# INTERNATIONAL PLANT PROTEOMICS ORGANIZATION (INPPO) WORLD CONGRESS 2014

EDITED BY: Joshua L. Heazlewood, Jesús V. Jorrín-Novo, Ganesh Kumar Agrawal, Silvia Mazzuca and Sabine Lüthje PUBLISHED IN: Frontiers in Plant Science

### *Frontiers Copyright Statement*

*© Copyright 2007-2016 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 978-2-88945-060-2 DOI 10.3389/978-2-88945-060-2

# 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

# **INTERNATIONAL PLANT PROTEOMICS ORGANIZATION (INPPO) WORLD CONGRESS 2014**

Topic Editors:

**Joshua L. Heazlewood,** The University of Melbourne, Australia **Jesús V. Jorrín-Novo,** University of Cordoba, Spain **Ganesh Kumar Agrawal,** Research Laboratory for Biotechnology and Biochemistry, Nepal **Silvia Mazzuca,** University of Calabria, Italy **Sabine Lüthje,** University of Hamburg, Germany

View across the river Elbe from the Stage Theater im Hafen of the Elbphilharmonie (concert hall) in Hamburg, Germany. Photo by Sabine Luthje

The field of proteomics has advanced considerably over the past two decades. The ability to delve deeper into an organism's proteome, identify an array of post-translational modifications and profile differentially abundant proteins has greatly expanded the utilization of proteomics. Improvements to instrumentation in conjunction with the development of these reproducible workflows have driven the adoption and application of this technology by a wider research community. However, the full potential of proteomics is far from being fully exploited in plant biology and its translational application needs to be further developed.

In 2011, a group of plant proteomic researchers established the International Plant Proteomics Organization (INPPO) to advance the utilization of this technology in plants as well as to create a way for plant proteomics researchers to interact, collaborate and exchange ideas. The INPPO conducted its inaugural world congress in mid 2014 at the University of Hamburg (Germany). Plant proteomic researchers from around the world were in attendance and the event marked the maturation of this research community. The Research Topic captures the opinions, ideas and research discussed at the congress and encapsulates the approaches that were being applied in plant proteomics.

# Table of Contents

*08 Editorial: International Plant Proteomics Organization (INPPO) World Congress 2014*

Joshua L. Heazlewood, Jesús V. Jorrín-Novo, Ganesh K. Agrawal, Silvia Mazzuca and Sabine Lüthje

# **Chapter 1: Technical Advances in Plant Proteomics**

*14 Scientific standards and MIAPEs in plant proteomics research and publications* Jesús V. Jorrín Novo

# **i. Two-Dimensional Gel Electrophoresis-Based Approaches**


Claudia-Nicole Meisrimler, Alexandra Schwendke and Sabine Lüthje


M. Cristina Romero-Rodríguez, Nieves Abril, Rosa Sánchez-Lucas and Jesús V. Jorrín-Novo

# **ii. Mass Spectrometry-based Approaches**


Harriet T. Parsons and Joshua L. Heazlewood

*65 Characterization of protein* **N***-glycosylation by tandem mass spectrometry using complementary fragmentation techniques*

Kristina L. Ford, Wei Zeng, Joshua L. Heazlewood and Antony Bacic

# **iii. Sample-based Approaches**


Elisabeth Stes, Kris Gevaert and Ive De Smet


Fangping Gong, Xiaolin Wu and Wei Wang

*121 The role of proteomics in progressing insights into plant secondary metabolism* María J. Martínez-Esteso, Ascensión Martínez-Márquez, Susana Sellés-Marchart, Jaime A. Morante-Carriel and Roque Bru-Martínez

# **iv. Data-based Approaches**


Valérie Poncet, Charlie Scutt, Rémi Tournebize, Matthieu Villegente, Gwendal Cueff, Loïc Rajjou, Thierry Balliau, Michel Zivy, Bruno Fogliani, Claudette Job, Alexandre de Kochko, Valérie Sarramegna-Burtet and Dominique Job

# **Chapter 2: Plant Abiotic Stress and Proteomics**

*144 Proteomic analysis of crop plants under abiotic stress conditions: where to focus our research?*

Fangping Gong, Xiuli Hu and Wei Wang

*149 Advances in plant proteomics toward improvement of crop productivity and stress resistance*

Junjie Hu, Christof Rampitsch and Natalia V. Bykova

*164 Proteomics of stress responses in wheat and barley—search for potential protein markers of stress tolerance*

Klára Kosová, Pavel Vítámvás and Ilja T. Prášil

*178 Gamma-glutamyl cycle in plants: a bridge connecting the environment to the plant cell?*

Antonio Masi, Anna R. Trentin, Ganesh K. Agrawal and Randeep Rakwal

*182 2-DE proteomics analysis of drought treated seedlings of* **Quercus ilex** *supports a root active strategy for metabolic adaptation in response to water shortage* Lyudmila P. Simova-Stoilova, Maria C. Romero-Rodríguez, Rosa Sánchez-Lucas,

Rafael M. Navarro-Cerrillo, J. Alberto Medina-Aunon and Jesús V. Jorrín-Novo

*198 Phosphoproteomic analysis of the response of maize leaves to drought, heat and their combination stress*

Xiuli Hu, Liuji Wu, Feiyun Zhao, Dayong Zhang, Nana Li, Guohui Zhu, Chaohao Li and Wei Wang


and Silvia Mazzuca

*275 Quantitative proteomics reveals role of sugar in decreasing photosynthetic activity due to Fe deficiency*

Sajad M. Zargar, Ganesh K. Agrawal, Randeep Rakwal and Yoichiro Fukao

*279 Protein profile of* **Beta vulgaris** *leaf apoplastic fluid and changes induced by Fe deficiency and Fe resupply*

Laura Ceballos-Laita, Elain Gutierrez-Carbonell, Giuseppe Lattanzio, Saul Vázquez, Bruno Contreras-Moreira, Anunciación Abadía, Javier Abadía and Ana-Flor López-Millán


# **Chapter 3: Plant Biotic Stress and Proteomics**


Leonor Guerra-Guimarães, Rita Tenente, Carla Pinheiro, Inês Chaves, Maria do Céu Silva, Fernando M. H. Cardoso, Sébastien Planchon, Danielle R. Barros, Jenny Renaut and Cândido P. Ricardo

# **Chapter 4: Proteomics to Profile Plant Cultivars**


Alfredo S. Negri, Bhakti Prinsi, Osvaldo Failla, Attilio Scienza and Luca Espen

*383 Quantitative analysis of proteome extracted from barley crowns grown under different drought conditions*

Pavel Vítámvás, Milan O. Urban, Zbynek Škodácˇek, Klára Kosová, Iva Pitelková, Jan Vítámvás, Jenny Renaut and Ilja T. Prášil

# **Chapter 5: Proteomics and Post-Transcriptional Regulation**

*401 The conundrum of discordant protein and mRNA expression. Are plants special?*

Isabel Cristina Vélez-Bermúdez and Wolfgang Schmidt

*405 Regulation of mRNA translation controls seed germination and is critical for seedling vigor*

Marc Galland and Loïc Rajjou

# Editorial: International Plant Proteomics Organization (INPPO) World Congress 2014

Joshua L. Heazlewood1, 2 \*, Jesús V. Jorrín-Novo<sup>3</sup> , Ganesh K. Agrawal 4, 5, Silvia Mazzuca<sup>6</sup> and Sabine Lüthje<sup>7</sup>

<sup>1</sup> Lawrence Berkeley National Laboratory, Physical Biosciences Division, Joint BioEnergy Institute, Berkeley, CA, USA, <sup>2</sup> Australian Research Council Centre of Excellence in Plant Cell Walls, School of BioSciences, The University of Melbourne, Melbourne, VIC, Australia, <sup>3</sup> Agricultural and Plant Biochemistry and Proteomics Research Group, Department of Biochemistry and Molecular Biology, University of Cordoba, Cordoba, Spain, <sup>4</sup> Research Laboratory for Biotechnology and Biochemistry, Kathmandu, Nepal, <sup>5</sup> Global Research Arch for Developing Education Academy Private Limited, Birgunj, Nepal, <sup>6</sup> Laboratorio di Biologia e Proteomica Vegetale, Dipartimento di Chimica e Tecnologie Chimiche, Università della Calabria, Rende, Italy, <sup>7</sup> Oxidative Stress and Plant Proteomics Group, Biocenter Klein Flottbek and Botanical Garden, University of Hamburg, Hamburg, Germany

Keywords: plant proteomics, mass spectrometry, 2-DE, world congress

### **The Editorial on the Research Topic**

### **International Plant Proteomics Organization (INPPO) World Congress 2014**

The discipline of proteomics has undergone considerably advances over the past two decades. Our ability to delve deeper into complex proteomes, identify post-translational modifications, and profile protein abundance has greatly expanded the utilization of mass spectrometry in biology. The plant research community has enthusiastically embraced proteomic approaches and has applied these technologies to explore a multitude of research questions in the field of plant biology (Jorrín-Novo et al., 2015). In 2011, a group of plant proteomic researchers established the International Plant Proteomics Organization (INPPO) to advance the application of this technology in plants and agriculture (Agrawal et al., 2011). The INPPO conducted its inaugural world congress in the autumn of 2014 at the University of Hamburg (Germany) (Lüthje et al., 2015). The meeting brought together leading international experts in plant proteomics and provided a critical mass for the discussion of proteomic technologies and their application in all aspects of plant biology. This Research Topic arose from this meeting as a means to capture current research, views, and approaches from the wider plant proteomics community.

### Edited and reviewed by:

Dominique Job, Centre National de la Recherche Scientifique, France

\*Correspondence:

Joshua L. Heazlewood jheazlewood@unimleb.edu.au

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 27 June 2016 Accepted: 25 July 2016 Published: 05 August 2016

### Citation:

Heazlewood JL, Jorrín-Novo JV, Agrawal GK, Mazzuca S and Lüthje S (2016) Editorial: International Plant Proteomics Organization (INPPO) World Congress 2014. Front. Plant Sci. 7:1190. doi: 10.3389/fpls.2016.01190 TECHNICAL ADVANCES IN PLANT PROTEOMICS

As the field of proteomics has evolved, many analytical approaches have relied on advancements in instrumentation as well as the progression of techniques to exploit these changes. Accordingly, this Research Topic highlights a range of updated approaches and provides a number of viewpoints on current technical limitations that are especially pertinent to plant proteomics researchers. Often it is necessary for plant researchers to adapt approaches or push techniques and concepts developed in other species into the field of plant biology. The development of proteomics standards or MIAPEs (Minimum Information About a Proteomics Experiment) were developed by the mammalian field as part of the Proteomics Standards Initiative (Taylor et al., 2007). The strict adoption of these reporting guidelines vary significantly amongst journals and there are various opinions in the plant community as whether they should be strictly adopted. However, in order to make valuable contributions to the proteomics field as a whole, the plant proteomics community should be more willing to embrace these guidelines (Jorrin Novo).

# Two-Dimensional Gel Electrophoresis-Based

The limited research dollars in plant biology has resulted in the persistence of older proteomic technologies. The approach of arraying and quantifying samples by 2-DE has been employed since the 1970s (O'farrell, 1975), however the past decade has seen the adoption of gel-free or shotgun approaches dominating the field. Thus, it is not surprising that many feel it is time to reassess the dominance of 2-DE in plant biology (Anguraj Vadivel), albeit with a view to synergy rather than completely exorcizing the past. Nonetheless, submissions employing and examining new ways to exploit 2-DE were still a common theme. Adaptations to the standard separation protocols where a non-reducing first dimension coupled to a clear native electrophoresis in the second dimension highlighted a means to produce phos-tag zymograms, enabling the in-gel detection of protein phosphorylation (Meisrimler et al.). The technical and laborious nature of standard 2-DE has seen researchers explore the potential of small scale 2-DE using reduced immobilized pH gradient strips (7 cm) and precast mini-gels to identify relevant proteins in a complex lysate, namely from wheat grain extracts (Fekecsova et al.). The success of the approach is highly relevant to plant science given its use of 2-DE and could enable high throughput screens due to its simplicity and reduced sample requirements. A major limitation with 2DE (and to a lesser degree with gel-free approaches) is the inability to survey deep into the total proteome. The hunt for low-abundance proteins from complex plant samples requires the depletion of abundant plantspecific polypeptides such as RuBisCO which is highly abundant in photosynthetic material to enable the visualization of low abundance proteins (Gupta et al.).

In recent years, most phosphoproteomic studies have been conducted using gel-free or shotgun approaches with enriched samples. However, historically the use of 2-DE was seen to hold some advantages for the analysis of phosphorylation, least of which was the visual change in isoelectric point of a protein when its phosphorylation state changed. The advent of sensitive fluorescent dye for phosphorylation (Pro-Q), has more readily enabled the visualization of phosphoproteins by 2- DE. To advance this approach to enable simultaneous protein quantification, a multiplexed approach employing Pro-Q, and the fluorescent protein dye SYPRO Ruby was undertaken on germinating seeds and seedlings of a non-orthodox plant species, Quercus ilex (Romero-Rodriguez et al.). The approach was capable of simultaneously determining both protein changes and phosphorylation changes from samples.

# Mass Spectrometry-Based

The past decade has seen major advances in instrumentation used for proteomic analyses. This has included improvements in sample delivery systems (e.g., nanoflow Ultra-High Performance Liquid Chromatography), mass analyzer (e.g., orbitrap mass analyzer) and fragmentation methods (e.g., electron-transfer dissociation). Collectively these advances have improved sensitivity, dynamic range, and mass resolution which have dramatically improved our capacity to identify and characterize proteins (Heazlewood, 2011). While developments in instrumentation have contributed significantly to advancing the field, there are still applications for "older" technologies such as MALDI-ToF. These instruments were at the forefront of the proteomics revolution, but in the past decade their use has diminished. However, they are still manufactured, relatively simple to use and can be acquired at a reasonable cost. Consequently, they have gained traction for their ability to easily profile samples, discriminate between species and identify pathogens through specific biomarker profiling (Mehta and Silva). These approaches are only now being applied for species determination in plants, but could also be used to assist with breeding and plant biotechnology.

Subcellular proteomics seeks to determine the functions and constituents of organelles, complexes and compartments within the plant cell. Insight into the biology of these systems is often hampered by technical limitations associated with subcellular enrichment strategies. The assessment of organelle purity is usually undertaken by immunoblotting; however, the field of plant biochemistry is significantly restricted by the lack of commercial antibodies that can be used to assess contamination. Targeted proteomic approaches (multiple reaction monitoring) provides an alternative approach to assess the relative abundance of organelle marker proteins in a plant lysate (Parsons and Heazlewood).

The biosynthesis of N-glycans is a well-characterized pathway of plant endomembrane with nearly all biosynthesis members elegantly characterized by molecular genetics techniques. In plants, few studies have applied advanced mass spectrometry-based approaches to survey N-glycans. Thus far, only the characterization of sites harboring N-glycans have been undertaken, with no analysis of site specific N-glycan heterogeneity occurring. It has recently been shown that combinatorial fragmentation approaches involving electrontransfer dissociation (ETD) and higher-energy collisional dissociation (HCD) are capable of being used to identify and characterize N-glycopeptides. This technique has been highlighted on a single plant N-glycopeptide (Ford et al.) and the authors further highlight how advances to this approach can be used to characterize this post-translational modification in complex protein lysates.

# Sample-Based

Sample preparation is at the core of all proteome surveys in plant science. The advanced analytical capabilities found in current mass spectrometers are of little value if sample integrity is compromised. Pertinent examples are surveys of plant subcellular proteomes, where sample preparation is much more important than any downstream characterization. The cell wall proteome is a unique compartment in plants and can be difficult to enrich from contamination cellular proteins. Thus, it is not surprising that the utilization of sequential extraction procedures involving CaCl2, EGTA, and LiCl-complemented buffers is crucial to achieve adequate enrichment (Printz et al.). Such intricate sample preparation procedures represent one of the only ways to confidently assign subcellular locations to a protein identified through proteomic characterizations.

Efforts to survey the entire proteome are significantly restricted by the limitations of current instruments to handle the sample complexity found in higher eukaryotes. This complexity is exemplified by the analysis of protein phosphorylation, an often difficult post-translational modification to confidently detect and quantify. While various phosphoproteomic techniques have been developed, progress in sample enrichment has been at the forefront of enabling reliable phosphoproteomic surveys. Many of these advances, as applied to plant biology have been summarized (Li et al.); the requirement for phosphopeptide and/or phosphoprotein enrichment is central to these highlighted approaches. A case in point is the analysis of signal transduction by mitogen-activated protein kinases (MAPKs). Profiling the phosphoproteomes of transgenic plants with manipulated MAPK pathways is at the forefront of investigations into dissecting these signaling pathways (Takác and Šamaj ˇ ). Such approaches require reproducible enrichment strategies and reliable phosphopeptide assignments and quantification procedures.

The manipulation of samples to enable reproducible quantification has been a feature of proteomics and include peptide labeling (e.g., iTRAQ) and metabolic labeling (e.g., SILAC). These approaches have been frequently employed to monitor changes in phosphopeptides and the latter approach, <sup>15</sup>N metabolic labeling, was recently used to identify the RAPID ALKALINIZATION FACTOR (RALF) peptide receptor in Arabidopsis. The success of metabolic labeling to identify reliable changes in the phosphorylation states of plasma membrane proteins was highlighted by Stes et al.

While the plastid represents a unique structure found in plants, studying this organelle also provides information on the evolution of eukaryotic plants. Since the initial endosymbiotic event, the majority of plastid derived genes were relocated into the nuclear genomes of their host, resulting in the evolution of an N-terminal targeting mechanism in plants. The cleavage of transit peptides during protein import into the plastid can be studied through the application of a sample preparation procedure known as terminal amine labeling of substrates (TAILS). This technique was applied to a freshwater alga (Cyanophora paradoxa), a representative of one of the main lineages after endosymbiosis of the ancestral pre-plastid (Köhler et al.). An intriguing finding from this analysis was that low abundant proteins may have evolved an alternative import mechanism in this lineage.

Pollen coat proteins are essential for a range of factors involved in the interaction with the stigma, including adhesion, recognition, hydration and germination. These features are essential requirements for successful reproduction and ultimately the development of seeds and fruits. The pollen coat proteins of maize, an economically important crop, have only had limited attention, with only a handful of proteins (14) characterized to date (Gong et al.). Pollen coat proteins are prepared by initially treating with an organic solvent then extracted using detergent buffers. These approaches have only recently been applied to maize pollen, and given the initial success, this enrichment approach should yield a deeper proteome in the future.

The diversity of secondary metabolites found in plants is immense and is of great interest due to their involvement in the biosynthesis of many essential plants-specific compounds (e.g., lignin). Since the biosynthetic pathways of secondary metabolites can occur across multiple compartments or can be specific to organs and can often represent minor components within the cell, a targeted approach is crucial for their study by proteomic technologies (Martinez-Esteso et al.). As such, proteomic surveys are often conducted on specific plant components, for example trichomes, tomato, or plastids to ensure enrichment of specific pathways. Such targeted approaches necessitate very specific sample preparation procedures for protein extraction and enrichment as well as working with species with limited genomic resources for data matching.

# Data

The access and availability to next generation sequencing technologies has enabled the application of proteomics to a wide variety of plant species. The application of proteomics to these newly assembled genomes and transcriptomes can also provide support to computationally derived gene models. The verification of proteins along with the confident assignment of spectral data can be used to create resources that can provide a framework for a proteome atlas; a resource for a plant species displaying expression and organ-specific profiles. While these types of repositories have been created for reference species such as Arabidopsis (Mann et al., 2013), they are only now emerging for economically important plant species such as the common bean (Zargar et al.).

The ability to more readily generate a reference database through next generation sequencing technologies has opened up proteome analyses on species which would have previously been ignored, even if of great scientific interest. Such an example is Amborella trichopoda, a shrub endemic to New Caledonia, thought to represent a sister lineage to flowering plants. The recent completion of its genome has enabled the evolutionary exploration protein families, such as the vacuolar processing enzymes which play important roles during seed maturation (Poncet et al.). Such studies can reveal selection biases on specific protein families in these isolated plant species.

# PLANT ABIOTIC STRESS AND PROTEOMICS

One of the most widely used applications of proteomics in plant biology is to investigate protein changes under abiotic stresses. These investigations are motivated by the desire to better understand plant adaptation to external stresses, especially in the context of improving agricultural productivity. This includes the application of quantitative proteomics to dissect cell specific responses, the identification of proteins involved in stress, the characterization of post-translational modifications involved in the stress response and protein–protein interactions to identify signaling networks that regulate processes for individual and combinations of stresses (Gong et al.; Hu et al.).

Both wheat and barley are major cereal crops and productivity is greatly affected by both abiotic and biotic stresses. A plethora of proteomic studies have been conducted on these two crop species, many with a focus on attempting to identify markers for stress tolerance, such as reactive oxygen species scavengers (Kosová et al.). Reactive oxygen species can be produced during unfavorable conditions, such as stress, and are often found in the apoplast, a compartment that connects the plant to its environment. Extracellular glutathione is sensitive to reactive oxygen and its presence within the apoplast is regulated via a gamma-glutamyl-transferase (GGT). Recently, the link between the regulation of apoplastic glutathione by GGT and the presence of defense enzymes was determined through a proteomic analysis of a GGT mutant (Masi et al.). Thus, it is conceivable that apoplastic glutathione is a major sensing component of the apoplast, connecting the external environment to the cell.

A popular developmental stage selected for abiotic stress studies is germination and seedling establishment, as this is when plants are most susceptible to disease and changing conditions. With increased temperatures and changing rainfall patterns, examining the effects of drought was a common theme for studies in this Research Topic. Examining plant species that are adapted or resistant to a stress is a common approach, and such a study was carried out with drought exposed Holm oak seedlings, a hot and dry adapted tree of the western Mediterranean region (Simova-Stoilova et al.). The 2-DE survey of root extracts from drought effected seedlings indicated that the seedlings responded by adjusting basic metabolic pathways and mobilizing defense systems to counteract the stress. The assessment of drought and heat stress in crops provides a direct link to mechanisms in agricultural relevant species. Such an approach in maize sought to identify early signaling responses in maize seedlings through a phosphoproteomic survey using iTRAQ technologies (Hu et al.). A myriad of phosphoproteins were detected from a range of functional classes, notably there was a significant overlap in the response of drought and heats stressed seedlings at the level of phosphorylation.

Another developmental stage with much attention in this area is that of seed development and storage. The conditions used for seed storage or seed aging has a major effect on seed viability. Proteomics was used to assess changes in germinating Brassica napus proteins after seeds were exposed to a treatment of 40◦C and 90% relative humidity (Yin et al.). Elevated levels of peroxiredoxin supported prior work indicating a role for reactive oxygen species in contributing to the seed aging process.

There is an inextricable link between abiotic stresses and hormone response in plants. These interactions have been extensively studied using proteomic technologies in an effort to uncover the resulting hormone induced adaptations. The acquisition of desiccation tolerance is a component of a plants life cycle, namely seed maturation. The maize ABA deficient mutant viviparous-5 was used to dissect ABA-mediated responses to seed maturation by 2-DE from the embryo and endosperm (Wu et al.). A number of proteins were identified as changing in vp5 seeds, significantly small heat shock proteins (sHSPs) increased and late embryogenesis abundant (LEA) proteins decreased. These findings suggest that sHSPs may be more loosely regulated by ABA during seed maturation.

The effects of temperature conditioning on seeds can significantly impact plant germination and growth and can greatly affect resultant yields. The emergence of sprouts from garlic can be influenced by low temperature treatments, but depending on the duration, can also effect growth rates and yield. A 2-DE survey of the garlic clove subjected to low temperature revealed major changes to metabolic processes which established a new cellular homeostasis that impacts growth rate, plant weight, and yields (Dufoo-Hurtado et al.).

Seagrasses are found in tropical and temperate environments and have important ecological roles in marine habitats. An important factor effecting their distribution is salinity, however little is known about their adaptation and tolerance mechanisms in the saline marine environment. To explore this adaptation process, the impact of a hypersaline environment on the Cymodocea nodosa proteome was assessed using 1D-PAGE and tandem mass spectrometry (Piro et al.). The results indicated that the hypersaline treatments increased glycolytic protein levels and vacuolar components (e.g., Na+/H+-antiporter) to deal with these conditions.

Iron and zinc are essential micronutrients for normal plant growth and development and are predominantly taken up from the soil. Iron deficiency is a major problem due to its requirement in redox reactions associated with photosynthesis and respiration and as an essential co-factor for a range of cellular enzymes. The effects of iron deficiency on respiration and other aspects of plant development have been extensively explored using a myriad of analytical techniques including proteomics (Zargar et al.). However, few studies have documented the effect on the apoplast, the compartment initially likely to detect a deficiency. Interestingly, an analysis of the leaf apoplast proteome of Beta vulgaris subjected to iron deficiency revealed few changes, indicting the apoplastic proteome is primed to deal with changes in iron concentrations (Ceballos-Laita et al.). Elevated zinc levels can result in zinc salts in the soil, which produce an osmotic response similar to that of saline stresses. To examine the effect of elevated zinc levels, it is thus important to also examine the effects of salt stress. Lettuce (Lactuca sativa L.) is a popular and economically important vegetable crop which is sensitive to salt stress and grown in a variety of soil types. The leaves of lettuce exposed to zinc and salt stress were analyzed by tandem mass spectrometry and revealed an accumulation in proteins related to glycolysis, nitrogen metabolism, hormone biosynthesis and protein metabolism (Lucini and Bernardo). The overlap of proteins identified between elevated zinc and the salt stress were similar, although the zinc stress appeared to enhance the effects.

The ultimate abiotic stress for plants is high levels of ionizing radiation, conditions which exist in the radio-contaminated areas around Chernobyl and Fukushima. Proteomic studies to assess the effects of these environments have been conducted on seeds harvested from soybean and flax from radio-contaminated areas of Chernobyl during two successive generations (Rashydov and Hajduch). The seed proteomes had altered abundances of glycine betaine, seed storage proteins, and proteins associated with carbon assimilation.

# PLANT BIOTIC STRESS AND PROTEOMICS

Investigating plant pathogen interactions is a major focus in plant biology as it attempts understand this critical and often deleterious interaction. This is often driven by the desire to reduce the impact of disease in crops, which is thought to result in 10–20% reductions in yield per year. Much of the work focuses on determining the biochemical and molecular parameters associated with host resistance or susceptibility. These can often manifest itself as specific resistance traits (e.g., a specific receptor) or general properties (e.g., thickened cell walls). Plant proteomics has played a significant role in extending our knowledge in this area as it attempts to uncover mechanisms of resistance.

The apoplast plays a central role in between plants and pathogens as it represents the initial interaction and communication point during infection. Thus, it is no surprise to find a plethora of proteomic surveys applied to the dissection of this compartment associated with plant pathogen interactions (Gupta et al.). This includes multiple analyses of the secretome from a range of plant species infected with specific pathogens. A specific example is the proteomic analysis of the apoplastic fluid of the Coffea arabica in an attempt to identify resistance markers. The coffee industry has been devastated for over a century by coffee leaf rust caused by the fungus Hemileia vastatrix. Employing 2-DE in combination with a susceptible and resistant variety of C. arabica, a range of resistance associated proteins were identified, including glycohydrolases, proteases, and other pathogenesis-related proteins (Guerra-Guimarães et al.).

# PROTEOMICS TO PROFILE PLANT CULTIVARS

The variations in response, growth, and yield of different cultivars has been fundamental in our ability to expand the range and adaptability of crop plants. However, the confident assignment and detection of markers in various cultivars for factors such as pathogen resistance and stress tolerance is challenging. While omic technologies have attempted to play a role in marker discovery and determination between cultivars, these approaches have had varying success. However, they likely represent one of the most promising options for breeders when used effectively (Zivy et al.).

Processes such as cold acclimation can significantly improve the tolerance of a many plants to freezing temperatures. The process of cold acclimation in alfalfa (Medicago sativa) is essential to enable tolerance during severe winter months. A proteomic analysis was conducted to uncover the acclimation process in alfalfa of a freezing-tolerant cultivar compared to a freezingsensitive cultivar (Chen et al.). The analysis revealed that more proteins changed in the resistant cultivar when subjected to cold acclimation supporting the notion of priming of the cultivar for freezing conditions.

The grape industry produces over 75 million metric tons annually. Grapevine production can be significantly affected by drought, climate and salinity. The completion of the grapevine genome in 2007 has enabled a range of proteomic surveys to be confidently conducted on the effects of abiotic stress at the developmental and varietal level (George and Haynes), again demonstrating the wealth of knowledge in genetic varieties. The grape berry skin is rich in secondary metabolites and is an important mechanical barrier against pathogens and damage by injury. Various grape cultivars are known to vary in their anthocyanin content, and thus a metabolomic and proteomic analysis was performed on the grape cultivars Riesling Italico, Pinot Gris, Pinot Noir, and Croatina (Negri et al.). The analysis indicated a relationship between secondary metabolism and pathways associated with primary metabolism in the development of the grape berry skin.

Barley is relatively salt and drought tolerant, however varieties can respond differently. The response of crowns from the barley cultivar Amulet to varying soil water capacities was examined using 2D-DIGE (Vitamvas et al.). A range of metabolic and protective enzymes were found to respond, however physiologically the cultivar was found to be sensitive to drought stress.

# PROTEOMICS AND POST-TRANSCRIPTIONAL REGULATION

The discrepancies observed between transcriptional and protein responses in plants have been amplified as proteomic technologies have advanced. While this difference does not appear to be strongly observed in mammals and yeast, the question arises "Are plants special?" (Velez-Bermudez and Schmidt). The observations that plants have specialized ribosomes with the potential for high levels of heterogeneity and distinct gene splicing mechanisms supports the proposition that plants may have additional controls that explain the observed discordance. A case in point is that of seed germination and aging, where it would appear that translation of stored mRNA is an essential component of seed quality. Thus in many species, factors effecting stored mRNA make a more significant contribution than those effecting transcription. The control of germination is thus likely to constitute mRNA translation and protein post-translational modifications rather than transcriptional regulation (Galland and Rajjou). Therefore, the application of proteomic technologies to seed germination represents an essential tool to dissect this vital aspect of plant growth and development.

# CONCLUSION

The 1st INPPO World Congress in 2014 marks the initial steps in bringing plant proteomic research to a level currently undertaken in the areas of human and microbial research. The collection of submissions to this Research Topic, to specifically support the inaugural INPPO congress, highlights the range of activities currently being undertaken by plant proteomic researchers and reflects the current state of the field. These submissions highlight the fact that plant proteomics only touches the surface of biological complexity and we are likely far from exploiting the full potential of this technique. At the methodological level, much work still employs first generation approaches (2-DE), but the community is slowly moving to more advanced approaches. There are still important challenges yet to be faced, including revealing more of the proteome, confident protein identifications with quantification, data validation, and addressing orphan and recalcitrant species. Ultimately the community needs to improve its adoption of advanced techniques to enable us to better address biological questions and develop translational outcomes. These questions will be key discussion points during the 2nd INPPO World Congress, due to be held in September of 2016 in Bratislava (Slovakia). It is anticipated that the diversity of

# REFERENCES


approaches and quality of science will meet and likely exceed that of the inaugural INPPO of 2014.

# AUTHOR CONTRIBUTIONS

JH wrote the draft. JJ, GA, SM, and SL edited and contributed to the initial draft. JH compiled the final version.

# ACKNOWLEDGMENTS

The Congress was supported by DFG Lu-668/16-1 and by the companies listed on the web-page (https://www.inppo2014.unihamburg.de/).


**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.

Copyright © 2016 Heazlewood, Jorrín-Novo, Agrawal, Mazzuca and Lüthje. 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.

# Scientific standards and MIAPEs in plant proteomics research and publications

Jesús V. Jorrín Novo\*

Agroforestry and Plant Biochemistry and Proteomics Research Group, Department of Biochemistry and Molecular Biology-ETSIAM, University of Cordoba-CeiA3, Córdoba, Spain

Keywords: plant proteomics, scientific standards, MIAPE, scientific publications, comparative proteomics

"I suspect that the authors are capable of doing a much better job of preparing a scientific Ms., and I wish that they had applied more effort with this one. If the writing and Ms. preparation are this sloppy and amateurish, can the research be trusted? The authors simply list quantitative results, and make some broad, generalized, pedestrian comments. Hardly a Discussion."

(Anonymous referee)

In this opinion paper it is my intent to briefly discuss some key issues related to scientific standards in plant proteomics research and the "Minimal Information About a Proteomics Experiment" (MIAPEs) requested for derived publications. It is mainly aimed at beginners rather than scientists who have established a long trajectory and experience within the field, trying to rationally connect proteomics and plant biology. The content was presented at the "1st INPPO World Congress on Plant Proteomics: Methodology to Biology," September 2014, and has been discussed in reviews published by the author, with the most recent referenced herein (Jorrin Novo et al., 2009, 2014, 2015; Valledor and Jorrin, 2011). As an opinion paper it should be, what else¡, subjected to comments, disagreements, or criticisms, but at the very least open discussion. It reflects 12 years of active research, as an author (who has had some experience with rejected manuscripts), a reviewer, and an editor who has handled around 400 manuscripts (about 50% of which were ultimately accepted).

MS-based Proteomics, as an analytical tool, has developed to an unanticipated level in a very short period of time. Even so, its potential remains far from being fully exploited especially as a component of plant biology in comparison with other organisms (i.e., humans, yeast, bacteria). Some areas (PTMs, interactomics) or techniques (targeted, arrays, imaging) are minimally represented in the current plant literature, while others (protein trafficking, degradation, protein function at the –omics level) remain absent or anecdotal. Despite being a powerful technique, it has limitations (quantitation, orphan organisms). As my friend Juan Pablo Albar (recently deceased) used to say, "Jesus, real proteomics is only possible when studying organisms with a sequenced genome as we should pretend to identify gene products." It is not a panacea, or a miracle. By itself it is almost impossible to unravel biological processes. Results must be validated, and contrasted with those obtained by using biochemical, molecular (classical, other –omics), or cellular biology approaches. We now appreciate that the protein world is much more complex from a structural and functional point of view than ever imagined. It is increasingly clear that in its present state, proteomics is mostly descriptive and to a great extent speculative. While description is valuable by itself, it is not always adequate to support biological speculations or support speculative conclusions. While this opinion might be considered controversial by some, it is shared by others and is well presented in the last review by Paola Picotti (Boersema et al., 2015). Because of this evolution in the nature of (plant) proteomics, we have chosen to present a philosophical rather than data-based contribution.

### Edited by:

Nicolas L. Taylor, The University of Western Australia, Australia

Reviewed by:

Brian Mooney, University of Missouri, USA

> \*Correspondence: Jesús V. Jorrín Novo, bf1jonoj@uco.es

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 15 April 2015 Accepted: 15 June 2015 Published: 30 June 2015

### Citation:

Jorrín Novo JV (2015) Scientific standards and MIAPEs in plant proteomics research and publications. Front. Plant Sci. 6:473. doi: 10.3389/fpls.2015.00473

For those with limited prior experience, but with a welldesigned biological project, it is important to remember that proteomic analysis is much more than just sending samples to a Proteomics Service (you should not pretend that mass spectrometists are knowledgeable about plant biology), then blindly accepting the results, and preparing a more or less confident protein identification and quantification table. Only if one understands both the experimental system and the proteomics techniques applied can we understand the results well enough to speculate about how, why, and what insight the results provide? Proteomics has innate limitations which must be taken into account; data must be critically evaluated, correctly validated and interpreted, and finally, submitted manuscripts should fit into the general scientific and particular proteomics standards or MIAPEs (Minimum Information About a Proteomics Experiment). Such a MIAPEs have been translated to a number of documents elaborated by the Proteomics Standard Initiative within the Human Proteomics Organization (HUPO; http://www.psidev.info/node/91); they are related to "community standards for data representation in proteomics to facilitate data comparison, exchange and verification" (Orchard et al., 2003; http://www.psidev.info/), making reference to each of the steps in a standard proteomics workflow (gel electrophoresis, gel informatics, MS general, MS informatics, MS quantitation, column chromatography, capillary electrophoresis, molecular interaction). These standards are requested by the four top-ranking journals in the field (by year of appearance, Proteomics, Molecular, and Cellular Proteomics, Journal of Proteome Research, and Journal of Proteomics), and briefly summarized in the corresponding journal instruction to authors. Other journals that publish proteomics data are typically more concerned with the biological contribution than a complete description of the methods used, and thus do not highlight these standards. However it behooves us all in the plant proteomics community (researchers, reviewers, editors) to strictly adhere to both these standards and MIAPEs.

When considering a manuscript for publication, apart from formal aspects (English edition and format requirements as indicated in the instruction to authors), the failure to meet general scientific standards is the most obvious and immediate reason for rejection. It is important to keep these standards in mind throughout the processes of planning, conducting, interpreting, and describing an experiment: (i) Experimental design (number of experiments, biological or analytical replicates, sample size, sample homogeneity); (ii) Method optimization and validation (the employed techniques must be validated from an analytical point of view, and specificity, precision, accuracy, dynamic range, limit of detection, limit of quantitation, should be known); (iii) Analysis of the data and statistics); and (iv) Interpretation of the data (taking into account the experimental design, the employed methods, and the statistical analysis). Reproducibility and bias minimization, as well as validity of the data from a biological point of view (the extent to which similar findings are reported using other experimental systems and/or approaches) are also key issues. The proteome is dynamic, even for clonal and synchronized cells, and because of this the mean coefficient of variance of a proteome is quite variable. Furthermore, it is crucial to remember that the proteomic results we describe and interpret from a biological point of view are but a single fixed photograph of a whole movie. We cannot pretend to fathom very complex biological processes from the results of a single experiment, even if we have resolved and identified thousands of protein species. Typically, the plant samples being analyzed include a complex mixture of tissues and cell types, each of which has its own protein signature and not all of which respond identically to the experimental variables.

The performed work can be translated to an acceptable manuscript if: (i) The main contributions to the experimental system or biological process are clearly presented and summarized in the abstract and introduction, ensuring its understanding (going beyond just a description of the proteome as far as possible without pretending to review the covered topic); (ii) proper terminology is correctly used; (iii) the methods section is written such as it ensures the repetition of the experiments by any who, anywhere; (iv) results, original in preference to very elaborately analyzed data, are presented; and (v) the discussion section does not contain unwarranted speculations but rather conclusions and hypotheses supported by the data presented. In the table accompanying this opinion paper (**Table 1**) is summarized the major causes of rejection of a submitted manuscripts or at least the main criticisms based on the author's experience as editor and/or author. One specific issue deserves some emphasis; the use of scientific terms. It is critical that we are all (authors, reviewers, editors, and readers) considering the same thing! Scientific terminology must be very precise and unambiguous, although it is also true that archaic can be productively adapted and properly interpreted in the context of contemporary techniques. The literal translation of specific terms from genomics to proteomics can also generate confusion. The scientific community should discuss and agree on that, and the creation of a nomenclature committee is a need. For example, the use of "protein species" or "protein forms" rather than just "proteins," as previously proposed (Jorrin et al., 2006; Schlutter et al., 2009; Smith et al., 2013). As far as possible it should be clearly stated which gene product is referred to, and whether the protein species corresponds to a multigene family, isogene, or allelic variant. In the case of orphan organisms whose genome is not sequenced reference to the orthologs should be made. Also, based on proteomics experiments we can only describe differences in protein species abundance. Terms such as differences in protein expression (in fact the genes are expressed), up or down-regulation, induction, repression, must be avoided. Up or down gene expression is just one of the possible mechanisms explaining differences in protein abundance.

Finally, I advise being modest and humble when dealing with proteomics research. Contemporary MS-based proteomics methods can generate huge datasets in a relatively short period of time. This is quite different from even the recent past. For my thesis I spent 4 years working with just one protein, the enzyme phenylalanine ammonia-lyase. Final analyses will depend on the comparisons made. In a single experiment, the best results we can imagine will include less than 10% of the total proteome. Because of this, I suggest replacing "proteome" with "extractome" in most instances. This is based upon the following considerations:



These comments mostly apply to comparative proteomics papers that represent the higher percentage of the submitted manuscripts. The table could be completed with the MIAPE standards (http://www.psidev.info/node/91).


While there are many other points that should be described/discussed/argued about in terms of the future of plant proteomics. Like the proteins we address, the field itself if dynamic and should be continuously evolving. Other specific points have not been incorporated due to the length restrictions by the journal for an opinion paper.

There is life beyond descriptive proteomics, and it is increasingly important that we consider biological context. If the results obtained do not fit our hypotheses (or preconceptions?) or fail to provide satisfactory answers to our research questions, then we must avoid the temptation to "tread water," and move on to application of additional approaches be they experimental or computational.

# Acknowledgments

To my mom, Justina, who recently died. Great person and better mother. I am where I am, as person and scientist, thanks to her.

# References


Mine and my brother's happiness and education were her life priority. Thanks a lot to my colleague and friend Jan Miernyk for sharing thoughts, encouragement, and, more important, critical reading and improvement of the manuscript.


**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.

Copyright © 2015 Jorrín Novo. 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.

# Gel-based proteomics in plants: time to move on from the tradition

Arun K. Anguraj Vadivel\*

*Department of Biology, Western University, London, ON, Canada*

Keywords: gel-based proteomics, plant proteomics, gel-free proteomics, mass spectrometry, technical fusion

In more than 20 years of the proteomics era, considerable technical developments and scientific discoveries have contributed to the advancement of this field of research. Gel-based proteomics has been popular for studying the proteomic changes during growth and development of plants as well as for the analysis of responses to different biotic and abiotic stimuli. The most widely used gel-based technique is two dimensional (2D) gel electrophoresis, in which around 2000 protein spots can be clearly visualized and processed prior to identification by mass spectrometry. Although many techniques have been developed to improve the quality and number of spots in a 2D gel, these enhancements are still not good enough to study the entire cellular proteome. It is important to also consider low abundant proteins, many of which could play critical roles in particular biological processes. Considering that thousands of proteins are expressed at a given time and each protein can undergo one or more post-translational modifications—are we able to capture all this information and the entire proteome using gel-based proteomic techniques? Are gel-free alternatives capable of addressing these challenges? The answer is no for both approaches. So, how long are we going to run traditional 1D and 2D gels in our research? Though we have been successful in developing many advanced gel-free proteomic techniques to study the cellular proteome, we still lack advances that enable the examination of each and every protein expressed in a system as well as any post-translational modification (PTM) at a given time. Gel-free and gelbased techniques may complement each other, however there needs to be synergy or a "technical fusion" in which researchers can find a balance between quick and efficient methods for studying whole cell proteomes in plants.

# Proteins and Proteomics

Many plants contain more genes than the human genome, which we consider one of the most complex organisms. However, the number of genes cannot be easily correlated with the number of functional proteins since an organism has multiple layers of gene regulation. Gene expression can be regulated during transcription, splicing, translation and post-translational maturation of the protein. Several physical factors such as tissue types, developmental stages, environmental stimuli and stresses also affect gene expression in plants. In addition, due to the number of ploidy levels in some plants, and the presence of several protein isoforms, studying a large plant proteome can be more complex than studying an animal proteome of apparently similar size. Different proteomic tools have been developed in the past few decades, most of which were first used for studying animal proteomes and then later adopted for plants. Chromatography and protein electrophoresis have been ideally used for the separation of proteins whereas mass spectrometry (MS) has been used for protein identification. Different kinds of MS have been developed that use variable ionization sources, fractionation and mass analyzers. Among these, the orbitrap mass analyzer is now the most widely used for high resolution MS. Generally, MS based proteomics is used for protein identification and quantification studies either in combination with gel-based (1D, 2D or 3D) or gel-free techniques and can be used with label free or tag-based techniques (ICAT, iTRAQ etc.).

### Edited by:

*Jesus V. Jorrin Novo, University of Cordoba, Spain*

### Reviewed by:

*Roque Bru-Martinez, Universidad de Alicante, Spain Luis Valledor, Universidade de Aveiro, Portugal*

### \*Correspondence:

*Arun K. Anguraj Vadivel, arunkumaran07.ak@gmail.com*

### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *31 December 2014* Accepted: *09 May 2015* Published: *27 May 2015*

### Citation:

*Anguraj Vadivel AK (2015) Gel-based proteomics in plants: time to move on from the tradition. Front. Plant Sci. 6:369. doi: 10.3389/fpls.2015.00369* The workflow of MS based proteomics has recently been reviewed by Jorrin-Novo et al. (2015). The review covers all research articles in plant proteomics that have been published in the journal Proteomics since 2000. During that period there have been many modifications to proteomic methods used to obtain high protein coverage and improve the number of proteins identified per sample. This article covers some of the more recent studies in the MS workflow before raising the question of how can we improve the output by merging different techniques i.e., development of different methods in gel-based and gel-free techniques to improve protein separation, digestion and recovery of peptides, and MS or MS/MS data analysis.

# Gel-based Proteomics

The most widely used methods in gel-based proteomics comprise the separation of proteins by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). This can be conducted in a single dimension (1D) based on molecular weight or in two dimensions (2D) based on a proteins isoelectric point (using immobilized pH gradient gel strips) and molecular weight (SDS-PAGE). Identification of proteins in a sample follows; separation by SDS-PAGE, in-gel digestion with an enzyme and MS. 2D gel electrophoresis is the most popular approach in plant proteomics as it allows separation of proteins by isoelectric point and molecular mass. Over 4000 proteins can be identified using 2D gel electrophoresis combined with MS (Imin et al., 2001). PTM specific stains have been developed for phosphoproteins (e.g., Pro-Q Diamond) and glycoproteins (e.g., Pro-Q Emerald, Dansylhydrazine), both of which have emerged as good techniques to study specific PTMs using 2D gels (Marondedze et al., 2013; Wang et al., 2014). However, 2D gels have limitations in terms of separation resolution of complex proteomes, number of spots and protein recovery; all which can be improved by merging methods developed separately for each step in the gel-based proteomic work flow. However, current 2D separation techniques are still not adequate, thus some recent developments in this area are outlined.

Recently, Colignon et al. (2013) developed a protocol for the three dimensional separation of proteins. 3D separation is an advanced 2D gel separation method and their protocol was developed to addresses co-migration interferences. It uses isoelectric focusing and sample fractionation followed by two consecutive separations by SDS-PAGE, with two different buffer systems to evade co-migration associated drawbacks that can affect protein resolution. In this study, MS/MS analyses from both 2D and 3D gels were compared to validate the protocol. There are previously reported methods available for 3D gels to study particular protein families (Jiang et al., 2009) but the method reported by Colignon et al. (2013) provides a wide range of applications in quantitative profiling of complex proteomes and in the identification of PTMs such as protein phosphorylation and sumoylation. Interestingly, a 3D-western blot is also possible. The authors suggest that by combining this method with a differential gene expression approach, the relative abundance of modified proteins can be studied in two different samples. However, overcoming extraction problems after protein separation still remains i.e., recovery of proteins from the gel.

One of the critical steps is protein digestion with proteases like trypsin, chymotrypsin, etc. There have been a number of methods developed for peptide recovery after digestion. A tube gel digestion has been developed for protein recovery that does not require protein electrophoresis to study membrane proteins (Lu and Zhu, 2005). Takemori et al. (2014) developed a method to improve peptide recovery from polyacrylamide gels in a tube using a disulfide-containing analog of bis-acrylamide called bisacrylylcystamine (BAC). In this technique, release of peptides from the gel was enhanced by adding tris-(2-carboxyethyl) phosphine (TCEP) instead of DTT for complete dissolution of the BAC-cross-linked acrylamide gel. BAC gels can be used for complex membrane proteins and can improve recovery prior to MS analysis. However, it is difficult to detect less abundant proteins in the samples even when used in conjunction with nanoLC–MS/MS. However, merging this method with 2D gel technology to create a 2D BAC gel would improve protein recovery from in-gel digestion of 2D gel spots and improve identifications. Moreover, developing 3D separation techniques with BAC gels with two different buffer systems, as discussed earlier, would give improved separation resolution to obtain more spots in conjunction with peptide recovery. This represents an excellent fusion of techniques if downstream process can be successfully applied.

The next critical step in the MS workflow is MS or MS/MS data analyses in order to accurately match the correct peptide to the spectra. Silva et al. (2014) discussed several ways to improve data visualization and analyses from both MALDI-TOF and ESI-MS approaches. The study also includes suggestions about feature selection or spot detection in gel-based proteomics, which could be helpful for researchers in this field. However, the data visualization methods discussed in this article only work when upstream processes are carefully monitored. The study also demonstrated the effectiveness of statistical and multivariate tools. As discussed before, developing methods for 3D separation of proteins using BAC gels for higher resolution of separation and recovery of peptides in combination with perfect feature selection for data visualization and multivariate analyses tools could be worth considering in the area of gel-based proteomics. It is undeniable that advances in technical development can take research to the next level; however there is always room for further improvements.

# Gel-Free Proteomics

Some of the limitations in gel-based proteomics can be ignored when employing gel-free proteomics. However, both techniques can still complement each other, and their selection of either is highly dependent on the sample or the question. The most popular gel-free approach among researchers is multi-dimensional protein identification (MudPIT) comprising strong cation-exchange (SCX) fractionation, reversed-phase (RP) chromatography and tandem mass spectrometer (MS/MS) (Link et al., 1999; Washburn et al., 2001). It comprises insolution digestion instead of in-gel digestion. The digested peptide mixture is loaded onto chromatography columns which are in line with the MS/MS. At least 2000 proteins can be identified in a sample using the MudPIT approach (Hernandez et al., 2012). Over 12,000 proteins have been identified in different organs of Arabidopsis and in maize leaf (Baerenfaller et al., 2008; Facette et al., 2013). Recently, Link and Washburn (2014) established two approaches to yield high quality tandem mass spectra from complex protein solutions. A multidimensional system is considered comprehensive and highly sensitive approach for protein identification in a complex sample.

A recent study compared an automated (online) or manual (offline) format for MudPIT as well as different quantitative MudPIT strategies using label-free and tandem mass tag (TMT) isobaric tagging (Magdeldin et al., 2014). The study concluded that higher sequence coverage and more peptide/protein identifications can be achieved using online MudPIT rather than when employing offline sample fractionation approaches prior to MS. Despite the recent advancement in gel-free MS techniques, there are some inherent shortcomings that come with this method. MudPIT experiments can be relatively lengthy processes due to the number of fractions produced and the time it takes to analyze each fraction by MS using the reverse phase gradient. Duration of the experiments is the major problem to be resolved in gel-free methods.

# Technical Fusion

Several studies that specifically compared gel-free and gel-based proteomics strategies emphasized the complementary nature of the two approaches. Nearly 4000 spots can be processed from a single gel with gel-based methods and over 12,000 proteins can be identified in a sample in the advanced gel-free method. Thus, each method separately produces a significant number of proteins. However, would a combination of both methods (technical fusion) allow us to study an even larger

# References


number of proteins? A combination of multiple experiments and analyses has produced a great number of protein identifications (Feng et al., 2009), so a technical fusion would definitely result in a greater number of identified proteins. Moreover, it is not just about the number of proteins/spots, the amount of information obtained about each protein is also important e.g., PTMs. Within gel-based proteomics, BAC gel methods should be fused with the 3D separation techniques to produce greater separation resolution and protein recovery after digestion. 3D BAC separation using two buffer systems would probably be a good technique to consider for not only studying the expressed proteins in a system but also PTMs. On the other hand, similar to the subtraction library system used for creating cDNA libraries, there could be a strategy of protein subtraction and pooling from two different methods. For instance, in a sample processed by both gel-based and gel-free methods, the peptides/proteins (m/zvalue) identified in the gel-based methods could be subtracted using exclusion lists from the in-solution digested samples during tandem MS. However, one of the main reasons behind the lack of method development in plant proteomics is budget. The issue of low budgets in plant science laboratories and their effect on proteomic research activities has been mentioned in a recent review by Jorrin-Novo et al. (2015). For example, gel-based techniques, which have been traditionally used in plant proteomics, are now lagging behind in terms of proteome coverage. We urge our fellow plant researchers to move on from our traditional approaches and develop novel strategies like 3D BAC gels or other fusions of techniques and merge different proteomic approaches to better capture proteomic information from the cell.

# Acknowledgments

I thank Dr. Sangeeta Dhaubhadel (AAFC, London, Canada) for her support and Mr. Kishor Duwadi and Mr. Hemanta Raj Mainali for their critical comments on this manuscript.

Proteomics 1, 1149–1161. doi: 10.1002/1615-9861(200109)1:9<1149::AID-PROT1149>3.0.CO;2-R


loss in LC/LC-MS/MS. J. Proteome Res. 13, 3826–3836. doi: 10.1021/pr50 0530e


Washburn, M. P., Wolters, D., and Yates, J. R. III. (2001). Largescale analysis of the yeast proteome by multidimensional protein identification technology. Nat. Biotechnol. 19, 242–247. doi: 10.1038/ 85686

**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.

Copyright © 2015 Anguraj Vadivel. 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.

# Two-dimensional phos-tag zymograms for tracing phosphoproteins by activity in-gel staining

### Claudia-Nicole Meisrimler 1, 2 \*, Alexandra Schwendke<sup>1</sup> and Sabine Lüthje<sup>1</sup> \*

<sup>1</sup> Plant Physiology, Biocenter Klein Flottbek and Botanical Garden, University of Hamburg, Hamburg, Germany, <sup>2</sup> Laboratoire de Biologie du Développement des Plantes, Commissariat à l'Energie Atomique et aux Energies Alternatives, Institut de Biologie, Environnementale et de Biotechnologie, Saint-Paul-lez-Durance, France

Protein phosphorylation is one of the most common post-translational modifications regulating many cellular processes. The phos-tag technology was combined with two-dimensional zymograms, which consisted of non-reducing IEF PAGE or NEPHGE in the first dimension and high resolution clear native electrophoresis (hrCNE) in the second dimension. The combination of these electrophoresis methods was mild enough to accomplish in-gel activity staining for Fe(III)-reductases by NADH/Fe(III)-citrate/ferrozine, 3,3′ -Diaminobenzidine/H2O<sup>2</sup> or TMB/H2O<sup>2</sup> in the second dimension. The phos-tag zymograms can be used to investigate phosphorylation-dependent changes in enzyme activity. Phos-tag zymograms can be combined with further downstream analysis like mass spectrometry. Non-reducing IEF will resolve proteins with a pI of 3–10, whereas non-reducing NEPHGE finds application for alkaline proteins with a pI higher than eight. Advantages and disadvantages of these new methods will be discussed in detail.

Keywords: phosphorylation, IEF, phos-tag, NEPHGE, high resolution clear native electrophoresis, in-gel activity staining

### Introduction

Protein phosphorylation, one of the most common post-translational modifications, can alter enzyme activity and subcellular localization as well as target proteins for degradation and can effect changes in protein-protein interactions (Cousin et al., 2013; Gerbeth et al., 2013; Uhrig et al., 2013). Monitoring the phosphorylation status of proteins is, thus, very important for the evaluation of diverse biological processes. Methods to quantify particular phosphorylation events include radioactive labeling, immunodetection of site-specific phosphorylations, phospho-specific site mapping in peptide mass fingerprinting, chemical labeling but also in-gel phospho stainings (e.g., Pro-Q Diamond <sup>R</sup> , all blue and quercetin staining) (Ferrão et al., 2012; Wang et al., 2014).

Complex protein samples are often separated by polyacrylamide gel electrophoresis (PAGE), before mass spectrometry (MS) analysis. After PAGE, immunodetection or phospho staining are the most commonly applied techniques to detect phosphorylated proteins.

Edited by:

Jesus V. Jorrin Novo, University of Cordoba, Spain

### Reviewed by:

Ning Li, The Hong Kong University of Science and Technology, China Jesus V. Jorrin Novo, University of Cordoba, Spain Yueming Yan, Capital Normal University, China

### \*Correspondence:

Claudia-Nicole Meisrimler, Plant Physiology, Biocenter Klein Flottbek and Botanical Garden, University of Hamburg, Ohnhorststraße 18, D-22609 Hamburg, Germany; Laboratoire de Biologie du Développement des Plantes, Commissariat à l'Energie Atomique et aux Energies Alternatives, Institut de Biologie, Environnementale et de Biotechnologie, Saint-Paul-lez-Durance, F-13108, France claudia.meisrimler@cea.fr; Sabine Lüthje, Plant Physiology, Biocenter Klein Flottbek and Botanical Garden, University of Hamburg, Ohnhorststraße 18, D-22609 Hamburg, Germany s.luthje@botanik.uni-hamburg.de

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

> Received: 15 December 2014 Accepted: 23 March 2015 Published: 14 April 2015

### Citation:

Meisrimler C-N, Schwendke A and Lüthje S (2015) Two-dimensional phos-tag zymograms for tracing phosphoproteins by activity in-gel staining. Front. Plant Sci. 6:230. doi: 10.3389/fpls.2015.00230

**Abbreviations:** BNE, blue native electrophoresis; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate; CNE, clear native electrophoresis; DOC, sodium deoxycholate; DOMA, n-Dodecyl-beta-D-Maltoside; hrCNE, high resolution clear native electrophoresis; IEF, isoelectric focusing; NEPHGE, non-equilibrium pH gel electrophoresis; TMB, 3,3′ ,5,5′ -Tetramethylbenzidine.

The fluorescent stain, Pro-Q Diamond <sup>R</sup> by life technologies, gives the opportunity to detect phosphoserine-, phosphothreonine-, and phosphotyrosine-containing proteins without sequence or context specificity (Miller et al., 2006). Currently Pro-Q Diamond <sup>R</sup> is a standard staining for SDS-PAGE. In contrast it is not often combined with native PAGE (Tsunaka et al., 2009) and no literature can be found on the combination with zymograms. An alternative to phospho staining is phos-tag PAGE, a phosphate affinity electrophoresis for the mobility shift of phosphoproteins (Kinoshita et al., 2006; Kinoshita-Kikuta et al., 2007; Kinoshita and Kinoshita-Kikuta, 2012). A dinuclear metal [Mn(II) or Zn(II)] complex of 1,3-bis[bis(pyridin-2 ylmethyl)-amino]propan-2-olato acts as a phosphate-binding tag molecule, phos-tag, in an aqueous solution under physiological conditions. Recently Mn(II)-phos-tag Blue Native PAGE (BNE) was accomplished in the first dimension (Deswal et al., 2010).

Native PAGE methods in combination with phosphorylation analysis are mainly needed for the characterization of proteinprotein interactions, complex assembly and activity regulation, which are prerequisite for the understanding of cellular processes. A variety of native PAGE methods exist (BNE, CNE, native Tris-PAGE, native Acetate-PAGE) and the most optimal can be chosen depending on the sample and the scientific question to be answered. Native PAGEs, with more or less modified protocols, are often used for zymograms because of reduced denaturizing conditions, e.g., high salt concentrations, reducing agents, and strong detergents (Wittig and Schägger, 2005, 2008; Wittig et al., 2006, 2007; Burré et al., 2009; Führs et al., 2009, 2010) which can effect the activity of a protein. It is likely that strong detergents (e.g., SDS), reductants [dithiothreitol (DTT), 2-mercaptoethanol] or heating could not only influence the protein activity but also the phosphorylation.

