recently, Yahashi et al. (90) reported that the peripheral effects of ghrelin on food intake and locomotor activity in goldfish are mediated via one of the four ghrelin receptor isoforms, GHS-R2a-1. In addition, ghrelin has the ability to stimulate GH and LH release from goldfish pituitary (64, 79, 80, 91). GHS-R1a-2 mRNA shows the most abundant expression in this structure, suggesting that the receptor is involved in the regulation of pituitary hormone release. Changes in GHS-R1a or 2a expression depending on the energy state suggest the involvement of ghrelin in energy homeostasis, as observed in frogs and goldfish (19, 22). However, no change was observed in the case of tilapia (60). In chickens and quails, the distributions of the receptor are consistent with its role

in gut contraction (32). However, although the ghrelin receptor is expressed throughout the intestinal tracts of goldfish and rainbow trout, ghrelin has no effects on intestinal motility (92). This result is in contrast to that seen in zebrafish, in which rat and human ghrelin stimulate gut contraction (93). Further studies are necessary to determine the nature of the relationship between ghrelin receptors and physiological function.

#### **ROLES OF GHS-R1b**

In contrast with GHS-R1a, little is known about the functions of the GHS-R1b isoform. Mammalian and non-mammalian GHS-R1b show no apparent intracellular Ca2<sup>+</sup> signaling response to ghrelin or GHSs (32, 86). Co-expression of GHS-R1a and 1b reduces the signaling capacity of GHS-R1a via heterodimerization (28, 86, 94), suggesting that GHS-R1b acts as a dominant-negative mutant during signaling via GHS-R1a (86). Intriguingly, GHS-R1b forms heterodimeric associations with other GPCRs such as neurotensin receptor 1 (NTSR1) (95). This heterodimeric receptor binds to peptide hormones other than ghrelin and affects intracellular signaling, i.e., the GHS-R1b/NTSR1 heterodimer binds neuromedin-U and induces cAMP production instead of Ca2<sup>+</sup> signaling.

Although GHS-R1b exists in the same gene as GHS-R1a, the sites, patterns, levels, and regulation of GHS-R1b expression differ from those of GHS-R1a. Therefore, elucidation of the physiological function of the receptor is awaited.

#### **PERSPECTIVE**

In this review, we assembled current knowledge about ghrelin receptors in non-mammalian vertebrates. Many questions remain unanswered because receptor genes have been identified only in a limited number of species. However, the functional importance of the ghrelin system is gradually becoming understood in species where the receptor distribution is clear. Presence of unique GHS-Rs such as GHS-R2a, GHS-R1a-LR, or variants found only in non-mammalian vertebrates are interesting in the divergence of the ghrelin system; therefore, examining the structural relationship and function of non-mammalian GHS-Rs based on comparisons with mammalian GHS-Rs is important for understanding the significance of the ghrelin system in vertebrates. However, the ghrelin system of an animal studied may also need to be considered without preconceptions or making comparisons with mammalian data. Thus, the study of non-mammalian GHS-Rs should be interesting and attract many researchers in the future.

#### **ACKNOWLEDGMENTS**

We thank Dr. Christopher A. Loretz (University of Buffalo, Buffalo, NY, USA) for valuable comments on this manuscript. We thank Mrs. Azumi Ooyama for excellent technical assistance. Hiroyuki Kaiya, Mikiya Miyazato, and Kenji Kangawa were supported by a Grant-in-Aid for Scientific Research from the Ministry of Education,Culture,Science,Sports,and Technology (MEXT,KAKENHI) of Japan and by the Takeda Science Foundation.

#### **REFERENCES**


hexapeptide that acts on the pituitary to specifically release growth hormone. *Endocrinology* (1984) **114**:1537–45. doi:10.1210/endo-114-5-1537


is modified by *n*-octanoic acid at its third threonine residue. *J Biol Chem* (2001) **276**:40441–8. doi:10.1074/jbc.M105212200


*Comp Neurol* (2006) **494**:528–48. doi:10.1002/cne.20823


*rerio*) brain. *Comp Biochem Physiol A Mol Integr Physiol* (2010) **156**:190–200. doi:10.1016/j.cbpa.2010.01.019


**12**:481–5. doi:10.1046/j.1365- 2826.2000.00446.x


*Oreochromis mossambicus*. *Comp Biochem Physiol A Mol Integr Physiol* (2009) **154**:541–6. doi:10.1016/j.cbpa.2009.08.018

63. Shiiya T, Nakazato M, Mizuta M, Date Y, Mondal MS, Tanaka M, et al. Plasma ghrelin levels in lean and obese humans and the effect of glucose on ghrelin secretion. *J Clin Endocrinol Metab* (2002) **87**:240–4. doi:10.1210/jc.87.1.240


in rat somatotropes by two mechanisms. *Endocrinology* (1994) **135**:1100–8. doi:10.1210/en.135.3.1100


*Endocrinol* (2009) **160**:148–57. doi:10.1016/j.ygcen.2008.11.006


memory facilitation implicates nitric oxide synthase activation and decrease in the threshold to promote LTP in hippocampal dentate gyrus. *Physiol Behav* (2010) **101**:117– 23. doi:10.1016/j.physbeh.2010. 04.026


does not affect gastrointestinal contractility in rainbow trout and goldfish in vitro. *Gen Comp Endocrinol* (2012) **178**:539–45. doi:10.1016/j.ygcen


U-growth hormone secretagogue receptor 1b/neurotensin receptor 1 oncogenic signaling pathway as a therapeutic target for lung cancer. *Cancer Res* (2006) **66**:9408–19. doi:10.1158/ 0008-5472.CAN-06-1349


100. Yu HJ. (2011, direct submission, database only).

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 27 November 2012; paper pending published: 10 December 2012; accepted: 20 June 2013; published online: 17 July 2013.*

*Citation: Kaiya H, Kangawa K and Miyazato M (2013) Ghrelin receptors in non-mammalian vertebrates. Front. Endocrinol. 4:81. doi: 10.3389/fendo.2013.00081*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Kaiya, Kangawa and Miyazato. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Ancient grandeur of the vertebrate neuropeptideY system shown by the coelacanth Latimeria chalumnae

# **Dan Larhammar \* and Christina A. Bergqvist**

Unit of Pharmacology, Department of Neuroscience, Science for Life Laboratory – Uppsala University, Uppsala, Sweden

#### **Edited by:**

Jae Young Seong, Korea University, South Korea

#### **Reviewed by:**

Hervé Tostivint, Natural History Museum, France Stacia A. Sower, University of New Hampshire, USA

#### **\*Correspondence:**

Dan Larhammar, Unit of Pharmacology, Department of Neuroscience, Science for Life Laboratory – Uppsala University, Box 593, SE-751 24 Uppsala, Sweden. e-mail: dan.larhammar@neuro.uu.se

The neuropeptide Y (NPY) family receptors and peptides have previously been characterized in several tetrapods, teleost fishes, and in a holocephalan cartilaginous fish. This has shown that the ancestral NPY system in the jawed vertebrates consisted of the peptides NPY and peptide YY (PYY) and seven G-protein-coupled receptors named Y1–Y8 (Y3 does not exist). The different vertebrate lineages have subsequently lost or gained a few receptor genes. For instance, the human genome has lost three of the seven receptors while the zebrafish has lost two and gained two receptor genes. Here we describe the NPY system of a representative of an early diverging lineage among the sarcopterygians, the West Indian Ocean coelacanth Latimeria chalumnae. The coelacanth was found to have retained all seven receptors from the ancestral jawed vertebrate. The receptors display the typical characteristics found in other vertebrates. Interestingly, the coelacanth was found to have the local duplicate of the PYY gene, called pancreatic polypeptide, previously only identified in tetrapods. Thus, this duplication took place very early in the sarcopterygian lineage, before the origin of tetrapods. These findings confirm the ancient complexity of the NPY system and show that mammals have lost more NPY receptors than any other vertebrate lineage. The coelacanth has all three peptides found in tetrapods and has retained the ancestral jawed vertebrate receptor repertoire with neither gains or losses.

**Keywords: G-protein-coupled receptor, neuropeptide Y, peptide YY, pancreatic polypeptide, coelacanth, Latimeria chalumnae**

#### **INTRODUCTION**

Neuropeptide Y (NPY) and its related peptides named peptide YY (PYY) and pancreatic polypeptide (PP) comprise a system of neuronal and endocrine peptides that act on several G-proteincoupled receptors in vertebrates. They are involved in the regulation of a broad range of functions including appetite/satiety, gut motility, cardiovascular activity, pituitary release of hormones, circadian rhythm, and many more (see Pedrazzini et al., 2003; Brumovsky et al., 2007; Mercer et al., 2011; Zhang et al., 2011 for reviews). NPY and PYY arose by duplication of a common ancestral peptide gene before the vertebrate radiation (Larhammar, 1996). In mammals NPY is almost exclusively neuronal whereas PYY is primarily expressed in endocrine cells in the gastrointestinal tract. In ray finned fishes, PYY too is expressed in the nervous system (Cerdá-Reverter et al., 2000; Söderberg et al., 2000). PP is a local duplicate of PYY (Hort et al., 1995) previously found only in tetrapods (Larhammar, 1996; Cerdá-Reverter and Larhammar, 2000; Conlon, 2002; Sundstrom et al., 2008) and is expressed in pancreatic islets.

Mammals generally possess four receptors for the NPY-family peptides, namely subtypes Y1, Y2, Y4, and Y5 (Larhammar and Salaneck, 2004). Subtype Y3 was postulated from pharmacological experiments but does not exist as a separate gene (Herzog et al., 1993; Jazin et al., 1993). A fifth receptor gene, Y6, is expressed in a few mammals such as mouse and rabbit but is a pseudogene in many others including primates (Matsumoto et al., 1996), pig (Wraith et al.,2000), and guinea-pig (Starbäck et al.,2000).With its

three peptides and four receptors, the NPY system in humans and most other mammals displays a degree of complexity resembling many other vertebrate peptide-receptor systems for neuronal and endocrine peptides, for instance the melanocortin system (Dores and Baron, 2011; Liang et al., 2013), the opioid system (Sundstrom et al., 2010), the oxytocin-vasopressin system (Ocampo Daza et al., 2011; Yamaguchi et al., 2012), and the somatostatin-cortistatin system (Tostivint et al., 2008; Ocampo Daza et al., 2012). However, previous evolutionary studies of the NPY receptor family have shown that a larger number of receptors existed in the early stages of vertebrate evolution before the emergence of jawed vertebrates,*Gnathostomata*: by sequence-based phylogenetic analyses and comparison of gene locations on chromosomes, we were able to deduce an ancestral vertebrate set of no less than seven NPY-family receptors (Larhammar and Salaneck, 2004; Larsson et al., 2008), more than for any other known peptide-receptor family. Subsequently, this repertoire was confirmed by our identification of all seven receptor subtypes in a cartilaginous fish (*Chondrichthyes*), the holocephalan elephant shark, or ghost shark, *Callorhinchus milii* (Larsson et al., 2009). Thus, the following evolutionary scenario was corroborated: an ancestral pre-vertebrate chromosome carried the genes for a Y1-like, a Y2-like, and a Y5 like receptor subtype. The two basal vertebrate tetraploidizations (Nakatani et al., 2007; Putnam et al., 2008) quadrupled this chromosome, thereby generating four similar chromosomal regions that probably had as many as 12 (4 × 3) family members, unless some were lost already after the first tetraploidization. In extant


**Table 1 | The table lists Ensemble gene ID or NCBI accession number for the coelacanth NPY-family receptor and peptide genes as well as the orthologs genes in human, chicken, zebrafish, and elephant shark.**

For Y6, the mouse sequence is included instead of the human pseudogene. The human somatostatin receptor SSTR1 that was used as outgroup in the phylogenetic tree is also included.

The coelacanth entries are shown in bold.

vertebrate lineages a total of seven family members have been found to remain: four Y1-subfamily genes with one on each of the four chromosomes resulting from the tetraploidizations (Y1, Y4, Y6, and Y8), two Y2-like genes (Y2 linked to Y1, Y7 linked to Y6), and a single surviving Y5 gene (linked to Y1 and Y2) (Larsson et al., 2008, 2009).

Of the vertebrates investigated to date only the elephant shark has maintained all seven of these ancestral vertebrate receptor genes (Larsson et al., 2009). All of the other gnathostome lineages seem to have suffered losses, although some of the genome databases may be incomplete. Among amphibians, the western clawed frog *Silurana (Xenopus) tropicalis* seems to have lost Y6 which appears to be a pseudogene in the frog *Pelophylax esculentus* (previously called *Rana esculenta*; Sundstrom et al., 2012). Amniotes have lost Y8, or possibly this gene was lost independently in birds and mammals (Larsson et al., 2009). The mammalian lineage subsequently lost also Y7. The Y6 gene is a pseudogene in several mammals as mentioned above. In the large and heterogeneous group of rayfinned fishes (*Actinopterygii*), the teleosts are the most carefully studied. In the true teleost species (*Euteleostei*) with sequenced genomes, the receptors Y1, Y2, Y4, Y7, and Y8, as well as a duplicate of Y8 called Y8b, have been identified, albeit only the zebrafish genome contains all of these (Larsson et al., 2008; Salaneck et al., 2008). Thus, the Y8 duplicate seems to be the only surviving copy resulting from the teleost-specific third tetraploidization, 3R (Jaillon et al., 2004). Also, this suggests that Y5 and Y6 are missing in *Euteleostei*. A local duplicate of Y2, named Y2-2, has been found in zebrafish and medaka (Fallmar et al., 2011). In addition, Y5 and Y6 have been identified in more basally diverging rayfinned fish species, namely a sturgeon and a bichir, as well as an early diverging teleost, the silver arowana, *Osteoglossum bicirrhosum* (Salaneck et al., 2008). Also,we have previously cloned the genes for Y5 and Y6 from the coelacanth *Latimeria chalumnae* (Lch; Larsson et al., 2007). The receptor gene duplication scenario is further supported by findings in lampreys of members of the Y1, Y2, and Y5 lineages (Salaneck et al., 2001; Larsson et al., 2009; Xu et al., 2012). In summary, of all gnathostome species that have been previously investigated only the elephant shark seems to have retained the complete ancestral repertoire of seven NPY receptor genes.

TheWest Indian Ocean coelacanth, Lch, is an important species for studies of vertebrate evolution as it is one of only two extant and closely related coelacanth species representing a very early *Actinistia* branch among the lobe finned fishes, *Sarcopterygii*. The relationship of coelacanths to lungfishes (Dipnoi) and tetrapods has been difficult to resolve. Recent data suggest that the coelacanths are slightly more closely related to the lungfishes than either of these groups is to tetrapods (Shan and Gras, 2011). The Actinopterygii-Sarcopterygii divergence took place approximately 424 million years ago (Mya) according to a recent estimate based on molecular data (Chen et al., 2012). The coelacanth and tetrapod lineages may have parted only a few tens of millions of years later, well before the split of the amphibian and amniote lineages approximately 350 Mya. Thus, it is of great interest to see which genetic and phenotypic Sarcopterygian characters had already arisen before the coelacanth-tetrapod divergence. In September 2011, a new genome assembly became available for the coelacanth. We present here a genomic analysis of the NPY system in Lch,identifying not only the receptor repertoire but also the genes for the three peptide ligands that bind to these receptors.

# **MATERIALS AND METHODS**

The PreEnsembl sequence database (ENSEMBL LatCha1 released September 2011, database version 69.1) for the West Indian Ocean coelacanth Lch was searched with Blast/Blat using as query sequences the human NPY-family receptors Y1, Y2, and Y4, and the zebrafish receptors Y7 and 8b. Five new receptors were identified, all but Y4 are annotated in Ensembl68. Amino acid sequences for NPY receptors from the following species were used for alignments: human (*Homo sapiens*), chicken (*Gallus gallus*), zebrafish (*Danio rerio*), elephant shark (*C. milii*), mouse (*Mus musculus*), and coelacanth (*L. chalumnae*). A list of genes with corresponding accession numbers is provided in **Table 1** with the coelacanth NPY receptor sequences in bold.

Amino acid alignments were made in Jalview 2.8 version 14.0 (Waterhouse et al., 2009) using the MUSCLE and Clustal W web tool with standard settings. Phylogenetic Neighbor-Joining (NJ; Saitou and Nei, 1987) trees were made by using Clustal X version 2.0.12 (Larkin et al., 2007), standard settings and 1000 bootstrap replicates were applied. The tree shown in **Figure 2** was rooted with the human somatostatin receptor sequence SSTR1. Phylogenetic trees were also made with the Maximum Likelihood (ML) method (Guindon et al., 2010).


**FIGURE 3 | Organization of theY1 andY5 genes relative to one another in coelacanth, human, and chicken.** Exons are shown as tall boxes for the coding region and low boxes for the 5<sup>0</sup> - and 3<sup>0</sup> -untranslated regions where known. The 3<sup>0</sup> -untranslated region is shown until the position of the first consensus poly(A) signal. The size of the intron in the

coding region of Y1 is shown. In all three species, the genes have a "head-to-head" orientation and may share regulatory upstream elements. The distance between the coding regions of the two genes are shown, but it has not yet been fully explored where promoters and any separate 5 0 -untranslated exons may be located.

### **RESULTS**

#### **LATIMERIA CHALUMNAE NPY-FAMILY RECEPTORS**

We have previously reported the coelacanth Y5 and Y6 genes (Larsson et al., 2007). To identify additional NPY-family receptors, we searched the PreEnsemble Lch database with sequences from other species as queries. The identified Lch sequences were aligned with known NPY-family receptors from other vertebrates and subjected to phylogenetic analyses. Some of the sequences had to be manually curated as indicated in **Table 1** which lists all the identified Lch sequences and their Ensemble ID or NCBI accession numbers. The table also lists the sequences for the other species that are included in the alignment and the phylogenetic tree described below. The searches allowed us to identify all the NPY-family receptors known to have arisen before the divergence of the jawed vertebrates, i.e., also Y1, Y2, Y4, Y7, and Y8.

An alignment of all Lch and human NPY-family receptors is shown in **Figure 1**. All of the sequences contain one to three consensus sequences for *N*-linked glycosylation in the aminoterminal region of the receptors, before transmembrane region 1 (TM1), i.e., the sequence NXS/T. The Lch Y6 and Y8 sequences also have a consensus glycosylation sequence in extracellular loop 2 (EL2), as do human Y1 and Y4. All sequences have a cysteine residue in EL1 and one cysteine in EL2 that presumably form a disulfide bond. The four Y1-subfamily sequences Y1, Y4, Y6, and Y8 also have a cysteine in the aminoterminal region and one in EL3, expected to form an additional disulfide bond, as in Y1-subfamily sequences from other species. Finally, one ore more cysteines are present in the cytoplasmic tail after TM7, expected to serve as attachment sites for palmitate to anchor the tail to the inner side of the cytoplasmic membrane. Also many other positions that are known to be highly conserved among NPY-family receptors, either between species or between receptor subtypes (or both), are conserved in the Lch sequences.

Phylogenetic analyses were performed with a more extensive alignment that included also the sequences of the six chicken



The coelacanth entries are shown in bold.

receptors and the seven elephant shark receptors, plus the mouse Y6 sequence as the human Y6 gene is a pseudogene. The resulting tree is shown in **Figure 2**. The seven Lch sequences could be clearly identified as orthologs to each of the seven previously described receptors in jawed vertebrates. The tree has a few minor deviations from the established order of divergence for the included species, mostly because of higher evolutionary rates for some lineages as explained in the discussion.

In all other vertebrates with assembled genomes, the genes for Y1 and Y5 are located close together in a head-to-head orientation, implying that their promoter regions may share some regulatory elements. The distance between the start codons for the human genes is 26.7 kb and for the chicken genes 16.8 kb. In the coelacanth, the distance is 23.6 kb (**Figure 3**). The promoter regions have not been characterized in detail for Y1 and Y5 in any species and furthermore seem to vary between species with

**FIGURE 4 | Pairwise prepropeptide sequence alignments between coelacanth and human for NPY, PYY, and PP (pancreatic polypeptide).** Identical positions are marked with dots for each pairwise comparison. The mature peptide sequences are marked with the horizontal line and the positions are numbered 1 through 36.

regard to the possibility of multiple promoters and alternatively spliced 50UTR exons (Wraith, 1999), why a detailed comparison with Lch is not meaningful at this point. The Y1 gene is the only member of the NPY receptor family that has an intron in the coding region. This intron is very small in mammals (97 bp in the human gene, 110 bp in opossum), chicken (121 bp), anole lizard (698 bp), and the frog *Silurana tropicalis* (92 bp; Sundstrom et al., 2012), but considerably larger in zebrafish with approximately 40 kb (in preparation), a sturgeon (Salaneck et al., 2008), and the two cartilaginous fishes spiny dogfish (Salaneck et al., 2003) and elephant shark (Larsson et al., 2009), the last-mentioned having an intron >3 kb. In the coelacanth, this intron is approximately 2.5 kb (**Figure 3**). Thus, the intron is large both in cartilaginous fishes, a teleost, and the coelacanth. The most parsimonious interpretation is that large size is ancestral and that the intron contracted in the ancestor of tetrapods and has remained small in this lineage.

The synteny groups of the NPY-family receptor genes and their neighboring genes in the same chromosome regions could not be analyzed in the coelacanth due to the small size of the scaffolds as a result of low-coverage sequencing.

# **LATIMERIA CHALUMNAE NPY-FAMILY PEPTIDES**

The coelacanth genome database was searched with the three human members of the NPY peptide family as queries. Hits were found for the most highly conserved parts of both NPY, PYY, and somewhat surprisingly also PP, although this peptide has not previously been identified outside of the tetrapods. Further analyses revealed the presence of the two small 3<sup>0</sup> exons encoding the carboxyterminal extensions of the prepropeptides for NPY and PYY. However, for the more rapidly evolving PP, these exons could not be found, most probably because they are missing in the Lch assembly. The Ensemble gene IDs are listed in **Table 2**.

An alignment of the coelacanth and human prepropeptides is shown in **Figure 4**. The mature NPY sequences share 31 och the 36 positions and three of the remaining five are highly conservative replacements, as has been observed for NPY for other vertebrates (Larhammar, 1996; Cerdá-Reverter and Larhammar, 2000). The mature Lch and human PYY sequences share 27 of the 36 positions. The coelacanth PYY sequence is identical to the deduced ancestral vertebrate PYY sequence (not shown), thus the differences are due to divergence of PYY in the lineage leading to humans and other mammals. In contrast, PP has diverged

considerably between coelacanth and human and the two species share only 18 of the 36 residues, a degree of identity found also for comparisons between birds or frogs and mammals (Larhammar, 1996; Cerdá-Reverter and Larhammar, 2000; Conlon, 2002). For both NPY and PYY, the differences that exist between coelacanth and human are predominantly located in the middle part of the peptides, suggesting that both ends are kept conserved for functional reasons, presumably interactions with the NPY-family receptors.

The genes for Lch NPY and PYY have the coding region distributed over three exons like in all other vertebrates (**Figure 5**). The introns in the coelacanth PYY gene are much larger, >13.75 and 2.2 kb, respectively, than in the mammalian orthologs, with only 106 and 128 bp in human. The introns in chicken and zebrafish are of intermediate size with 2.2 and 0.3 kb in chicken and similar sizes in zebrafish (Söderberg et al., 2000). For the PP gene, only the exon containing most of the mature PP sequence, preceded by the signal peptide, could be identified. In chicken and mammals, the PYY and PP genes are located in tandem approximately 10 kb apart. The intergenic distance is of a similar magnitude or greater in the coelacanth (the genomic sequence has a gap in the assembly). All three peptide genes have a separate exon for 50UTR in mammals and this exon has not been identified in any of the coelacanth genes.

#### **DISCUSSION**

The seven NPY-family receptor genes identified in the coelacanth genome database display orthology to the subtypes previously deduced to have arisen in the ancestor of the jawed vertebrates, namely subtypes Y1, Y2, Y4, Y5, Y6, Y7, and Y8 (**Figure 3**). Thus, the coelacanth has retained the complete ancestral repertoire like the elephant shark, but in contrast to mammals, chicken, and teleost fishes, all of which has lost one or more of the ancestral receptors. Our findings in the coelacanth corroborate our previous gene duplication scenario for the family of NPY receptors (**Figure 6**), with an ancestral local gene triplication, defining the three subfamilies of Y1, Y2, and Y5, followed by the chromosome quadruplication resulting from the two basal vertebrate tetraploidizations (Larhammar and Salaneck, 2004). Recent studies of other receptorfamiliesfor neuroendocrine peptide-receptors have also revealed surprising ancestral vertebrate complexity, with six ancient somatostatin receptors (Ocampo Daza et al., 2012), six ancestral oxytocin-vasopressin receptors (Ocampo Daza et al., 2011;Yamaguchi et al., 2012; Ocampo Daza, 2013), and four ancestral GnRH (gonadotropin-releasing hormone) receptors (Kim et al., 2011), followed by losses of genes in mammals as well as other vertebrate lineages.

The clades in the NJ tree shown in **Figure 2** corresponding to the seven receptor subtypes have high bootstrap support, except for the Y4–Y8 clade which often has been found to display poor resolution (Larsson et al., 2008, 2009) due to faster evolution of the Y4 sequences (Lundell et al., 2002) and the deviating amino acid composition of the zebrafish Y4 receptor, initially named Ya (Starbäck et al., 1999; see Salaneck et al., 2003). Analyses with the ML method give similar topology with this repertoire of species (not shown). Also when additional species are included for NJ and ML, these receptor subtypes fail to form completely separate

clades with high bootstrap values, but the orthology relationships between tetrapods and teleosts have been confirmed by conserved synteny for Y4 and Y8 (Larsson et al., 2008). The seven coelacanth receptors can clearly be identified as orthologs of the previously identified seven ancestral vertebrate receptors, as shown by the tree in **Figure 2**.

Within each clade in **Figure 2**, some deviations from the established evolutionary relationships between species can be seen except Y7 which does indeed conform to the conventional species topology. As has been observed for several other gene or protein families, the zebrafish branch is more basal than the elephant shark, namely for Y1 and Y2. This is probably due to a higher evolutionary rate in the teleost fish lineage as compared to cartilaginous fishes and sarcopterygians, as has been observed for conserved non-coding elements (Lee et al., 2011). Also the mouse Y6 branching can probably be accounted for by its higher evolutionary rate. Analyses using the ML method displays similar minor deviations from the species relationships (not shown).

The coelacanth genome project has insufficient coverage for assembly into large scaffolds and therefore does not allow analyses of conserved synteny. Nevertheless, the Y1 and Y5 genes were found to be located close together in a head-to-head orientation like in mammals and chicken, with a similar distance as in human, supporting coregulation of the genes. Y1 and Y5 are known to be coexpressed in brain regions of rat (Parker and Herzog, 1998, 1999; Wolak et al., 2003) and mouse (Naveilhan et al., 1998; Oberto et al., 2007). In the paraventricular region of the hypothalamus, both Y1 and Y5 are known to contribute to the appetite-stimulating effect of NPY as reviewed in Mercer et al. (2011), probably in slightly different ways (Lecklin et al., 2002, 2003). However, the most important promoters and regulatory elements remain to be functionally identified. Indeed, comparison of the pig promoter regions (Wraith, 1999) with those in human (Herzog et al., 1997) suggested that multiple alternative start sites and/or 50UTR exons exist that may differ even between these mammals, indicating complicated regulation of the two genes.

Based upon previous studies of the highly conserved peptides NPY and PYY, the identification of these in the coelacanth was expected and both of the mature peptides were found to be highly conserved. Indeed, the coelacanth PYY sequence appears to be identical to that of the deduced vertebrate ancestral PYY sequence (Larhammar, 1996), whereas mammalian PYY has undergone a few amino acid replacements. Possibly, the strong conservation of PYY in the coelacanth indicates a broader expression, like in teleost fishes and lamprey where PYY is expressed in both endocrine cells and neurons (Söderberg et al., 1994, 2000; Cerdá-Reverter et al., 2000; Kurokawa and Suzuki, 2002; Montpetit et al., 2005; Murashita et al., 2006, 2009; Wall and Volkoff, 2013), whereas mammalian PYY is almost exclusively expressed in endocrine cells. Broad distribution and/or usage has been found to correlate with higher conservation (Jordan et al., 2005; Khaitovich et al., 2005).

Our finding of PP in the coelacanth genome was unexpected because this peptide has previously only been identified in tetrapods (Larhammar, 1996; Cerdá-Reverter and Larhammar, 2000; Conlon, 2002; Sundstrom et al., 2008). The discovery of PP in the coelacanth could push back its origin by approximately 50 Myr, from a minimum of 340 Myr to a minimum of 390 Myr (Blair and Hedges, 2005). This may even raise the possibility that PP arose before the divergence of lobefinned fishes and rayfinned fishes, and that the PP gene has been lost in the rayfinned fish lineage. The NPY-family peptides so far described in teleosts have been assigned as NPYa, NPYb, PYYa, and PYYb (previously called

#### **REFERENCES**

Amores, A., Catchen, J., Ferrara, A., Fontenot, Q., and Postlethwait, J. H. (2011). Genome evolution and meiotic maps by massively parallel DNA sequencing: spotted gar, an outgroup

for the teleost genome duplication. *Genetics* 188, 799–808.

Berglund, M. M., Lundell, I., Eriksson, H., Söll, R., Beck-Sickinger, A. G., and Larhammar, D. (2001). Structure-activity analyses of the

PY), with the a and b duplicates having resulted from the teleost tetraploidization 3R (Sundstrom et al., 2008). PP in mammals binds to Y4 with higher affinity than either NPY or PYY, showing that partner preference has evolved between PP and Y4 in this lineage (Lundell et al.,1995,1996;Berglund et al.,2001). In chicken,in contrast, all three peptides bind to Y4 with similar affinities (Lundell et al., 2002). Interestingly, Lch PP does not have the change from the ancestral residue Gln-34 to Pro-34 that is found in all PP sequences, except a few bird sequences that have a histidine at this position and the caecilian *Typhlonectes natans* (an amphibian) which has a serine (Conlon et al., 1998; Conlon, 2002). The coelacanth has retained the ancestral Gln-34 found also in all NPY and PYY sequences. This may indicate that Lch PP has a binding profile that differs from PP in tetrapods with regard to receptor subtype preferences.

Detailed studies of peptide-receptor interactions will require cloning and functional expression of the coelacanth receptors and synthesis of the peptides, especially Lch PP. Hopefully future crystallization of NPY-family receptors will provide good templates for structural modeling of the coelacanth receptors and thereby help explain how this peptide-receptor system has evolved in the vertebrates. Unfortunately, studies of the tissue distribution of the mRNA for the NPY-family peptides and receptors cannot be easily performed with this endangered species.

The large repertoire of NPY-family receptors in the coelacanth may either mean that it has retained the ancestral functions for the seven vertebrate receptors or evolved novel functions for some of them. If ancestral functions are maintained, this might give clues to what functions may have disappeared in mammals with the loss of receptors Y7 and Y8, as well as Y6 in several mammalian lineages. Another possibility is that any such functions are partially or completely mediated by the four to five receptors that still exist in mammals. Information on such scenarios may be obtained from studies of species that represent additional basally diverging lineages, such as lungfishes (which unfortunately have very large genomes, thereby complicating bioinformatic analyses) and basal actinopterygians like the spotted gar whose genome was recently assembled (Amores et al., 2011). This species is especially interesting as it diverged before the third tetraploidization took place in the teleost lineage. Also basal lineages among the teleosts have recently added valuable information on gene family evolution, especially the recently assembled genome for the European eel (Henkel et al., 2012). Clearly considerable genetic complexity arose at an early stage in vertebrate evolution that can be deduced by comparing distantly related vertebrate lineages. An inevitable conclusion is that for several gene families, mammals must be considered to have degenerated by gene loss.

#### **ACKNOWLEDGMENTS**

This work was supported by a grant from the Swedish Research Council.

neuropeptide Y-family receptor Y4; studies of the human, rat, and the guinea pig receptors using NPY analogues and two distinct radioligands. *Peptides* 22, 351–356.


and sequence analysis of the neuropeptide Y receptors Y5 and Y6 in the coelacanth Latimeria chalumnae. *Gen. Comp. Endocrinol.* 150, 337–342.


receptor. *Proc. Natl. Acad. Sci. U.S.A.* 93, 5111–5115.


Larhammar and Bergqvist NPY receptors of coelacanth

a7e31d1a81174723ce69892e197e? pid=diva2:587276


lamprey. A potential ancestral gene. *Eur. J. Biochem.* 268, 6146–6154.


D. (2008). Evolution of the neuropeptide Y family: new genes by chromosome duplications in early vertebrates and in teleost fishes. *Gen. Comp. Endocrinol.* 155, 705–716.


deduced from the cloning of the five receptor subtype genes in pig. *Genome Res.* 10, 302–310.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 January 2013; accepted: 15 February 2013; published online: 08 March 2013.*

*Citation: Larhammar D and Bergqvist CA (2013) Ancient grandeur of the vertebrate neuropeptide Y system shown by the coelacanth Latimeria chalumnae. Front. Neurosci. 7:27. doi: 10.3389/fnins.2013.00027*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Neuroscience.*

*Copyright © 2013 Larhammar and Bergqvist. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Isolation of the bioactive peptides CCHamide-1 and CCHamide-2 from Drosophila and their putative role in appetite regulation as ligands for G protein-coupled receptors

*<sup>1</sup> <sup>2</sup> <sup>1</sup> <sup>3</sup> <sup>3</sup> Takanori Ida \*,Tomoko Takahashi , Hatsumi Tominaga ,Takahiro Sato , Hiroko Sano , Kazuhiko Kume4, Mamiko Ozaki 5,Tetsutaro Hiraguchi 5, Hajime Shiotani 5, Saki Terajima5,Yuki Nakamura5, Kenji Mori 6, MorikatsuYoshida6, Johji Kato7, Noboru Murakami 8, Mikiya Miyazato6, Kenji Kangawa6 and Masayasu Kojima<sup>3</sup>*


#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Young-Joon Kim, Gwangju Institute of Science and Technology, South Korea Liliane Schoofs, Catholic University of Leuven, Belgium

Christian Wegener, Universität Würzburg, Germany

#### *\*Correspondence:*

Takanori Ida, Interdisciplinary Research Organization, University of Miyazaki, Miyazaki 889-1692, Japan. e-mail: a0d203u@cc.miyazaki-u.ac.jp There are many orphan G protein-coupled receptors (GPCRs) for which ligands have not yet been identified. One such GPCR is the bombesin receptor subtype 3 (BRS-3). BRS-3 plays a role in the onset of diabetes and obesity. GPCRs in invertebrates are similar to those in vertebrates. Two Drosophila GPCRs (CG30106 and CG14593) belong to the BRS-3 phylogenetic subgroup. Here, we succeeded to biochemically purify the endogenous ligands of Drosophila CG30106 and CG14593 from whole Drosophila homogenates using functional assays with the reverse pharmacological technique, and identified their primary amino acid sequences. The purified ligands had been termed CCHamide-1 and CCHamide-2, although structurally identical to the peptides recently predicted from the genomic sequence searching. In addition, our biochemical characterization demonstrated two N-terminal extended forms of CCHamide-2. When administered to blowflies, CCHamide-2 increased their feeding motivation. Our results demonstrated these peptides actually present as the major components to activate these receptors in living Drosophila. Studies on the effects of CCHamides will facilitate the search for BRS-3 ligands.

**Keywords: GPCR, novel bioactive peptide,** *Drosophila***, CCHamide, bombesin receptor subtype 3**

"fendo-03-00177" — 2012/12/27 — 18:13 — page 1 — #1

#### **INTRODUCTION**

G protein-coupled receptors (GPCRs) constitute a large protein superfamily that shares a 7-transmembrane motif as a common structure. Human genome sequencing has identified several hundred orphan GPCRs for which ligands have not yet been identified (Vassilatis et al., 2003). GPCRs play crucial roles in cell-to-cell communication involved in a variety of physiological phenomena and are the most common target of pharmaceutical drugs. Therefore, the identification of endogenous ligands for orphan GPCRs will lead to clarification of novel physiological regulatory mechanisms and potentially facilitate the development of new GPCR-targeted therapeutics. Many bioactive molecules have been discovered or identified as endogenous ligands of orphan GPCRs through reverse pharmacology to date (Civelli et al., 2012). These molecules include nociceptin, prolactin-releasing peptide, orexin, apelin, ghrelin, metastin, and neuromedin S. The discovery of

novel endogenous ligands for orphan GPCRs in mammals is currently challenging, possibly because of the restricted timing of expression or distribution of GPCR ligands. One orphan receptor in mammals is the bombesin receptor subtype 3 (BRS-3). BRS-3 is primarily expressed in the hypothalamus and plays a role in the onset of diabetes and obesity (Ohki-Hamazaki et al., 1997). Although several small molecules that are agonists and antagonists for BRS-3 have been synthesized, the native ligand of BRS-3 has not yet been identified (Majumdar and Weber, 2012).

The recent sequencing of the *Drosophila melanogaster* genome has enabled the identification of at least 160 fly GPCRs (Brody and Cravchik, 2000). *Drosophila* is an excellent animal model for genetic analysis of developmental and behavioral processes, as it is a small, genetically modifiable organism with a relatively short lifecycle and can be bred easily under laboratory conditions. Structural or sequence comparison of newly discovered peptides in *Drosophila* with candidate molecules in mammals may lead to the discovery of new peptide signaling modules. We recently reported the discovery of dRYamide-1, dRYamide-2, and trissin as ligands

**Abbreviations:** BRS-3, bombesin receptor subtype 3; GPCR, G protein-coupled receptors.

for *Drosophila* orphan GPCRs (Ida et al., 2011a,b). We consider it likely that additional novel bioactive peptides can be discovered for orphan GPCRs. Two *Drosophila* GPCRs (CG14593 and CG30106) belong to the BRS-3 phylogenetic subgroup (Hewes and Taghert, 2001).

Here, we report the identification of CCHamide-1 and CCHamide-2, which are ligands for GPCRs CG30106 and CG14593, respectively, in *D. melanogaster*. Injection of CCHamide-2 resulted in the stimulation of feeding motivation in blowflies. These bioactive peptides may provide new insights in the search for BRS-3 ligands and the elucidation of *D. melanogaster* feeding mechanisms.

# **MATERIALS AND METHODS**

# **PURIFICATION OF** *Drosophila* **CCHamide-1 AND CCHamide-2**

An assay system using CG30106- or CG14593-expressing cells was prepared as previously described (Ida et al., 2011a,b). The fulllength cDNA of *Drosophila* CG30106 (GenBank accession number: NM\_136355; residues −31 to 1700) and CG14593 (GenBank accession number: NM\_136355; residues 656–2185) was obtained by RT-PCR using *Drosophila* cDNA as the template. The sense and antisense primers for CG30106 were 5- -aaatcgagcggactcagtacat-3 and 5- -gtggcctgtaattcctgtaaactc-3- , respectively. The sense and antisense primers for CG14593 were 5- -tgagacatcttgcccaggag-3 and 5- -gtgtttcggtacctccatttat-3- , respectively. The amplified cDNA was ligated into the pcDNA3.1 vector (Invitrogen). The expression vector, i.e., CG30106 or CG14593-pcDNA3.1, was transfected into Chinese hamster ovary (CHO) cells by using with Fugene6 transfection reagent (Roche), and stably expressing cells were selected using 1 mg/ml G418. The selected cell line, i.e., CHO-CG30106-line 2-4 or CHO-CG14593-line 10-1, showed the highest expression of CG3106 or CG14593 mRNA, respectively. Cells were cultured in a humidified environment of 95% air and 5% CO2. Changes in intracellular Ca<sup>2</sup><sup>+</sup> concentrations ([Ca2+]i) were measured using the FlexStation 3 fluorometric imaging plate reader to conduct high-throughput measurements of intracellular Ca2<sup>+</sup> concentration (Molecular Devices, CA, USA; Marshall et al., 2005). CHO-CG30106 or CHO-CG14593 cells (3 <sup>×</sup> 104 cells) were plated into 96-well black-wall microplates (Corning, NY, USA) 20 h before each assay. The cells were incubated with 100 μl of Calcium 4 assay kit reagent (Molecular Devices) for 1 h, and then 50 μl of each sample was added to the CHO-CG30106 or CHO-CG14593 cells to induce changes in fluorescence. The maximum [Ca2+]i changes were recorded.

*Drosophila melanogaster* flies (Canton S.; 350 g) were collected on dry ice. The whole body of each fly was boiled for 10 min in 10 volumes of water to inactivate intrinsic proteases. The solution was adjusted to 1 M AcOH. Peptides were extracted by homogenization using a Polytron mixer. The supernatant of the extracts, obtained after 30 min of centrifugation at 11,000 rpm, was concentrated to approximately 1/10 by an evaporator. The residual concentrate was subjected to acetone precipitation using 66% acetone. After the precipitates were removed, the supernatant acetone was evaporated and loaded onto a 40-g cartridge of Sep-Pak C18 (Waters), which was pre-equilibrated with 0.1% trifluoroacetic acid (TFA). The Sep-Pak cartridge was washed with 10% CH3CN/0.1% TFA, and then eluted with 60% CH3CN/0.1% TFA. The eluate was evaporated and lyophilized. The residual materials were redissolved in 1 M AcOH and then adsorbed on a column of SP-Sephadex C-25 (H+ form) that had been pre-equilibrated with 1 M AcOH. Successive elutions with 1 M AcOH, 2 M pyridine, and 2 M pyridine–AcOH (pH 5.0) provided three fractions of SP-I, SP-II, and SP-III. A basic peptide fraction (SP-III) was fractionated on a Sephadex G-50 gel filtration column (2.9 cm × 142 cm; GE Healthcare, Tokyo, Japan). A portion of each fraction, equivalent to 1.16 g of flies, was subjected to the assay using CHO-CG30106 or CHO-CG14593 cells. The active fraction was separated by carboxymethyl (CM)-ionexchange high-performance liquid chromatography (HPLC) on a TSK CM-2SW column (4.6 mm × 250 mm; Tosoh, Tokyo, Japan) with an ammonium formate (HCOONH4; pH 6.5) gradient of 10 mM to 1 M in the presence of 10% acetonitrile (ACN) at a flow rate of 1 ml/min for 160 min. The active fractions were separated by reverse-phase (RP)-HPLC with a μBondasphere C18 column (3.9 mm × 150 mm, Waters, MA, USA) by using a 10–60% ACN/0.1% TFA linear gradient at a flow rate of 1 ml/min for 80 min. The active fractions were further purified by RP-HPLC using a diphenyl column (2.1 mm × 150 mm, 219TP5125; Vydac, Hesperia, CA, USA) for 80 min by using a linear gradient of 10–60% ACN/0.1% TFA at a flow rate of 0.2 ml/min. Fractions corresponding to absorption peaks were collected, and an aliquot of each fraction (2 g tissue equivalent) was assayed by using the FLEX system. The active fractions were further purified by RP-HPLC by using a Chemcosorb 3ODSH column (2.1 mm × 75 mm; Chemco, Osaka, Japan) for 80 or 160 min by using a linear gradient of 10–60% ACN/0.1% TFA at a flow rate of 0.2 ml/min. Fractions corresponding to absorption peaks were collected, and an aliquot of each fraction (2 g tissue equivalent) was assayed by using the FLEX system. Approximately 20 pmol of the final purified peptides was analyzed using a protein sequencer (model 494; Applied Biosystems, CA, USA), and approximately 1 pmol of each active fraction was subjected to determination of molecular weight by matrix-assisted laser desorption–ionization time of flight (MALDI-TOF) mass spectrometry by using a Voyager-DE PRO instrument (Applied Biosystems).

### **CLONING OF** *Drosophila* **PREPRO-CCHamide-1 AND CCHamide-2 cDNA**

A tBLASTn search of the *Drosophila* genome resources was performed by using sequence of the purified peptides, and we obtained *D. melanogaster* mRNA sequences [CG14358 (CCHamide-1), NM\_001104314; and CG14375 (CCHamide-2), NM\_142028] derived from an annotated genomic sequence. We searched for open reading frames upstream and downstream of the genome sequences of CCHamide-1 and CCHamide-2 by using specific primers 5- -cgtgcagcttgcgaaataata-3 and 5- -cttctggcttagctagcgtgttatc-3 for CCHamide-1 and 5- -caccagccaagtgcaagtatc-3 and 5- -cggtttttaatgtacgttgtgg-3 for CCHamide-2. The candidate PCR product was subcloned into the pCR-II TOPO vector and sequenced. The nucleotide sequence of the isolated cDNA fragment was determined by automated sequencing (DNA sequencer model 3100; Applied Biosystems) according to the protocol for the BigDye terminator cycle sequencing kit (Applied Biosystems).

"fendo-03-00177" — 2012/12/27 — 18:13 — page 2 — #2

# **PEPTIDES**

CCHamide-1 (SCLEYGHSCWGAH-NH2), CCHamide-2 (GCQ-AYGHVCYGGH-NH2), CCHamide-1 C-terminal free (SCLEYG-HSCWGAH), CCHamide-2 C-terminal free (GCQAYGHV-CYGGH), long-form CCHamide-2 (AQQSQAKKGCQAYGHVC-YGGH-NH2), and long-form CCHamide-2 C-terminal free (AQQSQAKKGCQAYGHVCYGGH) were synthesized by Peptide Institute Inc. (Osaka, Japan).

# **PROBOSCIS EXTENSION REFLEX TEST FOR APPETITE MEASUREMENT**

The proboscis extension reflex (PER) test and feeding test were performed for the blowfly *Phormia regina* as previously described (Nisimura et al., 2005; Ida et al., 2011a). CCHamide-2 was dissolved in blowfly linger solution at a concentration of 10 pμol/ml. Twenty flies were secured by their wings using washing pins, and the first PER test was performed by using 12 steps of sucrose concentrations that had been prepared by twofold serial dilutions in distilled water, beginning from a sucrose concentration of 1 M. We investigated the PER in three different groups of 20 flies each: no injection, fly linger injection, and fly linger plus peptide injection. The PER tests were performed 30 min after 1 μl of blowfly linger solution with or without peptide was injected into the shoulder of each fly. We repeated five sets of PER tests each, in which 20 flies were used in each batch.

# **STATISTICAL ANALYSIS**

Results are presented as the mean ± SEM for each group. To compare the PER thresholds among the three groups, we used a non-parametric Steel–Dwass test. The criterion for statistical significance was *p* < 0.05 for all tests. The statistical software program GraphPad PRISM (GraphPad software, CA, USA) was used for analyses.

# **RESULTS**

# **STRUCTURAL DETERMINATION OF CCHamide-1 FOR CG30106**

[Ca2+]i assays were performed by using the gel filtration samples to isolate the endogenous ligands of CG30106 (**Figure 1A**). The active fractions were observed in eight sequential fractions (numbers 48–55). The fractions (51–55) with particularly high activity were separated by CM-ion-exchange HPLC at pH 6.5. The active fractions were separated by RP-HPLC. The active fraction was purified as a single peak in the final RP-HPLC (**Figure 1B**, P1). The amino acid sequence of the purified peptide was determined as SXLEYGHSXWGAH (P1; where X is a position that was not identified) using a protein sequencer. To elucidate the complete amino acid sequence of this peptide, *Drosophila* cDNA encoding the purified peptides was isolated by RT-PCR. The cDNA encoded a 182-residue protein (CG14358; **Figure 1C**) that contained features characteristic of an N-terminal signal peptide immediately preceding the purified peptide sequence. Every X residue was a cysteine, and the rest of the sequence was identical to that determined by peptide sequencing (**Figure 1C**). Sequencing resulted in a very low yield of phenyl thiohydantoin (PTH) at the steps involving X, which suggests that two cysteines may form disulfide bonds (S–S bonds). The preproprotein contained a potential processing site at the C-terminal end of the purified peptide sequence. This peptide contained Gly residues that presumably serve as an

amide donor for C-terminal amidation. We therefore deduced the primary structure of the peptide to be SCLEYGHSCWGAH-NH2. This peptide had been named CCHamide-1 (Roller et al., 2008). Mass spectrometric analysis revealed that the observed monoisotopic *m/z* value of the purified peptide (1445.30) was very similar to the theoretically predicted value for this peptide (1445.55) when including an intrachain disulfide bond and C-terminal amidation. We generated the synthetic peptide SCLEYGHSCWGAH-NH2 (CCHamide-1). The retention time of the P1 active fraction was identical to that of the synthetic SCLEYGHSCWGAH-NH2 peptide (which has an intrachain disulfide bond) on RP-HPLC (**Figure 1D**). Thus, these data suggest that both natural peptides have an intrachain disulfide bond and C-terminal amidation. **Figure 1E** shows the active fractions of each chromatography and the amino acid sequence of CCHamide-1.

# **STRUCTURAL DETERMINATION OF CCHamide-2 FOR CG14593**

The endogenous ligands of CG14593 were isolated in the same manner as those of CG30106 (**Figure 2A**). Three separate active fractions were revealed (**Figure 2G**, P2, P3, and P4), and each active fraction was purified as a single peak in the final RP-HPLC (**Figures 2B–D**). From the results obtained by using a protein sequencer and *Drosophila* cDNA encoding the purified peptides (**Figure 2E**), we deduced the primary structure of the peptide to be AQQSQAKKGCQAYGHVCYGGH-NH2 (P2), GCQAYGHVCYGGH-NH2 (P3), and KKGCQAYGHVCYGGH-NH2 (P4; **Figure 2G**). All of these cysteines may form S–S bonds. The shortest peptide (P3) had been named CCHamide-2. The cDNA encoded a 136-residue protein (CG14375; **Figure 2E**) that contained features characteristic of an N-terminal signal peptide immediately preceding the purified longest peptide sequence (P2). All peptides were derived from the same precursor (CG14375), but the length of the N-terminal peptide was different. Mass spectrometric analysis revealed that the observed monoisotopic *m/z* values of the purified peptides (P2, 2216.80; P3, 1347.69; and P4, 1603.60) were similar to the theoretically predicted values (2216.99, 1347.52, and 1603.71, respectively) for a peptide that has an intrachain S–S bonds and C-terminal amidation. We generated the synthetic peptides AQQSQAKKGCQAYGHVCYGGH-NH2 (long-form CCHamide-2) and GCQAYGHVCYGGH-NH2 (CCHamide-2). The retention times of the P2 and P3 active fractions were identical to those of the synthetic AQQSQAKKGCQAYGHVCYGGH-NH2 and GCQAYGHVCYGGH-NH2 peptides (which have an intrachain disulfide bond) on RP-HPLC, respectively (**Figure 2F**). Thus, these data suggest that both natural peptides have an intrachain disulfide bond and C-terminal amidation.

# **PHARMACOLOGICAL CHARACTERIZATION**

"fendo-03-00177" — 2012/12/27 — 18:13 — page 3 — #3

The interaction of CCHamide-1 and CCHamide-2 with CG30106 or CG14593 was examined using synthetic peptides. CCHamide-1 induced concentration-dependent, robust increases in [Ca2+]i in CHO-CG30106 cells, with a half-maximal response concentration (EC50) of 1.80×10−<sup>11</sup> M (**Figure 3A**). CCHamide-2 potently activated CG30106 (EC50; 4.86 <sup>×</sup> <sup>10</sup>−<sup>9</sup> M (**Figure 3A**). CCHamide-2 induced dose-dependent, robust increases in [Ca2+]i in CHO-CG14593 cells, with an EC50 of 4.80 <sup>×</sup> <sup>10</sup>−<sup>11</sup> M (**Figure 3B**).

"fendo-03-00177" — 2012/12/27 — 18:13 — page 4 — #4

**FIGURE 1 | Purification of CCHamide-1 from fly extracts.** Black bars indicate changes of [Ca2+]i fluorescence signal in CHO-CG30106 cells. **(A)** G-50 gel filtration of the SP-III fraction of fly extracts. The active fraction was subjected to one step of CM-ion-exchange HPLC and three steps of RP-HPLC. **(B)** Final purification of the active fraction by RP-HPLC. **(C)** Nucleotide sequence and deduced amino acid sequence of CCHamide-1 cDNA. CCHamide-1 cDNA encode 182-residue peptides. The asterisk indicates a glycine residue that serves as an amide donor for C-terminal amidation. The

CCHamide-1 sequence is underlined as (1). **(D)** Chromatographic comparison by RP-HPLC of natural CCHamide-1 and synthetic CCHamide-1. Black bar (P1) indicates the changes of [Ca2+]i fluorescence signal in CHO-CG30106 cells. Each peptide was applied to a Symmetry C18 column (3.9 mm × 150 mm, Waters, MA, USA) with a 10–60% ACN/0.1% trifluoroacetic acid (TFA) linear gradient at a flow rate of 1 ml/min for 80 min. P1 represent active fraction containing natural CCHamide-1. (a) Synthetic CCHamide-1. **(E)** Active fractions of each chromatography and the amino acid sequence of CCHamide-1.

CCHamide-1 potently activated CG14593 (EC50; 3.32 <sup>×</sup> <sup>10</sup>−<sup>8</sup> <sup>M</sup> (**Figure 3B**). Neither CCHamide-2 nor CCHamide-1 induced a response in CHO cells transfected with the vector alone (data not shown). In the investigation of the interaction between non-Cterminal amidated synthetic peptides or long-form CCHamide-2 and CG30106, the EC50 values were as follows: non-C-terminal amidated CCHamide-1, 1.66 <sup>×</sup> <sup>10</sup>−<sup>10</sup> M; non-C-terminal amidated long-form CCHamide-2, 8.93 <sup>×</sup> <sup>10</sup>−<sup>8</sup> M; long-form CCHamide-2, 6.45 <sup>×</sup> <sup>10</sup>−<sup>8</sup> M; and non-C-terminal amidated CCHamide-2, 1.22 <sup>×</sup> <sup>10</sup>−<sup>7</sup> M (**Figure 3C**). For CG14593, the EC50 values were as follows: long-form CCHamide-2, 1.49 <sup>×</sup> <sup>10</sup>−<sup>10</sup> M; non-C-terminal amidated long-form CCHamide-2, 1.18 <sup>×</sup> <sup>10</sup>−<sup>9</sup> M; non-C-terminal amidated CCHamide-2, 1.13 <sup>×</sup> <sup>10</sup>−<sup>8</sup> M; and non-C-terminal amidated CCHamide-1, 1.02 <sup>×</sup> <sup>10</sup>−<sup>7</sup> M (**Figure 3D**).

#### **PER TEST FOR MEASURING FEEDING SENSITIVITY**

As shown in **Figure 4**, a significant decrease was observed in the mean PER threshold, which was defined as the sucrose concentration at which 50% of flies show PER, after the injection of 10 pmol

"fendo-03-00177" — 2012/12/27 — 18:13 — page 5 — #5

**※** C C disulfide bond

**FIGURE 2 | Purification of CCHamide-2 from fly extracts.** Black bars indicate changes of [Ca2+]i fluorescent signal in CHO-CG14593 cells. **(A)** G-50 gel filtration of the SP-III fraction of fly extracts. The active fraction was subjected to one step of CM-ion-exchange HPLC and three steps of RP-HPLC. **(B–D)** Final purification of the active fraction by RP-HPLC. **(E)** Nucleotide sequence and deduced amino acid sequence of CCHamide-2 cDNA. CCHamide-2 cDNA encodes a 136-residue peptides. The asterisk indicates a glycine residue that serves as an amide donor for C-terminal amidation. The CCHamide-2 sequence is underlined as (4).

The other long-form of CCHamide-2 is translated from (2) or (3). **(F)** Chromatographic comparison by RP-HPLC of natural CCHamide-2 and synthetic CCHamide-2. Black bars (P2, P3) indicate the changes of [Ca2+]i fluorescence signal in CHO-CG14593 cells. Each peptide was applied to a Symmetry C18 column with a linear gradient elution for 80 min. P2 and P3 represent active fractions containing natural CCHamide-2. (b) Synthetic long-form of CCHamide-2. (c) Synthetic CCHamide-2. **(G)** Active fractions of each chromatography and the amino acid sequence of CCHamide-2.

"fendo-03-00177" — 2012/12/27 — 18:13 — page 6 — #6

replicates for each experiment.

changes in [Ca2+]i for various peptides, CCHamide-1 (open circle), and

of CCHamide-2: the mean PER threshold decreased from 236 mM (30 min after linger solution injection) to 77.2 mM (30 min after CCHamide-2 injection; *p* < 0.05). In contrast, no difference was observed between the mean PER threshold without any injection and 30 min after linger solution injection (222 and 236 mM, respectively, *p* > 0.05).

#### **DISCUSSION**

In this study, we biochemically purified 2 *Drosophila* peptides (CCHamide-1 and CCHamide-2) as endogenous ligands for *Drosophila* GPCRs CG30106 and CG14593. Recently, Hansen et al. (2011) independently identified these peptides from genome database and reported that synthetic CCHamide-1 and CCHamide-2 potently activated CHO/G-16 cells expressing recombinant CG30106 and CG14593. Then, Reiher et al. (2011) characterized CCHamide-1 and CCHamide-2 from the *Drosophila* midgut by capillary offline RP-HPLC coupled with MALDI-TOF MS/MS. Our biochemical characterization, however, for the first time, demonstrated three forms of CCHamide-2. The CCHamide-2 preproprotein is 136 amino acid residues long and contains three forms of CCHamide-2. The CCHamide-1 preproprotein is 182 amino acid residues long and contains one form of CCHamide-1. Pharmacological characterization by using CHO cells expressing GPCRs indicated that CCHamide-1 had a high potency for activating recombinant CG30106, but CCHamide-2 rather potently activated CG30106. In contrast, CCHamide-2 had a high potency for activating recombinant CG14593, but CCHamide-1 rather potently activated CG14593. Long-form CCHamide-2 and CCHamide-2 shared a highly similar potency for activating recombinant CG14593. Although we did not generate synthetic KKGC-QAYGHVCYGGH-NH2, it is predicted to have a high potency similar to that of other forms of CCHamide-2 for activating CG14593 because of the relationship between the amount of purified peptide and the specific activity. KKGCQAYGHVCYGGH-NH2 (P4) and AQQSQAKKGCQAYGHVCYGGH-NH2 (P2) may be incomplete processing intermediates of GCQAYGHVCYGGH-NH2 (P3), originating from two alternative signal peptide cleavage sites and incomplete KK prohormone convertase processing. The quantity of the purified peptide could not be accurately measured at the time of the experiments. Because the gel filtration fractions with particularly high activity were separated by CM-ion-exchange HPLC at pH 6.5, we did not purify all peptides for their receptors from the flies collected. However, we purified peptide KKGCQAY-GHVCYGGH-NH2 (P4) > AQQSQAKKGCQAYGHVCYGGH-NH2 (P2) > GCQAYGHVCYGGH-NH2 (P3) in amount. Therefore, in this study, we cannot conclude whether P4 and P2 are mature peptides or incomplete processing intermediates of P3. Because both CCHamide-1 and CCHamide-2 have a disulfide bond and a YGH motif, the disulfide bond is predicted to be an important structure for GPCR activation. Additionally, both peptides have a GXG-NH2 motif at the C-terminus. Therefore, we synthesized non-C-terminal amidated peptides to determine whether the C-terminal amide was necessary for the activation of each receptor. These results show that these peptides are considered to require both disulfide bonds and Cterminal amides to activate their respective GPCRs. Because we biochemically purified these ligands for the receptors by using the reverse pharmacological technique, we propose that no further modified forms or unknown ligands exist for these receptors in the fruit fly. CCHamide-1 is a cognate ligand for CG30106 and the three forms of CCHamide-2 are cognate ligands for CG14593.

BRS-3 is a mammalian orphan receptor (Ohki-Hamazaki et al., 1997). *Drosophila* CG30106 and CG14593 belong to the BRS-3 phylogenetic subgroup (Hewes and Taghert,2001). To provide new insights into the search for BRS-3 ligands, we examined whether CCHamides activate BRS-3, but we did not find any effect (data not shown).

CCHamide-1 and CCHamide-2 have been shown to be expressed predominantly in the brain and midgut (by FlyAtlas; http://www.flyatlas.org/; Chintapalli et al., 2007). In addition, CCHamide-1 and CCHamide-2 have been detected in the nervous system and midgut in a mass spectrometry study performed by Reiher et al. (2011). Therefore, CCHamides are suggested to be brain–gut peptides in insects. It is generally accepted that brain–gut peptides regulate feeding behavior in mammals (Williams et al., 2001). These peptides include neuropeptide Y, peptide YY, gastrin-releasing peptide, vasoactive intestinal peptide, adrenomedullin, cholecystokinin, galanin, glucagon-like peptide-1, and neuromedin U (Zimanyi et al., 1998; Beck, 2001). In addition, CCHamide-2 was distributed in the larval fat body (by FlyAtlas). The insect fat body is a functional counterpart of the mammalian adipose tissue and liver (Gutierrez et al., 2007). In mammal adipose tissue, leptin and adiponectin are important for feeding modulation. Therefore, we evaluated the effects of CCHamide on feeding by using the PER test in the blowfly *Phormia regina*. In flies and certain other insects, the PER test has long been used to investigate behavioral sensitivity to phagostimulative tastes (Nisimura et al., 2005). Flies extend their proboscis when the contact chemosensilla on their labella detects sweetness of sugar above a certain threshold concentration. Thus, we estimated the appetite or feeding motivation of the flies on the basis

"fendo-03-00177" — 2012/12/27 — 18:13 — page 7 — #7

of the PER test for sucrose, in which the threshold concentration of sucrose was evaluated as an indicator of feeding sensitivity. The injection of CCHamide-2 decreased the threshold for feeding on a sucrose solution. These data suggest that CCHamide-2 stimulates the feeding motivation of flies. Indeed, administration of CCHamide-2 significantly increased the sucrose intake (Hiraguchi et al., paper in preparation). In the presence of amino acids in the diet, target-of-rapamycin complex 1 (TORC1) signaling in fat cells generates a positive messenger that is released into the hemolymph (Colombani et al., 2003). This signal reaches the brain insulin-producing cells (IPCs), where it remotely controls the secretion of *Drosophila* insulin-like peptides (Dilp). Insulinlike peptides couple growth, metabolism, longevity, and fertility with changes in nutritional availability (Géminard et al., 2009). If CCHamide is a humoral factor that is secreted from the fat body like unpaired 2, it may play an important role in the modulation of nutrient status and growth (Rajan and Perrimon, 2012). Mice lacking functional BRS-3 develop metabolic defects and obesity (Ohki-Hamazaki et al., 1997). Therefore, the natural ligand of

#### **REFERENCES**


BRS-3 is expected to be a prominent inhibitor of appetitive behavior. The difference between CCHamide and the unknown ligand for BRS-3 with regard to feeding behavior is not clear. Further studies should de-orphanize BRS-3 by considering CCHamide by using bioinformatics or antibodies for CCHamide or *Drosophila* GPCRs.

#### **ACKNOWLEDGMENTS**

We thank Kaori Shirouzu,Masako Naito, and Eri Iwamoto for their technical assistance. This work was financially supported in part by the Improvement of Research Environment for Young Researchers Program of the Ministry of Education, Culture, Sports, Science and Technology; a grant for Scientific Research on Priority Areas from the University of Miyazaki; grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology, Japan; the Program for Promotion of Basic Research Activities for Innovative Bioscience (PROBRAIN); Suzuken Memorial Foundation; Shimadzu Scientific Foundation; and Uehara Memorial Foundation.

tyramine in appetite regulation. *J. Neurosci.* 25, 7507–7516.


"fendo-03-00177" — 2012/12/27 — 18:13 — page 8 — #8

feeding behavior by neuropeptide Y. *Curr. Pharm. Des.* 4, 349–366.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 25 September 2012; accepted: 14 December 2012; published online: 31 December 2012.*

*Citation: Ida T, Takahashi T, Tominaga H, Sato T, Sano H, Kume K, Ozaki M, Hiraguchi T, Shiotani H, Terajima S, Nakamura Y, Mori K, Yoshida M, Kato J, Murakami N, Miyazato M, Kangawa K and Kojima M (2012) Isolation of the bioactive peptides CCHamide-1 and CCHamide-2 from Drosophila and their putative role in appetite regulation as ligands for G protein-coupled receptors. Front. Endocrin. 3:177. doi: 10.3389/fendo.2012.00177*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Ida, Takahashi, Tominaga, Sato, Sano, Kume, Ozaki, Hiraguchi, Shiotani, Terajima, Nakamura, Mori, Yoshida, Kato, Murakami, Miyazato, Kangawa and Kojima. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# More than two decades of research on insect neuropeptide GPCRs: an overview

# *Jelle Caers , Heleen Verlinden, Sven Zels , Hans Peter Vandersmissen , Kristel Vuerinckx and Liliane Schoofs\**

Animal Physiology and Neurobiology, Department of Biology, Zoological Institute, KU Leuven, Leuven, Belgium

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Dick R. Nässel, Stockholm University, Sweden Hoffmann Klaus H., University of Bayreuth, Germany

#### *\*Correspondence:*

Liliane Schoofs, Department of Biology, Research Group of Functional Genomics and Proteomics, Naamsestraat 59, KU Leuven, 3000 Leuven, Belgium. e-mail: liliane.schoofs@ bio.kuleuven.be

This review focuses on the state of the art on neuropeptide receptors in insects. Most of these receptors are G protein-coupled receptors (GPCRs) and are involved in the regulation of virtually all physiological processes during an insect's life. More than 20 years ago a milestone in invertebrate endocrinology was achieved with the characterization of the first insect neuropeptide receptor, i.e., the Drosophila tachykinin-like receptor. However, it took until the release of the Drosophila genome in 2000 that research on neuropeptide receptors boosted. In the last decade a plethora of genomic information of other insect species also became available, leading to a better insight in the functions and evolution of the neuropeptide signaling systems and their intracellular pathways. It became clear that some of these systems are conserved among all insect species, indicating that they fulfill crucial roles in their physiological processes. Meanwhile, other signaling systems seem to be lost in several insect orders or species, suggesting that their actions were superfluous in those insects, or that other neuropeptides have taken over their functions. It is striking that the deorphanization of neuropeptide GPCRs gets much attention, but the subsequent unraveling of the intracellular pathways they elicit, or their physiological functions are often hardly examined. Especially in insects besides Drosophila this information is scarce if not absent. And although great progress made in characterizing neuropeptide signaling systems, even in Drosophila several predicted neuropeptide receptors remain orphan, awaiting for their endogenous ligand to be determined. The present review gives a précis of the insect neuropeptide receptor research of the last two decades. But it has to be emphasized that the work done so far is only the tip of the iceberg and our comprehensive understanding of these important signaling systems will still increase substantially in the coming years.

**Keywords: insects, neuropeptides, G protein-coupled receptors, signal transduction, neurobiology**

#### **INTRODUCTION**

The class of Insecta, which consists of more than 30 orders, forms the most diverse animal group on earth. With about one million documented species and presumably 10–30 million awaiting to be described, insects probably account for 50–70% of all existing animals (Scherkenbeck and Zdobinsky, 2009; Bellés, 2010; Van Hiel et al., 2010). Basically all the physiological processes during an insect's life cycle are regulated by neuropeptides, including developmental processes, behavioral functions, metabolic events and reproduction. As such, neuropeptides are the largest (very versatile) class of extracellular signaling molecules that are involved in communication between insect cells (Gäde and Goldsworthy, 2003; Meeusen et al., 2003; Claeys et al., 2005a). The insect neuropeptides and their actions have extensively been reviewed in the past (Nässel, 2002; Gäde and Auerswald, 2003; Gäde and Goldsworthy, 2003; Meeusen et al., 2003; Altstein, 2004; Gäde, 2004; Isaac et al., 2004; Simonet et al., 2004; Claeys et al., 2005a; Coast and Garside, 2005; Ewer, 2005; Predel and Wegener, 2006; Mertens et al., 2007; Stay and Tobe, 2007; De Loof, 2008; Audsley and Weaver, 2009; Scherkenbeck and Zdobinsky, 2009; Verleyen

et al., 2009; Verlinden et al., 2009; Weaver and Audsley, 2009; Altstein and Nässel, 2010; Bendena, 2010; Nässel and Winther, 2010; Van Hiel et al., 2010; Van Loy et al., 2010; Nässel and Wegener, 2011; Herrero, 2012; Spit et al., 2012; Taghert and Nitabach, 2012).

Neuropeptides exert their physiological functions by interacting with specific signal-transducing membrane receptors, resulting in intracellular responses (Zupanc, 1996). Most of these neuropeptide receptors belong to the G protein-coupled receptors (GPCRs), the largest family of cell surface proteins. However, there are some exceptions like the prothoracicotropic hormone (PTTH), which executes its role in metamorphosis through the activation of a receptor tyrosine kinase (RTK) (Rewitz et al., 2009). Most of the insulin-like peptides (ILPs) also interact with RTKs (Fernandez et al., 1995; Graf et al., 1997; Brogiolo et al., 2001; Wheeler et al., 2006; Wen et al., 2010; Iga and Smagghe, 2011). The eclosion hormone (EH), involved in ecdysis, interacts with a membrane-bound guanylate cyclase receptor (Chang et al., 2009) as does the neuropeptide-like precursor peptide 1 (NPLP1) (Overend et al., 2012).

The functional characterization of the first insect neuropeptide receptor, the *Drosophila melanogaster* tachykinin-like receptor (DTKR) took place in 1991 (Li et al., 1991). Subsequently, another *Drosophila* tachykinin-like receptor (NKD) (Monnier et al., 1992) and a neuropeptide Y (NPY)-like receptor (Li et al., 1992) were identified. The latter has recently been deorphanized as the *Drosophila* RYamide receptor (Collin et al., 2011; Ida et al., 2011a). In the following years only a few more insect GPCRs were cloned, e.g., the diuretic hormone receptors of *Manduca sexta* and *Acheta domesticus* (Reagan, 1994, 1996), the *Drosophila* gonadotropinreleasing hormone receptor (Hauser et al., 1998), which later on was deorphanized as an adipokinetic hormone (AKH) receptor (Staubli et al., 2002) and the *Drosophila* allatostatin (AST) receptor (DAR-1) (Birgül et al., 1999).

The real breakthrough in the field of insect neuropeptide receptor research came with the publication of the *Drosophila* genome in 2000 (Adams et al., 2000). This opened the opportunity to predict receptors based on genomic data (Hewes and Taghert, 2001), which clearly boosted the receptor deorphanization rate. At present, 35 GPCRs are functionally characterized in *Drosophila*. One receptor (Dmel\SPR) is activated by seemingly different neuropeptides, the myoinhibitory peptide (MIP) and the sex peptide SP. The others mainly respond to one neuropeptide type, which underlines the specificity of the receptor/ligand couples. Another 14 GPCRs are predicted to be involved in neuropeptide signaling pathways, but their ligands are still unknown and therefore they are classified as "orphan" receptors (**Table 1**) (Meeusen et al., 2003; Hauser et al., 2006, 2008; Clynen et al., 2010a). In section "Methuselah (CG6936) and Methuselah-like Receptors" the methuselah receptor is also briefly discussed. In spite the fact that several studies have been performed on this receptor, it still is not clear if it is really a neuropeptide receptor and if the *stunted* gene really encodes for its endogenous ligands.

Despite the diversity in their endogenous ligands, GPCRs have been rather well conserved during evolution. This has facilitated the search for neuropeptide receptors in newly released genomes like those of *Apis mellifera* (Hauser et al., 2006), *Tribolium castaneum* (Hauser et al., 2008), and *Bombyx mori* (Yamanaka et al., 2008; Fan et al., 2010). Research in other insects also revealed a set of new neuropeptide signaling systems that are not present in *Drosophila*, e.g., AKH/corazonin-related peptide (ACP) discovered in *Anopheles gambiae* (Hansen et al., 2010), allatotropin (AT) discovered in *B. mori* (Yamanaka et al., 2008), and inotocin discoverd in *T. castaneum* (Stafflinger et al., 2008) (**Table 1**).

Hitherto, 149 insect genome projects are either completed or in progress (http://www*.*ncbi*.*nlm*.*nih*.*gov/sites/entrez? db=bioproject) and in 2011, the i5K project was initiated, which aims to sequence 5000 insect genomes in the next 5 years (Robinson et al., 2011). With this overload of genomic information coming up, we intend to give the reader a clear overview of what is currently known on insect neuropeptide receptors. First, we will discuss some general characteristics of GPCRs and the deorphanizing strategies. Next, we will highlight the area of peptidomics, which facilitated the prediction and detection of ligands enormously, followed by a genetics part to discuss some commonly used tools to unravel the physiological functions of the neuropeptide-receptor systems. Thereafter, the current status of the insect neuropeptide GPCRs will be reviewed. To conclude, a short discussion about the importance of neuropeptide research in insects will be given.

#### **G PROTEIN-COUPLED RECEPTORS**

Several GPCR-(sub)families originated prior to the divergence of protostomian and deuterostomian animals. This led to a great diversification in chemical specificity to external stimuli like neuropeptides, glycoproteins, nucleotides, biogenic amines, odorants, taste ligands, and photons. Although GPCRs do not share any overall sequence homology, they do expose a similar topographical structure which is remarkably well conserved during evolution. They are typically composed of seven transmembrane (7TM) α-helices, each consisting of 20–30 hydrophobic amino acids, and three extracellular and intracellular loops connecting the different helices. The N-terminus is located at the extracellular site and often possesses several glycosylation sites; the C-terminus, on the other hand, is orientated into the cytoplasm and offers some potential phosphorylation sites. The extracellular parts are involved in ligand-specific binding, while the intracellular areas interact with a member of the family of heterotrimeric GTP-binding proteins (G proteins), consisting of an α-, β-, and γ-subunit (Bockaert and Pin, 1999). Based on shared sequence motifs, the GPCRs are categorized into at least six subfamilies. The evolutionary relationship between the different families is still unclear because of the lack of significant sequence homology. They probably evolved independently of each other or have adopted the G protein signal transduction pathways through convergent evolution (Brody and Cravchik, 2000; Gether, 2000; Horn et al., 2000). All the neuropeptide GPCRs belong to the rhodopsin-like (family A) or the secretin-like (family B) subfamily.

When a GPCR becomes activated by its ligand, the extracellular signal will be transduced into intracellular physiological responses. An activated receptor will undergo a conformational change, which in turn leads to the activation of the associated G protein. This promotes the release of GDP from the α-subunit, followed by binding of GTP. Next, the GTP-bound α-subunit dissociates from the βγ-dimer and both will be released in the cytoplasm. Subsequently, they can interact with their specific effector proteins to elicit cellular signaling pathways. The effector proteins involved depend on the type of the α-subunit. The most common α-subunits are Gq, Gs, and Gi*/*o. The Gq subunits interact with phospholipase Cβ (PLCβ) in order to initiate the hydrolysis of the membrane-bound phosphoinositolbiphospholipid-bisphosphates resulting in diacylglycerol (DAG) and inositol triphosphate (IP3). DAG activates protein kinase C (PKC) and IP3 mobilizes Ca2<sup>+</sup> from intracellular stores like the endoplasmic reticulum. The Gs and Gi*/*<sup>o</sup> subunits, respectively, activate or inhibit adenylyl cyclase provoking a subsequent increase or decrease of the cyclic adenosine monophosphate (cAMP) concentration within the cell. The Gs proteins are also capable of activating Ca2<sup>+</sup> channels, while the Gi*/*<sup>o</sup> proteins are able to interact with K+-channels. The intrinsic GTPase activity of Gα induces the hydrolysis of GTP to GDP, resulting in the reassociation of the subunits (Hepler and Gilman, 1992; Lustig et al., 1993; Vanden Broeck, 1996, 2001; Brody and Cravchik, 2000).


**neuropeptide**

 **receptors.**




Accessory gland peptide 70A

(Acp70A)/Ductus

 ejaculatorius

 peptide 99B (Dup99B).

# **DEORPHANIZING STRATEGIES**

There is a clear distinction between the techniques used to deorphanize receptors before and after the genomic era. In the past, one started with a bioactive ligand, purified from tissue extracts, in order to identify its corresponding receptor (the classic approach). Nowadays, an orphan receptor is used to explore its activating ligand from a library of synthetic compounds consisting of predicted neuropeptides (reverse pharmacology) (Meeusen et al., 2003). This strategy makes use of appropriate cellular expression systems used to express orphan receptors of interest. These systems hold the opportunity to measure one of the many second messenger reporter molecules released after receptor activation. The most commonly used expression systems are mammalian cell lines (Chinese Hamster Ovary [CHO] cells or Human Embryonic Kidney [HEK] 293 cells) and *Xenopus* oocytes. These are used in the bioluminescence-based assay (CHO cells), the fluorescence-based assay (HEK293 cells), the luciferase-based assay (HEK293 cells) and the electrophysiological assay (*Xenopus* oocytes).

Because it is nearly impossible to predict which kind of G protein interacts with an orphan receptor, a universal tool was required to predict the signaling cascade. This problem was circumvented with the discovery of the promiscuous G protein α subunits Gα<sup>16</sup> (human) and Gα<sup>15</sup> (murine). These Gα proteins regulate PLCβ, and possess the ability to interact with most GPCRs and, as such, their signaling pathways are redirected toward the release of Ca2<sup>+</sup> (Offermanns and Simon, 1995). Both, the bioluminescence and the fluorescence assay are based on the measurement of the release of intracellular Ca2<sup>+</sup> upon receptor activation. The bioluminescence assay makes use of bioluminescent proteins such as aequorine, purified from the jellyfish, *Aequoria victoria*, that interact with Ca2<sup>+</sup> (Prasher et al., 1987; Stables et al., 1997). In the fluorescence assay usually HEK293 cells are charged with a Ca2<sup>+</sup> sensitive fluorophore that serves as readout (Bender et al., 2002). The luciferase assay makes use of a reporter gene plasmid consisting of a cAMP response element (CRE) as readout for measuring intracellular cAMP levels (Janssen et al., 2008; Horodyski et al., 2011; Vuerinckx et al., 2011). For the electrophysiological assay, *Xenopus* oocytes are injected with a mix of the orphan receptor and the G protein gated inwardly rectifying K+ (GIRK) channels that are activated upon ligand binding. This leads to subsequent inward K+ currents that can be measured (Kofuji et al., 1995; Ho and Murrell-Lagnado, 1999; Ulens et al., 1999).

Besides the use of these heterologous expression systems, one can also make use of a homologous expression system in which the orphan receptor is expressed in *Drosophila* Schneider-2 (S2) cells (Vanden Broeck et al., 1998). The use of heterologous expression systems, however, prevents that compounds present in an insect extract, or predicted insect ligands would activate endogenous mammalian or amphibian receptors (for reviews, see: Meeusen et al., 2003; Mertens et al., 2004; Beets et al., 2011; Bendena et al., 2012).

#### **NEUROPEPTIDES AND PEPTIDOMICS**

An important feature of the currently used deorphanizing strategies is the ability to screen orphan receptors with compound libraries containing potential neuropeptides. The possibility to create such libraries coincided with the availability of the first whole genome databases. This also launched the era of peptidomics, which encloses the purpose to simultaneously identify and/or visualize all peptides present in a cell, tissue, body liquid, or organism. Peptidomics studies are based on two major elements, the *in silico* prediction of neuropeptides and the discovery and identification of neuropeptides using mass spectrometric devices (Baggerman et al., 2002; Predel et al., 2004; Wegener et al., 2006).

Endogenous neuropeptides are enclosed in larger preprohormones, mostly between 50 and 500 amino acids long (Baggerman et al., 2005a). They can code for multiple structurally related or unrelated neuropeptides, as well as for just one neuropeptide. The only common feature of preprohormones is the presence of an amino-terminal signal peptide, with exception of a predicted AST CC neuropeptide in *Drosophila* which has an amino-terminal peptide anchor (Veenstra, 2009a). This peptide is immediately cleaved off after arrival in the endoplasmic reticulum. The residual prohormone undergoes enzymatic cleavage at mono- or dibasic amino acid residues to release the neuropeptides (Hook et al., 2008; Rholam and Fahy, 2009). Most neuropeptides require posttranslational modifications to become bioactive or to improve stability.

Because of the poor sequence conservation between preprohormones and the short length of the neuropeptides, the majority consists only of 4–20 amino acids, their prediction from genome databases is not straightforward. Nevertheless, classical BLAST analyses have revealed 36 neuropeptide genes in *D. melanogaster* (Hewes and Taghert, 2001; Vanden Broeck, 2001), and 35 in *A. gambiae* (Riehle et al., 2002). Later on, the combined use of different bioinformatic tools, to overcome the low sensitivity of a BLAST analysis alone, revealed a total of 119 potential neuropeptide-coding genes in *Drosophila* (Liu et al., 2006; Clynen et al., 2010a). All neuropeptides predicted by these methods can be synthesized to construct synthetic peptide libraries applied in the reverse pharmacology assays.

The bioinformatic predictions, though, do not reveal which neuropeptides are ultimately produced, and endogenous bioactive neuropeptides may be overlooked in the genomic data. The processing of a precursor can also differ during developmental stages or between tissues, and post-translational modifications are hard to predict based on sequence information. Therefore, a biochemical characterization of neuropeptides is necessary. There are several possible peptidomics methods available to provide in these needs, all based on mass spectrometry. The most common tool is a combination of liquid chromatography, tandem mass spectrometry and database mining, which allows the detection and sequencing of low concentrations of neuropeptides from complex mixtures (Clynen et al., 2010b). Mass spectrometry applications led to the discovery of hundreds of neuropeptides. As is often the case, *Drosophila* peptidomics (Baggerman et al., 2002; Baggermanet al., 2005b; Schoofs and Baggerman, 2003) paved the way for peptidomic studies in other insects, e.g., *A. mellifera* (Hummon et al., 2006; Boerjan et al., 2010a), *Nasonia vitripennis* (Hauser et al., 2010), *T. castaneum* (Li et al., 2008), and *Aedes aegypti* (Predel et al., 2010). Also in insects with no completely sequenced genome, peptidomics may prove useful, e.g., *Locusta migratoria* (Clynen et al., 2006; for reviews, see: Hummon et al., 2006; Boonen et al., 2008; Menschaert et al., 2010).

# **FUNCTIONAL GENOMICS**

Upon the characterization of a neuropeptide receptor and its ligand, the question remains which function they possess in a specific organism. These functions can be determined with genetic tools. In the classic approach the phenotype of interest is chosen first and then attempts are made to identify the genes responsible for this phenotype (forward genetics). With the rise of the whole genome era, a tremendous number of genes with unknown functions were identified. This made it possible to start with a gene of interest and to study its function (reverse genetics). Currently, the most used techniques to perform reverse genetics are silencing of genes of interest by RNA interference (RNAi), generating knockouts, and overexpressing specific genes using the GAL4/UAS system.

The generation of loss-of-function phenotypes through the application of RNAi is a fairly new technique as it was described for the first time in 1998 in *Caenorhabditis elegans* (Fire et al., 1998), immediately followed by a report of RNAi usage in *D. melanogaster* (Kennerdell and Carthew, 1998). RNAi studies are widely used in the field of insect research and have proven to be appropriate to unravel functions of neuropeptides and their receptors in various species (Bellés, 2010; An et al., 2012). There is a genome-wide transgenic RNAi library available for *Drosophila*, consisting of short gene fragments cloned as inverted repeats and expressed using the binary GAL4/UAS system (Dietzl et al., 2007). The usage of the GAL4/UAS system to perform RNAi experiments becomes also more established in other insects like *B. mori* (Dai et al., 2008) and *T. castaneum* (Schinko et al., 2010).

RNAi can not entirely impede the expression of a gene of interest. To generate a complete knockout of a gene, mutagenic or homologous recombination tools are frequently used. Mutagenesis relies on the incorporation of mutations, which can be obtained by the application of chemical mutagenesis or by transposable element mutagenesis, followed by a thorough screen to detect the samples containing mutations in the gene of interest. Homologous recombination is based on the host DNA repair system for the alteration of a target sequence in the genome by a donor sequence. This donor sequence exhibits homology to the target sequence, but contains the desired genetic modifications. The alteration is preceded by the generation of a double strand break in the target or donor sequence, inducing the homologous recombination repair system (for reviews, see: Reumer et al., 2008; Wesolowska and Rong, 2010; An et al., 2012).

Besides studying the effects of a knockdown or a complete knockout of a certain gene, overexpressing a gene can also yield important information about its function. To obtain overexpression, the gene of interest can be coupled to a binary GAL4/UAS system as well.

The previous described techniques to identify, deorphanize and determine the functions of neuropeptide signaling systems are widely applied in insect research, yielding an enormous amount of information. **Table 2** summarizes the neuropeptide receptors that have been predicted and/or functionally characterized for a selection of model insects. In the next section we aim to give a brief summary of what is known so far relating to these insect neuropeptide receptors. For convenience all intertitles are accompanied with the corresponding computed gene (CG) numbers of the *Drosophila* receptors. These numbers were used for genes identified during the annotation of the whole *Drosophila* genome sequence. For those receptor genes not annotated in *Drosophila*, the accession number of the receptor gene for the insect in which it was first deorphanized is added.

# **DEORPHANIZED NEUROPEPTIDE RECEPTORS**

# **ADIPOKINETIC HORMONE RECEPTORS (CG11325 ORTHOLOGS)**

The first structural characterization of an AKH neuropeptide was achieved in 1976 (Stone et al., 1976). Currently, around 55 isoforms, derived from various insect species, have been described (Gäde, 2009; Caers et al., 2012; Gäde and Marco, 2012; Jedlicka ˇ et al., 2012; Malik et al., 2012; Weaver et al., 2012). They consist of 8–10 amino acids, and are characterized by a blocked N-terminus (pyroglutamate) and C-terminus (amidation) (Gäde and Auerswald, 2003). The main function of AKH is the regulation of the energy metabolism. During energy requiring processes like flight, the AKH neuropeptides are released from the corpora cardiac (CC) and will interact with their receptors, present in the membrane of the fat body adipocytes. This will induce the release of energy rich substrates (lipids, trehalose, or proline) (Lorenz and Gäde, 2009). The kind of substrates released, depends on the coupled G protein. When AKH binds to a Gq protein-coupled receptor, glycogen phosphorylase will be activated and trehalose will be set free. If the signaling pathway acts by a Gs proteincoupled receptor, triacylglycerol lipase will be activated, resulting in the production of DAG or free fatty acids (Gäde and Auerswald, 2003). The last years it became clear that the function of AKH is not restricted to locomotory activity alone, but that it acts as a general regulator of homeostasis in insects, influencing all energy requiring processes (e.g., egg production, feeding behavior, larval growth, molting, and immune response) (Goldsworthy et al., 2002, 2003; Lorenz, 2003; Lee and Park, 2004; Isabel et al., 2005; Grönke et al., 2007; Bharucha et al., 2008; Lorenz and Gäde, 2009; Arrese and Soulages, 2010; Attardo et al., 2012; Konuma et al., 2012). AKH also serves as an anti-stress hormone in oxidative stress situations (Kodrík et al., 2007; Veceˇ ra et al., 2007; Kodrík, ˇ 2008; Huang et al., 2011a).

The AKH receptors (AKHR) are closely related to the ACP receptors and constitute the invertebrate AKH/ACP receptor family. Together with the invertebrate corazonin/gonadotropin releasing hormone (GnRH) receptor family and the vertebrate/protochordate GnRH receptor family they compose the GnRH receptor superfamily (Lindemans et al., 2011; Roch et al., 2011). The first AKHR was determined in *M. sexta* by using fat body fractions to ascertain the optimal binding conditions for tritium-labeled *Manse*-AKH (Ziegler et al., 1995). The *Drosophila* AKHR was the first to be cloned and was deorphanized by making use of the electrophysiological assay (Park et al., 2002), and its characterization was confirmed by Staubli et al. (2002) using a bioluminescence assay. Later, AKHRs were also identified and characterized in other insect species: *Periplaneta americana* (Hansen et al., 2006; Wicher et al., 2006a), *A. gambiae* (Belmont


 **| Characterized and predicted neuropeptide receptors in insect species of varying insect**

**orders.**

**Table**

**2**


**Table**  Black box: functionally characterized neuropeptide receptors. box:

Gray previously predicted neuropeptide receptors.

White box: predicted neuropeptide receptors based on our own blast analysis.

na: not annotated. et al., 2006), *B. mori* (Staubli et al., 2002; Zhu et al., 2009; Huang et al., 2010), and *T. castaneum* (Li, unpublished data). Two putative AKHR variants have been predicted in *A. aegypti* (Kaufmann et al., 2009) and one AKHR is identified in the *Apis* genome (Hauser et al., 2006). However, it remains doubtful if this receptor is really functional in the honeybee, because mass spectrometric techniques have failed to detect the predicted *Apis* AKH neuropeptide (Veenstra et al., 2012). Besides expression in the fat body (Kaufmann and Brown, 2006; Ziegler et al., 2011), the AKHR is also expressed in various neurons of *P. americana*, including the abdominal dorsal unpaired medial (DUM) neurons, which are responsible for the release of octopamine. As such, octopamine may be the link between elevated AKH-titers and the increase in locomotion (Wicher et al., 2006b, 2007; Verlinden et al., 2010).

#### **ADIPOKINETIC HORMONE/CORAZONIN-RELATED PEPTIDE RECEPTOR (XP\_321591 ORTHOLOGS)**

In 2006, an *A. gambiae* receptor was annotated and cloned that was closely related to the AKH and corazonin receptors, but could not be activated by these neuropeptides (Belmont et al., 2006). Hansen et al. (2010) detected a neuropeptide closely related to both AKH and corazonin and named it ACP. This neuropeptide was able to activate the receptor expressed in CHO/Gα<sup>16</sup> cells in a dose-responsive manner (Hansen et al., 2010). Subsequently, the ACP receptor was also characterized in *T. castaneum* (Hansen et al., 2010). Recently, two predicted ACP receptors of *B. mori* (Yamanaka et al., 2008; Hansen et al., 2010) were also characterized, but were indicated as AKHR (Shi et al., 2011). The ACP neuropeptides were in fact already described in *L. migratoria* (Siegert, 1999) and in *A. gambiae* (Kaufmann and Brown, 2006), but were classified as AKH neuropeptides with unknown functions. ACP and its receptor are structurally intermediate between the AKH and corazonin neuropeptides and their receptors, which is a prominent example of receptor/ligand co-evolution. An ancestral receptor and ligand gene have probably duplicated several times followed by mutations and evolutionary selection, leading to three signaling systems. However, the ACP signaling system is absent in all investigated *Drosophila* species as well as in *A. mellifera*, *Acyrthosiphon pisum*, *Pediculus humanus*, and in the crustacean *Daphnia pulex*, suggesting that it may have been lost several times during arthropod evolution (Hansen et al., 2010). So far no functions are assigned to the ACP signaling system, but the high expression shortly before and after hatching of *T. castaneum* suggests a role in early larval development (Hansen et al., 2010).

# **ALLATOSTATIN A RECEPTORS (CG2872 AND CG10001 ORTHOLOGS)**

The endogenous ligands of the AST A receptor are the A-type AST neuropeptides which belong to the group of allatoregulatory neuropeptides together with the B-, and C-type allatostins and the ATs (for a review, see: Weaver and Audsley, 2009) and the recently discovered AST CC neuropeptides (Veenstra, 2009a). Allatoregulatory peptides are named after their ability to either inhibit (ASTs) or stimulate (ATs) juvenile hormone (JH) synthesis (Audsley et al., 2008). The B-type ASTs are also known as myoinhibiting peptides and were found to activate

The A-type AST-As, or FGLamides were first isolated of brain extracts of cockroaches (Woodhead et al., 1989; Pratt et al., 1991), and have since been found in every investigated insect species, except for *T. castaneum* (Li et al., 2008). They are characterized by a conserved (Y/F)XFG(L/I)-NH2 sequence (Hayes et al., 1994; Audsley et al., 1998). AST-As regulate JH biosynthesis in cockroaches, crickets, and termites (Pratt et al., 1989, 1991; Woodhead et al., 1989, 1994; Bellés et al., 1994; Weaver et al., 1994; Lorenz et al., 1995, 1999; Yagi et al., 2005; for a review, see: Stay and Tobe, 2007). A property attributed to all AST-As is myoinhibition of visceral muscles (Hoffmann et al., 1999; Stay, 2000; Aguilar et al., 2003; Weaver and Audsley, 2009; Zandawala et al., 2012). Recently, *Drosophila* AST-A was linked to food intake and foraging behavior (Hergarden et al., 2012; Wang et al., 2012). In *Drosophila*, two A-type AST receptors are identified: DAR-1 and DAR-2 (Birgül et al., 1999; Larsen et al., 2001). DAR-1, when expressed in *Xenopus* oocytes was shown to couple to a G-protein of the Gi*/*<sup>o</sup> family. When expressed in CHO cells, DAR-1 and -2 are activated by AST-A and mobilize intracellular Ca2<sup>+</sup> (Larsen et al., 2001). AST-A receptors were also characterized in *P. americana* (Auerswald et al., 2001), *B. mori* (Secher et al., 2001), and *Diploptera punctata* (Lungchukiet et al., 2008). Northern blot experiments showed that the *B. mori* receptor is expressed in the midgut of fifth larval instars and to a much lesser extend in the brain (Secher et al., 2001).

#### **ALLATOSTATIN C RECEPTORS (CG7285 AND CG13702 ORTHOLOGS)**

The first C- or PISCF-type AST was characterized in the late pupae of *M. sexta*. AST-Cs contain a typical C-terminal PISCF-OH sequence, a blocked N-terminus and a disulfide bridge linking Cys-7 and Cys-14 (Kramer et al., 1991). Orthologs are found in other lepidopteran, dipteran and coleopteran species (Li et al., 2006). In several insects, C-type or C-type-like ASTs can have both allatostatic and allatotropic properties, depending on the age of the animal (Abdel-Latief et al., 2004; Clark et al., 2008; Griebler et al., 2008; Abdel-Latief and Hoffmann, 2010). In Diptera two AST-C receptors have been characterized for *Drosophila* using *Xenopus* oocytes (Kreienkamp et al., 2002) and for *Aedes* using HEK cells (Mayoral et al., 2010). Only one AST-C receptor was found to be present in *Bombyx* (Yamanaka et al., 2008) and in *Tribolium* (Audsley et al., 2012). Activation of the *Bombyx* AST-C receptor elicits an increase in intracellular cAMP levels (Yamanaka et al., 2008), while the *Tribolium* receptor was deorphanized in HEK cells, inducing a Ca2<sup>+</sup> response (Audsley et al., 2012). In adult fruit flies, both *drostar* genes are expressed in the optic lobes and the pars intercerebralis, where the AST-C neuropeptide was also found to be present. This suggests a function in the modulation of visual information processing. In the last larval stage, receptor expression was found in the brain and corpora allata (CA) (Kreienkamp et al., 2002). In *Aedes* significant differences were observed in tissue distribution and expression levels for the two receptor paralogs (Mayoral et al., 2010). In *Tribolium* the highest transcript levels were noticed in the head and the gut, with variable amounts in the fat body and reproductive organs. These transcript levels were also shown to be sex-dependent (Audsley et al., 2012).

The recently discovered AST CC neuropeptide (AST CC) (Veenstra, 2009a) was also identified in *Tribolium* and showed to be capable of activating the AST-C receptor in a dose-dependent manner (Audsley et al., 2012). A knock out of the *Drosophila Ast-CC* gene is embryonic lethal, suggesting that it is an essential gene (Veenstra, 2009a).

#### **ALLATOTROPIN RECEPTORS (NP\_001127714 ORTHOLOGS)**

AT was named after its ability to stimulate JH biosynthesis in the CA but is also linked to other functions like myostimulation, cardio-acceleration, regulation of photic entrainment, ion exchange regulation, and the up-regulation of the secretion of digestive enzymes (Veenstra et al., 1994; Würden and Homberg, 1995; Lee et al., 1998; Koladich et al., 2002; Petri et al., 2002; Homberg et al., 2003; Hofer and Homberg, 2006; Lwalaba et al., 2010; Sterkel et al., 2010), of which the myotropic role of AT is probably the most ancestral (Elekonich and Horodyski, 2003). ATs are found in several invertebrate EST and genomic databases (for reviews, see: Clynen and Schoofs, 2009; Weaver and Audsley, 2009) and they all have a TARGF/Y motif at the C-terminus. In *Manduca* and *Bombyx*, also AT-like (ATL) neuropeptides were found, which arise by alternative splicing of the AT gene (Horodyski et al., 2001; Nagata et al., 2012a). In 2008, the AT receptor (ATR) was characterized in *B. mori* (Yamanaka et al., 2008). Remarkable, this receptor was mainly localized in the Short neuropeptide F (sNPF)-producing cells in the CC, but not in the JH producing CA. It was suggested that AT regulates the production and/or release of sNPFs from the CC and that these sNPFs are responsible for some of the allatotropic functions assigned to the ATs (Yamanaka et al., 2008). In 2011, the ATRs of *M. sexta*, *T. castaneum,* and *A. aegypti* were characterized (Horodyski et al., 2011; Vuerinckx et al., 2011; Nouzova et al., 2012) and show, unlike the ligand, remarkable similarity with the vertebrate orexin receptors (Yamanaka et al., 2008; Vuerinckx et al., 2011). Upon activation by AT or ATLs, *Manse-*ATR, and *Trica-*ATR elevate both intracellular Ca2<sup>+</sup> and cAMP concentrations in cellular expression systems (Horodyski et al., 2011; Vuerinckx et al., 2011). Expression of ATs and ATRs in the different insect species is likely to be strongly regulated, since large differences were measured between developmental stages, sexes, feeding conditions, *etc*. (Elekonich and Horodyski, 2003; Horodyski et al., 2011; Vuerinckx et al., 2011; Nouzova et al., 2012). Possibly additional ATRs may be present in some insect species, since very similar additional receptors have been predicted from *Bombyx* and *Aedes* genomes (Yamanaka et al., 2008; Nouzova et al., 2012).

#### **CALCITONIN-LIKE DIURETIC HORMONE RECEPTORS (CG32843/CG17415/CG17043 ORTHOLOGS)**

The first calcitonin-like diuretic hormone (CT/DH), called *Dippu*-DH31 was identified in *D. punctata* (Furuya et al., 2000). More orthologs were discovered by phylogenetic analysis (Zandawala, 2012). CT/DH stimulates fluid secretion by Malpighian tubules and seems to work via a Ca2+-dependent mechanism in *D. punctata* (Furuya et al., 2000). In *Drosophila*, CT/DH stimulates fluid secretion by activating the apical membrane V-ATPases via cAMP as second messenger (Coast et al., 2001) and in *Anopheles* the fluid excretion in Malpighian tubules is also cAMP driven (Coast et al., 2005). In *Rhodnius*, diuresis by CT/DH seems to be independent of cAMP (Te Brugge et al., 2011). CT/DH is also involved in contractions of the gut and associated glands (Te Brugge et al., 2009) and may play a role in ecdysis (Kim et al., 2006a,b). The *Drosophila* CT/DH receptor (DH31-R1) is activated by *Drome*-DH31 and is expressed in the Malpighian tubules. The signaling in HEK293 cells was dependent upon co-expression of the receptor component protein (RCP), which is critical for downstream signaling from the mammalian calcitonin-like receptor (Johnson et al., 2005). One CT/DH receptor has been predicted in *A. aegypti*, *A. gambiae*, *A. mellifera*, *N. vitripennis,* and *T. castaneum* and two CT/DH receptors were found in *A. pisum*, although it is not yet clear whether both paralogues encode a functional CT/DH receptor.

#### **CAPA RECEPTORS (CG14575 ORTHOLOGS)**

The insect capa neuropeptides, or periviscerokinin peptides, usually possess the C-terminal sequence FPRVamide. The insect *capability* gene encodes a preprohormone containing two capa neuropeptides (capa-1 and capa-2) and one or more pyrokinin-1 (Kean et al., 2002), but they do not activate each other's receptors (Iversen et al., 2002a; Rosenkilde et al., 2003; Cazzamali et al., 2005). Capa neuropeptides have a diuretic effect on the Malpighian tubules of *Drosophila* (Pollock et al., 2004), but in *R. prolixus* and other insects they act antidiuretic (Coast and Garside, 2005; Paluzzi and Orchard, 2006). Recently, it was shown that the *Aedes* capa neuropeptide can induce either diuretic or antidiuretic effects depending on the dose (Ionescu and Donini, 2012). In addition, capa neuropeptides have myotropic effects in a variety of insects (Wegener et al., 2002; Predel and Wegener, 2006). Capa receptors have been characterized in *Drosophila* and in *Anopheles* (Iversen et al., 2002a; Park et al., 2002; Olsen et al., 2007; Terhzaz et al., 2012). Both capa-1 and capa-2 elicited a dose-dependent response. The gene encoding the *Drosophila* capa receptor is highly expressed in larval and adult tubules (Terhzaz et al., 2012).

Capa receptors are found in different mosquito species, although not in *A. aegypti*. In other holometabolous insects, orthologs are found in representatives of the major orders, including Hymenoptera, Coleoptera, and Lepidoptera. The honey bee genome contains two paralogues, as does the *B. mori* and *M. sexta* genome. Also *N. vitripennis* contains a paralogue (XP\_001600587.2), formerly suggested lacking this receptor (Hauser et al., 2006, 2010; Yamanaka et al., 2008). Also in Coleoptera, a capa receptor is found in *Tribolium* (Hauser et al., 2008).

#### **CCHAMIDE-1 AND -2 RECEPTORS (CG30106/CG14484 AND CG14593 ORTHOLOGS)**

The first CCHamide neuropeptide has only recently been identified in *B. mori* and it was found to be expressed in the central nervous system and the midgut (Roller et al., 2008). Subsequently, two CCHamide neuropeptides were detected in all insects with a sequenced genome (Hansen et al., 2011). In *D. melanogaster*, cognate receptors have been identified for both CCHamide neuropeptides. CG30106 expressed in CHO/Gα<sup>16</sup> cells was activated by CCHamide-1 at nanomolar concentrations but also responded to high concentrations of CCHamide-2. CG14593 was activated by nanomolar concentrations of CCHamide-2 as well as by micromolar concentrations of CCHamide-1 (Hansen et al., 2011). Previously, CG30106 had been described as a receptor for myoinhibiting neuropeptides (Johnson et al., 2003b), but as several independent attempts to repeat this result were fruitless, this was likely an erroneous characterization.

#### **CORAZONIN RECEPTORS (CG10698 ORTHOLOGS)**

The first corazonin was isolated and identified from the CC of *P. americana* and was presented as a new cardioaccelerating neuropeptide (Veenstra, 1989). Corazonin is present in most insects (excluding beetles and aphids) (for reviews, see: Gäde et al., 2008; Li et al., 2008; Weaver and Audsley, 2008; Huybrechts et al., 2010) and the most common corazonin sequence among insects is pQT-FQYSRGWTNamide (Predel et al., 2007). The role of corazonin, however, is not restricted to cardio-excitatory actions. In locusts, corazonin is involved in cuticular melanization in the gregarious phase (Tawfik et al., 1999; Tanaka et al., 2002), in *M. sexta* a role in the initiation of ecdysis behavior is noticed (Kim et al., 2004; Žitnan et al., 2007 ˇ ) and it has been suggested that corazonin is involved in sex-dependent stress responses (Zhao et al., 2010) and in the regulation of insulin producing cells in *Drosophila* ((Kapan et al., 2012); for reviews, see: Veenstra, 2009b; Boerjan et al., 2010b).

The corazonin receptor was first characterized in *Drosophila* by making use of a bioluminescence assay (Cazzamali et al., 2002), which was confirmed using *Xenopus* oocytes (Park et al., 2002). Subsequently, the corazonin receptors for *M. sexta* (Kim et al., 2004), *A. gambiae* (Belmont et al., 2006) and *B. mori* (Shi et al., 2011) were characterized, and a putative corazonin receptor for *Musca domestica*, was cloned (Sha et al., 2012). Neither the corazonin neuropeptide nor its receptor could be identified in *Tribolium* (Hauser et al., 2008) or *Acyrthosiphon*. In *N. vitripennis*, despite the presence of a corazonin neuropeptide, so far no corazonin receptor could be predicted (Hauser et al., 2010). The invertebrate corazonin receptors are part of the GnRH receptor superfamily [see section "Adipokinetic Hormone Receptors (CG11325 Orthologs)"] (Lindemans et al., 2011; Roch et al., 2011). The *Drosophila* receptor is expressed in all developmental stages (Cazzamali et al., 2002). The *Manduca* corazonin receptor is present in endocrine Inka cells, the source of preecdysis- and ecdysis-triggering hormones, suggesting a role upstream of ecdysis triggering hormone (ETH) (Kim et al., 2004). In *Anopheles*, there are pronounced spikes of corazonin receptor expression in 2nd instar larvae and around the transition from pupa to adult (Hillyer et al., 2012). In *Musca*, a high level of corazonin receptor expression was noticed in the larval salivary glands and a moderate level in the central nervous system. In adults, the receptor was expressed both in the head and body (Sha et al., 2012).

#### **CRF-LIKE DIURETIC HORMONE RECEPTORS (CG8422 AND CG12370 ORTHOLOGS)**

The first corticotropin-releasing factor like diuretic hormone (CRF/DH) was identified in *M. sexta* as a diuretic peptide (DP) consisting of 41 amino acids that shows sequence similarity to corticotropin releasing factor, urotensin I and sauvagine (Kataoka et al., 1989). A second CRF/DH was also discovered in *M. sexta* (Blackburn et al., 1991). CRF/DHs are also referred to as DH44, after the number of amino acids in the CRF/DH of *D. melanogaster* (Cabrero et al., 2002). CRF/DH increases fluid excretion *in vivo* (Kataoka et al., 1989) and *in vitro* (Kay et al., 1991, 1992; Lehmberg et al., 1991; Clottens et al., 1994) and increases cAMP levels in Malpighian tubules (Lehmberg et al., 1991; Kay et al., 1992; Clottens et al., 1994; Furuya et al., 1995). Besides its diuretic function, CRF/DH negatively influences feeding and reproduction (Keeley et al., 1992; Van Wielendaele et al., 2012) and stimulates gut contractions (Te Brugge et al., 2009). The *M. sexta* CRF/DH receptor was the first to be cloned and was activated by *Manse*-DH, making use of cAMP as second messenger (Reagan, 1995). Also the CRF/DH receptor in *A. domesticus* uses cAMP as secondary messenger (Reagan, 1996). The first *D. melanogaster* CRF/DH receptor (DH44-R1), encoded by CG8422, may couple to multiple second messengers as both cAMP and Ca2<sup>+</sup> were stimulated upon binding of *Drome*-DH to the receptor (Johnson et al., 2004). The second *D. melanogaster* CRF/DH receptor (DH44-R2), encoded by CG12370, is also activated by *Drome*-DH resulting in an increase of intracellular cAMP and causes specific β-arrestin translocation to the plasma membrane. DH44-R2 is probably the receptor that modulates DH sensitivity at the level of the microtubules (Hector et al., 2009). A CRF/DH receptor was also cloned in *B. mori* and in *A. aegypti* (Ha et al., 2000). The *Aedes* DH-I receptor is by far the most abundant receptor in Malpighian tubules and its transcript levels increase after a blood meal (Jagge and Pietrantonio, 2008). More CRF/DH receptor orthologs were found in *T. castaneum* and *A. pisum*, but only one orthologue is found in *A. gambiae*, *A. mellifera,* and *N. vitripennis* up to date. Although the number of receptors seems to differ, CRF/DH signaling is likely to be conserved in all major insect orders.

#### **CRUSTACEAN CARDIOACTIVE PEPTIDE RECEPTORS (CG33344/CG6111/CG14547 ORTHOLOGS)**

Crustacean cardioactive peptide (CCAP) was originally identified in the shore crab *Carcinus maenas* and exhibited an acceleratory effect on semi-isolated heart tissue (Stangier et al., 1987). An identical neuropeptide was subsequently isolated from *L. migratoria* (Stangier et al., 1989). The structure of CCAP is identical in all examined insects and consists of the cyclic nonapeptide PFCNAFTGCamide. CCAP stimulates heart contractions (Cheung et al., 1992; Furuya et al., 1993; Li et al., 2011a) and contractions of visceral muscles (Stangier et al., 1989; Donini et al., 2001, 2002; Donini and Lange, 2002), and promotes the release of AKH (Veelaert et al., 1997) and digestive enzymes (Sakai et al., 2006). CCAP also plays a role in ecdysis in several insects (Gammie and Truman, 1997; Ewer et al., 1998; Kim et al., 2006a,b; Arakane et al., 2008). *Drosophila* and *Anopheles* CCAP receptors have been expressed in CHO/Gα<sup>16</sup> cells and are activated by CCAP (Cazzamali et al., 2003; Belmont et al., 2006). In *T. castaneum*, two genes encode for CCAP receptors (Hauser et al., 2008) and both showed a dose-dependent response to CCAP (Li et al., 2011a). Functional analysis using RNAi revealed that only TcCCAPR-2 is essential for cardioacceleratory activity (Li et al., 2011a). CCAP receptor orthologs have been found in *A. mellifera* (Hauser et al., 2006), *A. aegypti*, *A. pisum*, *B. mori,* and *N. vitripennis* and thus the CCAP receptor seems to be conserved in many insect orders.

#### **ECDYSIS TRIGGERING HORMONE RECEPTORS (CG5911 ORTHOLOGS)**

To be able to grow and undergo metamorphosis, insects need to shed their exoskeleton, the process known as ecdysis (Truman, 1996). This process is initiated and regulated by the ETH (for a review, see: Žitnan et al., 2007 ˇ ). The *eth* gene encodes for two active neuropeptides named pre-ETH and ETH in moths and ETH1 and ETH2 in other insects. The ETHs have a common PRX1-amide (X1 is I, V, L, or M) sequence at the C-terminus (Park et al., 2002). In *Drosophila*, *Manduca,* and *Bombyx*, the two ETHs differ in length. In *Drosophila* and *Manduca* the short form only can elicit a part of the ecdysis behaviors, whereas the long one can elicit whole ecdysis (Žitnan et al., 1999; Park et al., ˇ 2002). In *Bombyx* and *Aedes*, both neuropeptides seemed to be equally potent (Žitnan et al., 2002; Dai and Adams, 2009 ˇ ). In *Apis*, *Nasonia,* and *Acyrthosiphon* only one form is found, that in *Apis* is shown to be sufficient to elicit ecdysis (Žitnan et al., ˇ 1999; Park et al., 2002). These neuropeptides are released in the bloodstream and activate the ETH receptors (ETHRs) situated in the central nervous system. The *ethr* gene encodes for two splice variants of the receptor, ETRH-A and ETRH-B (Iversen et al., 2002b; Park et al., 2002; Dai and Adams, 2009; Roller et al., 2010), and the first ETHRs were identified in *Drosophila* (Iversen et al., 2002b; Park et al., 2002). The two forms are expressed in different central neurons (Kim et al., 2006a,b). ETHR-A is expressed in inhibitory and/or excitatory neuropeptide producing neurons, releasing the neuropeptides in response to ETH to regulate ecdysis (Kim et al., 2006a,b). In *B. mori* ETHR-B is highly expressed in the CA, pointing to a possible allatoregulatory function (Yamanaka et al., 2008). In *Drosophila*, *Manduca,* and *Aedes* activation of both receptors expressed in CHO cells could increase intracellular Ca2<sup>+</sup> levels (Iversen et al., 2002b; Park et al., 2002; Kim et al., 2006a,b; Dai and Adams, 2009). In *Bombyx*, ETHR-B was expressed in HEK293 cells and was shown to be able to increase intracellular cAMP levels (Yamanaka et al., 2008). In *Tribolium*, the function of the ETRHs was confirmed through RNAi experiments (Arakane et al., 2008). ETRHs were also found in several holo- and hemimetabolous insects (Riehle et al., 2002; Žitnan ˇ et al., 2003; Clynen et al., 2006; Roller et al., 2010).

#### **FMRFAMIDE RECEPTORS (CG2114 ORTHOLOGS)**

The family of (N-terminally extended) FMRFamides is named after the tetrapeptide FMRFamide that was identified in the sunray venus clam *Macrocallista nimbosa* (Price and Greenberg, 1977), but not all extended FMRFamides retain the exact Cterminal motif. The first extended FMRFamide in insects was cloned and characterized in *D. melanogaster* (Nambu et al., 1988; Schneider and Taghert, 1988). More extended FMRFamides were detected by mass spectrometric analysis in various major insect orders (Verleyen et al., 2004a; Neupert and Predel, 2005; Li et al., 2008; Ons et al., 2009; Rahman et al., 2009; Huybrechts et al., 2010; Audsley et al., 2011; Zoephel et al., 2012). FMRFamides modulate heart and gut contractions in insects (Banner and Osborne, 1989; Robb and Evans, 1990; Duttlinger et al., 2002). The FMRFamide neurons become active at the early stages of pre-ecdysis in *D. melanogaster*, suggesting a role in the ecdysis process (Kim et al., 2006b). The *Drosophila* FMRFamide receptor is the only deorphanized insect FMRFamide receptor so far and was found to be activated by six of the seven endogenous *D. melanogaster* extended FMRFamides (Cazzamali and Grimmelikhuijzen, 2002; Meeusen et al., 2002). Orthologous FMRFamide receptors are found in *A. gambiae* (Duttlinger et al., 2003), *A. mellifera*, *N. vitripennis*, *T. castaneum*, *A. pisum,* and *B. mori* but have not been characterized up to date. FMRFamide receptors are conserved throughout insects, but our knowledge about these receptors is very limited.

#### **INOTOCIN RECEPTOR (NP\_001078830 ORTHOLOGS)**

This neuropeptide was first discovered in the 1980s in *L. migratoria* and showed similarity to the oxytocin/vasopressin peptide family in Mammalia. The antiparallel dimer of the neuropeptide was described to have diuretic properties (Proux et al., 1987). Although the neuropeptide could not be identified in most insect species with sequenced genomes, it was recently found in *T. castaneum* and *N. vitripennis*. The mature neuropeptide shows C-terminal amidation. The *T. castaneum* inotocin receptor was characterized in CHO/Gα<sup>16</sup> cells displaying strong activation in the nanomolar range. For both the neuropeptide precursor and its receptor transcript levels have been reported throughout development of *T. castaneum*, but in larvae and the head of adult beetles high levels were detected (Aikins et al., 2008; Stafflinger et al., 2008). Inotocin was shown to act indirectly as a diuretic factor on *Tenebrio molitor* Malpighian tubules in the presence of central nervous system and CC-CA (Aikins et al., 2008).

#### **KININ (MYOKININ) RECEPTORS (CG10626 ORTHOLOGS)**

Insect kinins are small neuropeptides that function as myotropic, neuromodulatory, and diuretic hormones in the insect Malphigian tubules (Hayes et al., 1989; Terhzaz et al., 1999; Coast and Garside, 2005). These neuropeptides, which are characterized by the C-terminal sequence FX1X2WGamide (where X1 is F, H, N, S or Y and X2 is A, P, or S), were first isolated from *Leucophea maderae* (Holman et al., 1987; Hayes et al., 1989). The *Drosophila* kinin receptor was deorphanized in S2 cells using a bioluminescence assay (Radford et al., 2002). Antibodies raised against the receptor identified sites of myokinin action like stellate cells of the Malphigian tubules, two triplets of cells in the pars intercerebralis of the adult central nervous sytem and additional cells in the larval nervous system. Western blots and reverse transcription-PCR confirmed these locations, but also identified expression in male and female gonads. These tissues also displayed elevated Ca2<sup>+</sup> in response to myokinin, demonstrating novel roles for these neuropeptides (Radford et al., 2002). In *A. aegypti* the myokinin receptor was shown to be critical for *in vivo* fluid excretion post blood feeding (Kersch and Pietrantonio, 2011). In *Drosophila* the receptor was shown to be involved in appetite, chemosensory responses, and metabolism (Al-Anzi et al., 2010; de Haro et al., 2010; Cognigni et al., 2011; López-Arias et al., 2011). Receptor orthologs are also present in *A. mellifera* (Hauser et al., 2006), *A. gambiae*, *Culex quinquefasciatus*, *A. pisum*, *P. humanus,* and *B. mori*, but seem to be absent in *N. vitripennis* and *T. castaneum*.

#### **LEUCINE-RICH REPEATS CONTAINING GPCRs (LGRs)**

These receptors, which belong to the rhodopsin-like GPCRs, can be considered "the odd ones out" within this receptor family as they display ectodomains that are much larger than is generally the case for rhodopsin-like GPCRs. Based on the structure of the ectodomain and the hinge region which links the ectodomain to the serpentine domain, three major types can be identified within the LGR family (Hsu et al., 2000; Van Hiel et al., 2012).

#### *Type A LGRs (CG7665 orthologs)*

Type A LGRs typically have 7–9 leucine-rich repeats (LRRs) in their ectodomain. Although little data are available on these receptors in insects, they are thought to be of significant importance as they are homologous to the three vertebrate receptors for the glycoprotein hormones (follicle stimulating hormone, thyroid stimulating hormone, luteinizing hormone, and choriogonadotropin). In contrast to the situation in vertebrates, invertebrate genomes encode only one type A LGR and the receptor is conserved in most sequenced insect genomes, but seems to be lost in Hymenoptera (Hauser et al., 2006, 2010; Fan et al., 2010). Another exception is the *T. castaneum* genome which encodes two type A LGRs (Hauser et al., 2008; Van Hiel et al., 2012).

LGR1 from *D. melanogaster* is activated by a heterodimer formed by GPA2 and GPB5 (Sudo et al., 2005) which are produced in neuroendocrine cells of the ventral nervous system (Sellami et al., 2011). As is the case for the vertebrate glycoprotein hormones, both of these subunits are cystine knot proteins with complex three dimensional structures (Vitt et al., 2001). Based on transcript studies, *dLgr1* gene expression has been detected throughout all developmental stages of the fruit fly (Hauser et al., 1997; Graveley et al., 2011). In wandering larvae and adults, high transcript abundance has been reported for the hindgut and the salivary glands (Chintapalli et al., 2007).

# *Type B LGRs (CG8930 orthologs)*

LGRs from type B feature 16–18 LRRs, about twice the number found in the other two types (Van Hiel et al., 2012). In vertebrates, three type B LGRs can be identified, whereas in insect genomes only one type B has been found. The *D. melanogaster* member of the type B LGRs, LGR2 (*rk*) was cloned in 2000 (Eriksen et al., 2000) and was activated by bursicon (Luo et al., 2005; Mendive et al., 2005). Analogous to the known ligands of the LGR type A receptors, this hormone is a heterodimer of cystine knot glycoproteins. The bursicon hormone itself had already been described in the 1960s (Fraenkel et al., 1966), but it took until 2004 before its sequence was unraveled (Dewey et al., 2004; Honegger et al., 2004). Bursicon was found to induce the hardening and darkening of the cuticle of newly eclosed adult flies as well as the expansion of the wings (Luo et al., 2005; Mendive et al., 2005). More recently, bursicon has been shown to be responsible for the maturation of the wing, driving the epithelial-mesenchymal transition of the wing epithelial cells (Natzle et al., 2008), but the authors reported that apoptosis associated with wing maturation was not bursicon-regulated in contrast to previous results (Kimura et al., 2004). With regard to wing expansion, it has been proposed that the bursicon secreting neurons in the abdominal ganglion are responsible for neurohemal release, whereas the bursiconpositive neurons in the subesophageal ganglion would orchestrate wing expansion behavior (Peabody et al., 2008). Also, there are indications that bursicon is released preceding the initiation of larval ecdysis and that it is responsible for tanning the pupal case (Loveall and Deitcher, 2010). Additionally, recent data indicate that homodimers of the bursicon α- and β-subunits induced innate immunity genes in the fruit fly (An et al., 2012).

In addition to *D. melanogaster*, LGR2 homologues have been identified in representatives of most insect orders including in *A. mellifera*, *T. castaneum,* and *A. pisum* (Hauser et al., 2006, 2008, 2010; Van Hiel et al., 2012). Interestingly, in *A. mellifera* a single gene was found to encode bursicon. This protein features two cystine knot domains similar to the dimer of two cystine-knot proteins as is the case in the fruit fly and the silk moth (Mendive et al., 2005).

# *Type C LGRs (CG31096/CG6857 and CG34411/CG4187 orthologs)*

In contrast to the vertebrate type C LGRs which are activated by members of the insulin-relaxin peptide family, in insects these receptors are largely uncharacterized. In *D. melanogaster*, two members of the type C LGRs can be identified, dLGR3 and dLGR4. In contrast, in *A. mellifera* and *T. castaneum*, only one receptor has been found which is, respectively, most closely related to dLGR3 and dLGR4 (Hauser et al., 2008). The ligands of these receptors are still unknown.

# **MYOSUPPRESSIN RECEPTORS (CG8985 AND CG43745/CG13803 ORTHOLOGS)**

Myosuppressins have a conserved C-terminal FLRFamide. The first myosuppressin was isolated from *L. maderae* (Holman et al., 1986). Myosuppressins inhibit gut contractions and regulate heart contractions (Holman et al., 1986; Lange and Orchard, 1998; Wasielewski and Skonieczna, 2008; Maestro et al., 2011). They also contribute to the regulation of digestive processes by controlling the release of several digestive enzymes in the alimentary canal (Harshini et al., 2002; Hill and Orchard, 2005). Furthermore, myosuppressins inhibit food uptake and thus seem to classify as anorexic factors (Matthews et al., 2008; Vilaplana et al., 2008; Down et al., 2011; Nagata et al., 2011). The first putative myosuppressin receptor was characterized in *L. migratoria*. Cold competition binding studies and kinetic binding assays with a radiolabeled ligand were used to calculate the dissociation constant of the receptor (Kwok and Orchard, 2002). *D. melanogaster* possesses two myosuppressin receptors, DMSR-1 (CG8985) and DMSR-2 (CG43745/CG13803), and were activated by *D. melanogaster* myosuppressin in a dose-dependent manner. Another myosuppressin receptor was characterized in *A. gambiae* (Schöller et al., 2005). Additional myosuppressin receptors have been annotated in *A. aegypti*, *A. mellifera*, *N. vitripennis*, *T. castaneum*, *A. pisum*, and *B. mori*. DMSR-2 is expressed in the head and the body and possibly regulates the actions of myosuppressin on visceral muscles. DMSR-1 is only expressed in the head (Egerod et al., 2003a). Myosuppressin receptors are not

evolutionary related to FMRFamide receptors and both represent two separately evolved signaling systems, despite the resemblance of their ligands (Schöller et al., 2005).

#### **NEUROPEPTIDE F RECEPTORS (CG1147 ORTHOLOGS)**

Invertebrate neuropeptide F (NPF) peptides are structural homologues of the vertebrate NPY family. The *Drosophila* NPF neuropeptide was the first full length member of the NPY/NPF family identified in insects (Brown et al., 1999). The insect NPF neuropeptides are characterized by the consensus sequence xnPxRxnYLx2Lx2YYx4RPRFamide (Nässel and Wegener, 2011). NPF is involved in various processes in *Drosophila* like foraging, feeding, alcohol sensitivity, stress, aggression, reproduction, learning, and locomotion (Shen and Cai, 2001; Wu et al., 2003, 2005a,b; Wen et al., 2005; Lee et al., 2006; Dierick and Greenspan, 2007; Lingo et al., 2007; Chen et al., 2008; Krashes et al., 2009; Xu et al., 2010; Hermann et al., 2012; Shohat-Ophir et al., 2012, for a review, see: Nässel and Wegener, 2011). In several other insects NPF is also (predicted to be) involved in feeding behavior (Zhu et al., 1998; Stanek et al., 2002; Garczynski et al., 2005; Gonzalez and Orchard, 2008, 2009; Nuss et al., 2008, 2010; Ament et al., 2011; Huang et al., 2011b). NPF has also an effect on cardiac activity in the blowfly *Protophormia terraenovae* (Setzu et al., 2012). The *Drosophila* NPF receptor was characterized by means of a radioreceptor approach. The signaling pathway probably acts via Gi and adenylate cyclase as determined by NPF-induced inhibition of forskolin-stimulated cAMP production (Garczynski et al., 2002). The NPF receptor was also characterized in *Anopheles* (Garczynski et al., 2005) and has been predicted in several other insects like *Bombyx* and *Tribolium* (Hauser et al., 2008; Yamanaka et al., 2008; Fan et al., 2010). The proposed *Nasonia* NPF receptor (Hauser et al., 2010) is more likely to be a short NPF receptor; consequently there is probably no NPF receptor present in *Nasonia*. Expression of the *Drome-*NPF receptor was observed in cells of the midgut and numerous neurons in the brain and ventral nerve cord of the third instar larva (Garczynski et al., 2002). The NPF receptor was also located in the adult brain (Wen et al., 2005; Krashes et al., 2009). The *Anoga-*NPF receptor was detected in all life stages except for eggs (Garczynski et al., 2005).

# **PIGMENT DISPERSING FACTOR RECEPTORS (CG13758 ORTHOLOGS)**

The first pigment dispersing factor (PDF) neuropeptide in insects was characterized in *Romalea microptera* (Rao et al., 1987). The best know function of PDF is its role in the circadian clock as a network coordinator, output factor and regulator of its plasticity similar to the vertebrate vasoactive intestinal peptide (VIP). Further processes that where associated with PDF are activity, reproduction, arousal, and geotaxis (for a review, see: Meelkop et al., 2011). Recently, also a role for PDF in the control of visceral physiology in *Drosophila* was described, thereby extending the similarities between fly PDF and VIP in mammals (Talsma et al., 2012). In 2005, three research groups simultaneously identified the PDF receptor in *Drosophila*. Mertens et al. (2005) found the receptor to be specifically responsive to PDF and to couple with Gs, leading to an elevated cAMP concentration upon receptor activation. Mutants showed an aberrant behavioral rhythmicity and a severe negative geotaxis. In a large-scale temperature preference behavior screen in *Drosophila*, Hyun et al. (2005) identified a mutant that preferred colder temperatures during the night and named it *han* (Korean for cold). *Han* seems to be a mutant of a P element controlling the CG13758 gene. But mutations in the latter gene did not cause temperature preference difference. Instead it shows arrhythmic circadian behavior in constant darkness as seen in *pdf* null mutants. PDF specifically binds to S2 cells expressing HAN and thereby elevates the cAMP level. The third research group also identified a mutant with the same disrupted circadian behavior as *pfd* mutants and named it *groom-of-PDF* (*gop*) (Lear et al., 2005). Later studies showed, however, that only the advanced evening activity is common with the *pdf* mutants. *pdfr* mutants, in contrast to *pdf* mutants, did have a morning peak (Im and Taghert, 2010). There are several indications that *pdfr* is regulated at steady-state level by the clock gene *period* (Lear et al., 2005; Mertens et al., 2005). Localization studies showed PDFR expression in the brain and visual system in close correspondence to PDF expression. PDFR expression shows also similarities to the clock pacemaker network of neurons. Furthermore expression is found dispersed in the anterior and posterior surfaces of the central brain and subesophageal ganglion (Shafer et al., 2008; Im and Taghert, 2010). In embryos no expression was noticed (Hyun et al., 2005). *Drosophila* is the only insect where the PDFR has been deorphanized so far. However, homologous sequences are found in many insects like several *Drosophila* species, *A. gambiae*, *A. mellifera*, *N. vitripennis*, *B. mori,* and *T. castaneum*.

#### **PROCTOLIN RECEPTORS (CG6986 ORTHOLOGS)**

Proctolin or RYLPT is a myo- and neurostimulatory neuropeptide of which the appearance seems to be restricted to arthropods (Starratt and Brown, 1975; Nässel, 2002). It stimulates or potentiates muscle contraction, is cardio-acceleratory and acts as a neurohormone (Orchard et al., 1989; Lange, 2002; Clark et al., 2006; Lange and Orchard, 2006; Nässel and Winther, 2010). The *Drosophila* gene for the proctolin receptor was identified and cloned (Egerod et al., 2003b; Johnson et al., 2003a,b; Taylor et al., 2004; Orchard et al., 2011). When the receptor was stably expressed in CHO/Gα<sup>16</sup> cells, a dose-dependent response was measured for proctolin (Egerod et al., 2003b). In competitionbased studies, the proctolin receptor binds proctolin with high affinity (Johnson et al., 2003a). The proctolin and/or proctolin receptor gene was found in the genomes of only a few insect species, including *T. castaneum*, *T. molitor*, *P. humanus,* and *A. pisum* (Hauser et al., 2008; Li et al., 2008; Weaver and Audsley, 2008; Huybrechts et al., 2010). No proctolin gene has been identified in genomes of *A. aegypti*, *A. gambiae*, *A. mellifera*, *N. vitripennis, B. mori,* or *Acromyrmex echinatior* and three other ant species (Hauser et al., 2006, 2010; Roller et al., 2008; Predel et al., 2010; Nygaard et al., 2011), where proctolin and its receptor are now considered absent.

#### **PYROKININ RECEPTORS (CG8784, CG8795 AND CG9918 ORTHOLOGS)**

Pyrokinins are characterized by the C-terminal sequence FXPRLamide (X = S, T, K, A, or G) (Holman et al., 1986; Predel et al., 2001). They are involved in the stimulation of gut motility, the production and release of sex pheromones, diapause, and pupariation (Holman et al., 1986; Predel et al., 2001; Nässel, 2002; Altstein, 2004; Verleyen et al., 2004a; Homma et al., 2006). The pyrokinins can be subdivided into two groups, pyrokinin-1 (C-terminus WFGPRLamide) and pyrokinin-2 (C-terminus PFKPRLamide) (Cazzamali et al., 2005). The first identified insect pyrokinin receptors were those of *D. melanogaster*, where three pyrokinin receptors occur. CG9918 seems to be specific for pyrokinin-1 and CG8784 and CG8795 for pyrokinin-2 (Park et al., 2002; Rosenkilde et al., 2003; Cazzamali et al., 2005). Two pyrokinin receptors were cloned and pharmacologically characterized in *A. gambiae*, one being more specific for pyrokinin-1, the other for pyrokinin-2 (Olsen et al., 2007). The pyrokinin-2 receptor orthologue of *Helicoverpa zea* expressed in *Spodoptera frugiperda* (Sf) 9 cells also responded to pheromone biosynthesisactivating neuropeptide (PBAN) in the low nanomolar range (Choi et al., 2003).

*A. mellifera* has two pyrokinin receptor orthologs, but since they both have the same sequence identities (55–56%) to the *Drosophila* genes, it is difficult to classify them as pyrokinin-1 or -2 receptors (Hauser et al., 2006). The *T. castaneum* genome contains probably three pyrokinin receptors, which are currently classified according to their highest amino acid residue identities (Hauser et al., 2008). Pyrokinin receptors have been found in all insects so far, but it is difficult to classify them as pyrokinin-1 or -2 receptors (Jurenka and Nusawardani, 2011). This will remain problematic until *in vivo* studies using genetics will have solved this issue (Melcher et al., 2006).

#### **RYAMIDE RECEPTORS (CG5811 ORTHOLOGS)**

In 2010, a new class of neuropeptides was discovered from the genome of *N. vitripennis*. These RYamides are characterized by the C-terminal motif FFxxxRYamide (Hauser et al., 2010). Thereupon, RYamides were identified for all insects with a sequenced genome, except for some ant species (Hauser et al., 2010; Nygaard et al., 2011). Recently, the RYamide receptors for *D. melanogaster* and *T. castaneum* were characterized using CHO/Gα<sup>16</sup> cells. Both *Drosophila* RYamides were capable of activating the receptor in the nanomolar range. For *T. castaneum* it was observed that *Trica*-RYamide-2 is somewhat more potent than *Trica-*RYamide-1 to activate the receptor (Collin et al., 2011; Ida et al., 2011a). Although the *Drosophila* receptor was also activated by high concentrations of mammalian NPY and NPYY (Li et al., 1992), a phylogenetic analysis seems to indicate that there is no significant structural relationship between NPY and RYamide receptors (Collin et al., 2011). A first study to unravel the function of the RYamides was performed in *Phormia regina*. Injections of *Drosophila* RYamide-1 attenuate the feeding motivation of these flies (Ida et al., 2011a). The receptor is mainly expressed in the hindgut, while it is not, or hardly present in other investigated tissues in *Drosophila* males and females. This strengthens the hypothesis that the signaling system has a role in digestion, or maybe water reabsorption (Collin et al., 2011; Ida et al., 2011a).

#### **SEX PEPTIDE/MYOINHIBITING PEPTIDE RECEPTOR (CG16752/CG12731 ORTHOLOGS)**

SP induces the post-mating effects that occur in female fruit flies (Kubli and Bopp, 2012). It is produced in the male accessory glands and transferred with the seminal fluid during copulation. It induces egg laying and loss of receptivity for additional mating (Chen et al., 1988), alters the female's sleep pattern (Isaac et al., 2010) and provokes antimicrobial peptide expression (Peng et al., 2005; Domanitskaya et al., 2007). Additionally, the food uptake and preference of females is altered after copulation (Carvalho et al., 2006; Barnes et al., 2007; Kubli, 2010; Ribeiro and Dickson, 2010; Vargas et al., 2010). The SP receptor (SPR) from *D. melanogaster* has been characterized and homologues of this receptor were identified in various insects with the exception of Hymenoptera (Yapici et al., 2008; Kim et al., 2010). Expression of this receptor is found in the female reproductive organs, especially the spermatheca, and the central nervous system of both males and females in very similar patterns (Yapici et al., 2008; Häsemeyer et al., 2009; Poels et al., 2010).

In addition to SP, the related ductus ejaculatorius peptide (DUP) 99B (Saudan et al., 2002) can activate SPR (Yapici et al., 2008). Although both SP and DUP99B have only been identified in *Drosophila* species, they can also elicit physiological responses in the lepidopteran *Helicoverpa armigera* (Fan et al., 1999). As SP and DUB99B so far have only been found in most—not all—*Drosophila* species, the receptor's evolutionary conservation was a puzzle that was only recently solved. SPR can be activated not only by SP, but also by myoinhibiting peptides (MIPs, also known as B-type ASTs) (Kim et al., 2010; Poels et al., 2010). These neuropeptides show the same evolutionary conservation as SPR and therefore likely correspond to the ancestral ligands of SPR. MIPs display a characteristic WX6Wamide Cterminal motif and were first purified from *L. migratoria* (Schoofs et al., 1991), but members of the neuropeptide family were also identified in other species such as *Gryllus bimaculatus*, *D. melanogaster*, and *R. prolixus* (Lorenz et al., 1995; Williamson et al., 2001; Lange et al., 2012). MIPs display myoinhibiting activity in visceral muscle preparations *in vitro* (Schoofs et al., 1991; Blackburn et al., 1995, 2001; Predel et al., 2001). In *G. bimaculatus*, they inhibit JH biosynthesis (Lorenz et al., 1995), and in *D. melanogaster* and *M. sexta*, MIP may silence neurons that are not required during the ecdysis program (Kim et al., 2006a,b). Evidence from *B. mori* indicates that expression of the MIP receptor is strongly upregulated following a sudden decline of the 20-hydroxyecdysone titer. Therefore, MIP receptor signaling may be involved in the fine-tuning of ecdysteroid titers (Yamanaka et al., 2010).

# **SHORT NEUROPEPTIDE F RECEPTORS (CG7395/CG18639 ORTHOLOGS)**

sNPF neuropeptides were first identified in *A. aegypti* and indicated as "*Aedes* head peptides" (Matsumoto et al., 1989). Nowadays, sNPFs are predicted in all insect with a sequenced genome and they are characterized by the C-terminal consensus sequence xPxLRLRFamide (Nässel and Wegener, 2011). The main functions of sNPF seem to be linked to the regulation of feeding behavior (Lee et al., 2004, 2008, 2009; Chen and Pietrantonio, 2006; Kahsai et al., 2010; Ament et al., 2011; Lu and Pietrantonio, 2011; Nagata et al., 2011, 2012b; Root et al., 2011; Hong et al., 2012; Mikani et al., 2012). Other processes in which sNPF is probably involved in are diapause, learning behavior, ovarian growth stimulation, metabolic stress, cardiac activity, the circadian rhythm, and the regulation of hormone production and hormonal release (Schoofs et al., 2001; Huybrechts et al., 2004; Johard et al., 2008; Nässel et al., 2008; Kahsai et al., 2010; Lu and Pietrantonio, 2011; Kapan et al., 2012; Setzu et al., 2012; for a review, see: Nässel and Wegener, 2011). As previously discussed in section "Allatotropin Receptors (NP\_001127714 Orthologs)," sNPF peptides may also possess allatotropic activity (Yamanaka et al., 2008). The first sNPF receptor was identified in *Drosophila* and all four predicted *Drosophila* sNPF peptides activate the receptor in physiological concentrations (Mertens et al., 2002; Feng et al., 2003). Also in *Solenopsis invicta* (Chen and Pietrantonio, 2006), *A. gambiae* (Garczynski et al., 2007), and *B. mori* (Yamanaka et al., 2008) the sNPF receptor has been deorphaned. When co-expressed in *Xenopus* oocytes, the *Drosophila* sNPF receptor activates exogenously expressed inwardly rectifying K+ channels (Reale et al., 2004). The sNPF receptor is present in a limited number of neurons in the nervous system of all developmental stages. Throughout development, the receptor is also expressed in peripheral tissues including the gut, Malpighian tubules, fat body, and ovaries as has been shown in various insects (Mertens et al., 2002; Feng et al., 2003; Chen and Pietrantonio, 2006; Garczynski et al., 2007; Yamanaka et al., 2008; Lu and Pietrantonio, 2011; Kahsai et al., 2012; Nagata et al., 2012b).

#### **SIFAMIDE RECEPTORS (CG10823 ORTHOLOGS)**

SIFamides are highly conserved during evolution and have been isolated from various insects (Verleyen et al., 2004b; Audsley and Weaver, 2006). SIFamide is present in four neurons in the insect pars intercerebralis and this specific pattern suggests a neuromodulatory role in combining visual, tactile and olfactory input. Targeted cell ablation and RNAi has revealed that SIFamide modulates sexual behavior in fruit flies (Terhzaz et al., 2007). The *Drosophila* SIFamide receptor is activated by the SIFamide (Jørgensen et al., 2006). The identification of well-conserved SIFamide receptor orthologs in all insects with a sequenced genome, suggests that SIFamide signaling regulates an essential function in arthropods (Hauser et al., 2006, 2008; Jørgensen et al., 2006; Verleyen et al., 2009).

#### **SULFAKININ AND CHOLECYSTOKININ (CCK)-LIKE RECEPTORS (CG32540/CG6894/CG6881 AND CG42301/CG6857 ORTHOLOGS)**

Sulfakinins (SKs) are the insect homologues of the vertebrate cholecystokinin (CCK) and gastrin neuropeptides (Nachman et al., 1986a,b; Staljanssens et al., 2011). They are named after the sulfated tyrosyl residue in their active core sequence YGHMRFamide that is usually required for biological activity. The first insect SKs were isolated from *L. maderae* and stimulated hindgut contractions (Nachman et al., 1986a,b). Peptidomic techniques elucidated SK peptides in all major insect orders (Verleyen et al., 2004a; Li et al., 2008; Ons et al., 2009; Hauser et al., 2010; Huybrechts et al., 2010; Audsley et al., 2011; Zoephel et al., 2012). SK regulates food uptake and works as a satiety factor that inhibits feeding in several insect species (Wei et al., 2000; Maestro et al., 2001; Downer et al., 2007; Meyering-Vos and Müller, 2007). Drosulfakinins are coreleased with ILPs and influence food choice in *D. Melanogaster* (Söderberg et al., 2012). It stimulates hindgut contractions (Nachman et al., 1986a,b; Marciniak et al., 2011), but inhibits contractions of the heart, oviduct and ejaculatory duct (Marciniak et al., 2011). Only one insect SK receptor, the *D. melanogaster* SK receptor 1 (*Drome*-SKR1) has been deorphanized so far. It is activated by [Leu7]-*Drome*-SK-1 at nanomolar concentrations. [Leu7]-*Drome*-SK-1 was tested instead of the endogenous [Met7]-*Drome*-SK-1 for stability reasons. The sulphate residue is essential for high-affinity receptor binding in all tested cellular assays (Kubiak et al., 2002). SK receptors are widespread in insects: *T. castaneum* and *A. gambiae* have two SK receptors, while *A. aegypti*, *A. mellifera,* and *B. mori* contain at least one.

*Drosophila* contains a second, recently characterized, SK receptor, the CCK-like receptor (Chen et al., 2012). As both SK receptors probably arose through a gene duplication and because of the high homology between the two, it is likely that they also display similar ligand specificity (Hewes and Taghert, 2001; Kubiak et al., 2002). Both, CCKLR and DSK are strong positive growth regulators of the *D. melanogaster* larval neuromuscular junction (Chen and Ganetzky, 2012), by signaling via the cAMP-protein kinase A (PKA)-CRE binding protein (CREB) pathway, known for its role in structural synaptic plasticity in learning and memory (Chen and Ganetzky, 2012). A β-arrestin translocation assay in HEK cells was used to show that sulfated drosulfakinins are the endogenous ligands for CCKLR-17D1. Binding of DSK-1S or DSK-2S to the receptor promotes larval locomotion and evokes stress-induced larval escape behavior (Chen et al., 2012).