The standard zymograms (non-reducing SDS-PAGE, without heating of the protein sample) are commonly used for proteolytic enzymes (Vandooren et al., 2013), but combination of different electrophoresis methods and various in-gel activity staining makes the method applicable for different enzyme activities (Manchenko, 2002). In the past, activity in-gel stainings after isoelectric focusing (IEF) slab gels were reported for different enzymes, e.g., malate dehydrogase, peroxidase, quinone reductase, Fe(III)-reductase, superoxide dismutase, catalase and others (Mika et al., 2010; Meisrimler et al., 2011; Kukavica et al., 2012; Lüthje et al., 2014). In the standard IEF-PAGE, protein separation is based on their pI and oriented from basic to acidic pH. A related method, the non-equilibrium pH gel electrophoresis (NEPHGE), also separates proteins by their pI. In NEPHGE, the protein separation is reversed in comparison to IEF-PAGE. NEPHGE was developed to resolve proteins with extremely basic pI (pH 8.5–12.0) (Lopez, 2002). During NEPHGE, proteins are not focused to their pI as in the standard IEF-PAGE. Instead proteins move through the gel based on their charge. For this reason, the accumulated volt hours (Vh) determine the protein pattern across the gel and have to be kept constant to ensure reproducibility.

Furthermore, different native PAGE procedures, e.g. blue native PAGE (BNE), high resolution clear native PAGE (hrCNE) (Wittig et al., 2006, 2007; Wittig and Schägger, 2008; Burré et al.,

To date, native PAGE methods, as the described above, are well-established systems but none of them is usually named a standard method. Especially the combination of non-reducing IEF/NEPHGE with one or the other native PAGE in the second dimension has rarely been performed and is scarcely found in the literature, but has been important for two-dimensional zymograms (Lüthje et al., 2014). After multiple modifications and trials we now developed a protocol which we report in the present paper. It offers good resolution for the combination of nonreducing IEF or NEPHGE in the first dimension with hrCNE in the second dimension. The hrCNE was combined with the phos-tag to separate proteins depending on their phosphorylation. Various activity in-gel stainings can be accomplished in the first dimension and in the second dimension hrCNE or phos-tag hrCNE. For the first time, we attempted to directly link the phosphorylation status of an enzyme to its activity using 2D zymograms by combining several gel electrophoresis methods based on size, charge and affinity.

# Materials and Methods

# Plant Material

Proteins were obtained from leaves of 4 week old corn plants (Zea mays L. cv. Goldener Badischer Landmais, Saatenunion, Hannover, Germany) and roots of 19 day old pea (Pisum sativum L.) plants (Sperli cv. vroege, Lüneburg, Germany). Soluble proteins of corn and pea were separated by differential centrifugation from the microsomal fraction as described elsewhere (Meisrimler et al., 2011; Lüthje et al., 2014) and stored at −76◦C until use. Total protein extracts from corn roots (12 days) were acquired by grinding with liquid nitrogen, followed by extraction in Tris-HCl buffer pH 7.6 (50 mM NaCl, 1 mM DTT, 1% Triton X-100) for 1 h at 4◦C. Extraction was followed by centrifugation at 10,000 g for 10 min (Beckman, Avanti, Germany). All extraction buffers contained protease inhibitors (Sigma Aldrich, France) and phosphatase inhibitors (Sigma Aldrich, France). Protein amounts were quantified as described by Bradford (1976) in the presence of 0.01% Triton X-100 using bovine serum albumin as the standard.

## First Dimension–Non-Reducing IEF and NEPHGE

Similar gels were used for IEF and NEPHGE. Gels consisted of 4.5% acrylamide, 2% ampholytes pH 3–10 (Serva, Heidelberg, Germany), 4 M urea and 2% CHAPS. Gels were always prepared maximum 24 h before use. Minimum polymerization time was 1.5 h at 34◦C. Polymerization was triggered by 0.1% ammoniumpersulfat (APS) and 0.01% N,N,N′ ,N′ -tetramethylethylendiamin (TEMED). Sample buffer was prepared as 4× buffers for both separation methods (IEF, NEPHGE). Samples loaded on the gel contained 1 M urea, 10% glycerol, 0.5% CHAPS and 2% ampholytes. Before samples were applied, a pre-run of the gels was accomplished for 45 min at 30 V with no further restrictions. Electrophoresis conditions were described by Lüthje et al. (2014). For NEPHGE, the polarity and the IEF buffer system was reversed (**Figure 1**). Electrophoresis

reached their pI within the pH gradient of the gel [alkaline (blue) to acidic (red)]. During NEPHGE, proteins moved in the direction of the cathode due to their pI. NEPHGE was stopped before the pH equilibrium was reached to keep proteins with an alkaline pI in the gel. After separation by pI, gel lanes were sliced out, equilibrated

of separation. Low (L) and high (H) pH are labeled on the right of the first dimensions. Three steps are indicated for first and second dimension: (i) loading the sample, (ii) separation of proteins by electrophoresis, and (iii) final position of the proteins after stopping the electrophoresis.

conditions for native NEPHGE were constantly set to 450 Vh, to keep highly alkaline proteins in the gel (step 1: 100 V, 150 Vh; step 2: 250 V, 250 Vh; step 3: 500 V, 200 Vh).

# Second Dimension–hrCNE and Phos-tag hrCNE

Standard hrCNE was casted as a gradient gel of 4–16% as described by Lüthje et al. (2014). The content of the phos-tag hrCNE was similar to the standard hrCNE containing additionally phos-tag (Wako Chemicals GmbH, Neuss, Germany) and Mn(II)Cl2. Different phos-tag concentrations (0.5µM, 1µM, and 10µM) were tested using phosvitin as standard protein. The phos-tag: MnCl<sup>2</sup> ratio was always kept at 1:2 as recommended for phos-tag SDS-PAGE. hrCNE and phos-tag hrCNE were casted as 1 mm continuous gradient gels with an acrylamide concentration of 4–16% (Lüthje et al., 2014). Gel slices from the first dimension non-reducing IEF or NEPHGE were equilibrated in hrCNE equilibration buffer (0.1% Triton X-100, 0.07% DOC, 20% glycerol, 0.001% Ponceau S, 50 mM imidazol, 1 M 6-aminohexanoic acid) for 30 min at room temperature. Cathode buffer (50 mM tricine, 7.5 mM imidazol, 0.05% DOC, 0.05% Triton X-100) and anode buffer (25 mM imidazol-HCl, pH 7.0) were used as described by Wittig et al. (2007). The electrophoresis conditions were 30 min 100 V, 45 min and 10 mA per gel limited to 500 V at 4◦C.

# In-Gel Staining Procedures

Proteins were stained by Coomassie Colloidal Blue (CCB) for total protein content (Neuhoff et al., 1988). Native in-gel staining was done for heme and Cu proteins with 3,3′ ,5,5′ tetramethylbenzidine (TMB) and H2O<sup>2</sup> in Na-acetate buffer, pH 5.0 (Lüthje et al., 2014). For native in-gel staining, gels were documented after 1–5 min of incubation in staining solution. NADH dependent Fe(III)-chelate reductase staining was accomplished using 250µM ferrozine, 125µM NADH, and 250µM Fe(III)-citrate (Holden et al., 1991; Meisrimler et al., 2011). Gels were documented before the staining saturated. 3,3-Diaminobenzidine (DAB) staining was accomplished with 300µM DAB and 100µM H2O<sup>2</sup> in 250 mM Na-acetate buffer, pH 5.0. NADH/NBT staining was accomplished for NADH dependent flavin reductase family proteins in Tris-HCl at pH 7.4 (Meisrimler et al., 2011). Scanning of the gels was done with 600 dpi resolution (Epson Photo scanner V 700, Epson, Germany) and files were saved in TIF format.

Pro-Q Diamond <sup>R</sup> staining for phosphoproteins was accomplished after in-gel activity staining and fixation using the fast staining protocol as recommended by the provider. IEF or NEPHGE gels were fixed in 20% TCA and hrCNE or phos-tag hrCNE were fixed in 40% MeOH and 10% acetic acid overnight. All gels were washed once for 30 min and twice for 10 min in ultrapure water before phospho staining. After destaining gels were washed three times with ultrapure water, followed by detection using a CCD camera at 560 nm (Biorad, Chemdoc, Germany).

At least three independent technical replicates were accomplished per staining in the second dimension to show specificity of the spots in relation to their phosphorylation (Supplemental Datas 1, 2). Students T-test was used to statistically test the protein separation shift between hrCNE and phos-tag hrCNE for significance.

### Protein Digestion and Mass Spectrometry

Gel spots were cut out and proteins reduced with DTT, alkylated with iodoacetamide and digested with trypsin by standard protocol described in Meisrimler et al. (2014). After digestion, the gel pieces were repeatedly extracted (50% acetonitrile/5% formic acid) and the combined extracts dried down in a vacuum concentrator.

For QTOF, Premier tandem MS analysis peptide extracts were dried in a vacuum concentrator and resuspended in 20 mL 0.1% formic acid. The samples were centrifuged at 16,000 rpm and 2–4µL of the digest were used for LC-MS runs which were done on a QTOF Premier tandem mass spectrometer (Waters-Micromass, Eschborn, Germany) equipped with an Aquity UPLC (Waters, Eschborn, Germany). Samples were applied onto a trapping column (Waters nanoAquity UPLC column, C18, 180µm × 20 mm), washed for 10 min with 5% acetonitrile, 0.1% formic acid (5µL/min) and then eluted onto the separation column (Waters nanoAquity UPLC column, C18, 1.7µm BEH130, 75µm × 200 mm, 200 nL/min) with a gradient (A, 0.1% formic acid; B, 0.1% formic acid in acetonitrile, 5–50% B in either 60 or 120 min). The spray was done from a silica emitter with a 10µm tip (PicoTip FS360-20-10, New Objective) at a capillary voltage of 1.5 kV. For data acquisition, the MSE technique was applied: alternating scans (0.95 s, 0.05 s interscan delay) with low (4 eV) and high (ramp from 20 to 35 eV) collision energy was recorded (Silva et al., 2005; Li et al., 2009). The data were evaluated with the software package Protein Lynx Global Server version 2.5.2 (Waters, Eschborn, Germany) searching the Uniprot database and Uniprot tremble (Jan 2014 update). At intervals of 10 s, a lockspray spectrum (1 pmol/µL [Glu1] Fibrinopeptide B (Sigma)) was recorded. Using lockspray correction, a mass accuracy of <7 ppm was achieved in the MS mode.

Orbitrap measurements were performed in an Orbitrap Fusion Tribrid instrument (LC-ESI-OT-MS, Orbitrap Fusion, Thermo Scientific) equipped with a HPLC (Ultimate 3000, Thermo, LC parameter: RP C18 Column (Acclaim PepMap RSLC, Thermo, 75µm × 250 mm, 2µm, 100Å), flow: 0.3µl/min, solvent A: H2O/0.1% formic acid, solvent B: acetonitril/0.1% formic acid, gradient: 2–30% B in 30 min). The Orbitrap was operated with a resolution of 120,000 in positive ion mode. Precursor ions were selected using data dependent acquisition mode (DDA) and fragmented with a normalized HCD (high-energy collision dissociation) energy of 35%. The fragment ions were detected in the linear ion-trap (rapid mode).

The LC-ESI-OT-MS data were processed with Proteome Discoverer v1.4.1.14 (Thermo Scientific) using the following parameters: precursor mass tolerance 10 ppm, fragment mass tolerance 0.2 Da, 1 missed cleavage, carbamidomethylation on Cys as fixed and oxidation on Met and phosphorylation on Ser, The and Tyr as variable modifications. All peptide assignments were verified by manual inspection.

# Results and Discussion

### Two-Dimensional Zymograms

For the separation in the first dimension non-reducing IEF and NEPHGE were accomplished, separating proteins based on their pI. For NEPHGE the pH gradient was directed in the opposite direction (acidic to alkaline) than for IEF (**Figure 1**). Protein separation by NEPHGE was stopped before the pH equilibrium was reached. Therefore, NEPHGE could not be used to calculate the pI of a protein. NEPHGE is normally used for highly alkaline proteins (e.g., membrane proteins) that otherwise would be lost for any analysis by PAGE and the following MS identification. To reach comparability of NEPHGE replicates, the Vh were kept constant between different gel runs (Lopez, 2002). IEF and NEPHGE could be used to separate the same, differently phosphorylated, enzyme, based on their pI shift (Zhu et al., 2005). The shift is introduced by the extra negative charge of the phosphorylation and was also used for separation of phosphoproteins in IPG-strip/SDS-PAGE (Larsen et al., 2001).

The non-reducing IEF sample buffer contained 1 M urea and only CHAPS as detergent, resulting in a clear resolution in the first dimension of soluble proteins and microsomes (**Figure 2**). The pre-run before IEF and NEPHGE increased resolution and activity of the bands. Similar effects have been shown for native Tris-PAGEs in the past (Weydert and Cullen, 2010).

Urea can denature proteins because it diminishes the hydrophobic effect by displacing water in the solvation shell and because it specifically binds to amide units. It has been shown that urea interacts differently with different functional groups, resulting in heterogenic effects on the protein activity. Also, effects of urea have been shown to be reversible, if not used directly in an assay. Therefore, inhibitory concentrations of urea on protein activities strongly rely on the type of protein (Rajagopalan et al., 1961; Kim and Woodward, 1993; Zou et al., 1998; Garfin, 2003; Choi et al., 2004). For more urea (denaturing compounds) sensitive proteins, the optimal urea concentration of gels and sample buffers have to be adjusted based on the level of enzyme activity assayed in the urea-containing enzyme reaction buffer.

followed by Pro-Q Diamond® staining in the same gel. Type of staining is indicated at the bottom and pI of the bands indicated on the left of the gels. Cathode and anode are indicated as (−) and (+), respectively. Arrows show the direction of protein separation.

After separation by pI, gel lanes were sliced out, equilibrated and transferred to the second dimension as described earlier by Lüthje et al. (2014).

One of the most critical points for two-dimensional PAGE was the transfer of the proteins from the first to the second dimension. For the equilibration of the first dimension IEF/NEPHGE, the second dimension hrCNE gel-buffer was supplemented with 0.1% Triton X-100 and 0.07% DOC and gels were equilibrated by continuous shaking at room temperature. This equilibration buffer was applicable for all soluble samples, whereas microsomal fractions showed inferior separation (Supplemental Data 3) due to increased hydrophobicity often observed with membrane protein samples (Meisrimler and Lüthje, 2012). Higher concentrations of detergent had a negative influence on the separation of the proteins and produced irregularities in the separation pattern (data not shown). Sample-dependent adaptions on the presented method are possibly needed for strongly hydrophobic proteins, e.g., testing different detergent combinations, concentrations and solubilization time.

Second dimension standard hrCNE separates proteins based on the size of a protein. In phos-tag hrCNE phosphoproteins were separated by their affinity to the phos-tag under native conditions that has been shown to be highly specific by Kinoshita et al. (2006) and Kinoshita-Kikuta et al. (2007).

In comparison to standard two- dimensional gel electrophoresis (e.g., IPG-strip/SDS-PAGE) the combination of non-reducing IEF/NEPHGE with phos-tag hrCNE excludes the effects of DTT, precipitation and heating. These treatments can affect the activity of a protein and their phosphorylations. In-gel staining like Pro-Q Diamond <sup>R</sup> , all blue and quercetin, most commonly applied after IPG-strip/SDS-PAGE, only show the current form of phosphoproteins in the gels (Orsatti et al., 2009; Wang et al., 2014). In case of phosphorylation loss before staining, the information would be lost for further analysis. Also, multiple post-translational modifications per protein could affect the pI shift in the first dimension and analysis will be difficult. Phostag hrCNE was focused only on phosphoproteins comparable to affinity chromatography, e.g., IMAC (Machida et al., 2007). Other post-translational modifications were excluded as effectors in the second dimension and therefore results were easier to interpret.

Binding abilities and optimal concentrations of the phos-tag in the second dimension hrCNE were tested using phosvitin as a standard for protein phosphorylations (Samaraweera et al., 2011). Alongside, partially dephosphorylated phosvitin was used as a control. First dimension non-reducing IEF confirmed the theoretical pI of 4.5 of phosvitin, showing a pI of 4.4–4.6 for the phosphorylated phosvitin. The partially dephosphorylated protein showed bands with pI of 5.2 and higher (**Figure 2A**). Combinability of the non-reducing IEF with Pro-Q Diamond <sup>R</sup> was first tested with phosvitin and was followed for the combination of native in-gel staining followed by Pro-Q Diamond <sup>R</sup> (**Figure 2**).

In the second dimension phosvitin was observable in the 0.5µM, 1µM, and 10µM phos-tag hrCNE (**Figure 3**). Phosvitin was not visible in 0.1µM, similar to the standard hrCNE without phos-tag or the dephosphorylated protein (**Figure 3**). The concentration of 0.1µM phos-tag was under the limit of the binding ability for phosphoproteins. Overall, best resolution of the phosvitin was achieved in the 0.5 µM phos-tag hrCNE.

The fact that phosvitin was only detectable in its phosphorylated form in the second dimension was caused by the resolution of the hrCNE. hrCNE is normally used for the separation of native protein complexes which have fairly high molecular masses. Proteins with lower molecular masses than 45 kDa were not found after separation in the hrCNE or move very close to the separation front (Lüthje et al., 2014). Based on this fact, the combination of non-reducing IEF/NEPHGE with hrCNE was most useful for proteins with a size above 50 kDa. This fact was one of the major constraints of the presented method. This restriction could possibly be overcome by exchanging the hrCNE to a native Tris-PAGE (Weydert and Cullen, 2010). This has to be further investigated in the future.

The optimal concentration of 0.5–1µM phos-tag used in the presented protocol was found to be in the range reported for first dimension BNE (Deswal et al., 2010). However, the needed phostag concentration is much lower than in phos-tag SDS-PAGE. Deswal et al. (2010) discussed the difference in the needed phostag concentration between first dimension phos-tag BNE and phos-tag SDS-PAGE, speculating that it might be related to the difference in bis-acrylamid to acrylamide ratio used in the two methods. For the presented hrCNE protocol we used a similar bis-acrylamid to acrylamid ratio than used for standard SDS-PAGE, showing that the effect of decreased need of phos-tag was not related to this ratio, but more to the fact that proteins were closer to native conditions. It is highly possible that the treatment

of samples with SDS, reducing agents like DTT and heating in the standard protocol affects the phosphorylation sites or the accessibility of the phosphorylation sites that bind to the phos-tag, similar to the treatment before standard IPG-strip/SDS-PAGE.

Phosvitin was not detectable in non-reducing NEPHGE (450 Vh) optimized for highly alkaline proteins. Therefore, phosvitin cannot be used as standard for the pre-separation by non-reducing NEPHGE in the first dimension. An ideal standard for the separation in the alkaline range has still to be found.

NEPHGE protocols found in the literature normally use higher Vh than in the present protocol (Lopez, 2002). Preliminary work with plant samples showed that strong alkaline bands already moved out of the gel for higher Vh (data not shown). Based on these results, separation in NEPHGE was done constantly at 450 Vh to make replicates comparable (Supplemental Data 4).

### Colorimetric Staining and Identification of Proteins

Functionality of the two dimensional zymograms was tested with soluble proteins from corn leaves, soluble proteins of pea roots and total protein extracts of corn roots (**Figure 4**). The sample

NADH-dependent Fe(III)-reductase staining with ferrozine for soluble proteins (75 µg) of pea roots. (C) DAB/H2O2 staining for total protein extracts (50µg) from corn roots. Separation type for the first dimension (IEF or NEPHGE) is indicated on top and second dimension on the side of the gels. The pH of the IEF (3–10) is indicated on the top of the first dimension. Arrows show the direction of protein separation. Cathode and anode are indicated as (−) and (+), respectively. Numbered spots were analyzed for the shift on phos-tag hrCNE compared to hrCNE and analyzed by LC-MS/MS, numbers of potentially phosphorylated proteins are written in italic letters.

variety showed the independence of the method from the origin and age of a sample.

TMB, DAB, and ferrozine staining were chosen to test the properties of the conventional two dimensional zymograms and the phos-tag zymograms. Before accomplishing staining procedures in the second dimension compatibility with the first dimension non-reducing IEF was tested (**Figure 2**). Different samples were separated on non-reducing IEF and stained with TMB, ferrozine (**Figures 2B**, **4**), DAB (**Figure 4**), and NBT (Supplemental Data 5). Detectable bands in TMB and ferrozine staining are indicated with their corresponding pI (**Figure 2**). All stainings have been published to be specific for different protein groups. TMB/H2O<sup>2</sup> staining has commonly been used for the detection of Fe and Cu containing proteins, e.g., flavocytochromes, peroxidases, blue copper proteins. DAB/H2O<sup>2</sup> has been shown to be specific for oxygen radical producing enzymes, mostly hemecontaining proteins, e.g., peroxidases, but not for Cu-containing enzymes (Lüthje et al., 2014). The Fe(III)-reductase staining with ferrozine and NADH is highly specific for enzymes that are able to reduce Fe(III) to Fe(II) at the given pH using NADH as a co-substrate (Holden et al., 1991; Meisrimler et al., 2011). After reduction, the Fe(II) is bound in a stable complex with ferrozine (Viollier et al., 2000). NBT/NADH staining was also accomplished in the first dimension. This staining was published to be specific for NADH using reductases like quinone reductases (Yan and Forster, 2009; Meisrimler et al., 2011). The formazan salt formed in the reaction with NBT was too stable to be removed from the gel and stainings were not compatible with Pro-Q Diamond <sup>R</sup> staining (Supplemental Data 5).

TMB, DAB and Fe(III)-reductase staining procedures were accomplished in the second dimension with and without phostag, proving the functionality of non-reducing IEF/NEPHGE with the second dimension hrCNE and phos-tag hrCNE as zymograms. However, separation in the second dimension appeared to be the most problematic step. Especially the highly sensitive TMB staining for the relatively extensive group of heme and Cu proteins showed higher backgrounds (**Figure 4B**). The phos-tag hrCNE exhibited the highest background, making it difficult to analyze gels.

Migration of proteins was compared for hrCNE and phostag hrCNE. The phosphorylation of a protein causes a slower migration in phos-tag hrCNE due to their affinity to the phos-tag, leading to a measurable shift between hrCNE and phos-tag hrCNE. However, six spots (9, 13–16) showed no significant shift on the phos-tag hrCNE, when compared to the hrCNE. These proteins had no affinity for the phos-tag and were not phosphorylated. For spot 1 and 2, the shift was not computable due to the high background in the top of the phos-tag hrCNE gel. All other spots (4–8, 10–12) had a significant migration shift of more than 10% of the total migration distance (gel length).

The main spots, showing a significant shift on the phostag hrCNE compared to the hrCNE with a clear appearance on both gels, were picked and identified by LC-MS (**Table 1**). Spots 4–6 were identified as fructose bisphosphate aldolase on both gels. Spot 8 was identified as fruit protein (B4FRC8) and as an uncharacterized protein on the phos-tag hrCNE (**Table 1**; Supplemental Table 1). Based on the small amount of peptides detectable it was not possible to detect specific phosphopeptides in the analyzed spots. The fruit protein was identified also in a former phosphoproteome study available at http://www.ebi.ac.uk/pride/archive/projects/PRD000721 (Bonhomme et al., 2012). Spot 8 was additionally analyzed using LC-MS Orbitrap. Further proteins were significantly identified but not all were related to the TMB staining (Supplemental Table 2).

Phosphorylation sites were verified by in-silico analysis for all proteins identified (**Table 1**). Over all, MS based identification after zymograms is often the biggest challenge. The low abundance of proteins stained in zymograms is based on the high sensitivity of these staining methods [e.g., Fe(III)-reductase or


Protein spots picked from the second dimension hrCNE and phos-tag hrCNE (Figure 4) were digested by trypsin and analyzed by MS. Protein names listed in UniProt are used in the table. No., spot number; ID, accession number in UniProt (http://www.uniprot.org/); Shift, difference in the separation distance of the spot on phos-tag hrCNE and hrCNE in % (?, if unclear). Significant changes (student's T-test) were indicated by \* ≤ 0.01 and \*\* ≤ 0.001 for n = 3 technical replicates; P, phosphorylated form detected (+, positive; −, negative); Pept, number of identified peptides—manual sequencing was marked with an asterisk; pI, theoretical isoelectrical point; MW, theoretical molecular weight in kD; NetPhos, theoretical phosphorylation sites calculated with NetPhos 2.0 (http://www.cbs.dtu.dk/services/NetPhos/); Literature, available literature on phosphorylations of the identified protein (if not available, n.a.).

TMB staining] which is often higher than for silver staining. If primary MS results enable good protein identification, phosphopeptide enrichment is recommendable in a second MS analysis to verify the results from the phos-tag zymograms (Dunn et al., 2010). In contrast to the TMB staining, specific protein activities like the NADH-dependent Fe(III) reduction and the DAB staining led to a clear separation of proteins (**Figure 4**) but were more problematic for protein identification. The protocol for non-reducing IEF or NEPHGE/hrCNE presented is also the first functional protocol for Fe(III)-reductase detection in the second dimension. This staining was only published for the first dimension IEF to date (Holden et al., 1991; Meisrimler et al., 2011). Fe(III)-reductase activity was really sensitive and samples had to be treated carefully, avoiding multiple freeze thawing cycles. For spot 11 on the ferrozine stained phos-tag hrCNE only the peptide ISEYVTQLR was identified for ferritin, which is possibly regulated by phosphorylation under different physiological conditions (Beazley et al., 2009). Ferritin has also a Feoxidoreductase function, therefore it is potentially stainable by the ferrozine method. Overall, spots 10–12 were only detectable in the phos-tag hrCNE but not in the hrCNE. Therefore, the calculated shift of these spots was 100%. Detected proteins might have been of small size and migrated close to the front in the hrCNE. In phos-tag hrCNE, they have a strong affinity to the phos-tag and migrate slower. To understand the migration of the ferrozine spots further investigations are needed.

Interestingly, ferrozine activity detected in the soluble fraction was exclusively detectable in the alkaline pH using NEPHGE (Supplemental Data 4), whereas in microsomal fractions it was only detectable at more acidic pH with the pI 5.6, 6.7, and 7.2 (**Figure 2**). The band with a pI of 5.6 was identified as quinone reductase family protein NP\_194457 with the peptide AFLDATGGLWR (sequence coverage 5%, score of 26) by manual sequencing.

Spot 9 was identified as 6,7-dimethyl 8-ribityllumazine synthase with two peptides (FNEIITRPLLEGAVATFK and GAEAALTAIEMASLFEHHLK) that has no activity correlated to the Fe-reduction stained in the gel. In both cases, spot 5 and 6, found peptides were verified manually but final scores were to low for significant identification. Overall, multiple proteins per spot can be a problem for the identification of low abundant proteins responsible for an activity detected in zymograms and MS data have to be handled critically.

Spot 15 and 16 were identified on the DAB stained hrCNE as peroxidases. For all the DAB stained spots no significant shift appeared when the sample was separated by phos-tag hrCNE compared to the hrCNE. The identified peroxidases were part of the class III peroxidases, which have not been shown to be the aim of phosphorylation events. Especially, peroxidases of the excretory pathway seem not to be regulated by phosphorylation possibly due to the lack of excreted kinases and phosphatases.

If a specific protein with known activity has to be analyzed proteins can be pre-separated by chromatography (e.g., affinity, ion exchange). Phosphoprotein enrichment is another option and different variations of the technique are available (different IMACS, phosphoprotein enrichment kits). For plant samples IMAC was successfully applied (Tang et al., 2008), but the protocol might need adaption, as application was not relying on activity preservation.

Furthermore, non-reducing IEF/NEPHGE were stained with Pro-Q Diamond <sup>R</sup> directly after in-gel activity for ferrozine, TMB, and NBT, resulting in a low amount of phosphoproteins detected in the IEF. No phosphoproteins were detected in NEPHGE, which is possibly due to the high alkaline pH (ampholytes) (**Figure 2**). Pro-Q Diamond <sup>R</sup> was also applied in the second dimension after in-gel staining (ferrozine). A few spots were detectable in the standard hrCNE but in phostag hrCNE no signal could be found at all (Supplemental Data 3). The reason for the incompatibility is not clear but the phos-tag possibly blocks the phosphorylation site for the staining.

In any case, to identify detected proteins, spots should be picked and analyzed by MS and/or by Western blot. Other specific staining methods are available for different enzyme activities (e.g., malate dehydrogenase, lipoxygenase, superoxide dismutase and others) (Manchenko, 2002). Combination of native two-dimensional gel electrophoresis with the phos-tag technology and the use of specific activity stainings has the benefit that changes of these activities by phosphorylation can be directly monitored. Applications of the method can be the observation of specific activities by phosphorylation under differential stress conditions (Supplemental Data 6). The method itself should not be used as a stand-alone technique but together with Western blot, MS and specific point mutation of phosphorylation sites it can be used for a dynamic analysis of reactions to stress factors.

# Concluding Remarks

In the past years, various MS based approaches have been developed to identify phosphorylated peptides and proteins. In several techniques, phos-tag related molecules were used for the enrichment of phosphorylated peptides. In contrast to MS methods, phos-tag gels can easily be performed using general gel electrophoresis equipment and radioactivity is avoided. Furthermore, all phosphorylations can be detected. Different phosphorylated forms of the same protein can be distinguished. The combination of phos-tag with zymograms allows estimation of the effects of phosphorylation on protein activity. This allows following activation of proteins by phosphorylation and dephosphorylation. The combination with native IEF for low alkaline to acidic proteins and NEPHGE for highly alkaline proteins is helpful to separate proteins by pI, resulting in a higher resolution of different iso-enzymes. Phos-tag gels were not compatible with Pro-Q Diamond <sup>R</sup> . Protein identification is possible by MS and results can be confirmed by Western blot. In some cases phosphoprotein enrichment by IMAC or alternative might be needed before phos-tag zymograms to get better identifications by MS.

# Acknowledgments

The authors would like to thank Margret Vielhaben (University of Hamburg, Germany) for technical assistance in cultivation of plants, PD Dr. Friedrich Buck (UKE Hamburg, Germany) for the Mass spectrometry analysis, PD Dr. Harthwig Lüthen and MSc Jenny Köppe for text corrections. This work was supported by the excellence initiative of the University of Hamburg (Postdoc grant to C.N.M.) and DFG (Lu 668/4-4). The authors have declared no conflict of interest.

# References


# Supplementary Material

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fpls.2015. 00230/abstract


**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.

Copyright © 2015 Meisrimler, Schwendke and Lüthje. 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.

# Using 7 cm immobilized pH gradient strips to determine levels of clinically relevant proteins in wheat grain extracts

*Sona Fekecsová1,2, Maksym Danchenko3, Lubica Uvackova1, Ludovit Skultety3 and Martin Hajduch1,3\**

*<sup>1</sup> Department of Developmental and Reproduction Biology, Institute of Plant Genetics and Biotechnology, Slovak Academy of Sciences, Nitra, Slovakia, <sup>2</sup> Faculty of Natural Sciences, Comenius University, Bratislava, Slovakia, <sup>3</sup> Institute of Virology, Slovak Academy of Sciences, Bratislava, Slovakia*

### *Edited by:*

*Jesus V. Jorrin Novo, University of Cordoba, Spain*

### *Reviewed by:*

*Benjamin Schwessinger, University of California, Davis, USA Loïc Rajjou, AgroParisTech – Paris Institute of Technology for Life, Food and Environmental Sciences, France*

### *\*Correspondence:*

*Martin Hajduch, Department of Developmental and Reproduction Biology, Institute of Plant Genetics and Biotechnology, Slovak Academy of Sciences, Akademicka 2, P.O. Box 39A, Nitra, Slovakia hajduch@savba.sk*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 26 February 2015 Accepted: 27 May 2015 Published: 12 June 2015*

### *Citation:*

*Fekecsová S, Danchenko M, Uvackova L, Skultety L and Hajduch M (2015) Using 7 cm immobilized pH gradient strips to determine levels of clinically relevant proteins in wheat grain extracts. Front. Plant Sci. 6:433. doi: 10.3389/fpls.2015.00433* The aim of the work was to test a relatively simple proteomics approach based on phenol extraction and two-dimensional gel electrophoresis (2-DE) with 7 cm immobilized pH gradient strips for the determination of clinically relevant proteins in wheat grain. Using this approach, 157 2-DE spots were quantified in biological triplicate, out of which 55 were identified by matrix-assisted laser desorption/ionization – time of flight tandem mass spectrometry. Clinically relevant proteins associated with celiac disease, wheat dependent exercise induced anaphylaxis, baker's asthma, and food allergy, were detected in 24 2-DE spots. However, alcohol-soluble gliadins were not detected with this approach. The comparison with a recent quantitative study suggested that gelbased and gel-free proteomics approaches are complementary for the detection and quantification of clinically relevant proteins in wheat grain.

Keywords: *Triticum aestivum*, gel-based, quantification, MALDI-TOF/TOF, 2-DE, 7 cm IPG, grain, allergen

# Introduction

The main component of the wheat grain are storage proteins with gluten as the major part representing as much as 80% of total protein content (Ferranti et al., 2007). Gluten is a mixture of gliadins and glutenins that differ in their electrophoretic mobility (Payne et al., 1985; Jacobsen et al., 2007). Gluten is also the main allergen in the wheat grain and is responsible for nutritive intolerances such as celiac disease or gluten-sensitive enteropathy (Rubio-Tapia et al., 2009), and various allergies (Battais et al., 2008; Sapone et al., 2012; Mauro Martin et al., 2014). In addition to storage proteins, wheat grain allergens include enzymatic and structural proteins such as prolamins, cupins, and Bet v1 protein family (Breiteneder and Mills, 2005). Out of these, prolamins are dominant and include α-amylase and protease inhibitors, 2S albumins, and non-specific lipid transfer proteins (nsLTPs; Breiteneder and Radauer, 2004; Mills et al., 2004).

Protein two-dimensional gel electrophoresis (2-DE) has been extensively used to characterize wheat grain proteins. For instance, 2-DE followed with immunoblotting and tandem mass spectrometry (MS/MS) resulted into the identification of nine subunits of low molecular weight (LMW) glutenins, serpin, α-amylase inhibitor, and α-gliadin in wheat flour (Akagawa et al., 2007). The combination of 2-DE and MS/MS identified several allergenic proteins, such as serpins, in dough liquor of four wheat cultivars under abiotic stress (Sancho et al., 2008). Similar approaches detected heat responsible allergenic proteins, such as α-amylase inhibitors or serpins, in the endosperm of developing wheat grains under heat stress (Hurkman et al., 2009). Additionally, several allergenic proteins were detected in Korean sprouting wheat cultivars using 2-DE and matrix assisted laser desorption/ionization-tandem Time of Flight (MALDI-TOF) MS/MS (Kamal et al., 2009). Importantly, 20 allergenic proteins in wheat grains were detected using proteomics approach based on 2-DE in combination with 17 cm immobilized pH gradient (IPG) strips and MS/MS (Yang et al., 2011). Similarly, 2-DE in combination with isoelectric focusing (IEF) capillary tube gels, three different proteases, and MS/MS resulted in the detection of 476 2-DE spots out of which 233 were identified, including well-known allergens (Dupont et al., 2011). The 2-DE was also used to analyze wheat with genetically altered omega-5 gliadin content (Altenbach et al., 2014). Interestingly, this study showed that unique genetic transformation events with the same RNA interference construct may have differential effects on the wheat grain proteome (Altenbach et al., 2014). This study highlights the importance of proteomic analyses in the study of genetic transformations (Altenbach et al., 2014).

The above studies showed that a 2-DE approach is effective in the characterization of wheat grain proteins. However, 2-DE can be labor, resources, and time consuming, especially when long IPG strips are used. The aim of this study was to test a relatively simple 2-DE approach based on 7 cm IPG strips for the detection of clinically relevant proteins of wheat grain.

# Materials and Methods

## Plant Material and Protein Extraction

Seeds of winter wheat cultivar Viginta were obtained from SELEKT LtD, Bucany, Slovak Republic. Proteins were extracted ˇ in biological triplicate from 500 mg of dry seeds. Seeds were ground in liquid nitrogen and proteins were extracted with phenol-based extraction media [50% (v/v) phenol, 0.45 M sucrose, 5 mM EDTA, 0.2% (v/v) 2-mercaptoethanol, 50 mM Tris–HCl, pH 8.8]. Sample was stirred and homogenized for 30 min at 4◦C. The phenol phase was removed after centrifugation at 5000 × *g* for 10 min at 4◦C. Proteins were precipitated from the phenol phase by the addition of five volumes of ice-cold 0.1 M ammonium acetate in 100% methanol, and incubated at −20◦C overnight. The protein pellet was extensively washed twice using 0.1 M ice cold ammonium acetate in 100% methanol, followed by 80% ice cold acetone, and finally with 70% ice cold ethanol and precipitates were collected by centrifugation for 15 min., 5000 × *g* at 4◦C. Total protein concentration was determined using the Bradford (1976) assay with Bovine Serum Albumin as the standard.

### Two-Dimensional Gel Electrophoresis

Samples (50 μg protein) were diluted in 100 μl of IEF buffer [8 M urea, 2 M thiourea, 2% (w/v) CHAPS, 2% (v/v) Triton X-100, 50 mM dithiothreitol], 3 μl of ampholytes were added, and loaded onto 7 cm IPG strips of pH 3–10 (ReadyStripTM IPG Strips BioRad) for IEF. Isoelectric focusing was carried out using Protean IEF Cell (Bio-Rad) with the following conditions: 150 V for 150 VH, 500 V for 500 VH, and 4000 V for 15,000 VH including initial active rehydration for 12 h at 50 V. For the second dimension (SDS-PAGE), IPG strips were incubated in SDS equilibration buffer [1.5 M Tris-HCl pH 6.8, 6 M urea, 30% (v/v) glycerol, 5% (w/v) SDS) for 15 min with 2% (w/v) dithiothreitol] followed by a second equilibration step of 15 min with the equilibration buffer containing 2.5% (w/v) iodoacetamide. The equilibrated strips were loaded on the top of 10% polyacrylamide gel and the electrophoresis was run at 80 V until the dye reached the bottom of the gel. Gels were stained for 16 h with Coomassie Brilliant Blue G-250 at room temperature. The 2-DE gels were digitalized using a GS-800 Calibrated Densitometer (Bio-Rad) at 300 dpi and 16 bit grayscale. Digitalized gels were analyzed with PDQuest 8.0 software (Bio-Rad).

### Protein Digestion and Mass Spectrometry

Excised 2-DE plugs were washed with 300 μl destaining solution (50% acetonitrile in 50 mM ammonium bicarbonate) and dehydrated in 100% acetonitrile. After removal of acetonitrile, 2- DE spots were rehydrated with trypsin (Promega) and digested at 37◦C overnight. The digestion was stopped with formic acid and extracted tryptic peptides were stored at −80◦C until MS/MS analysis with a TOF/TOF mass spectrometer in combination with MALDI using an ultrafleXtreme instrument equipped with a 355 nm smartbeam-2 laser, capable of pulsing frequency 1 kHz (Bruker). Peptides were concentrated to 20 μl using Concentrator plus (Eppendorf). After that, concentrated peptides were desalted by μ-C18 ZipTips (Merck Millipore). Next, 1 μl of purified digests were spotted onto 800 μm AnchorChip MALDI target (Bruker) and α-cyano-4-hydroxycinnamic acid (CHCA) matrix (0.7 mg·ml−<sup>1</sup> in 85% acetonitrile, 0.1% trifluoroacetic acid, 1 mM ammonium phosphate) was added.

The mass spectrometer was operated by flexControl 3.3 software (Bruker). For every position 4000 shots were summed in positive reflector mode in the range of 700–3500 mass to charge (m/z). Following that, up to 25 of the most intense precursor peaks per sample were selected for the MS/MS analysis with the minimal signal to noise (S/N) ratio set to 15. Abundant trypsin and keratin peaks were specified in the exclusion list. Fragmentation spectra were acquired by accumulation of 3000 laser shots in positive reflector LIFT mode. Fragmentation was achieved by laser induced dissociation (LID) mechanism by 50% increase in laser power, without the introduction of a collision gas. Simultaneously detector voltage was boosted by 80%.

### Processing of MS/MS Data

Acquired spectra were processed by flexAnalysis 3.3 software (Bruker). A sophisticated numerical annotation procedure (SNAP) algorithm was used for peak picking to calculate exact monoisotopic masses. For the precursor spectra the S/N threshold was set to 10 and the resulting spectra were externally recalibrated against data from an adjacent spot containing nine peptides of the Peptide calibration standard 2 (Bruker). For the fragment spectra S/N threshold was set to 5, also baseline subtraction (TopHat algorithm), and smoothing (Savitzky-Golay algorithm 3 cycles with 0.15 m/z width) were applied.

The MS/MS peak lists were imported into the ProteinScape 2.1 proteomic data management software (Bruker). Peptide identification was performed by an in-house Mascot 2.3 server (Matrix Science), querying against the non-redundant Triticeae plant protein UniProt database downloaded on April, 2014 (100 981 entries). Additionally, protein assignments were verified by searches against the SwissProt database from June 2014 (545 388 sequences) that included major contaminants such as trypsin or keratin. Search parameters were the following: fixed cysteine carbamidomethylation, variable methionine oxidation, one missed trypsin cleavage site, 40 ppm precursor mass tolerance, 0.5 Da fragment mass tolerance. Protein identifications were accepted if at least two different matched peptides had ion score higher than 30, meaning *p <* 0.05.

For allergenicity assessments, identified proteins were queried against the Allergome database1 containing 2994 allergen entries, using Allergome Aligner module with an embedded NCBI blastp v.2.2.18 algorithm. Only hits with 100% sequence identity were accepted as clinically relevant allergens. The MS/MS proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD002067.

# Results

The 2-DE-based proteomics approach in combination with 7 cm IPG strips (**Figure 1**) quantified 157 2-DE spots in biological triplicate (Supplementary Figure S1; Supplementary Table S1) out of which 55 were identified (**Table 1**). Identified proteins were classified according to previous apporaches (Bevan et al., 1998) into five functional classes (**Figure 2**). The most abundant class was 37 proteins associated with destination and storage, followed by nine proteins associated with metabolism and five energy proteins (**Figure 2**). This study also detected two signaling proteins and two proteins associated with disease/defense (**Table 1**). All identified proteins were assigned on the 2-DE gel (Supplementary Figure S2). The most abundant protein on this reference map is the high molecular weight (HMW) glutenin subunit (GS) with a relative volume (%V) of 8.8 (spot number 2909) followed by 11-S seed storage domain containing protein (3302) with %V of 5.6 (**Table 1**).

### As Much as 45% of Identified Proteins were Associated with Various Allergies or Food Intolerances

To determine clinical relevance of the identified proteins, sequences were queried against the Allergome database1 which contains 2994 allergen entries (Supplementary Table S2). This approach detected clinically relevant proteins in 24 2-DE spots which represented 13 non-redundant accession numbers (**Table 1**). Out of these, nine 2-DE spots were identified as HMW GS, seven as serpins, three as α-amylase inhibitors, two as LMW GS, two as glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and one as Cys peroxiredoxin (PER1; **Table 1**). All

these proteins were assigned on the 2-DE gel in order to establish the reference map of clinically relevant proteins of wheat grain (**Figure 3**). All 24 2-DE spots presented on this reference map were color-coded based on protein identification to visualize regions of the 2-DE gel with a prevalence of clinically relevant proteins (**Figure 3**). The most abundant protein is HMW GS (2-DE spot number 2909) followed LMW GS (7211; **Table 1**). The 2-DE spots identified as serpin (2202), GAPDH (3216), GAPDH, and α-amylase inhibitor (4009) (3203) showed the lowest abundance on this reference map (**Figure 3**; **Table 1**). To reveal overall abundances of the detected clinically relevant proteins, relative volumes of each 2-DE spot were summed based on protein identifications. Using this approach, the summed (total) relative volumes for each of eight detected clinically relevant proteins was established (**Figure 4**). It was revealed that HMW GS, LMW GS, and serpins are highly abundant in wheat grain (**Figure 4**).

contained clinically relevant proteins. These proteins were assigned onto a

# Discussion

2-DE reference map.

The aim of this work was to test a 2-DE proteomics approach with 7 cm IPG strips and phenol-based protein extraction for the detection of clinically relevant proteins in wheat grain. Classical methods for protein isolation from wheat grain are based on isopropanol extraction (van den Broeck et al., 2009). We successfully

<sup>1</sup>http://www*.*allergome*.*org/




implemented this method and have previously determined quantities of wheat grain proteins using gel-free proteomics approach (Uvackova et al., 2013a,b). In the present study we tested the phenol-based extraction protocol (Hurkman and Tanaka, 1986), which also solubilizes membrane proteins often excluded from alcohol-based protein extractions. Previously, our group efficiently used this protocol for the characterization of seed proteins in soybean (Hajduch et al., 2005; Danchenko et al., 2009; Klubicova et al., 2012), canola (Hajduch et al., 2006), castor (Houston et al., 2009), *Arabidopsis* (Hajduch et al., 2010), and flax (Klubicova et al., 2010, 2013).

In the present study we detected nine 2-DE spots as HMW GS (**Table 1**; **Figure 3**), which influence the viscoelastic properties of wheat flour (Masci et al., 1998), and may cause wheat dependent exercise-induced anaphylaxis (WDEIA) when digested (Hofmann et al., 2012). The present study was particularly successful in the detection of wheat grain allergens associated with Baker's asthma (Salcedo et al., 2011; Olivieri et al., 2013). Five 2-DE spots were identified as serpin (**Table 1**; **Figure 3**), which are involved in food allergy and Baker's asthma (Salcedo et al., 2011; Mameri et al., 2012). Three 2-DE spots were identified as an α-amylase inhibitor, important contributors to Baker's asthma (Tatham and Shewry, 2008; Salcedo et al., 2011),

food allergies (James et al., 1997), and WDEIA (Hofmann et al., 2012). Additionally, one 2-DE spot was detected as PER1 which is a confirmed wheat allergen likely associated with Baker's asthma (Pahr et al., 2012). However, this study did not detect the 27 kDa albumin, which was shown to be associated with Baker's asthma (Weiss et al., 1993) or the alcohol-soluble gliadin proteins involved in celiac disease (Wieser, 1996; Allred and Ritter, 2010).

The majority of wheat grain allergenic proteins detected in the present study were not quantified in our recent MS-based study (Uvackova et al., 2013b). This finding is in agreement with a recent investigation of soybean under flooding stress, where only 9 out of 115 proteins were detected by both gel-based and gel-free proteomics approaches in root tips (Yin et al., 2014). Similar results have been shown in the analysis of the honey bees hemolymph proteome, where only 27% of proteins were detected with both approaches (Bogaerts et al., 2009).

Based on this, it is tempting to speculate that gel-based and gel-free approaches are complementary for the detection and quantification of wheat grain allergenic proteins. However, the complementarity of gel-based and gel-free proteomics approaches was suggested previously (Luque-Garcia et al., 2011; Abdallah et al., 2012). The combination of gel-based and gel-free proteomics was shown to be effective for the analyses of soybean under flooding (Yin et al., 2014), phytopathogenic fungus *Botrytis cinerea* (Gonzalez-Fernandez et al., 2013), *Nicotiana tabacum* trichomes (Van Cutsem et al., 2011), the honeybee hemolymph proteome (Bogaerts et al., 2009), or during soybean seed filling (Agrawal et al., 2008).

# References


# Conclusion

This study has demonstrated that phenol-based protein extraction in combination with 2-DE and 7 cm IPG strips is capable of determining clinically relevant proteins in wheat grain extracts. However, important clinically relevant proteins, such as alcohol-soluble gliadins were not detected with this approach. The comparison of these data with previous work suggests that gel-based and gelfree proteomics are complementary approaches for the determination of clinically relevant proteins in wheat grain extracts.

# Acknowledgments

This research was supported by the Scientific Grant Agency of the Ministry of Education of the Slovak Republic and Slovak Academy of Sciences (VEGA-2/0016/14) and European Community under project no 26220220180: Building Research Centre "AgroBioTech."

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal*.*frontiersin*.*org/article/10*.*3389/fpls*.* 2015*.*00433/abstract


patients suffering from wheat-induced respiratory allergy. *Clin. Exp. Allergy* 42, 597–609. doi: 10.1111/j.1365-2222.2012.03961.x


**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.

*Copyright © 2015 Fekecsová, Danchenko, Uvackova, Skultety and Hajduch. 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.*

# Time to dig deep into the plant proteome: a hunt for low-abundance proteins

### *Ravi Gupta1, Yiming Wang2, Ganesh K. Agrawal 3,4, Randeep Rakwal 3,4,5,6, Ick H. Jo7, Kyong H. Bang7 and Sun T. Kim1 \**

*<sup>1</sup> Plant Functional Genomics Laboratory, Department of Plant Bioscience, Life and Industry Convergence Research Institute, Pusan National University, Miryang, South Korea*

*<sup>2</sup> Plant Proteomics Group, Max Planck Institute for Plant Breeding Research, Cologne, Germany*

*<sup>3</sup> Research Laboratory for Biotechnology and Biochemistry, Kathmandu, Nepal*

*<sup>4</sup> Global Research Arch for Developing Education (GRADE) Academy Pvt. Ltd, Birgunj, Nepal*

*<sup>5</sup> Organization for Educational Initiatives, University of Tsukuba, Tsukuba, Japan*

*<sup>6</sup> Department of Anatomy I, Showa University School of Medicine, Tokyo, Japan*

*<sup>7</sup> Department of Herbal Crop Research, Rural Development Administration, Eumseong, South Korea*

*\*Correspondence: stkim71@pusan.ac.kr*

### *Edited by:*

*Joshua L. Heazlewood, The University of Melbourne, Australia*

### *Reviewed by:*

*Holger Eubel, Leibniz Universität Hannover, Germany Christian Lindermayr, Helmholtz Zentrum München - German Research Center for Environmental Health, Germany*

**Keywords: low-abundance proteins, high-abundance proteins, two-dimensional gel electrophoresis, RuBisCO, post-translational modifications**

### **INTRODUCTION**

Two-dimensional gel electrophoresis (2- DGE) has come a long way since its introduction around 40 years by the pioneering work of these three researchers (Klose, 1975; O'Farrell, 1975; Scheele, 1975). 2-DGE was one of the major breakthroughs in proteomics, enabling researchers to detect, analyze and identify the whole set of proteins of a cell or tissue, simultaneously. With the advancement in technology, some modifications to this technique like development of immobilized pH gradient (IPG) strips were introduced, which undoubtedly, made this technique more simple, rapid and autonomous (Bjellqvist et al., 1982). After its introduction to the present, 2-DGE has been the method of choice for analyzing the complex proteomes of plants. 2-DGE has been used extensively to investigate the effects of biotic and abiotic stress, role of hormones, and developmental changes of plants, among others (Agrawal and Rakwal, 2008). However, it was slowly realized that identification of the plant proteins led to the repeated detection of high-abundance proteins (HAPs) including ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) and other housekeeping proteins, which are present at 106-105 order of magnitude (Gygi et al., 2000; Patterson and Aebersold, 2003; Görg et al., 2004). Signaling and other regulatory proteins are generally present 100 molecules per cells. Subsequently, these proteins are difficult to identify by either gel-based or gel-free proteomic approaches, even with access to the latest mass spectrometers. RuBisCO comprises of a large percentage in the total proteins and thus hinders the absorption of low-abundance proteins (LAPs) on the IPG strips, which subsequently results in the poor detection and identification of LAPs on 2D gels and by mass spectrometry (MS), respectively. Therefore, the time has come for all plant proteomers to realize the need to hunt for the LAPs, moving one step ahead from the present. As RuBisCO is the major HAP in plant leaves, here we recommend the incorporation of a RuBisCO depletion/removal method in every plant protein extraction step to look deeper into the plant proteome. RuBisCO depletion will definitely improve the proteome coverage and will lead to the detection of novel unidentified LAPs.

### **OVERVIEW OF THE METHODS DEVELOPED FOR ENRICHMENT OF LAPs: MERITS AND DE-MERITS RuBisCO DEPLETION METHODS**

A number of protein extraction methods with numerous depletion steps were developed in the last decade to remove RuBisCO, which accounts for nearly half of the total leaf protein content (summarized in **Figure 1**). Kim and co-workers initially developed a poly-ethylene glycol (PEG)-based method for the depletion of RuBisCO from rice leaves (Kim et al., 2001). Application of 20% PEG significantly precipitated the RuBisCO protein (large and small subunits) in the pellet fraction, resulting in the enrichment of LAPs in the supernatant fraction. Later, it was shown that 16% PEG was sufficient to deplete the RuBisCO from Arabidopsis leaves (Xi et al., 2006) indicating this method can be applied to all plants for the removal of RuBisCO. However, this PEG method is laborious and time consuming. Following the PEG fractionation approach, a new method using calcium and phytate was introduced for the removal of RuBisCO from leaves of soybean. Results revealed that a 10 min incubation of the leaf extract with 10 mM calcium and 10 mM phytate at 42◦C, depleted 86% of the RuBisCO protein in the pellet fraction (Krishnan and Natarajan, 2009). As incubation of the protein extract at 42◦C is absolutely essential for significant depletion of RuBisCO, this temperature condition can lead to the denaturation of some heat labile proteins. Incubation at lower temperatures significantly reduces

the RuBisCO precipitation ability of this method. For example, only 44% RuBisCO depletion was achieved at 4◦C (Krishnan and Natarajan, 2009). More recently, a protamine sulfate-based specific RuBisCO depletion method was introduced (Kim et al., 2013). It was shown that addition of 0.1% protamine sulfate differentially precipitates the RuBisCO in the pellet fraction and enriches the LAPs in the supernatant fraction. Using Western blotting, no RuBisCO was detected in the supernatant fraction, suggesting this method is able to deplete RuBisCO below the detection limit. 2-DGE analysis showed that application of this method in soybean resulted in visualization of 423 new spots in the supernatant fraction which were not discernible in the total fraction. Furthermore, in addition to soybean, this method was also applicable to another dicot Arabidopsis, and monocots rice and maize, suggesting that it can be universally applied in plants for the removal of RuBisCO. This protamine sulfate-based method is rapid, reliable, cost effective, and highly efficient and is more specific than the previously published PEG and calcium-phytate based methods (Kim et al., 2013).

### **IMMUNO-AFFINITY BASED METHODS**

Other than the above mentioned precipitation methods, an affinity-based method has been developed for RuBisCO depletion. This method utilizes the anti-RuBisCO antibodies, which are commercially available and supplied as columns (IgY RuBisCO column, Sigma Aldrich; Cellar et al., 2008). The beauty of this method is that it is highly specific to RuBisCO. However, this method is very expensive limiting its wide acceptance among the scientific community, especially laboratories in the developing countries.

In addition to the RuBisCO-depletion methods, other techniques have also been introduced for the enrichment of LAPs. "Combinatorial peptide ligand library" (CPLL) technology, developed over the years by the group of P.G. Righetti, is commercially available under the trade name of Proteominer (BioRad) (Boschetti and Righetti, 2013). Briefly, CPLLs consists of several million hexapeptides (prepared using 16 different amino acids) that are able to recognize complementary amino acid sequence in a bait protein harvesting it from the sample matrix. To put is simply, when the protein extract is loaded onto a CPLL column under large overloading conditions, beads having affinity with the abundant proteins saturate first and therefore, the major fractions of these proteins are washed out due to limited binding capacity of the beads. However, due to low concentrations of LAPs, these proteins keep on binding with their partner beads when additional protein extract is loaded to it and thus get enriched (Boschetti and Righetti, 2013). This method is not based on the specific removal of RuBisCO and removes all the general HAPs from the plant extract.

### **RuBisCO DEPLETION AND POST-TRANSLATIONAL MODIFICATION ANALYSIS**

In addition to the detection of LAPs under normal conditions, depletion of RuBisCO can also be fruitful for post-translational modification (PTM) analysis. RuBisCO being phosphorylated and nitrosylated, hinders the detection of PTMs of LAPs. Recently, it was shown that removal of RuBisCO from *Brassica juncea* leaves significantly improves the detection of novel nitrosylated LAPs (Sehrawat et al., 2013). Similarly, the plant phosphoproteome coverage can also be increased by incorporating the RuBisCO depletion step during the protein extraction step, as indicated in soybean. Application of calcium phytate during protein extraction in soybean led to the identification of 28 new phosphorylated proteins which were previously undetectable, suggesting the application of RuBisCO depletion methods in PTMs discovery as well (Krishnan and Natarajan, 2009).

### **CONCLUSIONS**

The methods (precipitation- and affinitybased) discussed here can deplete the RuBisCO protein, a major HAP in plants. These methods are schematically presented in **Figure 1**. We recommend the incorporation of RuBisCO depletion step during the sample preparation for proteomics analyses given the fact that RuBisCO depletion enriches the LAPs/rare proteins. Identification of these proteins will enrich our knowledge on plant biology.

### **ACKNOWLEDGMENTS**

This work was supported by grants from National Agenda Programs for Agricultural R&D (FTA, grant#: PJ01010401), Rural Development Administration (RDA), Republic of Korea.

### **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: 05 December 2014; accepted: 12 January 2015; published online: 30 January 2015.*

*Citation: Gupta R, Wang Y, Agrawal GK, Rakwal R, Jo IH, Bang KH and Kim ST (2015) Time to dig deep into the plant proteome: a hunt for low-abundance proteins. Front. Plant Sci. 6:22. doi: 10.3389/fpls.2015.00022*

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

*Copyright © 2015 Gupta, Wang, Agrawal, Rakwal, Jo, Bang and Kim. 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.*

# **Multiplex staining of 2-DE gels for an initial phosphoproteome analysis of germinating seeds and early grown seedlings from a non-orthodox specie:** *Quercus ilex* **L. subsp.** *ballota* **[Desf.] Samp.**

*M. Cristina Romero-Rodríguez 1, 2, 3, Nieves Abril 1, Rosa Sánchez-Lucas 1, 2 and Jesús V. Jorrín-Novo1, 2\**

### *Edited by:*

Nicolas L. Taylor, The University of Western Australia, Australia

### *Reviewed by:*

Martin Hajduch, Slovak Academy of Sciences, Slovakia Deyou Qiu, Chinese Academy of Forestry, China

### *\*Correspondence:*

Jesús V. Jorrín-Novo, Department of Biochemistry and Molecular Biology, University of Cordoba, Campus de Rabanales, Ed. Severo Ochoa, Planta Baja, 14071 Cordoba, Spain bf1jonoj@uco.es

### *Specialty section:*

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

*Received:* 07 April 2015 *Accepted:* 27 July 2015 *Published:* 11 August 2015

### *Citation:*

Romero-Rodríguez MC, Abril N, Sánchez-Lucas R and Jorrín-Novo JV (2015) Multiplex staining of 2-DE gels for an initial phosphoproteome analysis of germinating seeds and early grown seedlings from a non-orthodox specie: Quercus ilex L. subsp. ballota [Desf.] Samp. Front. Plant Sci. 6:620. doi: 10.3389/fpls.2015.00620 <sup>1</sup> Department of Biochemistry and Molecular Biology, University of Cordoba, Cordoba, Spain, <sup>2</sup> Agricultural and Plant Proteomics Research Group, Department of Biochemistry and Molecular Biology, Escuela Técnica Superior de Ingenieros Agrónomos y de Montes, University of Cordoba, Cordoba, Spain, <sup>3</sup> Centro Multidisciplinario de Investigaciones Tecnológicas, Universidad Nacional de Asunción, San Lorenzo, Paraguay

As a preliminary step in the phosphoproteome analysis of germinating seeds (0 and 24 h after seed imbibition) and early grown seedlings (216 h after seed imbibition) from a non-orthodox sp. Quercus ilex, a multiplex (SYPRO-Ruby and Pro-Q DPS) staining of high-resolution 2-DE gels was used. By using this protocol it was possible to detect changes in protein-abundance and/or phosphorylation status. This simple approach could be a good complementary alternative to the enrichment protocols used in the search for phosphoprotein candidates. While 482 spots were visualized with SYPRO-Ruby, 222 were with Pro-Q DPS. Statistically significant differences in spot intensity were observed among samples, these corresponding to 85 SYPRO-Ruby-, 20 Pro-Q-DPS-, and 35 SYPRO-Ruby and Pro-Q-DPS-stained spots. Fifty-five phosphoprotein candidates showing qualitative or quantitative differences between samples were subjected to MALDI-TOF-TOF MS analysis, with 20 of them being identified. Identified proteins belonged to five different functional categories, namely: carbohydrate and amino acid metabolism, defense, protein folding, and oxidation-reduction processes. With the exception of a putative cyclase, the other 19 proteins had at least one orthologous phosphoprotein in Arabidopsis thaliana, Medicago truncatula, N. tabacum, and Glycine max. Out of the 20 identified, seven showed differences in intensity in Pro-Q-DPS but not in SYPRO-Ruby-stained gels, including enzymes of the glycolysis and amino acid metabolism. This bears out that theory the regulation of these enzymes occurs at the post-translational level by phosphorylation with no changes at the transcriptional or translational level. This is different from the mechanism reported in orthodox seeds, in which concomitant changes in abundance and phosphorylation status have been observed for these enzymes.

**Keywords: holm oak, recalcitrant seeds, germination, phosphoproteomics, post translational modification**

# **Introduction**

Holm oak (*Quercus ilex* L. subsp. *ballota* [Desf.] Samp.) is the dominant tree species in natural forest ecosystems over large areas of the Western Mediterranean Basin (Pulido et al., 2001). Nowadays, forest restoration and reforestation are high priority objectives, with *Q. ilex* being one of the major tree species for such a purpose (MAPA, 2006), this requiring its nursery production at a high scale. *Q. ilex* forest maintenance and sustainability are facing important problems and challenges related to seed viability/conservation and mortality of adult trees and plantlets after field transplantation resulting from adverse environmental conditions like drought and the so-called decline syndrome (Gallego et al., 1999). As natural, non-domesticated, plant species with a great plasticity and phenotypic variability, a key challenge prior to massive clonal propagation is the establishment of techniques for the cataloging and selection of genotypes among provenances with a high survival percentage and productivity under adverse environmental conditions. In our group, a Proteomics Research Program with *Q. ilex* has been carried out in order to study variability of holm oak populations and response to stresses, and to select elite individuals to be used in reforestation programmes (Jorge et al., 2006; Valero Galván et al., 2011, 2012a,b).

Holm oak is a recalcitrant plant species whose germination and viability loss during storage, has been poorly studied at the molecular level if compared with to orthodox ones. This knowledge will help to understand biochemistry and metabolic status before and after the germination process, which could be important for the development and optimization of strategies for large scale propagation, germplasm conservation and seed conservation practices (Balbuena et al., 2011; Walters et al., 2013).

The germination process of plant seeds has been analyzed by using a proteomics approach both for comparative purposes and for characterisation of posttranslational modifications (PTMs), mainly phosphorylation. Phosphorylation is a ubiquitous and reversible PTM, which determines protein conformation, stability and activity (Kersten et al., 2006; Hunt et al., 2007; Bond et al., 2011). Phosphorylation events modulate a wide range of biological processes in plants and other organisms (Nakagami et al., 2010). Thus, in seed germination, phosphorylation has proven to be one of the mechanisms underlying the signaling cascade pathway mediated by ABA (Fujii et al., 2009; Cutler et al., 2010; Umezawa et al., 2013). Quantitative and qualitative profiling of phosphoproteins during seed germination and seedling development has been performed using different proteomic approaches (gel based and gel-free) in different plant species such as *Arabidopsis thaliana* (Sugiyama et al., 2008; Kersten et al., 2009; Reiland et al., 2009), *Medicago truncatula* (Kersten et al., 2009; Rose et al., 2012a), *Phaseolus vulgaris* (Alonso and Zapata, 2014), *Zea mays* (Lu et al., 2008), and *Oryza sativa* (Chen et al., 2014; Han et al., 2014). It is important to highlight that to the best of our knowledge, all previously investigated species produced orthodox seeds and no data on phosphoproteomic analysis of non-orthodox or recalcitrant seeds have been published.

The characterisation of the phosphoproteome includes the detection and identification of phosphoproteins and phosphopeptides, localisation of the exact phosphorylation sites and the quantitation of phosphorylation status, which can be performed by gel-based and gel-free approaches. Although several MS-based approaches for studying phosphoproteins, including down and bottom-up ones (Kaufmann et al., 2001; Woods Ignatoski, 2001; Agrawal and Thelen, 2005) have been used, phosphopeptides are notoriously difficult to analyse, especially in the presence of their non-phosphorylated counterparts. This is due, among other factors, to the low stoichiometry of phosphorylated proteins arising from the fact that only a small fraction of the protein will exist in a particular phosphorylated form (Wu et al., 2011; Rigbolt and Blagoev, 2012).

Phosphoproteomic experiments are being perfomed by using a phospho-protein/peptide enrichment preliminary step (Thingholm et al., 2009). These protocols require an excessive manipulation of the sample, thus reducing the confidence of the comparative results. It is for that reason, and as a complementary protocol, we propose the use of multiplexing (SYPRO-Ruby and Pro-Q DPS) staining of high-resolution 2-DE gels for a simultaneous analysis of protein changes in abundance and/or phosphorylation status.

In the present work we describe the use of that a technique to detect changes in the phosphoprotein profile throughout the *Q. ilex* seed germination and early seedling growth stages. After MALDI-TOF-TOF MS analysis, we have identified 20 proteins whose phosphorylation status varies during the seed developmental process, with seven of them showing no differences in abundance. This last group included enzymes of the glycolytic and amino acid pathways that were, respectively, more and less phosphorylated in seedlings than in seeds. This pattern was different from the one reported for orthodox seed species, in which concomitant changes in abundance and phosphorylation have been observed for enzymes of these two pathways.

# **Materials and Methods**

### **Plant Material**

Mature acorns were harvested during October–November from healthy holm oaks from Cerro Muriano-Córdoba (Córdoba, Spain 37◦59 57.74N, 4◦46 57.93W). Germination and seedling growth were performed at 22 ± 1◦C for up to 10 days in darkness as described in Liu et al. (2012). Undamaged, mature acorns were sterilized by immersion in 2.5% sodium hypochlorite, washed abundantly with water and finally dried with filter paper. In order to achive a homogeneous and synchronized germination (Liu et al., 2012), acorns were peeled, removing the pericarp and cutting off parts of the distal ends of the acorns, and then placed in plastic boxes containing one sheet of whatman No3 filter paper over wet perlite. The system was covered with filter paper to avoid water loss (Figure S1). Analyzed time course/periods, corresponding to different seed developmental stages (Figure S2) was selected based on morphology as assessed by microscopic observations (Romero-Rodríguez, 2015; Ph. D Thesis); these stages were selected because they were representative of the morphological changes that occur during germination and seedling growth (24 h after imbibition the emergence of radicle was visible and 216 h shoot seedling started to grow). The embryonic axis was removed from seeds at 0 and 24 h after imbibition and the whole seedling at 216 h after imbibition. Samples from each time were abundantly washed with water, blot dried and frozen in liquid nitrogen and stored at −80◦C until protein extraction. Three pooled samples per stage, each one corresponding to a biological replicate (1–2 g fresh weight per pool coming from 20 to 100 individuals), were performed.

## **Protein Extraction, 2-DE Electrophoresis and Multiplex Staining of the Gels**

Tissue samples were ground to a fine powder in liquid nitrogen using a mortar and pestle (three biological replicates per stage). Protein extracts were obtained from embryo axes of mature (0 h, un-imbibed seeds) and germinated seeds (24 h after imbibition when the radicle just emerged) and from seedling radicles (4.5–5 cm length, 216 h after imbibition), (Figure S2). Proteins were extracted using TCA/acetone-phenol according to the protocol of Wang et al. (2006). Protein content in samples was estimated by the method of Bradford (Bradford, 1976) with bovine serum albumin as a standard. Samples (400μg of protein) of each biological replicate per gel, were focused on 17 cm, 5–8 pH IPG strips using a Bio–Rad Protean IEF Cell system (Görg et al., 2004; Maldonado et al., 2008; Valero Galván et al., 2011). The second dimension, SDS-PAGE (Laemmli, 1970) was performed on 12% polyacrylamide gels (PROTEAN<sup>R</sup> Plus Dodeca Cell). Gels were double stained, first with Pro-Q DPS and then with SYPRO-Ruby (Figure S3) following the procedure described in Agrawal and Thelen (2006) and Berggren et al. (2000). Images were captured with Molecular Imager FX (Bio-Rad Laboratories, Inc.). Experimental *Mr*-values were calculated by mobility comparisons with protein standard markers (SDS-PAGE Standards, 161-0304, Bio-Rad) run in a separate marker lane on the SDS-gel, while pI was determined by using a 5–8 linear scale over the total length of the IPG strips.

### **Gel Image Analysis and Statistical Tests**

Gel image (Pro-Q DPS and SYPRO-Ruby) analysis was performed with PDQuest 8.0.1 software (Bio-Rad) (Valledor and Jorrín, 2011). As reported by Agrawal and Thelen (2006) and in order to eliminate false positives, phosphoproteins spots (revealed with Pro-Q DPS) were only considered if the Pro-Q DPS/SYPRO-Ruby volume ratios were higher than those obtained for negative control, non-phosphorylated markers (βgalactosidase and serum albumin) and with ratios equal to or higher than those obtained for phosphorylated ovoalbumin used as positive control. Consistent spot volumes (those present in all biological replicates) were normalized based on total quantity in valid spots, calculated for each 2-DE gel and used for statistical assessments of differential phosphoprotein and total protein abundance. For statistical analysis (ANOVA, PCA), the webbased software NIA array analysis tool (http://lgsun.grc.nia.nih. gov/anova/index.html) (Sharov et al., 2005; Sghaier-Hammami et al., 2013) was employed.

### **MALDI-TOF/TOF Analysis**

Spots with differential abundance were automatically excised (Investigator ProPic, Genomic Solutions), transferred to multiwell 96 plates, and digested with modified porcine trypsin (sequencing grade; Promega) by using a ProGest (Genomics Solution) digestion station. In-gel digestion was performed as decribed by Shevchenko et al. (1996). Peptides were extracted from gel plugs by adding 10μL of 10% (v/v) trifluoracetic acid (15 min at room temperature). Solubilized peptides were desalted and concentrated by using μC-18 ZipTip columns (Millipore). Eluate was directly loaded onto the MALDI plate using α-cyano hydroxycinnamic acid as a matrix. Peptide mass analysis was performed with a MALDI-TOF/TOF (4800 Proteomics Analyzer, Applied Biosystems). The most abundant peptide ions were then subjected to fragmentation analysis (MS/MS), providing information that can be used to determine the peptide sequence. Proteins were assigned identification by peptide mass fingerprinting and confirmed by MS/MS analysis. Mascot 2.0 search engine (Matrix Science Ltd., London; http://www.matrixscience.com) was used for protein identification running over non-redundant NCBI protein, UniprotKB, and *Quercus* (Romero-Rodriguez et al., 2014) databases. The following parameters were allowed: taxonomy restrictions to *Viridiplantae* in public databases, one missed cleavage, 100 ppm mass tolerance in MS and 0.5 Da for MS/MS data, cysteine carbamidomethylation as a fixed modification, methionine oxidation, and the phosphorylation of Ser, Thr, and Tyr residues as a variable modification. The confidence in the peptide mass fingerprinting matches (*p* < 0.05) was based on the MOWSE score, and confirmed by the accurate overlapping of the matched peptides with the major peaks of the mass spectrum. Proteins with statistically significant (*p* < 0.05) hits were positively assigned identification. Identified phosphoprotein sequences downloaded from UniprotKB, NCBI nr or available in Quercus\_DB (Romero-Rodriguez et al., 2014) were subjected to BLAST analysis by using the phosphoprotein BLAST tool in the Plant Protein Phosphorylation DataBase (P3DB) (Gao et al., 2009) available at http://www.p3db.org/, to find orthologous proteins whose phosphorylation sites were described previously in other species. Proteins identified by MALDI TOF/TOF analysis were extracted and classified based on their putative function according to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, using Blast2GO (Conesa et al., 2005) based on BLASTp results against NCBI nr protein database (*e* < 10−3), or according to annotations in UniProtKB protein database.

# **Results**

By using a multiplex double staining of the gels it was possible to detect changes in the protein abundance (SYPRO-Ruby-stained spots) and phosphorylation status (Pro-Q DPS-stained spots) throughout the seed germination and early seedling growth of *Q. ilex* in three different stages, mature seeds (just before imbibition), germinated seeds (24 h after imbibition), and early grown seedlings (216 h after imbibition) (Figure S2). In the analyzed stages the protein yield per fresh weight was around 2– 15 mg of protein per g, diminishing during the seedling growth (**Table 1**).

Proteins in the extract were separated by 2-DE, and were evenly distributed throughout along the whole pH (5–8) and *Mr* (6–116 kDa) ranges (**Figure 1**). A total of 482 spots were resolved after SYPRO-Ruby staining, with 222 of them also being stained with Pro-Q DPS, these corresponding to putative phosphoproteins (Table S1).

Consistent Pro-Q DPS stained spots, present in all the three biological replicates, were subjected to statistical, ANOVA and PCA, analysis, with 55, out of the 222, showing significant variations (spot volume) between samples (Table S1). Both qualitative and quantitative changes were observed (**Table 1**). Taking as a reference the mature seed phosphoprotein-profile, big changes occur after radicle emergence, with small differences in germinated seeds. At the seedling stage (216 h post imbibition), 33 qualitative (7 newly appeared and 26 disappeared), and 20 quantitative (12 up and 8 down) changes were observed (**Table 1**).

Two-dimensional biplots indicating associations between experimental samples and protein spots were generated by principal component analysis (PCA) in NIA array analysis tools (**Figure 2**). The consistent Pro-Q-DPS stained spots were different enough to establish groups of the samples analyzed. The three analyzed stages were separated from each other; the first component separated the mature (0 h, un-imbibed) and germinated (24 h after imbibition) stage from seedling stages (216 h post imbibition), and the second component separated all the three stages. PCA results showed that PC1 and PC2 explained 88.57 and 11.42% of total variance, respectively. The 55 putative phosphoprotein spots were selected for MALDI-TOF/TOF MS analysis. ANOVA tests of SYPRO-Ruby stained spots (Table S1 and Figure S4) revealed that 20 out of the 55 did not show any differences in abundance while 35 did.

### **Protein Identification**

After MALDI-TOF/TOF analysis, 20 putative phosphoproteins were identified (**Table 2**). Out of the 20, 13 changed in abundance, while seven did not. For the former the variation in phospho-signal could be simply due to a change in abundance while for the other seven results can only be explained by a modification in their phosphorylation status. To validate the phosphoprotein character/nature of the identified proteins, a BLAST against entries at Plant Protein Phosphorylation DataBase (P3DB; http://www.p3db.org/) (Kersten et al., 2009) was performed. With the only exception of spot 3103, a putative cyclase family protein, the other *Q. ilex* proteins had at least one orthologous phosphoprotein in *A. thaliana*, *M. truncatula*, *Nicotiana tabacum,* and *Glycine max*. **Table 3** lists the orthologous proteins and their host species (Sugiyama et al., 2008; Jones et al., 2009; Li et al., 2009; Reiland et al., 2009; Grimsrud et al., 2010; Nakagami et al., 2010; Fíla et al., 2012; Rose et al., 2012b).

**FIGURE 1 | A virtual 2-DE gel showing the protein profile of** *Q. ilex* **mature seed embryo axis (0 h, un-imbibed) obtained by successive Pro-Q DPS and SYPRO-Ruby staining.** Proteins stained with SYPRO-Ruby appear in green, while Pro-Q DPS stained proteins appear in red. The statistically significant differential phosphoprotein spots are indicated with circles for quantitative differences and with triangles for qualitative (absence/presence) differences. Numbers in red indicate the protein spots that were identified by MALDI TOF/TOF.

**TABLE 1 | Electrophoretic analysis of changes in the protein and phosphoprotein profile during germination and seedling growth.**


Up and down accumulated proteins were calculated respect to the mature (0h) stage.

\* Phosphorylation profile was considered changed when no difference was observed in SYPRO-Ruby staining but was statistically different in Pro-Q DPS. In contrast, it was considered unchanged when a difference was observed in both staining methods.

Number of spots detected by SYPRO-Ruby and Pro-Q DPS in different analyzed stages and number of differential spots in total protein and phosphoprotein are shown.

The identified proteins were grouped into functional categories based on the KEGG pathways database (**Table 2**): carbohydrate and amino acid metabolism, defense, protein folding and oxidation-reduction processes.

# **Discussion**

As a preliminary step in the phosphoproteome analysis during the seed germination and early seedling growth processes of a non-orthodox sp. *Q. ilex*, a multiplex (SYPRO-Ruby and Pro-Q DPS) staining of high-resolution 2-DE gels was used. With this protocol it was possible to detect changes in protein-abundance and/or phosphorylation status, identifying, at the same time, candidate phosphoproteins. This simple technique could be a good complementary alternative to the enrichment protocols used in the search for phosphoprotein (Subba et al., 2013; Han et al., 2014; Li et al., 2015). Phosphoprotein enrichment apart from providing, as Pro-Q staining does, false positives, involves excessive manipulation of the sample that results in protein and PTM losses and possible biases. It is true that phosphoprotein validation requires the identification of the phosphorylated peptide, this not being possible or being moret difficult through the MALDI-TOF-TOF MS strategy employed in this work (Thingholm et al., 2009). The protocol presented suffers from the inherent limitations of the 2-DE coupled to the MALDI-TOF-TOF strategy, such as the possible existence and identification of commigrating spots. In at least one case, that of the cyclase, a phosphopeptide was identified, thus confirming its phosphoprotein nature. In any case, rather than identifying the site of phosphorylation, our objective was to search for putative phosphoroteins that showed changes in the phosphorylation status and interpreted those changes from a biological point of view. Different evidence confirmed that most of the Pro-Q–DPS stained spots identified corresponded to real phosphoproteins. Thus: (i) they are Pro-Q stained; (ii) they were only considered if the Pro-Q DPS/SYPRO-Ruby volume ratios were higher than those obtained for a negative control, non-phosphorylated markers (β-galactosidase and serum albumin in this work) and with a ratio equal to or higher than to those obtained for phosphorylated ovoalbumin used as a positive control (Agrawal and Thelen, 2006); (iii) phosphoprotein orthologs have been reported for four different plant species, including *A. thaliana, M. truncatula, N. tabacum,* and *G. max;* (iv) biological interpretation of the data, as discussed below, fits in very well with what is known about the regulation of the identified proteins by phosphorylation. The percentage of phosphorylated proteins detected in our experimental system was of 46%, with similar figures reported for chickpea seedlings (300, Subba et al., 2013) but lower than those reported for germinating rice seeds (500, Han et al., 2014). This could be due to the different methodological approaches and the experimental system used, rather than the system itself or the biological process used or the experimental conditions rather than differences in the number of detectable phosphorylated proteins.

The percentage of protein identification was lower than that obtained in Coomassie stained spots (Valero Galván et al., 2011, 2012b), this being related to the amount of protein present beyond the absence of sequences for *Quercus* in databases. The most important functional categories are discussed. In this work, we paid special attention to those proteins that showed variations in the phosphorylation pattern with no changes in protein abundance, so that they were supposed to be regulated at the post-translational levels. Independent (protein abundance or phosphorylation pattern) or simultaneous, multiplex (protein abundance and phosphorylation pattern) proteomics analysis by using a similar (bottom-up, 2-DE based) strategy has been used in the analysis of mature orthodox seed and seed germination process in model and crop plant species, including *Arabidopsis*, rice, soybean, rapeseed and maize (Lu et al., 2008; Meyer et al., 2012; Han et al., 2014).

Three proteins belonged to the *carbohydrate metabolism* category: pyrophosphate-dependent phosphofructokinase (PPi-PFK, spot 7502), phosphoglycerate kinase (PGK, spot 7318) and glucose-1-phosphate adenylyltransferase (AGP, spot 4304) (**Table 2** and **Figure 3**). PPi-PFK is a cytosolic enzyme that catalyzes the phosphorylation of fructose-6-phosphate to fructose-1,6-bisphosphate in the glycolytic direction, using inorganic pyrophosphate as the phosphoryl donor. This process makes fructose flow into glycolysis to provide energy. PGK catalyzes the conversion of 1,3-diphosphoglycerate to 3-phosphoglycerate, the first substrate-level phosphorylation reaction in the glycolytic pathway for production of ATP. AGP catalyzes the synthesis of ADP-glucose, which is the active glucoside for starch synthesis (**Figure 3**). Overall, the phosphorylation states of these three enzymes increased throughout the germination process. Phosphorylation modification of many glycolytic enzymes has been reported to cause a significant increase in enzyme activity (Li et al., 2011), incrementing the glycolysis rate and the generation of energy to supply the needs of the developing seedling. These results are in agreement with previous studies on rice germination and seedling (Nakagami et al., 2010; Chen et al., 2014; Han et al., 2014). An increased glycolytic activity in germinating *Q. ilex* seeds is supported by a decrease in sucrose content (Romero-Rodríguez, 2015).



**48**

(Continued)






**TABLE 3 | Continued**

On the contrary, for enzymes of the amino acid metabolism (Glutamate decarboxylase, spot 3612) and chaperones (Heat shock protein 60, spot 2606) a decrease in their phosphorylation signal was observed (**Table 2**). Glutamate decarboxylase (GDC) catalyzes the decarboxylation of glutamate to GABA, a nonprotein amino acid involved in stress tolerances that accumulates in germinating seeds of rice and tomato (Taji et al., 2002; Leitner et al., 2011). Some isoforms of this enzyme are inhibited by phosphorylation (Bao et al., 1995). If applicable to GDC, the reduction in its phosphorylation status observed here might imply an increase in the activity of this enzyme, to eliminate the

# **References**


excess of glutamate and glutamine originated by the high rates of stored proteins degradation occurring during germination.

The phosphorylation status of heat shock proteins (HSPs), involved in *protein folding*, has been described decreasing during rice germination (Han et al., 2014). In agreement with that, HSP60 showed high levels of phosphorylation in nonimbibed (0 h after imbibition) and germinated seeds (24 h after imbibition).

In conclusion, over 200 putative phosphoproteins spots were detected in our analysis. Among them, 20 proteins exhibited significant changes in their phosphorylation status, seven of which were identified. Identified enzymes of the glycolytic (pyrophosphate-dependent phosphofructokinase and phosphoglycerate kinase) and amino acid metabolic pathways (glutamate decarboxylase) and protein folding (heat shock protein 60) did not change in abundance during germination and growth but their phosphorylation status increased suggesting regulation at the post-translational level. Alterations in the phosphorylation status of proteins related to glycolysis and amino acid metabolism are in agreement considering that these pathways must increase from mature seeds to germinated seeds and seedling. To test these hypotheses it is necessary identify the phosphorylation sites, but this work constitutes an initiation in the study of the molecular mechanism involved in *Q. ilex* seed germination. The phosphoproteome analysis suggested that the metabolic machinery present in the recalcitrant seeds receives a signal to activate and resume/summarize the most important metabolic pathways in *Q. ilex* to start the germination and the establishment of the seedlings. In orthodox seeds, changes in abundance, together with differences in their phosphorylation status, were observed for these enzymes. Thus, this is one of the differences between orthodox and non-orthodox seeds that may explain their different behavior (Han et al., 2014).

# **Supplementary Material**

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00620


synthase in stress tolerance in *Arabidopsis thaliana*. *Plant J.* 29, 417–426. doi: 10.1046/j.0960-7412.2001.01227.x


**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.

*Copyright © 2015 Romero-Rodríguez, Abril, Sánchez-Lucas and Jorrín-Novo. 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.*

# MALDI-TOF MS profiling approach: how much can we get from it?

Angela Mehta\* and Luciano P. Silva\*

Embrapa Recursos Genéticos e Biotecnologia, Parque Estação Biológica, Brasília, Brazil

Keywords: MALDI-TOF MS profiling, plants, microorganisms

Mass spectrometry has brought unprecedented possibilities in the field of proteomics. The advances obtained in the last 10 years have been outstanding and have enabled faster and more reliable data acquisition and comparison. One powerful method developed was the matrix-assisted laser desorption/ionization time-of-flight mass spectrometry profiling (MALDI-TOF MS profiling) approach, primarily used for the rapid and accurate identification of microorganism species. At least three commercial manufacturers developed systems (instrument, processing method, and databases) for this purpose with relatively similar results although not completely comparable or exchangeable. These methods were conceived for species level identification; however we explored these approaches with some few modifications for expression profiling purposes. We tested different samples, from diverse organisms (not only microorganisms) and found that MALDI-TOF MS profiling methods also have the ability to differentiate samples submitted to different biological conditions (e.g., biotic or abiotic stresses). In the case of MALDI-TOF MS profiling traditional approach, the extraction procedure is based on the enrichment of ribosomal proteins. By using different extraction protocols, samples can also be enriched with different types of proteins. Indeed, when proteins considered for profiling were further analyzed by MALDI TOF/TOF methods, other proteins could also be detected. Although MALDI-TOF MS profiling methods have been used in several tissues and samples, this approach has been rarely employed in plants. One of the applications in plant research is the identification of biomarkers associated to disease, which could, for example, help on quarantine procedures. Overall, MALDI-TOF MS profiling has a high potential to contribute for sample differentiation and biomarker identification and should be better explored in plant proteomics.

# Advances and Applications of MALDI-TOF MS Profiling

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry profiling (MALDI-TOF MS profiling) is an emerging approach based on rapid and high-throughput screening of ions from molecules directly detected in biological samples, including intact cells and crude extracts (Albrethsen, 2011). Although MALDI-TOF MS profiling is one of the most promising mass spectrometry techniques, it detects only the most intense molecular ions of low molecular mass (typically < m/z 20,000) and does not enable direct protein identification. MALDI-TOF MS profiling commonly leads to the detection of a large number of biomarkers and captures the fingerprints of cells, tissues, and biological fluids under normal or altered conditions. MALDI-TOF MS profiling has been initially targeted for the classification of clinically relevant medical applications, including cancer (Sidransky et al., 2003) and pathogenic microorganism (Ilina et al., 2009) diagnostics. Recently, it has proved to be a versatile tool toward an unprecedented number of applications. Clinical bacterial (Eigner et al., 2009) and mycobacterial (Oswald-Richter et al., 2012) isolates, entomopathogenic soil fungi (Lopes et al., 2014), environmental yeasts (Agustini et al., 2014), and viruses (Calderaro et al., 2014) are some of the biological systems which were identified

### Edited by:

Jesus V. Jorrin Novo, University of Cordoba, Spain

### Reviewed by:

Tiago Santana Balbuena, State University of São Paulo, Brazil Jesus V. Jorrin Novo, University of Cordoba, Spain Manuela Peukert, Institut für Pflanzengenetik und Kulturpflanzenforschung Gatersleben, Germany

### \*Correspondence:

Angela Mehta and Luciano P. Silva, angela.mehta@embrapa.br; luciano.paulino@embrapa.br

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

> Received: 21 December 2014 Accepted: 06 March 2015 Published: 23 March 2015

### Citation:

Mehta A and Silva LP (2015) MALDI-TOF MS profiling approach: how much can we get from it? Front. Plant Sci. 6:184. doi: 10.3389/fpls.2015.00184 and clustered by MALDI-TOF MS profiling. In addition, it has been applied for quality control of foods including honey (Wang et al., 2009) and chocolate (Bonatto and Silva, 2014a).

Interestingly, it has also been demonstrated by several authors that MALDI-TOF MS profiling can exhibit resolution even toward bacterial strain-level identification (reviewed in Sandrin and Goldstein, 2013). MALDI-TOF MS profiling has also been applied for the detection of methicillinresistant strains of Staphylococcus aureus (Sparbier et al., 2013). Recently, Bonatto and Silva (2014b) reported the use of this technique for the differentiation of yeast cultures submitted to metal nanoparticle stress, showing that this method can have additional applications beyond microorganism species identification. We have further tested MALDI-TOF MS profiling for bacterial and plant samples and also successfully differentiated specific biological conditions (unpublished data).

The main prerequisite for improved identification of biological samples by MALDI-TOF MS profiling is a curated database of mass spectra. Commercial and public mass spectral libraries typically contain hundreds to thousands of entries which are clustered according to a subject category or biological taxa. Although there are several companies manufacturing MALDI instruments, there are only three major commercial systems available in the market (BioTyper— Bruker Daltonics, SARAMIS—Shimadzu and Anagnostec, and MicrobeLynx™—Waters Corporation) for which, equipment, software, and database are integrated aiming at the identification and classification of organisms. These systems typically apply robust algorithms that are associated with multivariate statistic approaches and show distinct quantitative levels of reliability. Indeed, some authors have compared the results obtained using two of the above mentioned approaches and their performances were overall similar (Lohmann et al., 2013).

# MALDI-TOF MS Profiling in Plant Proteomics

The use of MALDI-TOF MS profiling in plants has been restricted to metabolite profiles and has been rarely reported (Fraser et al., 2007). Therefore, to date there is no database sharing spectra from protein profiles of plant tissues or organs under physiological or altered states. However, the application of this technique can have high impacts in plant proteomic studies, contributing for sample differentiation and/or protein marker discovery in agriculture and industry. The identification of biological markers is an important field of study and can significantly help breeders select cultivars better adapted to diverse biotic and abiotic stresses or in different developmental stages.

One of the main challenges in applying MALDI-TOF MS profiling to plant studies is the high complexity observed in plant tissues. For example, leaves have a high ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) content, which seems to interfere in the reproducibility of protein profiles. We have analyzed Brassica oleracea leaves infected with the bacterium Xanthomonas campestris pv. campestris in comparison to non-infected leaves and found that the results did not always show a clear distinction and sometimes clustered different samples together (unpublished data). Reproducibility limitations, specially related to peak intensities, have also been reported in human disease diagnosis when using MALDI TOF MS (Albrethsen, 2007). However, we are confident that a fine tuning of the methodology may be able to solve this problem. Sample preparation and tissue origin, for example, may be crucial for the high reproducibility required during MALDI-TOF MS profiling experiments. The removal of highly abundant proteins such as RuBisCO may result in more reliable and consistent results. In addition, plant organs and tissues showing more constant proteomes (e.g., pollen and seeds) could be preferable when the molecular signature of a given species is the most important challenge. On the other hand, leaves and other more variable plant organs must be the choice if the main interest is to evaluate a specific biological state.

Another challenge is the identification of the proteins detected in the MALDI-TOF MS profiles since this approach is merely comparative. One possibility is the use of enzymatic hydrolysis to generate peptide fingerprints representing a given physiological stage. We recently used this approach in order to generate a representative pool of the peptides hydrolysable by trypsin from the B. oleracea proteins and further detected as a shotgun by the MALDI-TOF MS profiling approach. However, the proteins containing the peptides detected in the MALDI-TOF MS profiling could not be inferred based only on predicted peptide masses obtained by the hydrolysis due to the low mass accuracy and mainly the signal suppression associated with the high abundance of some molecular components.

The species identification based on MALDI-TOF MS profiling approaches is based on ribosomal proteins, which are enriched by the extraction method with formic acid. Indeed, for accurate identification purposes, this enrichment is crucial. However, when other extraction methods are used, it is possible to get a broader range of protein types, allowing a more global view of the proteins being expressed. We have already tested phenol extraction method followed by precipitation in ammonium acetate and methanol for bacterial and plant samples and have found that a higher diversity of proteins was obtained. Therefore, when using this approach for protein expression profiling, the use of such extraction methods seems preferable. It is noteworthy that ions from other biological molecules (e.g., lipids and secondary metabolites) and mass ranges (e.g., lower than m/z 1000) can be used for identification of biological conditions. Ion sources that do not require conventional MALDI matrices or approaches that are based on surfaceenhanced laser desorption/ionization (SELDI) or matrix-free methods are being currently developed by researchers and companies worldwide and forthcoming years will show significant evolution.

Overall, MALDI-TOF MS profiling is a powerful technique that can have various potential applications in plant proteomics, such as protein marker discovery, and provide substantial contributions for genetic breeding programs and biotechnology. Efforts need to be made in order to adapt the methodology for different plant tissues, however, our studies have shown that the use of this technology is feasible, fast and reliable and can be successfully applied in plant proteomic studies.

# References


# Acknowledgments

We thank Embrapa, CNPq and CAPES for the financial support. We acknowledge the Laboratory of Mass Spectrometry of Embrapa Recursos Genéticos e Biotecnologia for the support in the mass spectrometry analysis.


**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.

Copyright © 2015 Mehta and Silva. 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.

# **Beyond the Western front: targeted proteomics and organelle abundance profiling**

### *Harriet T. Parsons <sup>1</sup> and Joshua L. Heazlewood 2,3 \**

*<sup>1</sup> Section for Plant Glycobiology, Department of Plant and Environmental Sciences, University of Copenhagen, Frederiksberg, Denmark, <sup>2</sup> Joint BioEnergy Institute, Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA, <sup>3</sup> The Australian Research Council Centre of Excellence in Plant Cell Walls, School of BioSciences, The University of Melbourne, Melbourne, VIC, Australia*

The application of westerns or immunoblotting techniques for assessing the composition, dynamics, and purity of protein extracts from plant material has become common practice. While the approach is reproducible, can be readily applied and is generally considered robust, the field of plant science suffers from a lack of antibody variety against plant proteins. The development of approaches that employ mass spectrometry to enable both relative and absolute quantification of many hundreds of proteins in a single sample from a single analysis provides a mechanism to overcome the expensive impediment in having to develop antibodies in plant science. We consider it an opportune moment to consider and better develop the adoption of multiple reaction monitoring (MRM)-based analyses in plant biochemistry.

### *Edited by:*

*Sabine Lüthje, University of Hamburg, Germany*

### *Reviewed by:*

*Martin Hajduch, Slovak Academy of Sciences, Slovakia Stefanie Wienkoop, University of Vienna, Austria*

### *\*Correspondence:*

*Joshua L. Heazlewood, The Australian Research Council Centre of Excellence in Plant Cell Walls, School of BioSciences, The University of Melbourne, Swanston Street, Melbourne, VIC 3010, Australia joshua.heazlewood@unimelb.edu.au*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 17 March 2015 Accepted: 15 April 2015 Published: 05 May 2015*

### *Citation:*

*Parsons HT and Heazlewood JL (2015) Beyond the Western front: targeted proteomics and organelle abundance profiling. Front. Plant Sci. 6:301. doi: 10.3389/fpls.2015.00301* **Keywords: multiple reaction monitoring (MRM), organelle abundance, immunoblotting,** *Arabidopsis***, quantitative proteomics, proteomics**

Higher eukaryotic genomes encode tens of thousands of genes and after considering splice variants and post-translational modifications, likely produce hundreds of thousands of distinct protein products. In eukaryotic cells, proteins are found distributed amongst membrane bound organelles that undertake a multitude of specialized functions and often partition metabolic pathways. Understanding the functional roles of these organelles has given us a comprehensive overview of plant physiology, on to which the complex details of the dynamic regulation of plant-environment interactions can be mapped.

Subcellular fractionation and enrichment by density centrifugation has played a central role in elucidating the functional roles of subcellular compartments. The main biochemical processes were described years before the advent of electrophoretic transfer of proteins on to membranes or the use of antibodies to probe homogenates (Packer et al., 1970; Towbin et al., 1979; Burnette, 1981). Purity was typically assessed by a combination of electron microscopy and enzyme assays (Stocking, 1959; Douce et al., 1977; Mettler and Leonard, 1979) or, in some cases, radiolabeling (Galbraith and Northcote, 1977). Above a certain threshold of purity, maintenance of structural integrity and enzyme activity was the most important prerequisite during the fractionation process. However, for compartments that were less easily enriched than discrete organelles like the plastid or mitochondrion, assessment of contamination levels became more pressing and researchers turned toward immunoblotting, as well as enzyme assays and microscopy (Norman et al., 1986; Hahn et al., 1987; Meyer et al., 1988).

The advent of modern mass spectrometry and proteomics meant that not only could the main biochemical reactions or constituents of a compartment be investigated but many potentially functionally associated proteins could be identified, making organelle proteomics a valuable tool for reducing the complexity of the eukaryotic cell. The likelihood that a protein is correctly assigned to a location is either a function of the purity of the subcellular isolate (for review, see Millar and Taylor, 2014) or of the migration profile of an organelle on a continuous gradient, relative to other subcellular compartments (Dunkley et al., 2006; Nikolovski et al., 2012; Groen et al., 2014).This meant that accurate estimation of the organelle composition of samples became a critical question in this field as well as for biochemical analyses (Aronsson and Jarvis, 2002; Eubel et al., 2007; Parsons et al., 2012; Millar and Taylor, 2014). Although enzyme assays have proven useful in some contexts, as a general method for assessing organelle purity they are not suitable; maintenance of enzyme activity cannot always be assumed and, as reliable assays do not exists for all compartments, not all contaminants can be excluded. Purity is assayed most directly by electron microscopy, although as membranous vesicles can be difficult to distinguish, it cannot always provide a reliable answer. Furthermore, it is dependent on a considerable technical investment and knowledge that is not always possible for many research groups.

Immunoblotting provides a better means by which to address this question, but the qualitative nature of the signal detection makes it a poor choice for accurately assessing the proportional enrichment of a compartment. The availability of antibodies is not evenly distributed across the subcellular compartments in plants with some only being represented by one or two antibodies. Using multiple antibodies as representative markers for an organelle is an important control in situations where purity is paramount to the confidence placed in newly assigned proteins. Using publically available *Arabidopsis* proteomics data and spectral counting, it has been possible to estimate these confidence levels (Reumann et al., 2009; Parsons et al., 2012). However, quantification during fractionation nevertheless remains a limiting factor in this process.

The recent development of protein quantification methods by targeted mass spectrometry has revived discussions regarding the most efficient methods for the quantification of a protein in a sample (Lehmann et al., 2008; Aebersold et al., 2013). Targeted proteomics techniques aim to detect and determine the quantity of a limited set of predefined peptides in a complex mixture of peptides following enzymatic digestion of a protein samples by, e.g., trypsin. This is in contrast to data-dependent acquisition (commonly referred to as "shotgun proteomics") where the aim is to identify as many peptides, and therefore proteins, in a sample as possible. This, however, introduces a certain element of randomness into peptide detection, particularly for lower-abundance peptides and so makes for poor protein quantitation. In multiple reaction monitoring (MRM), or selected reaction monitoring (SRM), a triple quadrupole mass spectrometer is used to select a precursor ion and its resultant product ion(s) after fragmentation (Kondrat et al., 1978). Selection of the parent ion occurs in the first mass analyzing quadrupole (Q1), which is set to a narrow mass window according to the masses of the ion(s) of interest. Collision induced disassociation in the second quadrupole (q2) results in fragmentation of the parent ion in to product ions which are detected in the third quadrupole (Q3) which, again, is set to an appropriately narrow mass window. By focussing machine time on a defined number of peptides, and by requiring both the parent and product ion to be detected, this technique is sufficiently

sensitive and the background signal sufficiently low, that quantitation is possible for both high and moderately low-abundance peptides within the same complex starting mixture in a way that cannot be achieved using shotgun proteomics. The approach has been developed for proteomic studies, as demand for quantitative workflows has increased (Barnidge et al., 2003; Picotti et al., 2010; Maiolica et al., 2012). In recent years advocates have posited the technique as a superior alternative to immunoblotting (Maiolica et al., 2012; Aebersold et al., 2013; Picotti et al., 2013). Indeed, the application of MRM at the individual protein and protein isoform level has proved its ability to detect and quantify proteins against which raising antibodies would have been difficult (Zulak et al., 2009; Taylor et al., 2014). In *Arabidopsis* (Lehmann et al., 2008) and *Chlamydomonas* (Recuenco-Munoz et al., 2015), spiking samples with stable isotope-labeled versions of peptide targets has allowed absolute quantitation of proteins, referred to as a mass western as the results resemble the theoretical output of quantitative immunoblot but done using mass spectrometry.

Given the history of using approaches like westerns and enzyme assays to assess organelle contributions in a sample, the MRM technique could be extended from the individual protein to the compartment level by designing suites of peptide transitions covering marker proteins for multiple subcellular compartments. This would be akin to undertaking multiple immunoblots with suites of antibodies against major plant cellular compartments, like those currently available commercially (e.g., Agrisera AB) and would quickly and easily enable the estimation of the subcellular composition of a given sample. This perspective seeks to explore MRM as an alternative to immunoblotting for assessing the relative abundance of organelles in plant homogenates.

Unlike many targeted approaches using mass spectrometry where protein abundance is assayed in the context of a response, this survey describes the relative abundance of marker proteins between compartments in the same sample, without reference to their function. Several marker proteins and representative peptides per compartment were selected to ensure the overall signal would be representative of the compartment as a whole. Once adequately developed with a collection of reliable transitions that had been assessed for parameters such as limits of detection, limits of quantitation, matrix effects, ion suppression and linearity, the adoption of this technique could greatly benefit the plant community. The ability to assess both the contamination levels of an organelle preparation and track organelle migration during centrifugation would be incredibly useful, but it is imagined that it could also provide means for the rapid monitoring of changes in organelle populations (Yan et al., 2005; Castillo et al., 2008). Consequently, we sought to highlight the potential of the approach by developing an initial set of transitions for specific organelle marker proteins to assess the potential of this approach.

An organelle abundance profile was generated for the reference plant *Arabidopsis* by selecting and analyzing candidate MRM peptide transitions for a number of organelle marker proteins (**Figure 1**). Only proteins repeatedly localizing to a subcellular compartment (Tanz et al., 2013) and generating nonredundant peptides were selected as markers. As far as was possible, selected proteins were functionally unrelated, not coexpressed and within the top 40 most expressed transcripts for

rosettes **(B,D)**. MRM assays **(A,B)** were performed using two to five marker proteins per compartment, except the plastid where six marker proteins were used, including both the light-harvesting complex candidates (three proteins) and non-light harvesting complex candidates (three proteins). Error bars show standard error for *n* = 3 biological replicates. Spectral counts **(C,D)** were

**TABLE 1 | Summary of representative marker proteins and peptides used for detection of subcellular compartments by MRM.**


*<sup>1</sup>LHC, light harvesting complex.*

an organelle or compartment. This last point was important for comparisons between compartments. For this proof-of-concept study a minimum of three marker proteins per compartment was applied (with the exception of the vacuole); in some instances up to five were employed (e.g., for the PM) when available MRM transitions were readily identified (**Table 1**). A ribosomal category was included with the 10 major subcellular categories (**Table 1**) as these can be an appreciable source of sample contamination in subcellular proteomics. Using *in vitro* synthesis techniques (Brownridge et al., 2011), we have thus far validated the identity (retention time and fragment ions) of at least one peptide per subcellular compartment, i.e., 25 of the 72 peptides.

organelle/subcompartment were summed and expressed as a percentage of the total number of identified proteins. Cyt, cytosol; ER, endoplasmic reticulum; ExC, extracellular; Mt, mitochondria; Ncl, nucleus; Prx, peroxisome; Pld, plastid;

PM, plasma membrane; RP, ribosomal proteins; Vac, vacuole.

Typical differences in organelle abundance detected using this MRM method are demonstrated in two very different but popular experimental systems; heterotrophic *Arabidopsis* cellsuspension culture (**Figure 1A**) and 4-week old *Arabidopsis* rosettes (**Figure 1B**). As growth conditions varied dramatically between systems, particularly with respect to light and carbon source, both light-harvesting complex and non-light harvesting complex plastid markers were included. These MRM profiles of subcellular compartments were then compared to profiles generated by spectral counts of several 100 compartment marker proteins from data-dependent analyses of total protein extracts (**Figure 1**; **Table 1**). Although data-dependent acquisition approaches are known to favor medium/high abundance proteins (Wienkoop and Weckwerth, 2006; Ahn et al., 2007), since relatively abundant proteins had been selected as organelle markers for MRM such a comparison was considered meaningful. Both MRM and data-dependent analyses produced similar organelle profiles for each system (**Figure 1**), showing that using MRM to estimate the relative abundance of subcellular compartments is conceptually valid. Changes in relative abundance were detected in all subcellular categories, demonstrating the quantitative capacity and sensitivity of MRM. As expected given the physiological differences between the two systems, plastids were much less abundant, and mitochondria more so, in cell cultures compared to rosettes (**Figures 1A,B**). Ribosomal proteins and lower-abundance organelles such as the Golgi and peroxisome appeared more abundant in cell cultures (**Figure 1A**), as expected for cytoplasmic-dense, rapidly-dividing, undifferentiated cells grown in a relatively high-oxygen environment.

Some differences between spectral counting and MRM were observed. Lower-abundance organelles such as the Golgi, peroxisome and ER appeared lower when estimated by spectral counting (compare **Figures 1A,C**). The ratio of plastidic proteins to proteins from other compartments also appears lower in the MRM results compared to standard spectral counting approaches (**Figures 1B,D**). Detection is biased against very small, low abundance, or hydrophobic proteins using data-dependent acquisitions, particularly in complex samples, whilst heavily post-translationally modified proteins may never be detected. Undoubtedly, this will affect compartments disproportionately, potentially leading to misrepresentation using techniques such as spectral counting, which could explain these discrepancies between results such as over-representation of the plastid in photosynthetic tissues. This analysis demonstrates a proof of concept for this application of MRM in determining relative organelle abundance, and shows how it could potentially lead to a more accurate estimation of organelle abundance when compared to immunoblotting, enzyme assays, and other mass spectrometry techniques such as spectral counting. However, these results do also point to some potential drawbacks of this technique in its current format.

This technique relies on the assumption that changes in abundance of an entire subcellular compartment can be represented by a small number of proteins. Therefore, this makes the appropriate selection of proteins a critical consideration when designing suites of transitions for detecting subcellular compartments. Tissue- or environment specific changes in gene expression can be largely avoided by consulting publically available microarray data; however how this compares to proteins levels is harder to predict. The disproportionate decrease in light-harvesting complex proteins (**Figures 1A,B**) may reflect environmental influence on protein expression. However, the similarities between **Figures 1A,C** and **Figures 1B,D** suggest that by applying stringent criteria during MRM marker selection, such effects may be minimized. For

# **References**


example, had markers been selected entirely from non-light harvesting complex proteins, our results would have looked very different.

Through this proof-of-concept study, which paves the way for an in-depth analysis of the applicability of MRM to estimating the subcellular composition of samples, we demonstrate that relative quantitation can provide a sufficient overview. One established such approaches could enable profiling of organelles abundance between treatments or as a means to assess the purity of a subcellular fraction. Although this technique does not offer absolute peptide quantitation, as can be achieved by using labeled peptides (Lehmann et al., 2008), this technique likely represents an appreciable increase in accuracy compared to immunoblotting and would already be applicable to a well-annotated species such as *Arabidopsis*. By avoiding isotope-labeled peptides used for absolute quantification, e.g., AQUA (Gerber et al., 2003), which requires additional levels of cost, we aim to increase accessibility of this technique. The triple quadrupole mass spectrometers required are relatively affordable, further increasing the potential popularity of MRM for such approaches. Future design of peptide transitions will be facilitated by the MRMaid and *Arabidopsis* Proteotypic Predictor resources (Fan et al., 2012; Taylor et al., 2014) and validation of transitions facilitated by the development of synthetic peptides libraries (Picotti et al., 2010) and the emergence of techniques such as QconCAT (Beynon et al., 2005; Brownridge et al., 2011). The investment required for quality antibody development will likely prevent production keeping pace with the number of emerging organisms of research interest. Genomic annotation of several more plant species is approaching the level of completion required for non-redundant peptide selection (Goodstein et al., 2012), whilst ever accumulating data in PRIDE (Vizcaino et al., 2013) will facilitate the design of marker peptides for other species, meaning that MRM-based techniques will have a sizable impact on plant cellular biology in the coming years.

# **Acknowledgments**

This work was supported by the U. S. Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the U. S. Department of Energy. JH is supported by an Australian Research Council Future Fellowship [FT130101165]. HP is supported by a Marie Curie Intra European Fellowship 2012 [FP7-PEOPLE-2011-IEF 301401].


**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.

*Copyright © 2015 Parsons and Heazlewood. 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.*

# Characterization of protein N-glycosylation by tandem mass spectrometry using complementary fragmentation techniques

### Kristina L. Ford<sup>1</sup> , Wei Zeng<sup>1</sup> , Joshua L. Heazlewood1, 2 and Antony Bacic<sup>1</sup> \*

<sup>1</sup> ARC Centre of Excellence in Plant Cell Walls, School of BioSciences, The University of Melbourne, Melbourne, VIC, Australia, <sup>2</sup> Physical Biosciences Division, Joint BioEnergy Institute, Lawrence Berkeley National Laboratory, Berkeley, CA, USA

### Edited by:

Katja Baerenfaller, Swiss Federal Institute of Technology Zurich, Switzerland

### Reviewed by:

Cécile Albenne, Université Toulouse III - Paul Sabatier, France Friedrich Altmann, University of Natural Resources and Life Sciences, Austria

### \*Correspondence:

Antony Bacic, ARC Centre of Excellence in Plant Cell Walls, School of BioSciences, The University of Melbourne, Building 122, Melbourne, Victoria 3010, Australia abacic@unimelb.edu.au

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 08 April 2015 Accepted: 15 August 2015 Published: 28 August 2015

### Citation:

Ford KL, Zeng W, Heazlewood JL and Bacic A (2015) Characterization of protein N-glycosylation by tandem mass spectrometry using complementary fragmentation techniques. Front. Plant Sci. 6:674. doi: 10.3389/fpls.2015.00674 The analysis of post-translational modifications (PTMs) by proteomics is regarded as a technically challenging undertaking. While in recent years approaches to examine and quantify protein phosphorylation have greatly improved, the analysis of many protein modifications, such as glycosylation, are still regarded as problematic. Limitations in the standard proteomics workflow, such as use of suboptimal peptide fragmentation methods, can significantly prevent the identification of glycopeptides. The current generation of tandem mass spectrometers has made available a variety of fragmentation options, many of which are becoming standard features on these instruments. We have used three common fragmentation techniques, namely CID, HCD, and ETD, to analyze a glycopeptide and highlight how an integrated fragmentation approach can be used to identify the modified residue and characterize the N-glycan on a peptide.

Keywords: glycosylation, fragmentation, electron-transfer dissociation, post-translational modification, tandem mass spectrometry

The identification, characterization and quantification of post-translational modifications (PTMs) in proteins are a major challenge in the field of proteomics (Heazlewood, 2011). Protein glycosylation is one of the most commonly occurring PTMs with estimates of around 50% of the cellular proteome predicted to be glycosylated (Van den Steen et al., 1998). As well as being a commonly occurring PTM, protein glycosylation is one of the more difficult protein modifications to investigate due to the heterogeneity of the glycan structure (Song et al., 2011). Indeed, the examination of protein glycosylation is further complicated by the variations in the glycan structure (glycoforms) that can occur on a given polypeptide (Rudd and Dwek, 1997). Aberrations in protein glycosylation result in severe developmental abnormalities in diverse species including mammals (Furuichi et al., 2009) and plants (Lerouxel et al., 2005). Protein glycosylation primarily occurs in the secretory system (ER and Golgi) and can result in functional changes, influence subcellular localization, and protein stability (Oxley et al., 2004; Zhou et al., 2005). Identifying and profiling protein glycosylation is thus essential to define the underlying subtleties of a complex proteome.

The application of tandem mass spectrometry (MS<sup>n</sup> ) as a technique for the characterization of protein glycosylation has occurred for decades (Reddy et al., 1988). However, it has been the development of proteomics-based MS that has enabled the high-throughput analyses of protein glycosylation (glycoproteomics) to occur (Morelle and Michalski, 2007; Zhou et al., 2007; Zielinska et al., 2012). To secure peptide sequences, these high-throughput approaches usually require the enzymatic removal of the glycan prior to analysis by MS (Zhang and Aebersold, 2006), thus losing information on the glycan structure and the specific site of glycosylation. The removal of the glycan is necessary since commonly applied fragmentation techniques produce spectra yielding information from the carbohydrate structure rather than the peptide backbone. This is generally due to the labile nature of the glycan moiety (Ruiz-May et al., 2012b; Saba et al., 2012) and without removal of the glycan, it is difficult to match resultant fragmentation spectra using a standard data analysis workflow. As a result, these types of approaches have thus far dominated the area of glycoproteomics.

As with other eukaryotes, plants contain an array of glycoproteins that are mainly produced in the secretory pathway and contain both N- and/or O-linked glycan structures (Ruiz-May et al., 2012a; Strasser, 2014). A number of targeted plant glycosylation studies have revealed the importance of these motifs to protein function, including enzyme activity (Kimura et al., 1999), thermal stability and folding (Lige et al., 2001), and protein solubility (Welinder and Tams, 1995). However, the prevalence of O-linked glycosylation and the diversity of N-glycan structures found in plant glycoproteins poses distinct challenges for their global analysis in plants (Gomord et al., 2010). The range of N-glycans identified on plant proteins are separated into three broad types; each type represents a variety of structural permutations. The high mannose type comprises structures of GlcNAc2Man5–9 and they mainly occur in the ER and represent immature or precursors of the complex N-glycan structures. The complex type of plant N-glycan structures represents the mature form found in the late Golgi and on extracellular proteins and have been described as comprising GlcNAc4Xyl1Fuc1Man<sup>3</sup> to GlcNAc4Xyl1Fuc3Man3Gal2. Lastly, the paucimannose type represents the processing of terminal residues (GlcNAc/Fuc) from complex N-glycan structures resulting in a processed N-glycan comprising GlcNAc2Xyl1Fuc1Man3; these structures are thought to be present in vacuolar localized proteins (Rayon et al., 1998). Detailed knowledge about O-linked glycan structures found on plant proteins is still limited. The most widely studied proteins with O-linked glycans structures in plants are the extensins and arabinogalactan proteins (AGPs), which belong to the hydroxyproline-rich glycoprotein (HRGP) superfamily, where the O-glycan is mainly attached to modified Pro residues (hydroxyproline). The extensins are glycosylated on a series of adjacent hydroxyprolines with Ara4−<sup>1</sup> followed by an O-linked Gal on a Ser residue (Gomord et al., 2010). In contrast, the AGPs contain large branched Gal structures also O-linked through hydroxyprolines that contain terminal Ara but can also include Fuc, GlcA, Rha (Nguema-Ona et al., 2014). The complexity and diversity of O-glycan structures and resulting glycopeptides derived from these O-linked glycoproteins has resulted in few MS-based studies; indeed their characterization usually requires a multifaceted approach (Hijazi et al., 2012). Consequently, we intend to address the role of complementary fragmentation techniques by MS for the characterization of N-glycans.

The distinct structural feature found in complex plant N-glycans e.g., the presence of an α-1,3-linked fucose in N-glycans (Tretter et al., 1991), has resulted in few glycoproteomic studies as the established workflows require some adaptation (Song et al., 2013). Consequently, the first attempt at applying the emerging glycoproteomic technologies to plant material was only undertaken a few years ago (Zielinska et al., 2012). Prior to this, some plant glycoproteomics had been conducted including in Arabidopsis (Minic et al., 2007) and tomato (Catalá et al., 2011), however these studies only characterized the lectin enriched sub-proteomes from these species. An attempt to characterize the modified residue and the N-glycan structures by MS represented the first real advancement in the area (Zhang et al., 2011). Using a combination of various glycoprotein enrichment strategies and informatics, 127 putative glycoproteins were identified by MS, with N-glycan sites and structures determined by prediction and re-analysis of MS1 scans (Zhang et al., 2011). Only a year later, a significant advance in the identification of N-glycosylation sites in plant proteins occurred as part of a large-scale survey of model eukaryotic organisms applying the developed glycoproteomic strategies (Zielinska et al., 2012). While a standard workflow of lectin affinity to enrich glycopeptides followed by treatment with peptide-Nglycosidase (PNGase) F was used for non-plant species, enriched plant glycopeptides from Arabidopsis were treated with PNGase A to remove the more complex fucose containing N-glycans. Significantly, removal of N-glycan structures was undertaken in the presence of H<sup>18</sup> <sup>2</sup> O resulting in an isotopic signature on the modified residue. Thus, in combination with the deamidation of Asn to Asp by the PNGase reaction, both the site and presence of the N-glycan could be validated in the resultant peptides after analysis by MS. Over 2000 N-glycosylation sites were mapped to Arabidopsis proteins by this study, however, no structural information about the glycans was achieved (Zielinska et al., 2012). While lectin enrichment of glycans has been widely employed in glycoproteomic studies, complementary technologies have been developed including the enrichment of glycopeptides by crosslinking chemically activated carbohydrates to hydrazide beads (Zhang et al., 2003). The technique has been applied to plant samples (Arabidopsis) to both assess the specificity of PNGase (F and A) and profile N-glycans from wild-type and the cgl N-glycosylation mutant (Song et al., 2013). A total of 330 glycopeptides from 173 Arabidopsis proteins were identified using this approach. Surprisingly, it was found that the activities of commercially available PNGase A were ineffective with glycoproteomic workflows and that the identified glycopeptides were likely to have harbored immature or mannose-type glycans, like those found in the Arabidopsis cgl mutant (Song et al., 2013). It should be noted that this ineffective removal of complex N-glycans with PNGase A occurred with enriched glycopeptide fractions rather than with glycoproteins, which are known to be difficult to digest with PNGases. The focus in plant glycoproteomics has thus far mainly concentrated on the reference plant Arabidopsis with about 2500 N-glycan sites mapped (Mann et al., 2013). Nonetheless, the small number of glycoproteomic studies undertaken in plants and the limited species analyzed indicate that the current workflows are providing limited opportunities to study plant glycoproteomes and new or complementary approaches need to be explored.

The development of different fragmentation techniques available in the current generation of MS instruments may provide an alternative approach to the limitations of current glycoproteomic procedures (Scott et al., 2011). The most widely employed fragmentation technique in proteomic surveys to produce tandem mass spectra is collision-induced dissociation (CID). The process involves the acceleration of molecules (peptides) which are collided with a neutral gas (e.g., nitrogen) resulting in the breaking of molecular bonds and the generation of tandem mass spectra (Sleno and Volmer, 2004). The term CID generally encompasses trap-type (derived from an ion trap instrument) and beam-type [derived from a quadrupoletime-of-flight (Q-ToF) instruments]. With the development of the Orbitrap MS, a CID-based approach termed higherenergy collisional dissociation (HCD) was developed where fragmentation spectra are produced outside the iontrap, namely in the C-trap (Olsen et al., 2007). The resultant HCD fragmentation spectra are similar to those when producing CID in a Q-ToF mass spectrometer (Michalski et al., 2012). A further fragmentation technique used in proteomics is electron-transfer dissociation (ETD) or electron-capture dissociation (ECD) which causes fragmentation via the transfer of electrons to the gas phase ion (Zubarev et al., 1998; Syka et al., 2004). The application of ETD in proteomics is mainly associated with the analysis of PTMs since fragment ions tend to retain the modifications which are often lost in CID approaches (Wiesner et al., 2008). All these fragmentation techniques are now common options for most modern tandem MS.