#### **TACHYKININ RECEPTORS (CG6515 AND CG7887 ORTHOLOGS)**

Insect tachykinins differ from mammalian tachykinins by their C-terminal consensus sequence, which is FX1GX2Ramide, rather than FXGLMamide as in mammals. There are many different tachykinin isoforms in each insect, which are all encoded by a single gene (Siviter et al., 2000). They play various roles in neuronal signaling and gut activity (Vanden Broeck, 2001; Nässel, 2002; Coast and Garside, 2005; Predel et al., 2005; Van Loy et al., 2010). The first insect GPCR capable of sensing tachykininrelated neuropeptides was cloned from *Drosophila* and is termed *Drosophila* tachykinin receptor (DTKR and CG7887) (Li et al., 1991). *Drosophila* tachykinin-related neuropeptides (*Drome*-TKs) are the endogenous ligands of DTKR and dose-dependently increased intracellular Ca2<sup>+</sup> concentrations, as well as cyclic AMP levels, when applied on DTKR-expressing HEK293 or S2 cells (Birse et al., 2006; Poels et al., 2007). DTKR is involved in the regulation of insulin signaling and the olfactory sensory processing in the antennal lobe (Ignell et al., 2009; Birse et al., 2011).

A second tachykinin receptor in *Drosophila* is the neurokinin receptor (NKD and CG6515) (Monnier et al., 1992). *Drome*-TK-6 (with an Ala instead of Gly) is the only known fly neuropeptide with clear agonist activity on NKD-expressing cells (Poels et al., 2009), which suggests that NKD is able to discriminate between Ala- and Gly-containing isoforms of tachykinin ligands, a feature that does not apply to DTKR (Van Loy et al., 2010). A similar tachykinin receptor has been cloned from *Stomoxys calcitrans* (STKR) (Guerrero, 1997). Its endogenous ligand, *Stoca-*TK, which contains an Ala-residue instead of the highly conserved Gly-residue, behaves as a partial agonist (Poels et al., 2009; Van Loy et al., 2010).

A putative tachykinin receptor has been cloned from brain tissue of *L. maderae* (Johard et al., 2001). One or two tachykinin receptor orthologs have been identified in all insects with a sequenced genome (Hauser et al., 2006, 2008), pointing at an indispensable role of these proteins.

#### **TRISSIN RECEPTOR (CG34381/CG14003 ORTHOLOGS)**

Trissin is a recently identified neuropeptide that contains six Cys residues which form three intramolecular disulfide bridges. Trissin has been shown to activate the *D. melanogaster* GPCR CG34381 stably expressed in CHO/Gα<sup>16</sup> cells at picomolar concentrations. Given the toxic and antimicrobial properties of many Cys containing neuropeptides, one hypothesis is that trissin may have an antimicrobial function. Transcript profiling data for trissin and its receptor on the other hand indicated that transcripts for both were present in the central nervous system of third instar larvae and adults, suggesting that trissin might be a neuropeptide (Ida et al., 2011b). Neuropeptides with high sequence similarity to trissin have been found in several *Drosophila* species and in three mosquitoes, *A. aegypti*, *A. gambiae,* and *C. quinquefasciatus* as well as in *B. mori* (Ida et al., 2011b). In *A. mellifera*, raalin displayed sequence similarity with trissin but this neuropeptide only features 5 Cys residues (Kaplan et al., 2007). Since currently no signal sequence or C-terminal processing sites have been identified for this neuropeptide, its sequence may still be incomplete.

# **REMAINING ORPHAN** *DROSOPHILA* **RECEPTORS**

At least 14 GPCRs predicted to have a neuropeptide as cognate ligand are still orphan and include CG4313, CG12290, CG32547/CG12610, CG13229, CG13995, CG33696/CG16726, CG33639/CG5936, CG30340, and CG13575.

The orphan receptor *hector* (CG4395) is involved in the regulation of *Drosophila* male courtship behavior. It is expressed in numerous brain cells, mainly in the mushroom bodies, the central complex and, at lower levels, in a subset of glomeruli in the antennal lobes. However, only those cells that co-express *fruitless* (*fru*—one of the two main regulators of male courtship behavior besides *doublesex*) and *hector* are critical for male courtship (Li et al., 2011b).

The *moody* gene (CG4322) encodes two splice variants, Moody-α and Moody-β that differ in their long carboxy-terminal domains. Both receptors are coexpressed in glial cells that surround and insulate the nervous system, which is required for the formation and maintenance of the *Drosophila* blood-brain barrier (Bainton et al., 2005; Schwabe et al., 2005; Hatan et al., 2011). The moody receptors are also involved in drug sensitivity (Bainton et al., 2005; for a review, see: Daneman and Barres, 2005).

The receptor encoded by *trapped in endoderm 1* (*tre1*, CG3171) is a functional analog of the CXCR4 receptor of vertebrates, which is involved in tumor metastasis (Kamps et al., 2010). Tre1 is essential for the transepithelial migration of *Drosophila* germ cells from the posterior midgut toward the gonads (Kunwar et al., 2003). The receptor probably plays a role in three phases of early migration: polarization of germ cells, dispersal into individual cells, and transepithelial migration (Kunwar et al., 2008). Tre1 also regulates the relative orientation of cortical polarity in embryonic *Drosophila* neural stem cells (neuroblasts) (Yoshiura et al., 2012).

# **METHUSELAH (CG6936) AND METHUSELAH-LIKE RECEPTORS**

The *methuselah* (*mth*) gene encodes a family B GPCR and is involved in stress response and biological ageing in *Drosophila* (Lin et al., 1998). Also 15 *methuselah-like* (*mthl*) relatives were identified in *Drosophila*, most of them characterized by a unique motif in the extracellular domain consisting of up to ten cysteine residues and several glycosylation sites (West et al., 2001; Patel et al., 2012). Two peptides were identified capable of activating the Mth receptor in a dose-dependent manner. These peptides correspond with the splice variants A and B of the *Drosophila* gene *stunted* (*sun*), which codes for the ε-subunit of mitochondrial ATP synthase (Cvejic et al., 2004; Kidd et al., 2005). The EC50 values for both polypeptides, however, were quite high with 4 μM for Sun A and 3.8μM for Sun B. A second splice variant of the receptor was activated in somewhat lower doses with an EC50 value of 0.6μM for Sun A and Sun B. The *stunted* gene was also shown to be involved in ageing and oxidative stress (Cvejic et al., 2004). Later on, the *Drosophila* SP and a non-physiological peptide with a randomly generated sequence, the serendipitous peptide activator of Mth (SPAM), were also identified as agonists of the Mth receptor. The peptides share almost no sequence homology with Sun A and B, indicating the promiscuity of Mth for activation (Ja et al., 2009). However, *mth* mutants do not affect behaviors controlled by SP, soitis doubtful that activation of theMth receptor by these ligands is of any biological significance (Ja et al., 2009). An extensive developmental expression and sequence divergence study was performed by Patel et al.(2012). More studies are needed to affirm which are the cognate ligands of the Mth receptor and to unravel the physiological roles of the methuselah-like receptors.

# **REMAINING ORPHAN NEUROPEPTIDES**

For several neuropeptides, including the ion transport peptides (ITPs), neuroparsins, orcokinins, and amnesiac, the cognate receptor is unknown at the moment. ITPs function as antidiuretic hormones in locusts (Audsley et al., 1992; Phillips et al., 2001). ITPs are found in the genomes of many insect orders including dipterans, lepidopterans, and coleopterans (Dircksen et al., 2008; Begum et al., 2009; Dircksen, 2009).

Neuroparsins are pleiotropic neuropeptides and are inter alia involved in reproduction and serve as molecular markers of the process of phase transition in locusts (Brown et al., 1998; Girardie et al., 1998; Claeys et al., 2005b, 2006; Badisco et al., 2007). It is noteworthy that in the genus *Drosophila*, the gene coding for the neuroparsins is absent from the *melanogaster* subgroup of the subgenus *Sophophora*, although present in other species of the genus (Veenstra, 2010).

Insect orcokinins were first identified in *B. germanica* and *S. gregaria* (Pascual et al., 2004; Hofer et al., 2005) and were subsequently detected in various other insects, excluding *Drosophila* and *Tribolium* (Roller et al., 2008). A study in *L. maderae*indicates that orcokinins are involved in circadian behavior (Hofer and Homberg, 2006). In *B. mori* it was demonstrated that orcokinins act as prothoracicotropic factors and as such are involved in ecdysteroidogenesis (Yamanaka et al., 2011). Recently, a new family of neuropeptides was discovered in *R. prolixus*, named Orcokinin B, because it arises due to alternative splicing of the *orcokinin* gene. Orcokinin B expression is observed in several insects, except *Drosophila* spp. and *A. pisum* (Sterkel et al., 2012).

The *amnesiac* (*amn*) gene, which encodes a putative neuropeptide precursor (Feany and Quinn, 1995; Moore et al., 1998), is important for stabilizing olfactory memory, and is involved in various aspects of other associative and non-associative learning (Quinn et al., 1979; Gong et al., 1998; DeZazzo et al., 1999; Keene et al., 2004, 2006; Yu et al., 2006; Motosaka et al., 2007). Additional studies have indicated that *amn* is also involved in ethanol sensitivity, sleep, temperature preference behavior and nociception (Moore et al., 1998; Hong et al., 2008; Liu et al., 2008; Aldrich et al., 2010). The *amn* gene was found to code for neuropeptides closely related to the vertebrate pituitary adenylate cyclase-activating polypeptide (PACAP) and submammalian glucagon/growth hormone-releasing hormone (GHRH) and were shown to possess phylogenetically conserved functions (Hashimoto et al., 2002).

# **DISCUSSION AND FUTURE PROSPECTS**

This review clearly shows that during the last two decades a tremendous progress has been made on the field of insect neuropeptide signaling systems. This progress is mostly attributable to the increased availability of insect genomes and the advancing fields of genomics and peptidomics. It became clear that several of these systems are well conserved in all insect species, suggesting that they are indispensable in general insect physiological functions. Other neuropeptides and their receptors were apparantly lost during evolution in several insect species or orders, suggesting that they were otiose, or that their functions were taken over by other ligands. To gain more insight into the evolution of neuropeptide GPCRs across the Insecta, more insect genomes need to be sequenced, which may soon be accomplished due to the i5K project (Robinson et al., 2011). But despite the great progression made in insect endocrinology, the knowledge about the functions of many of the neuropeptides and their receptors involved is still scarce. Furthermore, even in *Drosophila*, the preeminent insect model organism, various receptors are still orphan and the physiological roles they play are still a mystery. It may be obvious that a lot of work has to be performed before the functions of the different signaling systems will be clearly understood and to unravel how these systems are intertwined with each other. This information is also necessary to get a better view on the evolutionary origin of the peptide-receptor couples and how they changed during evolution among species. It should be emphasized that sequence similarity between different insects does not necessarily implies functional similarity or *vice versa*. So, it remains a prerequisite to functionally characterize neuropeptide GPCRs in several insect model species. Reverse genetic tools including RNAi, or the application of the fairly new technique of genome editing using engineered zinc finger nucleases (Urnov et al., 2010) are only some of the methods being developed in several insects, which will likely boost GPCR functional research. Nevertheless, cross genome clustering of receptors based on sequence homology may be a good starting point to acquire a better view on their putative functions (Metpally and Sowdhamini, 2005).

The usefulness of research on insect neuropeptide signaling systems goes beyond the world of insects as several mammalian neuropeptides and/or their receptors have orthologs in insects. Well studied examples of such conserved signaling systems are the GnRH (Lindemans et al., 2011; Roch et al., 2011; De Loof et al., 2012), the tachykinin (Pennefather et al., 2004; Van Loy et al., 2010), the NPF/NPY (Nässel and Wegener, 2011), the capapyrokinin/neuromedin (Melcher et al., 2006; Terhzaz et al., 2012), the AST C/somatostatin (Birgül et al., 1999; Veenstra, 2009a), the myoinhibiting peptide/galanin (Blackburn et al., 1995), the PDF/VIP (Vosko et al., 2007; Talsma et al., 2012), the diuretic hormone/corticotropin releasing hormone (Lovejoy et al., 2009; De Loof et al., 2012), the diuretic hormone/calcitonin (Zandawala, 2012), and the SK/CCK (Staljanssens et al., 2011) orthologs, meaning that these signaling systems arose before the divergence of the Proto- and Deuterostomia (more than 700 million years ago). This strengthens the reasons to study insect endocrinology as these studies can help and learn vertebrate endocrinologists more about the current vertebrate receptors (Grimmelikhuijzen and Hauser, 2012a).

It should be clear that the importance of neuropeptides and their receptors in insect physiology can hardly be overestimated. The great variety among the neuropeptides and their receptors between different insect species makes them also potential targets for the development of a new generation of insecticides with high species specificity (Grimmelikhuijzen et al., 2007; Bendena, 2010; Grimmelikhuijzen and Hauser, 2012b). Such insecticides would ideally only be harmful for pest insects like insects acting as vectors for diseases or herbivorous insects detrimental for agriculture, while beneficial insects would be unharmed. The development of such new controlling agents is necessary because of the detrimental effects of the currently used insecticides on the environment and their toxicity to non-target organisms. The increasing resistance of pest insects against the used insecticides is also an expanding problem (Casida and Quistad, 1998; Van Hiel et al., 2010). As such, more and more research is performed on neuropeptides in order to develop synthetic ligands that can disturb the proper functioning of neuropeptide signaling systems, provoking detrimental effects on the insect's fitness. Once again we want to emphasize the importance of elucidating the biochemical pathways and the functions of the neuropeptides in order to be able to design these so called peptidomimetics. Hitherto, no neuropeptide-based insecticides are in use, but a lot of progress is made on a number of interesting neuropeptides (Teal et al., 1999; Gäde and Goldsworthy, 2003; Altstein, 2004; Scherkenbeck and Zdobinsky, 2009; Altstein and Nässel, 2010; Nachman and Pietrantonio, 2010).

# **ACKNOWLEDGMENTS**

The authors acknowledge the Research Foundation Flanders (FWO-Vlaanderen, Belgium, G.0417.08) and the KU Leuven Research Foundation GOA/11/002. Heleen Verlinden is a postdoctoral research fellow of the FWO-Vlaanderen. Sven Zels is supported by the Flemish government agency for Innovation by Science and Technology (IWT-Vlaanderen, Belgium). Hans Peter Vandersmissen and Kristel Vuerinckx are supported by the KU Leuven.

# **REFERENCES**


during molting by activating the NF-kappaB transcription factor relish. *PLoS ONE* 7:e34510. doi: 10.1371/journal.pone.0034510


peptide receptor DTKR. *J. Exp. Biol.* 214(Pt 24), 4201–4208.


of a *Drosophila* brain-gut peptide related to the neuropeptide Y family. *Peptides* 20, 1035–1042.


neurons mediates ethanol intoxication in *Drosophila melanogaster*. *Neuroscience* 156, 42–47.


roles of *Dippu*-allatostatin in the modulation of hormone release in *Locusta migratoria*. *J. Insect Physiol.* 54, 949–958.


discovered insect RYamide neuropeptides. *Biochem. Biophys. Res. Commun.* 412, 578–583.


inactivation in *Drosophila*. *Nature* 448, 151–156.


Grimmelikhuijzen, C. J. P. (2003a). Molecular cloning and functional expression of the first two specific insect myosuppressin receptors. *Proc. Natl. Acad. Sci. U.S.A.* 100, 9808–9813.


encodes two receptor isoforms with different signaling potential. *EMBO J.* 14, 3373–3384.


potential application for pest control. *Pest Manag. Sci.* 59, 1063–1075.


of the blood-feeding hemipteran, *Rhodnius prolixus*. *J. Insect Sci.* 9, 1–14.


CG14593 and CG30106 code for G-protein-coupled receptors specifically activated by the neuropeptides CCHamide-1 and CCHamide-2. *Biochem. Biophys. Res. Commun.* 404, 184–189.


from *Drosophila melanogaster* structurally related to members of the thyroid-stimulating hormone, follicle-stimulating hormone, luteinizing hormone/choriogonadotropin receptor family from mammals. *J. Biol. Chem.* 272, 1002–1010.


*Schisto*FLRFamide. *J. Insect Physiol.* 51, 1–9.


cAMP signaling in mushroom bodies modulates temperature preference behaviour in *Drosophila*. *Nature* 454, 771–775.


peptide revisited. *Peptides* 32, 483–492.


An evolutionarily conserved arginine is essential for Tre1 G proteincoupled receptor function during germ cell migration in *Drosophila melanogaster*. *PLoS ONE* 5:e11839. doi: 10.1371/journal.pone.0011839


Lehmann, R. (2008). Tre1 GPCR initiates germ cell transepithelial migration by regulating *Drosophila melanogaster* E-cadherin. *J. Cell Biol.* 183, 157–168.


gene in *Drosophila*. *Proc. Natl. Acad. Sci. U.S.A.* 103, 12580–12585.


the mosquito *Aedes aegypti*. *J. Biol. Chem.* 281, 34048–34055.


biological effects of an adipokinetic/hypertrehalosemic peptide in the stick insect, *Baculum extradentatum*. *Peptides* 34, 51–56.


bioinformatics angle. *J. Proteome Res.* 9, 2051–2061.


regulation of physiology and behavior. *Prog. Neurobiol.* 92, 42–104.


*Endocrinol. (Lausanne)* 2:59. doi: 10.3389/fendo.2011.00059


Pierantoni (Bologna: Medimond), 745–752.


*Aedes aegypti*. *J. Proteome Res.* 9, 2006–2015.


diuretic hormone receptor from the house cricket, *Acheta domesticus*. *Insect Biochem. Mol. Biol.* 26, 1–6.


et al. (2003). Molecular cloning, functional expression, and gene silencing of two *Drosophila* receptors for the *Drosophila* neuropeptide pyrokinin-2. *Biochem. Biophys. Res. Commun.* 309, 485–494.


Expression and functional characterization of a *Drosophila* neuropeptide precursor with homology to mammalian preprotachykinin A. *J. Biol. Chem.* 275, 23273–23280.


tribute to Berta Scharrer. *Insect Biochem. Mol. Biol.* 30, 653–662.


predicted prohormone processing. *J. Neurobiol.* 58, 379–391.


receptors reveals a novel LGR subtype. *Peptides* 34, 193–200.


*melanogaster*. *PLoS ONE* 7:e36059. doi: 10.1371/journal.pone.0036059


family members from the mosquito *Aedes aegypti* exhibit differential biological and receptor binding activities. *Mol. Cell. Endocrinol.* 328, 47–55.


new allatostatins from the brains of *Diploptera punctata*. *Insect Biochem. Mol. Biol.* 24, 257–263.


that mediates the post-mating switch in *Drosophila* reproductive behaviour. *Nature* 451, 33–37.


*PLoS ONE* 7:e41543. doi: 10.1371/journal.pone.0041543


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 October 2012; paper pending published: 11 October 2012; accepted: 14 November 2012; published online: 30 November 2012.*

*Citation: Caers J, Verlinden H, Zels S, Vandersmissen HP, Vuerinckx K and Schoofs L (2012) More than two decades of research on insect neuropeptide GPCRs: an overview. Front. Endocrin. 3:151. doi: 10.3389/fendo. 2012.00151*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Caers, Verlinden, Zels, Vandersmissen, Vuerinckx and Schoofs. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Neuropeptide GPCRs in C. elegans

#### **Lotte Frooninckx† , Liesbeth Van Rompay† , Liesbet Temmerman, Elien Van Sinay, Isabel Beets, Tom Janssen, Steven J. Husson and Liliane Schoofs\***

Laboratory of Functional Genomics and Proteomics, Department of Biology, Katholieke Universiteit Leuven, Leuven, Belgium

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Pascal Favrel, University of Caen Lower Normandy, France William Bendena, Queen's University, Canada

#### **\*Correspondence:**

Liliane Schoofs, Laboratory of Functional Genomics and Proteomics, Zoological Institute, Naamsestraat 59, 3000 Leuven, Belgium. e-mail: liliane.schoofs@ bio.kuleuven.be

†Lotte Frooninckx and Liesbeth Van Rompay have contributed equally to this work.

Like most organisms, the nematode Caenorhabditis elegans relies heavily on neuropeptidergic signaling. This tiny animal represents a suitable model system to study neuropeptidergic signaling networks with single cell resolution due to the availability of powerful molecular and genetic tools. The availability of the worm's complete genome sequence allows researchers to browse through it, uncovering putative neuropeptides and their cognate G protein-coupled receptors (GPCRs). Many predictions have been made about the number of C. elegans neuropeptide GPCRs. In this review, we report the state of the art of both verified as well as predicted C. elegans neuropeptide GPCRs. The predicted neuropeptide GPCRs are incorporated into the receptor classification system based on their resemblance to orthologous GPCRs in insects and vertebrates. Appointing the natural ligand(s) to each predicted neuropeptide GPCR (receptor deorphanization) is a crucial step during characterization.The development of deorphanization strategies resulted in a significant increase in the knowledge of neuropeptidergic signaling in C. elegans. Complementary localization and functional studies demonstrate that neuropeptides and their GPCRs represent a rich potential source of behavioral variability in C. elegans. Here, we review all neuropeptidergic signaling pathways that so far have been functionally characterized in C. elegans.

**Keywords: nematoda, Caenorhabditis elegans, G protein-coupled receptor, neuropeptidergic signaling, GPCR deorphanization**

#### **INTRODUCTION**

*Caenorhabditis elegans* is a free-living, microscopic soil nematode. Since its isolation in 1963, this bacterivorous animal acquired the status of model organism in neurobiology (Brenner, 1974). *C. elegans* is easy to cultivate and has a short life cycle of about 3 days at 20˚C. Before reaching its adult form, it goes through four larval stages (L1–L4). Hermaphrodites can either self-fertilize or mate with males, a feature that is commonly exploited during highthroughput genetic studies. The publication of its approximately 100 Mb genome in 1998 made *C. elegans* the first multicellular organism to have its entire genome sequenced (The *C. elegans* Sequencing Consortium, 1998). Genome-wide comparison of the predicted*C. elegans* genes and their vertebrate equivalents revealed an unexpected but notable resemblance between their nervous systems (Bargmann, 1998). The *C. elegans* nervous system comprises only 302 small neurons in adult hermaphrodites. It might seem simple at first sight, but appears to be chemically complex, equivalent to most vertebrate nervous systems. Chemical signaling in *C. elegans* occurs through a group of classical neurotransmitters which are important for synaptic communication and includes acetylcholine (ACh), γ-aminobutyric acid (GABA), glutamate, nitric oxide, serotonin, and other monoamines (Brownlee and Fairweather, 1999). These small-molecule neurotransmitters are packed into synaptic vesicles and subsequently released by exocytosis (Gasnier, 2000;Weimer and Jorgensen, 2003; Scalettar, 2006). In addition to classical neurotransmitters, cell-to-cell communication via chemical signaling in *C. elegans* also occurs through

neuropeptides. Both bioinformatic predictions and peptidomic analyses demonstrated that the *C. elegans* genome comprises a rich diversity of small neuropeptide bioregulators (Li et al., 1999; Nathoo et al., 2001; Pierce et al., 2001; Husson et al., 2005, 2007a; Li and Kim, 2010). This abundant group of over 250 signaling molecules is derived from neuropeptide precursor genes and outnumbers the classical neurotransmitters. As observed in other animal species, one or multiple mature bioactive neuropeptides are generated out of each preproprotein precursor by proteolytic processing and extensive post-translational modifications (Husson et al., 2005, 2006, 2007b; Husson and Schoofs, 2007). Besides their role in key physiological processes, it seems that *C. elegans* neuropeptides are implicated in the modulation of essentially all behaviors including locomotion, reproduction, social behavior, mechano- and chemosensation, learning and memory (Li and Kim, 2008); and may be important for behavioral adaptation throughout evolution (Avery, 2010). Neuropeptides are primarily thought to act as neuromodulators but can also act as fast neurotransmitters. Despite the lack of a circulatory system in *C. elegans*, a neurohormonal role is also ascribed to neuropeptides, and its nervous system harbors a significant number of peptidergic neurosecretory cells (Hartenstein, 2006). It is assumed that possibly all *C. elegans* neurons synthesize and secrete neuropeptides (Holden-Dye andWalker, 2012). Currently, 119 neuropeptide precursor genes are known which can be subdivided into three major families according to the sequence and structural similarities of their derived peptides. Thirty-one neuropeptide-encoding genes

are assigned to the FMRFamide (Phe-Met-Arg-Phe-amide)-like peptide (*flp*) gene family, while 40 genes belong to the family of insulin-like peptide (*ins*) genes. Peptides that bear no resemblance to FMRFamide- or insulin-like peptides are encoded by the family of neuropeptide-like protein (*nlp*) genes. So far, 48 *nlp* precursor genes are known. G protein-coupled receptors (GPCRs) are the principal neuropeptide targets through which intracellular signaling transduction pathways are triggered. GPCRs are defined as seven transmembrane receptors that signal through G proteins. They are found in almost any eukaryotic organism indicating they have an early evolutionary origin (Krishnan et al., 2012). GPCRs have a diverse array of ligands ranging from light, Ca2<sup>+</sup> and odorants to small molecules such as amino acid residues, nucleotides, peptides, and proteins (Pin, 2000). About 7% of all predicted protein-coding genes in *C. elegans* are GPCRs (Bargmann, 1998; Fredriksson and Schiöth, 2005). Most of them (∼1300) encode nematode-specific chemoreceptors,which are thought to compensate for the absence of visual and auditory systems in *C. elegans* (Thomas and Robertson, 2008). The remaining GPCRs can be classified according to the GRAFS classification system and comprise the Glutamate, Rhodopsin, Adhesion, Frizzled, and Secretin families (Schiöth and Fredriksson, 2005). In this review we will focus on the *C. elegans* neuropeptide GPCRs, which belong to the rhodopsin and secretin families. To date, only a limited number of nematode neuropeptide GPCRs of these families have been deorphanized and functionally characterized.

# **UNRAVELING NEUROPEPTIDERGIC SIGNALING IN THE MODEL ORGANISM C. ELEGANS**

The flexible genetic tool box that comes with the use of *C. elegans* as a model has greatly expedited the functional characterization of neuropeptide GPCRs in this organism. Genome-wide RNA interference (RNAi) and mutant analyses have been used to shed light on the behavioral output of neuropeptidergic signaling (Keating et al., 2003; Rual et al., 2004; Ceron et al., 2007). With these techniques, it has become clear that neuropeptides and their receptors influence many if not all of the worm's behaviors. This conclusion is further supported by results from genetic studies on orphan *C. elegans* GPCRs. Recently, Jee et al. (2012) generated mutants for the SEB-3 receptor, an orphan corticotropin-releasing factor (CRF)-related GPCR, and showed that it is strongly implicated in the worm's stress response and ethanol tolerance. This is a perfect example of an earlier finding that several neuropeptide pathways are involved in *C. elegans* responses to ethanol (Mitchell et al., 2010). A genome-wide RNAi study of predicted *C. elegans* GPCRs was performed byKeating et al. (2003) amongst others,which were able to identify a number of neuropeptide receptors involved in reproduction and locomotion.*In vivo* localization of neuropeptide signaling components isfacilitated by the worm's transparency and its simple but well-defined anatomy. The adult hermaphrodite has exactly 959 somatic nuclei ordered in fully differentiated tissues (Sulston et al., 1983; WormAtlas et al., 2002–2012). The developmental origin of every *C. elegans* neuron and the wiring diagram of its roughly 7000 synapses have been completely mapped (White et al., 1986; Jarrell et al., 2012). Using selective promoters that target gene expression to a cell or tissue of interest, combined with laser ablation and cell imaging techniques, gene function and

neural activity can be studied at the level of individual neurons, of which several examples are described below. Together, these powerful molecular and genetics tools enable the dissection of neural networks underlying the neuropeptidergic regulation of behavior with single cell resolution.

# **NEUROPEPTIDE GPCRs IN C. ELEGANS**

Since the publication of the *C. elegans* genome, many predictions have been made about the number of neuropeptide GPCRs it contains. These predictions are usually based on sequence similarities to vertebrate and insect neuropeptide GPCRs (Bargmann, 1998; Fredriksson and Schiöth, 2005). Compared to other *C. elegans* GPCRs, neuropeptide GPCRs appear to be less closely related to their vertebrate counterparts. This agrees with the apparently low conservation of the *C. elegans* neuropeptides (Bargmann, 1998). A crucial step in the characterization of a predicted neuropeptide GPCR is the identification of its natural ligand(s). For this purpose, a reverse pharmacology approach (see Finding Neuropeptide Ligands for Orphan GPCRs) can be applied (Mertens et al., 2004; Beets et al., 2011). Another way to predict neuropeptide GPCRs is to use all deorphanized neuropeptide GPCRs as a seeding set in a Multiple Expectation Maximization for Motif Elicitation/Motif Alignment and Search Tool (MEME/MAST) analysis. Doing so, a list of 125 potential neuropeptide receptors could be obtained (Janssen et al., 2010). Since this prediction, the number of deorphanized neuropeptide GPCRs has increased from 6 to 23. We enhanced this MEME/MAST analysis by use of an updated seeding set containing all newly deorphanized neuropeptide GPCRs and merged our predictions with the list of neuropeptide GPCRs from WormAtlas et al. (2002–2012). Manual verification of every receptor ensured a reliable, revised list of potential neuropeptide GPCRs in *C. elegans* (**Table 1**).

All predicted neuropeptide GPCRs can be grouped in the rhodopsin and secretin families according to the GRAFS classification system. Rhodopsin GPCRs are subdivided based on their resemblance to insect and mammalian neuropeptide GPCRs. The neuropeptide Y (NPY)/RFamide-like receptor family, containing 41 receptors, represents the best characterized group. Twelve of its representatives have been deorphanized and all are activated by FMRFamide like peptides (NPR-1, NPR-3, NPR-4, NPR-5a/b, NPR-6, NPR-10a/b, NPR-11, FRPR-3, and FRPR-18a/b). Their corresponding signaling pathways are involved in a multitude of functions such as locomotion, feeding, energy metabolism, and reproduction. So far, none of the 24 receptors belonging to the somatostatin and galanin-like receptor group have been deorphanized. Only RNAi phenotypes with respect to locomotion and fat metabolism have been observed for this poorly studied group. The tachykinin (neurokinin)-like receptor group contains 12 receptors. Mertens et al. (2006) deorphanized one of these receptors, namely NPR-22a. Remarkably, this GPCR was not activated by the predicted *C. elegans* tachykinin-like peptide but by a handful of FMRFamide-related peptides (FaRPs). The cholecystokinin (CCK)/gastrin-like receptor and gonadotropin releasing hormone (GnRH), oxytocin (OT), vasopressin (VP)-like receptor groups contain 3 and 14 receptors, respectively. Deorphanization of CKR-2a/b and GNRR-1a supports the theory of receptor-ligand coevolution (Janssen et al., 2010). Although no clear CCK or






GPCRs are ranked according to the GRAFS classification system. Deorphanized neuropeptide GPCRs are marked in gray. Underlined neuropeptides are most potent to be the endogenous ligands (<sup>a</sup> see **Table A1** in Appendix for neuropeptide ligand sequences). <sup>b</sup>Putative roles supported by RNAi. Neuropeptide GPCRs are predicted in <sup>c</sup> Janssen et al. (2010), <sup>e</sup>Hewes and Taghert (2001), <sup>f</sup>Harmar (2001), and <sup>g</sup>Cardoso et al. (2006). <sup>d</sup>To complete the list of neuropeptide GPCR we performed a MEME/MAST analysis as described in Janssen et al. (2010).

sulfakinin orthologs could be identified through *in silico* searches, library-based screening led to the identification of NLP-12a and NLP-12b as the endogenous ligands of CKR-2a/b. Alignment of these peptides to vertebrate CCK/gastrin hormones and arthropod sulfakinins revealed their similarity. The endogenous ligand of GNRR-1a was found using an *in silico* approach. GNRR-1a and its ligand, NLP-47, are both involved in reproduction as shown by RNAi experiments (Lindemans et al., 2009a). The VP/OT-like receptor NTR-1 has only recently been identified and deorphanized. The VP/OT-related signaling system is involved in gustatory associative learning and male reproduction (Beets et al., 2012; Garrison et al., 2012). The group of neurotensin, neuromedin U (NMU), growth hormone secretagogue, and thyrotropin releasing hormone (TRH)-like receptors contains 17 receptors, of which three have been deorphanized: NMUR-2 and EGL-6a/b. Only the EGL-6a/b receptors are functionally characterized and they proved to be involved in the regulation of egg-laying (Ringstad and Horvitz, 2008). The secretin family of GPCRs contains nine receptors. Of these, the three pigment dispersing factor (PDF) GPCRs are deorphanized and play a role in locomotion and egg-laying (Janssen et al., 2008a; Meelkop et al., 2012).

# **G PROTEIN SIGNALING IN C. ELEGANS**

G protein-coupled signaling pathways are highly conserved among *C. elegans* and mammals. In the classical G protein signaling pathway (**Figure 1**), the inactive receptor is bound to the heterotrimeric Gαβγ protein. Upon binding of an activating ligand, the receptor changes its confirmation and acts as a guanine nucleotide exchange factor (GEF) by catalyzing the release of GDP and binding of GTP by the Gα subunit. The now activated heterotrimeric Gαβγ protein dissociates from the receptor and splits into a Gα-GTP and a Gβγ subunit. Gα-GTP regulates different effectors depending on the Gα subtype (Gα<sup>s</sup> , Gαi/o, Gαq, and Gα12/13). Gα<sup>q</sup> is known for its activation of phospholipase Cβ (PLCβ), which splits phosphatidylinositol 4,5-bisphosphate (PIP2) into diacylglycerol (DAG) and inositol-1,4,5-trisphosphate (IP3). Binding of IP3 to IP3 dependent calcium channels leads to an increase in calcium, and DAG will bind and activate protein kinase C (PKC). Gα<sup>s</sup> and Gαi/o act through adenylyl cyclase by stimulating (Gαs) or inhibiting (Gαi/o) its activity and thereby regulating the concentration of cyclic AMP, which activates protein kinase A (PKA). Gα12/13 activates Rho dependent pathways. The Gβγ subunit also regulates certain downstream effectors such as ion channels and

PLCβ. G protein signaling is terminated by internalization of the GPCR, which is initiated by phosphorylation through GPCR kinases (GRKs; Ritter and Hall, 2009).

*C. elegans* has homologs for most of the above described G proteins and downstream second messengers. The worm has 21 Gα, 2 Gβ (GPB-1 and GPB-2), and 2 Gγ (GPC-1 and GPC-2) proteins. GPB-1 and GPC-2 seem to be mediators in the classical G protein signaling as the homologs of Gβ and Gγ respectively. For each of the four mammalian Gα subtypes there is a homologous Gα protein in *C. elegans* [GSA-1 (Gαs), GOA-1 (Gαi/o), EGL-30 (Gαq), and GPA-12 (Gα12/13)]. The remaining *C. elegans* Gα subtypes are believed to be specific for chemosensory GPCRs (Jansen et al., 1999; Bastiani and Mendel, 2006). EGL-30 and GSA-1 are the only Gα proteins for which the conservation of their downstream targets has been demonstrated. The classical role of the EGL-30 Gα<sup>q</sup> protein is intensively studied in neuromuscular junctions where it stimulates the release of the neurotransmitter ACh. EGL-30 binds and activates EGL-8, the PLCβ homolog, which splits PIP2 into IP3 and DAG. In neuromuscular junctions, DAG binds to UNC-13 which regulates synaptic vesicle release of ACh through syntaxin (Lackner et al., 1999). IP3 on the other hand can bind to the IP3 dependent calcium channel ITR-1 which leads to a calcium response (Bastiani et al., 2003; Baylis and Vázquez-Manrique, 2012). The Gα<sup>s</sup> protein homolog GSA-1 seems to function through the adenylate cyclase ACY-1. GSA-1 is an essential protein but constitutive activation of GSA-1 in the presence of ACY-1 causes neurodegeneration (Korswagen et al., 1997; Berger et al., 1998). Constitutive expression of rat Gα<sup>s</sup> correspondingly causes the same neurodegenerative phenotype.

# **FINDING NEUROPEPTIDE LIGANDS FOR ORPHAN GPCRs**

To find the activating ligand(s) of a GPCR, a reverse pharmacology approach can be applied. In this approach, the orphan GPCR is expressed in a heterologous expression system. Often Chinese hamster ovary (CHO) or human embryonic kidney (HEK) cells are the recombinant systems of choice because of their ease of use and proven history of functional GPCR expression (Szekeres, 2002). Subsequently, receptor expressing cells are challenged with a library of compounds and activation of the GPCR of interest is measured. The compound library is usually compiled based on bioinformatic predictions and peptidomic analyses of RP-HPLC fractions of a tissue extract (Beets et al., 2011). In the past few years, several successful strategies have been developed for receptor deorphanization (Mertens et al., 2004; Beets et al., 2011). One of the most frequently used methods is probably the calcium mobilization assay based on the detection of intracellular calcium that is released from storage sites upon receptor activation. This method is often combined with the co-expression of a promiscuous G protein, such as the Gα<sup>16</sup> subunit, which can direct the intracellular signaling cascade of the activated receptor through a calcium flux (Offermanns and Simon, 1995). Alternatively, chimeric G proteins can be used to lead the signal cascade to a pathway of choice (Milligan and Rees, 1999). The resulting calcium flux can then be detected by bioluminescent proteins such as aequorin, or by fluorescent calcium indicators (e.g., Fluo-4). In the bioluminescent assay, cells expressing the apoaequorin protein are charged prior to the assay with the cofactor coelenterazine to form a calcium-sensitive aquorin complex.When calcium binds to aequorin, the complex is oxidized and blue light is omitted. Similar to the luminescence assay, receptor expressing cells can be loaded

with a fluorophore, of which the fluorescence increases upon binding of calcium (Mertens et al., 2004). Thanks to the development of automated systems for simultaneous compound addition and signal detection in various well-plate formats, such as the FLEXstation® (Molecular Devices, CA, USA) fluorescent plate reader, calcium mobilization methods can be used in high-throughput screening assays. Once the activating ligand(s) of a receptor are found, the endogenous Gα signaling protein is identified by omitting the promiscuous Gα<sup>16</sup> protein. Coupling of a receptor with Gαq, Gα<sup>s</sup> , or Gα<sup>i</sup> can be visualized by respectively measuring the calcium increase or cAMP in-/decrease.

#### **CHARACTERIZED NEUROPEPTIDERGIC SIGNALING PATHWAYS**

#### **NPR-1 SIGNALING: INHIBITION OF AGGREGATION AND AEROTAXIS**

The neuropeptide receptor 1 (NPR-1) was the first neuropeptide GPCR to be deorphanized in *C. elegans* (Kubiak et al., 2003;Rogers et al., 2003)*.* This receptor shows homology to the vertebrate NPY receptor family that is implicated in a variety of physiological processes such as food intake and stress (Heilig, 2004; Arora and Anubhuti, 2006). In the nematode *C. elegans*, NPR-1 is involved in a multitude of functions such as food-dependent behaviors, thermal avoidance, ethanol tolerance, and innate immunity (de Bono and Bargmann, 1998; Davies et al., 2004; Gray et al., 2004; Cheung et al., 2005; Rogers et al., 2006; Gloria-Soria and Azevedo, 2008; Styer et al., 2008; Glauser et al., 2011; Milward et al., 2011; Jang et al., 2012).

The most explicit function of NPR-1 was elucidated with the observation of aggregating and solitary feeders in wild type isolates of *C. elegans* (de Bono and Bargmann, 1998). This behavioral difference could be attributed to a single amino acid difference. Aggregating isolates carry an *npr-1* Phe-215 allele whereas solitary feeders possess an *npr-1* Val-215 allele. Since a functional null mutation of *npr-1* converts the solitary wild type N2 lab strain into an aggregating one, NPR-1 activity is suggested to suppress aggregating behavior. The RMG inter/motor neuron seems to be the cellular hub of this NPR-1 mediated feeding behavior, as demonstrated by the full rescue of the solitary behavior through RMG-specific expression of NPR-1 in an *npr-1* knockout mutant (Macosko et al., 2009). The RMG neuron is the hub of a gap junction network that connects five sensory neurons which are known to trigger aggregation, while NPR-1 inhibits this gap junction driven activation of RMG (**Figure 2**).

Reduced NPR-1 activity in the RMG-hub-and-spoke circuit also contributes to thermal avoidance and sex-specific pheromone responses in *C. elegans*. Deletion of the NPR-1 receptor increases the threshold for heat avoidance, and cell-specific rescue of *npr-1* demonstrates the role of the RMG interneuron in the regulation of heat avoidance behavior (Glauser et al., 2011). Similarly, RMGspecific rescue of *npr-1* restores pheromone avoidance defects in the *npr-1* mutant background (Jang et al., 2012). Therefore, the RMG neural network can be considered a multifunctional sensory circuit that uses neuropeptide GPCR signaling amongst others to coordinate behavioral output.

In insects and mollusks FaRPs are reported as ligands for NPR-1-like receptors (Tensen et al., 1998; Feng et al., 2003). In 2003, two independent groups were able to deorphanize the NPR-1 receptor by using *C. elegans* and other invertebrate FaRPs. Both FLP-21 and

FLP-18 peptides activated the solitary Val-215 receptor. The social Phe-215 receptor variant could only be activated by FLP-21. NPR-1 signaling occurs through a Gαi/o type of G protein (Kubiak et al., 2003; Rogers et al., 2003). Deorphanization of the NPR-1 receptor supports its role in repressing aggregation, since the solitary Val-215 receptor variant displayed higher binding and functional activity than the Phe-215 receptor variant.

Besides its role in feeding behavior, NPR-1 also regulates aerotaxis (**Figure 3**). *C. elegans* exhibits a strong behavioral preference for 5–12% oxygen, avoiding higher and lower oxygen levels. Oxygen levels are sensed by the URX, PQR, AQR, and SDQ neurons (Gray et al., 2004). Oxygen sensing in these neurons is mediated by soluble guanylate cyclase homologs (GCY-35 and GCY-36). When ambient oxygen levels decrease, cGMP levels rise and the cGMP gated TAX-2/TAX-4 ion channel opens, leading to the depolarization of the neurons. Activation of NPR-1 in the presence of food inhibits the activation of these neurons (Cheung et al., 2005; Chang et al., 2006; Rogers et al., 2006). Oxygen binding globins such as GLB-5 further tune the behavioral responses to varying oxygen concentrations, and this effect is again modified by the NPR-1 receptor (Persson et al., 2009). In addition, it has also been shown that the NPR-1 expressing neurons AQR, PQR, and URX contribute to the enhancement of the worm's sensory perception under hypoxic conditions (Pocock and Hobert, 2010). PQR, AQR, and URX were recently reported to act as tonic receptors that cause long-lasting changes in neural circuit activity that sets *C. elegans* behavior according to ambient oxygen concentrations (Busch et al., 2012), which is also reflected in optimal foraging strategies (Milward et al., 2011). A Ca2<sup>+</sup> relay involving the Ltype voltage-gated Ca2<sup>+</sup> channel subunit EGL-19, the ryanodine receptor UNC-68, and the inositol-1-4,5-trisphosphate receptor

ITR-1 mediate tonic signaling from AQR, PQR, and URX, evoking continuous neuropeptide release.

The role of NPR-1 in the worm's innate immunity was elucidated by Styer et al. (2008), who uncovered an immune inhibitory function for this receptor. Mutations in the *npr-1* gene directly affect the expression of innate immunity markers, suggesting that neuropeptide GPCRs participate in the neuronal regulation of immune responses. In addition, polymorphisms in the *npr-1* gene have been correlated with the worm's pathogen avoidance and susceptibility (Aballay, 2009; Reddy et al., 2009, 2011).

#### **RFAMIDE-LIKE RECEPTOR SIGNALING: FLP-18 SIGNALING THROUGH NPR-4 AND NPR-5**

Both a reverse pharmacology study expressing orphan receptors in CHO cells and an independent*Xenopus laevis* oocyte assay demonstrated that the *flp-18*-encoded peptides are the most potent ligands of NPR-5a and NPR-5b, the splice variants of *npr-5* (Kubiak et al., 2008; Cohen et al., 2009). The latter study also showed that another member of the GPCR rhodopsin family, NPR-4, is also activated by FLP-18 peptides (Cohen et al., 2009), which in addition to their activation of NPR-1 (Kubiak et al., 2003) indicates these are widely deployed ligands of GPCRs. NPR-5a and NPR-5b seem to transduce the FLP-18 signal mainly through a Gα<sup>q</sup> type G protein, while NPR-4 might use a different cellular signaling machinery.

*flp-18(db99)* loss-of-function mutants display chemosensory, dauer formation, and foraging defects, accumulate excess intestinal fat and exhibit reduced aerobic metabolism. Distinct subsets of these phenotypes are phenocopied by *npr-4(tm1782)* and *npr-5(ok1583)* deletion mutants. Each one of the FLP-18 receptors regulates fat metabolism in response to the release of FLP-18 peptides from AIY and RIG interneurons in the head, some of the multiple expression sites of *flp-18*. NPR-4 mediated regulation of intestinal fat occurs at the level of the gut, while NPR-5 modulates the activity of a number of amphid sensory neurons. FLP-18

**FIGURE 4 | Hypothetical model in which the detection of nutrition by sensory neurons (AWC, AFD, and ASE) is coupled to the release of FLP-18 neuropeptides from AIY interneurons and subsequent signaling through the RFamide-like receptors NPR-4 and NPR-5.** By acting on NPR-4 in the intestine and NPR-5 in ciliated neurons, FLP-18 neuropeptides control fat storage. Signaling through NPR-4 in RIV and AVA neurons also modulates responses to odor and foraging behavior. Another food-dependent decision, dauer formation, is regulated by FLP-18 action on NPR-5 in the ASJ neurons (figure adapted from Cohen et al., 2009).

neurohormones released from AIY interneurons act on NPR-4 in AVA and RIV interneurons and appear to be implicated in odor responses and foraging behavior. The chemosensory ASJ neurons regulate dauer formation through activation of NPR-5. All of these observations led to the proposition of a model (**Figure 4**) in which sensory detection of nutritional availability is coupled to adequate responses such as foraging behavior and metabolic alterations via RFamide-like receptor signaling (Cohen et al., 2009).

#### **OFF-FOOD SEARCH BEHAVIOR: FEEDBACK SIGNALING THROUGH NPR-11**

Characterization of the neuropeptide GPCR NPR-11 is a good example of how the knowledge of the entire neuronal wiring diagram makes *C. elegans* a favorable model organism. When worms are removed from a food source, they display a local search behavior characterized by increased turning rates during the first 15 min. This behavior is known to depend on the activity of the AWC olfactory neurons, which release both glutamate and the neuropeptide NLP-1. Glutamate is necessary for increased turning rates during the off-food search behavior of the worm, a behavioral change that is also observed in knockout mutants of *nlp-1*. In glutamatedepleted mutants no increase is noticed, suggesting that NLP-1 acts as a co-transmitter for glutamate by decreasing its effect (Chalasani et al., 2010).

To identify the receptor through which NLP-1 is signaling,Chalasani et al. (2010)looked for orphan GPCRs expressed in neurons that are connected to the AWC sensory neurons. A knockout mutation of NPR-11, resulted in a similar phenotype as displayed by the *nlp-1* mutant. A calcium based assay confirmed the NLP-1/NPR-1 interaction.

Comparison of the calcium response of AWC neurons during the local search behavior upon food removal suggested that NPR-11 activation by NLP-1 evokes a negative feedback loop which dampens AWC activity (**Figure 5**). NPR-1 is expressed in the AIA interneurons which also express the insulin-like peptide INS-1. Indeed, an *ins-1* mutant shows the same increase in turning rates upon food removal as the *nlp-1* and *npr-11* mutants. Calcium imaging of the AWC neurons could confirm the role of INS-1 as a suppressor of AWC activity (Chalasani et al., 2010).

#### **CONSERVATION OF GnRH SIGNALING**

Gonadotropin releasing hormone is mainly known for its role in reproduction in vertebrates (Kah et al., 2007). The GnRH receptor and its ligand are highly conserved in vertebrates and homologs of the receptor are predicted in a variety of invertebrates (Roch et al., 2011). Remarkably, insect GnRH receptor orthologs are activated by adipokinetic hormone (AKH), corazonin, and AKH/corazonin-related peptide (ACP), which are known to be involved in energy metabolism, pigmentation, and cardiac regulation (Park et al., 2002; Staubli et al., 2002; Hansen et al., 2010; Lindemans et al., 2011;Roch et al., 2011). The genome of *C. elegans* is predicted to encode for a family of eight GnRH-related receptor genes (*gnrr-1* to *gnrr-8*). Only one of these receptors (GNRR-1, isoform a) has been deorphanized (Lindemans et al., 2009a). Since *Drosophila melanogaster* AKH (*Dm-*AKH) was capable of activating this receptor, the authors performed an *in silico* search for an AKH-GnRH-like peptide in *C. elegans.* This way, they were able to identify the decapeptide NLP-47 (pQMTFTDQWT) as the endogenous ligand for GNRR-1a (EC<sup>50</sup> = 150 nM; Lindemans et al., 2009a). AKH is known to regulate lipid mobilization during flight in insects (Gäde and Auerwald, 2003). Fat contents were examined by performing an RNAi knockdown of *gnrr-1* and/or *nlp-47*. Unfortunately, no significant differences between knockdowns and wild type were observed (Lindemans et al., 2009a). Nevertheless, injection of synthetic *Ce*-AKH-GnRH into the cockroach *Periplaneta americana* resulted in a significant increase in the levels of hemolymph carbohydrates. A delay in egg-laying could be observed after both *gnrr-1* and *nlp-47* knockdown (Lindemans et al., 2009a). The identification of an AKH-GNRH-like signaling system involved in reproduction is an interesting finding and could be a key to the interplay between reproduction and energy metabolism.

Since no clear ortholog for GnRH was found in insects and nematodes, it was proposed that GnRH has been preserved in lophotrochozoans, but lost in the ecdysozoans (Tsai and Zhang, 2008). Nevertheless, phylogenetic analysis of the ligands of the ecdysozoan GnRH receptors suggests that AKH and corazonin share a common ancestor with GnRH (Lindemans et al., 2011; Roch et al., 2011).

#### **THE NMU-LIKE SIGNALING PATHWAY**

In vertebrates, NMU is a highly conserved neuropeptide that plays a fundamental role in key physiological processes such as smooth muscle contraction, regulation of blood pressure, feeding

chloride channel GLC-3 upon release of the neurotransmitter glutamate from the AWC neurons. Alternatively, when odor is sensed, the AWC neurons release NLP-1, which in turn activates NPR-11 on the AIA interneurons. Upon activation of NPR-11, INS-1 is released, inhibiting AWC activity and thereby reducing its inhibition on AIA (adapted from Chalasani et al., 2010).

and energy homeostasis, stress responses, and immune regulation (Brighton et al., 2004). All NMU peptides isolated in vertebrates have an identical C-terminal pentapeptide (FRPRNamide; Brighton et al., 2004). The presence of an NMU-like receptor in invertebrates was first reported for the fruitfly *D. melanogaster*. The fruitfly genome encodes four NMU receptor homologs. These receptors are activated by pyrokinin neuropeptides (PRXamide) and are involved in many functions such as feeding behavior and visceral muscle contraction (Schoofs et al., 1993; Park et al., 2002; Melcher and Pankratz, 2005).

The *C. elegans* genome encodes four NMU receptor homologs. To date, only NMUR-1 has an assigned phenotype. Wild type *C. elegans* display an altered lifespan depending on the type of food source they live on*.* In 2010, NMUR-1 was demonstrated to be involved in this food source dependent regulation of lifespan (Maier et al.,2010). Sofar, the activating ligand of NMUR-1 has not yet been identified. In contrast, though still a receptor of unknown function, NMUR-2 has recently been deorphanized based on an *in silico* search for *C. elegans* homologs of the *Drosophila* pyrokinin peptides. This revealed three putative PRXamide peptides, all encoded by the same peptide precursor gene *nlp-44.* Only one of these peptides, AFFYTPRI-NH2, could activate NMUR-2 (Lindemans et al., 2009b).

#### **PDF-LIKE SIGNALING: LOCOMOTION AND REPRODUCTION**

In *C. elegans*, the G protein-coupled PDF receptors of the secretin receptor family PDFR-1a, b, c, d, and e represent five splice isoforms of *pdfr-1* (Janssen et al., 2008a; Barrios et al., 2012). Their endogenous neuropeptide ligands PDF-1a, PDF-1b, and PDF-2, encoded by *pdf-1* and *pdf-2*, are all of the NLP-type. All three PDF peptides are able to bind PDFR-1a, b and c; though with significant differences in affinity (Janssen et al., 2008a). PDFR-1a and PDFR-1b signaling occurs via a Gα<sup>s</sup> type of G protein, while PDFR-1c signaling occurs through a Gαi/o type of G protein (Janssen et al., 2009). PDFR-1d and e were only recently recovered from cDNA (Barrios et al., 2012), and have not yet been characterized in detail. The PDF-like neuropeptide pathway is highly conserved in nematodes, and PDF neuropeptides are also found in insects and crustaceans. In the latter, they were initially discovered and named pigment dispersing hormones (PDHs; Rao and Riehm, 1993; Janssen et al., 2009; Meelkop et al., 2011; Temmerman et al., 2011). Furthermore, all three *C. elegans* PDF receptors are closely related to insect orthologs, such as the *D. melanogaster* PDF receptor, and are distantly related to the vertebrate calcitonin GPCRs and vasoactive intestinal peptide (VIP) receptors (Janssen et al., 2008a). In *C. elegans*, the *pdfr-1* gene is expressed in every body wall muscle cell and, like *pdf-1* and *pdf-2*, in neuronal cells that are involved in the sensing and integration of environmental stimuli and the control of locomotion (Janssen et al., 2008a, 2009).

So far, functional characterization reveals that the PDF signaling system of *C. elegans* is involved in both locomotion and egg-laying, which stresses the pleiotropic nature of its biological functions. The *pdf-1(tm1996)* loss-of-function mutant shows locomotion defects by moving slower and executing more reversals than wild type worms. This locomotion phenotype is recapitulated by the overexpression of PDF-2 (Janssen et al., 2008a). *pdf-2 (tm4393)* deletion mutants conduct fewer backward/forward transitions than wild type animals, suggesting that PDF-1 and PDF-2 neuropeptides exert antagonistic effects on locomotion via PDFR-1. Furthermore, the PDF receptor loss-of-function mutant *pdfr-1(lst34)* turned out to have similar locomotion defects as the *pdf-1(tm1996)* mutant. Three splice variants of *pdfr-1* (a, b, c) were proven to be involved in the regulation of locomotion (Meelkop et al., 2012). The PDF system is also implicated in reproduction, as the timing of egg-laying appears to be delayed in *C. elegans pdf-1(tm1996)*, *pdf-2(tm4393)*, and *pdf-2(tm4780)* deletion mutants (Meelkop et al., 2012). The b and d isoforms could rescue a malespecific defect in mate-searching behavior. This defect is mediated through PDF-1 peptides, but not PDF-2; and seems to be needed in gender-shared neurons for the regulation of this sex-specific behavior (Barrios et al., 2012). Functions for PDF signaling in locomotion and reproduction have been demonstrated in other invertebrate species as well (Renn et al., 1999; Helfrich-Forster et al., 2000; Hamanaka et al., 2005). Recently, proteomic analysis proposed the involvement of PDF signaling in lipid metabolism and stress resistance (Temmerman et al., 2012).

#### **THE CCK/GASTRIN-LIKE SIGNALING SYSTEM: FOOD METABOLISM**

Cholecystokinin and gastrin are well-characterized peptide hormones in vertebrates. By acting on two conserved GPCRs, CCK1R, and CCK2R; they are implicated in a variety of digestive functions including the stimulation of digestive enzyme production, intestinal motility, and the promotion of satiety in order to regulate food

intake (Konturek et al., 2003; Dufresne et al., 2006; Clerc et al., 2007). In arthropods, the sulfakinin (SK) family of neuropeptides is both structurally and functionally related to the well-conserved vertebrate CCK and gastrin peptides (Schoofs and Nachman, 2006). A Basic Local Alignment Search Tool (BLAST) analysis of the *C*. *elegans* genome revealed *ckr-1* and *ckr-2* as the homologous genes of the vertebrate CCK/gastrin receptors and their SK counterparts in insects (Kubiak et al., 2002; Meeusen et al., 2003; McKay et al., 2007). The *ckr-2* gene encodes two splice isoform receptors, CKR-2a and CKR-2b, which belong to the rhodopsin GPCR family. By use of a reverse pharmacology approach, the endogenous *C. elegans* NLP-12a and NLP-12b neuropeptides – encoded by the *nlp-12* gene – were appointed the CCK/gastrin-like ligands of CKR-2a and CKR-2b (Janssen et al., 2008b). Signaling of the CCK receptors occurs through a Gα<sup>q</sup> type of G protein. The *nlp-12* gene is expressed in a single tail neuron, identified as DVA, while *ckr-2* is expressed in cholinergic and GABAergic motor neurons (Janssen et al., 2008b; Hu et al., 2011).

The *C. elegans ckr-2(tm3082)* receptor mutant displays decreased intestinal amylase activity/secretion relative to wild type worms, suggesting the involvement of CKR-2 signaling in the stimulation of digestive enzyme secretion. The CCK/gastrin signaling system also appears to be involved in the control of fat storage since *ckr-2(tm3082)* as well as *nlp-12(ok335)* deletion mutants show an increased fat content compared to wild type animals (Janssen et al., 2008b). Both observations are in accordance with the functions attributed to the CCK/gastrin signaling system in vertebrates and the SK signaling system in arthropods. Recently, Hu et al. (2011) suggested a mechanosensory feedback loop (**Figure 6**) for proprioceptive control of normal locomotion, whereby muscle contraction aids the secretion of NLP-12 by the stretch-activated DVA neuron. Subsequent signaling of NLP-12 through CKR-2 enhances presynaptic ACh release to potentiate transmission at neuromuscular junctions and as such adjust the pattern of locomotion. Correspondingly, a significantly reduced locomotion rate was observed for both the *ckr-2(tm3082)* and *nlp-12(ok335)* mutants compared to wild type worm.

#### **AN FaRP SIGNALING PATHWAY INVOLVED IN EGG-LAYING BEHAVIOR**

The *egl-6* gene encodes two GPCR isoforms that are both involved in the inhibition of egg-laying. In comparison with wild type, *egl-6(n592)* and *egl-6* overexpression mutants display slower egglaying rates and longer retention of embryos. Allele *n592* appeared to be a gain-of-function mutation in *egl-6*, increasing its inhibiting activity (Ringstad and Horvitz, 2008). Similar to NPR-11, ligands for EGL-6 were first suggested by looking at neuropeptides displaying the defective egg-laying phenotype of *egl-6* overexpression when they are overexpressed in wild type worms but not in *egl-6* deletion mutants. This way, *flp-10* and *flp-17* turned out to encode for the ligands of EGL-6. In addition, a *Xenopus laevis* oocyte assay demonstrated that FLP-10, FLP-17-1, and FLP-17-2 were able to unambiguously activate the EGL-6 GPCR at nanomolar concentrations (Ringstad and Horvitz, 2008). These FaRPs signal from multiple cell types via EGL-6 in a Gαi/o-dependent manner to inhibit egg-laying (**Figure 7**). In response to environmental cues, FLP-17 neurohormones are principally expressed in BAG sensory neurons and thought to modulate egg-laying behavior by acting

NLP-12 receptor, CKR-2, potentiates transmission at cholinergic neuromuscular junctions, thereby providing a mechanism for proprioceptive control of locomotion (Hu et al., 2011). NLP-12 signaling through CKR-2 also appears to be involved in the regulation of fat storage and digestive enzyme production (Janssen et al., 2008b).

in HSN neurons in order to inhibit egg-laying. Release of FLP-10 by the vulva and spermatheca along with subsequent signaling via EGL-6 also inhibits egg-laying. How parts of the hermaphrodite's reproductive system might inhibit egg-laying is not yet fully understood. In parallel to peptidergic inhibition, cholinergic signals inhibit egg-laying upon unfavorable conditions (Ringstad and Horvitz, 2008).

on EGL-6 in HSN motor neurons. The latter neurons are known to stimulate the action of vulval muscles and are involved in egglaying (Trent et al., 1983; White et al., 1986). The non-neuronal expression of FLP-10 peptides in parts of the hermaphrodite's reproductive system also inhibits egg-laying (Kim and Li, 2004; Ringstad and Horvitz, 2008). Upon unfavorable conditions, signaling through ACh also inhibits egg-laying in parallel to the aforementioned peptidergic inhibition (Ringstad and Horvitz, 2008).

# **VP/OT SIGNALING: GUSTATORY ASSOCIATIVE LEARNING AND REPRODUCTION**

Recently, a VP/OT-related signaling system has been identified in *C. elegans*. In mammals, this system is involved in a plethora of peripheral hormonal functions including water homeostasis, reproduction, and stress responses (van Kesteren et al., 1995; Aikins et al., 2008). These neuropeptides also function as neuromodulators in the central nervous system influencing social cognition and behavior, memory and learning (de Wied et al., 1993; Young and Wang, 2004; Meyer-Lindenberg et al., 2011). In the roundworm, a single VP/OT-like peptide, named nematocin (NTC-1), and two nematocin receptors (NTR-1 and NTR-2) are identified. The NTR-1 receptor is activated by the nematocin peptide in a dose-dependent way. On the other hand, the NTR-2 receptor is not directly activated by NTC-1 but co-expression of NTR-1 and NTR-2 is suggested to affect the intracellular levels of cAMP upon nematocin binding (Beets et al., 2012; Garrison et al., 2012).

In hermaphrodites, the *ntc-1* gene is mainly expressed in the DVA and AVK neurons. Since *ntr-1* is expressed in the left ASE (ASEL) gustatory neurons, the ASH and ADF chemosensory neurons, which function in chemotaxis toward water-soluble cues, Beets et al. (2012)studied the salt chemotaxis behavior of *ntc-1* and *ntr-1* mutants. Similar to wild type worms,*ntc-1* and *ntr-1* mutants are attracted to low NaCl concentrations. When pre-exposed to these low NaCl concentrations in the absence of food, wild type worms show reduced attraction to or avoidance of NaCl, a behavioral switch termed gustatory plasticity (Hukema et al., 2008). However, the aversive response of pre-exposed worms is reduced in *ntc-1* and *ntr-1* mutants. These results indicate that nematocin signaling is implicated in gustatory associative learning, similar to the effects of VP and OT on mammalian cognition. Moreover, AVKspecific expression of *ntc-1* and ASEL-specific expression of *ntr-1* in the *ntc-1* and *ntr-1* mutant background, respectively, partially restored gustatory plasticity. Genetic analysis and supplementation studies indicated that the TRPV channel protein OSM-9, the Gγ-subunit GPC-1 and serotonin and dopamine signaling interact with the nematocin pathway in regulating gustatory plasticity (Beets et al., 2012).

*ntc-1*, *ntr-1*, and *ntr-2* are expressed in sexually dimorphic patterns and have been shown to function in male mating behavior. *ntc-1, ntr-1*, and *ntr-2* mutant males perform poorly in several types of mating behaviors compared with wild type worms. Mutations in the NTR-1 and NTR-2 receptor cause partly overlapping defects in the mating response. Remarkably, cell-specific knockout of nematocin in the mechanosensory DVA neuron, which is not male-specific, seems to be responsible for most of the male mating defects. These findings indicate that nematocin signaling is necessary to coordinate male mating behaviors (Garrison et al., 2012).

# **CONCLUSION**

Despite the simplicity of its nervous system, *C. elegans* displays complex behaviors with a high level of plasticity and striking similarities to the functioning of "higher" nervous systems. The completely defined anatomy and wiring of the worm's nervous system, combined with a rapid life cycle and powerful molecular and genetic tools, have allowed the dissection of neuropeptidergic signaling networks with single cell resolution. As in other animals, neuropeptides in *C elegans* signal through GPCRs. Many predictions have been made about neuropeptide GPCR encoding genes in the *C. elegans* genome. This review brings the number of predicted neuropeptide GPCRs up to 128. High-throughput RNAi and mutagenesis studies of orphan neuropeptide GPCRs in *C. elegans* revealed their involvement in a broad repertoire of behaviors. However, cognate ligands for only 22 of the predicted neuropeptide GPCRs have been identified and only eight of these receptors are functionally characterized. These well-defined

# **REFERENCES**


neuropeptidergic signaling systems play crucial roles in key physiological processes such as reproduction, locomotion, and lipid metabolism, as well as in social and foraging behaviors. The broad functioning of these neuropeptide GPCRs in nematode physiology emphasizes the pivotal role of neuropeptidergic signaling in *C. elegans*. The development of high-throughput deorphanization systems in combination with an advanced genetic toolbox will allow further functional characterization of neuropeptide GPCRs in *C elegan*s, likely increasing our understanding of peptidergic signaling systems in other organisms as well.

#### **ACKNOWLEDGMENTS**

The authors acknowledge the Research Foundation Flanders (FWO-Vlaanderen, Belgium, G.0767.09 and G.0601.11) and the KU Leuven Research Foundation GOA/11/002. Isabel Beets, Liesbet Temmerman, Tom Janssen, and Steven J. Husson are research fellows of the FWO-Vlaanderen.

the neurotransmitter labyrinth in nematodes. *Trends Neurosci.* 22, 16–24.


et al. (2007). Involvement of cholecystokinin 2 receptor in food intake regulation: hyperphagia and increased fat deposition in cholecystokinin 2 receptordeficient mice. *Endocrinology* 148, 1039–1049.


al. (2012). Neuromodulatory state and sex specify alternative behaviors through antagonistic synaptic pathways in *C. elegans. Neuron* 75, 585–592.


A., Siney, E. J., et al. (2003). Whole-genome analysis of 60 G protein-coupled receptors in *Caenorhabditis elegans* by gene knockout with RNAi. *Curr. Biol.* 13, 1715–1720.


PRXamide peptides, CCAP, corazonin, and AKH supports a theory of ligand-receptor coevolution. *Proc. Natl. Acad. Sci. U.S.A.* 99, 11423–11428.


responses and aggregation in *C. elegans. Curr. Biol.* 16, 649–659.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 02 October 2012; paper pending published: 16 October 2012; accepted:* *04 December 2012; published online: 21 December 2012.*

*Citation: Frooninckx L, Van Rompay L, Temmerman L, Van Sinay E, Beets I, Janssen T, Husson SJ and Schoofs L (2012) Neuropeptide GPCRs in C. elegans. Front. Endocrin. 3:167. doi: 10.3389/fendo.2012.00167*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Frooninckx, Van Rompay, Temmerman, Van Sinay, Beets, Janssen, Husson and Schoofs. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# **APPENDIX**

**Table A1 | List of sequences of neuropeptide ligands produced by neuropeptide precursors and sorted according to the neuropeptide C. elegans GPCRs they activate.**



**REVIEW ARTICLE** published: 04 February 2013 doi: 10.3389/fendo.2013.00005

# NeuropeptideY receptors: how to get subtype selectivity

# *Xavier Pedragosa-Badia, Jan Stichel and Annette G. Beck-Sickinger\**

Institute of Biochemistry, Faculty of Biosciences, Pharmacy and Psychology, Universität Leipzig, Leipzig, Germany

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Janine A. Danks, Royal Melbourne Institute of Technology-University, Australia William Colmers, University of Alberta, Canada

#### *\*Correspondence:*

Annette G. Beck-Sickinger, Institute of Biochemistry, Faculty of Biosciences, Pharmacy and Psychology, Universität Leipzig, Brüderstraße 34, 04103 Leipzig, Germany. e-mail: beck-sickinger@uni-leipzig.de

The neuropeptide Y (NPY) system is a multireceptor/multiligand system consisting of four receptors in humans (hY1, hY2, hY4, hY5) and three agonists (NPY, PYY, PP) that activate these receptors with different potency.The relevance of this system in diseases like obesity or cancer, and the different role that each receptor plays influencing different biological processes makes this system suitable for the design of subtype selectivity studies. In this review we focus on the latest findings within the NPY system, we summarize recent mutagenesis studies, structure activity relationship studies, receptor chimera, and selective ligands focusing also on the binding mode of the native agonists.

#### **Keywords: GPCR, NPY,YR, subtype selectivity, ligand side, receptor side**

"fendo-04-00005" — 2013/1/31 — 23:14 — page 1 — #1

# **INTRODUCTION TO THE NEUROPEPTIDE Y FAMILY**

The neuropeptide Y (NPY) family is a multireceptor/multiligand system consisting of four receptors in humans and three polypeptides that bind and activate them with different affinity and potency. The NPY receptors belong to the class A or rhodopsinlike G-protein coupled receptors (GPCR). Five receptors have been cloned from mammals so far, Y1, Y2, Y4, Y5, and y6 but only four of the members are functional in humans (hY1, hY2, hY4, hY5; **Table 1**). The y6 receptor however is active in rabbit and mouse (Starback et al., 2000). The existence of an additional receptor subtype (Y3) was suggested by pharmacological studies of several human, rat, and rabbit tissues including the human adrenal medulla. This receptor subtype is characterized by a much lower affinity for PYY, compared to NPY (Gehlert, 1998; Lee and Miller, 1998). However, since all attempts to clone this receptor subtype were unsuccessful so far, the existence of Y3 is not very likely.

Neuropeptide Y receptors (NPYR) generally couple to Gi or G0 proteins, which leads to the inhibition of adenylate cyclase and finally to the inhibition of cAMP accumulation (Cabrele and Beck-Sickinger, 2000) and modulation of Ca2<sup>+</sup> and K<sup>+</sup> channels (Holliday et al., 2004). Besides this, it has been described that Y2 and Y4 receptors also couple to the Gq protein increasing inositol 1,4,5 phosphate (IP3) production via the activation of the phospholipase C-β (PLC) in rabbit smooth muscle cells (Misra et al., 2004).

Neuropeptide Y, peptide YY (PYY), and pancreatic polypeptide (PP) are the native ligands of the NPY family. NPY is the most abundant peptide in the mammalian brain and has been suggested to adopt a largely open structure. In surface association with phospholipid micelles a flexible N-terminus and a C-terminal alpha helix were identified (Lerch et al., 2004; Parker et al., 2011). However, PYY and PP are suggested to form the typical hairpinlike structure also called PP-fold, a suggestion for pPYY supported by NMR (Keire et al., 2000a; Neumoin et al., 2007), and for PP by the X-ray structure of the peptide (Blundell et al., 1981). Despite some structural differences between the ligands, these polypeptides have a common length of 36 amino acids (**Table 2**) and an amidated C-terminus. Furthermore, these polypeptides share high sequence identity. Whereas NPY and PYY show the highest percentage of common residues with 70%, NPY and PP share only 50% identity (Blomqvist et al., 1992; Zhang et al., 2011). Seven positions in NPY, PYY, and PP are strongly conserved throughout all species: Pro5, Pro8, Gly9, Ala12, Tyr27, Arg33, and Arg35. Apart from these, highly conserved positions are: Pro2, Tyr20, Thr32, and Tyr36 (Cabrele and Beck-Sickinger, 2000). Regarding its pharmacological properties, NPY acts as a neurotransmitter whereas PYY and PP act as neuroendocrine hormones.

The first identified member of the family PP, was isolated from avian pancreas in 1975 (Kimmel et al., 1975). This polypeptide is secreted in the pancreas by PP cells in the Langerhans islets after food ingestion in proportion to the caloric content (Boguszewski et al., 2010; Suzuki et al., 2010). It is thought to act mainly in brain stem and vagal nerve where it promotes appetite suppression, inhibition of gastric emptying and increases in energy expenditure (Asakawa et al., 2003) in addition to direct responses in the gut.

The second member of the ligand family PYY, was isolated from porcine intestinal extracts in 1980 (Tatemoto and Mutt, 1980) and is expressed by entero-endocrine L cells of the distal gut (Lundberg et al., 1982). PYY1−<sup>36</sup> is released in proportion to nutrient intake along the gut and cleaved to PYY3−<sup>36</sup> by the dipeptidyl aminopeptidase VI. The ligand PYY3−36, the predominant form released in the circulation, is selective for Y2 and produces anorexigenic effects (Pittner et al., 2004). This polypeptide acts on peripheral receptors but also on those located in the CNS (Hankir et al., 2011; Schloegl et al., 2011). The last family member, NPY, was isolated from porcine brain in 1982 (Tatemoto,1982) and is one of the most broadly distributed peptides of the central and peripheral nervous system. This peptide is well conserved among different species. It stimulates food intake in response to negative energy balance (Stanley et al., 1986). Additional roles of NPY are decreased bone



"fendo-04-00005" — 2013/1/31 — 23:14 — page 2 — #2

**Table 2 | NPYR: sequence length and ligand preference.**


formation (Baldock et al., 2009; Sousa et al., 2012), regulation of mood and anxiety disorders, the modulation of stress responses (Heilig, 2004), and ethanol intake (Thiele et al., 1998).

Neuropeptide Y family peptides mediate their activity in humans via four receptors. Structurally, these receptors contain two Cys residues in the extracellular regions that form a disulfide bond between extracellular loop I and II. This disulfide bond is a common feature of class A GPCRs and has been confirmed by X-ray crystallography for several members including bovine rhodopsin and the human β2 adrenergic receptor (Palczewski et al., 2000; Cherezov et al., 2007).

The evolution of this system shows that vertebrate ancestors probably had three receptor genes. These genes, possibly located in close proximity in the same chromosomal segment, would be the precursors of the receptor subfamilies. The Y1 subfamily includes the Y1, Y4, and y6 receptors, the Y2 subfamily comprises Y2 and Y7 (in zebrafish and frogs), and the Y5 subfamily consists of only the Y5 due to lack of close relatives of this receptor (Larhammar and Salaneck, 2004). Although theY1 andY2 receptor subtypes have a common pharmacological profile, they only share 27% of sequence identity. Y1 and y6 receptors share the highest sequence identity (51%), whereas Y1 and Y4 receptors share 44% identity, increasing to 56% identity in transmembrane regions (Larhammar et al., 2001). The Y4 receptor conserves 75% of overall identity between human and rat suggesting this protein may be the most rapidly evolving member of the family and the only member that has a selective agonist, the pancreatic polypeptide (Larhammar, 1996; Blomqvist and Herzog, 1997). The Y5 receptor displays low sequence identity, around 30%, with all members of the family. Compared with other GPCRs, neuropeptide Y receptors share a high sequence identity with NPFF1 and NPFF2 receptors, which are members of the RFamide receptor family (Bonini et al., 2000).

The Y1 receptor has 384 amino acids and its main agonists are NPY and PYY. It can be also activated by PP with a minor potency (**Table 2**). The receptor is expressed in the hypothalamus, hippocampus, neocortex, and thalamus (Caberlotto et al., 1997), but is also present in adipose tissue (Castan et al., 1993; Hausman et al., 2008), blood vessels (Cabrele and Beck-Sickinger, 2000), colon, kidney, adrenal gland, heart, and placenta (Wharton et al., 1993). It plays a role in the regulation of food intake (Kanatani et al., 2000b), vasoconstriction of blood vessels (Cabrele and Beck-Sickinger, 2000), heart rate, anxiety (Balasubramaniam, 2002), and bone homeostasis (Sousa et al., 2012).

The Y2 receptor is predominantly expressed in hippocampal neurons, in the thalamus, hypothalamus, and parts of the peripheral nervous system (Widdowson, 1993; Cabrele and Beck-Sickinger, 2000). It is mainly found in pre-synaptic neurons and exerts its action through the regulation of neurotransmitter release (Wahlestedt et al., 1986; Potter et al., 1989). Typical effects correlated with activation of this receptor include enhanced memory retention, the regulation of the circadian rhythm, angiogenesis (Flood and Morley, 1989; Golombek et al., 1996; Gribkoff et al., 1998; Zukowska-Grojec et al., 1998) and bone formation (Baldock et al., 2002). This receptor consists of 381 amino acids and its preferred agonists are NPY and PYY (**Table 2**).

The Y4 receptor subtype is the only member of the family with the endogenous agonist PP, while PYY and NPY can still activate this receptor with minor potency (**Table 2**). It consists of 375 amino acids and is mainly expressed in the gastrointestinal tract (Lundell et al., 1995; Ferrier et al., 2002) but also in the brain (Bard et al., 1995), as well as pancreas and prostate (Lundell et al., 1995). It plays a role in the regulation of feeding (Asakawa et al., 1999; Sainsbury et al., 2010), circadian ingestion and energy homeostasis (Edelsbrunner et al., 2009), colonic transit (Moriya et al., 2010), and stimulation of the luteinizing hormone release (Jain et al., 1999).

The Y5 receptor subtype is expressed in two different splice variants, composed of 445 and 455 amino acids, respectively (**Table 1**). The N-terminus of the longer isoform is extended by 10 amino acids. However, these differences in the sequence of the receptor isoforms do not result in differences in their pharmacological profile (Rodriguez et al., 2003). Both receptor isoforms bind NPY and PYY with comparable affinities. The affinity for PP is slightly lower, but still in the nanomolar range (Gerald et al., 1996). Y5 receptors are mainly expressed in the central nervous system. Tissues with high receptor density include the hippocampus and hypothalamus. The Y5 receptor subtype has been shown to be strongly involved in food intake (Gerald et al., 1996). Other possible roles of the Y5 receptor are the regulation of the circadian rhythm (Matsumoto et al., 1996b; Gribkoff et al., 1998) and reproduction through inhibition of LH release (Raposinho et al., 2001).

The y6 receptor encodes a 371 amino acid protein that has been cloned from rabbit, mouse, and chicken among others (Bromee et al., 2006). However, the sequence in humans and monkeys contains aframe shift mutation in the third intracellular loop, resulting in a non-functional truncated receptor protein (Matsumoto et al., 1996a; Michel et al., 1998).

Taken together, this multireceptor/multiligand system mediates many relevant physiological and pathological processes. This makes the NPY family truly attractive for the design of subtype selective analogs and receptors. Even if selective ligands are pharmacologically the most attractive approach to tackle subtype selectivity, development of receptor chimeras or receptor mutants will also help to understand how the receptors refined their binding pockets during evolution and, as a result, will show how the ligands tend to have distinct affinities for one or the other receptor subtype.

# **DEVELOPING SELECTIVE LIGANDS FOR NPY RECEPTORS**

As it has been previously described, the binding affinity of each peptide differs from receptor to receptor and the role that each receptor plays in regulating physiological processes is different. In light of this, the NPY system is a perfect candidate in which to develop selective ligands and selective receptors to modulate these characteristics.

#### **GENERAL STRATEGIES**

The most conventional way of investigating subtype selectivity is the synthesis of selective ligands. Consequently, to obtain subtype selective ligands, the peptides have to be modified in key positions allowing the investigator to modulate the ligand preference for a receptor. Although the peptides of the NPY family share high sequence homology, they do not necessarily have the same binding mode. The truncation of certain fragments can direct the selectivity to a certain receptor subtype providing information about essential fragments of the peptide. Therefore, one of the approaches to investigate important positions on the peptides are N- or C- terminal truncations.

Another approach to investigate subtype selectivity is the alanine-scan or Ala-scan: this means that each residue in the sequence is one by one individually substituted with Ala. When an Ala occurs naturally in a certain position, this residue is then changed to Gly. In this scan, only the functional groups are substituted permitting the investigation of ionic interactions as well as dipole-dipole and hydrophobic interactions. Once all the analogs are synthesized they must be tested at all the receptor subtypes to determine how the substitution of the native amino acid affects the binding or the activation. In case a residue shows a great loss in binding or activation for a certain receptor, further exchanges in this position can be done. For example the exchange of a certain residue of NPY by the residue present in PP can achieve Y4 receptor binding with the analog. The use of D-amino acids in a scan can provide information about the side chain orientation and steric information concerning ligand binding too, Pro-scans reveal favorable turn-structures and Phe-scans hydrophobic interactions (Lindner et al., 2008a).

As small peptides can adopt several active conformations and these conformations can be recognized by different receptor subtypes in structure-activity relationships, the knowledge of these binding subtypes is of great interest. Furthermore, to investigate the binding mode and receptor preference of small antagonists or non-peptidic drugs, knowledge of the bioactive conformation is of major importance. Constraining the ligand conformation and testing the peptide on several receptor subtypes, can provide information about its bioactive conformation and receptor selectivity. Several strategies can be used to investigate structure activity relationships constraining the conformation of small peptides (Beck-Sickinger, 1997). First of all, non-proteinogenic amino acids can be incorporated, reducing the number of angle combinations that a natural amino acid could adopt, and thereby decreasing the flexibility of the peptide. One example of a non-proteinogenic amino acid is Aib (aminoisobutyric acid). This residue is one the most commonly used in this kind of study. Secondly, the use of several templates and amino acid linkers to induce a desired conformation might be also a good strategy, although this does not always lead to the desired effect because of other amino acids within the sequence. The use of more flexible linkers such as Ahx (6-aminohexanoic acid) or ω-amino alkanoic acids might be a better method to determine the distance between two segments. Finally, the use of cyclization can significantly constrain the conformation of a ligand. Several cyclization techniques can be applied, the most commonly used are: cyclization by disulfide formation between two Cys residues, cyclization by lactamization of N- and/or C-terminus or by the N- and C-group-containing side chains Lys, Orn, Dab, Asp, and Glu and backbone to sidechain cyclization. Recent studies also use click reactions to cyclize peptides using triazoles to mimic disulfide bridges (Holland-Nell and Meldal, 2011) and peptide stapling to increase the propensity toform α-helices, therefore improving pharmacological properties (Verdine and Walensky, 2007).

# **Y1 RECEPTOR**

"fendo-04-00005" — 2013/1/31 — 23:14 — page 3 — #3

N-/C-terminal truncations of NPY confirm the importance of these two segments for Y1 receptor binding. N-terminally truncated analogs are not well accepted by theY1 receptor as can be seen in studies using the shortened sequences NPY(3–36), (13–36), and (18–36). These show only micromolar affinities for this receptor and even the truncation of the first amino acid NPY(2–36), results in a loss of affinity (Beck-Sickinger and Jung, 1995). C-terminal truncations show the importance of the amide group in the binding with the receptor (Hoffmann et al., 1996). Centrally truncated analogs containing the spacer Ahx and structurally constrained analogs showed that the N- and C-terminal fragments must have a certain length to bind with a good affinity to the receptor (Kirby et al., 1993b). Furthermore, using an Ala-scan it was found that, Pro2, Pro5, Arg19, and Tyr20 are important for ligand affinity. Also the amino acids from positions 27 to 36 were found to be crucial for the peptide, especially position 27. Moreover, positions 33 and 35 showed to be extremely important, as Ala analogs at these positions produced a dramatic loss in binding of >5000-fold over wt (**Figure 3**; Beck-Sickinger et al., 1994; Cabrele and Beck-Sickinger, 2000; Lindner et al., 2008b). The importance of Arg<sup>35</sup> was further confirmed as this residue was found to form a subtype-specific ionic interaction with Asp6.59 of the receptor (Merten et al., 2007). The Tyr on position 36 was also found to be relevant for the ligand

binding; this position does however tolerate the exchange to Phe, but not Ala, Bpa, or His. Similar results were obtained using a D-amino acid scan (Kirby et al., 1993a).

Positions 7, 25, 26, 31, and 34 were revealed to be important for subtype selectivity (**Figure 1A**). Modifications in positions 25 and 26 showed that [D-Arg25]NPY and [D-His26]NPY bind selectively to the Y1 receptor (Mullins et al., 2001). Also the introduction of Pro in position 34, present in pancreatic polypeptide, redirected the affinity of the peptide to Y1/Y5 receptors. Apart from Gln34, an additional exchange in Asn7 introducing Phe at this position, a similar residue like the Tyr present on the hPP, yielded [Phe7, Pro34]pNPY. This is a selective Y1 receptor binder and illustrated the importance of an aromatic residue in this position (Soll et al., 2001). Also the combination of Pro34 with an exchange in position 31 by Leu contributes to aY1/Y4/Y5receptor selective profile (Fuhlendorff et al., 1990; Cabrele et al., 2000). All this strongly indicates the importance of N- and C-terminal fragments for the Y1 receptor subtype.

The synthesis of small selective ligands has also been a topic of interest in the past years and many peptides have been synthesized and characterized. Although the first experiments with shortor medium-sized pNPY truncations showed low binding affinity at the hY1 receptor, in recent years several short antagonists, mimicking the NPY C-terminus have been synthesized such as, GR231118 (1229U91 or GW1229), T-241, and T-190. Unfortunately, these ligands also have Y4 agonistic properties (**Figure 2**; Parker et al., 1998). Taking the short NPY analog NPY (28–36) and the antagonist GR231118, Zwanziger et al. (2009) designed a set of 19 short peptide analogs. Only [Pro30, Nle31, Bpa32, Leu34]NPY(28–36) displayed hY1 receptor selectivity and was able to activate the receptor (**Figure 1B**). Follow-up investigations were made by Hofmann and colleagues (Neuropeptides, accepted) on position 32. The authors could further stabilize the peptide by replacing Bpa by Bip (biphenylalanine) and could switch the activity from hY1 receptor to hY2/hY4 receptors by introducing an ortho-carbaboranyl moiety. Other small peptide antagonists are BW1911U90 and [32−34βACC]-NPY(25–36)]; **Figure 2**; Koglin et al., 2003), and examples of known non-peptidic antagonists are BIBP3226, BIBO3304, LY357897, J-104870 (**Figure 4A**; Rudolf et al., 1994; Hipskind et al., 1997; Wieland et al., 1998; Sjodin et al., 2006; Antal-Zimanyi et al., 2008).

# **Y2 RECEPTOR**

As with the human Y1 receptor, the Y2 receptor binds NPY and PYY with comparable affinities. Beside these two native highaffinity ligands, a number of Y2-selective NPY-derived peptide

"fendo-04-00005" — 2013/1/31 — 23:14 — page 4 — #4

"fendo-04-00005" — 2013/1/31 — 23:14 — page 5 — #5

agonists have been synthesized in the past. Interestingly, in contrast to all other Y receptors,Y2 receptors allow large truncations of the peptidic ligands without loss of affinity (Beck-Sickinger and Jung, 1995) and also cyclizations between N- and C-terminally located residues are tolerated (Kirby et al., 1993b). Most commonly usedY2 receptor selective NPY-analogs are the N-terminally truncated NPY(3–36) and NPY(13–36). Even larger N-terminal truncations and centrally truncated analogs can bind to the Y2 receptor with nanomolar affinity [e.g., NPY(18–36), NPY(22–36), [Ahx5−24]NPY; **Figure 1B**; Beck et al., 1989; Fournier et al., 1994; Beck-Sickinger and Jung, 1995; Keire et al., 2000b]. An Ala-scan of the complete NPY peptide revealed only few positions to be highly important (**Figure 3**).

The substitution of Pro<sup>5</sup> to Ala led to a 600-fold loss of affinity. Accordingly, all other important residues except Pro5 are located in the C-terminal part of NPY. The individual substitution of Arg19, Tyr20, Tyr27, and Asn29 in the NPY peptide showed a 30- to 40-fold lower affinity. A more dramatic effect could be observed for the residues Leu<sup>31</sup> (1000-fold lower affinity), Arg<sup>33</sup> (1350 fold), Gln34 (150-fold),Arg<sup>35</sup> (75000-fold), and Tyr36 (17500-fold; **Figure 3**; Cabrele and Beck-Sickinger, 2000; Eckard et al., 2001). Interestingly, the introduction of a Pro residue at position 34 is not tolerated at the Y2 receptor, which is in contrast to the effect observed on the other Y receptor subtypes (Beck-Sickinger et al., 1994; Keire et al., 2000b; Eckard et al., 2001). Although Tyr<sup>36</sup> may not be substituted by Ala, the introduction of Hty (homotyrosine) or p-substituted Phe in PYY(3–36) is well tolerated at the Y2 receptor, but almost completely abolishes binding of the modified NPY analogs at Y1 or Y4 receptors (Pedersen et al., 2009). Taken together, these data underline the importance of the Cterminal part of the peptide ligand for high-affinity binding to the Y2 receptor, despite the fact that the binding pocket for NPY at the Y2 receptor seems to be less narrow than the ones of Y1 or Y4 receptors.

A number of selective high-affinity antagonists at the Y2 receptor have been published so far. The most widely used compound in pharmacological studies is BIIE0246 (**Figure 5A**; Doods et al., 1999). In order to identify compounds with improved biostability, bioavailability, and brain permeability, further studies have been conducted. A number of molecules and scaffolds have been reported as highly selective and affine small molecule Y2 receptor antagonists (**Figure 5A**) including JNJ-527787 (Bonaventure et al., 2004; Jablonowski et al., 2004), SF-11, SF-21, SF-22, SF-31, SF-41 (Brothers et al., 2010), ML072 to ML075 (Saldanha et al., 2009), JNJ-31020028 (Shoblock et al., 2010; Swanson et al., 2011), a series of substituted 3-chloranilides (Lunniss et al., 2009, 2010), CYM 9484, and CYM 9552 (Mittapalli et al., 2012).

# **Y4 RECEPTOR**

The Ala-scan of the NPY (Eckard et al., 2001) revealed that again Arg<sup>33</sup> and Arg<sup>35</sup> are crucial for receptor affinity. Ala substitutions in these positions led to a dramatic loss in binding. Positions

"fendo-04-00005" — 2013/1/31 — 23:14 — page 6 — #6

Tyr20, Tyr27, Arg25, Thr32, and Tyr36 are also important residues in the ligand and showed a loss in binding affinity (30- to 60-fold), whereas Pro5, Pro8, and Tyr21 proved to be less relevant, causing only a slight loss in affinity (5- to 10-fold) when changed to Ala (**Figure 3**).

In a follow-up study using hPP (Merten et al., 2007), Arg residues 33 and 35 were confirmed to be essential for receptor activation, showing a dramatic effect when exchanged to Ala in position 33. Position Arg35 was found to interact with Asp6.59 of the receptor.

Because this receptor subtype has its own selective ligand, peptide research is more focused on improving proteolytic stability and increasing bioavailability of the peptide. However, a number of specific ligands have been published in the past years for this receptor. As previously described, position 34 of NPY peptides is a key residue to introduce Y4 receptor selectivity to NPY and PYY, whereas in PP when exchanging Pro<sup>34</sup> for Gln the peptide acquires Y2 agonistic properties without losing Y4 receptor activity. Some of the analogs published like [Gln34]-hPP, the so called Obinepitide (Schwartz, 2006), which is selective for Y2 and Y4 receptors, contains this exchange (**Figure 1C**). Other small peptide agonists described also as Y1 receptor antagonists are: GR23- 1118 (1229U91 or GW1229), T-241, and T-190 (**Figure 2**; Schober et al., 1998).

To our knowledge, only one ligand with antagonistic properties at the hY4 receptor has been published up to now. UR-AK49 (**Figure 4B**) is a weak hY4 receptor antagonist but unselective, because it can also bind to hY1 and hY5 receptors (Ziemek et al., 2007).

# **Y5 RECEPTOR**

The hY5 receptor subtype does not tolerate large truncations of NPY. While the deletion of the first amino acid is accepted by the hY5 receptor, further N-terminal truncation of NPY results in a decreased affinity of the peptides. Similarly, larger central truncations of NPY are not tolerated by Y5 receptors. The only centrally truncated analog of NPY with high Y5 receptor affinity is [Ahx9−17]pNPY with a <sup>∼</sup>15-fold decreased affinity compared to pNPY (Cabrele and Beck-Sickinger, 2000).

An Ala scan of the complete peptide revealed the Pro residues 2, 5, and 8 to be important for the affinity of NPY at the Y5 (**Figure 3**). Haack et al. (2008) could confirm the importance of the peptide N-terminus for high-affinity binding at the Y5 receptor by a pyridone dipeptide scan. In addition to these findings, the individual substitution of Tyr residues 20, 21, 27, and 36 to Ala led to a loss of affinity; Arg<sup>25</sup> was also shown to be important for the ligand affinity. The highest impact could be observed for Tyr27 (∼400-fold) and Arg<sup>35</sup> (1000-fold; Cabrele and Beck-Sickinger, 2000), these findings fit with the fact that these two positions are involved in interactions with the receptor.

A number of selective high-affinity analogs of NPY and PYY have been developed in the past. It has been shown that especially the substitution of position 32 of the peptide ligands is critical for Y5 selectivity (**Figure 1A**). [D-Trp32]NPY and [Ala31, Aib32]NPY have been reported to be highly selective and potent agonists of Y5 receptor (Parker et al., 2000; Cabrele et al., 2002). However, the most potent and selective activator of the Y5 receptor

"fendo-04-00005" — 2013/1/31 — 23:14 — page 7 — #7

subtype is a chimeric peptide derived from chicken PP, human NPY, and human PP ([cPP1−7, NPY19−23, His34]hPP; Cabrele and Beck-Sickinger, 2000). In addition, Y5 receptors display high-affinity to some NPY analogs which also have a considerable affinity for other Y receptor subtypes. [Leu31,Pro34]pNPY is a Y1/Y4/Y5 selective agonist (Fuhlendorff et al., 1990; Widdowson et al., 1997), whereas the deletion of the first Tyr residue results in the Y2/Y5 selective agonist NPY (2–36) (Gerald et al., 1996).

Since NPY has been shown to stimulate feeding via the Y5 receptor, intensive research has been performed to identify small molecule antagonists of the humanY5 receptor as potentialfeeding suppressors. The first compound that was published as an antagonist of Y5 receptor was CGP71683A (**Figure 5B**), which displayed high-affinity and selectivity at the rY5 receptor (Criscione et al., 1998). Later studies confirmed the high-affinity and selectivity also for the hY5 receptor subtype (Rueeger et al., 2000). Other selective small antagonists at the human Y5 receptor include L152,804 (Kanatani et al., 2000a), FMS586 (Kakui et al., 2006), MK-0557 (Erondu et al., 2007), and SCH 500946 (Mullins et al., 2008). In the last 5 years, more than 10 studies have been published presenting newly identified or improved small molecule antagonists of the Y5 receptor. This clearly shows the importance of the Y5 receptor as an anti-obesity target.

# **HOW TO IDENTIFY RELEVANT RESIDUES ON THE RECEPTOR FOR BINDING AND SUBTYPE SELECTIVITY GENERAL STRATEGIES**

As these receptors consist of 350–450 residues it is impossible to perform a single mutagenesis approach to investigate each amino acid. In order to overcome this problem chimeric receptors can be used, in which fragments of a receptor (e.g., extracellular loops or transmembrane helices) can be exchanged between receptor subtypes. Testing these new constructs with the main agonist from both receptor subtypes can provide information about the importance of one or the other segment, in terms of interaction with the ligand. As soon as an important area in a receptor has been identified, a more detailed study can be carried out using single and multiple mutants, where certain residues of a receptor subtype are exchanged by the ones present on the other receptor subtype of investigation. This strategy allows to find amino acids that may play a role in selectivity to a certain agonist.

When investigating the relevance of the N-terminus, successive truncations or substitutions using tags or spacers can be an elegant method. Furthermore, C-terminal truncations can provide information about segments relevant for internalization. It is known that this receptor part is involved in arrestin-dependent internalization processes of Y1, Y2, and Y5 receptors (Walther et al., 2011). However, single mutagenesis techniques can be used to investigate important residues for the structure or for ligand receptor interactions. Using this approach, certain residues located in extracellular areas of the protein are mutated to Ala or other amino acids in single substitutions. The residues can be chosen according to its location, charge, aromaticity, hydrophobicity. Moreover, 3-D models are also a good tool to select new targets, although mutagenesis data are needed to refine the models and make them more reliable. Once a relevant residue has been identified, double

cycle mutagenesis can be used to find the type of interaction that includes both positions. In this technique, peptide analogs containing modifications in positions of interest are investigated with receptor mutants. The aim is to form artificial bonds to proof a native interaction. The introduction of charged residues in the peptide and receptor positions to create a repulsion/attraction, or aromatic residues and hydrophobic residues are feasible ways to prove a ligand receptor interaction. In order to finally prove a ligand-receptor interaction, a reciprocal mutation approach can be followed, where the residue of interest on the peptide side is exchanged by the residue present on the receptor side and *vice versa*. In the case of a critical position or segment, the binding affinity of the native ligand should significantly decrease, whereas in signal transduction assays the EC50 or half maximal activation value should increase. Despite all the advantages that these approaches provide, it has to be taken into account that they also present some disadvantages. Thus, when constructing receptor chimeras or receptor mutants, alterations in the receptor structure may arise due to misfolding and therefore may lead to impaired receptor export. Moreover, these modifications might result in a reduced receptor retention time at the cell surface and an enhanced degradation. All together, this might lead to a loss in binding or receptor activity. In order to analyze such phenomena, fluorescence microscopy, cell surface ELISA or radioligand binding studies are a good tool to ensure cell surface expression.

## **Y1 RECEPTOR**

In the past years, much effort has been made to characterize this receptor. Using N-terminal truncations and receptor chimera, it could be elucidated that the N-terminal part of NPY receptors does not participate in the binding pocket. N-terminal truncation in the hY1 receptor disrupts the membrane expression; however any eight residues are enough to recover the membrane expression (Lindner et al., 2009).

From all these studies, a number of residues emerged as important for the receptor. First of all, two negatively charged residues are able to establish electrostatic interaction. Asp2.68 and Asp6.59 were found to be important for the receptor as the peptide loses affinity when mutated to Ala (Sautel et al., 1995, 1996; Kannoa et al., 2001; Sjodin et al., 2006). Furthermore Asp6.59 was shown to bind Arg35 of pNPY being the first and only proved interaction for this receptor (Merten et al., 2007). Other residues that appeared to be important in several studies are Tyr2.64, Phe6.58, and His7.31 (Sautel et al., 1995, 1996; Kannoa et al., 2001; Sjodin et al., 2006), although a direct interaction was never established for any of the amino acids of the ligand. It was suggested that these residues form a hydrophobic pocket in the receptor. Further investigations using the Y1 receptor antagonist BIBP 3226 showed that Tyr2.64 and His7.31 did not affect the conformation of the receptor in a major way (Sautel et al., 1996) as the antagonist was perfectly bound. Taken together, it is very likely that position 6.58 and 7.31 interact with a C-terminally located amino acid, on the other hand it is unlikely that Tyr2.64 interacts with the C-terminus as it seems to be too far from the other two residues.

Other relevant residues of the receptor are Trp6.60, Asn6.55, and Asn7.32 (Kannoa et al., 2001). Although position 6.55 is in a slightly deeper position, Asn6.55 and Asn7.32 showed a loss in PYY binding and also in antagonist binding suggesting that they could play a role in ligand binding (**Figure 6A**).

Studies with antagonists indicate that the binding of these compounds differs depending on the ligand between transmembrane helices 3 and 7. Taking all the data into consideration, it can be assumed that the binding pattern of the native ligands and the small antagonists overlaps in TM6 because several residues have been found to be relevant in both cases.

# **Y2 RECEPTOR**

"fendo-04-00005" — 2013/1/31 — 23:14 — page 8 — #8

In this receptor subtype, the N-terminus does not play a role in membrane expression and it does not participate in a subtype specific binding pocket. However, it does play a role in agonist induced internalization processes since the complete truncation slowed down the process, although it could be seen that the exchange of the N-terminal fragment by the hY1 receptor or hY5 receptor fragment did not affect ligand dependent internalization (Lindner et al., 2009).

Mutagenesis studies to identify residues that contribute to ligand binding in the Y2 receptor were initially motivated by the finding that human and chicken Y2 receptors show a significantly different pharmacological profile. The chicken Y2 receptor is able to bind [Leu31,Pro34]-NPY, a peptide agonist selective for mammalian Y1/Y4/Y5 receptors, but was unable to bind BIIE0246, a small molecule antagonist for mammalian Y2 receptors (Salaneck et al., 2000). Sequence comparison and reciprocal mutagenesis revealed three residues in transmembrane helices 3, 5, and 6 that contribute to the binding of BIIE0246. Individual and combined substitution of Gln3.37, Leu5.51, and Leu6.51 in the hY2 receptor decreased the affinity for BIIE0246 to a chY2-like level, whereas substitution of the corresponding residues in the chY2 by the human residues increased the affinity for BIIE0246 (Berglund et al., 2002). Further mutagenesis studies on the human Y2 receptor revealed interaction partners for the native peptidic ligand NPY. Several acidic residues have been tested for their importance for NPY binding. Glu5.27 and Asp6.59 turned out to be highly important for the binding of NPY (**Figure 6C**). While Asp6.59 is important for all Y receptor subtypes, Glu5.27 only plays a role in the Y2 receptor. Both receptor mutants were tested in a signal transduction assay using pNPY, [Ala25]pNPY, [Ala33]pNPY, and [Ala35]pNPY to identify the interaction partner of the two acidic residues in the peptide. It could be shown that Asp6.59 interacts with Arg33 of the peptidic ligand in the Y2 and Y5 receptors, whereas the interaction partner in Y1 and Y4 receptors is Arg35. However, no direct interaction partner could be identified for Glu5.27 (Merten et al., 2007). More recent studies investigated additional residues in the Y2 receptor for their impact on the binding of pPYY, pNPY, hPYY (3–36), pNPY(13–36), and the nonpeptidic antagonist BIIE0246 (Akerberg et al., 2010; Fallmar et al., 2011). The residues tested, namely Tyr2.64, Gly2.68, Thr3.40, Leu4.60, Gln6.55, Val6.58, and Tyr7.31, were chosen by similarity to residues in the Y1 receptor subtype, which were proven to be important for ligand binding in this receptor subtype. It could be shown, that of the tested residues, only Tyr2.64 participates in the binding of all tested peptidic ligands and

"fendo-04-00005" — 2013/1/31 — 23:14 — page 9 — #9

the non-peptidic antagonist BIIE0246. The substitution of this residue to Ala resulted in a five- to ninefold reduction in affinity (Akerberg et al., 2010).

The individual substitution of Tyr7.31 by Ala and Gly2.68 by the bigger and more polar residue Asn revealed a lower affinity only for the truncated peptide agonists. The authors hypothesize that Tyr7.31 does not play a role in binding of the full-length peptide, but may contribute to a compensatory interaction for ligands that lack the N-terminal residues. Furthermore, the authors could show that an introduction of a His residue in position 7.31 (the corresponding residue in Y1 receptor) completely abolished the binding of [125I]-pPYY (Akerberg et al., 2010). These findings are somewhat unexpected, since this His residue was shown to be involved in ligand binding in the Y1 receptor (Sjodin et al., 2006). This indicates that position 7.31 is important in both receptor subtypes, but may have different modes of action (Akerberg et al., 2010). For position 2.68, a mode of binding is proposed in which the lack of Asp (a residue common to all other Y receptor subtypes at this position) contributes to the selectivity of truncated peptides [e.g., NPY(3–36)] for the Y2 receptor (Fallmar et al., 2011). The Leu4.60Ala mutant showed a slightly decreased affinity for hPYY(3–36) and a strong loss of affinity for BIIE0246, which may be caused by a weakened or lost hydrophobic interaction. This indicates that this residue is highly important for antagonist binding. The corresponding position in Y1 receptor (Phe4.60) has been shown to be involved in the binding of [125I]NPY, [3H]BIBP3226 (Sautel et al., 1996) as well as [3H]J-104870 (Kannoa et al., 2001). This indicates that position 4.60 is involved in the binding of small molecule antagonists at both receptor subtypes, Y1 and Y2.

The Y2 receptor mutants Thr3.40Ile and Gln6.55Ala showed increased affinity for pNPY and hPYY(3–36), but decreased affinity for the non-peptidic antagonist BIIE0246. Taken into account that these positions are located deeper in the transmembrane part of the receptor, an indirect effect on the binding of peptidic ligands would be the most likely explanation. However, the decreased binding of BIIE0246 may be also explained by a different binding pocket for small molecule antagonists, located more deeply in the transmembrane (Fallmar et al., 2011) and surrounded not only by Thr3.40 and Gln6.55, but also by the nearby residue Gln3.37, which was earlier shown to participate in the binding of BIIE0246 earlier (Berglund et al., 2002).

# **Y4 RECEPTOR**

N-terminal truncations and substitutions revealed the importance of this fragment for membrane expression and indicated that the N-terminus is not involved in forming a specific binding pocket. It is likely that this part could stabilize the TM1 to ensure the correct receptor structure (Lindner et al., 2009).

Two positions were investigated to find the binding partners on the ligand side (Merten et al., 2007). Glu5.24 was mutated to Ala in order to test the influence of the side chain, Glu5.24Ala showed a threefold loss in potency. On the other hand, Asp6.59 was mutated to Ala, Glu, Asn, and Arg to test the influence of charge and length of the side chain. The mutation to Ala showed a complete loss of both binding and activity, the exchange to Glu displayed wildtype-like binding and activation. In addition, the mutation to Asn showed indeterminate binding and a 200-fold loss in activation. Finally the exchange to Arg resulted in a dramatic loss in potency (>600-fold) and in no detectable binding (**Figure 6B**).

#### **Y5 RECEPTOR**

The Y5 subtype N-terminus could play a role in ligand binding, since the partial truncation of the segment produced a loss in activation. Interestingly, the receptor remains on the membrane even when the complete N-terminus is removed (Lindner et al., 2009).

Only few mutagenesis studies have been published so far for the human Y5 receptor. Merten et al. (2007) exchanged three acidic residues in the extracellular domains of Y5 receptor. While the Asp6.62Ala mutant showed wild-type-like pharmacological properties, Asp6.59Ala and Glu5.27Ala displayed a dramatically reduced affinity for NPY. Additional residues were investigated by Lindner et al. (2008b), resulting in identification of a third acidic residue (Asp2.68) which is important for ligand binding at the Y5 receptor. These Ala-mutants have also been tested with NPY analogs in which theTyr27, Tyr36, and the Arg residues at position 25, 33, and 35 were individually substituted byAla (**Figure 6D**). This approach revealed no further loss of affinity for [Ala33]pNPY on the Asp6.59Ala mutant of the receptor, indicating a direct interaction between Ala<sup>33</sup> of the peptide and Asp6.59 of the receptor. Similarly, Arg25 of the NPY peptide could be identified as the interaction partner for Asp2.68 of the receptor (Lindner et al., 2008b).

#### **CONCLUSION AND PERSPECTIVES**

The NPY system has been extensively characterized in the last years. The modulation of actions mediated by the distinct receptors like, e.g., its involvement in obesity, cancer, and epilepsy are of great importance. Therefore, the development of receptor subtype-selective ligands and structure-activity relationship studies have been a major objective in the past years. Primarily, amino acid scans and truncations have identified the important residues and areas of the ligand with respect to binding at each receptor. The Y receptors have been extensively studied, several important residues have been characterized and some of the binding pockets have been partially characterized. Two subtype-selective interactions have been elucidated so far. A similar binding mode has been identified on NPY receptors, where a common residue Asp6.59 binds to one of the two C-terminally located Arg of the peptide depending on the receptor subtype. Moreover, a second binding interaction has been found on the Y5 receptor where Asp2.68 located at the top of TM2 interacts with Arg<sup>25</sup> of the peptide (Merten et al., 2007; Lindner et al., 2008b). This finding would suggest that probably a second interaction could take place in other receptor subtypes. Nevertheless, further investigations have to be performed. It is likely that more interactions between the receptors and the peptides could occur, therefore structure activity relationship studies are still a focus of interest.

The design of short analogs and antagonists have confirmed these findings, indicating that this is a great tool to modulate and study the receptors. Some promising progresses have been achieved in cancer diagnosis using Y1 receptor selective short ligands. However the development of short analogs for treatment of this pathology still remains challenging. Also in anti-obesity drugs, Y2/Y4receptor selective agonists are in progress and currently in clinical trials of Phase I/II. On the basis of well studied characteristics accounting for receptor subtype selectivity, it is likely that subsequent investigations could be focused on the improvement of pharmacological properties such as stability and half-life. In addition the development of more potent selective ligands might be a focus of interest.

#### **ACKNOWLEDGMENT**

The financial contribution of the German Research Foundation (DFG) for SFB 610/3, project A1 is kindly acknowledged.

#### **REFERENCES**


characterization and appetite suppressive properties of BMS-193885, a novel and selective neuropeptideY(1) receptor antagonist. *Eur. J. Pharmacol.* 590, 224–232.


M., et al. (2003). Characterization of the effects of pancreatic polypeptide in the regulation of energy balance. *Gastroenterology* 124, 1325–1336.


"fendo-04-00005" — 2013/1/31 — 23:14 — page 10 — #10

antagonists based on [Pro(30), Tyr(32), Leu(34)]NPY(28–36)-NH2 (BW1911U90). *J. Med. Chem.* 44, 1479–1482.

Baldock, P. A., Lee, N. J., Driessler, F., Lin, S., Allison, S., Stehrer, B., et al. (2009). Neuropeptide Y knockout mice reveal a central role of NPY in the coordination of bone mass to body weight. *PLoS ONE* 4:e8415. doi: 10.1371/journal.pone.0008415

Baldock, P. A., Sainsbury, A., Couzens, M., Enriquez, R. F., Thomas, G.

P., Gardiner, E. M., et al. (2002). Hypothalamic Y2 receptors regulate bone formation. *J. Clin. Invest.* 109, 915–921.


tract, and the brain. *Endokrynol. Pol.* 61, 194–206.


"fendo-04-00005" — 2013/1/31 — 23:14 — page 11 — #11

Thogersen, H., Melberg, S. G., et al. (1990). [Leu31, Pro34]neuropeptide Y: a specific Y1 receptor agonist. *Proc. Natl. Acad. Sci. U.S.A.* 87, 182–186.


*J. Physiol. Gastrointest. Liver Physiol.* 279, G126-G131.


internalization. *Cell. Signal.* 21, 61–68.


"fendo-04-00005" — 2013/1/31 — 23:14 — page 12 — #12

molecule neuropeptide Y Y2 receptor antagonists. *Bioorg. Med. Chem. Lett.* 22, 3916–3920.


Jodka, C. M., et al. (2004). Effects of PYY[3-36] in rodent models of diabetes and obesity. *Int. J. Obes. Relat. Metab. Disord.* 28, 963–971.


NBK47354/ [updated August 6, 2010].


the hypothalamus: a powerful neurochemical inducer of hyperphagia and obesity. *Peptides* 7, 1189–1192.


"fendo-04-00005" — 2013/1/31 — 23:14 — page 13 — #13

of [Leu31,Pro34]NPY-sensitive, BIBP3226-insensitive [125I]PYY(3– 36) binding sites in rat brain: possible relationship to Y5 NPY receptors. *Brain Res.* 778, 242–250.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 29 August 2012; paper pending published: 02 October 2012; accepted: 09 January 2013; published online: 04 February 2013.*

*Citation: Pedragosa-Badia X, Stichel J and Beck-Sickinger AG (2013) Neuropeptide Y receptors: how to get subtype selectivity. Front. Endocrin. 4:5. doi: 10.3389/ fendo.2013.00005*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Pedragosa-Badia, Stichel and Beck-Sickinger. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# The melanocortin receptors and their accessory proteins

# *Shwetha Ramachandrappa, Rebecca J. Gorrigan, Adrian J. L. Clark and Li F. Chan\**

Centre for Endocrinology, William Harvey Research Institute, Queen Mary University of London, Barts and The London School of Medicine and Dentistry, London, UK

#### *Edited by:*

Jae Young Seong, Korea University, South Korea

#### *Reviewed by:*

Akiyoshi Takahashi, Kitasato University, Japan Robert Dores, University of Minnesota, USA

#### *\*Correspondence:*

Li F. Chan, Centre for Endocrinology, William Harvey Research Institute, Queen Mary University of London, Barts and The London School of Medicine and Dentistry, Charterhouse Square, London EC1M 6BQ, UK. e-mail: l.chan@qmul.ac.uk

# ligands the melanocortin peptides, there is now a growing list of important peptides that can modulate the way these receptors signal, acting as agonists, antagonists, and inverse agonists. The discovery of melanocortin 2 receptor accessory proteins as a novel accessory factor to the MCRs provides further insight into the regulation of these important G protein-coupled receptor. **Keywords: melanocortin receptors, accessory proteins, adrenal cortex, familial glucocorticoid deficiency, obesity**

The five melanocortin receptors (MCRs) named MC1R–MC5R have diverse physiological roles encompassing pigmentation, steroidogenesis, energy homeostasis and feeding behavior as well as exocrine function. Since their identification almost 20 years ago much has been learnt about these receptors. As well as interacting with their endogenous

# **THE MELANOCORTIN SYSTEM**

The melanocortin receptors (MCRs) are a family of five G proteincoupled receptors (GPCRs; MC1R–MC5R) expressed in diverse tissues, which serve discrete physiological functions. Early studies delineated a preliminary model where the receptors were activated by melanocortin peptides, derived from post-translational processing and proteolytic cleavage of the precursor protein proopiomelanocortin (Eipper and Mains, 1980). Receptor activation increased intracellular cyclic adenosine monophosphate (cAMP) levels triggering several downstream signaling pathways. Subsequent research into melanocortin signaling has added several additional tiers of complexity to this basic schema. It has emerged that many of the MCRs bind to endogenous peptides beyond the melanocortin peptides, which can act as agonists, antagonists, partial agonists, and even inverse agonists at these receptors (Cone, 2006). Furthermore, although the cAMP pathway continues to serve as the predominant readout of MCR function in many studies, numerous other intracellular signaling cascades for example the mitogen-activated protein kinase pathway have also been implicated in melanocortin signaling (Vongs et al., 2004; Sutton et al., 2005; Patten et al., 2007; Rodrigues et al., 2009). An additional feature of MCR signaling which has come to light over recent years is the presence of accessory proteins which regulate receptor function. These observations have broadened our understanding of MCR signaling, unveiling a highly specialized system, which on a cellular level enables individual cells to generate a complex co-ordinated response to the unique complement of ligands in their microenvironment (**Table 1**).

#### **MC1R**

MC1R is expressed in the melanocytes of the skin and hair follicles. Activation of MC1R results in a switch from red/yellow pheomelanin pigment production to the production of brown/black eumelanin pigment; it also promotes cell proliferation, DNA repair, and cell survival. Variants in MC1R are associated with red hair color, fair skin, and increased skin cancer risk (Valverde et al., 1995). Many of these variant receptors have been studied *in vitro* and their function, as measured by cAMP production in response to [Nle4,D-Phe7]-α-MSH (NDP-MSH), has been shown to be impaired compared to wild-type. Until recently MC1R was thought to be exclusively activated by alpha-MSH endogenously produced by keratinocytes in response to ultraviolet radiation. Human β-defensin 3 has since emerged as a novel endogenous MC1R ligand whose binding appears to initiate a discrete complement of intracellular signaling pathways *in vitro* (Beaumont et al., 2012; Swope et al., 2012). The physiological importance of this interaction remains to be seen. MC1R is also expressed in macrophages and in adipocytes where its role is less clearly defined. At extremely high plasma adrenocorticotropin hormone (ACTH) concentrations, ACTH activation of MC1R leads to hyperpigmentation observed in patients with familial glucocorticoid deficiency (FGD) (Turan et al., 2012). The study of the genetic determinants of coat color in animal models led to the discovery of agouti (also known as agouti signaling peptide), a high-affinity antagonist of the MC1R (Lu et al., 1994). The identification of an endogenous, physiologically relevant GPCR antagonist marked the beginning of a paradigm shift in our understanding of GPCR signaling. Later work demonstrated the complexity of this molecule by exploring its ability to act as an inverse agonist (Vage et al., 1997), inhibiting constitutively active MC1R.

#### **MC2R**

"fendo-04-00009" — 2013/2/7 — 11:57 — page 1 — #1

MC2R is predominantly expressed in the adrenal gland where it promotes the expression of steroidogenic enzymes in response to binding plasma ACTH. Mutations in MC2R result in FGD (Clark et al., 1993; Tsigos et al., 1993). This is a rare, lifethreatening autosomal recessive disorder of adrenal resistance to ACTH wherein affected individuals have low serum levels of cortisol despite the presence of extremely high circulating levels of plasma ACTH. Patients typically present within a few months


**Table 1 |The members of the melanocortin receptor family, expression, action, and phenotype of individuals with mutations.**

of life with symptoms of cortisol deficiency including recurrent infections, hypoglycemia, convulsions, failure to thrive, and shock. Milder forms of the disease have been observed to present later in life (Metherell et al., 2009; Hughes et al., 2010; Meimaridou et al., 2012). The classical presentation of FGD comprises isolated perturbation of the glucocorticoid axis; however, a small number of cases have been described where this is accompanied by a disorder of mineralocorticoid secretion (Lin et al., 2007; Chan et al., 2009a).

Mutations in MC2R account for 25% of cases of FGD. The remaining 75% of patients with FGD in whom mutations in MC2R have been excluded have formed the subject of genetic approaches to elucidate other causative loci. To date, mutations in three further genes have been associated with FGD. Mutations in the melanocortin 2 receptor accessory protein (MRAP), a single pass transmembrane protein implicated in MC2R function account for approximately 15–20% of cases of FGD (Metherell et al., 2005). Mutations in steroidogenic acute regulatory (STAR) protein are known to give rise to lipoid congenital adrenal hyperplasia, a severe form of adrenal insufficiency characterized by both glucocorticoid and mineralocorticoid deficiency together with gonadal deficiency; however, STAR mutations have also been identified in a number of patients with FGD suggesting that mutations in this gene can give rise to a spectrum of clinical phenotypes encompassing FGD (Metherell et al., 2009). Recently, mutations in nicotinamide nucleotide transhydrogenase (NNT) a mitochondrial membrane constituent which is involved in detoxification of reactive oxygen species were also associated with FGD (Meimaridou et al., 2012). Notably, mutations in minichromosome maintenance-deficient 4 (MCM4) which forms part of a protein complex which is essential for DNA replication and genome stability are associated with a variant of FGD found in the Irish Traveller population where adrenal failure is accompanied by short stature, chromosome instability, and natural killer cell dysfunction (O'Riordan et al., 2008; Gineau et al., 2012; Hughes et al., 2012).

#### **MC3R**

MC3R is primarily expressed in the central nervous system where it is found in the hypothalamus and the limbic regions (Roselli-Rehfuss et al., 1993). Targeted deletion of MC3R in murine models results in animals with increased fat mass, reduced lean mass, and reduced physical activity in the absence of hyperphagia (Butler et al., 2000; Chen et al., 2000). Additionally, these animals exhibit accelerated weight gain when placed on a high fat diet which is independent of hyperphagia implying that MC3R may be involved in nutrient partitioning. An emerging aspect of the MC3R deficient phenotype is that when these animals are subjected to food restriction regimes they exhibit impaired synchronized oscillation of the transcription factors which regulate liver clock activity and metabolism, utilize carbohydrates inappropriately, and fail to demonstrate the food anticipatory behaviors which are seen in wild-type mice (Sutton et al., 2008, 2010). These observations have led to the suggestion that MC3R is involved in coordinating appropriate homeostatic metabolic and behavioral responses to nutrient cues. Interestingly, despite focused investigation, rare variants at the MC3R locus have not been definitively associated with monogenic obesity in humans (Seng Lee et al., 2007; Mencarelli et al., 2011).

#### **MC4R**

"fendo-04-00009" — 2013/2/7 — 11:57 — page 2 — #2

MC4R is widely expressed in the central nervous system where it is abundant in several regions including the paraventricular nucleus (PVN) of the hypothalamus (Mountjoy et al.,1994). Targeted deletion of MC4R in a mouse model results in animals which are severely obese, with hyperphagia and reduced energy expenditure (Huszar et al., 1997). In an informative study, Cre-lox technology was used to selectively re-express MC4R at endogenous levels in the PVN and the amygdale of MC4R knockout animals (Balthasar et al., 2005). This was able to partially rescue their obese phenotype by normalizing their food intake suggesting that MC4Rs in these regions are responsible for regulating eating behavior while MC4Rs in anatomically distinct neuronal populations are relevant to its control of energy expenditure.

The PVN is a key integration center for diverse signals which impact upon energy balance and is densely innervated by the arcuate nucleus of the hypothalamus. The release of proopiomelanocortin (POMC)-derived peptides from neurons in the arcuate nucleus, which are stimulated by leptin results in the activation of MC4R signaling in the PVN (Cheung et al., 1997; Schwartz et al., 1997). Uniquely for a biological system, a physiologically relevant endogenous antagonist and inverse agonist of MC4R (as well as MC3R) signaling is also produced in the hypothalamus, agouti-related peptide (AGRP; Nijenhuis et al., 2001; Breit et al., 2006). ARGP is produced by leptin responsive neurons in the arcuate nucleus that are inhibited by leptin. The presence of endogenous agonists and antagonists for this receptor enables MC4R signaling in the PVN to be delicately balanced.

Mutations in the MC4R are the most prevalent form of monogenic obesity identified to date accounting for up to 6% of patients with severe obesity (Vaisse et al., 1998; Yeo et al., 1998). The signaling capacities of specific mutant receptors studied *in vitro* correlates with the severity of the phenotype of corresponding MC4R deficient individuals (Farooqi et al., 2003). This demonstrates that the system is sensitive to degrees of loss in function at the level of the receptor. In addition to rare variants which cause highly penetrantforms of monogenic obesity, common variants inMC4R have been associated with body mass index (BMI) in genome wide association scans suggesting that variation at this locus also contributes to obesity in the general population (Willer et al., 2009).

#### **MC5R**

The MC5R is widely expressed in peripheral tissues. Mice lacking MC5R are unable to produce the full complement of sebaceous lipids which constitute the water repellent component of their coats (Chen et al., 1997). As a result they demonstrate impaired water repulsion and thermoregulation. Studies in these mice have also suggested that MC5Rs play a role in the generation of pheromones which in turn influences aspects of behavior (Morgan et al., 2004; Morgan and Cone, 2006). High levels of MC5R are found in multiple exocrine tissues where they are thought to regulate the synthesis and secretion of a diverse range of exocrine products.

#### **THE MELANOCORTIN RECEPTOR ACCESSORY PROTEINS THE DISCOVERY OF MRAP**

The functional properties of the MCRs have been extensively examined in *in vitro* expression systems. Comparable studies of MC2R function were initially impeded by difficulties in expressing functional receptor in conventionally used cell lines. It became apparent that unlike other MCRs functional MC2R could only be expressed in cell lines of adrenal origin, suggesting that its expression was contingent upon the presence of a tissue-specific protein. This factor was identified in a study of genetic loci associated with FGD. Mutations in MC2R were known to lead to this inherited syndrome of ACTH resistance, but were only able to account for approximately 25% of cases. Whole genome single nucleotide polymorphism (SNP) analysis in two affected consanguineous families in whom mutations in MC2R had been excluded identified a region of interest encompassing a novel protein, which was highly expressed in the adrenal gland. *In vitro* studies confirmed that expression of this protein alongside MC2R enabled functional MC2R receptor expression in heterologous cell types and it was subsequently named MRAP (Metherell et al., 2005). Fifteen to twenty percent of cases of FGD are now known to be caused by mutations in MRAP.

#### **MRAP STRUCTURE**

HumanMRAP is a single pass transmembrane protein which exists as two isoforms which differ in their C-termini. Both isoforms have been shown by reverse transcription polymerase chain reaction (RT-PCR) in a cDNA panel derived from human tissues to be expressed at high levels in the adrenal gland (Metherell et al.,2005). MRAP expression has also been demonstrated in a variety of other human tissues including testis, breast, thyroid, lymph nodes, ovary, and skin (Metherell et al., 2005). Immunofluorescence microscopy staining of cultured cells which endogenously expressMRAP, using antibodies raised against N-terminal and C-terminal MRAP peptides, have shown that both regions of MRAP are present at the cell surface (Sebag and Hinkle, 2007). *In vitro* studies where differentially epitope-tagged forms of the protein were co-expressed in cells and their interaction and orientation were studied show that MRAP monomers form unique anti-parallel homodimeric structures (Sebag and Hinkle, 2007; Cooray et al., 2008). These dimers can be visualized in the endoplasmic reticulum and at the plasma membrane in live cells using biomolecular fluorescence complementation techniques (Sebag and Hinkle, 2009b, 2010). The deletion of amino acids 31–37 in mouse MRAP which directly precede the transmembrane portion of the protein abolishes its ability to homodimerize suggesting that this region of the protein constitutes a critical component of the interface between the two monomers (Sebag and Hinkle, 2009b). This mutant is unable to assist trafficking of the MC2R to the cell surface, suggesting that this unique conformation is essential to certain aspects of MRAP function (**Figure 1**).

# **MRAP AND MC2R**

Cell lines derived from adrenocortical tissue express both MC2R and MRAP and generate cAMP in response to treatment with ACTH. This response is abrogated by MRAP knockdown in these cells (Cooray et al., 2008). MRAP expression is necessary and sufficient for the functional expression of MC2R constructs in heterologous cell lines such as chinese hamster ovary (CHO) cells and SK-N-SH cells which are derived from ovarian and neuroblastoma tissue, respectively and do not endogenously express measurable levels of MRAP (Metherell et al., 2005; Sebag and Hinkle, 2007).

#### **MRAP FACILITATES TRAFFICKING OF THE MC2R TO THE CELL SURFACE**

When epitope-tagged MC2R is expressed in CHO cells it is visualized in the endoplasmic reticulum, however, when it is expressed alongside MRAP it is predominantly localized to the plasma membrane (Metherell et al., 2005; Sebag and Hinkle, 2007; Webb et al., 2009). Consistent with this finding, in cells of adrenal origin with endogenously expression of MRAP, MC2R can be demonstrated at the cell surface. Western blotting of cell lysates yields a different complement of MC2R species in the presence and absence of MRAP, suggesting that MRAP may enhance receptor trafficking to the cell surface by promoting post-translational modification (Sebag and Hinkle, 2007). Deletion mutants of the N-terminal region of human MRAP which lack amino acids 1–24 or 1–35 are unable to promote MC2R trafficking to the cell surface suggesting that the N-terminal region plays a prominent role in receptor trafficking (Webb et al., 2009).

#### **MRAP FACILITATES MC2R SIGNALING**

"fendo-04-00009" — 2013/2/7 — 11:57 — page 3 — #3

*In vitro* experiments examining MC2R signaling in the presence of various MRAP mutants suggest that the ability of MRAP to

"fendo-04-00009" — 2013/2/7 — 11:57 — page 4 — #4

promote MC2R trafficking to the cell surface does not wholly account for its role in facilitating MC2R signaling. Amino acids within the N-terminal region of MRAP appear to influence MC2R signaling independently of enhancing receptor trafficking. Specifically, MC2R receptor is able to traffic to the cell surface but is unable to bind to ACTH in the presence of a deletion mutant of mouse MRAP in which amino acids 18–21 have been mutated to alanine residues (Sebag and Hinkle, 2009b).

#### **MRAP AND MC2R COMPLEXES**

Protein complexes containing both MRAP and MC2R can be reciprocally isolated using co-immunoprecipitation techniques in transiently transfected cells (Metherell et al., 2005; Webb et al., 2009). Analogous experiments using MRAP deletion mutations indicate that the transmembrane region of MRAP interacts with MC2R (Webb et al., 2009). A model has emerged wherein MRAP– MC2R complexes interact with one another to form higher order complexes which facilitate MC2R function. BRET techniques have been used to explore some of the interactions within these complexes in more detail in live cultured cells with specific reference to the effect of the MC2R agonist ACTH. These experiments have shown that ACTH enhances the interaction between MC2R homodimers and MRAP–MC2R heterodimers (Cooray et al., 2011). Schematic representation of the role of MRAP in MC2R signaling is shown in **Figure 2**.

# **MRAP INTERACTIONS WITH OTHER MEMBERS OF THE MELANOCORTIN RECEPTOR FAMILY**

MRAP can be demonstrated to form protein complexes with all five members of the MCR family using co-immunoprecipitation techniques in cultured cells (Chan et al., 2009b). Aside from MC2R, the remaining members of the MCR family are capable of efficiently trafficking to the cell surface when expressed in CHO cells. In a series of experiments using human constructs, co-expression of MRAP alongside either MC1R or MC3R in CHO cells was not found to alter cell surface expression of the receptors or influence their ability to generate cAMP in response to agonist treatment. In contrast, the cell surface expression and signaling capacities of both MC4R and MC5R were found to be significantly reduced in the presence of MRAP in these cells (**Table 2**). Signaling in these experiments was assessed by measuring cAMP generation using a competitive protein binding assay in response to treatment with a single concentration of agonist (Chan et al., 2009b). When the signaling properties of MC4R in the presence of MRAP were studied by measuring cAMP generation with a luciferase assay across a range of agonist concentrations using mouse constructs, co-expression of MRAP was not found to influence MC4R signaling (Sebag and Hinkle, 2010). Further examination of this interaction is needed to definitively understand whether MRAP interacts with MC4R *in vivo* and to determine the impact of this interaction on MC4R signaling. The *in vitro* interaction of

MRAP and MC5R has been explored more fully using bimolecular fluorescence complementation techniques (Sebag and Hinkle, 2009a). When MC5R constructs are expressed in CHO cells MC5R monomers interact to from homodimers or oligomers. When MRAP is expressed alongside MC5R fewer multimeric complexes are formed and fewer receptors are expressed at the cell surface. This data suggests that MRAP is capable of interacting with members of the MCR family beyond MC2R and may differentially regulate aspects of their function. However, patients with MRAP mutations present with isolated glucocorticoid deficiency apparently without any other additional symptoms (Chung et al., 2010). One patient with FGD caused by a mutation in MRAP was described as being concomitantly obese, although the significance of this is unclear as the rest of the family had similarly high BMIs (Rumie et al., 2007).

#### **MRAP IN ADIPOCYTES**

Prior to its advent as an accessory factor for MC2R in the adrenal gland, MRAP was identified as a novel protein expressed in mouse adipocytes (Xu et al., 2002). Both isoforms of MRAP have subsequently been shown by RT-PCR to be expressed in human adipocyte tissue (Metherell et al., 2005). The function of MRAP in these cells is not known. Given its established role in facilitating MC2R function in the adrenal gland it is likely that its function is related to the melanocortin system. The distribution of epitopetagged MRAP within cultured adipocytes is responsive to insulin treatment suggesting that its role is contingent upon metabolic cues. Notably, gross metabolic dysfunctions have not been reported in patients with mutations in MRAP; however, this may reflect functional redundancy with the related protein MRAP2 in this tissue whose expression in adipocytes has not yet been investigated. The components of the melanocortin system which are expressed in adipocyte tissue show species-specific differences. 3T3L1 cells which are cells of murine origin that are commonly used in the study of adipocyte function express both MC2R and MC5R. Human adipose tissue samples have been shown to express MC1R and low levels of MC4R and MC5R (Hoch et al., 2007). Treatment of 3T3L1 cells with melanocortin ligands promotes lipolysis and cytokine production *in vitro*, however, this observation has not been replicated in human mesenchymal cell derived adipocytes (Hoch et al., 2008; Jun et al., 2010; Moller et al., 2011).

#### **IDENTIFICATION OF MRAP2**

Following the discovery of MRAP a related gene of unknown function with a comparable structure and a high degree of N-terminal and C-terminal sequence similarity was identified on

**Table 2 |** *In vitro* **effects of MRAP on melanocortin receptors, based on data derived from Metherell et al. (2005) Sebag and Hinkle (2007, 2009a, 2010), Chan et al. (2009b) and Cooray et al. (2011).**


"fendo-04-00009" — 2013/2/7 — 11:57 — page 5 — #5

chromosome 6 and was named MRAP2 (Metherell et al., 2005; Chan et al., 2009b). MRAP2 is a single pass transmembrane protein, in humans it has been shown by RT-PCR to be expressed in the brain and the adrenal gland (Chan et al., 2009b). Western blotting of mouse tissues resolves a single band of MRAP2 reactive species whose molecular weight corresponds to a dimer suggesting that endogenous MRAP2 exists in dimeric complexes (Chan et al., 2009b). MRAP2 undergoes N-linked glycosylation at its N-terminus and is thought to form anti-parallel homodimeric complexes analogous to those formed by MRAP (Chan et al., 2009b; Sebag and Hinkle, 2009a, 2010). MRAP2 constructs can be visualized in the endoplasmic reticulum and the plasma membrane of transiently transfected CHO cells (Chan et al., 2009b).

# **MRAP2 AND MC2R SIGNALING**

The physiological role of MRAP2 is not yet known. In view of the structural similarities between MRAP and MRAP2 and its concurrent expression in the adrenal gland, the interaction between MRAP2 and MC2R has been explored in detail. As discussed previously, when MC2R constructs are expressed in CHO cells in isolation the receptors are unable to reach the cell surface. Concurrent expression of MRAP facilitates MC2R trafficking to the cell surface. At the cell surface MC2Rs are activated by ACTH binding and generate a cAMP response, the magnitude of which is dependent on the ambient ligand concentration. Expression of MRAP2 alongside MC2R promotes MC2R trafficking to the cell surface to a similar extent as MRAP; however, the resultant complex has an extremely low affinity for ACTH binding. As a result the response curve of cAMP generated plotted against ligand concentration is significantly shifted to the right when MC2R is co-expressed with MRAP2 compared to when MC2R is expressed with MRAP (Sebag and Hinkle, 2009a, 2010; Gorrigan et al., 2011). It was suggested that this difference was due to lack of the leucine, aspartic acid, tyrosine, isoleucine (LDYI) motif in MRAP2, which was present in MRAP, and insertion of the LDYI residues enabled the MRAP2/MC2R complex to respond to lower concentrations of ACTH (Sebag and Hinkle, 2009b). MRAP and MRAP2 are able to form heterodimeric complexes *in vitro*, however, whether such a complex is physiologically relevant is not known (Chan et al., 2009b; Sebag and Hinkle, 2010). In one study MC2R was expressed alongside both MRAP and MRAP2, the resulting dose response curve appeared sensitive to the ratio of MRAP to MRAP2, with increasing proportions of MRAP2 shifting the curve further to the right (Sebag and Hinkle, 2010). Importantly, this effect has not been observed in other studies (Chan et al., 2009b; Agulleiro et al., 2010). Interestingly, Y1 cells which are of murine adrenocortical origin and endogenously expresses MRAP, MRAP2 and MC2R generate cAMP in response to ACTH treatment and this effect can be abrogated by transient expression of MRAP2 within these cells (Sebag and Hinkle, 2010). Taken together, further work is required to determine if the expression levels of MRAP and MRAP2 in the adrenal gland could act as a physiological mechanism dictating ACTH responsiveness.

Mutations in MC2R and MRAP lead to the inherited condition FGD. In contrast, mutations in the MRAP2 gene have not been linked with this condition suggesting that it may not be involved in MC2R signaling *in vivo* and that any degree of functional redundancy between MRAP and MRAP2 is unable to completely compensate for MRAP dysfunction. In-keeping with this observation *in situ* hybridization studies and quantitative RT-PCR using rat tissue have demonstrated that MRAP is expressed at much higher levels than MRAP2 in adult adrenal glands (Gorrigan et al., 2011).

# **MRAP2 BEYOND MC2R SIGNALING**

Consistent with the findings discussed above, exploring the interaction of MRAP with other members of the MCR family, MRAP2 has also been shown to interact with all five members of the MCR family by co-immunoprecipitation assays in transiently transfected cultured cells (Chan et al., 2009b). In one series of experiments when cell surface expression of individual MCRs was quantified in the presence or absence of MRAP2, expression levels of MC1R and MC3R were not affected by MRAP2 expression, however, expression levels of MC4R and MC5R were significantly reduced when they were co-expressed with MRAP2. Furthermore, when MCR signaling was examined in the presence of MRAP2, its expression was detrimental to MC3R,MC4R, and MC5R signaling but had no effect on MC1R signaling. In these experiments signaling was studied by quantifying cAMP generation in response to treatment with a single concentration of ligand. The effect of MRAP2 on MC4R signaling has been explored using a luciferase reporter assay to measure cAMP production across a range of ligand concentrations. In this study the authors concluded that although MC4R surface expression was reduced in the presence of MRAP2 that MC4R signaling did not appear to be affected (Sebag and Hinkle, 2010). The experiments discussed here have been conducted in CHO cells which do not endogenously express any of the components of the melanocortin system, further research will address whether MRAP2 interacts with MCRs beyond this experimental system and will further delineate the nature of these interactions. As the main site of MRAP2 expression is in brain and specifically in the hypothalamus it is likely that the main physiological function of MRAP2 is in the central nervous system, perhaps involving MC4R and/or MC3R.

# **THE FUTURE OF MRAPs**

Numerous lines of genetic evidence suggest that subtle perturbations of signaling capacity at the level of MCRs can have pathological consequences. As our appreciation of the complex nature of the melanocortin system grows, a large family of molecules is emerging which impact upon signaling and confer the ability to integrate extracellular and intracellular cues into a co-ordinated biological signal. The discovery of two proteins which may interact with multiple members of the MCR family in distinct ways represents a significant advance in our understanding of melanocortin signaling and paves the way for further lines of research to explore the physiological roles of these proteins.

#### **ACKNOWLEDGMENTS**

"fendo-04-00009" — 2013/2/7 — 11:57 — page 6 — #6

Li F. Chan holds an MRC/Academy of Medical Sciences Clinician Scientist Fellowship [G0802796] and Rebecca J. Gorrigan is supported by a WellcomeTrust Research Training Fellowship [092024/Z/10/Z].

# **REFERENCES**


"fendo-04-00009" — 2013/2/7 — 11:57 — page 7 — #7

glucocorticoid deficiency type 2. *Nat. Genet.* 37, 166–170.


Clinical and biological phenotype of a patient with familial glucocorticoid deficiency type 2 caused by a mutation of melanocortin 2 receptor accessory protein. *Eur. J. Endocrinol.* 157, 539–542.


homozygous mutations in MC2R (T152K) and MC1R (R160W). *J. Clin. Endocrinol. Metab.* 97, E771– E774.


"fendo-04-00009" — 2013/2/7 — 11:57 — page 8 — #8

C., et al. (2002). Identification of novel putative membrane proteins selectively expressed during adipose conversion of 3T3-L1 cells. *Biochem. Biophys. Res. Commun.* 293, 1161– 1167.

Yeo, G. S., Farooqi, I. S., Aminian, S., Halsall, D. J., Stanhope, R. G., and O'Rahilly, S. (1998). A frameshift mutation in MC4R associated with dominantly inherited human obesity. *Nat. Genet.* 20, 111–112.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 02 November 2012; accepted: 25 January 2013; published online: 08 February 2013.*

*Citation: Ramachandrappa S, Gorrigan RJ, Clark AJL and Chan LF (2013) The melanocortin receptors and their accessory proteins. Front. Endocrin. 4:9. doi: 10.3389/fendo.2013.00009*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Ramachandrappa, Gorrigan, Clark and Chan. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

**REVIEW ARTICLE** published: 17 January 2013 doi: 10.3389/fendo.2012.00184

# Neurotensin and its high affinity receptor 1 as a potential pharmacological target in cancer therapy

# *ZheruiWu1, Daniel Martinez-Fong2, Jean Trédaniel 1,3 and Patricia Forgez1\**

<sup>1</sup> INSERM-UPMC UMR\_S938, Hôpital Saint-Antoine, Paris, France

<sup>2</sup> Departamento de Fisiología, Biofísica y Neurociencias, Centro de Investigación y de Estudios Avanzados del Instituto Politécnico Nacional,

Mexico City, Mexico

<sup>3</sup> Unité de Cancérologie Thoracique, Groupe Hospitalier Paris Saint-Joseph/Université Paris Descartes, Paris, France

#### *Edited by:*

Jae Young Seong, Korea University, South Korea

#### *Reviewed by:*

Jean Mazella, Centre National de la Recherche Scientifique, France Aixa R. Bello, University of La Laguna, Spain

#### *\*Correspondence:*

Patricia Forgez, INSERM-UPMC UMR\_S938, Hôpital Saint-Antoine, Bâtiment Raoul Kourilsky, 184 rue du Faubourg St-Antoine, 75571 Paris Cedex 12, France. e-mail: patricia.forgez@inserm.fr

Cancer is a worldwide health problem. Personalized treatment represents a future advancement for cancer treatment, in part due to the development of targeted therapeutic drugs. These molecules are expected to be more effective than current treatments and less harmful to normal cells. The discovery and validation of new targets are the foundation and the source of these new therapies. The neurotensinergic system has been shown to enhance cancer progression in various cancers such as pancreatic, prostate, lung, breast, and colon cancer. It also triggers multiple oncogenic signaling pathways, such as the PKC/ERK and AKT pathways. In this review, we discuss the contribution of the neurotensinergic system to cancer progression, as well as the regulation and mechanisms of the system in order to highlight its potential as a therapeutic target, and its prospect for its use as a treatment in certain cancers.