Consequently, with the recent availability of differential MS-based fragmentation techniques, we were interested in exploring these approaches for the structural characterization of glycoproteins. The amount and purity of a specific glycoprotein is crucial for the adoption of the LC-MS<sup>n</sup> approach when applying complementary fragmentation techniques. As a result, we have been using in planta synthesis in Nicotiana benthamiana to transiently express proteins with suspected N-glycan modifications followed by immunoprecipitation as a means to adequately enrich proteins prior to analysis by MS. Recent glycoproteomic analyses in Arabidopsis (Zielinska et al., 2012) had suggested that many enzymes found in the plant Golgi apparatus and involved in cell wall biosynthesis may themselves be glycosylated. Consequently, we have been attempting to identify N-glycans from recently characterized plant glycosyltransferases (Song et al., 2015) after in planta synthesis and immunoprecipitation. Agrobacterium harboring a construct comprising IRREGULAR XYLEM 9 from Asparagus (AoIRX9, KJ556998) conjugated to the yellow fluorescent protein (Venus) was used to infiltrate N. benthamiana leaves and after 4 days, infected leaves were harvested and microsomal protein isolated. The AoIRX9 protein was enriched by immunoprecipitation with GFP-Trap <sup>R</sup> (ChromoTek) and digested with trypsin. The analysis of the digested protein lysate by standard LC-MS<sup>n</sup> (CID) resulted in the identification of AoIRX9 with high sequence coverage and indicated it was considerably enriched in the lysate. The utilization of different fragmentation techniques provides complementary information to sequence a glycopeptide. For example, the utilization of ETD in combination with CID (MS<sup>3</sup> ) has been used to produce relevant structural and sequence information from an N-glycopeptide (Catalina et al., 2007). The AoIRX9 digested lysate was analyzed by HCD and spectra manually inspected for the presence of an oxonium ion (204.09 m/z) likely derived from a HexNAc. A putative glycopeptide was identified with multiple charge states (770.94 m/z [M + 5H]5+, 963.43 [M + 4H]4<sup>+</sup> and 1284.23 [M + 3H]3+) with [M + 5H]5<sup>+</sup> the predominant form. The 770.94 m/z [M + 5H]5<sup>+</sup> ion was subsequently analyzed using ETD to determine the peptide sequence. The three fragmentation modes, (CID, HCD, and ETD) used to characterize this enriched plant glycopeptide from AoIRX9 are outlined in **Figure 1**.

The application of CID (trap-type) produced poor peptide and glycan fragmentation resulting in little peptide backbone information (b<sup>6</sup> and y2) and glycopeptide fragments with high charge states, as previously reported (Catalina et al., 2007; Desaire, 2013). The glycan derived peaks could only be assigned after analysis of the complementary fragmentation spectra was undertaken (**Figure 1A**). The HCD-derived spectra (**Figure 1B**) revealed glycan signatures, such as the HexNAc (GlcNAc) oxonium ion (204.09 m/z) and importantly resulted in an ion representing the mass of the peptide without the glycan structure (1137.54 m/z [M + 2H]2+). This information is essential for the calculation of the mass of the glycan, 1576.6 Da, corresponding to a glycan structure of HexNAc4Pent1Deoxy1Hex3, which is most likely GlcNAc4Xyl1Fuc1Man3, a commonly found plant N-glycan (Oxley et al., 2004; Song et al., 2011). To accurately determine the structure of the glycan, a coupled approach employing detailed analysis of fragment ions and enzymatic sequencing (e.g., exoglycosidase) are necessary (Harvey, 2005). Similar to CID, the HCD spectrum provides minimal information about the peptide backbone, yielding masses for b2, y3, and y5. In contrast, the ETD spectrum (**Figure 1C**) was essential for fragmenting the peptide backbone revealing virtually every z- and a majority of c-ions, enabling confirmation of the peptide sequence as has been previously demonstrated for ETD (Catalina et al., 2007). Lastly, the ETD spectra enabled the determination of the residue (Asn, N) where the glycan structure was attached. In this instance, only a single likely site of N-glycan attachment is possible, however this situation is not always the case.

A considerable bottleneck when attempting to characterize protein glycosylation by MS is the limitations of commonly used software for high confidence matching of glycopeptide fragmentation spectra (Woodin et al., 2013). This issue becomes even more problematic when tandem spectra are derived from fragmentation techniques such as ETD. Although there is an assortment of software options ranging from commercial to open source, each has its limitations depending on the workflow (Dallas et al., 2013). Nonetheless, high throughput matching of glycopeptide spectra is still a considerable challenge which requires extensive manual curation. Generally, we have found that the presence of oxonium ions by manual inspection of HCD derived spectra is still the most reliable approach, however the identity of the glycopeptide is still required. We have attempted to analyze ETD spectra derived from glycopeptides with commonly used programs such as Mascot (Matrix Science, UK), such software is often limited in the number of modifications and the expanded search space results in significant increases in processing time. Consequently, we have found that "boutique

method incorporating a specific activation type (A) CID, (B) HCD, and (C) ETD. The resultant MS<sup>n</sup> data (CID) were used to confirm the presence of the purified protein in the sample. Unmatched spectra from

.

revealed the sequence of the peptide. This complementary approach identified the peptide HLTYKENFTDAKAEADHQR with a complex N-linked glycan comprising HexNAc4Pent1Deoxy1Hex3

software," such as Byonic (Protein Metrics, USA) outperforms many of the more widely utilized programs for the analysis of ETD spectra containing a highly variable PTM. Overall, the issue of separate analyses e.g., HCD and ETD, the requirement to cross-reference data and variable modifications creates a complicated workflow which may be overcome in the future through software enhancements.

The application of complementary fragmentation techniques to characterize glycopeptides provides an effective approach in glycoproteomics (Desaire, 2013). The utilization and success of complementary fragmentation techniques for the analysis of PTMs including glycopeptides has been widely discussed (Catalina et al., 2007; Chi et al., 2007; Alley et al., 2009; Sobott et al., 2009; Snovida et al., 2010; Scott et al., 2011; Ruiz-May et al., 2012b), however the approach is yet to be adopted by the plant proteomics community. The glycopeptide example outlined here further highlights and confirms the effectiveness of HCD and ETD in revealing crucial information necessary for the characterization of both the glycan and the peptide (Scott et al., 2011). The utilization of CID is generally not of great assistance, although could contribute important structural information about the glycan from resultant fragment ions. The presence of the oxonium ion derived from the glycan in the HCD spectra (e.g., HexNAc, 204.09 m/z) should enable more advanced workflows to be employed for the characterization of glycopeptides (Wu et al., 2007; Scott et al., 2011). For example, such an application could involve the presence of the oxonium ion in the HCD fragmentation spectra triggering the ETD fragmentation of the precursor (Saba et al., 2012). A further advancement could be gained by coupling hydrophilic interaction liquid chromatography (HILIC) as the

# References


pre-MS separation procedure (Scott et al., 2011). There is a clear advantage for glycopeptide separation as features like oligosaccharide branching provide the necessary requirements for hydrophilic interactions (Zauner et al., 2011).

Finally, with advancements in hardware, it is simple to envision future instruments that simultaneously undertake a variety of fragmentation processes during a standard proteomics workflow. This improvement would also likely result in the development of software that could better deal with this complicated workflow. Such a feature could enable a global analysis of PTMs in conjunction with protein identification and label free quantification. The example outlined above applying complementary analytical approaches will likely be an essential part of future proteomic workflows for the accurate characterization of protein structure and function.

# Acknowledgments

The interpretation of ETD spectra was conducted with the assistance of Ms. Yin Ying Ho (The University of Melbourne). This work was funded by grants from the Australian Research Council (ARC) to the ARC Centre of Excellence in Plant Cell Walls [CE110001007] and the U. S. Department of Energy, Office of Science, Office of Biological, and Environmental Research, through contract DE-AC02- 05CH11231 between Lawrence Berkeley National Laboratory and the U. S. Department of Energy. JH is supported by an ARC Future Fellowship [FT130101165]. The MS spectra were acquired at the Mass Spectrometry and Proteomics Facility (MSPF), Bio21 Institute, The University of Melbourne with the help of Dr. Ching-Seng Ang.

spondylodysplastic dysplasias group diseases. J. Med. Genet. 46, 562–568. doi: 10.1136/jmg.2008.065201


**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.

Copyright © 2015 Ford, Zeng, Heazlewood and Bacic. 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.

# An improved protocol to study the plant cell wall proteome

Bruno Printz 1, 2, Raphaël Dos Santos Morais <sup>1</sup> , Stefanie Wienkoop<sup>3</sup> , Kjell Sergeant <sup>1</sup> , Stanley Lutts <sup>2</sup> , Jean-Francois Hausman<sup>1</sup> \* and Jenny Renaut <sup>1</sup>

<sup>1</sup> Environmental Research and Innovation Department, Luxembourg Institute of Science and Technology, Belvaux, Luxembourg, <sup>2</sup> Groupe de Recherche en Physiologie Végétale, Earth and Life Institute Agronomy, Universiteì catholique de Louvain, Louvain-la-Neuve, Belgium, <sup>3</sup> Department for Molecular Systems Biology, University of Vienna, Vienna, Austria

Cell wall proteins were extracted from alfalfa stems according to a three-steps extraction procedure using sequentially CaCl2, EGTA, and LiCl-complemented buffers. The efficiency of this protocol for extracting cell wall proteins was compared with the two previously published methods optimized for alfalfa stem cell wall protein analysis. Following LC-MS/MS analysis the three-steps extraction procedure resulted in the identification of the highest number of cell wall proteins (242 NCBInr identifiers) and gave the lowest percentage of non-cell wall proteins (about 30%). However, the three protocols are rather complementary than substitutive since 43% of the identified proteins were specific to one protocol. This three-step protocol was therefore selected for a more detailed proteomic characterization using 2D-gel electrophoresis. With this technique, 75% of the identified proteins were shown to be fraction-specific and 72.7% were predicted as belonging to the cell wall compartment. Although, being less sensitive than LC-MS/MS approaches in detecting and identifying low-abundant proteins, gel-based approaches are valuable tools for the differentiation and relative quantification of protein isoforms and/or modified proteins. In particular isoforms, having variations in their amino-acid sequence and/or carrying different N-linked glycan chains were detected and characterized. This study highlights how the extracting protocols as well as the analytical techniques devoted to the study of the plant cell wall proteome are complementary and how they may be combined to elucidate the dynamism of the plant cell wall proteome in biological studies. Data are available via ProteomeXchange with identifier PXD001927.

### Keywords: proteomics, cell wall, plant, glycosylation, EGTA

# Introduction

Cell walls are biological composites developing outside the cells and forming a rigid frame protecting the cell. Plant cell walls fulfill a wide variety of roles which differ between cell types, plants, and species (Cosgrove, 2005; Guerriero et al., 2014a,b). Cell walls are generally composed of cellulose, lignin, and hemicellulose embedded in an aqueous glue of pectins. The growth of the cell wall is further determined by the presence of minerals (in particular calcium) and the activity of enzymes and structural proteins that account for up to 10% of the mass of the wall of growing cells (Wolf et al., 2012).

Since the first study directed at the proteome of the plant cell wall, more than 55 papers (http:// www.polebio.lrsv.ups-tlse.fr/WallProtDB/index.php/links) have been published and extensive

### Edited by:

Ganesh Kumar Agrawal, Research Laboratory for Biotechnology and Biochemistry, Nepal

### Reviewed by:

Laurence Veronique Bindschedler, Royal Holloway University of London, UK Sixue Chen, University of Florida, USA

### \*Correspondence:

Jean-Francois Hausman, Environmental Research and Innovation Department, Luxembourg Institute of Science and Technology, 5, Avenue des Hauts-Fourneaux, L-4362 Esch/Alzette, Belvaux, Luxembourg jean-francois.hausman@list.lu

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

> Received: 28 January 2015 Accepted: 25 March 2015 Published: 10 April 2015

### Citation:

Printz B, Dos Santos Morais R, Wienkoop S, Sergeant K, Lutts S, Hausman JF and Renaut J (2015) An improved protocol to study the plant cell wall proteome. Front. Plant Sci. 6:237. doi: 10.3389/fpls.2015.00237 research has been carried out on the model species Arabidopsis thaliana (Albenne et al., 2013). Nonetheless, the remarkable diversity in composition and function of the wall across cells, organs and species and the regain of interest in plant by-products in the industrial field [as source of bioethanol (Sreenath et al., 2001), building components (Nozahic et al., 2012), and biopolymers (Hühns and Broer, 2010)] foster the analysis of the cell wall proteome.

The extracellular nature of the wall and the range of bindingaffinities that proteins have for the extracellular matrix make the purification of cell wall proteins in a one-step procedure difficult. The major steps made to improve the enrichment of cell wall proteins (CWPs) have previously been reviewed (Feiz et al., 2006; Jamet et al., 2008; Albenne et al., 2013). Although, CWPs enrichment can be done using non-destructive techniques which preserve membrane integrity, the use of destructive methods that require the grinding of the plant material and consequently the disruption of the plasma membranes is commonly preferred. In these protocols, CWPs are extracted from the ground plant material by washes in buffers of various ionic strengths. In 2004, Watson et al. used a washing procedure using sodium acetate, sodium chloride, and ascorbic acid followed by successive vacuum filtrations on nylon mesh membranes with sodium chloride, water, acetone, and sodium acetate (Watson et al., 2004). In 2006, Feiz et al. introduced a procedure which combines extractions with low ionic strength acidic buffers with different washes in increased sucrose concentration shown to considerably limit the contamination of the wall fraction with intracellular proteins, probably by helping the elimination of organelles and other vesicles less dense than cell wall polysaccharides (Feiz et al., 2006).

In the first studies on CWPs (Bozarth et al., 1987), proteins were extracted with a CaCl<sup>2</sup> solution, later studies proposed the enrichment of CWPs and the reduction of the complexity of the extracts by using CaCl2, cyclohexylenedinitrilotetraacetate (CDTA), DTT, NaCl and borate buffers to sequentially extract proteins with various wall-binding affinities (Robertson et al., 1997). Quicker methods involving only a two-steps fractionation using sodium acetate buffers with CaCl<sup>2</sup> or LiCl, known to be efficient extractants of CWPs, were then developed on crushed plant material (Watson et al., 2004; Feiz et al., 2006). Recently, CaCl<sup>2</sup> was replaced by the chelating agent EGTA to remove the proteins associated with the pectin fraction (Verdonk et al., 2012). Regarding the high degree of variability of the cell wall across species, organs and growing conditions, a broad comparison of these protocols starting from the same initial material appears essential.

Independent of the extraction procedure, CWPs can be identified using different methods involving either 2D LC-MS/MS analysis of the total digested proteins or a separation on gels, followed by a digestion step and MS/MS analysis (Jamet et al., 2008). Although, the basic glycoproteins that are found in the cell wall may be poorly resolved on 2D-gels, this method allows the separation and the relative quantification of different isoforms of a protein and eases the identification of post-translational modifications that may have occurred during the maturation of these proteins.

In this study, the protocols used in the two major studies dealing with the alfalfa stem cell wall proteome (Watson et al., 2004; Verdonk et al., 2012) were tested and compared with a third, hybrid protocol. In this latter protocol, adapted from Verdonk et al. (2012) and Feiz et al. (2006), a sequential three-step extraction based on low ionic strength buffers with additional CaCl2, EGTA, and LiCl is performed. The three protocols are compared by a direct analysis of the digested proteins using LC-MS/MS. Two-Dimensional electrophoresis were further carried out on the extracts from the hybrid protocol to highlight how this technique can complement LC-MS/MS analysis of plant CWPs.

# Methods

### Plant Material

Alfalfa stems (Medicago sativa L.) were harvested from a local field (49◦ 33′ 39.1′′N, 5◦ 41′ 38.0′′E, Musson, Belgium) in early spring 2014. After removal of the leaves, stems were ground to a homogeneous powder in a mortar filled with liquid nitrogen. In this study, about 30 g of field-grown fresh alfalfa stems were frozen in liquid nitrogen, ground, and divided in 3 × 2 (protocols × replicates) samples of about 5.0 g and stored at −80◦C prior to analysis.

# Cell Wall Protein Extraction

CWPs were extracted as described in Watson et al. (2004) (protocol 1) and in (Verdonk et al., 2012) (protocol 2). A third hybrid protocol adapted from Feiz et al. (2006) and Verdonk et al. (2012) was established (protocol 3) to analyze the role of the chelating agent EGTA used by Verdonk et al. (2012). Each extraction procedure was carried out in 2 replicates and performed as summarized in **Figure 1**. Minor modifications were done to the previously described protocols, these were done to focus on the differences induced by the extraction and less on differences in the first steps of the sample preparation. Another adaptation done was the use of the same, 2D-DIGE compatible, buffer for the final resolubilization of the extracted proteins.

## Protocol 1 (Adapted from Watson et al., 2004)

The CWPs isolation and extraction were adapted from Watson et al. (2004), Modifications include the absence of PVPP in the grinding buffer (Watson and Sumner, 2007), the filtering of the plant material through 30µm2 pore size nylon mesh and the resuspension of the proteins in 7 M urea, 2 M thiourea, 2% w/v CHAPS, 30 mM Tris. The plant material was placed in a 50 ml Falcon <sup>R</sup> tube with 10 ml of buffer (50 mM Na acetate, 50 mM NaCl, and 30 mM ascorbic acid, pH 5.5, 4◦C). After vigorous shaking (24 Hz, 2 min), the slurry was filtered through a nylon mesh membrane (30µm<sup>2</sup> pore size) under vacuum and washed sequentially with (a) 100 mL of 50 mM Na acetate, 50 mM NaCl, and 30 mM ascorbic acid, pH 5.5, 4◦C, (b) 50 mL of NaCl (0.1 M, 4◦C), (c) 100 mL of cold water (4◦C), (d) 250 mL of cold acetone (4◦C), and (e) 100 mL of cold water. The retentates were transferred to 30 ml tubes prior to protein extraction.

Cell wall protein extraction was carried out by resuspending the retentates in 7.5 mL of extraction buffer 1 (50 mM Na acetate, 200 mM CaCl2, pH 5.5, 4◦C). Samples were placed on a rocking platform (30 min, 4◦C), centrifuged (10,000 g, 15 min, 4◦C) and the supernatants were saved. The pellets were re-extracted once with extraction buffer 1 and both supernatants of the same sample were pooled to form the CaCl<sup>2</sup> fraction.

The pellets were resuspended in 15 mL of extraction buffer 2 (50 mM Na acetate, 3 M LiCl, pH 5.5, 4◦C), placed on a rocking platform (overnight, 4◦C) and centrifuged (10,000 g, 15 min, 4 ◦C). The supernatants were saved, forming the LiCl fraction of the samples.

### Protocol 2 (Adapted from Verdonk et al., 2012)

Cell wall proteins were extracted as presented in Verdonk et al. (2012) with the following adaptations. Only 5 g of fresh matter were used for extraction, proteins were precipitated and washed using ReadyPrep™ 2-D Cleanup Kit (Bio-Rad) and proteins were resuspended in 7 M urea, 2 M thiourea, 2% w/v CHAPS, 30 mM Tris.

Briefly, the plant material was placed in a 50 ml Falcon <sup>R</sup> tube with 20 mL of buffer A (5 mM Na acetate, 0.4 M sucrose, pH 4.6, 4◦C), shaken vigorously (24 Hz, 2 min) and placed on a rocking platform (overnight, 4◦C). Samples were then centrifuged (1000 g, 15 min, 4◦C) and supernatants were discarded. Both pellets were resuspended in 10 mL of buffer B (5 mM Na acetate, 0.6 M sucrose, pH 4.6, 4◦C) and placed on a rocking platform (30 min, 4◦C) and centrifuged again (1000 g, 15 min, 4◦C). Supernatants were discarded. This washing step was repeated respectively with buffer C (5 mM Na acetate, 1 M sucrose, pH 4.6, 4 ◦C) and twice with buffer D (5 mM Na acetate, pH 4.6, 4◦C). The isolated cell wall fractions (pellet) were then transferred to 30 mL tubes.

Proteins were extracted with 10 mL of extraction buffer 3 (5 mM Na acetate, 50 mM EGTA, pH 4.6) and samples were shaken vigorously at 37◦C for 1 h. After centrifugation (10,000 g, 15 min, 4◦C), supernatants were saved. This extraction step was repeated twice and all supernatants were pooled, leading to the EGTA fraction.

The remaining pellet was resuspended in 10 mL of extraction buffer 4 (5 mM Na acetate, 3 M LiCl, pH 4.6, 4◦C), placed on a rocking platform (overnight, 4◦C) and centrifuged (10,000 g, 15 min, 4◦C). Supernatants were saved and pellets were reextracted twice using the same procedure with a shaking step lasting at least 8 h. All supernatants were pooled, leading to the LiCl fraction.

## Protocol 3 (Adapted from Feiz et al., 2006 and Verdonk et al., 2012)

The isolation of the cell wall fraction was carried out using sequential washes in increased sucrose concentration as described in protocol 2.

The proteins from the isolated cell wall fraction (pellet) were extracted with 7.5 mL of extraction buffer 5 (5 mM Na acetate, 200 mM CaCl2, pH 4.6, 4◦C) and placed on a rocking platform (30 min, 4◦C). Samples were then centrifuged (10,000 g, 15 min, 4 ◦C) and supernatants saved. This step was repeated once and supernatants were pooled, leading to the CaCl<sup>2</sup> fraction.

Proteins were further extracted with 10 mL of extraction buffer 3 (5 mM Na acetate, 50 mM EGTA, pH 4.6) and shaken vigorously at 37◦C for 1 h. After centrifugation (10,000 g, 15 min, 4 ◦C), supernatants were saved. This extraction step was repeated twice and supernatants were pooled leading to the EGTA fraction.

The remaining pellet was finally resuspended in 15 mL of extraction buffer 4 (5 mM Na acetate, 3 M LiCl, pH 4.6, 4◦C), placed on a rocking platform (overnight, 4◦C) and centrifuged (10,000 g, 15 min, 4◦C). Supernatants were saved, forming the LiCl fraction.

### Concentration and Desalting of the Extracts

Each cell wall enriched protein fraction was concentrated by using an Amicon Ultra-15 10K Centrifugal Filter Device (Millipore) and centrifuged (4700 g, 4◦C) until reaching a final volume of approximately 200µL. Proteins were further washed and desalted with a ReadyPrep 2-D Cleanup Kit (Bio-Rad) according to manufacturer instructions. After drying, proteins were solubilized in 100µL labeling buffer (7 M urea, 2 M thiourea, 2% w/v CHAPS, 30 mM Tris) and protein concentrations were assessed by using the Bradford protein assay with BSA as standard (Bradford, 1976).

### Protein Analysis SDS-Page

The reproducibility of the extractions was assessed by SDS-PAGE. Proteins, 20µg of each sample (2 replicates by fraction) were loaded on Criterion™ XT precast 1D gel 12% Bis-Tris, 12 + 2 wells, 45µL, 1.0 mm (Bio-Rad) according to manufacturer's instructions. Proteins were allowed to migrate for 1 h at a constant voltage of 200 V. After migration, gels were stained for 45 min with 100 ml of InstantBlue solution (Expedeon). Gels were rinsed twice with deionized water and scanned using a Typhoon FLA 9500 scanner (GE Healthcare).

## Focus on the Hybrid Three-Steps Protocol Using 2-Dimensional Gel Electrophoresis Protein Separation

CWPs, 50µg per fraction (1 replicate), were mixed with 9µL Servalyte, pH 3-10 (Serva Electrophoresis GmbH) and 2.7µL of Destreak Reagent (GE Healthcare) and volumes were completed to 450µL with lysis buffer (7 M urea, 2 M thiourea, 0.5% (w/v) CHAPS). Samples were loaded onto Immobiline™ DryStrip 3-10 NL, 24 cm (GE Healthcare) during overnight rehydration.

Isoelectric focusing was carried out in a five step-program: (1) constant 100 V for 3 h, (2) linear gradient from 100 to 1000 V for 4 h, (3) constant 1000 V for 6 h, (4) linear gradient from 1000 to 10,000 V for 6 h, and (5) constant 10,000 V until reaching a total of 95,000 Vh. During IEF, the current was limited to 75µA per strip.

Strips were then equilibrated 15 min in equilibration buffer (Serva Electrophoresis GmbH) complemented with 6 M Urea and 1% w/v DTT and further 15 min in equilibration buffer complemented with 6 M Urea and 2.5% w/v IAA. Strips were loaded on 2D-HPE™ Large-Gels NF 12.5% (Serva Electrophoresis GmbH) and electrophoresis was carried out using an HPE™ Tower System according to manufacturer's instructions. After the front reached the bottom of the gel, the gels were placed in fixation solution containing 15% v/v ethanol complemented with 1% (m/v) of citric acid. Gels were subsequently placed for 90 min in a LavaPurple (Serva Electrophoresis GmbH) staining solution (0.005% v/v) containing 100 mM NaOH, 100 mM boric acid. After staining, gels were washed twice with 15% EtOH for 15 min, acidified again for 15 min in fixation solution and rehydrated in deionized water. Gels were subsequently scanned at 473 nm using a Typhoon FLA 9500 scanner (GE-Healthcare), and quantitative analysis was carried out using the DeCyder software (v7.0, GE-Healthcare).

### Protein Digestion and Analysis

Following spot detection, all visible spots of each of the 3 gels were stored in a picklist and picked with an Ettan Spotted Picker (GE Healthcare). Digestion and MALDI spotting were carried out using a Freedom EVO II workstation (Tecan). Briefly, gels plugs were washed for 20 min in a 50 mM ammonium bicarbonate solution in 50% v/v MeOH/MQ water and dehydrated for 20 min with 75% ACN. After dehydration, proteins were digested with trypsin Gold (Promega), 8µl of a solution containing 5 ng/µL trypsin in 20 mM ammonium bicarbonate (overnight, 37◦C). After digestion, peptides were extracted from the gel plugs with 50% v/v ACN containing 0.1% v/v TFA and dried. Peptides were then solubilized in 2µL of 50% v/v ACN containing 0.1% v/v TFA and 0.7µL was spotted on MALDI-TOF targets. A volume of 0.7µL of 7 mg/mL αcyano-4-hydroxycinnamic acid in 50% v/v ACN containing 0.1% v/v TFA was added. A MALDI peptide mass spectrum was acquired using the AB Sciex 5800 TOF/TOF (AB Sciex), and the 10 most abundant peaks, excluding known contaminants, were automatically selected and fragmented. MS analyses were carried out as described by Printz et al. (2013). MS and MS/MS spectra were submitted for NCBInr database-dependent identification using the taxonomy viridiplantae (http://www.ncbi. nlm.nih.gov) downloaded on September 23, 2013 and containing 32,770,904 sequences on an in-house MASCOT server (Matrix Science, www.matrixscience.com). A second search was carried out against an EST fabacea database downloaded on December 17, 2013 and containing 19,932,450 sequences. The parameters used for these searches were mass tolerance MS 100 ppm, mass tolerance MS/MS 0.75 Da, fixed modifications cysteine carbamidomethylation, and variable modifications methionine oxidation, double oxidation of tryptophan, and tryptophan to kynurenine. Proteins were considered as identified when at least two peptides passed the MASCOTcalculated 0.05 threshold scores (respectively a score of 50 for all NCBI viridiplantae queries and 57 for the EST fabacea queries).

# Liquid Chromatography Protein Digestion

The digestion of proteins was performed using Amicon Ultra-4 10K Centrifugal Filter Devices (Millipore) (Abdallah et al., 2012). CWPs, 25µg per fraction (1 replicate), were reduced for 20 min in 200µL 10 mM DTT dissolved in 100 mM ammonium bicarbonate. After centrifugation, (30 min, 4700 g, 4◦C) the sample was washed with 200µL of 100 mM ammonium bicarbonate and again centrifuged. The reduced proteins (at the top of the filter) were alkylated with 50 mM iodoacetamide in 100 µL of 100 mM ammonium bicarbonate for 30 min in the dark and after centrifugation washed twice with 100µL 100 mM ammonium bicarbonate. After the last centrifugation, 50µL of trypsin Gold (Promega), 5 ng/mL trypsin in 50 mM ammonium bicarbonate, was added and the filter device incubated overnight at 37◦C. Following digestion, 100µL of deionized H2O were added, filter devices were centrifuged (40 min, 4700 g, 4◦C), and the peptides collected at the bottom of the tube. The mixture of peptides was dried under vacuum and solubilized in 45µL of a solution containing 2% v/v ACN and 0.1% v/v formic acid.

### Peptide Separation and Analysis

A volume of 5µL of the extracted peptides were desalted and separated by reverse phase separation using an Eksigent nano 1DLC (AB Sciex) coupled with a LTQ-OrbiTrap XL mass spectrometer (Thermo scientific) operated with Xcalibur (2.0.7 SP1). Peptide desalting was carried out on C18 OMIX tips (100µl, Agilent Technologies) and separation was carried out at a flow rate of 400 nl.min−<sup>1</sup> on a Peptide ES-C18 column (15×0.1 mm, 2.7µm; Sigma-Aldrich) using a linear binary gradient (solvent A: 0.1% formic acid (FA); solvent B: 80% ACN 0.1% FA). MS and MS/MS analyses were performed online, in data-dependent mode with automatic switching between MS and MS/MS. Full scan MS spectra (300–1500 m/z) were acquired at 30,000 (m/z 400) resolution. Internal mass calibration was performed using Cyclomethicone (m/z 371.101230) as lock mass. Dynamic exclusion was enabled with exclusion size list of 500 and exclusion duration of 90 s. The eight most intense precursors were selected for subsequent fragmentation with normalized collision energy of 35%. Fragmentation spectra were acquired in the ion trap with an isolation window of 2.0 m/z, a target value of 1000, an activation Q of 0.25 and an activation time of 30 ms.

CID spectra were processed in an in-house Mascot server (Version 2.1, Matrix Science, www.matrixscience.com, London, UK) using Proteome Discoverer (version 1.4.0.288, Thermo scientific) by searching against the NCBInr database using the taxonomy viridiplantae (http://www.ncbi.nlm.nih.gov) downloaded on June 06, 2014 and containing 40,910,947 sequences. The searches were performed with the following parameters: used enzyme: trypsin, 2 missed cleavages, mass accuracy precursor: 10 ppm, mass accuracy fragments: 0.8 Da, fixed modifications: Carbamidomethyl (C), dynamic modifications: Dioxidation (W), Gln->pyro-Glu (N-term Q), Glu->pyro-Glu (N-term E), Oxidation (HW), Trp-> Kynurenin (W). Identifications were filtered using the following settings; high peptide confidence (minimum confidence: 95%, peptide decoy database search: Target FDR (Strict): 0.01; Target FDR (Relaxed): 0.05, Validation based on: q-Value), with minimum two peptides per protein. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Vizcaino et al., 2014) via the PRIDE partner repository with the dataset identifier PXD001927 and 10.6019/PXD001927. Finally, all identifications obtained based on the NCBI database were matched on the Medicago truncatula reference genome using an online platform available at http://plantgrn.noble.org/LegumeIP/v2/blasttranscript. jsp. Sequences were imported as FASTA sequences and blasted to the Mt4.0v1\_GenesCDSSeq\_20130731\_1800 database. Searches were performed with an E-value cut-off of 1e-04 and blast results were accepted in case more than 50% identity was reported.

## Protein Localization

Proteins were considered to be secreted in case the 2 servers SignalP 4.1 (http://www.cbs.dtu.dk/services/SignalP) and TargetP 1.1 (http://www.cbs.dtu.dk/services/TargetP) predicted the presence of a signal peptide cleavage site and an extracellular location with the standard search parameters.

# Results and Discussion

In 2006, Feiz et al. performed a comparative analysis of CWPs isolation protocols (Feiz et al., 2006). This comparison, based on published data, remains theoretical since all protocols were not tested in the same conditions, with a same analytical method and on a same initial plant material. So far two different protocols have been applied to study the cell wall proteome of alfalfa stems, both start with a purification of the cell wall followed by two steps of cell wall protein extraction (Watson et al., 2004; Verdonk et al., 2012). Watson et al. (2004) proposed the sequential use of 200 mM CaCl<sup>2</sup> and 3 M LiCl in 50 mM sodium acetate buffers and succeeded to limit the contamination with intracellular proteins to less than 50%. In the second strategy, described by Verdonk et al. (2012), CWPs are sequentially extracted with 50 mM EGTA and 3 M LiCl in 5 mM sodium acetate buffers and the percentage of proteins predicted to be targeted to the cell wall increased significantly to reach about 70%. Although, the sequential use of CaCl<sup>2</sup> and LiCl buffers to extract CWPs and glycosylated proteins has been frequently reported (Irshad et al., 2008; Day et al., 2013; Calderan-Rodrigues et al., 2014), the use of a chelating agent as first extractor as proposed by Verdonk et al. is rarely depicted.

In our study, we compare these two previously published methods with a third "hybrid" protocol based on the sequential use of low ionic strength buffers (5 mM sodium acetate) complemented with 200 mM CaCl2, 50 mM EGTA, and 3 M LiCl respectively. This three-steps fractionation should first allow the release of the most loosely attached proteins by saturating the pectin-fraction with Ca2<sup>+</sup> ions. The subsequent use of the chelating agent EGTA, that exhibits a high affinity for Ca2<sup>+</sup> ions, loosens the pectin network and frees up proteins associated with it. Finally, the last extraction with a high concentration of LiCl should release proteins that are more tightly bound to the wall matrix (Verdonk et al., 2012). The isolation of the cell wall and washing were performed according to Watson et al. (2004) for protocol 1—or using the washes in different sucrose

### TABLE 1 | Amount of proteins extracted by fraction and by protocol.


Proteins were extracted from approximately 5 g of fresh alfalfa stem crushed into powder in liquid nitrogen. \*The mass is expressed in µg per g of fresh material.

concentrations proposed by Feiz et al. (2006) for the protocols 2 and 3. For each protocol and fraction, the total mass of proteins extracted per g of fresh weight is presented in **Table 1**.

As already described by Feiz et al. (2006), the use of NaCl salt in an early step of the protocol, as proposed by Watson et al. (2004), decreased the amount of proteins extracted and thus potentially the number of CWPs present in the extract. The difference in the amount of protein extracted in the two replicates of protocol 1 is remarkable, and in our opinion stems from the fact that the isolation of the cell wall fraction using a Büchner filter is more difficult to control compared to the use of sucrose washes. In comparison the washes with sucrose led to extract the highest amount of proteins. The mass of extracted proteins was however significantly higher for the three-step hybrid protocol. For each protocol, the SDS-PAGE profiles of the two replicates were compared (**Figure 2**). Both fractions (CaCl<sup>2</sup> and LiCl) of the protocol adapted from Watson et al. (2004) have a similar protein profile although the intensity of some gel bands varies between fractions (**Figure 2A**). In contrast, the use of EGTA followed by LiCl in protocol 2 (adapted from Verdonk et al., 2012) and 3 (adapted from Feiz et al., 2006 and Verdonk et al., 2012) results in distinct protein patterns between the fractions (**Figure 2B**). This indicates that the use of the chelating agent allows the extraction of set of proteins different from those extracted by CaCl2. The similarity

et al. (2004) (B) protocol 2, adapted from Verdonk et al. (2012) (C) protocol 3 (hybrid), adapted from Verdonk et al. (2012) and Feiz et al. (2006): each lane was loaded with 20µg of proteins on Criterion™ XT precast 1D gel 12% Bis-Tris (Bio-Rad). The gels were stained with Coomassie blue (InstantBlue, Expedeon).

between the replicate profiles indicates the reproducibility of the different extraction steps.

# LC-MS/MS Shotgun Analysis

To screen the efficiency of the protocols in extracting alfalfa stem CWPs 1 of the 2 replicates was selected, the extracts were digested with trypsin and analyzed by LC-MS/MS. The total number of identified proteins, all fractions taken together, was the highest (458) for samples treated with protocol 2, which uses EGTA followed by LiCl (Verdonk et al., 2012). For the hybrid protocol a total of 331 proteins were identified while protocol 1 allowed the identification of 106 proteins. The highest ratio "cell wall protein"/"total proteins" was obtained with the protocol 1, i.e., 83% of the proteins identified in the two fractions were predicted to be secreted. However, since the total number of identified proteins is low, only 88 (out of 106) potential CWPs were identified using this method. In contrast, when proteins were extracted according to protocol 2, the number of proteins predicted to target the secretory pathway reached 212 (out of 458), which is in the range of what was identified in the original paper of Verdonk et al. (188 out of 272). Compared to the specificity for CWPs described in the original publication, in the current dataset the number



Percentages are calculated respectively relatively to the total number of accessions identified in the study (247, column "Total"), or relatively to the number of accession found by fraction in each protocol [192 (Hybrid), 176 (adapted from Verdonk et al.) and 78 (adapted from Watson et al.), column "% Range"]. Only the minimal and the maximal values from this last calculation are presented.

of cytosolic contaminants was rather high; 54% of the identified proteins vs. 31% in the original study. The higher number of identified non-cell wall proteins in protocol 2 in comparison with the hybrid protocol is however not surprising. Indeed, compared to the other extraction buffers the amount of protein extracted with the EGTA-complemented buffer is relatively low (**Table1**, protocol 2 first step and protocol 3 second step). In addition, the bulk of the proteins identified in the EGTA fraction of protocol 2 are non-cell wall proteins. This, together with the fact that all analyses start with a defined amount of protein (5µl for LC-MS/MS analysis and 50µg for gel-based analysis), makes that non-cell wall proteins are more abundant in protocol 2 to allow a significant identification using mass spectrometry.

The use of the three-steps fractionation in the hybrid method led to a relative decrease of the number of identified non-CWPs to 30%. Altogether, analysis of the identified sequences with SignalP and TargetP designated 242 out of the 331 proteins identified in the fractions of the hybrid protocol (73.1%) as putative cell wall proteins; indicating that the latter protocol combines a good selectivity for cell-wall proteins with a high yield of extraction. Globally, 601 different NCBInr accessions were identified in this study, among which 322 were predicted to be cell wall located (Supplementary Table 1).

One shortcoming of the use of the NCBInr database is the redundancy, leading to an overestimation of the real number of CWPs identified. To circumvent this, all NCBInr gene identifiers were matched to the Mt4.0v1 Medicago truncatula reference genome. The 322 (NCBI gi) putative CWPs found in this study matched to 247 (∼77%) non-redundant M. truncatula Gene Accessions. The effectiveness of the different protocols was then determined by analyzing the overlap between the total number of identified CWPs (the set of 247 M. truncatula matched gene accessions) and the number of proteins identified in each of the protocols. The three-steps fractionation scored best in this comparison, allowing the identification of 192 of the 247 M.

truncatula Gene Accessions (77.7%). The protocol proposed by Verdonk et al. (2012) presented a similar result (176; ∼71.3%), whereas the protocol described by Watson et al. (2004) allowed only the identification of 31.2% (78) of the total set of nonredundant M. truncatula gene accessions found in this study (**Figure 3**). Surprisingly, more than 43% of the M. truncatula gene accessions were only identified in extracts from one of the 3 protocols, suggesting that combining different protocols to study a sample increases the number of identified CWPs (**Figure 3**).

From the LC-MS/MS analysis it is concluded that the protocols adapted from Verdonk et al. and the hybrid protocol, respectively protocols 2 and 3, have a higher efficiency in extracting alfalfa stem CWPs. **Table 2** and **Figure 4** show the functional classification of the cell wall proteins identified by LC-MS/MS analysis of the different extracts, the functional classification was done based on previously listed functional classes (Jamet et al., 2008). Classification was performed according to the list of domain hits proposed in the NCBI blast after amino acid sequence comparison (http://blast.ncbi.nlm.nih.gov/ Blast.cgi), from protein functional analysis using the web tool (http://www.ebi.ac.uk/Tools/pfa/iprscan5) and from published data. However, the class "defense" was added, although most of the proteins from this category may also be classified as "proteins with interaction domains" or "proteins acting on carbohydrates." Both protocols allowed the identification of more than 200 putatively-secreted proteins based on the NCBInr database search, most of them being classified in the functional classes of the oxido-reductases and in the class of proteins acting on carbohydrates (**Table 2**). The protocol adapted from Verdonk et al. however led to the identification of a higher proportion of non-CWPs.

### 2-DE Analysis

Given the lowest proportion of cellular contaminants and the highest yield of CWPs the three-steps fractionation using washings with sucrose and extraction with CaCl2, EGTA, and LiCl complemented buffers was selected for a more detailed proteomic characterization of the 3 fractions using 2D-PAGE.

The main benefit of this technique would be to visualize different isoforms of a protein in case these isoforms present variations in their isoelectric point or mass. Confirming the 1D-analysis, the 2D-profile of each fraction showed clear distinctions (**Figure 5**). The 172 spots detected in the CaCl<sup>2</sup> fraction are mainly located in the basic part of the gel. The sequential extractions using EGTA and LiCl complemented buffers allowed the detection of respectively, 207 and 59 spots. In particular, a supplemental acidic cluster of proteins is present in the EGTA fraction. The final cell wall protein fraction (LiCl) presents a limited number of spots, most of them being localized at pH ranging from about 5–8 and being not present in the 2 previous fractions.

For each fraction of the hybrid protocol, all spots were picked, the proteins digested and the peptides analyzed using MALDI-TOF-TOF (all the identified peptides are represented in Supplementary Tables 2a–d). A set of 194 NCBInr accessions were identified significantly, among which 186 originated from plants and 8 from fungi (Supplementary Table 2a and **Figure 6**). The identification of fungal proteins is not surprising since these are field-grown samples and the fungi from which proteins are identified are known pathogens of alfalfa. Interestingly, most of these fungal proteins (5 out of 8) were also predicted to carry a signal peptide which targets the protein to the secretory pathway. Although, some spots contained more than one protein, a unique and significant protein was identified in 87 out of 172 spots of the CaCl<sup>2</sup> fraction, 141 out of 207 in the EGTA fraction and 43 out of 59 in the LiCl fraction.

Confirming the results obtained with LC-MS/MS, a large majority of the accessions identified by MALDI TOF-TOF [74.8% (CaCl<sup>2</sup> fraction), 73.1% (EGTA fraction), and 88.9% (LiCl fraction)] was predicted to carry a signal peptide for targeting to the cell wall. As already mentioned when discussing the LC-MS/MS results, some proteins, notably pectin methylesterases (PME), may not have a classical signal peptide but only a putative transmembrane (TM) domain. In PME, the presence of a TM domain in absence of peptide signal might nonetheless be sufficient to target the protein to the cell wall (Pelloux et al., 2007). Other proteins without a predicted signal peptide identified in the current study may similarly be localized in the cell wall, suggesting that the proportion of CWPs might be underestimated (Albenne et al., 2013).

In contrast to the results obtained with LC, a higher proportion of the predicted cell wall proteins was specific to one of the fractions, for instance 57.1% of the proteins identified in the CaCl<sup>2</sup> fraction were only identified in this fraction. For the EGTA

2D-HPE™ Large-Gels NF 12.5% (Serva Electrophoresis GmbH). Proteins were post-stained with LavaPurple (Serva Electrophoresis GmbH).

and LiCl fractions this proportion is 54.5 and 43.3% respectively, while LC-MS/MS analysis gave 24.1, 22.6, and 35.6% respectively (**Table 3**). Such differences may be explained by the lower sensitivity of the 2DE-appraoches in detecting proteins that are of low abundance in the extract. This interpretation is confirmed by the significantly lower number of total proteins identified on the 2DE-gels (194 NCBInr identifiers), in comparison with the 331 NCBI identifiers obtained from the LC-MS/MS analysis.

TABLE 3 | Information about the cell wall proteins (CWPs) detected in the fractions of the hybrid protocol by 2D-electrophoresis.


However, one main benefit of 2D-electrophoresis resides in the possibility to differentiate in one analysis the behavior of various isoforms of the same protein. In each fraction, a panel of proteins was identified in separate spots (**Figure 7** and Supplementary Table 3). In the CaCl<sup>2</sup> fraction, the accession NCBI/gi:169147017, which corresponds to a putative thaumatinlike protein, was found in 5 different spots. The accession NCBI/gi:358348728 attributed to be an alpha-amylase/subtilisin inhibitor was significantly identified in 11 different spots on the gel of the EGTA fraction. In the LiCl extract, the peroxidase NCBI/gi:537317 was identified in 10 different spots than were visibly present in three groups with a different pI.

These groups were arbitrary named "Group A," "Group B," and "Group C" as presented in **Figure 7**. In terms of volumes, these spots in which this peroxidase was identified represent more than 55% of the total volume of all spots visible on the LiCl gels. Since the same protein was automatically assigned in all spots of the 3 groups, the MS and MS/MS spectra were checked, additional peptides were fragmented and manual de novo sequencing was performed to

TABLE 4 | Characterization of three groups of peroxidase identified in the LiCl fraction of the hybrid protocol.


Symbols "X," "−" and "+" refer to the intensity of the peaks of the peptides obtained following MALDI-TOF/TOF analysis. "X": absence of the peak; "−": low intensity of the peak; "+": high intensity of the peak. Modifications in peptide sequence are indicated in bold.

determine whether subtle sequence variations could be identified (**Table 4**). The peptide at 2457 Da corresponds to the peptide **NFDR**QGLDTTDLVALSGAHTIGR and is mainly observed in


TABLE 5 | MALDI-TOF/TOF identification of some glycosylation events detected in the fractions of the hybrid protocol.

"putative" is added in case the identification has been manually determined and not confirmed by database searching. Glycosylations are indicated in bold.

group C spots. In "Group B" and in "Group A," the peak at 2516 Da corresponds to the same region of the protein but with the sequence **SNFDK**QGLDTTDLVALSGAHTIGR. Similarly, the peptide at 1025 Da corresponds to the predicted N-terminus of the protein (after removal of the signal peptide), with the sequence QLDNSFY**R**, and is the only form present in the spots belonging to "Group C." In the "Groups A and B" the same region of the protein is observed as a peak at 997 Da with as sequence QLDNSFY**K**. The same N-terminus, but with a missed cleavage, is confirmed in the peptide at 2276 Da which is not observed in the spots of "Group C." A peptide at 2591 Da is shared between all groups but a higher signal was recorded for the spots of "group C." Part of the sequence of this peptide was manually determined as PTL**N**TTYLQTLR with a glycosylation on the Asn-residue. The MS/MS spectrum matches with the determined peptide sequence of a paucimannosidic-type N-glycan HexNac(Fuc)HexNacHex(Xyl)Man(3), a known plant glycosylation structure, on the asparagine (**Table 4** and Supplementary Figure 1). None of these differences between the spots explains the observed shift in pI, due to the lack of genome sequence for alfalfa it is furthermore not possible to hypothesize on different functions for peroxidases identified in the different groups.

Manual checking and de novo sequencing were similarly performed when high quality peptides spectrum resulted in low scores with the common MASCOT search parameters. In case putative posttranslational modifications were manually identified [HexNAc (N), HexNAc(1)Hex(1) (N), HexNAc(2)Hex(1) (N), . . . ], the MS/MS spectra was submitted again against the NCBInr databases by adding the putative glycosylation events as variable modification in the MASCOT search parameters. If the supposed form of glycosylation however was not listed in the MASCOT posttranslational modifications, the manual identification of the glycosylation was maintained and "putative" was added in the description (**Table 5**). In our study, the presence of HexNAc residues (with or without additional carbohydrates) on Asn residues was observed in each fraction. Glycosylated CWPs included expansins (or expansin-like), xyloglucan-specific endoglucanase inhibitor proteins, peroxidases, Kunitz-type trypsin inhibitor alpha chain, disease resistance response proteins, low homolog to polygalacturonase inhibitor, receptor-like kinase, and pectinesterase/pectinesterase inhibitor inhibitor 40-like (**Table 5**). Uncommon glycosylation events were additionally observed on kynurenine, an oxidation product of tryptophan, and on the primary amino group of some peptides (some examples are presented in **Table 5** and Supplementary Figure 2). Regarding the switch in molecular mass of 162 Da between the unglycosylated and the glycosylated form of the peptide, an addition of one hexose is likely to have occurred on these residues. There is however no obvious biological significance for this glycosylation in plants, suggesting that these events may result from technical artifacts occurring during the extraction/analysis procedure rather than resulting from an in vivo processing of the nascent proteins (Rayon et al., 1998).

The difficulty of resolving basic glycoproteins on 2D-gels has limited the use of this approach to routinely analyze the plant cell wall proteome (Minic et al., 2007; Irshad et al., 2008). It remains nonetheless a powerful approach to elucidate how members of a multigenic family can be differentially translated into proteins. Most interesting is the detection of post-translational modifications such as glycosylations. This is particularly important when studying cell wall proteomes, since N-linked glycosylation is the

# References


most prominent modification of secretory proteins (Aebi et al., 2010). In contrast, in LC-approaches proteins are digested prior their separation on column, which often limits the possibility to discern closely related proteins, certainly in a non-model crop such as alfalfa. In our study, more than 85% of all accessions detected to be present in the hybrid protocol were identified by shotgun LC-MS/MS, whereas only 51% were identified using gels, suggesting that shotgun LC-MS/MS should be favored in the mapping of cell wall proteomes, while detailed information on groups of proteins can subsequently be obtained using gel-based approaches.

The analysis of the 3 protocols tested here has highlighted the complementarity of the 3 methods of cell wall protein extraction. The number of proteins that were only identified with the technique developed by Watson et al. (2004) was however relatively low, suggesting that the two other protocols should be preferred when doing global cell wall proteome analyses. In terms of number of identified CWPs, the washes with different sucrose concentrations and the further extraction of the proteins in two- (EGTA-LiCl) or three-steps (CaCl2-EGTA-LiCl) gave similar results. The degree of purity of the wall fraction and the yield of cell wall protein extraction varies nonetheless according to the method of extraction. Globally, only the three-steps extraction combines a good purity of the wall fraction and a high yield of protein extraction, two characteristics that favor this protocol for biological studies since the amount of material is often limited. Finally, when the cell wall proteome is divided in 3 subproteomes, the complexity of the cell wall extracts is reduced which helps to detect low-abundant proteins.

# Acknowledgments

We are thankful to Dr. Lucien Hoffmann for the careful reading of the article. We acknowledge in particular Céline Leclercq, Sébastien Planchon, and Laurent Solinhac for their day-to-day technical support. The authors thank partial financial support obtained through the National Research Fund, FNR Project CANCAN C13/SR/5774202.

# Supplementary Material

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fpls.2015. 00237/abstract


cell suspension cultures. Proteomics 14, 738–749. doi: 10.1002/pmic.2013 00132


**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.

Copyright © 2015 Printz, Dos Santos Morais, Wienkoop, Sergeant, Lutts, Hausman and Renaut. 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.

# Phosphoproteomics technologies and applications in plant biology research

Jinna Li <sup>1</sup> , Cecilia Silva-Sanchez <sup>2</sup> , Tong Zhang<sup>3</sup> , Sixue Chen1, 2, 3 and Haiying Li <sup>1</sup> \*

<sup>1</sup> College of Life Sciences, Heilongjiang University, Harbin, China, <sup>2</sup> Proteomics and Mass Spectrometry, Interdisciplinary Center for Biotechnology Research, University of Florida, Gainesville, FL, USA, <sup>3</sup> Plant Molecular and Cellular Biology Program, Department of Biology, UF Genetics Institute, University of Florida, Gainesville, FL, USA

Protein phosphorylation has long been recognized as an essential mechanism to regulate many important processes of plant life. However, studies on phosphorylation mediated signaling events in plants are challenged with low stoichiometry and dynamic nature of phosphorylated proteins. Significant advances in mass spectrometry based phosphoproteomics have taken place in recent decade, including phosphoprotein/phosphopeptide enrichment, detection and quantification, and phosphorylation site localization. This review describes a variety of separation and enrichment methods for phosphoproteins and phosphopeptides, the applications of technological innovations in plant phosphoproteomics, and highlights significant achievement of phosphoproteomics in the areas of plant signal transduction, growth and development.

### Edited by:

Sabine Lüthje, University of Hamburg, Germany

### Reviewed by:

Christof Rampitsch, Agriculture and Agrifood Canada, Canada Mohammad-Zaman Nouri, Rice Research Institute of Iran in Mazandaran, Iran

### \*Correspondence:

Haiying Li, College of Life Sciences, Heilongjiang University, 74 Xuefu Rd, Harbin 150080, China lvzh3000@sina.cn

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 17 March 2015 Accepted: 27 May 2015 Published: 16 June 2015

### Citation:

Li J, Silva-Sanchez C, Zhang T, Chen S and Li H (2015) Phosphoproteomics technologies and applications in plant biology research. Front. Plant Sci. 6:430. doi: 10.3389/fpls.2015.00430 Keywords: phosphoproteomics, enrichment, quantification, phosphorylation site mapping, plant biology

# Introduction

Phosphorylation is one of the most important post-translational modifications (PTMs) of proteins (Pawson and Scott, 1997). Approximately one-third of the proteins are modified by phosphorylation (Hubbard and Cohen, 1993). The kinase mediated covalent addition of a phosphate group to serine, threonine, and tyrosine residues in eukaryotes, and other amino acids such as histidine, aspartate, glutamate, lysine, arginine, and cysteine in prokaryotes and the subsequent removal of the phosphate groups by protein phosphatases constitute important signaling and regulatory mechanisms in living organisms (Batalha et al., 2012). Reversible protein phosphorylation regulates a wide range of cellular processes such as transmembrane signaling, intracellular amplification of signals, and cell-cycle control. Protein phosphorylation often leads to protein structural changes that can directly modulate protein activity, and induce changes in interaction partners or subcellular localization (Jørgensen and Linding, 2008). The cascade of protein phosphorylation in a signaling pathway provides the backbone for complex signaling networks and regulatory processes in plant cells, including hormone sensing (Park et al., 2009), and environmental stress responses (Mishra et al., 2006). Thus, the analysis of signaling pathways in plants has often been focused on protein kinases. Traditional studies, however, described the phosphorylation of a single substrate by a particular kinase. Based on genome annotation, protein kinases were found to make up about 5.5% of the Arabidopsis genome (The Arabidopsis Genome Initiative, 2000), which is nearly twice as many as in the human genome (Manning et al., 2002). This indicates high specificity and complex networks of phosphorylation events in plants Li et al. Phosphoproteomics and applications

(Schulze, 2010). Many plant protein kinases have been identified to play essential roles in response to a variety of stresses including salt stress, cold stress, and pathogen invasion. Deciphering the molecular events occurring in stress responses will enhance our understanding of the biological processes in plants (De la Fuente van Bentem et al., 2006; Stecker et al., 2014).

The combination of phosphoprotein/phosphopeptide enrichment techniques, along with technological advancement in tandem mass spectrometry has been employed as a powerful tool to study protein phosphorylation and its biological relevance (Chen and White, 2004). In this review, a variety of separation and enrichment methods for phosphoproteins and phosphopeptides, their features as well as applications in phosphoproteomics research are described.

# Phosphoproteomics Technologies

Low stoichiometry of phosphorylated proteins and low ionization efficiency of phosphopeptide are two major challenges for protein phosphorylation detection. To reduce sample complexity, it is necessary to enrich the modified proteins and/or peptides before mass spectrometry (MS) analysis. Commonly used enrichment techniques were summarized in **Table 1**, with enrichment at the peptide level as a popular strategy. A successful phosphoproteomics study depends not only on the selective enrichment of phosphopeptides, but also on accurate detection and quantitation of the peptides, as well as precise mapping of the phosphorylation sites. Advances in these areas have been extensively reviewed (Batalha et al., 2012; Fíla and Honys, 2012; Kline and De Luca, 2014; Silva-Sanchez et al., 2015). Most of the technologies were developed in animal and yeast systems, and subsequently applied in plants. Here we briefly describe the advancement of phosphoproteomics technologies in plant research (**Table 2**).

## Enrichment Strategies

The most widely used enrichment method for phosphopeptides takes advantage of the affinity binding between negatively charged phosphate and positively charge metal ions (Fíla and Honys, 2012). Immobilized metal affinity chromatography (IMAC) is often coupled with strong cation exchange (SCX) for two-step phosphopeptide enrichment. For example, in a SCX-IMAC experiment, three times more phosphopeptides were identified when compared to the use of SCX or IMAC alone (Trinidad et al., 2006). The first reported SCX-IMAC application in plants resulted in identification of 283 phosphopeptides (Nuhse et al., 2004). In addition, Polymer-based Metal-ion Affinity Capture (PolyMAC) is a variant of IMAC, also showed high selectivity. For instance, employment of complementary PolyMAC-Titanium (Ti) and PolyMAC-Zirconium (Zr) ion affinity chromatography lead to identification of 5386 unique phosphopeptides (Wang et al., 2013a).

Metal dioxide especially titanium dioxides (TiO2) and zirconium dioxides (ZrO2) are gaining popularity for phosphopeptide enrichment. A comparison of the performance of TiO<sup>2</sup> and ZrO<sup>2</sup> performed with α-casein and β-casein as standard proteins showed that TiO<sup>2</sup> tends to enrich multiply phosphorylated peptides, and ZrO<sup>2</sup> tends to enrich singly phosphorylated peptides. A serial enrichment procedure with both TiO<sup>2</sup> and ZrO<sup>2</sup> can significantly increase the efficiency of capturing phosphopeptides in biological samples (Kweon and Håkansson, 2006; Gates et al., 2010). Metal dioxide enrichment could also be coupled with other peptide fractionation methods. For instance, a combination of TiO<sup>2</sup> enrichment and hydrophilic interaction liquid chromatography (HILIC) resulted in identification of 2305 phosphopeptides belonging to 964 proteins in wheat (Yang et al., 2013). Electrostatic repulsion hydrophilic interaction chromatography (ERLIC), a variation of HILIC that uses electrostatic repulsion as an additional chromatography stationary phase, had also been used successfully for selectively enrichment of phosphopeptides (Gan et al., 2008; Loroch et al., 2013).

# Quantitative Phosphoproteomics

Quantitative phosphoproteomics is aimed to enable a better understanding of phosphorylation regulated biological events. Comparative phosphoproteomics of wild-type and mutant plants or control and treated plants could be conducted in many ways. In general, the approaches can be grouped into gel-based, gelfree, stable isotope labeling, or label-free. Two-dimensional gel electrophoresis (2-DE) has been a widely used technology that resolves thousands of proteins by isoelectric point and molecular weight. Pro-Q Diamond is a fluorescent stain that provides a convenient method for selectively staining phosphoproteins in acrylamide gels. The result shows a global map of the modified proteins and their relative abundances compared to nonphosphorylated counterparts when a total protein staining is used after Pro-Q Diamond staining. Differentially phosphorylated proteins in wild-type and snk2.8 mutant Arabidopsis plants were analyzed using 2-DE and Pro-Q, and putative substrates of SnRK2.8 were identified (Shin et al., 2007).

Stable isotope labeling has been applied in plant phosphoproteomics successfully using a gel-free approach, for example, stable isotope labeling of amino acids in cell culture (SILAC).The first SILAC in plants was done by introducing <sup>15</sup>N in Arabidopsis suspension cells (Benschop et al., 2007) (**Table 2**). The methodology has been improved over the years and found more applications (Schütz et al., 2011; Stecker et al., 2014). Another labeling approach introduces multiplex isobaric tags to isolated proteins or digested peptides in vitro. Commonly used tags include isobaric tags for relative and absolute quantification (iTRAQ) and tandem mass tags (TMT). The tags are designed to be isobaric during MS and fragment to reveal differential low mass ion reporters during MS/MS. Due to its capability of multiplexing up to 10 samples in a single experiment and the enrichment effect for low abundance proteins, iTRAQ/TMT labeling has become popular in plant phosphoproteomics (Jones et al., 2006; Yang et al., 2013; Fan et al., 2014).

While both in vivo and in vitro label methods are limited by the number of samples, label free approaches enable quantitative phosphoproteomics of unlimited number of samples. There are two main methods in label free quantitation. The first is based on precursor ion peak intensity/area, and the second is based on the number of MS/MS spectra acquired for a

### TABLE 1 | Phosphopeptide/phosphotprotein enrichment methodologies.


given peptide (known as spectral counting). Both methods were used in plant phosphoproteomics (Reiland et al., 2011; Engelsberger and Schulze, 2012; Wang et al., 2013a). For instance, Reiland et al. (2011) characterized the function of a thylakoidassociated kinase STN8 in the fine-tuning of cyclic electron flow, which is regulated by the phosphorylation/dephosphorylation event.