**Keywords: cancer therapy, cancer progression, carcinogenesis, neurotensin, neurotensin receptor**

#### **INTRODUCTION**

Cancer is one of the first, if not the first cause, of death in the world. Conventional treatment methods include surgery, radiotherapy, and chemotherapy. In many cases, only a supportive treatment can be offered to the patient. The last few decades have been marked by the accumulation of knowledge about the inner workings of the normal and cancer cell. Thus arose the therapeutic arsenal against cancer, and the so-called targeted biological therapies. Due to these new drugs, great progress has recently been achieved in the treatment of cancers considered refractory to previous therapies, including cancers of the liver, kidney, and melanoma skin. But other tumors, such as breast and lung cancer, have also greatly benefited from these advances. However, remissions are often transient and do not yet provide definitive cures of the patient. In this context the neurotensin system is proposed to be a candidate for therapeutical development.

Neurotensin (NTS) is a tridecapeptide originally isolated from calf hypothalamus (Carraway and Leeman, 1973). It acts as neurotransmitter or neuromodulator in the central nervous system and as a local hormone in the periphery, mainly the gastrointestinal tract (Vincent et al., 1999). In the brain, NTS modulates dopaminergic transmission in the nigrostriatal and mesocorticolimbic pathways as well as hormone secretion from the anterior pituitary (Vincent et al., 1999). It also exerts potent hypothermic and analgesic effects when injected into the central nervous system (Popp et al., 2007). NTS has also been related to central nervous system pathology such as Parkinson disease and schizophrenia (St-Gelais et al., 2006; Mustain et al., 2011). In the periphery, NTS is released by endocrine-like N cells predominantly in the small intestine (Reinecke, 1985). NTS has a dual function, as a paracrine and endocrine modulator of the digestive tract and as a growth factor on a variety of normal or cancer cells (Vincent et al., 1999; Carraway and Plona, 2006). The NTS/NTSR1 complex has been proposed to contribute to cancer progression because of the various oncogenic effects induced by NTS in tumors and in cancer cells from diverse origins (Thomas et al., 2003; Dupouy et al., 2011). In this review, we elaborate on how the NTS/NTSR1 complex could be developed as a possible target for cancer therapy.

#### **NEUROTENSIN/NEUROTENSIN RECEPTOR COMPLEX NEUROTENSINERGIC SYSTEM**

The biological effects of NTS are known to be mediated through three receptors, two G protein-coupled receptors, NTSR1 and NTSR2, and a single transmembrane domain sorting receptor, the NTSR3 (Vincent et al., 1999; Mazella and Vincent, 2006). NTSR3 is a member of the receptor family related to the yeast sorting receptor Vps10p (Petersen et al., 1997).

#### *Neurotensin receptor 1, NTSR1*

"fendo-03-00184" — 2013/1/12 — 17:45 — page 1 — #1

The effects of NTS are primarily transmitted through its high affinity receptor, the NTSR1 which has a sub-nanomolar affinity for NTS. This 424 amino acid receptor has been identified in the brain and in various cancer cells (Tanaka et al., 1990). The signaling pathways induced by the NTS/NTSR1 complex have been studied in different cellular types, such as N1E-115, HT-29, and NTSR1-transfected CHO overexpressing NTSR1. The stimulation of NTSR1 by its ligand (NTS) leads to activation of phospholipase C (PLC) via its coupling to the prime Gαq/11 subunit (Wang and Wu, 1996; Najimi et al., 2002). The activation of PLC leads to the production of inositol triphosphate (IP3) and diacylglycerol from membrane phospholipids (PIP2). These two second messengers induce the activation of PKC and the mobilization of intracellular calcium which are key oncogenic effectors (Bozou et al., 1986; Snider et al., 1986; Turner et al., 1990).

Several signaling pathways potentially involved in cell proliferation, survival, migration, and invasion are described after NTSR1 stimulation. The signaling mechanisms mediating the effects of neurotensinergic system involve multiple pathways and are cell-dependent. The NTS-induced PKC/ERK signaling pathway is the most well studied. The PKC activation stimulated by NTS/NTSR1 was demonstrated by using broad isotype inhibitors, such as Gö6976 which specifically inhibit the conventional PKCs α and β1 in CHO-NTSR1-transfected cells and in endogenously NTSR1-expressing cells from colon, lung, and pancreatic cells (Poinot-Chazel et al., 1996; Seufferlein and Rozengurt, 1996; Ehlers et al., 2000; Muller et al., 2011). The NTSR1 contribution in this activation was confirmed by its inhibition by a specific NTSR1 antagonist, the SR 489692 (Gully et al., 1993). PKC activation induce mitogen-activated protein kinase (MAPK) via direct stimulation of Raf-1, independently of Ras, or by transactivation of the epidermal growth factor receptor (EGFR) (Guha et al., 2003). EGFR transactivation by NTS/NTSR1 complex has been observed in several cell lines. In prostatic cancer PC3 cell line, NTS activates mitogenesis through EGFR transactivation in a PKC-dependent pathway, and the stimulation of the Raf–MEK– ERK. This effect was also PI3 kinase (PI3K)-dependent (Hassan et al., 2004). NTS also induces a time-dependent increase in Tyr845 EGFR phosphorylation, c-Src phosphorylation and signal transducer. NTS is also an activator of transcription 5b (Stat5b), a downstream effector of Tyr845 EGFR phosphorylated (Amorino et al., 2007). In colonic HT-29 cells, the EGFR tyrosine kinase inhibitor, gefitinib, blocks NTS-stimulated phosphorylation of both MAPK and Akt, indicating the transactivation of EGFR independently of PKC activation. However, in the colonic HCT116 cells, NTS/NTSR1 induces a PKC-dependent MAPK phosphorylation and an EGFR metalloproteinase-mediated transactivation, that is associated with a gefitinib-sensitive phosphorylation of the downstream adaptor protein Shc. The activation of Akt is only partly inhibited by gefitinib, suggesting an additional mechanism to EGFR transactivation (Muller et al., 2011). The mechanism of NTS-induced EGFR transactivation is still not clearly elucidated. The release of EGFR ligands-like (TGF-α, Hb-EGF, or amphiregulin), pre-existing at the plasma membrane, as proligand, by NTS has been proposed. These ligands are released by proteolytic cleavage involving enzymes of the metalloproteinase family, including ADAMS (disintegrin and metalloprotease) and MMP (matrix metalloproteinase; Sanderson et al., 2006; Zhao et al., 2007; Kataoka, 2009). Once released, these ligands bind to EGFR and activate the downstream signaling cascades of EGFR activation (Hassan et al., 2004). These results suggest a cooperative relationship between the neurotensinergic system and EGFR pathway.

Activation of MAPK pathway by NTS results in gene transcription stimulation, due to transcription factor activation, such as the induction of the early growth response gene-1 (Egr-1), the Ets family factors ELK1, and the AP-1 transcription factor family (Poinot-Chazel et al., 1996; Portier et al., 1998; Ehlers et al., 2000; Zhao et al., 2007). In colonic HCT166 cells, inhibition of PKC was shown to block NTS-induced DNA synthesis (Muller et al., 2011). In a tumor-initiating cell line derived from hepatocellular carcinoma (HCC) which is characterized by membrane expression of

CD133, addition of exogenous NTS resulted in concomitant upregulation of IL-8 and CXCL1 with simultaneous activation of MAPK and Raf-1, and promotion of angiogenesis, tumorigenesis, and self-renewal (Tang et al., 2012). The activation of MAPK via NTSR1 is mainly associated with uncontrolled cell growth, which can aggravate the growth of tumors (Harikumar et al., 2010; Kisfalvi et al., 2010).

Neurotensin also induces RhoGTPases and the non-receptor kinases focal adhesion kinase (FAK) and Src. Neurotensinergic system stimulation can modulate the activity of small RhoGT-Pases Rac1, Cdc42, and RhoA, which are partly responsible for the dynamics of the cytoskeleton. This modulation has an effect on cell migration. In the cell line U373 glioblastoma, the NTS has been associated with the stimulation of the activity of protein Rac1, RhoA, and Cdc42 (Zhao et al., 2003; Servotte et al., 2006). In addition, in small cell lung and prostate cancer cell lines, it has been shown that NTS can enhance the activity of focal adhesion kinase (FAK) (Tallett et al., 1996; Lee et al., 2001).

# *Neurotensin receptor 2, NTSR2*

The low affinity receptor of NTS, NTSR2, is a protein of 410 amino acids, with a high homology to NTSR1 (64%) (Chalon et al., 1996). NTSR2 exhibits a low binding property for NTS and this binding can be inhibited by levocabastine, a non-peptide histamine H1 receptor antagonist (Schotte et al., 1986). SR48692, which has a lower affinity for NTSR2 than for NTSR1, can stimulate the activity of this receptor (Yamada et al., 1998). When the cloned mouse NTSR2 coding sequence is expressed in *Xenopus laevis* ovocytes, NTS, neuromedin N, levocabastine, and SR48692, are capable of triggering an inward current which is calcium-dependent (Mazella et al., 1996). Using CHO cells transfected with the cloned rat or human NTSR2 cDNA, levocabastine and SR 48692 can mobilize intracellular Ca2<sup>+</sup> more intensively than NTS agonists and phosphorylate Erk1/2, suggesting that NTSR1 and NTSR2 receptors present distinct functional characteristics (Botto et al., 1997; Yamada et al., 1998; Gendron et al., 2004). In CHO cells transfected with human NTSR2 cDNA, both NTSR1 antagonists, SR48692 and SR142948A, enhance inositol phosphate (IP) formation with subsequent [Ca2+] immobilization, induce arachidonic acid release, and stimulate MAPK activity. Interestingly, these activities were inhibited by NTS and levocabastine in a dose-dependent manner. In summary, the signaling pathway triggered by NTSR2 is cell-dependent, and mainly based on its overexpression. This response is far different from that of the physiological endogenous expression.

# *Neurotensin receptor 3, gp85/sortilin, NTSR3*

"fendo-03-00184" — 2013/1/12 — 17:45 — page 2 — #2

NTSR3 functions as a modulator of neurotensinergic signaling when it is co-expressed with another receptor of NTS, and as a functional receptor involved in the migration when expressed alone. This receptor is not NTS-specific. It can bind other ligands such as lipoprotein lipase, proneurotrophins, protein RAP (receptor-associated protein), or protein SAP (sphingolipid activator protein) (Nielsen et al., 1999; Lefrancois et al., 2003). NTSR3 may act as a co-receptor to participate in true NTS/NTSR1 signaling. The study by immunoprecipitation using the adenocarcinoma cell line HT29, demonstrated that the NTSR3 forms heterodimers with the NTSR1. Additionally, upon NTS stimulation, the NTSR1/NTSR3 complex is internalized and the interaction between the two receptors modulates both the NTSinduced phosphorylation of MAPK and the phosphoinositide (PI) turnover mediated by NTSR1 (Martin et al., 2002).

In the human microglial cell line C13NJ, NTSR3 is the only known endogenous NTS receptor. In these cells, NTS elicited cell migration by a mechanism dependent on both PI3K and MAPK pathways (Martin et al., 2003). The NTS/NTSR3 complex has been shown to phosphorylate both Erk1/2 and Akt kinases in a murine microglial cell line (Dicou, 2008).

# **NEUROTENSIN/NEUROTENSIN RECEPTOR COMPLEX AND CANCER BIOLOGY**

Few years after its discovery, high-level expression of NTS was found in the plasma of pancreatic tumor patients (Gutniak et al., 1980). This discovery inspired investigations on the relationship between NTS and cancer. Many studies have since been performed to clarify the role of NTS in carcinogenesis in diverse cancer cells.

# **PANCREATIC CANCER**

Pancreatic cancer is the eighth leading cause of cancer death in the world (Yabushita et al., 2012). It has the poorest prognosis amongst all human malignant solid tumors, mainly due to its high rate of metastasis (Cheng et al., 2012). The growth promoting action of NTS has been observed in pancreatic cancer cell lines both *in vitro* and *in vivo*. In both cases, the NTSR1 antagonist, SR48692, inhibited the NTS-induced effects (Sumi et al., 1993; Iwase et al., 1997). Recently, NTS was shown to protect insulin producing cells (b-TC3, INS-1E) against apoptosis induced by IL-1b and staurosporine (STS) (Coppola et al., 2008). NTSR2 and NTSR3 have been shown to be essential for the NTS mediated survival of these cells (Beraud-Dufour et al., 2009). NTS also influenced the migratory ability of pancreatic cancer cells, while NTS significantly reduced the migration levels of collectively migrating cells on vitronectin, NTS significantly increased the levels of individually migrating cells (Mijatovic et al., 2007). Thus, NTS-induced migration is dependent on the extracellular matrix environment and their respective migratory mode.

# **COLORECTAL CANCER**

Colorectal cancer is the third most common cancer worldwide and the fourth most common cause of death (Jemal et al., 2011). NTS stimulates the growth of mouse and human colon cancer cell lines in tissue culture and after being xenografted into nude mice (Maoret et al., 1999). *In vivo*, systemic NTS administration stimulates tumor size and weight, DNA, RNA, and the protein contents of the murine colon cancer, MC26 (Yoshinaga et al., 1992). Recently, the expression of NTSR1 was significantly correlated to an increase in the number of tumors when sporadic cancer was generated in mouse models by inflammation. However, no effect of NTSR1 expression was noticed on the number of aberrant crypt foci or tumor size, suggesting that the NTS/NTSR1 signaling complex has a major role in tumor progression (Bugni et al., 2012). NTS is also known to enhance colon cancer cell migration by increasing IL-8 expression and secretion. These effects are blocked by NTSR1 antagonists and curcumin, a diet-derived chemopreventive and/or chemotherapeutic agent which blocks AP-1 and NF-κB induction (Wang et al., 2006a). The role of IL-8 identified as an integral part of the metastasis process, was shown due to its triggering of the release of enzymes [MMPs, and uroplasminogen activator (uPA)], responsible for extracellular matrix degradation (Xie, 2001).

# **PROSTATE CANCER**

Prostate cancer is the most common male malignancy in Western countries and the second most common cause of male cancer-related death in the UK and USA (Jemal et al., 2011). The androgen-dependent human prostate cancer LNCaP cell line has been shown to exhibit an autocrine growth response to NTS in androgen-deprived only conditions (Sehgal et al., 1994). Vias et al. (2007) found that long-term anti-androgen treatment of LNCaP cells produced a sub-line exhibiting upregulated expression of NTS and NTS receptors, which increased the proliferation rate, accelerated cell cycle progression, and increased invasiveness through Matrigel. These effects are sensitive to NTS siRNA. NTSR1 expression was found at very high levels in the human androgen-independent PC3 cell line, derived from prostate cancer metastasized to bone. The growth responses of these cells to NTS were found at concentrations close to human postprandial blood levels (Seethalakshmi et al., 1997). These studies proposed that NTS is a potential autocrine, paracrine, and endocrine regulator of prostate cancer growth in humans, after androgen ablation therapy and during the devastating final stages of the disease.

# **LUNG CANCER**

Lung cancer is the most common cause of cancer-related deaths throughout the world (Jemal et al., 2011). High concentration of NTS is present in and secreted from half classic small cell lung cancer cells (SCLC; Moody et al., 1985). NTS is one of the 73 genes overexpressed in the highly metastatic human lung cell line, H460-M, as compared to control cells (de Lange et al., 2003). In lung cancer cell lines, NCI-H209 and H345, SR48692 inhibited NTS-mediated calcium mobilization cells and c-fos mRNA induction and proliferation in a dose-dependent manner (Moody et al., 1985). SR48692 also inhibited tumor growth of NCI-H209 xenografts.

#### **BREAST CANCER**

"fendo-03-00184" — 2013/1/12 — 17:45 — page 3 — #3

Breast cancer is the most common cancer in women, and the second leading cause of cancer deaths in women worldwide (Jemal et al., 2011). NTS immunoreactivity has been observed in breast cancers *in vivo*. NTSR1 expression has also been demonstrated in several breast cancer cell lines (Elek et al., 2000). The NTS anti-apoptotic effect has been described in the cell line MCF-7 originating from breast adenocarcinoma. Prolonged exposure to JMV449, a NTS-specific agonist, protected MCF-7 cells from serum deprivation-induced death, and reduced the number of apoptotic cells by two to three times. These effects have been predicted to be due, in part, to NTS-mediated induction of Bcl-2 mRNA and protein levels which depends on stimulation of MAPK (Somaï et al., 2002). By using NTSR1 Sh-RNA and SR 48692, tumor growth was significantly decreased when NTSR1 expression

was abolished or blocked in experimental tumors of the breast (Souaze et al., 2006a).

# **REGULATION OF THE NEUROTENSINERGIC SYSTEM REGULATION OF NTS**

Neurotensin deregulation has been observed in many cancers such as in colonic adenocarcinomas, small cell lung carcinomas, nonsmall cell lung adenocarcinomas, medullary thyroid carcinomas, and in fibrolamellar HCCs (Baca and Schmidt-Gayk, 1981; Ulich et al., 1983; Moody et al., 1985; Dammrich et al., 1988; Ehrenfried et al., 1994). DNA methylation has been shown to play a crucial role in the expression of the gut endocrine gene neurotensin/neuromedin N (NT/N). In the human hepatoma cell line Hep G2 cells, methylation of a NT/N promoter construct resulted in a severe reduction of the promoter activity, whereas treatment with the demethylating agent 5-azacytidine-induced NT/N expression (Dong et al., 1998). These observations have been confirmed in different human colon cancer cell lines, either expressing or not the gene coding for NTS (Dong et al., 2000). Interestingly, NTS gene can be regulated by Ras while expressed. Both wild-type and activated Ras enhances expression of NTS in the gut-derived CaCo-2 cell line, by activating the proximalAP-1/CRE motif (Evers et al., 1995). More recently, the PI3K catalytic subunit, p110alfa, was demonstrated to negatively regulate NTS secretion *in vitro* and *in vivo*. This process involves several regulatory proteins such as α-tubulin deacetylase, small GTPase, and kinase D-interacting substrate (Li et al., 2012).

#### **REGULATION OF NTSR1**

The mechanisms involved in NTSR1 deregulation in cancer cells have been studied in the context of colorectal carcinogenesis. These studies implicate an important role of the Wnt/β-catenin pathway deregulation. The mutation or loss of the protein APC (adenomatous polyposis coli) causes a dysfunction in the degradation of β-catenin. The accumulation of the latter in the cytoplasm, and its subsequent translocation to the nucleus induces NTSR1 gene expression via its association with transcription factors Tcf/Lef (T cell factor/lymphoid enhancing factor) (Souaze et al., 2006b). The NTSR1 promoter can be activated by the complex β-catenin/Tcf because it contains a consensus site for the transcription factors Tcf. In agreement with this result, it has been demonstrated that inhibitors of GSK-3β (protein kinase involved in the phosphorylation of β-catenin and its degradation) which cause the significant accumulation of β-catenin, upregulates the level of NTSR1 transcription (Wang et al., 2010). Similar results have been obtained in other cancers such as lung, prostate, and breast cancers (Chesire et al., 2002; Turashvili et al., 2006).

#### **REGULATION OF NTSR1 BY ITS OWN LIGAND**

Upon acute agonist exposure, and under physiological conditions, initiated by β-arrestin-1 (βARR1), and β-arrestin-2 (βARR2), the NTS/NTSR1 complex is internalized and degraded in lysosomes through clathrin-coated vesicles. Cell resensitization occurs from *de novo* receptor synthesis a few hours after agonist removal (Souaze and Forgez, 2006; Law et al., 2012). However, some studies on cellular models such as the murine neuroblastoma cell line N1E-115 and human colon cancer cell line HT-29, showed a change in the traffic situation when the cell had a prolonged exposure to saturating doses of agonist (Souaze et al., 1997; Najimi et al., 1998). Instead of being degraded in the lysosome, NTSR1 accumulated transiently with NTS in the perinuclear recycling compartment (PNRC) where it was latter recycled to the plasma membrane (Toy-Miou-Leong et al., 2004). More recent research has shown the activity of endothelin-converting enzyme-1 (ECE-1) and βARRs being crucial for NTSR1 recycling and enhance NTS degradation (Law et al., 2012). Thus, NTS stimulation induces cellular adaptation by altering the degradation process of NTSR1. This phenomenon leads to permanently sensitizing cells to the neurotensinergic signal. The implementation of this mechanism could lead to deregulation of multiple signaling pathways involved in the cancer progression such as MAPK and its target genes.

#### **NEUROTENSIN/NEUROTENSIN RECEPTORS AND THERAPY**

The implications of the previous sections suggest a more direct role for NTS/NTSR1 in cancer growth and progression, than has been previously attributed. Nevertheless, the ability to develop therapeutic strategies, around this complex, remain a challenge. Yet, despite them, the characteristics and qualities associated with this system should provide new pharmaceutical approaches as the system becomes further studied.

In the periphery and in the central nervous system, NTS mainly modulates the action of other molecules which are the principal effectors. Support for this view was confirmed by experiments with NTS- or NTSRs-deficient mice. These mice do not present any physiological disorder, are viable, and show normal growth and overt behavior. The deficient mice were also useful to implicate the NTS system in body temperature control (Pettibone et al., 2002), feeding regulation, weight control, and locomotion (Remaury et al., 2002). Mice were also less responsive or sensitive to exogenous agents or conditions. NTSR1 ko mice were less sensitive to the anorectic effects caused by leptin (Kim et al., 2008), and less responsive to morphine-induced analgesia (Roussy et al., 2010). Experiments with NTSR1- and NTSR2-deficient mice indicate a role in the regulation of ethanol consumption. However, NTSR1 regulates ethanol intoxication while NTSR2 is involved in ethanolinduced hypnosis (Lee et al., 2010, 2011). This suggests that the disruption of NTS system in the periphery is unlikely to generate side effects superior to the benefits expected in cancer treatment. On the other hand, the quasi exclusive alteration of the NTS system in tumors would confer an advantage in its use as a therapeutic target.

#### **BIOLOGICAL TOOLS USED TO INTERACT WITH THE NTS/NSTR SYSTEMS**

Few biological tools have been developed to specifically target NTS system. The two most studied ones are the NTSR1 antagonists, SR 48692, and SR142948A (Gully et al., 1996, 1997). These molecules were developed to counteract NTS effects in central and peripheral nervous system. For this purpose, both molecules are specific and efficient. These compounds have been extensively used to successfully counteract exogenous NTS oncogenic actions. Nevertheless, these molecules remain weak and problematic to use to counteract intense autocrine regulation. For example, at high concentrations, SR48692 is toxic for the cells, and as no significant information exists on the biodegradability of these molecules, the dosage levels

"fendo-03-00184" — 2013/1/12 — 17:45 — page 4 — #4

are unknown. In addition, SR 48692 acts as a low affinity agonist with NTSR2 (Yamada et al., 1998). NTRS2 expression also occurs in the gastric mucosa, in neuroendocrine cells of the stomach, and the small and large intestine, and in cells of the exocrine pancreas. NTSR2 is rarely detected in human-derived tumors (Schulz et al., 2006). Recently, NTSR2 was found overexpressed in B cell leukemia patient's cells (Saada et al., 2012), and upregulated in prostate cancer cells luminal phenotype (Swift et al., 2010). It will be difficult to estimate NTSR2 impact on tumorigenesis until more information on its expression in human tumors becomes available.

In contrast, NTSR3 exhibits an ubiquitous expression in normal and tumoral cells. Its contribution on neoplastic progression is not known. The exact impact of SR48692 on NTSR3 remains complex. The most advanced hypothesis is that SR 48692 inhibits the oncogenic effects through NTSR1, when NTSR1 is expressed and in conjunction or not with NTSR3. In this context it was shown that SR 48692 inhibited the tumor growth of colon, breast, pancreas, and small lung cancer cells xenografted in mice (Iwase et al., 1997; Maoret et al., 1999; Moody et al., 2001; Souaze et al., 2006a). SR 48692 counteracts exogenously treated NTS in pancreas cells MIA PaCa-2 cells, bearing high and low affinity sites (Iwase et al., 1997), or circulating NTS as in the case of SW480, MDA-MB 231, and NCI-H209 cells (Maoret et al., 1999; Moody et al., 2001; Souaze et al., 2006a). It has to be noted that the use of SR 48692 does not stop tumor growth, but delays the growth rate by two to three times. The creation of the experimental tumor in mice results in tumors having an abnormal ratio (up to 10×) between the tumor size and body size, an observation uncommon and not seen in human tumors. Summarizing the effects induced by NTS, at the onset of cancer, NTS acts on the progression rather than on cellular transformation. Those actions, mediated by NTS include proliferation, survival, and metastatic effects. It is, therefore, not surprising to observe that the abolition of NTS/NTSR1 expression or signaling enhances the effect of several anti-tumoral treatments. For example, NTSR1 inhibition efficiently sensitized prostate regulation in orthotopic human tumor xenografts in mice to radiotherapy by significantly reducing tumor size, and in prostate cancer cells bearing NTS autocrine regulation (Valerie et al., 2011).

Another approach employed in the development of specific biological tools, uses Sh-RNA or Si RNA targeting NSTR1. The total or partial abolition of the target results in a decrease in cellular invasion, migration in HNSCC (Shimizu et al., 2008), or tumor growth of the human breast cancer cells (Souaze et al., 2006a). The two experimental designs demonstrate a contribution of NTS in these oncologic processes, through its interaction with NTRS1. At present, more knowledge is needed on the distribution and the coupling of NTSR2 in tumoral cells, as well as on the pharmaceutical tools specifically inhibiting NTSR2, to determine if NTSR2 is implicated in any oncogenic process.

#### **TARGETED APPROACHES FOR NTS/NTSR1**

A recent approach under investigation is to use NTSR1-specific overexpression as a means to target the delivery of therapeutic or visualizing molecules in tumoral cells. This approach is made possible since NTSR1 endocytosis permits the introduction of molecules of interest inside the targeted cell and several strategies have been developed. In the first case, a non-viral gene transfer particle bearing six NTS molecules covalently bound to poly-lysine, enabled the transfer of a reporter and therapeutic gene to those tumor cells expressing NTSR1 when injected in the tumor or into the blood circulation of xenografted nude mice (Arango-Rodriguez et al., 2006). Interestingly, both injection sites showed a high proportion of transfected cells. In a second case, a therapeutic effect was also detected on experimental neuroblastoma tumors transfected with a thymidine kinase gene associated with ganciclovir. A severe reduction of tumor growth was also observed (Rubio-Zapata et al., 2009). In the same vein, a protein interfering with the cell synthesis machinery by inducing irreversible ADP-ribosylation of elongation factor 2, exotoxin A, was fused with NTS. This recombinant protein was able to specifically recognize NTSR1 positive cells and to exert a real *in vitro* cytotoxicity (Wick and Groner, 1997). In another strategy, oligo-branched peptides were used because of their proteolysis resistant nature. Tetra-branched NTS, armed with 5-FdU, increased *in vitro* and *in vivo* tumoral cell cytotoxicity as compared to thefree drug (Falciani et al., 2010a,b). The same idea was applied to carry the *cis*,*cis*,*trans*diamminedichloridodisuccinatoplatinum(IV)–neurotensin bioconjugate specifically to tumor cells. The cytotoxicity of this molecule was tested on cell proliferation and found efficient when compared to the non-targeted platinum (Gaviglio et al., 2012).

Neurotensin-based radiopharmaceuticals have also been developed for tumor localization and therapy. Several strategies using different NTSR1 ligands were constructed whose main objectives were to limit the degradation of the molecule while maintaining the ligand specificity and a good affinity for NTSRs. Thus, the analog NT-XIX, with the three enzymatic cleavage sites stabilized showed a high specificity for NTSR1. A low accumulation of activity in the kidneys and a proper tumor-to-tissue ratio radioactivity clearance was observed *in vivo*, as well as a decrease in xenograft tumor growth (Garcia-Garayoa et al., 2009). More recently, new NTS analogs, DOTA-NT-20.3, and DOTA-NT-20.4 showed promising characteristics for imaging of NTS receptor-positive tumors and therapy (Alshoukr et al., 2009, 2011).

#### **NTSR1/NTS COMPLEX EXPRESSION IN HUMAN TUMORS**

"fendo-03-00184" — 2013/1/12 — 17:45 — page 5 — #5

Several experiments have investigated the expression of NTSR1 in human tumors, but most of these studies remain too small or local. In normal colonic tissue, NTSR1 was not detected by immunohistochemistry, whereas in human colonic adenomas expression, NTSR1 was associated with cytoplasmic beta-catenin localization. This was one of the first suggestions that NTSR1 expression was an early event in colon carcinogenesis (Souaze et al., 2006b). NTSR1 mRNA expression studied by *in situ* hybridization showed a higher level of expression in adenocarcinoma as compared to adenomas. In addition, a higher intensity of NTSR1 expression was observed in filtrated adenocarcinomas into and beyond the muscularis propria as compared with tumors that were localized to the mucosa or submucosa, suggesting a contribution of NTS/NTSR1 complex in tumor expansion (Gui et al., 2008).

In pancreas, NTS binding sites were first studied by autoradiography, and found specifically in pancreatic cancer but not in normal pancreas and chronic pancreatitis. Nevertheless, northern blots showed NTSR1 mRNA in normal pancreatic cells, but an increase of transcript level in chronic pancreatitis and pancreatic cancer. Within the tumors, NTSR1 expression was higher in advanced stages as compared to early stages (Wang et al., 2000). NTSR1 and NTS protein expression was confirmed by immunohistochemistry in normal and tumoral tissue. In this latter case 80% expressed both NSTR1 and NTS (Wang et al., 2011).

More recently, a correlation between NTSR1 or NTS expression with outcome of the disease was evoked. In non-small lung cancer cells, NTS and NTSR1 expression were not detectable, whereas 60% of patients with stage 1 adenocarcinoma expressed either NTS or NTSR1 and 40% expressed both markers. Patients treated only by surgery did not receive adjuvant therapy and, in this cohort, NTRS1 expression was correlated with a worse prognosis of the disease (Alifano et al., 2010b). In contrast, NTS was found to be expressed in normal epithelial cells of breast tissue, whereas NTSR1 was absent. In breast invasive carcinomas high expression of NTSR1 was correlated with pejorative clinical parameters and disease outcome (Dupouy et al., 2009). In head and neck squamous cell carcinomas, NTS and NTSR1 mRNA high levels were significantly correlated with higher rates of distant metastasis as well as with the survival rate (Shimizu et al., 2008). In normal pleura, NTS and NTSR1 were found in 30 and 77%, respectively. In malignant pleural mesothelioma, expression increased to 71.1 and 90.4%, respectively. Interestingly in benign tissue, NTSR1 was located at the cell membrane whereas in tumoral cells, NTSR1 expression was granular and mainly restricted to the cytoplasm. The lack of presence at the cell surface suggested a state of permanent activation. In these cases, NTS high expression was associated with poor disease outcome (Alifano et al., 2010a).

# **CONCLUSION**

Cancer is more often regarded as a generalized disease and its treatment is mainly based on chemotherapy. In openly metastatic cancers, healing the sick is rarely achieved. In these cases only a prolongation of survival associated with a better quality of life can be expected. Therefore, it is necessary to improve our strategies. Recent treatment developments include the customization and selection of drugs specifically directed against biological and genetic abnormalities expressed by the tumor. However, as with conventional chemotherapy, these treatments are still insufficiently targeted and are often accompanied by serious side effects.

Two potential approaches (amongst others) to improve this situation include the highlighting of abnormalities specific to cancer cells, or the delivery of specific drugs against these pathological processes. This latter avenue would allow for effective treatment without side effects or benefit from a metabolic pathway in which it would be possible to connect conventional medicines and transfer them inside the cancer cells.

The neurotensin and neurotensin 1 receptor system have provided insights for certain oncological situations. As seen above, NTS/NTSR1 is overactive in many tumors (**Figure 1**). The development of active antagonists, and those well tolerated by the patient, in human clinical trials will be a necessary step. NTS/NTSR1 may also be considered to become a carrier of chemotherapy drugs for tumor cells. In this regard, platinum salts, such as cisplatin, whose digestive and renal toxicity is well documented, are a prime target for the development of such tests. In all cases, the effectiveness of treatment can only be improved if the selection of patients is likely to benefit from such treatment. The last 45 years of work on the NTS/NTSR1 system have allowed for a better understanding of normal and cancer cells. This system can be further exploited for therapeutic purposes and for improving the management of cancer.

"fendo-03-00184" — 2013/1/12 — 17:45 — page 6 — #6

# **REFERENCES**


GTP-binding component of adenylate cyclase. *Mol. Pharmacol.* 29, 489–496.


B. M. (1998). DNA methylation contributes to expression of the human neurotensin/neuromedin N gene. *Am. J. Physiol.* 274, G535– G543.


"fendo-03-00184" — 2013/1/12 — 17:45 — page 7 — #7

trans-diamminedichloridodisuccinatoplatinum(IV)-peptide bioconjugates. *Metallomics* 4, 260–266.


Gully, D., et al. (1997). Inhibition of neurotensin-induced pancreatic carcinoma growth by a nonpeptide neurotensin receptor antagonist, SR48692. *Cancer* 79, 1787–1793.


et al. (2012). PI3K p110alpha/Akt signaling negatively regulates secretion of the intestinal peptide neurotensin through interference of granule transport. *Mol. Endocrinol.* 26, 1380–1393.


and pathologic processes. *Curr. Opin. Endocrinol. Diabetes Obes.* 18, 75–82.


"fendo-03-00184" — 2013/1/12 — 17:45 — page 8 — #8

and endocrine hormone. *Prog. Histochem. Cytochem.* 16, 1–172.


*Neuropathol. Appl. Neurobiol.* 32, 575–584.


pancreatic cancer, MIA PaCa-2. *Surg. Oncol.* 2, 267–272.


"fendo-03-00184" — 2013/1/12 — 17:45 — page 9 — #9

T., et al. (2012). Circulating microR-NAs in serum of human K-ras oncogene transgenic rats with pancreatic ductal adenocarcinomas. *Pancreas* 41, 1013–1018.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 23 October 2012; accepted: 26 December 2012; published online: 17 January 2013.*

*Citation: Wu Z, Martinez-Fong D, Trédaniel J and Forgez P (2013) Neurotensin and its high affinity receptor 1 as a potential pharmacological target in cancer therapy. Front. Endocrin. 3:184. doi: 10.3389/fendo.2012.00184*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Wu, Martinez-Fong, Trédaniel and Forgez. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *David Chatenet 1,2\*, Thi-Tuyet M. Nguyen1,2, Myriam Létourneau1,2 and Alain Fournier 1,2\**

<sup>1</sup> Laboratoire d'études moléculaires et pharmacologiques des peptides, INRS – Institut Armand-Frappier, Université du Québec, Ville de Laval, QC, Canada <sup>2</sup> Laboratoire International Associé Samuel de Champlain (INSERM/INRS-Université de Rouen), France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Stacia A. Sower, University of New Hampshire, USA Ishwar S. Parhar, Monash University, Malaysia

#### *\*Correspondence:*

David Chatenet and Alain Fournier, Laboratoire d'études moléculaires et pharmacologiques des peptides, INRS – Institut Armand-Frappier, Université du Québec, 531 Boulevard des Prairies, Ville de Laval, QC H7V 1B7, Canada. e-mail: david.chatenet@iaf.inrs.ca; alain.fournier@iaf.inrs.ca

The urotensinergic system plays central roles in the physiological regulation of major mammalian organ systems, including the cardiovascular system. As a matter of fact, this system has been linked to numerous pathophysiological states including atherosclerosis, heart failure, hypertension, diabetes as well as psychological, and neurological disorders. The delineation of the (patho)physiological roles of the urotensinergic system has been hampered by the absence of potent and selective antagonists for the urotensin II-receptor (UT). Thus, a more precise definition of the molecular functioning of the urotensinergic system, in normal conditions as well as in a pathological state is still critically needed. The recent discovery of nuclear UT within cardiomyocytes has highlighted the cellular complexity of this system and suggested that UT-associated biological responses are not only initiated at the cell surface but may result from the integration of extracellular and intracellular signaling pathways. Thus, such nuclear-localized receptors, regulating distinct signaling pathways, may represent new therapeutic targets. With the recent observation that urotensin II (UII) and urotensin II-related peptide (URP) exert different biological effects and the postulate that they could also have distinct pathophysiological roles in hypertension, it appears crucial to reassess the recognition process involving UII and URP with UT, and to push forward the development of new analogs of the UT system aimed at discriminating UII- and URP-mediated biological activities. The recent development of such compounds, i.e. urocontrin A and rUII(1–7), is certainly useful to decipher the specific roles of UII and URP in vitro and in vivo. Altogether, these studies, which provide important information regarding the pharmacology of the urotensinergic system and the conformational requirements for binding and activation, will ultimately lead to the development of potent and selective drugs.

**Keywords: urotensin II, urotensin II-related peptide, allosteric modulation, biased agonist, nuclear receptors**

#### **THE UROTENSINERGIC SYSTEM**

During the last decade, the urotensinergic system has drawn the attention of the scientific community due to its marked involvement in various pathological states including cardiovascular diseases. Initially isolated from the caudal neurosecretory system of the teleostean fish *Gillichthys mirabilis*, urotensin II (UII), a somatostatin-like peptide, was first characterized as a spasmogenic agent (Pearson et al., 1980). During more than 15 years, this peptide and its unknown receptors were thought to be restricted to fishes until it was demonstrated that UII was able to induce the relaxation of the mouse anococcygeus muscle (Gibson et al., 1984) and provoke the contraction of rat aortic strips (Gibson, 1987). These results, suggesting the presence of an homologous peptide in higher vertebrates, led to the isolation and characterization of UII in the frog *Rana ridibunda* (Conlon et al., 1992). Following this discovery, UII isoforms were either characterized or isolated in various vertebrate species including humans (Vaudry et al., 2010). A few years later, a peptide paralog, termed urotensin II-related peptide (URP), was isolated in rat brain extracts and subsequently identified in other mammalian species (Vaudry et al., 2010). Sequence comparison of all UII and URP isoforms revealed a striking conservation of the C-terminal cyclic hexapeptide (Vaudry et al., 2010). Conversely, the N-terminal region is highly variable both in length, ranging from 11 residues in humans to 17 residues in mice, and sequence composition (**Figure 1**) (Vaudry et al., 2010). In the human genome, UII and URP genes are respectively found at position 1p36 and 3q29 (Sugo et al., 2003). Those two genes are primarily expressed in motoneurons located in discrete brainstem nuclei and in the ventral horn of the spinal cord (Vaudry et al., 2010). However, UII and URP mRNAs have also been detected, although at a much lower level, in various peripheral tissues including the pituitary, heart, spleen, lung, liver, thymus, pancreas, kidney, small intestine, adrenal, and prostate (**Figure 2**) (Vaudry et al., 2010).

Both peptides are endogenous ligands of a G protein-coupled receptor initially identified as the orphan GPR14 receptor (Ames et al., 1999; Liu et al., 1999; Mori et al., 1999; Nothacker et al., 1999). Structural studies of this urotensin II receptor (UT) showed that, in addition to the common features found in the 1A GPCR family, such as the existence of a disulfide bridge between extracellular loops 1 and 2, N-linked glycosylation sites in the N-terminus portion, and phosphorylation sites in intracellular


**FIGURE 1 | Amino acid sequences of UII and URP in mammalian species;** *<***Gln, pyroglumatic acid.** Modified from Vaudry et al. (2010).

loops (Douglas et al., 2000), this protein also possesses a palmitoylation site located in the C-terminal segment of the rodent isoform that is not present in the human isoform (**Figure 3**). Worth to mention, the rat UT, consisting of 386 amino acids, shows only 75% homology with the human protein while sequences of human and monkey receptors, comprising 389 residues, are almost identical (Elshourbagy et al., 2002). Like UII and URP, UT is widely expressed in the central nervous system as well as

in various peripheral organs including the cardiovascular system, kidneys, bladder, pancreas, and adrenal gland (**Figure 2**) (Vaudry et al., 2010).

The urotensinergic system plays a seminal role in the physiological regulation of major mammalian organ systems, including the cardiovascular system (Vaudry et al., 2010). As a matter of fact, UII exerts potent haemodynamic effects (Krum and Kemp, 2007), positive inotropic and chronotropic responses (Watson et al., 2003), and osmoregulatory actions (Song et al., 2006), induces collagen and fibronectin accumulation (Dai et al., 2007; Zhang et al., 2008), modulates the inflammatory response (Shiraishi et al., 2008), plays a role in the induction of cardiac and vascular hypertrophy (Papadopoulos et al., 2008), causes a strong angiogenic effect (Guidolin et al., 2010) and inhibits the glucose-induced insulin release (Silvestre et al., 2004). Thus, the urotensinergic system was linked to numerous pathophysiological states including atherosclerosis, heart failure, hypertension, pre-eclampsia, diabetes, renal and liver diseases, variceal bleeding, ulcers, as well as psychological, and neurological disorders (Ross et al., 2010).

The present review focuses on the latest findings about the urotensinergic system in terms of receptor localization and pharmacology as well as receptor activation with the conception of new urotensinergic ligands aimed at discriminating UII- and/or URP-mediated biological actions.

# **DISCOVERY OF AN INTRACRINE PHARMACOLOGY OF THE UROTENSINERGIC SYSTEM**

#### **PRESENCE OF NUCLEAR UT IN THE HEART AND IN THE CENTRAL NERVOUS SYSTEM**

In many ways, UII exhibits actions similar to other key neurohormonal factors, *i.e.* angiotensin II (Ang-II) and endothelin-1 (ET-1), in driving a variety of cardiac and vascular disease processes (Maguire and Davenport, 2002). These include vasoconstriction as well as mitogenic, trophic and pro-fibrotic effects (Vaudry et al., 2010). A clear interaction of the urotensinergic system with the renin-angiotensin-aldosterone and endothelin systems is acknowledged in terms of regulation of systolic and diastolic functions (Fontes-Sousa et al., 2009). However, key differences were observed between these systems. In particular, UII induces a rather weak or absent vasoconstriction in a variety of human vascular beds (Maguire et al., 2000; Hillier et al., 2001) while it can also acts as a vasodilator in some vascular beds, such as those in the pulmonary vasculature (Stirrat et al., 2001). The recent discovery of specific intracellular receptors associated with the physiological and pathophysiological actions of Ang-II and ET-1 highlighted a high level of complexity for these peptidergic systems in the regulation of cardiovascular homeostasis. Traditionally, GPCRs are located at the plasma membrane where they modulate the activity of membrane-associated second messengers. As such, GPCRs can exert their effects through the regulation of ion channels, second messenger production, and protein kinase cascades in order to control cellular activity, gene expression, plasticity, differentiation, morphogenesis, and migration. However, in the recent years, the presence of functional intracellular receptors has almost become "a classic GPCR paradigm" (Boivin et al., 2008). These intracellular GPCRs could be involved in the control of several cellular processes including regulation of gene transcription, ionic homeostasis, cellular proliferation, and remodeling (Boivin et al., 2008). Intracellular GPCRs may be constitutively active, or may be activated by ligands internalized from the extracellular space or synthesized within the cell (**Figure 4**). Besides, they can regulate signaling pathways distinct from those activated by the

same receptor at the cell surface (Re, 1999). As such, biological outcomes might result from the integration of extracellular and intracellular signaling events (Terrillon and Bouvier, 2004; Hanyaloglu and von Zastrow, 2008; Sorkin and von Zastrow, 2009). This new paradigm for cellular signaling provides more complexity to study the function and physiological roles of GPCRs.

In a recent report, specific UII binding sites were observed on heart and brain cell nuclei from rat and monkey tissues (Doan et al., 2012; Nguyen et al., 2012). Except those two tissues and the spinal cord, none of the tested tissues including kidneys, lung, and skeletal muscle, all expressing UT at the cell surface, presented a subcellular localization of UT (Doan et al., 2012; Nguyen et al., 2012). Supporting the presence of such nuclear expression also in humans, the presence of nuclear UT was also observed in two human cell lines, *i.e.* SH-SY5Y neuroblastoma and U87 astrocytoma cell lines (Nguyen et al., 2012).

# **NUCLEAR UT ACTIVATION CAN MODULATE TRANSCRIPTION INITIATION**

As previously reported for nuclear Ang-II (Eggena et al., 1993), β3-adrenergic (Boivin et al., 2006; Vaniotis et al., 2011), and ET-1 receptors (Boivin et al., 2003), nuclear UT receptors can initiate transcription (Doan et al., 2012; Nguyen et al., 2012). Although UII and URP stimulated the transcription in isolated brain cell nuclei (Nguyen et al., 2012), only UII was able to trigger a similar effect in rat cardiac nuclei (Doan et al., 2012). Two-dimensional gel electrophoresis clearly indicated the occurrence of different immunoreactive species in both brain and heart membrane and nuclear fractions (Doan et al., 2012; Nguyen et al., 2012). Nuclear and membrane proteins extracted from heart tissues expressed three major UT-immunoreactive spots with an apparent molecular weight of 60 kDa at a pI value of 6–7 (Doan et al., 2012). Interestingly, a different pattern was observed in brain tissue (Nguyen et al., 2012). Since the UT gene is intronless, the various immunoreactive species were principally ascribed to post-translational modifications (Doan et al., 2012; Nguyen et al., 2012). Whether or not these UT species are involved in distinct UII-associated biological activities will require further investigation. However, it is well-known that glycosylation can modulate the cellular compartmentalization and functionality of the receptor, thereby influencing its intracellular trafficking and biological activity (**Figure 4**) (Duverger et al., 1995; Rondanino et al., 2003; Gobeil et al., 2006).

A growing body of evidence supports the presence of GPCRs at the surface of the nuclear membrane, their orientation within this membrane, however, remains controversial. If they maintain the topology adopted in the endoplasmic reticulum during protein synthesis, the ligand binding site would be located in the lumen of the nuclear envelop (perinuclear space) with the C-terminal of the protein being localized either within or outside the nucleus. In fact, the topology of the nuclear membrane lumen is very similar to the extracellular space, which makes it a favorable environment for a binding site (Jong et al., 2005; Bootman et al., 2009). Since signaling starts with the recruitment of specific proteins to the C-terminal portion of the receptor, signals would be sent toward the cytosol or into the nucleus in accordance with the adopted GPCR orientation within the nuclear shell (**Figure 5**). Hence, the orientation of those nuclear GPCRs would determine the direction in which the signal is transmitted. As recently reported, nuclear UT receptors are able to regulate gene transcription (Doan et al., 2012; Nguyen et al., 2012). Furthermore, it is well-known that calcium ions play an important role in the control of gene expression (Bootman et al., 2009). In isolated nuclei, nuclear calcium levels can regulate gene transcription by interacting with the cyclic AMP response element-binding protein (CREB) and the downstream regulatory element antagonist modulator (DREAM), which are constitutively present in the nucleus. Changes in nucleoplasmic calcium can be achieved by triggering inositol(1,4,5)-triphosphate receptors (IP3Rs) located on the inner nuclear membrane (Bootman et al., 2009). Because it is generally accepted that UT activation is associated with the recruitment of Gαq/11 proteins to its C-terminal tail resulting in an IP3 increase (Proulx et al., 2008), it is highly probable that this portion of the receptor is located into the nucleoplasm (**Figure 5**). Interestingly, IP3Rs are concentrated in the nuclear membrane of heart ventricular cells and their activation was shown to initiate a pro-hypertrophic pathway (Arantes et al., 2012). These findings are well-correlated with the UII-induced cardiomyocyte hypertrophy and the presence of nuclear UT receptors in cardiac tissues (Gruson et al., 2010; Doan et al., 2012). Nevertheless, these intracellular UT receptors may have the capacity to regulate signaling pathways that differ from those of their plasma membrane counterparts, as recently demonstrated for the metabotropic glutamate receptor 5 (Jong et al., 2009), and the renin–angiotensin system (De Mello, 2008). As such, this intracrine pharmacology of the urotensinergic system represents a complementary system that could potentially involve the regulation of physiological functions.

#### **INTRACELLULAR TRAFFICKING OF UT**

This new intracrine pharmacology clearly highlights the complexity of this peptidergic system where UII and URP can trigger not only common but also different biological activities (Prosser et al., 2008; Jarry et al., 2010; Doan et al., 2012). Previous studies have detected the presence of GPCRs, such as Ang-II receptors, at the nucleus in an agonist-independent manner (Lee et al., 2004). Confocal microscopy of heart and brain tissue sections as well as various non-transfected cell lines clearly revealed a constitutive nuclear localization for UT (Doan et al., 2012; Nguyen et al., 2012). However, it is also possible that following their agonist-stimulated internalization, GPCRs relocate at the nuclear membrane (Lee et al., 2004). In such a case, the translocation is initiated by the presence of a nuclear localization signal (NLS), a short stretch of basic amino acid residues often localized within the intracellular loops that is recognized by importins α and/or β (**Figure 4**). For example, a NLS was observed in the seventh transmembrane domain and the carboxy-terminal segment of the Ang-II receptor subtype1 (Lys-Lys-Phe-Lys-Arg) and the third intracellular loop of the apelin receptors (Arg-Lys-Arg-Arg-Arg) (Lee et al., 2004). Interestingly, a similar sequence, *i.e.* Lys-Arg-Ala-Arg-Arg, is also observed in the third intracellular loop of human and monkey UT isoforms (**Figure 3**) while a Lys-Gln-Thr-Arg-Arg segment is observed in rat and mouse UT. However, it is important to note that many NLS signals are still unknown and that the presence of an obvious NLS motif may mask the existence of still uncharacterized NLS sequences. Specific post-translational modifications such as palmitoylation were reported to be involved in the addressing of the receptor either to the membrane or the nuclei. For instance, it was demonstrated that de-palmitoylation of GRK6A promoted its translocation from the plasma membrane to both the cytoplasm and nucleus (Jiang et al., 2007). Such a putative palmitoylation site is also found within the seventh transmembrane domain (Cys339) of rat and mouse UT isoforms but is absent in the primate (human and monkey) receptor (Marchese et al., 1995; Tal et al., 1995; Ames et al., 1999). Under chemically mediated hypoxic conditions, an increase of total UT expression, was observed suggesting that hypoxia might induce *de novo* synthesis of the peptide receptor. However, a significant decrease in nuclear UT expression was reported that could be interpreted as an increase in translocation of the protein to the membrane or a decrease of internalization with concomitant nuclear translocation (Nguyen et al., 2012). Altogether, it could be noted that the subcellular UT localization could be either attributed to translocation from the cell surface and/or *de novo* synthesis (**Figure 4**).

#### **UII AND URP AS INTRACRINE LIGANDS**

UII, and by extension URP, were originally thought to act as autocrine and paracrine modulators rather than as hormones (Yoshimoto et al., 2004). The term "intracrine" ligand relates to intracellular molecules binding to and activating intracellular receptors (**Figure 4**). Such ligands can be synthesized and targeted to the Golgi apparatus for secretion or act intracellularly either before secretion or following reuptake. The intracrine gene product might also arise from an alternative transcription initiation site, differences in mRNA maturation or translation leading to a gene product lacking secretory signals and consequently active only in the intracellular space (**Figure 4**) (Kiefer et al., 1994; Lee-Kirsch et al., 1999; Xu et al., 2009). To this extent, it is interesting

to note that two isoforms of the human UII precursor, differing mostly by their peptide signal, were discovered (Coulouarn et al., 1998; Ames et al., 1999).

A recent study demonstrated that FITC-conjugated hUII and URP were both internalized in non-expressing UT cell lines through receptor-independent mediated endocytosis (Doan et al., 2012) (**Figure 4**). This receptor-independent endocytic mechanism brought a new perception of the pseudo-irreversible binding characteristics often described for the urotensinergic system. Indeed, the lack of rapid UT desensitization through classic mechanisms (acid wash or trypsin treatments) was thought to reflect a strong, pseudo-irreversible binding of the ligands (Douglas and Ohlstein, 2000). However, this pseudo-irreversible character could also be due to the ability of both endogenous peptides to reach the internal compartment of the cell. Moreover, it is yet possible that following the internalization of ligand-receptor complexes, ligands are subsequently released from internalized endosomes within the cell. As such, internalized peptide-receptor complexes can be dissociated under the acidic environment found in endosomes, giving rise to receptor recycling at the plasma membrane (**Figure 4**) (Giebing et al., 2005). At this point, the fate of the peptide is unknown but based on the results published by Doan et al. (2012), it is conceivable that UII, and to a lesser extend URP, could leak from the vesicle and ultimately activate intracellular receptors.

#### **PROSPECTIVE ROLES OF NUCLEAR UT**

The precise role of this new intracrine urotensinergic system has yet to be elucidated both in physiological and pathological conditions. However, as for other GPCRs including Ang-II and ET-1 receptors, these intracellular receptors are important regulators of physiological and pathological functions and could therefore represent new targets for therapeutic interventions (Boivin et al., 2008; Tadevosyan et al., 2012).

Elevated UII plasma levels were observed in numerous disease conditions, including hypertension, atherosclerosis, heart failure, pulmonary hypertension, diabetes, renal failure, and metabolic syndrome (Ross et al., 2010). As demonstrated, the cellular uptake of UII but not URP is increased at lower pH (Doan et al., 2012). Pathological conditions such as cancer, ischemic stroke, inflammation, and atherosclerotic plaques often induce an increase in metabolic activity and hypoxia associated with an elevated extracellular acidity (Andreev et al., 2010). In these conditions, UII would enter more easily than URP inside the cell triggering transcription of UII-associated genes by activating the nuclear receptor. Thus, the elevated concentration of UII observed during the etiology of various diseases could sustain specific cellular responses while an intracellular feedback loop could maintain a particular cellular state (Petersen et al., 2006). Interestingly, known intracrines do not present any structural or chemical similarities but are generally growth regulators that can directly or indirectly modulate angiogenic or anti-angiogenic actions. Therefore, the angiogenic actions of the urotensinergic system, reported both *in vivo* and *in vitro* (Spinazzi et al., 2006), could thus involve the activation of nuclear UT.

The urotensinergic system is also highly expressed in the central nervous system, but its physiological function is still poorly understood. UT was observed in cortical astrocytes (Castel et al., 2006), a ubiquitous type of glial cell that greatly outnumbers neurons and occupies 25% to 50% of brain volume (Bignami et al., 1991). It is noteworthy that glioblastoma multiform (GBM) is characterized by exuberant angiogenesis, a key event in tumor growth and progression and that UII, URP, and UT mRNAs were systematically found to be expressed in different glioma and glioblastoma tumors (Diallo et al., 2007). These results support a role for the urotensinergic system, and in particular nuclear UT, in human brain tumorogenesis possibly via angiogenesis regulation. Finally, in the CNS, UII is able to induce norepinephrine, dopamine, and serotonin release in noradrenergic neurons (Ono et al., 2008). Intracerebroventricular UII administration modulates cardiac homeostasis via β-adrenoreceptor activation (Hood et al., 2005). These observations bring up the idea that the presence of nuclear UT receptors could also be associated to excitatory neurotransmission. In accordance with this hypothesis, various intracrines were reported to act as neurotransmitters within the CNS (Re, 2004).

Whether specific UII or URP biological actions on the CNS and the cardiovascular system are mediated totally or in part by the nuclear UT will need further studies as well as the development of specific nuclear UT probes. Although still poorly understood, the diverse functions exerted by agonists and hormones acting on intracellular GPCRs suggest that intracrine signaling might activate cellular responses distinct from those at the cell surface for a given receptor. In the last decade, biological actions of intracrines in heart and vasculature, including those of the renin–angiotensin-system in cardiac pathology, dynorphin B in cardiac development, as well as endothelin, highlighted the importance of intracrine physiology in pathological processes such as left ventricular hypertrophy, diabetic cardiomyopathy, and arrythmogenesis. So, the presence of functional UT receptors at the cell membrane and at the nucleus will probably be a new aspect to take into account during the development of therapeutic compounds for the treatment of pathologies associated with the urotensinergic system.

# **NEW INSIGHTS INTO UT ACTIVATION**

The precise definition of the (patho) physiological roles of the urotensinergic system *in vivo* was hampered by the absence of potent and selective UT antagonists. Indeed, the lack of efficacy observed with Palosuran (ACT-058362) (Clozel et al., 2004, 2006), the only UT antagonist that reached a phase II clinical trial in patients with diabetic nephropathy, was clouded by its low antagonist potency (Behm et al., 2008). Therefore, drug discovery programs continued to focus on the identification of potent and selective UT antagonists suitable for assessment in both preclinical species and man (Maryanoff and Kinney, 2010). As reported earlier this year, Sanofi launched a phase I clinical trial regarding a long acting UT antagonist, derived from a 5,6-bisaryl-2-pyridine-carboxamide scaffold (European patent application *EP2439193*), for the treatment of diabetic nephropathy. Similarly, GlaxoSmithKline started a phase I clinical trial for the use of an UT antagonist, *i.e.* SB1440115 (United States Patent application *12,373,901*), for the treatment of asthma. Finally, over the past few years, Boehringer-Ingelheim (European patent application *EP2155748*) as well as Janssen Pharmaceutical (United States Patent application *8,193,191*) filled several patents regarding UT antagonists but no phase I clinical trial was yet reported. With the recent discovery that UII and URP could exert common as well as different biological activities (Prosser et al., 2008; Hirose et al., 2009; Jarry et al., 2010), development of selective UT antagonists has become a more complex task.

# **UT AS SHAPESHIFTING PROTEINS**

GPCRs represent the largest and most diverse family of cell surface receptors. These plasma membrane proteins bind their endogenous ligands in order to activate an intracellular signaling cascade that will result in a biological action. Conventional views of ligand-receptor activation considered all components of the signaling cascade to be linearly related, *i.e.* to emanate from the initial activation of the receptor. However, multiple studies pointed out the ability of some ligands to selectively trigger specific signaling pathways, therefore having collateral and not linear efficacy (Roettger et al., 1997; Kohout et al., 2004). As such, GPCRs cannot be considered as pharmacological on/off switches anymore. Their intrinsic nature rather suggests that dynamic changes in the receptor conformation, resulting from ligand binding, are a mean of information transfer (Kenakin and Miller, 2010). Hence, the propensity of GPCRs to assume multiple conformations make them allosteric proteins that are able to select specific subsets of secondary messengers depending on the ligand-induced adopted conformation. As such, various ligands were reported to possess differential functional profiles for a given receptor, as it was initially described for the CCR7 receptor (Kohout et al., 2004).

The URP sequence, strictly conserved throughout species, supports the concept that specific receptor interactions were maintained despite variation in the receptor amino acid sequence (Elshourbagy et al., 2002). Based on the specific expression of

URP mRNA in several cerebral structures (rostroventrolateral medulla) and tissues (heart, seminal vesicle), it was suggested that URP rather than UII would be the biologically active peptide in the UT-associated regulation of autonomic, cardiovascular and reproductive functions (Dubessy et al., 2008). Moreover, distinct pathophysiological roles for UII and URP in hypertension have been suggested (Hirose et al., 2009). Indeed, mRNA expression of both UII and URP was up-regulated in the atrium of spontaneously hypertensive rats (SHR) when compared with age-matched Wistar Kyoto (WKY) rats. However, the specific upregulation of URP but not UII mRNA in aorta and kidney of SHR rats supported the idea that these peptides might act individually in various biological systems (Hirose et al., 2009). Accordingly, it was demonstrated that UII and URP were able to exert not only common but also divergent physiological actions clearly suggesting the propensity of these two endogenous ligands to select a specific UT conformation (Prosser et al., 2008; Jarry et al., 2010; Doan et al., 2012). The concept of biased agonism has recently emerged from various studies, putting forward the notion that specific ligand-induced conformational changes can lead to particular signaling (Patel et al., 2010). In isolated ischaemic heart experiments, UII and URP were both able to reduce myocardial damages through creatine kinase reduction but only UII was able to reduce atrial natriuretic peptide (ANP) production (Prosser et al., 2008). Supporting the idea that UII and URP interact with UT in a distinct manner, it was recently demonstrated that URP, an equipotent UII paralog, was able to accelerate the dissociation rate of membrane bound 125I-hUII while hUII had no noticeable effect on URP dissociation kinetics (Chatenet et al., 2012a). Altogether, these results suggest that each ligand is able to select a specific UT conformation that triggers definite biological activities for each of these two peptides.

The peptide N-terminal region was initially pointed out as potentially involved in the observed biological activity differences between UII and URP (Prosser et al., 2008). However, it is only recently that this region was clearly associated with a putative differential binding mode of hUII and URP (Chatenet et al., 2012a). Using exocyclic Ala-derivatives of hUII, acting as very potent ligands of the UT receptor (Brkovic et al., 2003), dissociation kinetics experiments revealed a putative interaction between UT and the glutamic residue at position 1 of hUII. Indeed, it was observed that the replacement of this residue by an alanine moiety, *i.e.* [Ala1]hUII, provoked an increase of the dissociation rate of hUII but not URP (Chatenet et al., 2012a). In agreement with this view, an important electrostatic interaction between Glu1 of hUII and its receptor was previously reported in docking studies (Lescot et al., 2008). No other substitution was able to induce such pharmacological changes. Because this compound was reported to exert almost equipotent contractile activity compared to hUII and URP, the lost of a putative specific interaction with UT was thought to have generated an analog behaving as an URP derivative therefore acting at the URP-associated orthosteric binding site. Supporting this hypothesis, hUII(4–11), considered as the minimal hUII fragment to exert full biological activity, was also able to alter hUII but not URP dissociation rate (Chatenet et al., 2012a). This N-terminal segment could thus be crucial for signaling pathway selection upon activation and its deletion could lead to signaling mechanism misinterpretation, all UII truncated analogs potentially acting as URP derivatives.

Overall, these results support the presence of specific pockets/interactions within UT, aimed at selecting distinct UT conformations that can differentiate UII and URP biological activities (**Figure 6**). Briefly, it is hypothesized that upon the initial UII-UT interaction, involving the N-terminal region of UII, UT undergoes conformational changes aimed at welcoming the C-terminal domain of UII, characterized by an intracyclic β-turn (Carotenuto et al., 2004). To the opposite, URP, lacking this N-terminal portion and characterized by the presence of an intracyclic γ-turn (Chatenet et al., 2004), would bind UT in a slightly different manner; ultimately triggering a slightly different subset of signaling pathways. These observations have clearly highlighted the crucial need to reassess the development of pan UT antagonists, *i.e.* blocking UII- and URP-mediated receptor activation, and to develop new analogs of the urotensinergic system aimed at discriminating UII- or URP-mediated biological activities. Such compounds would allow a better understanding of the pathophysiological roles of the urotensinergic system and also expand our knowledge on allosteric modulation of class A GPCRs.

#### *Allosteric modulation of the urotensinergic system*

As demonstrated, the urotensinergic system is far more complex than previously thought with the presence of nuclear receptors and UII/URP specific as well as common actions. For the past two decades, identification of peptidic and non-peptidic agonists and antagonists of the urotensinergic system has gathered much interest for the treatment of various cardiovascular pathologies. An extensive review regarding the various peptidic and non-peptidic ligands of the urotensinergic system, all acting as competitive compounds, is beyond the scope of this review but more details can be found elsewhere (Maryanoff and Kinney, 2010).

Additional biological complexity, but also novel opportunities for drug discovery, has arisen from the fact that many GPCRs possess allosteric binding sites (Christopoulos and Kenakin, 2002). Similar to the initial concept of agonism, *i.e.* linear efficacy, antagonism has been historically viewed as a simple "turning off " of the receptor. As such, this non-accommodating mechanism does not allow any agonist to impart information to the receptor, the orthosteric binding site being occupied by the competitive antagonist. However, an allosteric modulator binds to its own site, different from the orthosteric site, forming a complex characterized by the concomitant presence of the endogenous agonist and the allosteric modulator. Such modulators can alter the biological properties of the endogenous orthosteric ligand either via changing its affinity, its efficacy, or both (Leach et al., 2007; May et al., 2007). This type of antagonism, termed permissive, can modify the reactivity of the receptor toward the agonist probably through conformation selection and stabilization of one or

part of the receptor states. Targeting receptor allosteric sites can offer the possibility of greater selectivity due to a lower sequence conservation within allosteric pockets across subtypes of a given GPCR, as well as the potential to fine-tune physiological signaling in a more spatial and temporally-selective manner (Kenakin, 2011).

The urotensinergic system, encompassing two endogenous peptides, provides potential for allosteric compounds to differentially modulate individual peptide responses, a behavior termed "probe dependence" (Kenakin, 2008). During the course of structure-activity relationship studies on URP derivatives, two compounds, *i.e.* [Bip4]URP and [Pep4]URP, termed urocontrin and urocontrin A (UCA) respectively, showed a specific behavior that has set them apart from known UT antagonists (**Figure 7**) (Chatenet et al., 2012a,b). Indeed, these compounds were able to selectively and significantly reduce hUII-induced contraction without altering URP-mediated vasoconstriction (Chatenet et al., 2012a,b). For instance, the efficacy of hUII-induced rat aortic ring vasoconstriction was significantly reduced (∼31%) by a pretreatment with a nanomolar concentration of UCA (**Figure 8**). Interestingly, this ability to selectively and significantly reduce UII-induced contraction was not specie-dependent since a similar effect was observed on cynomolgus monkey aortic rings. To the best of our knowledge, only two other UT ligands exerted insurmountable activity (Herold et al., 2003; Behm et al., 2010). However, none of them could differentially alter hUII and URP biological activity. The insurmountable nature of urocontrin and UCA antagonism was attributed to an allosteric modulation of UT. Indeed, an excess of UCA accelerated the 125I-hUII dissociation rate, thus suggesting that the binding of the antagonist changes the receptor conformation in such a way that the radioligand is released from the receptor. Accordingly, no difference in 125I-URP dissociation kinetics was observed in similar conditions (**Figure 8**). The apparent absence of effect on the URP pharmacological profile by UCA was attributed to its ability to select a receptor conformation through functional allosteric modulation that impairs hUII-associated actions but not URP-mediated

biological activities. As stated above, for a given receptor, an allosteric modulation that depends on the type of orthosteric ligand used is referred to as "probe dependence" (Kenakin, 2005; Keov et al., 2011). This probe dependence phenomenon supports the idea that the two endogenous ligands, despite depicting a high structure homology and recognizing a similar binding pocket, represent chemically distinct entities interacting in different structural environments within the orthosteric pocket. Because hUII and URP differ only by the length and composition of their N-terminal domain (Vaudry et al., 2010), it was postulated that this region could be involved in their putative different binding modes. Corroborating this hypothesis, the hUII counterpart of UCA, *i.e.* [Pep7]hUII,

**FIGURE 8 | Schematic representation of a proposed allosteric modulation of the urotensinergic system by urocontrin A and rUII(1–7).** By acting at an allosteric binding site, UCA is able to modify the receptor topography preventing the proper interaction of UT with the linear UII N-terminal region ultimately leading to an inefficient activation characterized

by a reduced efficacy. On the opposite, such a receptor conformational change has no effect on URP-mediated action. Conversely, binding of the rUII(1–7) N-terminal segment, initiates a topographical change that antagonizes the effect of URP, but not UII. Modified from Chatenet et al. (2012a).

acted as a weak but full agonist of the UT receptor (Chatenet et al., 2012a). This N-terminal domain is thus able to modulate the topology of the receptor in such a manner that the C-terminal domain of UII is able to trigger receptor activation. However, could this N-terminal segment be biologically active? As an agonist, the N-terminal domain of rat UII [rUII(1–7)], *i.e.* Pyr-His-Gly-Thr-Ala-Pro-Glu-amide (**Figure 7**), was unable to induce the contraction of rat aortic rings (Chatenet et al., 2012a). Amazingly, pre-treatment of rat aortic rings with rUII(1–7) induced an apparent increase in rUII contractile efficacy while reducing the potency and the efficacy of URP-mediated vasoconstriction (Chatenet et al., 2012a). These results clearly suggested that rUII(1–7) acted as a probe dependant allosteric modulator on rUII- and URP-mediated vasoconstriction (**Figure 8**) (Chatenet et al., 2012a). Since all UII isoforms possess different N-terminal domains, it is hypothesized that these regions could act as specie-selective specific URP modulators but there is currently no clue regarding an endogenous production of those N-terminal UII domains *in vivo*.

# **CONCLUSIONS**

These latest findings about the urotensinergic system will probably generate a considerable interest within the scientific

# **REFERENCES**


intact cells and vascular tissues. *Br. J. Pharmacol.* 155, 374–386.


community. First, the discovery of UT on the nuclear membrane and the presence of intracellular ligands open up new avenues in UT signaling physiology. In general, nuclear-localized receptors may regulate distinct signaling pathways, suggesting that biological responses mediated by GPCRs are not only initiated at the cell surface but might result from the integration of extracellular and intracellular signaling pathways. These receptors are therefore well-positioned to play major roles in the physiological and pathophysiological responses associated with their endogenous ligands. Finally, the discovery of allosteric modulators of the urotensinergic systems such as urocontrin, UCA, and rUII(1–7), will surely enable a better understanding of the urotensinergic system by allowing to discriminate *in vitro* and *in vivo* specific biological actions mediated by UII and/or URP. Therefore, these unique derivatives will be useful as chemical templates for the rational design of novel UT receptor ligands, as well as pharmacological tools for *in vitro* and particularly *in vivo* studies aimed at clarifying the role(s) played by the UII/URP/UT receptor system in physiology and pathology.

# **ACKNOWLEDGMENTS**

This work was supported by the Canadian Institutes for Health Research (awarded to Alain Fournier).


allosterism and complexing. *Pharmacol. Rev.* 54, 323–374.


transforming growth factor-beta1 secretion in urotensin II-induced collagen synthesis in neonatal cardiac fibroblasts. *Regul. Pept.* 140, 88–93.


Molecular and pharmacological characterization of genes encoding urotensin-II peptides and their cognate G-protein-coupled receptors from the mouse and monkey. *Br. J. Pharmacol.* 136, 9–22.


T., et al. (2009). Increased expression of urotensin II, urotensin II-related peptide and urotensin II receptor mRNAs in the cardiovascular organs of hypertensive rats: comparison with endothelin-1. *Peptides* 30, 1124–1129.


Competition between nuclear localization and secretory signals determines the subcellular fate of a single CUG-initiated form of FGF3. *EMBO J.* 13, 4126–4136.


M., Nishioka, R. S., et al. (1980). Urotensin II: a somatostatin-like peptide in the caudal neurosecretory system of fishes. *Proc. Natl. Acad. Sci. U.S.A.* 77, 5021–5024.


pancreatic extracts and inhibits insulin release in the perfused rat pancreas. *Eur. J. Endocrinol.* 151, 803–809.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 26 September 2012; paper pending published: 30 October 2012; accepted: 10 December 2012; published online: 02 January 2013.*

*Citation: Chatenet D, Nguyen T-TM, Létourneau M and Fournier A (2013) Update on the urotensinergic system: new trends in receptor localization, activation, and drug design. Front. Endocrin. 3:174. doi: 10.3389/fendo.2012.00174*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Chatenet, Nguyen, Létourneau and Fournier. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Taltirelin is a superagonist at the human thyrotropin-releasing hormone receptor

# *Nanthakumar Thirunarayanan, Bruce M. Raaka and Marvin C. Gershengorn\**

Laboratory of Endocrinology and Receptor Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, USA

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Hubert Vaudry, University of Rouen, France

Ludovic Galas, Institut National de la Santé et de la Recherche Biomédicale, France

#### *\*Correspondence:*

Marvin C. Gershengorn, Laboratory of Endocrinology and Receptor Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, 50 South Drive, Room 4134, Bethesda, MD 20892, USA. e-mail: marving@intra.niddk.hih.gov

Taltirelin (TAL) is a thyrotropin-releasing hormone (TRH) analog that is approved for use in humans in Japan. In this study, we characterizedTAL binding to and signaling by the human TRH receptor (TRH-R) in a model cell system. We found that TAL exhibited lower binding affinities than TRH and lower signaling potency via the inositol-1,4,5-trisphosphate/calcium pathway than TRH. However, TAL exhibited higher intrinsic efficacy than TRH in stimulating inositol-1,4,5-trisphosphate second messenger generation. This is the first study that elucidates the pharmacology of TAL at TRH-R and shows that TAL is a superagonist at TRH-R. We suggest the superagonism exhibited by TAL may in part explain its higher activity in mediating central nervous system effects in humans compared to TRH.

**Keywords: taltirelin, thyrotropin-releasing hormone, human TRH receptor**

"fendo-03-00120" — 2012/10/6 — 19:44 — page 1 — #1

# **INTRODUCTION**

Based on evidence that thyrotropin-releasing hormone (TRH) modulates a number of central nervous system (CNS) activities including arousal, antidepressant activity, anxiolytic effects, increase in locomotor activity, antagonism of pentobarbitalinduced sedation, thermoregulation, and cardiovascular and gastrointestinal autonomic functions (Horita et al., 1986; Horita, 1998; Khomane et al., 2011), many analogs of TRH were synthesized and studied. Taltirelin (TAL) hydrate [(1-methyl-(S)-4,5 dihydroorotyl)-histidyl-prolinamide, TA-0910] is an analog that showed improved CNS activity (Suzuki et al., 1990; Yamamura et al., 1991, 1997) and lower thyrotropin (TSH)-releasing activity (Yamamura et al., 1997) compared to TRH in rodents. Based on these characteristics, TAL was studied as a treatment for neurodegenerative disorders and is the only TRH analog that has been approved for use in humans; it is used in Japan to treat patients with adult spinal muscular atrophy (Ceredist®).

In some mammals, including rodents, there are two subtypes of G protein-coupled (or seven-transmembrane-spanning) receptors for TRH. These are TRH receptor 1 (TRH-R1), which is the primary (or only) receptor in TSH-secreting cells, and TRH-R2, which is expressed throughout the CNS along with TRH-R1 but typically in different areas of the brain (O'Dowd et al., 2000; Sun et al., 2003). In humans, by contrast, only a single type of receptor for TRH, TRH-R, is expressed that is more similar to TRH-R1 than TRH-R2 (Duthie et al., 1993; Matre et al., 1993). It has been reported that TAL binds with lower affinity than TRH to receptors in rat or mouse pituitary and in brain tissue preparations (Kinoshita et al., 1997; Asai et al., 1999, 2005). However, the pharmacology of TAL at TRH-R has not been characterized.

In this study, we characterized TAL binding and signaling by TRH-R in a model cell system and show that TAL is a superagonist at TRH-R.

# **MATERIALS AND METHODS MATERIALS**

Dulbecco's modified Eagle's medium (DMEM) and fetal bovine serum were purchased from Biosource (Rockville, MD, USA). TRH (pyroGlu-His-ProNH2) and MeTRH (pGlu-His(1(τ ) methyl)-ProNH2) were purchased from Sigma (St. Louis, MO, USA). [3H]MeTRH was purchased from PerkinElmer (Waltham, MA, USA). TAL (N-[[(4S)-Hexahydro-1-methyl-2,6-dioxo-4 pyrimidinyl]carbonyl]-L-histidyl-L-prolinamide) was obtained from Tocris (San Diego, CA, USA).

# **CELL CULTURE AND GENERATION OF CELLS STABLY EXPRESSING TRH-R**

HEK-EM 293 (human embryonic kidney) cells stably expressing TRH-R were generated as follows. The human TRH-R cDNA in pcDNA3.1(+) was obtained from the Missouri S&T cDNA Resource Center (Rolla, MO, USA) and was subcloned into the pcDNA3.1(+)/hygromycin vector. HEK-EM 293 cells were transfected with the cDNA of TRH-R using FuGENE 6 transfection reagent (Roche Diagnostics GmbH, Mannheim, Germany) and the cell clones stably expressing TRH-R were selected using hygromycin (250 μg/ml). HEK-EM 293 cells stably expressing TRH-R were grown in DMEM containing 10% fetal bovine serum, 100 U/ml penicillin, 10 μg/ml streptomycin, and 200 μg/ml hygromycin B (Invitrogen, Carlsbad, CA, USA) at 37◦C in a humidified 5% CO2 incubator.

#### **COMPETITION BINDING**

Competition binding assays were performed in monolayers of intact HEK cells expressing TRH-Rs. The cells (220,000 cells/well in 24-well plates) were preincubated with various concentrations of unlabeled TAL, TRH, or MeTRH for 15 min before addition of radioligand and then incubated at 37◦C for 1 h with 4 nM [3H]MeTRH as described elsewhere (Engel et al., 2006). Non-specific binding was determined in incubations with excess non-radiolabeled MeTRH. IC50 is the concentration of unlabeled ligand that reduces specific binding of [3H]MeTRH by 50%. The receptor number per cell was calculated from competition binding curves of various doses of unlabeled MeTRH and 4 nM [3H]MeTRH (**Figure 1**) and found to be 16,000/cell.

# **MEASUREMENT OF INTRACELLULAR CALCIUM MOBILIZATION**

Cells stably expressing TRH-R were seeded in black-walled, clearbottomed 96-well plates (Corning, NY, USA) at a density of 60,000 cells/well in DMEM with 10% fetal bovine serum and incubated for 24 h at 37◦C in 5% CO2. The following day, the culture media was replaced with 100 μl of Hank's balanced salt solution with 20 mM HEPES, pH 7.5 and the cells were loaded with 100 μl of calcium 4 fluorescent dye (Molecular Devices, Sunnyvale, CA, USA) for 1 h at room temperature before addition of compounds. Transient changes in intracellular [Ca++] induced by TAL, TRH, or MeTRH were measured using the FLIPRTETRA system (Molecular Devices, Sunnyvale, CA, USA). Changes in fluorescence were detected at the emission wavelength ranges from 515 to 575 nm. The agonistic responses of ligands were assessed

immediately upon their addition in a concentration range from 0.1 nM to 30 μM. Responses were measured as peak fluorescent intensity minus basal fluorescent intensity at each compound concentration and are presented as % of the maximum response.

# **MEASUREMENT OF IP1 PRODUCTION**

Cells were seeded at 220,000/well in white, solid bottom, tissue culture-treated 24-well plates and cultured at 37◦C with 5% CO2 overnight. Serial dilutions of TAL, TRH, or MeTRH, in Hank's balanced salt solution with 20 mM HEPES and 50 mM LiCl, pH 7.4, were added at 200 μl/well on the second day. After 60 min incubation at 37◦C in 5% CO2, inositol monophosphate (IP1) content was measured using the IP-One ELISA kit (Cisbio International, France) according to the manufacturer's protocol. The results were calculated as IP1 nanomoles/well and are presented as % of the maximum response.

# **DATA ANALYSIS**

The dose–response data were analyzed by non-linear regression of curve fit with one-site competition using GraphPad Prism software version 4 (GraphPad, Inc., San Diego, CA, USA) and the significance was determined by *t*-test or ANOVA.

# **RESULTS**

"fendo-03-00120" — 2012/10/6 — 19:44 — page 2 — #2

The pharmacology of TAL binding and signaling was studied in HEK-EM 293 cells, which do not endogenously express TRH-Rs, engineered to express 16,000 TRH-Rs/cell (**Table 1**). In competition binding assays using 4 nM [3H]MeTRH and various doses of unlabeled MeTRH, the concentration of MeTRH that halfmaximally inhibited [3H]MeTRH binding (IC50) was 3.0 nM (**Figure 1**). The half-maximally effective concentration (EC50) of MeTRH for stimulating an increase in cytosolic Ca2<sup>+</sup> concentration (Ca2<sup>+</sup> release) was 7.2 nM, which is not different from the IC50 (*p* > 0.1). Thus, MeTRH is an agonist at TRH-R with high potency. We compared the effects of TAL and TRH in competing for [3H]MeTRH binding and on stimulating Ca2<sup>+</sup> release (**Figure 2**). The IC50 values were 910 and 36 nM for TAL and TRH, respectively, and the EC50 values were 36 and 5.0 nM for TAL and TRH, respectively. Thus, TRH was a high potency agonist and TAL was a moderate potency agonist at TRH-R. We noted, moreover, that the IC50/EC50 ratio was 25 for TAL (*p* < 0.02), 7.2 for TRH (*p* < 0.05), and approximately 0.4 for MeTRH. When comparing two agonists, the agonist with the higher IC50/EC50 ratio has a greater intrinsic efficacy (Engel et al., 2006). These findings suggested that TAL may be a more efficacious agonist than TRH and that TRH is more efficacious than MeTRH. We previously showed that TRH exhibited a higher intrinsic efficacy than MeTRH at

#### **Table 1 | Pharmacological parameters for TRH-R.**


mouse TRH-Rs (Engel et al., 2006). As TRH is the natural, full agonist, TAL is termed a superagonist, and MeTRH is a partial agonist.

sigmoidal dose–response method.

As Ca2<sup>+</sup> release is a rapid and transient response to TRH-R activation, it is easier to compare relative intrinsic efficacies by quantifying activation of the inositol-1,4,5-trisphosphate pathway. Inositol-1,4,5-trisphosphate production is the step prior to Ca2<sup>+</sup>

release in signal transduction by TRH-Rs (Gershengorn, 1986) and can be quantified by measuring accumulation of its metabolic product IP1 over time by inhibiting IP1 degradation (**Figure 3A**). The EC50 values for IP1 production were found to be 150 nM for TAL and 3.9 nM for TRH. More importantly, TAL was clearly more efficacious than TRH in that TAL stimulated an increase in IP1 production that was 180% of that stimulated by TRH (*p* < 0.001).

Another way of demonstrating relative intrinsic efficacies of agonists is to show that the maximal response of a more efficacious agonist is inhibited by a less efficacious agonist (Engel et al., 2006). **Figure 3B** illustrates the IP1 responses to maximally effective doses of MeTRH, TRH, and TAL. As is evident, TAL (3.5 fold over MeTRH) is more efficacious than TRH (2.1-fold over MeTRH); MeTRH is the least efficacious. As predicted, the least efficacious agonist MeTRH inhibited the response to TRH (full agonist) and to TAL (superagonist).

#### **DISCUSSION**

In this study, we characterized the binding and signaling properties of TAL at TRH-R that, to our knowledge, have not been previously reported. The binding properties of TAL at rodent TRH receptors have been studied previously (Asai et al., 1999; Brown, 1999). In agreement with the findings with rodent receptors, we found that TAL binds to TRH-R with lower affinity than TRH. Our most interesting observation, however, is that TAL exhibits higher intrinsic efficacy than TRH; that is, TAL can stimulate the same level of signaling as TRH but at lower levels of receptor occupancy and could induce higher levels of signaling than TRH at full occupancy (**Figure 3**). Since TRH is the natural, full agonist for TRH-R, TAL is termed a superagonist. We previously reported that other TRH analogs displayed higher intrinsic efficacies than TRH at rodent TRH receptors (Engel et al., 2006). In the same study, we showed that MeTRH, the only TRH analog with higher affinity and potency than TRH, was a partial agonist that displayed lower intrinsic efficacy than TRH and that when high levels of TRH and MeTRH were added simultaneously the level of signaling was lowered to that of MeTRH. This is the predicted effect of adding a

"fendo-03-00120" — 2012/10/6 — 19:44 — page 3 — #3

**(B)** MeTRH inhibits IP1 formation stimulated by TAL and TRH. IP1 production was determined in the presence of TRH (0.1 μM), TAL (3 μM), MeTRH (3 μM), TRH + MeTRH, or TAL + MeTRH. The data are expressed as mean ± SEM performed in duplicate of two experiments.

partial agonist along with a full agonist. We used a similar experimental design herein to confirm that TAL is a superagonist at TRH-R; MeTRH antagonized IP1 production stimulated by both TRH and TAL (**Figure 3B**).

Previous reports in rodents showed that TAL displayed more activity in stimulating CNS effects than TRH (Suzuki et al., 1990; Brown, 1999). TAL may have similar CNS effects in humans (Gary et al., 2003; Khomane et al., 2011). The differences in the activities of TAL and TRH in the CNS have been attributed to the higher stability in blood and increased penetration of the blood–brain barrier of TAL compared to TRH. Although this is likely true, our new findings of the signaling efficacy of TAL at TRH-R suggest that the higher intrinsic efficacy of TAL may be contributing to its CNS activity in humans also.

# **REFERENCES**


(TRH) hypothesis of homeostatic regulation: implications for TRHbased therapeutics. *J. Pharmacol. Exp. Ther.* 305, 410–416.


In summary, we have described characterization of the pharmacology of TAL at TRH-R. Most importantly, we showed that TAL is a superagonist when signaling at TRH-R via the Gq/<sup>11</sup> protein-phospholipase C-phosphatidylinositol-4,5-bisphosphateinositol-1,4,5-trisphosphate-calcium pathway. We suggest the superagonism exhibited by TAL may, in addition to its relative metabolic stability and ability to cross the blood–brain barrier, explain its higher activity in mediating CNS effects in humans compared to TRH.

#### **ACKNOWLEDGMENTS**

This work was supported by the Intramural Research Program of the National Institutes of Diabetes and Digestive and Kidney Diseases, US National Institutes of Health (1 Z01 DK011006).

human thyrotropin-releasing hormone receptor. *Biochem. Biophys. Res. Commun.* 195, 179–185.


"fendo-03-00120" — 2012/10/6 — 19:44 — page 4 — #4

*Yakurigaku Zasshi* 110(Suppl. 1), P33–P38.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 10 July 2012; paper pending published: 20 August 2012; accepted: 20 September 2012; published online: 09 October 2012.*

*Citation: Thirunarayanan N, Raaka BM and Gershengorn MC (2012) Taltirelin is a superagonist at the human thyrotropin-releasing hormone receptor. Front. Endocrin. 3:120. doi: 10.3389/ fendo.2012.00120*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Thirunarayanan, Raaka and Gershengorn. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# **Patricia M. Hinkle<sup>1</sup>\*, Austin U. Gehret <sup>2</sup> and BrianW. Jones <sup>3</sup>**

<sup>1</sup> Department of Pharmacology and Physiology, University of Rochester Medical Center, Rochester, NY, USA

<sup>2</sup> Department of Science and Mathematics, National Technical Institute for the Deaf, Rochester Institute of Technology, Rochester, NY, USA

<sup>3</sup> Department of Pharmacology, University of Washington, Seattle, WA, USA

#### **Edited by:**

Liliane Schoofs, Catholic University of Leuven, Belgium

#### **Reviewed by:**

Perry Barrett, University of Aberdeen, UK Leo T. Lee, The University of Hong Kong, Hong Kong

#### **\*Correspondence:**

Patricia M. Hinkle, Department of Pharmacology and Physiology, University of Rochester Medical Center, 601 Elmwood Avenue Rochester, Rochester, NY 14642, USA. e-mail: patricia\_hinkle@

urmc.rochester.edu

The pituitary receptor for thyrotropin-releasing hormone (TRH) is a calcium-mobilizing G protein-coupled receptor (GPCR) that signals through Gq/11, elevating calcium, and activating protein kinase C. TRH receptor signaling is quickly desensitized as a consequence of receptor phosphorylation, arrestin binding, and internalization. Following activation, TRH receptors are phosphorylated at multiple Ser/Thr residues in the cytoplasmic tail. Phosphorylation catalyzed by GPCR kinase 2 (GRK2) takes place rapidly, reaching a maximum within seconds. Arrestins bind to two phosphorylated regions, but only arrestin bound to the proximal region causes desensitization and internalization. Phosphorylation at Thr365 is critical for these responses. TRH receptors internalize in clathrin-coated vesicles with bound arrestin. Following endocytosis, vesicles containing phosphorylated TRH receptors soon merge with rab5-positive vesicles. Over approximately 20 min these form larger endosomes rich in rab4 and rab5, early sorting endosomes. After TRH is removed from the medium, dephosphorylated receptors start to accumulate in rab4-positive, rab5 negative recycling endosomes.The mechanisms responsible for sorting dephosphorylated receptors to recycling endosomes are unknown. TRH receptors from internal pools help repopulate the plasma membrane. Dephosphorylation ofTRH receptors begins whenTRH is removed from the medium regardless of receptor localization, although dephosphorylation is fastest when the receptor is on the plasma membrane. Protein phosphatase 1 is involved in dephosphorylation but the details of how the enzyme is targeted to the receptor remain obscure. It is likely that future studies will identify biased ligands for theTRH receptor, novel arrestin-dependent signaling pathways, mechanisms responsible for targeting kinases and phosphatases to the receptor, and principles governing receptor trafficking.

**Keywords: dephosphorylation, desensitization, internalization, phosphorylation, Rab, thyrotropin-releasing hormone, trafficking,TRH receptor**

# **INTRODUCTION**

Thyrotropin-releasing hormone (TRH), a hypothalamic tripeptide, sits atop the hypothalamic/pituitary/thyroid axis, stimulating release of thyrotropin from the anterior pituitary gland. TRH also stimulates prolactin secretion (**Figure 1**). Initially characterized in clonal rat pituitary cell lines, TRH binding sites were later identified in brain membranes (Hinkle and Tashjian, 1973; Burt and Snyder, 1975). The link between TRH receptors and G proteins was made at a time when heterotrimeric G proteins were thought to transduce signals only from receptors coupled to adenylyl cyclase (Hinkle and Kinsella, 1984; Hinkle and Phillips, 1984). It quickly became clear that TRH receptors belong to the G protein-coupled receptor (GPCR) superfamily. A single TRH receptor gene has been found in humans and higher mammals and two genes encoding homologous receptors, TRHR1 and TRHR2, in rodents (Sun et al., 2003). TRHR1 predominates in the anterior pituitary gland while both TRHR1 and TRHR2 are found in rodent CNS (O'Dowd et al., 2000). Thyroid hormones exert powerful feedback inhibition over the TRH response system by inhibiting TRH synthesis and processing in TRH neurons in the paraventricular region of the hypothalamus and decreasing TRH receptors and responses in the pituitary gland (Gershengorn, 1978; Perrone and Hinkle, 1978; Hinkle and Goh, 1982; Segerson et al., 1987; Fekete and Lechan, 2007; Costa and Hollenberg, 2012). Prolonged hypothyroidism leads to a 40-fold increase in TRH receptor mRNA levels in pituitary glands (Costa et al., 2012). In addition, pyroglutamyl peptidase, a highly specific TRH-degrading ectoenzyme, is dramatically increased in hyperthyroid animals (Schomburg and Bauer, 1997; Marsili et al., 2011).

Genetic deletion of either the TRH peptide precursor or TRHR1 results in central hypothyroidism and mild growth retardation in mice, consistent with the established role of TRH in thyroid physiology (Yamada et al., 1997; Rabeler et al., 2004; Zeng et al., 2007). Interestingly, loss of the TRH receptor appears to result in a more severe phenotype than loss of the TRH peptide, possibly because the TRH receptor displays physiologically significant constitutive activity. Mice lacking TRHR1 also exhibit low prolactin and hyperglycemia, as well as increased anxiety and depression. Some of these behavioral effects may be secondary to hypothyroidism. Mice lacking TRHR2 display a subtle phenotype

consistent with depression and decreased anxiety (Sun et al., 2009). In humans,absence of the TRH receptor causes low free T3 and free T4 with thyroid stimulating hormone (TSH) levels inappropriately low for the degree of hypothyroidism (Bonomi et al., 2009). Lack of a TRH receptor does not result in infertility or failure to lactate.

This review focuses on aspects of TRH receptor signaling that have been elucidated in either pituitary cell models expressing endogenous receptors (pituitary tissue, primary cultures of pituitary cells, cell lines derived from pituitary tumors) or generic cell lines expressing transfected TRH receptors (HEK293, CHO, COS, Hela). In rat anterior pituitary tissue, TRH receptor mRNA is found not only in TSH-secreting cells but also in prolactin- and/or growth hormone-secreting cells (Konaka et al., 1997). Although the great majority of pituitary cells that bind rhodamine-labeled TRH and respond to TRH with an increase in intracellular calcium stain for prolactin or TSH-β, responses are occasionally observed in unexpected cell types (Ashworth et al., 1995; Villalobos et al., 1997; Yu et al., 1998). Signal transduction of endogenous TRH receptors has not been characterized in neurons, where no suitable model system has been identified. Even though there is only one TRH receptor in higher mammals, that receptor may be regulated quite differently in pituitary cells and neurons because of different patterns of TRH stimulation and different sets of G protein

subunits, effectors, receptors kinases, and phosphatases, arrestins, and downstream targets.

As outlined in **Figure 2**, TRH receptors are textbook calciummobilizing receptors: they are coupled to Gq and G11, which activate phospholipase Cβ (PLCβ), hydrolysis of phosphatidylinositol (4,5)bisphosphate, and release of the second messengers inositol (1,4,5)triphosphate (IP3) and diacylglycerol (DG) that in turn mobilize intracellular calcium and activate protein kinase C (Drummond et al., 1984; Drummond, 1985; Gershengorn, 1989). Two of the four forms of phospholipase Cβ (PLCβs 1 and 3) are found in anterior pituitary tissue, and calcium responses to TRH are intact in mice lacking PLCβ3 (Romoser et al., 2001). Thus PLCβ1, which is stimulated by Gαq/11-GTP but not Gβγ, generates second messengers in response to TRH. Maximal concentrations of intracellular free calcium are attained at concentrations of TRH at least an order of magnitude below the apparent *K*<sup>d</sup> of the receptor and the EC<sup>50</sup> for IP3 production. These initial events are followed by complex changes that result in depolarization and a sustained influx of extracellular calcium through voltage-gated L-type calcium channels (Hinkle et al., 1996; Barros et al., 1997). When TRH receptors are expressed in embryonic fibroblasts from mice lacking Gα<sup>q</sup> and Gα11, TRH does not stimulate any increase in intracellular calcium.

The TRH receptor is a rhodopsin family GPCR with typical features including seven transmembrane domains, several extracellular glycosylation sites, an essential disulfide bond between the first and second extracellular loops, a fairly short third intracellular loop, and a cytoplasmic tail. The receptor has a canonical (D/E)RY at the cytoplasmic end of the third transmembrane segment, the usual NPxxY at the cytoplasmic end of the seventh transmembrane helix, and an intracellular eighth helix anchored by two palmitoylated Cys residues in the carboxyl tail (Du et al., 2005). Except for the most distal region, the cytoplasmic tail of the receptor is conserved among species. The C-terminal amino acids of numerous GPCRs form classical PDZ ligands, sequences that can interact with proteins bearing a PDZ domain; PDZ is an acronym for three proteins containing the domain, PSD98, Dlg1, and zo-1 (Romero et al., 2011). TRH receptors, however, do not contain sequences predicted to interact with PDZ domains. Using a combination of experimental data obtained with receptor mutants, multiple ligands, and molecular modeling, Gershengorn and his colleagues have developed a sequential model for receptor binding and activation in which the TRH tripeptide first interacts weakly with extracellular loops and then moves deeper into the receptor and contacts transmembrane domains to form a high affinity complex. The conformational changes that occur upon receptor activation involve movements in transmembrane helices 5 and 6 and the third intracellular loop that result in G protein coupling (Gershengorn and Osman, 1996).

#### **DESENSITIZATION OF THE TRH RECEPTOR**

Thyroid stimulating hormone displays a circadian pattern that is attributed to rhythmic release of hypothalamic TRH. As a consequence, desensitization and resensitization of the TRH signaling pathway are likely to play central roles in regulating TRH actions *in vivo*. TRH rapidly stimulates a large increase in IP3 mass, providing a proximal, easily quantified response. Measurements of IP3 production have established that the TRH receptor undergoes classical desensitization. Continuous application of TRH leads to a transient burst of IP3 that peaks within seconds and falls within a minute, though to levels that remain above baseline (Drummond et al., 1984; Gershengorn and Osman, 1996; Yu and Hinkle, 1997, 1998). When cells are exposed to TRH intermittently, the size of the

IP3 response diminishes with successive stimulations. The extent of desensitization depends on the method used to measure it and the cell type under study (Falck-Pedersen et al., 1994). Decoupling the receptor from G proteins contributes to the transient nature of the IP3 elevation, which is more sustained in cells expressing a TRH receptor that lacks most of the cytoplasmic tail and is thereby spared from usual desensitization processes (Yu and Hinkle, 1998; Jones and Hinkle, 2005), as shown in **Figure 3**. Changes further downstream in the signal pathway can also decrease TRH responses by mechanisms including protein kinase C-mediated inhibition of phospholipase Cβ and slow refilling of intracellular calcium stores (Yu and Hinkle, 1997). TRH receptors were the first GPCRs shown to undergo what is now termed homologous downregulation (Hinkle and Tashjian, 1975; Gershengorn, 1978). Incubation with TRH decreases the number of TRH binding sites without changing receptor affinity in pituitary GH3 cells. The molecular basis for downregulation is still not completely understood.

Like many rhodopsin family GPCRs (Bulenger et al., 2005; Milligan, 2009; Lohse, 2010), TRH receptors form oligomers when they are overexpressed in non-native cells (Kroeger et al., 2001; Hanyaloglu et al., 2002; Zhu et al., 2002). Bioluminescence resonance energy transfer (BRET), co-precipitation and biochemical analyses have all been used to document TRH receptor homodimers, and the two rodent TRH receptors (TRHR1 and TRHR2) form heterodimers (Hanyaloglu et al., 2002). TRH increases the apparent fraction of receptors in dimers quantified by all of these techniques, perhaps because TRH causes receptors to concentrate in clathrin-coated pits and endosomes. Truncated receptors that do not recruit arrestin or internalize can form dimers

incubated with TRH and IP3 mass was measured at intervals (graphs). To visualize arrestin, receptors were expressed with GFP-labeled arrestin3. To with them but it does not interact with truncated receptors, which do not internalize.

constitutively but not in response to TRH (Zhu et al., 2002). When receptors are coerced to form oligomers by addition of a chemical dimerizing agent, signaling is not altered but internalization is accelerated and recycling blunted (Song and Hinkle, 2005). It has been difficult to assess the physiological relevance of TRH receptor dimers because no mutations that prevent oligomerization have been identified and available methods are inadequate to characterize endogenous receptors in pituitary cells. Recent studies (Whorton et al., 2007) have provided compelling evidence that at least some rhodopsin family GPCRs can activate G proteins as monomers, yet there are clear examples of receptors that form heterodimers with unique pharmacology (Jordan and Devi, 1999). One possibility is that that only one member of a GPCR pair signals to a single G protein. The TRHR1/TRHR2 heterodimer interacts with arrestins and internalizes in a different pattern from either receptor alone (Pfleger et al., 2004), but the intriguing possibility that the TRH receptor forms heteromers with other GPCRs to impart novel pharmacology has not been explored.

# **OVERVIEW OF GPCR RECEPTOR DESENSITIZATION**

Most GPCRs become phosphorylated after they are activated, and the TRH receptor is no exception. GPCR phosphorylation is catalyzed by members of a family of Ser/Thr protein kinases, GPCR kinases, or GRKs. There are seven GRKs, four of them widely expressed (GRKs 1, 2, 5, and 6; Moore et al., 2007; Gurevich et al., 2012). What sets the GRKs apart from other kinases is their ability to discriminate between GPCRs in inactive versus activated conformations. They act preferentially on agonist-occupied and constitutively active receptors and do not have strict requirements for the amino acid sequence surrounding phosphorylation sites. There are important differences in how different GRKs are recruited to the plasma membrane, however, and examples of GPCRs that are phosphorylated at different sites by different GRKs. GRKs exert a number of kinase-independent effects on GPCR signaling and are known to phosphorylate substrates other than receptors (Moore et al., 2007; Premont and Gainetdinov, 2007; Evron et al., 2012; Gurevich et al., 2012). GPCRs can also be phosphorylated by kinases that become activated in response to signaling such as cAMP-dependent protein kinase and protein kinase C. Phosphorylation carried out by such downstream kinases can influence GPCR-G protein coupling, desensitization, and trafficking.

Once activated and phosphorylated, most GPCRs bind to arrestin, which interrupts the interaction between an activated receptor and its cognate G protein and terminates signaling via the G protein-mediated pathway. Arrestins 2 and 3 (also referred to as β-arrestin1 and β-arrestin2) are ubiquitously expressed. Like G proteins and GRKs, arrestins distinguish between inactive and activated receptor conformations. Arrestins bind preferentially to activated GPCRs; they also contain a positively charged pocket that interacts strongly with negatively charged phosphates (Gurevich and Gurevich, 2006; DeWire et al., 2007; Moore et al., 2007; Premont and Gainetdinov, 2007; DeFea, 2011; Shenoy and Lefkowitz, 2011). The arrestins therefore tend to bind with highest affinity to receptors that are both activated and phosphorylated. When arrestin engages receptor, a buried helical region in the arrestin carboxyl terminus undergoes a major translocation to expose binding

sites for AP2 and clathrin, targeting the arrestin-receptor complex to coated pit regions of the membrane destined to pinch off into endosomes in a dynamin-dependent process (Moore et al., 2007; Shenoy and Lefkowitz, 2011).

Arrestin is a scaffolding protein capable of anchoring a wide array of kinases, phosphatases, and other proteins to the arrestin-GPCR complex (Xiao et al., 2007; DeFea, 2011). A rapidly expanding body of literature shows that arrestin-GPCR complexes can initiate alternate signaling pathways, notably those leading to MAP kinase activation (Lefkowitz and Shenoy, 2005; Luttrell and Gesty-Palmer, 2010; Reiter et al., 2012). The Lefkowitz group has proposed that the phosphorylation pattern of a GPCR can act as a barcode in which different phosphosites recruit different arrestins to turn on different signal pathways (Liggett, 2011; Nobles et al., 2011). This model is appealing because it adds tremendous diversity to GPCR signaling, and it is supported by evidence that arrestin2 and arrestin3 bind preferentially to distinct phophosites on several receptors (Ren et al., 2005). There are also many examples of what are termed biased ligands that favor particular pathways (Rajagopal et al., 2010; Reiter et al., 2012). For example, biased ligands may preferentially activate G protein or arrestin cascades, activate one arrestin signaling function but not another (such as kinase activation and internalization), or act as an agonist in one pathway and an antagonist in another. Biased agonists and antagonists open new doors for the development of therapeutically useful GPCR ligands.

G protein-coupled receptors have been broadly divided into two groups based on their interactions with arrestins (Oakley et al., 1999, 2000). Class A receptors bind arrestin 3 more strongly than arrestin 2 or visual arrestin, internalize without associated arrestin, and recycle rapidly. Examples include the well-characterized β2 adrenergic receptor, the µ-opioid receptor and dopamine D1 receptor. Class B receptors bind well to both arrestin 2 and arrestin 3, internalize with arrestin, and traffic to lysosomes where they undergo intracellular degradation or slow recycling. Examples include the V2 vasopressin receptor, type 1a angiotensin II receptor, and the TRH receptor.

#### **PHOSPHORYLATION OF TRH RECEPTORS**

Thyrotropin-releasing hormone receptors become phosphorylated rapidly once they are activated. Receptor phosphorylation has been demonstrated by <sup>32</sup>P incorporation, an upward mobility shift on gel electrophoresis, and reactivity with phosphositespecific antibodies (Hanyaloglu et al., 2001; Zhu et al., 2002; Jones and Hinkle, 2005; Jones et al., 2007). All of these changes are reversed by phosphatase treatment. Phosphorylation sites in the TRH receptor have been partially mapped by characterizing <sup>32</sup>P incorporation and antibody reactivity of wildtype receptors and receptors lacking various Ser and Thr residues in the cytoplasmic tail. Jones et al. generated antisera against multiply phosphorylated peptides from five different conserved regions of the TRH receptor tail and validated their specificity (Jones and Hinkle, 2008). None of the antibodies recognize TRH receptors from unstimulated cells, but four of them react with receptors from TRH-treated cells. Endogenous TRH receptors in pituitary cells are phosphorylated strongly at residues between amino acids 355 and 365 and less efficiently at two regions farther downstream. Current information about phosphosites in the TRH receptor C-terminus is presented in **Figure 4**. As discussed below, phosphorylation of Thr365 is particularly important for arrestin recruitment, internalization, and desensitization. There is no evidence for phosphorylation at Tyr residues. The distal regions of the TRH receptor are more heavily phosphorylated when the receptor is expressed in HEK293 or CHO cells compared to native receptors. The finding of different site usage in different settings suggests a need for caution when charactering phosphorylation of overexpressed epitope-tagged GPCRs in heterologous cells, an approach that is normally required for analysis by mass spectrometry.

Thyrotropin-releasing hormone receptor phosphorylation does not appear to be hierarchical because the rates of phosphorylation are the same for different phosphosites. TRH does not promote <sup>32</sup>P incorporation into receptors truncated before the palmitoylation site and TRH does not alter mobility of truncated receptors, leading to the conclusion that either intracellular loops are not phosphorylated in response to TRH or phosphorylation in the intracellular loops somehow requires the receptor tail. Following TRH addition, nearly all receptors appear to be phosphorylated based on the mobility shift and near quantitative immunoprecipitation with phosphosite-specific antibodies (Jones et al., 2007).

In pituitary cells, endogenous receptors on the plasma membrane are not detectably phosphorylated in the basal state, but they become strongly phosphorylated within 10 s of TRH addition and endocytosis of phospho-receptors is readily apparent by 10 min (**Figure 5**). Interestingly, when rat pituitary tissue from

residues are shown in pink or red in regions known to undergo TRH-dependent phosphorylation. Phosphorylation at Thr365, in red, is essential for desensitization and internalization, and phosphorylation at other sites between amino acids 355 and 365 has been documented and shown to Phosphorylation in this region is not essential for desensitization and internalization, but it may be important for different, as yet unidentified signaling functions. Phosphorylation in the non-conserved distal region has not been examined.

an untreated animal is examined, phosphorylated TRH receptor is clearly visible. The intensity of the phospho-receptor signal increases greatly minutes after animals are injected with TRH and, as expected, the signal is seen cells that stain for TSH-β and prolactin (Jones et al., 2007).

Thyrotropin-releasing hormone receptor phosphorylation is exceedingly fast, with a half-time of 0.2 min (Jones et al., 2007; Gehret and Hinkle, 2010). This rapid phosphorylation is unusual among GPCRs and marks the TRH receptor as a superior substrate. Rapid phosphorylation is not simply a consequence of the sequence of the receptor tail. The transmembrane helices and intracellular loops of the TRH receptor are likewise important for rapid phosphorylation of the receptor tail even though they are not phosphorylated themselves. A chimeric TRH receptor with the β2-adrenergic receptor tail is phosphorylated at β2-adrenergic receptor GRK sites with kinetics typical of the TRH receptor, not the β2-adrenergic receptor (Gehret and Hinkle, 2010). The ratelimiting step for TRH receptor phosphorylation may be recruitment of GRKs to the activated receptor. In support of this idea, dimerization of a mutant TRH receptor that cannot undergo activation with a truncated TRH receptor that signals well but cannot undergo phosphorylation results in phosphorylation of the signaling-incompetent partner (Song et al., 2007). This is most easily explained if the activatable receptor recruits a kinase that acts on its partner in the receptor dimer. The predicted amphipathic eighth helix between the canonical NPxxY at the end of the seventh transmembrane domain and the palmitoylated Cys-X-Cys motif is not phosphorylated but the positively charged residues in this region are essential for phosphorylation at downstream sites and subsequent internalization (Gehret et al., 2010).

Considerable evidence points toward GRK2 as the kinase responsible for TRH receptor phosphorylation. Phosphorylation is inhibited by dominant negative, kinase-dead forms of GRK2 and by siRNAs targeting GRK2 (Jones and Hinkle, 2005; Jones et al., 2007). The effects of siRNA knockdown are seen both in pituitary cells and heterologous cells expressing transfected receptors. In addition, paroxetine, which has recently been recognized as an effective and relatively specific GRK2 inhibitor, delays phosphorylation of the TRH receptor (Thal et al., 2012). In keeping with the rapid rate of TRH receptor phosphorylation, GRK2 translocates to membrane fractions within 10 s of TRH addition (Jones and Hinkle, 2005). All of these experiments implicate GRK2 in TRH receptor phosphorylation, but because GRK2 knockdown and inhibition do not block receptor phosphorylation completely it seems likely that other GRKs are also able to phosphorylate the receptor.

Inhibitor data suggest that casein kinase II, which is predicted to phosphorylate Thr365 and two other downstream sites, may contribute to receptor phosphorylation (Hanyaloglu et al., 2001). Pharmacological activation of conventional protein kinase C isoforms leads to weak phosphorylation of some but not all sites in the TRH receptor tail, but protein kinase C inhibitors do not alter TRH-dependent phosphorylation (Jones et al., 2007). Furthermore, TRH receptor phosphorylation and internalization occur normally in cells lacking Gα<sup>q</sup> and Gα11, indicating that calcium- and diacylglycerol-activated kinases are not essential (Yu and Hinkle, 1999; Jones and Hinkle, 2005). Together these results suggest little role for downstream kinases in TRH receptor phosphorylation.

#### **ARRESTIN INTERACTIONS WITH THE TRH RECEPTOR**

Arrestin interactions with TRH receptors have been monitored by a variety of approaches including translocation of GFP-arrestin (Groarke et al., 1999, 2001; Oakley et al., 1999, 2000; Yu and Hinkle, 1999; Smith et al., 2001; Hanyaloglu et al., 2002), coprecipitation of arrestin and receptor (Jones et al., 2007), effect of arrestin on agonist affinity (Jones and Hinkle, 2005, 2008), and BRET (Kroeger et al., 2001; Hanyaloglu et al., 2002). GFP-arrestin is diffusely localized in the cytoplasm of cells expressing TRH receptors. When TRH is added, GFP-arrestin2 and GFP-arrestin3 move to the plasma membrane within a minute or two (see **Figure 3**). The ability of TRH receptors to recruit arrestins 2 and 3 places it in the Group B category of GPCRs. Although arrestin translocation is detectable with most GPCRs, arrestin movement seen with the TRH receptor is exceptionally robust (Oakley et al., 2000). GFP-arrestin translocation is not observed in cells expressing TRH receptors lacking palmitoylation sites or truncated before the major phosphorylation sites (Vrecl et al., 1998; Yu and Hinkle, 1998; Hanyaloglu et al., 2001; Smith et al., 2001; Jones and Hinkle, 2005, 2008).

As shown in experiments depicted in **Figure 6**, GFP-arrestin does translocate to receptors with Ala substitutions for the four Ser and Thr residues between 355 and 365, proving that downstream phosphosites are sufficient to bind arrestin (Jones and Hinkle, 2008). This is important because distal phosphosites are not sufficient for desensitization and internalization, as discussed below. Arrestin co-precipitates with activated TRH receptors even if Ser/Thr residues in the 355–365 region are Ala-substituted, again showing that arrestin interacts with at least two regions of the receptor. BRET studies document a close interaction between the type 1 TRH receptor and arrestins 2 and 3 that is lost if the receptor is truncated before the palmitoylation sites (Pfleger et al., 2004). **Figure 7** summarizes current understanding about the importance of different TRH receptor phosphorylation sites.

Because arrestins bind preferentially to activated GPCRs, theoretical considerations dictate that arrestin will increase the affinity of a GPCR for agonists, a prediction borne out by experimental evidence (Gurevich et al., 1997). Consistent with this, the apparent affinity of the potent agonist [3H]MeTRH, measured at room temperature or above in intact cells, is highly dependent on arrestin levels (Jones and Hinkle, 2005, 2008). In arrestin2/3 knockout cells, cotransfection with wildtype arrestin increases the apparent affinity for [3H]MeTRH 14-fold. Arrestin increases the agonist affinity of truncated and Ala-substituted receptors, supporting the concept of multiple arrestin binding sites and raising the possibility of phosphorylation-independent arrestin interactions.

#### **ROLE OF ARRESTIN IN TRH RECEPTOR SIGNALING, DESENSITIZATION AND INTERNALIZATION**

Thyrotropin-releasing hormone receptors expressed in arrestin2/3 knockout cells generate strong IP3 responses to TRH. TRHstimulated IP3 levels are much lower in cells expressing arrestin2, arrestin3, or both, proving that arrestin is important for desensitization and that either arrestin is sufficient for this function.

**FIGURE 6 | Arrestin recruitment and arrestin-dependent desensitization.** Cells expressing the receptors shown were incubated with GFP-arrestin and imaged before or after 3 min exposure to TRH. The right panels show TRH-stimulated IP3 production in arrestin-null cells expressing receptor with no arrestin (No Arr), wildtype arrestin 3 (wtArr), or R169E-Arr, which binds

activated GPCRs in a phosphorylation-independent fashion. TRH receptors are: Wt, wildtype; 4Ala, Ala substituted for 4 Ser/Thr residues from amino acids 355 through 365; and 4Ala-371Stop, 4Ala receptor truncated at amino acid 370. Even though arrestin binds robustly to receptors lacking phosphosites at amino acids 355–365, arrestin does not desensitize signaling.

As expected, arrestins are not required for G protein-dependent signaling. Arrestins do not desensitize IP3 responses if receptors are missing phosphorylation sites between amino acids 355 and 365. Ala substitution of Thr365 severely impairs TRH receptor desensitization and internalization. In contrast, receptors truncated at amino acid 370 are still subject to arrestinmediated desensitization (Jones et al., 2007; Jones and Hinkle, 2008).

A similar pattern emerges when internalization is examined. Expression of dominant negative arrestin in normal cells or expression of receptors in arrestin-null cells severely restricts the rate and extent of endocytosis (Hanyaloglu et al., 2002; Jones and Hinkle, 2005, 2008). Endogenous arrestin levels control the desensitization and internalization of heterologously expressed TRH receptors; for example, HEK293 and even more so COS-7 cells have lower concentrations of arrestin and correspondingly less desensitization and internalization than pituitary cells (Falck-Pedersen et al., 1994; Vrecl et al., 1998). Arrestin binding to phosphorylated Thr365 and surrounding sites is absolutely required for internalization, but arrestin binding to phosphosites beyond amino acid 370 is not necessary. The TRH receptor is one of several examples where arrestin can bind to a GPCR without leading to internalization (Krasel et al., 2008). Conversely, expression of the R169E arrestin mutant, which binds activated GPCRs in a phosphorylation-independent fashion, restores TRH-dependent internalization to receptors lacking Thr365 and nearby phosphorylation sites (Hanyaloglu et al., 2001; Jones and Hinkle, 2008; see **Figure 6**). Time- and temperature-dependent conversion of receptor-bound radiolabeled peptide to an acid-resistant state has often been taken as a measure of receptor endocytosis. It is worth noting that this is not valid for receptor-bound [3H]MeTRH (Jones and Hinkle, 2008). Acid resistance precedes internalization and occurs at a reduced level with mutant receptors that do not internalize at all and in cells where internalization is blocked. Endocytosis can be documented readily by microscopy or by the TRH-driven loss of surface binding sites for antibodies to an N-terminal epitope tag on the receptor.

One question arising from these observations is whether internalization of TRH receptors contributes to early desensitization. To address this question, TRH responses were quantified in settings where receptor endocytosis was effectively blocked: in the presence of hyperosmolar sucrose, in cells infected with vaccinia virus encoding dominant negative dynamin, and in cells expressing receptors truncated just before the palmitoylation site (Yu and Hinkle, 1998). Signaling was also measured in arrestin-null cells transfected with a mutant arrestin that can bind to receptor but is incapable of interacting with clathrin and AP2 (∆LIELD/F391Aarrestin) and therefore incapable of promoting internalization (Jones and Hinkle, 2005, 2008). Initial Gq/11-mediated responses and subsequent desensitization of intact receptors were unaffected by the lack of internalization, but the lack of a receptor tail resulted in more persistent elevations in IP3 (Yu and Hinkle, 1998). Together these results confirm the importance of arrestin for both desensitization and internalization and prove that internalization is not the cause of desensitization. In highly sensitive assays, persistent IP1 production can be demonstrated for up to an hour after TRH is washed out in HEK293 cells expressing TRH receptors (Boutin et al., 2012). This sustained response does not depend on internalization. The question of whether there are G proteinindependent signaling pathways and if so, whether they persist following receptor endocytosis, has not been answered. For example, TRH activates MAP kinase. A strong early activation depends on protein kinase C-dependent raf phosphorylation but not on internalization of the receptor itself, although it does require an intact endocytic machinery (Ohmichi et al., 1994; Smith et al., 2001). Additional work is needed to establish whether the weak sustained activation of MAP kinase generated by TRH depends on arrestin or continues following internalization in pituitary cells. The known and possible additional roles of arrestin are shown schematically in **Figure 8**.

#### **MECHANISM OF TRH RECEPTOR INTERNALIZATION**

Thyrotropin-releasing hormone promotes rapid and extensive internalization of endogenous receptors in pituitary cells (Ashworth et al., 1995). Endocytosis of the activated TRH receptor proceeds via a classical arrestin- and dynamin-dependent pathway and is blocked by pharmacological inhibition of endocytosis and by dominant negative dynamin (Drmota et al., 1998; Yu and Hinkle, 1998). Gq/11-dependent signaling is not required, because internalization occurs normally when TRH receptors are expressed in fibroblasts from mice with genetic deletion of both Gα<sup>q</sup> and Gα11, where there is no calcium response (Yu and Hinkle, 1999). Regions of the cytoplasmic receptor tail important for internalization correspond to regions where phosphorylation takes place (Hanyaloglu et al., 2001, 2002; Jones et al., 2007). Endocytosis is much slower although not entirely absent when the TRH receptor is expressed in cells with either no arrestin or dominant negative arrestins (Vrecl et al., 1998; Groarke et al., 2001; Smith et al., 2001; Hanyaloglu et al., 2002; Jones and Hinkle, 2005). The mechanism of the arrestin-independent component is unknown, but TRH receptors appear to be excluded from caveolae (Rudajev et al., 2005).

Ubiquitination is required for endocytosis of several GPCRs (Hanyaloglu and von Zastrow, 2008; Hislop and von Zastrow, 2011) and a fraction of TRH receptors are ubiquitinated, but ubiquitin is not added to receptors on the plasma membrane and ubiquitination does not occur in response to TRH. Furthermore, receptor endocytosis continues at the non-permissive temperature in cells with a temperature-sensitive E1 ubiquitin-activating enzyme (Cook et al., 2003). It is clear, then, that ubiquitination does not tag TRH receptors for internalization, although ubiquitin-mediated degradation serves an important quality control function during TRH receptor biosynthesis.

#### **TRH RECEPTOR TRAFFICKING**

Trafficking of TRH receptors fused at the C-terminus to GFP has been characterized by several groups (Drmota et al., 1998; Yu and Hinkle, 1999; Scott et al., 2002). Shortly after TRH addition, receptors cluster on the cell surface, and over the course of about 10 min they internalize in pre-assembled vesicles without transferrin receptors or Gαq. These earliest vesicles soon merge with others containing transferrin receptors. Some Gα<sup>q</sup> is removed from the plasma membrane following TRH receptor activation, but this occurs quite slowly and internalized receptors and Gα<sup>q</sup> are not colocalized (Drmota et al., 1999; Yu and Hinkle, 1999). After 30 min or longer, GFP-labeled TRH receptors are found in much larger vesicles deep inside the cell. TRH receptors internalize together with arrestin (Groarke et al., 1999; Oakley et al., 2000; Smith et al., 2001; Jones and Hinkle, 2008).

As described above, the TRH receptor follows a familiar pattern of phosphorylation, arrestin binding, desensitization, and endocytosis. Much less is understood about what happens next:

how is the TRH receptor (and other GPCRs) dephosphorylated, sorted and trafficked back to the plasma membrane or targeted for degradation? In a study that capitalized on the availability of antibodies specific for phosphorylated TRH receptors, trafficking of phosphorylated, and dephosphorylated TRH receptors was monitored during internalization and recycling. Endosomal compartments were identified with GFP-labeled Rab proteins, and movements of fluorescently labeled arrestin were followed simultaneously (Jones and Hinkle, 2009). Cell surface receptors were selectively labeled with antibody against an HA tag on the extracellular N-terminus of the rat TRH receptor ("antibody feeding") and tracked over time following addition or withdrawal of TRH. Rab5 marks an early endosomal population, and TRH receptors appeared in Rab5-positive vesicles within a few minutes of TRH stimulation. Receptors in these vesicles were almost entirely phosphorylated and associated with arrestin. Over the course of 20 min, receptors moved to vesicles that were both rab4- and rab5 positive, the early sorting endosomes. Both phosphorylated and non-phosphorylated receptors were seen in this pool. Internalized receptors subsequently trafficked to a population of vesicles that were rab4-positive but rab5-negative, typical of rapidly recycling endosomes. These recycling vesicles were enriched in *dephosphorylated* receptors, i.e., receptors that began the experiment on the plasma membrane but were no longer phosphorylated. They were essentially devoid of phosphorylated receptors.

This result raised the question: were receptors able to move into this rab4-positive, rab5-negative "recycling" vesicle population because they were dephosphorylated, or were receptors quickly dephosphorylated once they reached these vesicles? This question was addressed by interrupting normal trafficking with dominant negative rabs. Dominant negative rab5 completely blocked movement of the receptor out of very early endosomes, yet it did not change the rate of receptor dephosphorylation. If dephosphorylation occurred preferentially in a later endosomal population, dominant negative rab5 would have delayed phosphatase action. These results lead to the conclusion that dephosphorylation takes place in sorting endosomes and permits trafficking of the TRH receptor into recycling vesicles.

A small subset of phosphorylated TRH receptors eventually appear in Rab11 vesicles, traditionally viewed as a late recycling compartment, suggesting that the long recycling pathway is taken by some receptors. Rab7 vesicles, which are associated with lysosomes, contained very little TRH receptor. It seems plausible that a fraction of intracellular receptor is degraded with each round of internalization, possibly contributing to the phenomenon of downregulation, but this remains speculative. The trafficking of internalized TRH receptors is summarized in **Figure 9**.

# **REPOPULATING THE PLASMA MEMBRANE: RECYCLING AND RECRUITMENT**

When cells are incubated with TRH to drive internalization and the hormone is removed, receptors reappear at the plasma membrane with a half-time of 20–30 min based on the amount of radioactive TRH able to bind, the amount of epitope-tagged receptor on the membrane determined by FACS or ELISA, or the amount of GFPlabeled receptor at the surface. These results are consistent with receptor recycling, but the story has recently grown more complicated with the discovery that although the total receptor number at the plasma membrane is restored quite quickly, this is largely due to recruitment of new receptors (Cook and Hinkle, 2004; Jones and Hinkle, 2009). An early piece of evidence appeared when the TRH receptor was fused to a derivative of DS-Red dubbed

**FIGURE 9 | Intracellular trafficking ofTRH receptors.** TRH activation is rapidly followed by receptor phosphorylation, arrestin binding, and recruitment of the arrestin-phospho-receptor complex to clathrin-coated pits that pinch off in a dynamin-dependent process to form clathrin-coated vesicles (CCVs). Vesicles containing phospho-receptor soon merge with rab5-positive vesicles where they colocalize with transferrin receptors. These vesicles gradually merge with others to form larger endosomes rich in rab4 and rab5, early sorting endosomes. After TRH is removed from the medium, but not before, dephosphorylated receptors start to become detectable in rab4-positive but rab5-negative

recycling endosomes. Dephosphorylated receptors then recycle to the plasma membrane. Phosphorylated receptor is rarely seen in these fast recycling vesicles. After long incubations with TRH, small amounts of phospho-receptor are detected in rab11 vesicles, considered to be a slow recycling pool. Protein phosphatase 1 (PP1) acts on the TRH receptor, but it is not known how the removal of extracellular TRH triggers receptor dephosphorylation. The mechanism that permits dephosphorylated receptors to move to recycling endosomes, or alternatively the mechanism that prevents phosphorylated receptors from exiting early sorting vesicles, are also not clear.

"Timer." The unique feature of the Timer protein is that it changes color from red to green with a half-time of 10 h. When TRH was added to drive internalization and then removed to allow recycling, receptors that moved to the plasma membrane were much redder (younger) than those that had been internalized (Cook and Hinkle, 2004). A second approach was an antibody feeding study which showed that intracellular receptors were recruited to the membrane before the internalized receptors had recycled (Jones and Hinkle, 2009). To avoid problems inherent in antibody feeding experiments, TRH receptors were fused at the N-terminus to a 15 amino acid tag that serves as a biotin ligase acceptor. Plasma membrane receptors were selectively labeled with biotin by adding purified bacterial biotin ligase, biotin, and ATP to the medium. TRH receptors on the plasma membrane were essentially the only proteins biotinylated, and biotin-labeled TRH receptor was visualized with fluorescent streptavidin. Again, after TRH-driven endocytosis and several hours of recovery, the surface receptor pool was repopulated initially with recruited rather than recycled (biotin-labeled) receptors (Jones and Hinkle, 2009). Finally, cells were labeled with [3H]TRH (not the higher affinity, slower dissociating [3H]MeTRH), which internalized with the receptor. When [3H]TRH was removed, plasma membrane receptor levels were restored at a time when most [3H]TRH still remained inside the cell. In each of these approaches, most internalized

receptors eventually returned to the plasma membrane, but only after an hour or more (Jones and Hinkle, 2009). These results provide another question for future study: How are receptors in an intracellular pool prompted to move to the plasma membrane?

#### **GPCR DEPHOSPHORYLATION**

Once a GPCR is activated, phosphorylated, and bound to arrestin, return to its resting state requires dissociation or degradation of the agonist, dissociation of arrestin and dephosphorylation. With rare exceptions, the phosphatases responsible for GPCR dephosphorylation have not been fully characterized. There are many more GPCRs than Ser/Thr phosphatases, necessitating involvement of enzymes with broad substrate specificity. In mammalian cells, the majority of proteins phosphorylated on Ser and Thr residues are dephosphorylated by members of two ubiquitous enzyme families, protein phosphatase 1 (PP1), and protein phosphatase 2A (PP2A; Barford et al., 1998). These protein phosphatases are multisubunit enzymes that achieve specificity by interacting with a wide array of scaffolding and targeting subunits. The catalytic subunits of PP1s can bind directly to substrates through a number of conserved motifs or localize through targeting subunits. Over 150 proteins have been shown to interact with PP1 (Bollen et al., 2010). PP2As are heterotrimers composed of a catalytic subunit, a scaffolding or structural subunit, and a targeting subunit. On the basis of high

sensitivity to inhibitors, the PP2A family of phosphatases has been implicated in the dephosphorylation of numerous GPCRs, and PP1, calcineurin (PP2B), and PP2C for others (Croci et al., 2003; Flajolet et al., 2003; Mao et al., 2005).

Pitcher et al. (1995) reported that the β2-adrenergic receptor was dephosphorylated by an endosomal phosphatase in the PP2A family following internalization . This model predicted an essential role for receptor endocytosis: a GPCR had to cycle through endosomal compartments to be dephosphorylated and resensitized. Subsequent work has uncovered many variants on this theme. Many GPCRs can be dephosphorylated while localized on the cell surface. Dephosphorylation of PKA and GRK sites on the β2-adrenergic receptor can take place on the plasma membrane (Iyer et al., 2006). Some sites on the somatostatin 2A receptor are not dephosphorylated until the receptor has internalized, while others can be dephosphorylated regardless of receptor location; different enzymes appear to be involved (Ghosh and Schonbrunn, 2011). Receptors for several peptide hormones cannot recycle until the peptide is degraded in acidified endosomes in a reaction catalyzed by ectopeptidases that cointernalize with the peptidereceptor complex (Roosterman et al., 2008;Cattaruzza et al., 2009). Although the TRH-degrading enzyme is an ectoenzyme present in pituitary tissue, there is no evidence that it plays a role in TRH receptor cycling.

# **TRH RECEPTOR DEPHOSPHORYLATION**

When TRH is removed from the medium or an inverse agonist such as chlordiazepoxide is added, receptor dephosphorylation rapidly ensues (Jones and Hinkle, 2005; Jones et al., 2007; Gehret and Hinkle, 2010, 2012). Once TRH dissociates, receptors return to an inactive conformation, arrestin affinity declines, and arrestin dissociation provides freer access of phosphatases. A number of studies have sought to identify factors controlling the rate of receptor dephosphorylation. The amino acid sequence surrounding the phosphosite does not seem to be a major factor because the rates of TRH receptor dephosphorylation are the same at different phosphosites (Jones et al., 2007). When the β2-adrenergic receptor tail was spliced onto the TRH receptor, the chimeric receptor underwent TRH-dependent phosphorylation at sites that are phosphorylated by GRKs in the β2-adrenergic receptor. The dephosphorylation kinetics of the chimera resembled those of the natural TRH receptor (Gehret and Hinkle, 2010). On the other hand, the location of the TRH receptors is important (Jones et al., 2007; Gehret and Hinkle, 2010). Dephosphorylation occurs more rapidly when receptors are on the plasma membrane than when receptors have undergone endocytosis. In pituitary cells, endogenous receptors are dephosphorylated with a half-time of about 45 s if TRH exposure lasts only a minute (sufficient for complete phosphorylation but not for internalization) but is 3 min if TRH exposure continues for 30 min (when receptors have undergone internalization; Jones et al., 2007). A similar situation holds in heterologous model systems, although dephosphorylation is not nearly as fast.

These data raise an important question: how does removing TRH from the outside of the cell trigger dephosphorylation of receptors in endosomes deep in the cytoplasm? The simplest idea is that some receptors are always at the plasma membrane and as long as TRH is present, they are signaling and somehow maintaining receptor phosphorylation. Removing extracellular TRH terminates signaling and leads to dephosphorylation. If this model is valid, however, the responsible signaling system cannot be traditional Gq/11-mediated activation of PLCβ, because dephosphorylation of internalized receptors is not altered by treatment with an intracellular calcium chelator and protein kinase C inhibitor (Jones et al., 2007).

Recently, the problem of identifying the TRH receptor phosphatase was tackled using an unbiased screen with an siRNA library directed against phosphatase subunits (Gehret and Hinkle, 2012). While this approach had the potential to identify a targeting or scaffolding subunit, only the catalytic subunit of PP1α came up as a bone fide hit in the screen. The results of the screen, which was performed in HEK293 cells, concur with inhibitor effects on dephosphorylation of endogenous TRH receptors in pituitary cells. Dephosphorylation of TRH receptors is powerfully inhibited by calyculin A, which acts on the catalytic subunits of PP1 and PP2A, but insensitive to fostriecin, a highly selective inhibitor of PP2A. Many details, such as the mechanism that targets PP1α to the receptor, remain to be elucidated. The discovery of a specific phosphatase subunit acting on the TRH receptor provides a starting point for dissecting an important aspect of the resensitization process.

# **FUTURE DIRECTIONS**

The TRH receptor has been viewed as a prototypical "calciummobilizing" GPCR and the mechanisms of TRH signaling have been the subject of intense investigation for decades. More recently, attention has also been focused on events that turn signaling off. In every instance, the pathways involved have turned out to be more complex than anticipated. This review has sought to review recent work and highlight some of the unknowns pertaining to TRH receptor signaling, desensitization, and trafficking. Once activated by hormone, the TRH receptor is phosphorylated in multiple regions; arrestins bind at different phosphosites with different consequences. It seems likely that the TRH receptor signals via its diverse arrestin interactions in ways that are not yet recognized. The trafficking studies described here show that dephosphorylation must take place before internalized TRH receptors can recycle, yet the nature of the interaction between receptor and phosphatase remains elusive. The recent identification of the protein phosphatase involved opens the door for future studies seeking to describe phosphatase targeting mechanisms. The principles governing receptor trafficking are not well understood, and it is unclear how phosphorylated and desphosphorylated receptors are sorted in endosomes. High throughput screening techniques have aided in the discovery of biased ligands with the capacity to control specific aspects of receptor function, and it is to be hoped that designer ligands for TRH receptors will provide new tools for research or even novel drugs for treating thyroid disease. Finally, long awaited information on the structure of the TRH receptor can be expected in the not too distant future.

# **ACKNOWLEDGMENTS**

Research in the author's laboratory was supported by NIH Grant DK19974.

# **REFERENCES**


with beta-arrestin 1-green fluorescent protein. *Mol. Pharmacol.* 59, 375–385.


lactotrophs and GH cells. *Trends Endocrinol. Metab.* 7, 370–374.


resensitization. *J. Biol. Chem.* 274, 32248–32257.


receptor affects receptor trafficking but not signaling. *Mol. Endocrinol.* 19, 2859–2870.


hypothalamic-releasing hormones: a cellular basis for paradoxical secretion. *Proc. Natl. Acad. Sci. U.S.A.* 94, 14132–14137.


gene. *Proc. Natl. Acad. Sci. U.S.A.* 94, 10862–10867.


phosphorylation of thyrotropinreleasing hormone receptors are modulated by agonist stimulation. *J. Biol. Chem.* 277, 28228–28237.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 02 October 2012; paper pending published: 02 November 2012; accepted: 26 November 2012; published online: 13 December 2012.*

*Citation: Hinkle PM, Gehret AU and Jones BW (2012) Desensitization, trafficking, and resensitization of the pituitary thyrotropin-releasing hormone receptor. Front. Neurosci. 6:180. doi: 10.3389/fnins.2012.00180*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Neuroscience.*

*Copyright © 2012 Hinkle, Gehret and Jones. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

REVIEW ARTICLE published: 18 September 2013 doi: 10.3389/fendo.2013.00128

# Relaxin-3/RXFP3 signaling and neuroendocrine function – a perspective on extrinsic hypothalamic control

# **Despina E. Ganella1,2, Sherie Ma<sup>1</sup> and Andrew L. Gundlach1,3,4\***

<sup>1</sup> The Florey Institute of Neuroscience and Mental Health, The University of Melbourne, Melbourne, VIC, Australia

<sup>2</sup> Department of Biochemistry and Molecular Biology, The University of Melbourne, Melbourne, VIC, Australia

<sup>3</sup> Florey Department of Neuroscience and Mental Health, The University of Melbourne, Melbourne, VIC, Australia

<sup>4</sup> Department of Anatomy and Neuroscience, The University of Melbourne, Melbourne, VIC, Australia

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Michiru Hirasawa, Memorial University, Canada Barbara McGowan, Guys and St. Thomas Hospital, UK

#### **\*Correspondence:**

Andrew L. Gundlach, The Florey Institute of Neuroscience and Mental Health, 30 Royal Parade, Parkville, VIC 3052, Australia e-mail: andrew.gundlach@florey. edu.au

Complex neural circuits within the hypothalamus that govern essential autonomic processes and associated behaviors signal using amino acid and monoamine transmitters and a variety of neuropeptide (hormone) modulators, often via G-protein coupled receptors (GPCRs) and associated cellular pathways. Relaxin-3 is a recently identified neuropeptide that is highly conserved throughout evolution. Neurons expressing relaxin-3 are located in the brainstem, but broadly innervate the entire limbic system including the hypothalamus. Extensive anatomical data in rodents and non-human primate, and recent regulatory and functional data, suggest relaxin-3 signaling via its cognate GPCR, RXFP3, has a broad range of effects on neuroendocrine function associated with stress responses, feeding and metabolism, motivation and reward, and possibly sexual behavior and reproduction. Therefore, this article aims to highlight the growing appreciation of the relaxin-3/RXFP3 system as an important "extrinsic" regulator of the neuroendocrine axis by reviewing its neuroanatomy and its putative roles in arousal-, stress-, and feeding-related behaviors and links to associated neural substrates and signaling networks. Current evidence identifies RXFP3 as a potential therapeutic target for treatment of neuroendocrine disorders and related behavioral dysfunction.

**Keywords: relaxin-3, oxytocin, arginine vasopressin, CRH, feeding, metabolism, stress, reproduction**

#### **INTRODUCTION**

Precise regulation of complex neural circuits in the hypothalamus governs essential autonomic processes *and* associated behaviors, such as metabolism, growth, and feeding; stress responses, arousal, and locomotor activity; as well as reproduction and social/sexual behavior (1–7). These intrinsic and often interacting neural circuits utilize various neuroendocrine peptides/hormones, such as thyrotropin-releasing hormone (TRH), growth hormone-releasing hormone (GHRH), somatostatin, orexins, melanin-concentrating hormone (MCH), agouti-related peptide (AgRP), pro-opiomelanocortin (POMC) gene products [alpha-melanocyte-stimulating hormone (α-MSH)], neuropeptide Y (NPY), corticotropin-releasing hormone (CRH) and urocortins, gonadotropin-releasing hormone (GnRH), arginine vasopressin (AVP), and oxytocin (8–16). The majority of these peptides and hormones signal via G-protein coupled receptors (GPCRs) and often multiple receptors exist for different members of a peptide family or for the same peptide modulator [e.g., Ref. (17–19)].

This combination of a large number of ligands and multiple receptors results in a vast diversity in the potential regulation of different populations of hypothalamic neurons. For example, a recent survey revealed more than 300 different GPCRs are expressed by the heterogeneous neurons in the paraventricular (PVN) and supraoptic nuclei (SON) alone (20). This diversity of potential functional regulation provides a challenge for neuroscientists and neuroendocrinologists to document the anatomical distribution and dissect the primary and integrative actions of different signaling systems, both within hypothalamic circuits and via their descending and ascending inputs. Importantly, modern experimental approaches including conventional and viral-based tract-tracing (21) and other viral-based methods, such as optogenetics and DREADD technology (22–25), combined with molecular genetics and complementary methods for measuring changes in physiology and behavior, are successfully dissecting the role of individual neuron populations and the key mediators involved. In turn, this is allowing a reappraisal of the "dogma" related to the function of several established neural transmitter and hormone networks in the hypothalamus and the integration of new "chemical players" into the existing circuitry.

Just over a decade ago, the final member of the relaxin and insulin-like peptide superfamily was discovered and named H3 relaxin (human) or relaxin-3 (rodents), in line with the prior discovery and characterization of two other relaxin genes in humans (26). However, unlike its related peptide, H2 relaxin or relaxin, which is widely distributed within the brain *and* peripheral tissues [see Ref. (27, 28) for review], relaxin-3 was found to be most highly expressed in brain (26, 29). In 2003, GPCR135 (now known as RXFP3) was identified as the cognate relaxin-3 receptor (30, 31) and was shown to be highly localized in various rat brain areas (30, 32), which were later confirmed to contain relaxin-3-positive

axonal projections and terminations (33). A similar central distribution of relaxin-3 neurons and projections to that reported in the rat was subsequently observed in the mouse [Ref. (34); Allen Brain Atlas<sup>1</sup> ] and macaque brain (35, 36), suggesting that this neuropeptide system has been highly conserved throughout evolution. Indeed, bioinformatic studies revealed that a relaxin-3-like ancestral peptide gave rise to the relaxin and insulin-like peptide superfamily and its sequence has been highly conserved by strong purifying selection, consistent with a highly conserved function in the central nervous system (37, 38).

After their discovery, characterization of the neuroanatomical distribution of relaxin-3- and RXFP3-expressing neurons provided insights into putative functions of relaxin-3; and a growing number of experimental studies have subsequently confirmed roles for relaxin-3/RXFP3 signaling in arousal, feeding, stress responses, and cognition [see Ref. (39) for review]. Several of these actions of relaxin-3 likely involve effects on RXFP3-positive hypothalamic neuron populations. Therefore, in this article we will provide a summary of the hypothalamic distribution of RXFP3

<sup>1</sup>www.brain-map.org

mRNA and protein, and relaxin-3 projections, a concise review of experimental data indicating that this neuropeptide/receptor system is a modulator of hypothalamic function, and a perspective on the future studies required to better understand this system and to exploit its therapeutic potential.

# **NEUROANATOMY OF THE RELAXIN-3/RXFP3 SYSTEM**

Relaxin-3 is a 5 kDa peptide that shares common structural features with all relaxin and insulin-like peptide family members – an A- and B-chain held together by three disulfide bonds (26, 40, 41). The native peptide is synthesized as a pre-prohormone that is subsequently cleaved by proteolytic processing of the signal and C-peptides to form the mature peptide [see e.g., Ref. (42, 43)]. Early *in situ* hybridization studies revealed that relaxin-3 mRNA was highly expressed by a cluster of neurons in the rat pontine central gray, identified as the *ventromedial dorsal tegmental area* (29), more commonly known as the *nucleus incertus* [NI; (44, 45)]. Smaller dispersed populations were also identified in the medial periaqueductal gray (PAG), pontine raphe (PR), and a region dorsal of the substantia nigra (dSN) in the rat (33, 46), mouse (34), and macaque (36) (**Figure 1**).

and RXFP3 binding sites (green) are illustrated [see (32–34, 46, 62) and (28)]. Amy, amygdala; Arc, arcuate nucleus; BNST, bed nucleus of the stria terminalis; CM, centromedial thalamic nucleus; CPu, caudate putamen; Cx, cerebral cortex; DBB, diagonal band of Broca; DG, dentate gyrus; DMH, dorsomedial nucleus of hypothalamus; DR, dorsal raphe nucleus; DTg, dorsal tegmental nucleus; Hb, habenula; Hi, hippocampus; IC, inferior

paraventricular thalamic area; PVN, paraventricular nucleus of hypothalamus; SC, superior colliculus; SFO, subfornical organ; SON, supraoptic nucleus; SN, substantia nigra; SuM, supramammillary nucleus; Th, thalamus; VP, ventral pallidum. Adapted from a figure kindly provided by Dr. Craig Smith (The Florey Institute of Neuroscience and Mental Health, Melbourne, Australia).