In addition to these large scale discovery phosphoproteomics approaches, multiple reaction monitoring (MRM) has been used for quantification of targeted phosphopeptides (Glinski and Weckwerth, 2006; Schulze et al., 2012; Minkoff et al., 2015). A triple quadrupole is typically used for the MRM measurement, in which the first quadrupole (Q1) is set as a filter for the precursor ion with predetermined mass and Q3 is set to measure a specific fragment ion. The specific combination between a precursor ion and a fragment ion is called a transition and multiple transitions can be used to determine the relative and absolute (with synthesized peptide standards) levels of phosphopeptides (Schulze et al., 2012).


TABLE 2 | Representative plant phosphoproteomics work in the past decade.

LTQ, linear ion trap; VEMS, Virtual Expert Mass Spectrometrist; MRM, multiple reaction monitoring; TOF, time of flight; FT-ICR, Fourier transform ion cyclotron resonance. Please refer to the text for other abbreviations.

# Applications of Phosphoproteomics in Plant Biology Research

### Phosphoproteomics of Signal Transduction

Protein phosphorylation in signal transduction is an important area of current plant biology research. Many key proteins such as kinases, transcription factors, and ubiquitin ligases function through reversible protein phosphorylation in the signal transduction cascade (Hunter, 2000). In recent years, it has become apparent that analysis of signaling networks is required for the understanding of the dynamic and complex mechanisms underlying cellular functions and outputs. Most of the studies in plants have often been focused on protein kinases and identification of the phosphorylated substrates.

The mitogen-activated protein kinases (MAPKs) constitute one of the most important signaling mechanisms in plants, and they play essential roles in linking the perception of different stimuli with cellular adaptive responses. The MAPK signal transduction pathways are evolutionarily conserved in all eukaryotic organisms such as plants, yeast, fungi, insects, nematodes, and mammals (Mishra et al., 2006). A MAPK cascade is minimally composed of distinct combinations of at least three protein kinases: a MAPKKK (MAP3K), a MAPKK (MAP2K) and a MAPK, which activate in a sequential manner via phosphorylation (**Figure 1**). An activated MAPKKK firstly phosphorylates two serine and/or threonine residues (S/T-X3- 5-S/T) located within the activation loop of the MAPKK. Activated MAPKKs in turn trigger MAPK activation through dual phosphorylation of a highly conserved T-X-Y motif in the activation loop of MAPKs (Hamel et al., 2012). In a proteomic analysis of plasma membrane isolated from maize roots, four isoforms of Pto-interacting-like kinase 1 (PTI1) showed increased levels in response to low and high iron conditions (Hopff et al., 2013). Interestingly, a previous oxidative stress study in Arabidopsis demonstrated that interaction of a PTI1-like kinase (PTI1-4) with another serine/threonine protein kinase, oxidative signal-inducible 1 (OX1), mediates oxidative stress signaling. In addition, PTI1-4 was found to interact with MPK6 in the same protein complex (Forzani et al., 2011). These results imply that the PTI signals may function through the OXI1 and MPK6 signaling cascades. Recently, Hoehenwarter et al. (2013) developed a two-step chromatography combining phosphoprotein enrichment using Al(OH)3-based metal oxide affinity chromatography with phosphopeptide enrichment using TiO2-based metal oxide affinity chromatography to enrich phosphopeptides from complex A. thaliana protein samples. The

method was successfully applied to identify MAPK substrates. A large number of novel phosphorylation sites and 141 MAPK substrate candidates (mostly novel) have been identified. For example, time for coffee (TIC) and non-phototropic hypocotyl 3 (NPH3), which are involved in circadian clock and phototropism, were found to be MPK3/6 substrates. The result suggests that plant circadian rhythm and phototropism may be regulated by the MAPK signaling network.

Abscisic acid (ABA) is a phytohormone that plays an important role in many aspects of plant life. For example, ABA is essential for regulating seed maturation and stomatal closure under abiotic and biotic stresses (Hubbard et al., 2010). Protein phosphorylation and dephosphorylation play a central role in ABA signaling. Multiple signaling components have been found to undergo phosphorylation/dephosphorylation regulation to control stomatal movement in response to ABA (Zhang et al., 2014). A simplified ABA signaling model consists of the soluble ABA receptors (members of the pyrabactin resistance 1 (PYR1) and PYR1-like (PYL) proteins, also known as regulatory component of ABA receptor (RCAR) family, and collectively referred to as PYR/PYL/RCAR), a subgroup of type 2C protein phosphatases (PP2Cs), and the SNF1-related protein kinase 2 (SnRK2) family (Umezawa et al., 2010). Umezawa et al. (2013) studied protein phosphorylation networks in ABA signaling using phosphoproteomics of Arabidopsis treated with ABA and dehydration stress, as well as snrk2 mutants to identify SnRK2-dependent protein components. Comparative analysis between ABA treatment and dehydration stress revealed that dehydration stress induced multiple protein phosphorylation pathways in addition to the ABA-dependent pathway, supporting that multiple protein kinases are involved in dehydration stress signaling, including SnRK2s, MAPKs, and calcium-dependent protein kinases (CDPKs) (Umezawa et al., 2013). Further studies will be required for understanding how multiple kinases mediate dehydration stress signaling. It appeared that subclass III SnRK2s may be uniquely employed during ABA responses, and subclass II SnRK2s are the main subclass that regulates dehydration stress responses, although they are also activated by ABA. By integration of genetics with phosphoproteomics, it is possible to connect protein kinases with their in vivo signaling pathways. In particular, this study provided insights into the ABA signaling pathway by identifying potential substrate proteins of SnRK2s (Umezawa et al., 2013).

# Phosphoproteomics of Subcellular Compartments

Phosphoproteomics studies were often performed in a shotgun fashion, with the identification of hundreds and thousands of proteins that lead to a very complicated set of phosphoproteins across subcellular compartments and organelles (**Table 2**), leading to a poor understanding of the networks that regulate the cellular activities (Jung et al., 2000). Compartmentalization in eukaryotes offers a practical approach to study subcellular phosphoproteomics networks, with a reduced population of identified proteins. There are about 3000 proteins in the chloroplasts of Arabidopsis, but only four kinases were previously identified. It may be feasible to find specialized kinases or families of kinases that can potentially show differential activities in the chloroplasts (Millar and Taylor, 2014; van Wijk et al., 2014). A meta-analysis of 27 publications of phosphoproteomics data sets in Arabidopsis comprises 60,366 phosphopeptides matched to 8141 non-redundant proteins. The phosphoproteins showed predicted subcellular distribution in the following categories: nucleus, secretory (containing endoplasmic reticulum, Golgi, plasma membrane, cell wall, and vacuolar), cytosol, other/unknown, intra-plastid, mitochondria, and peroxisome (van Wijk et al., 2014). The study of phosphoprotein compartmentalization supports the hypothesis that a fine mechanism helps to maintain and regulate protein translation, post-translational metabolism, signaling, and trafficking through the cells (Millar and Taylor, 2014). Some studies have already started to focus on PTMs in subcellular compartments and here we describe a few examples.

Jones et al. (2009) performed a phosphoproteomic analysis of the nuclei-enriched fractions prepared from suspension cell cultures and seedlings of A. thaliana. The work led to the identification of 416 phosphopeptides from 345 proteins. Two thirds of the proteins are known or predicted to be nuclear localized, and one half of the nuclear localized proteins have novel phosphorylation sites. Many phosphorylation sites and kinase motifs were identified on proteins involved in nuclear transport (e.g., Ran-associated proteins), and on transcription factors, chromatin remodeling proteins, and spliceosome components. Surprisingly, many novel phosphopeptides from proteins involved in vesicle trafficking such as components of the exocyst complex (SEC10, SEC51, and SEC5a-like) were identified. How phosphorylation of these SEC proteins plays a role in vesicle trafficking is intriguing. Recently, phosphorylation of Sec31 by a casein kinase 2 was found to control the duration of COPII vesicle formation, decrease its association with ER and promote ER-to-Golgi trafficking (Koreishi et al., 2013).

Subcellular proteomics can address conserved mechanisms underlying plant responses to stresses. By analyzing the phosphorylation changes in proteins of microsomal fractions from A. thaliana and Oryza sativa, Chang et al. (2012) found similar phosphoproteins between the species including photosystem II reaction center protein H PsbH. Both Arabidopsis and rice showed an increased ratio for a diphosphorylated peptide (ApTQpTVEDSSRSGPR) of PsbH as a response to salt stress. Interestingly, the two phosphorylation sites (Thr2 and Thr4) are found to be evolutionarily conserved in many plants using sequence alignment.

Light plays a crucial role in the regulation of protein phosphorylation in photosynthetic thylakoid membranes. In Arabidopsis, the thylakoid Ser/Thr protein kinase 7 (STN7) and STN8 kinases are light regulated and participate in phosphorylation of thylakoid membrane proteins and stroma proteins. Ingelsson and Vener (2012) performed a thylakoid phosphoproteomics study using Arabidopsis wild-type and the STN mutants stn7, stn8, and stn7stn8. The results showed that STN7 is required for the phosphorylation of pTAC16 at the Thr451, and pTAC16 was found to be distributed between thylakoids and nucleoid. In addition, the results suggest that pTAC16 could anchor DNA to the thylakoid membrane, and it was proposed that STN7-dependent phosphorylation of pTAC16 may regulate membrane-anchoring functions of the nucleoid.

# Phosphoproteomics of Plant Growth and Development

Sucrose non-fermenting 1 related kinase (SnRK1) acts as a sensor of energy levels in plant development, and regulates plant growth by maintaining energy homeostasis during stress conditions (Tsai and Gazzarrini, 2014). It is activated by sugar depletion, energy depletion in the dark and hypoxia (Baena-González and Sheen, 2008). Trehalose 6-phosphate (T6P) is a signaling molecule involved in the regulation of embryonic and vegetative development, flowering time, and meristem determinacy. An increase in the levels of T6P led to metabolic changes that promote plant growth. However, T6P regulates SnRK1 by inhibiting its activity. SnRK1 and T6P are global regulatory molecules that also interact with plant hormones, and along with ABA modulate several crucial cellular activities such as seed maturation and germination, ABA sensitivity and signaling, vegetable growth, and flowering regulation (Tsai and Gazzarrini, 2014). Seed germination is known to be controlled by phytohormones, including gibberellins (GAs) and ABA, which play antagonistic roles as positive and negative regulators, respectively (Seo et al., 2006). Many protein kinases and phosphatases participate in ABA signaling to regulate seed germination. Recently, Han et al. (2014) used PolyMAC phosphopeptide enrichment and gel-free proteomics identified a total of 933 phosphorylated peptides corresponding to 413 proteins in rice embryos during early stages of germination. By quantitative normalization of phosphoprotein abundance and One-Way ANOVA testing, 149 phosphorylated proteins were found to be significantly changed in abundance during germination. Among the phosphoproteins, seven brassinosteroid (BR) signaling pathway-related proteins were identified and three (BR signaling kinase 1, BR-insensitive 2, and BR-insensitive 1 suppressor 1) showed significant increases in phosphoprotein abundance during the early stages of germination. In addition, treatment with brassinolide promoted the rice seed germination. These results suggest that brassinosteroid signal transduction plays an important role in triggering seed germination.

Plant vegetative growth is important for biomass accumulation and potential biofuel applications. A recent phosphoproteomic study of Brachypodium distachyon as a model biofuel plant using TiO<sup>2</sup> enrichment and LC-MS/MS has identified a total of 1470 phosphorylation sites in 950 phosphoproteins (Lv et al., 2014). Among the phosphoproteins, there were 58 transcription factors, 84 protein kinases, 8 protein phosphatases, and 6 cellulose synthases. Through bioinformatic analysis, a protein kinase and phosphatase centered network related to rapid vegetative growth was deciphered. For example, a MAPK signaling cascade might play an important role in leaf growth and development (Lv et al., 2014). This finding is very interesting, considering MAPK cascade is generally involved in plant stress responses (Mishra et al., 2006).

# Identification and Functional Analysis of Novel Phosphorylation Sites

The identification of protein phosphorylation sites has been difficult in the past. Nowadays, high throughput modern technologies such as tandem MS have promoted large-scale discoveries of new phosphorylation sites and phosphoproteins in recent years. Rao and Møller (2012) initiated a large-scale study of phosphorylation site occupancy in eukaryotic proteins. They analyzed the occurrence and occupancy of phosphorylation sites in a large number of eukaryotic proteins, and provided insights into protein phosphorylation and related processes. Phosphorylation probability was found to be much higher in both termini of protein sequences (much more in the C-terminus) than middle parts of the sequences. A large proportion (51.3%) of the occupied sites had a nearby phosphorylation within a distance of 10 amino acid residues. This proportion is very high compared to the expected value of 16.9%. More than half of the phosphorylated sites fall within a small number of motifs.

A large phosphoprotein, the RNA surveillance protein UPframeshift 1 (Upf1) in Saccharomyces cerevisiae, has only been partially characterized for phosphorylation sites, but the functional relevance of the phosphorylation has not been studied. Lasalde et al. (2014) used tandem MS and in vitro phosphorylation assays to identify novel phosphorylation sites in UPF1. A total of 11 phosphorylated residues of UPF1 were identified. Sequence alignment of UFP1 from lower and higher eukaryotes showed complete conservation of the phosphorylated residue Y-754. Residues corresponding to S. cerevisiae UPF1 T-194, S-492, Y-738, and S-748 were similar to those in the homologs of Homo sapiens, Mus musculus, Drosophila melanogaster, and A. thaliana. Since the phosphorylated residues in UPF1 were clustered in four small regions, each one was tested to determine its importance by independently deleting the four individual regions. The deletion mutant lacking phospho-motif-4 was not able to complement the Nonsense-Mediated mRNA Decay (NMD) defect as revealed by Northern blot analysis. To test the role of phospho-motif-4 in translation termination efficiency, a well-established dual luciferase assay was used. The deletion-mutant lacking phospho-motif-4 was not able to rescue this defect, indicating that this motif has a role in translation termination efficiency. To dissect the sequences within phosphomotif-4 required for NMD activity, PCR-mediated mutagenesis was used to generate three additional deletion mutants (736–745, 746–750, 751–751). The results revealed that deletion of residues 736–745 reduced NMD activity as measured by Northern blot analysis. To test the functional role of Y738 and Y742, sitedirected mutagenesis was used to create phosphorylation mimic mutants. Interestingly, the Y738F and Y742F fully rescued NMD activity of a chromosomal UPF1 deletion-mutant strain, indicating that they are not compromised in their ability to function in NMD. These results provided strong evidence that UPF1's ability to promote translational termination fidelity is depended on the conserved C-terminal phosphorylation motif, which is important for its NMD activity.

## Concluding Remarks

Phosphopeptide enrichment and MS have been essential tools for studying protein phosphorylation. It is challenging to directly detect phosphoproteins in biological samples due to the low abundance and low stoichiometry of phosphorylation in different biological processes. The enrichment methods of phosphoproteins/phosphopeptides from complex mixtures have

# References


greatly improved over the years. For instance, combining the titanium (Ti4+)-based IMAC and the reverse phase (RP) strong cation exchange (RP-SCX) biphasic trap column-based online RPLC is a great example of the advancements (Bian et al., 2012; Wang et al., 2013b). The recent development of specific labeling techniques has greatly aided the quantification of phosphorylation profiles and their stress-induced changes. Especially, iTRAQ and TMT in vitro labeling and SILAC in vivo labeling have shown to be successful in combination with IMAC and MS (Isner et al., 2012; Yang et al., 2013; Zhang et al., 2013; Stecker et al., 2014). These studies have revealed novel nodes and edges in signaling pathways and regulatory processes that are dependent on phosphorylation. Despite many new insights gained from quantitative phosphoproteomics, improvements are required to enable a comprehensive description of total and PTM proteomes. Currently, LC-MS/MS based phosphoproteomic technologies have established as an indispensable tool in identification of novel phosphorylation sites and signaling pathways. As large data sets accumulate, informatics tools will be indispensable, e.g., informatics has revealed phosphorylation probability to be frequent at the termini of protein sequences. Taken together, researchers have provided not only new insights into the complex phosphorylation regulatory networks in plants, but also important resources for future functional studies of protein phosphorylation in plant growth and development.

# Acknowledgments

Research in the HL lab was supported by the National Science Foundation of China (Project 31471552: The response of antioxidant enzymes to salt stress in sugar beet M14, and Project 31401441: Identification of root variation related proteins in sugar beet (Beta vulgaris L.) monosomic addition line M14 using iTRAQ analysis), the National Science Foundation of Heilongjiang Province (Project C201202: Comparative proteomics analysis of sugar beet M14 under salt stress), and the Common College Science and Technology Innovation Team of Heilongjiang Province. The paper represents serial 016 from our innovation team at the Heilongjiang University (Hdtd2010-05).


is a common target of the oxidative signal-inducible 1 and mitogen-activated protein kinases. FEBS J. 278, 1126–1136. doi: 10.1111/j.1742-4658.2011.08033.x


**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.

Copyright © 2015 Li, Silva-Sanchez, Zhang, Chen and Li. 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.

# Advantages and limitations of shot-gun proteomic analyses on Arabidopsis plants with altered MAPK signaling

# *Tomáš Takác and Jozef Šamaj ˇ \**

*Department of Cell Biology, Faculty of Science, Centre of the Region Haná for Biotechnological and Agricultural Research, Palacký University, Olomouc, Czech Republic*

*\*Correspondence: jozef.samaj@upol.cz*

### *Edited by:*

*Joshua L. Heazlewood, The University of Melbourne, Australia*

### *Reviewed by:*

*Justin Lee, Leibniz Institute of Plant Biochemistry, Germany Gerold J. M. Beckers, RWTH Aachen University, Germany*

**Keywords: mitogen-activated protein kinase (MAPK), signaling, proteomics, phosphoproteomics, shot-gun proteomics, validation**

# **INTRODUCTION**

The molecular mechanism of signal transduction by mitogen-activated protein kinases (MAPKs) in plants is an intensive research field of contemporary plant biology. Signaling in plants involves perception of changing environment and the intracellular transduction of such generated information leading to the activation of adaptive mechanisms or programmed cell death, as it is in the case of hypersensitive response to some biotrophic pathogen attack. At the molecular level, distinct signaling pathways, such as MAPK modules, involve reversible phosphorylation, enzymatic activation, protein-protein interaction and finally transcriptional activation via nuclear-resident transcription factors or cell reorganization via regulation of cytoplasmic or plasma membraneassociated proteins. All these processes are strictly regulated and subcellularly compartmentalized.

Modern OMICS technologies encompassing transcriptomic (Krysan et al., 2002; Frei dit Frey et al., 2014), metabolomic (Lassowskat et al., 2014), proteomic (Miles et al., 2009; Conroy et al., 2013; Lassowskat et al., 2014; Takác et al., 2014 ˇ ) and phosphoproteomic (Hoehenwarter et al., 2013; Lassowskat et al., 2014) approaches have been employed to gain functional information on plant signaling. The main challenges of advanced (phospho)proteomic studies are: reliable identification of all signaling constituents of a particular MAPK pathway as well as identification and validation of respective target proteins. In this opinion article, we aim to briefly evaluate benefits and pitfalls of various proteomic approaches that are used to study MAPK signaling in plants. We emphasize on the advantages of comparative shot-gun proteomic analyses of Arabidopsis mutants and transgenic plants with altered MAPK signaling. The importance of critical data validation and interpretation is also discussed.

### **PROTEOMIC APPROACHES ON MAPK SIGNALING**

Modern proteomics offers a variety of approaches to study cell signaling events in living organisms (Jørgensen and Locard-Paulet, 2012). They have been quite rarely used also for investigation of molecular mechanisms in plant MAPK signaling (**Table 1**). Nevertheless, these studies substantially helped to better describe complex processes accompanied with signaling and to find new kinase targets regulated by reversible phosphorylation.

Targeted phosphoproteomics is a powerful tool for the identification of putative downstream substrates of MAPKs, but also for mapping of phosphorylation sites detected upon activation or inactivation of particular MAPK pathways (de la Fuente van Bentem et al., 2008; Hoehenwarter et al., 2013; Lassowskat et al., 2014). Diverse phosphoproteomic approaches have been applied, mostly employing genetically modified kinase expression systems. For example, a recent phosphoproteomic study of transgenic *Arabidopsis thaliana* plants harboring a gene encoding a constitutively active *MEKDD* from *Nicotiana tabacum* (NtMEK2*DD*) expressed under the control of the dexamethasone (DEX)-inducible GVG promoter identified 141 putative MAPK substrates with the help of complementary enrichment of phosphoproteins and consecutive phosphopeptide enrichment (Hoehenwarter et al., 2013). In a similar approach, early and late putative substrates of MPK3 and MPK6 were identified by phosphoproteomics performed on *Arabidopsis thaliana* plants transformed with a constitutively-active variant of *MKK5* from *Petroselinum crispum*, expressed under the control of a DEXinducible promoter (Lassowskat et al., 2014). The expression of heterologous MAP2Ks in Arabidopsis might result in the identification of somehow unspecific or indirect targets. This problem might be solved by validation of putative MAPK substrates by both bioinformatic and independent analytical methods.

The phosphoproteomics following stable isotope labeling in Arabidopsis (SILIA) efficiently elucidated a protein complex containing ethylene receptor ETR1 and constitutive triple response 1 (CTR1) MAP3K (Yang et al., 2013). Using the loss of function ethylene response Arabidopsis mutant *ctr1-1*, novel CTR1 targets involved in ethylene response were identified including calcium-sensing receptor and plastidal transcriptionally active protein.

Previously, microarray containing 1690 proteins from Arabidopsis was generated and incubated with MAPKs in the presence of [γ−33P]ATP. Quantitative evaluation of radioactive signals identified 48 potential substrates of MPK3 and 39 substrates of MPK6 (Feilner et al., 2005).

### **Table 1 | Overview of proteomic approaches on MAPK signaling in plants.**



A more extended protein microarraybased approach for MAPK phosphorylation target analysis was published by Popescu and coworkers (Popescu et al., 2009). First, purified wild type MAPKs (MPK1–MPK8, MPK10, and MPK16) were tested by *in vitro* phosphorylation assays for their activation by certain MAPKKs co-expressed in *Nicothiana benthamiana* leaves. Subsequently, 2156 proteins were placed on protein microarray and probed by above MAPKs. This approach allowed the identification of new signaling modules composed of MAPKKs, MAPKs and their target proteins.

Peptide microarray analysis based on screening of hundreds of synthetic peptides containing experimentally verified phosphorylation sites for different types of human kinases was used to discriminate between *Solanum lycopersicum* LeMPK1, LeMPK2, and LeMPK3 (Stulemeijer et al., 2007). The incubation of peptide microarrays with recombinant LeMPK1, LeMPK2, and LeMPK3 in the presence of [γ−33P]ATP was followed by *in vitro* kinase assay which displayed both similar, but also different peptides phosphorylated by these MAPKs.

In our opinion, another promising tool for MAPK substrate identification was provided by recent study, which used a library of 377 synthetic peptides, representing previously identified phosphorylation sites in developing seeds of *Arabidopsis thaliana* and *Brassica napus,* for screening with purified kinases while the resulting phospho-sites were analyzed subsequently by mass spectrometry (Ahsan et al., 2013). It is of great value that all these diverse phosphoproteomic approaches helped to identify new putative targets of MAPKs. However, in our view they lack the capability of shotgun comparative proteomics to identify mid- and long-term complex protein profiles and networks associated with the transcriptional impact of deregulation of MAPK pathways, as this seems to be often the case in the mutant and overexpressor plant lines (e.g., Krysan et al., 2002). We consider this point quite important because changed complex protein profiles revealed by shot-gun proteomics might be closely linked to phenotypes of respective mutant and transgenic lines, including their characteristics such as modified stress responses or developmental defects.

### **BENEFITS AND WEAKNESSES OF COMPARATIVE SHOT-GUN PROTEOMICS ON ARABIDOPSIS MAPK MUTANTS AND TRANSGENIC OVEREXPRESSION PLANTS**

Complex molecular processes affected by deregulation of MAPK signaling cascades *in planta* are often focused on transcriptomic studies (Frei dit Frey et al., 2014) while global protein changes are still underestimated. Since conditionallymodulated transcript levels can be rectified at the protein level, it is of essential importance to perform proteomics rather than speculate on protein abundance based on transcriptomic data (Gygi et al., 1999). Shot-gun proteomic analyses on mutant and transgenic plants with genetically modified MAPKs and their upstream activators such as MAP2Ks and MAP3Ks provide a very useful proteome readout and quite unique information because they might not be fully consistent with transcriptomic data from the same plant lines (e.g., Krysan et al., 2002 cf. Takác et al., 2014 ˇ ). Few studies used shot-gun proteomic approaches to better explain stress-related and developmental phenotypes in Arabidopsis mutants and transgenic lines with altered MAPKs expression. One earlier study employing ICAT (isotope-coded affinity tag) quantitative proteomic analysis on *MPK6 RNAi* transgenic Arabidopsis plants unveiled molecular mechanisms accompanying response to ozone fumigation in relation to MPK6 downregulation (Miles et al., 2009). The redox response pathway was different between untreated plants with suppressed MPK6 expression and the corresponding control wild type plants. After ozone treatment, a more pronounced antioxidant defense was observed in plants with downregulated MPK6 (Miles et al., 2009). Recently, the two-dimensional LC MS/MS analysis of Arabidopsis double mutant deficient in two redundant MAP3Ks called ANP2 and ANP3 revealed constitutive upregulation of a protein functional network involved in oxidative stress response in the double mutant *anp2anp3* plants (Takác et al., ˇ 2014). This suggested an increased resistance of the *anp2anp3* mutant against oxidative stress which was corroborated by showing increased viability of this mutant growing on paraquatsupplemented medium. The activation of a protein network consisting of superoxide dismutase isoforms, chaperonin 20 and enzymes of the ascorbate-glutathione cycle was validated through detailed biochemical, physiological and histochemical analyses showing connection of proteomic data with resilience of the above mutant to oxidative stress. Importantly, this shot-gun proteomic study uncovered a previously unknown function of ANP2 and ANP3 in Arabidopsis antioxidant defense.

Next, we reported that the heterologous overexpression of alfalfa SIMKK-YFP leads to salt hypersensitivity of Arabidopsis seedlings. The proteomic analysis showed decreased levels of several key proteins involved in salt tolerance in the roots of these SIMKK-overexpressor transgenic plants (Ovecka et al., 2014 ˇ ). Thus, the overexpression of SIMKK repressed the levels of catalase, peroxiredoxin, glutathione S-transferase, nucleoside diphosphate kinase 1, endoplasmic reticulum luminal-binding protein 2, and finally those of plasma membrane aquaporins. Downregulation of these proteins most likely contributed to the increased salt sensitivity of transgenic Arabidopsis seedlings overexpressing SIMKK (Ovecka et al., ˇ 2014).

In addition to the analysis of stress responses, shot-gun proteomic analyses on mutant and transgenic plants might help to explain unique growth or developmental phenotypes regulated by members of particular MAPK pathway. Recently, we have found that both inactivation and constitutive activation of MAP3K called YODA caused pronounced root phenotypes in Arabidopsis (Smékalová et al., 2014). Analysis of endogenous auxin levels showed that these phenotypes might be related to increased IAA production in respective *yda1* and -*Nyda* mutants. Again, shot-gun proteomics provided insights to the molecular mechanism underlying auxin overproduction by detection of elevated levels of proteins involved in auxin biosynthesis such as tryptophan synthase in -*Nyda* and nitrilases in both mutants (Smékalová et al., 2014).

Finally, a recent elegant study used integrated metabolomics, shot-gun proteomics and phosphoproteomics approaches on leaves of Arabidopsis Col-0 ecotype as well as *mpk3* and *mpk6* mutants, all expressing DEX-inducible constitutively active *Petroselinum crispum* MKK5 (MKK5*DD*). Heterologous MKK5*DD* expression activated MPK3 and MPK6 at equal levels in Col-0 plants. The induction of MKK5*DD* in the *mpk6* mutant resulted in more robust proteome changes compared to *mpk3.* In this case, shot-gun proteomic data were in positive correlation with metabolomic as well as phosphoproteomic analyses (Lassowskat et al., 2014).

All these shot-gun proteomic studies helped to identify new molecular processes linked to MAPK pathways. This capability could be assigned to phosphoproteomics as well, as it was shown for example for nitrogen metabolism, control of circadian clock and phototropism (Hoehenwarter et al., 2013), and for biosynthesis of antimicrobial defense substances (Lassowskat et al., 2014).

Proteomics on total protein extract has technically limited capability for identification of all molecular mechanisms connected with particular MAPK pathway. These limitations arise from limited proteome resolution especially in terms of detection of low abundant proteins and inability of this approach to resolve phosphorylation events. Nevertheless, we consider these studies feasible in order to elucidate molecular networks composed from abundant proteins, including those involved in antioxidant defense (Miles et al., 2009; Takác et al., 2014 ˇ ) or stress response (Ovecka et al., 2014 ˇ ). In our view shot-gun proteomics can also provide important initial insights into new biological processes regulated by MAPKs, which should be later investigated by targeted application of genetic and cell biological methods. In ideal case, general shot-gun differential proteomics may be perhaps quite efficiently complemented by targeted phosphoproteomic (Lassowskat et al., 2014). In this respect, dual-enrichment techniques (Hoehenwarter et al., 2013; Lassowskat et al., 2014) might represent a crucial tool in order to overcome the limitation of the large dynamic range of protein abundance in shot-gun proteomic approaches.

### **EVALUATION AND VALIDATION OF PROTEOMIC DATA**

The complexity of proteomic analysis and the high inter-replicate variability of the proteomic data necessitate proper independent validation methods to confirm proteomic results. This significantly increases the biological value of the proteomic analysis. For example, protein abundance might be verified using immunoblotting with monospecific primary antibodies (Takác et al., 2014 ˇ ). As an alternative, immunolocalization of protein in intact respective tissue have been also employed (Takác et al., 2013 ˇ ) and provide useful information about spatial protein localization. The live cell microscopy imaging using fluorescently tagged proteins may be also used to study protein localization (Takác et al., 2012 ˇ ). Nevertheless, to study abundance of fluorescently tagged protein of interest it should be expressed under the control of its own promoter in the respective mutant background (Smékalová et al., 2014). Transgenic mutant lines with such construct have to show rescue of the mutant phenotype which is considered as a proof of construct functionality. Subsequently, such rescued mutants bearing tagged protein can be used for isolation of functional protein complexes and MS analysis. To validate enzymatic activity, specific spectrophotometric measurements and/or activity staining on native polyacrylamide gels might be used (Takác et al., ˇ 2014). The validation of protein phosphorylation is possible via kinase assays (Feilner et al., 2005; Yang et al., 2013; Lassowskat et al., 2014), by using Phos-tag technology (Kinoshita et al., 2006; Komis et al., 2014) or by isoelectric focusing combined with immunoblotting (Anderson and Peck, 2008). Proteinprotein interactions can be tested by BIFC or co-immunoprecipitation assays (Singh et al., 2012).

The efficiency of shot-gun proteomics for the exploration of MAPK signaling pathways could be significantly increased by bioinformatic evaluation of proteomic data. There are several commercial software and web-based applications providing various analyses including functional categorization of differentially abundant proteins (Conesa and Götz, 2008), clustering of commonly expressed proteins (de Hoon et al., 2004), metabolic pathway analysis tools (Iersel et al., 2008), applications summarizing protein-protein interactions (Jensen et al., 2009), coexpressing protein candidates (Obayashi and Kinoshita, 2010), tools for prediction of protein localization (Horton et al., 2007), detection of signaling peptides (Petersen et al., 2011), search for most abundant sequence motifs (Chou and Schwartz, 2011) and others.

### **GOAL-ORIENTED APPROACHES AND COMPLEMENTARITY IN COMPLEX CELL SIGNALING ANALYSIS**

As explained above, the high throughput analyses of MAPK signaling such as transcriptomics, proteomics and phosphoproteomics are not always necessarily converging to a common output. They have, at least partially, different scopes and their results might be either complementary or contradictory (Gygi et al., 1999). Protein array surveys identify a list of putative substrates of MAPKs but outside of a physiological context. In this sense, phosphoproteomic analyses will decipher phosphoproteins in MAPK mutants or under conditions triggering MAPK activity but with uncertainty as to whether they truly constitute exclusive MAPK substrates (they might be eventually phosphorylated also by alternative kinases). Finally, proteomics (including shot-gun approaches) has potential to corroborate the transcriptional transactivation role of MAPK signaling and should provide factual data on the protein levels. Although proteomics will not reveal mechanistic insights of MAPK signaling (i.e., identify MAPK substrates) it will provide the landscape of its physiological consequences which can be efficiently followed by targeted biochemical, genetic cell biological and physiological assays.

Altogether, this article summarizes recent evidence that shot-gun proteomic analysis of Arabidopsis mutants and transgenic lines with genetically-manipulated MAPK expression is a reliable method for the detection of some molecular networks accompanying developmental or conditional phenotypes of these mutants and lines. We strongly recommend using proper independent methods, including biochemical, cell biological and phosphoproteomic ones, for validation and corroboration of proteomic data. Such detailed studies may eventually provide new links between MAPK-related changes in protein abundance and direct regulation of cellular processes through MAPK-dependent phosphorylation of target proteins.

# **ACKNOWLEDGMENTS**

This work was supported by National Program for Sustainability I (NPU grant no. LO1204) to the Centre of the Region Haná for Biotechnological and Agricultural Research. We would like to thank George Komis for useful suggestions and critical reading of the manuscript.

# **REFERENCES**


and substrates by acrylamide pendant Phos-TAG™. *Methods Mol. Biol.* 1171, 47–63. doi: 10.1007/978-1-4939-0922-3\_5


target networks in *Arabidopsis thaliana* revealed using functional protein microarrays. *Genes Dev.* 23, 80–92. doi: 10.1101/gad.1740009


and biochemical analyses show functional network of proteins involved in antioxidant defense of *Arabidopsis anp2anp3* double mutant. *J. Proteome Res.* 13, 5347–5361. doi: 10.1021/pr50 0588c

Yang, Z., Guo, G., Zhang, M., Liu, C. Y., Hu, Q., Lam, H., et al. (2013). Stable isotope metabolic labeling-based quantitative phosphoproteomic analysis of Arabidopsis mutants reveals ethylene-regulated time-dependent phosphoproteins and putative substrates of constitutive triple response 1 kinase. *Mol. Cell. Proteomics* 12, 3559–3582. doi: 10.1074/mcp.M113. 031633

**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 December 2014; accepted: 09 February 2015; published online: 25 February 2015.*

*Citation: Takáˇc T and Šamaj J (2015) Advantages and limitations of shot-gun proteomic analyses on Arabidopsis plants with altered MAPK signaling. Front. Plant Sci. 6:107. doi: 10.3389/fpls.2015.00107*

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

*Copyright © 2015 Takáˇc and Šamaj. This is an openaccess 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.*

# Phosphoproteomics-based peptide ligand-receptor kinase pairing. Commentary on: "A peptide hormone and its receptor protein kinase regulate plant cell expansion"

Elisabeth Stes 1, 2, 3, 4, Kris Gevaert 3, 4 and Ive De Smet 1, 2 \*

<sup>1</sup> Department of Plant Systems Biology, Vlaams Instituut voor Biotechnologie, Ghent, Belgium, <sup>2</sup> Department of Plant Biotechnology and Genetics, Ghent University, Ghent, Belgium, <sup>3</sup> Department of Medical Protein Research, Vlaams Instituut voor Biotechnologie, Ghent, Belgium, <sup>4</sup> Department of Biochemistry, Ghent University, Ghent, Belgium

Keywords: peptide ligand, receptor kinase, phosphorylation, mass spectrometry, phosphoproteomics

### **A commentary on**

**A peptide hormone and its receptor protein kinase regulate plant cell expansion** by Haruta, M., Sabat, G., Stecker, K., Minkoff, B. B., and Sussman. M. R. (2014). Science 343, 408–411. doi: 10.1126/science.1244454

### Edited by:

Sabine Lüthje, University of Hamburg, Germany

### Reviewed by:

Ian Max Møller, Aarhus University, Denmark Sixue Chen, University of Florida, USA

\*Correspondence: Ive De Smet, ivsme@psb.vib-ugent.be

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

> Received: 17 December 2014 Accepted: 21 March 2015 Published: 09 April 2015

### Citation:

Stes E, Gevaert K and De Smet I (2015) Phosphoproteomics-based peptide ligand-receptor kinase pairing. Commentary on: "A peptide hormone and its receptor protein kinase regulate plant cell expansion". Front. Plant Sci. 6:224. doi: 10.3389/fpls.2015.00224

Small signaling peptides and their receptors play essential roles in plant growth and development (Murphy et al., 2012; Czyzewicz et al., 2013; Matsubayashi, 2014). Arabidopsis thaliana encodes over 600 putative receptor-like kinases and more than 1000 potential secreted peptides (Shiu and Bleecker, 2001; Lease and Walker, 2006), and similar numbers can be expected in other plant species (Shiu et al., 2004; Lehti-Shiu et al., 2009). Taking into account that one peptide can bind or signal through multiple receptors and one receptor can recognize several peptides (Ogawa et al., 2008; Kinoshita et al., 2010; Lee et al., 2011; Shinohara et al., 2012), there is an enormous number of possible peptide ligand-receptor kinase pairs. However, this number of possibilities is in stark contrast with the very few peptide ligand-receptor kinase pairs that have been identified (Butenko et al., 2009; Murphy et al., 2012; Czyzewicz et al., 2013; Endo et al., 2014). Evidently, unambiguous identification of a ligand is crucial to fully understand receptor kinase–mediated signaling pathways (Hirakawa et al., 2008; Jia et al., 2008; Ogawa et al., 2008; Lee et al., 2012; Uchida et al., 2012; Okamoto et al., 2013; Tabata et al., 2014).

The identification of ligand-receptor pairs is technically very challenging, as the genes encoding them regularly belong to gene families with multiple members and are often low expressed, and this only in certain cell types or during specific developmental stages. Therefore, various strategies were followed to identify candidate pairs, such as transcriptional analyses at cellular resolution, microscopic characterization of loss and gain-of-function plants, genetic interaction studies, and biochemical assays demonstrating direct physical interactions (see e.g., Murphy et al., 2012 and Czyzewicz et al., 2013 for more details).

To study the physical interaction of ligands with their receptors, Butenko et al. (2014) recently developed a rapid cellular bioassay that uses the oxidative burst response in Nicotiana benthamiana leaves as readout for activation of (ectopically expressed) receptors by synthetic peptides. However, while a broad range of receptor kinases might be able to activate an oxidative burst, it is likely that this approach cannot be applied to all peptide ligands and/or receptor kinases. In addition, prior knowledge on potential receptor candidates facilitates such studies and there is the requirement for expressing the receptor in N. benthamiana. A similar approach relies on chimeric receptors and monitoring of luciferase activity of known targets in protoplasts transiently expressing signaling components (Albert et al., 2010; Mueller et al., 2012).

Another approach was used by Tabata et al. (2014). Here, overexpression of individual A. thaliana leucine-rich repeat receptor kinases (LRR-RKs) from subfamilies X and XI in tobacco BY-2 cells was combined with photoaffinity labeling of these LRR-RKs by a biologically active small signaling peptide analog derivatized with <sup>125</sup>I-labeled photoreactive 4-azidosalicylic acid ([125]ASA). This revealed that two related LRR-RKs of subfamily XI directly and specifically interact with C-TERMINALLY ENCODED PEP-TIDE 1 (CEP1). Subsequently, this interaction was confirmed by demonstrating that cepr1 cepr2, a double loss-of-function mutant in the identified CEP1 receptor genes, was insensitive to synthetic CEP1 peptide in a root growth assay. This novel approach represents a major leap forward regarding ligand–receptor pairing, but does not take the ligand-receptor interactions into account that rely on a protein complex status involving co-receptors and/or interacting proteins.

In this General Commentary, we would like to highlight an original proteomics-driven approach overcoming the abovedescribed limitations, including the need for constructs, which was employed by the Sussman lab in their quest for the receptor of the secreted RAPID ALKALINIZATION FACTOR (RALF) peptide (Haruta et al., 2014) (**Figure 1**). Specifically, the phosphorylation status of plasma membrane proteins—in their natural in planta environment—was studied in response to treatment with recombinant, biologically active RALF peptide. To obtain quantitative phosphoprotein profiles, <sup>15</sup>N metabolic labeling of A. thaliana seedlings was combined with mass spectrometry– based phosphoproteomics. This strategy allowed the identification of five plasma membrane proteins that displayed a RALF-induced change in phosphorylation pattern, of which two receptor kinases. These can be considered as putative RALF receptors, given that ligand binding likely results in an immediate change in phosphostatus, as illustrated through phosphorylation of the brassinosteroid receptor BRI1 upon ligand binding (Wang et al., 2005). Among these was the FERONIA (FER) receptor kinase, which, following necessary biochemical and functional characterization, was indeed confirmed as a RALF receptor. It is hypothesized that, upon recognition of the peptide ligand, phosphorylation at the C-terminus activates the kinase and initiates a RALF-induced signaling cascade. Additionally, the phosphoproteome analysis pinpointed a plasma membrane H+-ATPase, AHA2, as a potential downstream protein and/or putative FER substrate. Their findings support a model where RALF recognition by the FER kinase affects proton transport, and consequently cell elongation and plant development (see also Murphy and De Smet, 2014). This powerful and straightforward approach hence identified RALF and FER as peptide–receptor pair and

shed light on the molecular mechanism that regulates cell elongation. However, it should be taken into account that depending on the experimental set-up, the selected treatments, and the developmental and physiological growth stage, this approach might not give the complete picture. This is exemplified by the fer4 knockout mutant, which is not completely insensitive to RALF at higher concentrations, suggesting there are likely other RALF receptors. Nevertheless, this is a powerful approach that allows identifying receptors for selected ligands and/or universal downstream responses, without (transient) expression of components.

## References


In conclusion, enormous progress has been made in the matching of peptide ligand–receptor kinase pairs in the last years, with mass spectrometry-driven phosphoprotein profiling as a promising strategy that can be applied to identify receptors for orphan ligands and to progressively close the gap in the plant peptide–receptor field.

## Acknowledgments

ES is a Postdoctoral Research Fellows of the Fund for Scientific Research (FWO)-Flanders (Belgium).


**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.

Copyright © 2015 Stes, Gevaert and De Smet. 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.

# Identification of protein N-termini in Cyanophora paradoxa cyanelles: transit peptide composition and sequence determinants for precursor maturation

Daniel Köhler <sup>1</sup> , Dirk Dobritzsch<sup>1</sup> , Wolfgang Hoehenwarter <sup>2</sup> , Stefan Helm<sup>1</sup> , Jürgen M. Steiner <sup>3</sup> and Sacha Baginsky <sup>1</sup> \*

Institute of Biology, Martin-Luther-University Halle-Wittenberg, Halle (Saale), Germany

<sup>1</sup> Plant Biochemistry, Institute of Biochemistry and Biotechnology, Martin-Luther-University Halle-Wittenberg, Biozentrum, Halle (Saale), Germany, <sup>2</sup> Proteomeanalytik, Leibniz Institute of Plant Biochemistry, Halle (Saale), Germany, <sup>3</sup> Plant Physiology,

### Edited by:

Joshua L. Heazlewood, The University of Melbourne, Australia

### Reviewed by:

Fabio Facchinelli, Heinrich Heine Universität Düsseldorf, Germany Beverley R. Green, The University of British Columbia, Canada

### \*Correspondence:

Sacha Baginsky, Institute of Biochemistry and Biotechnology, Martin Luther University Halle-Wittenberg, Weinbergweg 22, 06120 Halle (Saale), Germany sacha.baginsky@ biochemtech.uni-halle.de

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

> Received: 17 May 2015 Accepted: 07 July 2015 Published: 22 July 2015

### Citation:

Köhler D, Dobritzsch D, Hoehenwarter W, Helm S, Steiner JM and Baginsky S (2015) Identification of protein N-termini in Cyanophora paradoxa cyanelles: transit peptide composition and sequence determinants for precursor maturation. Front. Plant Sci. 6:559. doi: 10.3389/fpls.2015.00559 Glaucophyta, rhodophyta, and chloroplastida represent the three main evolutionary lineages that diverged from a common ancestor after primary endosymbiosis. Comparative analyses between members of these three lineages are a rich source of information on ancestral plastid features. We analyzed the composition and the cleavage site of cyanelle transit peptides from the glaucophyte Cyanophora paradoxa by terminal amine labeling of substrates (TAILS), and compared their characteristics to those of representatives of the chloroplastida. Our data show that transit peptide architecture is similar between members of these two lineages. This entails a comparable modular structure, an overrepresentation of serine or alanine and similarities in the amino acid composition around the processing peptidase cleavage site. The most distinctive difference is the overrepresentation of phenylalanine in the N-terminal 1–10 amino acids of cyanelle transit peptides. A quantitative proteome analysis with periplasm-free cyanelles identified 42 out of 262 proteins without the N-terminal phenylalanine, suggesting that the requirement for phenylalanine in the N-terminal region is not absolute. Proteins in this set are on average of low abundance, suggesting that either alternative import pathways are operating specifically for low abundance proteins or that the gene model annotation is incorrect for proteins with fewer EST sequences. We discuss these two possibilities and provide examples for both interpretations.

Keywords: TAILS, cyanelle, transit peptide, quantitative proteomics, evolution

# Introduction

Photosynthetic organisms originated from a primary endosymbiotic event in which a freeliving ancient cyanobacterium was taken up by a mitochondria-containing eukaryotic host cell and modified to evolve into plastids. Subsequent to primary endosymbiosis, three distinct evolutionary branches emerged and evolved into today's rhodophyta, chloroplastida and glaucophyta (Bhattacharya et al., 2004; Adl et al., 2005). Most research has been devoted to members of the chloroplastida branch, since it represents around 99% of plant biomass on the mainland Köhler et al. Cyanelle transit peptides

and performs most of the photosynthetic activity on this planet. However, recent sequencing efforts established the genome sequence of Galdieria sulphuraria and Cyanophora paradoxa as representatives for rhodophytes and glaucophytes, making comparative sequence analyses and functional genomics approaches possible (Price et al., 2012; Schönknecht et al., 2013). With proteomics tools at hand, the evolutionary divergence of fundamental plastid processes can now be studied at the protein level, which has the advantage over comparative genomics that such analyses are closer to gene function.

The key to establishing an endosymbiosis is the distribution of tasks between the cell organelles, and their coordination by the nucleus as the central gene expression system. Thus, the endosymbiont has lost much of its original genome coding capacity to the nucleus (Martin et al., 2002). Since plastids are the site of many complex metabolic processes, there is clearly a need for the re-import of nuclear encoded plastid proteins from the cytosol in a specific and regulated manner. At the outer and inner plastid envelope membranes, a specialized import machinery has evolved that comprises high molecular mass protein complexes designated as translocases at the outer (TOC) and inner (TIC) chloroplast envelope membrane (Agne and Kessler, 2009). A basic set of common TOC and TIC components is maintained in all descendants from primary endosymbiosis, representing most likely the basic functional module in plastid protein import (Steiner and Löffelhardt, 2005). This supports the single origin of primary plastids and furthermore suggests that plastid protein import has been established prior to the divergence of the three distinct evolutionary branches.

Therefore, basic characteristics of import specificity must have been in place prior to the evolutionary branching. Proteins that are destined for plastids possess an N-terminal transit peptide that is cleaved off after plastid protein import by a processing peptidase. At present, most of the information on transit peptide design is available for the chloroplastida, mostly Arabidopsis, rice, and Chlamydomonas (Patron and Waller, 2007; Chotewutmontri et al., 2012). In the case of Arabidopsis, an N-terminal HSP70-binding motif enriched in hydrophobic amino acids is required for protein translocation (Chotewutmontri et al., 2012). Chloroplastida transit peptides are generally devoid of acidic amino acids and enriched in hydroxylated amino acids. Serine is the most frequently occurring amino acid in Arabidopsis transit peptides, while it is alanine in rice and Chlamydomonas (Kleffmann et al., 2007; Patron and Waller, 2007; Zybailov et al., 2008). Surprisingly, apart from a modular structure, transit peptides are not well conserved and only few functional modules were identified that are involved in protein translocation (Lee et al., 2008).

Initial attempts were made to compare chloroplastida transit peptides with those of rhodophyta and glaucophyta but these analyses were hampered by the constrained datasets available for the latter two evolutionary branches. The most striking feature of their transit peptides is the occurrence of a phenylalanine at the N-terminus, that is prevalent in most known rhodophyte transit peptides (85% carry an F in their N-terminal region) and in all known glaucophyta transit peptide (100%, however only 23 sequences were considered) (Patron and Waller, 2007). The functional importance of F is apparent from import experiments that showed that chloroplastida transit peptides are only functional in glaucophytes when phenylalanine is added to the N-terminal region (Steiner et al., 2005b). Apparently, phenylalanine in cyanelle transit peptides is an ancestral feature that was lost during the evolution of the chloroplastida branch.

Since glaucophytes were the first lineage to diverge after plastid primary endosymbiosis (Qiu et al., 2012) the composition of the cyanelle proteome and the mechanisms for assembling it are of particular interest for comparative analyses. We report here a targeted analysis of N-terminal peptides in cyanelle proteins from Cyanophora paradoxa using a proteomics method called "terminal amine labeling of substrates" (TAILS) (Doucet et al., 2011). This allowed us to assess protein N-termini as they accumulate in vivo and identify sequence requirements for processing peptidase cleavage and functional protein import.

# Materials and Methods

# Biological Material

Cyanophora paradoxa LB555UTEX was grown photoautotrophically at 26◦C under continuous fluorescent white light using a liquid mineral medium continuously gassed with a mixture of 5% (v/v) CO2 in air as previously described (Mucke et al., 1980).

### Isolation of Intact and Periplasm-free Cyanelles

Import-competent cyanelles were isolated according to Steiner et al. (2000). In brief, cultures in exponential growth phase were centrifuged at 3500 × g for 5 min at room temperature. The blue-green pellet was then gently resuspended at 4◦C using a buffer containing 25 mM HEPES/NaOH pH 7.6, 2 mM EDTA and 0.35 M Sucrose. C. paradoxa cells were broken in a Waring blender 5 × 1 min at maximal speed with 1 min cooling on ice in between. Intact cyanelles were sedimented by centrifugation (2000 × g, 5 min, 4◦C) and resuspended twice to remove residual cell debris and contaminant cytosolic proteins. For preparation of periplasm-free cyanelles, plastids were treated with 1% Triton X-100 for 10 min on ice, prior to centrifugation.

## Extraction of Proteins and Enrichment of N-terminal Peptides

For the analysis reported here we used total C. paradoxa celles as well as isolated cyanelles as biological material. For total protein extraction a buffer containing 100 mM HEPES/NaOH (pH 7), 4% (w/v) SDS, 1 mM PMSF and 0.1% protease inhibitor cocktail was used for both preparations (Sigma-Aldrich). The enrichment of protein N-termini from C. paradoxa samples was performed by terminal amine isotopic labeling of substrates (TAILS) as described by Köhler et al. (2015) based on the protocol of Doucet et al. (2011). The workflow was performed with 100µg of each sample. The protein-N-termini were chemically blocked by dimethylation. The chemically modified proteins were digested with trypsin and the resulting peptides carrying unblocked primary amines covalently bound to a polymer. Removal of the polymer by filtration results in a sample enriched in N-terminal peptides, which was analyzed by mass spectrometry after sample desalting.

# Setup for Mass Spectrometric Measurements of TAILS Samples

The desalted and completely dried samples were resuspended in water with 2% acetonitrile and 0.1% formic acid. Samples were separated by liquid chromatography using C18 columns from Proxeon (guard column: 2 cm, ID 100µm, 5µm, analytical column: 10 cm, ID 75µm, 3µm). For separation, a gradient consisting of water with 0.1% formic acid (A) and acetonitrile with 0.1% formic acid (B) was used (method: 0-150 min 5-40% B, 150-160 min 40-80% B, 160-170 min 80% B). The samples were analyzed on an LTQ Orbitrap Velos (Thermo Scientific), precursor ion detection was done in an m/z-range from 400-1850 and the 20 most intensive signals were used for MS2.

### TAILS Data Processing and Analysis

The MS-raw-data were processed as previously described (Köhler et al., 2015) using Proteome Discoverer 1.2 (Thermo Scientific) and MaxQuant (Cox and Mann, 2008). We used a precursor mass tolerance of 7 ppm and a fragment ion mass tolerance of 0.5 Da with a maximal number of 3 missed cleavages. In all searches, semi tryptic peptides were accepted. MaxQuant searches were performed with a protein FDR of 0.01 and ProteomeDiscoverer searches with a strict FDR of 0.01 and a relaxed FDR of 0.05. The C. paradoxa protein database was retrieved from the "Cyanophora Genome Project" (Price et al., 2012). Searched Nterminal modifications were "dimethylation" or "acetylation." Further, modifications allowed were dimethylation on lysines, carbamidomethylation of cysteines and methionine oxidation. For further analysis we only used peptides with a tryptic C-terminus (Arg, Lys) and where the assigned proteins in the Cyanophora paradoxa protein database contained a startmethionine. Cyanelle localized proteins were identified from the cyanelle preparation performed here, by cross-comparing the data against recent proteomics results (Facchinelli et al., 2013), and by homology to proteins in the Arabidopsis chloroplast proteome reference table (Reiland et al., 2009; van Wijk and Baginsky, 2011).

# Nano-LC Separation, HD-MS<sup>E</sup> Data Acquisition and Protein Identification/Quantification

LC separation (140 min gradient) and HD-MS<sup>E</sup> data acquisition was performed using 1µl of the digested total protein samples of TritonX-100 treated or untreated cyanelles on an ACQUITY UPLC System coupled to a Synapt G2-S mass spectrometer (Waters, Eschborn, Germany). MS acquisition was set to 50– 2000 Da. Data analysis was carried out by ProteinLynx Global Server (PLGS 3.0.1, Apex3D algorithm v. 2.128.5.0, 64 bit, Waters, Eschborn, Germany) with automated determination of chromatographic peak width as well as MS TOF resolution. Lock mass value for charge state 2 was defined as 785.8426 Da/e and the lock mass window was set to 0.25 Da. Low/high energy threshold was set to 180/15 counts, respectively. Elution start time was 5 min, intensity threshold was set to 750 counts. Databank search query (PLGS workflow) was carried out as follows: Peptide and fragment tolerances was set to automatic, two fragment ion matches per peptide, five fragment ions for protein identification, and two peptides per protein. Maximum protein mass was set to 250 kDa. Primary digest reagent was trypsin with one missed cleavage allowed. According to the digestion protocol, fixed (carbamidomethyl on Cys) as well as variable (oxidation on Met) modifications were set. The false discovery rate (FDR) was set to 4% at the protein level. MS<sup>E</sup> data were searched against the same C. paradoxa database as for the TAILS data analysis, and rabbit glycogen phosphorylase B (P00489) was used as internal quantification standard. Quantification was performed based on the intensity of the three most abundant proteotypic peptides.

### Availability of Mass Spectrometry Data

All mass spectrometry data have been deposited to the ProteomeXchange Consortium (http://proteomecentral. proteomexchange.org) via the PRIDE partner repository with the submission number PXD002187.

# Results and Discussion

# Identification of Cyanelle Proteins and their Transit Peptides

We analyzed the composition and the processing site of Cyanophora paradoxa cyanelle transit peptides by a specialized proteomics method for the specific and sensitive identification of protein N-termini. Using terminal amine isotopic labeling of substrates (TAILS) we enriched N-termini of proteins isolated from cyanelles and full cells from a Cyanophora paradoxa culture. The TAILS method works by blocking the primary amine of the protein N-terminus by dimethylation prior to tryptic digest (Boersema et al., 2009; Doucet et al., 2011; Huesgen et al., 2013). Trypsin digestion generates new "unblocked" N-termini that are subsequently coupled to a high molecular weight dendritic hyperbranched polyglycerol-aldehyde (HPG-ALD) polymer with their primary amines. The polymer with the bound internal peptides is removed, while the dimethyl-labeled N-terminal peptides remain in the sample. Thus, native in vivo N-termini of C. paradoxa proteins are enriched by a subtractive approach and can be unambiguously identified by their dimethyl label in mass spectrometry.

We identified non-redundant N-terminal peptides for 303 proteins in an Orbitrap Velos (**Figures 1A,B**, Supplemental Table S1). Plastid proteins were identified from the cyanelle preparation performed here and from the complete cell preparation by cross comparing the data to a recent cyanelle proteome study (Facchinelli et al., 2013) and by homology to proteins in the Arabidopsis thaliana plastid proteome map (Reiland et al., 2009; van Wijk and Baginsky, 2011). For the homology searches we accepted the best Arabidopsis hit at a threshold of e-10, and extracted chloroplast proteins from the list of homologs with a chloroplast proteome reference table (Reiland et al., 2009). With these criteria, we identified 123 nucleus encoded cyanelle proteins in the TAILS dataset (Supplemental Table S1) of which 74 identifications originated from the cyanelle preparation performed here (**Figure 1B**). The minimal start

position of peptides in the identified nucleus encoded proteins is presented in a histogram in **Figure 1A**, separately for nucleus encoded cyanelle and non-cyanelle proteins (**Figure 1A**). As reported for other systems, the largest bin for non-cyanelle proteins is that of position "2," revealing N-terminal methionine excision as an important posttranslational process in Cyanophora paradoxa (Giglione et al., 2015). A local maximum at starting positions 21–40 and 51–80 in the set of non-cyanelle proteins comprises most likely mature mitochondrial proteins and so far unknown cyanelle proteins that were generated by target peptide cleavage following import. Indeed, 13 out of 53 proteins falling into these bins carry a phenylalanine within the Nterminal 10 amino acids and therefore fulfill a basic requirement for cyanelle targeted proteins (Supplemental Table S1). Proteins with the start position >150 are most likely derived from proteolytic events that occur during protein activity control and breakdown.

## Characteristics of Cyanophora paradoxa Transit Peptides

Together, 77 nucleus encoded plastid proteins were identified matching the minimal starting range 11–110. (**Table 1**, Supplemental Table S1). Compared to the set of 136 Arabidopsis proteins identified with the same method (Köhler et al., 2015), Cyanophora transit peptides are slightly longer with a median length of 63 amino acids compared to 55 amino acids for Arabidopsis, however, this difference is not significant (p = 0.2, two-tailed Welch's-test). Transit peptides of higher plants possess several distinct modules that are combined and exchanged to result in tissue- and age-specific plastid protein import, a complexity that has been summarized elegantly in the "M and M" model (Li and Teng, 2013). Tissue-specificity of transit peptides is irrelevant for unicellular organisms and it was therefore expected that cyanelle transit peptides are rather shorter than those of Arabidopsis, e.g., more similar in size to those of Chlamydomonas (around 32 amino acids) (Patron and Waller, 2007). One possible explanation for the lengths of Cyanophora transit peptides is the peptidoglycan wall between the outer and inner envelope membrane that must be crossed by imported proteins.

The overall amino acid composition is largely similar except that serine is prevalent in Arabidopsis and alanine in Cyanophora paradoxa transit peptides (**Figures 2A,B**). In this respect, the latter are similar to transit peptides of rice and Chlamydomonas, while Arabidopsis transit peptides are exceptional (Kleffmann et al., 2007; Patron and Waller, 2007). Because alanine tends to form alpha-helices, it is conceivable that alpha-helical structures might be functional determinants of Cyanophora paradoxa transit peptides (Otaki et al., 2010). However, alanine is overrepresented along the entire transit peptide (**Figure 2B**), speaking against the formation of specific regulatory helical structures. At present, only limited information is available for the role of secondary structure elements in precursor recognition and organellar import and most analyses of transit peptide secondary structures were performed under non-physiological conditions (von Heijne et al., 1989; Franzen et al., 1990; Bruce, 2001). The situation is further compounded by the fact that most transit peptides will form a random coil under aqueous conditions, while functionally important structures only form upon contact with lipid-containing hydrophobic surfaces at the envelope membranes (Wienk et al., 1999; Bruce, 2001; Ambroggio et al., 2007).

### Cyanophora paradoxa Transit Peptides have a Modular Structure

Transit peptides of both Cyanophora and Arabidopsis have a tripartite modular structure consisting of a hydrophobic Nterminus, a less hydrophobic middle region and a C-terminal region that contains the cleavage site for the processing peptidase (**Figure 3**). The hydrophobic N-terminal region is required for interaction with cytosolic HSP70 (Ivey et al., 2000; Chotewutmontri et al., 2012), which is essential for plastid

TABLE 1 | List of all nucleus encoded cyanelle proteins that match the minimal starting range 11–110 as identified by TAILS.



The proteins are represented by their CDS-Identifier and the identified peptide representing the minimal start identified for the protein. Phenylalanine (F) within the first 10 amino acids is highlighted and underlined. The data underlying the assignment of proteins as cyanelle proteins are indicated with a plus (+).

"published": Cyanelle proteins previously identified by proteomics (Facchinelli et al., 2013).

"BLAST": C. paradoxa proteins identified as cyanelle proteins by BLAST searches against an A. thaliana chloroplast proteome reference table (van Wijk and Baginsky, 2011). "This study": Proteins identified by mass spectrometry in samples of purified cyanelles in this study.

protein import (Pilon et al., 1995). One prevalent amino acid in the N-terminus of cyanelle transit peptides is a phenylalanine residue that occurs within the first 10 amino acids (**Table 1**, **Figure 2B**). N-terminal phenylalanine is detectable in 49 out of 77 cyanelle transit peptides identified here, thus, phenylalanine is clearly the most distinguishing feature of cyanelle transit peptides compared to those of the chloroplastida branch (i.e., rice, Arabidopsis and Chlamydomonas, **Figure 2B**) (see discussion below).

In addition to phenylalanine, valine is overrepresented in the N-terminal region of cyanelle transit peptides (**Figure 2B**) while it is leucine in Arabidopsis transit peptides. Together these three amino acids (V, L, F) together with the weakly hydrophobic amino acid alanine establish the hydrophobicity of the N-terminal module in the transit peptides of both organisms (**Figures 2A,B**). The hydrophobic N-terminal part and the less hydrophobic middle region of cyanelle transit peptides are separated by several proline residues between amino acid positions 10–20. Proline is present in loops and in regions between secondary structures (Otaki et al., 2010). It is conceivable that the frequent occurrence of proline up to amino acid 20 constitutes a structural separation of the N-terminal 10–20% of the cyanelle transit peptide from the rest. Bruce (2001) hypothesized that such proline containing regions may be necessary to separate two functional regions from each other or function as motifs by themselves (Bruce, 2001). Following the hydrophobic N-terminal region, hydrophobicity decreases as the amount of charged amino acids increases in both organisms. In the identified cyanelle transit peptides, this region contains on average more acidic amino acids than in Arabidopsis (**Figure 2B**). It is possible that these are required for the efficient transfer of proteins through the periplasm, but functional data on the translocation process are currently not available.

Arabidopsis and Cyanophora. For every amino acid, the average occurrence in transit peptides is provided along with the standard deviation. (B) Topology heat map of amino acid distribution in transit peptides of Arabidopsis thaliana

peptide was set to 100% and the relative position of amino acids in the transit peptide was calculated on this basis. The darker the color the higher the frequency of occurrence.

### The Processing Peptidase Cleavage Site

We aligned the identified N-termini of cyanelle proteins with WebLogo 3 (Crooks et al., 2004) and ICE Logo (Colaert et al., 2009) using the 10 amino acids up- and downstream of the most N-terminal amino acid. Both procedures produced essentially identical alignments and we therefore present only the WebLogo 3 results. This alignment revealed a local "bit" maximum (yaxis) N-terminal to the processing site resulting mostly from the enrichment of valine, alanine, and arginine within the positions -3 to -1 relative to the cleavage site (**Figures 4A,B**). The occurrence of alanine as C-terminal amino acid is coupled to the occurrence of valine in the -3 position (**Figure 4B**). The transit peptides with a C-terminal arginine show no further "bit" maximum (**Figure 4B**). We can exclude that these peptides are artifacts from the tryptic digest because we accepted only peptides that were identified with a dimethyl-label or an acetyl-group at their N-terminus (the latter modification occurs in vivo and blocks the N-terminal dimethyl labeling). Thus, our data suggest that arginine alone is capable of orchestrating the processing peptidase cleavage. The composition of the cleavage site in

Arabidopsis transit peptides identified with TAILS is similar, especially the enrichment of valine, arginine, and alanine close to the processing site (**Figure 4B**). Additionally, Arabidopsis transit peptides are enriched for isoleucine residues that have most likely the same function as valine in Cyanophora (Köhler et al., 2015).

The processing peptidase cleavage releases an N-terminal amino acid, the nature of which may have regulatory impact on protein stability (Apel et al., 2010). Interestingly, the occurrence of amino acids at the N-terminus is similar for both Arabidopsis and Cyanophora (**Figure 5**) with the majority of proteins carrying serine or alanine in the first position. The greatest difference between these two organisms is the relatively high frequency of cysteine residues around the cleavage site in Arabidopsis that is missing in Cyanophora (**Figure 4A**). Although transit peptide cleavage prediction programs predict cleavage C-terminal of cysteine (Emanuelsson et al., 2000), it does not occur as N-terminal amino acid in the mature protein (**Figure 4**). This observation can be explained by postulating an additional cleavage by an intermediate cleavage peptidase

(ICP55) (Huang et al., 2015) that removes destabilizing residues from the N-terminus of mature proteins in Arabidopsis. At present we do not have any indication that a similar cleavage occurs in cyanelles.

### Quantitative Proteomics Supports the Prevalence of Phenylalanine, but Reveals Some Deviation for low Abundance Proteins

We next generated a quantitative proteome dataset for isolated cyanelles to set the above-described observations into a biological context. In order to focus the analysis on transit peptides in proteins that have crossed two membranes, we generated two different quantitative proteomics datasets, (i) one set from isolated cyanelles including the periplasmic space and (ii) another set from isolated cyanelles from which the periplasm was removed by Triton X-100 washes. Absolute protein quantification was performed by HD-MS<sup>E</sup> as described previously (Helm et al., 2014). All data were acquired in three replicate measurements and only proteins consistently identified and quantified in at least 2 out of 3 replicates in both preparations were considered for further analyses. To identify periplasmic proteins and to remove them from further analysis, we included only those proteins in our analysis whose abundance is less than 2-fold different between the two preparations. With this constraint, the final dataset comprises 276 quantified proteins (Supplemental Table S2). The most abundant proteins are cyanelle-encoded phycocyanin subunits, phycobilisome linker proteins and other photosynthetic proteins, while several uncharacterized proteins, as well as enzymes involved in amino acid and nucleotide metabolism are the least abundant proteins in this dataset (for a detailed list see Supplemental Table S2).

Altogether, 19 proteins with an obviously incomplete gene model (no start methionine) and the remaining 70 cyanelleencoded proteins were removed from further analysis. The remaining nucleus-encoded proteins carry a phenylalanine within position 1–18 in 145 out of 187 cases. The 42 proteins lacking phenylalanine are predominantly low abundance proteins with a mean abundance of around 5 fmol on column, compared to 14 fmol on column average abundance calculated over all identified proteins. There are two ways to interpret these findings. First, it is possible that there are different import pathways that differ by their specificity. Since high abundance proteins are easier to detect, a potential detection bias for phenylalanine-containing transit peptides exists while low abundance proteins remained unidentified until recently (this study, Facchinelli et al., 2013). The second possibility is that the low abundance protein dataset is enriched for incorrect gene models, because more and better EST sequences are available for high abundance proteins.

# Analysis of Non-F Transit Peptide Containing Proteins and Examples for Potential Non-F Precursors

To distinguish between these two possibilities, we analyzed the 42 proteins without phenylalanine at their N-terminus for additional evidence to support one or the other conclusion. The results of this search are summarized in Supplemental Table S3. In summary, we identified supporting EST sequences for 11 out of 42 non-F precursors. Of these, four ESTs supported the original gene model without F, while seven led to curation of the gene model. BLAST searches suggest a truncation for 15 out of the 31 remaining gene models.

In the set of proteins that do not carry F at their Ntermini are three highly abundant opsins (Frassanito et al., 2010, 2013). Indirect immunofluorescence with antibodies against Cyanophopsin\_1 in C. paradoxa cells showed labeling of the cyanelle envelope. Since knowledge about the targeting of proteins to the cyanelle envelope is missing, it may be hypothesized that phenylalanine is not required for the envelopetargeting of proteins. It is currently unknown whether a Toc75 related import channel is used for their translocation process. It has been speculated that the N-terminal phenylalanine attenuates the interaction with the cyanelle surface containing mono- and digalactosyldiacylglycerol, thus promoting contact with proteins present in the outer membrane such as a cyanelle Toc75 (Wunder et al., 2007). Therefore, it seems possible that at least some of the envelope-localized opsins are integrated into the membrane by a mechanism similar to the integration of AtOEP80, which contains the targeting information within its mature sequence (Inoue and Potter, 2004). The lack of a canonical transit peptide is supported by the identification of an unprocessed opsin with start position 2 in the TAILS experiment explaining the lack of F in the N-terminal region of these proteins (Supplemental Table 1).

Another non-F precursor is gene model 37109\_cds:7008 that encodes for the 50S ribosomal protein L27. Many ESTs from this gene (e.g., EG947820.1) are shifted in their start 43 residues to the C-terminus compared to the annotated sequence. The EST-supported start MAHKKGSGSTKNGRDSN resembles the start of the mature part of the protein without any targeting signal. Interestingly, this N-terminal part of the sequence is 100% identical to the corresponding part of L27 of C. merolae where it appears to be encoded on the rhodoplast genome. Therefore, 50S ribosomal protein L27 might be imported into the cyanelle without any targeting signal. The only peptidoglycan-related protein found in this analysis is 54819\_cds:11202 encoding UDP-N-acetylglucosamine-N-acetylmuramyl- (pentapeptide) pyrophosphoryl-undecaprenol N- acetylglucosamine transferase (murG), which is a stroma-localized peripheral inner envelope membrane protein. It appears to carry a targeting signal of about 80 residues without F when compared to bacterial proteins. A chaperonin GroEL homolog (37904\_cds:7901), carries the aromatic residue tyrosine in the N-terminal region of the precursor (MEAAYVTSSP). When the GroEL precursor is compared to a homologous EST (EC658348.1) of the same species it appears that the latter carries a phenylalanine in the same position as the tyrosin (MDHAFVVP). It was shown that tyrosine can substitute for phenylalanine to some extent in in vitro import experiments (Steiner et al., 2005a). Thus, it is possible that F is functionally replaced by Y similar to the situation in diatoms where the aromatic amino acids phenylalanine, tryptophan, tyrosine allow efficient plastid protein import (Gruber et al., 2007).

The data presented here clearly support the prevalence of F in the N-terminal region of cyanelle transit peptides. The majority of the putative non-F precursors are represented by an incorrect gene model (Supplemental Table 3) and the abundant

# References


non-F cyanelle opsins do not carry a canonical transit peptide (see the TAILS data in Supplemental Table 1). Apart from the distinguishing N-terminal F, cyanelle transit peptides are similar in structure and amino acid composition to those of Arabidopsis, suggesting that basic principles guiding the protein translocation process into plastids were in place before the divergence of the three lineages after primary endosymbiosis.

# Acknowledgments

The authors are grateful for financial support from DFG project Ba1902/3–1 (DK) and from the European Regional Development fund of the European commission via grant W21004490 "Land Sachsen-Anhalt" (SH).

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00559


targeted to the chloroplastic outer envelope by different mechanisms. Plant J. 39, 354–365. doi: 10.1111/j.1365-313X.2004.02135.x


**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.

Copyright © 2015 Köhler, Dobritzsch, Hoehenwarter, Helm, Steiner and Baginsky. 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.

# Diversity and function of maize pollen coat proteins: from biochemistry to proteomics

*Fangping Gong, Xiaolin Wu and Wei Wang\**

*State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou, China*

Maize (*Zea mays* L.) is globally cultivated as one of the most important grain crops. As a wind-pollinated species, maize produces a large quantity of pollen grains that heavier and larger compared to *Arabidopsis*. Maize is an important model plant in pollen biology of monocots. The pollen coat, the outermost layer of pollen, plays a vital role in pollen–stigma interactions and successful fertilization. Pollen coat proteins (PCPs), which confer species specificity, are required for pollen adhesion, recognition, hydration, and germination on the stigma. Thus, PCPs have attracted intensive research efforts in plant science for decades. However, only a few PCPs in maize have been characterized to date, whereas the functions of most maize PCPs remain unclear. In this review, we summarize the current knowledge of maize PCPs with regard to protein constituents, synthesis and transport, and functions by comparison with the model plant *Arabidopsis thaliana* and *Brassica* plants. An understanding of the comprehensive knowledge of maize PCPs will help to illuminate the mechanism by which PCPs are involved in pollen–stigma interactions in maize and other crop plants.

Keywords: pollen coat proteins, maize, pollen–stigma interaction, pollen germination, proteomics

# Introduction

The pollen coat, also called pollenkitt (Dobson, 1988) and tryphine (Murphy and Ross, 1998), is the outermost layer of pollen (**Figure 1**). Pollenkitt is most common in angiosperms, whereas tryphine refers to the pollen coat in insect-pollinated Brassicaceae plants (Pacini and Hesse, 2005). Despite the difference in the constituent and formation, both types of pollen coats originate from the anther tapetum and share some functions. The pollen coat, which confers species specificity, is composed of lipids, proteins, pigments, and aromatic compounds (Bih et al., 1999; Mayfield et al., 2001; Murphy, 2006), fills the sculptured cavities of the exine (Heslop-Harrison, 1968; Chay et al., 1992) and thus is highly heterogeneous and extremely hydrophobic. The study of pollen coat constituents dates back to the 1960s (Heslop-Harrison, 1968). The constituents of the pollen coat, especially pollen coat proteins (PCPs), are thought to play vital roles in aiding pollen–stigma recognition, adhesion, and hydration and pollen initial germination on the stigma (Doughty et al., 1993, 1998; Suen and Huang, 2007; Dresselhaus and Franklin-Tong, 2013). For example, SP11/SCR was found to determine pollen S-specificity in the self-incompatibility of *Brassica* species (Cabrillac et al., 2001; Shiba et al., 2001); in addition, xylanase facilitates maize pollen tube penetration into the silk via enzymatic xylan hydrolysis (Suen and Huang, 2007). Therefore, PCPs have attracted intensive research efforts in plant science for decades, especially in Brassicaceae plants (e.g.,

### *Edited by:*

*Silvia Mazzuca, Università della Calabria, Italy*

### *Reviewed by:*

*Giampiero Cai, Università degli Studi di Siena, Italy Stefano Del Duca, University of Bologna, Italy*

### *\*Correspondence:*

*Wei Wang, State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou 450002, China wangwei@henau.edu.cn*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

> *Received: 29 December 2014 Paper pending published: 16 February 2015 Accepted: 12 March 2015 Published: 30 March 2015*

### *Citation:*

*Gong F, Wu X and Wang W (2015) Diversity and function of maize pollen coat proteins: from biochemistry to proteomics. Front. Plant Sci. 6:199. doi: 10.3389/fpls.2015.00199*

*Arabidopsis* and *Brassica napus*). However, only a few studies on

PCPs in crop plants (e.g., maize and rice) are available to date. Maize (*Zea mays*), belonging to the Poaceae family, is one of the most important cereal crops worldwide, and the completion of the maize B73 genome sequence has greatly promoted the progress of maize proteomics. Compared to *Arabidopsis*, maize pollen grains are heavier and larger (150–500 ng, 60–125 μm in diameter; Kalinowski et al., 2002), and adequate amounts of maize pollen grains can easily be obtained in high purity for biological studies. Therefore, maize serves as an important model for exploring the mechanisms of pollen germination, pollen–sigma interaction and pollen tube growth in monocots.

To our knowledge, only a few major PCP constituents in maize, e.g., cysteine protease, β-expansin (Zea m 1), xylanase and β-glucanase (Bih et al., 1999; Suen et al., 2003; Wang et al., 2006; Li et al., 2012), have been identified and characterized using biochemical and molecular techniques, whereas most PCPs in maize remain to be characterized. Proteomics is a powerful tool for analyzing complex mixtures of proteins and identifying biomarkers. For the first time, we have recently identified the overwhelming majority of the maize PCPs via a gel-based proteomic technique (Wu et al., 2015). Based on informatics analyses, many maize PCPs have been proposed to interact with stigma surface constituents.

In this review, the current knowledge of maize PCPs, including the protein constituents, synthesis and transport, and functions, are summarized. By comparison with the model plant *Ar. thaliana* and *Brassica* plants, we highlight

tapetum or pollen interior to the pollen coat.

the specificity of maize PCPs and the potential roles of specific PCPs in pollen–stigma interactions during early pollen germination.

# The Isolation of Maize PCPs

The lipidic pollen coat can be readily removed by treatment with organic solvents such as cyclohexane, diethyl ether, and chloroform (Doughty et al., 1993; Murphy and Ross, 1998). PCPs in such a pollen coat preparation are then extracted and purified using detergent-containing buffers. The extraction effects of different organic solvents on PCPs have been evaluated. Carbon tetrachloride and chloroform were reported to extract relatively more species of maize PCPs compared to six other organic solvents (hexane, heptane, cyclohexane, benzene, diethyl ether, and methanol; Bih et al., 1999). However, due to its high protein extractability and low pollen hydration rates, cyclohexane was found to be the best solvent (out of 24 organic solvents, ranging from non-polar to polar) for Bermuda grass (*Cynodon dactylon*) pollen coat extraction (Bashir et al., 2013).

Previously, three major maize PCPs (endoxylanase, βglucanase, and cysteine protease; Bih et al., 1999; Suen et al., 2003) and a small amount of Zea m 1 (Wang et al., 2006) were extracted from a pollen coat preparation using diethyl ether, and we recently used chloroform to extract maize PCPs for a proteome analysis (Wu et al., 2015).

# The Constituents of Maize PCPs

At present, the known PCPs in maize include a total of 14 protein species, including allergens, hydrolases, and other proteins (**Table 1**). With the exception of the previously characterized endoxylanase, β-glucanase, cysteine protease and β-expansin 1 (Zea m 1), most PCPs have been identified via gel-based proteomics (Wu et al., 2015). Endoxylanase, β-glucanase, Zea m 1, and cysteine protease are present in high abundance in the maize pollen coat. In addition, many maize PCPs (e.g., profilin, exopolygalacturonase, and ABA-induced caleosin) exist in several isoforms.

The proteins identified in maize pollen coat are greatly different from those in Brassicaceae plants. In *A. thaliana*, 10 PCPs (*>*10 kDa) have been identified, including two kinases, one caleosin-like proteins, two lipase proteins, and five oleosins (Mayfield et al., 2001). In *B. napus*, 12 PCPs have been identified, most of which are oleosin isoforms (Murphy, 2006). In particular, the amphipathic oleosin, originating from storage tapetosomes in tapetum cells (Wu et al., 1997; Ting et al., 1998), has been demonstrated to be absent from the maize pollen coat (Li et al., 2012).

This largely difference in PCPs between maize and Brassicaceae plants is largely related to the physiological property and formation, and fundamentally the species specificity. Indeed, the pollen coat constituents in wind-pollinated species such as maize are quite different from that in insect- or self-pollinated species. *Brassica* and *Arabidopsis* are insect- or self-pollinated species, and their pollen has a thick coat, which is sticky and contains abundant lipids. In contrast, maize pollen has a thinner pollen coat that is non-sticky and contains a reduced amount of lipids (Bih et al., 1999). In addition, maize PCPs can originate from the tapetum or the pollen interior.

# Synthesis and Transport of Maize PCPs

The tapetum, forming the innermost sporophytic cell layer of the anther and enveloping the developing pollen (**Figure 1**), plays a central role in pollen coat and exine formation (Ariizumi and Toriyama, 2011; Liu and Fan, 2013). In *Arabidopsis* and *Brassica* species, the tapetum is packed with two predominant storage organelles: elaioplasts and tapetosomes (Hsieh and Huang, 2007; Li et al., 2012; Quilichini et al., 2014a,b). The constituents of the pollen coat, such as steryl esters, lipids, alkanes, lipidassociated proteins, oleosin proteins, and flavonoids, are mainly derived from the elaioplasts and tapetosomes. However, electron microscopy studies have shown that at a late developmental stage, maize tapetum cells do not possess elaioplasts and tapetosomes (Li et al., 2012; Liu and Fan, 2013), indicating that maize PCPs synthesized in the tapetum may be delivered to the pollen coat via other mechanisms.

Although relatively more maize PCPs have been reported of late, our knowledge about their exact origin and transport routes is limited. Glucanase, xylanase, and cysteine protease have been demonstrated to be synthesized in the adjacent tapetum and transported via the endoplasmic reticulum (ER), Golgi, vacuoles, and vesicles (**Figure 1**); some remain in the tapetum, eventually depositing in the sculptured cavities of the pollen exine upon programmed cell death of the tapetum (Li et al., 2012). The synthesis of glucanase and xylanase begins at the middle stage of anther development, and these enzymes are then stored in vesicles and the cytosol, respectively; in contrast, cysteine protease first emerges at the late stage and is stored in vacuoles. Both glucanase and cysteine protease contain an ER-targeting signal peptide. Xylanase is initially synthesized via a tapetum mRNA with a long 5 leader (Bih et al., 1999) as a 60-kDa precursor, which is then converted to the active 35-kD xylanase (Wu et al., 2002).

Zea m 1, together with Cyn d 1, Sor h 1, Lol p 1, and Phl p 1, are pollen-specific group-1 allergens (Valdivia et al., 2007; Bashir et al., 2013). Phl p 1 is mainly present in the pollen intine (Grote et al., 1994; Behrendt et al., 1999) and also in the pollen coat and cytosol of pollen vegetative cells (Staff et al., 1990). By immunoelectron microscopy, Wang et al. (2006) found Zea m 1 in the pollen coat fraction, in the tectum and the foot layer of the exine. Besides, a substantial amount of β-expansin (Zea m 1) was found localized in pollen interior (Suen et al., 2003; Wang et al., 2006). Moreover, two allergens profilin (Zea m 12) and exopolygalacturonase (Zea m 13) were found to be easily released from pollen into aqueous solution (Suen et al., 2003), implying a pollen surface



localization. Our recent work showed that profilin and exopolygalacturonase exist in the coat of maize pollen (Wu et al., 2015). Thus, it is possible that these allergens in the maize pollen coat may be synthesized within the pollen interior (**Figure 1**).

Sporopollenin is the main component of the exine and is exported from intact tapeta during the tetrad stage to the early bicellular pollen stage (Quilichini et al., 2014a,b). Within this time frame, lipid transfer proteins are abundant in the locule fluid (Huang et al., 2013), which is in direct contact with both the tapetal cells and developing pollen grains. Interestingly, lipid transfer proteins are also found in the pollen coat after tapetum programmed cell death (Huang et al., 2013; Quilichini et al., 2014a). Therefore, it is speculated that some PCP constituents appear in the exine and pollen coat via sporopollenin traffic from the tapeta to the developing pollens in flowering species with secretory tapeta (Quilichini et al., 2014a). However, the exact transport pathway of many PCPs in maize still needs to be verified.

### The Functions of Maize PCPs

As maize pollen grains usually germinate within 5 min when landing on the stigma (Heslop-Harrison, 1979), it is difficult to systematically study the functions of maize PCPs in such a short time window. To elucidate the functions of maize PCPs during pollen–stigma interaction, pollen germination and tube growth, mutant pollen grains were generated using antisense or RNAi techniques to measure the resulting phenotype of pollen grains *in vitro* and *in vivo*. Among the identified maize PCPs, only three (xylanase, β-glucanase, and β-expansin 1) have been functionally characterized in maize, though the homologs of other PCPs have been studied in other tissues or species (Bashir et al., 2013; Zhang et al., 2014).

According to their synthesis pathways, the maize PCPs discussed herein can be divided into three groups. Group 1 includes tapetum-synthesized hydrolases, e.g., xylanase, βglucanase, cysteine protease, and subtilase. For successful germination, pollen gains need to overcome the mechanical resistance from a thin protein layer and the polysaccharide-rich wall in the style (stigma). It may be the result of evolution that these enzymes become part of pollen coats to facilitate the penetration of pollen tube into the style tissues during sexual reproduction.

After tapetum cells rupture, their contents are scattered and enveloped by the pollen. Xylanase on the pollen coat, together with other hydrophilic components, initially helps to provide sufficient water along the pollen surface to the aperture for pollen germination. At the same time, xylanase begins the hydrolysis of carpel wall xylan to create an opening for pollen tube entry (Suen and Huang, 2007). Along with coat xylanase, coat β-glucanase also hydrolyzes the stigma wall during pollen germination (Suen et al., 2003). Pollen coat β-glucanase is notably different from the enzyme that hydrolyzes the callose wall of the microspore tetrad (Suen et al., 2003; Takeda et al., 2004). After the pollen tube penetrates into the stigma, β-glucanases appear to hydrolyze these glucans, playing an important role in the regulation of pollen

b*Molecular function annotated in UniProtKB* 

*(http://www.uniprot.org/).*

Gong et al. Maize pollen coat proteome

tube elongation (Takeda et al., 2004). In addition, xylanase and cysteine protease on the pollen coat of Bermuda grass have dual functions: IgE-binding capacity and proteolytic activity, which disrupts the integrity of the human airway epithelial cell barrier (Bashir et al., 2013). It would be interesting to examine the potential proteolytic activity of both PCPs on the surface of maize stigma cells.

Cysteine protease exists widely in animals, plants and parasites. In *Arabidopsis*, cysteine protease is a key executor involved in tapetal programmed cell death and thus regulates pollen development (Zhang et al., 2014). In maize, cysteine protease is the only known tapetum protease that appears at a very late stage of anther development (Li et al., 2012). Based on its substantial abundance in maize pollen coat (Wu et al., 2015), cysteine protease may function by interacting with (or hydrolyzing) a thin layer of proteins on the stigma surface during pollen germination.

Group 2 comprises pollen-synthesized proteins, including various allergens, such as Zea m 1 (β-expansins 1, 10), Zea m 12 (profilin), and Zea m 13 (exopolygalacturonase). Profilin is the main monomer actin-binding protein involved in cytoskeleton organization and is essential for tip growth of plant cells (Vidali et al., 2007). In addition to the existence in cytosol, profilin exists in plasma membrane (Marmagne et al., 2007) and cell wall (Van Damme et al., 2004)*.* In maize pollen coat, profilin may strengthen the protective role of the coat via binding phospholipids.

After pollen landing on the stigma, these pollen-synthesized proteins may be released to assist in pollen tube penetration into the stigma and in the subsequent hydrolysis and modulation of the carpel interior wall. For example, polygalacturonase (Zea m 13) hydrolyzes the pectin between adjacent cells in the transmitting track and facilitates tube advance (Suen et al., 2003). Zea m1(β-expansin) can loosen stigmatic cell walls and aid pollen tube penetration of the stigma (Cosgrove et al., 1997). A reduction in Zea m 1 level in maize pollen by insertional mutation was found to greatly affect pollen tube growth rates and thus pollen competition *in vivo* (Valdivia et al., 2007). Moreover, the pollen deficient in Zea m 1 gene expression tended to form large aggregates, leading to poor pollen dispersal upon anther dehiscence, and the emerging pollen tubes had difficulty entering the style (Valdivia et al., 2009). So, Zea m 1 is required for pollen separation and stigma penetration and plays an important role in determining the outcome of pollen competition *in vivo* for access to ovules.

# References


Group 3 includes various proteins, such as ABA-induced caleosin, rho GDP-dissociation inhibitor 1 (RhoGDI 1) and rasrelated protein Rab-2-A (RAB2A). Caleosin has been detected in the pollen coats of *A. thaliana* (Mayfield et al., 2001), *B. napus* (Murphy, 2006), and maize (Wu et al., 2015). Structurally, caleosin contains an N-terminal region with a single Ca2+-binding EF-hand domain, a central hydrophobic region, and a C-terminal region with several putative protein kinase phosphorylation sites. These signaling-related motifs suggested that caleosin may play a role in pollen–stigma communication. RhoGDI is involved in Rac/Rop GTPase-regulated pollen tube growth through prevents the formation of transversal actin bands (Fu et al., 2001; Chen et al., 2003). Similarly, RAB2A belongs to the rab GTPase family that is important for pollen tube tip growth (de Graaf et al., 2005; Szumlanski and Nielsen, 2009). Thus, the biological functions of RhoGDI 1 and RAB2A in maize pollen coat are worth to be studied.

# Conclusion and Perspectives

Despite the diversity of known maize PCPs, only a few PCPs have been functionally characterized. Indeed, the functions of other PCPs, especially those in large abundance in the maize pollen coat, remain unclear, though some functional clues are available through informatics analyses. The precise biological role of uncharacterized maize PCPs can be established in the future through the analysis of transgenic mutant pollen by gain-offunction or loss-of function approaches. An understanding of the comprehensive knowledge of maize PCPs will help to illuminate the mechanism by which PCPs are involved in pollen–stigma interactions in maize and other crop plants.

# Acknowledgments

Our research was funded by the National Natural Science Foundation of China (Grant No. 30971705), the Plan for Scientific Innovation Talent of Henan Province (Grant No. 144200510012) and Program for Innovative Research Team (in Science and Technology) in University of Henan Province (Grant No. 15IRTSTHN015).

L.) pollen as allergen carriers and initiators of an allergic response*. Int. Arch. Immunol.* 118, 414–418. doi: 10.1159/000024151


death, regulates pollen development in *Arabidopsis*. *Plant Cell* 26, 2939–2961. doi: 10.1105/tpc.114.127282

**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.

*Copyright © 2015 Gong, Wu and Wang. 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.*

# **The role of proteomics in progressing insights into plant secondary metabolism**

*María J. Martínez-Esteso <sup>1</sup> , Ascensión Martínez-Márquez <sup>1</sup> , Susana Sellés-Marchart 1,2 , Jaime A. Morante-Carriel <sup>3</sup> and Roque Bru-Martínez <sup>1</sup> \**

*<sup>1</sup> Plant Proteomics and Functional Genomics Group, Department of Agrochemistry and Biochemistry, Multidisciplinary Institute for Environmental Studies "Ramon Margalef", University of Alicante, Alicante, Spain, <sup>2</sup> Biotechnology and Molecular Biology Group, Quevedo State Technical University, Quevedo, Ecuador, <sup>3</sup> Proteomics and Genomics Division, Research Technical Facility, University of Alicante, Alicante, Spain*

### *Edited by:*

*Joshua L. Heazlewood, The University of Melbourne, Australia*

> *Reviewed by: Marc Boutry, KU Leuven, Belgium Sebastien Carpentier, KU Leuven, Belgium*

### *\*Correspondence:*

*Roque Bru-Martínez, Plant Proteomics and Functional Genomics Group, Department of Agrochemistry and Biochemistry, Multidisciplinary Institute for Environmental Studies "Ramon Margalef", University of Alicante, Carretera San Vicente del Raspeig s/n, 03690 Alicante, Spain roque.bru@ua.es*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 28 February 2015 Accepted: 23 June 2015 Published: 07 July 2015*

### *Citation:*

*Martínez-Esteso MJ, Martínez-Márquez A, Sellés-Marchart S, Morante-Carriel JA and Bru-Martínez R (2015) The role of in proteomics progressing insights into plant secondary metabolism. Front. Plant Sci. 6:504 doi: 10.3389/fpls.2015.00504* The development of omics has enabled the genome-wide exploration of all kinds of biological processes at the molecular level. Almost every field of plant biology has been analyzed at the genomic, transcriptomic and proteomic level. Here we focus on the particular contribution that proteomic technologies have made in progressing knowledge and characterising plant secondary metabolism (SM) pathways since early expectations were created 15 years ago. We analyzed how three major issues in the proteomic analysis of plant SM have been implemented in various research studies. These issues are: (i) the selection of a suitable plant material rich in secondary metabolites of interest, such as specialized tissues and organs, and *in vitro* cell cultures; (ii) the proteomic strategy to access target proteins, either a comprehensive or a differential analysis; (iii) the proteomic approach, represented by the hypothesis-free discovery proteomics and the hypothesis-driven targeted proteomics. We also examine to what extent the mostadvanced technologies have been incorporated into proteomic research in plant SM and highlight some cutting edge techniques that would strongly benefit the progress made in this field.

**Keywords: secondary metabolism, polyphenols, alkaloids, terpenoids, cell cultures, shotgun proteomics, DIGE, MRM**

# **Introduction**

The development of omics has enabled the genome-wide exploration of all kinds of biological processes at the molecular level. In the field of plant biology, growth and development, organogenesis, stress response or fundamental processes, such as organelle biogenesis, cell cycle or metabolism, have been analyzed at the genomic and proteomic levels (Jorrín et al., 2007; Yuan et al., 2008; Fukushima et al., 2009; Jorrín-Novo et al., 2009, 2015; Gapper et al., 2014). Here we focus on the particular contribution of proteomic technologies to the progress made in the knowledge and characterization of plant secondary metabolism (SM) pathways since the early expectation created 15 years ago (Jacobs et al., 2000). Major interest in this research relies on the discovery of the genes and enzymes involved in the biosynthesis of bioactive and high-value natural products to bridge in the still many knowledge gaps in their respective pathways. On top of all this, the discovery and characterization of the transcription factors involved in the regulation of SM pathways is the ultimate frontier in this field (Yang et al., 2012). Movement of metabolites within cell compartments, between cell types, and even between plant organs, confers a three-dimensional character to the metabolic map. Thus the proteins that act as transporters and carriers have to be included in the pathways. A direct application of such knowledge consists in developing or improving metabolic engineering strategies in different host organisms, including whole plants or plant cells, or microbial systems that are especially suitable for producing the desired metabolites under the cell factory concept (Verpoorte et al., 2000; Yang et al., 2014; Farré et al., 2015). Perhaps one major hurdle for scientists to approach this problem through proteomics is the generally lower abundance of SM proteins, including enzymes, transporters, and especially transcription factors. Other reasons also include: (i) some pathways occur in specialized cells, tissues and organs (Kutchan, 2005), such as the monoterpene biosynthetic pathway in glandular trichomes of peppermint (Turner and Croteau, 2004), which may be difficult to collect in abundance; (ii) fragmented pathways also exist, and each fragment occurs in a different cell type, which can be quite distant within the plant, and some steps can even occur during transport (Kutchan, 2005); e.g., five key enzymes of alkaloid formation in opium poppy occur in multiple cell types and different plant organs (Weid et al., 2004); (iii) each cell or tissue usually contains normal levels of primary metabolism enzymes and housekeeping proteins that involve a high background "noise" of non-target proteins; (iv) some pathways or key enzymes are inducible and are present at a significantly lower level than other enzymes of the same or pathways; (v) most pathways of interest are specific of the under-represented species in sequence databases. Thus modelplant genomic and proteomic information may not be useful as they lack these pathways, at least in part, which precludes extensive protein identification. These issues have been taken into account in some of the studies carried out, but not in them all, and should always be borne in mind for future studies, as well as the limitation of proteomic techniques, which are common to the analysis of samples of any biological origin.

# **Plant Material and Secondary Metabolism**

One first issue is to select suitable plant material that is rich in secondary metabolites of interest (**Figure 1**). In a literature survey about proteomics-based research into plant SM, the first impression is that major families, i.e., phenolics, alkaloids, and terpenes, have been investigated to some extent. However, the high diversity of the metabolic pathways within them is quite under-represented, thus it is necessary to broaden the proteomic research scope in this field. The success of a proteomic approach in finding and discovering new potential pathway enzymes very much relies on the choice of a suitable plant part where the target pathway is over-represented. In grape berry skin, a flesh proteomic analysis has provided extensive coverage of not only shikimate, phenylpropanoid and flavonoid pathway enzymes, but also of their relative abundance profiles during berry development (Martínez-Esteso et al., 2011a,b, 2013). Isolation and comprehensive analyses of trichomes in tobacco (Van Cutsem et al., 2011) and tomato (Schilmiller et al., 2010) have led to the identification of the enzymes involved in the methylerythritol phosphate (MEP) synthesis pathway of terpenoid precursors, terpenoid synthesis and modification, and also potential transporters. The study of tomato, which was combined with a transcriptomic analysis, has also revealed pathways of flavonoid and volatile aldehydes synthesis which occur in trichomes; most interestingly, morphologically identical trichomes from different plant parts appeared to be specialized in the terpenoid metabolite type produced by a sesquiterpene synthase, only found at the protein level in leaf, but not stem, trichomes (Schilmiller et al., 2010). Parts of complex biosynthetic pathways, such as that for alkaloids, may occur in conducting fluids (e.g., phloem and latex). An analysis of latex has revealed the presence of enzymes of SM. Yet a major factor for their successful identification is the availability of the extensive structural annotation of sequences in the databases of the target species. For *Euphorbia kansui* (Zhao et al., 2014), only four of the 19 identified proteins had a functional description, while several hundreds of proteins, including various morphine synthesis steps, have been identified and described for opium poppy (Onoyovwe et al., 2013). The preparation of subcellular fractions specialized in secondary metabolite synthesis, such as chromoplasts from orange fruit pulp (Zeng et al., 2011), has allowed the identification of most of the enzymes of the MEP pathway and lycopene synthesis, and also one enzyme involved in vitamin E. However, it has been noted that it was not possible to identify the enzymes that catalyze the regulated steps in each pathway.

Besides organ and tissues specialized in particular SM pathways, *in vitro* cell cultures have been considered the ideal biological material for equivalence with specialized tissues in which a metabolic pathway occurs or can be induced through elicitation or stress under laboratory controlled conditions. In fact most proteomic studies about SM have been carried out with elicited cell cultures. Among polyphenolics biosynthesis, stilbenoid in grapevine (Martínez-Esteso et al., 2011c; Ferri et al., 2014), flavonolignan in *Silybum marianum* (Corchete and Bru, 2013), lignans in *Podophyllum hexandrum* (Bhattacharyya et al., 2012), isoflavones in *Medicago truncatula* (Lei et al., 2010), and chalcone derivatives in *Boesenbergia rotunda* (Tan et al., 2012) have been analyzed at proteome level under the induction of elicitors, such as chitosan, cyclodextrins, methyl jasmonate or yeast extract, and either individually or combined; i.e., cyclodextrin and methyl jasmonate. In addition to the expected enzymes of the biosynthetic pathway, the bonus proteins, which are potentially involved in movement or modification of end products, have been found to be co-induced; e.g., secretory peroxidases (Martínez-Esteso et al., 2009), glutathione-S transferase (Martínez-Esteso et al., 2011c), Rab11C and ABC transporter (Corchete and Bru, 2013) and laccase (Lei et al., 2010). Alkaloid synthesis has been investigated at the proteome level in cell cultures of California poppy: benzophenanthridinetype (Oldham et al., 2010), opium poppy: benzylisoquinolinetype (Desgagné-Penix et al., 2010), and Madagascar periwinkle: terpenoid indole-type (Champagne et al., 2012), and elicitation and phenotype comparison strategies have been adopted for differential analyses. As these pathways are large and complex, the metabolic profile of cell suspensions may differ from that of plant tissues, even under elicitation. Thus the cell culture approach is useful for analyzing specific parts or branches of the target pathway; e.g., secoiridoid as part of terpenoid indole alkaloid (TIA) synthesis (Champagne et al., 2012), or the sanguinarine branch of benzylisoquinoline alkaloids (BIAs; Desgagné-Penix et al., 2010).

# **Proteomic Strategies and Approaches in Secondary Metabolism**

A relevant issue that hampers the detection of the specific proteins involved in SM in non-model plant species is the fact that fully sequenced genomes are not available, which represents a significant challenge. Two main approaches have been adopted to overcome the problem in the studies reviewed. Highly conserved proteins can be identified by sequence homology to *Arabidopsis thaliana* and other plant species (Jacobs et al., 2005; Cheng and Yuan, 2006; Bhattacharyya et al., 2012; Nagappan et al., 2012). Alternatively, specific EST databases (Desgagné-Penix et al., 2010; Schilmiller et al., 2010; Champagne et al., 2012) have been created and used for protein identification. Sequence annotation remains a huge challenge as model organisms have only a limited set of SM. The retrieval of Gene Ontology terms from the closest related protein via a BLAST similarity search with the BLAST2GO tool (Conesa et al., 2005) has proven a successful strategy to functionally annotate proteomic experiments in grapevine (Martínez-Esteso et al., 2011a,b) early after the genome sequence release. Although caution is necessary since many annotations are too general, and a minority is manually curated and based on experimental evidence (Yon Rhee et al., 2008).

As depicted in **Figure 1**, the strategy to find proteins of interest to SM, and the technical approach to achieve this, are in fact common to any proteomic analysis type. Essentially, the strategy consists of a comprehensive analysis, in which the identification of the largest possible number of proteins is intended; or a differential analysis, which focuses only on those proteins with differential abundance across samples. Much of the information obtained in a proteomic comprehensive analysis can be provided through mRNA sequencing at a significantly lower cost and with better dynamic range coverage (Zubarev, 2013), which might be the main reason for it being used much less frequently than differential proteomics as a strategy (**Figure 1**). Yet the structural and quantitative information obtained through different proteomic strategies still remains an essential tool to identify the acting metabolic enzymes co-localized with the target metabolites in the cell, and to also identify post-translational modifications (PTMs), including proteolytic processing, characterization of protein complexes and, to some extent, the analysis of protein–protein interactions. All of this has given rise to a new way of annotating the genomes called proteogenomics (reviewed in Nevizhskii, 2014). Therefore, the scientific value of undertaking a comprehensive analysis of the global protein expression has been recognized (Baginsky et al., 2010). This strategy has been applied to massively identify and characterize protein content in plant organelles, such as plastids and chromoplast, in specialized plant organs such as trichomes, or in elicited cell cultures.

Since the proteome of a cell or tissue, and even of a purified organelle or cell compartment, is extremely complex, pre-fractionation and protein or peptide separation steps are crucial for broadening SM proteome coverage. The analysis of purified plastid proteomes, followed by differential protein solubilization, has led to the ability to resolve approximately 1,000 different protein species per fraction displayed on 2DE gels (von Zychlinski and Gruissem, 2009). Moreover from purified orange pulp chromoplasts, 418 proteins have been identified as plastid proteins following a shotgun analysis of the protein extract digests (Zeng et al., 2011), of which about 50 were classified as SM biosynthesis. The identification of membrane proteins, especially ABC transporters for their role in SM transport, is particularly important for shedding light on metabolite trafficking across cell and organelle membranes. The shotgun approach is quite suitable for achieving this goal since 2-DE performance for the separation of hydrophobic proteins is compromised. Extensive fractionation and identification of peptides by LC MALDI MS/MS of the microsomal fraction of *Nicotiana tabacum* trichomes, followed by a sequence analysis for transmembrane span prediction, have led to the identification of 165 membrane proteins, of which 39 were putative transporters, and eight were of the ABC type (Van Cutsem et al., 2011). The analysis of whole protein extract digests requires exhaustive separations at the peptide level in order to go more deeply into the proteome. Using multidimensional LC-separation of the peptides mixtures coupled to MS, an increased proteome depth of Madagascar Periwinkle cells has been achieved, compared to conventional 1D-LC peptide separation in shotgun proteomics, and has increased the number of the identified proteins by 40%. Of the 1,663 identified proteins, 22 were involved in TIA biosynthesis and 16 transporters were potentially implicated in SM transport. About 30% of the identified proteins perform an unknown function, which indicates the large gap in SM knowledge. Apart from identification by analogy with other plant species of already characterized proteins, which act in well-defined steps of the TIA biosynthesis pathway, several proteins, including alcohol dehydrogenases, terpene cyclases, cytochrome P450-dependent monooxygenases, glycosyl- and methyl-transferases, have been proposed as candidates to be involved in other unknown steps of this pathway (Champagne et al., 2012). The availability of speciesspecific nucleotide databases and the technical approach have proved critical to successfully access the SM proteome, which thus demonstrates the power of combining deep mRNA and protein sequencing to access plant-specific SM. Only one enzyme involved in the biosynthesis of BIAs has been identified after a 2-DE-based workflow in opium poppy (*Papaver somniferum*) cell cultures (Zulak et al., 2009a). In contrast, the former creation of a specific EST database from a deep transcriptome analysis, followed by an LC-MS/MS analysis of 1DE-fractionated whole protein extracts, has enabled the identification of 1,004 proteins, which included almost all the enzymes of the pathway (Desgagné-Penix et al., 2010). The creation of a large trichome-specific EST collection and a translated protein database, followed by a similar proteomic workflow on the proteins extracted from mixed and purified type VI tomato trichomes, has led to 1,973 database hits corresponding to 1,552 proteins (Schilmiller et al., 2009, 2010).

**FIGURE 1 | Major issues in the proteomic analysis of plant secondary metabolism.** Three major issues have been considered in the proteomic analysis of plant secondary metabolism. Two are common to any type of proteomic analysis, i.e., the strategy to find proteins of interest, and the technical approach to achieve it. In the scheme above, we have included the ways in which such issues have been resolved to date and the corresponding number of representative studies (figure in brackets). So the *first and specific issue* for accessing the plant secondary metabolism proteome is the selection of suitable plant material, which is rich in secondary metabolites of interest. If using whole plants as a source, collection of specialised tissues-roots, fruit exocarp and mesocarp-, organs –trichomes-, fluids –milky sap- or preparation of organelles –chromoplasts- before starting protein extraction has been a successful strategy to access the target proteome. Alternatively, *in vitro* cell culture has been a smart option to easily generate an abundant population of homogeneous cells that produce SM whenever they were stimulated through different treatments, such as elicitation, precursor feeding or physical stress. A *second issue* is the proteomic strategy to find target proteins; i.e., enzymes and transporters specifically involved in the metabolic pathway of interest. One is a comprehensive analysis in which the identification of the largest possible number of proteins is intended. The other typical strategy is differential proteomics. In this case, the proteome complements obtained from two

In differential proteomics analyses, a significant amount of information can be collected, as reflected in the literature. In the top-down approach, classical 2-DE continues to be the choice for most comparative proteomic studies which aim to find deregulated proteins that act on the metabolic pathways involved in the biosynthesis of important secondary metabolites. A few key enzymes directly related to anthocyanin biosynthesis, and their accumulation on the skin of grape berries, have been detected while studying the effect of a low leaf-to-fruit ratio (Wu et al., 2013), sunlight exclusion (Niu et al., 2013b) and the difference of the red-to-white cultivar (Niu et al., 2013a). In two other studies, experimental groups or more, which differ in secondary metabolite content, are compared. Proteins with differential abundance are selected. In both cases, a bioinformatics-based analysis of the protein lists follows to classify proteins according to their molecular and (potential) biological function, and to select the candidate proteins involved in SM for further functional characterization using biochemical and genetic tools. Eventually, a *third major issue* is the proteomic approach. As the initial goal is to find the new enzymes and transporters involved in secondary metabolite synthesis and biology, a hypothesis-free type discovery proteomics approach, either top-down or bottom-up, is usually undertaken. A number of applications of classical and advanced gel-based and gel-free proteomic techniques to investigate plant SM pathways have been reported. Having identified the proteins of interest, a hypothesis-driven targeted proteomics approach is the next step to profoundly characterize the pathway under different experimental conditions. For this purpose, proteomic workflows have utilized MRM. Indeed a number of technological developments of immediate applicability that are currently used in proteomics from which SM proteomic research would very much benefit are suggested. These may introduce advantages in handling plant material to obtain cleaner and higher yield protein or peptide samples, and to provide improvements in analytical times, protein identification rates, and quantification of protein changes at either the whole or targeted proteome level.

no clear results on the successful finding of SM enzymes have been obtained by 2-DE differential proteomics (Nagappan et al., 2012; Zhao et al., 2014). This could be due, in part, to the limitations of 2DE to access a lower abundance protein population when whole protein extracts are being analyzed. Apart from exhaustive fractionation, the use of advanced enrichment strategies to detect low abundant proteins such as combinational hexapeptide ligand libraries (CPLL) combined with 2-DE has been applied to analyze the effect of UV stress on *Mahonia bealei* leaves in relation to the induction of the BIAs alkaloid production. As a result, 91 deregulated proteins have been identified, but only after CPLL enrichment, which has led to the finding of several new SM proteins (Zhang et al., 2014).

In the cell cultures that produce SM in response to stimuli, the extent of SM identified proteins in whole protein extracts has slightly improved compared to plant tissues. For example, several phenylpropanoid and monolignol biosynthetic enzymes accumulated in methyl jasmonate-elicited *Podophyllum hexandrum* cell cultures (Bhattacharyya et al., 2012); isoflavonoid biosynthetic enzymes and a putative laccase have been found to accumulate in yeast-elicited *Medicago truncatula* cell cultures (Lei et al., 2010); soluble and microsomal isoforms of stilbene synthase have been identified in chitosan-elicited *Vitis vinifera* cell cultures (Ferri et al., 2014); up to eleven proteins involved in the phenylpropanoid biosynthetic pathway have been found to be up-regulated upon feeding *B. rotunda* callus with phenylalanine to induce the accumulation of chalcone derivative SMs (Tan et al., 2012).

Differential proteomics methods that better perform than single staining 2-DE have also been applied to conduct discovery studies in plant SM. These include differential in gel electrophoresis (DIGE), as a top-down protein-centric approach (Martínez-Esteso et al., 2011c; Champagne et al., 2012; Corchete and Bru, 2013), and isobaric tags for relative and absolute quantitation (iTRAQ) (Martínez-Esteso et al., 2011b, 2013; Robbins et al., 2013) and label-free (Oldham et al., 2010), as bottom-up peptide-centric approaches. DIGE has allowed the identification of three TIA biosynthetic enzymes in *Catharanthus roseus* cell cultures and five others that are putatively involved in the pathway among the 172 identified deregulated proteins (Champagne et al., 2012). The comparison made of DIGE and iTRAQ results has shown a good quantitative correlation between commonly identified proteins (Robbins et al., 2013), but iTRAQ is clearly superior in proteome coverage. In developing and ripening grape berries, DIGE has identified six flavonoid pathway enzymes present in 15 spots (Martínez-Esteso et al., 2011b), while 38 proteins from the shikimate, phenylpropanoid and flavonoid pathways have been identified with iTRAQ (Martínez-Esteso et al., 2013). However, the observed partial overlap between both types of proteome analyses emphasizes the use of different approaches to gain a better picture of the target process under study.

The question as to what approach to use very much depends on the type of information sought. Gel-free approaches combined with extensive fractionation results in broader proteome coverage as compared to gel-based methods. Gel-based techniques greatly benefit the detection of isoforms and post-translationally modified proteins.

Having identified the proteins involved in a SM pathway of interest, a targeted proteomics approach with a multiple reaction monitoring (MRM) methodology in triple quadrupole instruments can be accomplished to study their quantitative behavior under different experimental conditions or to functionally analyze protein isoforms. And so it was that Norway spruce isoforms of terpene synthases and 1-deoxy- xylulose-5-phosphate synthase (DXS) (Zulak et al., 2009b), and loquat polyphenol oxidase isoforms (Martínez-Márquez et al., 2013) have been quantitatively analyzed. Pre-fractionation of peptides can also be implemented to improve signal quality in MRM. In the MRM analysis of strawberry proteins related to flavonoid and anthocyanin biosynthesis during ripening, peptides have been previously fractionated based on pI (OFFGEL; Song et al., 2015).

# **Future Prospects**

Based on all this information, it is possible to find some important messages: (i) databases need to be enriched in species-specific sequences as this clearly increases the rate of identification of the proteins that act on SM pathways; (ii) gel-based differential analyses are less effective than Gel-free ones to access lower abundance SM proteins, but as the information provided by each technique is only partly redundant, a combined approach is recommended; (iii) plant material selection and sample fractionation are critical for good coverage of SM pathways. Moreover, there are a number of technological developments that are currently used in proteomics of immediate applicability from which SM proteomic research would greatly benefit. Some important ones are described below.

Proteomic approaches and workflows have evolved very quickly in recent years and are ready to be exploited to analyze the plant SM proteome more profoundly; i.e., better resolution chromatography in relatively longer columns accommodated in ultra-HPLC systems, which enables similar performance as extensive 2D-LC fractionation to be achieved in a single shot (Nagaraj et al., 2012).

Metabolic isotope labeling strategies, which are still not used in SM proteomics, are especially suitable for using plant cell cultures fed with isotopic variants of amino acids (Gruhler et al., 2005), known as SILAC, or fed with cost-effective <sup>15</sup>N-KNO<sup>3</sup> salt as the sole N source (Engelsberger et al., 2006). The latter is also applicable to whole plants grown in hydroponics (Ippel et al., 2004). Compared to chemical labeling, i.e., iTRAQ, metabolic labeling has the advantage of simplifying sample preparation and handling as heavy and light variants can be mixed at the plant material level to thus reduce technical variability. A major bottleneck of using <sup>15</sup>N-labeling as a differential proteomics strategy is the variable mass shift between the light and heavy forms, dependent on amino acid composition, which implies adding considerable complexity to the quantitative data analysis. The application of SILAC to plant cell cultures has been shown to be restricted to incomplete isotopic label incorporation because plants are able to "*de novo*" synthesize all proteinogenic amino acids, and to thus compete with the labeled amino acids fed into the culture medium. Recently, a smart modification of the SILAC protocol has enabled the accurate reproducible application of SILAC to plant cells grown in the dark (Schütz et al., 2011), thus providing a new powerful tool to analyze the SM proteome.

The spectacular development of instrumentation for LC-MS of peptides over the last decade has almost left protein sample preparation, including extraction and digestion, as the one major critical point in proteomic workflows in the overall performance of proteomic experiments. Cleanness of samples in relation to non-protein contaminants dramatically affects the protein identification rate. The current trend in simplifying sample preparation steps and handling minimal quantities of biological material has led to the integration of protein extraction, digestion, and fractionation in a single pipette tip that holds a small disk of membrane-embedded separation material, the so-called StageTip (Rappsilber et al., 2003; Kulak et al., 2014). Extrapolating these protocols to plant material is challenging given protein scarcity and the abundance of interfering compounds in plant cells, but it is an exciting challenge because the benefits for research of SM will outweigh development efforts.

Targeted proteomics using MRM is a powerful proteomic approach that has been scarcely used to investigate plant proteomics, and more specifically SM proteomics. This hypothesis-driven relatively new approach requires the previous selection of the target proteins and their proteotypic peptides to monitor and/or quantify them either in a relative or an absolute fashion (Lange et al., 2008). Since the availability of genome-wide sequences, as well as annotated spectral libraries, are critical in the application of this methodology, plants are currently at a disadvantage compared to mammalian and microbial organisms. Future efforts in plant proteomics should generally focus on this issue. In addition to the above-cited examples (Zulak et al., 2009b; Martínez-Márquez et al., 2013; Song

# **References**


et al., 2015), MRM might become the gold standard to validate proteomic experiments as an alternative to using western blotting (Lehmann et al., 2008; Aebersold et al., 2013), and it could be particularly relevant in plant SM proteomics thanks to the highly limited availability of specific antibodies. Quantification of target enzymes allows a much larger number of conditions analyzed per experiment, which accelerates the process to reveal key enzymes. As the number of unveiled target proteins, i.e., enzymes and transporters, directly involved in the SM pathways increases, it is envisaged that this approach will be used extensively in the near future.

# **Acknowledgments**

This work has been supported by grants from the Spanish Ministry of Science and Innovation (BIO2011-29856-C02-02; BIO2014-51861-R), European Funds for Regional Development (FEDER) and Conselleria d'Educacio, Cultura I Sport de la Generalitat Valenciana (FPA/2013/A/074). JM-C acknowledges a postdoctoral and research grants from SENESCYT GOVERNMENT OF ECUADOR (006-IECE-SMG5-GPLR-2012 and Programa1\_Senescyt\_2014) and a grant from UTEQ (UTEQ-Ambiental-9-FCAmb-IFOR-2014-FOCICYT002).


**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.

*Copyright © 2015 Martínez-Esteso, Martínez-Márquez, Sellés-Marchart, Morante-Carriel and Bru-Martínez. 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.*

# Towards a common bean proteome atlas: looking at the current state of research and the need for a comprehensive proteome

Sajad M. Zargar <sup>1</sup> \*, Muslima Nazir <sup>2</sup> , Vandna Rai <sup>3</sup> , Martin Hajduch<sup>4</sup> , Ganesh K. Agrawal <sup>5</sup> and Randeep Rakwal 5, 6

*<sup>1</sup> School of Biotechnology, SK University of Agricultural Sciences and Technology of Jammu, Jammu, India, <sup>2</sup> Department of Botany, Jamia Hamdard University, New Delhi, India, <sup>3</sup> National Research Centre on Plant Biotechnology, Indian Agricultural Research Institute, New Delhi, India, <sup>4</sup> Reproduction and Developmental Biology, Institute of Plant Genetics and Biotechnology, Slovak Academy of Science, Nitra, Slovakia, <sup>5</sup> Research Laboratory for Biotechnology and Biochemistry, Kathmandu, Nepal, <sup>6</sup> Organization for Educational Initiatives, University of Tsukuba, Tsukuba, Japan*

Keywords: nutrition, proteome atlas, common bean, human health, genomics

### Edited by:

*Joshua L. Heazlewood, The University of Melbourne, Australia*

### Reviewed by:

*Subhra Chakraborty, National Institute of Plant Genome Research, India Michael A. Grusak, United States Department of Agriculture-Agricultural Research Service Children's Nutrition Research Center, USA*

### \*Correspondence:

*Sajad M. Zargar, smzargar@gmail.com*

### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

> Received: *29 December 2014* Accepted: *13 March 2015* Published: *27 March 2015*

### Citation:

*Zargar SM, Nazir M, Rai V, Hajduch M, Agrawal GK and Rakwal R (2015) Towards a common bean proteome atlas: looking at the current state of research and the need for a comprehensive proteome. Front. Plant Sci. 6:201. doi: 10.3389/fpls.2015.00201*

# Common Bean, Human Health, and Food Security

The common bean (Phaseolus vulgaris L.) is a legume (Leguminosae; Cronquist, 1981) with high nutritional value that may be a critical food source in our battle to have food for all, i.e., food security worldwide (FAOSTAT, 2013). Nutrition-wise, the common bean is an important protein-rich, low-fat, nutrient-dense food containing high amounts of energy as carbohydrates. It is also rich in minerals (mainly Fe and Zn) (Pinheiro et al., 2010), and potentially disease preventing with health-promoting compounds exhibiting pharmaceutical properties (Hayat et al., 2014). Legumes, via their ability to fix atmospheric nitrogen, also play an important role in sustainable agriculture (Liu et al., 2011). Like other crops, the common bean bears the brunt of a changing climate resulting in diverse abiotic and biotic stresses which reduces its yield and nutritional status. These factors necessitate new research into its biology. An exponential increase in population size and current trends in bean consumption due to the presence of vital nutrients demands both its cultivation and high productivity in many parts of the world.