Ultrastructural analysis of relaxin-3 immunoreactivity in the rat NI identified the peptide in the rough endoplasmic reticulum and Golgi apparatus in the cell soma and within dense-core vesicles adjacent to synapses in nerve terminals of distant target regions such as the lateral hypothalamus (46) and medial septum (47), indicating that relaxin-3 is processed and released as a transmitter.

The efferent and afferent connections of the rat NI have been characterized (44, 45, 48–50) and many NI projection target regions, including the hypothalamus, contain relaxin-3 immunoreactive fibers and terminals, and neurons expressing RXFP3 (33, 50) (**Figure 1**; **Table 1**), suggesting many of these areas are innervated by NI relaxin-3 neurons. Not surprisingly, NI relaxin-3 neurons have been the focus of the majority of functional studies to date, which indicate they are highly responsive to neurogenic stressors and CRH (46, 51–53).

Less is known about the connections, regulation and function of other relaxin-3 neuron populations, but a recent study demonstrated that PAG relaxin-3 neurons strongly innervate and modulate neuronal activity in the intergeniculate leaflet [IGL; (46, 54, 55)], a region that is known to contribute to the regulation of arousal and circadian activity (56–58). In brain slice studies, patch-clamp recordings of IGL neurons, revealed that activation of RXFP3 by bath application of the agonist peptide, R3/I5, produced depolarization of identified NPY-containing neurons (55), which are known to project to the suprachiasmatic nucleus via the geniculohypothalamic tract (56, 58). Neurons in the IGL also project to a number of other hypothalamic areas, including

the anterior and lateral hypothalamic areas, and the dorsomedial nucleus (58).

There are, however, several brain regions like the aforementioned IGL and the amygdala that contain dense relaxin-3 immunoreactivity and/or RXFP3, but sparse NI projections; suggesting they are also more strongly innervated by other relaxin-3 populations. In fact, there is anatomical evidence, chiefly from neural tract-tracing studies, to suggest the various RXFP3-positive regions in the hypothalamus are also innervated by relaxin-3 neurons in the NI *and* other relaxin-3 groups. For example, a recent study of brainstem inputs to the PVN and surrounding area in the rat (59) revealed projections from the medial PAG, PR, and "dorsal to substantia nigra" regions, which contain relaxin-3 neurons.

RXFP3 has been localized in various subregions of the hypothalamus in the rat (32, 33) mouse (34, 60) (**Table 1**), and macaque (36). The highest densities are present in the PVN, SON and adjacent medial (MPO) and lateral preoptic (LPO) nuclei, but there are RXFP3-positive neurons in other hypothalamic areas, including the periventricular nucleus (33, 34) (**Table 1**), which is consistent with a putative role of relaxin-3/RXFP3 signaling in the control of a range of homeostatic and autonomic behaviors via modulation of related hypothalamic networks.

# **RELAXIN-3 RECEPTOR BINDING AND ACTIVATION IN BRAIN**

Effects of endogenous relaxin-3 are predicted to be mediated by its cognate receptor, RXFP3, but relaxin-3 is also an agonist at the relaxin-family receptors, RXFP1 and RXFP4, when administered at pharmacological doses, albeit with lower potency than at RXFP3


**Table 1 | Comparative distribution of relaxin-3 and its receptor, RXFP3, in hypothalamic regions of rat and mouse brain.**

Relative abundance values are given: − not detectable, + low density, ++ moderate density, +++ high density, n.r. not reported). Adapted from Ma et al. (33), Smith et al. (34), Sutton et al. (32), Tanaka et al. (46), Allen Brain Institute Brain Atlas (http:// mouse.brain-map.org/ experiment/ show/ 71358555). RLN3-LI, relaxin-3 like immunoreactivity.

(30, 61). RXFP1 and RXFP4 are likely expressed in the human brain, along with RXFP3, but based on animal studies these receptors and the peptides which bind and activate them (relaxin, insulin-like peptide 5, and relaxin-3, respectively) are expressed in quite distinct regions of the brain and at very different levels [e.g., Ref. (26, 29, 30, 33, 46, 62)], so a current working hypothesis that can be tested is that these receptors mediate distinct functional effects in the brain, which modulate different homeostatic processes and behaviors.

Cell signaling events associated with the relaxin-family receptors have been studied in different cell lines transfected with the human receptors (28, 63) and activation of RXFP1 and RXFP3/4 by H3 relaxin leads to different intracellular responses *in vitro*. In Chinese hamster ovary (CHO) cells, RXFP3 and RXFP4 couple to the inhibitory Gαi/Gαo-protein system and receptor activation leads to sequestration of these G-proteins and inhibition of adenylate cyclase (AC), and subsequent cAMP accumulation (30, 64). The intracellular signaling pathway of the relaxin-3-RXFP3 interaction has also been studied in the SN56 neuronal-like cell line, in which the Gαi/Gα<sup>o</sup> pathway was recruited, suggesting an inhibitory intracellular pathway may be activated *in vivo* when relaxin-3 binds to RXFP3-expressing neurons within the brain (63), an idea that can be tested in different functional networks [see Ref. (54, 55)]. In contrast, RXFP1 activation by either relaxin or relaxin-3 in mammalian cell expression systems initiates a downstream accumulation of cAMP, as it is predominantly coupled to the stimulatory Gs-protein. These *in vitro* studies suggest activation of different receptors may lead to different downstream effects in neurons *in vivo*. Little is known, however, about the native intracellular signaling of any of the relaxin family of receptors (RXFP1–RXFP4) in specific neuronal populations, apart from a recent *in vitro* study that revealed activation of RXFP3 by the agonist peptide, R3/I5 (65), produced depolarization of identified NPY-containing neurons and hyperpolarization of adjacent non-NPY neurons (55).

Although there is no definitive evidence that major biological effects mediated by relaxin-3 are caused by activation of either RXFP1 and/or RXFP4, the ability of relaxin-3 to activate RXFP1 and RXFP4 as well as RXFP3 must be considered as a confounding factor when using pharmacological doses of peptides *in vivo*, in attempts to study neuropeptide function in the rodent. From a practical viewpoint, the rat is suited to studies of the neurobiology of relaxin-3/RXFP3 signaling, since RXFP4 is a pseudogene in this species, and so not a "confound." However, *in situ* hybridization and radioligand binding site studies indicate RXFP1 is expressed in the rat brain in a number of regions positive for RXFP3, including the cerebral cortex, amygdala, thalamus, and hypothalamus (32, 33, 62). Consequently RXFP1 activity must be considered in studies of exogenous relaxin-3 peptide administration [e.g., Ref. (40, 66–70)]. The relative degree to which RXFP1 activation has impacted outcomes in studies of relaxin-3 actions within the hypothalamus and other brain areas is hard to gage, although in some cases comparative effects of relaxin were reported [see Ref. (71) below]. Fortunately, the development more recently of agonist and antagonist peptides which selectively activate or inhibit RXFP3 [e.g., Ref. (65, 72–74)] has facilitated investigations of the behavioral and physiological effects of specific relaxin-3/RXFP3

interactions. Indeed, several studies have used these peptides to assess relaxin-3/RXFP3 related functions associated with the hypothalamus, although their use is not as widespread as it might be [e.g., Ref. (69, 70)], particularly in light of the availability of a chemically less-complex, single-chain peptide antagonist for RXFP3 (73).

# **ACTIONS OF THE RELAXIN-3/RXFP3 SYSTEM – FOCUS ON HYPOTHALAMUS**

#### **FEEDING AND ENERGY BALANCE**

The PVN and arcuate nucleus (ARC) are two hypothalamic nuclei which tightly regulate food intake and energy homeostasis [e.g., Ref. (3, 4, 12, 15, 75)]. Relaxin-3 immunoreactivity and RXFP3 mRNA and binding sites have been identified in the PVN and ARC (30, 32, 33) (**Table 1**) and extensive research has demonstrated that H3 relaxin can alter feeding and appetite in rats [see Ref. (76)]. Central (icv) administration of H3 relaxin caused hyperphagia in satiated male Wistar rats in the first hour after treatment in both the light and the dark phase, while equivalent doses of H2 relaxin did not produce a similar effect (66). The doses of H3 relaxin administered icv (180 pmol) or directly into the PVN (18 pmol) at which a significant increase in feeding was observed (see below) were similar to doses of other "feeding" peptides which elicit an increased feeding response upon similar administration [e.g., NPY ∼80 pmol iPVN; (77)], consistent with a physiological role for relaxin-3/RXFP3 signaling in modulating appetite and feeding behavior. Importantly, icv administration of the selective agonist, R3/I5, produced an increase in first-hour food intake and this effect was inhibited by co-administration of the selective antagonist, ∆R3/I5 (72), further suggesting RXFP3 involvement. In these feeding studies, the food intake was monitored for only 24 h postinjection, as it was predicted the injected peptide would be quite rapidly degraded by proteolysis. In a study of longer term effects of relaxin-3 on feeding and body weight, H3 relaxin (600 pmol/day) was infused icv into rats for 14 days via osmotic mini-pump, which produced a significant increase in food intake and body weight gain, and plasma leptin and insulin levels, compared to vehicle (78). This data indicated that the relaxin-3 induced increase in food intake could be sustained and lead to an increase in body weight and associated biochemical changes in the rat.

Given the primary role of the hypothalamus in energy balance, the effect on feeding of local hypothalamic injections of relaxin-3 was assessed. Acute H3 relaxin injection into the PVN (iPVN) increased food intake over the first hour (67). Sub-chronic iPVN H3 relaxin administration in *ad libitum* fed rats also produced an increase in food intake and cumulative body weight gain *c.f.* vehicle (67). These authors also used "Fos-activation" mapping after icv administration of H3 relaxin to identify activated hypothalamic nuclei. In addition to the PVN, the SON, ARC, and anterior preoptic area displayed increased Fos staining (79). When injected directly into these and other hypothalamic nuclei H3 relaxin stimulated a significant increase in food intake within the first hour, relative to control (79), suggesting multiple hypothalamic nuclei may mediate these potent orexigenic effects. Unfortunately the potential involvement of RXFP1 activation in the observed effects cannot be reliably excluded, as RXFP3-selective peptides were not utilized. It is also possible the injected peptide is able to diffuse

from the injection site to adjacent areas that are primarily involved in the activation of feeding, although some targeted sites were not associated with feeding (79).

In order to circumvent issues associated with acute peptide administration and cross-reactivity of H3 relaxin at RXFP1, we used a viral strategy to investigate the effect of chronic R3/I5 mediated RXFP3 activation within the hypothalamic PVN (43). Using a recombinant adeno-associated virus (rAAV) engineered to locally secrete bioactive R3/I5 (rAAV-R3/I5), we demonstrated an increase in food intake in the R3/I5 expressing rats (∼5.2 g/day more than control) which was sustained for up to 2 months, leading to an ∼23% increase in cumulative body weight gain (43). In an attempt to identify targets of RXFP3 activation within the hypothalamus, levels of mRNA for a number of genes were also assessed in dissected hypothalamic tissue blocks using quantitative reverse transcription PCR. Notably, no major differences were identified in expression of some "major" feeding peptide genes (NPY, AgRP, POMC) between rAAV-R3/I5 and rAAV-control treated groups, whereas the levels of oxytocin and AVP mRNA were altered. This is, however, consistent with the strong expression of RXFP3 by neurons in the PVN and SON (30, 32, 33) (**Table 1**), which contain oxytocin- and AVP-containing magnocellular neurons. Chronic viral-mediated expression of R3/I5 in the hypothalamus for up to ∼14 weeks led to a marked reduction in oxytocin mRNA expression in PVN (∼50%) and a smaller reduction in AVP (∼25%), compared to control expression (43). These data suggest the acute orexigenic effect of RXFP3 activation may be mediated by changes in oxytocin release (and perhaps AVP) [see Ref. (80)], an idea that can be tested experimentally. Furthermore, there is a substantial literature relating to oxytocin as a satiety factor that also supports this putative acute and chronic mechanism of action.

#### **INTERACTION WITH OXYTOCIN AND ARGININE VASOPRESSIN SYSTEMS IN THE PVN**

Oxytocin is a peptide hormone highly expressed in magnocellular and some parvocellular neurons of the hypothalamic PVN and SON (81–83). Oxytocin is classically known as a reproductive hormone with roles in parturition, lactation, sexual behavior, and pair bonding and attachment [see Ref. (84, 85) for review]. However, early studies revealed that central administration of oxytocin dose-dependently reduced food consumption and time spent eating and increased the latency to the first meal in pre-fasted rats; an effect prevented by co-administration of an oxytocin receptor antagonist (86). A number of studies have since confirmed central oxytocin administration inhibits food intake, strengthening the hypothesis that oxytocin serves a key role in appetite control (87–89). More recent data is consistent with this view, as oxytocin null mutation mice display enhanced intake of sweet and non-sweet carbohydrate solutions (90, 91) and develop late-onset obesity (92).

The magnitude of the increase in feeding and body weight gain observed in our hypothalamic R3/I5 expression studies (43) was modest relative to those seen after similar viral-mediated NPY and AgRP over-expression, consistent with actions independent of direct effects on these neurons. The down-regulation of oxytocin and AVP mRNA expression observed (43), suggests the R3/I5 agonist peptide is activating RXFP3 on oxytocin- and AVP-containing neurons, which results in downstream effects on the transcription of oxytocin and AVP mRNA in these neurons (e.g., Ref. (93, 94)] and subsequent effects on production and release of this anorexigenic hormone. If RXFP3 associated cell signaling in these neurons is similar to effects reported *in vitro*, this could be associated with reduction in cellular cAMP levels (via inhibitory Gi-protein coupling) and inhibition or hyperpolarization of target neurons – a possibility that can be assessed experimentally using *in vitro* or *in vivo* electrophysiological and biochemical methods [e.g., Ref. (55, 95)].

It is also necessary to establish whether it is the proposed reduction in oxytocin production and release which produces the observed increase in food intake in AAV-R3/I5-treated rats, or whether the chronic increase in food intake caused by hypothalamic RXFP3 activation over time leads to down-regulation of oxytocin mRNA via a distinct mechanism. This could be explored by conducting acute peptide administration studies or shorter time course viral infusion studies, and assessing oxytocin mRNA, peptide and release levels, before any marked changes in body weight have occurred.

Indeed, a recent study examined the effect of icv H3 relaxin on anxiety-like behavior in rats and observed an anxiolytic effect in the elevated plus maze test and the shock probe-burying test (70), consistent with our studies demonstrating that icv administration of a selective RXFP3 agonist peptide reduced anxiety-like behavior in the light-dark box and elevated plus maze (96). Notably, these authors used microarray and peptidomics approaches to identify associated downstream signaling targets in the hypothalamus altered by icv H3 relaxin and detected a relatively acute (6–24 h) and quite specific *increase* in the level of hypothalamic oxytocin mRNA and peptide levels (70). This data is, however, more consistent with the ability of H3 relaxin to *activate* neurons via RXFP1 [see e.g., Ref. (40, 67, 68)] rather than via RXFP3, which based on predicted signaling (28, 63) might be expected to decrease oxytocin neuron activity (and mRNA and peptide levels), as seen in our study (43). Therefore, further studies are required to identify specific effects of RXFP3 (and RXFP1) activation in hypothalamus and other areas in altered feeding and metabolism and anxiety-like behavior (76, 96).

Further evidence for an association of a down-regulation/ inhibition of PVN oxytocin activity with a hyperphagic phenotype was reported recently. Optogenetic electrophysiological studies revealed that GABA/AgRP neurons in the ARC project to a population of oxytocin neurons in the PVN and strongly inhibit their activity, and this suppression of oxytocin neurons by AgRP neuronal activation drives evoked feeding (23). This study also demonstrated that increased food seeking and consumption in response to GABA/AgRP neuron activation is similarly induced by suppressing the activity of the PVN neurons that (selectively) express the *single-minded 1* (SIM1) transcription factor (23, 97).

Importantly from a clinical perspective, a number of disease states in which hyperphagia is a symptom are associated with hypothalamic oxytocin dysregulation. A small population of PVN oxytocin neurons is selectively lost in Prader–Willi syndrome, which is a condition involving insatiable hunger (98); and disruption of synaptic release of hypothalamic oxytocin results in overeating (99). Oxytocin deficits in SIM1 haploinsufficient mice and mutations in the SIM1 gene in humans lead to hyperphagic obesity (97, 100, 101). In the mouse model, an ∼80% reduction in oxytocin and ∼30% reduction in AVP was observed (97). These studies illustrate that modulation of relaxin-3 signaling and associated alterations in oxytocin neuron activity may be a fruitful area to explore for treating disease states associated with eating disorders. For example, a recent clinical cross-sectional study reported that female patients with anorexia nervosa, characterized by food restriction, low weight, and hypoleptinemia, had higher mean circulating oxytocin levels than healthy controls at all times assessed (102). The severity of disease psychopathology was also positively associated with circulating oxytocin levels (102).

#### **HYPOTHALAMIC-PITUITARY-ADRENAL AXIS AND STRESS RESPONSES**

Integration of the stress response via the hypothalamic-pituitaryadrenal (HPA) axis occurs via interaction between brain areas which are sensitive to stress and neuroendocrine neurons of the hypothalamic PVN, particularly those in the parvocellular region producing CRH [see Ref. (1, 6, 103) for review]. CRH stimulates the secretion and synthesis of adrenocorticotropin hormone (ACTH) from the pituitary and is the main regulator of HPA axis activity during stress.

Early regulatory studies revealed that neurons in the NI and specifically relaxin-3 expressing neurons respond to stress and CRH (46, 104), and that relaxin-3-containing neurons in the NI express CRH type 1 receptors (CRH-R1) (46). Upon icv administration of CRH, 65% of relaxin-3-positive neurons underwent activation (detected using Fos-immunostaining 2 h post-infusion) (46). Electrophysiological characterization of NI neurons revealed that a significant population increased firing following icv administration of CRH, of which the majority were relaxin-3-positive, though an almost equal number of non-relaxin-3 neurons exhibited a decrease in firing in the anesthetized rat (53). These findings are consistent with semi-quantitative immunohistochemical studies of the NI revealing this heterogeneous neuron population consists of relaxin-3 neurons that all co-express CRH-R1, though not all CRH-R1 contain relaxin-3, in addition to a significant population that are CRH-R1 negative (53). Rats tested in a water immersion-restraint stress paradigm, displayed increased Fos-immunostaining and an up-regulation of relaxin-3 mRNA in NI neurons (46). Subsequently, the effect of a repeat forced swim on relaxin-3 expression in the NI was examined and led to a rapid increase in relaxin-3 mRNA expression (51). This increase was largely mediated by CRH activation of CRH-R1 located on NI neurons, as pre-treatment with the CRH antagonist, antalarmin, reduced the increase in relaxin-3 mRNA expression by 70–80%. Levels of relaxin-3 heteronuclear (hn) RNA were also increased in NI neurons after the repeat forced swim (51). Changes in hnRNA are a measure of transcriptional activity and are thought to reflect the level of encoded peptide synthesis (105), suggesting in this case, an increase in relaxin-3 utilization.

Initial insights into the hypothalamic action of relaxin-3 in relation to the stress response have been obtained by monitoring the effect of icv administration of relaxin-3 in male rats (69). Increased Fos staining in the PVN and SON was observed 1 h post-administration and levels of c-*fos* and CRH mRNA in the PVN were also increased 2 h after H3 relaxin administration. Central H3 relaxin administration also elicited an increase in circulating plasma ACTH (69). These data *suggest* a role for relaxin-3 in the acute hypothalamus-pituitary CRH-ACTH system response, but these studies did not directly identify the presence of RXFP3 on CRH neurons or measure the direct acute excitatory or inhibitory effect of RXFP3 activation on these neurons [see Ref. (55)].

Notably, a recent report also suggests sex-specific regulation of CRH and relaxin-3 systems in response to combined stressors. Chronically stressed and repeatedly food-restricted female rats consumed more standard chow during recovery and had an increased bodyweight relative to controls, whereas male rats exposed to this regime had a reduction in bodyweight (106). Stressed/food-restricted female rats had elevated plasma corticosterone and low PVN CRH mRNA levels. CRH neurons in the medial preoptic area were identified as a source of increased CRH production/release during stress in female brain, i.e., CRH mRNA levels in this area were – higher in female than male rats, increased by chronic stress, and increased in female, not male, rats after repeated food restriction (106).

Further studies are now required to determine precisely how the robust, consistently observed stress and CRH-induced activation of NI and relaxin-3 neurons (46, 51, 53, 106) directly or indirectly activates (or possibly inhibits) the PVN and the main components of the HPA axis; and whether and how these processes are regulated by steroid and other hormones under different conditions.

#### **EFFECTS ON REPRODUCTIVE NEUROENDOCRINE SYSTEMS**

Preliminary studies have indicated a putative role for relaxin-3 in reproductive physiology. Following injection into the PVN and surrounds, H3 relaxin increased levels of marker hormones of the hypothalamic-pituitary-gonadal (HPG) axis (71, 107); and anatomical studies have identified RXFP3 in many areas in the hypothalamus relevant to reproductive neuroendocrinology, including the preoptic area and the PVN and SON (33). H3 relaxin administered icv or iPVN significantly increased plasma luteinizing hormone (LH) levels 30 min post-injection in male Wistar rats, an effect blocked by peripheral pre-treatment with a GnRH antagonist, consistent with increased central GnRH release (71). In these initial studies, H2 relaxin administration via the same routes produced a small,non-significant increase in LH,suggesting a stronger relaxin-3/RXFP3-mediated effect (71). Activity of the endogenous peptide at RXFP1 cannot be completely excluded from an involvement in these effects, however, as RXFP1 is expressed in the anterior hypothalamus and PVN of the rat (62). If so, high levels of endogenous relaxin-3 may act via RXFP1 to stimulate the HPG axis, an idea that could be tested experimentally. It would also be of interest to observe the effects of icv or iPVN administration of RXFP3-selective peptides (65, 72–74) on the HPG axis in female rats to check for similar hormone changes to those observed in male rats.

Notably, treatment of mouse hypothalamic neuron-like (GT1– 7) cells or hypothalamic explants with synthetic H3 relaxin, produced a dose-dependent increase in secretion of GnRH (71). However, it is again unclear if this is an RXFP3-mediated effect, as GT1–7 cells and hypothalamic explants also express RXFP1; and H3 relaxin binds and activates these receptors. As these cells express both RXFP1 and RXFP3, the relaxin-3 mediated secretion

of GnRH should be assessed for sensitivity to blockade with an RXFP3 antagonist (71–73).

In our recent study, we assessed the effect of chronic (∼14 weeks) viral-mediated expression of R3/I5 in the hypothalamus on GnRH mRNA levels (43). Although the difference between groups was not statistically significant, there was a trend for increased GnRH mRNA levels in AAV-R3/I5-treated compared to control rats, with considerable variability in individual values. Given the dispersed nature of GnRH expressing neurons in the hypothalamus, the variability observed suggests that different numbers (populations) of GnRH expressing neurons may have been dissected in these assays and/or there may be a genuine increase in GnRH transcription as a result of the treatment. This important question needs to be further investigated using suitable quantitative methods, in conjunction with further assessments of relaxin3/RXFP3 indices relative to GnRH and other reproductive peptides and receptors during different stages of pregnancy, birth and lactation.

#### **OTHER HYPOTHALAMIC SITES OF RELAXIN-3/RXFP3 ACTION**

The relaxin-3/RXFP3 system may have actions in other hypothalamic nuclei, demonstrated by the presence of both relaxin-3 immunoreactive fibers and RXFP3 mRNA/protein in the lateral and medial preoptic areas, anterior, posterior, dorsomedial, and ventromedial regions, and the adjacent supramammillary nucleus (SuM) (33) (**Table 1**). The functions of these regions will be briefly reviewed, in view of the postulated role of relaxin-3 in arousal, feeding, behavioral state, and cognition (39).

The SuM receives a moderate relaxin-3 innervation in the rat (33) and the mouse (34), and this nucleus represents a major target of the peptide network in the macaque brain (35). In contrast to other "hypothalamic" regions, this nucleus is best characterized as a key input to and modulator of the hippocampus and hippocampal theta rhythm (108, 109), a synchronous activity between 4 and 12 Hz associated with active waking behavior and REM sleep (110), mnemonic processing (111, 112), spatial navigation and exploration (113), and sensorimotor integration (114, 115). There is growing evidence that the NI is a key player in mediating brainstem-elicited theta rhythm (47, 53, 116). In addition to projections to the SuM, there are strong relaxin-3-containing NI projections to other regions subserving theta rhythm generation, such as medial septum (48–50), *reticularis pontine oralis*, and the hippocampus, as well as the median raphe, which is involved in theta desynchronization [see (52) for review]. In fact, relaxin-3 projections to the medial septum have been shown to promote hippocampal theta rhythm and are necessary for normal spatial navigation of rats in a spontaneous alternation task (47). Thus, the SuM is likely a further node at which this neuropeptide system acts to regulate hippocampal theta activity and associated cognitive/autonomic processes.

The lateral and medial preoptic areas of hypothalamus are also rich in relaxin-3 projections and RXFP3. The LPO area contains populations of neurons that have identified roles in thermoregulation (117, 118) and sleep-wakefulness (ventrolateral neurons) (119), so it will be of interest, to assess the precise topography of RXFP3 in these areas and to assess the effects of RXFP3 activation/inhibition in these regions on these physiological parameters, particularly as our laboratory has anatomical and functional data consistent with effects of relaxin-3 on sleep-wake activity in mice (34, 120). Similarly, as discussed, the role of relaxin-3/RXFP3 signaling in the medial preoptic area may be important in stress responses in female rats and the possible regulation of CRH neurons in the area (106).

# **RELAXIN-3/RXFP3 IN MOUSE HYPOTHALAMUS – SPECIES DIFFERENCES**

The distribution of relaxin-3 neurons and projections, as well as the distribution of RXFP3 mRNA and binding sites in the mouse brain, is regionally similar to that in the rat both generally throughout the forebrain and within the hypothalamus (33, 34). Furthermore, the distribution of Rxfp3 mRNA in the C57BL6/J mouse detailed in the Allen Brain Institute Brain Atlas (see text footnote 1) using a digoxigenin-labeled RNA probe is similar to that observed in our studies using radioactively labeled oligonucleotides (34) (**Table 1**). However, a more detailed analysis is required to determine whether identical populations of neurons are targeted within key hypothalamus areas, such as the arcuate, periventricular, PVN, and SON of rat and mouse to assess whether relaxin-3 signaling might regulate similar neuroendocrine processes in these species.

We have conducted several studies to assess the possible role of relaxin-3/RXFP3 in food intake in mice (121). After the administration of similar or higher levels of relaxin-3 or various RXFP3 selective agonists (R3/I5, RXFP3-A2) used in studies on rats [see (76) for review], we did not observe an increase in food intake of satiated C57BL/6J mice. In more recent studies, we have observed that the icv administration of the RXFP3 antagonist, R3(B1–22)R (73) produced a reduction in the robust feeding that occurred in mice offered access to regular chow after a 4 h period of isolated housing in a novel cage without bedding and in the absence of food (Smith and Gundlach, unpublished data). These particular conditions are presumed to induce a level of stress, hypothermia, and energy deficiency in the mice, which motivates their feeding; and suggests that relaxin-3/RXFP3 signaling in mice may not influence feeding significantly under basal conditions, but may play a role under altered stress conditions. These possibilities are currently being explored experimentally.

Consistent with these pharmacological studies, however, there is no genotypic difference between relaxin-3 KO and wild type littermates in bodyweight, total food consumption, or circadian rhythm of food consumption (120,122). These findings are in contrast to data obtained by Sutton et al. (123) in a separately derived colony of relaxin-3 KO mice on a mixed C57BL6/J/SV129 background, which displayed a markedly reduced body weight relative to wild type mice when both genotypes were fed a diet with a moderately elevated fat content. However, subsequent studies of a null mutation Rxfp3 KO mouse strain revealed no body weight-related phenotype, but did reveal an identical circadian hypoactivity phenotype to the relaxin-3 KO mouse (120,124,125). This suggests the relaxin-3 KO phenotype reported by Sutton et al. (123) was associated with genetic differences independent of relaxin-3/RXFP3 [see (39)].

With regard to the specificity of pharmacological studies of RXFP3 signaling in the mouse, while the presence of insulinlike peptide 5 (INSL5) and RXFP4 in the mouse brain has been reported (126), these findings have not been independently validated; and in a separate study, the presence of INSL5-sensitive receptor binding sites could not be identified (127). Furthermore, the ligand/receptor expression profile suggests the INSL5/RXFP4 system acts primarily within the gastrointestinal tract and large intestine (28, 128). While it is possible that peripheral INSL5 signaling may alter central (hypothalamic) function, at this stage no such data is available.

# **CONCLUSION AND PERSPECTIVES**

A decade of research has revealed the basic structural framework of central relaxin-3/RXFP3 networks and their likely functional importance in mammalian brain (27, 39, 76); and several studies have highlighted the interaction between hypothalamic homeostatic systems and relaxin-3/RXFP3 signaling, predominantly in pharmacological and regulatory studies in the rat. Together, these studies point to a role for relaxin-3/RXFP3 in regulating hypothalamic activity, with evidence suggesting it does so via interactions with oxytocin, AVP, and/or CRH, to modulate neuroendocrine function associated with stress,

#### **REFERENCES**


(1995) **20**:209–49. doi:10.1016/ 0165-0173(94)00013-F


feeding, and metabolism, and motivation and reward. Further studies are now required to clarify the nature of these various effects and how they are coordinated and regulated by hormonal and transmitter inputs onto relaxin-3 neurons and integrated signaling at the level of hypothalamic loops (e.g., HPA, HPG axes). Recent research strongly suggests the relaxin-3/RXFP3 system is an important regulator of the neuroendocrine axis, with potential as a novel therapeutic target for the treatment of neuroendocrine disorders and associated behavioral dysfunction.

#### **ACKNOWLEDGMENTS**

Research in the authors' laboratory is supported by grants from the National Health and Medical Research Council (NHMRC) of Australia (Andrew L. Gundlach, Sherie Ma), the Besen Family and Pratt Foundations (Andrew L. Gundlach), and by the Victorian Government Operational Infrastructure Support Program. Despina E. Ganella was the recipient of a Commonwealth Australian Postgraduate Award. Andrew L. Gundlach is the recipient of an NHMRC (Australia) Research Fellowship.


amygdala attenuates fear response. *Neuron* (2012) **73**:553–66. doi:10. 1016/j.neuron.2011.11.030


immunoreactivity, and RXFP3 binding sites in the brain of the macaque, *Macaca fascicularis*. *Ann N Y Acad Sci* (2009) **1160**:256–8. doi:10.1111/j.1749-6632.2009. 03954.x


(2005) **21**:1659–70. doi:10.1111/j. 1460-9568.2005.03980.x


rat thalamus – neuronal tracttracing and *in vitro* electrophysiological studies. *Eur J Neurosci* (2013) **37**:1284–94. doi:10.1111/ ejn.12155


GPCR135 and GPCR142 over relaxin receptor LGR7: in vitro and in vivo characterization. *J Biol Chem* (2007) **282**:25425–35. doi: 10.1074/jbc.M701416200


*Front Neuroendocrinol* (2004) **25**:150–76. doi:10.1016/j.yfrne. 2004.05.001


an oxytocin gene deletion. *Am J Physiol Regul Integr Comp Physiol* (2005) **289**:R1798–806. doi:10. 1152/ajpregu.00558.2005


programming of body weight and energy balance. *Neuron* (2011) **69**:523–35. doi:10.1016/j.neuron. 2010.12.036


**1160**:250–5. doi:10.1111/j.1749- 6632.2008.03796.x


food-entrainable arousal. *Proceedings of 6th International Conference on Relaxin and Related Peptides*. Florence: Florence University Press (Forthcoming 2013).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 22 July 2013; paper pending published: 06 August 2013; accepted: 02 September 2013; published online: 18 September 2013.*

*Citation: Ganella DE, Ma S and Gundlach AL (2013) Relaxin-3/RXFP3 signaling and neuroendocrine function – a perspective on extrinsic hypothalamic control. Front. Endocrinol. 4:128. doi: 10.3389/fendo.2013.00128*

*This article was submitted to Neuroendocrine Science, a section of the journal Frontiers in Endocrinology.*

*Copyright © 2013 Ganella, Ma and Gundlach. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Galanin receptors and ligands

#### **Kristin E. B.Webling<sup>1</sup>\*, Johan Runesson<sup>1</sup> ,Tamas Bartfai <sup>2</sup> and Ülo Langel 1,3**

<sup>1</sup> Department of Neurochemistry, Arrhenius Laboratories for Natural Science, Stockholm University, Stockholm, Sweden

<sup>2</sup> Molecular and Integrative Neurosciences Department, The Scripps Research Institute, La Jolla, CA, USA

3 Institute of Technology, University of Tartu, Tartu, Estonia

#### **Edited by:**

Jae Young Seong, Korea University, Korea

#### **Reviewed by:**

Jong-Ik Hwang, Korea University, Korea Sebastien G. Bouret, University of Southern California, USA

#### **\*Correspondence:**

Kristin E. B. Webling, Department of Neurochemistry, Arrhenius Laboratories for Natural Science, Stockholm University, Svante Arrheniusv. 21A, 10691 Stockholm, Sweden. e-mail: kristin.webling@neurochem. su.se

**THE GALANIN FAMILY**

Since the discovery of galanin 30 years ago, several bioactive peptides have been reported to be part of the galanin family. The discovery of galanin was followed by the characterization of a second peptide originating from the same prepropeptide as galanin, the galanin message associated peptide (GMAP). Furthermore, a third peptide, GALP, was identified with capacity to bind to the galanin receptor subtypes, GalR1-3, followed by the characterization of a splice variant of GALP named alarin.

# **GALANIN**

Galanin was discovered among several other bioactive peptides with C-terminal α-amide motif, using a new method by Professor Viktor Mutt and colleagues at Karolinska Institute, Stockholm (Tatemoto et al., 1983; Hökfelt, 2005; Lang et al., 2007). The 29 amino acid long peptide (30 amino acids in humans) was named galanin after its N-terminal glycine and its C-terminal alanine. The N-terminal end of galanin is crucial for its biological activity and the first 15 amino acids are conserved in all species (the tuna fish being the exception; Kakuyama et al., 1997). Interestingly, the C-terminal region (residues 17–29) varies among species and it lacks receptor affinity (**Table 2**), which is also true for N-terminal fragments shorter than galanin (1–11) (Land et al., 1991b). The C-terminus is believed to primarily serve as a protector against proteolytic attacks (Land et al., 1991a; Bedecs et al., 1995). In a membrane-mimicking environment, galanin adopts a horseshoelike shape, where the N-terminus is organized in an α-helical conformation,followed by a β-bend around the proline in position 13 and a more uncertain configuration of the C-terminal region (Wennerberg et al., 1990; Morris et al., 1995, Öhman et al., 1998).

Galanin has been ascribed a large range of different functions. To accomplish these, the galanin gene has a highly plastic

The neuropeptide galanin was first discovered 30 years ago. Today, the galanin family consists of galanin, galanin-like peptide (GALP), galanin-message associated peptide (GMAP), and alarin and this family has been shown to be involved in a wide variety of biological and pathological functions. The effect is mediated through three GPCR subtypes, GalR1-3. The limited number of specific ligands to the galanin receptor subtypes has hindered the understanding of the individual effects of each receptor subtype. This review aims to summarize the current data of the importance of the galanin receptor subtypes and receptor subtype specific agonists and antagonists and their involvement in different biological and pathological functions.

**Keywords: galanin, galanin-like peptide, GMAP, alarin, epilepsy**

expression pattern, which has been portrayed numerous times in the literature. Galanin was early shown to be induced by estrogens (Vrontakis et al., 1987, 1989; Kaplan et al., 1988), and later, three copies of estrogen responsive element, ERE, were identified in the promoter region of the human galanin gene (Kofler et al., 1995). Thereafter, the galanin expression has also been shown to be up-regulated by the leukemia inhibitory factor (LIF; Corness et al., 1996; Sun and Zigmond, 1996), and down-regulated by the nerve growth factor (NGF; Verge et al., 1995).

Galanin is widely expressed in the central and peripheral nervous system as well as in the endocrine system and co-exists with a number of classical neurotransmitters, including acetyl choline, serotonin, glutamate, GABA, noradrenalin, and dopamine (Melander et al., 1986; Hökfelt et al., 1987; Xu et al., 1998; Liu et al., 2003). Galanin also co-exists with other neuropeptides like enkephalin, NPY, substance P, vasopressin, calcitonin generegulated peptide, and gonadotropin-releasing hormone (Rökaeus and Carlquist, 1988; Merchenthaler et al., 1990; Zhang et al., 1993a,b, 1995).

An extensive up-regulation of galanin was seen during development of sensory and motor systems (Gabriel et al., 1989; Xu et al., 1996) and after nerve injury, both in PNS and CNS (Hökfelt et al., 1987) and also, an extensive up-regulation in the basal forebrain of patients with Alzheimer's disease (AD; Chan-Palay, 1988a,b). Epileptic seizures have been shown to rapidly deplete galanin (Mazarati et al., 1998).

Galanin has also been shown to be expressed in keratinocytes, eccrine sweat glands and around blood vessels (Kofler et al., 2004). Furthermore, galanin has been proposed to be expressed in macrophages of the dermis (reviewed in Bauer et al., 2010).

# **GALANIN MESSAGE ASSOCIATED PEPTIDE**

There are very few studies regarding the localization,function, and pharmacological potential of GMAP. It was early shown that the sequence of GMAP displays a much greater divergence between species than galanin (Lundkvist et al., 1995). Immunohistochemistry has shown that GMAP distribution generally parallels that of galanin (Hökfelt et al., 1992) although heterologous distribution was observed in certain areas. Xu et al. (1995a,b) showed that GMAP has a pharmacological action in spinal nociceptive transmission in rat spinal cord (Andell-Jonsson et al., 1997; Hao et al., 1999). GMAP has also been assigned anti-microbial activities and hypothesized to be part of the innate immune system, since it

suppresses *Candida albicans* growth and the budded-to-hyphalform transition of *C. albicans* (Rauch et al., 2007) (**Table 1**). Recently, in an extended study, it was shown that GMAP could significantly reduce growth in six out of seven *Candida* strains (Holub et al., 2011).

#### **GALANIN-LIKE PEPTIDE**

Ohtaki et al. (1999) characterized a third peptide, isolated from porcine hypothalamus, that was recognized to induce GTPbinding to a membrane preparation of GalR2-transfected cells. They named this new peptide galanin-like peptide, or GALP. Porcine GALP was shown to act as an agonist in a GTPγS binding

#### **Table 1 | A short summary of the involvement of the galanin family in different physiological and pathological functions.**



**Table 2 | Affinities of galanin, GALP, GMAP, and alarin, as well as fragments of galanin and GALP, for the three galanin receptor subtypes, determined as K<sup>i</sup> .**

Displacement is performed on the rat galanin receptor unless indicated otherwise. (h) human; <sup>a</sup>presented as IC<sup>50</sup> values; – not determined.

assay and to have a preferential binding (20 times) toward GalR2 (Ohtaki et al., 1999). A later study using human GALP showed that GALP interacts with GalR3 with three times preferential selectivity as compared to GalR2 (Lang et al., 2005) (**Table 2**).

The amino acid sequence of GALP-(9–21) is identical to that of galanin (1–13).

Galanin-like peptide distribution in the CNS appears to be rather restricted, disparate to the much broader expression pattern seen for galanin. Cells identified to produce GALP mRNA and protein have only been found in the hypothalamic arcuate nucleus (ARC), the median eminence and infundibular stalk, and the posterior pituitary when studied in the rat, mouse, and macaque (Juréus et al., 2000, 2001; Kerr et al., 2000; Larm and Gundlach, 2000; Takatsu et al., 2001; Cunningham et al., 2002; Fujiwara et al., 2002). GALP-immunoreactive (IR) fibers were shown to project to several regions of the forebrain (Takatsu et al., 2001).

Galanin-like peptide has also been shown to be expressed by specialized glia-like cells known as pituicytes in the neuronal lobe of dehydrated and salt loaded rats, where the expression is strongly regulated by osmotic stimuli (Shen et al., 2001; Fujiwara et al., 2002; Saito et al., 2003; Shen and Gundlach, 2004). Furthermore, studies show that the GALP gene expression, especially in the pituicytes, is induced by both acute and chronic inflammatory stimuli (Saito et al.,2003,2005). Central administration of GALP increases IL-1α and IL-1β and it has been suggested that IL-1 mediates both the anorectic and febrile actions of GALP (Man and Lawrence, 2008b).

Intracerebroventricular (i.c.v.) injection of GALP profoundly stimulates male sex behaviors in rat (Fraley et al., 2004), seemingly independent of the testosterone milieu (Stoyanovitch et al., 2005) (**Table 1**). Interestingly, the opposite is seen in mice were GALP instead inhibits male sex behavior (Kauffman et al., 2005). Recently, Taylor et al. (2009) presented evidence supporting the hypothesis that this effect of GALP depends upon hypothalamic dopamine input to the medial preoptic area (mPOA).

Several studies have proposed that GALP does not solely interact with the three known galanin receptor subtypes (Man and Lawrence, 2008a). Krasnow et al. (2004)reported that GALP injection affect food intake and body weight in a similar manner in both GalR1-KO and GalR2-KO mice compared to wild type littermates. Furthermore, to somewhat exclude the possibility that this effect was mediated through GalR3, the authors showed that the GALP fragment, GALP (1–21), failed to mimic the effect of full length GALP (Krasnow et al., 2004).

#### **ALARIN**

The newest member of the galanin peptide family, alarin, a 25 amino acid long peptide named after its N-terminal alanine and its C-terminal serine originating as a splice variant of the GALP mRNA (Santic et al., 2006). The alarin peptide has been isolated from murine brain, thymus, skin (Santic et al., 2007), human neuroblastic tumors, and human skin (Santic et al., 2006, 2007) and has no detectable affinity toward either of the three galanin receptor subtypes (Boughton et al., 2010) (**Table 2**). Recently, two publications characterized in more detail the alarin-LI in the murine brain (van Der Kolk et al., 2010; Eberhard et al., 2012). Alarin-LI has a much broader expression pattern than GALP and was found in such diverse areas as the accessory olfactory bulb, different nucleus in the hypothalamus, within the locus coeruleus (LC) and locus subcoeruleus of the midbrain.

When first discovered, alarin was ascribed vasoconstrictive and anti-edema activities (Santic et al., 2007) (**Table 1**). Contradictory to the effect of GALP, alarin has neither an effect on body temperature nor an effect on male sex behaviors in rodents (van Der Kolk et al., 2010; Fraley et al., 2012). Recently, it was shown that alarin stimulates acute food intake and some studies have reported a significant increase in body weight after 24 h, although other studies were unable to confirm this (Boughton et al., 2010; van Der Kolk et al., 2010; Fraley et al., 2012). Central injection of alarin elicit a gonadotrophin-releasing hormone (GnRH)-mediated increase in leutizing hormone (LH)-levels in both rats and mice (Boughton et al., 2010; van Der Kolk et al., 2010; Fraley et al., 2012).

#### **GALANIN RECEPTOR SUBTYPES**

All three galanin receptor subtypes are members of the GPCR superfamily but the subtypes have substantial differences in sites of expression as well as their functional coupling and subsequent signaling activities. These differences between the receptor subtypes contributes to the diversity of possible physiological effects and the plausible pharmacological relevance of targeting the galanin family (**Table 1**).

#### **GALANIN RECEPTOR TYPE 1**

The first known galanin receptor, galanin receptor type 1 (GalR1), was isolated from the Bowes human melanoma cell line (Habert-Ortoli et al., 1994) and subsequently rat (Burgevin et al., 1995; Parker et al., 1995) and mouse (Jacoby et al., 1997; Wang et al., 1997c) receptor was cloned.

The human GalR1 gene contains three exons and the hGalR1 gene translates into a 349 amino acid long protein (Jacoby et al., 1997). The homology between species is rather high, as 93% of the residues in rat GalR1 are identical to those of human GalR1 (Jacoby et al., 1997). The expression of GalR1, but neither GalR2 nor GalR3, is regulated by cyclic adenosine monophosphate (cAMP) through the transcription factor CREB (cAMP regulatory element binding protein; Zachariou et al., 2001; Hawes et al., 2005). The GalR1 expression does not fluctuate during development (Branchek et al., 2000; Burazin et al., 2000).

GalR1 mRNA was initially identified by northern blot to be found in the fetal brain and small intestinal tissues (Habert-Ortoli et al., 1994). It has, thereafter, been identified by reverse transcript polymerase chain reaction (RT-PCR) in the gastrointestinal tract (Lorimer and Benya, 1996). However, a later study identified the GalR1 expression to be exclusively in the central and peripheral nervous system (Waters and Krause,2000),where it was detected in hippocampus, hypothalamus, amygdala, thalamus, cortex, brainstem (medulla oblongata), spinal cord, and dorsal root ganglia (DRG; Gustafson et al., 1996; Waters and Krause, 2000), even if broader central and peripheral tissue distribution has also been reported (Sullivan et al., 1997).

Activation of GalR1 results in a pertussis toxin (PTX) sensitive inhibition of adenylate cyclase (AC) through interaction with Gαi/α<sup>o</sup> types of G-proteins (Habert-Ortoli et al., 1994; Parker et al., 1995; Wang et al., 1997c) which leads to opening of GIRK channels. Activation of GalR1 can also stimulate a mitogen associated protein kinase (MAPK) activity, through a PKC-independent mechanism, consistent with that the mediator is the βγ-subunit of Gα<sup>i</sup> (Wang et al., 1998).

#### **GALANIN RECEPTOR TYPE 2**

The second galanin receptor type (GalR2) was identified in rat hypothalamus, spinal cord, and DRG (Fathi et al., 1997; Howard et al., 1997; Smith et al., 1997; Ahmad and Dray, 2004) and subsequently in mouse spleen (Pang et al., 1998) as well as from various human tissues (Bloomquist et al., 1998; Borowsky et al., 1998). The human GalR2 has rather high sequence identity to rat GalR2 (92%), although there is one notable difference; the 15 amino acid extension of the C-terminal end in human GalR2 (Kolakowskim et al., 1998; Waters and Krause, 2000).

GalR2 is able to activate the stimulatory pathway of Gαq/11 class of G-proteins, i.e., PTX-insensitive. This triggers PLC activity and intracellular phosphoinositol turnover, mediating the release of Ca2<sup>+</sup> into the cytoplasm from intracellular stores and opening Ca2+-dependent channels (Smith et al., 1997; Kolakowskim et al., 1998; Wang et al., 1998). GalR2 is also able to activate MAPK through a PKC and Gα<sup>o</sup> class of G-proteins dependent mechanism (Wang et al., 1998). This may in turn lead to the downstream PI3K-dependent phosphorylation of Protein Kinase B (PKB) leading to suppression of caspase-3 and caspase-9 activity (Ding et al., 2006; Elliott-Hunt et al., 2007). GalR2 activation may also inhibit forskolin stimulated cAMP production in a PTX-sensitive manner, suggesting the activation of Gαi/α<sup>o</sup> types of G-proteins (Fathi et al., 1997; Wang et al., 1997a). Consequently, both GalR1 and GalR2 activation can inhibit CREB (Badie-Mahdavi et al., 2005).

GalR2 is expressed in a wider pattern, compared to GalR1, as it is found in several peripheral tissues including the pituitary gland, gastrointestinal tract, skeletal muscle, heart, kidney, uterus, ovary, and testis as well as in regions in the CNS (Smith et al., 1997; Bloomquist et al., 1998; Waters and Krause, 2000). In the brain, the highest levels of GalR2 are detected in hypothalamus, dentate gyrus, amygdala, piriform cortex, and mammillary nuclei (Mitchell et al., 1999; O'Donnell et al., 1999; Waters and Krause, 2000).

Interestingly, GalR2 expression levels vary during the development of the rat brain with a broader distribution with a peak in expression before postnatal day 7, particularly in cortex and thalamus, and much reduced levels after postnatal day 14 (Burazin et al., 2000).

#### **GALANIN RECEPTOR TYPE 3**

Galanin receptor type 3 (GalR3) was first isolated from rat hypothalamic cDNA libraries (Wang et al., 1997b) and later from human cDNA (Kolakowskim et al., 1998; Smith et al., 1998). The 368 amino acid long hGalR3 shares 36% amino acids identity with hGalR1 and 58% with hGalR2 and approximately 90% with rGalR3 (Kolakowskim et al., 1998).

The distribution pattern of GalR3 is somewhat unclear but it is assumed that this receptor has a more restricted expression pattern in relation to the other two receptors. Transcript levels is most prominent in the hypothalamus (Wang et al., 1997b; Smith et al., 1998; Mennicken et al., 2002) although, some studies report a wider distribution of GalR3 throughout central and peripheral tissues (Kolakowskim et al., 1998; Waters and Krause, 2000).

Signaling properties of GalR3 are still ill-defined. Activation of GalR3 expressed in *Xenopus* oocytes or *Xenopus* melanophores leads to the activation of Gαi/α<sup>o</sup> type of G-proteins inhibiting AC which results in the opening of GIRK channels (Kolakowskim et al., 1998; Smith et al., 1998).

#### **PEPTIDE LIGANDS FOR THE GALANIN RECEPTORS**

Endogenous galanin has high affinity for all three galanin receptors (Wang et al., 1997b). The N-terminal part of galanin is crucial for receptor interaction and the galanin fragment galanin (1–16) retains the high affinity of its parental peptide. When galanin (1–16) underwent an L-alanine scan and subsequent testing on rat hypothalamus membranes, Gly<sup>1</sup> , Trp<sup>2</sup> , Asn<sup>5</sup> , Tyr<sup>9</sup> , and Gly<sup>12</sup> were identified as pharmacophores (Land et al., 1991b). A later study, which tested an identical set of peptides on separated GalR1 and GalR2 membranes, identified Trp<sup>2</sup> , Tyr<sup>9</sup> , and Leu<sup>10</sup> as pharmacophores on both receptor subtypes (Carpenter et al., 1999).

Several N-terminal truncated galanin fragments have been shown to have a preference for GalR2 (Wang et al., 1997b; Liu et al., 2001), in concurrence with the fact that Gly<sup>1</sup> is of great importance for ligand binding to GalR1. Further truncation, with as little as two amino acids, leads to a complete loss of receptor affinity to all receptor subtypes (Wang et al., 1997a).

Liu et al. (2001) published the galanin fragment galanin (2–11) as a GalR2 selective agonist, although they did not test it on GalR3 (**Table 3**). Later publication has unfortunately shown that it has similar affinity toward GalR3 (Lu et al., 2005a), without testing receptor signaling, even so, it has been used extensively as a non-GalR1 agonist. Lundström and colleagues showed that Trp<sup>2</sup> , Asn<sup>5</sup> , Gly<sup>8</sup> , Tyr<sup>9</sup> , and Leu<sup>10</sup> were identified as crucial for interactions with GalR2 by performing Ala-scan on the peptide (Lundström et al., 2005a).

The interaction between the galanin receptor subtypes and GALP has received less attention. GALP, isolated from porcine tissues, was original published as a GalR2 preferring ligand, with a 20 times difference in affinity between GalR1 and GalR2 (Ohtaki et al., 1999). Later it was shown, using human GALP, that GALP also interacts with GalR3. In this study GALP was ascribed a GalR3 preferential selectivity (3 times differences; Lang et al., 2005). Recently, Boughton et al. (2010) showed a more than 10 times preferential binding toward GalR3 for the rat GALP (**Table 2**).

Several chimeric ligands have been synthesized, conjugating galanin (1–13) to other bioactive molecules, yielding M15 (also called galantide; Bartfai et al., 1991),M32 (Wiesenfeld-Hallin et al., 1992b), M35 (Wiesenfeld-Hallin et al., 1992b, Ögren et al., 1992, Kask et al., 1995), C7 (Langel et al., 1992), and M40 (Langel et al., 1992; Bartfai et al., 1993). Although, they all maintain antagonistic properties *in vivo* at doses between 0.1 and 10 nmol when delivered i.c.v. or intrathecally (i.t.; Parker et al., 1995; Lu et al., 2005b), they all have a partial agonistic nature *in vivo* at doses higher than 10 nmol when delivered i.c.v. or i.t. (Kask et al., 1995; Lu et al., 2005b).

The first introduced chimeric peptide which acts as an antagonist of the galanin receptor family was M15 (Bartfai et al., 1991).


#### **Table 3 | Published ligands and their affinities for the galanin receptor subtypes.**

The sequences and structures of the ligands are listed in**Table 4**.

Displacement was performed on the rat galanin receptor unless indicated otherwise. (h) human; – not determined.

#### **Table 4 | The sequences for the galanin family peptides along with the discussed analogs.**


(Continued)

Here, the galanin (1–13) fragment, was coupled to a C-terminal fragment in substance P (residue 5–11), reported to have agonistic effect on the substance P receptor. M15 showed an about 10-fold higher affinity than the endogenous galanin to unspecified subtypes of the galanin receptor family in membrane preparations of rat tissues. Later, M35 was synthesized (Ögren et al., 1992) with an improved *in vivo* stability (Wiesenfeld-Hallin et al., 1992b). M15, M32, M35, and M40 have similar affinity as galanin and have been valuable tools in galanin research but are limited by their relative non-specificity toward the different galanin receptors (Ögren et al., 1992) and by their weak interactions with other receptors than the galanin receptors (Wiesenfeld-Hallin et al., 1992a).

M617 resembles the M35 peptide, with the substitution of proline at position 14 to a glutamine, which results in a 25 fold selectivity for GalR1 over GalR2 *in vitro* (**Table 3**). M617

#### Sar, sarcosine.

has thereafter been shown to produce anti-nociceptive effects (Jimenez-Andrade et al., 2006) and to delay the development of seizure in an animal model (Mazarati et al., 2006). The M871 peptide is N-terminally truncated and has two additional amino acid residues compared to the M40 peptide and function as a partial agonist, selective for GalR2 (Sollenberg Eriksson et al., 2006, 2010). M871 has been used in several *in vivo* studies (Jimenez-Andrade et al., 2006; Alier et al., 2007; Kuteeva et al., 2008). Several GalR2 selective agonists have been reported over the years (Pooga et al., 1998; Runesson et al., 2009; Saar et al., 2011). Small changes in the N-terminus of galanin have been associated with lost binding affinity. However, recently analogues with modifications at both N-terminus and C-terminus have been presented, namely M1145 (Runesson et al., 2009) and M1153 (Saar et al., 2011). M1145 was reported as the first specific GalR2 agonist with a 90-fold binding preference for GalR2 over GalR1 and 76-fold preference over GalR3 (Runesson et al., 2009). The importance of the development of M1145 and M871 and other subtype selective agonists and antagonists can almost not be overestimated and is the key to a successful delineation of galaninergic system and to identify its potential as a therapeutic target.

Recently, several galanin analogs, all modified by introducing several cationic amino acid residues and a palmitoyl moiety was shown to exhibit improved bioavailability after systemic administration (Bulaj et al., 2008; White et al., 2009). One of these, the Gal-B2, with a slight selectivity toward GalR1 (**Table 3**), was shown to have anticonvulsant effect in several tested animal models (White et al., 2009). In a later study, Bulaj and colleagues modified Gal-B2 to obtain a ligand with an 18 times preferential binding toward GalR2, which displayed similar anticonvulsant activity as the parental peptide (Robertson et al., 2010). Future characterization will probably identify other potential application of Gal-B2 and other systemically active galanin analogs.

#### **NON-PEPTIDE LIGANDS**

The non-peptide ligand galnon was identified after screening a combinatorial peptidomimetic library (**Table 5**). It acts as an agonist in functional studies both *in vitro* and *in vivo* (Saar et al., 2002; Bartfai et al., 2004). It has been evaluated in models of anxiety and depression (Rajarao et al., 2007), feeding (Abramov et al., 2004), and pain (Wu et al., 2003). Galmic (**Table 5**) is a nonpeptide agonist with higher affinity for GalR1 compared to GalR2, which under conditions of intrahippocampal administration was 6-fold more potent than galnon in inhibiting self-sustaining status epilepticus (SE), an *in vivo* model for epilepsy (Bartfai et al., 2004; Ceide et al., 2004). Nevertheless, both galnon and galmic potentials are limited by the fact that they have multiple sites of interactions, i.e., D2 dopamine receptors, grehlin and melanocortin receptors, which produce unwanted physiological effects (Florén et al., 2005; Lu et al., 2005b).

The metabolite Sch 202596 (**Table 5**), originated from an *Aspergillus* sp. culture found in an abandoned uranium mine in Tuolemene County California, was found to have a modest affinity to GalR1 *in vitro* (Chu et al., 1997). Sch 202596 was characterized as a molecule with a spirocoumaranone skeleton and has only partly been synthesized so far (Katoh et al., 2002). Several 1,4-dithiins and dithiipine-1,1,4,4-tetroxides with binding affinity to GalR1 were identified at the R. W. Johnson Pharmaceutical Institute (Scott et al., 2000). The compound 2,3-dihydro-2- (4-methylphenyl)-1,4-dithiepine-1,1,4,4-tetroxide (**Table 5**) was shown to be a submicromolar antagonist. It has an IC<sup>50</sup> of 190 nM for GalR1 and above the highest tested concentration (30µM) for GalR2. However, its reactive nature and its low solubility makes it unattractive from a therapeutic point of view. Nevertheless, it has been used and evaluated in several studies (Mahoney et al., 2003; Kozoriz et al., 2006).

A series of 3-imonio-2-indolones were identified as specific GalR3 antagonists, with Ki-values for GalR3 as low as 17 nM


**Table 5 | Affinities of non-peptidergic galanin receptor ligands for the three galanin receptor subtypes, determined as K<sup>i</sup> on human receptor subtypes.**

The structures of the ligands are listed in**Table 4**.

Displacement is performed on the rat galanin receptor unless indicated otherwise.

<sup>a</sup>presented as IC<sup>50</sup> values; – not determined.

and above the tested 10µM for the other receptors studied (Konkel et al., 2006a). One of these was referred as SNAP37889 (Swanson et al., 2005) (**Table 5**). One drawback of the above mentioned indolones is the low aqueous solubility (less than 1µg/ml) which motivated further studies, leading to the identification of a compound with an increased water solubility and selectivity, 1,3-dihydro-1-[3-(2-pyrrolidinylethoxy)phenyl]- 3-[[3-(trifluoromethyl)phenyl]imino]-2*H*-indol-2-one, referred as SNAP398299 (Swanson et al., 2005; Konkel et al., 2006b) (**Table 5**). Another of the synthesized indolones (**Table 5**) was evaluated *in vivo* by Barr et al. (2006), which together with the other articles and several patent applications (Konkel et al., 2004) indicates that specific GalR3 ligands are in development.

A series of 2,4,6-triaminopyrimidines were recently introduced by The Scripps Research Institute (Sagi et al., 2011). They present both GalR1 and GalR2 selective compounds with Ki-values starting from 330 nM. Further development of these compounds is likely ongoing and published in due course. Studies from the same institute led to characterization of the first identified allosteric modulator, named CYM2503, for the galanin receptor family, i.e., GalR2 (Lu et al., 2010). CYM2503 failed to displace galanin in binding studies and showed no detectable signaling by itself, but potentiated the effect of galanin when administered simultaneously (Lu et al., 2010).

#### **GALANIN LIGANDS AS POSSIBLE THERAPEUTICS FOR EPILEPSY**

Among the early reported biological effects of galanin were the decreased excitability of myenteric neurons (Tamura et al., 1988) and cardiac ganglia (Konopka et al.,1989). These findings, together with reports that the hippocampus, which is a key structure for the initiation and maintenance of seizures, have a considerable amount of galaninergic innervation (Lu et al., 2005b) draw attention to galanin as a possible anticonvulsant (Mitsukawa et al., 2008).

Mazarati et al. (1992) reported that galanin had an anticonvulsant effect in a picrotoxin-kindled seizure model. Since then, galanin has been shown to up-regulated in several models of SE (in adult rats), i.e., in kainic acid-induced SE (Wilson et al., 2005) and after perforant path stimulation-induced SE (Mazarati et al., 1998). Galanin administrated i.c.v. had anticonvulsant activity in rodents exposed to either PTZ or Li-pilocarpine (Chepurnov et al., 1998; Mazarati et al., 1998, 2000). Similar results were obtained when SE was induced by perforant path stimulation (Mazarati et al., 1998, 2004a).

The galanin receptor subtypes present in the hippocampus have been investigated and both GalR1 and GalR2 are present in relatively high levels (Lu et al., 2005b) with GalR1 mRNA in CA-fields and GalR2-mRNA in the dentate gyrus (Burazin et al., 2000). The involvement of GalR3 in hippocampus is still not well characterized.

GalR1-KO mice displayed a more severe seizure phenotype when SE is induced by either perforant path stimulation or Lipilocarpine exposure but not when induced by KA exposure compared to WT (Mazarati et al., 2004b). Li-pilocarpine exposure resulted in cell death in CA1, an effect that was elevated in GalR1-KO mice (Mazarati et al., 2004b). Inbred mice with a lower expression of GalR1 has a larger cell loss than wildtype littermates in several hippocampal regions when exposed to KA (Kong et al., 2008; Schauwecker, 2010) without any alteration in seizure parameters. Some studies has also reported that GalR1-KO mice exhibit spontaneous epilepsy (Jacoby et al., 2002; Fetissov et al., 2003; McColl et al., 2006) although other studies could not replicate this phenotype (Mazarati et al., 2004b).

GalR2-KO mice display no difference in seizure susceptibility in two model of SE compared to WT (Gottsch et al., 2005). In contrast to the knockout mice, application of a putative GalR2 specific ligand shorten the SSSE duration and decreased the seizure density and seizure episodes in the perforant path stimulation model, but not the duration of single seizure episodes (Mazarati et al., 2004a). Similar effects were reported after addition peptide nucleic antisense (PNA) oligonucleotide that mediated transient downregulation of GalR2. PNA-treatment resulted in an increase in the severity of SSSE after perforant path stimulation (Mazarati et al., 2004a). Increased damage to hilar interneurons was also seen after PNA-application (Mazarati et al., 2004a).

Acute administration of two systemically active non-selective subtype galanin receptor agonists, galnon, and galmic, has been shown to prevent self-sustained seizure activity (Saar et al., 2002; Bartfai et al., 2004) and penthylenetetrazole (PTZ)-induced seizures (Saar et al., 2002). Galnon has shown to interact with several other receptors (Florén et al., 2005), although the anticonvulsant effect seems to be mediated via GalR1, as pretreatment with a GalR1-specific PNA attenuates its anticonvulsant properties (Saar et al., 2002).

In concordance with this, application of non-selective subtype galanin receptor antagonists has been shown to worsen the severity of SE in several models, i.e., kainic acid-induced seizures (Reiss et al., 2009), hippocampal kindling model (Kokaia et al., 2001), self-sustained SE (SSSE), and PTZ-induced convulsions (Chepurnov et al., 1998; Mazarati et al., 1998, 2000; Saar et al., 2002). A recent study showed that M15, a non-selective subtype galanin receptor antagonist significantly induced cell death in several hippocampal areas although no differences in the latency of onset or duration of severe seizures were seen (Schauwecker, 2010).

Galanin-KO mice have a lower threshold for developing SE after perforant path stimulation or KA exposure compared to WT (Mazarati et al., 2000). Furthermore, Gal-KO mice displayed a neuronal injury in the CA3-region that was not present in WT littermates (Mazarati et al., 2000). In concordance with this, Galanin-OE mice have a higher threshold for SE induced by either perforant path stimulation or PTZ and KA exposure compared to WT (Mazarati et al., 2000). Gal-OE mice have been shown to be less affected during hippocampal kindling, a model for human complex partial epilepsy (Kokaia et al., 2001).

Utilizing a recombinant adeno-associated viral (AAV) system that overexpresses galanin resulted in a dramatic reduction in KA-induced seizure episodes and the total time spent in seizures although no reduction of cell damage was seen (Lin et al., 2003). The same vector delayed the initiation of convulsions at generalized seizure stages and shortened the duration of

#### **REFERENCES**


37889, reduces operant responding for ethanol in alcohol-preferring rats. *Regul. Pept.* 166, 59–67.


electrographic after discharges in rats undergoing hippocampal kindling (Kanter-Schlifke et al., 2007). A similar AAV system that overexpresses galanin together with the fibronectin secretory signal sequence succeeded to the attenuation of KA-induced seizures and the neuronal death after KA exposure (Haberman et al., 2003).

A recent study showed that a GalR2 allosteric modulator increased the latency to the first electrographic seizure, decrease the total time in seizure and decreased the mortality in the Li-pilocarpine SE-model (Lu et al., 2010).

Furthermore, acute administration of the systemically active subtype galanin receptor agonist, Gal-B2, with a moderate GalR1 preferential binding, prevents seizures in the 6 Hz mouse model of pharmacoresistant epilepsy (Bulaj et al., 2008). It was later shown to be active also in other seizure and epilepsy models (White et al., 2009). An analog with a moderate GalR2 preferential binding [Nme, des-Sar]Gal-B2, also prevent seizure in the 6 Hz mouse model (Robertson et al., 2010). The authors conclude that these GalR1 and GalR2 preferential analogs (with 15 and 18 times selectivity, respectively) exhibit similar levels of anticonvulsant activity in the 6 Hz mouse model.

In summary, the wide involvement of galanin family peptides in physiological and pathological conditions has drawn attention to this neuropeptide family. Among the earliest areas of interests was the usage of galanin as a possible anticonvulsant.

Due to the three different galanin receptors specific expression in the CNS, several attempts have been made trying to characterize the contribution of each receptor and delineate their effects. Unfortunately, more selective or specific ligands are still needed.

Recent publications of stable peptide ligands have made new administration routes available as well as attract attention from the pharmaceutical industry.


and dehydrated rats. *J. Physiol. Pharmacol.* 54, 625–641.


of a galanin-like peptide (GALP) in pituicytes of the rat posterior pituitary gland. *Neurosci. Lett.* 317, 65–68.


(2011). Effects of galanin messageassociated peptide and neuropeptide Y against various non-albicans Candida strains. *Int. J. Antimicrob. Agents* 38, 76–80.


product of the rat anterior pituitary. *Proc. Natl. Acad. Sci. U.S.A.* 85, 7408–7412.


in health and disease. *Pharmacol. Ther.* 115, 177–207.


galanin receptor 2. *Neuropeptides* 39, 169–171.


seizures by galanin. *J. Neurosci.* 20, 6276–6281.


*Candida albicans*. *Antimicrob. Agents Chemother.* 51, 4167–4170.


the galanin-like peptide gene. *J. Mol. Neurosci.* 29, 145–152.


and M871 to galanin receptor type 3 reveals characteristics of galanin receptor-ligand interaction. *Int. J. Pept. Res. Ther.* 16, 17–22.


neuropeptide expression in vivo: a novel role in peptide suppression in adult sensory neurons. *J. Neurosci.* 15, 2081–2096.


galanin analog, in epilepsy models. *Neurotherapeutics* 6, 372–380.


and NPY-like immunoreactivities in locus coeruleus and noradrenergic nerve terminals in the hippocampal formation and cortex with notes on the galanin-R1 and -R2 receptors. *J. Comp. Neurol.* 392, 227–251.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 24 August 2012; accepted: 08 November 2012; published online: 07 December 2012.*

*Citation:Webling KEB, Runesson J, Bartfai T and Langel Ü (2012) Galanin receptors and ligands. Front. Endocrin. 3:146. doi: 10.3389/fendo.2012.00146*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Webling , Runesson, Bartfai and Langel. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *Anne-Laure Schang, Bruno Quérat, Violaine Simon, Ghislaine Garrel, Christian Bleux, Raymond Counis, Joëlle Cohen-Tannoudji and Jean-Noël Laverrière\**

Physiologie de l'Axe Gonadotrope, Biologie Fonctionnelle et Adaptative, EAC CNRS 4413, Sorbonne Paris Cité, Université Paris Diderot-Paris 7, Paris, France

"fendo-03-00162" — 2012/12/11 — 20:28 — page 1 — #1

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Jae Young Seong, Korea University, South Korea Ursula B. Kaiser, Brigham and Women's Hospital, USA

#### *\*Correspondence:*

Jean-Noël Laverrière, Physiologie de l'Axe Gonadotrope, Biologie Fonctionnelle et Adaptative, EAC CNRS 4413, Sorbonne Paris Cité, Université Paris Diderot-Paris 7, Bâtiment Buffon, case courrier 7007, 4 rue MA Lagroua Weill-Hallé, 75205 Paris Cedex 13, France. e-mail: jean-noel.laverriere@univ-parisdiderot.fr

The GnRH receptor (GnRHR) plays a central role in the development and maintenance of reproductive function in mammals. Following stimulation by GnRH originating from the hypothalamus, GnRHR triggers multiple signaling events that ultimately stimulate the synthesis and the periodic release of the gonadotropins, luteinizing-stimulating hormone (LH) and follicle-stimulating hormones (FSH) which, in turn, regulate gonadal functions including steroidogenesis and gametogenesis. The concentration of GnRHR at the cell surface is essential for the amplitude and the specificity of gonadotrope responsiveness. The number of GnRHR is submitted to strong regulatory control during pituitary development, estrous cycle, pregnancy, lactation, or after gonadectomy. These modulations take place, at least in part, at the transcriptional level. To analyze this facet of the reproductive function, the 5- regulatory sequences of the gene encoding the GnRHR have been isolated and characterized through in vitro and in vivo approaches. This review summarizes results obtained with the mouse, rat, human, and ovine promoters either by transient transfection assays or by means of transgenic mice.

**Keywords: GnRH receptor, promoter regions, transcription, gonadotrope cell lines, steroidogenic factor 1, homeodomain proteins, transgenic mice**

#### **INTRODUCTION**

Two GnRH systems (GnRH decapeptides and specific receptors) originally existed in the common mammalian ancestor but a large number of extant species have lost either the second type GnRH or its receptor (Stewart et al., 2009). Humans possess GnRH-II but lack a functional type-II receptor whereas mouse and rats lack both of them. In this paper, we will be dealing with the only GnRH system that has been proved to be involved in the reproductive hormonal cascade and that was retained in either humans or the two rodent models. The pituitary GnRH receptor (GnRHR) plays a central role in mammalian reproductive function since it establishes a unique molecular link between its ligand, the decapeptide GnRH originating from the hypothalamus, and the gonadotrope cells in the anterior pituitary. The hypothalamic GnRH is released into the portal hypophyseal vasculature in a periodic manner through a pulse generator that, likely by means of both gap junctions and voltage-gated calcium channels, coordinates the activity of individual neurons diffusely distributed in the hypothalamus (Vazquez-Martinez et al., 2001). Within the anterior pituitary, GnRH binds to specific high-affinity receptors present at the surface of the gonadotrope cells and induces increase in the synthesis and pulsatile release of the gonadotropins, luteinizing and follicle-stimulating hormones (LH and FSH). The pituitary hormones, which are composed of a common α- and distinct β-subunits, then enter the systemic circulation and, *via* specific receptors, modulate gonadal functions including gametogenesis, steroidogenesis, and ovulation. The amplitude and frequency of GnRH pulses relayed by the GnRHR appear critical in the development of the reproductive function, in the onset of puberty and throughout the menstrual or estrous cycle.

#### **INTRACELLULAR SIGNALING AND GnRHR-MEDIATED EFFECTS**

The activation of GnRHR may trigger several intracellular signaling pathways in gonadotrope cells depending on the cellular context. Two mouse gonadotrope tumor-derived cell lines expressing the GnRHR, αT3-1 and LβT2 cell lines, have been used as homogenous cell models to study gonadotrope function (Windle, et al., 1990; Alarid et al., 1996; Thomas et al., 1996; Turgeon et al., 1996). In both cell lines, GnRH activates the protein kinase C (PKC)-dependent signaling pathway through coupling to G proteins of the Gq/G11 family (see review in Anderson, 1996). In primary culture of rat pituitary cells under sustained GnRH stimulation, a cAMP/PKA pathway is preferentially recruited (Garrel et al., 2010). In the gonadotrope-derived LβT2 cell line, sustained stimulation of GnRHR activates the cAMP signaling pathway through PKC δ and ε (Larivière et al., 2007). Also, in these cells, GnRHR signaling has been shown to further involve the three mitogen-activator protein kinases (MAPK) subfamilies (Liu et al., 2002). In non-gonadotrope cells like insect or rat lactosomatotrope cells stably transfected with GnRHR, the signaling mechanisms may involve the PKA pathway *via* Gs or Gi (Stanislaus et al., 1996; Delahaye et al., 1997; see review in Kraus et al., 2001). GnRHR activation generates several intracellular processes in the gonadotropes leading to enhanced transcription of the common α and specific LH and FSHβ subunit genes and secretion of these gonadotropins. Among these processes have been described activation of the neuronal nitric oxide

synthase, induction of *cfos* expression, increase in the early growth response protein 1, increase in mRNA for annexin 5 and activation of translation and phosphorylation of the pituitary adenylate cyclase activating polypeptide (PACAP) type 1 receptor (see review in Garrel et al., 1995, 1997, 1998; Lozach et al., 1998; Brown and McNeilly, 1999; Sosnowski et al., 2000; Bachir et al., 2001, 2003; Duan et al., 2002; Kawaminami et al., 2002; Liu et al., 2002; Larivière et al., 2008). A global profile of genes regulated by GnRH in the LβT2 gonadotrope cell line was established through microarray analysis showing that more than 200 genes are either up- or down-regulated after GnRH agonist treatment (Kakar et al., 2003). The amplitude of these events is strictly dependent on the number of GnRHR molecules at the surface of the pituitary gonadotropes that is itself dependent, at least in part, on the transcriptional level of GnRHR gene (*Gnrhr*).

#### **GENE STRUCTURE AND PATHOLOGIES ASSOCIATED WITH MUTATIONS AFFECTING THE HUMAN GENE**

The cloning of its encoding cDNA, firstly in the mouse and subsequently in other mammalian species, revealed that GnRHR is 325–328 amino acid long (327 in mouse and rat, 328 in human) and confirmed that this receptor belongs to the serpentine family of transmembrane G protein-coupled receptors (review in Norwitz et al., 1999a and references therein). The cDNA probes have subsequently been used to characterize the gene and its chromosomal localization in various mammalian species. *Gnrhr* is composed of three exons and two introns and is approximately 15–31 kb in size depending on the species (**Figure 1**). It is localized to chromosome 4q21.2 in human, chromosome 4, 5, 6, or 8 in bovine, murine, ovine, and porcine species, respectively (Fan et al., 1994; Kaiser et al., 1994; Morrison et al., 1994; Kakar and Neill, 1995; Kottler et al., 1995; Leung et al., 1995; Montgomery et al., 1995; Connor et al., 1999; Jiang et al., 2001). Cloning of the human *GNRHR* cDNA has led to the identification of mutations in the coding sequence that are associated with variable clinical features ranging from partial to complete hypogonadotropic hypogonadism (Achermann et al., 2001; de Roux and Milgrom, 2001). Two separate groups initially reported loss-offunction mutations in the *GNRHR* gene in patients with isolated hypogonadotropic hypogonadism (de Roux et al., 1997; Layman et al., 1998). A natural knockout of the human *GNRHR* resulting from aberrant splicing that eliminates exon 2 and creates a frame shift in the coding sequence was reported (Silveira et al., 2002). The woman affected by this homozygous mutation presented with primary amenorrhea, absence of gonadotropin pulsatility and did not respond to exogenous pulsatile or acute GnRH administration. To date, at least 20 additional mutations in *GNRHR* have been identified in patients with sporadic and familial isolated hypogonadotropic hypogonadism (Chevrier et al., 2011; Noel and Kaiser, 2011). These data again lay emphasis on the predominant role of *GNRHR* in the development of reproductive function in humans.

# **TISSUE- AND CELL-SPECIFIC EXPRESSION OF THE GnRHR**

The isolation of the cDNA also provided the basic tools for initiating studies on the regulation of *Gnrhr* expression by *in situ* hybridization, northern blotting, and semi quantitative RT-PCR. The presence of transcripts for the GnRHRs have notably been observed in brain regions previously shown to bind radiolabeled GnRH, thus confirming that *Gnrhr* is expressed in non-pituitary tissues, notably in the hippocampus and hypothalamus (see review in Jennes et al., 1997). *Gnrhr* is also expressed in granulosa cells of atretic follicles in the ovary, in Leydig cells in the testis, in prostate, breast, and placenta (Stojilkovic et al., 1994). In the pituitary, the expression of *Gnrhr* is restricted to the gonadotrope cells despite the common origin of the six different secreting cell types that composed the anterior pituitary (Scully and Rosenfeld, 2002).

"fendo-03-00162" — 2012/12/11 — 20:28 — page 2 — #2

(+523/+738).

regions (UTR) are designated by thick black lines whereas the two introns are

These diverse and discrete sites of expression raise the question of the mechanisms that determine the tissue-specific expression of *Gnrhr*. In addition, *Gnrhr* is regulated by extracellular signals, the first of them being GnRH itself. Endocrine factors such as estradiol (E2), progesterone, or testosterone also regulate the levels of *Gnrhr* mRNA (Stojilkovic et al., 1994). Altogether, these data led several groups to investigate the mechanisms that underlie the tissue-specific expression and regulation of *Gnrhr* (see review in Hapgood et al., 2005).

To this aim, the 5 regulatory sequences of the rat, mouse, and human receptor genes have been isolated and characterized essentially by transient transfection assays and gel-shift analyses. A limited number of studies have also been conducted *in vivo* through the elaboration of transgenic mouse lines harboring a reporter gene under the control of the mouse, rat, or ovine promoter. This review focuses on the present state of knowledge related to the cell-specific and regulated activity of the promoter of the *Gnrhr in vitro* and *in vivo*.

# *IN VITRO* **ANALYSIS OF THE MOUSE, RAT, AND HUMAN PROMOTERS**

The tissue-specific and regulated activity of the rodent promoters have been evaluated using three main cellular models; the αT3-1 and LβT2 mouse gonadotrope-derived cell lines that are commonly used for studying the expression of marker genes of the gonadotrope lineage (Windle, et al., 1990; Alarid et al., 1996; Thomas et al., 1996; Turgeon et al., 1996) and a rat cell line of lactosomatotrope origin, the GGH(3) cells which were obtained by stable transfection with GnRHR of the tumor-derived GH3 cell line (Janovick et al., 1994; Kaiser et al., 1994; Stanislaus et al., 1994). αT3-1 cells express early marker genes of the developing gonadotrope in mouse, the gene encoding the common α-subunit of the three pituitary glycoprotein hormones (*Cga*) detected at embryonic day 11.5 (E11.5), *Gnrhr* detected at E13.5 and steroidogenic factor 1 (*Sf1*, formally nuclear receptor subfamily 5, member 1, NR5A1) detected at E14.5. The LβT2 cell line further expresses later marker genes, namely those encoding the β-subunits of LH (*Lhb*), from E16.5 onward, and FSH (*Fshb*), from E17.5. These data together with the way by which these cell lines were generated – by targeted oncogenesis using the promoter of *Cga* and *Lhb*, respectively – strongly suggest that the αT3-1 cell line may be representative of the gonadotrope lineage at early developmental stages, between E14 and E16.5. In contrast, the LβT2 cell line would be derived from cells at later stage, beyond E17.5 (Ingraham et al., 1994; Japón et al., 1994; Granger et al., 2004). It has to be emphasized that GGH3 cells express different tissue-specific transcription factors such as pituitary-specific transcription factor 1 (PIT1 formally POU domain class 1 transcription factor 1, POU1F1) that directs the expression of *Prl*, *Gh*, and *Tshb* (Bodner et al., 1988; Ingraham et al., 1988; Li et al., 1990; Ngô et al., 1995). Also, gonadotrope and lactosomatotrope cells likely express ubiquitous transcription factors at different levels. Moreover, as already mentioned above, the GnRHR seems to be coupled to slightly different signal transduction pathways in these cell lines. Consequently, the results obtained with these cellular models may reflect these major differences.

# **ELEMENTS INVOLVED IN THE GONADOTROPE-SPECIFIC ACTIVITY OF THE MOUSE AND RAT** *Gnrhr*

## *The mouse promoter (Figure 2A)*

*SF1, AP1, and GRAS elements.* The 1.2 kb promoter of the mouse *Gnrhr* was the first to be isolated and characterized in 1994 by Albarracin et al. (1994) Shortly after, a 1.9 kb mouse promoter was further isolated and functionally characterized (Clay et al., 1995). Both studies allowed identification of multiple transcription start sites within the first 100 bp immediately upstream from the ATG codon, the major one being located 62 nucleotides from the translation start codon. This transcription start site was referred to as position +1 in papers dealing with the mouse promoter. However, owing to the presence of multiple transcription start sites in GnRHR promoters and in order to facilitate the comparison between promoters of different species, we chose to use the translation start site as the position +1 in this review. Consequently, numbering in this review differs by +63 nucleotides from that currently used for the mouse promoter. This core promoter domain does not exhibit any typical CAAT or TATA boxes close to the transcription start sites. Both studies also demonstrated that the highest levels of promoter activity after transient transfection were obtained in mouse αT3-1 gonadotrope-derived cells. In rat lactosomatotrope cell line (GH3) or human JEG-3 placental cells, the activity of the mouse promoter was considerably weaker and represented only 17 and 5% of that observed in αT3-1 cells, respectively. Progressively, it appeared that all the positive regulatory elements involved in the tissue-specific activity of the mouse promoter in gonadotrope cells were confined into the 500 bp proximal region (Duval et al., 1997a,b). Three distinct motifs making up a tripartite basal enhancer are essentially required: a SF1 (NR5A1) response element (−243 5- - TGGCCTTCA-3- −235), a canonical activating protein 1 (AP1) motif (−336 5- -TGAGTCA-3- −330), and a novel element termed GnRHR activating sequence (GRAS; −391 5- -CTAGTCACAACA-3- −380; **Figure 2**). Mutation of all three elements fully abrogates promoter activity. The identity of SF1 was further corroborated by gel-shift and antibody super-shift assays. Using similar approaches, the AP1 complex has been found to involve factors likely belonging to the FOS and JUN families. The nature of the factors that bind the GRAS element were also deciphered. They consist of SMAD3 (mothers against decapentaplegic homolog 3) which interacts with SMAD4 and bind to the 5 end of the GRAS motif, AP1 that binds within the middle of the motif, and FOXL2, a member of the forkhead family, that interacts with the 3 end of the GRAS motif. GRAS is thus a composite regulatory element whose functional activity is dependent on a multi-protein complex (Ellsworth et al., 2003a).

*Homeobox transcription factors.* Several homeobox factors further participate in the combinatorial code that directs gonadotrope-specific expression of the mouse *Gnrhr* promoter. The paired-like homeodomain transcription factor 1 (PITX1) belongs to the expandingfamily of bicoid-related vertebrate homeobox genes. It appears very early, being involved in the specification of the adenohypophyseal placode and its expression in the anterior pituitary persists afterward until adulthood (Drouin et al., 1998; Schlosser, 2006). PITX1 regulates different marker genes in

"fendo-03-00162" — 2012/12/11 — 20:28 — page 3 — #3

"fendo-03-00162" — 2012/12/11 — 20:28 — page 4 — #4

the adult pituitary notably *Cga*, *Lhb*, *Fshb*, *Tshb*, *Gh*, *Prl*, *Pomc*, and *Gnrhr* (Drouin et al., 1998). PITX1 activates the *Gnrhr* promoter by interacting with c-JUN in the AP1 complex and by binding to several low affinity response elements scattered within the −370/−326 region of the mouse promoter that includes the AP1 response element (Jeong et al., 2004). Octamer-binding transcription factor 1 (OCT1, formally POU2F1) belongs to the other members of homeodomain transcription factors that regulate the *Gnrhr* promoter activity. This was demonstrated by several convergent approaches including targeted mutagenesis coupled with gel-shift assays and transient transfection in gonadotrope cell lines, chromatin immunoprecipitation (ChIP) assay, and short interfering RNA strategy (Kam et al., 2005). Together with the discovery of the OCT1 response element that contains a core TAAT motif at position −352/−349, a functional binding site for nuclear factor Y (NFY) was identified in the course of this study. It partially overlaps the OCT1 response element. Both transcription factors contribute to the tissue-specific activity of the mouse *Gnrhr* promoter. Several other TAAT motifs are present in the sequence of the mouse promoter. One of them, located at position −164/−159, mediates OCT1 trans-acting stimulation of *Gnrhr* promoter in addition to the distal OCT1 interacting motif (Lents et al., 2009). Additional homeodomain factors are involved in promoter activity. LHX3 that belongs to the LIM homeodomain (LIM-HD) proteins is expressed in the early stages of pituitary ontogenesis at E9.5 and its expression persists in the adulthood (Ericson et al., 1998). Similarly to PITX1, LHX3 is involved in the expression of several pituitary marker genes. An ATTA motif located at −360/−357 was found to mediate the mouse *Gnrhr* promoter responsiveness to LHX3. Also, the presence of LHX3 on the mouse promoter was demonstrated *in vivo* by ChIP assay (McGillivray et al., 2005). A complementary study showed that responsiveness to LHX3 was further dependent on a second TAAT motif located only 4 bp downstream (Cherrington et al., 2006). The integrity as well as the specific helical orientation of the two motifs are required for full LHX3 activity.

*E-box transcription factors.* Seven non-canonical E-boxes have been identified along the *Gnrhr* promoter sequence suggesting that basic/helix-loop-helix (b/HLH) transcription factors might be involved in tissue-specific promoter activity. b/HLH transcription factors (124 in numbers in mouse) have been classified into six groups based on their evolutionary relationship, protein structure characteristics, and sequence binding affinities (Li et al., 2006). Four of them, belonging to class A, B, and C, are suspected to modulate the activity of the mouse *Gnrhr* promoter. Indeed, mutation of each E-box in the mouse promoter resulted in full or severe decrease in promoter activity as measured by transient transfection assays using the αT3-1 gonadotrope cell line (Resuehr et al., 2007). One at least out of the three most proximal E-boxes (E-box 1, 2, and 3), binds the CLOCK (circadian locomotor output cycles kaput) transcription factor as demonstrated by ChIP assay. This strongly suggested that the mouse promoter is under the control of CLOCK/BMAL1 (brain and muscle arnt-like protein 1) heterodimer. In addition, transcripts encoding *Drosophila* homolog of period 1 (PER1), PER2, cryptochrome 1 (CRY1), and CRY2 were detected, signifying that dynamic inhibition by CRY/PER of CLOCK/BMAL1-directed gene transcription may take place in this gonadotrope cell line. Although the actors of the molecular loop essential for rhythmic timekeeping are all present, circadian rhythms of *Gnrhr* expression were not observed in cultured gonadotrope αT3-1 cells. Subsequent study considering E-box transcription factors has demonstrated the involvement of two other b/HLH transcription factors, neurogenic differentiation 1 (NEUROD1) and mammalian achaete-scute homolog 1 (MASH1), that both interact with E-box 3, suggesting that the CLOCK/BMAL1 heterodimer bind to either E-box 1 or 2 but not to E-box 3 (Cherrington et al., 2008). Interestingly, the αT3-1 gonadotrope cell line preferentially expresses NEUROD1 whereas LβT2 cells preferentially express MASH1. Consequently, *Gnrhr* E-box 3 binds NEUROD1 from αT3-1 cells, but binds MASH1 from LβT2 cells. This difference may be related to the differentiation state of these cell lines as mentioned above.

#### *The rat promoter (Figure 2B)*

*SF1, CRE, SAP, and AP1 elements in the proximal active domain.* The 5 flanking sequence of the rat promoter has been successively isolated by Reinhart et al. (1997) and by Pincas et al. (1998) in our laboratory. Comparative analyses reveal 82–84% sequence identity with the corresponding 1.9 and 1.2 kb mouse promoter region. Sequence homology is well illustrated by dot matrix pairwise alignment of rat versus mouse sequence (**Figure 3A**). The analysis further reveals full lack of homology beyond 2.1 kb, suggesting that master regulatory domains in rodents are constrained within 2 kb upstream of the ATG codon. Furthermore, the transcriptional start sites appear similarly clustered within the 110 bp proximal region. A first set of major sites is located between −110 and −103, depending on the authors and a second set of both major and minor sites is located around position −30. The rat promoter was proved to be strongly active in αT3-1 and LβT2 gonadotropederived cells as well as in GT1-7 hypothalamic-derived cell line. Within the proximal domain, two elements that are identical to the AP1 (−352/−346) and SF1 (−245/−237) response elements of the mouse promoter are required to mediate gonadotropespecific activity (Pincas et al., 2001a). However, despite these remarkable similarities, the rat promoter displays functional characteristics that distinguish it from the mouse promoter. Two additional response elements are required in the proximal active domain for full cell-specific activity (Granger et al., 2006) a CRE (cAMP response element) element (−110 TGACGTTT−103) that binds CRE binding protein (CREB) with a lower affinity than the consensus TGACGTCA sequence (Pincas et al., 2001b) and an element located at −252/−245 that binds a yet unidentified factor referred to as SAP for SF1 adjacent protein. Despite its immediate proximity with the SF1 response element, the SAP element acts independently of SF1. Furthermore, a GRAS-like element is *Homeodomain factor response elements in the distal and proximal regulatory domains.* The distribution of response elements along the promoter constitutes the most important difference between the rat and mouse species. In contrast with the mouse promoter where all known response elements are clustered within the 500 bp proximal region, an additional regulatory region containing a bipartite enhancer that we named *Gnrhr*-specific enhancer (GnSE) lies on a more distal part of the rat promoter (−1135/−753). Two major response elements located at positions −994/−960 and −871/−862 are responsible for GnSE action (Pincas et al., 2001a; Granger et al., 2006). Interestingly, GnSE activity is dependent on both cellular and promoter contexts. Maximal activity required the presence of the SF1 response element that lies on the proximal promoter domain, 600 bp downstream. Thus, the GnSE belongs to a class of promoter-specific enhancers, capable of acting at a distance as classical enhancers do, but requiring a specific context, in our case, a SF1 element containing promoter. Regarding the proximal GnSE element (−871/−862), it was found to bind the pair of LIM-HD proteins LHX3 and ISL1 (Granger et al., 2006). This was demonstrated by several convergent approaches, notably using overexpression of the LIM-HD proteins and a LHX3 dominant negative in co-transfection assays, gel-shift and super-shift assays, and DNA affinity chromatography. This element displays high sequence identity in the antisense orientation with the −363/−342 region of the mouse promoter that was itself shown to interact with LHX3 (McGillivray et al., 2005; Cherrington et al., 2006). Using antibodies against LHX3 and ISL1 in ChIP assays, we were able to show that the mouse sequence did recruit ISL1 in addition to LHX3 (Schang et al., 2012b) as already demonstrated by McGillivray et al. (2005). Nevertheless, deletion of the LIM-HD response element in the rat promoter does not fully abolish the trans-acting effect of the ISL1/LHX3 pair. This led us to identify the upstream element of the GnSE as a second LIM-HD response element. It displays similar properties and specificity as the first characterized one (Schang et al., 2012b). These elements were then referred to as proximal and distal LIM-HD response elements, P-LIRE and D-LIRE, respectively. Finally, recent work from our laboratory has permitted to the identification of a prophet of PIT1 (PROP1)/orthodenticle drosophila homolog 2 (OTX2) response element in the proximal regulatory domain of the rat promoter, located at −368/−357. Interestingly, OTX2 (or OTX1) stimulates promoter activity in synergy with PROP1 by interacting with this element in gonadotrope-derived cell lines. In contrast, in non-gonadotrope cells, notably in the neutral Chinese hamster ovary (CHO) cell line, OTX2 acts synergistically with CREB through another OTX2 response element involving a ATTA core motif located at position −163/−160, close to the CRE element (Schang et al., 2012a).

#### *Negative regulatory elements in the proximal part of the mouse promoter*

"fendo-03-00162" — 2012/12/11 — 20:28 — page 5 — #5

As underlined above, the GH3 lactosomatotrope-derived cell line expresses the mouse promoter at a low but detectable level. The

"fendo-03-00162" — 2012/12/11 — 20:28 — page 6 — #6

**mammalian species. (A)** Dot matrix representation of pairwise alignment of rat versus mouse (left panel) or human (right panel) sequences. A strong sequence conservation between mouse and rat 5 upstream flanking regions extends over 2 kb whereas that between rat (or mouse) and human sequences is limited to a proximal 0.5 kb region upstream of the ATG codon. Relaxed sequence identity between rat and human sequences is further observed between 0.5 and 1.2 kb upstream of the ATG codon. **(B)** Multiple

identity are highlighted in gray. Some motifs of interest are boxed in the sequences of the species where they have been characterized (e.g., SURG1 in the mouse, AP1 in rat and mouse). Note that the SF1 response element characterized in the rat and mouse promoter sequences (−243/−235) is particularly well conserved among species contrasting with the human SF1 response element (−134/−142). The latter appears to be shared by bovine, porcine, ovine and canis species but not by rat and mouse.

stably transfected GGH(3) cell, first designed in order to evaluate GnRH action, was thereafter used to identify the elements that allow constitutive expression in this non-gonadotrope, SF1 lacking cell line (Maya-Nunez and Conn, 1999). A CRE element (−109 5- -TGACGTTT-3- −102) contributed to basal promoter activity, suggesting that constitutive activation of the PKAdependent signaling pathway may also participate in cell-specific activity in pituitary gonadotropes as well as in other GnRHR expressing tissues. Importantly, several negative regulatory regions were identified (−213/−207; −268/−259; −354/−342) that may serve to repress the activity of the *Gnrhr* in non-gonadotrope pituitary cells. Among them the −268/−259 (5- -TCTGCTGA-3- ) region is strictly conserved among bovine, porcine, ovine, canis, human, rat, and mouse sequences (**Figure 3B**). However, the inhibitory effects of these sequences are modest and cannot account for the whole lack of *Gnrhr* expression in nongonadotrope pituitary cells. Furthermore, they partially overlap regulatory regions characterized as positive in a gonadotrope cell context.

#### **RESPONSE ELEMENTS INVOLVED IN TISSUE-SPECIFIC MOUSE AND RAT PROMOTER ACTIVITY ARE ALSO INVOLVED IN REGULATION BY EXTRA-CELLULAR SIGNALS**

It is often noticed that tissue-specific transcription factors participate either directly or as co-regulators in the modulation of promoter activity by extra-cellular signals. This is also true for the rodent *Gnrhr* since, as we shall see in more detail below, SF1 is involved in PACAP- and PKA-regulated activity,AP1 in the homologous up-regulation by GnRH, and GRAS in the up-regulation by activin.

# *GnRH homologous up-regulation in gonadotrope-derived cells (Figure 4)*

Two independent studies using transient transfection in αT3-1 cells demonstrated a time- and dose-dependent increase in the activity of the mouse *Gnrhr* promoter under GnRH treatment (Norwitz et al., 1999b; White et al., 1999). By using 3 and 5- deletion analyses combined with functional transfection studies, White et al. (1999) showed that a 54 bp fragment extending from −370 to −326 and thus including the AP1 element was sufficient to mediate GnRH response when placed upstream of a minimal heterologous promoter. Further analysis by Norwitz et al. (1999a) using linker scanning mutagenesis led to identifying two sequences of 8 bp, SURG-1 (sequence underlying responsiveness to GnRH-1; −354/−347) and SURG-2 (−338/−331) that are necessary for full GnRH responsiveness (**Figure 3B**). The SURG-2 element corresponds to the AP1 element and is absolutely required since mutations within this sequence totally abolished GnRHinduced stimulation of promoter activity whereas mutations within the SURG-1 element (5- -GCTAATTG-3- ) only attenuated the response. Based on targeted mutagenesis of each of the GRAS, AP1, and SF1 elements that compose the tripartite basal enhancer, the study byWhite et al. (1999) has led to a rather similar conclusion regarding the AP1 element involvement. Both studies, using antibodies directed against conserved protein domains of either the JUN or FOS families in electrophoretic mobility shift assays strongly suggest that an AP1/DNA complex mediates GnRH homologous up-regulation. The MAPK kinase (MEK1/2) inhibitor PD98059 abrogates both GnRH-induced stimulation of promoter activity and GnRH-induced phosphorylation of extracellular signal-regulated kinases (ERK) 1 and 2 (White et al., 1999) and of c-Jun N-terminal kinase (JNK; Ellsworth et al., 2003b). GnRH induction of mouse *Gnrhr* promoter activity may thus require functional activation of a PKC-dependent pathway that involves the ERK- or JNK-signaling cascade or both and ultimately converges to AP1. However, accurate analysis of the intracellular events induced by GnRH treatment of αT3-1 cells strongly suggested that the JNK-signaling cascade is predominantly involved (Ellsworth et al., 2003b). Accordingly, in pituitarytargeted ERK1/2 double-knockout mice, *Gnrhr* expression was not altered as compared with wild-type animals (Bliss et al., 2009), suggesting that ERK signaling is not primarily involved in gonadotrope-specific expression of the *Gnrhr*. Moreover, the activation of the JNK-signaling cascade induced post-translational modifications of FOSB and JUND that led to increased binding of both factors on the SURG-2, the AP1 response element of the *Gnrhr* promoter (Ellsworth et al., 2003b). More recently, the factors that interact with SURG-1 have been identified as OCT1 and NFY. They are not only involved in the tissue-specific expression as mentioned above but also in GnRH homologous up-regulation. ChIP assays demonstrated that both factors are recruited on the mouse promoter in a time-dependent manner after GnRH treatment (Kam et al., 2005). It is important to point out that, in the context of the 600 bp mouse promoter, the adenylate cyclase activator forskolin inhibits GnRH-induced stimulation, suggesting a cross-talk between these two pathways (White et al., 1999). However, in the context of the chimera promoter made up with the mouse −370/−326 bp promoter fragment linked to the minimal GH heterologous promoter, forskolin is inefficient even if the GnRH response is maintained. This may indicate that the cross-talk takes place at the promoter level and requires response element(s) that are present within the 600 bp mouse promoter fragment and absent from the −370/−326 bp promoter fragment, notably the CRE element located at −109/−102 bp.

*GnRH homologous up-regulation in non-gonadotrope cells (Figure 4)* The GnRHR is known to be expressed in several extra-pituitary tissues, notably in some areas of the brain, as well as in gonads and placenta. In this regard, it is important to consider the modalities of expression and regulation of the *Gnrhr* in cells distinct from gonadotropes. The availability of the stably transfected GGH3 cell line has permitted evaluating GnRH action on the transfected mouse promoter in a non-gonadotrope context. Consistent with the data obtained with αT3-1 cells, the activity of the mouse promoter was significantly increased in GGH3 cells following GnRH treatment (Lin and Conn, 1998). Subsequent analysis suggested that GnRH action is likely mediated by both PKA- and PKC-dependent signaling pathways (Lin and Conn, 1998, 1999; Maya-Nunez and Conn, 1999, 2001). Accordingly, GnRH stimulation was partly prevented by the adenylate cyclase inhibitor SQ22536 while treatment with the PKC inhibitor GF109203X led to nearly complete blockade. These data are in accordance with other results obtained using cholera toxin or cAMP analogs as well as following treatment with phorbol ester (Lin and Conn,

"fendo-03-00162" — 2012/12/11 — 20:28 — page 7 — #7

"fendo-03-00162" — 2012/12/11 — 20:28 — page 8 — #8

1998, 1999). It is important to specify that all these treatments resulted in the stimulation of the MAPK pathway in GGH3 cells (Han and Conn, 1999). However, agents that inhibit the activation of the MAPK pathway such as the drug PD98059 or ERK1 and ERK2 mutants together increased both tissue-specific and GnRH-stimulated promoter activity. Conversely, transfection of

vectors expressing agents that mimic the activation of the MAPK pathway such as Raf1 or wild-type ERK1 and ERK2 inhibited tissue-specific and GnRH-stimulated promoter activity. Altogether, these data indicate that activation of the MAPK pathway is inhibitory in the lactosomatotrope GGH3 cell line whereas it is stimulatory in gonadotropes. Furthermore, GnRH action in

GGH3 cells resulted in two opposite effects on the activity of the mouse *Gnrhr* promoter: a negative effect through the activation of the MAPK pathway and a positive one through the activation of PKA- and PKC-dependent pathways. The CRE element (−109/−102) that is involved in tissue-specific activity seems to be also involved in the GnRH and PKA-dependent regulation. The stimulatory effect of GnRH was indeed partially abolished and that of cAMP analog fully abrogated by targeted mutations of the CRE element. It is then rather probable that GnRH action also occurs through the SURG2 element in GGH3 cells.

# *Activin up-regulation*

Activin, a member of the transforming growth factor-β superfamily, has first been shown to enhance the GnRH-induced FSH and LH release and to enhance the rate of GnRHR synthesis in rat pituitary cells (Braden and Conn, 1992; Weiss et al., 1993). This led to question its potential action at the transcriptional level. Exogenous activin B treatment of αT3-1 cells was shown to increase *Gnrhr* promoter activity as demonstrated by nuclear run-off assays or after transient transfection of the 1.2 kb mouse promoter linked to the luciferase reporter gene (Fernandez-Vazquez et al., 1996). However, since αT3-1 cells and pituitary gonadotropes both produce activin B, responsiveness to activin has then been mainly monitored by follistatin-induced inhibition of activin action in subsequent analyses (Duval et al., 1999). Using transient transfection in αT3-1 cells, follistatin inhibition of promoter activity was demonstrated to be mediated through the GRAS element. Indeed, mutation of this motif abrogated follistatin inhibition whereas alteration of the SF1 or AP1 elements was minimally efficient. Consistent with these results, three copies of the GRAS element were sufficient to confer activin responsiveness to a minimal heterologous promoter. The GRAS motif also exhibits sequence similarities with the element (GTCTAGAC) that binds the receptor-activated SMAD2/SMAD3 and the common SMAD4 (Stopa et al., 2000), involved in activin signaling (Walton et al., 2012). The autocrine/paracrine action of activin also explains why the GRAS motif has been classified as a paracrine/autocrine cell-specific response element. Comparison with the rat promoter sequence that does not exhibit activin responsiveness led to the characterization of a second region involved in activin-mediated up-regulation of the mouse promoter, a region referred to as "downstream activin regulatory element" or DARE (Cherrington et al., 2005). This 18 bp-long region lies between −366 and −349 and overlaps LHX3, OCT1, and NFY response elements. However, none of these factors seems to be involved in activin responsiveness. The integrity of the two TAAT motifs identified in this sequence was mandatory for functional DARE. Furthermore, appropriate helical orientation of this region relatively to the GRAS element located 20 bp downstream is required suggesting that factors that bind to DARE interact with GRAS binding factors (SMAD, AP1, FOXL2; Cherrington et al., 2006).

Norwitz et al. (2002a,b) have investigated the potential synergy between GnRH and activin treatment on the mouse promoter activity. They found that activin was able to increase the stimulatory effect of GnRH by approximately twofold. Interestingly, deletion of the promoter region that contains the GRAS element as well as mutation that alters the 5part of the GRAS element

"fendo-03-00162" — 2012/12/11 — 20:28 — page 9 — #9

abrogated the effect of activin on GnRH stimulation. They also showed that overexpression of SMAD4 together with SMAD2 or SMAD3 increased both tissue-specific promoter activity and GnRH homologous up-regulation. However, it remains to be established if mutation of the presumed Smad binding element (SBE box −393/−386) in the GRAS element may prevent the stimulatory effect induced by SMAD overexpression. Competition and super-shift experiments further indicated that a second factor, belonging to the AP1 family, may bind the 3 part of the GRAS element, suggesting that JUN/FOS related factors and SMAD protein functionally interact.

# *PACAP up-regulation*

Specific type 1 receptors for PACAP are present in gonadotrope cells of the anterior pituitary gland as well as in mouse gonadotrope-derived αT3-1 cells. By transient transfection in αT3-1 cells, Pincas et al. (2001b) provided evidence that PACAP stimulates rat *Gnrhr* promoter activity. The EC50 of this stimulation was compatible with PACAP activation of the cyclic AMPdependent signaling pathway and, consistently, co-transfection of an expression vector expressing the PKA inhibitor caused reduction in PACAP – as well as cholera toxin –stimulated promoter activity. Deletion and mutational analyses indicated that PACAP activation necessitated a bipartite response element that consists of a first region (−272/−237) termed PACAP response element (PARE) I that includes the SF1-binding site and a second region (−136/−101) referred to as PARE II that contains the CRE (TGACGTTT). Gel-shift experiments revealed that SF1 and SAP together bind to PARE I while a protein immunologically related to the CREB interacted with PARE II. Altogether, these findings indicated that PACAP regulates the rat *Gnrhr* at the transcriptional level in αT3-1 cells. Interestingly, the mouse promoter was similarly proved to be regulated by PACAP (Sadie et al., 2003). However, the response elements were not identified, nor were those involved in the PKA-induced stimulation of the mouse promoter. Notably, the ability of the CRE located at −109/−102 (i.e., equivalent to the rat promoter one) to mediate the PKA-dependent stimulation of the mouse promoter was not assessed. Nevertheless, two other response elements were shown to modulate PKA responsiveness (Sadie et al., 2003), the SF1 response element and a newly identified response element located at −15/−7, close to the translation initiation site (ATG). Both elements were able to bind NUR77 (NR4A1) and SF1 and exerted opposite effects. Mutation of the element close to the ATG codon together with overexpression of SF1 dramatically amplified PKA-induced stimulation. In contrast, overexpression of NUR77 repressed PKA-dependent regulation, most probably through the −15/−7 element. PACAP has been shown to stimulate follistatin (*Fst*) gene expression which restrains activin signaling and thus represses *Fshb* and *Gnrhr* expression as well as other activin-responsive genes. Therefore, in mouse gonadotrope cells, the PACAP direct stimulatory effect on *Gnrhr* promoter may be weakened by the indirect inhibitory action of PACAP on activin-stimulated promoter activity. The rat promoter is not submitted to this down-regulation because both the GRAS element and the DARE region are inactive. Whether PACAP regulation also occurs *in vivo* and is a common trait in mammals remains to be elucidated however.

# **FUNCTIONAL CHARACTERISTICS OF THE HUMAN** *Gnrhr* **PROMOTER** *IN VITRO*

The human *Gnrhr* promoter has been independently isolated by two groups (Fan et al., 1995; Kakar, 1997). It displays several features that distinguish it from rodent promoters (Cheng and Leung, 2005). Firstly, a significant sequence identity with the above-described promoters does not extend beyond approximately −1200 bp from the ATG codon (**Figure 3A**). Secondly, in contrast to the rodent promoters, the transcription start sites are located far upstream of the ATG codon and disperse within two domains localized between −578 and −826 and between −1347 and −1751, indicating the existence of two core promoter regions (Fan et al., 1995; Kakar, 1997). A further difference with mouse and rat promoters is the presence of several TATA boxes well scattered among the transcription start sites, suggesting that transcription initiation occurs through different mechanisms in rodent and human.

Given that GnRHR is expressed in several tissues in addition to the pituitary gland, functional studies with the human promoter have been performed in non-pituitary cell lines, notably cells derived from placenta, endometrium, trophoblast, and ovary. This strategy has also been motivated by the absence of pituitary cell line of human origin. These studies led to the discovery of several positive and negative regulatory regions, the efficacy of which being strongly dependent on the considered cell line. They essentially confirmed the existence of distinct promoters within the 2 kb human 5flanking sequence.

# *The proximal core promoter is predominantly active in gonadotrope-derived cells (Figure 5A)*

Following transfection in αT3-1 cells, it appeared that crucial elements are located in the close vicinity of the translation start codon because the −173/+1 deletion abolished promoter activity (Ngan et al., 1999; Kang et al., 2000). Targeted mutagenesis of a SF1 response element located at −142/−134 (5- -CAGGGACAA-3- ) altered overall promoter strength suggesting that it is likely contributing to the gonadotrope-specific activity. The involvement of SF1 has been further evidenced by electrophoretic mobility shift and super-shift assays as well as by overexpression of SF1 (Ngan et al., 1999). However, silencer elements present in the middle part of the human promoter (−1018/−707) weaken promoter strength. Indeed, 5 deletion to −707 increased tissuespecific activity in αT3-1 cells, the deleted promoter (−707/+1) being approximately threefold more active than the entire promoter (−2200/+1; Cheng and Leung, 2001). These data are in agreement with the existence of a proximal core promoter region (−816/−577) previously documented (Fan et al., 1995) and with the presence of a positive regulatory region (−771/−557) identified by 3 deletion analysis (Kang et al., 2000). This also suggests that positive regulatory regions either overlap the core promoter region or are situated downstream from the transcription start sites as is the SF1 element cited above. Subsequent analysis by scanning mutations in αT3-1 cells has allowed precise delimitation of this proximal promoter that extends from −607 to −568. It contains two pyrimidine-rich initiator elements (−602/−597 and −589/−584) that bind similar proteins notably TFIID (Hoo et al., 2003). The distal promoter of the human gene is constitutively active in both gonadotrope and non-pituitary cell lines (**Figure 5B**).

The existence of a distal promoter active in non-pituitary cell lines has been strongly suggested in the early study by Kakar (1997). In this work, it is indeed shown that the −2300/−1020 human flanking sequence is highly active in human endometrial (HEC-1A) and breast tumor cell (MCF-7) lines. This upstream promoter was further characterized in either gonadotrope or nonpituitary cells by Ngan et al. (2000) and Cheng et al. (2001a), respectively. It appears to be less active in αT3-1 pituitary than in JEG-3 placental cells (Cheng et al., 2001a) and thus contrasts with the core proximal promoter that displays opposite characteristics. Also, the distal promoter seems to be more ubiquitous than the proximal promoter since it was active in gonadotrope, ovarian, endometrial, mammary, placental, and kidney cells. The in-depth analysis of the distal promoter in gonadotrope-derived αT3-1 cells (Ngan et al., 2000) and placental JEG-3 cells (Cheng et al., 2001a) reveals cell-specific differences.

In gonadotrope-derived cells, the active elements of the upstream promoter are located within a −1705/−1674 region and are not functional in the reverse orientation. They consist of a pyrimidine-rich initiator (−1685/−1676) and additional upstream motifs (−1705/−1687) that include a CCAAT box. Multiple proteins ranging from 40 to 54 K are capable of binding this upstream promoter. Interestingly, a downstream silencer element (−1674/−1577) prevents upstream promoter activity, at least in gonadotrope cells. This silencer is efficient only when positioned downstream of the promoter in either reverse or forward orientation. Two sequence-specific repressor motifs of 28 and 8 bp (−1670/−1643 and −1630/−1623) are essential for silencer activity as evidenced by linker-scanning mutagenesis and gel mobility shift assays (Ngan et al., 2001).

In JEG-3 and IEVT placental cell lines, a similar organization seems to occur, combining positive regulatory sequence and downstream silencer, however with different localization and distinct response elements (Cheng et al., 2001a). The placental cell-specific upstream promoter is located within a −1737/−1346 region and thus overlaps the gonadotrope cell-specific promoter. In JEG-3 cells, four motifs related to OCT1 (−1718/−1710), CRE (−1649/−1641), GATA (−1602/−1597), and AP1 (−1518/−1511) response elements are contributing to promoter activity. The characterization of these motifs has been further assessed by gel mobility shift and antibody super-shift assays. The downstream silencer is located between −1017 and −771 and abrogates upstream promoter activity in a manner similar to what is observed with the gonadotrope promoter. The silencer was further characterized in the context of the −1300/−1018 promoter described below. Under these conditions, it appeared that most of the silencing activity of this negative regulatory element resides in an octamer responsive sequence activated by OCT1 (Cheng et al., 2002b).

# *Distal promoter usages in granulosa and neuronal cell lines*

"fendo-03-00162" — 2012/12/11 — 20:28 — page 10 — #10

A third distal promoter was characterized in human ovarian granulosa-luteal cells. It extends from −1300 to −1018 and exhibited the highest promoter activity in SVOG-4o and SVOG-4m immortalized granulosa-luteal cells (Cheng et al., 2002a).

"fendo-03-00162" — 2012/12/11 — 20:28 — page 11 — #11

However, it was also active in placental, ovarian, and gonadotrope cells. Two CCAAT enhancer binding protein (C/EBP) located at −1244/−1232 and −1157/−1144 and one GATA response element located at −1176/−1168 were required for full promoter activity because simultaneous mutations of these three elements led to more than 75% decrease in promoter activity in granulosa cells. Interestingly, the same mutations induced minimal or no significant decrease in promoter activity in ovarian, placental, and gonadotrope cells suggesting that different response elements and different transcription factors are involved in these cell contexts.

Human GnRHR was further expressed in the human cerebellar medulloblastoma cell line TE-671. In these neuronal cells, promoter activity was dependent on an upstream region located between −2197 and −1018. Important *cis*-acting regulatory elements were identified at −1300/−1018 and −2197/−1900, as deletion of either region induced a dramatic decrease in promoter activity (Yeung et al., 2005).

#### **REGULATION OF HUMAN PROMOTER ACTIVITY BY EXTRACELLULAR SIGNALS (FIGURE 5)**

#### *Homologous up- and down-regulation by GnRH and the PKC-dependent pathway*

Surprisingly, when compared to rodents, a 6–24 h-long treatment with 0.1 μM GnRH reduced activity of the 2.3 kb human *GnRHR* promoter by 31–71% in αT3-1 cells. This inhibitory effect was mimicked by phorbol ester and prevented by GnRH antagonists or specific PKC inhibitors (Cheng et al., 2000) strongly suggesting that GnRH action is mediated through receptor-dependent activation of the PKC pathway. Subsequent 5 deletion analyses combined with site-directed mutagenesis revealed that an AP1 related element located at −1000/−994 (TTAGACA) is responsible for the homologous down-regulation of promoter activity by forming an AP1 complex involving the c-JUN and c-FOS protooncogenes. This AP1 element is unrelated to the AP1/CRE element involved in tissue-specific expression. It is worth stressing that similar GnRH treatment applied to placental JEG-3 or IEVT cell lines led to significant increases in promoter activity, contrasting again with data obtained in gonadotrope-derived cells (Cheng et al., 2000). The mechanisms underlying these positive and negative regulations remain to be elucidated, notably whether differential promoter usage (upstream versus proximal) may be related to such difference in GnRH regulation.

#### *Up-regulation by PACAP and PKA-dependent pathway*

The regulation of the human promoter by PACAP and the PKA-dependent activation pathway has been investigated in two different studies and led to the identification of distinct regulatory domains located within either the upstream or the proximal promoter, respectively (Cheng and Leung, 2001; Ngan et al., 2001). The former study (Ngan et al., 2001) indicated that both the gonadotrope cell-specific upstream promoter as well as the downstream silencer are involved in PACAP and PKA-dependent regulation. In support, PACAP treatment enhanced the formation of a complex with the 28 bp sequence-specific repressor motif (see above). Whether this may contribute to suppress silencer function and in turn to activate promoter remains to be demonstrated. The mechanisms underlying this positive regulation are not yet clearly identified. The latter study by Cheng and Leung (2001) led to the conclusion that PACAP and CREs reside within a 412 bp fragment (−577/−167) of the proximal promoter. Indeed, mutation of the (−568/−561 and −340/−333) AP1/CRE related motifs nearly abrogated the stimulation.

# *Up- and down-regulation by steroid hormones*

Similarly to GnRH, progesterone treatment led to either inhibition or stimulation of promoter activity in gonadotrope versus placental cells, respectively (Cheng et al., 2001b). A progesterone binding site located at −535 to −521 has been shown to mediate, at least partially, both the inhibitory and stimulatory action of progesterone. Overexpression of the human progesterone receptor A and B isoforms increased the negative effect of progesterone in αT3-1 cells, whereas only B isoform mediated its positive action in JEG-3 placental cells. It is important to point out that the A isoform mediated inhibitory action in both cell lines, suggesting that the balance in the expression of A and B isoforms may be critical for GnRHR expression in the pituitary. The involvement of the progesterone receptor in binding the −535/−521 motif has been further substantiated by gel-shift and antibody super-shift assay.

In OVCAR3 ovarian carcinoma and MCF7 breast carcinoma cell lines, E2-induced repression of *Gnrhr* promoter activity through an E2 receptor α (ERα)-mediated mechanism (Cheng et al., 2003). This occurred in a dose- and time-dependent manner, the maximal effect being obtained at 10−<sup>8</sup> M E2 after a 48 hlong treatment (80% reduction of promoter activity). Repression occurred *via* an AP1-like element located at position −130/−124. This element was able to bind c-FOS and c-JUN but not ERα as demonstrated by EMSA. DNA binding by ERα was not needed for the E2 repression of *Gnrhr* promoter activity. Further experiments strongly suggested that recruitment of AP1 and ERα to CREB binding protein (CBP) was mutually exclusive leading to inhibition of AP1-stimulated promoter activity by E2.

# *IN VIVO* **ANALYSIS OF THE MOUSE, OVINE, AND RAT PROMOTERS IN TRANSGENIC MICE**

Since *in vitro* studies of promoter functions often suffer from limitation of the cellular models used in transient transfection analyses, the next and complementary approach consists in analyzing promoter activity in an *in vivo* context through the generation of transgenic mice.

Several studies have been published that use this procedure. The initial study by McCue et al. (1997) demonstrated the homologous up-regulation of *Gnrhr* promoter activity by GnRH. A 1900 bp fragment corresponding the mouse *Gnrhr* promoter fused to the luciferase reporter gene has been microinjected in mouse embryos and mice harboring the transgene were selected. Analysis of luciferase activity in various tissues showed that the fusion gene was expressed essentially in the pituitary gland, to a 10-fold lower level in the brain and to a limited extend in the testis. These data are in agreement with those deduced from autoradiographic studies using 125I-labeled GnRH or *in situ* hybridization (Jennes et al., 1988, 1995; Leblanc et al., 1988; Ban et al., 1990; see review in Jennes et al., 1997). Surprisingly, the expression of the transgene was not detected in the ovary contrasting with a number of studies reporting the presence of GnRHR in this tissue, specifically

"fendo-03-00162" — 2012/12/11 — 20:28 — page 12 — #12

in granulosa cells of atretic follicles. Using an antibody directed against GnRH, the authors further showed a decrease in transgene expression in the pituitary gland but not in the brain or testis. The GnRH antibody inhibitory action could be reversed by a GnRH agonist treatment. As expected from *in vitro* studies, the 1900-bp promoter contains essential GnRH response elements. However, E2 treatment of these transgenic mice did not alter the activity of the transgene. This diverges with a number of reports demonstrating that, *in vivo*, E2 increases GnRHR number and steady-state level of *Gnrhr* mRNA in the pituitary (Hamernik et al., 1995; Turzillo et al., 1995; Kirkpatrick et al., 1998).

The same group has thus reinvestigated this regulation with novel lines of transgenic mice harboring a fusion construct containing the ovine receptor gene promoter linked to the luciferase reporter gene (Duval et al., 2000). Using either a 9.1 or a 2.7 kb promoter, they first showed that the ovine transgene displayed the same tissue-specific expression as the mouse promoter. The ovine transgene was indeed consistently expressed in the pituitary, brain, and gonads across the three analyzed transgenic lines. Pituitary expression was higher in females than in males whereas ovarian expression was significantly lower than the testicular one. In agreement with results obtained with the mouse promoter, immunoneutralization of GnRH reduced pituitary-specific expression of the luciferase reporter gene in ovariectomized mice, expression that could be restored by E2 treatment. Similar results were observed in castrated, male transgenic mice. Finally, combined treatments with GnRH agonist and E2 increased pituitary expression of the luciferase fusion gene to an extent greater than the sum of each individual treatment, suggesting synergistic activation of the ovine promoter.

The third study involving transgenic mice was different in at least two facets. Firstly, in this case, the reporter gene was the simian virus 40 (SV40) large T antigen and, secondly, the mouse promoter was the shortest used since it was 1.2-kb long only (Albarracin et al., 1999). With this fusion construct, the resulting heterozygote transgenic female mice were infertile preventing the generation of homozygote animals. Interestingly, all transgenic animals developed large intracranial tumors by 4–5 months of age that were of pituitary origin. As evidenced by *in situ* hybridization, these tumors expressed the marker genes of the gonadotrope phenotype, i.e., GnRHR, glycoprotein hormone α-subunit as well as FSHβ and LHβ subunits. The large T antigen was also expressed in the pituitary tumors but neither in the gonads or in the brain. This may indicate that the 1.2-kb, as compared with the 1.9-kb promoter fragment, is not able to direct transgene expression to sites other than the pituitary, notably brain and gonads. Surprisingly, the expression of the SV40 large T antigen transgene driven by the GnRHR gene promoter results in female-specific infertility due to disruption of gonadotropin production and secretion even before tumor development (Jeong et al., 2009).

In our laboratory, the 3.2 kb rat promoter was placed upstream of the human placental alkaline phosphatase reporter gene (*ALPP*) and inserted into the mouse genome. The resulting transgenic mice expressed the ALPP exclusively in gonadotropes within the pituitary gland (Granger et al., 2004). Transgene expression was detected within the developing pituitary at E13.5 in some of the cells that also expressed the *Cga*. It showed that *Gnrhr* is the

"fendo-03-00162" — 2012/12/11 — 20:28 — page 13 — #13

earliest marker of gonadotrope cell differentiation, before SF1 that is detected at E14.5. At E15.5, ALPP expression was localized in LHX3- and ISL1-immunoreactive positive cells in agreement with *in vitro* data showing that these LIM-HD proteins are involved in rat and mouse promoter activity. Importantly, the 3.2 kb promoter was also able to direct *ALPP* expression to several brain areas, in particular the hippocampus and the lateral septum, in agreement with previous data (Jennes et al., 1988, 1995; Leblanc et al., 1988; Ban et al., 1990; see review in Jennes et al., 1997). In contrast with the pituitary gland, transgene activity in the hippocampus was detected only after birth, increasing gradually until 14–20 days after birth and then remaining at constant level until 60 days after birth (Schang et al., 2011b). This occurred simultaneously in fibers extending from the hippocampus to the lateral septum, suggesting that GnRHR may be involved in post-natal maturation of the septo-hippocampal system. The same time-course was observed by measuring*Gnrhr* mRNA in rat hippocampus by quantitative RT-PCR, indicating that transgene expression directed by the rat promoter recapitulates rat endogenous *Gnrhr* expression in a mouse context. Further experiments using transient transfection of *Gnrhr* promoter luciferase fusion constructs in primary culture of rat hippocampus cell revealed that a RE1-silencing transcription factor (REST also known as NRSF)-like element located at −2.5 kb may be involved in the inhibition of *Gnrhr* gene expression before birth. Finally, GnRH treatment of rat hippocampus primary cell culture led to stimulation of several markers of neurogenesis such as EGR1, synaptophysin, and spinophilin. Using this transgenic mouse model, several additional sites of transgene expression were detected, notably in the oculomotor pathway (Schang et al., 2011a). Interestingly, transgene expression was also detected in two functionally and evolutionary related organs, the pineal gland and the retina. Again, the onset of transgene expression was specific to these tissues, being detected at E13.5 and E17.5 in the pineal gland and retina, respectively (Schang et al., 2012a). In the pineal gland, transgene expression persisted until adulthood whereas it strongly decreased in the retina. During development, transgene expression was the strongest in the neuroblast cell layer of the retina and less marked in the ganglion cell layer. In the pineal gland, transgene expression was constantly observed in approximately 50% of the cells whatever the developmental stage. Finally, transgene expression was strongly detected in the testes of transgenic mice in some Leydig cells (Schang et al., 2011a).

Recent studies involving functional genomic approaches have shown that regulatory domains may be located within the gene locus at several kilobases, either 5 upstream or 3 downstream as well as within intronic regions. This is well demonstrated particularly for genes coding for transcription factors, notably SF1 (Shima et al., 2012 and references therein), or genes encoding for factors involved in early developmental processes such as the LIM-HD proteins LHX3 and ISL1 (Kappen and Salbaum, 2009; Mullen et al., 2012). Conversely, other studies have shown that the most important genetic regulation may be contained within only 1 kb 5 upstream of the transcription start site, including informations regarding epigenetic modifications such as pattern of DNA methylation (Lienert et al., 2011). Furthermore, species-specific informations would be encoded into DNA sequence and appears to prevail over species-specific cellular environment. A human

chromosome introduced into a mouse cell line directs genetic and epigenetic modifications as it does in a human cell context and thus differently from its mouse counterpart (Wilson et al., 2008). Therefore, data obtained using transgenic mice as well as those described above must be considered bearing in mind the fact that all regulatory sequences are not necessarily contained within the 5 upstream region of the gene only, even if relatively short 5- upstream sequences may contain master regulatory informations. This could explain why the human *GNRHR* promoter responded negatively while mouse or rat promoters are stimulated by GnRH.

To overcome these limitations, a double knock-in mice model was recently developed. In this binary genetic approach, the Crerecombinase was inserted into the *Gnrhr* locus whereas the YFP reporter gene, preceded by a floxed stop signal, was inserted into the ubiquitously expressed ROSA26 locus. The Cre-mediated excision of the stop signal led to a constitutive YFP expression in about 15% of the cells in the male anterior pituitary. Furthermore, the fluorescent signal was only colocalized with LHβ and/or FSHβ subunits as demonstrated by immunohistochemistry. These YFP mice have allowed single-cell analysis on living cells which have revealed unexpected heterogeneity in the resting properties of gonadotrope cells as well as in their secretory, electrophysiological, and calcium responses to GnRH (Wen et al., 2008).

# **SF1, A PAN-SPECIES TRANSCRIPTION FACTOR CRITICAL FOR GONADOTROPE-SPECIFIC EXPRESSION**

An obvious conclusion that emerges from this overview of the mechanisms involved in *Gnrhr* expression is the important role of the orphan nuclear receptor SF1. It is undoubtedly involved in the tissue-specific expression of the human, mouse, and rat *GnRHR*. It is also likely involved in the tissue-specific expression of the ovine gene since the promoter is strongly homologous to its human counterpart. SF1 is crucial not only for tissue-specific expression but also for the regulation by extracellular signals as evidenced by PACAP and PKA-dependent regulations of the rat promoter. This may be related to the other important functions of SF1 in the tissue-specific and regulated expression of *Cga* and *Lhb* genes, two other markers of the gonadotrope lineage. Consistent with these data, the SF1 knockout mice exhibit impaired expression of LHβ, FSHβ, and GnRHR in pituitary gonadotropes in addition to other major defects such as agenesis of the ventro-medial hypothalamus, complete adrenal and gonadal agenesis and maleto-female sex reversal with persistence of Müllerian structures in male (Ingraham et al., 1994; Ikeda et al., 1995; Luo et al., 1995). To bypass the influence of hypothalamic deficiencies inherent to this transgenic model, SF1 has been inactivated specifically in the anterior pituitary using the Cre-loxP system (Zhao et al., 2001). These pituitary-specific SF1 knockout mice are sterile and never mature sexually. Consistent with a direct role of SF1 at the pituitary level, FSH and LH are markedly decreased in the pituitary-specific SF1 knockout mice as well as are gonadotropin β-subunit and *Gnrhr* mRNAs as assessed by semi-quantitative RT-PCR.

The important role of SF1 has been further confirmed in humans by the discovery of a heterozygous missense mutation in the DNA binding domain of SF1 in a patient with complete sex reversal, testicular dysgenesis, and adrenal failure (Achermann et al., 1999). However, this patient displayed a relative preservation

"fendo-03-00162" — 2012/12/11 — 20:28 — page 14 — #14

of gonadotrope function. This may be due either to its heterozygous genotype or to a relative autonomy of the gonadotrope function with respect to SF1. In agreement with the latest hypothesis, GnRH treatment can restore LH and FSH secretion in SF1 knockout mice (Ikeda et al., 1995), suggesting that SF1 is not required for the emergence and maintenance of gonadotrope function. Furthermore, some tissues known to express GnRHR such as amygdala and hippocampus are deprived of SF1 (Schang et al., 2011b). This suggests that other factor(s) than SF1 are needed for the tissue-specific transcriptional onset of the *Gnrhr*. To identify this (or these) factor(s), several parameters must be considered. From an ontogenetic point of view, the date of appearance of the *Gnrhr* transcripts must be correlated with the presence of the factor in the fetal pituitary or in fetal tissues that will express the GnRHR in the adult. For instance, binding sites for GnRH have been detected as early as 12–13 days of gestation in the rat developing pituitary and GnRHR mRNA at 14.5 and 15.5 days post-conception in testis and ovary from rat fetuses (Aubert et al., 1985; Jennes, 1990; Botté et al., 1998). This factor should also be able to activate the promoter of the *Gnrhr* in an SF1-independent manner in transient transfection assays.

# **CONCLUSIONS AND PERSPECTIVES**

An increasing number of studies on gene regulations and promoter activities aim to demonstrate the existence of specific regulatory units that bring together several response elements. From one tissue to another, the same elements may bind different factors resulting in tissue- or cell-specific modulations of gene expression. This ability to bind different factors is likely facilitated by sequence degeneracy of the response elements that lie within these regulatory units. Consequently, the degenerated response elements may be less efficient than their canonical counterparts when they are isolated from their promoter context. Efficiency is recovered in the original promoter context through synergistic interactions with other transcription factors belonging to the regulatory unit. Interestingly, this synergy may be attenuated if the wild-type degenerated element is artificially replaced by its canonical counterpart (Ito et al., 2000). In the *Gnrhr* promoters, several elements display such characteristics. In particular, the SF1 binding site and the proximal CRE in the rat and mouse promoter differ from the consensus sequence (TGACGTTT instead of TGACGTCA for the later). In gel-shift experiment, both the SF1 element and the CRE display an apparent weaker affinity for their cognate transcription factors than the SF1 binding site from the rat aromatase gene or the canonical CRE sequence (Pincas et al., 2001a,b). PACAP- and PKA-dependent regulations of the rat promoter are mediated by these two elements that may thus be considered as a cAMP regulatory unit (Roesler, 2000). The SURG-1 and SURG-2 elements establish another functional regulatory unit since they are both required for the GnRH up-regulation of the mouse promoter. The distal GnSE together with the proximal domain containing the SF1 binding site also form a regulatory unit for the tissuespecific activity of the rat promoter. Different regulatory units may also bind different combinations of factors depending on the tissue or the cell type that expresses the *Gnrhr*. We have recently demonstrated that OTX2 interacted with two distinct partners depending on the cell context (Schang et al., 2012a). In a gonadotrope cell context, OTX2 interacted with PROP1 on two adjacent response elements located between −388 and −357, each of them encompassing a core TAAT motif. In a neutral cell context, OTX2 acted in synergy with CREB and formed a regulatory unit that involved CRE, an AP1 response element together with a new OTX2 binding site containing a TAAT core motif located at −163/−160. This motif is thus located 50 bp upstream of the CRE and 180 bp downstream of the AP1 element establishing a regulatory unit extending over 200 bp within the proximal promoter, markedly larger than the PROP1/OTX2 regulatory unit (Schang et al., 2012a). Whether such difference in size of these regulatory units as well as the other cited above is functionally relevant remains to be determined. In this respect, an important point that must be considered is the chromatin structure that depends on epigenetic modifications. At least, two histone H3 modifications are predominantly involved in transcriptional regulation: tri-methylation of Lys 4 of histone H3 (H3K4me3) and tri-methylation of Lys 27 of histone H3 (H3K27me3; see review in Greer and Shi, 2012). They are linked to gene activity and gene silencing, respectively. H3K4me3 is catalyzed by trithorax group (trxG) proteins, whereas H3K27me3 is catalyzed by polycomb-group (PcG) proteins (Schuettengruber et al., 2007; Ku et al., 2008). PcG proteins reside in two main complexes, termed Polycomb repressive complexes 1 and 2 (PRC1 and PRC2). PRC2 catalyzes H3K27me3 and components of PRC1 are recruited to the modified histone leading to ubiquitinylation of histone H2A at lysine 119 which in turn prohibits RNA polymerase II elongation (Simon and Kingston, 2009). This results in stable transcriptional repression. Analysis of H3K4me3 and H3K27me3 modifications by ChIP assay showed that in gonadotrope cell lines, the *Gnrhr* promoter preferentially displayed the active mark whereas in lactosomatotrope or corticotrope cell lines, it displayed preferentially the repressive mark

#### **REFERENCES**


characterization of the 5- -flanking region of the mouse gonadotropinreleasing hormone receptor gene. *Endocrinology* 135, 2300–2306.


(unpublished data). This is well correlated with *Gnrhr* mRNA levels that are not detected in lactosomatotrope and corticotrope cell lines. Constitutive *Gnrhr* transcription is thus likely controlled by mechanisms involving trxG proteins in gonadotrope cells whereas in other pituitary cells, PcG proteins stably repress *Gnrhr* expression. In mammals, composition of trxG and PcG protein complexes displays broad diversity depending on the considered family of genes. A critical point would be to determine what are the main actors involved in histone methyltransferase and demethylase activity regarding the *Gnrhr*. Furthermore, PcG complexes are known to be clustered into subnuclear structures that look like nuclear foci termed polycomb bodies (Pirrotta and Li, 2012). Conversely, active genes are preferentially located in different subnuclear structures referred to as transcription factories that contain active RNA polymerase II or interchromatin granule clusters (also termed nuclear speckles) that contain histone-modifying enzymes, transcription elongation factors, and splicing machinery (Meldi and Brickner, 2011; Spector and Lamond, 2011). In the developing pituitary gland, lineage specification likely requires relocation of marker genes like the *Gnrhr* into these subnuclear structures (Yang et al., 2011). The elucidation of the mechanisms underlying this reorganization would permit to shed new light on *Gnrhr* expression, on gonadotrope cell lineage specification and differentiation, and more generally on organ development.

#### **ACKNOWLEDGMENTS**

This study was supported by grants from the Centre National de la Recherche Scientifique, the Paris 7 University, and the Agence Nationale pour la Recherche (ANR-08-CES-011-03). Anne-Laure Schang is supported by a fellowship from the Ministère de l'Education Nationale, de la Recherche et de la Technologie (MENRT).

and characterization of a rat nitric oxide synthase type I gene promoter that confers expression and regulation in pituitary gonadotrope cells. *Endocrinology* 142, 4631–4642.


"fendo-03-00162" — 2012/12/11 — 20:28 — page 15 — #15

L. (1998). Fetal expression of GnRH and Gnrhrs in rat testis and ovary. *J. Endocrinol.* 159, 179–189.


Cheng, C. K., Yeung, C. M., Chow, B. K., and Leung, P. C. (2002a).

Characterization of a new upstream GnRH receptor promoter in human ovarian granulosa-luteal cells. *Mol. Endocrinol.* 16, 1552–1564.


(1995). Cell-specific expression of the mouse gonadotropin-releasing hormone (GnRH) receptor gene is conferred by elements residing within 500 bp of proximal 5 flanking region. *Endocrine* 3, 615–622.


receptor gene. *Biol. Reprod.* 56, 160–168.


"fendo-03-00162" — 2012/12/11 — 20:28 — page 16 — #16


kinase activation in response to gonadotropin-releasing hormone receptor activation. *Endocrinology* 140, 2241–2251.


hormones. *Brain Res. Mol. Brain Res.* 33, 104–110.


hormone agonist using microarray analysis. *Gene* 308, 67–77.


"fendo-03-00162" — 2012/12/11 — 20:28 — page 17 — #17


chromosome 6 outside of the region of the FecB locus. *Mamm. Genome* 6, 436–438.


A augments GnRH-mediated transcriptional activation of the mouse Gnrhr. *Endocrinology* 143, 985–997.


"fendo-03-00162" — 2012/12/11 — 20:28 — page 18 — #18

modulated by steroidogenic factor-1 and Nur77. *Endocrinology.* 144, 1958–1971.


translation in pituitary gonadotrope cells by gonadotropin-releasing hormone. *Mol. Endocrinol.* 14, 1811– 1819.


TGF-beta response element of the promoter requires functional Smad binding element and E-box sequences for transcriptional regulation. *J. Biol. Chem.* 275, 29308–29317.


(1993). Perifusion of rat pituitary cells with gonadotropin-releasing hormone, activin, and inhibin reveals distinct effects on gonadotropin gene expression and secretion. *Endocrinol-*


"fendo-03-00162" — 2012/12/11 — 20:28 — page 19 — #19

in human neuronal cells. *Mol. Hum. Reprod.* 11, 837–842.

Zhao, L., Bakke, M., and Parker, K. L. (2001). Pituitary-specific knockout of steroidogenic factor 1. *Mol. Cell. Endocrinol.* 185, 27–32.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 13 September 2012; paper pending published: 12 November 2012; accepted: 28 November 2012; published online: 13 December 2012.*

*Citation: Schang A-L, Quérat B, Simon V, Garrel G, Bleux C, Counis R, Cohen-Tannoudji J and Laverrière J-N (2012) Mechanisms underlying the tissue-specific and regulated activity of the Gnrhr promoter in mammals. Front. Endocrin. 3:162. doi: 10.3389/fendo. 2012.00162*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Schang, Quérat, Simon, Garrel, Bleux, Counis, Cohen-Tannoudji and Laverrière. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Sensitivity of cholecystokinin receptors to membrane cholesterol content

# *Aditya J. Desai and Laurence J. Miller\**

Department of Molecular Pharmacology and Experimental Therapeutics, Mayo Clinic, Scottsdale, AZ, USA

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Pedro A. Jose, Georgetown University, USA Akiyoshi Takahashi, Kitasato University, Japan

#### *\*Correspondence:*

Laurence J. Miller, Department of Molecular Pharmacology and Experimental Therapeutics, Mayo Clinic, 13400 E. Shea Blvd., Scottsdale, AZ 85259, USA. e-mail: miller@mayo.edu

Cholesterol represents a structurally and functionally important component of the eukaryotic cell membrane, where it increases lipid order, affects permeability, and influences the lateral mobility and conformation of membrane proteins. Several G protein-coupled receptors have been shown to be affected by the cholesterol content of the membrane, with functional impact on their ligand binding and signal transduction characteristics. The effects of cholesterol can be mediated directly by specific molecular interactions with the receptor and/or indirectly by altering the physical properties of the membrane. This review focuses on the importance and differential effects of membrane cholesterol on the activity of cholecystokinin (CCK) receptors. The type 1 CCK receptor is quite sensitive to its cholesterol environment, while the type 2 CCK receptor is not. The possible structural basis for this differential impact is explored and the implications of pathological states, such as metabolic syndrome, in which membrane cholesterol may be increased and CCK1R function may be abnormal are discussed. This is believed to have substantial potential importance for the development of drugs targeting the CCK receptor.

**Keywords: cholecystokinin, G protein-coupled receptors, cholesterol**

# **INTRODUCTION**

Biological membranes consist of a variety of lipids and proteins that establish diffusional boundaries for the cell and its organelles. An important lipid component of the plasma membrane is cholesterol that may be present in concentrations as high as 20–40 mol%, which is known to have substantial impact on the physical properties of the membrane and on the structure and function of various intrinsic membrane proteins (Mouritsen and Zuckermann, 2004).

The hydrophilic hydroxyl groups of the amphiphilic cholesterol molecules are intercalated into the lipid bilayers, with one cholesterol molecule spanning approximately half of the bilayer (Mouritsen and Zuckermann, 2004). Cholesterol has the unique ability to increase order in such membranes by organizing the arrangement of the surrounding lipids, while maintaining fluidity and lateral diffusion within the membrane. A region of the membrane in which cholesterol is absent typically exhibits disorder and rapid lateral diffusion with randomly packed lipid molecules (Miao et al., 2002; Mouritsen and Zuckermann, 2004). Lowering the temperature can result in a transition from a liquiddisordered phase toward a solid-ordered phase having slower lateral diffusion and greater ordering of the lipid chains. The presence of cholesterol can also induce order in the liquid phase by increasing the density of the packing of fatty acyl chains. This is also identified as a liquid-ordered phase (Ipsen et al., 1987).

Cholesterol can also affect the permeability function of the membrane by changing its lateral density fluctuations and structural heterogeneity and thereby regulating its "leakiness." The effect of cholesterol on lipid order also affects the thickness of the bilayer (Ohvo-Rekila et al., 2002; Mouritsen and Zuckermann, 2004). In the presence of low or absent cholesterol, sodium ions are able to passively permeate the lipid bilayer, while such permeability is inhibited in the presence of high (40%) cholesterol (Ohvo-Rekila et al., 2002). Cholesterol can also bind and inhibit some solutes such as ethanol, which become strongly adsorbed as a result of density fluctuations and the heterogeneous composition of the bilayer (Schroeder et al., 1996).

Cholesterol can also contribute to specialized microdomains within the plasma membrane that are known as planar lipid rafts. These are small, low-density regions in the outer leaflet of the bilayer that are enriched in cholesterol and glycosphingolipids. These structures have been identified based on their insolubility in non-ionic detergents at low temperature and their high buoyancy in density gradients (Brown and Rose, 1992). These planar structures may also give rise to caveolae, representing flaskshaped invaginations that can ultimately form free intracellular organelles (Cohen et al., 2004; Shaw, 2006). It is notable that lipid rafts and caveolae can act as organizational platforms for elements involved in signal transduction (Okamoto et al., 1998; Pike, 2003; Cohen et al., 2004; Ostrom and Insel, 2004).

# **EFFECTS OF MEMBRANE CHOLESTEROL ON G PROTEIN-COUPLED RECEPTORS (GPCRs)**

Several GPCRs have been reported to be sensitive to the concentration of cholesterol in the membrane, with different effects

**Abbreviations:** CCK, cholecystokinin; cAMP, cyclic adenosine monophosphate; CCM, cholesterol consensus motif; CRAC, cholesterol recognition/interaction amino acid consensus; DAG, diacylglycerol; FRET, fluorescence resonance energy transfer; GPCRs, guanine nucleotide-binding protein-coupled receptors; MβCD, methyl-beta-cyclodextrin; IP3, inositol triphosphate.

observed for different receptors (**Table 1**). Cholesterol can affect receptor conformation, thereby affecting its ligand binding and signaling characteristics. It can affect lateral mobility within the bilayer that is critical for G protein coupling. It can also affect receptor trafficking and sequestration that contribute to desensitization. However, currently there are no well-established rules for which receptors might be influenced by cholesterol and how these receptors might be affected. Two distinct groups of mechanisms have been proposed: (A) direct binding of cholesterol molecules to GPCR molecules, including the possibility of interacting with specific sites or motifs (Albert et al., 1996; Li and Papadopoulos, 1998; Paila et al., 2009) and/or (B) indirect effects of the cholesterol by altering the physical properties of the membrane in which the GPCRs reside (Lee, 2004; Mouritsen and Zuckermann, 2004).

The effects of membrane cholesterol on rhodopsin, the photoreceptor of the retinal rod cells, have been extensively studied. Rhodopsin exists in various conformations known as metarhodopsins. The equilibrium between these conformational states is sensitive to the amount of membrane cholesterol present, with increased cholesterol shifting the equilibrium toward the inactive states of the receptor (Mitchell et al., 1990; Bennett and Mitchell, 2008). The influence of membrane cholesterol on rhodopsin function has been attributed to both direct and indirect mechanisms. Spatial approximation between tryptophan residues of rhodopsin and cholesterol has been demonstrated using fluorescence resonance energy transfer (FRET), with an estimation of one sterol molecule interacting with one rhodopsin molecule (Albert et al., 1996). These observations have been supported by the crystal structure of metarhodopsin I that includes a cholesterol molecule bound between the tryptophan residues of transmembrane segment four of one protomer and transmembrane segments five, six, and seven of the other protomer (Ruprecht et al., 2004). An indirect effect of cholesterol has also been attributed to its effect on the partial free volume of the membrane. The conversion of metarhodopsin I to metarhodopsin II involves an expansion of the protein in the plane of the lipid bilayer (Attwood and Gutfreund, 1980), thereby occupying the partial free volume of the surrounding bilayer. As membrane cholesterol is increased, the formation of metarhodopsin II is decreased by reducing the partial free volume in the membrane (Niu et al., 2002).


Similarly, cholesterol has been shown to be bound to the β2 adrenergic receptor in its crystal structure (Hanson et al., 2008), and this lipid has been shown to be necessary for ligand binding, G protein interaction, and signal transduction at that receptor (Ben-Arie et al., 1988). Two cholesterol molecules appear to be bound to sites on transmembrane segments one, two, three, and four of a β2-adrenergic receptor molecule (Hanson et al., 2008).

Other examples include the oxytocin receptor, where the amount of cholesterol in the membrane is directly related to the ligand binding affinity of the receptor (Gimpl et al., 1997, 2002; Gimpl and Fahrenholz, 2002) [43% decrease in membrane cholesterol caused a sharp decline in ligand binding (Gimpl et al., 1997)]. This has been shown to be a highly co-operative process where more than six molecules of cholesterol can be bound to one oxytocin receptor molecule (Gimpl et al., 2002). There is clear structural specificity for this interaction, since only cholesterol analogs that are structurally similar to cholesterol are able to reproduce this effect on oxytocin receptor function (Gimpl et al., 1997). Cholesterol is also believed to provide stability against thermal and pH alterations, and to protect some receptors from proteolytic degradation (Gimpl and Fahrenholz, 2002). Similarly, in the case of the galanin-2 receptor, cholesterol affects ligand binding process in a positively co-operative manner (Pang et al., 1999). Only a limited number of cholesterol analogs are able to exhibit similar effects to those of cholesterol, again supporting structural specificity of this interaction (Pang et al., 1999). Other examples of GPCRs affected by membrane cholesterol are listed in **Table 1**.

Some interactions of membrane cholesterol with GPCRs have been attributed to the presence of consensus motifs within these receptors. Several proteins that are known to interact with cholesterol have a characteristic amino acid sequence, termed the cholesterol recognition/interaction amino acid consensus (CRAC) motif, in their transmembrane segments. This is defined by the pattern—L/V-(X)1–5-Y-(X)1–5-R/K-, in which (X)1–5 represents between one and five residues of any amino acid (Li and Papadopoulos, 1998). This sequence is present in rhodopsin, β2-adrenergic, serotonin1A, and cholecystokinin (CCK) receptors (Jafurulla et al., 2011; Potter et al., 2012). Another consensus motif, the strict-cholesterol consensus motif (CCM), was described in transmembrane segments by Hanson et al. (2008), and was later expanded by Adamian et al. (2011). This motif [4.39–4.43 (R,K)] [4.50 (W,Y)] [2.45 (S)] [4.46 (I,V,L)] [2.41 (F,Y)] that incorporates the Ballesteros and Weinstein numbering system based on residue position relative to most conserved residues within a given transmembrane segment (Ballesteros and Weinstein, 1992), was recognized from the analysis of the 2.8 Å structure of the human β2-adrenergic receptor and is present in 21% of the class A GPCRs (Hanson et al., 2008). A less-restrictive variant of this motif is also found in 44% of class A GPCRs, where the aromatic residue is absent at the position 2.41. The presence of this motif, suggests that specific sterol binding may be important to the structure and stability of receptors in this family.

#### **CHOLECYSTOKININ (CCK) PEPTIDES AND PHYSIOLOGY**

The GPCRs that are responsive to the gastrointestinal and brain peptide, CCK, are the major focus of this review, with one of the CCK receptors affected by cholesterol and the other unaffected. CCK is a polypeptide hormone synthesized in the I-cells of the small intestine (Rehfeld, 1978) that is released in response to protein and fat in the lumen that plays an important role in nutrient homeostasis. CCK was identified based on its ability to stimulate gallbladder contraction, and it was also eventually recognized as being identical to pancreozymin, a hormone that stimulates pancreatic exocrine secretion (Harper and Raper, 1943). CCK also contributes to post-cibal satiety (Kissileff et al., 1981; Smith and Gibbs, 1985; Beglinger et al., 2001), which is a critically important role that could provide the basis of a treatment of obesity. This hormone is also one of the most abundant neuropeptides present in the brain (Miller et al., 1984), and it has been shown to have effects on enteric smooth muscle and nerves at various locations in the peripheral and central nervous system. It also has been described to have direct natriuretic effects on the kidney, and to decrease renal excretion of calcium and magnesium (Duggan et al., 1988; Ladines et al., 2001). CCK is present as a variety of different length peptides that are produced from a single 115-residue preprohormone precursor, all sharing their carboxylterminal sequence. These range from 58, 39, 33, and 8 residues, with each containing a sulfated tyrosine residue seven residues from the carboxyl terminus, as well as amidation of the carboxylterminal phenylalanine residue (Eysselein et al., 1990; Rehfeld et al., 2001; Miller and Gao, 2008).

# **CCK RECEPTOR STRUCTURE AND FUNCTION**

CCK exerts its physiological functions through the activation of two structurally-related class A GPCRs identified as CCK receptor type 1 (CCK1R) and CCK receptor type 2 (CCK2R) (also known as CCKAR and CCKBR, respectively, related to their prominent presence in "alimentary tract" and "brain") (Dufresne et al., 2006). CCK2R also binds gastrin, another structurally-related polypeptide hormone produced in the gastric antrum (Kopin et al., 1992). In contrast, CCK1R binds gastrin at a very low affinity (by 500-fold). Both these receptors are highly homologous and approximately 50% identical, particularly in the transmembrane regions where they are 70% identical (Miller and Gao, 2008). Although both types of CCK receptors bind and are activated by CCK and gastrin peptides, the molecular basis for peptide binding to these receptors appears to be distinct. The CCK1R requires the carboxyl-terminal CCK heptapeptide-amide that includes the sulfated tyrosine for high affinity binding and biological activity, whereas the CCK2R requires only the carboxyl-terminal tetrapeptide amide that is shared between all CCK and gastrin receptors (Miller and Gao, 2008). The activation of the CCK1R by CCK elicits a broad variety of important physiological functions, such as stimulation of gallbladder contraction and pancreatic exocrine secretion, delay of gastric emptying, relaxation of the sphincter of Oddi, inhibition of gastric acid secretion, and induction of post-cibal satiety (Kerstens et al., 1985; Schmitz et al., 2001). The CCK1R is localized in the human gastric mucosa within D cells (Schmitz et al., 2001) and muscularis propria of human gastric antrum, fundus, and pylorus (Reubi et al., 1997). CCK2R is localized mainly in brain and is also present in the gastric oxyntic mucosa, some enteric smooth muscle, and the kidneys (Noble and Roques, 1999; Von Schrenck et al., 2000). In addition to stimulation of gastric acid secretion, CCK2R also plays a role in anxiety and nociception (Noble and Roques, 1999; Noble, 2007).

The CCK receptors belong to the class A group of GPCRs [see review of CCK receptor structure and pharmacology (Miller and Gao, 2008)]. They share the typical signature sequences of this family, including E/DRY at the intracellular side of transmembrane segment three and NPxxY at the intracellular side of transmembrane segment seven. Both receptors are glycosylated, possess the conserved disulfide bond between extracellular loops one and two (CCK1R also has an extra disulfide bond within the amino-terminal extracellular domain), and include multiple sites for serine and threonine phosphorylation in intracellular loop three and in the carboxyl-terminal tail. The function of phosphorylation is to desensitize the receptor, interfering with its coupling to G proteins (Rao et al., 1997). These receptors also have cysteine residues representing sites of palmitoylation in the intracellular carboxyl-terminal tail, which help to attach an eighth helical segment to the cytosolic face of the bilayer.

A broad range of approaches, including ligand binding and signaling of chimeric and site-specific mutants (Kopin et al., 1995; Miller and Lybrand, 2002), photoaffinity labeling (Ding et al., 2001; Dong et al., 2005, 2007) and fluorescence-based techniques (Harikumar et al., 2004, 2005a, 2006; Harikumar and Miller, 2005) have provided important information regarding the molecular basis of CCK binding to these receptors. Information from these studies has shown that the determinants for CCK binding to the CCK1R are distributed throughout the extracellular loop and amino-terminal tail regions, but not within the predicted transmembrane domain bundle (Miller and Lybrand, 2002; Harikumar et al., 2004). However, a possible difference for CCK binding to these two receptors relates to the position of the carboxyl-terminal end of the peptide. This portion of CCK resides at the surface of the lipid bilayer adjacent to transmembrane segment one for the CCK1R, as directly demonstrated by siteselective photoaffinity labeling (Harikumar et al., 2004), while it may dip into the helical bundle for the CCK2R, although the latter is based on less definitive data coming from site-directed mutagenesis (Harikumar et al., 2006).

**Table 2** shows differences in microenvironment-dependent spectral properties of fluorescent CCK ligands docked to CCK1 or CCK2 receptors, supporting the suggested differential binding of the ligands to these two receptors.

Various selective and potent non-natural ligands for the CCK receptor have been developed [see reviews (Herranz, 2003; Kalindjian and McDonald, 2007)]. The most extensively studied in regard to mechanism of binding to CCK receptors is the group of benzodiazepine compounds (Aquino et al., 1996; Darrow et al., 1998; Gao et al., 2008; Cawston et al., 2012). Minor chemical modifications of these compounds have been shown to change receptor subtype selectivity and biological responsiveness (Aquino et al., 1996; Gao et al., 2008; Miller and Gao, 2008). It is now clear, based on receptor mutagenesis, photoaffinity labeling, and pharmacological manipulations, that these ligands bind to an allosteric site within the intramembranous helical bundle that is distinct from the orthosteric CCK peptide-binding site of the CCK1R (Kopin et al., 1994; Hadac et al., 2006; Gao et al., 2008). In fact, using two specific benzodiazepine antagonist ligands, it has been shown that transmembrane segments six and seven [residues 6.51, 6.52, and 7.39 (Ballesteros and Weinstein, 1992)] are most important for binding the CCK1R-selective ligand, whereas residues of transmembrane segments two and seven



Top, CCK analogs with a fluorescent alexa indicator at the amino terminus of a CCK-8 analog and a CCK-4 analog were used to examine the microenvironment of the fluorophore as docked to CCK1 or CCK2 receptors (Harikumar et al., 2005a).

Bottom, CCK analogs with a fluorescent aladan indicator at the amino terminus, mid-region, or carboxyl-terminus of CCK were used to examine the microenvironment of the fluorophore as docked to CCK1 or CCK2 receptors. Fluorescence quenching, anisotropy and red edge excitation shifts were examined (Harikumar et al., 2006).

(2.61 and 7.39) are most important for binding the CCK2R selective ligand (Cawston et al., 2012).

# **EFFECTS OF MEMBRANE CHOLESTEROL ON CCK RECEPTOR FUNCTION**

The ability of CCK to induce gallbladder contraction has been shown to be reduced in individuals with cholesterol gallstones (Fridhandler et al., 1983; Lamorte et al., 1985; Behar et al., 1989; Chen et al., 1997; Amaral et al., 2001; Kano et al., 2002). Initially, it was shown that high cholesterol diet (1–1.2%) induced cholesterol stone formation in the gallbladders of animals like prairie dogs and ground squirrels (Fridhandler et al., 1983). Gallbladder muscle contraction in these models was impaired in response to CCK, but it was also impaired in response to acetylcholine and the non-hormonal stimulant potassium (Lamorte et al., 1985). Studies using muscle strips from human gallbladders with cholesterol gallstones have also demonstrated reduced contractility in response to CCK (Behar et al., 1989). Similar effects were observed when studying *in vivo* gallbladder contraction in response to CCK in patients with gallbladder disease (Amaral et al., 2001). More specifically, using isolated human gallbladder smooth muscle strips and single muscle cells, it has been shown that specimens with cholesterol stones exhibit lower cAMP responses compared with those in gallbladders with pigment stones. Hence, it was suggested that the muscle defect responsible for this impairment was at the level of the plasma membrane (Chen et al., 1997).

Indeed the plasma membrane can incorporate excessive amounts of cholesterol in the presence of cholesterolrich environment by diffusion (Nichols and Pagano, 1981). Studies measuring the amount of membrane-bound esterified [3H]cholesterol in the presence of excessive amounts of unesterified [3H]cholesterol, had already confirmed the existence of this phenomenon in cell types such as erythrocytes (Lange and D'Alessandro, 1977) and rat arterial smooth muscle cells (Slotte and Lundberg, 1983). It was suggested from these observations that membrane cholesterol was a key factor in causing the gallbladder muscle impairment. This was confirmed in a study by Yu et al. (1996), in which they measured the cholesterol content of isolated single muscle cells and plasma membranes from gallbladder muscles from prairie dogs, finding an association of elevated cholesterol with reduced muscle contractility. It was reported that after feeding a high cholesterol (1.2%) diet, cholesterol content, and the molar ratio of cholesterol/phospholipid in plasma membranes of gallbladder muscle increased by 90%, with a parallel decrease in muscle contractility by 58% in response to CCK. Similar changes were observed when normal gallbladder muscle cells were incubated with cholesterol-rich liposomes, an effect which was reversed upon incubation with cholesterol-free liposomes (Yu et al., 1996). These observations were also confirmed in human gallbladder muscle from patients with cholesterol gallstones (Chen et al., 1999; Xiao et al., 1999), where the membrane cholesterol content and cholesterol/phospholipid ratio was significantly higher in gallbladders with cholesterol stones than in those with pigment stones. Membrane anisotropy was also higher in gallbladders from patients with pigment gallstones, reflecting

lower membrane fluidity in gallbladders from patients with cholesterol gallstones (Chen et al., 1999).

A report studying CCK signaling in isolated single muscle cells from human gallbladders with cholesterol gallstones demonstrated that the production of IP3 and diacylglycerol (DAG) was reduced when compared with gallbladders from patients with pigment gallstones (by 80–90% and 78%, respectively) (Yu et al., 1995). However, this effect could be circumvented by stimulating the cells with agents acting directly on G proteins. These results suggested that the defect was proximal to the G protein, at the level of the CCK receptor or its coupling with the G protein (Yu et al., 1995). An illustrative clinical report was published in 1995 that described a patient with morbid obesity and cholesterol gallstone disease in whom the CCK1R was misspliced to yield a non-functional variant missing its third exon (Miller et al., 1995). The frequency of such events is likely extremely low based on another study by the same group (Nardone et al., 1995), in which full length sequencing of the cDNA encoding the CCK1R was shown to be entirely normal in 12 patients with cholesterol gallstones undergoing cholecystectomy (Nardone et al., 1995). Despite normal molecular structure, the function of CCK receptors in patients with cholesterol gallstone disease has been directly shown to be abnormal (Xiao et al., 1999). Of particular interest, the apparent affinity of binding was higher than normal, while the ability of this hormone to stimulate a signaling response was decreased. These abnormalities were reversed after extraction of excess cholesterol from the membrane by incubation with cholesterol-free liposomes (Xiao et al., 1999).

Indeed the impact of membrane cholesterol on this receptor was further explored in manipulatable model systems in 2005 (Harikumar et al., 2005b). This study included extensive analyses of the CCK1R function in lipid-modified environments and showed that normal CCK1R function is dependent on the content of cholesterol in the membrane. In order to monitor the changes in conformation of the CCK1R in the presence of various levels of cholesterol, the study used two different fluorescence-based approaches by measuring anisotropy and lifetimes of a fluorescent full agonist ligand, which shows a decrease in both of these parameters when occupying a receptor in active conformation (Harikumar et al., 2002). Cholesterol depletion from CCK1R-bearing cells using chemical (MβCD chelator) or metabolic (lipoprotein deficient serum supplemented with hydroxymethylglutaryl-CoA reductase inhibitor) methods increased the movement of the label as reflected by decreased anisotropy and reduced lifetime, whereas cholesterol enrichment had the opposite effect. The fluorescence changes in the presence of increased cholesterol were similar to those observed in the presence of a non-hydrolyzable GTP analog which is known to shift the receptor into an inactive, G protein-uncoupled state (Harikumar et al., 2002).

The changes in the conformation of the CCK1R in response to varying amounts of membrane cholesterol are reflected in its ligand binding characteristics and in its biological activity (Harikumar et al., 2005b). Cholesterol depletion was shown to be associated with a reduction in CCK binding affinity, while augmentation of membrane cholesterol content actually increased CCK binding affinity. However, the higher binding in the presence of increased cholesterol was non-productive, resulting in lower biological responsiveness, like the membranes depleted in cholesterol. Notably, the defective intracellular calcium responses to CCK after cholesterol depletion were reversed upon cholesterol repletion. These observations support a defective coupling of the CCK receptor to its G protein in the presence of abnormal membrane cholesterol content (Chen et al., 1997; Xiao et al., 1999).

Other properties such as clathrin-mediated receptor internalization after agonist occupation, and receptor recycling were unaffected by modulation of the membrane cholesterol content, although these regulatory processes have been described to be abnormal after modification of membrane sphingolipid content. This indicates that two different lipid components by themselves or in combination with other lipid components of the plasma membrane probably induce different conformational changes to the CCK1R which can lead to either defective G protein coupling (in case of cholesterol), or disruption in the internalization and recycling pathways (sphingolipid) of the CCK1R. These are reflected by the differences in the observed effects on binding, signaling, and receptor internalization (Harikumar et al., 2005b). However, it is still unclear whether these effects of modification of cholesterol and sphingolipids reflect events occurring in the bulk phase of the plasma membrane or within rafts. Nevertheless, these insights have added a new dimension to our understanding of CCK receptor biology.

#### **POSSIBLE STRUCTURAL BASIS FOR THE DIFFERENTIAL SENSITIVITY OF CCK RECEPTORS TO CHOLESTEROL**

In contrast, the other subtype of CCK receptor, the CCK2R, is not sensitive to alterations in membrane cholesterol (Potter et al., 2012). As noted earlier, these receptors are highly homologous and, in fact, share all of the predicted cholesterol binding motifs (**Figure 1**). The key structural determinant for cholesterol sensitivity appears to reside in the third exon of the CCK1R which encodes most of the transmembrane segment three and four, including one CRAC motif and one CCM motif (Potter et al., 2012). It should be noted that the CCK1R and CCK2R do not contain the entire four-component strict-CCM, but a less restrictive CCM motif that is shared by 44% of human class A GPCRs (Hanson et al., 2008). This is due to the absence of an aromatic residue at 2.41. The relevant contributors to this effect were further studied with site-directed mutants (Potter et al., 2012). This suggested that three residues, Y140 (Y3.51), which is a part of the CRAC motif (Li and Papadopoulos, 1998) in transmembrane segment three, W166 (W4.50), which is a part of the CCM (Hanson et al., 2008) in transmembrane segment four and Y237 (Y5.66), which is part of the CRAC motif in transmembrane segment five, are important for the cholesterol sensitivity of the CCK1R. It is noteworthy that mutation of each of the three residues has negative effects on either CCK binding or CCK-induced signaling. Mutation of Y140 to alanine leads to reduced signaling but increased CCK binding affinity and is insensitive to reduction in membrane cholesterol levels. On the other hand, mutation of W166 to alanine leads to decreased CCK binding and signaling ability of the CCK1R, with a further reduction in signaling upon depleting membrane cholesterol. The Y237A mutant displays no change in signaling, but reduced CCK binding affinity, whereas cholesterol reduction results in the reduction of both parameters. Corresponding single residue mutations (Y153A, W179A, Y246A) within the CRAC and the CCM motifs of the CCK2R did not modify receptor function.

It is remarkable that mutation of only Y140 residue within the CRAC motif within transmembrane segment three resulted in loss of cholesterol sensitivity of the CCK1R (Potter et al., 2012). This implies a defective conformation of the receptor, particularly because the mutation alters the conserved (D/E) 3.49-R3.50- Y3.51 motif, and thus its ability to mediate the differential effects of cholesterol in the two structurally related receptors is compromised. However, studies with other class A GPCRs have shown that the mutation of Y3.51 of the D/ERY motif does not affect the ligand binding affinity or receptor trafficking, and has no or marginal effects on receptor signaling (Ohyama et al., 2002; Gaborik et al., 2003; Proulx et al., 2008). So, it appears that probably this residue is specifically responsible for the differential cholesterol sensitivity observed between the CCK receptors. This implies that the transmembrane segment three CRAC cholesterol binding motif could be responsible for the cholesterol sensitivity of the CCK1R, because the transmembrane segment four CCM (W166) and transmembrane segment five CRAC (Y237) mutants are still sensitive to cholesterol. However, the transmembrane segment three CRAC cholesterol-binding motif is essentially the same in both the CCK receptors, differing only in the presence of lysine in CCK1R and an arginine in the CCK2R. Both the residues which are a part of the consensus motif have similar properties.

While the sequence surrounding the motif is largely similar, they do exhibit minor differences between the two receptors that could contribute to the observed difference in the cholesterol sensitivity.

These observations show that the presence of a CRAC or/and CCM motif do not necessarily mean that the receptor is sensitive to cholesterol. At least in the case of the CCK receptors, it appears that the subtype specific conformational changes are conferred by the differences in direct interaction of the membrane cholesterol with the two receptors. It has been shown in the case of the human cannabinoid receptors (CBR), that the type 1 CBR (CB1R) is affected by manipulations in the membrane cholesterol content, in a way similar to the CCK1R exhibiting defective ligand binding and signaling, as well as its localization to membrane rafts (Bari et al., 2005a,b). On the other hand, similar to CCK2R, the type 2 CBR (CB2R) does not share this sensitivity to membrane cholesterol (Bari et al., 2005a,b). In fact it was found that CB1R possesses a CRAC motif in transmembrane segment seven that differs by one residue from that in the CB2R, and the corresponding mutation rendered CB1R insensitive to membrane cholesterol similar to CB2R (Oddi et al., 2011). This explains that cholesterol binding motifs outside of the CCM can have a significant influence on receptor signaling and trafficking.

In addition to this, other possibilities for the differential sensitivity of CCK receptors may be related to the indirect impact of the cholesterol on the processes like mechanism of ligand binding or receptor trafficking. As mentioned earlier, the CCK1R and the CCK2R bind the same CCK peptide ligand differently (Silvente-Poirot and Wank, 1996; Dong et al., 2005; Harikumar et al., 2005a, 2006). The peptide ligand binds to extracellular loops and the amino-terminal tail of the CCK1R, whereas in the CCK2R, the carboxyl-terminal end of the peptide ligand may dip into the helical bundle (Harikumar et al., 2005a, 2006). It can be postulated that in the case of CCK2R the peptide binding within the helical bundle could potentially stabilize it and consequently the intracellular regions as well, and prevent the negative impact of changing membrane cholesterol. On the other hand, the impact of membrane cholesterol cannot be overcome in the case of the CCK1R because the conformational changes affecting the extracellular regions binding the ligand are very different to that conferred to the part of the receptor within the lipid bilayer. So, these differential changes on the domains of the receptors which are physically further away, may dictate the observed effects of cholesterol sensitivity in the CCK1R.

Other processes occurring post ligand binding can also be affected by changes in membrane cholesterol, which can contribute to the observed effects. In the case of CCK1R, the receptor has been described to occupy a unique plasma membrane

#### **REFERENCES**


segment disk membranes. *Biochim. Biophys. Acta* 1285, 47–55.


compartment after CCK stimulation of rat pancreatic acinar cells (Roettger et al., 1995). This was called as a site of "insulation" that is comparatively devoid of G proteins, in which the lateral mobility of the CCK receptor is markedly reduced as another highly specialized cellular mechanism of desensitization (Roettger et al., 1999). Accordingly, it can be postulated that the membrane cholesterol can cause similar changes to the CCK1R, however, there is no information yet about the existence of a similar mechanism for CCK2R. Also, whether the CCK receptors are located in the lipid rafts, and if the cholesterol modulation effects are exclusive to these regions remains to be investigated.

## **CONCLUSION AND THERAPEUTIC PERSPECTIVES**

It is clear that lipid-protein interactions, especially cholesterolreceptor interactions, involved in the function of GPCRs can be extremely important. More specifically, this review has focused on the role of cholesterol in the function of the CCK1R, an important receptor likely to be involved in pathologic states such as obesity and metabolic syndrome. A better understanding of the mechanisms of appetite regulation that contribute to the development of obesity could provide insights into the prevention and more effective treatment of this disorder. The gastrointestinal hormones like CCK play important roles in servomechanisms involved in regulating nutritional homeostasis. The observations showing that elevated membrane cholesterol reduce the effectiveness of stimulus-activity coupling activated by CCK action at the CCK1R provide vital information regarding the pathological phenomenon associated with CCK receptor dysfunction. It is essential and would be very interesting to study the membrane environment of the CCK1R receptor in obese patients, in order to verify these observations in humans. Additionally, it would be very exciting to study whether the mutations affecting the cholesterol sensitivity might also be present in these patients. Overall, the observations so far may suggest that elevated levels of membrane cholesterol described in cholesterol gallstone disease and metabolic syndrome could reduce hormone-stimulated signaling and, thereby, reduce the feedback inhibition of food consumption that is an essential servomechanism. If this is proven true in patients, the CCK1R can become the target for future development of positive allosteric modulators that could "recalibrate," and consequently "normalize" this important regulatory mechanism.

#### **ACKNOWLEDGMENTS**

We thank Dr. K. G. Harikumar for critical input on the manuscript and insightful discussions. This work was supported by a grant from the National Institutes of Health (DK32878).


of the equilibrium between metarhodopsin I and II from bovine retinas. *FEBS Lett.* 119, 323–326.

Ballesteros, J. A., and Weinstein, H. (1992). Analysis and refinement of criteria for predicting the structure and relative orientations of transmembranal helical domains. *Biophys. J.* 62, 107–109.


the early stages of cholesterol gallstone formation. *Gastroenterology* 85, 830–836.


cholecystokinin receptor. *J. Biol. Chem.* 280, 2176–2185.


non-peptide CCK2 receptor agonist and antagonist ligand. *Curr. Top. Med. Chem.* 7, 1195–1204.


receptor signaling: implications for molecular pharmacology. *Br. J. Pharmacol.* 143, 235–245.


is cholecystokinin-33. *J. Clin. Endocrinol. Metab.* 86, 251–258.


et al. (2004). D1 dopamine receptor signaling involves caveolin-2 in HEK-293 cells. *Kidney Int.* 66, 2167–2180.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 04 August 2012; paper pending published: 23 August 2012; accepted: 01 October 2012; published online: 18 October 2012.*

*Citation: Desai AJ and Miller LJ (2012) Sensitivity of cholecystokinin receptors to membrane cholesterol content. Front. Endocrin. 3:123. doi: 10.3389/fendo. 2012.00123*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Desai and Miller. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Mutation of Phe318 within the NPxxY(x)5*,*6F motif in melanin-concentrating hormone receptor 1 results in an efficient signaling activity

#### *Akie Hamamoto1, Manabu Horikawa2, Tomoko Saho1 and Yumiko Saito1 \**

<sup>1</sup> Graduate School of Integrated Arts and Sciences, Hiroshima University, Hiroshima, Japan

<sup>2</sup> Bioorganic Research Institute, Suntory Foundation for Life Sciences, Osaka, Japan

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Jean A. Boutin, Institut de Recherches SERVIER, France Stewart Clark, University at Buffalo, SUNY, USA

#### *\*Correspondence:*

Yumiko Saito, Graduate School of Integrated Arts and Sciences, Hiroshima University, 1-7-1 Kagamiyama, Higashi-hiroshima, Hiroshima 739-8521, Japan. e-mail: yumist@hiroshima-u.ac.jp

Melanin-concentrating hormone receptor 1 (MCHR1) is a G-protein-coupled receptor (GPCR) that plays an important role in feeding by coupling to Gαq- and Gαi-mediated signal transduction pathways. To interrogate the molecular basis for MCHR1 activation, we analyzed the effect of a series of site-directed mutations on rat MCHR1 function. In the highly conserved NPxxY(x)5*,*6F domain of GPCRs, the phenylalanine residue is involved in structural constraints; replacement with alanine generally leads to impaired/lost GPCR function. However, Phe-to-Ala (F318A) mutation in MCHR1 had no significant effect on the level of cell surface expression and receptor signaling. By analyzing a further series of mutants, we found that Phe-to-Lys substitution (F318K) caused the most significant reduction in the EC50 value of MCH for calcium mobilization without affecting receptor expression at the cell surface. Interestingly, GTPγS-binding, which monitors Gα<sup>i</sup> activation, was not modulated by F318K. Our results, combined with computer modeling, provide new insight into the role of Phe in the NPxxY(x)5*,*6F motif as a structurally critical site for receptor dynamics and a determinant of Gα protein interaction.

**Keywords: GPCR, helix 8, melanin-concentrating hormone, NPxxY(x)5***,***6F motif, signal transduction**

#### **INTRODUCTION**

Mammalian melanin-concentrating hormone (MCH), a cyclic nonadecapeptide produced predominantly by neurons of the lateral hypothalamus, is involved in the regulation of food intake behavior and energy expenditure (Bittencourt et al., 1992; Rossi et al., 1997; Shimada et al., 1998). MCH acts *via* two G-proteincoupled receptors (GPCRs), Melanin-concentrating hormone receptor 1 (MCHR1), and MCHR2 (Chambers et al., 1999; Saito et al., 1999; An et al., 2001), of which MCHR2 is not functionally present in rodents (Tan et al., 2002). MCHR1 is widely expressed at high levels in the brain (Saito et al., 2001). Because mice lacking MCHR1 are lean, hyperactive, hyperphagic, and hypermetabolic (Chen et al., 2002; Marsh et al., 2002), MCHR1 is viewed as the physiologically relevant MCH receptor in rodents. In support of this belief, selective MCHR1 antagonists decrease food intake and body weight in both normal and dietinduced obese rats (Takekawa et al., 2002; Shearman et al., 2003). Moreover, some of these antagonists exhibit anti-depressant and anxiolytic effects (Borowsky et al., 2002; Georgescu et al., 2005). Therefore, the MCH-MCHR1 system could be an important target for the treatment of obesity and certain mood disorders.

In mammalian cells transfected with MCHR1, MCH is able to activate multiple signaling pathways including calcium mobilization, activation of extracellular signal-regulated kinase (ERK) and inhibition of cyclic AMP generation through Gαi*/*o- and Gαq-coupled pathways (Chambers et al., 1999; Saito et al., 1999; Hawes et al., 2000). Several studies have reported structural determinants of MCHR1 activation by MCH. Biochemical analysis of MCHR1 using molecular modeling identified Asp123 in the third transmembrane domain (TM3) as being crucial for ligand binding (Macdonald et al., 2000). In addition, Thr255, which is located at the junction of intracellular loop 3 (i3) and transmembrane domain 6 (TM6), is critically important for receptor folding and correct trafficking to the cell surface (Fan et al., 2005). We previously identified that Asn23 in the extracellular N-terminus contributed mainly to N-linked glycosylation of MCHR1 and is necessary for MCHR1 cell surface expression, ligand binding and signal transduction (Saito et al., 2003). We also showed that Arg155 in intracellular loop 2 (i2) and a proximal dibasic motif (Arg319/Lys320) in eighth cytoplasmic helix (helix 8: a common short amphiphilic helical domain in the proximal C-terminal tail) are important for signaling (Tetsuka et al., 2004; Saito et al., 2005), whereas the distal part of the C-terminal tail is necessary for receptor internalization (Saito et al., 2004). However, despite numerous mutagenesis studies, the residues that determine G protein selectivity (Gα<sup>q</sup> vs. Gαi) have yet to be identified.

The NPxxY(x)5*,*6F sequence, located at the junction between TM7 and the connecting cytosolic helix 8, is conserved in most

**Abbreviations:** ECL, enhanced chemiluminescence; ER, endoplasmic reticulum; ERK, extracellular signal-regulated kinase; FBS, fetal bovine serum; GPCR, G-protein-coupled receptor; HEK293T, human embryonic kidney 293; helix 8, eighth cytoplasmic helix; i1, intracellular loop 1; i2, intracellular loop 2; i3, intracellular loop 3; MCH, melanin-concentrating hormone; MCHR1, MCHR1 receptor; MCHR2, MCHR2 receptor; PTX, pertussis-toxin; TM, transmembrane.

rhodopsin family (class A) GPCRs, including the MCH receptor (Gether, 2000; Huynh et al., 2009). The high degree of conservation of this motif suggests that it must play very important roles in rhodopsin family GPCR functionality. Mutations in the NPxxY(x)5*,*6F motif are reported to affect ligand binding, G protein coupling and receptor phosphorylation. In rhodopsin, the prototypical GPCR, the Tyr and Phe residues within the motif were both found to be critical for proper light-induced conformational changes from the ground state (Acharya and Karnik, 1996). Moreover, the Phe residue is reported to be essential for export of the β1-adrenergic receptor (β1-AR), α2Badrenergic receptor (α2B-AR) and A1 adenosine receptor from the endoplasmic reticulum (ER) (Delos Santos et al., 2006; Duvernay et al., 2009; Málaga-Diéguez et al., 2010). Indeed, Phe-to-Ala substitution in the α2B-AR dramatically reduced cell-surface expression by 91% compared with their wild-type variants (Duvernay et al., 2009). To determine the role of the conserved Phe residue (F318) in the NPxxY(x)5*,*6F motif present in the MCHR1, we examined the effect of site-directed mutagenesis of this residue on receptor function, and noted a most significant increase in calcium mobilization relative to wild-type after substitution of F318 with a positively-charged lysine residue. Our analyses show that Lys replacement mutation (F318K) produces an efficient signaling property that selectively increases Gαq-mediated pathway without changing cell surface expression. We further discuss the significance of the position of Phe318 using a homology docking model of MCHR1 with Gα<sup>q</sup> and Gα<sup>i</sup> proteins, respectively. To date, this is the first study to provide meaningful insights into the relationship between conformational changes in MCHR1 and G protein activation.

#### **MATERIALS AND METHODS**

#### **cDNA CONSTRUCTS FOR MCHR1 AND MUTANT RECEPTORS**

The generation of a cDNA encoding a Flag epitope tag before the first methionine in rat MCHR1 (NM\_031758/GenBank/EMBL) was described previously (Saito et al., 2003). Single-substitution mutations of the NPxxY(x)5*,*6F domain were produced by oligonucleotide-mediated site-directed mutagenesis using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA). All mutations in the MCHR1 cDNA sequence were confirmed by sequencing analysis. Mutated MCHR1 cDNAs were excised by digestion with EcoRI and XhoI and inserted into the pcDNA3.1 expression vector.

#### **CELL CULTURE AND TRANSIENT TRANSFECTION**

HEK293T cells were cultured in DMEM containing 10% fetal bovine serum. The plasmid DNA was mixed with LipofectAMINE PLUS transfection reagent (Life Technologies Corporation, Carlsbad, CA, USA) and the mixture was diluted with OptiMEM and added to the cells (Saito et al., 2003). For western blotting and GTPγS-binding assays, the cells were re-seeded onto 6-well plates. Cells were replated onto LAB-TEK 8-well plates (Nunc, Rochester, NY, USA) for immunocytochemistry, and onto 24- and 96-well plates (BIOCOAT, Becton Dickinson, Belford, MA, USA) for FACScan flow cytometric analysis and the calcium influx assay, respectively. The re-plated cells were cultured for a further 18–24 h.

## **WESTERN BLOTTING FOR MCHR1**

To generate whole cell extracts, transiently-transfected HEK293T cells were lysed with ice-cold sodium dodecyl sulfate sample buffer [50 mM Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate, 50 mM β-mercaptoethanol, and 10% glycerol], then homogenized at 4◦C by sonication (SONICAOR Ultrasonic processor W-225, Wakenyaku Ltd., Kyoto, Japan) using 5×30 s bursts at 20% power. Aliquots containing 15μg of total protein were separated by SDS-PAGE and electro-transferred to Hybond-P PVDF membranes (GE Healthcare UK Ltd., Little Chalfont, UK). After blocking with 5% skim milk, membrane-expressed Flag-MCHR1 was detected using 0.5 μg/ml anti-Flag M2 antibody (Wako, Osaka, Japan), followed by a horseradish peroxidase-conjugated goat anti-mouse IgG antibody (Saito et al., 2005). Reactive bands were visualized with enhanced chemiluminescence (ECL) reagent (GE Healthcare UK Ltd.).

#### **FACScan FLOW CYTOMETRIC ANALYSIS OF CELL SURFACE RECEPTORS**

Transfected HEK293T cells in 24-well plates were fixed with 1.5% paraformaldehyde-PBS solution for 10 min at room temperature, then incubated with 0.25μg/ml anti-Flag M2 antibody in PBS containing 20% FBS for 1 h. The cells were washed three times with PBS and then incubated with Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody (Molecular Probes, Eugene, OR, USA) for 1 h (Tetsuka et al., 2004; Saito et al., 2005). The cells were washed, harvested with 5 mM EDTA and analyzed using a FACSCalibur flow cytometer (BD, Franklin Lakes, NJ). Cells were gated by light scatter or exclusion of propidium iodide, and 10,000 cells were acquired for each time point. The mean fluorescence of all cells minus the mean cell fluorescence with the Alexa Fluor 488-conjugated secondary antibody only was used for the calculations.

#### **IMMUNOFLUORESCENCE MICROSCOPY**

Transfected HEK293T cells were fixed in a 3.7% paraformaldehyde-PBS solution for 10 min. After two washes with PBS, the cells were transferred, either with or without permeabilization using 0.05% Triton X-100 in PBS for 15 min, into a blocking solution (20% goat serum in PBS) for 30 min, then incubated with 0.5μg/ml anti-Flag M2 antibody for 1 h. The anti-Flag M2 antibody was detected using Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody. Fluorescence imaging was performed using a BZ-9000 microscope (Keyence, Tokyo, Japan). For fluorescence imaging of MCH-induced receptor internalization, cells were pre-incubated at 37◦C in serum-free DMEM for 3 h. Cells were then incubated with 1μM rat MCH for 10, 30, and 60 min at 37◦C in a 5% CO2 incubator. Cells were fixed, permeabilized and then incubated with 0.5μg/ml anti-Flag M2 antibody in PBS containing 20% FBS for 1 h. The cells were washed three times with PBS and then incubated with Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody for 1 h. Fluorescence imaging was performed using a FLUOVIEW FV1000 confocal microscope (Olympus, Tokyo, Japan).

#### **MEASUREMENT OF INTRACELLULAR Ca2<sup>+</sup>**

Measurement of intracellular Ca2<sup>+</sup> was performed as described previously (Saito et al., 2003, 2005; Tetsuka et al., 2004). Transiently transfected cells seeded in 96-well plates were loaded with a non-wash calcium dye (Calcium Assay Kit 5, Molecular Devices Japan, Tokyo, Japan) in Hank's balanced salt solution containing HEPES (pH 7.5) for 1 h at 37◦C. For each concentration of MCH, the level of [Ca2+]i was detected using a FlexStation 3 Microplate Reader (Molecular Devices). The data were expressed as fluorescence (arbitrary units) vs. time. The EC50 values for MCH were obtained from sigmoidal fits using a non-linear curve-fitting program (Prism v3.0; GraphPad Software, San Diego, CA, USA). Rat/mouse/human MCH and Compound 15 were purchased from Peptide Institute (Osaka, Japan) and Bachem AG (Bubendorf, Switzerland), respectively.

# **GTPγS-BINDING ASSAY**

GTPγS-binding assay was performed as described previously (Saito et al., 2005). Aliquots (10μg) of membrane proteins were incubated in GTPγS binding buffer (20 mM HEPES-NaOH pH 7.5, 100 mM NaCl, 5 mM MgCl2, 0.2% BSA and 3μM GDP) containing 0.2 nM [35S]GTPγS (PerkinElmer, Santa Clara, CA, USA) and various concentrations of MCH for 30 min at 30◦C. To determine the non-specific binding, unlabeled GTPγS was added to the binding mixtures to a final concentration of 100μM. Bound [35S]GTPγS was separated from free [35S]GTPγS by rapid filtration through GF/C filters and washed with ice-cold binding buffer. Filters were then immersed in scintillation cocktail (Emulsion-Scintillator Plus; Packard Bioscience, Groningen, The Netherlands) and trapped radioactivity counted using a liquid scintillation counter.

# **MOLECULAR MODELING**

To generate homology models of rat MCHR1 mutant F318K, we used the X-ray structure of constitutively active rhodopsin bound to the C-terminus peptide of the α-subunit of the G protein, transducin (PDB code 2X72), as a structural template. Alignment analysis in sequences of rat MCHR1 and rhodopsin was performed using CLUSTALW2.0 installed in Genetyx v9.0 (Genetyx Corporation, Tokyo, Japan). Initial models of rat MCHR1 mutant F318K with K341L transducin C-terminus peptide (340–350; ILENLKDCGLF) was constructed using the Modelor module installed in Discovery Studio (DS) v3.0 (Accelrys, Tokyo, Japan). After replacement of L341 and E342 to Lys and Asn, the structure of rat MCHR1(F318K) bound to a C-terminus peptide (343–353; IKNNLKDCGLF) of the Gα<sup>i</sup> subunit was optimized using the molecular mechanics and molecular dynamics simulation with a CHARMm force field in the DS. Furthermore, replacement of the Gα<sup>i</sup> peptide with a C-terminus peptide to the Gα<sup>q</sup> subunit (349–359; LQLNLKEYNLV) and subsequent similar optimization in DS provided the model structure of rat MCHR1(F318K) bound to the Gα<sup>q</sup> C-terminus peptide. The X-ray structure of rhodopsin (PDB code 2X72) and the two constructed models (rat MCHR1(F318K)-Gα<sup>i</sup> peptide and rat MCHR1(F318K)-Gα<sup>q</sup> peptide) were superimposed for comparison.

# **RESULTS**

# **EFFECTS OF VARIOUS SINGLE-SITE ALA SUBSTITUTION MUTATIONS OF THE CONSERVED NPxxY(x)5***,***6F motif ON RECEPTOR EXPRESSION AND ACTIVITY**

First, to analyze the function of the NPxxY(x)5*,*6F motif, a series of Ala-substituted mutants were generated, as shown in **Figure 1A**. We transiently transfected Flag-tagged MCHR1 or mutant receptors into HEK293T cells, then examined receptor expression levels by western blotting analysis using an anti-Flag M2 antibody. Several immunoreactive bands were detected in the whole lysate isolated from cells expressing Flag-MCHR1 (**Figure 1B**), some of which corresponded to the predicted molecular masses of MCHR1 variants (approximately 35, 44, 45, and 60 kDa (Saito et al., 2003, 2005; Tetsuka et al., 2004), although additional immunoreactive bands were observed at 45–60 kDa.

**FIGURE 1 | Analysis of the effects of individual substitution mutations in the NPxxY(x)5***,***6F motif of MCHR1 on receptor function in HEK293T cells. (A)** Schematic representation of the conserved NPxxY(x)5*,*6F motif and the C-terminal tail of the rat MCHR1. The C-terminal tail of rat MCHR1 extends 42 residues from the plasma membrane (residue 312–353) and is predicted to include an eighth cytoplasmic helix (helix 8) (Tetsuka et al., 2004). Amino acid residues targeted for mutational analyses are underlined. TM7; transmembrane 7. **(B)** Protein expression of Flag-MCHR1 and mutant receptors. After lysis of transfected cells with SDS-sample buffer, 15μg total protein was separated by 15% SDS-PAGE, transferred to a polyvinyl difluoride membrane, and immunoblotted with an anti-Flag M2 antibody. Four major immunoreactive bands of 35, 44, 45, and 60 kDa are present in Flag-MCHR1 and individual mutant receptors. **(C)** Cells transfected with Flag-MCHR1 or substitution mutant receptors were stimulated with the indicated concentrations of MCH, and the subsequent changes in cytoplasmic free Ca2<sup>+</sup> levels were measured using a FlexStation. Results shown are representative of at least three-independent experiments.

Our previous study revealed that the 35-kDa band is the nonglycosylated form of MCHR1 (Saito et al., 2003), while the three higher molecular mass bands are different *N*-linked glycosylated forms. The migration patterns of N307A, Y311A, and F318A were very similar and no significant reductions in the intensity of the higher molecular mass bands were observed relative to Flag-MCHR1. However, the pattern in P308A was different, with the expression of the higher molecular mass band at 60 kDa apparently drastically reduced (**Figure 1B**, arrow). This phenomenon is likely caused by a lack of appropriate glycosylation of the mutant receptors, as previously shown in an-*N*-linked glycosylation study and other MCHR1 studies (Saito et al., 2003; Aizaki et al., 2009). The cell surface expression levels of Flag-MCHR1 and mutants containing Ala-substitutions in the NPxxY(x)5*,*6F motif were monitored by FACScan flow cytometry using an anti-Flag M2 antibody. Transient transfection of N307A and Y311A gave expression levels of 23% and 30%, respectively, relative to that of Flag-MCHR1, while the F318A mutation was expressed at approximately the same level as the non-mutated control (**Table 1**). Conversely, P308A expression was reduced by more than 90% compared to Flag-MCHR1, suggesting that the mutant was mostly retained intracellularly (**Figure 1B** and **Table 1**).

Next, we assessed the capacity of receptors containing alanine mutations in their NPxxY(x)5*,*6F motif to induce intracellular signals in response to MCH. MCH-induced calcium influx was quantified in transiently transfected cells using a FlexStation 3 Microplate Reader. Mock-transfected HEK293T cells acted as a negative control and did not respond to MCH stimulation (data not shown). Considerable evidence suggests that most singlesubstitution mutations of highly conserved amino acids (such as the DRY or NPxxY(x)5*,*6F motifs) lead to impairment or inactivation of receptor protein signaling. Indeed, substitution of conserved Pro308 with Ala (P308A) resulted in a dramatic attenuation of cell surface expression (**Table 1**). Therefore, this receptor did not respond to MCH by calcium mobilization (**Figure 1C**, **Table 2**), even when challenged with a high concentration (10 μM) of MCH.



The data represent the mean ± S.E.M of three or four-independent experiments performed in triplicates.

aP *<* 0.05, significantly different from Flag-MCHR1 by Student's t-test.

bP *<* 0.01, significantly different from Flag-MCHR1 by Student's t-test.

Substitution of either Asn307 or Tyr311 with Ala also significantly affected MCH-induced calcium mobilization. N307A and Y311A mutant receptors exhibited a maximal response that was 20–30% lower than wild-type receptors and had EC50 values that were 21- and 8-fold higher, respectively, than Flag-MCHR1 (**Table 2**). However, the effects of alanine mutation of Phe318 in the highly conserved motif were distinct from other mutants (Delos Santos et al., 2006; Duvernay et al., 2009; Málaga-Diéguez et al., 2010; Kaye et al., 2011). The EC50 and maximal response of MCH-induced calcium signaling in cells expressing the F318A mutant were essentially identical to those of Flag-MCHR1. This is consistent with our previous data (Tetsuka et al., 2004). These results indicate that the conserved Phe in MCHR1 has a distinct signaling role as compared with other conserved amino acids in the NPxxY(x)5*,*6F motif.

#### **EFFECTS OF INDIVIDUAL SUBSTITUTION OF HIGHLY CONSERVED PHE ON MCHR1 FUNCTION**

To further analyze the effect of the Phe318 substitution in the NPxxY(x)5*,*6F motif, we performed site-directed mutagenesis of Phe318 to proline, a change that is thought to disrupt helix formation and may cause drastic changes in receptor function. We also mutated Phe to Arg and Lys, because these substitutions impart a positive charge to the position. Analysis of receptor expression levels by western blotting analysis with the anti-Flag M2 antibody showed that the migration patterns of F318A, F318P, F318R, and F318K were very similar and no drastic reduction in the intensity of the higher molecular mass bands were observed compared to Flag-MCHR1 (**Figure 2A**). We next determined the expression characteristics of each mutant by quantifying cell surface expression and observing subcellular distribution. The cell surface expression levels of F318A, F318P, F318R, and F318K caused no significant decrease as compared to the Flag-MCHR1 (**Table 1**). Antibody staining in non-permeabilized cells revealed that F318A, F318P, F318R, and F318K were clearly localized in the plasma membrane, and their labeling intensities were approximately equivalent with that of Flag-MCHR1 (**Figure 3**, upper). In permeabilized cells, all four mutants were also predominantly detected in the plasma membrane, as was Flag-MCHR1 (**Figure 3**, bottom). These results confirmed that both the level of cell surface expression and the subcellular localization were unaffected by the Phe318 substitution in the NPxxY(x)5*,*6F motif. This is in marked

**Table 2 | Calcium mobilization stimulated by MCH via Flag-MCHR1 and variants [containing various single point mutations in the highly conserved NPxxY(x)5***,***6F motif] expressed in HEK293T cells.**


P308A does not respond to MCH (–). The results represent the mean ± S.E.M. of at least three-independent experiments performed in duplicate.

contrast with similar mutants of other GPCRs including the α2B-AR (Delos Santos et al., 2006; Duvernay et al., 2009; Málaga-Diéguez et al., 2010), in which membrane translocation was dramatically impaired by mutation of the conserved Phe residue in the NPxxY(x)5*,*6F motif. Using fluorescence microscopy, we also examined the features of receptor internalization following exposure to MCH, because rat MCHR1 undergoes rapid MCHinduced internalization during the 30 min after stimulation (Saito et al., 2004). Consistent with this previous study, treatment with MCH for 30 min caused a loss of membrane localization and the appearance of MCHR1-containing vesicles. This distribution was also observed following 60 min of incubation. A similar timecourse of MCH-induced receptor internalization was obtained for the F318K mutant (**Figure 4**).

The effects of Phe318 mutations on MCHR1 responsiveness to MCH in calcium mobilization are shown in **Table 3** and **Figure 2B**. In cells expressing single-substitution mutants, the MCH EC50 was 7.8 nM and 0.8 nM for F318P and F318R, respectively, with identical maximal responses. Importantly, MCH had

**FIGURE 3 | Confocal immunolocalization of Flag-MCHR1 and the mutant receptors using an anti-Flag M2 antibody in HEK293T cells.** Cell surface expression was compared using transfected non-permeabilized cells (-TX100, without Triton X-100; upper row) and permeabilized cells (+TX100, with Triton X-100; lower row). Vector-transfected cells incubated with the anti-Flag M2 antibody showed no significant staining (data not shown). Bar, 10 μm.

Cells expressing MCHR1 or the F318K mutant were stimulated with 1μM MCH for the time shown, fixed and imaged by confocal fluorescence microscopy. Prior to MCH addition, cells were incubated with serum-free DMEM for 3 h. Bar, 10μm.

increased potency in releasing calcium *via* the F318K mutant receptor (EC50 = 0*.*4 ± 0*.*1 nM, compared with 2*.*5 ± 0*.*8 nM for Flag-MCHR1, **Table 3**), although the maximal response was unchanged. We also mutated the uncharged hydrophobic Phe to other uncharged polar amino acid residues (Gly, Ser, and Cys), a different hydrophobic residue (Trp) and a different basic residue (His) and measured the responsiveness of the resultant mutant receptors in the calcium mobilization assay (**Table 3**). Among the mutants, F318K and F318R exhibited significantly enhanced cellular signaling, but F318K caused a higher responsiveness than F318R. In addition to using MCH itself, we also tested the activity of the mammalian MCH analog, Compound 15, which efficiently binds with high affinity to MCHR1 (Bednarek et al., 2002). The potency of Compound 15 was also enhanced in cells expressing F318K compared to Flag-MCHR1 (0*.*15 ± 0*.*10 nM vs. 1*.*3 ± 0*.*4 nM, respectively; mean ± S.E.M. from three-independent experiments). This enhanced response to MCH was also observed when F318K was transiently transfected into CHO cells and COS7 cells (data not shown).

To elucidate further the effect of Lys mutation, mutagenesis of single or multiple residues around Phe318 was performed, as shown in **Figure 5**. The E316K, T317K, and R319K/R321K **Table 3 | Calcium mobilization by MCH** *via* **Flag-MCHR1 and variants [containing single point mutations at F318 in the NPxxY(x)5***,***6F motif] expressed in HEK293T cells.**


Results represent the mean ± S.E.M. of at least three-independent experiments performed in duplicate.

aP *<* 0.05, significantly different from Flag-MCHR1 by Student's t-test.

bP *<* 0.01, significantly different from Flag-MCHR1 by Student's t-test.

mutations resulted in no higher potency or efficacy of MCH in the calcium mobilization assay. Instead, the T317K and R319K/R321K mutants had a significantly depressed maximal response relative to Flag-MCHR1 (**Table 3**, **Figure 5**). Taken together, the activity of MCHR1 in calcium mobilization is enhanced by the introduction of a positively charged Lys residue at the 318 position, but not at the adjacent amino acid residues, indicating that the 318 position seems to be critical for receptor conformation and/or receptor interaction with the Gα proteins that mediate calcium signaling. We then investigated the effects of a similar mutation in human MCHR2, the human ortholog of rat MCHR1. The corresponding Phe in MCHR2 was mutated to Lys (F313K) and tested for MCH responsiveness in calcium mobilization, since MCHR2 is known to couple exclusively to Gα<sup>q</sup> (An et al., 2001). As shown in **Table 4**, however, F313K was no more responsive than the non-mutated receptor, and rather showed a 22% reduced maximal response. These results imply that the effect of the Phe-to-Lys substitution on receptor signaling seems to be a specific and intrinsic feature of MCHR1.

#### **SELECTIVITY OF F318K FOR G PROTEINS IN MCHR1**

MCHR1-stimulated calcium signaling is mediated through both Gαq- and Gαi*/*o-dependent pathways (Hawes et al., 2000; Tetsuka et al., 2004). To confirm which G protein was primarily responsible for the enhancement of calcium signaling by the F318K receptor, we used pertussis-toxin (PTX) to uncouple Gαi*/*<sup>o</sup> from MCHR1. In cells expressing Flag-MCHR1, PTX pretreatment increased the EC50 value to 3*.*6 ± 0*.*6 nM (compared to 1*.*1 ± 0*.*3 nM in untreated cells), which equates to a +PTX (PTX-insensitive, Gαq-dependent response)/- PTX (combined Gαi*/*<sup>o</sup> and Gαq-dependent response) ratio of 3.30 (i.e., 3.6/1.1 nM). Following PTX pretreatment, the EC50 of MCH in cells expressing F318K was 0*.*48 ± 0*.*08 nM

**residues adjacent to Phe318 in HEK293T cells. (A)** Sequences of Flag-MCHR1 and three mutants for which native residues were replaced with Lys (E316K, T317K and R319K/R321K). **(B)** Dose-response relationship of MCH-stimulated calcium influx in HEK293T cells expressing Flag-MCHR1 or substitution mutant receptors. Cells transfected with Flag-MCHR1 or substitution mutant receptors were stimulated with the indicated concentrations of MCH, and the subsequent changes in cytoplasmic free Ca2<sup>+</sup> levels were measured using a FlexStation. Results shown are representative of at least three-independent experiments.

**Table 4 | Signaling of human MCHR2 and its single-substitution (F313K) mutant after transfection into HEK293T cells.**


Since the F313 residue corresponds to F318 in rat MCHR1, the F313K mutant of human MCH2R was analyzed for its effect on receptor signaling to calcium mobilization. The data represent the mean ± S.E.M of three-independent experiments performed in duplicate.

(compared to 0*.*13 ± 0*.*03 nM in untreated cells), giving a +PTX/−PTX ratio of 3.70. The maximal calcium response to MCH at either receptor was not affected by the addition of PTX. Because PTX reduced the potency of MCH but did not abolish its effect, it is clear that F318K and Flag-MCHR1 both drive calcium influx by a combination of Gαq- and Gαi*/*o-dependent pathways. Because MCH still stimulated calcium influx at a lower agonist concentration in F318K after treatment of PTX (EC50 = 3*.*6 nM at Flag-MCHR1 vs. 0.48 nM at F318K), it is clear that PTX-insensitive Gα<sup>q</sup> protein is involved in mediating calcium responses in both receptors. However, the ratio of MCH potency in the presence and absence of PTX was remarkably similar in both mutated and non-mutated receptors, suggesting that the selectivity of the mutant receptor is likely to be the same as in the wild-type receptor (i.e., that the mutation has not introduced a strong preference for Gαq). However, calcium signals by Gα<sup>q</sup> tend to dwarf those induced by Gαi*/*<sup>o</sup> in HEK293T cells (Tetsuka et al., 2004), so it is difficult to judge what effect the mutation has on the ability of the receptor to couple to Gαi*/*o. Thus, it was necessary to analyze other signal transduction events related to MCHR1-Gαi*/*<sup>o</sup> protein interactions.

It has been shown previously that MCH stimulation of cells expressing MCHR1 can activate a Gαi*/*o-mediated pathway to cause a decrease in adenylyl cyclase activity, thus reducing cAMP production (Chambers et al., 1999; Saito et al., 1999; Hawes et al., 2000). We therefore tried to measure the interaction between MCHR1 and Gαi*/*<sup>o</sup> in cells where cAMP accumulation had been induced by forskolin. Although stable clones for F318K were established, their signaling profiles were very different to transiently transfected cells. For instance, F318K clones were unable to inhibit forskolin-stimulated cAMP accumulation, even at MCH concentrations up to 100 nM MCH (data not shown). These results suggest that the F318K stable clones have lost the ability to couple to Gαi*/*<sup>o</sup> protein. Because the amount of total protein and the glycosylation pattern of these proteins in cells were equivalent to that of Flag-MCHR1 (as determined by western blotting), the uncoupling from Gαi*/*<sup>o</sup> in stably-transfected cells may be due to some aberration in structure and/or functionality when cultured long-term in the presence of antibiotic selection.

Given the difficulties in measuring Gαi*/*<sup>o</sup> protein activation using cAMP assays in stable clones, we employed an alternative technique. It is well-established that the key step in GPCR activation is the induction of guanine nucleotide exchange (GDP-GTP) on the G protein α-subunit. The nucleotide exchange process can be monitored by measuring the binding of non-hydrolyzable GTPγS analog, [35S]GTPγS. Because the Gα<sup>i</sup> family of G proteins has a substantially higher basal rate of guanine nucleotide exchange than other G proteins, this assay is ostensibly a measure of GPCR-mediated activation of Gαi*/*<sup>o</sup> proteins (Milligan, 2003). Indeed, the GTPγS binding assay has been widely used to measure the activation of Gα<sup>i</sup> proteins by various GPCRs. Therefore, we examined both basal and MCH-stimulated [35S]GTPγS binding using the membrane fraction of HEK293T cells transiently expressing Flag-MCHR1 and the F318K mutant (**Figure 6**). In cells expressing Flag-MCHR1, MCH dose-dependently stimulated binding of GTPγS with an EC50 value of 0*.*23 ± 0*.*07 nM, while the value for F318K was 0*.*47 ± 0*.*25 nM (mean ± S.E.M. from three-independent experiments). The values for F318K were slightly higher, but a significant difference was not observed. The maximal amount of binding for Flag-MCHR1 with 1μM MCH was 208*.*3 ± 25*.*2% of basal, while that for F318K was 190*.*6 ± 20*.*3% of basal. Overall, there was no difference in the amount of GTPγS binding between cells expressing Flag-MCHR1 and the F318K mutant, in contrast to the clear influence of the F318K mutation on Gαq-mediated signaling. There was no significant difference in basal GTPγS binding (100% for Flag-MCHR1 vs. 97*.*6 ± 0*.*4% for F318K), suggesting that the F318K mutation does not affect constitutive receptor activation.

**FIGURE 6 | MCH-induced [35S]GTPγS binding to Flag-MCHR1 and F318K.** HEK293T cells were transfected with Flag-MCHR1 or F318K. After 48 h, the cells were harvested and the membrane fractions recovered. Membrane proteins (10μg) were subsequently incubated with 0.2 nM [ 35S]GTPγS and 0.001–1000 nM MCH in GTPγS binding buffer for 30 min at 30◦C. The amounts of radioactivity bound to the membrane preparations are shown for Flag-MCHR1 (filled circles) and F318K (open circles). Results shown are representative of three-independent experiments.

#### **INTERPRETATION OF THE FUNCTIONAL IMPORTANCE OF F318K IN MCHR1**

To understand better how the signaling dynamics of F318K are related to its interaction with (and activation of) Gαq, we constructed a molecular model of F318K activation of G protein based on that for the active structure of rhodopsin in complex with a transducin peptide as a reference (Kleinau et al., 2010). Because MCHR1 belongs to the same GPCR subfamily as rhodopsin, the existing rhodopsin sequence alignment allowed us to construct a preliminary model of MCHR1 conformation. We refined this model to account for the specific properties of MCHR1 using software for protein model building and a molecular dynamics software package. Our model highlighted several putative amino acid contacts between the receptor and Gα<sup>q</sup> protein, notably at intracellular loop 1 (i1) and helix 8 (**Figure 7**, left). Lys318 is predicted to interact efficiently with Asp79 (located in i1) *via* a distinct hydrogen-bonding pattern, and these two residues create an accessible intracellular interface for the Asn357 residue in the Gα<sup>q</sup> protein. However, Asp79 in i1 also exists in human MCHR2, in which F313K had no effect on MCH potency in calcium signaling, as shown in **Table 4**. We therefore searched for alternative candidate residues that may be involved in the interaction with Asn357 by comparing sequences between rat MCHR1 and other GPCRs, including human MCHR2. The sequences we used are for GPCRs in which mutation at the position corresponding to Phe318 of MCHR1 did not produce increased signaling potency (Fritze et al., 2003; Delos Santos et al., 2006; Anavi-Goffer et al., 2007; Duvernay et al., 2009; Málaga-Diéguez et al., 2010; Kaye et al., 2011). By making multiple alignments of these sequences, we found several amino acid insertions in i1 that occurred only in rat MCHR1, and were absent in other GPCRs (**Table 5**). Notably, a sequence in i1 includes Trp73, which possesses a bulky, aromatic side chain. Hydrophobic

Homology model of rat MCHR1 mutant F318K was constructed using the crystal structure of rhodopsin E113Q mutant in complex with the C-terminal tail of transducin (PDB ID 2X72). Amino acid residues involved in coupling to Gα<sup>q</sup> or Gα<sup>i</sup> proteins in intracellular loop 1 (i1) and K318 in helix 8 are shown. **(A)** The model structure of MCHR1(F318K)-Gα<sup>q</sup> C-terminal tail. The area demarcated by the square is magnified below in panel **(C)**. **(B)** The model structure of MCHR1(F318K)-Gα<sup>i</sup> C-terminal tail. The area demarcated by the square is magnified below in panel **(D)**. **(C)** Positions of Lys318, Tyr311, Trp73, and Asp79 residues and their position relative to Asn357 of Gαq. This model depicts the hydrogen bond network between Lys318 and Asp79 in the receptor and Asn357 in Gα<sup>q</sup> C-terminal tail. Note that interactions between Trp73 in i1 loop and Asn357 in the G protein C-terminal tail. Green, red, and blue areas indicate carbon, oxygen and nitrogen atoms, respectively, within the key residues of the receptor. Carbon atoms in the Gα<sup>q</sup> protein are colored cyan. **(D)** Magnified view of Lys318, Tyr311, Trp73, and Asp79 residues in the receptor and their position relative to Gly351 of Gαi. Intramolecular hydrogen interactions predicted between Lys318 and Asp79 in the receptor are shown. Carbon atoms in the Gα<sup>i</sup> protein are colored yellow.

interaction of Trp73 with Asn357 in the Gα<sup>q</sup> C-terminal tail is predicted to stabilize the intracellular interaction between MCHR1 and Gαq. Conversely, the Gly351 residue in Gα<sup>i</sup> has no side chain to interact with Lys318 (**Figure 7**, right), which may explain the lack of enhancing effect of the F318K mutation on Gαi-mediated signaling. Collectively, a predicted model of F318K combined with sequence alignments suggests that the key residues of the MCHR1-Gα<sup>q</sup> interface are Trp73 and Asp79 in the receptor and Asn357 in the Gα<sup>q</sup> protein. We speculate that the F318K mutation somehow enhances this interface, either by promoting heightened interaction or by facilitating G protein activation, possibly by changing the binding affinity (see section "Discussion").

#### **DISCUSSION**

Recent studies have underscored that, the Phe residue is a key position in the conserved NPxxY(x)5*,*6F motif, which connects TM7 and helix 8 (Palczewski et al., 2000; Fritze et al., 2003). Indeed, of 180 class A GPCRs, this Phe residue is the most highly conserved in 135 of these receptors (Okuno et al., 2005). Mutation of the conserved Phe to Ala in rhodopsin and the M1 muscarinic acetylcholine receptor considerably reduced the potency in Gα*t*- and Gαq-mediated signaling, respectively (Fritze et al., 2003; Kaye et al., 2011). Furthermore, Phe-to-Ala substitution in this motif dramatically reduced cell surface expression of α2B-AR, resulting in extensive ER arrest (Duvernay et al., 2009). These data suggest that the highly hydrophobic Phe residue in the conserved NPxxY(x)5*,*6F motif of many GPCRs is necessary for the maintenance of proper receptor conformation, and is often involved in receptor export from the ER. In the present study, we showed that Phe-to-Ala substitution (F318A) in rat MCHR1 did not significantly affect either the level of cell surface expression or MCH-induced calcium mobilization. However, by analyzing a series of substitution mutants, Phe-to-Lys substitution led to the most increased MCH potency in stimulating calcium mobilization, a pathway largely mediated *via* Gαq. In contrast, F318K had no significant effect in the GTPγS-binding assay that measures receptor interaction with Gα<sup>i</sup> protein. These findings are surprising considering that they contradict findings in Phe substitutions of other GPCRs. Therefore, our study provides new insight into the role of the NPxxY(x)5*,*6F motif Phe residue in term of G protein activation. Although the consensus is that this residue is a key determinant of receptor structure and function, its effect may be dependent on the specific sequence of the receptor and the presence or absence of other residues that make up the receptor-G protein binding/activation interface.

We showed that MCH was the most potent in stimulating calcium mobilization in cells expressing F318K among our mutants. To date, many mutations or truncations in MCHR1 have been designed to study the roles of certain residues and sequences in the diverse functionality of the receptor (Macdonald et al., 2000; Saito et al., 2003, 2004, 2005; Tetsuka et al., 2004; Fan et al., 2005). Some caused a slight or moderate decrease in calcium mobilization (Saito et al., 2003, 2004) and the others resulted in more severe impairment or loss of function (Macdonald et al., 2000; Tetsuka et al., 2004; Fan et al., 2005; Saito et al., 2005; Aizaki et al., 2009). F318K is, therefore, the first mutation characterized by an active variant with more efficient signaling properties. We conclude that substitution of Phe318 with the moderately positively charged Lys is most responsible, at least in part, for producing increased MCH potency in stimulating calcium mobilization *via* MCHR1. Replacement with another positively charged amino acid, His, at the 318 position had no effect for signaling enhancement. Furthermore, the highly positively charged Arg could not induce equivalent effects to F318K (**Figure 2**, **Table 3**). These findings suggest that unique and strict physicochemical characteristics, including the extent of positive charge, polarity and the size of the side chain, may be required for this phenomenon to occur.

GPCRs that bind promiscuously to several Gα protein subtypes are useful tools for clarifying the determinants of G protein selectivity. To date, most studies of rhodopsin family GPCRs have emphasized the role of membrane-proximal regions in the i2 and i3 loops and/or the cytoplasmic loop of the receptor (Anavi-Goffer et al., 2007; Kunieda et al., 2007; Kleinau et al., 2010). Our present data suggest that MCHR1 couples to both Gα<sup>i</sup> and



Stars beneath sequence indicate sequence identity across the alignment, and single dot indicates conserved substitutions.

hMCHR2, human MCHR2; rmAChR1, rat muscarinic acetylcholine receptor M1; rA1AR, rat adenosine 1A receptor; ra(2B)-AR, rat α2B adrenergic receptor; hb1AR, human β1 adrenergic receptor; bovineRho, bovine rhodopsin; hCB1, human cannobinoid receptor 1.

Gα<sup>q</sup> proteins *via* overlapping intracellular regions but that Lys mutation of the 318 position in the NPxxY(x)5*,*6F motif appears to affect only the coupling of the receptor to Gαq. Phe318 is located in the juxtamembrane helix 8 in the C-terminal tail. The role of helix 8 is believed to be in regulating G protein activation by GPCRs, yet few studies have established the pivotal role of helix 8 in G protein selectivity. The position in cannabinoid receptor 1 that corresponds to the MCHR1 Phe318 residue is not a Phe but is a hydrophobic residue (Leu404), and the L404I mutation led to impaired activation of Gαi3 but not GαoA among the Gαi*/*<sup>o</sup> protein subgroup (Anavi-Goffer et al., 2007). Furthermore, the positively charged Arg687 in helix 8 of the thyrotropin receptor is important for selective interaction with Gα<sup>q</sup> (Kleinau et al., 2010). In helix 8 of MCHR1, three basic amino acid residues (Arg319, Lys320, and Arg321) are juxtaposed with Phe318. This basic region in helix 8 seems not to be involved in Gα protein coupling specificity, since alteration of the positive charge of these residues (R319Q/K320Q/R321Q) caused an equally strong inhibition of both Gα<sup>q</sup> and Gα<sup>i</sup> (Tetsuka et al., 2004). Taken together, we might conclude that each amino acid residue in helix 8 of MCHR1 may have different roles in determining receptor activity and G protein selectivity. Our clarification here of the significance of the Phe position will hopefully allow additional understanding of dual G protein coupling in other GPCRs.

Upon stimulation, GPCRs undergo a conformational change that results in G protein activation. Light-induced conformational changes in rhodopsin were elucidated by a series of biophysical studies. Evidence indicates that an important hydrophobic pairing between Tyr306 and Phe313 in the NPxxY(x)5*,*6F motif stabilizes the ground state of rhodopsin (Palczewski et al., 2000; Nygaard et al., 2009). Comparative molecular dynamics simulations of MCHR1 have also suggested that the release of intramolecular interactions between the Tyr and Phe residues in the motif is an important step in transitioning from an inactive to an active form (Vitale et al., 2004). We initially hypothesized that the Tyr311-Phe318-based switch in MCHR1, which is the homolog of a crucial relay point in rhodopsin, would be modified by substituting Phe with Lys, with disruption of an aromatic stacking interaction in the resultant ternary complex somehow affecting the efficacy of signaling. However, the activated F318K structure bound with Gα<sup>q</sup> protein (**Figure 7**) has suggested that an additional and distinct interface may exist between Phe318 and Asp79 in the i1 loop. Further analysis with multiple alignment data for other GPCRs showed that close proximity of Phe318, Asp79, and Asn357 is insufficient to induce such signaling changes and that Trp73 in the i1 loop may be important in stabilizing/maintaining the efficient binding/activation of the Gα<sup>q</sup> protein by the activated receptor. A series of combinational mutations involving F318K and several amino acid residues in the i1 loop (Trp, Cys, and Ser, as described in **Table 5**) may clarify the importance and role of each residue. A precise three-dimensional crystal structure of the MCHR1-G protein complex would also facilitate our understanding, but our present model helps to understand how a single amino acid substitution, F318K, can alter the intra-/inter-molecular interactions of the receptor-G protein complex to modulate G protein signaling *via* this receptor. At this time, we cannot exclude the possibility that F318K has an increased binding affinity, although the EC50 value of MCH in the GTPγS-binding assay showed no change between Flag-MCHR1 and F318K. The concept of modes in agonism of GPCRs has recently been expanding (Smith et al., 2011), implying that receptors can convey different signals through different pathways with varying degrees of potency or efficacy depending on the ligand used. Therefore, it may be possible that the functional selectivity observed in F318K results from enhancement of the binding affinity that favors Gαq-binding but not Gαi-binding. Overall, based on our data, we favor an interpretation that F318 is involved in receptor dynamics.

In conclusion, by analyzing a series of mutants, we report for the first time that F318K, a point mutation in the NPxxY(x)5*,*6F motif of MCHR1, most efficiently enhances the potency of MCH in stimulating calcium mobilization *via* MCHR1 without showing any increase in cell surface expression. We speculate that Phe318 might be involved in the interface between the receptor and G protein, and may regulate which G proteins the receptor can bind and activate. Because there are a limited number of GPCR mutant that significantly enhanced the signaling (Reinscheid et al., 2005; Kato et al., 2008), F318K may provide useful clues in understanding the process of intrinsic receptor dynamics. However, further research is required to demonstrate whether the role of Phe318 described here is of physiological and functional significance.

# **REFERENCES**


# **ACKNOWLEDGMENTS**

We thank Yuki Kobayashi, Akiko Kojo, Yui Funakoshi and Saori Utsuda for expert technical assistance and helpful support. Source of funding: This work was supported by research grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Kakenhi 20500337, 00215568 to Yumiko Saito).

receptors based on structurefunction studies on the type 1 angiotensin receptor. *Mol. Cell. Endocrinol.* 302, 118–127.


K. (2003). Functional role of *N*linked glycosylation on the rat melanin-concentrating hormone


modalities of GPCR ligands. *Mol. Cell. Endocrinol.* 331, 241–247.


are required for receptor function. *Endocrinology* 145, 3712–3723.

Vitale, R. M., Pedone, C., de Benedetti, P. G., and Fanelli, F. (2004). Structural features of the inactive and active states of the melaninconcentrating hormone receptors: insights from molecular simulations. *Proteins* 56, 430–448.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 14 August 2012; accepted: 08 November 2012; published online: 26 November 2012.*

*Citation: Hamamoto A, Horikawa M, Saho T and Saito Y (2012) Mutation of Phe318 within the NPxxY(x)*5*,*6*F motif in melanin-concentrating hormone receptor 1 results in an efficient signaling activity. Front. Endocrin. 3:147. doi: 10.3389/fendo.2012.00147*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Hamamoto, Horikawa, Saho and Saito. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# On the existence and function of galanin receptor heteromers in the central nervous system

*Kjell Fuxe1\*, Dasiel O. Borroto-Escuela1 ,Wilber Romero-Fernandez1, Alexander O. Tarakanov2, Feliciano Calvo1, Pere Garriga3, Mercé Tena3, Manuel Narvaez4, Carmelo Millón4, Concepción Parrado5, Francisco Ciruela6, Luigi F. Agnati 7,8, José A. Narvaez4 and Zaida Díaz-Cabiale<sup>4</sup>*

<sup>1</sup> Department of Neuroscience, Karolinska Institutet, Stockholm, Sweden

<sup>2</sup> St. Petersburg Institute for Informatics and Automation, Russian Academy of Sciences, Saint Petersburg, Russia

<sup>3</sup> Centre de Biotecnologia Molecular, Departament d´Enginyeria Química, Universitat Politécnica de Catalunya, Barcelona, Spain

<sup>4</sup> Department of Physiology, School of Medicine, University of Málaga, Málaga, Spain

<sup>5</sup> Department of Histology, School of Medicine, University of Málaga, Málaga, Spain

<sup>6</sup> Unitat de Farmacologia, Departament Patologia i Terapéutica Experimental, Universitat de Barcelona, Barcelona, Spain

<sup>7</sup> Department of Biomedical Sciences, University of Modena and Reggio Emilia, Modena, Italy

<sup>8</sup> Istituto di Ricovero e Cura a Carattere Scientifico, Lido Venice, Italy

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Wing-Ho Yung, Chinese University of Hong Kong, Hong Kong Joao C. Cardoso, University of Algarve, Portugal

#### *\*Correspondence:*

Kjell Fuxe, Department of Neuroscience, Karolinska Institutet, Retzius väg 8, 17177 Stockholm, Sweden. e-mail: kjell.fuxe@ki.se

Galanin receptor (GalR) subtypes 1–3 linked to central galanin neurons may form heteromers with each other and other types of G protein-coupled receptors in the central nervous system (CNS). These heteromers may be one molecular mechanism for galanin peptides and their N-terminal fragments (gal 1-15) to modulate the function of different types of glia–neuronal networks in the CNS, especially the emotional and the cardiovascular networks. GalR–5-HT1A heteromers likely exist with antagonistic GalR–5-HT1A receptor– receptor interactions in the ascending midbrain raphe 5-HT neuron systems and their target regions. They represent a novel target for antidepressant drugs. Evidence is given for the existence of GalR1–5-HT1A heteromers in cellular models with trans-inhibition of the protomer signaling. A GalR1–GalR2 heteromer is proposed to be a galanin N-terminal fragment preferring receptor (1-15) in the CNS. Furthermore, a GalR1–GalR2–5-HT1A heterotrimer is postulated to explain why only galanin (1-15) but not galanin (1-29) can antagonistically modulate the 5-HT1A receptors in the dorsal hippocampus rich in gal fragment binding sites. The results underline a putative role of different types of GalR–5-HT1A heteroreceptor complexes in depression. GalR antagonists may also have therapeutic actions in depression by blocking the antagonistic GalR–NPYY1 receptor interactions in putative GalR–NPYY1 receptor heteromers in the CNS resulting in increases in NPYY1 transmission and antidepressant effects. In contrast the galanin fragment receptor (a postulated GalR1– GalR2 heteromer) appears to be linked to the NPYY2 receptor enhancing the affinity of the NPYY2 binding sites in a putative GalR1–GalR2–NPYY2 heterotrimer. Finally, putative GalR–α2-adrenoreceptor heteromers with antagonistic receptor–receptor interactions may be a widespread mechanism in the CNS for integration of galanin and noradrenaline signals also of likely relevance for depression.

**Keywords: galanin receptors, heteromers, GPCRs, 5HT1A, NPY receptors, allosteric modulator**

"fendo-03-00127" — 2012/10/24 — 14:27 — page 1 — #1

# **INTRODUCTION**

Galanin is a neuropeptide (Tatemoto et al., 1983) widely distributed in neurons within the central nervous system (CNS; Jacobowitz et al., 2004). Three Galanin receptor (GalR) subtypes, GalR1–3, have been cloned and belong to the rhodopsin subfamily of G protein-coupled receptor (GPCR; Branchek et al., 2000; Lundstrom et al., 2005; Mitsukawa et al., 2008). GalR1 and GalR2 in particular are found in many regions of the CNS as demonstrated with *in situ* hybridization, radioligand binding, and immunohistochemical studies and have all a high affinity for galanin. GalR1 and GalR3 are coupled to Gi/o leading to inhibition of adenylate cyclase (AC), increases in MAPK activity and opening of G protein-coupled inwardly rectifying K+ channels. GalR2 is coupled to Gq/11 and its activation leads to increases in phospholipase C with formation of IP3 increasing intracellular calcium levels and of diacylglycerol (DAG) with the subsequent activation of the protein kinase C. These three GalR subtypes are involved in a number of functions in the CNS modulating neuroendocrine, cardiovascular and mood regulation, pain control, food intake, and seizure threshold (Lundstrom et al., 2005; Mitsukawa et al., 2008). Galanin has also been demonstrated to exert neurotrophic and neuroprotective actions. As early as 1988 (Fuxe et al., 1988b) indications for the existence GalR–5-HT1A receptor–receptor interactions were obtained in rat limbic membranes and in 1994 (Mazarati et al., 1994) in a neuropharmacological analysis results were obtained suggesting interactions of GalRs with glutamate receptors in the dorsal striatum.

This review serves to summarize the indications that GalR subtypes may form heteromers with each other and other types of GPCRs in the CNS as a molecular mechanism to modulate the function of different types of glia–neuronal networks in the CNS.

# **RECEPTOR HETEROMERS AND THEIR ALLOSTERIC RECEPTOR–RECEPTOR INTERACTIONS VIA THE RECEPTOR INTERFACE**

We began to test the hypothesis of intramembrane receptor– receptor interactions in 1980–1981 in membrane preparations of various CNS regions and found that neuropeptides could modulate the binding characteristics, especially the affinity of the monoamine receptors, in a receptor subtype specific way (Agnati et al., 1980; Fuxe et al., 1981, 1983). Thus, intramembrane receptor–receptor interactions did exist in addition to indirect actions via phosphorylation and changes in membrane potential. However, it took around 10 years before they began to have an impact in the receptor field. But the good news were that the results were in line with earlier findings by Limbird et al. (1975), showing negative cooperativity in beta-adrenergic receptors, which could be explained by the existence of receptor homodimers leading to receptor–receptor interactions. It was also clear that adapter proteins could be involved in mediating the receptor–receptor interactions in brain membranes. A logical consequence for the indications of direct physical interactions between neuropeptide and monoamine receptors, the term heteromerization was introduced to describe a specific interaction between different types of GPCRs (see Zoli et al., 1993; Fuxe et al., 2012a).

Fluorescence resonance energy transfer (FRET) and bioluminescence resonance energy transfer (BRET) methods gave the evidence needed to demonstrate heteromers among class A GPCRs. In FRET two putative proteins can, e.g., bear a "donor" (CFP or GFP2) or an "acceptor" (YFP). If these two proteins interact then the donor and acceptor fluorophores are likely in proximity (10 nm or less) and energy transfer between donor and acceptor can occur after donor excitation by demonstration of YFP emission. In BRET two putative proteins bear a "donor" (Rluc) or an "acceptor" fluorophore (YFP or GFP2). If these proteins are in proximity they do interact. Then, a donor–acceptor energy transfer can occur after Rluc substrate (h-coelenterazine) oxidation. The bioluminescence formed can activate YFP and YFP emission develops (Fernández-Dueñas et al., 2012; Fuxe et al., 2012a).

More recently, evidence has been presented by means of *in situ* proximity ligation assay (*in situ* PLA) that in HEK293 cells D2*L*R can form heteromers with D4.7R and especially with D4.2R and D4.4R (Borroto-Escuela et al., 2011). *In situ* PLA have the potential to enable a fuller understanding of GPCR receptor–receptor and could be highly suited to investigate GPCR heteromers in tissue, providing new insights in basic biological mechanisms, heteroreceptor levels and their locations, e.g., in the brain (Borroto-Escuela et al., 2012).

It is clear that allosteric mechanisms make possible the integrative activity taking place intramolecularly in monomers and intermolecularly in homomers and heteromers (see Fuxe et al., 2010b). Allostery is a mode of long distance communication between distal sites in proteins. There may or may not exist a conformational change at the binding site by the allosteric communication. The conformational change is only one of the possible scenarios. We have preferred pathways which vary with given conditions, through which the strain energy is released from the allosteric site following a perturbation event which can pass over the receptor interface into the other protomer of the receptor homomer or heteromer (see Fuxe et al., 2010a). Kenakin (2008) rightly regards "Seven TM receptors as Nature's prototype allosteric protein: de-emphasizing the geography of binding". The allosteric receptor–receptor interactions induced by activation of one receptor protomer can influence the recognition, G proteincoupling and signaling and trafficking of the other protomers in the receptor heteromer.

A high energy strength double arginine-phosphate electrostatic interaction has been found in the A2AR–D2R heteromer by Ciruela et al. (2004) and Woods et al. (2005) which possess a covalent-like stability as demonstrated with mass spectrometry in combination with collision-induced dissociation experiments and confirmed by pull-down techniques. Based on a mathematical approach, Tarakanov and Fuxe (2010) have deduced, based on 48 pairs of receptors thatform or notform heterodimers, a set of triplet amino acid homologies that may importantly participate in receptor– receptor interactions with an origin from integrin triplets of marine sponges and toll-like receptor triplets (Tarakanov et al., 2012a,b). We show how such triplets of amino acid residues and their "teams" may be utilized to construct a kind of code that determines (and/or predicts) which receptors should or should not form heterodimers. We propose a "guide-and clasp" manner for receptor–receptor interactions where "adhesive guides" may be the triplet homologies. Ciruela et al. (2004) showed an amazing stability of epitope–epitope electrostatic interaction based on arginine phosphate. Thus, we should note that two main players contain R (Arg): IDR and AAR. Others have distinct relationships to negatively and positively charged amino acids.

The receptor heteromers due to the receptor–receptor interactions (RRI) represent a new target for drugs since they allow *inter alia* a higher selectivity in drug actions. Thus, in principle, it is possible to have not simply drugs acting as agonists, or partial agonists or antagonists on a receptor A that is assembled in a heteromer. Rather it is also possible to modulate receptor A through an action on receptor B which belongs to the same heteromer and interacts, via allosteric RRI, with receptor A (**Figure 1**). Against this background, a new field for pharmacology can be considered. Let us list the possible new opportunities to be investigated in a very simple case that of A–B-receptor heterodimers with the aim to modulate the recognition/decoding processes of receptor A. Due to the RRI in the dimer it could be possible to use: (1) receptor B-agonists, (2) receptor B-partial agonists, (3) receptor B-antagonists, and (4) receptor B allosteric modulators. These treatments could change the recognition/decoding process of:

(a) The complex endogenous ligands/receptor A.

"fendo-03-00127" — 2012/10/24 — 14:27 — page 2 — #2

(b) The complexes receptor A-agonist/receptor A, receptor Apartial agonist/receptor A, or receptor A-antagonist/receptor A or receptor A allosteric modulator/receptor A.

"fendo-03-00127" — 2012/10/24 — 14:27 — page 3 — #3

**integration of transmitter signals.** Three-dimensional molecular models of the seven TM regions of two putative GPCR (A and B) were built by means of the homology modeling program Accelrys Discovery Studio 2.5 (San Diego, CA, USA) to show that dimerization of GPCR can result from either covalent and non-covalent unions between receptor protomers. (Helix–helix interaction) Seen from a lateral view, representation of the A(TM-IV/)–B(TM-V) interaction in the A–B receptor heterodimer is shown. (Electrostatic interactions) Illustration of positively charged arginine-rich epitope (red) in the N-terminal part of the third intracellular loop of receptor A electrostatically interacting with the negatively charged (blue) C-terminal epitopes of the receptor B. These electrostatic interactions may represent important hot spots in the receptor interface of some receptor heteromers like A2AR–D2R,

or intermolecularly in homo/heteromers (arrow with two directions in the interface). Intermolecular allosteric mechanisms take place through the formation of different types of receptor homo/heteromers and receptor/ protein complexes which can change the function of an individual receptor protomer present in a homomer or heteromer. One example of the novel pharmacology created by heteromers is the use of heterobivalent ligands containing an A pharmacophore and a B pharmacophore linked through a spacer of variable length which may function as useful molecular probes for targeting the A–B receptor heteromer and in this way counteracting or enhancing the receptor–receptor interactions in these heteromers. Such compounds may have a potential for use in pharmacotherapy of CNS diseases.

In the case of a trimer or even higher order oligomers that is of a receptor mosaic (RM) we should consider especially


# **POTENTIAL EXISTENCE OF GALANINR–5-HT1A HETEROMERS WITH ANTAGONISTIC GalR–5-HT1A RECEPTOR–RECEPTOR INTERACTIONS IN THE ASCENDING MIDBRAIN RAPHE 5-HT NEURON SYSTEMS AND THEIR TARGET REGIONS. A NOVEL TARGET FOR ANTIDEPRESSANT DRUGS**

A substantial density of high affinity GalRs was demonstrated in the dorsal raphe by the Jacobowitz group in 1986 (Skofitsch et al., 1986) after having shown the existence of galanin IR cell bodies in the dorsal raphe (DR) after colchicine treatment (Skofitsch and Jacobowitz, 1985). This work inspired the Fuxe group to perform intraventricular (ivt) injections with galanin and evaluate its effects on regional 5-HT levels and metabolism (Fuxe et al.,1988a). In 1986 the coexistence of 5-HT and galanin IR in DR cell bodies was found (Melander et al., 1986) and in 1990 only a proportion of the 5-HT nerve cell bodies in the DR was found to costore galanin and 5-HT IR after colchicine (Fuxe et al., 1990). Intraventricular galanin reduced 5-HT metabolism in ventral limbic cortex, hippocampal formation, and fronto-parietal cortex probably via direct inhibitory actions on DR 5-HT nerve cells reducing their firing rates (Fuxe et al., 1988a). These results for the first time suggested based on the 5-HT hypothesis of depression (see Carlsson et al., 1968) that galanin, via actions on GalRs, mainly GalR in the DR, may contribute to depression by reducing firing in the ascending 5-HT neurons (see also Kehr et al., 2002). Thus, GalR antagonists may represent novel antidepressant drugs.

The same year it was also discovered in limbic membrane preparations that galanin in nanomolar concentrations can reduce the affinity of postjunctional [3H]-5HT1A agonist binding sites suggesting that galanin can reduce 5-HT1A recognition and probably signaling in the limbic system (Fuxe et al., 1988b). This indicates that galanin also via such actions can contribute to development of a state of depression, since postjunctional 5-HT1A receptors likely is one of the 5-HT receptors elevating mood upon activation (see Fuxe et al., 1991). This receptor has also been implicated in the pathophysiology of depression based *inter alia* on analysis of brain samples from depressed suicides (Lowther et al., 1997). These results on GalR modulation of 5-HT1A agonist binding sites gave the first indication that brain GalR/5-HT1A heteromers may exist where GalRs antagonize postjunctional 5-HT1A recognition and signaling via intramembrane receptor–receptor interactions (Fuxe et al., 1990, 1991, 2008; Zoli et al., 1993). The antagonistic GalR/5-HT1A receptor interactions in putative receptor heteromers represented a novel integrative mechanism in 5-HT neurotransmission. The role of galanin in the modulation of 5- HT neurotransmission became, however, more complex with the observations that galanin can increase potassium-induced 5-HT release from synaptosomal preparations (Martire et al., 1991). This gave the first indication there may also exist GalRs that increase 5-HT neurotransmission and thus may be beneficial for treatment of depression. In fact, it has recently been proposed that activation of GalR2 can produce antidepressant actions (Lu et al., 2005; see Lundstrom et al., 2005; Mitsukawa et al., 2008). However, the overall inhibitory influence of GalRs was in dominance (see Fuxe et al., 1988a,b). The relevance of antagonistic GalR/5-HT1A interactions for depression were discussed especially in the frame of the 5-HT isoreceptor disbalance hypothesis of depression (see Fuxe et al., 1977, 1991; Ogren et al., 1979) and the galanin-induced reduction of activity in the ascending 5-HT neurons (Fuxe et al., 1991). GalRs counteract the postjunctional mood elevating 5-HT1A signaling while the signaling over other subtypes of 5-HT receptors like the 5-HT2-like receptors which may be blocked by classical antidepressant drugs (Fuxe et al., 1977, 1991; Ogren et al., 1979) were less affected. A behavioral correlate has been obtained to the postjunctional antagonistic GalR/5-HT1A receptor interactions (Razani et al., 2001). The activation of 5-HT1A receptors in membrane preparations and in brain sections was in contrast found to increase the affinity of the GalRs in diencephalic and telencephalic areas (Hedlund et al., 1991a). Thus, reciprocal intramembrane receptor–receptor interactions may exist in putative GalR/5-HT1A heteromers where an activation of 5-HT1A receptors increases GalR recognition and probably GalR signaling. Thus, if the 5- HT1A receptor is regarded as a hub receptor and the GalR as the accessory receptor in this receptor heteromer, it seems possible that the GalR mediates an inhibitory intramembrane feedback mechanism to dampen overactivity in 5-HT1A receptor signaling (Fuxe et al., 1988b, 1991; Hedlund and Fuxe, 1996). However, it is still unknown which GalR subtype protomers are modulated by the agonist-induced activation of the 5-HT1A protomer in the GalR– 5-HT1A heteromers. In other cases, however, the enhancement of the GalR signaling may be the major action, allowing the 5-HT1A protomer to bring down firing in the raphe nuclei further due to increased opening of the GalR regulated GIRK channels leading to increased hyperpolarization (Hedlund et al., 1991b).

# **GalR–5-HT1A AUTORECEPTOR INTERACTIONS IN THE DORSAL RAPHE 5-HT CELL BODIES/DENDRITES**

A time-dependent modulatory action by GalRs exist also on 5- HT1A autoreceptors in the DR nerve cells (Razani et al., 2000). Intraventricular galanin produces first a rapid decrease of the 5-HT1A autoreceptor recognition within 10 min (reduction of affinity) in line with the findings at postjunctional 5-HT1A receptors (Fuxe et al., 1998, 2008). Thus, intramembrane antagonistic GalR/5-HT1A autoreceptor interactions in putative GalR– 5-HT1A autoreceptor heteromers can be demonstrated at the soma-dendritic level of the DR 5-HT nerve cells. This will, however, not increase their nerve cell firing since galanin itself causes hyperpolarization (Xu et al., 1998) and it may represent only an intramembrane inhibitory feedback to avoid abnormal inhibition of the DR 5-HT nerve cells.

However, this change in the 5-HT1A autoreceptor is followed after 2 h by a delayed rise in 5-HT1A autoreceptor density probably through reduced internalization of the GalR/5-HT1A receptor heteromers (Razani et al., 2000) and/or a galanin-induced increase in maturation and insertion of these heteromers into the plasma membrane. This was associated with a reduction of galanin mRNA

"fendo-03-00127" — 2012/10/24 — 14:27 — page 4 — #4

levels and of 5-HT1A mRNA levels in the DR supporting this interpretation. Thus, GalR antagonists may exert antidepressant effects in the DR not only by blockade of GalR signaling but also via blockade of the galanin elicited increase in 5-HT1A autoreceptor density.

A major paper in pointing to a role of raphe GalRs in depression and for the use of GalR antagonists as novel antidepressant drugs was published by Bellido et al. (2002). The discovery was made of an increased density of GalRs in the DR of a genetic rat model of depression based on the use of Flinders Sensitive Line rats. This rise of GalR density was associated with a reduction of galanin IR in the DR and an increase in the immobility time in the forced swimming test. This may represent a primary disturbance contributing to development of human depression by reducing firing in the ascending 5-HT pathways to the limbic system and the diencephalon. These results emphasize the introduction of GalR antagonists targeting the DR in treatment of depression. However, the GalR subtypes appear to have a differential role in depression as demonstrated by Bartfai and colleagues (Bartfai et al., 2004; Lu et al., 2005; Barr et al., 2006). Thus, newly developed non-peptide GalR3 antagonists exert antidepressant like activity in various rodent models of depression (Barr et al., 2006). However, fluoxetine and electroconvulsive shock increase Gal mRNA levels in the DR accompanied by increases in GalR2 receptor binding sites (Lu et al., 2005) giving in contrast an antidepressant potential to GalR2 agonists. In fact, GalR2 may be the GalR involved in releasing 5- HT from synaptosomal preparations (Martire et al., 1991). Based on the available information it therefore seems that general GalR and GalR3 antagonists have antidepressant properties while GalR2 antagonists may reduce mood.

# **EVIDENCE FOR THE EXISTENCE OF GalR1–5-HT1A HETEROMERS IN CELLULAR MODELS**

Previous work has established homodimerization and internalization of GalR1 in living CHO cells using FRET and time lapse confocal imaging (Wirz et al., 2005). Thus, GalR1 can exist as a dimer in the plasma membrane which may undergo desensitization and internalization upon agonist activation with Gal1-29.

We have examined the possible existence of GalR–5-HT1A heteromers in HEK-293 cells co-transfected with GFP2-tagged 5- HT1A receptor and YFP-tagged GalR1 receptor using a proximitybased FRET assay (Borroto-Escuela et al., 2010). In addition, a novel bioinformatic approach to predict receptor–receptor interface interactions was used (Tarakanov and Fuxe, 2010) together with an analysis of signaling.

Upon co-expression of the 5-HT1A-GFP2 and GalR1-YFP cDNA, a significantly higher FRET signal was observed in comparison to the FRET signal obtained from a mixture of cells individually expressing one of the two receptors. It represented a constitutive heteromers since the observed FRET ratios were unaltered by agonist treatments. In cells co-expressing 5-HT1A-GFP2 and GalR1-YFP the two receptors became clearly co-distributed in the plasma membrane at 48 h. And the specificity of this heteromer is indicated by the observation that expression of GalR2 does not block the FRET signal from developing in the 5-HT1A-GFP2 and GalR1-YFP heteromer. Instead, the FRET signal is increased. The understanding of the mechanism underlying this increase may unravel whether the formation of a GalR1–GalR2–5- HT1A higher-order heteromer (RM) takes place (Borroto-Escuela et al., 2010).

Using CRE-luciferase and SRE-luciferase reporter assays it was found that signaling by either the MAPK or AC pathways by these heteromers results in a trans-inhibition phenomenon through their interacting interface via allosteric mechanisms that block the development of an excessive activation of Gi/o linked to each of the receptors and an exaggerated inhibition of AC or stimulation of MAPK activity (**Figure 2**; Borroto-Escuela et al., 2010). These receptor heteromers may exist in the ascending raphe 5-HT pathways in view of the demonstration of antagonistic GalR–5-HT1A receptor interactions in the limbic regions and in the raphe reducing the affinity of the 5-HT1A receptors (see above).

Based on a bioinformatics approach, Tarakanov and Fuxe (2010) have deduced a set of triplet homologies that may be responsible for receptor–receptor interactions. This set consists of two non-intersecting subsets: "pro-triplets" and "contra-triplets". Any pro-triplet appears as a homology in at least one heterodimer but does not appear as a homology in any non-heterodimer. The triplets SNS and LAR may have an important role in the GalR–5-HT1A receptor interface. The locations of the triplet SNS in the same transmembrane domains (TM7) of GalR1 and 5- HT1A were shown. The locations of the triplet LAR were found in the cytoplasmic (intracellular) domains of GalR1 (between TM1 and TM2) and 5-HT1A (between TM5 and TM6). These two triplets may therefore participate in the transmembrane and intracellular components of the interface of the GalR1–5-HT1A heteromer.

# **A GalR1–GalR2 HETEROMER IS PROPOSED TO BE A GALANIN N-TERMINAL FRAGMENT PREFERRING RECEPTOR (1-15) IN THE CNS**

The three cloned receptors are known to show a higher affinity for Gal than for galanin N-terminal fragments like Gal (1-15) (Branchek et al., 1998). A substantial further development of this field was the demonstration of specific N-terminal galanin fragment, galanin (1-15) binding sites in the rat brain emphasizing the powerful role of galanin fragments in galanin communication (**Figure 3**), especially in dorsal hippocampus, neocortex, and striatum having few high affinity galanin (1-29) binding sites (Hedlund et al., 1992). Our hypothesis is that these N-terminal Gal fragment preferring sites may be the result of formation of GalR1/GalR2 heteromers leading to conformational changes in their galanin recognition sites converting them into highly specific galanin fragment binding sites with markedly reduced affinity for galanin (1-29) (Fuxe et al., 2008). It is of high interest that gal (1- 15) given intraventricularly in the rat has been found to produce marked depression-like behavior in the FST and anxiogenic like effects in the open field (Millon Penuela et al., 2012).

# **POSTULATED EXISTENCE OF A GalR1–GalR2–5-HT1A HETEROTRIMER**

"fendo-03-00127" — 2012/10/24 — 14:27 — page 5 — #5

In agreement with above only galanin (1-15) but not galanin (1-29) can antagonistically modulate the 5-HT1A receptors in

"fendo-03-00127" — 2012/10/24 — 14:27 — page 6 — #6

**antagonistic receptor–receptor interactions in serotonin 5-HT1A and galanin receptor (GalR1) heteromers in HEK293 cells and postulated 5-HT1A–GalR1–GalR2 trimers. (A)** Schematic representation of the antagonistic interactions observed after co-treatment with 5-HT1A receptor agonist (8-OH-DPAT) and galanin (peptide 1-29) in co-transfected HEK293 cells. Signaling by either the mitogen-activated protein kinase (MAPK) or adenylyl cyclase (AC) pathways by the 5-HT1A–GalR1 heteromers indicates a trans-inhibition phenomenon through their interacting interface via allosteric mechanisms that block the development of an excessive activation of Gi/o with an exaggerated inhibition of AC or stimulation of MAPK activity. **(B)** A study of the subcellular localization of tagged 5-HT1A and tagged GalR1 in the presence of non-tagged GalR2 showed a co-distribution in the plasma membrane of tagged 5-HT1A and tagged GalR1 already at 36 h in comparison

the dorsal hippocampus and this effect may be blocked by a known GalR antagonist M35 (Hedlund et al., 1994). Thus, a RM of GalR1–GalR2/-5-HT1A receptors may exist especially in the dorsal hippocampus, neocortex, the striatum, and the raphe where galanin fragments may effectively antagonize postjunctional and autoreceptor 5-HT1A recognition and may function via activation of the postulated GalR1/GalR2 heteromer (**Figure 2**). Also a cross-inhibition of the GalR1–GalR2 heteromer by the 5-HT1A protomer in the trimer should be considered. The GalR1–GalR2–5-HT1A heterotrimer may have a higher trafficking rate from the endoplasmic reticulum to the plasma membrane than the GalR1–5-HT1A heteromer (**Figure 2**; Borroto-Escuela et al., 2010).

Thus, known GalR antagonists should be putative antidepressant drugs also by blocking galanin fragment preferring sites in addition to galanin binding sites increasing postjunctional 5- HT1A mediated 5-HT signaling and the firing of the ascending 5-HT pathways (Hedlund et al., 1994; Fuxe et al., 1998). Therefore, known GalR antagonists may have multiple targets and it would be of high interest to develop an antagonist for treatment of depression that selectively target the GalR1–GalR2 heteromer postulated to be the galanin N-terminal fragment receptor.

to cells expressing tagged 5-HT1A and tagged GalR1 only. In this case, the tagged 5-HT1A receptor showed a more reticular distribution while tagged GalR1 was mainly found in the plasma membrane. In the presence of GalR2 a co-distribution in the plasma membrane of the tagged 5-HT1A and tagged GalR1 is obtained already at 36 h and is maintained at 48 h. A facilitated interaction of the GalR1–5-HT1A heterodimer into the plasma membrane by expression of GalR2 can be explained based on the postulated receptor mosaic (5HT1A–GalR1–GalR2), where the endoplasmic reticulum (ER) trafficking rate is higher than the trafficking rate shown by the 5-HT1A–GalR1 heteromer. Available findings can be explained by postulating that a GalR1–GalR2 heteromer is a galanin fragment preferring receptor which in the postulated trimer with the 5-HT1A receptor exerts a strong antagonistic receptor–receptor interaction with the 5-HT1A protomer at the level of 5-HT1A recognition.

Evidence has been presented that N-terminal galanin fragments can more strongly and more potently reduce postjunctional 5- HT1A receptor recognition also in the ventral limbic cortex where also high affinity GalRs exist (Diaz-Cabiale et al., 2000a). These effects were also blocked by a GalRs antagonist. The galanin fragment preferring receptor may again be formed by the heteromerization of GalR1 and GalR2, since they are known to be present here like in the dorsal hippocampus (O'Donnell et al., 1999). The results underline an important role of different types of GalR–5-HT1A heteroreceptor complexes in depression.

# **POTENTIAL EXISTENCE OF GalR–NPYY1 RECEPTOR HETEROMERS IN THE NUCLEUS TRACTUS SOLITARIUS (NTS), THE HYPOTHALAMUS AND THE DORSAL RAPHE WITH ANTAGONISTIC GalR–NPYY1 RECEPTOR INTERACTIONS**

In a series of papers by Diaz-Cabiale and colleagues (Diaz-Cabiale et al., 2006, 2011; Parrado et al., 2007) evidence has been obtained for an antagonistic GalR modulation of NPY receptor mechanisms suggesting the existence of GALR–NPYY1 interactions involving a likely reduction of NPYY1 receptor agonist affinity probably taking place in GalR/NPYY1 receptor heteromers (**Figure 4A**) in NTS, hypothalamus, and DR (Diaz-Cabiale et al., 2006, 2011; Parrado

"fendo-03-00127" — 2012/10/24 — 14:27 — page 7 — #7

et al., 2007). The results open up the possibility that GalR/NPYY1 receptor–receptor interactions in putative heteromers is a frequent phenomenon in CNS with implications for the integrative functions of galanin and NPY in these regions. Thus, the GalR subtype involved has not been determined. This interaction may be based on a GalR/NPYY1 receptor heteromerization where the galanin induced conformational change in the GalR can cause a conformational change in the NPYY1 receptor via the GalR/NPYY1 interface leading to reduced NPYY1 recognition and G protein coupling and thus to reduced NPYY1 receptor signaling.

In the NTS (Diaz-Cabiale et al., 2006) the results suggest the existence of antagonistic GalR–NPYY1 receptor interactions in cardiovascular regions of this nucleus, reducing NPYY1 signaling and thus vasodepressor activity leading to enhanced vasopressor and tachycardic actions of GalR activation in the NTS.

In the hypothalamus (Parrado et al., 2007) evidence was, for the first time, obtained that GalR activation significantly reduced the NPYY1 receptor agonist binding in the hypothalamus without effects on NPYY2 receptor agonist binding. These GalR–NPYY1 receptor interactions have physiological implications since the food intake induced by the NPYY1 receptor agonist is blocked by galanin. The changes observed on c-Fos expression support the hypothesis that GalR activation modulates the response elicited by the NPYY1 agonist.

In the DR we may postulate, on the basis of the galanininduced decrease in NPYY1 binding, an inhibitory GalR/NPYY1 receptor–receptor interaction that modulates behavioral functions associated with mood and motivation. Behavioral and neurochemical studies support a role of galanin and NPY in mood disorders and GalR1-3 and NPYY1 receptors have been the receptors implicated in depression with GalR subtype specific antagonists and NPYY1 agonists having an antidepressant role (Lu et al., 2005; Lundstrom et al., 2005; Ishida et al., 2007; Jimenez-Vasquez et al., 2007; Fuxe et al., 2008). The decrease of NPYY1 agonist binding induced by galanin and the demonstrated crossinhibition of c-Fos expression in the dorsal raphe upon GalR and NPYY1 agonist co-activation provides one possible basis for the use of synergistic interactions of GalR subtype specific antagonists and NPYY1 receptor agonists as a strategy for treatment of depression. Thus, GalR antagonists may also have therapeutic actions in depression by blocking the antagonistic GalR–NPYY1 receptor interactions resulting in increases in NPYY1 transmission and antidepressant effects.

In the amygdala (Parrado et al., 2007) we may instead on the basis of the galanin-induced increase in NPYY1 binding postulate a facilitatory GalR/NPYY1 receptor interaction (**Figure 4B**) that could be expected to produce anxiolytic actions. Behavioral studies support a role of galanin and NPY in reducing anxiety in the amygdala and GalRs, NPYY1, and NPYY5 have been the receptors implicated in this effect (see Moller et al., 1999; Heilig, 2004). The increase of NPYY1 binding induced by galanin in this study provides a possible basis for synergistic interactions of GalR agonists and NPYY1 receptor agonists in counteracting anxiety behavior. The differential modulation of NPYY1 binding by galanin in the hypothalamus and in the amygdala could be explained by *inter alia* the involvement of different GalR subtypes in the interaction with the NPYY1 receptor in the hypothalamus and amygdala, respectively. It may also be that the same GalR subtype and NPY receptor subtypes are involved in the two areas but they may be part of different RMs (cluster of multiple receptors; higher order heteromers) leading to altered GalR/NPYY1 interactions in the two areas due to differences in the multiple receptor–receptor interactions (Rimondini et al., 1999) in discrete RMs (Agnati et al., 1982, 2003; Fuxe et al., 2010a,b). Thus, in this way the GalR may

"fendo-03-00127" — 2012/10/24 — 14:27 — page 8 — #8

subtypes1-3 linked to central galanin neurons may form heteromers with each other and other types of GPCRs in the CNS. These heteromers may be one molecular mechanism for galanin peptides and their N-terminal fragments (galanin 1-15) to modulate the function of different types of glia–neuronal networks in the CNS, especially the emotional and the cardiovascular networks. The allosteric receptor–receptor interactions induced constitutively or by activation of one receptor protomer can influence the recognition, G protein coupling, signaling and trafficking of the other protomers in the receptor heteromer. **(A)** NPYY1–GalRs heteromers likely exist with antagonistic NPYY1–GalR receptor interactions involving a reduction of NPYY1 receptor agonist affinity probably taking place in postulated GalR/NPYY1 receptor heteromers in NTS, hypothalamus, and dorsal raphe. GalR antagonists may also have therapeutic actions in depression by blocking the antagonistic GalR–NPYY1 receptor interactions in putative GalR–NPYY1 receptor heteromers in the CNS resulting in increases

NPYY1 binding induced by GalR activation in this region expected to produce anxiolytic actions. **(C)** Putative GalR–α2-adrenoreceptor heteromers with antagonistic receptor–receptor interactions reducing α2 function may be a widespread mechanism in the CNS for integration of galanin and noradrenaline signals, also of likely relevance for depression. **(D)** Galanin fragment preferring receptor (a postulated GalR1–GalR2 heteromer) appears to be linked to the NPYY2 receptor enhancing the affinity and signaling of the NPYY2 receptor via unknown signaling pathways in a putative GalR1–GalR2–NPYY2 heterotrimer; a reduction of Gi/o signaling may contribute. **(E)** Furthermore, the postulated GalR1–GalR2 heteromer, a postulated galanin N-terminal fragment preferring receptor (1-15), appears to be linked also to the AT1R in a postulated AT1R–GalR1– GalR2 receptor mosaic enhancing the signaling of the AT1R within the NTS via unknown signaling pathways; a reduction of Gi/o signaling may contribute.

reduce NPYY1 signaling in the hypothalamus and increase it in the amygdala.

# **POSTULATED EXISTENCE OF A GalR1–GalR2–NPY Y2 HETEROTRIMER IN THE NTS**

The presence of specific binding sites for the galanin fragment 1-15 in the nuclei involved in central cardiovascular regulation has been described (Hedlund et al., 1992). In the brainstem a high density of these galanin (1-15) binding sites appears within the NTS, supporting the hypothesis of the existence of a receptor with a higher affinity for the N-terminal fragment than for galanin (Diaz-Cabiale et al., 2005b). We have proposed (see above) that these N-terminal galanin fragment preferring sites may be the result of formation of GalR1/GalR2 heteromers leading to conformational changes in their galanin recognition sites converting them into highly specific galanin fragment binding sites with markedly reduced affinity for galanin (Fuxe et al., 2008). As a matter of fact galanin and N-terminal fragment galanin (1- 15) have specific and different roles (hypotensive and vasopressor responses, respectively) in cardiovascular regulation (Diaz-Cabiale et al., 2005b). The N-terminal fragment galanin (1-15) antagonized the cardiovascular effects of galanin (Narvaez et al., 1994) and galanin fragment (1-15) but not galanin decreases baroreceptor reflex sensitivity (Diaz et al., 1996). Galanin and galanin (1-15) also stimulate the expression of c-Fos with different temporal and spatial profiles, especially in the NTS and in the ventrolateral medulla (Marcos et al., 2001). The GalR antagonist M40 is able to block the cardiovascular responses elicited by the N-terminal fragment galanin (1-15) (Narvaez et al., 1994) giving evidence that regular GalR antagonists also can block the galanin fragment preferring receptors (postulated GalR1–GalR2 heteromer).

NPY operates in central cardiovascular regulation through the NPYY1 and Y2 receptor subtypes. Leu31Pro34NPY, a specific NPYY1 receptor agonist, microinjected into the NTS elicits vasodepressor and bradycardic responses (Yang et al., 1993), whereas the injection of the NPY C-terminal fragment (13-36), a specific NPYY2 receptor agonist, leads to vasopressor responses at low doses (Aguirre et al., 1990; Yang et al., 1993). In the study of Diaz-Cabiale et al. (2010) the co-injection of threshold doses of galanin (1-15) and of the NPYY2 agonist resulted in an increase of MAPK of the same magnitude as observed with threshold doses of NPY and galanin (1-15). Galanin (1-15) was also found to specifically increase the NPYY2 agonist binding in the NTS without inducing any effect on NPYY1 agonist binding (Diaz-Cabiale et al., 2010). The increase by galanin (1-15) of the NPYY2 agonist binding may indicate a galanin (1-15)-induced increase of Y2 receptor affinity in the NTS, since the concentration of the NPYY2 agonist used (25 pM) is in the range of the *Kd* value, where mainly affinity changes affect the binding level. This interaction may be based on GalR1–GalR2–NPYY2 receptor heteromerization (**Figure 4D**) where the galanin (1-15) induced conformational change in the GalR1–GalR2 heteromer via the interface with NPYY2 can cause a conformational change in the NPYY2 receptor leading to increased NPYY2 recognition and switching of G protein coupling likely to Gq and thus to increased novel NPYY2 receptor signaling producing the vasopressor activity (**Figure 4D**; Diaz-Cabiale et al., 2010). These results illustrate the high impact of the allosteric receptor–receptor interactions in heteromers in the integrative mechanisms responsible for the organization of the cardiovascular responses from the local circuit level of the NTS.

# **POSTULATED EXISTENCE OF GalR1–GalR2–AT1 HETEROTRIMER IN THE NTS**

Galanin (1-15)/angiotensin II (Ang II) interactions have also been observed in central cardiovascular control (Diaz-Cabiale et al., 2005a). Thus, intracisternal co-injections of threshold doses of Ang II with galanin (1-15) induce a significant vasopressor response that was maintained during the whole recording period, without any significant effect on heart rate. This response was blocked by the AT1 specific antagonist DuP753 (Diaz-Cabiale et al., 2005a). These data suggest the existence of a synergistic interaction between Ang II and galanin (1-15), in which the AT1 and Gal receptor subtypes participate. It may involve allosteric receptor–receptor interactions in a GalR1–GalR2–AT1 heterotrimer (**Figure 4E**) within the NTS and switching of the AT1R to Gq/11 mediated signaling.

# **POSTULATED EXISTENCE OF GalR–**α**2 ADRENORECEPTOR HETEROMERS IN THE CNS**

This proposal is *inter alia* based on the effects of galanin on α2-adrenoreceptor activation were evaluated on central cardiovascular regulation in the NTS using also quantitative receptor autoradiography (Diaz-Cabiale et al., 2000b). Central administration of threshold doses of galanin together with an effective vasodepressor dose of the α2-adrenoreceptor agonist clonidine was found to induce a rapid vasopressor and tachycardic response. On the contrary, the co-injection of threshold doses of clonidine and N-terminal galanin fragment (1-15) did not result in any significant cardiovascular change (Diaz-Cabiale et al., 2000b). These functional findings suggest that galanin, but not galanin fragment (1-15), antagonizes α2-adrenoreceptor signaling via a GalR–α2 adrenoreceptor interaction in a postulated heteromer built up of these receptor protomers (**Figure 4C**). The GalR subtypes involved in these heteromers are presently unknown.

Quantitative receptor autoradiography supported this view since galanin decreased the affinity of the α2-adrenoreceptor agonist [3H]p-aminoclonidine in the NTS and also increased significantly the density of the α2-adrenoreceptor agonist binding sites. These effects disappeared in presence of the specific GalR antagonist M35, demonstrating that this action is a direct consequence of GalR activation by galanin (Diaz-Cabiale et al., 2000b, 2005b). Galanin also reduced the affinity of the α2-adrenoreceptor agonist [3H]p-aminoclonidine in the tel- and diencephalon which was also blocked by the specific receptor antagonist M35 (Diaz-Cabiale et al., 2001). Thus, GalR–α2 adrenoreceptor heteromers with antagonistic receptor–receptor interactions may be a widespread mechanism in the CNS for integration of galanin and noradrenaline (NA) signals.

These antagonistic receptor–receptor interactions participate in the regulation of pre- and postjunctional central NA transmission and many antidepressant drugs exert part of their therapeutic effects by increasing NA transmission via blockade of the NA

"fendo-03-00127" — 2012/10/24 — 14:27 — page 9 — #9

transporter (Carlsson et al., 1966, 1969). Thus, these antagonistic GalR/α2-adrenoreceptor interactions are of relevance for depression and its treatment.

In the NA cell bodies, dendrites and terminal networks the α2 adrenoreceptor function as NA autoreceptors reducing NA cell firing and NA release from the terminal networks. At this level this interaction will therefore favor increases in NA release and thus NA transmission. At the postjunctional level this antagonistic interaction found all over the tel- and diencephalon will change the balance of the pattern of NA isoreceptor activation and favor the increased activation of beta-adrenergic receptor and α1-adrenoreceptor subtypes. In view of the calming influence of α2-adrenoreceptors at the behavioral level (see Fuxe et al., 2012b) the preferential activation of the other adrenergic receptor subtypes by galanin through the antagonistic GalR/α2-adrenoreceptor interaction may result in increased arousal. It will therefore be interesting to explore how this interaction may be altered in the locus coeruleus and in the limbic networks in models of depression.

# **EXISTENCE OF DopamineR–galaninR HETEROMERS IN THE VENTRAL HIPPOCAMPUS**

There exists evidence for the existence of D1R–GalR1 and D5R– GalR1 heteromers in cellular models and D1-like receptors upon agonist-induced activation were shown to enhance the GalR1 induced MAPK signaling (Moreno et al., 2011). D1-like receptor antagonists blocked galanin-induced MAPK activation in the ventral hippocampus and in synaptosomes from this region galanin facilitated acetylcholine release upon co-activation of the D1-like receptors. Thus facilitatory allosteric D1-like-GalR interactions in heteromers may exist in control of hippocampal acetylcholine

#### **REFERENCES**


A novel, systemically active, selective galanin receptor type-3 ligand exhibits antidepressant-like activity in preclinical tests. *Neurosci. Lett.* 405, 111–115.


release and electrophysiological experiments in hippocampal slices using field EPSP recordings suggest a modulatory role of the dopamineR–galaninR heteromers in cholinergic neurotransmission (Moreno et al., 2011).

# **CONCLUSIONS**

GalR subtypes may have a major role in modulating the emotional networks of the brain through heteromerization with 5-HT1A, NPYY1, and α2-adrenoreceptors leading to antagonistic allosteric receptor–receptor interactions producing reductions in 5-HT1A, NPYY1, and α2-adrenoreceptor pre and especially postsynaptic signaling in the central 5-HT and NA neurons. This may be one way in which the activity at certain GalR subtypes and at galanin fragment preferring receptors may contribute to a reduction of mood, which may lead to depression. The GalR heteromers also participate in cardiovascular functions, food intake and regulation of fear and anxiety. The hypothesis is introduced that the galanin fragment preferring receptor is formed by the GalR1–GalR2 heteromer which can mediate the strong depressant actions of Gal 1-15 upon intraventricular injections. Its postulated formation of a trimer with 5-HT1A receptors may represent a novel target for antidepressant drugs.

# **ACKNOWLEDGMENTS**

This work has been supported by grants from *Fundacio la Marato* de TV3, Generalitat de Catalunya, Barcelona, Spain and from the Swedish Research Council (04X-715) to Kjell Fuxe. Karolinska Institutets Forskningsstiftelser 2010 and 2011 to Dasiel O. Borroto-Escuela. Alexander O. Tarakanov has not received any support for this work.

et al. (2012). The existence of FGFR1- 5-HT1A receptor heterocomplexes in midbrain 5-HT neurons of the rat: relevance for neuroplasticity. *J. Neurosci.* 32, 6295–6303.


"fendo-03-00127" — 2012/10/24 — 14:27 — page 10 — #10


"fendo-03-00127" — 2012/10/24 — 14:27 — page 11 — #11

antidepressant-like effect via Y1 receptor. *Hippocampus* 17, 271–280.


receptors with glutamate receptors in the neostriatum: a neuropharmacological analysis. *Neurophysiology* 26, 230–235.


of the novel galanin receptor subtype GALR2 in the adult rat CNS: distinct distribution from GALR1. *J. Comp. Neurol.* 409, 469–481.


binding sites in the rat central nervous system. *Peptides* 7, 1029–1042.


"fendo-03-00127" — 2012/10/24 — 14:27 — page 12 — #12

into the nucleus tractus solitarius of the rat counteract the vasodepressor responses of NPY(1-36) and of a NPY Y1 receptor agonist. *Brain Res.* 621, 126–132.

Zoli, M., Agnati, L. F., Hedlund, P. B., Li, X. M., Ferre, S., and Fuxe, K. (1993). Receptor–receptor interactions as an integrative mechanism in nerve cells. *Mol. Neurobiol.* 7, 293–334.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 26 August 2012; paper pending published: 16 September 2012; accepted: 05 October 2012; published online: 26 October 2012.*

*Citation: Fuxe K, Borroto-Escuela DO, Romero-Fernandez W, Tarakanov AO, Calvo F, Garriga P, Tena M, Narvaez M, Millón C, Parrado C, Ciruela F, Agnati LF, Narvaez JA and Díaz-Cabiale Z (2012) On the existence and function of galanin receptor heteromers in the central nervous system. Front. Endocrin. 3:127. doi: 10.3389/fendo.2012.00127*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Fuxe, Borroto-Escuela, Romero-Fernandez, Tarakanov, Calvo, Garriga, Tena, Narvaez, Millón, Parrado, Ciruela, Agnati, Narvaez and Díaz-Cabiale. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Differential roles of orexin receptors in the regulation of sleep/wakefulness

# **Michihiro Mieda\*, Natsuko Tsujino and Takeshi Sakurai \***

Department of Molecular Neuroscience and Integrative Physiology, Graduate School of Medical Science, Kanazawa University, Kanazawa, Ishikawa, Japan

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Etienne Challet, Centre National de la Recherche Scientifique, France Christelle Peyron, INSERM U1028, France

#### **\*Correspondence:**

Michihiro Mieda and Takeshi Sakurai, Department of Molecular Neuroscience and Integrative Physiology, Graduate School of Medical Science, Kanazawa University, 13-1 Takara-machi, Kanazawa, Ishikawa 920-8640, Japan. e-mail: mieda@med.kanazawa-u.ac.jp; tsakurai@med.kanazawa-u.ac.jp

**INTRODUCTION**

The neuropeptides orexin A and orexin B (also known as hypocretin 1 and hypocretin 2), were originally identified as endogenous ligands for two orphan G-protein-coupled receptors (Sakurai et al., 1998). Both orexin A and orexin B are derived from a common precursor peptide, prepro-orexin. An mRNA encoding the same precursor peptide of hypocretin 1 (corresponding to orexin A) and hypocretin 2 (orexin B) was independently identified by de Lecea et al. (1998) as a hypothalamus-specific transcript.

The actions of orexins are mediated by two receptors, orexin 1 (OX1R) and orexin 2 (OX2R) receptors (also known as HCRTR1 and HCRTR2) (Sakurai et al., 1998). OX1R shows a higher affinity for orexin A than orexin B by one-order, while OX2R binds orexin A and orexin B with similar affinities. Both receptors are coupled to Gq/11 subclass of G proteins and caused strong excitatory effects on neurons examined thus far (Sakurai and Mieda, 2011). OX2R has also been reported to couple to Gi/o in a neuronal cell line when overexpressed (Zhu et al., 2003).

Orexins are produced in a specific population of neurons (orexin neurons) that are located exclusively in the hypothalamus, including the lateral hypothalamic area (LHA), perifornical area, and posterior hypothalamus (PH) (de Lecea et al., 1998; Peyron et al., 1998; Sakurai et al., 1998; Date et al., 1999; Nambu et al., 1999). Orexin neurons project in almost all brain areas with especially dense projections to monoaminergic neurons, which play important roles in the regulation of sleep/wake states. Orexins are thought to be a critical regulator of sleep/wake states. Intracerebroventricular (ICV) administration of orexin A and orexin B increases wakefulness and suppresses both rapid-eye-movement (REM) sleep and non-REM (NREM) sleep (Hagan et al., 1999). More importantly, orexin deficiency causes sleep disorder narcolepsy in humans and animals (Chemelli et al., 1999; Lin et al.,

Orexin A and orexin B are hypothalamic neuropeptides that play critical roles in the regulation of sleep/wakefulness, as well as in a variety of physiological functions such as emotion, reward, and energy homeostasis. The actions of orexins are mediated by two receptors, orexin 1 (OX1R) and orexin 2 (OX2R) receptors. OX1R and OX2R show partly overlapping but distinct distributions throughout the central nervous system, suggesting their differential roles.This review presents and discusses the current knowledge concerning the physiological roles of each orexin receptor subtype, focusing on the regulation of sleep/wakefulness.

**Keywords: orexin, hypothalamus, sleep, monoamine, narcolepsy, arousal**

1999; Peyron et al., 2000; Thannickal et al., 2000; Hara et al., 2001). At present, various reports suggest that orexins are involved not only in the sleep/wake regulation but also in other physiological and behavioral processes such as food intake, emotion, stress response, and reward via activation of OX1R and OX2R (Yamanaka et al., 2003; Akiyama et al., 2004; Mieda et al., 2004; Boutrel et al., 2005; Harris et al., 2005; Sakurai et al., 2005; Narita et al., 2006; Yoshida et al., 2006). Recent reports revealed that each orexin receptor has a different function in these processes. Thus, understanding such a difference is important for understanding the function of the orexin system.

This review will discuss the physiological roles of the orexin system in the regulation of sleep/wakefulness, especially focusing on the functions of each orexin receptor.

# **DISTRIBUTIONS OF OREXIN RECEPTOR mRNA**

According to data obtained by *in situ* hybridization histochemistry, many brain regions have been shown to differentially express *OX1R* and *OX2R* mRNAs (Trivedi et al., 1998; Lu et al., 2000; Greco and Shiromani, 2001; Marcus et al., 2001). Consistent with the broad projections of orexin neurons, *OX1R* and *OX2R* show wide distributions within the brain with partly overlapping but distinct and complementary distributions. For example, *OX2R* is expressed in layers 2 and 6 throughout the cerebral cortex, while *OX1R* is expressed in layers 5 and 6, except the cingulate cortex where *OX1R* mRNA is found in layer 3. In the ammon's horn of the hippocampus, CA1 and CA2 express *OX1R*, while CA3 expresses *OX2R*.

In the arcuate nucleus and paraventricular nucleus of the hypothalamus, which are involved in the regulation of feeding, energy homeostasis, autonomic and endocrine systems, *OX2R* is almost exclusively observed. Concerning the nuclei implicated

in sleep/wake regulation, the locus ceruleus (LC), laterodorsal tegmental nucleus (LDT), and pedunculopontine tegmental nucleus (PPT) mainly express *OX1R* mRNA, while the tuberomammillary nucleus (TMN) almost exclusively expresses *OX2R* mRNA. The dorsal raphe (DR) and median raphe (MnR) express both subtypes. These distributions suggest partly overlapping and partly distinct roles of these two receptors.

Monoaminergic (i.e., histaminergic, noradrenergic, and serotonergic) and cholinergic neurons in these nuclei have been considered critical for sleep/wakefulness regulation (Pace-Schott and Hobson, 2002; Zeitzer et al., 2006). Our recent work determined precise cellular localization of two orexin receptors in these nuclei (**Figure 1**) (Mieda et al., 2011). All histaminergic neurons in the TMN exclusively expressed *OX2R*, while all noradrenergic neurons in the LC exclusively expressed *OX1R*. In the DR and MnR, approximately 90% of serotonergic neurons expressed *OX1R* and/or *OX2R*. In addition, many non-serotonergic cells in the DR/MnR also expressed *OX1R* or *OX2R* mRNA. At least some populations of these cells were likely to be GABAergic, since a population of *Gad1*-positive cells also expressed detectable *OX1R* or *OX2R* mRNA. In the LDT and PPT, all cholinergic neurons expressed *OX1R* but not *OX2R* mRNA, while many *OX1R*-positive and/or *OX2R*-positive non-cholinergic neurons were intermingled with cholinergic neurons in the area. *Gad1* mRNA staining further revealed that *OX1R*- or *OX2R*-expressing cells included both GABAergic and non-GABAergic cells.

Orexin receptors-positive GABAergic neurons in the DR/MnR and LDT/PPT, which are intermingled with serotonergic and cholinergic neurons, respectively, are intriguing. Since GABAergic neurons are likely to function as local inhibitory interneurons and orexin receptors usually cause excitatory effects on neurons, simultaneous activation of serotonergic/cholinergic neurons with GABAergic interneurons by orexins within a nucleus may inhibit serotonergic/cholinergic neurons. Thus, balance of orexinergic activation between principal neurons and interneurons may be of importance. This point will be further discussed in a later section.

# **OREXIN AND MONOAMINERGIC/CHOLINERGIC SYSTEMS**

Orexin neurons project almost all brain regions with especially dense projections being seen in monoaminergic and cholinergic nuclei involved in the regulation of sleep/wakefulness. Various reports suggested that orexin excites these neurons directly and/or indirectly *in vivo* and *in vitro* (**Figure 1**). This orexinergic regulation may be important to control sleep/wakefulness. Here, we will review *in vitro* and *in vivo* electrophysiological studies of orexin receptors in monoaminergic and cholinergic neurons.

#### **SEROTONERGIC NEURONS OF THE DORSAL RAPHE**

In 2001, Brown et al. reported that orexin A strongly excited DR serotonergic neurons in brain slices of rats under the whole-cell patch clamp recording. The depolarization persisted in the presence of the voltage dependent Na<sup>+</sup> channel blocker, tetrodotoxin,

**FIGURE 1 | Schematic illustration of presumed pathways underlying orexin actions on NREM and REM sleep (Mieda et al., 2011)**. Orexins activate histaminergic (His)/GABAergic (GA), serotonergic (5HT), noradrenergic (NA), and cholinergic (ACh) neurons as well as GABAergic putative interneurons in wake-promoting nuclei, including the TMN, DR/MnR, LDT/PPT, and LC. These neurons differentially express OX1R and/or OX2R, and regulate wakefulness/NREM sleep and NREM/REM sleep transitions. OX1R and OX2R may be expressed in the same populations of GABAergic neurons, as shown in the figure, or may be expressed in distinct populations of these neurons in each area.

Wake/REM-on cholinergic neurons (ACh/W) are likely to suppress NREM sleep, but REM-on cholinergic neurons (ACh/R), are likely to induce REM sleep. Wake-active serotonergic and noradrenergic neurons in the DR/MnR and LC, respectively, counteract activation of REM-on cholinergic neurons in the LDT/PPT, as well as REM-on neurons in the brainstem reticular formation (Pace-Schott and Hobson, 2002; Sakurai, 2007). Previous reports have suggested contributions of GABAergic interneurons inhibiting PPT cholinergic and raphe serotonergic neurons (Liu et al., 2002; Takakusaki et al., 2005). LHA, lateral hypothalamic area; PH, posterior hypothalamus. Reproduced from Sakurai and Mieda (2011) with permission.

indicating a direct postsynaptic action on the serotonin cells (Brown et al., 2001). Moreover, orexin A and orexin B induced inward currents with similar amplitude and dose-dependency in most serotonergic neurons (Brown et al., 2001; Liu et al., 2002; Soffin et al., 2004) via activation of Na+/K<sup>+</sup> non-selective cation channels (Liu et al., 2002). These results may suggest that OX2R is mainly involved in the activation of serotonergic neurons, although single-cell PCR, *in situ* hybridization, and immunohistochemical studies revealed that serotonergic neurons express both *OX1R* and *OX2R* (Brown et al., 2001; Liu et al., 2002; Wang et al., 2005; Mieda et al., 2011).

A recent report revealed that orexin A depolarized DR neurons by activating a noisy inward current (Kohlmeier et al., 2004, 2008). This current was through non-selective cation channels that did not contribute to the somatic Ca2<sup>+</sup> influx. These reports suggested that orexin A has two independent actions: activation of non-selective cation channels and activation of a protein kinase C-dependent enhancement of Ca2<sup>+</sup> transients mediated by L-type Ca2<sup>+</sup> channels.

At higher concentrations, orexin also increased spontaneous inhibitory postsynaptic currents in serotonergic neurons indicating orexin excited GABAergic interneurons that project to serotonergic neurons in the DR (Liu et al., 2002). Such a structure of intra-DR circuit may execute a negative feedforward regulation of the activation of serotonergic neurons by orexin neurons.

Regulation of DR neurons by orexin seems to be even more complex. Haj-Dahmane and Shen found that orexin B depresses the evoked glutamate-mediated synaptic currents in DR serotonergic neurons. This effect was mediated by retrograde endocannabinoid release, which depended on the stimulation of postsynaptic orexin receptors and subsequent activation of phospholipase C and DAG lipase enzymatic pathways but not on a rise in postsynaptic intracellular calcium (Haj-Dahmane and Shen, 2005).

*In vivo* extracellular recording also revealed that application of orexin A increased firing frequency of serotonergic DR neurons in unanesthetized rats during REM sleep or slow-wave sleep, while during wakefulness, a similar amount of orexin A did not increase the firing rate (Takahashi et al., 2005). These reports suggest that orexin excited DR serotonergic neurons in a behavioral state-dependent manner.

#### **NORADRENERGIC SYSTEM**

Orexin A and orexin B increased the firing frequency of all LC neurons in the presence of tetrodotoxin in rat brain slices by intracellular recording (Horvath et al., 1999; Soffin et al., 2002). The effects of orexin A are fivefold more potent than orexin B. Moreover, the effect was inhibited by selective orexin 1 receptor antagonist, SB-334867 (Soffin et al., 2002, 2004). These data suggest that orexins directly excited LC noradrenergic neurons via OX1R, which is consistent with exclusive expression of *OX1R* in these neurons demonstrated by *in situ* hybridization (Mieda et al., 2011). The orexin-induced depolarization of LC noradrenergic neurons may be produced by augmentation of the non-selective cationic conductance and suppression of G-protein-coupled inward rectifier (GIRK) channel activity (Hoang et al., 2003; Murai and Akaike, 2005).

In addition to the action of orexins within the LC, orexin A and B also evoked noradrenaline release in the rat cerebrocortical slice, suggesting that orexins act on OX1R on the orexin fibers projecting from the LC in the cerebral cortex (Hirota et al., 2001). Orexin-evoked noradrenaline release was time- and concentration-dependent and partially extracellular Ca2+-dependent (Hirota et al., 2001).

Another report suggested that the orexin modulates LC functions via OX1R by regulating noradrenaline release not only from the axonal terminals, but also from the somatodendritic region of LC noradrenergic neurons (Chen et al., 2008). Application of orexin alone dose-dependently induced somatodendritic noradrenaline release. Orexin A also potentiated *N*-methyld-aspartate (NMDA) receptor-mediated somatodendritic noradrenaline release from LC neurons, which was blocked by SB-334867, an OX1R antagonist, or a PKC inhibitor, indicating the involvement of OX1R and PKC. Orexin A enhanced NMDAinduced intracellular Ca2<sup>+</sup> elevation as well (Chen et al., 2008). Taken together, activation of OX1R of LC noradrenergic neurons regulates not only noradrenergic input to their targets, but also noradrenergic communication in the soma and dendrites.

Excitatory effects of orexin A on LC neurons have been demonstrated also *in vivo* by measuring their firing rates (Bourgin et al., 2000) or noradrenaline release into the hippocampus (Walling et al., 2004). In addition, local administration of orexin A but not orexin B in the LC of rats suppressed REM sleep in a dosedependent manner and increased wakefulness at the expense of REM sleep and deep slow-wave sleep (Bourgin et al., 2000).

#### **HISTAMINERGIC SYSTEM**

Both patch clamp study using freshly isolated neurons and intracellular recording using brain slices revealed that orexin A and orexin B directly and dose-dependently depolarized TMN histaminergic neurons of rats (Bayer et al., 2001; Eriksson et al., 2001; Yamanaka et al., 2002). Orexin A and B showed almost the same potency, indicating that orexin-induced excitation of histaminergic neurons is mediated via OX2R. Consistent with electrophysiological data, *OX2R* was shown to be the principle subtype of orexin receptors by single-cell RT-PCR study, although both receptor mRNAs were detected in most tuberomammillary neurons (Eriksson et al., 2001). Histological studies have demonstrated that TMN histaminergic neurons express *OX2R* almost exclusively (Yamanaka et al., 2002; Mieda et al., 2011).

Orexin-induced depolarization was associated with a small decrease in input resistance. The mechanisms of depolarization of histaminergic neurons are likely to be related to the activation of both the electrogenic Na+/Ca2<sup>+</sup> exchanger and Ca2<sup>+</sup> current (Eriksson et al., 2001), as well as suppression of GIRK channels (Hoang et al., 2003).

On the other hand, a recent *in vitro* optogenetic study demonstrated that postsynaptic currents and rapid increase of firing rates in TMN histaminergic neurons evoked by stimulation of orexin neurons were mediated by glutamate rather than by orexins (orexin neurons are also glutamatergic) (Schone et al., 2012). This suggests a possibility that orexin might play as a modulator of the glutamatergic transmission in these cells.

*In vivo* application of orexin A to the TMN increased histamine release in both the medial preoptic area and the frontal cortex in a dose-dependent manner (Huang et al., 2001). Moreover, perfusion of orexin A into the TMN of rats through a microdialysis probe promptly increased wakefulness, concomitant with a reduction in REM and NREM sleep (Huang et al., 2001; Yamanaka et al., 2002). These findings indicate that orexins excite TMN histaminergic neurons mainly via OX2R and enhance histamine release from the TMN to maintain wakefulness.

#### **CHOLINERGIC SYSTEM**

#### **Laterodorsal tegmental nucleus**

Whole-cell recordings revealed that orexin exhibited actions both presynaptically and postsynaptically on LDT cholinergic neurons, as well as in non-cholinergic neurons in the same area. Orexin increased frequency and amplitude of spontaneous EPSCs in these neurons. Postsynaptically, orexin produced an inward current and an increase in membrane current noise, which were accompanied by a conductance increase (Burlet et al., 2002). Being similar to DR serotonergic neurons, orexin A has two independent actions on LDT cholinergic neurons: activation of non-selective cation channels and activation of a protein kinase C-dependent enhancement of Ca2<sup>+</sup> transients mediated by L-type Ca2<sup>+</sup> channels (Kohlmeier et al., 2004, 2008).