# Common Bean Genome Research Milestones at a Glance

Improvement of the common bean means possessing in-depth knowledge of its genetic diversity, the genome and gene functions, to enable the analysis of pathways and networks in response to fluctuating environmental conditions. Various genomic resources for common bean are available and include physical maps, bacterial artificial chromosome libraries, anchored physical and genetic maps, expressed sequence tags, and the recently published complete genome sequence (O'Rourke et al., 2014; Schmutz et al., 2014). The 473 Mb genome sequence of common bean will help scientists to understand the evolution of the crop, synteny with other legumes and is a repository of genetic information for molecular breeders. Establishment of a 2.12-Mb transposon database for the common bean (www.phytozome.org), which includes 791 representative transposon sequences, will serve as an important resource for understanding genome evolution and genetic variation (Gao et al., 2014). In combination, a systems biology approach is required for in-depth analysis of the molecular substrates and target moieties regulating various metabolic pathways which is made possible by coherently integrating "omics" data. Here, we dwell upon the necessity for developing a common bean (P. vulgaris) proteome resource and thereby help translate this information into improving its nutritional value and to develop a more sustainable crop.

# Common Bean Proteomes

During past 10 years, common bean proteome studies have ranged from abiotic to biotic stress, seed to storage proteins, and diversity. Research in abiotic stresses have examined the effect of the gaseous pollutant ozone, revealing distinct protein changes in affected leaves by two-dimensional electrophoresis (2- DGE; Torres et al., 2007). Phosphoproteomic analysis revealed that enhanced phosphorylation of dehydrin plays a major protective role in the reversibility of cell wall extensibility during recovery from osmotic stress induced physical breakdown of cell wall structures (Yang et al., 2013). 2D-DIGE analysis showed the impact of drought stress on different biological pathways and molecular functions in leaves of drought tolerant and sensitive cultivars (Zadražnik et al., 2013). Response to chilling stress was dependent on length and manner of exposure to low temperature, as determined by divergent proteomic alterations in roots in response to varying periods of low temperature stress (Badowiec and Weidner, 2014).

Rust fungus infection of leaves revealed R-gene based defense modulates proteins similar to those in the basal system defense (Lee et al., 2009). Approaches have been developed to improve protein extraction from seeds with TCA–acetone followed by a clean-up step resulting in the highest amount of storage and defense proteins over a phenol method (De La Fuente et al., 2011). The TCA-acetone method for seeds in combination with 2-DGE analysis was applied to the analysis of improved common beans through the alteration of protein components (Natarajan et al., 2013). A further study explained the role of posttranslational modifications of phaseolin proteins demonstrating phosphorylation during the transition from seed dormancy to an early germination stage (López-Pedrouso et al., 2014). Previously, seed proteomics had revealed that a lack of storage proteins leads to increased legumin, albumin-2, defensin and albumin-1, which contribute to elevated sulfur amino acid content and raffinose metabolic enzymes while simultaneously down-regulating

the secretory pathway (Marsolais et al., 2010). A 2-DGE analysis revealed differences among cultivated and wild-type common bean cultivars as evidenced from differences in number as well as abundance of protein spots on gels, and used these factors to classify these cultivars based on their Centre of Domestications (COD; Mensack et al., 2010). Recently, a legume specific protein database (LegProt, http://bioinfo.noble.org/manuscript-support/ legumedb) has been created which contains sequences of several legumes including the common bean. This resource significantly increases legume protein identification success rates and the confidence levels compared to the commonly used database NCBInr (Lei et al., 2011).

# Toward Common Bean Proteome Atlas: Understanding Global Regulation

These previous studies have all individually contributed to specific proteomes of the common bean. However, considering a focus on the improvement of the common bean, a comprehensive and systematic approach is required at the proteome level with an overall goal of creating a "PROTEOME ATLAS." As such, studying the proteome from all the organs of the plant (leaves, stems, roots, flowers, pods, and developing and mature seed) under specific conditions (disease or other environmental conditions) and at particular developmental stages along with the organelle and secreted proteomes, will collectively help to identify unexplored pathways that can be utilized and targeted to address specific problems associated with this legume. For example, our group has developed a collection of diverse germplasm of the common bean which mainly includes landraces from Jammu and Kashmir, India (Zargar et al., 2014). Availability of landraces from other specific geographical locations from around the world will also deliver unique genetic stocks for valuable traits (e.g., abiotic or biotic stress tolerance). A big challenge will be the large number of experiments to be performed using both gel-based and gel-free approaches along with new methods for unraveling low-abundance/rare proteins. **Figure 1** provides a snapshot of the work-plan, along with Phase I of the workflow as a first target for our research. Essentially, a particular trait and organ/tissue will be focused on under specific criteria to create a unique database (2D-map image and protein information) for that specific proteome. We are of the opinion that execution of this huge task will require collaboration among different laboratories that have expertise in various aspects of proteomic technologies. We intend to invite members of plant proteomics community through the INPPO platform (http://www. inppo.com/). Overall, proteomics linked with genomics and transcriptomics will likely enhance the agronomic merit as well as quality traits in the common bean by enabling us to first understand regulatory pathways and then enable the manipulation of these regulatory pathways to attain an improved and more sustainable crop.

## References


polyethylene glycol-induced osmotic stress in root tips of common bean (Phaseolus vulgaris L.). J. Exp. Bot. 64, 5569–5586. doi: 10.1093/jxb/ ert328


**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.

Copyright © 2015 Zargar, Nazir, Rai, Hajduch, Agrawal and Rakwal. 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.

# The Amborella vacuolar processing enzyme family

Valérie Poncet 1 †, Charlie Scutt 2 †, Rémi Tournebize<sup>1</sup> , Matthieu Villegente<sup>3</sup> , Gwendal Cueff 4, <sup>5</sup> , Loïc Rajjou4, 5, Thierry Balliau<sup>6</sup> , Michel Zivy <sup>6</sup> , Bruno Fogliani 3, 7 , Claudette Job<sup>8</sup> , Alexandre de Kochko<sup>1</sup> , Valérie Sarramegna-Burtet <sup>3</sup> \* and Dominique Job5, 8 \*

### Edited by:

Joshua L. Heazlewood, The University of Melbourne, Australia

### Reviewed by:

Manuel Martinez, Universidad Politécnica de Madrid, Spain Barend Juan Vorster, University of Pretoria, South Africa

### \*Correspondence:

Valérie Sarramegna-Burtet, Laboratoire Insulaire du Vivant et de l'Environnement, Université de la Nouvelle-Calédonie, BP R4, 98851 Nouméa, New Caledonia valerie.sarramegna@univ-nc.nc; Dominique Job, Centre National de la Recherche Scientifique/Bayer CropScience Joint Laboratory, Bayer CropScience, 14-20 rue Pierre Baizet, Lyon F-69263, France job.dominique@gmail.com

†Co-first authors.

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 07 May 2015 Accepted: 27 July 2015 Published: 21 August 2015

### Citation:

Poncet V, Scutt C, Tournebize R, Villegente M, Cueff G, Rajjou L, Balliau T, Zivy M, Fogliani B, Job C, de Kochko A, Sarramegna-Burtet V and Job D (2015) The Amborella vacuolar processing enzyme family. Front. Plant Sci. 6:618. doi: 10.3389/fpls.2015.00618 1 Institut de Recherche pour le Développement, UMR Diversité, Adaptation et Développement des Plantes, Montpellier, France, <sup>2</sup> Laboratoire Reproduction et Développement des Plantes, UMR 5667, Ecole Normale Supérieure de Lyon, Lyon, France, <sup>3</sup> Laboratoire Insulaire du Vivant et de l'Environnement, Université de la Nouvelle-Calédonie, Nouméa, New Caledonia, <sup>4</sup> Institut National de la Recherche Agronomique, Institut Jean-Pierre Bourgin, UMR 1318 Institut National de la Recherche Agronomique/AgroParisTech, ERL Centre National de la Recherche Scientifique 3559, Laboratoire d'Excellence "Saclay Plant Sciences" (LabEx SPS), RD10, Versailles, France, <sup>5</sup> AgroParisTech, Chaire de Physiologie Végétale, Paris, France, <sup>6</sup> Institut National de la Recherche Agronomique, Plateforme d'Analyse Protéomique de Paris Sud-Ouest, Institut National de la Recherche Agronomique/Université Paris-Sud/Centre National de la Recherche Scientifique/AgroParisTech, UMR 0320/UMR 8120 Génétique Quantitative et Evolution – Le Moulon, Gif-sur-Yvette, France, <sup>7</sup> Institut Agronomique Néo-Calédonien, Diversités Biologique et Fonctionnelle des Ecosystèmes Terrestres, Païta, New Caledonia, <sup>8</sup> UMR 5240 Laboratoire Mixte Centre National de la Recherche Scientifique/Institut National des Sciences Appliquées/Université Claude Bernard Lyon 1/Bayer CropScience, Lyon, France

Most vacuolar proteins are synthesized on rough endoplasmic reticulum as proprotein precursors and then transported to the vacuoles, where they are converted into their respective mature forms by vacuolar processing enzymes (VPEs). In the case of the seed storage proteins, this process is of major importance, as it conditions the establishment of vigorous seedlings. Toward the goal of identifying proteome signatures that could be associated with the origin and early diversification of angiosperms, we previously characterized the 11S-legumin-type seed storage proteins from Amborella trichopoda, a rainforest shrub endemic to New Caledonia that is also the probable sister to all other angiosperms (Amborella Genome Project, 2013). In the present study, proteomic and genomic approaches were used to characterize the VPE family in this species. Three genes were found to encode VPEs in the Amborella's genome. Phylogenetic analyses showed that the Amborella sequences grouped within two major clades of angiosperm VPEs, indicating that the duplication that generated the ancestors of these clades occurred before the most recent common ancestor of living angiosperms. A further important duplication within the VPE family appears to have occurred in common ancestor of the core eudicots, while many more recent duplications have also occurred in specific taxa, including both Arabidopsis thaliana and Amborella. An analysis of natural genetic variation for each of the three Amborella VPE genes revealed the absence of selective forces acting on intronic and exonic single-nucleotide polymorphisms among several natural Amborella populations in New Caledonia.

Keywords: Amborella trichopoda, vacuolar processing enzymes, seed, proteomics, plant evolution, genetic diversity

# Introduction

Evolutionary genetics is considered as a central part of biology (Charlesworth and Charlesworth, 2009). In plants, Amborella trichopoda (Amborella), an understory shrub endemic to New Caledonia, has been proposed to correspond to the single living representative of the sister lineage to all other extant flowering plants (Bremer et al., 2009; Jiao et al., 2011; Lee et al., 2011; Wickett et al., 2014). Hence the recent release of its genome sequence provides a pivotal reference for understanding genome and gene family evolution throughout angiosperm history (Amborella Genome Project, 2013).

In previous work (Amborella Genome Project, 2013), we characterized the Amborella seed storage proteins with the goal of identifying proteome signatures that could be associated with the origin and early diversification of angiosperms. In particular, we focused our attention on the abundant 11S globulins that have been characterized and compared across seed plants in evolutionary analyses (Häger et al., 1995; Adachi et al., 2003; Li et al., 2012). We found that the Amborella genome contains three distinct 11S globulin genes (Amborella Genome Project, 2013). In all plant species, 11S globulins are synthesized in the form of high molecular weight precursors that are processed by vacuolar processing enzymes (VPEs) during seed maturation. This limited proteolysis, which is regularly directed to an Asn-Gly (N-G) junction, yields the A (acidic)- and B (basic)-subunits of mature 11S globulins that is accompanied by further assembly of the trimer precursor-protein complexes into mature hexamers within the protein storage vacuoles (PSVs) (Chrispeels et al., 1982; Müntz, 1998; Shutov et al., 2003).

Although two of the three Amborella 11S globulins do contain a canonical N-G cleavage site, we observed that a third one deviates notably from the two others as it exhibits, in place of an N-G junction, an N-V-I sequence (Amborella Genome Project, 2013). Similar deviations from the N-G cleavage motif were observed for 11S globulins from Ginkgo biloba (Amborella Genome Project, 2013) and Metasequoia glyptostroboides (Häger and Wind, 1997), thus highlighting the possibly ancestral nature of this atypical Amborella 11S globulin.

Most vacuolar proteins (as is the case for the 11S globulins) are synthesized on the rough endoplasmic reticulum (ER) as proprotein precursors and then transported to the vacuoles where they are converted into their respective mature forms (Neuhaus and Rogers, 1998; Herman and Larkins, 1999) by the action of VPEs (EC 3.4.22.34). VPEs, also called legumains or asparaginyl endopeptidases, are cysteine proteases found in various organisms, including plants, mammals, and protozoans such as Schistosoma mansoni. They are classified as members of family C13 in the MEROPS protease database (Rawlings et al., 2008; http://merops.sanger.ac.uk/) that belongs to the CD clan, which also contains caspases (family C14A) and metacaspases (family C14B) (Misas-Villamil et al., 2013). Caspases are the main players in the regulation of programmed cell death (PCD) in animals, whereas metacaspases are involved in the same process in plants and fungi (Hara-Nishimura and Hatsugai, 2011; Tsiatsiani et al., 2011, 2012). Clan CD proteases contain a His– Cys catalytic dyad and have strict substrate requirements for the amino acid preceding the cleavable bond (P1 position) (Chen et al., 2004; Dall and Brandstetter, 2013; Misas-Villamil et al., 2013).

Plant VPEs are classified into vegetative and seed-expressed types (Gruis et al., 2002; Ariizumi et al., 2011; Radchuk et al., 2011; Julián et al., 2013; Kumar et al., 2015). The Arabidopsis thaliana (Arabidopsis) genome contains four VPE genes designated as α-VPE, β-VPE, γ-VPE, and δ-VPE (Rojo et al., 2003; Nakaune et al., 2005; Hatsugai et al., 2015). The seed-type β-VPE is essential for the proper processing of storage proteins (Shimada et al., 2003).

The co-existence in Amborella seeds of the angiosperm- and gymnosperm-type 11S globulins prompted us to characterize the VPE system in seeds of this plant. Here, we refine our understanding of this gene family with the characterization of several Amborella VPE homologs.

Phylogenetic analyses of plant VPEs and legumains have been previously reported. However these previous studies only considered selected sequences from monocots and eudicots and did not include sequences from gymnosperms or basal eudicots (Kato et al., 2003; Nakaune et al., 2005; Julián et al., 2013; Kang et al., 2013; Christoff et al., 2014; Pierre et al., 2014). To gain further insight in plant VPEs and benefiting from the present Amborella sequences, we reconstructed a phylogeny of VPE proteins based on the amino acid sequences of VPEs from a wide range of embryophytes (land plants). By using a comparative approach, combined with the principle of parsimony, data from this uniquely-placed angiosperm can help defining the condition of any character in the most recent common ancestor (MRCA) of the living angiosperms, and we have applied this method to the structural and functional evolution of the VPE family.

Another way to evaluate the functional relevance of genes is to examine the levels of naturally occurring genomic variations therein, i.e., polymorphism within populations (Koornneef et al., 2004). For this purpose we used next-generation sequencing data from the recently completed Amborella genome (Amborella Genome Project, 2013) to characterize singlenucleotide polymorphisms (SNPs) in VPE sequences and their distribution over the natural range distribution of Amborella in New Caledonia (Poncet et al., 2013).

# Materials and Methods

### Plant Material

Mature drupes of Amborella were collected from 10 individual trees located at "plateau de Dogny-Sarraméa" (New Caledonia; 21◦ 37′ 0 ′′ N, 165◦ 52′ 59′′ E). The fleshy part of the fruits was

**Abbreviations:** 1D, one dimensional; ACN, acetonitrile; AGP, Amborella Genome Project; CDS, coding sequence; CHAPS, 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate; DTT, dithiothreitol; ER, endoplasmic reticulum; HPLC, high-performance liquid chromatography; i.d., internal diameter; LC-MS/MS, liquid chromatography coupled to tandem mass spectrometry; MAF, minor allele frequency; ML, maximum likelihood; MRCA, most recent common ancestor; Mya, millions years ago; PCD, programmed cell death; PSV, protein storage vacuole; SDS-PAGE, sodium dodecylsulfate polyacrylamide gel electrophoresis; SNP, single nucleotide polymorphism; SSR, simple sequence repeats; TAIR, The Arabidopsis Information Resource; TFA, trifluoroacetic acid; VPE, vacuolar processing enzyme.

removed and pits (containing the seeds) were briefly dried on paper before being removed for seed isolation stricto sensu. Surface-sterilized seeds were cut longitudinally in two with a razor blade. A drop of sterile Milli-Q water was placed on the endospermic face of each half. Embryos were quickly extracted (in <1 min), with clean extra-thin needles and immediately frozen in liquid nitrogen.

### Preparation of Soluble Protein Extracts

For the preparation of soluble protein extracts, 100 isolated Amborella embryos were ground in liquid nitrogen using a mortar and pestle. Total soluble proteins were extracted at room temperature in 400µl thiourea/urea lysis buffer composed of 7 M urea, 2 M thiourea, 6 mM Tris-HCl, 4.2 mM Trizma <sup>R</sup> base (Sigma-Aldrich, Lyon, France), 4% (w/v) 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS, Sigma-Aldrich) supplemented with 50µl of the protease inhibitor cocktail Complete Mini (Roche Diagnostics France, Meylan, France). Then, 15µl of 1 M dithiothreitol (DTT, Sigma-Aldrich), 2µl of DNase I (Roche Diagnostics), and 5µl of RNase A (Sigma-Aldrich) were added to the sample. Following stirring for 2 h at 4◦C, protein extracts were centrifuged at 20,000 g at 4◦C for 15 min. The resulting supernatant was submitted to a second clarifying centrifugation, as above (Rajjou et al., 2011). The final supernatant was kept and protein concentrations in the various extracts were measured according to Bradford (1976) using Bovine Serum Albumin as a standard.

### Shotgun Proteomic Analysis

The Amborella seed proteome exploration was performed by LC-MS/MS analysis following preparation of soluble protein extracts (30µg protein; n = 3 biological replicates) that had been subjected to 1D-SDS-PAGE (http://pappso.inra.fr). Protein extracts were loaded in 1X Laemmli buffer (Laemmli, 1970) with DTT (50 mM) in a stacking gel [acrylamide 8% (w/v); Tris-HCl 0.56 M, pH 8.8; SDS 0.1% (w/v)]. After 15 min of migration at 10 mA, the gel was stained with colloidal blue (GelCode Blue Stain Reagent; Thermo Fisher Scientific Inc, Rockford, IL) and destained in Milli-Q water. The whole band corresponding to total proteins was excised and submitted to in-gel digestion with the Progest system (Genomic Solution, Huntingdon, UK) according to a standard trypsin protocol. Briefly, gel pieces were washed for 1 h at 37◦C in a solution containing 25% (v/v) acetonitrile (ACN) and 50 mM ammonium bicarbonate (pH 7.8), followed by dehydration in 100% ACN for 15 min. Gel pieces were rehydrated overnight at 37◦C with 1/50 (w/w) trypsin (Promega, Madison, WI, USA) in 20 mM ammonium bicarbonate, pH 7.8. Digestion was stopped by adding 0.4% (v/v) of trifluoroacetic acid (TFA).

HPLC was performed on a NanoLC-Ultra system (Eksigent). A 4-µl sample was loaded at 7.5µl/min−<sup>1</sup> on a precolumn cartridge (stationary phase: BIOSPHERE C18, 5µm; column: 100µm i.d., 2 cm; NanoSeparations) and desalted with 0.1% methanoic acid (HCOOH). After 3 min, the precolumn cartridge was connected to the separating PepMap C18 column (stationary phase: BIOSPHERE C18, 3µm; column: 75µm i.d., 150 mm; NanoSeparations). Buffers used were 0.1% HCOOH in water (A) and 0.1% HCOOH in ACN (B). Peptide separation was achieved with a linear gradient from 5 to 30% B for 30 min at 300 nl/min−<sup>1</sup> . Including the regeneration step at 95% B and the equilibration step at 95% A, one run took 45 min. Eluted peptides were analyzed on-line with a Q-Exactive mass spectrometer (Thermo Electron) using a nano-electrospray interface (noncoated capillary probe, 10 µ i.d.; New Objective). Xcalibur 2.1 interface was used to monitor data-dependent acquisition of peptide ions (http://www.thermoscientific.com/content/tfs/ en/product/xcalibur-software.html). This acquisition included a full MS scan covering 300 to 1400 range of mass-to-charge ratio (m/z) with a resolution of 70,000 and a MS/MS step (normalized collision energy: 30%; resolution: 17,500). MS/MS step was reiterated for the eight major ions detected during full MS scan. Dynamic exclusion was set to 45 s.

A database search was performed with XTandem (version 2011.12.01.1) (Bjornson et al., 2008; http://www.thegpm.org/ TANDEM/) for protein identification. Enzymatic cleavage was declared as a trypsin digestion with one possible miscleavage. Cys carbamidomethylation and Met oxidation were declared, respectively as fixed and variable modifications. Precursor mass and fragment mass tolerance were 10 ppm and 0.02 Th, respectively. The Amborella Genome database (http://www. amborella.org/) and a contaminant database (trypsin, keratins) were used. Identified proteins were analyzed using XTandem Pipeline (http://pappso.inra.fr/bioinfo/xtandempipeline/, version 3.3). Only peptides with an E-value smaller than 0.03 were validated.

### Phylogenetic Analyses

VPE protein sequences were obtained from the available databases by BLAST searching (Altschul et al., 1997). Multiple alignments were performed using MUSCLE (Edgar, 2004) and well-aligned sites were chosen using G-Blocks (Castresana, 2000) with settings to minimize the stringency of selection. Maximum likelihood (ML) phylogenetic reconstructions incorporating 500 bootstrap replicates were performed in PhyML (Guindon et al., 2010) using the LG substitution model (Le and Gascuel, 2008).

### Gene Sequence Annotation

To identify genes potentially encoding VPEs in Amborella, Arabidopsis VPE sequences retrieved from the Arabidopsis Information Resource (TAIR) (Swarbreck, 2008; http://www. arabidopsis.org) were blasted against the Amborella EVM 27 Predicted Protein database (http://www.amborella.org/). The annotations of the scaffolds containing the predicted VPE genes (NCBI accession # NW\_006499912.1 and NW\_006497648) were then manually checked and eventually re-annotated using the Lasergene Genomics Suite (http://www.dnastar.com/t-productsdnastar-lasergene-genomics.aspx; DNASTAR Inc., USA). The newly annotated VPE sequences were submitted to NCBI under the accessions n◦ BK009356 and BK009357. The previous annotated cds # XM\_006853855.2 was not modified.

### Gene Sequence Polymorphism

Next-generation resequencing data from the Amborella genome (Amborella Genome Project, 2013) were used to characterize SNP polymorphisms corresponding to the VPE sequences characterized in the present work. The genomic information has been generated for 12 individuals (named according to their location: Tonine, Ponandou, Pwicate, Tchamba, Ba, Aoupinié, Boregaou, Amieu, Dogny, Mé Ori, Mé Fomechawa, and Nakada) covering a wide range of Amborella's extant geographical distribution and natural genetic diversity (Poncet et al., 2013).

All the SNPs identified were defined as informative with a minor allele frequency (MAF) >0.08 and a missingness rate <0.42 by SNP. Biallelic exonic SNPs were annotated as synonymous or non-synonymous according to their reference/alternate allele nucleotides and to their position in the updated annotated coding sequences.

For each of the 12 individuals, the average MAF values were first computed across all SNPs for each of the three genes. Then, for each of the four genetic clusters inferred by a previous microsatellite (simple sequence repeats, SSR) analysis (Poncet et al., 2013), namely North (Tonine, Ponandou, Pwicate, Tchamba), Center (Ba, Aoupinié, Boregaou, Amieu, Dogny), Me (Mé Ori, Mé Fomechawa), and Nak (Nakada), we calculated for each gene (i) the mean proportion of polymorphic SNPs and (ii) the mean proportion of private alleles. Computations were performed on intronic and exonic SNP datasets independently by discarding missing data.

To assess any impact of natural selective pressures on VPE sequences, we examined the SNP patterns in comparison with 10 neutrally-behaving SSRs. Polymorphisms under neutral evolution co-vary with divergence between populations regardless of the mutation rate (Hartl and Clark, 2007). Marks of selection on SNP markers would be detected as polymorphism deviating from the expected neutral background polymorphism (influenced by forces such as divergence as well as other demographic events).

# Results

## Shotgun Proteomic Analyses

A shotgun proteomic analysis revealed 415 proteins from the isolated Amborella embryos (Villegente et al., unpublished results). In particular, this analysis confirmed the presence of three 11S globulin forms in the Amborella embryos, of which two contained canonical N-G VPE cleavage sites, while the third contained a variant cleavage site (N-V-I) (data not shown). This shotgun proteomic analysis also revealed the presence of two specific peptides (GIIINHPQGEDVYAGVPK and HQADVCHAYQLLLK) (Supplemental Table S1) matching with the amino acid sequences of Amborella VPE proteins encoded by sequences on the scaffold AmTr\_v1.0\_scaffold 00002, labeled 27.model.AmTr\_v1.0\_scaffold00002.262 and evm\_27.model.AmTr\_v1.0\_scaffold00002.263 (**Figure 1**).

## The Amborella VPE Family

To identify other genes potentially encoding VPEs in Amborella, we blasted the Arabidopsis VPE sequences retrieved from TAIR (Swarbreck, 2008; http://www.arabidopsis.org) against the Amborella EVM 27 Predicted Protein database (http://www. amborella.org/). Two additional loci were thus revealed, namely evm\_27.model.AmTr\_v1.0\_scaffold00036.100 (designated below as AmTr\_36.100) and evm\_27.model.AmTr\_v1.0\_scaffold 00002.265 (AmTr\_2.265). To confirm the automatic annotation (http://www.amborella.org/ and http://amborella.uga.edu/) a manual annotation was performed on the retrieved scaffold sequences. The automatic annotation process of the scaffold AmTr\_v1.0\_scaffold 00002 was actually wrong leading to truncated genes (**Figure 1**). Two full-length genes duplicated in tandem on the same scaffold (AmTr\_v1.0\_scaffold00002) were in fact identified and designated as AmTr\_2.262-1 and AmTr\_2.262- 2; (**Figure 1**). They are each composed of nine predicted exons and eight predicted introns like AmTr\_36.100. This structure is shared by almost all VPE genes from the available sequence databases representing green algae, bryophytes, lycophytes, gymnosperms, monocots, and eudicots (data not shown). The two duplicated genes (AmTr\_2.262-1 and AmTr\_2.262-2) show 84.5% similarity at the nucleotide sequence level and 98.8% at the protein level (Supplemental Figure S1). The amino acid sequences encoded by all three predicted Amborella VPE genes exhibit the two conserved amino acid residues (H and C), which are also present in the active sites of all known active VPEs (Chen et al., 1998, 2004) (Supplemental Figure S1). BLAST searching of the Amborella RNASeq Trinity Assembly using the online Amborella Genome Database (http://www.amborella.org)

annotated, supernumerary and truncated genes.

genes, AmTr\_2.262-1 and AmTr\_2.262-2, referred to 262-1 and 262-2,

identified expressed sequences corresponding to AmTr\_2.262- 2 (gnl|Ambo\_Trinity|comp32807\_c0\_seq1 and >gnl| Ambo\_Trinity|comp23371\_c0\_seq1) and AmTr\_36.100 (gnl| Ambo\_Trinity|comp847\_c0\_seq1 and gnl|Ambo\_Trinity|comp 777\_c0\_seq1), though not to AmTr\_2.262-1.

We conclude that the Amborella VPE family is composed of three genes: AmTr\_2.262-1, Am\_Tr2.262-2, and AmTr\_36.100 (**Figure 1**; Supplemental Figure S1), of which the former two are very closely related. AmTr\_2.262-2 and AmTr\_36.100 are both transcribed, while the transcription of AmTr\_2.262-1 remains to be demonstrated. It should be noted among others that no seed tissues were used to obtain transcribed sequences in the available Amborella transcriptome databases (see http://ancangio.uga. edu/content/amborella-trichopoda), and so it is possible that AmTr\_2.262-1 might be expressed, but shows an entirely seedspecific expression profile. Both of the peptide signatures identified using the proteomics approach described in the present work (GIIINHPQGEDVYAGVPK and HQADVCHAYQLLLK) show 100% identity to fragments of the predicted genes AmTr\_2.262-1 and AmTr\_2.262-2 (of which the latter is certainly transcribed and the former could be transcribed in other organs and tissues that were not surveyed in the available data, including fruits and seeds). No peptide signatures from AmTr\_36.100, which is most presumably transcribed in non-seed tissues, were detected in seeds in the proteomics approach described in the present work.

Most of our current knowledge of the genes encoding VPEs and their biological functions comes from molecular and genetic studies of the model plant Arabidopsis. In this species, four VPE homologs have been described: α-VPE and γ-VPE, which are specific to vegetative organs, β-VPE, which is specific to seeds, and δ-VPE, which is involved in seed coat formation (Nakaune et al., 2005; Hatsugai et al., 2015) and in the processing and degradation of various proteins in senescent Arabidopsis tissues (Rojo et al., 2003).

To reveal potential biological functions of the Amborella VPEs, the amino acid sequences of the three Amborella VPEs were blasted at TAIR against the Arabidopsis genome. An analysis of the scores obtained from this comparison disclosed that the AmTr\_36.100 gene would encode a γ-type VPE (Supplemental Figure S2), while the AmTr\_2.262-1 and AmTr\_2.262-2 genes would encode β-type VPEs (Supplemental Figure S2).

### Phylogenetic Analyses

To gain further insight in plant VPEs and benefiting from the present Amborella sequences, the amino acid sequences of VPEs identified by BLAST from a wide range of embryophytes (land plants) taxa were used for phylogenetic reconstruction. These sequences were first blasted at TAIR against the Arabidopsis genome. For all VPEs presently considered, the best hits corresponded to Arabidopsis VPEs, testifying of high amino acid sequence conservation among plant VPEs (data not shown).

The phylogenetic reconstruction of embryophyte VPE proteins (**Figure 2**; Supplemental Figure S3) shows two sister clades of VPEs from angiosperms, of which one clade contains the α-, γ-, and δ-VPE proteins from Arabidopsis ("Angiosperm α/γ/δ-VPE clade"), while the other contains the β-VPE protein from Arabidopsis ("Angiosperm β-VPE clade"). The presence of Amborella VPEs within each of these two clades is very well supported (100% bootstrap support in each case). Amborella proteins occupy basal positions in both the angiosperm α/γ/δand β-VPE clades, albeit with modest bootstrap support. Gymnosperm VPEs and non-seed plant VPEs group in two further clades, externally to the combined clade of angiosperm α/γ/ δ- and β-VPEs.

The above topology clearly indicates that the duplication event that generated the respective ancestors of the angiosperm α/γ/δand β-VPE clades occurred before the MRCA of the extant angiosperms. The absence of clearly distinguished α/γ/δ- and β-VPEs in gymnosperms, and the position of the gymnosperm VPE clade as sister to a clade containing all angiosperm VPEs (albeit with modest bootstrap support), suggests that the duplication that separated the angiosperm α/γ/δ- and β-VPE lineages occurred along the angiosperm stem lineage, after its separation from that of the living gymnosperms.

Arabidopsis δ-VPE occurs in a well-supported sub-clade of the angiosperm α/γ/δ-VPE clade, together with genes from widely diverged core eudicots including Medicago, Vitis, and Populus. By contrast, all α/γ/δ-clade VPEs from Amborella, monocots and basal eudicots (such as Papaver) group externally to the point of divergence of the δ-VPE sub-clade (**Figure 2**; Supplemental Figure S3). It thus appears that the δ-VPE lineage arose in a gene duplication event in a common ancestor of all, or a major part of, the living core eudicots. Branch lengths within the δ-VPE subclade are very long compared to those in the remainder of the angiosperm α/γ/δ-VPE clade, suggesting either strong positive or relaxed selection pressure to have operated on the δ-VPE lineage since its separation from that of the remaining α/γ/δ-VPE lineage (Guindon et al., 2004).

The Arabidopsis α- and γ-VPEs (Ath\_gi15225226 and Ath\_gi15233996) group with 100% bootstrap support in a small clade containing sequences only from closely related Brassicaceae, including Arabidopsis lyrata and Eutrema salsugineum (**Figure 2**; Supplemental Figure S3). This topology strongly suggests that the duplication that generated the α- and γ-VPE lineages occurred recently, within Brassicaceae. This conclusion accords well with the common expression pattern of Arabidopsis α- and γ-VPEs in vegetative organs.

The phylogeny in **Figure 2** and Supplemental Figure S3 shows that multiple, closely related VPE proteins are present in numerous taxa, including monocots, Amborella, eudicots, and gymnosperms. Thus, relatively recent gene duplications, such as that which generated the α- and γ-VPE lineages in Arabidopsis, appear to have occurred frequently within the VPE family in diverse plant groups.

### Gene Sequence Polymorphism and Amborella Population Analyses

Nucleotide sequence polymorphism was characterized at both the intron and exon levels for the Amborella VPE genes described in the present work, from 12 individuals that were considered to be representative of Amborella's extant geographical distribution and genetic diversity (Poncet et al., 2013) (**Figure 3**; Supplemental Figure S4).

### Impact of Exonic SNPs on Protein Structure

SNPs recorded across the studied individuals revealed a similar number of intronic and exonic polymorphisms for the three genes, with one to eight exonic SNPs and 22 to 34 intronic SNPs per gene. Among the 12 SNPs located in the proteincoding regions of the three genes, seven SNPs were synonymous and five were non-synonymous, though none was found located in the codons encoding catalytic residues (Supplemental Figure S4).

### Minor Allele Frequency (MAF) and Genetic Diversity within Amborella Distribution

We examined the MAF statistics of Amborella VPE genes within the previously described genetic groups North (Tonine, Ponandou, Pwicate, Tchamba), Center (Ba, Aoupinié, Boregaou, Amieu, Dogny), Me (Mé Ori, Mé Fomechawa), and Nak (Nakada). For each of the individuals and genetic groups studied, average MAFs were computed for each Amborella VPE gene (**Figures 3**, **4**; Supplemental Figure S4). For exonic SNPs, higher frequencies were observed for the northern (Ponandou, Pwicate, Tonine, and Tchamba) individuals for genes AmTr\_36.100 and AmTr\_2.262-2, and for Nakada individual for genes AmTr\_2.262-1 and AmTr\_2.262-2 (**Figure 3**). This trend was comparable to the diversity distribution observed with SSR microsatellites (Poncet et al., 2013) with individuals from the north exhibiting higher diversity and clustering in a same group. A similar distribution was also observed for intronic SNPs, with the northern group displaying a high percentage of polymorphic SNPs (71, 68, and 95% for genes AmTr\_36.100, AmTr\_2.262- 1, and AmTr\_2.262-2, respectively), the highest percentage of private SNPs (18, 6, and 18% for genes AmTr\_36.100, AmTr\_2.262-1, and AmTr\_2.262-2, respectively) and a mean MAF of 0.43 across all three genes, the highest among all groups (Supplemental Figure S4). In particular, changes in the mean proportion of exonic and intronic polymorphic SNPs across the four genetic groups follow a parallel progression with the mean SSR allelic richness (**Figure 4**). Levels of naturally occurring genomic variations in the VPE sequences thus appeared to co-vary with divergence and demographic history between populations and in particular with neutrally behaving polymorphisms (SSRs).

# Discussion

### The Amborella VPE Family

The present work shows that three VPE genes are present in the sister to all other living angiosperms, Amborella trichopoda. Two of these, AmTr\_2.262-1 and AmTr\_2.262-2, encode closely related β-VPEs, which are orthologs of the single β-VPE found in Arabidopsis. Our proteomics work shows that at least one

of these two genes is expressed in seeds, thus generating the peptide fragments GIIINHPQGEDVYAGVPK and HQADVCHAYQLLLK, which we detected. Transcriptomics data available through both the Amborella Genome Database (http:// www.amborella.org) and Ancestral Angiosperm Genome Project website (http://ancangio.uga.edu) indicate that AmTr\_2.262-2 is expressed in non-seed tissues, though this gene may additionally be expressed in seeds, as no seed tissues were used to obtain these transcriptomics data. We conclude that AmTr\_2.262-1 is either not expressed, or is specifically expressed in seeds. The third VPE gene present in Amborella, AmTr\_36.100, appears from phylogenetic studies to be a pro-ortholog of the α-, γ-, and δ-VPEs in Arabidopsis. From transcriptomics analyses, AmTr\_36.100 is transcribed in non-seed tissues, while the proteomics analysis presented here failed to show any peptide signatures derived from this gene in a seed-protein extract. We therefore conclude that AmTr\_36.100 is the Amborella pro-ortholog of the α-γ- and δ-VPEs from Arabidopsis and shows an entirely non-seed expression profile. The analysis of the Amborella VPE family performed in the present work has allowed us to draw solid conclusions (see following section) on the state of the VPE family in the MRCA of the extant angiosperms which no previous study has been able to make.

## A Partial Reconstruction of the Evolution of the VPE Family in Angiosperms

The Arabidopsis VPE family, whose expression patterns and functions are the best characterized of any plant species, consists of four genes encoding α-, β-, γ-, and δ-VPEs. Previous phylogenetic studies have succeeded in identifying several major clades of plant VPEs (Nakaune et al., 2005; Christoff et al., 2014). However, these studies did not clearly elucidate the deep evolutionary relationships between the major clades of VPEs identified, or the origins through gene duplication of the VPE lineages present in Arabidopsis or other established plant models.

The phylogenetic reconstructions presented here (**Figure 2**; Supplemental Figure S3), incorporating novel sequences from Amborella and a wide range of land plants, indicate the four VPE genes in Arabidopsis to have been generated through three duplication events that occurred at quite distinct evolutionary stages. The conclusions of this analysis are summarized in **Figure 5**. In non-seed plants, including bryophytes such as

Physcomitrella and non-seed, vascular plants such as Selaginella, several VPEs are found, though these group together in a clade that is positioned externally to all seed plant VPEs. Thus, the α-, β-, γ-, and δ-VPEs of Arabidopsis share no direct one-toone relationships of orthology with VPEs in non-seed plants. Similarly, multiple gymnosperm VPEs occur together in a clade that groups, though with modest bootstrap support, in a sister position to a clade containing all angiosperm VPEs. Accordingly, the α-, β-, γ-, and δ-VPEs of Arabidopsis appear to share no direct one-to-one relationships of orthology with VPEs from gymnosperms.

By contrast, in Amborella, the most basally diverging angiosperm, two VPE lineages are present with distinct relationships of orthology to the VPEs from Arabidopsis. Accordingly, proteins encoded by AmTr\_2.262-1 and AmTr\_2.262-2 are both orthologous to Arabidopsis β–VPE and, from our proteomics data, appear to have conserved a similar expression profile in seeds to that of Arabidopsis β–VPE. This result implies that angiosperm β–VPEs have conserved their seed-specific expression pattern since the MRCA of the living flowering plants, at least in the lineages leading to Amborella and to Arabidopsis. The protein encoded by AmTr\_36.100 is the putative Amborella pro-ortholog of all three other Arabidopsis VPEs (the α-, γ-, and δ-VPEs) and appears to have conserved a non-seed specific expression profile since the MRCA of living angiosperms.

The role of Arabidopsis β-VPE is to process the major protein reserves so these are correctly assembled within PSVs during seed maturation on the mother plant (Shimada et al., 2003). The need for such a role logically arises only in seed plants, and in this regard, the absence of a direct ortholog of β-VPEs in non-seed plants, such as bryophytes and lycophytes, is not incongruent. However, the duplication event that generated the β-VPE lineage appears to be specific to angiosperms, such that the remaining seed plants, the gymnosperms, also lack direct orthologs of the angiosperm β-VPE lineage. This apparent incongruity might be explained in at least two distinct ways, which relate to the functions and expression of VPE orthologs. Firstly, gymnosperms, including Pinus and Picea, appear like angiosperms to possess multiple VPE isoforms, and these are orthologous as a group to all angiosperm VPEs, including the β-VPEs. Therefore, a careful analysis of VPE expression patterns and/or protein accumulation in gymnosperms might show that some gymnosperm VPEs, like angiosperm β-VPEs, play a specific role in the mobilization of seed protein reserves. A second explanation for the differences between the VPE family in angiosperms and gymnosperms might relate to a possibly lesser functional specificity of VPEs in gymnosperms compared to angiosperms, such that the same VPEs in gymnosperms might be responsible both for the mobilization of seed protein reserves and for the processing of proteins in vegetative tissues.

The separate origins of the α-, γ-, and δ-VPE lineages, as these occur in Arabidopsis, have been considerably elucidated in our phylogenetic analyses (**Figure 2**; Supplemental Figure S3). Accordingly, the δ-VPE lineage appears to have arisen by a gene duplication in a common ancestor of widely diverged core eudicot taxa including Arabidopsis, Medicago, Populus, and Vitis. The δ-VPE lineage is thus not present as a separate lineage in basal eudicots, monocots or Amborella, but is represented in these taxa by single or multiple pro-orthologs of all core-eudicot α-, γ-, and δ-VPEs. Interestingly, the very long branches within the core eudicot δ-VPE clade indicate this gene lineage to have evolved very rapidly, which may have been due to intense positive selection pressure, or to a very relaxed selection, leading to rapid,

neutral evolution. Finally, the α- and γ-VPEs of Arabidopsis show direct one-to-one relationships of orthology only to VPEs from species of closely related Brassicaceae. These two genes therefore must have arisen through a gene duplication event within Brassicaceae.

The rapidity of evolution of the δ-VPE lineage compared to the remaining α/γ-lineage within the core eudicots suggests that the α- and γ-VPEs of Arabidopsis may have largely conserved the ancestral functions of the α/γ/δ lineage, while the δ-VPE lineage may have either acquired a novel function, or become specialized through sub-functionalization. Further experiments (e.g., RNASeq, RT-PCR) are needed to carefully analyze the expression of the three Amborella genes in seeds and vegetative tissues and different developmental stages. For example, it would be interesting to characterize the expression patterns of seed VPEs at the stage at which unprocessed pro-globulins forms accumulate in maturating seeds. Accordingly, it would be worth performing a thorough expression analysis of all three Amborella VPE genes in both vegetative and seed tissues. In particular, this would allow us to determine whether one of these genes is expressed in the seed coat, as is δ-VPE in Arabidopsis.

## Gene Sequence Polymorphism along VPE Genes in Natural Populations

The intra-specific genetic diversity of Amborella for the three VPE genes was explored at both the exon and intron levels using 12 individuals that are considered representative of Amborella's extant geographical distribution and genetic diversity in New Caledonia (Poncet et al., 2013). While SNPs in introns were most frequent among all SNPs located in the protein-coding regions of the three genes, most of them were synonymous or led to changes for amino acid with similar polarities, and never in the enzyme active sites, suggesting low impact on the protein function. Moreover, the polymorphism pattern across individuals was totally congruent with the genetic structure previously observed with neutrally behaving markers, SSRs (Poncet et al., 2012, 2013) or with SNPs distributed throughout the genome (Amborella Genome Project, 2013) on the same populations (**Figures 3**, **4**; Supplemental Figure S4).

If there were selective pressure acting on SNP polymorphisms, these three genes would not have co-varied with neutrallybehaving markers (SSRs and SNPs), and population divergence. Because no obvious signature of natural selection was detected either in exons or introns, the observed patterns of polymorphism were probably shaped by neutral demographic forces only, i.e., genetic drift and migration.

The present geographical distribution of Amborella in New Caledonia has been suggested to be a signature of the impact of ancient climatic changes, and notably of the severe restriction of favorable environments during the last glacial maximum (LGM) (ca. 22,000 years BP) and Holocene (ca. 12,000 years BP) when Amborella experienced a dramatic reduction (about 96%) in suitable area (Poncet et al., 2013). The survival and expansion of at least two lineages from putative refugia might have generated major diversity groups. The present data on VPE polymorphisms across the species provides an additional signature of the biogeographical history of Amborella.

# Acknowledgments

We are grateful to Claude dePamphilis and Joshua Der from Penn State University for helpful discussions. We would also like

# References


to thank the South Province of New Caledonia for supplying sampling permit for the Amborella seed proteomics analyses.

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00618

seed-type members of the vacuolar processing enzyme family of cysteine proteases. Plant Cell 14, 2863–2882. doi: 10.1105/tpc.005009


**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.

Copyright © 2015 Poncet, Scutt, Tournebize, Villegente, Cueff, Rajjou, Balliau, Zivy, Fogliani, Job, de Kochko, Sarramegna-Burtet and Job. 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.

# Proteomic analysis of crop plants under abiotic stress conditions: where to focus our research?

Fangping Gong, Xiuli Hu and Wei Wang\*

State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou, China

Keywords: crop stress proteomics, abiotic stress, subcellular proteome, stress proteins, initial proteome stress response, post-translational modifications

# Introduction

Approximately 80% of human food is composed of crops, which are dominated by cereals that collectively make up 50% of global food production (Langridge and Fleury, 2011). Among cereal crops, rice, wheat, and maize provide approximately half of the calories consumed worldwide. Nevertheless, crop production is seriously hampered by influential abiotic stresses like drought, climate fluctuations, and salinity. It is estimated that up to 50–70% decline in crop productivity is attributed to abiotic stress (Mittler, 2006). Therefore, to ensure the security of global food production, it is essential to produce sustainable crop varieties that can adapt to climate variability, and to develop a broad spectrum of abiotic stress tolerant crops (Tester and Langridge, 2010). This has driven much research into the study of crop responses to abiotic stresses.

Proteomics has been successfully used to study abiotic stress responses in a wide range of crops (Abreu et al., 2013; Barkla et al., 2013; Ngara and Ndimba, 2014), especially rice (Kim et al., 2014), wheat (Komatsu et al., 2014), and maize (Benesova et al., 2012; Gong et al., 2014). It is generally envisioned that at this stage, proteomic-based discoveries in rice are likely to be translated into improving other crop plants against ever-changing environmental factors (Kim et al., 2014).

Despite the potential role of proteomics to advance the study of stress tolerance in crops, thus far little useful information has been made available for crop improvement and breeding, even with the numerous proteomics studies undertaken in recent years. In our opinion, crop stress proteomics should be better focused on the following aspects: dissecting cell specific stress response (especially initial stress responses), identification of stress proteins, and the analysis of post translational modifications (PTMs) of proteins (**Figure 1**).

# Dissecting Cell or Tissue Specific Stress Response

Understanding how plant cells sense and respond to abiotic stress is not only fundamental to our understanding of stress tolerance, but has the potential to yield novel approaches to improve crop productivity. Cellular proteomics plays an essential role in determining the functions of cellular compartments and the mechanisms underlying protein/gene targeting and trafficking.

Currently, numerous organ-specific proteomic analyses of abiotic stress in crops have contributed to our understanding of the response mechanisms of crops to abiotic stresses (Komatsu and Hossain, 2013). Obviously, the specifics of proteomic response to abiotic stress vary from tissue to tissue within a plant. Therefore, the crop stress response should be analyzed at a cellular or subcellular level, integrated with studies on whole plants, organs or tissues, to discriminate the specific responses of different cell types to abiotic stress. At present, cell or subcellular proteomic

### Edited by:

Silvia Mazzuca, Università della Calabria, Italy

### Reviewed by:

Dipanjana Ghosh, National University of Singapore, Singapore Klára Kosová, Crop Research Institute, Czech Republic

\*Correspondence:

Wei Wang, wangwei@henau.edu.cn; wang\_w@aliyun.com

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 15 February 2015 Accepted: 24 May 2015 Published: 05 June 2015

### Citation:

Gong F, Hu X and Wang W (2015) Proteomic analysis of crop plants under abiotic stress conditions: where to focus our research? Front. Plant Sci. 6:418. doi: 10.3389/fpls.2015.00418

studies focus on relatively abundant, or easily isolated homogenous compartments (e.g., plastids, mitochondria, peroxisomes, and nuclei) mainly in Arabidopsis (Tanz et al., 2013), but also in rice, wheat, barley, maize (Huang et al., 2013; Millar and Taylor, 2014; Hu et al., 2015).

To increase the probability of identifying stress proteins (genes) from specific cells or tissues, an appropriate sampling method needs to be first developed to obtain relatively pure subcellular fractions from this material. A promising sampling method is laser capture microdissection (LCM), which can isolate specific cell types of interest from sectioned specimens of heterogeneous tissues under direct microscopic visualization with the assistance of a laser beam (Longuespée et al., 2014). LCM has been successfully used in transcriptome and microarray studies in maize (Nakazono et al., 2003; Rajhi et al., 2011) and rice (Suwabe et al., 2008; Kubo et al., 2013). Hopefully, combined with more sensitive protein staining technologies and more advanced mass spectrometers, LCM has the potential to promote crop stress proteomics at a cellular level. Another promising technique is free flow electrophoresis (FFE), which can isolate much purer membrane fractions and/or organelles. The FFE technique has been successfully applied in plant proteomics to the isolate tonoplast, mitochondria, plasma membranes, and Golgi apparatus (Barkla et al., 2013).

New approaches are beginning to enable the proteome to be analyzed at cell specific resolution. Not only are these studies leading to a better definition of the specialized proteomic behavior of certain cell types, but they also illustrate that information about proteome levels and responses gained from whole plants or organs can be misleading. This will surely help us to better understand the processes of crop stress acclimation and stress tolerance acquisition.

In addition, the studies on initial stress response of crops should be enhanced. Plant stress response consists of multiple phases, including an initial shock phase, an acclimation phase, a maintenance phase, an exhaustion phase, and/or a recovery phase (Kosová et al., 2011). Each phase of stress response is characterized by its unique transcriptome, proteome and metabolome changes. At 15 and 30 min after the onset of stresses (e.g., UV-B light, drought, and cold), the initial transcriptional changes in Arabidopsis are significant (Kilian et al., 2007). Currently, crop proteomic changes are often analyzed after several hours, even days after a stress onset (e.g., Meng et al., 2014). Thus, the initial proteome changes of crops under abiotic conditions should be further analyzed. These types of studies will give important insights in the signaling cascade activated immediately in crops in response to abiotic stresses. Sensitive proteomic approaches are capable of identifying low-abundance proteins (especially transcription factors and regulatory proteins) involved in the initial stress response in crops.

# Identification of Stress Proteins

The sequencing of major crops, especially rice, maize, and wheat represented a major breakthrough in crop proteomic research. For example, stress proteomic studies in maize have increased exponentially since 2009, after the release of the maize genome sequence (Schnable et al., 2009). Because knowledge of the genomic sequence alone does not indicate how a plant interacts with the environment, and not all open reading frames correspond to a functional gene (Ribeiro et al., 2013), proteomic approaches are critical to understand plant mechanisms of stress tolerance. A main aim of stress proteomics in crops is to identify stress proteins which can potentially be used for crop improvement and breeding.

The stress-tolerant phenotype in crops is a result of differential expression of unique proteins in resistant cultivars to protect them during stress periods. To develop better crop plants for sustainable food production, proteomic discovery of these unique stress proteins to further understand the stress-tolerance mechanisms at the molecular level is very important. We can potentially modify these key stress proteins to enhance a crops abiotic stress tolerance. A potential role for crop stress proteomic studies could be the identification and further characterization of key proteins underlying crop tolerance to a given abiotic stress, which can then be used as protein biomarker of a given stress. In such abiotic stress studies, it is common to analyze proteomes by contrasting stressed crop plants against control ones, attempting to correlate changes in protein accumulation with the phenotypic response (Abreu et al., 2013).

Rapid progress in proteome profiling methodologies, such as iTRAQ, DIGE, and high-resolution tandem mass spectrometry has enabled a more accurate comparison of crop stress responses and can detect more differentially abundant proteins than prior analyses. Unfortunately, ascribing a probable function to a newly identified stress responsive protein can be difficult in crops. Unlike Arabidopsis, it is difficult to determine the number of experimentally characterized genes in public databases for many crops. This is mainly due to the lack of high quality functional annotations for many crop genomes. For example, the maize genome sequence contains approximately 40,000 genes (Schnable et al., 2009), but little is known about the function of most genes. A search for maize protein sequences using the keyword "maize" retrieved 262,228 entries at NCBI and 85,389 entries in UniProt (14 May, 2015). This is indicative of the high level of redundancy and repetition, particularly in the NCBI database. In the UniProt database, only 840 maize protein entries have been reviewed (14 May, 2015), with most entries listed as "uncharacterized protein." Likewise, in rice, only 1% of the protein-coding genes have had a functional annotation based on experimental evidence (Rhee and Mutwil, 2014). Given this situation, the experimental validation of stress proteins and their role in stress tolerance is very important to bridge the gap between proteomic discovery of stress proteins and the selection of potential target proteins for future crop improvements.

More attention should be paid to up-regulated proteins in crop stress proteomics. In plants, translation efficiency can change dramatically in response to abiotic stress, leading to a massive bias in the pool of mRNAs that are actively translated (Mustroph et al., 2009; Juntawong et al., 2014). This might be related to the importance of stress- associated proteins that are required to recalibrate cellular metabolism. Thus, upregulated proteins are more important for crops to adapt to a stressful environment compared to down-regulated proteins; an important point when considering crop stress tolerance breeding. Of course, down-regulated proteins are also likely to contribute to an acquisition of enhanced plant stress tolerance. For example, some secondary metabolism related proteins affected by stress would likely decrease to conserve energy (Ghosh and Xu, 2014). In addition, due to discordant protein and mRNA expression, especially in plants (Vélez-Bermúdez and Schmidt, 2014), it is essential to identify up-regulated stress proteins rather than mRNA in order to better identify candidates which could be used for crop improvement.

Finally, it is worth noting that much of the stress proteomic research has been performed in laboratories under controlled conditions and relied on screening whole crops for their ability to survive severe stress. The effects of plant growth and gene expression in response to stress can be highly dose-responsive, indicating the existence of sensitive machinery in the plant for assessing the stress level and fine-tuning molecular responses (Claeys et al., 2014). Therefore, many proteomic analysis of abiotic stress in crops can be misleading and may not be useful.

# Analysis of PTMs of Proteins

PTMs can affect protein function, interactions with other proteins, subcellular targeting, and stability. In crop stress proteomics, the identification and quantification of PTMs will contribute the detailed functional characterization of a protein and will likely assist in our understanding of crop stress acclimation and stress tolerance acquisition.

Large-scale proteomics studies have revealed that PTMs are more widespread than previously estimated. For example, up to two-thirds of the metabolic proteins in yeast may be affected by protein phosphorylation (Breitkreutz et al., 2010). In Arabidopsis, PTMs include phosphorylation (Heazlewood et al., 2008; Vialaret et al., 2014), N-linked glycosylation (Zielinska et al., 2012), ubiquitination (Kim et al., 2013), methionine oxidation (Marondedze et al., 2013), S-nitrosylation (Fares et al., 2011), and acetylation (Finkemeier et al., 2011).

Few proteomic studies have been performed to specifically characterize PTMs in crops under abiotic stress; this includes the analysis of phosphorylation during salt and water stresses in maize (Zörb et al., 2010; Bonhomme et al., 2012; Hu et al., 2013) and characterization of protein glycosylation in soybean roots under flooding (Mustafa and Komatsu, 2014).

It is expected that in certain cases PTMs play a more important role than protein abundance changes. Thus, the analysis of PTMs of stress-responsive proteins in crops should be strengthened; however the ability to routinely identify and quantify PTMs represents a grand challenge in the field of proteomics (Heazlewood, 2011). In conjunction with improvements in methodological approaches, it would be expected that the study of PTMs will become more common in the area of crop stress proteomics in the future.

# Concluding Remarks and Outlook

Proteomics has an important role to play in assisting our understating at the molecular level of how crops respond to abiotic stress. In-depth proteomic analysis of crop stress responses will be essential for future crop improvements. Though proteomic characterization of cell and tissue specific stress responses, stress proteins and PTMs is still a difficult undertaking in crops, the development of more sensitive methodologies, particularly for the cell specific analysis of the proteome will be crucial for understanding stress responses at the cellular level. In addition, rapid advances in high-throughput omics technologies (e.g., proteomics, transcriptomics, genomics, and metabolomics) will make it possible to use a systems biology approach to understand crop responses to abiotic stresses.

# Acknowledgments

We thank Dr. Joshua Heazlewood for copy-editing our manuscript. We acknowledge the National Natural Science Foundation of China (Grant No. 31371543), the Plan for Scientific Innovation Talent of Henan Province (Grant No. 144200510012), and the Program for Innovative

References


Research Team (in Science and Technology) in University of Henan Province (Grant No. 15IRTSTHN015) for financial support.


**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.

Copyright © 2015 Gong, Hu and Wang. 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.

# Advances in plant proteomics toward improvement of crop productivity and stress resistance

*Junjie Hu1,2, Christof Rampitsch2 and Natalia V. Bykova2\**

*<sup>1</sup> Department of Biology, Memorial University of Newfoundland, St. John's, NL, Canada, <sup>2</sup> Cereal Proteomics, Cereal Research Centre, Agriculture and Agri-Food Canada, Morden, MB, Canada*

Abiotic and biotic stresses constrain plant growth and development negatively impacting crop production. Plants have developed stress-specific adaptations as well as simultaneous responses to a combination of various abiotic stresses with pathogen infection. The efficiency of stress-induced adaptive responses is dependent on activation of molecular signaling pathways and intracellular networks by modulating expression, or abundance, and/or post-translational modification (PTM) of proteins primarily associated with defense mechanisms. In this review, we summarize and evaluate the contribution of proteomic studies to our understanding of stress response mechanisms in different plant organs and tissues. Advanced quantitative proteomic techniques have improved the coverage of total proteomes and sub-proteomes from small amounts of starting material, and characterized PTMs as well as protein– protein interactions at the cellular level, providing detailed information on organ- and tissue-specific regulatory mechanisms responding to a variety of individual stresses or stress combinations during plant life cycle. In particular, we address the tissuespecific signaling networks localized to various organelles that participate in stressrelated physiological plasticity and adaptive mechanisms, such as photosynthetic efficiency, symbiotic nitrogen fixation, plant growth, tolerance and common responses to environmental stresses. We also provide an update on the progress of proteomics with major crop species and discuss the current challenges and limitations inherent to proteomics techniques and data interpretation for non-model organisms. Future directions in proteomics research toward crop improvement are further discussed.

Keywords: combinatorial stresses, quantitative techniques, crop productivity, tissue-specific proteomics, subcellular localization

# Introduction

Understanding all levels that regulate adaptive mechanisms and the resilience of crop plants in the context of climate changes is absolutely essential to reach significant achievements in genomics-driven breeding of major crops for high productivity and stress tolerance. A new pattern of frequently occurring extreme weather events has already been taking a toll on agricultural production systems. In addition to increasing amount of genomic information available for both model and non-model plants, the parallel development of bioinformatics techniques and analytical instrumentation makes proteomics an essential approach to reveal major signaling and biochemical pathways underlying plant life cycle, interaction with the

### *Edited by:*

*Sabine Lüthje, University of Hamburg, Germany*

### *Reviewed by:*

*Christian Lindermayr, Helmholtz Zentrum München, Germany Jesus V. Jorrin Novo, University of Cordoba, Spain*

### *\*Correspondence:*

*Natalia V. Bykova, Cereal Proteomics, Cereal Research Centre, Agriculture and Agri-Food Canada, 101 Route 100, Morden, MB R6M 1Y5, Canada natalia.bykova@agr.gc.ca*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

> *Received: 31 December 2014 Accepted: 16 March 2015 Published: 14 April 2015*

### *Citation:*

*Hu J, Rampitsch C and Bykova NV (2015) Advances in plant proteomics toward improvement of crop productivity and stress resistance. Front. Plant Sci. 6:209. doi: 10.3389/fpls.2015.00209* environment, and responses to abiotic and biotic stresses. Highthroughput proteomic studies have gone beyond simple identification of individual proteins to quantitative profiling, analysis of dynamic post-translational modifications (PTMs), subcellular localization and compartmentalization, protein complexes, signaling pathways, and protein–protein interactions (for reviews refer to Agrawal et al., 2011, 2013; Matros et al., 2011; Pflieger et al., 2011; Hossain et al., 2012).