In a study of *in vivo* extracellular recordings of LDT neurons in rats, the application of orexin A induced long-lasting excitation in five of seven cholinergic neurons (Takahashi et al., 2002). Furthermore, when orexin A was microinjected in the LDT of cats, time spent in wakefulness was significantly increased due to an increase in the duration of wake episodes, and time spent in REM sleep was significantly reduced due to a decrease in the frequency of active sleep episodes (Xi et al., 2001).

#### **Pedunculopontine tegmental nucleus**

Whole-cell patch clamp recordings revealed that orexin A and orexin B depolarized PPT cholinergic and non-cholinergic neurons dose-dependently in the presence of tetrodotoxin (Kim et al., 2009a). Approximately 80% of PPT neurons were depolarized by orexin A, and 20% of PPT neurons did not respond to orexin A (Kim et al., 2009b). SB-334867, a selective inhibitor for OX1R, significantly suppressed the orexin A-induced depolarization. These results suggest that orexin depolarized PPT neurons directly through OX1R. This OX1R-mediated depolarization was caused by a decrease of K<sup>+</sup> conductance and an increase of non-selective cationic conductance (Kim et al., 2009a). It was also indicated that it was blocked by D609, a phosphatidylcholine-specific PLC inhibitor, suggesting that the excitatory effects of orexin on PPT neurons are mediated by PLC-dependent pathway (Kim et al., 2009b).

An *in vivo* study demonstrates that orexin inhibited PPT neurons via GABAergic neurons (Takakusaki et al., 2005). In decerebrated cats, an injection of orexin into the PPT increased the intensity of electrical stimulation required at the PPT to induce muscle atonia. The effect of orexin on PPT was abolished by simultaneous injection of bicuculline, a GABA<sup>A</sup> receptor antagonist, into the PPT. These results suggest that orexin enhances GABAergic effects on presumably cholinergic PPT neurons *in vivo* to control postural muscle tone.

Cholinergic neurons in these areas include two functionally distinct types of neurons. Some are active during wakefulness and REM sleep (W/REM-on neurons) and others are specifically active during REM sleep (REM-on neurons) (Pace-Schott and Hobson, 2002; Sakurai, 2007). The latter population is likely to play a critical role in REM sleep-related physiological phenomena, including muscle atonia. Orexin A excites both cholinergic and non-cholinergic neurons of the LDT in slice preparations, as described previously (Burlet et al., 2002). Furthermore, orexin A microinjection in the cat LDT increases wakefulness and reduces REM sleep. In addition, as mentioned previously, GABAergic neurons in the PPT mediate suppression of REM sleep and muscle atonia following local injection of orexin A into this area (Takakusaki et al., 2005). Consistent with the actions of orexins on both cholinergic and non-cholinergic neurons, orexin receptors are differentially expressed in cholinergic and GABAergic neurons of the LDT/PPT, shown by a histological study (Mieda et al., 2011). Taken together, these results indicate that in the LDT/PPT, orexin may activate W/REM-on cholinergic neurons through OX1R to facilitate wakefulness. Simultaneously, orexin might activate GABAergic interneurons to inhibit REM-on cholinergic neurons in these nuclei (**Figure 1**).

#### **Basal forebrain**

Electrophysiological studies were performed on cultured nucleus basalis neurons from the basal forebrain and on rat brain slice (Eggermann et al., 2001; Hoang et al., 2004). In cultured neurons, orexin A induced a depolarization and increased firing accompanied by a decrease of whole-cell conductance in cholinergic neurons. The mechanism of this neuronal excitation was shown to be accompanied by inhibition of inward rectifier K<sup>+</sup> channel (KirNB) activity (Hoang et al., 2004). In rat brain slices, orexin has a direct excitatory effect on the cholinergic neurons of the contiguous basal forebrain (Eggermann et al., 2001). All cholinergic neurons were excited by orexin A and even more potently by orexin B. This result suggests that the action depends on OX2R. Thus, orexins may excite cholinergic neurons in the basal forebrain via OX2R and contribute to the cortical activation.

The medial septum-diagonal band of Broca (MSDB), via its cholinergic and GABAergic projections to the hippocampus, controls the hippocampal theta rhythm and associated learning and memory functions. MSDB receives a dense innervation of orexin neurons, and neurons of the MSDB express very high levels of OX2R. Septohippocampal cholinergic neurons were excited by orexin A and B with similar EC<sup>50</sup> in a concentrationdependent manner, mediated via a direct postsynaptic mechanism (Wu et al., 2004). The orexin effect is likely to be mediated by inhibition of a K<sup>+</sup> current, presumably an inward rectifier, and activation of the Na+–Ca2<sup>+</sup> exchanger. Thus orexin effects within the septum should increase hippocampal acetylcholine release and thereby promote hippocampal arousal (Wu et al., 2004).

# **DIFFERENTIAL ROLES OF EACH OREXIN RECEPTOR SUBTYPE IN THE REGULATION OF SLEEP/WAKEFULNESS**

#### **MOLECULAR GENETIC STUDIES**

Recent studies have established that the orexin (hypocretin) system is one of the most important regulators of sleep/wake states and that its deficiency results in the human sleep disorder narcolepsy (Zeitzer et al., 2006; Sakurai and Mieda, 2011). Narcolepsy is characterized by excessive daytime sleepiness that often results in "sleep attacks" (sudden onset of NREM sleep), cataplexy (sudden bilateral skeletal muscle weakening triggered by emotions without loss of consciousness), hypnagogic hallucinations, and sleep paralysis. These symptoms can be divided into two independent pathological phenomena. One is the inability to maintain a consolidated awake period characterized by abrupt transitions from wakefulness to NREM sleep (i.e., dysregulation of NREM sleep onset). This phenomenon manifests clinically as excessive daytime sleepiness or sleep attacks. The other key phenomenon is the pathological intrusion of REM sleep into wakefulness or at sleep onset (i.e., dysregulation of REM sleep onset). It is during these periods that patients experience cataplexy, hypnagogic hallucinations, and sleep paralysis.

Mice with targeted deletion of the *prepro-orexin* gene (*orexin*−*/*<sup>−</sup> mice) display a phenotype strikingly similar to narcolepsy: abrupt behavioral arrests with muscle atonia (cataplexy), fragmented wakefulness (inability to maintain consolidated wakefulness episodes), and direct transitions from wakefulness to REM sleep (Chemelli et al., 1999). In addition, functionally null mutations in the *OX2R* gene were found in two independent lines of familial narcoleptic dogs (Lin et al., 1999).

Based on studies on orexin receptor-deficient mice (*OX1R*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice), it has been suggested that deletion of *OX1R* produces no measurable effect on sleep/wakefulness states (Sakurai, 2007; Hondo et al., 2010). However, *OX2R*−*/*<sup>−</sup> mice have clear characteristics of narcolepsy, although their behavioral and EEG phenotype is less severe than that found in *orexin*−*/*<sup>−</sup> mice (Willie et al., 2003). In infrared videophotographic studies in the dark phase, *OX2R*−*/*<sup>−</sup> mice showed abrupt behavioral arrests. However, its frequency was much less than in *orexin*−*/*<sup>−</sup> mice (31-fold lower frequency in *OX2R*−*/*<sup>−</sup> mice as compared to *orexin*−*/*<sup>−</sup> mice). Instead, *OX2R*−*/*<sup>−</sup> mice showed a distinct variety of behavioral arrests with onsets that were more gradual in nature (gradual arrests). Moreover, *orexin*−*/*<sup>−</sup> mice also exhibit gradual arrests with a frequency similar to *OX2R*−*/*<sup>−</sup> mice in addition to plenty of abrupt arrests.

In contrast to abrupt arrests, gradual arrests typically began during quiet wakefulness and could be easily distinguished from the normal onset of resting behavior by (i) the absence of stereotypic preparation for sleep (e.g., nesting and/or assumption of a curled or hunched posture with limbs drawn under the body) and (ii) a characteristic ratchet-like"nodding"of the head over a period of several seconds with a transition to a collapsed posture (Willie et al., 2003).

Detailed observations of behaviors during EEG/EMG recordings found that abrupt arrests in *orexin*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice occurred during direct transitions from wakefulness to REM sleep, whereas EEG/EMG correlates of gradual arrests in both *orexin*−*/*<sup>−</sup>

and *OX2R*−*/*<sup>−</sup> mice invariably revealed the onset of attenuated muscle tone, but not atonia, and an EEG transition from wakefulness to NREM sleep (Chemelli et al., 1999;Willie et al., 2003). Pharmacologically, abrupt arrests in *orexin*−*/*<sup>−</sup> mice were suppressed by systemic administration of clomipramine, an anticataplectic agent used for treatment of human narcolepsy. Whereas, administration of caffeine, a psychostimulant used to treat excessive sleepiness in human narcolepsy, tended to produce a mild exacerbation of abrupt arrest frequency. In clear contrast, systemic administration of caffeine dose-dependently suppressed gradual arrests, while administration of an anticataplectic agent clomipramine did not affect the frequency of gradual arrests in both *orexin*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice. These observations showed strong similarity with those observed in human narcolepsy patients; psychostimulant drugs are effective for the sleepiness, but exacerbate cataplexy, which is treatable with antidepressants. This detailed characterization of behavioral, pharmacological, and electrophysiological features of *orexin*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice defined abrupt and gradual arrests as the presumptive mouse correlates of cataplexy and sleep attack in human narcolepsy, respectively.

In addition to the gradual behavioral arrests, *OX2R*−*/*<sup>−</sup> mice exhibit fragmentation of wakefulness, another sign of sleepiness, to the extent similar to that in *orexin*−*/*<sup>−</sup> mice (Willie et al., 2003). *OX1R*−*/*−; *OX2R*−*/*<sup>−</sup> mice appear to have the same phenotype with *orexin*−*/*<sup>−</sup> mice, implying that these two receptors are sufficient to mediate regulation of sleep/wakefulness by orexins (Sakurai, 2007; Hondo et al., 2010). These results of mouse reverse genetic studies suggest that normal regulation of wake/NREM sleep transitions depends critically on OX2R activation, whereas the profound dysregulation of REM sleep control are unique to narcolepsy from loss of signaling through both OX1R- and OX2R-dependent pathways.

Substantially lower frequency of cataplexy in *OX2R*−*/*<sup>−</sup> mice as compared to *orexin*−*/*<sup>−</sup> mice appears to be inconsistent with the fact that mutations of *OX2R* gene are solely responsible for an inherited canine model of narcolepsy, which demonstrate frequent occurrence of cataplexy as well as excessive sleepiness (Lin et al., 1999). This may result from species difference (e.g., the precise expression patterns of two orexin receptors). However, even in dogs, the absence of orexin peptides may cause severe narcoleptic symptoms as compared to *OX2R* mutation. Early studies of narcoleptic Dobermans and Labradors found these dogs to be 30 to 80-fold less severely affected with cataplexy than poodles with sporadic narcolepsy that showed literally hundreds of attacks a day (Baker et al., 1982), an effect previously attributed solely to differences in breed and breed size.

As an experiment complementary to the behavioral studies and baseline sleep/wakefulness recordings of *OX1R*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice, we elucidated that two orexin receptors play distinct and differential roles in the regulation of sleep and wakefulness states by comparing the effects of ICV orexin A administration in wild-type, *OX1R*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice (Mieda et al., 2011). The effects of orexin A on wakefulness and NREM sleep were significantly attenuated in both knockout mice as compared to wild-type mice, with substantially larger attenuation in *OX2R*−*/*<sup>−</sup> mice than in *OX1R*−*/*<sup>−</sup> mice. These results suggest that although the OX2R-mediated pathway has a pivotal role in the promotion of wakefulness, OX1R also plays additional roles in promoting arousal.

In contrast, suppression of REM sleep by orexin A administration was slightly and similarly attenuated in both *OX1R*−*/*<sup>−</sup> and *OX2R*−*/*<sup>−</sup> mice, suggesting a comparable contribution of the two receptors to REM sleep suppression (Mieda et al., 2011). In addition, our observations further suggest that orexin A directly suppresses transitions from NREM sleep to REM sleep, and that activation of OX1R is sufficient for this effect. Supplementary role of OX1R in the suppression of NREM sleep is consistent with the fact that *OX2R*−*/*<sup>−</sup> mice on a C57BL/6J genetic background show less fragmented wakefulness when compared to *orexin*−*/*<sup>−</sup> mice and *OX1R*−*/*−; *OX2R*−*/*<sup>−</sup> mice (but show similarly fragmented wakefulness on a C57BL/6J-129/SvEv mixed background) (Sakurai, 2007; Mochizuki et al., 2011), suggesting that OX1R is indispensable for the maintenance of wakefulness in the absence of OX2R.

Although application of exogenous orexins has been shown to excite many types of neurons (Sakurai and Mieda, 2011), neurons directly downstream to orexin neurons in physiological conditions (i.e., neurons influenced by endogenous orexins that mediate their wake-promoting and REM-suppressing effects), have remained uncertain. Several reports suggested that histaminergic neurons in the TMN play an important role in the arousal-promoting effect of orexin, supported by the facts that the effect of ICV orexin A administration is markedly attenuated by the histamine H1 receptor antagonist pyrilamine (Yamanaka et al., 2002) and is absent in *H1 histamine receptor* knockout mice (Huang et al., 2001) and that the TMN expresses OX2R abundantly (Marcus et al., 2001; Mieda et al., 2011), the subtype whose absence causes narcoleptic phenotype in mice and dogs (Lin et al., 1999; Willie et al., 2003). Mochizuki et al. (2011) produced a mouse model in which a *loxP*-flanked gene cassette disrupts production of the OX2R, but normal OX2R expression can be restored by Cre recombinase. They showed that targeted Cre expression (i.e., focal restoration of OX2R expression), in the TMN and adjacent regions rescues fragmentation of wakefulness in this mouse model, suggesting that the orexin signaling mediated by OX2R in the TMN (and possibly its surrounding area in the PH) is sufficient to prevent sleepiness caused by systemic OX2R deficiency.

On the other hand, this hypothesis is still controversial. Mice lacking both OX1R and histamine H1 receptors demonstrate no abnormality in sleep/wakefulness (Hondo et al.,2010). In addition, a recent optogenetic study showed that orexin-mediated sleepto-wake transitions do not depend on the TMN histaminergic neurons (Carter et al., 2009). Rather, another optogenetic study suggested the role of LC noradrenergic neurons as a direct downstream pathway of orexin neurons by demonstrating that optogenetic inhibition of LC noradrenergic neurons blocked arousal effects of optogenetic stimulation of orexin neurons (Carter et al., 2012).

#### **REFERENCES**

Akiyama, M., Yuasa, T., Hayasaka, N., Horikawa, K., Sakurai, T., and Shibata, S. (2004). Reduced food anticipatory activity in genetically orexin (hypocretin) neuron-ablated

mice. *Eur. J. Neurosci.* 20, 3054–3062.

Baker, T. L., Foutz, A. S., McNerney, V., Mitler, M. M., and Dement, W. C. (1982). Canine model of narcolepsy: genetic and

#### **PHARMACOLOGICAL STUDIES**

Antagonists for orexin receptors are drawing people's attention as novel medications for insomnia (Scammell and Winrow, 2011; Mieda and Sakurai, 2013). Several non-selective (dual) antagonists for orexin receptors, as well as subtype-selective antagonists, have been developed. These drugs are indeed useful also for studying the roles of each subtype in the regulation of sleep/wakefulness without considering any chronic compensatory changes that might complicate the phenotypes of genetically engineered models.

A study reported that an OX2R-selective antagonist JNJ10397049 has a better ability to promote NREM sleep than the dual antagonist almorexant in rats (Dugovic et al., 2009). They suggested that simultaneous inhibition of OX1R attenuates the sleep-promoting effects mediated by selective OX2R blockade, possibly correlated with dopaminergic neurotransmission. In contrast, another recent study led to a different conclusion (Morairty et al., 2012). It reported that an OX1R-selective antagonist SB-334867 produces small increases in REM and NREM sleep, and an OX2R-selective antagonist EMPA produces a significant increase in NREM sleep. But administration of almorexant increases NREM sleep more than these subtype-selective antagonists, leading to the conclusion that dual orexin receptor antagonism is more effective for sleep promotion than subtype-selective antagonism. The difference in these drugs' capabilities of crossing the blood-brain barrier, as well as their pharmacokinetic/pharmacodynamic characters, might influence the effectiveness. The latter view is also consistent with the conclusion derived from studies using subtypespecific knockout mice that OX2R plays a pivotal role, but OX1R has additional effects on promotion of wakefulness (Sakurai, 2007; Mieda et al., 2011).

#### **CONCLUSION**

The orexin system is a necessary component for the normal regulation of sleep/wakefulness. Nevertheless, broad expression of orexin receptors throughout the brain makes it difficult to identify neurons and orexin receptor subtypes that are directly regulated by endogenous orexins and mediate their effects on the physiology of interest in a natural context. Future studies using molecular genetic strategies such as brain region/cell type-specific deletions and brain region/cell type-specific rescues of orexin receptors, as well as pharmacological studies of focal administration of subtype-specific orexin agonists/antagonists, would further dissect the differential roles of orexin receptors in the regulation of sleep/wakefulness. Furthermore, the orexin system is also important for the regulation of a variety of physiological functions, such as feeding, reward, and emotions. Thus, precise information of the different roles of the two orexin receptors is beneficial not only for understanding the mechanisms underlying orexinergic regulations of physiology, but also for application of orexin agonists/antagonists as medications for various diseases.

developmental determinants. *Exp. Neurol.* 75, 729–742.

Bayer, L., Eggermann, E., Serafin, M., Saint-Mleux, B., Machard, D., Jones, B., et al. (2001). Orexins (hypocretins) directly excite

tuberomammillary neurons. *Eur. J. Neurosci.* 14, 1571–1575.

Bourgin, P., Huitron-Resendiz, S., Spier, A. D., Fabre, V., Morte, B., Criado, J. R., et al. (2000). Hypocretin-1 modulates rapid eye movement

sleep through activation of locus coeruleus neurons. *J. Neurosci.* 20, 7760–7765.


Orexins/hypocretins excite basal forebrain cholinergic neurones. *Neuroscience* 108, 177–181.


innervation of the locus coeruleus noradrenergic system. *J. Comp. Neurol.* 415, 145–159.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 10 January 2013; accepted: 25 April 2013; published online: 16 May 2013.*

*Citation: Mieda M, Tsujino N and Sakurai T (2013) Differential roles of orexin receptors in the regulation of sleep/wakefulness. Front. Endocrinol. 4:57. doi: 10.3389/fendo.2013.00057*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Mieda, Tsujino and Sakurai. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, providedthe original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Control of ventricular ciliary beating by the melanin concentrating hormone-expressing neurons of the lateral hypothalamus: a functional imaging survey

#### **Grégory Conductier 1,2† , Agnès O. Martin3,4,5† , Pierre-Yves Risold<sup>6</sup>† , Sonia Jego<sup>7</sup> , Raphaël Lavoie<sup>7</sup> , Chrystel Lafont 3,4,5, Patrice Mollard3,4,5‡ , Antoine Adamantidis <sup>7</sup>‡ and Jean-Louis Nahon1,2,8\* ‡**

<sup>1</sup> UMR7275, Institut de Pharmacologie Moléculaire et Cellulaire, Centre National de la Recherche Scientifique, Valbonne, France

<sup>3</sup> UMR5203, Institut de Génomique Fonctionnelle, Centre National de la Recherche Scientifique, Montpellier, France

<sup>4</sup> U661, INSERM, Montpellier, France


#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Gert Jansen, Erasmus Medical Centre, Netherlands Serge H. Luquet, University Paris Diderot, France

#### **\*Correspondence:**

Jean-Louis Nahon, UMR7275, Institut de Pharmacologie Moléculaire et Cellulaire, Centre National de la Recherche Scientifique, 660 Route des Lucioles, Sophia Antipolis, Valbonne, France e-mail: nahonjl@ipmc.cnrs.fr

†Co-Authors ‡Co-Directors

#### **INTRODUCTION**

First identified in the early 80s from chum salmon pituitaries, the melanin concentrating hormone (MCH) draw its name from its capability to induced the concentration of melanin in the skin melanophores (2). However, this function seems to be restricted to teleosts [reviewed in Ref. (3)]. In contrast with high MCH structural conservation, the neuronal distribution appears quite different, reflecting evolutionary changes in the prosencephalon across vertebrate species (4). In mammals, this cyclic peptide is mainly expressed in neurons of the lateral hypothalamic area (LHA), projecting widely throughout the brain (5); reviewed in Ref. (6). Accordingly, MCH is involved in a broad spectrum of cerebral functions [for recent reviews, see Ref. (7, 8)]. Nevertheless, all of these seem to converge to the adaptation of global physiologic state to metabolic needs by promoting memory processes and reward pathways activation on one hand and by decreasing arousal and thermogenesis on the other hand. Activation of these cognitive and neuroendocrine networks leads to an increase in food intake and energy storage, respectively [reviewed in Ref. (9, 10)].

The structure of the *Pmch* gene locus appears to be complex and sense/antisense transcripts could generate different proteinderivatives. Indeed, the precursor ppMCH may be processed mainly, but not exclusively, in two different peptides (MCH and

The cyclic peptide Melanin Concentrating Hormone (MCH) is known to control a large number of brain functions in mammals such as food intake and metabolism, stress response, anxiety, sleep/wake cycle, memory, and reward. Based on neuro-anatomical and electrophysiological studies these functions were attributed to neuronal circuits expressing MCHR1, the single MCH receptor in rodents. In complement to our recently published work (1) we provided here new data regarding the action of MCH on ependymocytes in the mouse brain. First, we establish that MCHR1 mRNA is expressed in the ependymal cells of the third ventricle epithelium. Second, we demonstrated a tonic control of MCH-expressing neurons on ependymal cilia beat frequency using in vitro optogenics. Finally, we performed in vivo measurements of CSF flow using fluorescent micro-beads in wild-type and MCHR1-knockout mice. Collectively, our results demonstrated that MCHexpressing neurons modulate ciliary beating of ependymal cells at the third ventricle and could contribute to maintain cerebro-spinal fluid homeostasis.

**Keywords: MCH, MCHR1, non-neuronal function, cilia, CSF flow**

NEI) in the brain and in several intermediates, including the dipeptide MCH-NEI, in peripheral organs (11–14). An additional protein, named MGOP, may be produced by an alternative splicing of the *Pmch* gene primary transcript in all cells producing MCH (15, 16). Finally a set of proteins, involved in DNA repair, may be synthesized by expression of the *AROM/PARI* gene located on the complementary strand overlapping the *Pmch* gene (8, 17). Based on this disparity in gene-products expression, it is quite difficult to associate a single molecular substrate responsible to the wide phenotypic changes observed in *Pmch* gene KO mice in which the full exon-intron sequences of the *Pmch* gene as well as the 30UTR region of spliced AROM/PARI gene transcripts were deleted. Meanwhile, the issue of developmental compensation (or adaptation) in these genetic models of *Pmch* gene inactivation should also be considered [see Ref. (9) for discussion of this point].

Efforts to identify the MCH receptor initially led to the discovery of a spliced variant of the seven-transmembrane G-coupled protein named SLC-1 (18) as a cognate MCH receptor and thereafter referred to as MCHR1 (19–23). MCHR1 is widely localized in brain regions involved in the control of neuroendocrine, reward, motivational, and cognitive aspects of feeding behavior (9, 10, 24– 26). Interestingly, MCHR1-deficient mice are lean due to hyperactivity and increased metabolism (27). A second MCH receptor,

<sup>2</sup> University of Nice Sophia Antipolis, Nice, France

named here MCHR2, was identified and characterized in human tissues and cell lines (27–33). This MCH receptor displayed a brain distribution that overlapped partially with that of MCHR1 in the primate and fish brain (32, 34). However, MCHR2 is lacking in rat and mouse genomes (35). Furthermore, in contrast to MCHR1 that signals to either Gai or Gaq, depending on the transfected or native cell systems, MCHR2 signaling operates apparently exclusively through Gaq protein [our unpublished data; reviewed in Ref. (35–37)].

Based on neuro-anatomical mapping and electrophysiological data, it was assumed that synaptic transmission represents the main mode of action of MCH in the brain. However, non-neuronal intercellular communication or "volume" transmission may also be involved but evidence were lacking. In a recently published study (1), we mapped numerous MCH fibers in close vicinity to MCHR1 expressed into ependymocytes of the ventral part of the third ventricle (3V). Developing new techniques to measure and analyze the ependymal cilia beat frequency (CBF) in acute mouse brain slice preparations, we also showed that the CBF is increased by MCH application or LHA stimulation, an effect blocked by a selective MCHR1 antagonist and absent in MCHR1-knockout (MCHR1-KO) mice. In addition, using *in vivo* brain MRI, we demonstrated that the volume of both the lateral and third ventricles is increased in MCHR1-KO mice compared to their wild-type (WT) littermates. Thus, our study revealed a previously unknown function of the MCH/MCHR1 signaling system in non-neuronal cells. Here, we first demonstrated MCH mRNA expression in the ventral 3V ependymal cells isolated by laser-capture and *in situ* hybridization. We then extended our previous work, by using *in vitro* optogenetic activation or inhibition of MCH neurons. Finally, we investigated *in vivo* tracking of fluorescent micro-beads through the 3V in WT and MCHR1-KO mice. Collectively, we demonstrate a dynamic control of MCH neurons on spontaneous CBF of MCHR1 mRNA-expressing ependymal cells and discuss the current strategies for measuring CSF flows in small animal models.

# **MATERIALS AND METHODS**

#### **ANIMALS**

The experiments were conducted with male C57BL/6J mice (for laser-captured cell mapping, *in situ* hybridization and cellular optogenetic measurements) and female KO MCHR1 mice (*in vivo* CSF flow experiments) of 10–12 weeks of age. The animals were obtained from heterozygous breeding in the local animal facilities and maintained on a 12-h dark/light cycle (7 a.m./7 p.m.) with food and water *ad libitum*. The MCHR1-KO mice were established as previously described (38).

All the protocols were carried out in accordance with French ethical guidelines for laboratory animals (Agreement N°75–178, 05/16/2000) and were approved by the IPMC care committee. Attention was paid to use only the number of animals requested and necessary to generate reproducible results.

#### **LASER MICRO-DISSECTION OF THIRD VENTRICLE EPITHELIUM**

After decapitation, each brain (*n* = 2) was dissected out in <2 min and immediately frozen at −80°C using a Snapfrost (Alphelys, France). Sections (10µm thick) were cut on a cryostat (Microm

HM 560; object holder and chamber were kept at −21°C). Eight sections passing through the posterior hypothalamus were collected on pen membrane slides. Slides, continuously maintained on dry ice, were dehydrated in three baths of increasing ethanol baths (70, 95, and 100%) and two baths of fresh xylene (Roth, France) for 5 min each. Sections were air dried and kept in the vacuum of a dessicator until dissection.

Dissections were performed using a PixCell® (Arcturus Engineering) with CapSure® HS LCM caps. The dissection time never exceeded 20 min/slide, starting from when the slide was removed from the dessicator. Laser parameters were calibrated for each dissection by measuring the impact of shots on the membrane of the slide adjacent to the tissue. The area of interest was then dissected and laser-captured using UV laser to cut the tissue and IR laser to capture the sample. Four samples were collected per cap (micro-dissection of two slides in <40 min in total) and only one cap per brain. As soon as the fourth sample was obtained, the cap was examined under the microscope to ensure the absence of unwanted debris. The sample lysis and the RNA extraction were performed using the RNAqueous®-Micro Kit (Ambion, France) following the manufacturer's instructions. The quality of the samples was finally evaluated with the Agilent 2100 Bioanalyzer (Agilent Technology). A rin of 6.1 was twice obtained.

After reverse transcription (Superscript III, Invitrogen), cDNA corresponding to 1 ng of RNA was used as input in a PCR reaction (GoTaq Green MasterMix, Promega, Charbonnières, France) for MCHR1 and HPRT as positive control (MCHR1 F: 5<sup>0</sup> -GCTCTATGCCAGGCTTATCC-3<sup>0</sup> , MCHR1 R: CAGCTGTCTGAGCATTGCTG-3<sup>0</sup> , amplicon size: 494 bp; HPRT F: CTCCGGAAAGCAGTGAGGTAAG, HPRT R: GGAGGGA-GAAAAATGCGGAGTG, amplicon size: 306 bp). Sample for which reverse transcriptase was omitted was used as negative control. PCR protocol used was designed as follow: initial denaturation: 95°C, 5 min follow by 40 cycles composed of 95°C, 30 s, 58°C, 30 s, 72°C, 1 min, and a final elongation for 7 min at 72°C.

#### **IN SITU HYBRIDIZATION**

Frozen sections were post-fixed in 4% paraformaldehyde in 0.1 M phosphate buffer and digested with proteinase K (1µg/mL,Roche) for 30 min at 37°C. Slides were incubated for 8 min in 0.1 M triethanolamine (TEA), pH 8.0, and then for 5 min at room temperature in 100 mL 0.1 M TEA + 500µL acetic anhydride followed by a decarboxylation in active diethyl pyrocarbonate (DEPC).

Sections were then rinsed briefly with 5× standard citrate sodium (SSC) buffer then incubated for 2 h in prehybridization buffer at 56°C. After rinsing in 0.2× SSC, the sections were incubated overnight at 56°C, in humid chambers, with 50µL hybridization buffer containing 5% Denhardt's and 50 ng labeled RNA probes. After rinsing with 5× SSC, sections were incubated successively in 2× SSC at 56°C (1 h 30 min) and 0.2× at room temperature (5 min). They were incubated in anti-DIG Fab fragments conjugated to alkaline phosphatase (1/1300, overnight) and revealed with enzyme substrate NBT-BCIP (overnight, at room temperature).

Two MCHR1 RNA probes were used; one probe was kindly provided by Drs Civelli and Chung (University of California, Irvine, CA, USA) and one made by reverse transcription/polymerase chain reaction from mouse genomic DNA. Control hybridization, including hybridization with sense DIG-labeled riboprobes was realized and did not reveal any signal.

#### **MEASURES OF CILIA BEAT FREQUENCY USING MCH NEURON-SPECIFIC OPTOGENETIC EXCITATION OR INHIBITION**

We have previously shown that electrical stimulation of the LHA induced an increase in the CBF in the 3V (1). To further extend and improve the specificity of the response, we used new models allowing the optogenetic control of MCH neurons activity. All procedures and controls were previously published in (39). Briefly, using cre-dependent Ef1a-DIO-ChETA-EYFP AAV mediated transduction, the fast mutant of the light-activated Channel rhodopsin-2 (ChETA) or the chloride pump halorhodopsin (NpHR) was specifically expressed in MCH neurons of two different groups of 3 week-old mice. Four weeks after stereotactic injection of the AAV vectors in the LHA (AP: −1.45 mm, ML: ±1 mm, DV: −5.5 mm), brain slices were made and recordings were made in CBF as described elsewhere (1) before, during, and after stimulation of ChETA-expressing MCH neurons (473 nm, stimulation frequency: 1, 5, 10, 20, and 40 Hz, pulse length: 10 ms, total stimulation duration: 3 min) or NpHR (590 nm, continuous stimulation during 8.5 s each 10 s, 5 min). For each slice and area of recording, the instantaneous CBF was calculated. All samples in which basal CBF was out of a range comprised between 5 and 20 Hz (considered as the natural basal frequency in healthy slices) were not included. The mean of the basal frequency during the first 5 min of recordings were used as a baseline for normalization of the experimental values. Results are expressed as the percentile variation of this baseline.

#### **MEASURE OF CSF FLOW INDEX USING FLUORESCENT MICRO-BEADS**

Littermate controls and female MCHR1-KO mice (aged 12 weeks old) were anesthetized by intraperitoneal injection of ketamine hydrochloride 50 mg/kg and xylazine 10 mg/kg and placed in a stereotaxic frame tip of a 26 gage needle was brought to the following coordinates relative to the bregma: 1.75 mm posterior, 2.5 mm ventral, and 0 mm right and left. About 10µL of polystyrene beads (diameter 3µm (sigma L4530) dilution 1:4 in 0.09% NaCl) was injected in the third ventricle.

Fibered confocal fluorescence microscopy (FCFM) (CellviZio; Mauna Kea Technologies, Paris, France) imaging was used to visualize the *in vivo* movement of the polystyrene beads in the CSF flux. FCFM provides an easy access to these regions of interest with low disturbance of brain structure (40, 41). Small-diameter fiber-optic probe consisting of tens of thousands of fibers was implanted in the brain of the mice and connected to a Laser scanning unit LSU-488 (FibroScan) that uses a laser source with a wavelength of 488 nm. We used a MiniZ probe of 300µm diameter with a working distance of 70µm. The probe was stereotaxically lowered in the third ventricle at 2.5 mm ventral. Sensitive, single-pixel detection of fluorescence stimulated by the photodiode laser pulse through each fiber element, combined with the high scan rate allows the visualization of beads movements. Four acquisition sessions of at least 10 min was recorded for each animal at a frame rate of 11 frames/s.

# **STATISTICS**

#### **Variation in the CBF**

Statistics were performed using Prism software (Graphpad Inc., La Jolla, CA, USA). The global mean for grouped time points (baseline, stimulation, and recovery) were compared Using One way ANOVA followed by Bonferroni's multiple comparison test (BMCT). N = number of mice, n = number of slices, *n* = number of cells considered. *p* Value <0.05 were considered significant.

#### **Measure of CSF flow index using fluorescent micro-beads**

Movies were visualized on ImageCell™ viewer. The speed of the beads was analyzed by tracking 10 beads/10 min films. The mean speed for an animal was the mean of the four films speeds. The movies displaying significant modifications of the speed over time were excluded as probably corresponding to pressure due to probe positioning or blood clot.

#### **RESULTS**

#### **LASER MICRO-DISSECTION OF THIRD VENTRICLE EPITHELIUM AND IN SITU HYBRIDIZATION**

As illustrated in **Figures 1A,B**; ependymal cell layer was carefully dissected and used for RNA extraction. RT-PCR results indicate that mRNA coding for MCHR1 were present in the 3V epithelium (**Figure 1C**). This was further confirmed by *in situ* hybridization with two specific probes recognizing MCHR1 mRNA. Indeed, numerous (but not all; see open arrowhead) ependymocytes were labeled with antisense probe (**Figure 1D**), within the cytoplasm (**Figure 1E**), while sense probe did not stain any cell types (**Figure 1F**). These results are in agreement with our immunohistochemical study (1) and recent data from Maratos-Flier's lab using a *MCHR1-cre/tdTomato* mouse strain (26).

#### **MCH NEURON-SPECIFIC OPTOGENETIC TOOLS**

The ChETA-NpHR system was used to dissect MCH neuronal circuitry reaching 3V ependymal cells and controlling CBF.

The stimulation of ChETA-expressing MCH neurons in the LHA induces an increase in the CBF reaching 134% of the basal value (**Figures 2A,B**; 1 Hz, N = 5, n = 7, *n* = 10, *F*5.069, 29 = 0.0131, BMCT: baseline vs. ChETA *t* = 2.896, *p* < 0.05). After 10 min recovery, the subsequent stimulation at 5 Hz tended to increase the CBF but did not reach the significance level (**Figure 2B**; 5 Hz, N = 5, n = 7, *n* = 10). For higher frequencies, no effect was observed (not shown, see Discussion). On the other hand, the stimulation of NpHR induced a marked decrease in the CBF reaching 76% of the basal (**Figures 2C,D**; N = 4, n = 8, *n* = 15, *F*4.616, 44 = 0.0154, BMCT: baseline vs. NpHR *t* = 3.027, *p* < 0.05).

#### **MEASURE OF CSF FLOW INDEX USING FLUORESCENT MICRO-BEADS**

In order to address the physiological relevance of MCH-driven ciliary beating regulation, we conducted *in vivo* flow measure. FCFM imaging was used to visualize the movement of the polystyrene beads in the CSF flux *in vivo* in groups of WT and KO MCHR1 mice (*n* = 5 each). The visualization of fluorescent beads movements in the third ventricle allow to approximate the speed of the CSF flux using *in vivo* brain imagery (**Figure 3**).

Overall, no significant statistical difference in the mean speed between the two groups was found. However, 3/5 animals in the

: after reverse transcription,

KO group displayed a twofold increase in the mean speed by comparison with the WT group (that shows consistently low speed). However, it should be stressed that we are measuring the bulk flow driven by the arterio-venous pressure gradients and arterial pulsations and that the laminar flow (close to the ventricle wall and therefore dependent upon cilia beating) remains too small to be measured. Furthermore, we believe that the technical caveat of the implantation of the probe and beads injection *in vivo* (blood clot, exact position, size of the probe . . .) may induce a methodological bias (see Discussion).

: negative control of the reverse transcription. HPRT: positive control. **(D)** Photomicrograph to illustrate the distribution of the MCHR1 in situ

MCHR1 mRNA in the ventricular epithelium. RT<sup>+</sup>

#### **DISCUSSION**

RT<sup>−</sup>

The MCH system is involved in a broad spectrum of function through mainly the synaptic release of the peptide(s) and neuronal activity modulation in mammalian brain. However, a growing body of data indicates that MCH may also have non-neuronal function, especially by regulating the activity of more or less specialized peripheral blood mononuclear cells [PBMCs (42)]. The expression of ppMCH and/or MCHR1 genes in pancreatic islets or in adipocytes (43–45) may highlight a metabolism-related function of MCH at the periphery. Nevertheless, such action may not be totally independent of the intracerebral and/or spinal MCH pathway (46). Based on our previous study (1) and the present set of data, we demonstrate a new role for the hypothalamic peptide MCH in modulation of CBF in the ventral part of the third ventricle through activation of ependymal cells.

In combining the laser micro-dissection of ependymal epithelium of the 3V followed by RNA extraction and RT-PCR experiments and *in situ* hybridization data, we demonstrate the presence of MCHR1 transcripts within the epithelium. This is consistent with the data obtained by immunohistochemistry (1) or in a

expression of MCHR1 mRNA in discrete ependymal cells. **(F)** A negative control using sense MCHR1 gene probe. Scale bar = 25µm in **(A)**; Scale

bar = 20µm in (D); Scale bar = 10µm in **(E)** and **(F)**.

*MCHR1-Cre/tdTomato* mouse model (26). The optogenetic stimulation of the MCH neurons (through ChETA activation) increased the CBF in the same extend than electrical stimulation did, compared to basal conditions. This confirms also with high temporal precision, the specificity of the response observed in the 3V. Indeed, only ciliated cells lining the ventral 3V that expressed MCHR1 were MCH-sensitive, while those from the dorsal third ventricle or the lateral ventricles were not (1). The spatial specificity of the MCH response adds a new level of complexity to the previously described characteristics of ciliated ependymal cells [orientation, size, beating mode of cilia along the ventricles (47, 48)], and suggests that ciliary beating in cerebral ventricles is fine-tuned to modulate CSF flow in response to metabolic, neurohormonal, and neuroimmune changes. Moreover, our present *in situ* hybridization experiments, previous immunohistochemical data (1) and mapping using *MCHR1-Cre/tdTomato* mouse (26) highly suggested a "cluster-like" distribution of the MCHR1 mRNA and proteins along the 3V. Since adjoining ependymal cells are known to be coupled through GAP junctions (49), it is tempting to speculate that the only few ependymal cells expressing MCHR1 may act as hubs responsible for the effect of MCH neuronal stimulation on the whole epithelium in the 3V.

The results obtained following genetic invalidation and pharmacological inactivation of MCHR1 suggested that the MCH system could exert a tonic positive control on CBF (1). Here, we demonstrate the validity of this hypothesis since the inhibition of MCH neurons activity (through NpHR stimulation) directly

expressing ChETA **(A)** or NpHR **(C)** from the LHA to the ventricular epithelium and their optogenetic stimulation paradigm. **(B)** Consequences of the stimulation of ChETA in MCH neurons, by light pulses of 10 ms at a rate of 1 and 5 Hz as shown by the bars, on the CBF recorded in the ventral 3V

numbers of slices, n: number of recording area. \*p < 0.05 **(D)** Consequences of the stimulation of NpHR in MCH neurons as shown by the bars, in MCH neurons on the CBF recorded in the ventral 3V. N: number of animals, n: numbers of slices, n: number of recording area.

affects the CBF. To our knowledge, MCH is the only known molecule exerting a tonic positive effect on cilia beating in the brain. Even more interestingly, this tonic control of MCH on the CBF in the 3V does not seem to be compensated through adaptive mechanisms during development, since the basal CBF in MCHR1-KO mice is also reduced (1).

Even if MCH neurons activity seems to be important in the regulation of cilia beating, we have not yet firmly established whether the communication between MCH neurons and ependymal cells involves a true asymmetric synapse or not. With respect to the anatomical distribution of MCH fibers around the v3V, the communication should more likely involve passing fibers "leaking" MCH close to the epithelium basal pole and/or release of the peptide directly into the CSF. Indeed, MCHR1 immunolabeling was observed at both the apex and the basal poles of ciliated ependymal cells as well as MCH fibers crossing the epithelium. The nature of the contacts between ciliated cells and fibers remains to be addressed using a detailed electronic microscopic analysis.

Genetic ablation of MCHR1 results in an increase in the volume of LVs and 3V (both ventral and dorsal) but no change in 4V as reported previously (1). In order to determine the flow of CSF using a non-invasive method, we have tried to transpose clinical tools (CINE-MRI) to mice. Unfortunately, the main limit of such technique is the speed of the flow. Preliminary experiments indicate that this speed is <10µm s−<sup>1</sup> , preventing the use of CINE-MRI in the mouse brain (Kober F., Troalen T., and Viola A. CRMBM. Marseille; personal communication).

The most used methods to study the flow of CSF in rodents consist in the injection of tracers (X-rays orMRI contrast agent)*in vivo* (50), or the use of fluorescent micro-beads or china ink *ex vivo* on dissected epithelia (51). Here, we show that it is possible to follow the migration of fluorescent micro-beads through the v3V. Unexpectedly, our data indicate that, in MCHR1-KO mice, the speed of the CSF flow tends to increase as compared to WT littermates, without reaching the statistical significance level. This paradoxical effect could be explained by the Poiseuille's law which postulates a direct link between the mean speed (*V*) of a viscous liquid (such as CSF) and the radius (*r*) of a small cylinder (such as a ventricle) (*V* = DP\*Π\**r* 4 /8*h* × *l* with *h* = viscosity, DP = difference of pressure between the extremities of the cylinder, *l* = length, and *r* = radius of the cylinder). Indeed, the flow measured at the center of the ventricle would increase when the radius expands. This fits quite well with an enlargement of the ventricle in the KO MCHR1 mice as observed using MRI (1). Another explanation for this discrepancy would be that the optic fiber used for the recording may block the CSF flow in the 3V ofWT animals but not inMCHR1-KO

(since the volume of this ventricle is enlarged in these animals). Moreover, an increase in the CSF pressure into the ventricles could not be excluded.

At this point, it seems important to dissociate the global flow of CSF, mainly resulting from cardio-respiratory activity, to the laminar flow imputable to cilia beating. This point is of prime importance since it has been shown that MCHR1-KO mice display an increase in heart and respiratory rate (52). As a consequence, this may be responsible for the increase in global CSF flow. Moreover, because the volume of the ventricle is increased in MCHR1-KO mice, according to the Poiseuille's equation (see above), the speed of the CSF close to the epithelium should be reduced as compared to what is observed in the center of the ventricle. Because of the cilia action, such an effect is reduced and the speed close to epithelial cells is increased. Thus, in MCHR1-KO mice in which the CBF is altered, an increase in the total CSF flow may compensate the local decrease of the flow at the level of cilia. Taking into account all of these parameters, it is not surprising to observe an increase in CSF speed in the 3V.

Our data further suggest that motile ciliated cells of the cerebral ventricles are chemosensory as primary cilia, similar to motile ciliate cells from the airway epithelia (53). Hydrocephalus is one of the features of Bardet–Biedl syndrome (BBS), a genetic disease caused by a mutation in one of several proteins involved in the development of primary cilia, BBS1 being the most frequently affected in humans (54). The characteristics of BBS include ventriculomegaly of the lateral and third ventricles, particularly marked in knockin mice expressing the mutated human BBS1 protein (55). As BBS1 is involved in the trafficking of MCHR1 (54), this ventriculomegaly may be partly due to a defect in MCHR1 expression by ciliated ependymal cells, However, it is worth mentioning that the whole distribution of MCHR1 throughout the brain of BBS mouse models is still lacking. The absence of MCHR1 targeting to primary cilium in BBS models does not seems to affect the ciliogenesis (54). In this context, we found no difference in the morphology of the cilia between WT and MCHR1-KO mice, suggesting that MCHR1 does not play an important role in the development of cilia, but only in the modulation of the CBF, once the cilia are in place (although some compensatory changes could occur during development). This fits quite well with the characterization of MCHR1 mutants in ciliated hREP1 cells and the discovery of a motif in the third intracellular loop that is mandatory for MCHR1 trafficking to the primary cilia but not ciliogenesis. Other GPCR such as, somatostatin 3 receptor (SST3) and serotonin receptor 6 (5-HTR6) are specifically targeted to the primary cilia. Based on the physiological roles of somatostatin, and since new genetically engineered models such as SST3:Cre – cilia GFP mice (56) have been generated, a complete study about SST involvement in CSF and/or CBF regulation should be considered.

In conclusion, this paper and our previous study (1) point to a new role for MCH in maintaining CSF flow and homeostasis in the mouse brain. It is worth noting that the main group of MCH neurons in primitive vertebrates (lampreys) and most fish species (but teleosteans) are located very close (and also projected) to the ventricular surface and could regulate general volume transmission, like in rodents (4, 57). This convergent anatomy could be associated with an ancestral function maintained during evolution. Indeed, MCH neurons could anticipate and initiate the acceleration of CSF circulation, for instance, under conditions of metabolic necessity (glucose withdrawal, fasting, . . .). The strategic location of the ciliated cells innervated by MCH fibers, at the base of the third cerebral ventricle, could allow them to act as a pump to initiate an increase in CSF flow, providing peptides and other messengers to several brain areas and prolonging the effects of these factors in conjunction with neuronal transmission. Moreover, if the same type of CSF flow regulation exists in humans, this work suggests that the chronic administration of brain penetrating MCHR1 antagonists may have long term side effects due to alterations of CSF flows, limiting the probability of their use as therapeutic agents.

#### **AUTHORS CONTRIBUTION**

Grégory Conductier conducted and analyzed brain slice imaging experiments (CBF measurements and optogenetics) and participated to writing. Agnès O. Martin, Chrystel Lafont, and Patrice Mollard performed and analyzed the FCFM experiments, and participated in manuscript writing. Pierre-Yves Risold performed laser-captured, RT-PCR and *in situ* hybridization experiments. He participates also to writing. Sonia Jego, Raphaël Lavoie, and Antoine Adamantidis performed vector designing and set up the optogenetics experiments, and participated in manuscript writing. Jean-Louis Nahon acquired funding, designed and analyzed experiments, coordinated collaborations, and participated in manuscript writing and editing.

## **ACKNOWLEDGMENTS**

We warmly thank Dr. Bernard Lakaye of University of Liège (Belgium), for the gift of MCHR1-KO mice and Dr Mounia Annour-Louet for careful reading of the manuscript. We wish to thank Drs Alice Guyon, Frédéric Brau, and Olivier Meste for development and validation of bioinformatics tools, Christophe Houdayer for collecting samples used in lasercaptured experiments, Gabrielle Franchi for help in *in situ* hybridization, Veronique Thieffin for animal care, Isabelle Larre for help with immunoassays, and Franck Aguila for excellent artwork. We also thank Mauna Kea Technologies (Paris, France) for providing access to the CellviZio system during the duration of the experiments. Funding sources: this work was funded by the French Government (National Research Agency, ANR) through the "Investments for the Future" LABEX SIGNALIFE:#ANR-11-LABX-0028-01 (Jean-Louis Nahon), ANR-08-MNPS-018-01 (Jean-Louis Nahon), and ANR-2010-BLAN-1415-01 (Patrice Mollard), the Centre National de la Recherche Scientifique (PEPS INSB; Jean-Louis Nahon); the sixth FP EU STREPS/NEST-APES Project no. 28594 (Jean-Louis Nahon), the Fondation de la Recherche Médicale (FRM) (Jean-Louis Nahon), the SFR Biocampus de Montpellier (IPAM platform). Financial support: Gregory Conductier was supported by postdoctoral fellowships from the ANR-08-MNPS-018-01 and the Centre National de la Recherche Scientifique (INSB) and by awards from the "Société Française de Nutrition 2011" and the "Institut Danone 2012."

# **REFERENCES**


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 13 August 2013; accepted: 07 November 2013; published online: 25 November 2013.*

*Citation: Conductier G, Martin AO, Risold P-Y, Jego S, Lavoie R, Lafont C, Mollard P, Adamantidis A and Nahon J-L (2013) Control of ventricular ciliary beating by the melanin concentrating hormone-expressing neurons of the lateral hypothalamus: a functional imaging survey. Front. Endocrinol. 4:182. doi: 10.3389/fendo.2013.00182 This article was submitted to Neuroendocrine Science, a section of the journal Frontiers in Endocrinology.*

*Copyright © 2013 Conductier,Martin, Risold, Jego, Lavoie, Lafont,Mollard, Adamantidis and Nahon. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Receptor oligomerization: from early evidence to current understanding in class B GPCRs

# *StephanieY. L. Ng, Leo T. O. Lee and Billy K. C. Chow\**

Endocrinology, School of Biological Sciences, The University of Hong Kong, Hong Kong, China

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Ralf Jockers, University of Paris, France Pedro A. Jose, Georgetown University, USA

#### *\*Correspondence:*

Billy K. C. Chow, Endocrinology, School of Biological Sciences, The University of Hong Kong, Pokfulam Road, Hong Kong, China. e-mail: bkcc@hku.hk

# **INTRODUCTION**

G protein-coupled receptors (GPCRs) comprise the largest subset of cell surface receptors which transduce signals via coupling to heterotrimeric G proteins and are extensively involved in the finetuning of physiological processes. They constitute at least 3% of the human genome and have great pathophysiological importance with an abundance of information indicating their dysfunction to be associated with diseases such as diabetes, visual disability, and chronic inflammation. As a result, at least 30–40% of pharmaceutical drugs developed are targeted at GPCRs. Traditionally, GPCRs were recognized to elicit physiological responses via coupling as monomeric units to G proteins in a 1:1 stoichiometric ratio. However there is now a growing body of evidence confirming that GPCRs can self-associate or associate with different receptors to form homo-and/or hetero-oligomers. With the wealth of information recently discovered, oligomerization has been implicated to play important roles in maturation, cell surface delivery, signaling, and internalization of GPCRs. The concept of oligomerization is groundbreaking as it not only widens our perspective in understanding the molecular determinants to receptor regulation and function, but also provides new opportunities in the development of personalized drug treatments. In recent years, the small 15 member class B [secretin (SCT) or class II] GPCRs are emerging as oligomerization candidates with efforts contributed predominantly by Laurence Miller's and Dominik Schelshorn's groups. In this article, we will provide an overview of the history of oligomerization and review the available information on class B GPCRs and its functional implications.

# **EARLY EVIDENCE OF GPCR DIMERIZATION BY PHARMACOLOGICAL AND BIOCHEMICAL APPROACHES**

Although GPCR oligomerization is now a widely accepted phenomenon with modern non-radiative techniques such as resonance energy transfer (RET) strategies that are effective in demonstrating homo- and/or hetero-oligomer interactions, early

Dimerization or oligomerization of G protein-coupled receptors (GPCRs) are known to modulate receptor functions in terms of ontogeny, ligand-oriented regulation, pharmacological diversity, signal transduction, and internalization. Class B GPCRs are receptors to a family of hormones including secretin, growth hormone-releasing hormone, vasoactive intestinal polypeptide and parathyroid hormone, among others. The functional implications of receptor dimerization have extensively been studied in class A GPCRs, while less is known regarding its function in class B GPCRs. This article reviews receptor oligomerization in terms of the early evidence and current understanding particularly of class B GPCRs.

**Keywords: GPCR, class B, secretin receptor, oligomerization, BRET**

"fendo-03-00175" — 2013/1/2 — 20:01 — page 1 — #1

clues to their existence were often indirect and overlooked. In 1975, a binding experiment of a potent antagonist to β2-adrenergic receptors (ADRB2) in frog erythrocyte membranes led to findings of negative cooperativity among binding sites (Limbird et al., 1975; Limbird and Lefkowitz, 1976). This was pioneering evidence for oligomerization and can be better explained as site–site interactions amongst ADRB2 oligomers based on today's knowledge. Soon after, radiation inactivation, cross-linking and co-immunoprecipitation studies have also provided biochemical evidence complementing earlier observations of oligomerization. For example, immunoaffinity chromatography and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) showed the subunit molecular weight of mammalian lung ADRB2 to be 59 kDa, while radiation inactivation of ADRB2 resulted in a functional subunit molecular weight of 109 kDa (Fraser and Venter, 1982). Taken together, the contrasting molecular sizes indicate that ADRB2s exist in the membrane as dimers of two subunits (Fraser and Venter, 1982). Similar suggestions have also been made for the rat liver membrane α1-adrenergic receptor (ADRA1; Venter et al., 1984) and human platelet α2-adrenergic receptor (ADRA2; Venter et al., 1983) which each have molecular masses of 160 kDa. Alternatively, cross-linking experiments, using photoaffinity labeling reagents, cell permeable cross-linkers, agonists, and antagonists (Capponi and Catt, 1980; Paglin and Jamieson, 1982; Guillemette and Escher, 1983; Rogers, 1984; Carson et al., 1987; Rondeau et al., 1990; Siemens et al., 1991) have been used to study the dimeric nature of angiotensin II (ANGII) receptors. Using similar methods, other GPCRs such as the dopamine D2 (Ng et al., 1996), calcium-sensing (Bai et al., 1998), and chemokine (Rodriguez-Frade et al., 1999) receptors have also been demonstrated to exist as dimeric units. Other experiments such as photoaffinity labeling of MRs from various brain regions and the heart have also provided evidence for the presence of inter convertible dimers and tetramers (Avissar et al., 1983). The idea of higher order oligomerization was further supported by comparable findings of thyrotropin hormone receptor complexes (Gennick et al., 1987).

In the 1990s, another line of evidence demonstrating GPCR oligomerization stemmed from *trans*-complementation, and dominant-negative and -positive effect studies on co-expression of chimeric and mutant receptor constructs. One of the first experiments of these series was performed in 1993 by co-expression of chimeric muscarinic and adrenergic receptors composed of their C-terminal units (Maggio et al., 1993). Although no binding activity was detected when the chimeras were expressed individually, co-expression restored ligand binding to a level comparable to wild-type receptors and allowed agonist-dependent phospholipase C (PLC) activation (Maggio et al., 1993). These observations were explained as molecular interaction of the chimeras restoring functionality in a heteromeric complex (Maggio et al., 1993). Other *trans*-complementation studies for GPCRs such as the ANGII type 1a (AGTR1a; Monnot et al., 1996), somatostatin (SST – SST1 and SST5; Rocheville et al., 2000) and dopamine (D2 and D3; Scarselli et al., 2001) receptors have also provided useful information in support of GPCR oligomerization. This concept was further explored in experiments showing dominant-negative or -positive effects such those observed with the antidiuretic V2 vasopressin (Zhu and Wess, 1998) and ADRB2 (Hebert et al., 1998), respectively. Although there are reports of different functional effects, the various studies described are useful in offering more clues to molecular interactions amongst GPCRs.

Though earlier experiments provided a large pool of information suggesting GPCR oligomerization, there was still no direct evidence to support this concept. Early western blot studies often showed presence of immuno-reactive bands with molecular masses two or more times greater than a single receptor (Harrison and van der Graaf, 2006). Western blot was later performed in combination with immunoprecipitation using antibodies to endogenous or epitope tagged-receptors, providing for the first time direct evidence of GPCR interaction. Using this technique, homodimerization of ADRB2 (Hebert et al., 1996) was demonstrated. For example, cross-species oligomerization was also detected in select serotonin receptors with lysophosphatidic acid receptors 1 and 3 and γ-aminobutyric acid B2 receptors (Salim et al., 2002), dopamine D1 and D2 receptors (Lee et al., 2004), and α1b and β<sup>2</sup> adrenergic receptors (Uberti et al., 2005).

# **RET APPROACH IN GPCR OLIGOMERIZATION**

Since the days of radioactive inactivation experiments, the concept of GPCR oligomerization has been confirmed with emergence of biophysical data providing more in-depth evidence. In the last decade, RET techniques have gained popularity and are one of the best resolution strategies developed for direct study of GPCR interactions. The basis of these techniques lies in non-radiative energy transfer from an excited "donor" molecule to an "acceptor" (Wu and Brand, 1994). Bioluminescence RET (BRET) and fluorescence RET (FRET) are two of the most widely used RET approaches. BRET relies on a naturally occurring biophysical process in marine species between a luminescent donor and fluorescent acceptor, while only fluorescent molecules are utilized in FRET (Wu and Brand, 1994). These have provided advantages over traditional biochemical methods including high signal to noise ratios, use of intact cells, precise targeting of fusion receptors and actual quantification of the proportion and type of GPCR oligomers formed, and therefore are currently the most commonly used methods. Based on these approaches, a wide range of GPCRs have since been identified to function as oligomers, including the yeast α-factor (Overton and Blumer, 2000), dopamine D2 (Wurch et al., 2001), thyrotropin (Latif et al., 2001, 2002), opioid (Ramsay et al., 2002), neuropeptide Y (Dinger et al., 2003), melatonin (Ayoub et al., 2004), adrenergic (Mercier et al., 2002; Ramsay et al., 2002; Carrillo et al., 2004), and chemokine (Percherancier et al., 2005) receptors. Despite the successes in identifying various oligomers, there are several drawbacks of the BRET and FRET techniques. With regards to these aforementioned RET techniques, a primary concern is that the signals generated cannot be used to discriminate between non-mature and mature proteins within the various cellular locations (e.g., intracellular complexes and plasma membrane; Gandia et al., 2008; Cottet et al., 2011). Therefore, the information provided is essentially limited and more steps may be needed to address this concern. Moreover, another drawback of BRET and FRET is potential bleed-through artifacts which may be due to spectral overlaps, substrate instability, and auto-luminescence from serum containing medium, which together skew the interpretation of the results (Pfleger et al., 2006; Gandia et al., 2008). Therefore, the careful use of appropriate controls and fluorescent molecules need to be considered when using these RET techniques for the study of oligomerization. More recently, homogeneous time-resolved fluorescence (HTRF) which combines FRET with time-resolved measurements has also been considered. HTRF provides the advantage of increasing assay sensitivity and accuracy via replacing donor molecules with rare-earth lanthanides (e.g., europium cryptate) whose properties include non-auto-fluorescence and long emission half-lives (Degorce et al., 2009). Using HTRF, a number of receptors have been identified to dimerize, e.g., GABA (Maurel et al., 2004) and metabotropic glutamate-like receptors (Rondard et al., 2006; Brock et al., 2007) and oligomerize, e.g., dopamine D2 (Guo et al., 2008) and histamine H4 (Albizu et al., 2006) receptors. Fluorescence life-time imaging microscopy (FLIM) is also useful in overcoming the drawback of the use of intact cells and bleed-through in BRET and FRET assays. Moreover, FLIM can also provide useful information on the proportion of non-dimerizing and dimerizing partners and also visual images indicating where the signals are localized, for example the dimerization of the transcription factor CCAAT and enhancer binding protein-α in live mouse pituitary cell nuclei (Sun et al., 2011).

# **OLIGOMERIZATION OF CLASS B GPCRs**

"fendo-03-00175" — 2013/1/2 — 20:01 — page 2 — #2

Although techniques in studying oligomerization have existed for more than three decades, information with regards to the class B GPCRs is still in its infancy and largely contributed by the newer RET techniques. In class A GPCRs, several important interfaces including the transmembrane (TM) domains have been identified to contribute toward oligomerization. Moreover, oligomerization has also been shown to play a role in modulatingfunction, via altering parameters such as ligand binding, signaling, and trafficking. Though the SCT receptor (SCTR) remains to be the most extensively studied class B GPCR, information regarding other class B

GPCRs continues to grow. Collectively, the SCT superfamily of hormones and receptors represent important drug targets due to their physiological roles in glucose homeostasis, feeding behavior, and vascular contractility. They are characterized by a large N-terminal extracellular domain (ECD) containing six cysteine residues for disulfide bond formation and *N*-glycosylation sites for receptor conformation and ligand binding. The N-terminal ectodomain is linked to seven hydrophobic TM domains arranged in α-helical bundles connected by three exoloops, three endoloops, and a short cytosolic C-terminus (Miller et al., 2007). Currently, there are intense efforts to link the various structural domains to oligomerization in hopes of unraveling the specific molecular determinants that may affect certain pathological conditions. Taken together, information regarding the oligomerization of class B GPCRs will be highly valuable in enhancing the specificity of future drug design.

#### **SECRETIN RECEPTORS**

The SCTR was first cloned from the rat in 1991 (Ishihara et al., 1991) and is now recognized as one of the most extensively studied class B GPCRs, and the first that was demonstrated to have the ability to form oligomers. By investigating wild-type and exon-3 splice variants using confocal imaging and BRET, SCTR has been shown to be capable of homo- and heterodimerization (Ding et al., 2002a,b; **Tables 1 and 2**). When expressed alone, the exon-3 splice variant receptors lack SCT binding and signaling activity but are able to traffic normally to the cell surface. However, co-expression of the variant with the wild-type SCT results in a dominantnegative effect, indicating direct physical interaction. Since the variant receptor is predominantly expressed, comprising up to 70% of SCTRs in gastrinoma, over the wild-type receptor, heterodimerization has been suggested to be an underlying factor in facilitating tumor growth through its dominant-negative effects.


Techniques used for study and domains indicated to be important are described. SCTR, secretin receptor; GCGR, glucagon receptor; GLP1R, glucagon-like peptide 1 receptor; GLP2R, glucagon-like peptide 2 receptor; GIPR, glucose-dependent insulinotropic polypeptide receptor; VIPR1, vasoactive intestinal peptide (VIP) receptor 1; VIPR2, VIP receptor 2; ADCYAP1R1, PACAP type 1 receptor; GHRHR, growth hormone-releasing hormone receptor; PTHR1, parathyroid hormone receptor type 1; CALCR, calcitonin receptor; CALCRL, CALCR-like receptor; CRHR, corticotropin-releasing hormone receptor; FRET, fluorescence resonance energy transfer; BRET, bioluminescence resonance energy transfer; n/a, information not available.

"fendo-03-00175" — 2013/1/2 — 20:01 — page 3 — #3



Techniques used for study are described.

This study in 2002 was a breakthrough for class B GPCRs, showing the first physiological relevance of class B GPCR oligomerization. Taken together with data from subsequent studies, SCTR homodimerization is now recognized to occur, independent of ligand binding (Ding et al., 2002a; Harikumar et al., 2006).

To decipher the molecular determinants of oligomerization, receptors with mutated N- and C- terminal domains were studied (Lisenbee and Miller, 2006). Mutant receptors, as expected, were unable to bind SCT or sort efficiently to the plasma membrane, but the ability to still produce BRET signals above background suggested that the TM spanning core alone was sufficient for oligomerization (Lisenbee and Miller, 2006). Moreover, SCTR dimerization was also found to be independent of the GxxxG helix–helix motif (Lisenbee and Miller, 2006), which is essential for oligomerization of class A receptors, such as the yeast α-factor (Overton et al., 2003) and β<sup>2</sup> adrenergic (Salahpour et al., 2004) receptors. In addition, FRET signals within organelles of the receptor biosynthetic pathway including the endoplasmic reticulum, Golgi clusters, and plasma membrane were also detected, suggesting the occurrence of oligomerization even during the maturation of nascent molecules (Lisenbee and Miller, 2006). To further explore the domains responsible for homodimerization, TM peptide competition experiments have been performed. Of the seven TM peptides applied, TM4 was the only segment that disrupted BRET signals, suggesting it to be a functionally important interface for oligomerization (Harikumar et al., 2007). Using alanine mutagenesis, the contribution of TM4 in SCTR oligomers was pinpointed to two lipid-facing residues: Gly243 and Ile<sup>247</sup>

(Harikumar et al., 2007). Although no direct pathophysiological connection has been made regarding these new data, reduced signaling of monomeric forms suggests that SCTR dimers are necessary in optimizing functionality (Harikumar et al., 2007). Most recently, bimolecular luminescence complementation and BRET experiments have clarified SCTR's existence as homodimers and not higher order oligomers (Harikumar et al., 2008a). Therefore, previous references to oligomerization of SCTRs should be referred to as homodimerization unless otherwise stated. Aside from homodimers, SCTRs have also been implicated to form heterodimers with a wide range of class B GPCRs, including the glucagon-like peptide 1 receptor (GLP1R), glucagon-like peptide 2 receptor (GLP2R), vasoactive intestinal peptide receptor 1 (VIPR1), vasoactive intestinal peptide receptor 2 (VIPR2), growth hormone-releasing hormone receptor (GHRHR), parathyroid hormone 1 receptor (PTHR1), parathyroid hormone 2 receptor (PTHR2), and calcitonin receptor-like (CALCRL; Harikumar et al., 2006, 2008b; **Table 2**).

#### **GCG, GLP1, GLP2, AND GIP RECEPTORS**

Receptors for glucagon (GCG), glucagon-like peptide 1 (GLP1), glucagon-like peptide 2 (GLP2), and glucose-independent insulinotropic polypeptide (GIP) are well-known regulators of carbohydrate, fat, and protein metabolism. Oligomerization within the GCG subfamily of receptors was recently identified but with some controversy (Roed et al., 2012; Schelshorn et al., 2012). Homodimerization of each GCGfamily receptor member has been reported through BRET experiments (Schelshorn et al., 2012), however, these events remained unconfirmed with the lack of appropriate negative controls and the fact that another group was unable to repeat some of these results (Roed et al., 2012). Amongst the receptors, the GCG receptor (GCGR) and GIP receptor (GIPR) exhibited the strongest ability to homodimerize but BRET signals were slightly reduced upon application of GCG or GIP, while the GLP receptors were weakly responsive and unaffected by agonist stimulation (Schelshorn et al., 2012). Further studies with BRET saturation experiments confirmed GLP1R and GIPR homodimerization, with some agonist-induced reduction (Schelshorn et al., 2012). On the basis of these results, it was suggested that ligand binding leads to a decrease in affinity and conformational change of the homodimers (Schelshorn et al., 2012). In a similar study, the same BRET saturation results were obtained for the GLP1R, however no changes were observed when GLP1 was added (Roed et al., 2012). This was similarly the case for GCGRs (Roed et al., 2012). Taken together, it is highly likely that the GCG subfamily receptors are able to form homodimers (**Table 1**), while further studies with appropriate use of controls are necessary to solidify this conclusion.

Despite the controversy, heterodimerization of GLP1R and GIPR has been confirmed by BRET saturation and kinetic experiments (Roed et al., 2012; Schelshorn et al., 2012; Whitaker et al., 2012; **Table 2**). The heterodimerization event between GLP1R and GIPR was found to be dose-dependently induced by GLP1 and inhibited by GIP (Schelshorn et al., 2012). Further application of GLP1 and its antagonists to this system showed heterodimerization to occur independently to GLP1R activation, but was induced by ligand binding, with BRET signals reaching a maximum 10–30 s

"fendo-03-00175" — 2013/1/2 — 20:01 — page 4 — #4

after addition of GLP1 (Schelshorn et al., 2012). More specifically, the heterodimers were suggested to play a role in allosteric modulation with arrestin recruitment. In basal conditions, GLP1R and GIPR are proposed to exist mainly as monomers or homodimers, GLP1, under low concentrations (<30 nM), acts as a high affinity ligand for functioning of GLP1R but switches to become a low affinity ligand to GIPR with high concentrations (>30 nM). This promotes formation of heterodimers with altered G protein coupling resulting in GIP-like activity that can later be dissociated in the presence of GIP (Schelshorn et al., 2012). Heterodimerization of GLP1R and GIPR was also confirmed by a later study with wild-type GLP1R rescuing mutant GIPR function (Whitaker et al., 2012).

#### **VIP/PACAP RECEPTORS**

To elicit physiological response, vasoactive intestinal peptide (VIP) and pituitary adenylate cyclase activating polypeptide (PACAP) bind to three receptors, namely VIPR1, VIPR2, and PACAP receptor type 1 (ADCYAP1R1). Discoveries of VIPR1 and VIPR2 homodimers were made during the study of SCTR oligomerization in 2006 (Harikumar et al., 2006; **Table 1**). These physical interactions were demonstrated by BRET experiments and found to be modulated by agonist but not antagonist binding (Harikumar et al., 2006). In the same study, BRET results also indicated that VIPR1 and VIPR2 to form hetero-oligomers and VIP binding was found to affect this association negatively (Harikumar et al., 2006; **Table 2**). Formation of these oligomeric complexes was also supported by co-immunoprecipitation studies (Langer et al., 2006). Besides formation of hetero-oligomers between the two subtypes, both VIPRs can oligomerize with SCTR. Interestingly, co-expression of SCTR with either VIPR resulted in strong intracellular FRET signals, suggesting the trapping of hetero-oligomers within organelles of the biosynthetic pathway (Harikumar et al., 2006). Since this retention occurred independent to SCT and VIP stimulation, it has been described as a regulatory step, allowing SCTR to have dominant-negative effects by inhibiting VIP's action on cells expressing both receptors (Harikumar et al., 2006). Within a physiological context, the SCT and VIPR are often co-expressed; therefore information regarding their physical interactions may provide some insights toward diseases such as pancreatic carcinoma (Estival et al., 1981; Ding et al., 2002a). Of the receptor trio, information regardingADCYAP1R1 oligomerization is scarce with only one study indicating its homodimerization by time-resolved FRET studies (Maurel et al., 2008; **Table 1**).

#### **GROWTH HORMONE-RELEASING HORMONE RECEPTORS**

Growth hormone-releasing hormone (GHRH) is a hypothalamic peptide responsible for stimulating growth hormone release and disruption of the peptide–receptor pair often results in abnormal growth, such as dwarfism or gigantism. GHRHRs have been suggested to form oligomers, with co-expression of mutant and wild-type receptors having dominant-negative effects (McElvaine and Mayo, 2006). The physical interaction amongst GHRHRs was implied with the reduction of ligand binding ability and cyclic adenosine monophosphate (cAMP) production to 60% in presence of truncated receptors (McElvaine and Mayo, 2006). These observations were further supported by detection of both receptor forms by differentially tagged co-immunoprecipitation studies (McElvaine and Mayo, 2006). The relevance of these findings has been hypothesized in the context of pituitary adenomas for example, with preferential expression of the mutant receptor down-regulating signaling and thus modulating growth hormone release (Matsuno et al., 1999).

#### **PARATHYROID HORMONE RECEPTORS**

Parathyroid hormone receptors (PTHRs) play an important role in calcium homeostasis and bone maintenance, mediating the effects of its natural ligands parathyroid hormone (PTH) and PTH-related protein (PTHrP). In 2008 and 2009, the crystal structures of PTH and PTHrP bound human PTHR1 were determined by engineering the receptor as a readily crystallizing maltose binding protein (Pioszak and Xu, 2008; Pioszak et al., 2009). A "hot dog bun" three-layer α-β-βα fold structure was described, forming the central hydrophobic groove within which the ligands in an amphipathic helical fashion docked (Pioszak and Xu, 2008; Pioszak et al., 2009). More importantly, the crystal structures depicted ligand binding to PTHR1 ECD monomers in a 1:1 stoichiometric ratio (Pioszak and Xu, 2008; Pioszak et al., 2009). Interestingly a year later, the crystal structure of ligand-free PTHR1 showed the ECDs as dimers (Pioszak et al., 2010). These ECD dimers were formed in a similar fashion to ligand-bound PTHR1, with the C-terminal of one ECD forming α-helices to occupy the peptide-binding groove of the other (Pioszak et al., 2010). Besides crystallography, BRET and FRET experiments also supported existence of these ECDmediated PTHR1 homodimers (Pioszak et al., 2010). To confirm the ligand-independent nature of ECD dimerization shown by the crystal structures, PTH was applied to PTHR1s in BRET studies and as predicted, led to agonist-induced dissociation of the PTHR1 homodimers (Pioszak et al., 2010). Furthermore, coexpression of mutant and wild-type ECD PTHR1s showed that monomeric units of the PTHR1 were sufficient for functionality (Pioszak et al., 2010). With the discovery of monomeric and dimeric units of PTHR1, this may be useful in understanding how the two biologically distinct peptides PTH and PTHrP are able to modulate function through binding to the same receptor (Pioszak et al., 2010).

# **CALCITONIN AND CORTICOTROPIN-RELEASING HORMONE RECEPTORS**

"fendo-03-00175" — 2013/1/2 — 20:01 — page 5 — #5

Targeting mainly osteoblasts, calcitonin is important in calcium homeostasis and has anorexic and analgesic effects in the central nervous system. In a study focused on rabbit calcitonin receptors (CALCRs), homo- and heterodimers of the alternatively spliced CALCR (Δe13) and CALCR type 1a isoform (CALCR1a) were detected by co-immunoprecipitation and FRET analyses (Seck et al., 2003). Co-expression resulted in a dominant-negative effect of the splice variant Δe13 over CALCR1a, resulting in reduced CALCR1a surface expression, cAMP response, and extracellular signal-regulated kinase (ERK) phosphorylation (Seck et al., 2003). The combined data highlight the importance of CALCR1a homodimers in proper expression and functioning, and also suggest a role of Δe13 in regulating CALCR1a expression (Seck et al., 2003). In contrast, both static and saturation BRET studies on the human CALCR were found unable to form homodimers (Harikumar et al., 2010). On the basis of SCTR studies where TM4 was identified to be an important interface (Harikumar et al., 2007), the same TM was also considered for CALCR (Harikumar et al., 2010). Comparing human and rabbit TM4 domains, only one single amino acid was found to differ between the two species, namely Arg and His at position 236, respectively (Harikumar et al., 2010). Synthetic TM4 peptides were constructed by exchange of these identified lipid-facing residues to those from respective CALCR species and human SCTR, and their application enhanced human CALCR BRET signals (Harikumar et al., 2010). Similar to the SCTR, lipid-exposed residues in TM4 are also key in facilitating homodimerization events in human CALCR (Harikumar et al., 2010). Though not studied in-depth, the closely related CALCRL has also been demonstrated to exist as homodimers (Heroux and Bouvie, 2005).

Within the class B GPCRs, the corticotropin-releasing hormone receptors (CRHRs) are the least studied. CRHRs are mediators of stress response, modulating the release of adrenocorticotropic hormone from the anterior pituitary. Two subtypes of the receptors exist, but only subtype 1 (CRHR1) has been studied for oligomerization. Data from FRET and co-immunoprecipitation studies show evidence of CRHR1 forming constitutive homodimers independent to ligand binding but their physiological relevance remains unexplored (Kraetke et al., 2005; Young et al., 2007).

# **FUNCTIONAL CONSEQUENCES OF OLIGOMERIZATION IN CLASS B GPCRs**

Within the large superfamily of cell surface proteins, obligatory oligomerization is unique only to class C GPCRs and recognized as a prerequisite for activation and allosteric modulation (Kniazeff et al., 2011; Chun et al., 2012). However, in the case of class A and B GPCRs where oligomerization has also been established as a real event, the implications on their functions remain unresolved. This is further complicated by the fact that monomeric GPCRs are demonstrated to be independently capable of regulating signal transduction via a series of processes including the activation of G proteins (White et al., 2007; Whorton et al., 2007, 2008; Kuszak et al., 2009), kinase phosphorylation and signal "switchoff" by arrestin binding (Xu et al., 1997; Bayburt et al., 2011). To decipher the underlying functional significance, the effect of GPCR oligomerization on various functional properties needs to be examined with emphasis on class B GPCRs, as examples.

# **LIGAND BINDING**

Ligand binding, either with an agonist or antagonist, is fundamental in activating subsequent downstream signaling cascades. Natural ligands for the class B GPCRs share common features, comprising of 25 or more amino acids and have diffuse pharmacophoric domains (Ulrich et al., 1998). Critical residues have been identified throughout the length of the peptides (Dong et al., 2011), with the N-terminus recognized as necessary for receptor binding and signaling, whilst the C-terminus contributes mainly to high affinity binding and receptor specificity. Crystallography (Sasaki et al., 1975; Parthier et al., 2007) and nuclear magnetic resonance (NMR; Grace et al., 2007; Venneti and Hewage, 2011) analyses have also been useful, elucidating the class B ligands to preferentially assume an α-helical conformation particularly in their C-terminal regions. Structure–activity studies have further implicated the specific interaction of ligand–receptor regions, with the C-terminal ligand domain first fitting into the receptor ECD peptide-binding cleft, thus facilitating the N-terminal ligand domain to associate with the receptor core helical bundle domain (Dong et al., 2004, 2008; Mann et al., 2007; Parthier et al., 2009). Though the molecular interactions described are on the basis of monomeric GPCRs, similar associations may be expected from oligomeric forms. Similarly, molecular modeling and crystallization studies have predicted the binding pockets in SCTR and PTHR homodimers, respectively, to be localized in the extracellular N-terminal domains (Gao et al., 2009; Pioszak et al., 2010). Moreover, the SCTR homodimer peptide-binding clefts are demonstrated to reside opposite to the dimerization interfaces that may allow structural freedom for each protomer to bind to its respective ligand in a 1:1 stoichiometric ratio, as expected for monomeric GPCRs (Gao et al., 2009). Though limited, the two models show oligomerization to have minimal structural modifications in terms of adequately exposing the binding pockets, allowing ligand–receptor interactions in class B GPCRs.

In majority of studies for this receptor class, the use of peptide or mutant receptors to inhibit dimerization events have been utilized to explore the contribution of such receptor interactions towardfunctionality. On the basis of these studies, oligomerization has been demonstrated to have differential effects on the ligand binding affinity of class B GPCRs. Using TM4 peptides that are known to disrupt SCTR homodimerization, both monomeric and dimeric SCTR forms were found to have analogous ligand binding properties (Harikumar et al., 2007). Similarly, oligomerization was also found to have null effects on the ligand binding affinity of PTHR1, with the non-dimerizing R179A/V183A double mutant maintaining wild-type PTH binding activity (Pioszak et al., 2010). In another study, CALCR mutants also bound calcitonin saturably with no statistically significant differences in binding affinity (Harikumar et al., 2010). In contrast, disruption of GLP1R homodimerization by TM4 peptides resulted in elimination of G protein-dependent high affinity binding to GLP1(7–36)NH2 (Harikumar et al., 2012). Moreover, co-expression of full-length and truncated GHRHRs resulted in a decrease in GHRH ligand binding and has been described to be resultant of the truncated receptor inducing a dominant-negative effect, leading to conformational changes which inhibit proper binding (McElvaine and Mayo, 2006). Nevertheless, these contrasting findings are consistent with recent reports of oligomerization having variable effects on the kinetics and efficiencies amongst receptors (Dorsch et al., 2009; Hern et al., 2010). Therefore, oligomerization may be important in modulating receptor activity via altering ligand binding affinity only in certain class B GPCRs.

# **ALLOSTERY**

"fendo-03-00175" — 2013/1/2 — 20:01 — page 6 — #6

In recent studies, a link between oligomerization and allosteric regulation has been established in class B GPCRs. Allostery concerns association of a ligand or molecule to the receptor other than its orthosteric site, modifying its functional properties in either a positive or negative manner (May et al., 2007). Using a classical radio-ligand binding approach, the dissociation of labeled SCT in the absence or presence of unlabeled SCT was monitored to study the negative binding cooperativity of SCTRs (Gao et al., 2009). The wild-type dimerizing receptors were found to shift from monophasic to biphasic dissociation upon presence of unlabeled SCT peptide, whilst the non-dimerizing mutant receptors maintained monophasic dissociation, regardless of treatment (Gao et al., 2009). On the basis of these results, the homodimeric SCTRs have been speculated to predominantly exist in a G protein-coupled high affinity state and the monomeric forms in a G protein-uncoupled low affinity state (Gao et al., 2009). Though the molecular basis responsible for the allosteric effects are still unknown, steric hindrance by the two large extracellular amino-terminal domain or selective coupling may be underlying factors contributing to the negative cooperativity observed in SCTRs (Miller et al., 2012). Similarly, the homodimerization event amongst the GCGR and GLP1R have also been identified to exhibit negative cooperativity with addition of labeled native peptides to their respective receptors leading to accelerated dissociation of unlabeled ligands (Roed et al., 2012). More recently, the ECD of the GCGR has been demonstrated to act as a negative regulator of receptor activity through interaction with the third extracellular loop (Koth et al., 2012), a region not only recognized to contribute to receptor affinity and efficacy, but may also aid in optimizing the arrangement of the TM bundle for efficient ligand binding (Peeters et al., 2011). As this case, agonistoccupied homodimers may induce negative cooperativity through structural modifications that affect ligand binding. Though the molecular determinants remain to be confirmed, there is a clear relation between dimerization and negative allostery. Aside from the typical ligand-induced allostery, physical coupling of each protomer in a oligomeric complex may also play a role (Haack and McCarty, 2011). Within the class B GPCRs, this has only been demonstrated in heteromers of the GLP1R and GIPR, where GIPR is suggested to act allosterically on the GLP1R to decrease maximal responses in calcium flux and β-arrestin recruitment assays (Schelshorn et al., 2012). In a later study, mutant GIPR function rescue by wild-type GLP1R was observed, with reduction in GLP1 responsiveness (Whitaker et al., 2012). Again GIPR may be exerting an allosteric effect on GLP1R, but the precise molecular basis needs to be explored with further structural and functional studies.

#### **EFFECTS OF AGONIST OCCUPATION**

Besides allostery, agonist occupation of class B GPCRs can also promote the association or dissociation of oligomeric complexes. In contrast to SCT homodimers that are unaffected by ligand binding (Ding et al., 2002a; Harikumar et al., 2006), GLP1 binding was found to encourage the dimerization between GLP1R and GIPR (Schelshorn et al., 2012). A model proposed describes GLP1 to act as a high affinity ligand for GLP1R at low concentrations, but also becomes a low affinity ligand for GIPR at higher concentrations to trigger formation of the GLP1 and GIPR dimer (Schelshorn et al., 2012). Interestingly, the presence of the high affinity GIP reverses the GLP1 induced association, dissolving the heteromer to a monomeric state (Schelshorn et al., 2012). As previously discussed, the GIPR has allosteric effects on the GLP1R when heterodimerized. Taken together with the agonist-induced association and/or dissociation of receptors, this may be a mechanism by which the receptors activate their G proteins and subsequent downstream signaling cascades (Gomes et al., 2001). Agonist-induced dissociation has also been reported for the PTHR1 via structural mimicry of PTH/PTHrP binding (Pioszak et al., 2010), however the functional relevance remains unclear. Other examples of agonist-induced disruption of oligomerization include VIPR1 and VIPR2 homodimers and VIPR1/VIPR2 heterodimers (Harikumar et al., 2006).

#### **SIGNALING**

Another prominent effect of oligomerization in class B GPCRs is on signaling selectivity and efficiency. In particular, cAMP and PLC are two of the predominant pathways in which physiological responses are elicited through in class B GPCRs. Therefore most oligomerization studies have extended their focus beyond information from ligand binding and BRET studies, but also rely on the cAMP and calcium mobilization responses of receptors. In particular, SCTRs are one of the first class B GPCRs demonstrated to have altered functionality due to homodimerization. By TM4 peptide competition studies, the disruption of SCTR dimerization was found to result in impairment of the generation of cAMP (Harikumar et al., 2007). The dimeric form of SCTR was, therefore, suggested be the optimal conformation for G protein activation (Harikumar et al., 2007). Similarly, disruption of the GLP1R homodimers was found to not only affect cAMP formation and ERK phosphorylation, but also lead to a complete loss of intracellular calcium mobilization response (Harikumar et al., 2012). In the case of GLP1R homodimers, signal transduction involves more than one G protein and/or accessory proteins such as β-arrestins. Nonetheless, dimerization is likely the basis in altering G protein coupling efficiency (Harikumar et al., 2012). This was also evident in the heterodimerization of GLP1 and GIP receptors, with the heteromer having more "GIP-like" (Schelshorn et al., 2012) pharmacological characteristics. Based on most of the studied class B GPCRs, oligomerization seems to exhibit a general trend in affecting G protein activation and signaling efficiency. However, oligomerization should not be a prerequisite for signaling, as monomers such as the PTHRs have been demonstrated to be capable of activating G proteins on their own (Pioszak et al., 2010). Therefore, oligomerization is likely important in providing structural framework to facilitate G protein coupling and thus modulate function.

#### **TRAFFICKING**

"fendo-03-00175" — 2013/1/2 — 20:01 — page 7 — #7

A number of studies have suggested the importance of oligomerization in cellular trafficking of the receptors and is a process that occurs early in the biosynthetic pathway (Bulenger et al., 2005). Within the class B GPCRs, this aspect of oligomerization has been most thoroughly explored for the SCTRs. Indeed using a combination of fluorescence and morphologic FRET microscopy studies, the SCTR homodimers were localized within organelles of the biosynthetic pathway including the tubular endoplasmic reticulum, Golgi clusters and the plasma membrane (Lisenbee and Miller, 2006). The homodimerization of SCTRs has therefore been suggested to occur as early as newly synthesized receptors and continue throughout their maturation process (Lisenbee and Miller, 2006). Moreover, non-dimerizing SCTR constructs were not expressed nor yielded a significant morphological FRET signal both intracellularly and on the plasma membrane (Harikumar et al., 2007). Taken together, the SCTR data support the idea that oligomerization may be necessary for receptors to by-pass quality checkpoints before being delivered to their site of action (Bulenger et al., 2005). Consistent with the SCTR data, the VIPRs have also been identified in the biosynthetic pathway and plasma membrane as homo- and heterodimeric complexes (Harikumar et al., 2006). Interestingly, heterodimerization involving the SCTR and VIPR impaired trafficking to the plasma membrane with morphological FRET analyses revealing these heteromers to be trapped within the biosynthetic pathway, lacking cell surface expression (Harikumar et al., 2006). In the case of VIPRs, heterodimerization with SCTR predominates over the homodimerization of VIPRs (Harikumar et al., 2006). This demonstrates dominant-negative inhibition of oligomerization on cellular trafficking, which may be useful for modulating function in cells expressing both the SCTR and VIPR (Harikumar et al., 2006). In a later study, a small amount of mutant GIPR completely lacking *N*-glycosylation and cell surface expression was rescued by coexpression with GLP1R but not GIPR, suggesting the possibility of a heteromer which has a dominant-positive effect in facilitating the trafficking of GIPR (Whitaker et al., 2012).

# **STRUCTURAL ELEMENTS GOVERNING OLIGOMERIZATION IN CLASS B GPCRs**

With the class B GPCRs key in a number of pathophysiological diseases and mounting evidence indicating their oligomerization to have influence in modulating various functional properties, this has fueled research in unraveling the structural determinants. To date, majority of studies have suggested the importance of different domains in GPCR oligomerization. However within the class B GPCRs, a common theme seems to prevail with several lines of evidence indicating the TM domains to be important. More specifically, it is the fourth TM domain, which has been implicated as a key molecular domain in class B receptor homodimerization (**Table 1**). This has been well documented by the disruption of SCTR homodimerization by TM4 peptides, with mutation studies identifying the lipid-facing residues namely Gly<sup>243</sup> and Ile<sup>247</sup> as important interfaces (Harikumar et al., 2007). Similarly, TM4 also provides the primary interface for GLP1R homodimerization, as supported by experimental data with TM4 peptide treatment dissociating the complex (Harikumar et al., 2012). Consistent with findings from the SCTR studies, lipidfacing residues within TM4 of the CALCR were also found to be necessary for dimerization (Harikumar et al., 2010). Though TM4 has been indicated to provide basis for dimerization in class B GPCRs, other domains including the ECD in PTHR (Pioszak et al., 2010) and C-terminal tail in CALCR (Harikumar et al., 2010) may also have some role. More recently, a pseudo signal peptide unique to the CRHR2a was identified to inhibit

#### **REFERENCES**

Albizu, L., Balestre, M. N., Breton, C., Pin, J. P., Manning, M., Mouillac, B., et al. (2006). Probing the existence of G protein-coupled receptor dimers by positive and negative liganddependent cooperative binding. *Mol. Pharmacol.* 70, 1783–1791.

Avissar, S., Amitai, G., and Sokolovsky, M. (1983). Oligomeric structure oligomerization (Teichmann et al., 2012). Steric hindrance by the bulky nature of the high mannose glycan protruding from this N-terminal domain has been suggested, thus preserving CRHR2a as monomers (Teichmann et al., 2012). Current prediction programs are unable to differentiate between canonical and pseudo signal peptides, as a result the presence of similar structures may also exist in other class B GPCRs (Rutz et al., 2006; Teichmann et al., 2012). Taken together, dimerization of class B GPCRs may not necessarily be dependent solely on one structural motif but instead is a highly complex interaction amongst several domains to achieve optimal structural basis for the fine-tuning of physiological processes.

# **FUTURE PERSPECTIVES**

Over the last three decades, there has been a rapid development of techniques for studying the molecular interaction between GPCRs. Though early evidence was largely overlooked and limited to the class A GPCRS, establishment of RET techniques, in particular, have led to reinterpretation and strongly supports the concept of GPCR oligomerization. In recent years, class B GPCRs are also gaining merit in their ability to form homoand hetero- oligomers, with the SCTRs being the most extensively studied. Though still nascent, the information available supports the role of dimerization in modulating function of class B GPCRs. Naturally, ligand binding to a receptor results in conformational changes in G protein arrangement to elicit downstream signaling pathways. When GPCRs associate, conformational arrangements also occur to modulate functionality. For example, SCTR homodimers have greater signaling responses than its monomeric counterparts. Furthermore, GPCR dimerization may also act as a repair mechanism, *trans*-complementing domains to restore binding sites and thus signaling, as observed for spliced SCTRs (Ding et al., 2002a,b). In terms of trafficking, oligomerization may also play a role with evidence of SCTR homodimers present as early as nascent molecules throughout the biosynthetic pathway (Lisenbee and Miller, 2006). Though seemingly simple when considering homodimeric units, evidence of GPCR heterodimerization, for example of SCTRs with a wide range of class B GPCRs, such as VIPR and GCGR, suggests a more complex role. Attempts have also been made to link these physical characteristics to pathophysiological responses in class B GPCRs, for example, association of SCTR homodimers to gastrinoma (Ding et al., 2002b) and GLP1R/GIPR heterodimers to diabetes (Whitaker et al., 2012). GPCR oligomerization is therefore functionally relevant and further research will be useful in providing more insights in understanding the molecular determinants to *in vivo* responses and for drug design.

#### **ACKNOWLEDGMENTS**

"fendo-03-00175" — 2013/1/2 — 20:01 — page 8 — #8

The present study was supported by the Hong Kong Government RGC grant GRF764510, CRFHKU6/CRF/11G to Billy K. C. Chow and GRF770212 to Leo T. O. Lee.

of muscarinic receptors is shown by photoaffinity labeling: subunit assembly may explain high- and lowaffinity agonist states. *Proc. Natl. Acad. Sci. U.S.A.* 80, 156–159.

Ayoub, M. A., Levoye, A., Delagrange, P., and Jockers, R. (2004). Preferential formation of MT1/MT2 melatonin receptor heterodimers with distinct ligand interaction properties


receptor in ductal pancreatic adenocarcinoma. *Cancer Res.* 62, 5223– 5229.


Christopoulos,A., et al. (2009). Functional importance of a structurally distinct homodimeric complex of the family B G protein-coupled secretin receptor. *Mol. Pharmacol.* 76, 264–274.


"fendo-03-00175" — 2013/1/2 — 20:01 — page 9 — #9

receptors. *Mol. Pharmacol.* 69, 363–373.


negative regulation of the glucagon receptor. *Proc. Natl. Acad. Sci. U.S.A.* 109, 14393–14398.


(2007). Peptide binding at the GLP-1 receptor. *Biochem. Soc. Trans.* 35, 713–716.


site by co-expression of two deficient mutants*. J. Biol. Chem.* 271, 1507–1513.


"fendo-03-00175" — 2013/1/2 — 20:01 — page 10 — #10


bovine adrenal zona glomerulosa. *Biochem. J.* 268, 443–448.


C1a isoform and inhibits its surface expression. *J. Biol. Chem.* 278, 23085–23093.


"fendo-03-00175" — 2013/1/2 — 20:01 — page 11 — #11


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 14 September 2012; accepted: 11 December 2012; published online: 04 January 2013.*

*Citation: Ng SYL, Lee LTO and Chow BKC (2013) Receptor oligomerization: from early evidence to current understanding in class B GPCRs. Front. Endocrin. 3:175. doi: 10.3389/fendo. 2012.00175*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Ng, Lee and Chow. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# The VPAC1 receptor: structure and function of a class B GPCR prototype

# *A. Couvineau\*, E. Ceraudo , Y.-V. Tan , P. Nicole and M. Laburthe\**

Faculté de Médecine Xavier Bichat, INSERM 773/Centre de Recherche Biomédicale Bichat Beaujon (CRB3), Université Paris 7, Paris Cedex 18, France

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

James Waschek, University of California at Los Angeles, USA Daniel Fourmy, University of Toulouse 3, France

#### *\*Correspondence:*

A. Couvineau and M. Laburthe, Faculté de Médecine X. Bichat, INSERM U773/CRB3, 16 Rue Henri Huchard, 75018 Paris, France. e-mail: alain.couvineau@inserm.fr; marc.laburthe@inserm.fr

The class B G protein-coupled receptors (GPCRs) represents a small sub-family encompassing 15 members, and are very promising targets for the development of drugs to treat many diseases such as chronic inflammation, neurodegeneration, diabetes, stress, and osteoporosis. The VPAC1 receptor which is an archetype of the class B GPCRs binds Vasoactive Intestinal Peptide (VIP), a neuropeptide widely distributed in central and peripheral nervous system modulating many physiological processes including regulation of exocrine secretions, hormone release, foetal development, immune response . . . VIP appears to exert beneficial effect in neurodegenerative and inflammatory diseases. This article reviews the current knowledge regarding the structure and molecular pharmacology of VPAC1 receptors. Over the past decade, structure–function relationship studies have demonstrated that the N-terminal ectodomain (N-ted) of VPAC1 plays a pivotal role in VIP recognition. The use of different approaches such as directed mutagenesis, photoaffinity labeling, Nuclear Magnetic Resonance (NMR), molecular modeling, and molecular dynamic simulation has led to demonstrate that: (1) the central and Cterminal part of the VIP molecule interacts with the N-ted of VPAC1 receptor which is itself structured as a - Sushi domain; (2) the N-terminal end of the VIP molecule interacts with the first transmembrane domain of the receptor where three residues (K143, T144, and T147) play an important role in VPAC1 interaction with the first histidine residue of VIP.

**Keywords: GPCR, photolabeling, VPAC1, VIP, mutagenesis, inflammation, neuroprotection, molecular modeling**

#### **INTRODUCTION**

Vasoactive Intestinal Peptide (VIP) discovered by Said and Mutt (1970) is an ubiquitous 28-aminoacid neuropeptide that is widely distributed in central and peripheral nervous system. During the past 10 years, VIP was also identified in the immune system where it plays the role of a "cytokine-like peptide" (Delgado et al., 2004). VIP plays an important role in human physiology (**Table 1**) such as in development, growth, immune responses, circadian rhythms, neuronal and endocrine control, neuroprotective actions, and in the functions of the digestive, respiratory, reproductive, and cardiovascular systems (Laburthe et al., 2007). Associated to its large distribution and biological functions, VIP may also play a role in various pathologies (**Table 1**). It has been identified as a very promising agent in the treatment of inflammatory and neurodegenerative diseases (Gozes et al., 2003; Delgado et al., 2004). Indeed, VIP appears to be a very potent anti-inflammatory peptide in animal models of Crohn disease (Abad et al., 2003), rheumatoid polyarthritis (Delgado et al., 2001), or septic shock. This neuropeptide belongs to the structural-related peptide named secretin/VIP family (**Table 2**) encompassing VIP, pituitary adenylate cyclase activating peptide (PACAP), secretin, growth hormone releasing factor (GRF), peptide having an histidine residue in N-terminal position and an isoleucine residue in C-terminal position (PHI and its human homolog PHM), helodermin, glucagon, gastric inhibitory polypeptide (GIP), glucagon-like peptide 1 and 2 (GLP-1 and GLP-2).

#### **VIP A POTENTIAL THERAPEUTICAL AGENTS**

Few years ago, VIP emerged as a potential therapeutic agent for various diseases including asthma (Groneberg et al., 2006), sexual impotence (Fahrenkrug, 2001), brain strokes (Dogrukol-Ak et al., 2004), chronic inflammation (Delgado et al., 2004), neurodegenerative disorders (Dejda et al., 2005), and cancers (Moody et al., 2011). Recently, a lot of reports have focused on the role of VIP and its receptors in chronic inflammation and neurodegenerative diseases.

VIP has been identified as a very promising agent in treatment of inflammatory diseases (Delgado et al., 2004). Indeed, VIP appeared to be a very potent anti-inflammatory peptide in animal models of various chronic inflammatory diseases (Couvineau and Laburthe, 2012a,b). The VIP anti-inflammatory effect has been widely studied (Delgado et al., 2004). These studies showed, in homeostasis condition, innate and adaptive immunity, that VIP can help to preserve the equilibrium between pro-inflammatory and anti-inflammatory response. In chronic

**List of non-standard abbreviations:** VIP, Vasoactive Intestinal Peptide; PACAP, Pituitary Adenylate Cyclase Activating Peptide; VPAC, VIP and PACAP receptor; Bpa, Benzophenone; N-ted, N-terminal ectodomain; ITF, Intrinsic Tryptophan Fluorescence.

#### **Table 1 | Major physiological and pathophysiological actions of VIP***a***.**


aReviewed in Gozes et al. (2003); Dickson and Finlayson (2009); Delgado and Ganea (2011); Moody et al. (2011); Harmar et al. (2012).

#### **Table 2 | Sequence alignments of class B GPCR ligands***b***.**


aBlack boxes represent sequence identity and light gray boxes represent sequence homology.

bNumbers indicate the length of the peptides.

inflammatory diseases (Crohn disease, rheumatoid polyarthritis, hepatitis, encephalomyelitis. . . ) a modification of this equilibrium can be induce by various stimuli such as pathogenic agents, auto-immunity, environment, genetic background. . . which lead to the stimulation of production of pro-inflammatory cytokines (IL-17, IL-1, IFNγ, TNFα. . . ) by macrophages and lymphocytes T (Th1 and Th17). Conversely, the anti-inflammatory response mediated by anti-inflammatory cytokines (IL-10, IL-4, IL-13, TGFβ. . . ) secreted by lymphocytes T (Th2 and Treg), was strongly inhibited. The VIP anti-inflammatory effect involves a "rebalancing" of immune system (Firestein, 2001) by inhibition of pro-inflammatory response (Th1 and Th17) and stimulation of anti-inflammatory response mediated by Th2 and Tregs. In parallel, VIP induces an inhibitory effect on innate immunity by inhibition of production of pro-inflammatory cytokines and chemokines secreted by macrophage. Moreover, VIP is able to strongly inhibit the production of reactive oxygen species (ROS) induced by fMLP in monocytes (personnal data). Moreover, various reports clearly demonstrate that VIP promotes tolerance by inducing expansion of Treg cells (Leceta et al., 2007). Whereas, some reports reveal that VIP-deficient mice are resistant to the development of induced-encephalomyelitis or inducedendotoxemia indicating that in these conditions VIP plays unexpected permissive and/or pro-inflammatory actions (Abad et al., 2010, 2012). Despite this effect, VIP represents a potential anti-inflammatory agent that could be used in human therapeutic treatment, although the VIP anti-inflammatory effects have been mainly described in animal models (Couvineau and Laburthe, 2012a,b). Whereas, the major obstacle to the use of VIP in clinic therapies is its high sensitivity to protease degradation. Indeed, removing of the first residues by peptidases, such as dipeptidyl peptidase IV (DPPIV), induces a drastic loss of affinity of VIP peptide family (Lambeir et al., 2001). To circumvent these labile properties, VIP can be modified to increase its resistance to degradation by N-acylation of the peptide N-terminal end or by substitution of residues involved in proteolytic consensus sequences (dibasic doublets). Recent data indicate that PACAP N-terminal modifications confer resistance to DPPIV (Bourgault et al., 2008). In the same way acetylation of the VIP N-terminal end increase its stability in the presence of human serum (personal data). Other strategies consist to protect peptide against degradation by insertion of VIP into micelles or nanoparticles (Fernandez-Montesinos et al., 2009; Onyüksel et al., 2009). Despite these limitations, VIP has been tested in a phase I clinical trial for the treatment of acute respiratory distress syndrome and sepsis (id: NCT00004494, http://www*.*clinicaltrials*.*gov).

In the mid-1980s, the first report of VIP neuroprotection, demonstrated that this peptide is able to prevent neuronal death associated with electrical blockade induced by tetrodotoxin (TTX) addition to primary spinal cord cultures (Brenneman and Eiden, 1986). Further studies have demonstrated that VIP plays a neuroprotective effects in various neurodegenerative diseases developed in animal models including Alzheimer's disease (Gozes et al., 1996), Parkinson's disease or encephalomyelitis (Gonzalez-Rey et al., 2005; Tan and Waschek, 2011). Some of these VIP neuroprotective actions were associated with glial cells possessing VPAC receptors. Clearly, VIP induced, on glial cells, a secretion of various trophic molecules having neuroprotective properties such as IL-1, IL-6, protease nexin-1, the chemokine RANTES and MIP (Dejda et al., 2005). Moreover, VIP inhibits the production of pro-inflammatory cytokines as TNFα and/or IL-1β secreted by activated microglia which is involved in neuroinflammation observed in Parkinson's disease or brain trauma models (Delgado et al., 2004). VIP also induces neuroprotective effect by increasing the secretion of activity-dependent neurotrophic factor (ADNF) and/or activitydependent neurotrophic protein (ADNP) (Gozes et al., 2003). These two protective proteins, which belong to the heat shock protein family, are able to prevent the neuronal death (Dejda et al., 2005) and represent one of the most potent neuroprotective agents secreted by astroglia in response to VIP. Recently, it was suggested that the VPAC2 receptor, which binds VIP and/or PACAP with the same affinity, could be a potential target for the development of anti-psychotic drugs. Effectively, the VPAC2 receptor gene has been found to be duplicated in schizophrenia (Vacic et al., 2011). Although VIP is able to cross the brain blood barrier (Dogrukol-Ak et al., 2004), no clinical trials in humans were developed to evaluate its neuroprotective role in brain diseases. However, some human clinical trials based on VIP vasoactive properties on cerebral arteries and hemodynamics have been performed (id: NCT00272896 and NCT00255320 http://www*.*clinicaltrials*.*gov) to evaluate its role in development of headache/migraine.

# **VPAC RECEPTORS, A REPRESENTATIVE MEMBERS OF CLASS B GPCR**

Biological responses induced by VIP are triggered by interaction with two receptors, VPAC1 and VPAC2, which are mainly coupled to the G-protein, Gs, resulting in the stimulation of cell adenylyl cyclase (Couvineau et al., 2010). Furthermore, some groups have reported the ability of VIP to increase calcium levels in different cells (Dickson and Finlayson, 2009). Moreover, VPAC1 receptor is able to interact with RAMP (Receptor Activity-Modifying Proteins) proteins, in particular RAMP2, inducing a significant enhancement of agonist-mediated inositol triphosphate production but do not modify the coupling to adenylate cyclase (Christopoulos et al., 2002). VPAC1 and VPAC2 receptors bind, with the same affinity, VIP and another neuropeptide named PACAP. It should be noted that VIP interacts also with the specific PACAP receptor (PAC1) but with a lower affinity (Couvineau and Laburthe, 2012a). Previous report indicate that VPAC1 receptor is able to homo-dimerize and hetero-dimerize with VPAC2 or secretin receptors (Harikumar et al., 2006). However, the relation between receptor oligomerization and the ability to VPAC1 receptor to interact with RAMPs remains unclear.

In the nineties secretin and VPAC receptors have been cloned (Ishihara et al., 1991, 1992; Lutz et al., 1993; Sreedharan et al., 1993; Couvineau et al., 1994) revealing a new G protein-coupled receptor (GPCR) sub-family termed class B GPCR. This GPCR sub-family shares with the other GPCR classes (A, C, D, E, F) the same general structural scheme characterized by the presence of seven-transmembrane helices denoted as TM I through TM VII which are interconnected by extracellular and intracellular loops (Fredrikson and Schiöth, 2006). The class B receptors family is composed of 15 members including receptors for VIP, PACAP, secretin, glucagon, glucagon-like peptide-1, glucagonlike peptide-2, GRF, GIP, and also include receptors for parathyroid hormone, calcitonin, calcitonin gene-related peptide, and corticotropin-releasing factor (CRF) (Couvineau and Laburthe, 2012b). Class B receptors display very low sequence homologies with others GPCRs (Laburthe et al., 2007) and share several specific characteristics: the presence of a large (*>*120 residues) and structured N-terminal ectodomain (N-ted) which is usually small in most class A GPCRs. The N-teds contain six highly conserved cysteine residues connected by three disulfide bridges, this sequence is the signature of class B GPCRs. The N-ted of class B receptor which represents the major binding site for its cognate natural peptide ligand, is characterized by; (1) the presence of a signal peptide probably involved in insertion of receptor in plasma membrane; (2) the absence of archetypical class A GPCR motifs such as E/D-R-Y or NP-xx(x)-Y; (3) a complex gene organization with many introns (Laburthe et al., 2002).

Currently, no data are available regarding the full-length structure of class B receptors as compare to class A receptors (Shoichet and Kobilka, 2012), although the structural properties of the class B GPCR N-ted, have recently been described, representing the first step toward better understanding of the binding receptor site at the atomic level. Recently, six N-ted structures, including those of the human PACAP receptor (PAC1), human PTH receptor (PTH1R), human GLP-1 receptor (GLP-1R), human GIP receptor (GIPR), and human type-1 and type-2 CRF receptor (CRFR1 and CRFR2) have been elucidated by Nuclear Magnetic Resonance (NMR) spectroscopy and X-ray crystallography in the presence of bound antagonist or agonists (Couvineau et al., 2010). These studies reveal the presence in the N-ted of a common core (**Figure 1**) formed by a Sushi domain (Parthier et al., 2009; Pal et al., 2012). This shared structure is composed of two anti-parallel β sheets (**Figure 1**) stabilized by (1) three disulfide bonds involving the typical six highly conserved cysteine residues (**Figure 1**); (2) a putative salt bridge involving acidic and basic residues, sandwiched between hydrophobic aromatic rings (**Figure 1**). The high conservation of the Sushi domain in the Nted of class B GPCRs supports the idea that this structure plays a crucial role for peptide recognition (Grace et al., 2004). A "twosite" model for the binding of native ligands to class B GPCRs has been postulated (Hoare, 2005). In short, the central and the C-terminal portions of the peptide ligand are captured by the Nted of the class B GPCRs. This step is essential for the peptide structuration, allowing the ligand N-terminus to interact with the transmembrane region of the receptor (Hoare, 2005).

As mentioned above VIP belongs to the secretin/VIP/PACAP family. The emergence of the class B GPCR has enlarged this peptide family (**Table 2**) by including parathyroid hormone (PTH), calcitonin, and CRF. All these natural ligands share some common properties: (1) they are all peptides with 27–44 amino acid residues; (2) they are synthesized and released by endocrine cells, neurons, and/or immune cells; (3) all these peptides exhibit a marked propensity to form α-helices; (4) all these peptides contain a N-Cap structure in the N-terminal part (Neumann et al.,

ionic and hydrophobic interactions (light gray sticks). All figures were obtained by using PyMOL software (http://www.pymol.org).

2008). The presence of this structural signature which includes a hydrophobic cluster between N-terminal hydrophobic residues and a hydrogen bond between two polar residues (**Figure 2**) have been recently confirmed (Watkins et al., 2012). All these peptides play an important role in physiological processes and strongly impact human physiopathology including chronic inflammation diseases, neurodegenerative disorders, schizophrenia, diabetes, osteoporosis, stress (Couvineau and Laburthe, 2012a).

Cloning of the human VPAC1 receptor (Couvineau et al., 1994) allowed its extensive studied for many years by site-directed mutagenesis and molecular chimerism (Laburthe et al., 2007) laying its molecular basis in terms of: (1) affinity (Couvineau et al., 1995); (2) specificity (Couvineau et al., 1996a); (3) cellular addressing (Couvineau et al., 2004); (4) desensitization (Marie et al., 2003); (5) association with RAMP proteins (Christopoulos et al., 2002); (6) adenylyl cyclase coupling (Couvineau et al., 2003). These studies have revealed that the receptor N-ted plays a crucial role in peptide agonist binding (Couvineau et al., 2010). In parallel, structure–function relationships analysis of VIP by a complete alanine scanning (Nicole et al., 2000) showed that the peptide has a diffuse pharmacophoric domain. In this study we have demonstrated that the N-terminal 1–5 segment plays a crucial role in receptor activation e.g., mainly adenylyl cyclase activation.

# **THE VPAC1 BINDING SITE, CONTRIBUTION OF PHOTOLABELING APPROACH**

The physical interaction sites between VIP and the VPAC1 receptor had remained elusive until the development of a photoaffinity labeling strategy, which allowed the demonstration that VIP side chains are physically in contact with the N-ted of VPAC1 (Couvineau et al., 2010). This strategy has two advantages over structural studies of purified recombinant receptors or receptor fragments: (1) the labeled ligand has an affinity for its receptor in the nanomolar range, which is similar to the high affinity measured under physiological conditions; and (2) the labeled ligand can interact with the glycosylated native receptor expressed in plasma membranes of eukaryotic cells. This is particularly important, given to the critical role of glycosylation in VPAC1 expression and function (Couvineau et al., 1996b). Addition of a benzophenone group (Bpa) to the VIP peptide has extensively contributed to the elucidation VIP biochemistry and of its receptor (Couvineau and Laburthe, 2012b). The use of photolabeling probes has clearly demonstrated that VIP residues in position 0, 6, 22, 24, or 28 were in physical contact with different amino acids in N-ted of the VPAC1 e.g., Gln135, Asp107, Gly116, Cys122, and Lys127 (**Figure 3**), respectively (Tan et al., 2003, 2004, 2006; Ceraudo et al., 2008, 2012). To dock VIP within the receptor N-ted, we determined the structure of VIP by NMR (**Figure 2**) and also developed a structural model of the VPAC1 receptor N-ted (Tan et al., 2006). Determination of VIP structure by NMR revealed that most of the 28 amino acids sequence has an α-helice structure (sequence 7–28) with the exception of the N-terminal 1–5 sequence, which has no defined structure in solution (**Figure 2**). In parallel, the structural model development of the VPAC1 receptor N-ted, by homology with the NMR structure of the CRF 2β receptor N-ted, allowed us to localize the VIP binding site in the N-ted. As expected, the structure contains two anti-parallel β sheets that are stabilized by three disulfide bonds between residues Cys<sup>50</sup> and Cys72, Cys63 and Cys105, and Cys86 and Cys122, and by a putative salt bridge involving Asp68- Arg103, sandwiched between the aromatic rings of Trp73 and Trp110 (**Figure 3**). The NMR structure of VIP has been docked in the VPAC1 receptor N-ted giving rise to a valid model in which, the N-ted C-terminal part, nicely accommodates the VIP molecule at least for the 6-28 sequence (**Figure 3**). This model has

structure. In magnified inset the N-capping motif is represented as (1) the hydrophobic interactions between side-chain groups of N' and N3 residues (dashed lines); (2) the hydrogen bond between side chain of N-cap residue and backbone atom of N3 residue. See ref. Neumann et al. (2008) for details.

obtained by using PyMOL software (http://www.pymol.org).

been submitted to molecular dynamic simulations over 14 ns in a water box and appears to be highly stable (Ceraudo et al., 2008).

Recently, using similar strategy, we have characterized the interaction site of the VPAC1 receptor-specific VIP antagonist, [Ac-His1, D-Phe2, K15, R16, L27] VIP(3-7)/GRF(8-27) or PG 97- 269 (Gourlet et al., 1997). This antagonist is a chimeric peptide between VIP (sequence 1–7) and GRF (sequence 8–27) having a D-phenylalanine residue in position 2. The use of Bpa0-PG97-269 affinity labeling probe revealed that the N-terminal part of antagonist physically interacted with Gly<sup>62</sup> residue of VPAC1 N-ted (Ceraudo et al., 2012). These observations clearly support that the N-terminal part of VIP (agonist) or PG97-269 (antagonist) were recognized by two different domains present in N-ted of VPAC1 receptor.

As mentioned above, the N-ted structure of different class B GPCRs has been obtained recently by X-ray crystallography or NMR spectroscopy (Parthier et al., 2009). These studies seem to indicate the existence of two different binding sites for ligands in class B receptor N-teds (Couvineau et al., 2010). Analysis of these structure and/or molecular models revealed that N-teds of GIPR, PTHR, CRF1R, CRF2R, and GLP-1R interact with ligands in regions encompassing the loop located between β1 and β2 sheets and the loop located between β3 and β4 sheets (Parthier et al., 2009). In contrast, the N-teds of PAC1R and VPAC1R bind peptides along β3 and β4 sheets of the sushi domain (Couvineau et al., 2010). However, a recent report based on the X-ray crystallography analysis of PAC1 receptor N-ted and the docking of PACAP indicates that PACAP could interact with its receptor as GIPR, PTHR, CRF1R, CRF2R, and GLP-1R (Kumar et al., 2011). The real significance of these differences were unclear but may be tentatively related to the following interpretations: (1) some structural determinations were carried-out in presence of ligands which have a low affinity (micromolar range) for the recombinant N-ted whereas in other studies ligand affinity was higher; it also could be hypothesized that low and high affinity binding occur at different sites in the N-ted structure; (2) the determination of interaction between N-teds and ligands was mainly obtained in the presence of antagonist but it some cases in the presence of an agonist; (3) moreover it could be hypothesized that agonists and antagonists bind to different domains in the N-teds. Finally, we cannot exclude the possibility that ligands can bind by two different ways to N-ted of class B GPCR.

#### **THE KEY ROLE OF THE FIRST TRANSMEMBRANE DOMAIN OF VPAC1 IN VIP BINDING**

VPAC1 domain interacting with the N-terminus of VIP (1–5) is still unknown. Up to now, no data are available regarding the fulllength structure of class B receptors. To circumvent this unavailability, a 3D-model of the receptor encompassing VIP/N-ted complex and the transmembrane core of the receptor (**Figure 4**) was developed (Ceraudo et al., 2012). The 3D-model of the transmembrane core was constructed by homology modeling based on the recent determination of the X-ray structure of the adenosine A2A receptor (Jaakola et al., 2008). The resulting 3D-model of VPAC1 revealed that the central and C-terminal residues of VIP are in contact with N-ted whereas the N-terminus of VIP lies in a pocket formed by the extracellular side of the first, second

and seventh transmembrane domains and the second extracellular loop of VPAC1 (**Figure 4A**). Based on distance calculation (*<*6Å) between residues of VPAC1 and VIP, substitutions by alanine of residues revealed that many residues are involved in the binding affinity of VIP to VPAC1. Three of them (H112, L131, and Q134) are present in the N-ted, and their substitution to alanine induced an affinity modification of about 100 times as compared to native receptor, indicating that these residues are probably involved in the interaction between the N-ted and the central and C-terminal parts of VIP (Ceraudo et al., 2012). Substitution to alanine of four other residues (K143, T144, T147, and L375) located in the extracellular side of TMI and VII, also induced a strong modification of receptor affinity for VIP (Ceraudo et al., 2012). Moreover site-directed mutagenesis experiments and reciprocal exchange between K143, T144, and T147 residues of VPAC1 and H<sup>1</sup> of VIP, shown that this interaction with (**Figure 4B**) the first histidine residue of VIP play a crucial role (Ceraudo et al., 2012). This step is important for the adenylyl cyclase activation (Couvineau et al., 1984). These observations were in good agreement with previous results indicating that D<sup>196</sup> present in the second extracellular loop (Du et al., 1997), K<sup>195</sup> and R<sup>188</sup> in TMII (Solano et al., 2001), N<sup>229</sup> in TMIII and Q<sup>380</sup> in TMVII of VPAC1 play an important role in VIP binding and probably could interact with the D<sup>3</sup> residue of VIP (Chugunov et al., 2010). Thus, these results along with our data clearly indicate that the N-terminus of VIP interacts with the extracellular side of the VPAC1 core.

#### **THE N-TED DETERMINES THE SPECIFICITY OF THE VPAC1 RECEPTOR**

As mentioned above, VPAC1 and VPAC2 receptors do not discriminate between the two neuropeptides, VIP and PACAP. Moreover, some others VIP related-peptides are able to bind to human VPAC1 receptor with low affinity, including peptide histidine methionineamide (PHM), secretin, helodermin and GRF (Laburthe et al., 2007). Potency order being VIP = PACAP*>*helodermin*>*PHM*>*GRF*>>*secretin. In this context, the development of specific ligands for VPAC1 and VPAC2 receptors represents a major goal. To develop a specific VPAC1 agonist, structure-function relationships analysis of VIP by a complete alanine scanning (Nicole et al., 2000) was used to rationally design the most potent and specific peptide for VPAC1 receptor currently available e.g., [Ala11*,*22*,*28]-VIP (Nicole et al., 2000). Indeed, this VIP derivative has an affinity 1000 times higher for the VPAC1 receptor, which is mainly involved in VIP anti-inflammatory action, than VPAC2 receptor (Delgado et al., 2004). As mentioned above a high selective antagonist of VPAC1 receptor (PG97-269) has been developed (Gourlet et al., 1997). Regarding VPAC2 receptor, the cyclic peptide analog of VIP [Ac-Glu8, OCH3-Tyr10, Lys12, Nle17, Ala19, Asp25, Leu26, -Lys27*,*<sup>28</sup> -VIP(cyclo 21–25)] or Ro 25–1392 is a potent and selective agonist (Xia et al., 1997). In our opinion, there is still no satisfactory VPAC2 receptor antagonist since PG 99–465, a VIP analog that antagonizes VIP action on VPAC2 receptor, which also has a significant agonist activity on human VPAC1 receptor (Moreno et al., 2000). Since recently, two non-peptide antagonists specific of VPAC1 (Harikrishnan et al., 2012) or VPAC2 (Chu et al., 2010) have been developed but they display a very low affinity for receptors.

The use of VIP photoaffinity probes associated to receptor mapping and Edman degradation demonstrated that VIP physically interacts with the N-ted of VPAC1 receptor (Couvineau et al., 2010). In order to get a high resolution structure of the VPAC1 receptor N-ted, the production of large quantities of recombinant N-ted protein in bacteria was performed (Couvineau et al., 2008). The 31–144 sequence of human VPAC1 receptor corresponding to the N-ted sequence in which the signal peptide (Couvineau et al., 2004) has been deleted was subcloned in front of 6xHis (His-tag) and behind the thioredoxin sequence containing a thrombin cleavage site (Couvineau et al., 2008). The construction of the thioredoxin-N-ted-6xHis (Trx-N-ted-6xHis) fusion protein was chosen in order to increase the solubility of recombinant proteins as previously described for production of recombinant N-ted of mouse CRF2β receptor (Grace et al., 2004). The soluble recombinant N-ted was purified onto Ni-NTA column and tested for its ability to bind VIP by using the influence of VIP binding on the intrinsic tryptophan fluorescence (ITF) of W67, W73, and W110 residues which are present in the N-ted sequence. Indeed, the presence of three tryptophan residues in the N-ted Sushi domain (Couvineau et al., 2008) represents a good fluorescent tag which can be used to measure the interaction between VIP and recombinant N-ted. Based to the ITF parameters, the estimation of dissociation constants revealed a Kd of 0.54 μM for VIP, 0.57μM for PACAP, and 1μM for PG96-269 (Couvineau et al., 2008). It should be noted that those Kd values were close to Kd values observed for others purified N-ted such as PAC1 receptor (Sun et al., 2007) and GIP receptor (Parthier et al., 2007). Moreover, the Kd of truncated VIP6–28 is very similar to Kd of native VIP (0.54 μM vs. 0.85μM) demonstrating that the 6–28 VIP sequence is sufficient to interact with a low affinity to **Table 3 | Binding of VIP related-peptides to recombinant N-ted.**


aThe ITF of W67, W73, and W110 residues from the purified N-ted was measured in 2 ml of HEPES buffer pH 7.5 containing 1 *µ*M purified N-ted, in absence or presence of increasing concentration of peptides. Dissociation constants were determined from titration curves using analytical procedure developed by Bechet et al. (Bechet et al., 1986).

bNot detectable.

recombinant N-ted (**Table 3**). In contrast, the deletion of large C-terminal (VIP1-12), central (VIP1-9/21-28) and N-terminal (VIP18-28) part of VIP abolishes totally the ability of truncated peptides to bind to recombinant N-ted (**Table 3**). These data clearly indicate that recombinant N-ted is able to recognize with a low affinity the central and C-terminal part of VIP molecule. Using the same approach, the ability of VPAC1 recombinant N-ted to discriminate VIP related-peptides was investigated (**Table 3**). As shown in **Table 2**, the estimation of Kd was of 0.54μM, 0.57μM, 2.54μM, 8μM, 10.12μM, and 16 μM for VIP, PACAP, helodermin, PHM, GRF, and secretin respectively, indicating that the order of potency is similar to native receptor i.e., VIP = PACAP*>*helodermin*>*PHM*>*GRF*>*secretin (Laburthe et al., 2007). Taken altogether these results reveal that: (1) the recombinant N-ted is able to bind with a low affinity and to discriminate VIP related-peptides suggesting that the VPAC1 N-ted contains residues involved in the VPAC1 specificity; (2) the first transmembrane domain of VPAC1 contains three residues (see above) which interact with the first residue of VIP and these three residues are probably involved in the high affinity and the activation of the receptor (**Figure 4B**).

#### **CONCLUSION**

The VPAC receptors, in particular VPAC1, are very promising targets for the development of therapeutic molecules in various pathologies including asthma, chronic inflammation diseases (Crohn's disease, rhumatoid arthritis, septic shock, multiple sclerosis. . . ) neurodegenerative disorders, schizophrenia. While new peptide derivatives specifically targeting VPAC receptor sub-types are now available, however, their very short half-life and the inconvenient related to their administration routes make them difficult to use in human therapy. The recent advance in the structural knowledge of the VPAC1 binding site should lead to the design of non-peptide receptor agonists and/or antagonists. The development of such molecules will represent an important overhang in the treatment of many human diseases.

#### **ACKNOWLEDGMENTS**

This work was supported by the Institut National de la Santé et de la Recherche Médicale, Centre National de la Recherche

#### **REFERENCES**


cyclase-activating peptide 1 receptor: photoaffinity, molecular modeling, and dynamics. *Mol. Endocrinol.* 22, 147–155.


Scientifique, Université Paris 7 and by a grant from the Association de Recherche sur la Polyarthrite. E. Ceraudo is supported by grant FDT20070910820 from the Fondation pour la Recherche Médicale and by an - Inter-university Poles of Attraction Program (PAI) from the Université Libre de Bruxelles (ULB), Belgium.

C., and Laburthe, M. (2008). Production and purification of large quantities of the functional N-terminal ectodomain of human VPAC1 receptor. *J. Mol. Neurosci.* 36, 249–253.


prevents experimental arthritis by downregulating both autoimmune and inflammatory components of the disease. *Nat. Med.* 7, 563–568.


therapeutic target for Parkinson's disease. *Expert Opin. Ther. Targets* 9, 923–929.


activation via helix induction? *Trends Biochem. Sci.* 34, 303–310.


signalling: new therapeutic strategies in multiple sclerosis. *ASN Neuro* 3, e00065.


pharmacology and drug development. *Drug Discov. Today* 17, 1006–1014.

Xia, M., Sreedharan, S. P., Bolin, D. R., Gaufo, G. O., and Goetzl, E. J. (1997). Novel cyclic peptide agonist of high potency and selectivity for the type II vasoactive intestinal peptide receptor. *J. Pharmacol. Exp. Ther.* 281, 629–633.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 August 2012; paper pending published: 26 September 2012; accepted: 26 October 2012; published online: 16 November 2012.*

*Citation: Couvineau A, Ceraudo E, Tan Y-V, Nicole P and Laburthe M (2012) The VPAC1 receptor: structure and function of a class B GPCR prototype. Front. Endocrin. 3:139. doi: 10.3389/ fendo.2012.00139*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Couvineau, Ceraudo, Tan, Nicole and Laburthe. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# PACAP inhibits β-cell mass expansion in a mouse model of type II diabetes: persistent suppressive effects on islet density

#### **Hiroaki Inoue<sup>1</sup>† , Norihito Shintani <sup>1</sup>† ,Yusuke Sakurai 1,2† , Shintaro Higashi 1,2, Atsuko Hayata-Takano1,3 , Akemichi Baba1,4 and Hitoshi Hashimoto1,3\***

<sup>1</sup> Laboratory of Molecular Neuropharmacology, Graduate School of Pharmaceutical Sciences, Osaka University, Suita, Osaka, Japan

<sup>2</sup> Japan Society for the Promotion of Science, Tokyo, Japan

<sup>3</sup> Department of Experimental Disease Model, Molecular Research Center for Children's Mental Development, United Graduate School of Child Development,

Osaka University, Kanazawa University, Hamamatsu University School of Medicine, Chiba University and University of Fukui, Suita, Osaka, Japan

<sup>4</sup> School of Pharmacy, Hyogo University of Health Sciences, Kobe, Hyogo, Japan

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Ricardo Borges, University of La Laguna, Spain M. Sue O'Dorisio, University of Iowa, USA

#### **\*Correspondence:**

Hitoshi Hashimoto, Laboratory of Molecular Neuropharmacology, Graduate School of Pharmaceutical Sciences, Osaka University, 1-6 Yamadaoka, Suita, Osaka 565-0871, Japan.

e-mail: hasimoto@phs.osaka-u.ac.jp

†Hiroaki Inoue, Norihito Shintani and Yusuke Sakurai have contributed equally to this work.

Pituitary adenylate cyclase-activating polypeptide (PACAP) is a potent insulinotropic Gprotein-coupled receptor ligand, for which morphoregulative roles in pancreatic islets have recently been suggested. Here, we evaluated the effects of pancreatic overexpression of PACAP on morphometric changes of islets in a severe type II diabetes model. Following cross-breeding of obese-diabetic model KKA<sup>y</sup> mice with mice overexpressing PACAP in their pancreatic β-cells, the resulting KKA<sup>y</sup> mice with or without PACAP transgene (PACAP/+:A<sup>y</sup> /+ or A<sup>y</sup> /+ mice) were fed with a high-fat diet up to the age of 11 months. Pancreatic sections from 5- to 11-month-old littermates were examined. Histomorphometric analyses revealed significant suppression of islet mass expansion in PACAP/+:A<sup>y</sup> /+ mice compared with A<sup>y</sup> /+ mice at 11 months, but no significant difference between PACAP/+ and +/+ (wild-type) mice, as previously reported. The suppressed islet mass in PACAP/+:A<sup>y</sup> /+ mice was due to a decrease in islet density but not islet size. In addition, the density of tiny islets (<0.001 mm<sup>2</sup> ) and of insulin-positive clusters in ductal structures were markedly decreased in PACAP/+:A<sup>y</sup> /+ mice compared with A<sup>y</sup> /+ mice at 5 months of age. In contrast, PACAP overexpression caused no significant effects on the level of aldehyde-fuchsin reagent staining (a measure of β-cell granulation) or the volume and localization of glucagon-positive cells in the pancreas.These results support previously reported inhibitory effects of PACAP on pancreatic islet mass expansion, and suggest it has persistent suppressive effects on pancreatic islet density which may be related with ductal cell-associated islet neogenesis in type II diabetes.

**Keywords:** β **cells, KKA<sup>y</sup> mice, high-fat diet, pituitary adenylate cyclase-activating polypeptide, islet neogenesis, type 2 diabetes**

#### **INTRODUCTION**

Pituitary adenylate cyclase-activating polypeptide (PACAP) is an extraordinarily potent insulinotropic peptide (Yada et al., 1994) belonging to the vasoactive intestinal polypeptide (VIP)/secretin/glucagon superfamily,which also includes glucagonlike peptide-1 (GLP-1) and glucose-dependent insulinotropic peptide (GIP) (Vaudry et al., 2009). PACAP and its receptors [PACAPspecific PAC1, andVIP-sharedVPAC1 andVPAC2 receptors,which belong to the class B (class II) G-protein-coupled receptor family] are highly expressed in neural elements, suggesting that it acts as a neurotransmitter and neuromodulator both in brain and peripheral tissues (Vaudry et al., 2009). There have been numerous studies on PACAP, in which its metabolic effects have been well documented (Ahrén, 2008; Vaudry et al., 2009). For example, PACAP has been shown to reduce food intake, increase glucose uptake in adipocytes by potentiating insulin action, stimulate the secretion of glucagon from the pancreas and norepinephrine from

the adrenal medulla, in addition to its insulinotropic activities in the pancreas. Based on these reports, several studies have evaluated the therapeutic potential of agonists or antagonists of the PACAP/VIP receptors (including PACAP and/or VIP themselves), and an inhibitor of dipeptidyl peptidase-4 (DPP-4), a common degradation enzyme for PACAP,VIP, GIP, and GLP-1,for the treatment of metabolic syndrome, including diabetes mellitus (Ahrén, 2008, 2009;Verspohl, 2009; Chapter et al., 2010). DPP-4 inhibitors are currently used as anti-diabetic agents (Holst, 2006; Verspohl, 2009).

As a potential cure for diabetes mellitus, a disease resulting from insulin insufficiency, recent studies have raised the possibility of the enhancement of endogenous β-cell mass, and transplantation of islets themselves, as a novel therapeutic strategy (Vaithilingam et al., 2008; Hanley, 2009; Verspohl, 2009; Dalle et al., 2011). With respect to this possibility, some studies have revealed that PACAP can stimulate β-cell proliferation and suppress the effects of harmful exogenous insults on the β-cell (Yamamoto et al., 2003; Nakata et al., 2010), although it has also been shown that PACAP has inhibitory effects on the pancreatic islet mass (see our review article, Sakurai et al., 2011). Recent studies using animal models in which PACAP/VIP signaling molecules have been knocked out have shown an increased mean islet area in PACAP-KO mice (Tomimoto et al.,2008) and altered islet architecture inVPAC1-KO mice (Fabricius et al., 2011), although no significant defects have been reported in VIP-KO mouse islets (Martin et al., 2010). These results suggest possible roles of PACAP/VIP signaling in islet morphoregulation, although it remains unknown how they regulate islet morphology, particularly in the case of type II diabetes.

Approximately a decade ago, we generated mice either lacking PACAP (Hashimoto et al., 2001) or overexpressing PACAP specifically in pancreatic β-cells (PACAP/+ mice;Yamamoto et al., 2003). To explore the long-term effects of PACAP in type II diabetes, we cross-bred PACAP/+ mice with agouti yellow KKA<sup>y</sup> mice, an obesity-induced type II diabetic model (Iwatsuka et al., 1970), and showed that pancreatic PACAP overexpression attenuated hyperinsulinemia and islet hyperplasia in KKA<sup>y</sup> mice, without any alteration of plasma glucose, glucose tolerance, or insulin tolerance (Tomimoto et al., 2004). Since the mild and delayedonset hyperglycemia in KKA<sup>y</sup> mice (Srinivasan and Ramarao, 2007) might mask the effects of PACAP in this model, we recently re-examined the phenotypic effects of PACAP overexpression in KKA<sup>y</sup> mice fed a high-fat diet (HFD) (Sakurai et al., 2012). The results showed that HFD feeding of KKA<sup>y</sup> mice induced severe, early-onset diabetes, but caused an unexpected recovery from hyperglycemia between 6 and 11 months of age, partly due to simultaneous (6–10 months of age) hyperinsulinemia. We also found that PACAP overexpression retained its previously observed suppressive effects, particularly those relating to hyperinsulinemia, in HFD-fed KKA<sup>y</sup> mice (Sakurai et al., 2012), however there has been no reported morphological information on the pancreatic islets of this model.

In the present study, we performed several morphometrical analyses of the islet phenotype of HFD-fed KKA<sup>y</sup> mice, including staining with hematoxylin-eosin (HE), aldehyde-fuchsin (AF), and anti-insulin and anti-glucagon antibodies. Here, we used PACAP/+ mice to evaluate the direct and local action of PACAP on the islet morphology, since PACAP can exert pleiotropic actions on adipocyte and adrenal medulla in addition to islets and secondary affect islets. The results obtained show that PACAP retains its inhibitory effects on pancreatic islet mass expansion, and provide a range of evidence suggesting that pancreatic PACAP affects ductal cell-associated islet neogenesis.

#### **MATERIALS AND METHODS**

#### **ANIMALS, DIETS, AND REAGENTS**

All animal care and handling procedures were approved by the Institutional Animal Care and Use Committee of Osaka University. Mice were housed in a temperature-, humidity-, and lightcontrolled room with a 12-h light/12-h dark cycle (lights on at 08:00 a.m.) and allowed free access to water and chow. Mating, genotyping, and feeding procedures were as previously described (Sakurai et al., 2012). In brief, F<sup>1</sup> mice (+/+, PACAP/+, A<sup>y</sup> /+, and A<sup>y</sup> /+:PACAP/+) were obtained by mating female transgenic

mice overexpressing PACAP in their pancreatic β cells (Yamamoto et al., 2003) with male KKA<sup>y</sup> mice (KK-A<sup>y</sup> /Ta mice, CLEA Japan Inc., Tokyo, Japan). F<sup>1</sup> males were individually housed after genotyping and weaning, and males with the A<sup>y</sup> allele (A<sup>y</sup> /+ and A y /+:PACAP/+) were fed with an HFD (HFD-32, CLEA Japan, Tokyo, Japan) from 4 weeks of age,while the other males continued on a normal diet (ND) (DC-8, CLEA Japan). These diets contain either 11.8% (DC-8) or 56.7% (HFD-32) of energy derived from fat.

#### **HISTOCHEMISTRY**

From each deeply anesthetized mouse, the pancreas was removed, weighed, and immediately fixed in 4% paraformaldehyde in phosphate buffered saline solution. Samples were embedded in paraffin, and 5µm sections were prepared for HE or AF staining, or immunohistochemical staining with anti-insulin (N1542, DAKO, Carpinteria, CA, USA) or anti-glucagon (N1541, DAKO) antibodies, in which signals were visualized using the diaminobenzidine method and were counterstained with cresyl violet. To investigate the architectural changes in the islets, two serial sections were prepared for immunostaining with anti-insulin and anti-glucagon antibodies, respectively.

#### **MORPHOMETRY**

Stained sections were photographed using a BIOREVO BZ-9000 microscope (Keyence, Japan), and morphometrical parameters were examined as follows. In HE-stained sections (*n* = 4–6 for 5 month-old mice, and *n* = 7–9 for 11-month-old mice), total islet number, size of each islet, and total pancreatic area were counted or measured, and analyzed as previously described (Tomimoto et al., 2004). Briefly, in each section from four F<sup>1</sup> groups, the mean islet size was determined by averaging the size of each islet, and the islet density by dividing total islet number by total pancreatic area (mm<sup>2</sup> ). Islet mass was calculated by multiplying the pancreas weight by the relative islet area per pancreas. The density per mm<sup>2</sup> of total pancreatic area, in addition to the frequency of the six groups of optical islet size (<0.003, 0.003–0.01, 0.01–0.03, 0.03– 0.1, 0.1–0.3, and >0.3 mm<sup>2</sup> ), was also determined. In AF-stained sections (*n* = 5 for 5-month-old mice, and *n* = 3 for 11-monthold mice), blinded observers evaluated the AF reagent staining in each islet of four F<sup>1</sup> groups. In sections stained with anti-insulin or anti-glucagon antibodies (*n* = 4 for each group), the positive area and the size of each insulin-positive cluster were measured using ImageJ software (version 1.30, http://rsb.info.nih.gov/ij). The number of insulin-positive clusters, and of glucagon cellinfiltrated islets (which exhibit glucagon-positive cells inside of islets as indicated by arrows in **Figure 2B**), were also counted and analyzed.

#### **PLASMA INSULIN LEVEL IN AN INTRAPERITONEAL GLUCOSE TOLERANCE TEST**

Two milligrams per kilogram glucose was intraperitoneally injected to each mouse after a 14-h food deprivation as described (Sakurai et al., 2012). Plasma samples were prepared just before (time 0) and at 10, 30, 60, 90, and 120 min after glucose load, and the insulin level in the samples was examined by a mouse insulin enzyme-linked immunosorbent assay (Morinaga, Tokyo, Japan).

#### **DATA ANALYSIS AND STATISTICS**

All data are expressed as mean ± standard error of the mean. Statistical evaluation was carried out using KaleidaGraph software (HULINKS, Tokyo, Japan). The statistical significance of differences was assessed using two-way ANOVA followed by the Tukey– Kramer test, χ 2 test, and Student's un-paired *t*-test. Differences with *P* < 0.05 were considered significant.

#### **RESULTS**

#### **EFFECTS ON SIZE AND DENSITY OF ISLETS**

In KKA<sup>y</sup> mice fed with a ND, we previously showed that pancreatic PACAP overexpression markedly suppresses the increase of mean islet size and of islet density (Tomimoto et al., 2004). Contrary to these observations, HE-stained pancreatic sections from 11 month-old HFD-fed KKA<sup>y</sup> mice showed that not only A<sup>y</sup> /+ mice, but also A<sup>y</sup> /+:PACAP/+ mice, exhibited a clear islet enlargement (**Figure 1A**). Quantitative analyses indicated that the mean size was significantly increased in both groups compared with their respective controls, but no significant difference was observed between A y /+ and A<sup>y</sup> /+:PACAP/+ mice at either 5 or 11 months of age (**Figure 1B**, upper graphs). In contrast, a significant increase in inlet density was observed in A<sup>y</sup> /+ but not in A<sup>y</sup> /+:PACAP/+ mice, and the density in A<sup>y</sup> /+:PACAP/+ mice was significantly suppressed compared with A<sup>y</sup> /+ mice at both 5 and 11 months of age (**Figure 1B**, middle graphs). With respect to islet mass, there was a large increase in A<sup>y</sup> /+ mice (e.g., it was 22-fold higher than +/+ mice at 11 months of age; **Figure 1B**, lower graphs). In a model showing such remarkable islet mass expansion, pancreatic

PACAP overexpression caused a 30% reduction in islet mass at 11 months of age (**Figure 1B**, lower right graph). Taken together with our previous results in ND-fed KKA<sup>y</sup> mice (Tomimoto et al., 2004), these results indicate that the inhibitory effects of PACAP on islet density are retained, but those influencing islet size are lost, in HFD-fed KKA<sup>y</sup> mice. Note that no significant differences were observed between +/+ and PACAP/+ mice for any of the parameters measured, as described (Tomimoto et al., 2004).

#### **EFFECTS ON ISLETS OF VARIOUS SIZES**

We next examined which size of islets was decreased by PACAP in HFD-fed KKA<sup>y</sup> mice (**Figure 1C**). In both 5- and 11-monthold groups, the density of larger (>0.03 mm<sup>2</sup> ) islets was preferentially increased in A<sup>y</sup> /+ mice compared with +/+ mice, and this increase was significantly inhibited in A<sup>y</sup> /+:PACAP/+ mice. Since the density of smaller (<0.003 mm<sup>2</sup> ) islets was also decreased inA<sup>y</sup> /+:PACAP/+ mice at 5 months of age,we also compared the density of islets sized <0.001 mm<sup>2</sup> between A<sup>y</sup> /+ and A y /+:PACAP/+ mice at that age (**Table 1**). The result showed an 80% decrease in the density of these tiny islets in A<sup>y</sup> /+:PACAP/+ mice, supporting the observation that a marked decrease in the smaller islets has indeed occurred in A<sup>y</sup> /+:PACAP/+ mice. In contrast, size distribution analysis indicated that the distribution in A<sup>y</sup> /+:PACAP/+ mice was almost the same as A<sup>y</sup> /+ mice at both ages (5 months old, χ <sup>2</sup> = 2.33, *P* = 0.802; 11 months old, χ <sup>2</sup> = 1.33, *P* = 0.932), whereas they were clearly shifted right compared to +/+ mice (for example, in A<sup>y</sup> /+ versus +/+ mice; 5 months old, χ <sup>2</sup> = 19.0, *P* < 0.01; 11 months old,

**FIGURE 1 | Quantitative islet histomorphometry of hematoxylineosin-stained pancreatic sections of 5- and 11-month-old F<sup>1</sup> mice. (A)** Representative images of pancreatic sections of +/+ (wild-type), PACAP/+, A<sup>y</sup> /+, and PACAP/+:A<sup>y</sup> /+ mice stained with hematoxylin-and-eosin (HE). Arrow heads indicate islets. Scale bar, 500µm. **(B,C)** Morphometric data of the HE-stained pancreatic sections. F<sup>1</sup> male littermates of 5-month-old

(n = 4–6) and 11-month-old (n = 7–9) mice were examined. **(B)** Mean islet size, density of islets (the number of islets per square millimeter of total pancreatic area), and the calculated total islet mass in F<sup>1</sup> mice. **(C)** Density of islets of the indicated size in F<sup>1</sup> mice. Data are expressed as the mean + SEM. \*P < 0.05, \*\*P < 0.01, versus representative control with or without PACAP, #P < 0.05, versus Ay/+ mice, one-way ANOVA followed by the Tukey–Kramer test.


#### **Table 1 | Parameters of islets in A<sup>y</sup> /**+ **and A<sup>y</sup> /**+**:PACAP/**+ **mice fed a high-fat diet.**

Pancreatic sections prepared from 5-month-old A<sup>y</sup> /+ and A<sup>y</sup> /+:PACAP/+ mice were stained with hematoxylin-and-eosin (HE, n = 5 for each group), anti-insulin antibodies, or anti-glucagon antibodies (n = 4 for each group), and were subjected to morphometric analyses. Statistical analyses were performed using an un-paired Student's t-test. †Values regarded as statistically significant. \*Tiny islets are those of area <0.001 mm<sup>2</sup> . # Infiltrated islets are those that have glucagon-positive cells inside, in addition to at their periphery.

χ <sup>2</sup> = 35.0, *P* < 0.0001). Collectively, these results suggest that PACAP overexpression does not affect the overall size distribution of islets in HFD-KKA<sup>y</sup> mice, but has clear inhibitory effects on islet density, particularly in the smaller islets (<0.003 mm<sup>2</sup> ) at an earlier age (5 months old).

#### **OBSERVATION OF AF-STAINED SECTIONS**

To explore the possible changes of islets in A<sup>y</sup> /+ and A y /+:PACAP/+ mice, we used AF staining to examine the degranulation of β-cells, a well known phenotypic change in obese mouse islets, including those of KKA<sup>y</sup> mice (Iwatsuka et al., 1970). As shown in **Figure 2A**, we observed a clear disappearance of AF staining in HFD-fed A<sup>y</sup> /+ mice, but not in +/+ mice, at 5 months of age. However, we unexpectedly observed that AF staining was obvious in islets of 11-month-old A<sup>y</sup> /+ mice,implying that a compensatory reaction had occurred. When comparing between A<sup>y</sup> /+ and A<sup>y</sup> /+:PACAP/+ mice, all islets in all samples at 5 months of age (*n* = 5 for each genotypes) lacked AF staining, whereas those at 11 months of age (*n* = 3 for each genotype) showed definite AF staining. These results suggest that an unexpected recovery in β-cell degranulation could be occurring in HFD-fed KKA<sup>y</sup> mice, and that pancreatic PACAP overexpression does not affect either degranulation, or compensatory re-granulation, of β-cells in this model.

#### **INTRAPERITONEAL GLUCOSE TOLERANCE TEST**

In line with the unexpected recovery in β-cell degranulation, we previously showed that the glucose disposal in HFD-fed KKA<sup>y</sup> mice is unexpectedly enhanced at 11 months of age compared with their age-matched wild-types (Sakurai et al., 2012). Therefore, we here checked the glucose-induced elevation of plasma insulin level in the 11-month-old F<sup>1</sup> groups (**Figure 3**). The results indicated that the insulin level in A<sup>y</sup> /+ but not A y /+:PACAP/+ mice is significantly elevated compared with +/+ and PACAP/+ mice even under fasted state. On the other hand, the first-phase insulin response (value dividing the insulin level

at 10 min by that at 0 min) was attenuated in both of A<sup>y</sup> /+ and A y /+:PACAP/+ compared with wild-type mice, but no significant difference was observed between two groups (The first-phase insulin response: +/+, 2.73 ± 0.46; PACAP/+, 2.72 ± 0.35; A<sup>y</sup> /+, 1.16 ± 0.11; A<sup>y</sup> /+:PACAP/+, 1.22 ± 0.26). These data suggest that the impaired first-phase insulin response seems to be persistently observed in both A<sup>y</sup> /+ and A<sup>y</sup> /+:PACAP/+ mice, and that pancreatic PACAP overexpression showed little effects on the glucose-induced insulin release at least at 11 months of age.

#### **EFFECTS ON INSULIN OR GLUCAGON CELLS**

The above data indicate that phenotypic differences between the 5-month-old groups were more obvious and are possibly causative for the changes in 11-month-old groups. Thus, we next performed detailed analyses on the islets of 5-month-old A y /+ and PACAP/+:A<sup>y</sup> /+ mice using two adjacent sections stained with anti-insulin and anti-glucagon antibodies, respectively (**Figure 2B**). In insulin-stained section, well-stained islets were commonly observed in both A<sup>y</sup> /+ and PACAP/+:A<sup>y</sup> /+ mice. Compared with +/+ mice, not only enlarged islets but also small insulin-positive clusters, some of which were located in the ductal structures (as indicated by arrowheads), were often observed in these two groups. Glucagon-stained sections revealed an apparent reduction of staining in both A<sup>y</sup> /+ and PACAP/+:A<sup>y</sup> /+ mice compared with +/+ mice. In addition, some islets showed an altered localization of glucagon-positive cells (as indicated by arrows); namely, the cells localized not only peripherally but also in the central area of the islets. In age-matched +/+ mice, this was rarely observed and glucagon-positive cells were predominantly localized at the periphery of the islets. Based on these observations, a range of parameters were examined (**Table 1**). In insulin-stained sections, quantitative analysis revealed a tendency toward a decrease, but no significant change, in the percent of insulin-positive area, the density, and the mean size of insulinpositive clusters between A<sup>y</sup> /+ and A<sup>y</sup> /+:PACAP/+ mice. However, the density of ductal insulin-positive clusters was significantly

**anti-insulin or anti-glucagon-stained pancreatic sections of F<sup>1</sup> mice. (A)** Pancreatic sections from 5 to 11 months old +/+ (wild-type), A<sup>y</sup> /+, and PACAP/+:A<sup>y</sup> /+ mice were stained with aldehyde-fuchsin (AF) reagent to detect granulated β-cells. Note that a lack of AF staining (degranulation of β-cells) is observed in islets of A<sup>y</sup> /+ and PACAP/+:A<sup>y</sup> /+ mice but not in +/+ mice at 5 months old, but clear staining is generally seen in

5-month-old +/+, A<sup>y</sup> /+, and PACAP/+:A<sup>y</sup> /+ mice were stained with anti-insulin and anti-glucagon antibodies, respectively. Arrowheads indicate ductal insulin-positive clusters, whereas arrows denote glucagon cell-infiltrated islets in which glucagon-positive cells reside inside in addition to at the periphery of the islet. Note both types of signals were rarely observed in +/+ mice. Scale bars, 500µm.

decreased (by 58%) in A<sup>y</sup> /+:PACAP/+ mice compared with A<sup>y</sup> /+ mice, whereas the total frequency of ductal clusters was not significantly different. In glucagon-stained sections, no significant difference was observed in the percent of glucagon-positive area and of glucagon cell-infiltrated islets.

# **DISCUSSION**

We examined the morphoregulative roles of PACAP on pancreatic islets using a severe and early-onset type II diabetes model (HFD-fed KKA<sup>y</sup> mice; Sakurai et al., 2012) over the course of approximately 1 year. In A<sup>y</sup> /+:PACAP/+ mice, the significant suppression of islet mass expansion at 11 but not at 5 months of age (**Figure 1B**) fits with previous data showing attenuation of enhanced hyperinsulinemia during 6–10 months of age (Sakurai et al., 2012). In contrast, between the age-matched A<sup>y</sup> /+ and A<sup>y</sup> /+:PACAP/+ mice, other morphometric analyses revealed no significant difference in β-cell granulation, insulin-, and the glucagon-positive area per pancreatic section, or the distribution of glucagon-positive cells in islets (**Figure 2**; **Table 1**). In addition, the data in intraperitoneal glucose tolerance test (ipGTT) also suggest that no significant effects of PACAP on the glucose-induced insulin release. These results therefore suggest that PACAP inhibits

morphological rather than functional changes in islets and thereby suppresses the increase of hyperinsulinemia in A<sup>y</sup> /+:PACAP/+ mice. In addition, taken together with results in ND-fed KKA<sup>y</sup> mice (Tomimoto et al., 2004), the present study provides additional evidence showing morphoregulative roles of PACAP on pancreatic islets, and establishes the inhibitory effects of PACAP on compensatory islet mass expansion in type II diabetes.

The present study revealed sustained suppression (at least between 5 and 11 months of age) of islet density, but not of mean islet size, in A<sup>y</sup> /+:PACAP/+ mice compared with A<sup>y</sup> /+ mice (**Figure 1B**). Taken together with our previous results (Tomimoto et al., 2004), these data indicate that PACAP universally and persistently inhibits the increase of islet density from the early postnatal period in type II diabetes models. It is unlikely that HFD and A<sup>y</sup> allele-boosted islet enlargement masked PACAP's inhibitory effects on mean islet size, because the mean islet size in 5-month-old A<sup>y</sup> /+ mice (0.35 ± 0.06 mm<sup>2</sup> ) was less than half of that observed in ND-fed KKA<sup>y</sup> mice (0.80 ± 0.08 mm<sup>2</sup> ; Tomimoto et al., 2004). Size-fractionated analyses indicated that the increase of larger-sized islets in A<sup>y</sup> /+ mice is preferentially suppressed by PACAP at both 5 and 11 months, but also revealed a large decrease in the smaller islets in A<sup>y</sup> /+:PACAP/+ mice at 5 months of age

(**Figure 1C**; **Table 1**). These data are consistent with PACAP's effects at earlier ages, and suggest PACAP-induced reduction of small-sized islets as a core phenotype of A<sup>y</sup> /+:PACAP/+ mice, which eventually contributes to the decrease of larger islets and of islet mass in these mice compared with A<sup>y</sup> /+ mice.

In A<sup>y</sup> /+:PACAP/+ mice fed with ND, we previously showed that the pancreatic PACAP content and the plasma insulin level are increased by 3.5- and 2.8-fold of PACAP/+ mice (Tomimoto et al., 2004), suggesting that A<sup>y</sup> allele-related increase in the plasma insulin boosts the PACAP expression in PACAP/+ mice by activating the human insulin promoter cassette of the transgene construct (Yamamoto et al., 2003). Although we did not checked the PACAP content in the present study, the pancreatic content in HFD-fed A y /+:PACAP/+ mice could be estimated as more than 10-fold compared with +/+ mice, because their plasma insulin level is 70 to 100-fold compared with+/+ mice (Sakurai et al.,2012) whereas ND-fed A<sup>y</sup> /+:PACAP/+ mice showed 10-fold increase in PACAP with 2.8-fold increase in plasma insulin compared with +/+ mice (Tomimoto et al., 2004). Since our previous study showed that the pancreatic PACAP content is 48 ± 12 pg/mg (Tomimoto et al., 2004), it could be translated that approximately 10 pM PACAP exists in pancreas or 1 nM PACAP locally exists around islets

#### **REFERENCES**


targets for treatment of type 2 diabetes. *Nat. Rev. Drug. Discov.* 8, 369–385.

Bonner-Weir, S., Toschi, E., Inada, A., Reitz, P., Fonseca, S. Y., Aye, T., et al. (2004). The pancreatic ductal epithelium serves as a potential pool of progenitor cells. *Pediatr. Diabetes* 5, 16–22.

(because PACAP is known to be produced only nearby islets that corresponds to 1% volume of pancreas). Therefore, if the PACAP content was increased by more than 10-fold in HFD-fed A y /+:PACAP/+ mice, it could be activate various intracellular signaling pathway, because higher concentration of PACAP is known to stimulate Gq-linked pathway in addition to Gs-linked pathway via binding to PAC1, VPAC1, and VPAC2 receptors (Vaudry et al., 2009). Thus, it should be noted that further studies are required to determine the molecular mechanism underlying the phenotypic changes observed.

Based on current knowledge, the number of small islets (β-cell clusters) is regulated via diverse processes including differentiation from ductal precursor cells, trans-differentiation from non-β-cells, fission or fusion between islets, and replication, hypertrophy, and apoptosis of the β-cells themselves (Bonner-Weir et al., 2004;Brennand and Melton, 2009). Although further studies on how these processes may be affected by PACAP should be performed, the significant decrease in the density of ductal insulin-positive clusters in the context of a normal frequency of these clusters in the duct (**Table 1**) strongly suggests that PACAP inhibits ductal precursor-related islet neogenesis in A<sup>y</sup> /+:PACAP/+ mice. If this is the case, it is likely that differentiation from precursor cells, and/or the cell-fate regulation of newly produced β-cells from these precursors, are the possible causative mechanisms explaining PACAP-induced reduction of small-sized islets in A<sup>y</sup> /+:PACAP/+ mice.

In conclusion, the present study provides additional evidence for the inhibitory effects of PACAP on pancreatic β-cell mass expansion, and suggested its possible effects on ductal precursorrelated islet genesis. An increased number of ductal insulinpositive cells has been reported in pancreatic biopsy samples from human type II diabetic patients, in which the increased β-cell mass was suggested to be due to increased islet neogenesis but not to islet enlargement (Butler et al., 2003). Since these results imply that the regulation of islet mass depends on species or disease state-dependent differences, future studies on (postnatal) islet neogenesis are important for a deeper understanding of islet homeostasis in type II diabetes, because a number of unrelated and sometimes contradictory results appear to have accumulated in this research field.

#### **ACKNOWLEDGMENTS**

This work was supported in part by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS); and the Funding Program for Next Generation World-Leading Researchers (Hitoshi Hashimoto). Yusuke Sakurai and Shintaro Higashi are JSPS research fellows and are supported by Research Fellowships for Young Scientists from JSPS.


with type 2 diabetes. *Diabetes* 52, 102–110.

Chapter, M. C.,White, C. M., DeRidder, A., Chadwick, W., Martin, B., and Maudsley, S. (2010). Chemical modification of class II G proteincoupled receptor ligands:frontiers in the development of peptide analogs as neuroendocrine pharmacological

therapies. *Pharmacol. Ther.* 125, 39–54.


mice. *J. Pharmacol. Exp. Ther.* 309, 796–803.


Overexpression of PACAP in transgenic mouse pancreatic betacells enhances insulin secretion and ameliorates streptozotocin-induced diabetes. *Diabetes* 52, 1155–1162.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 26 September 2012; paper pending published: 16 October 2012; accepted: 22 February 2013; published online: 11 March 2013.*

*Citation: Inoue H, Shintani N, Sakurai Y, Higashi S, Hayata-Takano A, Baba A and Hashimoto H (2013) PACAP inhibits* β*-cell mass expansion in a mouse model of type II diabetes: persistent suppressive effects on islet density. Front. Endocrinol. 4:27. doi: 10.3389/fendo.2013.00027*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Inoue, Shintani, Sakurai, Higashi, Hayata-Takano, Baba and Hashimoto. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *Ingrid Langer\**

Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Faculté de Médecine, Université Libre de Bruxelles, Brussels, Belgium

#### *Edited by:*

Hubert Vaudry, University of Rouen, France

#### *Reviewed by:*

Nicholas D. Holliday, University of Nottingham, UK Alain Couvineau, Institut National de la Santé et de la Recherche Médicale, France

Vasoactive intestinal peptide (VIP) plays diverse and important role in human physiology and physiopathology and their receptors constitute potential targets for the treatment of several diseases such as neurodegenerative disorder, asthma, diabetes, and inflammatory diseases. This article reviews the current knowledge regarding the two VIP receptors, VPAC1 and VPAC2, with respect to mechanisms involved in receptor activation, G protein coupling, signaling, regulation, and oligomerization.

**Keywords: VIP, VPAC1, VPAC2, mutagenesis, activation, signaling, regulation, oligomerization**

#### *\*Correspondence:*

Ingrid Langer, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Faculté de Médecine, Université Libre de Bruxelles, 808 route de Lennik CP602, B-1070 Brussels, Belgium. e-mail: ilanger@ulb.ac.be

# **VASOACTIVE INTESTINAL PEPTIDE**

Vasoactive intestinal peptide (VIP) is a 28 amino acids peptide isolated from porcine intestine (Said and Mutt, 1970) and that belongs to a family of structurally related peptide hormones that includes secretin, pituitary adenylate cyclase-activating polypeptide (PACAP), glucagon, glucagon-like peptides (GLP), gastric inhibitory peptide (GIP), and growth hormone-releasing hormone (GHRH). It has initially been shown that VIP has a diverse range of effects such as vasodilation, relaxation of smooth muscle in intestinal and pulmonary tissues, and stimulation of electrolyte secretion in the gut (Dickson and Finlayson, 2009; Harmar et al., 2012). As a consequence, VIP was first classified as a gut hormone. Later studies demonstrated that VIP has a more expanded distribution including peripheral (PNS) and central (CNS) nervous system (Larsson et al., 1976a,b) as well as cells and tissues of the immune system (Delgado et al., 2004), suggesting that VIP may also behave as a neuroendocrine hormone, a neurotransmitter and a cytokine-like peptide. Many studies supported this hypothesis and it is now well accepted that VIP plays important role in the CNS such as control of circadian rhythms, anxiety, response to stress, schizophrenia, learning, and memory (King et al., 2003; Dickson and Finlayson, 2009; Moody et al., 2011; Harmar et al., 2012). In the periphery, it is involved in the control of insulin secretionfrom the pancreas and the release of catecholaminesfrom the adrenal medulla (Dickson and Finlayson, 2009; Moody et al., 2011; Harmar et al., 2012). VIP also acts as a co-transmitter of non-adrenergic non-cholinergic relaxation of vascular and nonvascular smooth muscles (Said and Rattan, 2004). Finally, in the immune system VIP has potent effects on T cell differentiation and migration and modulates cytokine production by T helper cells (Delgado et al., 2004).

# **VPAC RECEPTORS**

"fendo-03-00129" — 2012/10/28 — 19:05 — page 1 — #1

The biological effects of VIP are mediated by two receptors named VPAC1 and VPAC2 that belong to the family B of G proteincoupled receptors (GPCRs) which also includes PAC1-, secretin-, glucagon-, GLP 1 and 2-, calcitonin-, GIP-, corticotropin-releasing factor (CRF) 1 and 2-, and parathyroid hormone receptors. Besides VIP, VPAC1 and VPAC2 receptors also bind with the same high affinity PACAP. The VPAC1 receptor was initially cloned from rat lung (Sreedharan et al., 1993) and the VPAC2 receptor from rat pituitary (Lutz et al., 1993). Like VIP, VPAC receptors are widely distributed throughout the body. In the CNS,VPAC1 receptors are abundantly localized in piriform cortex, cerebral cortex, suprachiasmatic nucleus, hippocampus, and pineal gland (Usdin et al., 1994) while VPAC2 receptors are mainly found in cerebral cortex, suprachiasmatic nucleus, thalamus, hypothalamus, and amygdala (Sheward et al., 1995). Although both receptors may be co-expressed in the same areas of the CNS, studies suggest that the two receptor subtypes have complementary distribution. In peripheral tissues, VPAC1 receptors have been found in breast, kidney, liver, lung, prostate, spleen, and mucosa of the gastrointestinal tract (Reubi, 2000). Similarly, VPAC2 receptors have a large distribution and have been localized in adrenal medulla, blood vessels, lung, pancreatic acinar cells, smooth muscle, and thyroid follicular cells (Harmar et al., 2004). In the immune system, VPAC1 receptors are constitutively expressed in T cells, monocytes, and macrophages while expression of VPAC2 receptors is induced upon activation of T cells and macrophages (Delgado et al., 2004).

Like all members of the GPCR-B family, VPAC receptors are heptahelical membrane proteins and are characterized by the presence of a large N-terminal extracellular domain. The recent solving

of the NMR or X-ray structure of the N-terminus of several family B receptors [CRF, PTH, PAC1, GIP, GLP-1, CLR/RAMP1, and VPAC2(PDB ID: 2X57)] clarified their role in ligand binding. The structure comprises a crucial sushi domain characterized by two antiparallel β sheets and stabilized by three disulfide bonds and a salt bridge sandwiched between aromatic rings of two tryptophan residues (Harmar et al., 2012). The data support the two-site model for peptide binding to family B GPCR, in which the Nterminal domain of the receptor is the principal binding site for the central and the C-terminal regions of the natural ligand and ensures correct ligand positioning, whereas binding of residues 1–6 of the ligand to the extracellular loops and transmembrane helices drives the receptor activation (Hoare, 2005). More recently, it has also been proposed that a helix N-capping motif, identified in the N-terminus of all GPCR-B family ligands and stabilizing their helical conformation, was probably formed upon receptor binding and could also constitute a key element in receptor activation (Neumann et al., 2008). Although the experimental structure of the VPAC1 has not been solved, a three-dimensional model obtained by homology modeling associated with photoaffinity experiments supports that VPAC1 shares the same features (Couvineau and Laburthe, 2012).

### **SIGNALING PATHWAYS ACTIVATED BY VPAC RECEPTORS**

Like all GPCRs, upon agonist binding VPAC receptors undergo physical/conformational changes that allow interaction of cytoplasmic domains with heterotrimeric G proteins and promote exchange of GDP for GTP on the Gα subunit. This initiates the dissociation of GTP-bound Gα subunit from Gβγ dimer and activation of downstream effector pathways. Signals are terminated following hydrolysis of Gα-bound GTP and reformation of inactive trimeric complex (Pierce et al., 2002). VPAC receptors are preferentially coupled to Gαs leading to activation of adenylate cyclase and subsequent cAMP production. Accumulation of intracellular cAMP also leads to activation of protein kinase A (PKA) that may activate the ERK signaling pathway to promote proliferation (Pechon-Vallee et al., 2000) and neuroendocrine cell differentiation (Gutierrez-Canas et al., 2005) as seen in pituitary cells and prostate cancer cell line, respectively. VIP-induced PKA activation is also responsible for most of the anti-inflammatory activity of VIP by regulating several signaling pathways and transcription factors thus increasing anti-inflammatory cytokines and reducing pro-inflammatory cytokines production (Gonzalez-Rey et al., 2007). Several studies also reported that VPAC receptors were able to activate phospholipase C (PLC) pathway and stimulate calcium levels, either in cells endogenously expressing VPAC receptors or in transfected cell lines. However, the precise mechanisms which contribute to VIP-induced [Ca2+]i increase remain unclear due to divergent results. Indeed, in transfected cell lines some studies observed that activation of PLC was partly sensitive to pertussis toxin (PTx; MacKenzie et al., 1996; Langer et al., 2002), thus involving both Gαi and Gαq coupling, while others did not observe coupling to Gαi (Sreedharan et al., 1994). Moreover, PTx sensitive mechanisms leading to PLC activation seems different for VPAC1 and VPAC2 receptors. Cross-linking studies demonstrated that physical interaction between VPAC1 and Gαi and VPAC1-induced [Ca2+]i increase are not affected by chelating of extracellular calcium (Langer et al., 2002), while PTx sensitive activation of PLC by VPAC2 relies on the availability of free Gβγ and on Ca2<sup>+</sup> entry through receptor-operated Ca2<sup>+</sup> channels (MacKenzie et al., 2001). Finally, it was also observed that both VPAC1 and VPAC2 are able to couple to Gα16, a G protein of the Gαq family, that enables the coupling of a wide variety of receptors to PLC and whose expression is restricted to hematopoietic cells (with the exception of the mature B cells; Langer et al., 2001). These findings indicate that the specific G protein/second messenger complement of the cell line/type being examined could alter the transduction pathways/pharmacology observed forVPAC receptors (**Figure 1**).

Additional coupling events that are not G protein-mediated may also elicit auxiliary signals. Both VPAC1 and VPAC2 are able to activate phospholipase D (PLD). PLD responses induced by VPAC2 are not affected by PLC inhibitors, PTx or PKA inhibitors but are sensitive to brefeldin A an inhibitor of ADP-ribosylation factor (ARF) known to act as a direct activator of PLD (McCulloch et al., 2000). In pancreatic β cells, VIP induced sustained stimulation of insulin secretion is mediated by phosphatidylinositol 3 kinase activation by VPAC2 receptors (Dickinson et al., 1999). In the rat pineal gland,VIP stimulates cGMP formation through activation of nitric oxide synthase, NO production and activation of cytosolic guanylate cyclase (Spessert, 1993).

# **MOLECULAR MECHANISMS INVOLVED IN VPAC RECEPTORS ACTIVATION**

The recent solving of the X-ray crystal structures of several GPCR-A family members in complex with agonist, antagonist, and G proteins provides clues to the transmembrane helix (TM) rearrangements that result from agonist binding and subsequent receptor activation. These include the disruption of an ionic interaction involving the E/DRY motif located at the cytoplasmic face of TM3 and maintaining the receptor preferentially in a ground inactive conformation in absence of agonist (ionic lock), a "rotamer toggle switch" (modulation of the helix conformation around a proline kink) in TM6 causing key sequences to be exposed to cytoplasmic binding partners and a conformational change of Y residue of the NPXXY motif located in TM7 stabilizing the active conformation (Rosenbaum et al., 2009, 2011; Rasmussen et al., 2011). Surprisingly, the predicted interaction between E/DRY motif located in TM3 and a glutamate residue located in TM6 ("ionic lock") was solely observed in rhodopsin inactive state crystals but not in other GPCR crystals available to date. Instead, as seen for the β2-adrenergic receptor (β2-AR), Arg of the E/DRY motif interacts primarily with the adjacent Asp residue in crystals of β2-AR stabilized with the inverse agonist carazolol (inactive conformation; Rasmussen et al., 2007; Rosenbaum et al., 2007) while in crystals of β2-AR/Gs protein complex (active conformation) this interaction is broken and Arg of the E/DRY motif packs against a Tyr residue of Gs (Rasmussen et al., 2011). In absence of X-ray crystal structure of the VPAC receptors, only model structures of VPAC1 have been reported which used as template the structures of the N-terminal domain of the CRF 2β receptor (Ceraudo et al., 2012) or structures of family A GPCRs for the transmembrane domains (Conner et al., 2005; Chugunov et al., 2010). However, the low sequence identity between the sequence

"fendo-03-00129" — 2012/10/28 — 19:05 — page 2 — #2

"fendo-03-00129" — 2012/10/28 — 19:05 — page 3 — #3

of the VPAC receptors and the templates used for homology modeling prevents direct transposition of molecular switches that drive GPCR-A members activation.

As all members of GPCR-B family, VPAC receptors lack the E/DRY sequence. On the basis of subtle changes observed when Y<sup>239</sup> and L240, located in TM3 of VPAC1, were substituted with alanine it was proposed that this YL sequence was equivalent to the E/DRY motif of GPCR-A family (Tams et al., 2001). Another model based on a three-dimensional analysis of the GLP-1 receptor proposed that a E/DRY motif could be formed by three nonadjacent residues consisting in R<sup>174</sup> in the cytoplasmic end of TM2, E236, and Y239 in the distal part of TM3 of VPAC1 (Frimurer and Bywater, 1999). However, another study showed that Y239A, L240A, E236A, Y239A, and R174A mutants were undistinguishable from the wild type receptor (Nachtergael et al., 2006). One possible explanation for the discrepancy can be the fact that Tams et al. (2001) studied cyclic AMP measurements in intact cells a more sensitive model than the adenylate cyclase assay on membrane used in the other study. Nevertheless, even if the YL motif of GPCR-B family and E/DRY motif of GPCR-A family have the same location, they certainly do not have the same importance for receptor activation (**Figure 2**).

As mentioned before structural data confirmed that GPCR activation is accompanied by modification of helix conformation. For the β2-AR, the largest difference between the inactive and active structures is a large outward movement of TM6 and an inward movement of TM7 such as Tyr of the conserved NPxxY sequence (located in TM7) moves into the space occupied by TM6 in the inactive state. In GPCR-A family conserved prolines located in TM6 and laying near the ligand binding pocket and in TM7 (NPxxY sequence) play crucial role in these rearrangements due to their cyclic pyrrolidine ring side chain that introduce kink into the α-helical structure of TM and allow TM flexibility important for G protein coupling and signaling. In this line, a study of Knudsen et al. (2001) investigated the role of prolines located in TM of VPAC1 (**Figure 2**). They found that mutation of P266 (TM4), P300 (TM5), and P348 (TM6) into alanine significantly decrease VPAC1 expression levels but preserve VIP binding. P266A showed decreased ability to stimulate cAMP, while P300A and P348A displayed an increased potency in cAMP production combined with a high sensitivity toward GTP compared to the wild type receptor, thus demonstrating that these prolines are important for overall structure of VPAC1 and receptor activation (Knudsen et al., 2001).

A more recent study, that combined pharmacological and *in silico* approaches, identified a network of interactions between residues located in helices 2, 3, and 7 of the VPAC1 receptor, which could be involved in the stabilization of the receptor in absence of agonist and in early steps of receptor activation. It was proposed that, in absence of ligand, interaction between R188, N229, and Q380 ties helices 2, 3, and 7 together (**Figure 3**). Upon VIP

"fendo-03-00129" — 2012/10/28 — 19:05 — page 4 — #4

binding, the interaction between R<sup>188</sup> and N<sup>380</sup> is broken and a stronger interaction (salt bridge) is established between R<sup>188</sup> and the D3 side chain of VIP. TM2 and probably other helices undergo conformational changes causing key sequences located in intracellular loops to be exposed and to interact with the G proteins. In the meantime, the interaction network involving N<sup>229</sup> and Q380 maintains TM7 in a conformation necessary for proper activation of G proteins. The three-dimensional model also suggested that Q<sup>380</sup> could function as a floating "ferry-boat", switching between R188 and N<sup>229</sup> residues' side-chains hence contributing to signal transduction propagation and activation of G proteins (Chugunov et al., 2010).

When considering other site-directed mutagenesis studies, it is likely that a complex and larger network of interaction between TM helices must be considered for stabilization of VPAC1 inactive and active conformations (**Figure 2**). Indeed, the mutation into arginine of H178 located at the bottom of TM2 led to a constitutively activated VPAC1 receptor (Gaudin et al., 1998). Similarly, mutation of T343, located at junction of the third intracellular loop and TM6 of VPAC1, into lysine, proline, or alanine also led to a constitutively activated receptor (Gaudin et al., 1999). Another study showed that Y<sup>146</sup> and Y150, located in TM1 of VPAC1, do not interact directly with VIP but stabilize the correct active receptor conformation (Perret et al., 2002). It was also observed that K195 and D196 located at junction of TM2 and the first extracellular loop were essential for VPAC1 activation but were not directly involved in VIP recognition (Langer et al., 2003). How all these residues cooperate to propagate signal transduction afterVIP binding remains to be elucidated and would require a model or a structure of the activated receptor in complex with VIP. Particularly the two N-terminal residues of VIP, H<sup>1</sup> and S2, are likely to affect, directly or indirectly, the interaction network involved in receptor activation. In this line, a very recent study suggested that K143, T144, and T147, located in TM1 of VPAC1 could interact with H<sup>1</sup> residue of VIP and play an important role in receptor activation (Ceraudo et al., 2012). Mutagenesis studies of VPAC2 are less

"fendo-03-00129" — 2012/10/28 — 19:05 — page 5 — #5

exhaustive, however, some studies identified key residues involved in receptor activation such as Y<sup>130</sup> and Y<sup>134</sup> located in TM1 (Perret et al., 2002), K<sup>179</sup> in TM2 (Vertongen et al., 2001), and N216 in TM3 (Nachtergael et al., 2006), suggesting that VPAC1 and VPAC2 share a similar pattern of activation. Moreover, so far as all residues that were identified as important for VPAC receptors activation are highly conserved among GPCR-B family members, they may therefore be involved in binding and activation mechanisms that are common to the whole family.

#### **MOLECULAR MECHANISMS INVOLVED IN VPAC/G PROTEIN BINDING AND ACTIVATION**

The α subunit of heterotrimeric G proteins has a central role in interaction with both the receptor and the effectors. Several studies have shown that the C-terminal part of Gα subunit can directly bind to the receptor and is involved in the coupling specificity (Conklin et al., 1996). The current model of GPCR activation, based on the study of family A GPCRs, proposes that when the receptor switches to its active conformation, TM movements are accompanied by intracellular loops switches leading to exposure of the G protein binding pocket to cytosol and efficient binding to G protein. However, the diversity of sequences and loop sizes as well as their flexibility has made difficult the identification of a specific set of residues defining the coupling profile.

For the VPAC1 receptor, Gα binding domains are mainly located in the third intracellular loop (IC3) that contains subdomains dedicated to the recognition of the different Gα subunits. K<sup>322</sup> located in proximal part of IC3 and E394 located at the junction of TM7 and the C-terminal tail are required for adenylate cyclase activation but not for the coupling to the inositol trisphosphate/calcium pathway. The former being involved in direct interaction with Gαs (G protein binding), as demonstrated by a reduced sensitivity to GTP, while E<sup>394</sup> triggering switch of Gαs from inactive to active state (G protein activation; Couvineau et al., 2003; Langer and Robberecht, 2005). Similarly, two other sequences located in IC3 have been identified as important for VIP-induced intracellular calcium increase but not cAMP production. A small sequence, I328-R329-K330-S331, located in the central part of IC3 is involved in efficient binding of VPAC1 to Gαi/o and Gαq (Langer et al., 2002), while R<sup>338</sup> and L339, located at the distal part of IC3, mediate interaction of VPAC1 with Gαi/o (Langer and Robberecht, 2005). Combining mutations in the proximal and distal part of IC3 together with mutation of E<sup>394</sup> gave rise to a completely inactive VPAC1 receptor with respect adenylate cyclase activation and intracellular calcium increase (**Figure 2**). In VPAC2 receptors, L310 located in the proximal part of IC3 contributes to Gαs activation and the proximal part of IC3 (R<sup>325</sup> and K328) is involved in both Gαs and Gα16 coupling. The combined mutations of these three amino acids generates an inactive VPAC2 receptor with respect to [Ca2+]i increase and adenylate cyclase activation (Langer et al., 2005).

Among the different members of the GPCR-B family, proximal and distal domains of IC3 share conserved sequences that could therefore represent common G protein binding motifs. In line with this hypothesis, studies performed on other members of the GPCR-B family identified the proximal domain of IC3 as essential for adenylate cyclase activation but the amino acids involved may differ and additional conserved sequences located in other intracellular regions of the receptor may also be necessary as seen for glucagon (IC2; Cypess et al., 1999) and CGRP receptors (R151 located in IC1; Conner et al., 2006). The junctions of IC3 loop are predicted to beα-helical and it is assumed that the correct positioning of charged amino acids plays an important role in G protein interaction. However, other data suggest that lipophilic and aromatic residues are also important for G protein interaction. It is possible that IC3 loop junctions activate G protein directly or that

they may serve as regions that control the loop conformation. As mutations may change both direct interaction site and secondary structure, it is difficult to define more precisely the mechanisms involved in IC3 loop/G protein interaction.

# **REGULATION OF VPAC RECEPTORS ACTIVITY**

Besides coupling to the effectors, GPCR activation by agonists also initiates the process of receptor desensitization, an adaptive response contributing to rapidly fade the G protein signaling. This process starts with phosphorylation of the receptors in intracellular loops and carboxyl terminus by activity dependent kinases (PKA and PKC) and/or by receptor activated dependent kinases (GPCR kinases or GRK). GRK mediated phosphorylation of the receptor promotes the high affinity binding of β-arrestins to the receptor, which both sterically interdicts further coupling of G protein and may act as a signal transducer to activate for instance MAPKs, AKT, and PI3 kinases. β-arrestins are also able to bind proteins of the endocytic machinery including clathrin and adaptor protein AP2 and promotes receptor internalization (Shenoy and Lefkowitz, 2011).

Like most GPCRs, VPAC1 and VPAC2 receptors are rapidly phosphorylated after agonist exposure. Absence of inhibitory effect of PKA and PKC inhibitors (Langer et al., 2005; Langlet et al., 2005), use of dominant negative GRK and GRK overexpression experiments (Shetzline et al., 2002; Huang et al., 2007), suggest that GRK are the main kinases involved in VIP receptors phosphorylation. VPAC1 receptor phosphorylation induces βarrestin translocation to cell membrane, however, over-expression of β-arrestin in HEK 293 cells only causes a minor decrease in cAMP, suggesting VPAC1 desensitization may occur through an arrestin-independent mechanism (Shetzline et al., 2002). By using site-directed mutagenesis studies, T429, S435, S447, S448, S449, S455 located in the C-terminus and S<sup>250</sup> located in IC2 were identified as potential candidates for VIP-induced VPAC1 receptor phosphorylation (Marie et al., 2003; Langlet et al., 2005; **Figure 2**). Combining the mutations of these identified residues indicated that the effect on phosphorylation was not additive and the mutants tested maintained a phosphorylation level of about 30% of that observed for the wild type receptor except when all the Ser/Thr residues of the C-terminus were mutated (Langlet et al., 2005). Although it was shown that VPAC1 forms a complex with β-arrestin and this complex is transported into the cell during endocytosis (Shetzline et al., 2002), direct correlation between the VPAC1 phosphorylation level and internalization was not obvious as in some mutated receptors a significant reduction in phosphorylation has no or little effect on receptor internalization. For instance, some VPAC1 truncated receptors were still internalized while VIP-induced phosphorylation was undetectable thus suggesting an arrestin-independent mechanism of internalization and only mutation of all the Ser/Thr residues of the C-terminus and S250 completely abolished VPAC1phosphorylation and internalization (Langlet et al., 2005). VPAC1 and VPAC2 receptors are both rapidly internalized following agonist exposure but differ in their trafficking pattern. After internalization, VPAC1 receptors are not re-expressed at the cell surface within 2 h after agonist washing (suggesting that VPAC1 binds with high affinity β-arrestin) while VPAC2 receptors are recycled back to the cell surface (suggesting that VPAC2 binds β-arrestin with low affinity; Langlet et al., 2004, 2005). In VPAC2 receptors, it was found that inactivating mutations that alter G protein coupling reduced both receptor phosphorylation and internalization in a manner that appeared directly linked to the alteration of the Gαs and Gα16 coupling (Langer et al., 2005). As mutants studied do not affect phosphorylatable residues, this suggests that impaired receptor phosphorylation and internalization directly reflect the reduced ability of the mutants to adopt active receptor conformation. However, another study showed that VIP stimulated phosphorylation of N229Q VPAC1 and N216Q VPAC2 (two mutants characterized by a decrease in potency and efficacy of VIP stimulated adenylate cyclase activity, by the absence of agonist stimulated [Ca2+]i increase, by a preserved receptor recognition of agonists and antagonist and by a preserved sensitivity to GTP) was only slightly decreased, that receptor internalization was comparable to that of the wild type receptors but unlike wild type VPAC1 the N229Q mutant was rapidly re-expressed at the cell surface upon VIP washing (Nachtergael et al., 2006). As N229and N216 residues are located in the middle of TM3, it seems unlikely that they could be in direct interaction with β-arrestin but probably contribute to receptor conformation necessary for high affinity binding with β-arrestin. This later study thus suggests that receptor conformation necessary for activation and regulatory mechanisms could be different (**Figure 4**). As mentioned before, β-arrestins are also able to act as a signal transducer and activate G protein-independent signaling pathways but to date such a mechanism has not been described for the VIP receptors.

# **VPAC RECEPTORS OLIGOMERIZATION**

It is now well accepted that, like for single transmembrane receptors, GPCRs are also able to form oligomeric complexes either with themselves (homo-oligomerization) or with other receptors (hetero-oligomerization) that may affect the properties of the receptor. When considering the large repertoire of GPCRs, and thus the multitude of potential oligomers that can be formed, a challenging task consists in identifying the specificity and the physiological relevance of the partners, and the functional consequences of these associations. By far, the GPCR-A/rhodopsin family has been the most extensively studied, and data collected indicate that the consequences of oligomerization may be detected at the level of both physiological and pharmacological ligand recognition, in signaling properties as well as at desensitization and internalization levels. Besides pharmacological and functional regulation, it was also demonstrated that GPCR oligomerization plays a role in receptor maturation and in expression at the cell surface (Milligan, 2009; Maurice et al., 2011).

Regarding VPAC receptors, two studies performed on the prostate cancer cell line LNCaP (Juarranz et al., 2001) and adipocytes (Akesson et al., 2005) demonstrated that those cells expressed both VPAC1 and VPAC2 receptors but differed in their pharmacological response with respect to the selective VPAC1 and VPAC2 agonists. LNCaP cells displayed a VPAC1-like phenotype (the VPAC2 selective agonist being inactive) while adipocytes displayed a VPAC2-like phenotype (the VPAC1 selective agonist being inactive). One possible explanation of these results might be the

"fendo-03-00129" — 2012/10/28 — 19:05 — page 6 — #6

**FIGURE 4 | Mechanisms involved in regulation of VPAC1 activity.** Upon VIP binding wt VPAC1 **(A)** is rapidly phosphorylated through a GRK-dependent mechanism and desensitized, phosphorylation promotes high affinity binding of β-arrestin to VPAC1 and internalization into endosome through an arrestin- and dynamin-dependent mechanism. After internalization VPAC1 is not re-expressed at the cell surface. In some truncated VPAC1 receptors **(B)**, VIP fails to induce GRK-mediated receptor phosphorylation but the receptor is still able to internalize through an arrestin-independent mechanism and is re-expressed at the cell surface. Some VPAC1 receptor mutants **(C)**, with decrease in potency and efficacy of VIP-stimulated adenylate cyclase activity, preserved receptor recognition of agonists and antagonist and a preserved sensitivity to GTP, are still phosphorylated and internalized and recycle back to cell surface.

formation of inactive receptors dimers neutralizing the receptor expressed at the lowest concentration and letting only one receptor active. First evidence that VPAC receptors form oligomers came from a study of Harikumar et al. (2006) who demonstrated by using biophysical methods that VPAC1 and VPAC2 receptors formed constitutive homo- and hetero-oligomers but also heterodimers with secretin receptors that remained trapped into the cells (Harikumar et al., 2006). In an attempt to evaluate the functional consequences of hetero-oligomerization of VPAC receptors, another study investigated the effect of coexpressing VPAC1 and VPAC2 receptors in Chinese hamster ovary (CHO) cells on ligand binding, adenylate cyclase activation, receptor internalization, and trafficking. Three agonists were used, the natural ligand VIP, and selective VPAC1[K15,R16,L27]VIP(1-7)/ GRF(8-27)] and VPAC2[Ro 25-1553] agonists. They found that pharmacological properties of cells expressing both receptors were not different from those obtained when mixing cells expressing each receptor individually. Similarly, VIP receptors co-expression did not modify receptor internalization and trafficking patterns following exposure to VIP or selective agonists (Langer et al., 2006). Although this study did not point out any pharmacological consequences of constitutive heterodimerization of VPAC1–VPAC2 receptors, it suggests that the pharmacological profile of the selective VPAC1 and VPAC2 receptors ligands that was established in cells expressing one receptor subtype is also applicable to cells that could express both receptors endogenously. Thus making possible the activation, antagonism, or downregulation of one receptor subtype without affecting the other. Additional experiments are now needed to demonstrate that these oligomers do exist in more physiological conditions and identify physiological consequences, if any, of VIP receptors oligomerization.

Some GPCRs are also able to form hetero-oligomers with receptor activity-modifying proteins (RAMPs), a family of single transmembrane accessory proteins (RAMP1, 2, and 3). More particularly, hetero-oligomerization of RAMPs with two GPCR-B family members, the calcitonin receptor and the calcitonin receptor-like receptor, generates multiple receptor phenotypes with different specificities for endogenous ligands. It has first been shown that VPAC1 receptors, but not VPAC2 receptors, were also able to interact with RAMP1, 2, and 3. VPAC1–RAMP complexes did not display altered ligand specificity compared with VPAC1 alone. However, specific interaction with RAMP2 led to a significant enhancement of VIP- and PACAP-induced inositol phosphate hydrolysis with unaltered cAMP production (Christopoulos et al., 2003). A more recent study, however, observed that VPAC2increases cell surface expression of all three RAMPs. As seen for VPAC1, VPAC2, and RAMP co-expression had no or little effect on agonist-stimulated cAMP production but RAMP1 and RAMP2 significantly enhanced basal coupling to Gαi (Wootten et al., 2012). The RAMPs being largely distributed throughout the nervous system and peripheral tissues, it would be interesting to investigate more deeply the properties of VPAC–RAMP complexes such asfor example selective agonists and antagonists specificity, receptor signaling through other pathways, receptor desensitization and internalization.

#### **CONCLUSION AND PERSPECTIVES**

"fendo-03-00129" — 2012/10/28 — 19:05 — page 7 — #7

Vasoactive intestinal peptide plays diverse and important role in the CNS, PNS, gastrointestinal tract, and immune functions and VPAC receptors constitute potential targets for the treatment of

several diseases such as neurodegenerative disorder, schizophrenia, asthma, diabetes, gastrointestinal motility disorder, Crohn's disease, and rheumatoid arthritis. A key limiting factor in this field is that all the currently useful pharmacological tools are peptides, and thus limit their use in human therapy. The first small molecule antagonists of VPAC1 (Harikrishnan et al., 2012) and VPAC2 (Chu et al., 2010) have been described recently and emerge

#### **REFERENCES**


molecular pharmacology and interaction with accessory proteins. *Br. J. Pharmacol.* 166, 42–50.


from high throughput screening of compound collection. However, all are low affinity, micro molar range, antagonists and need further improvement to expect to obtain druggable modulators of VPAC receptors activity. Better understand of structure, ligand binding site, and molecular pharmacology of VPAC receptors constitute thus a key information for the rationale design of VPAC receptors modulators and potential druggable compounds.

of immune tolerance by antiinflammatory neuropeptides. *Nat. Rev. Immunol.* 7, 52–63.


"fendo-03-00129" — 2012/10/28 — 19:05 — page 8 — #8

cell line maintains the expression and function of VIP and PACAP receptors. *Cell. Signal.* 13, 887–894.


Langer VPAC receptors

and analogs on activation and internalization of the recombinantVPAC2 receptor expressed in CHO cells. *Peptides* 25, 2079–2086.


phospholipase D by VPAC and PAC1 receptors. *Ann. N. Y. Acad. Sci.* 921, 175–185.


and diseased human tissues. Clinical implications. *Ann. N. Y. Acad. Sci*. 921, 1–25.


"fendo-03-00129" — 2012/10/28 — 19:05 — page 9 — #9

receptors expressed by stable transfectants couple to two distinct signaling pathways. *Biochem. Biophys. Res. Commun.* 203, 141–148.


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 27 August 2012; paper pending published: 14 September 2012; accepted: 11 October 2012; published online: 30 October 2012.*

*Citation: Langer I (2012) Mechanisms involved in VPAC receptors activation and regulation: lessons from pharmacological and mutagenesis studies. Front. Endocrin. 3:129. doi: 10.3389/fendo. 2012.00129*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Langer. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Alternative splicing of the pituitary adenylate cyclase-activating polypeptide receptor PAC1: mechanisms of fine tuning of brain activity

# **Janna Blechman and Gil Levkowitz\***

Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Hubert Vaudry, University of Rouen, France Atsuro Miyata, Kagoshima University, Japan

#### **\*Correspondence:**

Gil Levkowitz, Department of Molecular Cell Biology, Weizmann Institute of Science, P. O. Box 26, Rehovot 76100, Israel. e-mail: gil.levkowitz@weizmann.ac.il

# **INTRODUCTION**

G-protein-coupled receptors (GPCRs) represent the largest family of membrane-associated proteins mediating physiological responses in vertebrates by means of controlling metabolic, neural, and developmental functions (Lefkowitz, 2007; Frooninckx et al., 2012; Zhang and Eggert, 2013). These proteins are expressed in almost all types of tissues and are represented by over 1,000 membrane receptors for extracellular (EC) ligands including hormones, neurotransmitters, pheromones, lipids, and other proteins (Xue et al., 2008; Markovic and Grammatopoulos, 2009; Nordstrom et al., 2009; Hoyer and Bartfai, 2012; Jafurulla and Chattopadhyay, 2013). The basic structure of GPCR proteins consists of a seven-transmembrane (TM) domain with a complex EC structure composed of an N-terminal region and three EC loops involved in the diverse ligand recognition process. Three intracellular (IC) loops and a C-terminal domain transduce a signal into the cell's cytoplasm and nucleus. Typically, ligand binding to the EC loops induces conformational changes in the TM and IC domains of the receptor resulting in specific coupling to a set of cytoplasmic molecules, termed G-proteins, each composed of different isoforms of alpha, beta, and gamma subunits. G-proteins in turn regulate the activity of IC effector molecules, including adenylate cyclase (AC), phospholipase Cβ (PLC), and RhoGEF causing the activation of secondary messengers such as cyclic AMP (cAMP), inositol-1,4,5 triphosphate (InP3), and diacylglycerol leading to the initiation of distinct IC signaling pathways (Simon et al., 1991; Sah et al., 2000; Pierce et al., 2002; Oldham and Hamm, 2008; Mizuno and Itoh, 2009; Mahata et al., 2011; Maurice et al., 2011).

The diversity of the GPCR-initiated signal transduction pathways is determined by the multiplicity of cognate ligands with different receptor binding properties and by the receptors' membrane-interacting partners that form homo- and

Alternative splicing of the precursor mRNA encoding for the neuropeptide receptor PAC1/ADCYAP1R1 generates multiple protein products that exhibit pleiotropic activities. Recent studies in mammals and zebrafish have implicated some of these splice isoforms in control of both cellular and body homeostasis. Here, we review the regulation of PAC1 splice variants and their underlying signal transduction and physiological processes in the nervous system.

**Keywords: ADCYAP1R1, activity-dependent gene regulation, zebrafish model system, PACAP receptor, stress disorders, post-traumatic, hypothalamic hormones, homeostasis**

> hetero-dimerized GPCR complexes. These receptor dimers have unique binding properties to both EC ligands and a set of IC G-proteins (Milligan and Kostenis, 2006; Harikumar et al., 2008; Furness et al., 2012). Thus, ligand binding, along with receptor heterodimerization, can generate diversity in G-proteins coupling that can generate varied IC signaling functions.

> Beyond the complexity of the aforementioned EC and IC GPCRs interacting effectors, the genetic diversity of the GPCR family can be generated by means of alternative precursor mRNAs (pre-mRNA) splicing of exons encoding specific protein moieties and causing considerable functional differences of the resulting splicing products (Black, 2003; Jaillon et al., 2008; Furness et al., 2012). Alternative splicing mechanisms allow the generation of multiple mRNA transcript variants from a single gene by utilizing different combinations of exons by means of skipping or insertion of alternatively spliced exons. In particular, neuronal cells are known to exhibit high levels of alternative splicing, generating the basis for molecular and cellular diversity important for the patterning and maintenance of the central and peripheral nervous systems (Yeo et al., 2007; Betke et al., 2012; Norris and Calarco, 2012; Sun et al., 2012). In the case of GPCRs, this combinatorial exon assembly can lead to changes in the protein domains responsible for ligand binding, IC effector coupling, as well as receptor stability, and endocytosis.

> In this review we focus on one GPCR, the pituitary ACactivating polypeptide (PACAP) receptor (ADCAYP1R1), also known as PAC1. This receptor represents a fascinating example of how alternative splicing of a single GPCR gene leads to different physiological outcomes. We will describe the current knowledge regarding the role and mechanism of action of PAC1 splice variants in development, physiology, and diseases focusing on PAC1's role in the nervous system.

# **THE PAC1 RECEPTOR**

PAC1 belongs to the glucagon/secretin receptor family of GPCRs that consists of hormone and neuropeptide receptors. These receptors signal through coupling to the G-protein alpha subunits Gs and Gq that typically activate AC and phospholipase Cβ enzymes, respectively (McCulloch et al., 2002; Ahren, 2008; Dickson and Finlayson, 2009; Vallejo, 2009; Vaudry et al., 2009). Coupling of PAC1 to Gs may also lead to cAMPdependent accumulation of IC calcium (Braas and May, 1996; Mustafa et al., 2010). The PAC1 gene contains multiple exons that undergo extensive alternative splicing. The peptide PACAP is the high-affinity ligand for PAC1. Post-translational proteolytic processing of the PACAP precursor protein generates several polypeptides with varying sizes, including PACAP38 and PACAP27 (Miyata et al., 1989; Vaudry et al., 2009; Harmar et al., 2012; Watkins et al., 2012). PAC1 was found to play a pivotal role in the spatio-temporal regulation of proliferation, differentiation, or cell survival during development as well as in the regulation of synthesis and release of neuroendocrine hormones.

Phylogenetic analysis of the vertebrate PAC1 receptor family indicates their origin from a common ancestral gene and demonstrates a tree topology with species-specific paralogs belonging to different separated sub-groups (**Figure 1**). Bony fishes (*teleosts*) genomes typically contain more than one gene due to teleostspecific gene duplication event of both ligand and receptor molecules during the evolution of these species (Wei et al., 1998; Fradinger et al., 2005; Bjarnadottir et al., 2006; Cardoso et al., 2007; Machluf et al., 2011).

# **ALTERNATIVELY SPLICED PAC1 GENE PRODUCTS**

The alternative splicing of PAC1 has been extensively studied. In this review we use the exon numbers designation of the human *PAC1* and otherwise indicate cases where the exons numbers do not match. The human gene was shown to contain 18 exons with the open reading frame encoded by exons 2–18 (Chatterjee et al., 1997; Lutz et al., 2006). Ten of these exons are constitutively expressed (exons 2, 3, 7–13, 18), whereas the rest (exons 4–6, 14, 15, and possibly 16, 17) are regulated by the alternative splicing (**Figure 2**). The N-terminal part of PAC1's EC domain is encoded by six exons. Exons 7–17 encode the seven-TM domains including EC and IC loops and exon 18 encodes the C-terminal cytoplasmic tail including the 3<sup>0</sup> -untranslated region (**Figures 2** and **3**). PAC1 splicing variants were identified in other vertebrate species including rat, mouse, frog, and fish. In **Table 1** we assembled all known PAC1 splice isoforms from different vertebrate species. The functional outcome of these splicing events will be discussed later. By and large, PAC1 alternative splicing can be divided into four types of splicing events that impact receptor functions (**Figure 3**; **Table 1**): (1) Variations in the EC N-terminal domain altering the ligand-binding specificity and affinity. (2) Variations in exons encoding to part of the third IC loop (IC3) thereby affecting G-protein coupling and/or interaction with other IC signaling proteins. (3) Variations in the TM domains TM2 and TM4 contributing to the receptors heteromerization and IC transport. (4) Variations in the 5<sup>0</sup> UTR that may affect mRNA expression dynamics. Notably, splicing products

containing different combinations of N-terminal and IC3 splice variants were identified in mammalian species.

The terminology used in independent studies can be somewhat perplexing as different names were designated to PAC1 isoforms corresponding to the same splicing events identified in various species. In the following section we describe the various splice variants and indicate when a different (i.e., specie-specific) name was designated to the same isoform. A splice isoform that includes all N-terminal encoding exons but does not contain IC (IC3) insertion(s) is referred to as "PAC1-null" (**Table 1**). The ligand binding and signaling properties of PAC1-null are often used as a reference when assessing the activities of other splice variants.

#### **N-TERMINAL VARIATIONS**

The N-terminal variants were identified in rodents and humans but not in fishes to date. They are generated by alternative splicing of exon 5 (PAC1-δ5), exons 5–6 (PAC1-short or PAC1-δ5,6), or exons 4–6 (PAC1-very short or PAC1-δ4,5,6) leading to deletions of 7, 21, or 57 amino acids, respectively (Journot et al., 1994;

Pantaloni et al., 1996; Chatterjee et al., 1997; Dautzenberg et al., 1999; Lutz et al., 2006; Ushiyama et al., 2007, 2010). In the rat, an insertion of 24 amino acids is caused by splicing of a novel exon 3a located between exons 3 and 4 (PAC1-3a) (**Figure 3**) (Daniel et al., 2001; Ajpru et al., 2002; Pilzer and Gozes, 2006a). N-terminal splicing isoforms of PAC1 display alterations both in ligand-binding selectivity and coupling to second messengers compared to PAC1-null.

#### **INTRACELLULAR LOOP VARIATIONS**

PAC1 splice variants in the third IC loop (IC3) have been identified in human, amphibian, fishes, and rodents. These splice isoforms are characterized by the presence of one or two cassettes of 84 nucleotides (hip or hop1 variants) or 81 nucleotides (hop2 variant), or a combination of alternative spliced "cassettes" (hip-hop1 or hip-hop2) (Spengler et al., 1993; Journot et al., 1995; Grimaldi and Cavallaro, 1999; McCulloch et al., 2001; Ronaldson et al., 2002; Fradinger et al., 2005; Ushiyama et al., 2007; May et al., 2010; Holighaus et al., 2011). In human, hip and hop variants are also referred to as SV1 and SV2 or SV3 for hip-hop1 (Pisegna and Wank, 1996; Pisegna et al., 1996). In addition, a splice variant formed by a C-terminal deletion of 193 nucleotides (denoted"*hop1 novel*") including two amino acids that are essential for G-protein recognition was identified in rat cochlea samples (Abu-Hamdan et al., 2006). Similar alternative splicing variants of PAC1 are also found in non-mammalian vertebrates such as the frog and bony fishes. Three alternative splice variants were identified in the frog IC3 loop (Alexandre et al., 2002). PAC1-R25, with an insertion of 25 amino acids, corresponds to the mammalian hop cassette. PAC1-R41 contains a cassette with no homology to any other variant. PAC1-RMc with a unique cytoplasmic insertion of 13 amino acid into the TM7 domain is missing a canonical Gs recognition motif. PAC1 genes were also identified in zebrafish, goldfish, stickleback, fugu, sea bream, and several others bony fish species. Three zebrafish PAC1 alternative splice variants with insertions in the IC3 loop were identified. Two of them were found to be homologous to the hop1 (84 nucleotides) and hop2 (81 nucleotides) mammalian isoforms (Fradinger et al., 2005). A unique 107 nucleotides inclusion with a premature stop codon (PAC1-skip) resulting in a truncated PAC1 protein did not correspond to any previously detected PAC1 isoform (Fradinger et al., 2005). This isoform resembles an alternative splicing event in a human gene encoding another secretin family receptor, *VPAC1*. Interestingly, the hip cassette variant was not identified in the mouse and zebrafish genomes. Moreover, teleosts genomes contain two (duplicated)

PAC1 paralogs, *pac1a* and *pac1b,* however, only *pac1a* encodes for the hop splicing cassette.

#### **TRANSMEMBRANAL DOMAIN VARIATIONS**

A PAC1 variant, cloned from the rat cerebellum, has amino acids deletion/substitution in the TM4 domain along with two amino acid substitutions in the N-terminal (D136N) and TM2 (N190D) domains (Chatterjee et al., 1996;Ajpru et al., 2002). The exact molecular mechanism underlying these variations remains unclear. Notably, the TM4 domain of the secretin family receptors are involved in homo- and hetero-oligomerization of these receptors, associations with receptor activity-modifying proteins (RAMPs), and with GPCR kinases (GRKs) thereby suggesting that splice alterations in the PAC1-TM4 domain may affect receptor function (Morfis et al., 2003; Ritter and Hall, 2009; Magalhaes et al., 2012).

#### **5** <sup>0</sup> **UTR VARIATIONS**

Alternative splicing events in exons located at the 5<sup>0</sup> UTR were identified for rat *PAC1* gene (Chatterjee et al., 1997). These include different alternative usage of exons located upstream to the ATG translation start codon. Such variations in the 5<sup>0</sup> UTR organization and sequences may play a role in the regulation of mRNA expression.

# **ALTERNATIVE PAC1 SPLICING ALTERS LIGAND-BINDING PROPERTIES**

PAC1 is considered as being the high-affinity receptor for PACAP, while it displays low binding affinity to the vasoactive intestinal polypeptide (VIP) (Apostolakis et al., 2005; Vaudry et al., 2009; Harmar et al., 2012). Alternative splicing of PAC1 results in different protein products displaying different ligand-binding properties that may result in changes in affinity and selectivity (**Table 1**). Most studies use the PAC1-null (McCulloch et al., 2001; Holighaus et al., 2011) isoform as a reference for PACAP and VIP binding properties.

Modeling of ligand-receptor binding proposes that the Cterminal part of the ligand PACAP binds to the N-terminus of the PAC1 receptor and that the N-terminal part of PACAP binds to the receptor's EC loops and TM domains (Furness et al., 2012). Consequently, alterations in the EC domain of PAC1 are predicted to affect these ligand-receptor binding properties (**Table 1**). Thus, PAC1-very short (a.k.a. PAC1-δ4,5,6), which lacks 57 amino acids in the EC1 domain displays decreased affinity to PACAP27 and PACAP38 but its affinity toward VIP remains the same (Journot




et al., 1995; Pantaloni et al., 1996; Dautzenberg et al., 1999; Lutz et al., 2006). The binding affinities for PACAP38 of the δ5, δ5,6 (a.k.a. short) splice isoforms appeared to be very similar to that of PAC1-null, while δ5,6 isoform has increased affinity toward VIP. The rat-specific PAC1-3a isoform containing a 24 residue N-terminal insertion displays increased affinity to PACAP38 but not to PACAP27 (Daniel et al., 2001; Pilzer and Gozes, 2006a). It should be noted that in all of these examples, PACAP is still the better ligand for PAC1 when compared with VIP (**Table 1**).

The hop1 and hip-hop1 splice isoforms retain similar binding properties to those of PAC1-null. In contrast, insertions into the IC3 loop, namely hip- and hop2-cassettes lead to the elevation of VIP (Spengler et al., 1993; Pisegna et al., 1996; Pilzer and Gozes, 2006b) and the diminution of PACAP binding affinities, thereby suggesting that the alteration in IC3 causes an inside-out conformational change.

Notably, binding analysis of alternative spliced isoforms, which result from combined changes in the EC and IC protein domains yields interesting results. Thus, a splice variant composed of PAC1 short with the inclusion of hop1-cassette, displays similar PACAP and VIP binding properties to that of PAC1-null and different VIP binding properties than PAC1-short alone (Lutz et al., 2006; Ushiyama et al., 2007, 2010). Another combined splice variant composed of PAC1-δ5 or PAC1-δ5,6 together with the hip cassette had different ligand-binding properties to that of PAC1, PAC1-δ5, or PAC1-hip alone (Lutz et al., 2006) (**Table 1**). This imply that the IC3 domain contributes to the association of PAC1 with its cognate ligands.

These results may be explained by a model for agonist binding to the corticotropin-releasing hormone (CRH) receptor,which belongs to the same GPCR sub-family of PAC1 (Dong et al., 2005). According to this model the TM6 protein domain plays a structural role in ligand binding. Thus, insertions into IC3 loop, which is located between TM5 and TM6 region may cause a conformational change in TM6 and possibility affect binding of PACAP and VIP to the EC part of the receptor. A proof for this model awaits a solved three-dimensional protein structural of the different PAC1 isoforms.

# **ALTERNATIVE PAC1 SPLICING ALTERS INTRACELLULAR SIGNAL TRANSDUCTION**

Ligand binding to PAC1 results in allosteric changes in the IC docking sites for effector coupling. A variety of IC signal transduction cascades can be differentially transduced by multiple PAC1 splice variants with altered EC or IC protein domains (**Figure 4**) (Arimura, 1998; Dickson and Finlayson, 2009; Vaudry et al., 2009; Furness et al., 2012). The most studied pathways include coupling of PAC1 to Gs and Gq proteins and activation of AC and phospholipase Cβ PLCβ) which result in the respective stimulation of production of cAMP and/or IP3 (Milligan and Kostenis, 2006;

**FIGURE 4 | A scheme depicting the current knowledge on PAC1-mediated signaling cascades resulting in a variety of neuronal outcomes**. PKA was shown to induce ERK1/2 thereby contributing to PACAP neuroprotective (Villalba et al., 1997; Vaudry et al., 2003; Falluel-Morel et al., 2006; Stumm et al., 2007) and neurotrophic (Ravni et al., 2006; Botia et al., 2007; Monaghan et al., 2008) functions. Action potential (AP) firing induces a CRCT1/CREB mediated neuroprotective effect, presumably through NMDA receptor activation and was shown to be initiated by PKA activation (Baxter et al., 2011). NMDA receptor was also shown to be indirectly activated by PACAP (Llansola et al., 2004) or cAMP/PKA signaling (Costa et al., 2009). PAC1/PKA signaling controls cellular apoptosis through inhibition of potassium channels (Mei et al., 2004; Castel et al., 2006; Pugh and Margiotta, 2006) or induction of calcium channels (Pugh and Margiotta, 2006). PAC1/PKA-mediated activation of ion channels leads to activity-dependent neuronal differentiation and synaptic plasticity (Lee et al., 1999; Maturana et al., 2002; Grumolato et al., 2003; Nishimoto et al., 2007). The tumor suppressor gene Lot1 known as a negative regulator of neuronal precursor proliferation was shown to be controlled by PAC1/cAMP/ERK signaling pathway (Fila et al., 2009). PAC1/PKA-dependent phosphorylation of Tau is involved in the control of granule cell migration during cerebellar development (Falluel-Morel et al., 2006). PAC1-mediated cAMP/ERK-dependent neurite outgrowth was shown to be regulated via a novel neuritogenic factor NCS (Emery and Eiden, 2012) or Epac/ERK (Gerdin and Eiden, 2007) pathways. PAC1/Epac was shown to regulate neuronal differentiation via activation of p38 kinase along with mobilization of Ca<sup>2</sup><sup>+</sup> from intracellular stores (Ster et al., 2007) as well as Epac/Rit-dependent pathway involving CREB signaling (Shi et al., 2006). PAC1 was demonstrated to induce Rit through TrkA/Shc/SOS signaling initiated by

Src activation via dual Gs/Epac and Gi stimulation (Shi et al., 2010). PACAP also signals through Gq-linked PLC > IP3 > Ca<sup>2</sup><sup>+</sup> /DAG > PKC, and PLD > phosphatidic acid (PA) pathways (Journot et al., 1994; Makhlouf and Murthy, 1997; Dejda et al., 2006). PAC1 is a mediator of gene transcription, neuronal differentiation, and synaptic development (Masmoudi-Kouki et al., 2006; Vaudry et al., 2007; Andero and Ressler, 2012). As an example of functional diversity caused by PAC1 alternative splicing two additional PAC1-hop1 signaling pathways are presented on the scheme. They depict PAC1-hop1/ARF dependent PLD activation (McCulloch et al., 2001) and internalization-dependent engagement of PI3Kγ/Akt activation (May et al., 2010). AC, adenylate cyclase; AGS, activator of G protein signaling; Akt, serine-threonine protein kinase PKB; cAMP, cyclic adenosine monophosphate; ARF, ADP(adenosine diphosphate) ribosylation factor; Caln, calcineurin; CaMK, calcium calmodulin kinase; CBP, creb binding protein; CRCT1, cysteine-rich C-terminal 1; CREB, cAMP responsive element-binding protein; DAG, diacyl glycerol; Epac, exchange factor directly activated by cAMP; ERK, extracellular signal-regulate kinase; G, guanine nucleotide-binding regulatory protein; IP3, inositol-1,4,5-triphosphate; JNK, c-Jun oncogene N-terminal kinase 1; Lot1, lost on transformation 1; NCS, neuritogenic cAMP sensor; NGF, nerve growth factor; NMDAR, N-methyl-d-aspartate receptor; p38, p38 mitogen-activated protein kinase; PA, phosphatidic acid; PI3K, phosphatidylinositol 3<sup>0</sup> OH kinase; PKA, protein kinase A; PKC, protein kinase C;PLC, phospholipase; Raf, B-Raf proto-oncogene serine/threonine kinase;Ras, retrovirus-associated DNA sequences; Rap1, Rit, small GTPases of the RAS oncogene family; Sos, son of sevenless homolog 1; Src, sarcoma viral oncogen homolog; Tau, neuron-specific microtubule-associated protein; TrkA, tropomyosin-related kinase.

Couvineau and Laburthe, 2012) (**Figures 3** and **4**). In addition, PAC1 was shown to regulate the level of IC Ca2<sup>+</sup> in either Gprotein dependent or independent manners. The following section

reviews the role of PAC1 splice variants in the activation of the above pathways as well as other non-canonical signaling cascades (see also **Table 1**).

# **AC AND PLC**β

When compared to the PAC1-null isoform, some alternative spliced variants display altered mode of signaling. For example, PAC1 isoform 3a, δ4,5,6 (a.k.a. very short), hip, hip-hop, δ5-hip, and δ5,6-hip show reduction in the potency of agonists-mediated cAMP and IP3 production (Spengler et al., 1993; Dautzenberg et al., 1999; Parsons et al., 2000; Daniel et al., 2001; Germano et al., 2004; Lutz et al., 2006; Pilzer and Gozes, 2006a; Holighaus et al., 2011). Similarly, PAC-TM4, hop1 novel, δ5,6,14–17, skip, and Rmc have lost their ability to stimulate cAMP and IP3 production (Chatterjee et al., 1996; Daniel et al., 2001; Ajpru et al., 2002; Alexandre et al., 2002; Fradinger et al., 2005; Abu-Hamdan et al., 2006; Pilzer and Gozes, 2006a). Notably, all of the above splice variants have modified or deleted IC3, TM, and C-terminal protein domains that are known to be important for G-protein binding (Chatterjee et al., 1996).

In some studies, PAC1-null displays very low or no activation of PLCβ while PAC1-hop1 switches its mode of signaling from AC to PLCβ (Spengler et al., 1993; DiCicco-Bloom et al., 2000; Nicot and DiCicco-Bloom, 2001; Ronaldson et al., 2002; May et al., 2010). For example, PAC1-hop1 was reported to mediate the activation of both AC and phospholipase C signaling in cortical sympathetic neuroblasts, while PAC1-null merely signals via AC stimulated pathway (Braas and May, 1996; DiCicco-Bloom et al., 2000). Other studies report that both isoforms display dual activation of AC and PLCβ pathways, however the hop1 isoform has higher activation of these pathways. Thus, analyses of SV1/hip and SV2/hop1 show that these isoforms display increased coupling to cAMP and PLCβ, respectively (Spengler et al., 1993; Pisegna and Wank, 1996; Lu et al., 1998; Braas and May, 1999; Parsons et al., 2000; Nicot and DiCicco-Bloom, 2001; Germano et al., 2004; Lutz et al., 2006; May et al., 2010; Holighaus et al., 2011). The frog R25 and R41 isoforms containing amino acid insertions into IC3 have higher activation of cAMP compared to PAC1-null (Alexandre et al., 2002).

Consistent with the effect of IC3 loop insertions on ligandbinding properties, hip- and hop2-cassettes conferred comparable potency to induction of cAMP and IP3 accumulation by PACAP and VIP but at the same time characterized by much lower efficacy of this response (Spengler et al., 1993; Pisegna and Wank, 1996; Pisegna et al., 1996; Ronaldson et al., 2002; Lutz et al., 2006). On the other hand, PAC1-hip variant was shown to retain AC mediated signaling but demonstrated impaired coupling to the PLCβ pathway (May et al., 2010).

#### **CA2**+**SIGNALING**

The concentration of calcium ions (Ca2+) in the cytoplasm is controlled by its uptake and release by specific transporter proteins residing in the plasma membrane, the inner mitochondria membrane, and the endoplasmic reticulum (ER) (Clapham,2007). PAC1-mediated Ca2<sup>+</sup> signaling was shown to play important roles in regulating neurotransmitter release, and neurotransmitter receptors (Shioda et al., 1997; Taupenot et al., 1999; Germano et al., 2009; Mustafa et al., 2010; Pugh et al., 2010;Amir-Zilberstein et al., 2012; Smith and Eiden, 2012). The effects of PAC1 splice isoforms on the levels of cytoplasmic Ca2<sup>+</sup> were reported in the cases of PAC1-null hop1/2, δ5,6/short, 3a, and PAC1-TM4 (**Table 1**) (Dautzenberg et al., 1999; Lutz et al., 2006; Mustafa et al., 2007, 2010; Nishimoto et al., 2007; Germano et al., 2009; Hansson et al., 2009; Vallejo, 2009; Ushiyama et al., 2010; Holighaus et al., 2011). PACAP is known to modulate both EC Ca2<sup>+</sup> influx via voltagegated calcium channels (VGCC) and Ca2<sup>+</sup> release from IC ER stores through both IP3/PLCβ and AC pathways but also through other signaling cascades (Tanaka et al., 1997; Shioda et al., 1998; Grimaldi and Cavallaro, 1999).

PAC1-null was reported to modulate cytosolic Ca2<sup>+</sup> mobilization from both EC and IC stores (Shioda et al., 1997; Masmoudi et al., 2003; Payet et al., 2003; Nishimoto et al., 2007). In acutely dissociated rat melanotrophs, the increase of cytosolic Ca2<sup>+</sup> was dependent on the activation of non-selective cation channels and the facilitation of voltage-dependent Ca2<sup>+</sup> channels by PKCand PKA-dependent phosphorylation, respectively (Tanaka et al., 1997). cAMP-dependent entry of EC calcium was also reported in astrocyte cells (Vallejo, 2009). It was also suggested that PAC1 null-mediated elevation of cytoplasmic Ca2<sup>+</sup> levels in NG108-15 cells is due to IP<sup>3</sup> receptor-mediated Ca2<sup>+</sup> release from IC stores (Holighaus et al., 2011).

The PAC1-hop1 variant is responsible for both Ca2<sup>+</sup> mobilization from IC stores and influx through voltage-gated Ca2<sup>+</sup> channels in bovine chromaffin cells (Mustafa et al., 2007). Transfection of PAC1-hop to the adrenomedullary pheochromocytoma (PC12) cell line showed sustained IP3-mediated Ca2<sup>+</sup> release from IC stores and from store-operated Ca2<sup>+</sup> entry (SOCE) (Taupenot et al., 1999; Mustafa et al., 2007, 2010). In bovine adrenal chromaffin cells, PAC1-hop was shown to mediate cytosolic Ca2<sup>+</sup> release from ryanodine/caffeine-sensitive Ca2<sup>+</sup> stores that was not dependent on either cAMP or IP3 generation (Tanaka et al., 1998; Payet et al.,2003). Heterologous expression of the rat PAC1-hop1 but not PAC1-hip variant in NG108-15 and PC12 cells leads to an increase in IC Ca2<sup>+</sup> concentration (Mustafa et al., 2007; Holighaus et al., 2011). In both cell lines the response consisted of a rapid and transient rise of Ca2<sup>+</sup> reminiscent of IP<sup>3</sup> receptor-mediated Ca2<sup>+</sup> release from IC stores followed by a prolonged Ca2<sup>+</sup> accumulated from EC source.

When compared with PAC1-null, PAC1-hop1 was more potent in Ca2<sup>+</sup> elevation (Holighaus et al., 2011). However, following overexpression in Chinese Hamster Ovarian cells, activation of PAC1-null induced higher Ca2<sup>+</sup> levels when compared with activation of PAC1-hop1 (Ushiyama et al., 2010). PAC1-hop1 also mediated PACAP-induced Ca2<sup>+</sup> release from ER and PKCγ translocation to the nucleus and plasma membrane resulting in the astrocytic differentiation (Nicot and DiCicco-Bloom, 2001). Finally, stimulation of voltage-gated L-type or non L-type Ca2<sup>+</sup> channels following initiation of PAC1-hop1 signaling was also reported (Mustafa et al., 2010).

Somewhat limited data was reported with regards to Ca2<sup>+</sup> signaling via other PAC1 splice variants. Thus, δ5,6/short as well as the PAC1 isoform containing combined "short" deletion in the N-terminal EC1 domain along with hop1-cassette in the IC3 loop exhibit PACAP-induced cytoplasmic calcium elevation (Ushiyama et al., 2010). PAC1-3a was shown to induce Ca2<sup>+</sup> accumulation through coupling to Gs/cAMP rather than Gq/PLCβ pathway (Pilzer and Gozes,2006a). PAC1-TM4 that wasfound to be inactive when assayed for both cAMP and PLCβ activation but it displayed elevation of Ca2<sup>+</sup> in response to PACAP27. It was demonstrated that this effect involved the modulation of voltage-gated L-type Ca2<sup>+</sup> channels (Chatterjee et al., 1996).

Taken together, the apparent inconsistent results concerning the mechanisms underlying the aforementioned Ca2<sup>+</sup> signaling events are most likely due to the different cellular systems (e.g., cell types) employed by the above studies to analyze PAC1-mediated IC Ca2<sup>+</sup> changes.

#### **OTHER TRANSDUCTION PATHWAYS**

Signaling of PAC1 through interaction with cytoplasmic protein partners, other than the canonical Gs and Gq, was reported mainly in the case of the PAC1-hop1 isoform (McCulloch et al., 2001; Ronaldson et al., 2002). Both PAC1-null and -hop1 proteins were reported to activate phospholipase D (PLD). Although PAC1-nullmediated PLD stimulation involved Gq/11 > PLC > PLD pathway, PAC1-hop1 was capable of activating PLD through direct binding to ADP-ribosylation factor (ARF) (McCulloch et al., 2002; Dejda et al., 2006).

#### **EXPRESSION OF PAC1 SPLICE ISOFORMS IN THE NERVOUS SYSTEM**

Information on time- and region-specific distribution of PAC1 splice isoforms may shed light on how PAC1 gene products regulate a plethora of biological functions in developmental and physiological processes (D'Agata et al., 1996; Waschek et al., 2000; Waschek, 2003; Vaudry et al., 2009). The distribution of PAC1 has been examined in many species using different techniques that revealed widespread expression in different tissues, including the nervous, cardiovascular, endocrine, immune, and respiratory systems (Aino et al., 1995; Abu-Hamdan et al., 2006; Gomariz et al., 2006; Molnar et al., 2008; Jolivel et al., 2009; Shneider et al., 2010; Lugo et al., 2011; Buljan et al., 2012). However, knowledge concerning the expression of PAC1 splice variants is somewhat limited at it is mainly based on analyses of isolated brain areas and/or primary cell cultures. The reported changes in PAC1 receptor variant expression appear to be most evident during development. We have summarized the existing data concerning the expression of these variants by focusing on expression in the nervous system of different mammalian species.

The major mRNA isoform of PAC1 in the brain is PAC1-null, which contains no splice deletions or insertions. PAC1-null is predominantly expressed in neurons residing in different brain areas although it is also detected in glial cells of the cerebella cortex and in activated astrocytes (Pilzer and Gozes, 2006b; Dickson and Finlayson, 2009; Vaudry et al., 2009).

PAC1 variants with N-terminal deletions are well represented in human fetal brain tissues and in five human neuroblastoma lines (Lutz et al., 2006; Falktoft et al., 2009), thereby suggesting their role in immature nervous tissue. N-terminal splice variants were detected in different brain areas. The PAC1-short variant with 21 amino acids N-terminal deletion was predominantly found in the thalamus, hypothalamus, and the hypophysis, and more moderately in the amygdala and retina (Dautzenberg et al., 1999; Grimaldi and Cavallaro, 1999; Jamen et al., 2002; Girard et al., 2004; Lutz et al., 2006; Ushiyama et al., 2007, 2010; Hammack et al., 2010a). It was shown to be expressed in cochlea subfractions

along with the PAC1-δ5 isoform with a seven amino acid deletion (Abu-Hamdan et al., 2006). The PAC1-very short variant (57 amino acid deletion) was also detected at low levels in the mouse amygdala and cortex though not in the retina. All three N-terminal variants (null, short, and very short) were abundantly expressed in neuronal tissues (Dickson and Finlayson, 2009; Vaudry et al., 2009). The PAC1-3a variant containing N-terminal insertion of 24 amino acids was detected in rat cerebral cortex. This splice isoform is also highly expressed in distinct cell populations of the testis, including Sertoli cells, pachytene spermatocytes, and round spermatids (Daniel et al., 2001;Ajpru et al., 2002; Pilzer and Gozes, 2006a).

Contrary to the aforementioned N-terminal splice variants that were only reported in mammals, IC loops (i.e., IC3) splice products are found in all studied vertebrate species. In the postnatal cerebral cortex of rats, the expression of PAC1-null, hop1, hip, and hip-hop1 dramatically decreases during the first neonatal month suggesting a major role for these isoforms in embryonic development (Shneider et al., 2010). Interestingly, the expression of these isoforms was shown to be gender dependent displaying higher levels in female postnatal brain at least during the first 3 months of development. In the developing retina, the proportion of IC3 PAC1 isoforms changes as development proceeds: PAC1-hip-hop1 transcript demonstrated transient elevation at day P10, while a decrease in PAC1-null and hip along with elevation of PAC1 hop1 level was observed by P20 (Lakk et al., 2012). PAC1-hop1 expression was found in the olfactory bulb, hippocampus, cerebral cortex, cerebellum, and striatum of newborn rats. PAC1-hop2 was present in rat cerebral astrocytes from newborn brains,in neuronal enriched cultures, and in PC12 cells that undergo neuronal differentiation following NGF treatment (Jamen et al., 2002; Onoue et al., 2002; Hashimoto et al., 2003; Ravni et al., 2006; Mustafa et al., 2010).

PAC1-hip was detected at much lower levels in adult tissues and it was therefore speculated to be important for the embryonic development (Shneider et al., 2010; Holighaus et al., 2011). The relative expression of human and rat PAC1 splice variants in the frontal cortex was reported to have the following expression level hierarchy: null > SV1/hip > SV2/hop = SV3/hiphop (Pisegna and Wank, 1996; Pisegna et al., 1996; Germano et al., 2004; Lutz et al., 2006). PAC1-hop1 was found to be expressed in all neuroendocrine cells, suggesting a fine tuning of PAC1-mediated signaling in the neuroendocrine cells (Mustafa et al., 2010). PAC1 hop1 splice isoform was the major form expressed in the superior cervical ganglia (SCG) sympathetic neurons, which also express PAC1-short (Braas and May, 1999). Zebrafish IC3 splice isoforms are widely expressed in the adult tissues with PAC1-hop1 detected in brain and testis, PAC1-hop2 in the ovary and PAC1-skip variant in the gills (Fradinger et al., 2005). Lastly, although PAC1-TM4 isoform is mainly found in pancreatic β-cells it is also expressed in the cerebral cortex, cerebellum, and brain stem (Chatterjee et al., 1996; Ajpru et al., 2002).

In summary, changes in PAC1 receptor variant expression appear to be most evident during development. The fact that PAC1 splice variants display differential expression patterns in the nervous system suggests that alternative splicing of this gene product plays a role in fine tuning of PAC1 activity in these areas.

# **REGULATION OF PAC1 SPLICING BY NEURONAL-SPECIFIC RNA-BINDING PROTEINS**

The regulation of PAC1 splicing is not well understood. Alternative splicing of pre-mRNA is often regulated by conserved cis-regulatory RNA sequence elements that serve as recognition sites for different splicing factors. These RNA-binding splicing factors differ from the general spliceosome machinery proteins due to the formers' ability to either activate or repress the inclusion of alternatively spliced exons, depending on whether the factors' binding site is located upstream, downstream, or inside the exon. Neuronal-specific splicing factors include nova-1/2, Rbfox-1/2, and ELKV2 splicing factors, the neuronal-specific polypyrimidine

tract-binding protein (nPTB) as well as a set of heterogeneous nuclear ribonucleoproteins (hnRNP A1,L,F.H1). As the consensus cis-acting binding sequences for the aforementioned splicing factors have been identified, their involvement in regulating the alternative splicing of a given gene by sequence analysis can be predicted.We analyzed the presence of known consensus binding sites for thefollowing RNA-binding proteins: Rbfox-1/2,Nova-1/2 (single and repeatedly organized elements), SC-35, nPTB enhancers and silencers, and CaRRE for Ca2+/CaM kinase IV recognition sequences. Putative binding sites for neuronal-specific splicing factors in the respective PAC1 genes of human, mouse, and zebrafish are shown in **Figure 5**.


**FIGURE 5 | Regulation of PAC1 splicing**. **(A)** A chart depicting the predicted binding sites for neuronal-specific RNA-binding proteins that potentially regulate PAC1 gene splicing by binding in proximity to PAC1 exons. Consensus cis-acting binding DNA elements for the respective RNA-binding protein are depicted with respect to their position in the PAC1 genes of rat, mouse, human, and zebrafish

PAC1. We analyzed the presence of putative consensus binding sites located within 300 base-pairs upstream (upst) or downstream (dnst) and inside the exons. **(B)** Schematic representation of the PAC1 gene in which the location of the analyzed cis-acting elements is color-coded as above. Corresponding exons numbers and sizes (bp) are indicated.

Rbfox-1/2 can act either as a splicing enhancer or as an inducer of exon skipping (Zhang et al., 2008). Putative Rbfox-1/2 binding sites located within a short distance downstream of target exons are predicted to induce exon inclusion whereas upstream binding sites are predicted to cause exon skipping. Such Rbfox-1/2 recognition element located within a short distance downstream of hop1 encoding exon has been experimentally validated as a genuine Rbfox-1 binding site (Lee et al., 2009). Our own PAC1 bioinformatic analysis identified a putative Rbfox-1/2 binding element within −30 nucleotides downstream from conserved exon spanning zebrafish TM7 suggesting that this domain might be regulated by alternative splicing (**Figure 5**). Rbfox-1/2 motifs are also present in the sequences of all examined species downstream of exon 16 encoding parts of the TM7 domain. Another recognition site is located upstream of exon 4 consistent with a deletion of the exon resulting in the generation of PAC1-very short splice variant (**Figure 5**). Rbfox-1/2 recognition motif predicts potential deletion of exon 8 that may result in a generation of a soluble PAC1 receptor as in the case of the CRH receptor (Zmijewski and Slominski, 2009, 2010). Rbfox-1/2 binding site is also detected downstream of exon 11, which is known to encode the most hypervariable amino acid sequence among Secretin family GPCRs, as well as part of TM4 domain which undergoes alternative splicing in PAC1-TM4 isoform (Markovic and Grammatopoulos, 2009).

Calcium signaling plays an important role in neuronal development and is involved in the regulation of immediate and long-term neuronal responses to various stimuli such as stressors and hormones (Ghosh et al., 1994; West et al., 2001). Depolarization is known to be critical for modulating the neuronal activity that induces Ca2+-dependent gene regulation, including alternative splicing. Ca2+-dependent splicing is mediated by L-type calcium channels and by Ca2+/calmodulin-dependent protein kinases IV (CaMK IV) (Lee et al., 2007, 2009). The latter was shown to repress splicing of target genes containing specific recognition elements (CaRRE) located within the 3<sup>0</sup> splice site or inside the exon. Our own analysis detected additional putative CaRREs that can be responsive to CaMK IV. We predicted single, multiple, and tandemly organized CaRRE repeats upstream of exon 15 and 17 of the zebrafish, human, and mouse PAC1 receptors, suggesting the involvement of CaMK IV in regulating the alternative splicing of these exons. Putative CaRREs are also present proximal to mouse, rat, and zebrafish exon 4 (**Figure 5**).

Exonic and intronic hnRNPA1 and nPTB cis-elements that are important for activity-dependent splicing in neuronal cells were shown to repress the inclusion of target exons (Allemand et al., 2005; An and Grabowski, 2007; Donev et al., 2007; Resnick et al., 2008; Zheng et al., 2012). Putative hnRNPA1 and nPTB recognition sites are located proximal to exons 4, 5, and 6 consistent with the known deletions of these exons that result in the generation of PAC1-short and very short splice variants. Putative hnRNPA1 recognition motifs are also found near exons 8–10 encoding to IC1, IC2, and EC2 domains (**Figure 5**).

Nova-1/2 regulates target exon inclusion or skipping with broad distribution of binding sites across the gene sequence (Ule et al., 2005; Yano et al., 2010; Norris and Calarco, 2012). Predicted

Nova-1/2 recognition elements are found in human and mouse PAC1 and in both paralogs of zebrafish PAC1 gene (1a and 1b). Interestingly, higher frequency of putative splicing-inducing AUrich and Nova-specific sequences are found in intronic regions upstream of alternative spliced exons characteristic of the Secretin family splicing patterns, e.g., around exon 4, 8, 9, 10 (Zmijewski and Slominski, 2009) as well as exon 17. Exons 4–6 encode sequences covering known N-terminal PAC1 deletion regions (deletions of 7, 21, and 57 amino acids). Exon 17 encodes the last amino acids of the seventh TM domain and the beginning of the IC tail shown to represent the highly conserved exon within the secretin family (Markovic and Grammatopoulos, 2009). Skipping of this exon was not detected in the PAC1 receptor gene, but it has been proposed that such exon skipping in the human CRF1 and PTH1 receptors results in a six TM receptor displaying impaired trafficking (Shyu et al., 1996; Grammatopoulos et al., 1999; Markovic et al., 2008). The rabbit calcitonin receptor is also formed as a result of the skipping of the TM7 domain and displays slightly diminished cAMP activation along with deficiency in IP3 induction (Shyu et al., 1996). Taken together the above analysis of potential cis-elements spanning TM7 exons predicts that a PAC1 splice variant containing TM7 skipping may exist in neuronal tissues.

An important question raised in this chapter is whether the aforementioned RNA-binding factors regulate PAC1 splicing *in vivo*. It has been demonstrated that alternative splicing of the PAC1 receptor is regulated by the Rbfox-1 splicing factor in depolarized neurons. In this case, neuronal depolarization induced CaM-kinase dependent self-splicing of Rbfox-1, leading to the translocation of Rbfox-1/2 from the cytoplasm to the nucleus where Rbfox-1 could mediate the alternative splicing of neuronal-specific target genes, including PAC1. Interestingly, Rbfox-1 mRNA levels were shown to be regulated by the Orthopedia (Otp) homedomain-containing protein. The latter is a hypothalamic-specific transcriptional factor, which plays a role in hypothalamic neuronal-specification during development and in regulation of body homeostasis in the mature zebrafish brain (Acampora et al., 1999; Wang and Lufkin, 2000; Blechman et al., 2007; Ryu et al., 2007; Amir-Zilberstein et al., 2012; Fernandes et al., 2012). The involvement and regulation of other neuronal-specific RNA-binding factors in PAC1 splicing remains to be determined.

# **PLEIOTROPIC ROLES OF PAC1 SPLICE ISOFORMS IN THE NERVOUS SYSTEM**

The distribution of PAC1 and its ligands PACAP and VIP in a variety of cell types is reminiscent of the pleiotropic functions of PAC1 in development and physiology. PACAP is described in the literature as a hormone, neuropeptide, endocrine peptide, neurotransmitter, and neurotrophic factor. It affects the central nervous system (CNS), cardiovascular system, pituitary, thyroid, and adrenal glands and providesfunctional activities in the gonads, gastrointestinal tract, and pancreas (Waschek et al., 2000; Dickson and Finlayson, 2009; Vaudry et al., 2009). In the nervous system, PAC1 was shown to affect a variety of hormones and neuropeptides, including stimulation of oxytocin (Jamen et al., 2003) and melatonin (Nakahara et al., 2002) release and *de novo*

mRNA synthesis of CRH (Nakahara et al., 2002; Amir-Zilberstein et al., 2012), arginine-vasopressin (Murase et al., 1995; Gillard et al., 2006), GnRH (Kanasaki et al., 2009; Winters and Moore, 2011), somatostatin (Kageyama et al., 2007), and MSH (Mounien et al., 2006). This overabundance of PAC1-mediated physiological activities can be made possible through the signaling diversity of its alternatively spliced gene products. Moreover, different expression of PAC1 isoforms is common in neuronal ontogeny. Such differences in the expression of PAC1 splice variants might modulate final outputs of VIP and PACAP activities as neurotransmitters, neurotrophic, or differentiation factors. Examples for the involvement of PAC1 splice isoforms in mediating these activities are described below.

#### **NEUROGENESIS, NEUROPROTECTION, AND DIFFERENTIATION**

Neural progenitor cells (NPC) and astroglial cells express PAC1 null and PAC1-hop1 variants that mediate both cAMP- and Ca2+ dependent signaling pathways and induce production of a subset of neurotrophic factors resulting in neuronal proliferation and/or differentiation. PAC1-hop2 variant, detected in astrogenic and neuronal populations, promote neuroprotective function induced by VIP. In another study, VIP was also demonstrated to induce PAC1-hop2-mediated astrocytes neuroprotection against oxidative stress. Astrocytic expression of PAC1-hop2 isoform may therefore play a critical role in the NPC shift toward neuronal or astrocytes differentiation (Ashur-Fabian et al., 1997; Pilzer and Gozes, 2006b).

Expression of PAC1-null and PAC1-hop1 define region-specific neurogenesis in the CNS and peripheral nervous systems (Lu et al., 1998). Both variants are differently expressed in proliferating sympathetic (PAC1-hop1) and cortical precursors (PAC1-null) revealing opposing PACAP-mediated mitogenic regulation – either by stimulating sympathetic neuroblast proliferation or by inhibiting cortical precursor mitosis. Ectopic expression of PAC1-hop1 in cortical neuroblasts may transform the anti-mitotic effect of PACAP into promitotic. This promitotic signaling was shown to involve PLC signaling pathway (Nicot and DiCicco-Bloom, 2001). PAC1-hop1 was also shown to promote sympathetic neurons survival following growth factor withdrawal (May et al., 2010).

In conjunction with the abovementioned, PACAP may possess therapeutic potential for neurodegenerative pathologies, such as Parkinson's and Alzheimer's disease (AD). PACAP is enriched in rat mesencephalic dopaminergic neurons and protects them from neurotoxin-induced death (Reglodi et al., 2004). PACAP was shown to be downregulated in several mouse models for AD as well as in the human temporal cortex of AD patients (Kojro et al., 2006; Rat et al., 2011; Postina, 2012). Amyloidogenic processing of the amyloid precursor protein (APP) to αβ-peptides is responsible for the development of AD. However, non-amyloidogenic APP processing pathway results in the α-secretase-dependent cleavage within the αβ-peptide region, preventing AD pathology. Continual PACAP and PAC1 activation resulted also in a feed-forward autocrine elevation of both PACAP and PAC1 in mice that may further facilitate non-amyloidogenic APP cleavage. α-secretase activation was shown to be regulated by ERK1 and ERK 2 and PI-3 kinase, suggesting that PAC1 hop1 is the most efficient isoform with regard to the activation of these downstream PLC effectors (Kojro et al., 2006; Rat et al., 2011; Postina, 2012). It remains to be determined whether specific PAC1 splice variants are involved in these neuroprotective activities.

#### **NEUROSECRETION AND NEUROTRANSMISSION**

Pituitary AC-activating polypeptide was shown to elicit catecholamine synthesis and release. Expression of PAC1-hop1 in bovine chromaffin NG108-15 cell line, which lacks endogenous PAC1 receptors, induces the release of norepinephrine via a Ca2<sup>+</sup> influx-dependent mechanism (Mustafa et al., 2007). Transfected PAC1-hop1 triggers sustained catecholamine secretion by regulating Ca2<sup>+</sup> levels through both ER and EC Ca2<sup>+</sup> channels (Mustafa et al., 2010; Smith and Eiden, 2012). Moreover, acute and longterm met-enkephalin secretion and enkephalin biosynthesis were attributed to bovine chromaffin cells upon PAC1-hop1-mediated activation of L-type Ca2<sup>+</sup> channels. This implies a role of hop1 splice cassette in regulating neuroendocrine secretion. SCG sympathetic neurons were shown to predominantly express the PAC1 hop1 splice isoform and PACAP stimulates neuropeptide Y release in these neurons via a mechanism involving both AC and PLCβ (Braas and May, 1999).

#### **ROLE OF PAC1 SPLICE ISOFORMS IN BODY HOMEOSTASIS**

PAC1 is implicated in the regulation of homeostatic processes, including food and liquid consumption (Nomura et al., 1997; Mounien et al., 2006, 2009), sleep (Hannibal and Fahrenkrug, 2004), stress (Pilzer and Gozes, 2006b; Amir-Zilberstein et al., 2012), locomotion (Vaudry et al., 2000), memory and learning activities (Dong et al., 2010; Andero and Ressler, 2012; Holighaus et al., 2012), and circadian functions (Ajpru et al., 2002; Hannibal and Fahrenkrug, 2004). Recent studies have indicated that at least some of these activities are modulated by alternative splicing of PAC1 (Ajpru et al., 2002; Hannibal and Fahrenkrug, 2004; Pilzer and Gozes, 2006b; Holighaus et al., 2011, 2012; Amir-Zilberstein et al., 2012).

Pituitary AC-activating polypeptide/PAC1 signaling was recently implicated in abnormal stress responses underlying posttraumatic stress disorder (PTSD) pathology (Ressler et al., 2011; Andero and Ressler, 2012; Hauger et al., 2012). PACAP38 levels in females were strongly associated with PTSD symptoms. Ressler et al. and May et al. (2010) found that a single nucleotide polymorphism (SNP), rs2267735 in the PAC1 gene is correlated with PTSD in a gender-specific estrogen-dependent manner. This SNP, which resides in the putative estrogen response element within PAC1's promoter/enhancer is associated with sexdependent PTSD, fear discrimination, PAC1 mRNA expression, and methylation of PAC1. Using mouse models, these authors showed that PAC1 mRNA was induced by fear conditioning or estrogen hormone replacement in the bed nucleus of stria terminalis (BNST), which is a component of the extended amygdala involved in fear- and anxiety-like responses.

In connection to the above, steroid-induced changes in PAC1 splice isoforms were demonstrated in the case of PAC1-hop2 (Apostolakis et al., 2005). PAC1-hop2 and the N-terminal PAC1 short splice variants were detected in the hypothalamic ventromedial nucleus (VMN) upon estradiol or dual estradiol/progesterone treatment. The ratio of PAC1-hop2 mRNA to other PAC1 variants in VMN depends on steroid application, implying its importance for cumulative PAC1 isoforms expression and hence signaling properties. PACAP-induced sexual/mating behavior in female rats was dependent on progesterone receptor in the VMN. Estradiol stimulates synthesis of progesterone, which in turn facilitates PACAP synthesis and activation of PAC1-short and -hop2 signaling critical for the induction of animal sexual receptivity.

PAC1 signaling is known to be required for physiological stress response as activation of PAC1 by PACAP is required for stress-induced CRH transcription *in vivo* and *in vitro* (Agarwal et al., 2005; Kageyama et al., 2007; Hammack et al., 2010b; Stroth and Eiden, 2010; Tsukiyama et al., 2011). The above studies suggest that PAC1 splice variants and their relative expression ratio might be involved in the regulation of body homeostasis including physiological responses to stressor challenges. A recent study performed in our lab, employed various stress paradigms in mouse and zebrafish to demonstrate that activation and termination of CRH transcription caused by stressful stimuli is regulated by an interrelationship of PAC1-null and hop1 isoforms (Amir-Zilberstein et al., 2012). Thus, the generation of PAC1-hop1 by alternative splicing leads to the termination of CRH transcription, normal activation of the hypothalamicpituitary-adrenal axis, and adaptive anxiety-like behavior (Amir-Zilberstein et al., 2012). We therefore suggested that alternative splicing of the hop1-cassette serves as an ON/OFF stress switch (**Figure 6**).

PAC1 rs2267735 gene polymorphism is also associated with increased dark-enhanced startle (DES) in adult females but not males with PTSD (Ressler et al., 2011). Moreover, children of abused mothers show elevated DES and the same PAC1 gene polymorphism associated with PTSD risk in adult females is also associated with increased DES in these children (Jovanovic et al., 2011, 2012). Notably, zebrafish larvae show a strong aversion to the dark side of a two-compartment light/dark arena and this place preference can be mitigated by anxiolytic drugs, such as Diazepam, indicating that this assay measures an anxiety-like behavior (Steenbergen et al., 2011; Schnorr et al., 2012). We found that during the recovery phase that follows a stressful challenge (osmotic shock), wild type larvae decrease their dark-avoidance time, a phenotype indicative of decreased stress-related anxiety, while larvae with depleted PAC1a-hop1 display delayed dark-avoidance recovery (Amir-Zilberstein et al., 2012). Thus, the delayed behavioral response of PAC1a-hop1-depleted embryos correlates with their respective failure to terminate *crh* and cortisol levels following stressors.

# **CONCLUSION**

Alternative splicing is a major gene regulatory process involving cis- and trans-acting factors. PAC1 signaling controls a variety of cellular and physiological responses, such as differentiation, proliferation, cell cycle regulation, neurotransmitter, and hormone release and adaptation to stressful challenges. The PAC1 gene encompasses a relatively long genomic region, which consists of up to 18 exons and contains many putative splicing

altered exon (hop1) encoding to 28 amino acids of the third intracellular loop leading to the formation of the PAC1-hop1 splice variant. Generation of the PAC1-hop isoform terminates stress response by means yet to be uncovered (see text and Amir-Zilberstein et al., 2012).

factors recognition sites that might be activated during different phases of neuronal activation. PAC1 receptor signaling can be fine-tuned by the generation of a set of alternatively spliced variants produced in a spatio-temporal manner. Splicingdependent alterations in PAC1 protein domains modify its ligand binding and signaling properties leading to a range of cellular activities. Although the physiological function(s) of the vast majority of the alternatively spliced PAC1 gene products is still unknown, recent studies have implicated certain PAC1 splice variants in the regulation of homeostatic processes such as adaptation to stressful challenges. In view of the recent association of PAC1 with PTSD risk, the regulation of PAC1 splicing and its underlying physiological outcomes might prove to be relevant to the etiology of some neurological and psychiatric disorders.

#### **ACKNOWLEDGMENTS**

We thank Shifra Ben-Or for PAC1 gene structure and phylogenetic analyses and Danielle Sabah-Israel, Jacob Biran, and Michael Gliksberg for comments on the manuscript. The research in the Levkowitz lab is supported by the Simons Foundation Autism Research Initiative (SFARI); Israel Science Foundation; Kirk Center for Childhood Cancer and Immunological Disorders; The Krenter Institute and Estate of Lore Lennon.

# **REFERENCES**


modulate neuronal adaptation to stress. *Neuron* 73, 279–291.


PACAP in the cerebellar cortex. *Peptides* 28, 1746–1752.


the pituitary adenylate cyclaseactivating polypeptide (PACAP) receptor that stimulates calcium influx by activation of L-type calcium channels. *J. Biol. Chem.* 271, 32226–32232.


amyloid precursor protein isoforms. *Mol. Psychiatry* 12, 681–690.


receptor function. *Br. J. Pharmacol.* 166, 98–109.


and NGF regulate common and distinct traits of the sympathoadrenal lineage: effects on electrical properties, gene markers and transcription factors in differentiating PC12 cells. *Eur. J. Neurosci.* 17, 71–82.


calcium leading to splice variantspecific gene induction in neural cells. *Peptides* 32, 1647–1655.


receptor (ADCYAP1R1) genotype is associated with dark-enhanced startle in children. *Mol. Psychiatry*. doi:10.1038/mp.2012.98. [Epub ahead of print].


Pituitary adenylate cyclaseactivating polypeptide (PACAP) stimulates endozepine release from cultured rat astrocytes via a PKAdependent mechanism. *FASEB J.* 17, 17–27.


pituitary cells. *Biochem. Biophys. Res. Commun.* 164, 567–574.


in the circadian oscillator, in chick pineal cells. *Brain Res.* 939, 19–25.


human fetal adrenal gland chromaffin cells. *J. Biol. Chem.* 278, 1663–1670.


deficits in a rat model of Parkinson's disease. *Behav. Brain Res.* 151, 303–312.


to function of PACAP-27. *J. Mol. Neurosci.* 42, 341–348.


(2000). PACAP action in nervous system development, regeneration, and neuroblastoma cell proliferation. *Ann. N. Y. Acad. Sci.* 921, 129–136.


protein-coupled receptors in basic cell biology. *Mol. Biosyst.* 9, 586–595.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 25 November 2012; accepted: 24 April 2013; published online: 21 May 2013.*

*Citation: Blechman J and Levkowitz G (2013) Alternative splicing of the pituitary adenylate cyclase-activating polypeptide receptor PAC1: mechanisms of fine tuning of brain activity. Front. Endocrinol. 4:55. doi: 10.3389/fendo.2013.00055*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Blechman and Levkowitz. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, providedthe original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Role of PACAP in female fertility and reproduction at gonadal level – recent advances

#### **Dora Reglodi <sup>1</sup>\*, Andrea Tamas <sup>1</sup> , Miklos Koppan<sup>2</sup> , Donat Szogyi <sup>2</sup> and LauraWelke<sup>3</sup>**

<sup>1</sup> Department of Anatomy, Lendulet PACAP-Research Team of the University of Pécs and Hungarian Academy of Sciences, Pécs, Hungary

<sup>2</sup> Department of Obstetrics and Gynaecology, University of Pécs, Pécs, Hungary

<sup>3</sup> Department of Anatomy, Ross University School of Medicine, Roseau, Commonwealth of Dominica

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Rita Canipari, University of Rome – La Sapienza, Italy Khampoun Sayasith, University of Montreal, Canada

#### **\*Correspondence:**

Dora Reglodi, Department of Anatomy, University of Pécs, Szigeti u 12, 7624 Pécs, Hungary. e-mail: dora.reglodi@aok.pte.hu

Pituitary adenylate cyclase activating polypeptide (PACAP) is a pleiotropic neuropeptide, first isolated from hypothalamic extracts, but later shown in peripheral organs, such as endocrine glands, gastrointestinal system, cardiovascular system, and reproductive organs. PACAP plays a role in fertility and reproduction. Numerous studies report on the gonadal regulatory effects of PACAP at hypothalamo-hypophyseal levels. However, the local effects of PACAP at gonadal levels are also important.The present review summarizes the effects of PACAP in the ovary. PACAP and its receptors are present in the ovary, and PACAP plays a role in germ cell migration, meiotic division, follicular development, and atresia. The autocrineparacrine hormonal effects seem to play a regulatory role in ovulation, luteinization, and follicular atrophy. Altogether, PACAP belongs to the ovarian regulatory peptides.

**Keywords: PACAP, ovary, oocyte, ovulation, luteinization**

#### **INTRODUCTION**

Pituitary adenylate cyclase activating polypeptide (PACAP) was originally isolated from the hypothalamus, and named after its cAMP increasing effect in pituitary cells (Miyata et al., 1989; Arimura, 2007). PACAP belongs to the vasoactive intestinal peptide (VIP)/secretin/glucagon peptide family. It occurs in two amino acid forms: PACAP38 and PACAP27, the longer form is the predominant peptide in mammals. The receptors of PACAP are the VPAC1 and VPAC2 receptors, which bind VIP and PACAP with equal affinity and the specific PAC1 receptor, which binds PACAP selectively (Miyata et al., 1989; Sherwood et al., 2000; Arimura, 2007; Vaudry et al., 2009; Brubel et al., 2010). After the discovery of PACAP, numerous studies described the regulatory effects of PACAP at hypothalamo-pituitary levels, affecting the synthesis and release of several releasing and pituitary hormones. However, soon after its discovery, it was found that PACAP is, and its receptors are, expressed in the peripheral organs, and high levels of the peptide were found in the gonads (Arimura et al., 1991). This immediately drew attention to the peptide playing a central role in fertility and reproduction, not only at central hormonal, but also at local levels.

The last two decades in PACAP research have revealed that PACAP influences fertility and reproduction at several levels. The present mini-review aims to summarize findings on the effects of PACAP in the ovary and how PACAP influences female gonadal functions at the ovarian level.

#### **PACAP IN MAMMALIAN FOLLICULAR DEVELOPMENT**

Primordial germ cells (oogonia) migrate from the wall of the yolk sac to the ovaries during early mammalian development. Ovarian germ cells then undergo mitosis and develop further into primary oocytes. Developing oocytes are surrounded by epithelial cells, which build the ovarian follicles. At first, follicular epithelial cells are flattened and surround the primary oocytes entering the first meiotic division. They subsequently are arrested during fetal development until the female reaches sexual maturity. The first meiotic division is only completed during the follicular development before ovulation in each cycle, and the second meiotic division is completed only at the time of fertilization, resulting in the haploid gamete (Sadler, 2012).

A cohort of primordial follicles is activated at the beginning of the maturation process in cyclic manner. Follicular maturation involves changes in the oocyte, the follicular epithelial cells, and the surrounding ovarian stroma. Although significant inter-species differences exist in the follicular growth and its regulation, the major steps show basic similarities in mammalian species, and in most mammals, the vast majority of follicles undergo apoptosis similarly to humans (Matsuda et al., 2012). Therefore, we will focus on mammalian findings and only briefly mention other species. Early follicular development is independent from gonadotropins, but later follicular growth, ovulation, and luteinization are primarily regulated by the pituitary gonadotropins follicle-stimulating hormone (FSH) and luteinizing hormone (LH). However, their actions are also dependent on other peptidergic and non-peptidergic signaling pathways (Mayerhofer et al., 1997; Bodis et al., 2001; Matsuda et al., 2012). Accumulating evidence shows that PACAP is one of the regulatory peptides in the ovary.

#### **ACTIONS OF PACAP ON PRIMORDIAL GERM CELLS**

There is evidence for the involvement of PACAP in the proliferation of primordial germ cells (Pesce et al., 1996). The authors showed PACAP immunoreactivity in gonadal ridges, mostly on the germ cell surface. Primordial germ cells intensively proliferate during their migration to the gonadal ridges and PACAP was found to stimulate this proliferation process. The peptide binds to the germ cells and also to gonadal somatic cells (Pesce et al., 1996). These observations show that PACAP plays an important role in reproductive functions from the earliest stage of differentiation and development.

#### **PACAP IN THE OVARY**

In rat ovarian cells, PACAP positive cells were found in the preovulatory and ovulatory periods (Gras et al., 1996), with PACAP and PACAP mRNA positivity in granulosa and cumulus cells. Positivity was also found in stroma cells and theca cells and even nerve fibers innervating the ovary showed immunopositivity. These observations suggest that PACAP is involved in a number of processes regulated via hormonal as well as neuronal activity during follicular development in the ovary (Gras et al., 1996). Another group published on the existence of PACAP in the rat ovary the same year (Scaldaferri et al., 1996).

Pituitary adenylate cyclase activating polypeptide expression in the corpus luteum of the rat was subsequently shown by Kotani et al. (1997, 1998). Park et al. (2001) showed that PACAP mRNA was only expressed in the granulosa but not theca cells. This expression has a stage-specific regulation by GnRH in rat granulosa cells. While treatment of preovulatory granulosa cells with GnRH agonist stimulates PACAP mRNA expression, GnRH agonist treatment alone has no effect but reduces the FSH-induced PACAP mRNA levels. GnRH antagonist on the other hand had opposite effects: it inhibited induced PACAP gene expression in preovulatory cells while stimulating expression in immature granulosa cells. These results show that GnRH plays an important role in the regulation of ovarian PACAP expression and that PACAP expression is differentially regulated during the ovarian cycle (Park et al., 2001). The LH and FSH control of the PACAP expression was also confirmed by Lee et al. (1999). The involvement of progesterone receptors have also been shown in LH-induced PACAP gene expression (Ko et al., 1999). In addition, the gonadotropindependent regulation of PACAP mRNA has been confirmed in bovine preovulatory follicles (Sayasith et al., 2007). PACAP was found to be expressed in granulosa cells and its expression was found to be dependent on protein kinase A. A recent study has shown that PACAP mRNA expression gradually increases in pregnancy in the corpus luteum, suggesting its involvement in the maintenance of mid-term and late pregnancy (Zhao et al., 2010).

Transient expression of PACAP has also been observed in the mouse ovary within granulosa cells of preovulatory follicles after hCG treatment (Barberi et al., 2007). The contemporary induction of PACAP in preovulatory follicles suggests an important role for PACAP around the time of ovulation (Barberi et al., 2007). In human granulosa cells, it has been found that FSH and LH promote PACAP expression (Morelli et al., 2008). PACAP is synthesized as a preprohormone and is processed by prohormone convertases. It has been shown that in both testis and ovary, prohormone convertase 4 is the enzyme responsible for PACAP processing, supported by the finding that prohormone convertase 4 deficient mice expressed no PACAP in their ovaries (Li et al., 2000).

#### **EXPRESSION OF PACAP RECEPTORS IN THE OVARY**

Earlier studies examining binding sites for PACAP have already identified specific binding sites in the ovary (Gottschall et al., 1990). Scaldaferri et al. (1996) identified isoforms of the PAC1 receptor in the rat ovary. Later, PAC1 receptors were shown in the rat corpus luteum (Kotani et al., 1997, 1998). Park et al. (2000) have demonstrated stage-specific expression of PACAP receptors in the rat ovary. Northern blot analysis shows that the ovarian transcript of the PAC1 receptor appears at day three in the rat ovary, followed by a gradual increase later in development. There is a marked increase at puberty, at the age of 21 days, when compared to an age of 15 days, when only non-growing small follicles are present. *In situ* hybridization has revealed that PACAP receptor mRNA is present mainly in the granulosa cells of the large preantral follicles, while atretic follicles and mural granulosa cells are devoid of the receptor. The authors have also demonstrated that the dominant splice variant of the PAC1 receptor was the short variant in the ovary (Park et al., 2000). Gonadotropin stimulation has been shown to induce PACAP receptor mRNA expression in the granulosa cells of the preovulatory follicles (Ko and Park-Sarge, 2000; Park et al., 2000). Pregnant mare serum gonadotropin (PMSG), on the other hand, causes reduction of PACAP receptor gene expression. PMSG induces multiple follicular growth to the preovulatory stage. These observations suggest an involvement of PACAP in the follicular growth, and in ovulation, in a stage- and time-dependent manner. The observation that PACAP receptor expression is restricted to granulosa cells of the growing follicles at the time of puberty and to granulosa cells of preovulatory follicles after gonadotropin treatment indicates that PACAP may act in a limited time-window in the ovary. Progesterone receptors have been shown to be involved in the induced PAC1 receptor expression (Ko and Park-Sarge, 2000).

A more detailed analysis of the rat ovarian PACAP receptor expression has confirmed the expression of PAC1 receptors in the granulosa cells and, furthermore, the presence of VPAC2 receptors in these cells (Vaccari et al., 2006). Theca cells do not express PAC1 receptors, only VPAC1 and 2 receptors. Fully developed oocytes only express the PAC1 receptor. hCG stimulation has been found to induce PAC1 receptor expression in granulosa andVPAC2 receptor expression in theca cells. The VPAC receptor expression has been found to have a lower expression level than the PAC1 receptor. This study has also confirmed the previous findings of Park et al. (2000) describing receptor expression at 3 days after birth with a marked increase before puberty. In addition, they found that VPAC1 receptors decreased with age and VPAC2 receptors remained constant. Immunohistochemical analysis revealed the presence of VPAC1 receptors in association with stromal blood vessels in the vicinity of the follicles, especially at the entrance site of the ovarian arteries into the medulla. The expression of the VPAC2 receptors was more ubiquitous in the ovary. Denuded oocytes express only the PAC1 receptor, which could not be detected in Met-I and II phases in oocytes matured *in vivo*. Gras et al. (2000) also described PAC1 and VPAC2 receptors in preovulatory follicles. All three types of the PACAP receptors have been identified in human granulosa cells, although the main receptor types in these cells seem to be the VPAC receptors (Morelli et al., 2008).

All three receptors have also been identified in the mouse (Barberi et al., 2007). PAC1 receptors could be stimulated by hCG, VPAC2 only mildly stimulated, and VPAC1 downregulated. The authors suggested that the contemporary induction of PACAP

and PAC1 receptors by hCG in granulosa cells of preovulatory follicles indicates an important role for PACAP around the time of ovulation (Barberi et al., 2007).

#### **ACTIONS OF PACAP IN THE OVARY**

#### **Follicular development, hormone production, and ovulation**

In all mammalian species, the early stages of folliculogenesis, including the initiation of primordial follicle growth, are independent from gonadotropins (Latini et al., 2010). Fine balances between inhibitory and stimulatory signals regulate follicular recruitment. PACAP has been shown to inhibit primordial to primary follicle transition (Latini et al., 2010); PACAP did not influence granulosa cell viability at these stages but inhibited proliferation. PACAP also inhibited the growth of preantral follicles. These observations indicate an important role of PACAP in follicular recruitment (Latini et al., 2010). On the other hand, at later stages, PACAP can prime immature follicles (Gras et al., 2005). A 12 h PACAP priming stimulates FSH-induced estradiol production, which is an important step in cyclic recruitment, when a cohort of antral follicles escapes apoptosis and reaches the preovulatory stage (Gras et al., 2005). These results show a fine balance between factors, including PACAP, that play a role in follicular recruitment. Other studies have also shown that PACAP plays an important role in preantral follicular growth and differentiation (Cecconi et al., 2004). In the mouse ovary, PACAP or VIP alone did not affect follicular growth, but they both inhibited it when added to FSH-stimulated follicles. Both peptides caused a dose-dependent inhibition of follicle growth, antrum formation, granulosa cell proliferation, and estradiol production (Cecconi et al., 2004).

Studies on the endocrine effects of PACAP revealed that PACAP can stimulate cAMP accumulation and steroidogenesis in the rat ovary (Zhong and Kasson, 1994; Heindel et al., 1996;Vaccari et al., 2006). Granulosa cell cultures responded to PACAP treatment with increased production of estrogen, progesterone, and 20alphadihydroprogesterone. This effect was dose-dependent and more potent than similar effects of VIP and GHRH (Zhong and Kasson, 1994). The peptide was also able to augment FSH-induced progesterone and 20alpha-dihydroprogesterone accumulation, and in high doses that of estrogen. The PACAP-stimulated progesterone accumulation was minimal in the absence of androstendion and dramatically augmented in the presence of the androgen. This shows that androgens are required for the PACAP to stimulate steroid production and indicate that PACAP might play a role in modulating the production of steroids in the ovary (Zhong and Kasson, 1994). The effects of PACAP depend on the presence of LH. In cultured luteal cells of the rat,PACAP alone stimulates progesterone, 20alpha hydroxyl-4-pregnene-3-one along with cAMP accumulation (Usuki and Kotani, 2001, 2002). However, when LH is present, PACAP in high concentrations inhibits the LHinduced cAMP accumulation and progesterone production, while it enhances the LH-induced stimulation of 20alpha hydroxyl-4 pregnene-3-one production (Kotani et al., 1998). This decreases the ratio of progesterone to 20alpha hydroxyl-4-pregnene-3-one in LH-stimulated cells, which suggests an involvement in luteolysis. This is supported by the finding that PACAP suppresses increases in LH receptors in luteal cells (Usuki and Kotani, 2001). Gras et al. (2000) showed that PACAP can stimulate progesterone

production and cAMP synthesis, and the involvement of the PKA pathway. These data together provide evidence for the involvement of PACAP as an autocrine-paracrine regulator of ovarian hormone production, especially in preovulatory follicles (Gras et al., 2000).

Interestingly, one intracerebroventricular PACAP treatment was shown to inhibit ovulation in adult animals (Koves et al., 1998). Injecting PACAP before the critical period of the proestrous stage, blocks the ovulation and prevents the proestrous LH surge (Koves et al., 1996). Furthermore, neonatal administration of PACAP delays onset of puberty in female animals (Szabo et al., 2002).

The effects of PACAP on ovarian steroidogenesis have been confirmed in human studies as well (Apa et al., 1997), where PACAP was found to stimulate progesterone synthesis without synergistic effects with hCG. This was tested in corpora lutea obtained from non-pregnant women undergoing hysterectomy.

PLC was activated by PACAP in granulosa but not theca cells (Vaccari et al., 2006). Plasminogen activators are known to be involved also in ovulation. PACAP has been shown to increase the tissue-type plasminogen activator and to decrease the urokinasetype one (Apa et al.,2002).VIP did not show this effect in granulosa cells, but when entire follicles were exposed to the peptides, both VIP and PACAP exerted similar stimulatory effects on tissue-type plasminogen activator levels, supporting the observations of different PACAP/VIP receptors being expressed in different ovarian compartments (Apa et al., 2002).

#### **Apoptosis**

In addition to gonadotropins, factors secreted from granulosa cells, including steroids, growth factors, and cytokines, are essential for granulosa cell survival and follicular growth (Matsuda et al., 2012). Granulosa cells are the initial population to undergo apoptosis in atretic follicles, indicating their role in the initiation of follicular atresia (Matsuda et al., 2012). Besides estradiol, as the main ovarian steroid influencing follicular apoptosis, several other factors play an important role. Among those, insulin-like growth factor, epidermal growth factor, fibroblast-like growth factor, and interleukin-1 beta have prosurvival effects, while the Fas ligand system and the tumor necrosis alpha are proapoptotic. Pro- and antiapoptotic members of the mitochondrial Bcl family have also been described to play an important role (Matsuda et al., 2012). A few papers suggest that PACAP is also one of the factors influencing granulosa cell death and survival.

One of the main effects of PACAP is its cell survival-promoting effects, which was first described in neuronal cells, and thus, PACAP was designated as an important neuroprotective peptide (Vaudry et al., 2002; Somogyvari-Vigh and Reglodi, 2004). However, numerous studies have revealed that the antiapoptotic effects of PACAP are not restricted to neuronal cells, but can be shown in several other cell types, such as lymphocytes (Delgado and Ganea, 2000), prostate cancer cells (Gutierrez-Canas et al., 2003), endothelial cells (Racz et al., 2007), retinal pigment epithelial cells (Fabian et al., 2012), and even in the invertebrate mollusks (Pirger et al., 2008). Therefore, PACAP can be considered as a general cytoprotective peptide. VIP, structurally the closest peptide to PACAP has been described to have antiapoptotic effects in the ovary (Flaws et al., 1995).

Based on these data, it is expected that PACAP can also exert antiapoptotic effects in ovarian cells. Indeed, several studies have shown this in granulosa cells. In isolated granulosa cells, apoptosis can be induced by serum-free culturing. In accordance with the generally known antiapoptotic role of PACAP, the peptide could reduce the granulosa cell apoptosis in a dose-dependent manner, an effect that could be blocked by the PACAP antagonist PACAP6- 38 (Vaccari et al., 2006). These results are in accordance with earlier observations showing that PACAP could block the apoptotic cascade in granulosa cells (Lee et al., 1999) and that the LH suppression of follicle apoptosis was partially blocked by PACAP 6-38. The antiapoptotic effect of PACAP was also confirmed in human granulosa-luteal cells (Morelli et al., 2008). Both PACAP and VIP could reverse the decrease in procaspase-3 induced by the serum withdrawal. However, it was found that spontaneous apoptosis was not influenced by PACAP in another study (Gras et al., 2005).

Expression of PACAP in both benign and malignant ovarian tumors has been shown (Odum and Fahrenkrug, 1998). Higher levels were found in cancers than benign tumors. The authors suggest a trophic function of PACAP in ovarian cells (Odum and Fahrenkrug, 1998). Interestingly, PACAP does not influence the cisplatin-induced toxicity in proliferating ovary cells of the CHO ovarian cell line (Aubert et al., 2008). The same treatment, however, does prevent toxicity and related apoptosis in cerebellar granule cells. This dissociation in the antiapoptotic behavior of PACAP could be of high clinical importance: PACAP can decrease the cisplatin-induced neurotoxic side effects while leaving the antitumor therapeutic effect intact (Aubert et al., 2008).

#### **Effects in oocytes**

Fully developed oocytes express PAC1 receptor and the addition of nanomolar concentrations of PACAP induces calcium release. However, this answer could not be observed in Met-I and II phase oocytes, supporting that finding that receptors are expressed stage-specifically (Vaccari et al., 2006).

In *Xenopus* oocytes, PACAP has been described to modulate membrane potential by eliciting hyperpolarization-activated chloride current, thereby affecting oocyte physiology (Kato et al., 1997).

Very interesting results have been described by Apa et al. (1997) supporting both a direct and indirect effect of PACAP on oocyte maturation. Mammalian oocytes are known to arrest in the first meiotic division, which is resumed at the time of the preovulatory LH surge. The inhibition of oocyte maturation and its relief is mediated by gonadotropins in conjunct with several other factors, while only a few are known to act directly on oocytes. The authors described that PACAP accelerated meiotic maturation in follicleand cumulus-enclosed oocytes while inhibiting meiotic maturation in denuded oocytes (Apa et al., 1997). This result was not due to a direct cytotoxic effect because the inhibition on oocyte maturation was reversible when PACAP was removed from the medium. This difference in PACAP action on enclosed and denuded oocytes support the stage-dependent regulatory effects of PACAP. Other studies have also confirmed the effect of PACAP on the meiotic processes. In the mouse ovary, Cecconi et al. (2004) found that PACAP severely impaired meiotic maturation in oocytes isolated from the follicles.

Recently, mass spectrometric and radioimmunoassay analysis have shown that PACAP is present in human follicular fluid obtained from patients undergoing hyperstimulation treatment (Brubel et al., 2011; Koppan et al., 2012). PACAP could be identified in all human samples examined. Correlation was found between retrieved oocytes and PACAP levels in the follicular fluid drawing the attention to PACAP as an important factor in the medium of the developing oocyte and its possible future use as a biomarker in women with fertility problems (Koppan et al., 2012).

# **EVOLUTIONARY PERSPECTIVES**

Recent reports point to the important roles of PACAP in fish reproduction, pointing to the fact that the effects of PACAP on the hypothalamo-pituitary axis, as well as in the gonads, are conserved and biologically ancient functions (Levy and Degani, 2011, 2012). In fish, stage-specific expression of PACAP has also been revealed. For example, higher PACAP expression can be found in female blue gourami, with oocytes in the final maturation stage, than in vitellogenic individuals. Also, higher expression was found in mature males that are not reproductively active than in nest builders and juveniles (Levy and Degani, 2012). PACAP is thus, differentially expressed in females and males. PACAP acts in close association with GnRH in fish reproduction and the expression and actions of PACAP also depend on environmental factors, such as temperature (Levy et al., 2011; Levy and Degani, 2012). It has been proposed that PACAP is involved in the final oocyte maturation in females and in males it is mainly associated with sexual behavior (Levy et al., 2010). In the zebrafish, a new type of PACAP has been described in the ovary and its role as an ovarian factor mediating gonadotropin action have been suggested (Wang et al., 2003).

# **ROLE OF ENDOGENOUS PACAP**

Given the important functions of PACAP in reproduction outlined above, it would be expected that the lack of endogenous PACAP have major effects on fertility and reproduction. Indeed, PACAP deficient mice have been shown to have lower reproductive rates in several studies (Shintani et al., 2002; Sherwood et al., 2007; Isaac and Sherwood, 2008). To study the exact mechanism, a few studies examined the possible background for this lower reproductive rate. Interestingly, Isaac and Sherwood (2008) found that these mice had normal puberty onset, estrous cycle, and seminal plugs when paired with males. However, the birth rate after mating was only 20% when compared to the 100% in wild type mice. The authors found no defect in ovulation, ovarian histology, or fertilization, but they found major defects in implantation and associated hormone levels (Isaac and Sherwood, 2008). Similarly, low birth rates were found by Shintani et al. (2002), but they also found reduced mating and maternal behaviors. It is possible that in the ovary, compensatory mechanisms counteract the lack of PACAP and thus, no major alterations can be detected under normal circumstances. However, more studies are necessary to elucidate the exact role of endogenous PACAP in the ovary. Furthermore, in several other organs, including intestine, kidney, retina, and central nervous system, PACAP deficient mice do not exhibit major alterations unchallenged conditions (Reglodi et al., 2012). However, in case of stressors or under pathological

conditions, PACAP deficient mice react with increased vulnerability and the observed deficits have been shown to be more severe in mice lacking endogenous PACAP (Reglodi et al., 2012). Based on these observations, it is possible that endogenous PACAP has ovarian functions not compensated under challenged or pathological conditions. Further studies are required to shed light on this possibility.

# **CONCLUDING REMARKS**

The present review summarized findings on the presence and effects of PACAP in the ovary. It has to be emphasized that this is only one role in the plethora of actions that PACAP plays in fertility and reproduction and that the detailed discussion of these is beyond the scope of the present review. Briefly, PACAP, at the hypothalamic level, influences receptive behavior in female rodents, in association with GnRH and steroids (Apostolakis et al., 2004, 2005), and plays an important modulatory role in pituitary hormone production. The role of PACAP in the hypothalamopituitary-gonadal axis has been reviewed several times previously (Rawlings and Hezareh, 1996; Nussdorfer and Malendowicz, 1998; Sherwood et al., 2000;Koves et al., 2003;Vaudry et al., 2006;Counis et al., 2007).

Pituitary adenylate cyclase activating polypeptide also plays a role in the muscle contraction of the vaginal wall as well as that of the uterus and uterine tube (Steenstrup et al., 1995; Ziessen et al., 2002), and even decreased immunoreactivity has been shown in premenopausal and postmenopausal women in the vaginal wall (Hong et al., 2008). The PACAPergic innervation of the female genital tract has also been shown and has been associated with nerves originating from the paracervical ganglia (Fahrenkrug and Hannibal, 1996; Fahrenkrug et al., 1996).

#### **REFERENCES**


rats. *Mol. Endocrinol.* 19, 2798–2811.


Effects of PACAP have also been shown in the placenta where PACAP and its receptors are present (Scaldaferri et al., 2000; Koh et al., 2005; Brubel et al., 2010), where the peptide influences blood flow in the utero-placental unit (Steenstrup et al., 1996) and influences survival of trophoblast cells (Boronkai et al., 2009). The serum level of PACAP increases in the third trimester of pregnancy in healthy pregnant women and it markedly decreases during delivery, reaching prebirth levels 3 days after delivery (Reglodi et al., 2010). High concentrations of PACAP have been shown in human and animal milk (Borzsei et al., 2009; Csanaky et al., 2012), which suggests that the peptide is also an important nutritional source for the newborn. However, its exact function in breastfeeding has not been identified yet.

In summary, PACAP is one of the peptides regulating germ cell development in the ovary and it has several other regulatory functions in reproduction. PACAP influences ovarian hormone production, affects meiosis, and is an important local regulator of follicular development. The exact physiological role and its possible clinical importance still awaits further investigation, but the current evidence points to a possible clinical diagnostic or therapeutic use of the peptide in both human and veterinary fertility.

# **ACKNOWLEDGMENTS**

This work was supported by Hungarian National Scientific Grants OTKA104984, CNK 78480, SROP 4.1.2.B-10/2/KONV-20/0-0002, SROP-4.2.2/B-10/1-2010-0029, Bolyai Scholarship, MTA Lendulet Program, TAMOP-4.2.2.A-11/1/KONV-2012- 0024, 4.2.2A-11/1konv-2012-0053, Akira Arimura Foundation, Ross University School of Medicine Research Fund, PTE AOK Research Fund AOK KA/34039-81-8000.

Influence of melatonin on basal and gonadotropin-stimulated progesterone and estradiol secretion of cultured human granulosa cells and in the superfused granulosa cell system. *Gynecol. Obstet. Invest.* 52, 198–202.


serum withdrawal. *Br. J. Pharmacol.* 139, 1050–1058.


(PACAP) on progestin biosynthesis in cultured granulosa cells from rat ovary and expression of mRNA encoding PACAP type I receptor. *J. Reprod. Fertil.* 112, 107–114.


the blue gourami. *Gen. Comp. Endocrinol.* 166, 83–93.


Pituitary adenylate cyclase activating polypeptide (PACAP) stimulates adenylate cyclase and promotes proliferation of mouse primordial germ cells. *Development* 122, 215–221.


rats delays puberty through the influence of the LHRH neuronal system. *Regul. Pept.* 109, 49–55.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 October 2012; paper pending published: 21 October 2012; accepted: 20 November 2012; published online: 11 December 2012.*

*Citation: Reglodi D, Tamas A, Koppan M, Szogyi D and Welke L (2012) Role of PACAP in female fertility and reproduction at gonadal level – recent advances. Front. Endocrin. 3:155. doi: 10.3389/fendo.2012.00155*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Reglodi, Tamas, Koppan, Szogyi and Welke. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

**REVIEW ARTICLE** published: 06 December 2012 doi: 10.3389/fendo.2012.00158

# Glucagon-like peptide-1 receptor overexpression in cancer and its impact on clinical applications

# *Meike Körner1, Emanuel Christ 2, DamianWild3 and Jean Claude Reubi1\**

<sup>1</sup> Division of Cell Biology and Experimental Research, Institute of Pathology, University of Berne, Berne, Switzerland

<sup>2</sup> Division of Endocrinology, Diabetology and Clinical Nutrition, University Hospital of Berne, Inselspital, Berne, Switzerland

<sup>3</sup> Division of Nuclear Medicine, Department of Radiology, University Basel Hospital, Basel, Switzerland

#### *Edited by:*

Jae Young Seong, Korea University, South Korea

#### *Reviewed by:*

Jong-Ik Hwang, Korea University, South Korea Billy K. Chow, University of Hong Kong, China

#### *\*Correspondence:*

Jean Claude Reubi, Division of Cell Biology and Experimental Research, Institute of Pathology, University of Berne, Murtenstrasse 31, CH – 3010 Berne, Switzerland. e-mail: reubi@pathology.unibe.ch

Peptide hormones of the glucagon-like peptide (GLP) family play an increasing clinical role, such as GLP-1 in diabetes therapy. Moreover, GLP receptors are overexpressed in various human tumor types and therefore represent molecular targets for important clinical applications. In particular, virtually all benign insulinomas highly overexpress GLP-1 receptors (GLP-1R).Targeting GLP-1R with the stable GLP-1 analogs 111In-DOTA/DPTA-exendin-4 offers a new approach to successfully localize these small tumors. This non-invasive technique has the potential to replace the invasive localization of insulinomas by selective arterial stimulation and venous sampling. Malignant insulinomas, in contrast to their benign counterparts, express GLP-1R in only one-third of the cases, while they more often express the somatostatin type 2 receptors. Importantly, one of the two receptors appears to be always expressed in malignant insulinomas. The GLP-1R overexpression in selected cancers is worth to be kept in mind with regard to the increasing use of GLP-1 analogs for diabetes therapy. While the functional role of GLP-1R in neoplasia is not known yet, it may be safe to monitor patients undergoing GLP-1 therapy carefully.

**Keywords: glucagon-like peptide-1, glucagon-like peptide-1 receptor, insulinoma, 111In-DOTA/DPTA-exendin-4**

"fendo-03-00158" — 2012/12/4 — 21:05 — page 1 — #1

# **INTRODUCTION**

G protein-coupled peptide hormone receptors play an increasing role as tumor targets in cancer medicine (Reubi, 2003). The underlying molecular basis is primarily an overexpression of a specific peptide receptor on tumor cells, irrespective of receptor functions. This overexpression allows a receptor-targeted scintigraphic imaging and radiotherapy of tumors with adequate radiolabeled peptide analogs (Reubi, 2003). Historically, the somatostatin receptors were the first receptors identified for these purposes (Reubi, 2003). They are expressed in high incidence and at high levels in gastroenteropancreatic neuroendocrine tumors (Reubi, 2003). Somatostatin receptor scintigraphy using the somatostatin analog octreotide (OctreoScan-) represents nowadays a standard imaging procedure for patients with gut neuroendocrine tumors, while PET/CT with 68Ga-labeled somatostatin analogs turns out to be even superior to OctreoScan- (Gabriel et al.,2007). Furthermore, results from clinical studies performing somatostatin receptor-mediated radionuclide therapy are encouraging (Kwekkeboom et al., 2008; Imhof et al., 2011).

The clinical success of somatostatin receptor targeting of gut neuroendocrine tumors has stimulated the search for other peptide receptors suitable for similar applications. A promising candidate is the glucagon-like peptide 1 (GLP-1) receptor (GLP-1R). This receptor has been cloned almost 20 years ago (Thorens et al., 1993). It is a member of the class 2 G protein-coupled receptor family (Gros et al., 1993; Nauck, 2009). Only a single GLP-1R has been identified so far, which is structurally identical in all tissues (Thorens et al., 1993). Receptor activation upon agonist binding stimulates adenylate cyclase and phospholipase C, with subsequent activation of protein kinase A and C, respectively (Thorens et al., 1993).

Physiologically, the GLP-1R is expressed mainly in the alimentary tract, particularly in the pancreatic islet cells (Wei and Mojsov, 1995) where it mediates the actions of GLP-1 released from the small intestines in response to food intake. GLP-1 is considered to be one of the most important glucose-dependent insulin secretagogues (Nauck, 2009). Specifically, it stimulates glucosedependent insulin synthesis and secretion, inhibits glucagon secretion, decreases β-cell apoptosis and increases differentiation of β-cell precursor cells in the pancreas as well as inhibits gastric emptying and appetite at the hypothalamic level (Nauck et al., 2009). Exploiting GLP-1 pathways therefore represents an ideal therapeutic approach for patients with type 2 diabetes, as this interferes with the main pathophysiological mechanisms of the disease (Nauck et al., 2009). Indeed, synthetic GLP-1 analogs are FDA- and EMEA-approved for the treatment of type 2 diabetes (Nauck et al., 2009).

The GLP-1R is of clinical interest not only due to its physiologic expression and functions in pancreatic islet cells and its potential in diabetes therapy, but also because of its possible role in cancer. Indeed, 10 years ago, the GLP-1R was found to be expressed in insulin-producing islet cell tumors, i.e., insulinomas (Reubi and Waser, 2003). This discovery lead to an extensive evaluation of the potential of the GLP-1R for targeted tumor imaging and therapy analogous to somatostatin receptor targeting of gut neuroendocrine tumors. This evaluation included the characterization of human tumors and normal tissues for their GLP-1R expression, since an important prerequisite for a successful peptide receptor

targeting of tumors is a high receptor expression in tumors, but a low receptor expression in normal background tissues. Further activities included the development of adequate radiolabeled GLP-1 analogs, testing of such analogs in *in vivo* animal models and application of selected suitable candidate analogs to tumor patients in preliminary clinical studies. This review summarizes the knowledge on the *in vitro* and *in vivo* basis of GLP-1 receptor targeting of tumors accumulated in the last decade.

#### **GLP-1R IN TUMORS**

The GLP-1R expression has been systematically assessed in a broad spectrum of original human tumor tissues using *in vitro* receptor autoradiography (Reubi and Waser, 2003; Korner et al., 2007; Waser et al., 2011). The GLP-1R was thus identified in specific endocrine, embryonal, and brain tumors, but virtually not in carcinomas (**Table 1**). The most striking GLP-1R expression was found in insulinoma. This is an endocrine tumor of the pancreatic islet cells with mostly benign biological behavior, but characterized clinically by severe symptoms of hyperinsulinism due to insulin secretion. Benign insulinomas expressed GLP-1Rs in very high incidence (>90%) and extremely high density (Reubi

**Table 1 | GLP-1R expressing human tumors: receptor incidences and densities.**


\*dpm/mg tissue.

†Mean value of two tumors tested with in vitro GLP-1R autoradiography.

**Frontiers in Endocrinology** | Neuroendocrine Science December 2012 | Volume 3 | Article 158 |

and Waser, 2003; **Table 1**; **Figure 1**). In fact, no other peptide receptor has been found to exhibit such high expression levels in this tumor type (Reubi and Waser, 2003). On the contrary, malignant, metastasizing insulinomas expressed GLP-1Rs significantly less frequently. High GLP-1R levels were found in only 36% malignant insulinomas (Wild et al., 2011). In insulinoma cells, the GLP-1R may represent a mediator of insulin secretion: in a model of GLP-1R transfected insulinoma cells, glucose-mediated insulin release was increased compared to control cells, in parallel with an increase of the intracellular second messenger of the GLP-1R (cAMP; Montrose-Rafizadeh et al., 1997).

Also several other functioning endocrine tumors of the pancreas expressed GLP-1Rs, in particular gastrinomas, however in lower amounts compared with insulinomas (**Table 1**; Reubi and Waser, 2003). Moreover, GLP-1Rs were discovered in a number of extrapancreatic endocrine tumors, including ileal carcinoids, pheochromocytomas, paragangliomas, bronchial carcinoid tumors, and medullary thyroid carcinomas, while they were not identified in pituitary adenomas or adrenal cortical tumors (Korner et al., 2007). Pheochromocytomas are of particular clinical interest due to their high GLP-1R expression levels (**Table 1**). Furthermore, medullary thyroid carcinomas are noteworthy because of important species differences in their GLP-1R expression. In rats, virtually all medullary thyroid carcinomas expressed GLP-1Rs in high amounts (Waser et al., 2011), while in humans only 28% expressed GLP-1R at low density levels (**Table 1**).

Lower GLP-1R expression levels were found in embryonal tumors, including medulloblastoma, nephroblastoma, and neuroblastoma (**Table 1**). They showed GLP-1Rs in low density in 15–25% of the tumors (Korner et al., 2007). Similarly, tumors of the nervous system such as meningiomas and astrocytomas demonstrated an incidence of GLP-1Rs between 25 and 35%, whereas glioblastomas and ependymomas expressed GLP-1Rs in 9–16% (**Table 1**; Korner et al., 2007). Schwannomas were devoid of GLP-1Rs (Korner et al., 2007).

Conversely, carcinomas exhibited a very low or no GLP-1R expression. Only ovarian and prostate carcinomas rarely showed GLP-1R at low levels, while breast, colorectal, gastric, pancreatic, hepatocellular, and cholangiocellular as well as lung carcinomas (non-small and small cell carcinomas) were negative for GLP-1R (Korner et al., 2007). Likewise, non-Hodgkin lymphomas did not express GLP-1R (Korner et al., 2007).

Among all GLP-1R expressing tumor types, insulinomas are at present of highest clinical interest for an *in vivo* targeting in patients, based on several considerations. First, insulinomas exhibit particularly high GLP-1R expression levels with respect to both incidence and density. Second, benign insulinomas, in contrast to most other gastroenteropancreatic neuroendocrine tumors, show relatively low expression levels of somatostatin receptors. Consequently, OctreoScan is not a reliable tool to detect these tumors (Plockinger et al., 2004). Third, the exact intraoperative localization of insulinomas is critical in order to minimize the surgical intervention (Rostambeigi and Thompson, 2009). This is, however, difficult due to the small size of benign insulinomas (usually 10–20 mm). Conventional radiological procedures (endosonography, MR-, and CT-imaging) are not always

"fendo-03-00158" — 2012/12/4 — 21:05 — page 2 — #2

successful in localizing insulinomas (Chatziioannou et al., 2001). Moreover, [18F]DOPA PET shows at present controversial results, with sensitivities ranging between 17 and 90% (Kauhanen et al., 2007; Tessonnier et al., 2010). Although selective arterial stimulation and venous sampling is a reliable intraoperative tool to detect insulinomas in experienced institutions (Wiesli et al., 2004), it is an invasive procedure with the associated risks. Moreover, this procedure identifies only the region of the pancreas – depending on the vasculature – where the insulinoma should be located and not the tumor itself (Wiesli et al., 2004).

#### **GLP-1R IN NON-NEOPLASTIC TISSUES**

The GLP-1R expression was similarly characterized in human normal tissues. It has been found in the pancreatic islets and acini, stomach, duodenal Brunner's gland, small and large intestinal myenteric plexus, lung and kidney vasculature, breast parenchyma, heart, brainstem, hypothalamus, neurohypophysis, and meninges (Wei and Mojsov, 1995; Korner et al., 2007). *In vitro* receptor autoradiography revealed that GLP-1R levels were highest in the neurohypophysis, followed by Brunner's glands, meninges, and pancreatic islets (Korner et al., 2007). Of practical importance, with the exception of Brunner's glands, the different tissues in the pancreatic area (i.e., pancreas islets and acini, intestines, and kidney) exhibit far lower GLP-1R density levels than insulinomas. This results in a high tumor-to-background ratio in GLP-1R density levels for insulinomas, which is an important prerequisite for a GLP-1R-targeted scintigraphic imaging of insulinomas.

Prominent species differences in the physiological GLP-1R expression between humans and rodents are noteworthy. Indeed, autoradiography experiments indicate that GLP-1R density levels are considerably higher in the lungs of rats and mice than of humans (Korner et al., 2007). This has to be considered when interpreting results of *in vivo* testing of GLP-1R targeting in rodent models. Likewise, the GLP-1R expression in the thyroid gland is substantial in rodents, but virtually absent in humans (Korner et al., 2007). In rodent thyroids, GLP-1Rs are located in the medullary C-cells. Of interest, treatment with GLP-1 analogs in rats is known to occasionally lead to thyroid C-cell hyperplasia and medullary thyroid carcinoma, whereas in humans there is so far no evidence of such complications. It can be speculated whether the species differences in the GLP-1R density expression in the precursor cells of these tumors, the medullary C-cells, contribute to these controversial findings (Waser et al., 2011).

#### **RADIOLABELED GLP-1 ANALOGS**

"fendo-03-00158" — 2012/12/4 — 21:05 — page 3 — #3

In general, radioactively labeled peptide analogs represent pharmaceuticals with favorable characteristics. Due to their small size, they show fast diffusion and rapid blood clearance and lack

immunogenicity. Moreover, radiopeptides exhibit only rare side effects (Reubi, 2003). In addition, radiolabeling is easily feasible, preferably after attaching a chelator to the peptide (Reubi and Maecke, 2008). However, since peptides are physiologically degraded within minutes in the human blood by potent peptidases such as dipeptidyl-peptidase-4 (DPP-4; Baggio and Drucker, 2007), stable peptide analogs have to be used instead in clinical applications. As for GLP-1, a naturally occurring stable analog exists, namely exendin-4, which is a component of the Gila monster venom. It shares 53% homology with GLP-1 and similarly binds to GLP-1Rs, but is resistant to DPP-4 cleavage (Nauck, 2009). Exendin-4 is, therefore, a good candidate for the development of radiolabeled GLP-1R ligands.

The first radiopeptides tested for *in vivo* GLP-1R targeting were 125I-labeled GLP-1 and the GLP-1 analog exendin-3 (Gotthardt et al., 2002). However, the low peptide stability of GLP-1 and the low efficiency of radio-iodination of exendin-3 limited their clinical use. Further testing resulted in the development of 111In-labeled exendin-4 (Wild et al., 2006): exendin-4 was coupled via the Lys side chain to a chelator (DOTA, tetraazacyclododecane tetraacetic acid or DTPA, diethylenetriaminepentaacetic acid) using a spacer (Ahx, aminohexanoic acid) and then labeled with 111In. This radiopeptide was subsequently extensively tested *in vitro* and *in vivo* in insulinoma models and applied to insulinoma patients (see below). Lately, several studies have been published that describe GLP-1R ligands suitable for PET/CT imaging, such as 68Ga-, 64Cu-, or 18F-labeled exendin-4, or for SPECT/CT imaging like 99mTc-labeled exendin-4 (Brom et al., 2010; Wild et al., 2010; Wu et al., 2011; Kiesewetter et al., 2012). These novel radiopeptides have not yet been tested in insulinoma patients.

#### **GLP-1R TARGETING IN ANIMAL MODELS**

Initially, GLP-1R targeting was performed in the rat insulinoma cell line RINm5F and in a rat insulinoma animal model (NEDH rats) using 125I-labeled GLP-1 and exendin-3 (Gotthardt et al., 2002). Specific uptake was detected in the cell and animal models. This provided the proof of principle for GLP-1R targeting of insulinoma (Gotthardt et al., 2002).

Follow-up experiments were carried out in the Rip1tag2 mouse model with 111In-DTPA-exendin-4. These transgenic mice develop tumors of the pancreatic β-cells in a reproducible multistage fashion (Hanahan, 1985) and, therefore, represent an ideal model to study GLP-1R targeting *in vivo* and *in vitro*. Using GLP-1R multipinhole SPECT/MRI and SPECT/CT, *in vivo* GLP-1R imaging was performed in these animals following administration of 111In-DTPA-exendin-4 (Wild et al., 2006). In parallel, GLP-1R autoradiography of the tumors was carried out *in vitro*. Finally, biodistribution and pharmacokinetics as well as internalization and cellular retention of 111In-DTPA-exendin-4 were measured *in vitro* (Wild et al., 2006).

This preclinical study showed the following main findings: First, the GLP-1R density in the tumors was extremely high, resulting in a remarkably high uptake of 111In-DTPA-exendin-4 (287 ± 62% IA/g tissue) already 4 h after injection. Second, excellent visualization of tumors as small as 1 mm by pinhole SPECT/MRI and SPECT/CT was demonstrated. Third, the tumor-to-background ratio was very high (between 13.6 and 299), substantiating the high potential of this radiopeptide to specifically localize GLP-1R positive lesions within the pancreas. Lastly, *in vitro* studies in the cells derived from the tumor model demonstrated a specific internalization of 111In-DTPA-exendin-4, and biochemical investigations confirmed the high metabolic stability of the radiopeptide in the tumor cells as well as in the serum.

The same Rip1tag2 mouse model also provided preliminary data on GLP-1R-targeted therapy of insulinoma. Mice were injected with different doses of 111In-DTPA-exendin-4 (1.1, 5.6, and 28 MBq) and sacrificed 7 days after injection. Most impressively, a single injection lead to a reduction in tumor volume of up to 94% in a dose-dependent manner without significant acute organ toxicity. Histological examination revealed that the decrease in tumor mass was mainly due to an increase in tumor cell apoptosis and decreased proliferation (Wicki et al., 2007).

# **GLP-1R TARGETING IN HUMANS**

The first patient who underwent GLP-1R scintigraphy suffered from severe endogenous hyperinsulinemic hypoglycemia with non-convulsive seizures. MRI, CT scan, and endosonography did not detect any suspicious lesion. However, GLP-1R scintigraphy revealed an increased extrapancreatic uptake in the mesentery supplied by the anterior mesentery artery. Selective arterial stimulation and venous sampling correctly indicated the vascular territory, but since this patient had an ectopic insulinoma, the results of the invasive investigation without GLP-1R imaging would have been misleadingfor the surgical strategy (Wild et al., 2008).

In a proof of principle study, 111In-DOTA-exendin-4 was prospectively administered to a total of six patients (Christ et al., 2009). All of them presented with neuroglycopenic symptoms lasting for 4–26 months. Biochemical evaluation during a fasting test revealed endogenous hyperinsulinemic hypoglycemia in all patients.

Conventional imaging (CT or MRI) reliably detected the insulinoma in only two patients, whereas endosonography identified a possible lesion in four patients, in keeping with data in the literature (McAuley et al., 2005). In three patients, selective arterial stimulation and venous sampling was performed, with accurate localization in all (Christ et al., 2009). Remarkably, GLP1-R scintigraphy correctly detected the insulinoma in all six consecutive patients (**Figure 2**; Christ et al., 2009). Four patients underwent an enucleation of the insulinoma. In two patients, a Whipple procedure had to be performed due to the localization of the insulinoma. In all patients, a benign insulinoma was confirmed by histology. *In vitro* autoradiography studies showed GLP-1R densities in the range as previously described (between 2,600 to >10,000 dpm/mg tissue; Reubi and Waser, 2003), but low levels of somatostatin receptor type 1 in 2 patients only (Christ et al., 2009). Importantly, within a time frame of 2–14 days after injection of 111In-DOTA-exendin-4, intraoperative utilization of a gamma probe was highly beneficial for the *in situ* localization of the insulinoma in all patients, resulting in a successful enucleation where possible (Christ et al., 2009).

Fortunately, background uptake over the whole body was low with the exceptions of the kidneys, which were strongly labeled due to renal excretion of the radioligand (**Figure 2**). In two

"fendo-03-00158" — 2012/12/4 — 21:05 — page 4 — #4

"fendo-03-00158" — 2012/12/4 — 21:05 — page 5 — #5

patients demarcation between tumors (maximal diameter of 9– 11 mm) and kidneys was only possible after late scans, indicating an improved tumor-to-kidney ratio with time, in keeping with the fact that the effective half-life of 111In-DOTA-exendin-4 was longer in the tumor (38–64 h) than in the kidneys (31.2–31.8 h; **Figure 2**; Christ et al., 2009). This suggests that patients with negative early scans should have additional imaging 3–7 days after the injection.

In humans, 111In-DOTA-exendin-4 was initially administered, but later replaced by 111In-DTPA-exendin-4 due to the four times higher specific activity of the latter (Wild et al., 2009). A higher specific activity permits to reduce the amount of peptide (exendin-4), thereby decreasing the occurrence of possible side effects. In addition, radiolabeling of DTPA-exendin-4 can be performed at room temperature, whereas labeling of DOTA-exendin-4 has to be accomplished at high temperature (95◦C; Wild et al., 2006).

By now, data of a prospective, multicenter trial including 30 patients that underwent GLP-1R scanning are available. The inclusion criteria were proven endogenous hyperinsulinemic hypoglycemia and none or maximally one lesion on conventional imaging. Conventional imaging (CT, MRI) and endosonography – where available – was performed locally using a standard protocol. 111In-DTPA-exendin-4 was administered intravenously at a dose of 90–130 MBq over 2 min. Whole-body planar images and SPECT/CT of the abdomen were performed at 4, 24 and in some patients between 72 and 96 up to 168 h post-injection, the most important time point being the scan 24 h after injection. Diagnosis was confirmed by histology after surgical removal (Christ et al., 2012).

Conventional imaging (MRI, CT, endosonography) was positive in 17 patients. 111In-DTPA/DOTA-exendin-4 SPECT/CT detected 23 true positive benign insulinomas and five additional positive lesions (one malignant insulinoma; two islet cell hyperplasias; two uncharacterized lesions). True negative tests were detected in two patients (one malignant insulinoma; one islet cell hyperplasia). Malignant insulinomas were diagnosed based on the histological finding of a positive lymph node, not detected on conventional imaging preoperatively. There was no false negative result. Sensitivity was 100% and the positive predictive value was 82% (Christ et al., 2012). These findings are encouraging and suggest that *in vivo* GLP-1R imaging defines a new noninvasive diagnostic approach to successfully localize small benign insulinomas.

About 90% of insulinomas are benign and only 10% of patients present with malignant disease usually characterized by liver metastasis (Plockinger et al., 2004). Anecdotal evidence suggests that malignant insulinomas exhibit more often somatostatin receptors type 2 than benign ones and can, therefore, be visualized by OctreoScan- (Plockinger et al., 2004). A more extensive study with data from 10 patients with malignant insulinoma showed that somatostatin receptors type 2 were expressed in seven patients, whereas GLP-1R were present in four patients, and both receptors in only one patient (Wild et al., 2011). Importantly, one of the two imaging methods appears always to be positive in a malignant type of insulinoma (Wild et al., 2011). The consequences of the respective receptor expression in an insulinoma with regard to biological behavior (malignant or benign course) remains to be established.

# **SIDE EFFECTS AND LIMITATIONS**

In humans, the injection of 111In-DOTA-exendin-4 and 111In-DPTA-exendin-4 was well tolerated. Due to the small amount of exendin-4, the decrease in plasma glucose concentrations was only 1.4 ± 0.7 mmol/L after 40 min (Christ et al., 2009). By regularly monitoring glucose levels, no severe hypoglycemic episode occurred. One patient experienced a short episode of vomiting only with 111In-DOTA-exendin-4. Otherwise, no further side effects were observed (Christ et al., 2009).

In two patients, there was focal 111In-DOTA-exendin-4 uptake in the proximal duodenum. This may be related to the presence of Brunnner's gland of the duodenum which, as previously mentioned, are known to contain GLP-1Rs in a significant density (Korner et al., 2007). Brunner's glands may become hyperplastic (Levine et al., 1995), as observed in particular in patients with chronic pancreatitis (Stolte et al., 1981). Such hyperplastic glands may possibly be detected by GLP-1R imaging.

A differential diagnosis of endogenous hyperinsulinemic hypoglycemia includes nesidioblastosis, also known as "noninsulinoma pancreatogenous hypoglycemia syndrome" in a clinical setting (Thompson et al., 2000). Histopathologically, this entity is defined as a diffuse hyperplasia of β-cells occurring usually in children (Yakovac et al., 1971). Recent evidence suggests that this pathology can also be demonstrated in adults, in particular after bypass surgery for morbid obesity (Service et al., 2005). In the previously mentioned series of patients (Christ et al., 2012), islet cell hyperplasia was diagnosed in three patients, two were positive on GLP-1R imaging and one was negative. Based on these preliminary data GLP-1R imaging does not appear to be an appropriate tool to diagnose or exclude islet cell hyperplasia. These findings are further supported by the recent evidence that the *in vitro* density of GLP-1R in pancreatic tissues of patients with nesidioblastosis after bypass surgery for morbid obesity is much lower than in benign insulinomas (Reubi et al., 2010).

#### **REFERENCES**


of insulinomas. *J. Clin. Endocrinol. Metab.* 94, 4398–4405.


"fendo-03-00158" — 2012/12/4 — 21:05 — page 6 — #6

Recently, 18F-DOPA-PET has successfully been used to detect nesidioblastosis and benign insulinoma (Kauhanen et al., 2007). Although 18F-DOPA-PET may be helpful to diagnose nesidioblastosis, in benign insulinomas the tumor-to-background ratios are higher for 111In-DOTA-exendin-4 SPECT than for 18F-DOPA-PET (3.3 vs. 1.4), suggesting an increased sensitivity of targeting GLP-1Rs (Kauhanen et al., 2007; Christ et al., 2009) in benign insulinomas.

#### **SUMMARY AND CONCLUSION**

Because of the massive GLP-1R overexpression in selected gastrointestinal tumors, GLP-1 and GLP-1R play an increasing role in endocrine gastrointestinal tumor management. Targeting GLP-1R with 111In-DOTA-exendin-4 or 111In-DPTA-exendin-4 offers a new approach that permits the successful localization of small benign insulinomas pre- and intraoperatively. Since virtually all benign insulinomas express GLP-1Rs and the preliminary clinical data are very encouraging, it is likely that this approach will affect the algorithm of pre- and intraoperative localization of suspected insulinoma.

In contrast to benign insulinomas, where the exact localization of the tumor is the main goal, the clinical challenge in malignant, metastasizing insulinomas is to define the extension of the disease and – if possible – offer a targeted therapy (peptide receptor radionuclide therapy, PRRT). Contrary to benign insulinomas, malignant insulinomas more often express sst2 receptors than GLP-1R. Importantly, one of the two receptors seems to be always expressed.

With regard to the increasing and successful use of GLP-1 analogs for diabetes therapy, it is worth keeping in mind that selected cancers can overexpress GLP-1R. While the functional role of these receptors in these tumors is not known yet, it may be safe to monitor patients with such tumors carefully during their GLP-1-analog-based diabetes therapy.

peptide-1-(7-36) amide, oxyntomodulin, and glucagon interact with a common receptor in a somatostatin-secreting cell line. *Endocrinology* 133, 631–638.


patients. *J. Clin. Endocrinol. Metab.* 92, 1237–1244.


27 cases. *Am. J. Gastroenterol.* 90, 290–294.


peptide receptors in neuroendocrine tumors as molecular basis for in vivo multireceptor tumor targeting. *Eur. J. Nucl. Med.* 30, 781–793.


diabetes therapy. *Neuroendocrinology* 94, 291–301.


"fendo-03-00158" — 2012/12/4 — 21:05 — page 7 — #7


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 October 2012; accepted: 23 November 2012; published online: 06 December 2012.*

*Citation: Körner M, Christ E,Wild D and Reubi JC (2012) Glucagon-like peptide-1 receptor overexpression in cancer and its impact on clinical applications. Front. Endocrin. 3:158. doi: 10.3389/fendo. 2012.00158*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Körner, Christ, Wild and Reubi. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Structural and molecular conservation of glucagon-like peptide-1 and its receptor confers selective ligand-receptor interaction

#### **Mi Jin Moon<sup>1</sup> , Sumi Park <sup>1</sup> , Dong-Kyu Kim<sup>1</sup> , Eun Bee Cho<sup>1</sup> , Jong-Ik Hwang<sup>1</sup> , Hubert Vaudry <sup>2</sup> and JaeYoung Seong<sup>1</sup>\***

<sup>1</sup> Graduate School of Medicine, Korea University, Seoul, Republic of Korea

2 INSERM U982, Laboratory of Neuronal and Neuroendocrine Differentiation and Communication, University of Rouen, Mont-Saint-Aignan, France

#### **Edited by:**

Billy K. Chow, University of Hong Kong, China

#### **Reviewed by:**

Joao Carlos Dos Reis Cardoso, University of Algarve, Portugal Charlier Dominique Thierry, University of Liege, Belgium

#### **\*Correspondence:**

Jae Young Seong, Graduate School of Medicine, Korea University, Seoul 136-705, Republic of Korea. e-mail: jyseong@korea.ac.kr

Glucagon-like peptide-1 (GLP-1) is a major player in the regulation of glucose homeostasis. It acts on pancreatic beta cells to stimulate insulin secretion and on the brain to inhibit appetite.Thus, it may be a promising therapeutic agent for the treatment of type 2 diabetes mellitus and obesity. Despite the physiological and clinical importance of GLP-1, molecular interaction with the GLP-1 receptor (GLP1R) is not well understood. Particularly, the specific amino acid residues within the transmembrane helices and extracellular loops of the receptor that may confer ligand-induced receptor activation have been poorly investigated. Amino acid sequence comparisons of GLP-1 and GLP1R with their orthologs and paralogs in vertebrates, combined with biochemical approaches, are useful to determine which amino acid residues in the peptide and the receptor confer selective ligand-receptor interaction. This article reviews how the molecular evolution of GLP-1 and GLP1R contributes to the selective interaction between this ligand-receptor pair, providing critical clues for the development of potent agonists for the treatment of diabetes mellitus and obesity.

**Keywords: GLP-1, GLP1R, G protein-coupled receptors, evolution, paralog, ortholog, ligand-receptor interaction**

#### **INTRODUCTION**

Glucagon-like peptide-1 (GLP-1) is an intestinal incretin released in response to nutrient ingestion that stimulates insulin secretion in a glucose-dependent manner. The insulinotropic effect of GLP-1 on the pancreas has been demonstrated to be preserved in animal models of diabetes by stimulating insulin exocytosis (Shen et al., 1998; Drucker, 2001; MacDonald et al., 2002), and promoting insulin biosynthesis (Fehmann and Habener, 1992; Perfetti and Merkel, 2000; Moon et al., 2011). Recently, direct effects of GLP-1 on growth, survival (Xu et al., 1999; Stoffers et al., 2000; List and Habener, 2004), and differentiation of β-cells have been reported (Drucker, 2003). Beside its insulinotropic effects, GLP-1 inhibits glucagon (GCG) secretion in pancreatic α-cells (Nauck et al., 2002), attenuates gastric emptying, and ameliorates glucose excursion in the gastrointestinal tract (Nauck et al., 2002).

GLP-1 exerts its action through the G protein-coupled receptor (GPCR), GLP1R. This receptor belongs to the class B (or secretin-like) GPCR family. This GPCR family has only 15 members in humans and is characterized by a relatively long N-terminal extracellular domain (ECD) containing six conserved Cys residues that form a constraint structure necessary for ligand binding (Couvineau et al., 2004). GLP1R is mainly expressed in pancreatic β-cells, and upon stimulation by GLP-1, it induces the accumulation of cAMP and the influx of intracellular calcium which accelerate insulin release from secretory granules (Drucker et al., 1987; Fehmann et al., 1995).

GLP-1 and GLP1R are also expressed in the central nervous system. GLP-1 is synthesized largely in the brainstem and transported along axonal networks to diverse brain regions, including the hypothalamus (Vrang et al., 2007; Hisadome et al., 2010). GLP1R is expressed in cerebral cortex, hypothalamus, hippocampus, thalamus, caudate-putamen, and globus pallidum (Alvarez et al., 2005). In the brain, GLP-1 is known to reduce appetite, leading to significant reductions in body weight (Zander et al., 2002). In addition, GLP-1 is likely neuroprotective and involved in neurite growth and spatial learning ability in the brain (During et al., 2003; Perry et al., 2007).

Due to its combined beneficial effects, GLP-1 has been identified as a potential therapeutic agent for the treatment of diabetes mellitus and obesity. However, the molecular mechanisms leading to high affinity ligand-receptor binding and receptor activation have not been fully understood. Studies using alanine scanning mutagenesis, substitution/modification, and chimeric peptide construction of GLP-1 have explored the bioactive motifs of GLP-1(Adelhorst et al., 1994; Gallwitz et al., 1994; Hinke et al., 2004), yet none have been able to identify the mechanism through which individual residues in the peptide interact with residues in the receptor. Recent studies using X-ray crystallography have demonstrated that residues in the central α-helical region of GLP-1

**Abbreviations:** Bpa, *p*-benzoyl-l-phenyl alanine; CRFR, CRF receptor; Cys, cysteine; ECD, extracellular domain; ECL, extracellular loop; ER, endoplasmic reticulum; GCG, glucagon; GCGR, glucagon receptor; GIP, glucose-dependent insulinotropic polypeptide; GIPR, GIP receptor; GLP-1, glucagon-like peptide-1; GLP1R, GLP-1 receptor; GPCR, G protein-coupled receptor; PAC1R, PACAP receptor; PTHR, PTH receptor; TMH, transmembrane helix.

interact with residues in the N-terminal ECD of GLP1R (Runge et al., 2008; Underwood et al., 2010). However, as ligand-induced receptor activation is mainly governed by interactions between residues in the N-terminus of the peptide and residues within the transmembrane helices (TMH) and extracellular loops (ECL) of the receptor (Thorens et al., 1993; During et al., 2003), this crystal structure only accounts for the peptide binding to the ECD of the receptor. Indeed, virtually no progress has been made in exploring the receptor binding sites for the N-terminal moiety of the peptide which is responsible for ligand-induced receptor activation.

Peptide ligands and their receptors have become diversified through evolutionary processes that ultimately yield families of related, yet distinct, peptides and receptors (Hoyle, 1999; Cho et al., 2007; Lee et al., 2009; Kim et al., 2011, 2012). Specific diversification of peptides, conservation within orthologs but variation among paralogs, would often confer the selective interaction with their cognate receptors, allowing discrimination of paralogous receptors (Acharjee et al., 2004; Wang et al., 2004; Li et al., 2005). Thus, amino acid sequence comparison of the peptide and receptor with their orthologs and paralogs, along with mutational mapping approaches, are useful tools helping to determine specific residues in the peptide ligands and receptors that are essential for maintaining selective ligand-receptor interaction. Indeed, ligand binding domains identified in mammalian receptors are highly conserved in orthologous non-mammalian receptors, indicating that there is high evolutionary selection pressure to maintain selectivity for their ligands (Acharjee et al., 2004; Wang et al., 2004; Li et al., 2005). Recently, we reported that evolutionarily conserved amino acid residues in GLP-1 and core domains of the GLP1R confer selective ligand-receptor interaction and receptor activation (Moon et al., 2010, 2012). This article reviews how the molecular evolution of GLP-1 and GLP1R contributes to acquiring high affinity interaction between this peptide ligand and receptor.

#### **GENERAL STRUCTURE OF GLP-1 AND ITS FAMILY PEPTIDES**

GLP-1 is a product of the *GCG* gene which encodes a common GCG-GLP-1-GLP-2 precursor. All three peptides are encoded by different exons of the *GCG* gene, raising the possibility of exon duplications during early vertebrate evolution (Sherwood et al., 2000). The *GCG* gene produces one or two mature peptides by a tissue-specific alternative post-translational process (Kieffer and Habener, 1999; Irwin, 2001, 2009). For instance, in pancreatic αcells mature GCG, but not GLP-1 and GLP-2, is produced. In intestinal L cells, however, mature GLP-1 and GLP-2, but not GCG, are generated. One GLP-1 paralog is the glucose-dependent insulinotropic polypeptide (GIP) which is independently encoded by the *GIP* gene (McIntosh et al., 2009). GLP-1 and GIP share a high degree of amino acid sequence identity, particularly in their N-terminal moiety, and function similarly by inducing insulin secretion from β-cells. However, these peptides act through distinct yet related receptors, GLP1R and GIP receptor (GIPR), respectively. In addition, albeit absent in mammals, another GLP-1 paralog is exendin, which was first discovered in Gila monsters (*Heloderma suspectum*; Göke et al., 1993). Recently, the full-length and/or partial cDNAs for exendin were characterized in a few species, such as *Xenopus*, chicken, and Gila monster (Irwin and Prentice, 2011). Although the receptor for exendin has not yet

been identified in non-mammals, exendin exhibits high affinity binding for mammalian GLP1R (Göke et al., 1993). Recent evidence suggests that the *GCG*, *GIP*, and *exendin* genes were generated by genome duplication events during early vertebrate evolution, as these genes are flanked by similar neighboring genes in the genomes of vertebrates (Irwin, 2002; Irwin and Prentice, 2011).

GLP-1 and its family peptides are 30 ∼ 40 amino acids in length and share similarities in amino acid sequence and secondary structure. All these peptides tend to be disordered in aqueous solutions but exhibit a marked propensity to form α-helices under mild ambient conditions, such as in the presence of organic solvents or lipids (Braun et al., 1983; Gronenborn et al., 1987; Thornton and Gorenstein, 1994; Inooka et al., 2001; Neidigh et al., 2001; Chang et al., 2002; Tan et al., 2006; Alana et al., 2007), or upon crystallization (Sasaki et al., 1975). It is now well known that the N-terminal domains of GLP-1-related peptides form a random coil structure, while the central parts of these peptides have an α-helical structure. In addition, hydrophobic amino acids at positions 6 and 10, and a short-chain polar amino acid at position 7 form a helix N-capping motif through hydrophobic interaction and hydrogen bonding (Neumann et al., 2008). This capping motif is believed to introduce a specific local fold that facilitates receptor activation upon peptide-receptor binding (Neumann et al., 2008).

The N-terminus of GLP-1 and its family peptides share a high degree of sequence identity (**Figure 1**). Particularly, Gly<sup>4</sup> , Thr/Ser<sup>5</sup> , Phe<sup>6</sup> , and Asp/Glu<sup>9</sup> are conserved across all GLP-1 paralogs. Indeed, alanine scanning of these conserved residues of GLP-1 suggest that positions 4, 6, and 9 are crucial for either maintaining secondary structure of the peptide or for interaction with the receptor (Adelhorst et al., 1994; Gallwitz et al., 1994). His<sup>1</sup> and Thr/Ser<sup>7</sup> are common for most GLP-1 paralogs except for GIP, which has Tyr<sup>1</sup> and Ile<sup>7</sup> in these positions. Our recent observation using a chimeric GLP-1/GIP peptide revealed that His/Tyr<sup>1</sup> and Thr/Ile<sup>7</sup> are responsible for the selective interaction toward GLP1R and GIPR (Moon et al., 2010). The second position of the peptides is highly variable across paralogs, even within orthologs of vertebrates, and it is known to be of lesser importance for receptor binding. This residue can be modified to Ser or another amino acid to confer protection against cleavage by dipeptidyl peptidase IV (Hinke et al., 2002). The third position of the peptides is variable across paralogs but conserved within orthologs. Although the importance of this ortholog-specific third residue in receptor binding or peptide structure is not fully understood, Glu<sup>3</sup> of GIP is known to be critical for receptor interaction (Hinke et al., 2003; Gault et al., 2007; Yaqub et al., 2010).

The sequence similarity of the α-helix domain of the peptides is not pronounced among GLP-1 paralogs, and there are also many variable residues within orthologs (**Figure 1**). However, the Phe<sup>22</sup> , Ile/Val23, and Leu<sup>26</sup> residues that are part of the hydrophobic surface of the α-helix are highly conserved among GLP-1 and its paralogs. Interestingly, all these residues are known to interact with highly conserved residues in the ECD of the GLP1R and its paralogs (Parthier et al., 2007; Runge et al., 2008; Underwood et al., 2010). Some ortholog-conserved residues, Glu/Asp<sup>15</sup> of exendin and GIP, along with Ala18/19 and Lys<sup>20</sup> of GLP-1, are found to interact with the receptors as was revealed by the peptide-bound


glucagon, GIP, and Gila monster exendin are aligned. Residues in red are conserved sequences within orthologs of vertebrates including mouse, anole, chicken, Xenopus tropicalis, medaka, fugu, tetraodon,

ECD crystal structures (Parthier et al., 2007; Runge et al., 2008; Underwood et al., 2010).

#### **GENERAL STRUCTURE OF CLASS B GPCRs**

The class B GPCR family is composed of 15 members including receptors for VIP, PACAP, secretin, GCG, GLP-1, GLP-2, GHRH, GIP, PTH, calcitonin, calcitonin gene-related peptide, and CRH (Laburthe et al., 2007). This family shares the general GPCR architecture: seven TMH interconnected by intracellular loops with a C-terminal intracellular domain. Class B GPCRs differ from class A rhodopsin-like GPCRs in the structures of their TMH. The TMH of class B GPCRs do not contain conserved amino acid residues such as Asp/Asn2.50 in TMH2, Asn/Asp7.49-Pro7.50 x-x-Tyr7.53 (N/DPxxY) motif in TMH7, Asp/Glu3.49-Arg3.50 - Tyr/Trp3.51 (D/ERY/W) motif at the junction between TMH3 and intracellular loop 2 that are commonly found in class A rhodopsinlike GPCRs (Oh et al., 2005). Instead, class B receptors share a high degree of amino acid identity in TMH with one another. Further, they possess a large and structured N-terminal ECD of ∼120 residues.

Although no experimentally determined full-length class B receptor structure has been achieved to date, the structure elucidation of individual class B GPCR ECDs represents considerable progress toward a molecular understanding of their action. The first structure of the agonist-bound recombinant N-terminal ECD of the CRH2B receptor has been resolved using NMR (Grace et al., 2004). Six representative ECD structures of the class B family of GPCRs have been determined by X-ray crystallography or NMR spectroscopy in complex with bound ligand: the human VPAC1 receptor (Tan et al., 2006), a subtype of human PACAP receptor (PAC1R<sup>s</sup> ; Sun et al., 2007), human GIPR (Parthier et al., 2007), human GLP1R (Runge et al., 2008; Underwood et al., 2010), human PTH receptor (PTH1R; Pioszak and Xu, 2008), the human type-1 CRF receptor (CRFR1; Pioszak et al., 2008), and human GLP2R (Venneti and Hewage, 2011).

Class B GPCRs contain N-terminal signal peptides that are cleaved off by the signal peptidase of the endoplasmic reticulum (ER) during the translocation-mediated receptor insertion into the ER membrane. These signal peptides play a crucial role in the membrane expression of receptors (Couvineau et al., 2004;

variable sequences. Underlines indicate the α-helical conformation. The residues responsible for interaction with their receptor are shaded.

Alken et al., 2005; Huang et al., 2010). After cleavage of the signal peptide, the N-terminal helix at the beginning of the ECD and four β-strands forming two antiparallel sheets remain (**Figure 2**). Three disulfide bonds formed by a set of six Cys residues lock these secondary structural elements together. Cysteine residues are completely conserved across the receptors. The disulfide bond pattern seems to be conserved in all receptors, suggesting a similar three dimensional structure. There are three disulfide bonds between the first and third, the second and fifth, and the fourth and sixth cysteine residues (Cys<sup>1</sup> -Cys<sup>3</sup> , Cys<sup>2</sup> -Cys<sup>5</sup> , Cys<sup>4</sup> -Cys<sup>6</sup> ). The first bond (Cys<sup>1</sup> -Cys<sup>3</sup> ) links the N-terminal α-helix to the first β-sheet. The second (Cys<sup>2</sup> -Cys<sup>5</sup> ) connects the two β-sheets, whereas the third disulfide bond (Cys<sup>4</sup> -Cys<sup>6</sup> ) holds the C-terminus of the domain in close proximity to the central β-sheets (Tan et al., 2006; Parthier et al., 2007, 2009; Sun et al., 2007; Pioszak and Xu, 2008; Pioszak et al., 2008; Runge et al., 2008; Underwood et al., 2010; Venneti and Hewage, 2011).

In addition, this core folding is further stabilized by a salt bridge involving acidic and basic residues flanked by hydrophobic aromatic residues. This fold, called the Sushi domain, is conserved in all class B GPCRs (Grace et al., 2004). Five additional residues, Asp67, Trp72, Pro86, Gly108, and Trp<sup>110</sup> in GLP1R and the corresponding residues in class B GPCRs, are conserved. Particularly, Asp and the two Trp residues take part in forming the Sushi domain, suggesting that these residues are crucial for domain stability and ligand binding. The strongly conserved fold observed in the ECD of class B GPCRs suggests that a common mechanism underlies ligand recognition.

The crystal structures of the ligand-bound ECD reveal amino acid residues that interact with their cognate ligands (Parthier et al., 2007; Runge et al., 2008; Underwood et al., 2010). It is of interest to note that many ligand-interacting residues are highly conserved between GLP1R and its paralogs. For instance, Trp<sup>39</sup> in the α1-helix, Asp<sup>67</sup> in the β1-sheet, Tyr<sup>69</sup> between the β1- and β2-sheet, Arg<sup>121</sup> near TMH1 of GLP1R, residues 87–90 (Tyr-Leu-Pro-Trp) between the β3- and β4-sheet, Arg<sup>101</sup> in the β4-sheet, and Trp<sup>112</sup> near TMH1 of GIPR are all highly conserved across paralogs (**Figure 2**). There are also ortholog-specific residues that interact with the peptide ligands, such as Leu<sup>32</sup> in the α1-helix of GLP1R, and Gly<sup>110</sup> and His<sup>115</sup> near TMH1 of GIPR.


**FIGURE 2 | Amino acid sequence alignment of GLP1R and its paralogous receptors.** Amino acid sequences of human GLP1R, GLP2R, GCGR, and GIPR are compared. The signal peptides are indicated in gray. The residues colored in blue represent conserved sequences across the GLP-1-related peptide receptors. Residues in red are conserved

# **THE TWO-DOMAIN HYPOTHESIS FOR CLASS B GPCR LIGAND BINDING AND ACTIVATION**

Class B GPCRs likely share a similar secondary and tertiary structure with long N-terminal ECD and highly conserved TMHs. The orientation and mechanism of interaction of the peptide with their receptors has been investigated in studies using fragmented peptide/receptor, and chimeric peptides and receptors (Holtmann et al., 1995; Stroop et al., 1995; Bergwitz et al., 1996; Laburthe and Couvineau, 2002; Runge et al., 2003). For instance, the GLP1R ECD itself is able to bind with its peptide ligands (Graziano et al., 1996; Van Eyll et al., 1996; Wilmen et al., 1996, 1997; Runge et al., 2003;Parthier et al.,2007). This binding,however,may not account for ligand-induced receptor activation (Buggy et al., 1995; Holtmann et al., 1995, 1996; Hjorth and Schwartz, 1996; Xiao et al., 2000). The N-terminally truncated exendin(9–39) is unable to activate GLP1R even though it binds to the receptor with an affinity that is comparable to that of wild type exendin (Thorens et al., 1993). In contrast, exendin(1–9), a short N-terminal fragment of exendin, is able to activate GLP1R although its affinity to the receptor is quite low (During et al., 2003). Likewise, the GIP fragment, GIP (7–30), is able to bind to GIPR with high affinity but fails to induce receptor activation. In contrast, GIP (1–14) exhibits a very low affinity toward the receptor but fully activates the receptor at a micromolar concentration (Hinke et al., 2001, 2003; Gault et al., 2007). Thus, the two-domain model explaining ligand binding followed by receptor activation has emerged: the central α-helical and C-terminal portion of the peptide binds to the N-terminal ECD of the receptor (Al-Sabah and Donnelly,

sequences within orthologs of vertebrates such as mouse, anole, chicken, Xenopus tropicalis, medaka, fugu, tetraodon, stickleback, and zebrafish. The residues in black are variable sequences. The amino acid residues that interact with their peptides are shaded. The α-helix, β-sheets, and TMH domains of GLP1R are indicated.

2003; Dong et al., 2003; Lopez de Maturana et al., 2003) followed by binding of the N-terminal moiety of the peptide with the core domain – including the TMH and ECL – of the receptor, conferring receptor activation and G protein coupling (Runge et al., 2003; Lopez de Maturana et al., 2004; Castro et al., 2005; Wittelsberger et al., 2006). The two-domain model is generally consistent with photoaffinity crosslinking studies of several class B receptors. With a few exceptions, photoreactive side chains in the C-terminus of the peptide ligand interact with residues in the ECD of the receptor, whereas photoreactive side chains in the N-terminus of the ligand bind to the TMH domain (Gensure et al., 2001; Assil-Kishawi and Abou-Samra, 2002; Dong and Miller, 2002; Dong et al., 2004). Most recently, strong corroborating evidence for the two-domain model has been obtained by the structural characterization of the isolated ECDs of several class B GPCRs (Grace et al., 2004; Parthier et al., 2007; Sun et al., 2007; Pioszak and Xu, 2008; Pioszak et al., 2008; Runge et al., 2008; Underwood et al., 2010).

# **MOLECULAR EVOLUTION OF GLP-1 AND GLP1R FOR THEIR SELECTIVE INTERACTION**

Although GLP-1 and its paralogs share a high degree of sequence identity and structural similarly, they generally exhibit specific binding to their own cognate receptors with little cross-reactivity with paralogous receptors (Runge et al., 2003; Moon et al., 2010, 2012). This observation allows us to presume the presence of distinct amino acid residues within each peptide and receptor that allows for the selective interaction with their own partners. Further, these facts indicate that evolutionary selection pressure has

exerted a specific diversification of peptide and receptor: variation among paralogs but conservation within orthologs. However, there are some exceptional cross-reactivities among paralogous partners. For instance, exendin, a GLP-1 paralog in non-mammals, has a high affinity for the mammalian GLP1R (Göke et al., 1993). Until now, genetic orthologs for GLP1R have not been found in teleost fish, yet teleost fish have two copies of GLP-1 peptides which have been generated by teleost-specific genome duplication (Plisetskaya and Mommsen, 1996; Irwin and Wong, 2005). Interestingly, GLP-1 is able to activate fish GCG receptor (GCGR) orthologs (Yeung et al., 2002; Irwin and Wong, 2005), indicating that fish GCGRs have achieved functional response to GLP-1 through an evolutionary process. These exceptional cross-reactivities provide a unique opportunity to explore the identification of specific amino acid residues in the peptides and receptors responsible for specific ligand-receptor interaction. For instance, amino acid sequence comparison between GLP-1 and exendin allows us to predict which amino acid residues are important for the activation of GLP1R (Moon et al., 2010). Sequence comparison between tetrapod GLP1Rs and fish GCGRs has led to the identification of residues in these receptors that interact with GLP-1 (Moon et al., 2012).

#### **INTERACTION BETWEEN THE** α**-HELIX OF THE PEPTIDE AND THE ECD OF THE RECEPTOR**

The crystal structures of the ligand-bound ECD revealed that GLP-1 is a continuous α-helix from Thr<sup>7</sup> to Val27, with a kink around Gly16. Only the residues between Ala<sup>18</sup> and Val<sup>27</sup> interact with the ECD. The α-helical segment of GLP-1 is amphiphilic, allowing hydrophilic and hydrophobic interactions through opposite faces of the α-helix (Underwood et al., 2010). The hydrophilic face of GLP-1 comprises residues Gln17, Lys20, Glu21, and Lys<sup>28</sup> , of which only Lys<sup>20</sup> interacts directly with the ECD by forming a hydrogen bond with the side chain of Glu128. Interestingly, exendin also possesses a basic residue at position 20 (Arg20), allowing interaction with Glu<sup>128</sup> of the GLP1R ECD (Runge et al., 2008; Underwood et al., 2010). The hydrophobic face of GLP-1 includes Ala18, Ala19, Phe22, Ile23, Leu26, and Val27. The hydrophobic residues are exposed toward the complementary hydrophobic binding pocket in the ECD. Particularly, Phe22, Ile23, and Leu<sup>26</sup> of the peptide are found to interact with the highly conserved residues Val36, Trp39, Asp67, Tyr69, Arg121, and Leu<sup>123</sup> in the ECD of GLP1R (Underwood et al., 2010). The contribution of Phe<sup>22</sup> , Ile23, and Leu<sup>26</sup> to GLP1R binding has been demonstrated by Ala substitutions or mutations of these residues (Adelhorst et al., 1994; Wilmen et al., 1997). It is of interest to note that the GLP-1 family peptides exendin, GIP, GLP-2, and GCG also contain hydrophobic residues Phe22, Ile/Va23, and Leu26. Further, residues Trp39, Asp<sup>67</sup> , Tyr69, and Arg<sup>121</sup> in the GLP1R ECD are also highly conserved in GLP2R, GIPR, and GCGR (Parthier et al., 2007; Runge et al., 2008; Underwood et al., 2010;Venneti and Hewage, 2011). This observation suggests that these residues are evolutionarily conserved and likely contribute to the primary binding between the α-helix of the peptides and ECD of the receptors. This may also in part account for the cross-interaction of one α-helix of the peptide with the ECD of other partners (Parthier et al., 2007; Runge et al., 2008; Underwood et al., 2010). Indeed, this is supported by the observations

that chimeric peptides containing the GLP-1 N-terminus with the α-helix of GIP or GCG can induce GLP1R activation with a relatively high potency (Runge et al., 2003; Moon et al., 2010). However, the specific interaction of the α-helix of the peptide with the ECD of its own receptor may be of higher affinity than those with other related paralogous receptors. For instance, interaction of Ala<sup>19</sup> of GLP-1 with GLP1R-specific Leu32, and Gln<sup>19</sup> of GIP with GIPR-specific Ala32, may explain the higher affinity of each peptide toward its own receptor than toward paralogous receptors (Parthier et al., 2007; Underwood et al., 2010).

#### **INTERACTIONS BETWEEN THE N-TERMINUS OF THE PEPTIDE AND THE CORE DOMAIN OF THE RECEPTOR**

Specificity of ligand-receptor binding between a peptide and the corresponding receptor can be further conferred by binding between the N-terminus of the peptide and the receptor core domain. In addition, this interaction allows ligand-induced receptor activation (Thorens et al., 1993; Montrose-Rafizadeh et al., 1997; Hinke et al., 2001, 2003; During et al., 2003; Gault et al., 2007). Therefore, many approaches, such as alanine scanning, photoaffinity labeling, and molecular modeling-based approaches have explored the specific residues within the peptide and receptor responsible for ligand-receptor interaction (Adelhorst et al., 1994; Gallwitz et al., 1994; Xiao et al., 2000; Lopez de Maturana and Donnelly, 2002; Lopez de Maturana et al., 2004; Chen et al., 2009, 2010; Lin and Wang, 2009). Alanine scanning of GLP1R demonstrated that residues found between the TMH2 and ECL1 including Lys197, Asp198, Lys202, Met204, Tyr205, Asp215, and Arg<sup>227</sup> are likely important for the binding of the receptor to the Nterminal moiety of GLP-1, as mutations at these residues lead to a significant decrease in ligand affinity (Xiao et al., 2000; Lopez de Maturana and Donnelly, 2002; Lopez de Maturana et al., 2004). However, these observations did not define how these individual residues interact with in the N-terminal moiety of GLP-1. Further, Ala mutations in these residues can modify the receptor conformation which may interfere with binding to the ligand. Recently, using photoaffinity labeling, Chen et al. (2010) observed that Tyr<sup>205</sup> in ECL1 is in close proximity to the *p*-benzoyl-l-phenyl alanine (Bpa) at position 6 of GLP-1. However, the mutation of Tyr<sup>205</sup> to Ala in GLP1R does not alter either receptor activity or ligand binding, indicating that Tyr<sup>205</sup> is not the direct binding site for the N-terminal moiety of GLP-1. Furthermore, no ligand-bound crystal structure for the core domain of the class B GPCR family is currently available. Thus, our understanding of the molecular mechanism underlying the high affinity interaction between the N-terminal moiety of the peptide and the receptor core domain is primitive.

Recently, by comparing the amino acid sequences of GLP-1 and GLP1R with their orthologs and paralogs in vertebrates, we were able to obtain clues to help determine which amino acid residues may be responsible for ligand-receptor interaction (Moon et al., 2010, 2012). Although the GLP-1-related family of peptides shares a similarity in the amino acid sequence at the N-terminal moiety, there are specific, divergent amino acid sequences. For instance, the N-terminus of GLP-1 and its family peptides starts with either His<sup>1</sup> (for GCG, GLP-1, GLP-2, and exendin) or Tyr<sup>1</sup> (for GIP), and most of the peptides contain Thr at position 7 (GCG, GLP-1, and exendin), or Ser (GLP-2); only GIP contains Ile at this position. The first and seventh amino acid residues of each peptide are conserved within orthologs of vertebrate species. Thus, it may be postulated that His/Tyr<sup>1</sup> or Thr/Ile<sup>7</sup> residues of GLP-1 and GIP confer ligand selectivity to their cognate receptors. Indeed, our recent observation using chimeric GLP-1/GIP peptide reveals that the His/Tyr<sup>1</sup> and Thr/Ile<sup>7</sup> residues within these peptides confer differential ligand selectivity toward GIPR and GLP1R, respectively (Moon et al., 2010).

The chimeric GLP1R/GIPR approach together with chimeric GLP-1/GIP peptides offers a new strategy to determine which residues in the core domain are responsible for interacting with His<sup>1</sup> and Thr<sup>7</sup> of GLP-1 (Moon et al., 2012). For example, this approach enables us to determine crude motifs in TMH2, ECL1, and ECL2 of GLP1R which are likely to interact with His<sup>1</sup> and Thr<sup>7</sup> of GLP-1. Amino acid sequence comparison of these regions between those of tetrapod GLP1Rs, fish GCGRs, and vertebrate GIPRs further define amino acid residues that tentatively interact with His<sup>1</sup> and Thr<sup>7</sup> of GLP-1. In this case, we searched for residues which are conserved within GLP1R ortholog and fish GCGR but are different from those of GIPRs. We were able to identify Ile<sup>196</sup>

and Lys<sup>197</sup> of TMH2, and Met<sup>233</sup> of ECL1, and Asn<sup>302</sup> and Met<sup>303</sup> of ECL2 in GLP1R (Moon et al., 2012). It is noteworthy that fish GCGRs exhibit both a significantly high affinity for GLP-1 (Irwin and Wong, 2005) and these conserved residues, even though other regions are significantly different from tetrapod GLP1Rs (**Figure 3**). Mutational mapping together with application of chimeric GLP-1/GIP peptides reveals that His<sup>1</sup> -harboring peptides are sensitive for Asn<sup>302</sup> mutations, while Thr<sup>7</sup> -containing chimeric peptides are highly sensitive for the Ile<sup>196</sup> mutation, indicating a possible interaction of His<sup>1</sup> and Thr<sup>7</sup> of GLP-1 with Asn<sup>302</sup> and Ile<sup>196</sup> of GLP1R, respectively. Indeed, computer-aided molecular modeling showed interaction of His<sup>1</sup> with Asn<sup>302</sup> and of Thr<sup>7</sup> with a binding pocket formed by Ile196, Leu232, and Met<sup>233</sup> of GLP1R (Moon et al., 2012). Interestingly, Asn<sup>302</sup> is highly conserved at the corresponding position in GCGRs (Asn<sup>300</sup> for human) and GLP2Rs (Asn<sup>336</sup> for human), and these receptors have peptide ligands containing His<sup>1</sup> . In addition, GCGRs that respond to a Thr<sup>7</sup> -containing peptide ligands have conserved Val<sup>193</sup> and Met<sup>231</sup> at the corresponding positions of Ile196, and Met<sup>233</sup> of GLP1R, indicating that these residues may have a contact with Thr<sup>7</sup> of GCG. This possibility, however, needs to be further addressed.

vertebrate GLP1Rs, fish GCGRs (which are known to bind GLP-1), and human GIPR are shown in top and bottom panels. The residues in blue are conserved between GLP1Rs and GIPR. Residues shown in red are conserved sequences within the orthologous receptor of vertebrates. The

GLP-1, exendin, and GIP are shown in the middle. Residues which are identical among three peptides are in green. Residues common for GLP-1 and exendin are in blue. The interaction between peptides and receptors are indicated by dotted lines.

### **CONCLUSION**

Although GLP-1 may become a promising therapeutic agent for the treatment of type 2 diabetes mellitus and obesity, these peptide agonists cannot be administered orally due to their peptide nature. Thus, orally administered small molecules that regulate GLP1R need to be developed. However, a bottleneck slowing the development of these small molecules is the lack of information regarding the molecular structure of the ligand-bound GLP1R. Unfortunately, the crystal structure of the ligand-bound N-terminal ECD of GLP1R may not fully account for the interaction between peptide ligands and receptors. Rather, it is likely that the seven TMHs and ECLs of the receptors are more critical than the N-terminal ECD for peptide binding and receptor activation. Thus, exploring the domains or amino acid residues within TMHs and ECLs that confer ligand binding and receptor activation may greatly contribute to the design of the molecular model for the peptide ligand-receptor complex. In turn, this would facilitate the development of potent small molecules capable of regulating GLP1R. Determination of ligand-receptor interaction points (either by ligand-bound ECD crystal structure or by biochemical analysis using chimeric peptides and receptors) demonstrates that conserved residues across paralogous peptides tend to interact

# **REFERENCES**


and protein allows the effect of GLP-1 on glucose metabolism in the human hypothalamus and brainstem. *J. Neurochem.* 92, 798–806.


with conserved residues among the paralogous receptors while the same can be said of conserved residues within orthologous peptides that likely interact with conserved residues within orthologous receptors. Thus, evolutionary conservation across ortholog and specific diversification within paralogs may be required for maintaining the selective interaction of a peptide ligand with its cognate receptor. The crystal structure of ligand-bound GLP1R ECD explains how residues between Ala<sup>18</sup> and Val<sup>27</sup> interact with the ECD. Biochemical studies using chimeric peptides and receptors reveal receptor contact points only at His<sup>1</sup> and Thr<sup>7</sup> . Thus, the residues between the N-terminus through to position 17 of GLP-1 remain to be explored for their impact on binding to the receptor. Comparative biochemical approaches combined with computeraided molecular modeling will discover additional residues in the peptide and receptor sequences that help understanding the high affinity binding interaction between the ligand-receptor pair.

#### **ACKNOWLEDGMENTS**

This work was supported by grants (2011K00277) from the Brain Research Center of the 21st Century Frontier Research Program, the Brain Research Program of the National Research Foundation of Korea (NRF) funded by the Ministry of Education.

peptide-1: formation of a symmetric helical dimer. *FEBS Lett.* 515, 165–170.


the agonist-bound secretin receptor. *J. Biol. Chem.* 278, 48300–48312.


and the development and growth of pancreatic beta-cells. *Am. J. Physiol. Endocrinol. Metab.* 286, E875–E881.


cognitive functions, and insulin secretion during hyperinsulinemic, stepped hypoglycemic clamp experiments in healthy volunteers. *J. Clin. Endocrinol. Metab.* 87, 1239–1246.


exendin-4 is an agonist and exendin- (9-39) an antagonist of the receptor. *Diabetes* 42, 1678–1682.


glucagon-like peptide-1 receptorbinding determinants. *J. Mol. Endocrinol.* 25, 321–335.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 04 September 2012; paper pending published: 24 September 2012; accepted: 29 October 2012; published online: 19 November 2012.*

*Citation: Moon MJ, Park S, Kim D-K, Cho EB, Hwang J-I, Vaudry H and Seong JY (2012) Structural and molecular conservation of glucagon-like peptide-1 and its receptor confers selective ligandreceptor interaction. Front. Endocrin. 3:141. doi: 10.3389/fendo.2012.00141*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012Moon, Park,Kim,Cho, Hwang , Vaudry and Seong . This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Acting on hormone receptors with minimal side effect on cell proliferation: a timely challenge illustrated with GLP-1R and GPER

# **Véronique Gigoux \* and Daniel Fourmy**

Université de Toulouse, Université Paul Sabatier, Toulouse, France

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Lisa Chopin, Queensland University of Technology, Australia Terry Moody, National Cancer Institute, USA

#### **\*Correspondence:**

Véronique Gigoux, CHU Rangueil – INSERM, Université de Toulouse, Université Paul Sabatier, EA4552, 1 Avenue Jean Poulhès, BP 84225, 31432 Toulouse Cedex 4, France.

e-mail: veronique.gigoux@inserm.fr

G protein-coupled receptors (GPCRs) constitute a large family of receptors that sense molecules outside the cell and activate inside signal transduction pathways and cellular responses. GPCR are involved in a wide variety of physiological processes, including in the neuroendocrine system. GPCR are also involved in many diseases and are the target of 30% of marketed medicinal drugs.Whereas the majority of the GPCR-targeting drugs have proved their therapeutic benefit, some of them were associated with undesired effects.We develop two examples of used drugs whose therapeutic benefits are tarnished by carcinogenesis risks. The chronic administration of glucagon-like peptide-1 (GLP-1) analogs widely used to treat type-2 diabetes was associated with an increased risk of pancreatic or thyroid cancers. The long-term treatment with the estrogen antagonist tamoxifen, developed to target breast cancer overexpressing estrogen receptors ER, presents agonist activity on the G protein-coupled estrogen receptor which is associated with an increased incidence of endometrial cancer and breast cancer resistance to hormonotherapy. We point out and discuss the need of pharmacological studies to understand and overcome the undesired effects associated with the chronic administration of GPCR ligands. In fact, biological effects triggered by GPCR often result from the activation of multiple intracellular signaling pathways. Deciphering which signaling networks are engaged following GPCR activation appears to be primordial to unveil their contribution in the physiological and physiopathological processes. The development of biased agonists to elucidate the role of the different signaling mechanisms mediated by GPCR activation will allow the generation of new therapeutic agents with improved efficacy and reduced side effects. In this regard, the identification of GLP-1R biased ligands promoting insulin secretion without inducing pro-tumoral effects would offer therapeutic benefit.

**Keywords: receptor, G protein, arrestin, GLP-1, estrogen, biased ligands, adverse effects, cancer**

# **INTRODUCTION**

Seven transmembrane receptors, also termed G protein-coupled receptors (GPCR), form the largest class of the cell surface membrane receptors, involving 850 members in the human genome. GPCR are generally expressed in several different tissues in the same individual and involved in numerous physiological processes, including in the neuroendocrine system by playing a pivotal role in the control of feeding behavior, reproduction, growth, hydromineral homeostasis and stress response. At the cellular level, biological effects triggered by GPCR often result from the activation of multiple intracellular signaling pathways which are dependent or independent of G protein coupling (Rajagopal et al., 2010a).

Near 30% of therapeutic agents on the pharmaceutical market target GPCR (Hopkins and Groom, 2002). Whereas the majority of the GPCR-targeting drugs have proved their therapeutic benefit, some of them were associated with undesired effects (**Table 1**). Around 70% of the drugs which target GPCR are derived from the natural ligand and the use of agonist mimetics in clinical indication can act on the different tissues expressing the targeted GPCR and potentially induce undesired effects. Notably, prolonged treatment with GPCR-targeting agonist analogs was shown to induce preneoplastic and tumoral side effects. Here, we develop two examples of the use of GPCR-targeting drugs whose therapeutic benefits are tarnished by carcinogenesis risks. As a first example, glucagon-like peptide-1 receptor (GLP-1R) agonists used as anti-diabetic treatment were shown to induce preneoplastic lesions and/or cancers in the pancreas and the thyroid. The second example of ligands that we chose to develop does not initially target a GPCR, but the unexpected undesired effect is associated with a new GPCR target. Indeed, nuclear estrogen receptor antagonists such as tamoxifen are a breakthrough in the therapy and the prevention of breast cancer; however, long-term treatment was shown to be associated with an increased risk in endometrial cancer which was explained by the tamoxifen-induced activation of a GPCR, named G protein-coupled estrogen receptor (GPER).

We point out and discuss the need of more pharmacological studies to understand and overcome the undesired effects associated with the chronic administration of ligands which target GPCR. Deciphering the signaling networks engaged following GPCR activation appears to be primordial to unveil



The cells or organs targeted by the drug in the clinical indication or in the adverse effects are given in parenthesis.

their contribution in the physiological and physiopathological processes.

#### **THE GLUCAGON-LIKE PEPTIDE-1 RECEPTOR**

One of the main physiological roles of GLP-1 is to enhance insulin secretion in a glucose-dependent manner. Thus, GLP-1 is an incretin hormone released after meals by L cells in the intestine (**Figure 1**) (Mojsov et al., 1987). GLP-1 exerts its physiological effects through binding to its specific G protein-coupled receptor, GLP-1R, which is primarily and positively coupled to adenylate cyclase, through Gαs-containing heterotrimeric G proteins, leading to the activation of second messenger pathways such as protein kinaseA (PKA) and cAMP-regulated guanine nucleotide exchange factor II (cAMP-GEFII, also known as Epac2) signaling pathways (**Figure 2**) (Thorens, 1992; Kashima et al., 2001; Mayo et al., 2003; Holz, 2004; Seino and Shibasaki, 2005; Doyle and Egan, 2007; Holst, 2007). In addition to its stimulatory effect on insulin secretion, GLP-1 suppresses the secretion of glucagon, a counter-hormone to insulin, thus maintaining glucose homeostasis following a meal (Nauck et al., 1993). GLP-1 plays also a key role in the homeostasis of β-cell mass by inducing β-cell proliferation and protecting against apoptosis which favor an expansion of β-cell mass (**Figure 2**) (Doyle and Egan, 2007). These functions are mediated *via* the activation of the cAMP/PKA/CREB (cAMPresponsive element binding protein) and the transactivation of the EGF-R (epidermal growth factor receptor) leading to the activation of phosphatidylinositol-3 kinase (PI3K), Protein Kinase Cζ (PKCζ), Akt-protein kinase B, Extracellular Regulated Kinase (ERK1/2) signaling pathways and to the up-regulation of the expression of the cell cycle regulator cyclin D1 (Buteau et al., 2003; Drucker, 2003; Trumper et al., 2005; Park et al., 2006; Doyle and Egan, 2007). The antiapoptotic effect of GLP-1 in β-cells also involves β-arrestin1 recruitment by GLP-1R which mediates the ERK1/2 activation leading to the phosphorylation and inactivation of the pro-apoptotic protein Bad (Quoyer et al., 2010). The properties of GLP-1 on insulin secretion and β-cell proliferation make GLP-1 one of the most promising therapeutic agent to treat type-2 diabetes. Moreover, GLP-1 analogs offer the advantage of improved glycemic control of type-2 diabetic patients, without inducing severe hypoglycemia (Phillips and Prins, 2011).

On the other hand, GLP-1 receptor activation directly promotes cell proliferation and enhances cell survival in several tissues including neurons, fibroblasts, and cardiomyocytes (Brubaker and Drucker, 2004).

#### **COULD ANTI-DIABETIC TREATMENT WITH GLP-1 ANALOGS INDUCE CANCERS?**

Two GLP-1 mimetic drugs are now widely used to treat type-2 diabetes, exendin-4/exenatide and liraglutide, because of their optimal glucose lowering capacity with low risk of hypoglycemia (Chia and Egan, 2008; Buse et al., 2009; Nauck et al., 2009). Preclinical and clinical studies indicated that exenatide and liraglutide

exert a positive effect on insulin secretion, β-cell proliferation, and survival (Goke et al., 1993; Chang et al., 2003; Drucker, 2006;Vilsboll et al., 2007, 2008; Pratley and Gilbert, 2008; Madsbad, 2009; Vilsboll, 2009). On the other hand, recent studies showed that the use of these GLP-1R agonists in anti-diabetic treatment can be associated with an increase of cancer risk. The main organs where concerns exist about the trophic effects of GLP-1 analogs and their potential carcinogenic propensity are the pancreas and the thyroid, both organs expressing GLP-1R.

#### **The pancreas**

Recent studies reported that both treatments with exenatide and liraglutide are associated with an increased risk of pancreatitis in humans, a disease which represents a known risk factor for pancreatic cancer (Denker and Dimarco, 2006; Cure et al., 2008; Tripathy et al., 2008; Greer and Whitcomb, 2009). The chronic administration of GLP-1 agonists was also shown to be associated with increased serum lipase and amylase in many patients with type-2 diabetes, suggesting pancreatic damage and inflammation (Lando et al., 2012). Evaluation of the U.S. Food and Drug Administration (FDA) adverse events database by Elashoff et al. (2011), showed 10- and 3-fold increases in the incidence of pancreatitis and pancreatic cancer, respectively, in diabetic patients treated with exenatide as compared to other therapies (rosiglitazone, nateglinide, repaglinide, and glipizide) (Elashoff et al., 2011).

Undesired effects were also observed on different animal models. Indeed, chronic administration of exenatide during 12 weeks increased pancreatic acinar inflammation, sensitized to pancreatitis, and promoted pancreatic duct hyperplasia in rats or in the LSL-KrasG12D/+/Pdx1-Cre<sup>±</sup> murine model of pancreatic carcinogenesis (Nachnani et al., 2010; Gier et al., 2012a). The authors of this study related these adverse effects to the expression of GLP-1R in duct cells of the exocrine pancreatic tissue (Gier et al., 2012a). Whereas GLP-1R expression is clearly established in normal β-cells, its expression in the exocrine pancreas raises questions as it could be detected or not in the ductal or acinar cells according to the study (Horsch et al., 1997; Xu et al., 1999; Korner et al., 2007; Tornehave et al., 2008; Gier et al., 2012a). Importantly, inflammation and/or tissue damage can promote neoplasia by altering the fate of acinar and endocrine differentiated cells which can transdifferentiate to ductal cells, thus leading to ductal cell proliferation and preneoplastic lesion formation eventually progressing to pancreatic cancer (Jura et al., 2005; Means et al., 2005; Hernandez-Munoz et al., 2008; Gidekel Friedlander et al., 2009; Logsdon and Ji, 2009; Rebours et al., 2009; Perez-Mancera et al., 2012). Other studies carried on normal and diabetic mice and rats treated with exenatide or liraglutide with or without induction of experimental pancreatic injury did not find any relationship between incretin therapy and the development of pancreatic disease such as pancreatitis and pancreatic tumor (Koehler and Drucker, 2006; Koehler et al., 2009; Tatarkiewicz et al., 2010). But, in these last studies, GLP-1 agonists administration did not exceed 6 days or 4 weeks. Nevertheless, exenatide treatment upregulated PAP/Reg3b (pancreatitisassociated protein) expression as already observed in the course of pancreatic carcinogenesis and pancreatitis (Graf et al., 2006; Gigoux et al., 2008; Koehler et al., 2009; Tatarkiewicz et al., 2010). At last, Nyborg et al. (2012) did not observe pancreatitis in non-diabetic mice, rats, or monkeys after 2 years of liraglutide treatment at exposure levels up to 60 times higher than in humans.

There are very few data on GLP-1R-induced proliferative signaling in pancreatic duct cells. Gier et al. (2012a) showed that

cyclase leading to the activation of cAMP-regulated guanine nucleotide exchange factor II (cAMP-GEFII, also known as Epac2) signaling pathway. GLP-1 plays also a key role in the homeostasis of β-cell mass by inducing β-cell proliferation (blue) and protecting against apoptosis (red). These functions are mediated via the activation of the cAMP/PKA/CREB (cAMP-responsive element binding protein) and the transactivation of the epidermal growth factor receptor (EGF-R) leading to the activation of

exenatide induced proliferative signaling pathways in human pancreatic duct cell line by increasing CREB and ERK1/2 phosphorylation and cyclin D1 expression. ERK1/2 phosphorylation induced by exenatide is dependent of EGF-R activation (Buteau et al., 2003; MacDonald et al., 2003). Koehler and Drucker (2006) also showed that exenatide increased cAMP or induced ERK1/2 activation in some pancreatic cancer cell lines although the proliferation of these cell lines was not modulated.

#### **The thyroid**

Elashoff et al. (2011)showed a 4.7-fold increase in the incidence of thyroid cancer in diabetic patients treated with exenatide as compared to other therapies (rosiglitazone, nateglinide, repaglinide, and glipizide), by analyzing the U.S. FDA's database of reported adverse events. In contrast, Hegedus et al. (2011) reported no significant risk for the activation or growth of C-cell cancer in response to liraglutide over a 2-year period. Nevertheless, GLP-1R expression was found in thyroid glands of 20, 91, and 100%

GLP-1R agonists also improve β-cell function and survival during endoplasmic reticulum stress (purple) by enhancing of ATF-4 translation in a cAMP- and PKA-dependent manner, promoting the up-regulation of the endoplasmic reticulum stress markers CHOP and GADD34 expression and the dephosphorylation of eIF2α. Of note, there is considerable overlap between pathways induced by the GLP-1R activation. Reprinted from Gastroenterology (Baggio and Drucker, 2007).

of patients with papillary carcinoma, medullary thyroid cancer (MTC), and C-Cell hyperplasia, respectively (Gier et al., 2012b). GLP-1R could be also detected in human normal thyroids (Bjerre Knudsen et al., 2010; Gier et al., 2012b). Therefore, GLP-1 analogs might increase the risk of thyroid C-cell pathology, but this awaits confirmation in humans.

Preclinical studies carried out on rodents treated with liraglutide or exenatide showed a higher incidence of C-cell neoplasia and tumor formation in the thyroid [European Medicines Agency (EMA), 2006, 2009, 2011; Bjerre Knudsen et al., 2010; U.S. Food and Drug Administration, 2011; Bulchandani et al., 2012; Madsen et al., 2012; Victoza (Liraglutide) Injection, 2012]. Indeed, a continuous exposure to liraglutide or exenatide was associated with marked increases in plasma calcitonin and in the incidence of Ccell hyperplasia. These effects were mediated by the GLP-1R as they were not seen in GLP-1R knockout mice (Bjerre Knudsen et al., 2010; Madsen et al., 2012). C-cell hyperplasia is considered as a preneoplastic lesion that constitutes *in situ* carcinoma of the thyroid C-cells (LiVolsi,1997) and calcitonin, an hormone secreted by thyroid C-cells, is regarded as an important clinical biomarker for C-cell diseases such as MTC and hereditary C-cell hyperplasia because of its high sensitivity and specificity (Elisei et al., 2004; Costante et al., 2007; Machens et al., 2009). Neoplasms were not observed in monkeys after long-term liraglutide administration, indicating that GLP-1 induced C-cell proliferation in rodents but not in primates and suggesting that possible species-specific differences in GLP-1R expression and activation might occur in the thyroid (Bjerre Knudsen et al., 2010).

There are very few data on GLP-1R-induced proliferative signaling in thyroid C-cells. Chronic administration of liraglutide did not modify ERK phosphorylation, but increased ribosomal S6 phosphorylation, a downstream target of mTor and PI3K activation which plays a role in regulating cell proliferation and survival by growth factors (Sengupta et al., 2010; Madsen et al., 2012).

In conclusion, results obtained from preclinical and clinical studies tend to support a pro-tumoral action of GLP-1 in the pancreas and the thyroid, although few studies contradict this role. The relatively short time of chronic treatment with GLP-1 analogs in some studies could explain the absence of significative pro-tumoral effects. Moreover, this raises the question of whether GLP-1 can induce preneoplastic lesions and cancer alone or enable pre-existing lesions to progress to cancer. Further studies should be conducted to determine whether GLP-1 agonists induce or sensitize to pancreatic and thyroid diseases, by comparing chronic administration of GLP-1 mimetics in rodents presenting or not previous injury in the pancreas and the thyroid. However, it is important to note that diabetes is recognized to increase the incidence of pancreatitis and of a variety of cancers, including breast, pancreas, and colon cancers (Giovannucci et al., 2010; Girman et al., 2010; Pandey et al., 2011). Of note, GLP-1R is overexpressed in neuroendocrine pancreatic tumors, more particularly in insulinomas (Korner et al., 2007; Christ et al., 2010). In the current state of knowledge, GLP-1 agonists remain contra-indicated in patients with a personal or family history of MTC or multiple endocrine neoplasia type-2 (Anonymous, 2010; Victoza (Liraglutide) Injection,2012). Importantly,Risk Evaluation andMitigation Strategies program including a FDA safety warning published recommendations regarding the risk of thyroid cancer and pancreatitis after use of liraglutide and after dose increases (U.S. Food and Drug Administration, 2011).

Very few data are available on the proliferative intracellular pathways mediated by GLP-1R in pancreatic ductal cells and thyroid C-cells. Nevertheless, in the current state of knowledge, GLP-1R induces proliferation of these cells by same intracellular pathways as in the pancreatic β-cells. The identification of GLP-1 analogs that promote insulin secretion to treat type-2 diabetes without inducing pro-tumoral effects is therefore a timely challenging issue. Glucose-insulinotropic peptide (GIP) incretin could be also another alternative in type-2 diabetes treatment, especially as the GIP receptor (GIP-R) was not expressed in the normal thyroid and the exocrine pancreas unlike GLP-1R (Ahren, 2009;Waser et al., 2011, 2012). But, these clinical indication of GIP should be effective only after normalization of patient's glycemia which can restore the expression of GIP-R in β-cells (Holst et al., 1997; Vilsboll et al., 2002; Piteau et al., 2007; Younan and Rashed, 2007).

# **THE ESTROGEN RECEPTORS ER/GPER**

Estrogen hormone regulates the growth and the differentiation of many tissues playing a critical role in the development of the reproductive system but also in the nervous, immune, vascular, muscular, skeletal, and endocrine systems. The binding of 17β-estradiol, the natural endogenous estrogen, to the estrogen receptors ERα and ERβ (ER) is the main mechanism responsible for the diverse biological effects of the hormone (Pedram et al., 2006; Meyer and Barton, 2009; Meyer et al., 2009). These highly homologous receptors can shuttle between the cytoplasm and the nucleus and function as ligand-activated nuclear transcription factors that bind *cis*-acting estrogen response elements in the promoter and enhancer regions of hormonally regulated genes (genomic effects of estrogen) (**Figure 3**) (Ring and Dowsett, 2004; Edwards, 2005; Carroll and Brown, 2006). Estrogen also induces some rapid biochemical responses to estrogen stimulation which occur in seconds to minutes, such as the increase in intracellular free calcium and the activation of multiple intracellular kinases including ERK, PI3K, PKA, and PKC (non-genomic effects of estrogen) (Chen et al., 2008).

Estrogen is the one of the risk factors for breast tumors, which accounts for 40% of cancer among the women and approximately 50% of all breast cancers demonstrated elevated levels of ER expression (Pike et al., 2004). Consequently, anti-estrogen therapy has been extended such as the gold standard tamoxifen (**Figure 3**) (Deroo and Korach, 2006; Lorand et al., 2010). Unfortunately, long-term treatment with tamoxifen is associated with adverse effects such as an increased incidence of endometrial cancer and with breast cancer resistance to hormonotherapy. Moreover, these events were shown to be associated with G protein signalingor growth factor-mediated pathways which were not blocked by tamoxifen antagonist, leading to the prediction that an alternative membrane-bound estrogen receptor exists (Wehling, 1997; Hammes and Levin, 2007; Meyer and Barton, 2009). In fact, an orphan GPCR was identified as an estrogen-binding membrane GPCR from vascular and cancer cells and is now included in the official GPCR nomenclature and was designated GPER or GPR30 by the International Union of Pharmacology (Revankar et al., 2005; Thomas et al., 2005; Prossnitz et al., 2008a; Alexander et al., 2011). Its localization seems to be predominantly intracellular due to the constitutive internalization of plasma membrane GPER (**Figure 3**) (Revankar et al., 2005; Otto et al., 2008; Cheng et al.,2011; Sanden et al.,2011). GPER is widely expressed in cancer cell lines and primary tumors of the breast (Carmeci et al., 1997; Filardo et al., 2000; Revankar et al., 2005; Albanito et al., 2008a), endometrium (Vivacqua et al., 2006a; Leblanc et al., 2007; He et al., 2009), ovaries (Albanito et al., 2007, 2008b; Henic et al., 2009), thyroid (Vivacqua et al., 2006b), lung (Siegfried et al., 2009), prostate (Chan et al., 2010), and testicular germ cells (Franco et al., 2011). GPER does not only bind estrogens but also other substances such as tamoxifen which displays estrogenic agonist activity on GPER notably in the reproductive systems (**Figure 3**) (McDonnell, 1999; Filardo et al., 2000; Thomas and Dong, 2006; Jordan, 2007; Albanito et al., 2008b; Orlando et al., 2010; Chevalier et al., 2012).

accepted as mainly mediating gene transcriptional regulation. Tamoxifen is an ER antagonist in some tissue, such as breast cancer, while has agonistic effects in other tissues, such as endometrium. GPER was found

Indeed, tamoxifen stimulates the cell proliferation and growth of cell lines of thyroid, ovarian, endometrial, and breast cancers (Filardo et al., 2000; Thomas et al., 2005; Vivacqua et al., 2006a; Albanito et al., 2007; Prossnitz et al., 2008b; Pandey et al., 2009). The discovery of GPER-selective agents and the elaboration of GPER knockout mice helped to examine GPER signaling pathways and strongly supported that GPER is associated with cancer proliferation, migration, invasion, metastasis, differentiation, prognosis, and drug resistance (Prossnitz et al., 2008b; Wang et al., 2010).

#### **COULD ESTROGEN ANTAGONISTS USED IN BREAST CANCER TREATMENT INDUCE CANCER IN OTHER TISSUES?**

An increased incidence of uterine malignancies in association with tamoxifen treatment has been reported. The incidence and severity of endometrial cancer increased by 4- to 6.9-fold in women PI3K, and PLC, and other rapid cellular processes. Most of them are mediated by transactivation of EGF-R. Reprinted from Endocrinology (Wang et al., 2010).

with 5 years of exposure to tamoxifen (van Leeuwen et al., 1994; Bernstein et al., 1999; Bergman et al., 2000; Goldstein, 2001). Uterine sarcoma has been also reported to occur more frequently among long-term users (≥2 years) of tamoxifen than non-users (Wickerham et al., 2002). In support to these data, tamoxifen has been shown to stimulate the proliferation and the invasion of uterine cells *in vivo* and of human endometrial carcinoma cell lines and these effects were mediated by GPER (Gottardis et al., 1988; Jamil et al., 1991; Schwartz et al., 1997; Du et al., 2012a). Indeed, tamoxifen promoted cell proliferation and invasion of the human endometrial cancer cell lines ISHIKAWA and KLE, while the down-regulation of GPER partly or completely prevented these effects (Du et al., 2012a). GPER is widely expressed in primary tumors of endometrium including ER-negative endometrial carcinomas (He et al., 2009). High levels of GPER expression correlate with an increased incidence of endometrial cancer and with

tamoxifen-induced uterine pathology and predict poor survival in endometrial cancer (Smith et al., 2007; Ignatov et al., 2010a). All together, these data strongly support that tamoxifen treatment might have a cancer-promoting effect through GPER.

G protein-coupled estrogen receptor promotes carcinogenesis by endometrial cancer cells as down-regulation of GPER led to reduce growth and invasion by RL95 endometrial cancer cells treated with 17β-estradiol and to decrease tumorigenesis *in vivo* (He et al., 2009, 2012). GPER mediates the proliferative effects of estrogen and tamoxifen in endometrial cancer cells through EGF-R transactivation leading to the activation of ERKs and PI3K pathways (Vivacqua et al., 2006a; Prossnitz et al., 2008b; He et al., 2009, 2012; Du et al., 2012a; Lappano et al., 2012;Wei et al., 2012). GPER also mediates invasion by endometrial cancer cells through the stimulation of ERK pathway, as well as the increase of interleukin-6 secretion, leading to the production and activation of matrix metalloproteinases MMP-2 and MMP-9 known to degrade extracellular matrix components and to be involved in cancer invasion and metastasis (He et al., 2009, 2012; Du et al., 2012a).

#### **COULD GPER BE INVOLVED IN BREAST CANCER RESISTANCE TO HORMONOTHERAPY?**

The majority of breast cancers is ER-positive and depends on estrogen for growth. Therefore, blocking estrogen signaling remains the strategy of choice for the treatment and the prevention of breast cancer. Tamoxifen is the prototypical drug that targets ER. It presents potent anti-estrogenic properties and has been used extensively for the past 40 years to treat and prevent breast cancer (Jordan and Morrow, 1999). Tamoxifen treatment is very effective in tumors expressing ER receptors and significantly reduces the mortality of breast cancer patients (Jordan and Morrow, 1999; Powles et al., 2007). Many patients with ER-positive breast cancer have benefited from anti-hormonal treatment, but unfortunately, almost 30–50% of patients with advanced disease did not respond to first-line treatment with tamoxifen. Furthermore, long-term tamoxifen therapy causes the development of acquired resistance (Early Breast Cancer Trialists' Collaborative Group (EBCTCG), 2005). Indeed, development of resistance is very frequent and tamoxifen is not effective for more than 5 years (Saphner et al., 1996; Clarke et al., 2001; Early Breast Cancer Trialists' Collaborative Group (EBCTCG), 2005; Barron et al., 2007; Brewster et al., 2008).

The tumor resistance to tamoxifen treatment is associated to a decrease or a loss of ER expression and to an increase of GPER expression. GPER protein is expressed in ∼50% of all breast cancers including half of ER-negative tumors and correlates with increased tumor size and metastasis (Filardo et al., 2006; Ignatov et al., 2011). Moreover, GPER protein expression is increased in breast tumors of patients treated only with tamoxifen and in tamoxifen resistant tumor tissues correlating with a poor relapsefree survival in patients treated with tamoxifen (Filardo et al., 2006; Ignatov et al., 2011). *In vitro* prolonged tamoxifen treatment leads to an increased cell surface expression of GPER and also to clonal selection of GPER-positive MCF-7 breast cancer cells (Ignatov et al., 2010b). Thus, GPER expression is associated with an increased risk of resistance to tamoxifen and patients with breast cancer who have high GPER protein expression should not be treated with tamoxifen alone.

G protein-coupled estrogen receptor mediates the proliferative and tamoxifen-resistance effects through EGF-R transactivation leading to the phosphorylation of ERK and Akt (Filardo et al., 2000; Prossnitz et al., 2008b; Ignatov et al., 2010a,b). Thus, ERK andAkt canfurther stimulate transcription of different genes (even ER), leading to cell proliferation, and interfere with the activation of Smad proteins, known effectors of the TGF-β signaling, an important intracellular pathway involved in the inhibition of tumor progression (Clarke et al., 2001; Kleuser et al., 2008; Yoo et al., 2008; Ignatov et al., 2010b).

In conclusion, tamoxifen has been the only available hormonal option for the systemic treatment for breast cancer from 1973 to 2000. Despite the clinical success of tamoxifen, the development of drug resistance and endometrial cancers leads to the requirement of alternative hormonal therapy. In this regard, the knowledge of the contribution of GPER-mediated signaling in the undesired effects of estrogenic antagonist uses for breast cancer treatment should allow the future development of new molecules. Moreover, further researches are required to define the role of GPER signaling in estrogen undesired physiological effects and to elucidate the role of non-selective estrogen receptor ligands in health and disease.

#### **NEW HOPES TO OVERCOME UNDESIRED EFFECTS**

G protein-coupled receptors are generally expressed in several different tissues and involved in numerous physiological processes. Many natural ligands can bind and activate several subtypes of GPCR. This is illustrated, for example, with cholecystokinin and somatostatin receptors (Guillermet-Guibert et al., 2005; Dufresne et al., 2006). Ligands can also activate different classes of receptors as illustrated before with estrogen (Prossnitz et al., 2008a). Such a diversity of receptors activation following agonist administration can engage multiple intracellular signaling pathways and be responsible for adverse effects in treated patients. Furthermore, biological effects triggered by the same GPCR result from the activation of G protein-dependent and -independent intracellular signaling pathways. Recently, signaling engaged after GPCR recruitment of β-arrestin proteins have emerged as new G protein-independent intracellular signaling pathways (Luttrell and Gesty-Palmer, 2010; Rajagopal et al., 2010a). To increase the complexity, a single GPCR has pleiotropic signaling properties and each signal can crosstalk at different levels with the transactivation of cell surface receptor having tyrosine kinase activity (EGF-R, PDGF-R, FGF-R, for examples) or serine/threonine kinase activity (TGF-β, for example) or with the formation of multimers, thus potentially influencing the signaling pathways of the different receptors (Burch et al., 2012; Wang and Lewis, 2013). Indeed, numerous biochemical and biophysical studies supports that GPCRs can form physiologically relevant homo-, hetero-, or oligo-mers (Angers et al., 2002). Homodimerization of the GLP-1R was shown to be critical for selective coupling of the receptor to physiologically relevant signaling pathways (Harikumar et al., 2012). Indeed, disruption of GLP-1R homodimerization completely abrogated the intracellular calcium mobilization response whereas it slightly reduced cAMP formation and phosphorylation of ERK. Furthermore, GLP-1R dimerization can discriminate between peptide and non-peptide-mediated receptor activation. In the chemokine receptors family, antibodies

against the CCR2b promoted the receptor dimerization and second messenger production (Rodriguez-Frade et al., 1999), whereas an antibody directed against CCR5, that induces receptor dimerization, inhibits its function (Vila-Coro et al., 2000). Many studies was also conducted to analyze the role of GPCR heterodimerization and supported that heterodimerization could be the source of additional pharmacological properties which are different from those of the individual receptors. As a first example, the co-expression of the δ- and κ-opioid receptors in the same cell leads to an almost complete loss of binding to selective δ- and κ-ligands while preserving binding to nonselective ligands (Jordan and Devi, 1999). As a second example, somatostatin receptor SSTR1 displays internalization in cells when it is co-expressed with SSTR5, whereas monomeric SSTR1 is resistant to internalization in contrast to monomeric SSTR5, suggesting that the SSTR1 trafficking is modified by its heterodimerization with SSTR5 (Rocheville et al., 2000). Thus, homo- and heterodimerization between GPCR cause complexity in the receptor pharmacological properties that can be responsible of synergistic or antagonistic signaling cross-talks. This GPCR pharmacological and signaling complexity could account for unexpected pharmacological effects and have dramatic impacts on drug development. All together, these hallmarks indicate that undesired adverse effects can be expected with a prolonged agonist administration that targets a GPCR (**Table 1**). Moreover, many GPCR have already been shown to present proliferative and pro-tumoral properties (**Table 2**), suggesting that an increase of preneoplastic lesions and cancer incidence can potentially occur following chronic activation of GPCR. Thus, deciphering which signaling networks are engaged and orchestrated following GPCR receptor activation appears to be primordial to unveil their contribution in the cell fate.

One strategy to overcome these limitations would be to examine the initial steps following receptor activation. The release of X-ray structures of agonist/GPCR complexes (Chung et al., 2011; Lebon et al., 2011; Warne et al., 2011; Xu et al., 2011; Audet and Bouvier, 2012), the numerous biophysical and biochemical studies (Granier et al., 2007; Kahsai et al., 2011; Liu et al., 2012; Rahmeh et al., 2012) have enabled to show that different and selective ligands, named biased ligands, can induce or stabilize distinct receptor conformations and activate one (or several) signaling pathway(s) in contrast to non-biased agonists which activate all the signaling pathways (Vaidehi and Kenakin, 2010). Thus an understanding of the structure and dynamics of the ensemble of receptor conformations would greatly help the design of small molecules with functional selectivity or "biased signaling" properties and would provide more specific and efficient new drugs. Receptor structure/activity relationship studies, structureand docking-based virtual screening are now widely applied in drug discovery and must take in account the existence of different receptor conformations activating specific signaling pathways.

Although GPCRs can modulate a large variety of distinct signaling pathways, classification of biased ligands are actually restricted to two groups depending on their ability to activate two main transduction pathways (Whalen et al., 2011): (1) G protein-biased ligands which promote G protein activation without β-arrestin recruitment and (2) β-arrestin-biased ligands which recruit β-arrestin to the receptor and initiate consecutive signaling pathways in the absence of G protein activation.

The vast majority of biased ligands identified so far exhibits exclusive β-arrestin activity for a number of receptors (Rajagopal et al., 2010a; Whalen et al., 2011), including the AT1 angiotensin II receptor, β1- and β2-adrenergic receptors, or the CXCR7 decoy receptor (Wei et al., 2003; Wisler et al., 2007; Kim et al., 2008; Rajagopal et al., 2010b). The parathyroid hormone (PTH) analog, D-Trp(12),Tyr(34)-PTH(7-34), binds the PTH receptor 1 (PTHR1) and activates β-arrestin-dependent but not classical G protein-dependent signaling (Gesty-Palmer et al., 2009; Gesty-Palmer and Luttrell, 2011). In mice, this PTH biased agonist induces anabolic bone formation without stimulating bone resorption, comparatively with the non-selective agonist PTH(1- 34) which induces both functions. Thus, this PTHR1 biased ligand may present interesting properties for the treatment of metabolic bone diseases such as osteoporosis and is a proof of concept that the exploitation of β-arrestin biased agonism may offer therapeutic benefit.

Few ligands have been yet identified as perfect G protein-biased ligands, namely inducing G protein signal transduction without any β-arrestin recruitment (Whalen et al., 2011). GMME1 ligand binding to the CCR2 chemokine receptor leads to calcium mobilization, caspase-3 activation and consecutive cell death, but does not recruit β-arrestin2 (Rafei et al., 2009). Selective ligands that activate G protein-coupling by FSH-R (follicle-stimulating hormone receptor) and PTH-1R have been also reported (Bisello et al., 2002; Wehbi et al., 2010). Of note, some ligands classified as G protein-biased can induce a weak β-arrestin recruitment by the targeted GPCR (Whalen et al., 2011). For example, oxyntomodulin and glucagon are full agonists in GLP-1R-mediated cAMP accumulation but partial agonists in recruiting β-arrestins to this receptor, suggesting that oxyntomodulin and glucagon are biased ligands on the GLP-1R (Jorgensen et al., 2007).

Interestingly, some ligands are biased in regard to the different G protein families and can trigger opposite cellular responses (Reversi et al., 2005; Sensken et al., 2008). For example, oxytocin receptors (OTR) coupling to Gi inhibits cell proliferation, whereas its coupling to Gq stimulates cell proliferation. Atosiban, an oxytocin derivative, was shown to act as a competitive antagonist on OTR/Gq coupling, and to display agonistic properties on OTR/Gicoupling, thereby leading to the selective inhibition of cell growth (Reversi et al., 2005; Busnelli et al., 2012). SOM230 which activates the somatostatin receptor sst2A behaves as agonist for Gi coupling and inhibition of adenylyl cyclase, but antagonizes somatostatin's actions on intracellular calcium and ERK phosphorylation which can be activated by a Gi/Go independent process (Cescato et al., 2010).

Biased signaling can also exist with respect to other signaling proteins than G proteins and arrestins. The internalization of apelin receptor takes different signaling pathways depending of the apelin isoforms (Lee et al., 2010). Indeed, apelin-13-activated receptors dissociated rapidly from β-arrestin1 and were recycled to the cell surface through a Rab4-dependent mechanism, while the apelin-36-internalized receptors trafficked with β-arrestin1 to intracellular compartments and were targeted by Rab7 to lysosomes for degradation. CCL19 and CCL21 ligands both induce



β-arrestin2 recruitment by the receptor CCR7, but activate different GRK (G protein receptor kinase) isoforms (Zidar et al., 2009). Indeed, CCL19 leads to robust CCR7 phosphorylation and β-arrestin2 recruitment catalyzed by both GRK3 and GRK6 whereas CCL21 activates GRK6 alone. The functional consequences are that only CCL19 leads to classical receptor desensitization whereas both agonists are capable of signaling through GRK6 and β-arrestin2 to ERK kinases.

# **THE GLP-1R**

Results obtained from preclinical and clinical studies tend to support a pro-tumoral action of GLP-1 in the pancreas and the thyroid. Deciphering the signaling networks engaged following GLP-1R agonist administration in pancreatic ductal cells and thyroid C-cells comparatively to pancreatic β-cells is critical to unveil their contribution in the different cellular processes. The identification of GLP-1 analogs that promote insulin secretion to treat type-2 diabetes without inducing pro-tumoral effects is therefore a timely challenging issue. Like most GPCRs, the GLP-1R couples to different classes of heterotrimeric G proteins, including Gαs, Gαq, and Gαi, regulatory proteins such as the β-arrestins, and activates multiple signaling pathways such as cAMP production, intracellular calcium mobilization, phosphorylation of ERK1/2. While GLP-1 can activate all of these signaling pathways, some compounds were shown to present biased activity on GLP-1R. Oxyntomodulin and glucagon biased the GLP-1R toward cAMP accumulation over the recruitment of β-arrestins, BMS21 compound toward ERK1/2 activation and cAMP production over β-arrestins recruitment, BETP compound toward calcium mobilization and β-arrestins recruitment over cAMP production and ERK1/2 activation (Jorgensen et al., 2007; Wootten et al., 2013). Moreover, some molecules acting as allosteric modulators were shown to modulate GLP-1R agonist-mediated signaling pathways (Willard et al., 2012; Wootten et al., 2013). For example, BETP increases the affinity of GLP-1R to oxyntomodulin and potentiates the activation of cAMP production induced by oxyntomodulin.

The most crucial GLP-1R signaling pathway for enhancing glucose-dependent insulin secretion involved the receptor coupling to Gαs proteins and the activation of cAMP production (Baggio and Drucker, 2007). While GLP-1 analogs are currently tested for their capacity to activate Gαs protein and cAMP production, their effects on other signaling pathways, particularly those involved in cell proliferation, should be included. Moreover, the analysis of pharmacological ligand properties should be done on the main cellular target, the pancreatic β-cell, but also on cells involved in carcinogenic side effects. This could enable the design and development of improved therapeutics that have the ability to fine-tune receptor signaling leading to beneficial therapeutic outcomes while reducing side effect profiles. The use of allosteric ligands in addition to GLP-1R biased agonists could

#### **REFERENCES**


also provide a therapeutic advantage to target a specific receptor response toward signaling pathways promoting insulin secretion over cell proliferation.

#### **THE GPER**

Despite the clinical success of tamoxifen in breast cancer treatment, the development of drug resistance and endometrial cancers involving the GPER leads to the requirement of alternative hormonal therapy. In this regard, the contribution of GPERmediated responses estrogen antagonists must be considered in the future development of anti-estrogenic molecules. Recent studies based on pharmacological structure/function relationship on ER and/or GPER have identified selective GPER antagonists which completely block uterine epithelial cell proliferation mediated by GPER and which are poorly active or inactive on ER (Dennis et al., 2009, 2011; Burai et al., 2012). Future studies utilizing GPER-selective ligands will further define the role of this receptor *in vivo* and open the door to the generation of diagnostics and therapeutics directed at individual or both estrogen receptors. Such compounds might represent an important new approach for cancer therapy, thus increasing the armamentarium of drugs used to treat estrogen-sensitive and resistant cancers. On the other hand, aromatase inhibitors which act by preventing the enzyme aromatase to convert androgens into estrogen have been also brought forward as a potential alternative (Josefsson and Leinster, 2010; Abdulkareem and Zurmi, 2012).

In conclusion, GPCRs provide huge therapeutic opportunities; some are already in use. The progress in the knowledges of signaling pathways downstream of these receptors and the effects arising, their regulation by pharmacological agents, and the data from the receptor structure provide new opportunities which should lead to new generation of ligands with minimized side effects.

emerging concept for G proteincoupled receptor ontogeny and function. *Annu. Rev. Pharmacol. Toxicol.* 42, 409–435.


Group. Assessment of Liver and Endometrial cancer Risk following Tamoxifen. *Lancet* 356, 881–887.


G-protein-coupled receptor superfamily associated with estrogen receptor expression in breast cancer. *Genomics* 45, 607–617.


al. (2011). Conformational changes in the G protein Gs induced by the beta2 adrenergic receptor. *Nature* 477, 611–615.


2 diabetes or nondiabetic obese subjects treated with the human GLP-1 analog, liraglutide. *J. Clin. Endocrinol. Metab.* 96, 853–860.


al. (2009). Neuromedin U is overexpressed in pancreatic cancer and increases invasiveness via the hepatocyte growth factor c-Met pathway. *Cancer Lett.* 277, 72–81.


ICI 182 780 on survival of uterine cancer cell lines in the presence and absence of exogenous estrogens. *Int. J. Oncol.* 30, 477–487.


endocrine therapies for breast cancer. *Cancer Treat. Rev.* 36(Suppl. 3), S67–S71.


diabetes: role of GLP-1 receptor agonists and DPP-4 inhibitors. *Rev. Diabet. Stud.* 5, 73–94.


cancer cells. *Endocrinology* 146, 624–632.


and humans: relevance for incretinbased diabetes therapy. *Neuroendocrinology* 94, 291–301.


Christopoulos, A., et al. (2013). Differential activation and modulation of the glucagon-like peptide-1 receptor by small molecule ligands. *Mol. Pharmacol.* 83, 822–834.


replication and neogenesis, resulting in increased beta-cell mass and improved glucose tolerance in diabetic rats. *Diabetes* 48, 2270–2276.


effect of gastric inhibitory polypeptide (GIP) in obese and diabetic rats is related to the downregulation of its pancreatic receptors. *Gen. Physiol. Biophys.* 26, 181–193.

Zidar, D. A., Violin, J. D., Whalen, E. J., and Lefkowitz, R. J. (2009). Selective engagement of G protein coupled receptor kinases (GRKs) encodes distinct functions of biased ligands. *Proc. Natl. Acad. Sci. U.S.A.* 106, 9649–9654.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 11 November 2012; accepted: 10 April 2013; published online: 29 April 2013.*

*Citation: Gigoux V and Fourmy D (2013) Acting on hormone receptors with minimal side effect on cell proliferation: a timely challenge illustrated with GLP-1R and GPER. Front. Endocrinol. 4:50. doi: 10.3389/fendo.2013.00050*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2013 Gigoux and Fourmy. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

REVIEW ARTICLE published: 08 October 2012 doi: 10.3389/fendo.2012.00121

# The neuroendocrine functions of the parathyroid hormone 2 receptor

#### **Arpád Dobolyi <sup>1</sup>\*, Eugene Dimitrov <sup>2</sup> , Miklós Palkovits <sup>1</sup> and Ted B. Usdin<sup>2</sup>**

<sup>1</sup> Neuromorphological and Neuroendocrine Research Laboratory, Department of Anatomy, Histology and Embryology, Hungarian Academy of Sciences, Semmelweis University, Budapest, Hungary

<sup>2</sup> Section on Fundamental Neuroscience, National Institute of Mental Health, National Institute of Health, Bethesda, MD, USA

#### **Edited by:**

Hubert Vaudry, University of Rouen, France

#### **Reviewed by:**

Tamas Kozicz, Radboud University Nijmegen, Netherlands Sarah J. Spencer, Monash University, Australia

Kazuhiro Nakamura, Kyoto University, Japan

#### **\*Correspondence:**

Arpád Dobolyi, Neuromorphological and Neuroendocrine Research Laboratory, Department of Anatomy, Histology and Embryology, Hungarian Academy of Sciences, Semmelweis University, T˝uzoltó u. 58, Budapest H-1094, Hungary. e-mail: dobolyi@ana.sote.hu

The G-protein coupled parathyroid hormone 2 receptor (PTH2R) is concentrated in endocrine and limbic regions in the forebrain. Its endogenous ligand, tuberoinfundibular peptide of 39 residues (TIP39), is synthesized in only two brain regions, within the posterior thalamus and the lateral pons.TIP39-expressing neurons have a widespread projection pattern, which matches the PTH2R distribution in the brain. Neuroendocrine centers including the preoptic area, the periventricular, paraventricular, and arcuate nuclei contain the highest density of PTH2R-positive networks. The administration of TIP39 and an antagonist of the PTH2R as well as the investigation of mice that lack functional TIP39 and PTH2R revealed the involvement of the PTH2R in a variety of neural and neuroendocrine functions. TIP39 acting via the PTH2R modulates several aspects of the stress response. It evokes corticosterone release by activating corticotropin-releasing hormone-containing neurons in the hypothalamic paraventricular nucleus. Block of TIP39 signaling elevates the anxiety state of animals and their fear response, and increases stress-induced analgesia. TIP39 has also been suggested to affect the release of additional pituitary hormones including arginine-vasopressin and growth hormone. A role of the TIP39-PTH2R system in thermoregulation was also identified. TIP39 may play a role in maintaining body temperature in a cold environment via descending excitatory pathways from the preoptic area. Anatomical and functional studies also implicated the TIP39-PTH2R system in nociceptive information processing. Finally, TIP39 induced in postpartum dams may play a role in the release of prolactin during lactation. Potential mechanisms leading to the activation ofTIP39 neurons and how they influence the neuroendocrine system are also described.The unique TIP39-PTH2R neuromodulator system provides the possibility for developing drugs with a novel mechanism of action to control neuroendocrine disorders.

**Keywords: neuropeptide, thermoregulation, stress response, corticotropin-releasing hormone, somatostatin, neuroendocrine hypothalamic regulations, reproductive regulations, maternal adaptation**

# **INTRODUCTION**

The parathyroid hormone 2 receptor (PTH2R) is a member of the family B (type II) of G-protein coupled receptors. It was discovered based on its sequence similarity to other proteins belonging to this receptor family (Usdin et al., 1995). The novel receptor was named PTH2R because of its sequence similarity to the parathyroid hormone receptor and also because the human PTH2R can be activated by parathyroid hormone (Usdin et al., 2002). In rat, however, nanomolar concentrations of parathyroid hormone do not cause significant activation of the PTH2R (Hoare et al., 1999a; **Figure 1**). An additional difference between the parathyroid hormone 1 receptor (PTH1R) and the PTH2R is that the distinct polypeptide parathyroid hormone-related peptide is a coligand of the PTH1R but does not bind to the PTH2R (Hoare et al., 1999b). A breakthrough in the field of parathyroid hormone receptor research was the discovery of a novel peptide, tuberoinfundibular peptide of 39 residues (TIP39), an endogenous ligand of the PTH2R (Usdin et al., 1999b). TIP39 was purified from bovine hypothalamus based on its ability to elevate cAMP in a

PTH2R-expressing cell line (Usdin et al., 1999b). TIP39's sequence has veryfew amino acid residues in common with parathyroid hormone and parathyroid hormone-related peptide but it does have a similar three-dimensional structure to them (Piserchio et al., 2000). TIP39 is a high affinity and fully potent agonist for both the human and rodent PTH2R (Usdin et al., 1999b). Apart from elevating cAMP (presumably *via* Gs proteins), TIP39 was also shown to elevate intracellular Ca2<sup>+</sup> levels (presumably *via* Gq proteins) in some cell types (Goold et al., 2001; Della Penna et al., 2003).

PTH2R expression is greater in the brain than in peripheral tissues based on Northern blot, *in situ* hybridization histochemistry and immunohistochemistry (Usdin et al., 1995, 1996, 1999a). In the periphery, its expression pattern was also very different from that of the PTH1R, as only a low level of expression was found distributed in the following places: pancreatic islet somatostatin synthesizing D cells, large vessels in bronchi, and the parenchyma in the lung, cardiac endothelium, a small number of cells associated with the vascular pole of renal glomeruli, spermatids in the

head of the epididymis, atretic follicles of the ovary, chondrocytes in thyroid cartilage, a small number of cells in bone, and in some endocrine cells including thyroid parafollicular C cells, and some gastrointestinal peptide synthesizing cells. There is relatively little information on the effects of PTH2R signaling outside the brain and potential functions in the periphery are not discussed in this review.

Within the brain, the distributions of TIP39 axon terminals and PTH2R immunoreactivity show remarkable similarities as described below. Therefore, it was suggested that TIP39 is the endogenous ligand of the PTH2R. Consequently, it has been proposed that TIP39 and the PTH2R form a neuromodulator system in many brain regions. This is supported by the very similar phenotypes of mutant mice lacking either functional PTH2Rs or TIP39 or wild type (WT) animals administered a PTH2R antagonist as discussed below.

#### **THE DISTRIBUTION OF THE TIP39-PTH2R SYSTEM**

#### **NEURONS EXPRESSING TIP39, THE LIGAND OF THE PTH2R**

TIP39-expressing cells in the adult brain are restricted to the subparafascicular area of the thalamus and the medial paralemniscal nucleus of the pons as revealed by immunohistochemistry and *in situ* hybridization histochemistry (Dobolyi et al., 2003b) while the amygdala-hippocampal transitional zone also contains some TIP39 neurons during embryonic development (Brenner et al., 2008). The subparafascicular area TIP39 neurons were subdivided into the medially located group in the periventricular gray of the thalamus (PVG) and a laterally positioned group of the posterior intralaminar complex (PIL) of the thalamus (PIL; Dobolyi et al., 2010; **Figure 2**). Recent evidence supports the idea that the anatomical separation of these cell groups is concomitant with different functions. TIP39 appears in neurons in the PIL earlier during ontogenic development and largely disappears from them immediately after birth (Brenner et al., 2008). In turn, a marked induction of TIP39 in PIL cells but not in PVG cells can be observed in postpartum dams (Cservenak et al., 2010). Although the possibility that the PTH2R has another endogenous ligand apart from TIP39 is theoretically not excluded, the

similarity in their distributions argues that TIP39 is available to activate PTH2Rs wherever they are present in the brain. Thus, neurons in the three sites of TIP39 expression would provide all information necessary to act *via* the PTH2R in the brain. Some evidence is available from lesion and tracer studies that TIP39 neurons in the subparafascicular area project to forebrain limbic and hypothalamic regions while TIP39 neurons in the medial paralemniscal nucleus provides TIP39 fibers to the hindbrain and spinal cord potentially affecting auditory and nociceptive functions (Dobolyi et al., 2003a; Wang et al., 2006b; Palkovits et al., 2010).

#### **TIP39 AND THE PTH2R IN THE PITUITARY AND MEDIAN EMINENCE**

Neither TIP39 nor PTH2R mRNA or protein were found in the pituitary (Usdin et al., 1999a; Dobolyi et al., 2002). In contrast, PTH2R-containing fibers were abundant in the external zone of the median eminence. PTH2R was found here in somatostatin fibers but not in fibers containing growth hormone (GH)-, gonadotropin-, or corticotropin-releasing hormones (CRH), or arginine-vasopressin (AVP; Dobolyi et al., 2006a). As opposed to the PTH2R, only a few TIP39-ir fibers were present in the median eminence. This is in fact the most striking difference between the otherwise very similar distributions of TIP39 axon terminals and the PTH2R.

#### **THE SIMILARITIES BETWEEN THE DISTRIBUTION OF THE PTH2R AND TIP39 IN THE BRAIN**

The distribution of PTH2R fibers and PTH2R-expressing cells is generally similar in the brain and particularly so in the neuroendocrine hypothalamus (Wang et al., 2000; Faber et al., 2007). However, it has to be pointed out that while the labeling pattern of cell bodies provided by immunocytochemistry and *in situ* hybridization histochemistry was very similar in the rat (Wang et al., 2000), PTH2R-labeled cell bodies were mostly not visible by immunohistochemistry in the mouse, non-human primate and human (Faber et al., 2007; Bago et al., 2009). Nevertheless, *in situ* hybridization histochemistry and X-Gal histochemistry in mice expressing beta-galactosidase driven by the promoter of the PTH2R revealed a similar expression pattern of the PTH2R in these species as well (Faber et al., 2007; Bago et al., 2009). The finding that PTH2R-immunopositive fibers are often localized in the vicinity of PTH2R-expressing neurons suggests that these fibers may represent either axons or dendrites of local PTH2R-expressing neurons. Colocalization of the PTH2R with vesicular glutamate transporters suggests its axonal localization as described below in Section "Development of Fear." Interestingly, the distributions, and even the subregional distributions of TIP39- and PTH2Rcontaining axon terminals are remarkably similar (Dobolyi et al., 2006a; Faber et al., 2007). Thus, TIP39 axon terminals and the PTH2R are co-distributed in the very same brain structures allowing the presynaptic modulation of PTH2R axon terminals. Therefore, an axo-axonal action of TIP39 is plausible (Dobolyi et al., 2010). However, there is no evidence available at present based on which the dendritic localization of the PTH2R can be excluded. In particular, electron microscopic investigation of the PTH2R has not been reported yet.

subparafascicular area is a mediolaterally and to some degree rostrocaudally elongated region. Most TIP39 cells are located in its mediorostral portion, the periventricular gray of the thalamus **(A)** and in its caudolateral part, the posterior intralaminar complex of the thalamus **(B)**. The area where TIP39 neurons are distributed is shown in red. A few TIP39 neurons are scattered between these two regions below the fasciculus retroflexus and above the medial lemniscus. Additional abbreviations: ca, cerebral aqueduct; cc, corpus

#### **THE TIP39-PTH2R NEUROMODULATOR SYSTEM IN HYPOTHALAMIC AREAS EXPRESSING NEUROENDOCRINE HORMONES**

A high density of PTH2R-expressing cells and TIP39- and PTH2Rcontaining fibers is present in the medial preoptic nucleus and some surrounding parts of the medial preoptic area (Dobolyi et al., 2006a; Faber et al., 2007). A TIP39-PTH2R neuromodulator system is also abundant in the paraventricular and periventricular nuclei while other parts of the anterior hypothalamic region contain a lower density of labeling. Thus, TIP39 terminals are ideally positioned to influence somatostatin and CRH neurons. It has indeed been demonstrated that somatostatin neurons express the PTH2R (Wang et al., 2000; Dobolyi et al., 2006a) and that both TIP39- and PTH2R-containing terminals approximate CRH-expressing neurons in the parvicellular subdivision of the hypothalamic paraventricular nucleus (PVN; Bago et al., 2009; Dimitrov and Usdin, 2010). Similarly, in the tuberal region of the hypothalamus, the arcuate nucleus contains the highest density of PTH2R-expressing cells and TIP39- and PTH2R-containing fibers, providing an anatomical basis for influencing GH and prolactin release *via* GH-releasing hormone (GHRH) and dopamine neurons in the arcuate nucleus (**Table 1**).

#### **TIP39 AND THE PTH2R IN HYPOTHALAMIC, LIMBIC, AND SENSORY BRAIN REGIONS THAT POTENTIALLY EXERT INFLUENCE ON THE NEUROENDOCRINE SYSTEM**

In the nervous system, TIP39 fibers and the PTH2R have a widespread distribution pattern (Wang et al., 2000; Dobolyi et al., 2003b; Faber et al., 2007; Bago et al., 2009). A number of brain regions known to affect the neuroendocrine system *via* neuronal projections to hypophysiotropic neurons were shown to contain a high density of PTH2R-expressing cell bodies as well as

callosum; CG, central gray; DG, dentate gyrus; fr, fasciculus retroflexus; H, hippocampus; ic, internal capsule; MG, medial geniculate body; ml, medial lemniscus; mt, mamillothalamic tract; PH, posterior hypothalamus; Pir, piriform cortex; PVG, periventricular gray; SC, superior colliculus;SN, substantia nigra; SNl, substantia nigra, lateral subdivision; SPFl, lateral (parvocellular) subparafascicular nucleus; SPFm, magnocellular subparafascicular nucleus; VB, ventrobasal thalamus. The original drawings are modifications of panels from a rat brain atlas (Paxinos and Watson, 2007).

TIP39- and PTH2R-containing fibers. These brain regions include the medial prefrontal, especially the infralimbic cortex*,* the lateral septal nucleus, the bed nucleus of the stria terminalis, the amygdala, especially its medial and central nuclei, some midline and intralaminar thalamic nuclei, several hypothalamic nuclei, the periaqueductal gray, the lateral parabrachial nuclei, the locus coeruleus and subcoeruleus areas, and the nucleus of the solitary tract. Within the hypothalamus, the following nuclei contained a high level of TIP39 and PTH2R apart from the above mentioned neuroendocrine regions: the MnPO, the vascular organ of the lamina terminalis, the dorsomedial and perifornical hypothalamic nuclei, and some parts of the lateral hypothalamic area including the so-called far-lateral hypothalamus immediately next to the internal capsule, the medial subdivision of the supramamillary nucleus, the ventral, and dorsal premamillary nuclei, and the posterior hypothalamic nucleus, (Wang et al., 2000; Dobolyi et al., 2003b; Faber et al., 2007; Bago et al., 2009). In contrast, TIP39 and the PTH2R are scarce in a number of hypothalamic nuclei including the lateral preoptic area, the supraoptic, suprachiasmatic, and lateroanterior hypothalamic nuclei, the medial magnocellular part of the paraventricular nucleus, the ventromedial nucleus, and the medial and lateral nuclei of the mamillary body (**Table 1**).

# **TIP39 AND THE PTH2R IN REGIONS OF THE NERVOUS SYSTEM NOT INVOLVED IN NEUROENDOCRINE REGULATIONS**

Elements of the TIP39-PTH2R neuromodulator system are also expressed in some brain regions that are not known to be directly or indirectly involved in neuroendocrine regulation. Thus, TIP39 as well as PTH2R fibers were abundant in the fundus striati, the medial geniculate body, tegmental areas of the midbrain and pons, the deep layers of the superior colliculus, the external cortex of the

#### **Table 1 | The distribution of TIP39 axon terminals and the PTH2R in the nervous system.**


inferior colliculus, the periolivary area, the nucleus of the trapezoid body, the superficial layers of the dorsal horn and the lateral cervical nucleus of the spinal cord, and some cells in the dorsal root ganglia (Wang et al., 2000; Dobolyi et al., 2003b; Faber et al., 2007; Bago et al., 2009; Matsumoto et al., 2010). Several of these brain regions may participate in sensory, especially auditory information processing, which could represent non-neuroendocrine-related functions of the TIP39-PTH2R neuromodulator system. However, a potential role of TIP39 in the auditory stress pathway will be discussed later.

#### **NEUROENDOCRINE FUNCTIONS THAT MAY INVOLVE THE TIP39-PTH2R SYSTEM**

Based on the distribution of the TIP39-PTH2R system in the brain, its involvement in endocrine, limbic, nociceptive, and auditory functions have been hypothesized (Dobolyi et al., 2003a). These functions can of course be interrelated with each other. Initial studies in which PTH2Rs were activated with exogenous TIP39 as well as*in vitro* approaches provided some insights into the possible functions of the TIP39-PTH2R system. More recently, a selective and potent peptide antagonist of the PTH2R (Kuo and Usdin, 2007) and transgenic mice lacking functional TIP39 and PTH2R genes were developed (Fegley et al., 2008), which accelerated functional studies. It has indeed been established that the peptide neuromodulator system is involved in a variety of neuroendocrine functions including the stress response, thermoregulation, and prolactin release. Some evidence is also available for a role of the TIP39-PTH2R system in the regulation of AVP and GH release.

#### **STRESS RESPONSE**

The stress response is a complex reaction of the organism to stimuli that threaten homeostasis. The stress response includes an altered psychological state called anxiety and physiological effects that include catecholamine and glucocorticoid release as final common pathways (Kvetnansky et al., 2009). The anxiety level and the hormone secretions interact to provide the organisms stress response.

#### **The regulation of CRH function**

The first functional evidence that the PTH2R might be involved in the regulation of CRH release came from an *in vitro* study. TIP39 increased CRH secretion from medial basal hypothalamic explants (Ward et al., 2001). In addition, intracerebroventricular injection of TIP39 dose-dependently increased the plasma adrenocorticotropin (ACTH) level at 10 min after injection in rat (Ward et al., 2001). In another study, local injection of TIP39 above the paraventricular hypothalamic nucleus (PVN) in mice also elevated plasma corticosterone levels in addition to increasing the number of pCREB-containing activated cells in and around the PVN (Dimitrov and Usdin, 2010). Most importantly, these effects of local TIP39 were not present in PTH2R knockout (KO) animals (Dimitrov and Usdin, 2010) excluding non-specific actions of TIP39 injection. These results suggest that the TIP39-PTH2R system is positioned and available to potentially modulate activation of the hypothalamus-pituitary-adrenal (HPA) axis. In fact, the activation of TIP39 neurons has been reported in some stress situations as described below in Section "The Activation of TIP39 Neurons." Furthermore, the daily peak of the basal plasma corticosterone level was reduced in TIP39 KO mice suggesting that endogenous TIP39 plays a role in the circadian regulation of corticosterone levels (Dimitrov and Usdin, 2010). These findings are consistent with a high density of TIP39 fibers, PTH2R-expressing cells, and PTH2R-containing fiber terminals in the parvicellular subdivisions of the PVN (Faber et al., 2007). In fact, close apposition between CRH neurons and PTH2R-containing fibers has been demonstrated in mice (Dimitrov and Usdin, 2010) as well as in human (Bago et al., 2009).

#### **The regulation of noradrenergic function**

The involvement of the PTH2R in the regulation of the catecholamine systems is less well established even though TIP39 fibers are abundant in the locus coeruleus and subcoeruleus areas (Dobolyi et al., 2003b) where rostrally projecting noradrenergic neurons reside. Recent evidence brings up the possibility that TIP39 may interact with some noradrenergic pathways. TIP39 KO mice and WT mice injected with a PTH2R antagonist demonstrated selective impairment of memory performance during novelty-induced arousal (Coutellier et al., 2011a). Noradrenergic signaling has a biphasic, inverted U shaped, effect on cognitive functions (Arnsten, 2009). The impaired performance of mice without TIP39/PTH2R signaling was restored by propranolol, an antagonist of beta adrenoceptors, suggesting that PTH2R signaling influences the effect of novelty stress *via* an interaction with noradrenergic mechanisms (Coutellier et al., 2011a).

#### **Stress-induced analgesia**

In an experiment addressing the role of PTH2R signaling in stressinduced analgesia (SIA), hotplate tests were performed before and after an inescapable foot shock, used as the stressful stimulus (Dimitrov et al., 2010). As discussed in a later section, TIP39 KO and PTH2R KO mice as well as WT mice injected with a PTH2R antagonist had somewhat elevated pre-shock response latencies in the hotplate test. Following the foot shock, the response latencies increased in the WT mice as expected based on the phenomenon of SIA. Unexpectedly, a larger and dramatic increase was found in the response latency in mice without PTH2R signaling (Dimitrov et al., 2010). These findings suggest that signaling through the TIP39-PTH2R system may normally limit SIA.

Stress-induced analgesia induced by high intensity stressful stimuli, including inescapable foot shock, has been shown to have a predominant non-opioid component (Lewis et al., 1980; Terman et al., 1984). Indeed the opioid antagonist naloxone did not, but the CB1 cannabinoid receptor antagonist rimonabant did, decrease the SIA in WT mice (Dimitrov et al., 2010). The inhibitory effect of rimonabant on the SIA was much greater in the KO mice suggesting that the KO mice have greater endocannabinoid release or sensitivity and that endogenous TIP39, *via* the PTH2R, may negatively modulate release or an effect of endogenous cannabinoids during SIA.

#### **RELEASE OF ARGININE-VASOPRESSIN**

The role of the PTH2R on the release of AVP was investigated by intracerebroventricular injection of TIP39 in rats (Sugimura et al., 2003). Reduced AVP levels were found in plasma 5 min following TIP39 administration. TIP39 also suppressed the plasma AVP increase following dehydration by water deprivation for 48h, hyperosmolality following i.p. injection of hypertonic saline, and hypovolemia following i.p. injection of polyethylene glycol. These inhibitory effects cannot be attributed to a decrease in the level of osmotic or hypovolemic stimulation because plasma Na<sup>+</sup> and plasma total protein were not affected by the injection of TIP39 (Sugimura et al., 2003). The effect of TIP39 was also not a consequence of a change in blood pressure because injection of TIP39 produced afall in mean arterial blood pressure,which would rather stimulate AVP secretion. In turn, the opioid receptor antagonist naloxone significantly reversed the inhibitory effect of TIP39 on dehydration-induced AVP release while it had no significant effect on the plasma AVP level when injected alone (Sugimura et al., 2003). These results suggest that TIP39 inhibits AVP release by central action – possibly *via* an opioid system but without hemodynamic or osmotic influence. The effect was rapid and did not last long suggesting that TIP39 may play a role in the dynamic regulation of AVP release. TIP39 and PTH2R are scarce in the supraoptic nucleus and the part of the PVN where magnocellular AVP neurons are located. Thus, TIP39 may exert its AVP release-inhibitory effect by acting through other hypothalamic nuclei. The hypothalamic arcuate nucleus is a candidate as it contains a high density of TIP39 and PTH2R immunoreactivity (Faber et al., 2007) as well as

many opioid neurons, which are involved in the regulation of AVP release (Haaf et al., 1987; Heijning et al., 1991). Indirect action of TIP39 *via* hypothalamic angiotensin and atrial natriuretic peptide neurons affecting AVP release (Antunes-Rodrigues et al., 2004) is also conceivable.

#### **GROWTH HORMONE SECRETION**

In a preliminary experiment, TIP39 injection into the lateral ventricle of male rats blocked the appearance of GH almost completely in plasma for the next 3 h (Usdin et al., 2003). This finding is consistent with anatomical data showing a high density of TIP39-containing fibers around somatostatin neurons in the periventricular hypothalamic nucleus. Somatostatin neurons in this area project to the median eminence and inhibit the release of GH (Luque et al., 2008). PTH2R expression was demonstrated on many of these somatostatin neurons in the rat (Usdin et al., 1999b) as well as in human (Bago et al., 2009) providing the anatomical basis for TIP39 stimulating the release of somatostatin, which in turn inhibits GH secretion (Luque et al., 2008).

## **PHYSIOLOGICAL ACTIONS OF TIP39 NOT DIRECTLY RELATED TO ITS NEUROENDOCRINE EFFECTS EFFECTS ON THE ANXIETY LEVEL**

The anxiety levels of animals may also be affected by the TIP39- PTH2R system. In the elevated plus maze, a test of anxiety, intracerebroventricular injections of TIP39 in rats resulted in increased open arm entries and duration as compared to controls (animals receiving the inactive TIP (7–39) or saline) while no differences were observed between groups in the number of closed arm entries or total arm entries, suggesting an anxiolytic-like effect of TIP39 (LaBuda et al., 2004). The TIP39-PTH2R system in the infralimbic cortex, lateral hypothalamus, preoptic area, lateral septum, and the paraventricular thalamic nucleus could be involved in this action because TIP39 administration induced Fos activation in these brain regions (LaBuda et al., 2004). In the forced-swim test, a similar TIP39 administration reduced the duration of immobility, and increased the amount of climbing behavior (LaBuda et al., 2004) suggesting an anti-depressant-like action of TIP39. The phenotype of TIP39 KO mice is also consistent with an anxiolytic role of endogenous TIP39. TIP39 KO mice demonstrated increased anxiety in the shock-probe defensive burying test as compared to WT controls. In "standard/low stress" testing conditions, TIP39 KO mice did not differ from WT controls in the arm entries in the elevated plus maze or in the dark-light emergence test of spontaneous anxiety-like behaviors (Fegley et al., 2008). However, an increase in anxiety-like behavior became apparent in TIP39 KO mice that were tested in the elevated plus maze under conditions of mildly increased stress evoked by either brief prior restraint or bright illumination. These results are consistent with a role of endogenous TIP39 in limiting the consequences of stressful perturbations. Furthermore, mice lacking TIP39 or the PTH2R demonstrated increased anxiety- and depression-like behaviors 16–17 days but not 7–9 days after a foot shock in elevated-zero maze, open field, light-dark box and forced-swim tests (Coutellier and Usdin, 2011).

# **DEVELOPMENT OF FEAR**

Closely related to anxiety,fear has also been investigated in relation with the TIP39-PTH2R system using Pavlovian fear conditioning (Fegley et al., 2008). TIP39 KO mice showed more freezing than WT controls after only one tone-shock pairing during conditioning and, subsequently, more freezing during both tone- and context-recall tests (Fegley et al., 2008). However, based on a similar rate of decline in freezing responses to repeated tone presentation and a similar level of freezing during subsequent tone presentation, there did not appear to be an effect of TIP39 deletion on fear extinction learning or extinction recall, respectively (Fegley et al., 2008). Furthermore, foot shock conditioned fear recall was enhanced 14 days but not 6 days after the aversive stimulus in both TIP39 KO and PTH2R KO mice as compared to WT controls (Coutellier and Usdin, 2011). These results suggest that normal TIP39 signaling lessens the long-term consequences of a traumatic event while the absence of signaling *via* the PTH2R delays recovery. Since the amygdala is known to be involved in the fear response (LeDoux, 2003), the abundant TIP39-PTH2R system in the amygdala, especially in its central and medial nuclei (Faber et al., 2007), might be involved in these effects.

#### **THERMOREGULATION**

Body temperature regulation is a fundamental homeostatic function that in homeothermic animals. Thermal information on environmental temperature sensed by skin thermoreceptors ascends to the MnPO in the hypothalamic preoptic area. GABAergic inhibitory neurons in the MnPO integrate this input with local thermal influences and project to the dorsomedial hypothalamic nucleus and the rostral medullary raphe region to restore homeostasis (Nakamura, 2011). Injection of TIP39 into the lateral ventricle increased the core temperature of WT mice while TIP39 injection had no effect in PTH2R KO mice, excluding the possibility of non-specific inflammatory actions (Dimitrov et al., 2011). Furthermore, PTH2R KO mice had impaired heat production upon cold exposure, but no change in basal temperature and no impairment in response to a hot environment suggesting that the TIP39-PTH2R system plays a specific role in temperature conservation in a cold environment (Dimitrov et al., 2011). Since temperature sensation was normal in PTH2R KO mice, the PTH2R may play a role in the heat production signal or heat production ability. Both seem to be the case because acute intracerebral PTH2R antagonist administration also impaired the heat production response to a cold environment, though to a smaller extent. In addition, the weight of brown adipose tissue (BAT), and its capacity to increase body temperature were reduced. PTH2Rs in the MnPO seem to be involved in the thermoregulatory action of TIP39 because TIP39 injected locally into the MnPO produced a larger body temperature increase (2˚C) for longer periods of time than injection of the same amount of TIP39 into the lateral ventricle. Furthermore, local injection of TIP39 into the dorsomedial hypothalamic nucleus had no effect on the body temperature. The MnPO as a site of action is consistent with its high density of TIP39 terminals and PTH2R immunoreactivity (Faber et al., 2007; Dimitrov et al., 2011) as well as with the known role of MnPO neurons in the control body temperature (Baffi and Palkovits, 2000; Bratincsak and Palkovits, 2004) *via* descending systems regulating

BAT thermogenesis and cutaneous vascular tone (Morrison and Nakamura, 2011). The action of TIP39 on the HPA axis and in central thermoregulation allows associations with fever. Although no data are available in this regard, this potential role of the TIP39-PTH2R system will be an interesting line of research in the future.

# **NOCICEPTIVE FUNCTIONS**

The TIP39-PTH2R system may play a role at several levels of nociceptive processing. Its role in nociceptive sensation and spinal cord processing has been revealed by intrathecal administration of TIP39 and antagonizing its actions with the injection of an anti-TIP39 antibody (Dobolyi et al., 2002). Intrathecal injection of TIP39 stimulated a dose-dependent nocifensive response, caudally directed scratching, biting, and licking and decreased the tail-flick and paw-pressure withdrawal latencies. Intrathecal injection of the TIP39 antibody increased the response latency in the thermal tail-flick assay and in the paw-pressure test, corresponding to decreased sensitivity (Dobolyi et al., 2002). These actions are likely related to the facilitation of nociceptive transmission from the DRG to neurons in the spinal cord dorsal horn as intense PTH2R immunoreactivity is present in superficial layers of the spinal cord dorsal horn where most nociceptive afferents terminate (Wang et al., 2000).

The TIP39-PTH2R system may also be involved in the supraspinal regulation of pain processing. Intracerebroventricular injection of TIP39 reduced the latencies in tail-flick and hotplate tests while injection of a PTH2R antagonist had the opposite, anti-nociceptive effect in these tests as well as in the formalin test (Dimitrov et al., 2010). Furthermore, TIP39 and PTH2R KO mice also demonstrated reduced nociceptive responses in these tests, arguing for a pro-nociceptive function of endogenous TIP39 *via* the PTH2R (Dimitrov et al., 2010). These findings are consistent with the distribution of TIP39 and the PTH2R in a variety of brain regions known to be involved in the processing of nociceptive information, including the nucleus of the solitary tract, the parabrachial nuclei, the periventricular gray, the midline thalamic nuclei, the PVN, and the insular and infralimbic cortices. These areas are thought to be components of autonomic-limbic pain-related pathways including the ascending reticular activating system (Benarroch, 2006).

# **REGULATION OF POSTPARTUM EVENTS IN MOTHERS**

TIP39 expression is decreased in neurons in both the subparafascicular area and the MPL during the period of pubertal development in rat (Dobolyi et al., 2006b). Since the level of the PTH2R did not decrease, it was hypothesized that TIP39 is induced under some circumstances to act on the already available PTH2Rs. Indeed, a dramatic increase in the TIP39 mRNA levels was demonstrated in the postpartum period using real-time PCR. Upon removal of the pups, the level of TIP39 mRNA decreased to its basal level. *In situ* hybridization histochemistry confirmed the induction of TIP39 and revealed that within the subparafascicular area, TIP39 neurons in the PIL but not in the PVG demonstrated an increase in TIP39 expression (Cservenak et al., 2010). The elevated mRNA of TIP39 in the PIL and MPL is translated into

increased peptide levels, as demonstrated by immunohistochemistry (Varga et al., 2008; Cservenak et al., 2010). The functional significance of the elevated TIP39 was tested on the release of prolactin because pathway transections revealed projection of TIP39 neurons in the PIL toward tyrosine hydroxylase-containing neurons in the mediobasal hypothalamic regions known to regulate prolactin secretion. Retrograde labeling in nulliparous female rats also demonstrated a projection of subparafascicular TIP39 neurons to the arcuate nucleus (Szabo et al., 2010). In rodents, removal of the pups from the dams for 4h results in a decrease in prolactin level, which is in turn dramatically increased upon the return of the litter and the immediate onset of nursing. Injection of a PTH2R antagonist into the lateral ventricle 5 min before uniting the mothers with pups potently and dose-dependently inhibited suckling induced prolactin release in the rat (Cservenak et al., 2010). The physiological significance of this is supported by the observation that in a similar pup removal/return paradigm the weight increase (a measure of milk consumed) of pups suckling PTH2R KO mice was reduced 30 min after the onset of nursing as compared to pups suckling WT mice (Coutellier et al., 2011b). Also consistent with less effective suckling by PTH2R KO dams, pups reared by PTH2R KO mice had a lower body weight at the time of weaning than pups reared by WT mice (Coutellier et al., 2011b). To eliminate the effect of the genotype of the pups from the analysis, WT females were mated with PTH2R KO males while PTH2R KO females were mated with WT males so that all pups were heterozygous in these experiments (Coutellier et al., 2011b).

Additional influences of subparafascicular TIP39 neurons on reproductive neuroendocrine function cannot be excluded. TIP39 fibers and PTH2Rs are ideally positioned to affect gonadotropinreleasing hormone (GnRH) neurons, whose activity is suppressed during lactation. In addition, the TIP39-PTH2R neuromodulator system might also play a role in conveying the effect of suckling on other systems adapted in the postpartum period. PTH2Rs in the preoptic area, the lateral septum, and the periaqueductal gray could be involved in the control of maternal behaviors (Dobolyi, 2011). Emotional changes that take place in the postpartum period could also be affected by TIP39 based on the localization of the TIP39-PTH2R system in the infralimbic cortex, the medial, and central amygdaloid nuclei, the amygdalo-hippocampal transitional zone, the premamillary nuclei, the ventral subiculum, and the periaqueductal gray, which are parts of the circuitry of reproductive and emotional regulation (Lonstein and Stern, 1997; Lin et al., 1998; Li et al., 1999; Simerly, 2002; Numan and Insel, 2003; Hasen and Gammie, 2005).

# **POSSIBLE MODELS FOR ACTIONS VIA THE TIP39-PTH2R SYSTEM**

#### **THE ACTIVATION OF TIP39 NEURONS**

There is limited information available showing that TIP39 neurons are activated in stress situations. At present, only the effects of an h-long high intensity noise stress have been reported. Medial paralemniscal and posterior intralaminar TIP39 neurons demonstrate Fos induction together with CRH neurons in response to noise stress, suggesting that TIP39 neurons could be involved in the transmission of acoustic stress derived information to CRH neurons in the PVN (Palkovits et al., 2009). In agreement with these findings, medial paralemniscal and possibly also posterior intralaminar TIP39 neurons have afferent neuronal connections with the primary auditory cortex and the external cortex of the inferior colliculus (Varga et al., 2008) providing the anatomical basis for an auditory influence on the CRH neurons in the PVN.

Cold exposure but not warm ambient temperature induced *cfos* in some PVG neurons (Kiyohara et al., 1995; Miyata et al., 1995; Baffi and Palkovits, 2000; Bratincsak and Palkovits, 2004) providing the possibility that TIP39 neurons in this region could be activated by cold exposure leading to both stress responses and temperature regulation. It has also been reported that the *cfos* expression in the PVG significantly outlasts the cold exposure (Miyata et al., 1995), suggesting that it may have a role in the maintenance of homeostasis during adaptation to cold stress (Baffi and Palkovits, 2000).

The PVG is a site of stimulation-induced analgesia (Rhodes and Liebeskind, 1978; Peschanski and Mantyh, 1983). Potent analgesia is obtained in rats following electrical stimulation in the gray matter surrounding the caudal portion of the third ventricle and the midline area of the caudal thalamus that is comparable to that produced by stimulation of the caudal periaqueductal gray. Analgesia outlasts the period of brain stimulation, and is not due to a generalized motor debilitation of the animal (Rhodes and Liebeskind, 1978). In addition, some neurons in the area are activated by noxious stimuli (Dong et al., 1978; Sugiyama et al., 1992) providing the possibility that TIP39 neurons in this region could be activated by noxious stimuli leading to both stress and nociceptive responses.

The TIP39-PTH2R neuromodulator system is present in several brain areas that are activated in males following mating (Sachs and Meisel, 1988; Coolen et al., 1997; Veening and Coolen, 1998) including the PIL, the medial preoptic nucleus, the posteromedial part of the medial subdivision of the bed nucleus of the stria terminalis, and the posterodorsal subdivision of the medial amygdaloid nuclei. In addition, the PIL, which contains TIP39 neurons, has also been implicated in sexual functions *via* the facilitation of copulatory behavior by current injection into the area (Shimura and Shimokochi, 1991), by the decreased sexual behavior following lesion in the area (Maillard and Edwards, 1991), and by the demonstration of the activation of neurons in the area following sexual behavior using the Fos technique in rats (Coolen et al., 1997) and imaging techniques in human (Holstege et al., 2003). Indeed, TIP39 neurons in the PIL have been shown to exhibit *Fos* expression following ejaculation (Wang et al., 2006a). This suggests that these TIP39 neurons are part of the afferent circuits that process genital-somatosensory information related to ejaculation contributing to mating and mating-induced changes in reproductive behaviors.

Another reproductive influence reaching TIP39 neurons in the PIL was reported in mother rats. Dams deprived of their pups for a day demonstrated *c-fos* expression in the PIL but not in the PVG in response to reunion with their litter (Cservenak et al., 2010). Almost all TIP39 neurons in the PIL were involved in this response. Since the pups start to suckle very soon after they are returned to their mothers, it was hypothesized that suckling results

in the activation of TIP39 neurons in the PIL (Cservenak et al., 2010).

#### **THE POTENTIAL MECHANISMS OF ACTION OF TIP39**

There are no indications at present that TIP39 would have any action apart from activating the PTH2R. The PTH2R elevates cAMP and in some cells also increases Ca2<sup>+</sup> levels (Goold et al., 2001; Della Penna et al., 2003) suggesting an excitatory influence of TIP39 on the target cells. In some cases, the PTH2R may be located in cell bodies and dendrites, there is evidence for this for the somatostatin neurons in the rat periventricular nucleus. Most typically, the PTH2R may be located in axon terminals. Apart from the somatostatin terminals in the median eminence (Dobolyi et al., 2006a), a presynaptic location of the PTH2R at excitatory synapses has been demonstrated in several hypothalamic areas (**Figure 3**). In the PVN, vesicular glutamate transporter 2 (VGluT2)-containing terminals closely apposed to CRH neurons were shown to also contain the PTH2R (Dimitrov and Usdin, 2010). In addition, the close apposition by TIP39-containing terminals suggested that TIP39 from the subparafascicular area might also contribute to the activation of CRH neurons by increasing the efficacy of the excitatory input to the CRH neurons (Dimitrov and

**FIGURE 3 | Colocalization of PTH2R fibers and VGluT2 punctae in the anterior hypothalamus as detected by immunohistochemistry. (A)** a low magnification image of PTH2R (green) and VGluT2 (red) expression in the region of the medial preoptic area. **(B)** a high magnification image of PTH2R/VGluT2 colocalization, where some of the colocalized points are indicated with arrowheads. Abbreviations: 3V, third ventricle; ac, anterior commissural; AHA, anterior hypothalamic area; MPO, medial preoptic nucleus of the hypothalamus. Scale bars = 200 µm for A and 10 µm for B.

#### **FIGURE 4 | Colocalization of VGluT2 mRNA and PTH2R mRNA in the anterior hypothalamus as detected by fluorescent in situ hybridization. (A)** VGluT2 mRNA (red puncta) over cell nuclei (blue, DAPI).

**(B)** PTH2R mRNA (green puncta) over cell nuclei (blue, DAPI). **(C)** a merged image, where the arrows point to cells expressing both PTH2R/VGluT2 mRNAs and the arrowhead points to a neuron with only VGluT2 signal. Scale bar = 10 µm.

Usdin, 2010). In a similar fashion, median preoptic neurons projecting to the dorsomedial hypothalamic nucleus for the potential transfer of thermoregulatory information were also suggested to be modulated by TIP39 *via* presynaptic excitation of their afferents (Dimitrov et al., 2011). Expression of the PTH2R by glutamatergic neurons in several brain regions is supported by data from *in situ* hybridization histochemistry (**Figure 4**). In other cases, the effect of TIP39 on inhibitory neurons has also been postulated based on functional data (Sugimura et al., 2003; Cservenak et al., 2010). Thus, the inhibition of AVP neurons and dopaminergic neurons is expected from the actions of TIP39 on the serum levels of AVP and prolactin, respectively. This inhibition could be carried out *via* potentiation of excitatory inputs to inhibitory interneurons acting on the AVP and dopaminergic neurons, respectively.

**FIGURE 5 | Potential mechanisms of actions of TIP39 via the PTH2R.** TIP39 neurons in the PVG receive information on threats to homeostasis while those in the PIL receive various inputs related to reproduction. Axon terminals arising from these TIP39 neurons may facilitate excitatory synapses by presynaptic actions. A well-documented example is the action of TIP39 on glutamatergic synapses terminating on CRH neurons in the PVN. Another possible route to influence hypothalamic hormone release is via terminals of hypophysiotropic neurons in the median eminence (ME) as has been suggested for somatostatin. Neurons that express PTH2Rs in their cell bodies may be directly affected by TIP39. Some actions of TIP39 including those on prolactin and AVP release, seem to be mediated by inhibitory neurons. TIP39 terminals might innervate cell bodies of GABAergic cells or alternatively, presynaptic terminals leading to the activation of inhibitory neurons (not shown). Additional abbreviations: ACTH, adrenocorticotropin; Arc, arcuate nucleus; DA, dopamine; GH, growth hormone; Glu, glutamate; Pe, periventricular hypothalamic nucleus; PRL, prolactin; SOM, somatostatin.

# **CONCLUSION**

The TIP39-PTH2R neuromodulator system may play an important role in the regulation of several different aspects of neuroendocrinefunctions. TIP39 neurons in the PVG may be stimulated by stress inputs compromising the homeostasis of the animal while TIP39 neurons in the PIL may be stimulated by various reproductive events (**Figure 5**). Additional, as yet unexplored inputs to the TIP39 neurons are also plausible. These TIP39 neurons in the posterior thalamus relay their input toward neurons located in neuroendocrine and limbic brain areas to exert direct and indirect effects on neuroendocrine systems. So far, substantial evidence is available for the involvement of the TIP39-PTH2R system in the

# **REFERENCES**


regulation of corticosterone, AVP, and prolactin. These actions of TIP39 suggest that the PTH2R may be a future target to develop drugs for the treatment of some neuroendocrine abnormalities and disorders.

# **ACKNOWLEDGMENTS**

Support was provided by the Bolyai János Fellowship Grant of the Hungarian Academy of Sciences, the Hungarian Science Foundation OTKA K100319 and NFM-OTKA NNF85612 Research Grants for AD, and the Intramural Research Program of the NIH, National Institute of Mental Health for TBU.


al. (2008). Increased fear- and stressrelated anxiety-like behavior in mice lacking tuberoinfundibular peptide of 39 residues. *Genes Brain Behav.* 7, 933–942.


functional marker of nociceptive myelinated fibers responsible for neuropathic pain.*J. Neurochem.* 112, 521–530.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 11 July 2012; paper pending published: 06 August 2012; accepted: 20 September 2012; published online: 08 October 2012.*

*Citation: Dobolyi A, Dimitrov E, Palkovits M and Usdin TB (2012) The neuroendocrine functions of the parathyroid hormone 2 receptor. Front. Endocrin. 3:121. doi: 10.3389/fendo.2012.00121*

*This article was submitted to Frontiers in Neuroendocrine Science, a specialty of Frontiers in Endocrinology.*

*Copyright © 2012 Dobolyi, Dimitrov, Palkovits and Usdin. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*