Growing in the field or cultivated in the laboratory, plant development and productivity are inevitably controlled by various extreme environmental factors such as drought, heat, salinity, cold, or pathogen infection, which may delay or induce seed germination, reduce seedling growth, and decrease crop yields. Proteomics studies can substantially contribute to revealing virtually every aspect of cellular function in plant stress responses, unraveling possible relationships between protein abundance and/or modification and plant stress tolerance. An increasing number of studies have been discussing the contribution of proteomics to deeper insights into the molecular mechanisms of plant responses to stresses and signaling pathways linking changes in protein expression to cellular metabolic events, such as studies using model plants *Arabidopsis* (Wienkoop et al., 2010), rice (see for reviews Singh and Jwa, 2013; Kim et al., 2014) and sorghum (Ngara and Ndimba, 2014). Attributed to the improvement in diverse proteomic technology platforms that combined classical two-dimensional electrophoresis (2-DE) gel-based techniques with mass spectrometry (MS)-based quantitative approaches as well as the accessibility of protein databases of various plant species, major monocotyledonous cereals and dicotyledonous legumes (e.g., maize, wheat, barley, soybeans etc.) have been widely used to study quantitative changes in protein abundance related to different abiotic stresses (Li et al., 2013a; Jacoby et al., 2013a; Deshmukh et al., 2014; Wu et al., 2014a; Kamal et al., 2015).

In the agricultural environment crop plants are subject to a complex set of abiotic and biotic stresses. In addition to studying effects of various stresses applied individually under laboratory controlled conditions, recent evidence shows that simultaneous occurrence of multiple stresses affecting crop growth, yield and physiological traits can cause plants to activate intricate metabolic pathways involved in specific programming of gene expression that uniquely respond to different combinations of stresses (Atkinson and Urwin, 2012). Several different signaling pathways involved in multiple stress-responding mechanisms have been revealed in transcriptome, metabolome, and proteome analysis of various crop plants subjected to different stress combinations, suggesting a complex regulatory network orchestrated by hormone signals, transcription factors, antioxidants, kinase cascades, reactive oxygen species (ROS), and osmolyte synthesis (Suzuki et al., 2014).

Fundamentally, crop growth depends on efficient production of energy and nutritional compounds regulated through different organs, which are equipped with various organelles and organ-specific sets of cytosolic proteins, hormones and metabolites (Komatsu and Hossain, 2013). The responses of plant cells to abiotic stresses vary in different organs. Organ-specific proteomics combined with subcellular organelle proteomic studies of developmental mechanisms from leaf to root can provide more detailed information for understanding of cellular mechanisms that regulate stress response and signal transduction in various organelles (Hossain et al., 2012; Komatsu and Hossain, 2013; **Table 1**). Tissue-targeted seed proteomic studies of different developmental stages under abiotic stresses have contributed to increasing our depth of knowledge about the processes controlling seed development, dormancy and germination by analyzing spatial and functional sub-proteomes (Finnie et al., 2011a). In this article we provide an update on the progress of proteomics with major crop species and discuss the current challenges and limitations inherent to proteomics techniques and data interpretation for non-model organisms.

# Approaches and Challenges in Crop Plant Proteomics

With the completion of genome sequences in model species such as dicotyledonous plant *Arabidopsis thaliana*, monocotyledonous crop plant rice (*Oryza sativa* ssp. *japonica* and ssp. *indica*), model grass *Brachypodium distachyon*, legume species soybean (*Glycine max*), *Medicago truncatula* and *Lotus japonicus,* cereal crops *Zea mays,* and *Hordeum vulgare*, other crop species to follow (NCBI list of sequenced plant genomes), and with recently released chromosome-based draft sequence of the hexaploid bread wheat (*Triticum aestivum*) genome (International Wheat Genome Sequencing Consortium [IWGSC], 2014), attention has been focused on linking genomic data and transcriptomic profiles to the spatial and temporal expression, biological function and functional network of proteins. Moreover, dealing with complex and dynamic plant proteomes, it is crucial to choose suitable proteomic approaches targeting the identification of proteins and their modification that may contribute to crop improvement. In the recent years quantitative proteomic studies using high resolution and mass accuracy instruments have been contributing with important information to our understanding of plant growth, development, and interactions with the environment. This capability is especially useful for crops as it may not only contribute to increasing nutritional value and yield, but also to understanding mechanisms of crop adaptation to responses to abiotic stresses.

# Recent Innovative Technologies used in Quantitative Plant Proteomics

Driven by innovations in MS-based technologies and rapid development of quantitative methods, proteomics has emerged as a complementary technique to other approaches such as transcriptomics and metabolomics in the post-genomic era (Wienkoop et al., 2010). Proteomic analysis is achieved through (1) separation and identification of proteins based on 2-DE or coupled gel-free shotgun liquid chromatography tandem mass spectrometry (LC–MS/MS) platforms; (2) elucidation of protein functions and protein functional networks in plant metabolic and signaling pathways through the analysis of protein mapping, characterization of PTMs and protein–protein interactions; (3) bioinformatic strategies and the use of databases for both model



*For more comprehensive discussion on previously published organellar and sub-organellar proteomic studies, including model species, refer to Agrawal et al. (2011).*

and non-model plant species (Holman et al., 2013). Recently, the application of gel-free protein separation approaches and 'second generation' proteomic techniques such as multidimensional protein identification technology (MudPIT), quantitative proteomic approaches including isotope-coded affinity tags (ICATs), targeted mass tags (TMTs), and isobaric tags for relative and absolute quantitation (iTRAQ) have been widely used in descriptive and comparative proteomic studies of plant development and metabolic strategies in abiotic stress adaptation. Advances in liquid chromatography-based separation and label-free quantitative proteomic analysis of large number of proteins derived from complex plant samples have recently been discussed (Matros et al., 2011). Although modern gel-free quantitative proteomic approaches, label-based and label-free, are considered to be more advanced and can provide more information on comparative changes in protein expression than one and 2-DE gel-based methods, they have limitations. One obvious limitation in terms of global proteome coverage is the fact that they are designed for less hydrophobic, more aqueous buffer-soluble sub-proteomes, whereas the buffers and detergents used in gelbased protein separation techniques can be quite powerful and efficient in solubilization of more hydrophobic protein groups, especially in conjunction with organellar and sub-organellar pre-fractionation. The development and application of label-free quantitative analysis techniques in combination with one and 2-DE gels was the new realization that helped to solve one of the main drawbacks of gel-based separation and quantitative analysis of proteins related to co-focusing/co-localization of several proteins and their modified forms in one spot or band on the gel.

One of the most socio-economically important agricultural crops is wheat. However, until recently, large-scale proteomic studies with this organism were difficult to realize due to the lack of genomic information available to facilitate protein identifications. The publication of the first draft of completely sequenced wheat genome (International Wheat Genome Sequencing Consortium [IWGSC], 2014) is inspirational to plant proteomics researchers, although the annotation of complete wheat genome assembly will remain a challenging task. Until recently proteomic studies have been using alternative available bioinformatics resources (**Table 2**) that included wheat ESTbased databases (Bykova et al., 2011a,b; Pascovici et al., 2013), or available closely related *Brachypodium distachyon* model plant genome, or D-genome progenitor *Aegilops tauschii*, as well as a composite database of available cereals sorghum, maize and rice (Pascovici et al., 2013), a translated database of the low copy number genome assemblies of *T. aestivum* (Alvarez et al.,

### TABLE 2 | Recent large-scale gel-free quantitative proteomic studies on wheat.


*NSAF, normalized spectral abundance factor; FDR, false discovery rate.*

2014), and proteins from monocot family *Poaceae* (Kang et al., 2015). Pascovici et al. (2013) have evaluated the most effective pipeline for large-scale shotgun quantitative experiments using bread wheat (*T. aestivum*), iTRAQ multirun quantitative approach, and the available resources for bioinformatics data analysis and downstream functional interpretation. The study emphasized many challenges related to the repetitiveness/redundancy/polymorphism of bread wheat genome and therefore extremely large size of the corresponding EST-based database that could not be readily manipulated by the available bioinformatics tools, and the stochastic aspect of protein grouping across multiple runs. The use of smaller databases was demonstrated as alternative pragmatic approaches to reliably identify proteins and proceed with functional annotations (**Table 2**).

More recently, targeted MS-based quantitative approaches such as multiplexed selective reaction monitoring (SRM) have proven to be powerful for identification of specific proteins with causative functions in agronomically important traits (Jacoby et al., 2013b). Attributable to outstanding sensitivity of this methodology to selective quantitation of low abundance protein components in complex mixtures, SRM technique is seen by researchers as an alternative to antibody-based immunodetection assays (Picotti et al., 2013). This SRM approach is based on highly specific detection and quantitation of proteotypic couples comprised of target precursor and corresponding fragment ions and until recently it was exclusively based on triple quadrupole MS platforms due to the necessity of two stages of mass filters. The advantages and limitations of this application have been discussed and demonstrated experimentally elsewhere. The new generation of Orbitrap-technology instruments such as Q Exactive hybrid quadrupole-Orbitrap provided an efficient and user-friendly alternative for further application of SRM method in quantitative assay development with high potential for large-scale targeted proteomics experiments (Wu et al., 2014b). The purpose of SRM methodology in plant proteomics is biomarker validation in crops, which follows the discovery phase with more explorative qualitative and quantitative comparative proteomic studies aimed at finding potential candidates important in stress responses. This highly selective and sensitive quantitative approach can be powerful not only for biomarker validation but also for the development of new stress tolerance assessment methods, which will facilitate the identification of genotypes with improved resistance and ultimately discovery of gene targets for marker-assisted breeding.

Another approach has recently been developed for label-free shotgun proteomics based on data independent (MSE) acquisition protocols that has a potential to identify peptides from complex samples in a rapid, consistent, and sensitive way, and offers a higher dynamic range for peptide quantification (Buts et al., 2014). In this approach, ultra performance liquid chromatography is used, which is coupled to an LC–MS/MS run where an alternating energy level allows to obtain accurate precursor masses at low energy, and to take fragmentation spectra of all parent masses at high collision energy in one analytical run. However, this approach is rather at the early stages of development, and has many challenges related to the difficulties in interpretation of very complex composite fragmentation spectra resulting in poor protein identification rate. At present, parallel data-dependent runs are needed in order to acquire all necessary information, build the databases of individual peptide fragmentation spectra, and link them to the MS<sup>E</sup> (Buts et al., 2014). This approach was successfully used in quantitative analysis of important allergenic proteins in wheat grain extracts, in identification of gliadins and glutenins in wheat grain and quantitation of proteins associated with celiac disease and baker's asthma (Uvackova et al., 2013a,b).

An alternative strategy that combines high specificity dataindependent acquisition method with a novel targeted data extraction approach to mine the resulting fragment ion data sets was recently demonstrated (Gillet et al., 2012). This method, termed SWATH MS, is based on sequential time-and masssegmented acquisition, which generates fragment ion spectra of all precursors in two user-defined dimensions, retention time and *m/z* space, resulting in complex fragment ion maps. The interpretation of highly specific multiplexed data sets required the development of fundamentally different data analysis strategy, which uses previously acquired information contained in spectral libraries to mine the fragment ion maps for targeted extraction and quantitation of specific peptides of interest. The accuracy and consistency of SWATH MS was demonstrated to be comparable to SRM approach (Gillet et al., 2012). One of the important advantages of the former, alleviating most constrains of present proteomics methods, is the iterative retrospective remining of the acquired data sets for targeted extraction. This approach offers unprecedented possibilities for the qualitative and quantitative profiling not only in proteomics but also in metabolomics and lipidomics. One of the main bottlenecks for proteomics development is the lack of robust bioinformatics tools with novel algorithmic solutions to processing of MS data, which are lagging behind the substantial advances occurring in instrumentation and protocols (Cappadona et al., 2012; Smith et al., 2014). It remains to be seen how this breakthrough technology will evolve into a powerful tool utilized throughout plant sciences.

## From Cellular Proteome to Subcellular Protein catalogs

The major advances in organelle-based proteomics have not only provided a deeper insight into protein localization and organellespecific function in plant biological processes, but also a better understanding of the functions of organelles in metabolic processes involved in plant development and growth (for a recent review refer to Agrawal et al., 2011). The accurate description of an organelle proteome requires the ability to purify organelles from cellular mixture and identify low abundance proteome extracted from these organelles. Several quantitative proteomic studies have focused on spatial subcellular proteomes with particular methods designed for different organelles (references in **Table 1**).

Proteomic analysis of wheat chloroplasts using a combination of two complementary approaches Tricine SDS–PAGE and 2-DE coupled to a high throughput MS methods (LTQ-FTICR and MALDI-TOF/TOF) has contributed to a better understanding of the responsive proteins in photosynthesis during abiotic stress in plastids (Kamal et al., 2012, 2013). Chloroplast proteomics study of soybean leaves under ozone stress by 2-DE and MALDI-TOF MS approach identified increased expression level of proteins involved in antioxidant defense and carbon metabolism (Ahsan et al., 2010a). Proteomic studies on mitochondrial organelles provided information on both individual proteins and protein complexes that participate in salinity response mechanisms in rice through 2-DE and MALDI-TOF MS (Chen et al., 2009), and in wheat through 2-DE and LC– MS/MS analysis (Jacoby et al., 2010, 2013a). Moreover, the effects of drought, cold and herbicide treatments on mitochondrial proteomics in pea were also analyzed by 2-DE Blue Native PAGE with Q-TOF MS (Taylor et al., 2005). Cell wall proteomic studies of major crops such as rice (using 2-DE and nanoLC–MS/MS), soybean (using 2-DE, MALDI-TOF MS, and nanoLC–MS/MS), and maize (using 2-DE and Q-TOF MS), provided insights into either dehydration- or water stress-responsive proteomes involved in a variety of functions, including carbohydrate metabolism, cellular defense through redox mechanisms, cell wall modification, and cell signaling pathways (Zhu et al., 2007; Komatsu et al., 2010). Plasma membrane, as a primary interface between the cellular cytoplasm and the extracellular environment, plays a vital role in signaling, communication and uptake of nutrients (Alexandersson et al., 2004). Due to their low abundance and low solubility, nanoLC–MS/MS-based proteomic approaches have been used to study osmotic stressinduced proteins in soybean, which are mostly involved in antioxidative defense system (Komatsu et al., 2009; Nouri and Komatsu, 2010). Changes in nuclear proteins under the flooding stress in soybean root tips, studied by gel-free nano-LC MS/MS, revealed differentially regulated responding proteins (Komatsu et al., 2014). Moreover, reversed-phase chromatography, SDS– PAGE and LC–MS/MS have been used to prepare and analyze integral plasma membrane-enriched tissue fractions from barley aleurone layers and germinated embryos, providing more information about aleurone layer as a secretory tissue during seed germination (Hynek et al., 2006, 2009). The accurate description of an organelle proteome requires the ability to identify genuine protein residents (**Table 1**). Although many challenges remain, quantitative proteomic profiling of organelles has been developed to reliably identify the protein complement of whole organelles, as well as for protein assignment to subcellular location and relative protein quantification, which are improving our understanding of protein functions and dynamics in plant cells.

# Organ-Specific Proteome Analysis of Abiotic Stress Responses in Crop Plants

### Proteomics of Leaf Photosynthesis and Senescence to Understand Crop Productivity

Leaf photosynthesis is the main source of plant biomass influencing potential crop yield. Highly abundant chlorophyll in leaves plays essential roles in light harvesting and energy transfer during photosynthesis, therefore, chlorophyll metabolism contributes to photosynthetic efficiency during leaf development. Recently, Chu et al. (2015) analyzed changes in protein profiles upon the development of chlorophyll deficiency in *Brassica napus* leaves and provided new insights into the regulation of chlorophyll biosynthesis and photosynthesis in crops (Chu et al., 2015). Moreover, the levels of chlorophylls were also shown to be associated with the maintenance of photosynthetic rate of CO2 consumption during the grain-filling period, and with the rate of leaf senescence in different rice cultivars (Panda and Sarkar, 2013). Leaf senescence is featured with loss of photosynthetic activities and hydrolysis of macromolecules followed by the degeneration of chloroplasts and remobilization of the hydrolyzed nutrients to young leaves and developing seeds (Guo, 2013). Several studies have focused on the proteomics of leaf senescence, mainly on the investigation of nitrogen mobilization from leaves during leaf senescence (Bazargani et al., 2011; Avice and Etienne, 2014). The results of these studies suggest the importance of proteolysis, chloroplast degradation and nitrogen remobilization during this process (**Figure 1**). Chloroplast contains up to 75% of leaf nitrogen in the form of Rubisco enzyme components in the stroma and complex of photosystem II in the thylakoid membrane (Roberts et al., 2012). Advances in organelle proteomic studies integrated with largescale genomic approaches and determination of enzymes with proteolytic activity have addressed the complexity of chloroplastic proteolytic machinery during leaf senescence and investigated different classes of senescence-associated proteases with unique physiological roles according to their expression profiles along the senescence progress (Liu et al., 2008; Roberts et al., 2012). Moreover, the regulation of photosynthetic carbon metabolism has also been investigated during leaf senescence in rice by a comparative proteomic approach, contributing to a deeper insight into the enzymatic regulation involved in the Calvin cycle during senescence featured with down-regulated photorespiration (Zhang et al., 2010). Furthermore, sucrose as the main photosynthetic product is rich with carbon and energy, is a key component in carbon metabolism, and is essential for both plant growth and the synthesis of storage reserves, such as starch and oil (Barsan et al., 2012). Glycolytic enzymes involved in sucrose synthesis are of particular interest with respect to crop yield, and have been identified by subcellular proteomic studies of senescence, photosynthesis, and stress-responding processes in rice leaves (Zhang et al., 2010).

# Xylem and Phloem Proteomics of Root-to-Leaf Signaling Pathways During Stress

Maximizing crop yields also depends on the leaves receiving an optimal supply of nutrients from the root system via the xylem vessels. The xylem sap is a dynamic fluid circulated between root and leaf, featured with a high content of secreted proteins undergoing changes in its proteome upon abiotic and biotic stresses (Ligat et al., 2011; de Bernonville et al., 2014; Zhang et al., 2014a). Xylem proteomic and secretomic studies have recently become one of the major areas of interest in understanding plant development and responses to environmental perturbations, and illustrated several types of xylem sap-containing proteins that participate in cell wall development and repair process (Kalluri et al., 2009; Ligat et al., 2011; Zhang et al., 2014a), leaf senescence (Wang et al., 2012), abiotic stress responses (Alvarez et al., 2008), biotic stress defense mechanisms (González et al., 2012), and intercellular and intracellular communication (Agrawal et al., 2010). Additional studies of protein and metabolite composition of xylem sap and apoplast in soybean (*Glycine max*) provide further investigation of expression profiles and signaling roles of corresponding proteomes, and ultimately reveal more root contributions to pathogenic and symbiotic microbe interactions, and root-to-shoot communication (Djordjevic et al., 2007; Subramanian et al., 2009; Krishnan et al., 2011).

The phloem tissue isolated from broccoli (*Brassica oleracea*) was found to be enriched in proteins associated with biotic and abiotic stress responses and structural proteins (Anstead et al., 2013). Other proteomic studies were predominantly focused on the analysis of phloem sap exudates from agriculturally important plants oilseed rape (Giavalisco et al., 2006), rice (Aki et al., 2008), pumpkin (Lin et al., 2009; Cho et al., 2010; Fröhlich et al., 2012), and melon (Malter and Wolf, 2011), identifying several hundred physiologically relevant proteins and ribonucleoprotein complexes. The phloem sap proteomes showed enhanced presence of proteins involved in redox regulation, defense and stress responses, calcium regulation, RNA metabolism and G-protein signaling (**Figure 1**). Some of the important insights into the operation of the sieve tube system were revealed through proteomics studies. The findings indicate likely occurrence of protein synthesis and turnover within the angiosperm phloem translocation stream, processes that were thought to be absent in enucleate sieve elements (Lin et al., 2009). There is also indication that phloem may exert some level of control over flowering time (Giavalisco et al., 2006), seed development (Lin et al., 2009), other developmental processes via gibberellin biosynthesis or modification (Cho et al., 2010).

### Root Proteomics of Symbiotic Systems to Improve Legume Productivity

In associations between plants and soil microorganisms, only symbiotic interactions have beneficial effects for the host


FIGURE 1 | Proteomic studies-driven insights into cellular activities during abiotic stress response of major crops. Summary of the abiotic stress studies published in: (1) Heat in soybean (Ahsan et al., 2010b). (2) Cold in winter wheat (Vítámvás et al., 2012; Kosová et al., 2013). (3) Flooding and anoxia in soybean seedlings (Nanjo et al., 2011 review; Komatsu et al., 2009, 2010; Yin et al., 2014). (4) Drought tolerance in

wheat roots (Alvarez et al., 2014) and leaves (Ford et al., 2011; Kang et al., 2012; Zhang et al., 2014b). (5) Salinity in wheat (Guo et al., 2012; Jacoby et al., 2013a; Capriotti et al., 2014; Fercha et al., 2014), soybean (Ma et al., 2012), rice (Ruan et al., 2011; Song et al., 2011; Liu et al., 2013b), and barley (Witzel et al., 2009; Gao et al., 2013). (6) Metal ions stress in wheat (Li et al., 2013b; Kang et al., 2015).

plants while pathogenic interactions lead to severe damages (Schenkluhn et al., 2010). Rhizobia-legume symbiosis is the best characterized beneficial plant–bacterial mutualistic interaction, which represents one of the most productive nitrogen-fixing systems and effectively renders legumes being independent of other nitrogen sources (Mathesius, 2009; Colditz and Braun, 2010). Legume plants maintain control of nodulation to balance the nitrogen gains with their energy needs and developmental costs through a systemic mechanism known as the autoregulation of nodulation that involves peptide hormones, receptor kinases, and small metabolites for long-distance signaling control (Reid et al., 2011, 2012). Legumes are among the most economically important crops due to their high protein content. Model legume plant *M. truncatula* is closely related to major legume crops such as soybean (*Glycine max*) and pea (*Pisum sativum*), therefore its investigation is of high relevance for agriculture. Recently, comparative and quantitative proteomic studies provided in-depth characterizations of *M. truncatula* proteome focused on symbiosis- or pathogenesisinduced changes of root system (Schenkluhn et al., 2010; Molesini et al., 2013). Moreover, the establishment and maintenance of rhizobium–legume symbiosis require reciprocal recognition with exchanges of signal molecules and complex developmental programs between the organisms, which leads to the formation of nodules on the legume root and the differentiation of rhizobial cells into bacteroids (del Giudice et al., 2011). A number of proteomic studies have attempted to investigate mutual impacts between symbiotic or pathogenic bacteria and the root of host plant in the rhizosphere under a multitude of biotic and abiotic stresses from the soil (for a recent review refer to Knief et al., 2011). Especially under drought stress, symbiotic nitrogen fixation is one of the physiological processes to first show stress responses in nodulated legumes (**Figure 1**). Differential plant and bacteroid responses to drought stress have been revealed by proteomic analysis of root nodule in *M. truncatula* (Larrainzar et al., 2007). Large-scale phosphoproteome analyses focused on nitrogen-fixing root nodules in *M. truncatula* (Wienkoop et al., 2008; Grimsrud et al., 2010) and root hairs in soybean (Nguyen et al., 2012) investigated phosphorylationmediated signal transduction cascades. Symbiotic signals produced by the rhizobia during the initiation of symbiosis and the development of nodules were revealed, suggesting a complex network of kinase-substrate and phosphatase-substrate interactions in response to rhizobial infections. Furthermore, improved knowledge of root proteases associated with rhizosphere and drought tolerance reinforces the importance of their role in endocytosis of proteins, peptides and microbes, or root exudatemediated nitrogen uptake mechanisms, which can contribute to a systematic study of root proteases in crop improvement (Adamczyk et al., 2010; Kohli et al., 2012). Additionally, numerous studies have shown that root sensing of stresses from soil drying and salinization can regulate chemical root-to-leaf signaling through altering the complexity of constituents and their interactions in xylem sap, which ultimately reduce leaf transpiration and growth, and therefore, influence crop productivity (Bazargani et al., 2011; Martin-Vertedor and Dodd, 2011).

# Proteomic Studies of the Interaction Between Seed Viability, Seedling Growth and Abiotic Stress

# Tissue-Specific Proteomic Studies of Seed Developmental Stages During Abiotic Stresses

Seed developmental events are programmed to occur as a result of expression and activation of different proteins in distinct seed compartments (i.e., embryo, endosperm, and caryopsis coat) and even within specific regions (e.g., apical meristem), at distinct developmental stages (Itoh et al., 2005). Tissue- and organelle-specific proteomic studies relevant to seed development focused on characterization of temporal and spatial proteomes together with PTMs in metabolic and molecular events that occur at different seed developmental stages and the transit phases between stages (for review refer to Finnie et al., 2011a). These studies provided an insight into the physiological and biochemical pathways, stress-responsive mechanisms, nutrient accumulation and its regulation, redox regulation in programmed cell death, ferredoxin/thioredoxin-linked metabolic processes and signaling pathways, which are specific to embryo (Huang et al., 2012; Xu et al., 2012; Domínguez and Cejudo, 2014; Wolny et al., 2014), aleurone layer (Hägglund et al., 2010; Finnie et al., 2011b; Barba-Espín et al., 2014), endosperm (Vensel et al., 2005; Balmer et al., 2006), and peripheral layers (Jerkovic et al., 2010; Tasleem-Tahir et al., 2011; Miernyk and Johnston, 2013), and whole kernel solubility-based protein groups (Yang et al., 2011). These proteomic studies elucidated the molecular pathways underlying the control of seed development and physiological transitions, which contribute to the advancement of valuable and potentially agriculturally important strategies for improving yield, quality, and stress tolerance in cereals and legumes.

Pre-harvest sprouting (PHS) causes substantial losses in grain yield and quality, and therefore, it is one of the major factors negatively affecting the quality of crops in the areas with high levels of precipitation during grain maturation. Cereal crops with low levels of seed dormancy are susceptible to PHS when wet and moist conditions occur prior to harvest. A defined level of seed dormancy is under genetic, epigenetic regulation and environmental control, and it is an essential component of seed quality (Graeber et al., 2012). PHS resistance is a complex trait that is determined by genotype together with a number of other factors: stage of maturity, environmental conditions during grain ripening, crop morphology, biotic and abiotic stress (Fofana et al., 2008; Rikiishi and Maekawa, 2014). Proteomics offers the opportunity to examine simultaneous changes and to characterize temporal patterns of protein accumulation occurring during seed dormancy maintenance or release (Rajjou et al., 2011). Given that PHS is closely associated with seed dormancy, it is important to gain a deeper understanding of biomarker proteins and molecular signaling mechanisms involved in dormancy regulation, which will contribute to the breeding of PHS resistant cultivars during crop production. Recent proteomics and transcriptomics studies have indicated that antioxidant defense mechanisms, redox regulation of seed mRNAs and protein thiols, integrated with hormonal signaling play a key role in controlling seed dormancy in wheat (Bykova et al., 2011a,b; Liu et al., 2013a), barley (Bahin et al., 2011; Ishibashi et al., 2012), and rice (Huh et al., 2013; Liu et al., 2014). These findings will significantly contribute to the development of efficient strategies for breeding of PHS tolerant crops.

# Proteomic Analysis of Crop Seedlings Subjected to Abiotic Stress Conditions

Seeds break dormancy and restart their metabolism to prepare for germination then proceed with seedling establishment when environmental conditions are suitable for seedling growth and development (Arc et al., 2011). However, crop seedlings in their early growth stage are subjected to various abiotic stresses in the field, which will lead to lower yields and possible crop failure. Quantitative proteomic analysis of soybean seedlings and wheat roots subjected to the flooding or osmotic stresses has revealed the metabolic pathways of flooding-responsive proteome reacting to excess water supply and anoxia, while osmosisrelated proteins were responding to the various stresses such as drought, cold, and salinity stresses (Nanjo et al., 2011 and references therein). The initial changes in soybean root tip proteome under flooding stress indicated an important role of calcium signaling in the early responses (Yin et al., 2014), and the exogenous calcium treatment was shown to have a recovery effect on the expression of proteins involved in cell wall, hormone metabolisms, protein degradation and synthesis, and DNA synthesis in soybean roots under flooding stress (Oh et al., 2014). A comparative proteomic study of tissue-specific proteome in soybean seedlings under heat stress indicated common defense and adaptive mechanisms associated with elevated induction of several tissue-specific heat shock proteins and proteins involved in antioxidative defense (**Figure 1**). The down-regulated proteins were associated with photosynthesis, secondary metabolism, and amino acid and protein biosynthesis in response to heat stress (Ahsan et al., 2010b). Salinity stress-responsive proteins in seedlings have been investigated through a comparative proteomic analysis of salt-tolerant and salt-sensitive varieties of wheat (Guo et al., 2012) and soybean (Ma et al., 2012). A significant number of salt tolerance-related proteins were identified in wheat seedling roots, including signal transduction-associated proteins, carbon, amino acid and nitrogen metabolism proteins, detoxification and defense-related proteins (Guo et al., 2012). In a recent proteomics study of soybean seedling leaves, a salt stress-responsive protein network was proposed, including proteins responsible for redox homeostasis, as well as accelerated proteolysis, which was accompanied by reduced biosynthesis of proteins, impaired photosynthesis and energy supply, and enhanced ethylene biosynthesis (Ma et al., 2012). Recently, a comparative label-free shotgun proteomic analysis on wheat leaves of salt-stress tolerant genotype of durum wheat (*T. durum* Desf.; **Table 2**) subjected to increasing salinity levels revealed major changes in proteins involved in primary metabolism, energy production, protein metabolism and cellular defense, as well as tolerance-related high capacity for osmotic homeostasis, and increased cell wall lignification

allowing for higher growth recovery potential (Capriotti et al., 2014). In order to understand the role of ascorbate priming in boosting the salt stress tolerance during germination and early seedling growth in durum wheat, a label-free quantitative analysis of whole seeds (Fercha et al., 2013), as well as seed embryos and surrounding tissues was performed to study tissue-specific variation of metabolic proteomes (Fercha et al., 2014). It was demonstrated that ascorbate pretreatment prevents the effect of salinity by specifically changing the abundance of proteins involved in metabolism, protein destination and storage categories, which could be modulated by methionine, auxin and other hormones metabolism, as well as ROS managing and signaling systems (Fercha et al., 2014). These studies provided insights into possible management strategy of cellular activities occurring in salt-responding seedlings of major crops (**Figure 1**).

Among the cereals, rice (*Oryza sativa* L.) is a salt-sensitive crop. High salinity causes delayed seed germination, slow seedling growth, and reduced rate of seed maturation, leading to decreased rice yield. In the early period of growth, rice seedling roots, leaf sheath, and leaf blade are highly sensitive to salt stress signal, which is perceived via plasma membrane receptors and can rapidly initiate an intracellular signal to elicit an adaptive response (Ruan et al., 2011). Abbasi and Komatsu (2004) conducted a proteomic analysis of rice leaf sheath, leaf blades, and roots to investigate relative abundance of salt-responsive proteins, which changed according to intracellular ion homeostasis caused by continuous excessive ion uptake. In another study by Yan et al. (2005), it was suggested that energy production was activated in rice roots subjected to salt stress due to the disruption of enzyme activities and basic metabolism. Kim et al. (2005) elucidated differentially expressed proteins in rice seedling leaves during salt stress as being functionally important in major photosynthetic metabolic processes and in oxidative damage processes. Another study supported findings of the relationship between enzymes involved in carbohydrate and energy metabolism and the increased production of antioxidants that mediate maintenance of cellular homeostasis (Liu et al., 2013b). Song et al. (2011) predicted an extracellular salt stress-responsive apoplastic protein network in shoot stems of rice seedlings, suggesting candidate proteins involved in initial perception of salt stress, such as enzymes involved in carbohydrate metabolism, the intracellular equilibrium between ROS production and ROS scavenging, and protein processing and degradation. Moreover, transgenic rice seedlings overexpressing the cyclophilin OsCYP2 showed improved tolerance to salt stress, with an increased antioxidant enzyme activity and a decreased lipid peroxidation, which indicated a role for OsCYP2 in controlling oxidative damages by modulating activities of antioxidant enzymes at translational level (Ruan et al., 2011). Barley (*Hordeum vulgare* L.), on the other hand, is a salt-tolerant cereal crop and shows variation for tolerance toward salinity stress. MS-based proteomic analysis revealed cultivar-specific and salt stress-responsive protein expression patterns, indicating that proteins involved in the glutathione-based detoxification of ROS were highly expressed in the salt-tolerant genotype, while proteins involved in iron uptake were abundantly expressed in the salt-sensitive genotype. This study emphasized the role of proteins involved in ROS detoxification during salinity stress, and identified potential candidates for increasing salt tolerance in barley (Witzel et al., 2009). Moreover, a comparative transcriptional profiling of barley seedlings under salt stress elucidated a large number of hormone-related kinases and protein phosphatases involved in defense responses against salinity (Gao et al., 2013). These findings provide important information that will be used toward improving salt tolerance of cereals.

More recent proteomics research on the effect of environmental changes and abiotic stress in wheat can be grouped into studies focused on either water supply-responsive proteome reacting to drought (Bazargani et al., 2011; Ford et al., 2011; Ge et al., 2012; Kang et al., 2012; Budak et al., 2013; Alvarez et al., 2014; Zhang et al., 2014b), or temperaturerelated changes from extreme heat (Yang et al., 2011) to cold (Rinalducci et al., 2011a,b; Kosová et al., 2013) and frost tolerance (Vítámvás et al., 2012; Han et al., 2013), or salinity tolerance and stress (Kamal et al., 2012; Guo et al., 2012; Fercha et al., 2013, 2014; Jacoby et al., 2013a; Capriotti et al., 2014), and heavy metal toxicity stress (Li et al., 2013b; Kang et al., 2015). Availability of moisture during early growing season is critical to wheat production and prolonged drought conditions due to climate change have major impact on the crop yield. Using large-scale isobaric tags-based shotgun quantitative approach, differences in drought stress-responsive proteomes were studied in leaves from mature plants of three wheat varieties, fast and slow responding drought tolerant and intolerant (Ford et al., 2011; **Table 2**; **Figure 1**). An increase in proteins involved in ROS scavenging and a down regulation of proteins involved in photosynthesis and the Calvin cycle were observed in cellular response of all three varieties with the tolerant variety showing significant protein changes during early response and increases in proteins involved in cell detoxification. In a recent study Alvarez et al. (2014) used two wheat varieties adapted to different environmental conditions, drought tolerant and sensitive, to quantitatively evaluate inherent differences in protein expression patterns overall and in variety-specific effect of abscisic acid (ABA) on the root proteome (**Table 2**; **Figure 1**). It was revealed that the tolerant wheat variety had significantly higher number of ABAresponsive and ABA-induced proteins, most of those in the categories representing response to environmental stress and oxidation-reduction processes, which can play an important role in drought adaptation. Another study showed the effect of salicylic acid treatment on the induction of drought tolerance in wheat seedlings and identified salicylic acid-responsive protein interaction network suggesting more effective defense systems, efficient photosynthesis, active anabolism, and abundant energy supply (Kang et al., 2012). Significant changes in the phosphoproteome under drought stress were demonstrated in seedling leaves of two bread wheat cultivars using gel-free TiO2 phosphopeptide enrichment and label-free relative quantiative approach (Zhang et al., 2014b). Those mainly concerned proteins involved in RNA transcription/processing, stress/detoxification/defense, and signal transduction, drought tolerance and osmotic regulation.

Two wheat cultivars of contrasting temperature growth habits, frost tolerant winter wheat and frost sensitive spring wheat, were used recently to study cold-induced changes in the proteomes of crown tissues (Kosová et al., 2013). Proteins involved in the regulation of stress response, growth cessation and maintenance of vegetative stage were specifically increased in the winter cultivar, whereas proteins involved in restoration of cell division, plant growth and development, and progression to reproductive stage were induced by cold treatment in the spring cultivar. In an earlier study proteomic analysis of the changes induced by a prolonged treatment of vernalization-requiring winter wheat with low temperatures demonstrated down-regulation of several photosynthesis-related proteins and a concomitant increase in abundance of some Calvin and TCA cycle enzymes, and proteolysis (Rinalducci et al., 2011a). Another study on the effect of long-term cold acclimation and dynamics of cold tolerance on crown proteome composition in winter wheat cultivars (Vítámvás et al., 2012) showed that cold acclimation is an active energy-demanding process accompanied by profound changes in carbohydrate metabolism and ROS-scavenging enzymatic system, increased biosynthesis of methionine and *S*-adenosylmethionine, chaperones, enzymes involved in protein turnover and stress-responsive proteins. Molecular mechanisms helping a cold-sensitive spring wheat cultivar to survive long term exposure to suboptimal temperatures were also investigated from the proteomics point of view (Rinalducci et al., 2011b). Quantitative differences in protein abundance indicated reinforcement in ascorbate recycling, protein processing, and accumulation of the enzyme involved in tetrapyrrole resynthesis, whereas key Krebs cycle enzymes and many photosynthesis related proteins were down-regulated (**Figure 1**). A short term exposure of spring wheat plants to freezing stress caused extensive changes in major metabolic processes with enhanced accumulation of key proteins involved in ROS detoxification, signal transduction, stress and disease resistance (Han et al., 2013).

Recent plant proteomics studies have also been focused on heavy metal stress responses, biochemical mechanisms involved in cellular detoxification and heavy metal tolerance (Li et al., 2013b; Kang et al., 2015). The study of copper-induced stress responses reflected in proteomes of leaves and roots from young common wheat seedlings identified significant enhancement in the abundance of proteins involved in signal transduction, stress defense, and energy production, whereas carbohydrate, protein and photosynthetic metabolisms were severely reduced (Li et al., 2013b). Under copper stress conditions, exogenous jasmonic acid application had a protective effect, and activated transcription of glutathione-*S*-transferase. Organ-specific differences in adaptation to high-level mercury (Hg) stress were revealed for majority of proteins with differentially altered abundance in wheat seedling roots and leaves (Kang et al., 2015). The identified Hg-responsive proteins were associated with the main biological processes (**Figure 1**). Furthermore, application of exogenous ABA facilitated protection against Hg stress identifying potential network of key interacting proteins. These studies provide new insights and will lead to better understanding of heavy-metal stress responses in crop plants.

# Concluding Remarks and Future Challenges

Proteomic data that are particularly informative include quantitative protein profiles, profiles of regulatory modifications and protein interaction networks. Gel-based 2-DE proteomic approaches combined with gel-free MS-based quantitative proteomic techniques have been widely used for crop proteome analysis. The complex mixtures of proteins with the dynamic range of protein concentrations in plant cells have been analyzed more in-depth using a combination of separation techniques based on subcellular proteomics in different stress-responding organs and tissues. Improved protein extraction protocols, advanced gelfree quantitative techniques and bioinformatics approaches to the identification and analysis of complex proteomes at both subcellular and whole plant proteome levels in different crops have significantly advanced our understanding of developmental process, abiotic stress sensing mechanisms, and intracellular signal transduction mechanisms. The recent proteomic studies have contributed to elucidation of complex relationship between stress tolerance and crop productivity, which would enable the development of novel breeding strategies resulting in an increase in crop productivity and environmental performance.

Proteomics is rapidly becoming an indispensable tool for global phenotypic characterization at the molecular level that provides invaluable information about novel gene identifications, role of PTMs and protein interactions, linking genotype and its functionality. This type of information is especially useful in breeding programs offering specific identifications of potential biomarkers for isolation of candidate genes that can be integrated through proteomic-based marker-assisted selection and marker-based gene pyramiding (**Figure 2**). Furthermore, proteomics programs contribute to the analysis of advanced mapping populations such as hybrid doubled haploid lines, near isogenic lines and recombinant inbred lines, which further verify the correlations between responsive quantitative trait loci and stress tolerant phenotypes. One of the challenges is to convert massive data into knowledge that can be readily applied into crop improvement programs. The solution can be found in interdisciplinary approaches, creating sufficient genetic resources, robust bioinformatics tools with novel algorithmic solutions.

# Acknowledgment

The authors acknowledge Agriculture and Agri-Food Canada for funding support.

# References


during germination under salt stress. *J. Proteomics* 108, 238–257. doi: 10.1016/j.jprot.2014.04.040


the detection of novel nodule phosphoproteins by mass spectrometry. *J. Exp. Bot.* 59, 3307–3315. doi: 10.1093/jxb/ern182


**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.

*Copyright © 2015 Hu, Rampitsch and Bykova. 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.*

# Proteomics of stress responses in wheat and barley—search for potential protein markers of stress tolerance

# *Klára Kosová\*, Pavel Vítámvás and Ilja T. Prášil*

*Laboratory of Plant Stress Biology and Biotechnology, Division of Crop Genetics and Breeding, Department of Plant Genetics, Breeding and Product Quality, Crop Research Institute, Prague, Czech Republic*

### *Edited by:*

*Jesus V. Jorrin Novo, University of Cordoba, Spain*

### *Reviewed by:*

*Martin Hajduch, Slovak Academy of Sciences, Slovakia Jesus V. Jorrin Novo, University of Cordoba, Spain Arkadiusz Kosmala, Institute of Plant Genetics of the Polish Academy of Sciences, Poland*

### *\*Correspondence:*

*Klára Kosová, Laboratory of Plant Stress Biology and Biotechnology, Division of Crop Genetics and Breeding, Crop Research Institute, Drnovská 507, Prague 6 – Ruzyne,ˇ 161 06, Czech Republic e-mail: kosova@vurv.cz*

Wheat (*Triticum aestivum; T. durum*) and barley (*Hordeum vulgare*) agricultural production is severely limited by various abiotic and biotic stress factors. Proteins are directly involved in plant stress response so it is important to study proteome changes under various stress conditions. Generally, both abiotic and biotic stress factors induce profound alterations in protein network covering signaling, energy metabolism (glycolysis, Krebs cycle, ATP biosynthesis, photosynthesis), storage proteins, protein metabolism, several other biosynthetic pathways (e.g., S-adenosylmethionine metabolism, lignin metabolism), transport proteins, proteins involved in protein folding and chaperone activities, other protective proteins (LEA, PR proteins), ROS scavenging enzymes as well as proteins affecting regulation of plant growth and development. Proteins which have been reported to reveal significant differences in their relative abundance or posttranslational modifications between wheat, barley or related species genotypes under stress conditions are listed and their potential role in underlying the differential stress response is discussed. In conclusion, potential future roles of the results of proteomic studies in practical applications such as breeding for an enhanced stress tolerance and the possibilities to test and use protein markers in the breeding are suggested.

**Keywords: proteome, barley, wheat, abiotic stress factors, biotic stress factors, protein markers**

# **INTRODUCTION**

Wheat (*Triticum aestivum*; *T. durum*) and barley (*Hordeum vulgare*) represent major cereal crops grown in temperate climate areas. Cereal agricultural production is limited by a wide array of abiotic and biotic stress factors including drought (Cattivelli et al., 2008), cold (Thomashow, 1999; Kosová et al., 2008a), heat, salinity (Munns, 2005; Kosová et al., 2013a), imbalances in mineral nutrition, viral (Kosová et al., 2008b) and fungal pathogens such as *Fusarium* (Kosová et al., 2009; Yang et al., 2010a,b), leaf rust (*Puccinia triticina*; Rampitsch et al., 2006), blotch (*Septoria tritici*; Yang et al., 2013) and others, often acting in combinations under field conditions (Mittler, 2006). Proteome plays an important role in stress response since proteins are directly involved in several processes aimed at an enhancement of stress tolerance being "closer to phenotype" than transcripts.

During the past decades, the boom of high-throughput proteomics techniques has enabled the researchers to study plant proteome responses to various factors including stresses in a complex way. Despite numerous studies reporting identifications of a few thousand of proteins in plant samples, a complete description of plant proteome in a given tissue, developmental phase and environmental conditions still remains a great challenge (Jorrin-Novo et al., 2009).

Both abiotic and biotic stresses induce profound changes in plant proteomes aimed at an adjustment of metabolism to altered environment and an enhancement of plant stress tolerance. Plant stress response is a dynamic process and several phases with a unique proteome composition could be distinguished (Levitt, 1980; Larcher, 2003). Reviews on plant proteome responses to abiotic stresses (Kosová et al., 2011; Hossain et al., 2012) and pathogens (Sergeant and Renaut, 2010; Gonzalez-Fernandez and Jorrin-Novo, 2012) provide important overviews; however, several novel studies were published recently (**Table 1**).

Most proteomic papers aimed at an investigation of plant stress responses are comparative studies that are based on comparison of proteome composition in stressed plants vs. control ones, and also in differentially-tolerant genotypes exposed to stress. Moreover, studies on the roles of subcellular proteomes such as chloroplast (Kamal et al., 2012) and mitochondrial (Jacoby et al., 2010, 2013) proteomes as well as posttranslational modifications (PTMs) such as phosphoproteomics (Yang et al., 2013; Zhang et al., 2014) in wheat exposed to stress have been published recently. Considering the increasing amount of proteomic data, it is arising necessary to mine the data published in various proteomic studies in order to identify key proteins involved in plant responses to a wide array of stress factors (dehydrative stresses—drought, osmotic stress, salinity, frost, heat) as well as proteins induced only at specific stress conditions (e.g., phytochelatins and heavy metal stress). An attempt has already been published regarding proteomic studies under salinity (Zhang et al., 2012). Moreover, comparison of proteome responses in


**Table 1 | A list of proteomic studies focused on abiotic and biotic stress responses in wheat (***Triticum aestivum; T. durum***), barley (***Hordeum vulgare***), and related species.**






**170**


*Abbreviations: 2Cys-Prx, 2-cysteine peroxiredoxin; 2DE, two-dimensional electrophoresis; 2D-DIGE, two-dimensional differential in-gel electrophoresis;* β*-CAS,* β*cyanoalanine synthase; ABA, abscisic acid; ACP, acyl carrier protein; AGPase, ADP glucose pyrophosphorylase; AOX, alternative oxidase; APX, ascorbate peroxidase; AQP, aquaporin; AsA, ascorbic acid; CCOMT, caffeoyl-coenzyme A O-methyltransferase; COR, Cold-regulated (protein); CPN, chaperonin; CS, cysteine synthase; CDPK, calcium-dependent protein kinase; DAP, differentially abundant proteins; DH, double haploid (line); DHAR, dehydroascorbate reductase; DON, deoxynivalenol; ENO, enolase; FBP ALDO, fructose-1,6-bisphosphate aldolase; FWC, field water capacity; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GAPDH B, glyceraldehyde-3-phosphate dehydrogenase B form; GDH, glutamate dehydrogenase; GLP, germin-like protein; GPX, glutathione peroxidase; GS, glutamine synthetase; GST, glutathione S-transferase; HPLC, high performance liquid chromatography; Hsc70, heat shock cognate protein 70; iTRAQ, isobaric tag for relative and absolute quantification; LC, liquid chromatography; LEA, Late embryogenesis-abundant (protein); LOX, lipoxygenase; LTQ-FTICR, linear quadruple trap-Fourier transform ion cyclotron resonance; MALDI-TOF/TOF, matrix-assisted laser desorption ionization time-of-flight/time-of-flight (spectrometry); MAPK, mitogen-activated protein kinase; MDAR, monodehydroascorbate reductase; MDH, malate dehydrogenase; MS, mass spectrometry; MSSP2, monosaccharide sensing protein 2; NBS-LRR, nucleotide-binding site leucine-rich repeat protein; NEPHGE, non-equilibrium pH gel electrophoresis; NDPK, nucleoside diphosphate kinase; NIL, near-isogenic line; OEE, oxygen evolving enhancer (protein); PBS, phosphate buffer saline; PC, plastocyanin; PDI, protein disulfide isomerase; PDX, pyridoxal biosynthesis protein; PEG, polyethylene glycol; PGK, phosphoglycerokinase; PGM, phosphoglyceromutase; POX, peroxidase; PPDK, pyruvate phosphate dikinase; PRK, phosphoribulokinase; Prx, peroxiredoxin; PS, photosystem; PVP, polyvinyl pyrrolidone; qTOF, quadrupole time-of-flight; RubisCO, ribulose-1,5-bisphosphate carboxylase/oxygenase; RubisCO LSU, RubisCO large subunit; RubisCO SSU, RubisCO small subunit; RWC, relative water content; S, sensitive (genotype); SA, salicylic acid; SHMT, serine hydroxymethyltransferase; SnRK, sucrose non-fermenting-related protein kinase; SOD, superoxide dismutase; SUS1, sucrose synthase 1; T, tolerant (genotype); t, genotype less tolerant than T; TCA, trichloroacetic acid; TLP, thaumatin-like protein; TPI, triose phosphate isomerase; Trx, thioredoxin; V-ATPase, vacuolar ATPase; VDAC, voltage-dependent anion channel; WCS, Wheat Cold-specific (protein); WRAB, Wheat responsive-to-ABA (protein); XET, xyloglucan endo-transglycosylase.*

differentially-tolerant genotypes may help researchers to identify key proteins underlying the differences in stress tolerance.

The aim of this minireview is to summarize the major results obtained by proteomic studies in temperate cereal crops wheat and barley studied under abiotic and biotic stresses. Proteins affected by differential stress factors and proteins revealing a differential response between differentially-tolerant wheat and barley genotypes are discussed in a greater detail. Possibilities of utilization of proteins revealing a differential stress response between tolerant and sensitive genotypes as protein markers in breeding programs aimed at improvement of stress tolerance are suggested.

### **COMMON FEATURES OF STRESS RESPONSE AT PROTEOME LEVEL**

Plant stress response is a dynamic process aimed at an enhancement of plant stress tolerance and an establishment of a novel homeostasis between plant and environment (**Figure 1**). Several

phases of plant stress response could be distinguished including an alarm phase, an acclimation phase, a resistance phase, an exhaustion phase when stress is too severe or lasts too long, and a recovery phase after a cessation of the given stress factor (Levitt, 1980; Larcher, 2003; Kosová et al., 2011). At proteome level, profound alterations in protein relative abundance were found between stressed and control plants as well as between differential genotypes (**Table 1**). During an alarm phase, stress induces profound alterations in proteins involved in cell signaling although these proteins are detected scarcely on 2DE gels due to their low abundance. An increase in 14-3-3 proteins as well as translationally controlled tumor protein homologs was detected in copperand water-stressed wheat (Kang et al., 2012; Ghabooli et al., 2013; Li et al., 2013; Alvarez et al., 2014) and barley (Wendelboe-Nelson and Morris, 2012) and genotype-specific responses of β subunit of heterotrimeric G protein were found in salt-stressed wheat (Peng et al., 2009). Phosphorylation plays an important role in abiotic and biotic stress responses as shown on several kinases (calciumdependent protein kinases CDPK, mitogen-activated protein kinases MAPK, sucrose non-fermenting-related kinases SnRK2), phosphatases (PP2C) and other signaling proteins (calmodulin 2-2) regulation (Yang et al., 2013; Zhang et al., 2014).

Stress acclimation represents an adaptive process aimed at an enhancement of plant stress tolerance. An active stress acclimation requires relatively high energy costs as indicated by profound alterations in energy metabolism. Practically all stresses induce an increased relative abundance of enzymes of carbohydrate catabolism such as glycolysis (glyceraldehyde-3-phosphate dehydrogenase GAPDH, triosephosphate isomerase TPI, enolase ENO), Krebs cycle (mitochondrial NAD+-dependent malate dehydrogenase (MDH; Vítámvás et al., 2012), aconitase (Jacoby et al., 2010, 2013; Budak et al., 2013) and components of mitochondrial ATP-synthase, namely β subunit of CF1 complex (Bahrman et al., 2004; Patterson et al., 2007; Vítámvás et al., 2012; Budak et al., 2013; Kosová et al., 2013b; Rollins et al., 2013; Xu et al., 2013) indicating an enhanced demand for energy. Regarding photosynthesis, an increase or a decrease in several photosynthetic proteins (proteins involved in primary photosynthetic reactions, carbon fixation, and Calvin cycle) have been observed depending on the severity of stress (Caruso et al., 2008, 2009; Ye et al., 2013). A downregulation of photosynthesis reactions under severe stress is reflected at proteome level by a decrease in D1 and D2 proteins in photosystem II reaction center (RC PSII), proteins of oxygen evolving complex (OEC), a decrease in chlorophyll *a*-*b* binding proteins in both photosystem (PS) I and II, a decrease in Fe-S complex in PSI, a downregulation of RubisCO and key Calvin cycle enzymes phosphoglycerate kinase and phosphoribulokinase in cold- and waterlogging-treated winter wheat (Li et al., 2014), salt-treated durum wheat (Caruso et al., 2008) and in drought-treated barley (Ghabooli et al., 2013) while an increase in OEC protein OEE2 was found in salt-treated barley (Rasoulnia et al., 2011; Fatehi et al., 2012). In addition, an increase in proteins with stimulating and protective functions such as RubisCO activase A (Bahrman et al., 2004; Caruso et al., 2008, 2009; Fatehi et al., 2012; Budak et al., 2013), a thermostable RubisCO activase B (Rollins et al., 2013), carbonic anhydrase (Caruso et al., 2008) and RubisCO large and small subunit binding proteins CPN60-α and CPN60-β was observed under various stresses (Caruso et al., 2008; Sarhadi et al., 2010; Kang et al., 2012; Budak et al., 2013; Kosová et al., 2013b; Xu et al., 2013).

An increased demand on energy under stress acclimation corresponds with a decreased abundance of enzymes (fructokinase-2, sucrose synthase-1) involved in biosynthesis of energy-rich compounds such as starch and a decrease in storage proteins (11S seed storage protein 2-like, legumin-like protein; Vítámvás et al., 2012; Kosová et al., 2013b).

Stress acclimation also reveals an enhanced demand on protein metabolism including both protein biosynthesis and degradation. Changes in the levels of eukaryotic translation initiation factors eIF3 subunit I, eIF5A2 (Kosová et al., 2013b), eIF4A (Rollins et al., 2013) and elongation factor eEF1-α (Budak et al., 2013), several ribosomal proteins, e.g., 60S proteins P0, P2A, L3, L38 (Fercha et al., 2014), or chloroplastic ribosomal proteins 30S-3, 50S-L12 (Ghabooli et al., 2013; Gharechahi et al., 2014), as well as in proteasome subunits such as 20S proteasome subunit αtype 1 and 6 (Rampitsch et al., 2006; Rinalducci et al., 2011a; Fercha et al., 2013; Ghabooli et al., 2013) and proteins of ubiquitin pathway involved in proteasome targeting such as ubiquitin conjugating enzyme E2 variant IC like (Kosová et al., 2013b) were reported indicating an enhanced protein turnover during stress acclimation.

Stress represents an enhanced risk of protein damage due to imbalances in cellular homeostasis. Therefore, increased abundances of several proteins with chaperone and other protective functions have been reported. Extreme temperatures, but also drought, pathogens, and other stresses cause an enhanced risk of protein misfolding and they are thus associated with an enhanced accumulation of chaperones from HSP superfamily, namely HSP70 (Rampitsch et al., 2006; Li et al., 2013; Rollins et al., 2013), HSP100 (Clp protease; Ashoub et al., 2013), and small HSPs (Skylas et al., 2002; Majoul et al., 2004; Hajheidari et al., 2007), but also others such as chopper chaperone (Vítámvás et al., 2012; Hlavácková et al., 2013 ˇ ), serpins (Yang et al., 2010b; Fercha et al., 2013, 2014), and protein disulfide isomerase (Hajheidari et al., 2007; Vítámvás et al., 2012; Li et al., 2013). However, a decrease in HSP90 was reported under cold (Vítámvás et al., 2012). Disorders in cellular metabolism under stress lead to an enhanced risk of oxidative damage. At proteome level, an increased abundance of several reactive oxygen species (ROS) scavenging enzymes was found practically under each kind of stress (Hajheidari et al., 2007; Ford et al., 2011). Plants try to reduce a risk of ROS formation by several ways. The major one represents a downregulation of photosynthesis reactions which is associated with a decrease in D1 and D2 proteins in photosystem II reaction center (RC PSII), proteins of OEC, RubisCO small subunit and key Calvin cycle enzymes phosphoglycerate kinase, phosphoribulokinase and transketolase (Caruso et al., 2008; Ford et al., 2011; Ashoub et al., 2013). Other indirect ways how to reduce ROS lie in a reduced uptake of metal ions, especially iron, which can act as catalyzers of ROS formation. A reduced level of protein IDI2 and dioxygenases IDS2, IDS3 involved in iron uptake and phytosiderophore biosynthesis was found by Witzel et al. (2009) in salt-treated barley roots while an increased level of these enzymes was found by Patterson et al. (2007) in barley grown under elevated boron.

Several stresses including drought, heat, salinity, cold, but also mechanical wounding, induce an enhanced accumulation of proteins belonging to LEA superfamily. Late embryogenesisabundant (LEA) superfamily includes at least five subclasses, the most important being LEA-II (dehydrins) and LEA-III proteins whose transcript and protein levels, and also phosphorylation level, have been reported to correlate with wheat and barley tolerance to low temperatures (Crosatti et al., 1995; Vágújfalvi et al., 2000, 2003; Vítámvás et al., 2007; Kosová et al., 2008c, 2013c; Sarhadi et al., 2010), drought (Labhilili et al., 1995; Brini et al., 2007) and other stresses.

Several stresses, especially biotic ones, are associated with an induction of protective proteins from PR superfamily. Pathogenesis-related (PR) proteins encompass 16 groups involved in defense against microbial and fungal pathogens (Edreva, 2005). Many of PR proteins can resist acidic pH, they reveal enzymatic activities aimed at modifications of cell wall, and ROS scavenging functions (some germins and germin-like proteins reveal manganese superoxide dismutase (Mn-SOD) and oxalate oxidase activities). An enhanced abundance of several PR proteins was reported not only in cereals exposed to fungal pathogens such as *Fusarium* (class-II chitinase, β-amylase, thaumatin-like protein, PR9–peroxidase; Yang et al., 2010a,b; Eggert and Pawelzik, 2011; Eggert et al., 2011), but also under abiotic stresses such as cold (β-1,3-glucanase, chitinase, PR4, thaumatin-like protein; Sarhadi et al., 2010; Kosová et al., 2013b; Gharechahi et al., 2014), salinity (germin-like protein, PR10; Fatehi et al., 2012; Kamal et al., 2012; Witzel et al., 2014), and others.

Stresses also affect other aspects of cellular metabolism. An increased abundance of methionine synthase catalyzing formation of methionine or S-adenosylmethionine synthase (SAMS) catalyzing formation of S-adenosylmethionine (SAM) has been reported (Bahrman et al., 2004; Patterson et al., 2007; Witzel et al., 2009; Vítámvás et al., 2012; Kosová et al., 2013b; Xu et al., 2013). SAM represents not only a universal methyl donor in regulation of DNA heterochromatin formation and gene expression, but it is also a precursor of several stress-related metabolites as glycine betaine, polyamines, hydroxymugineic acids (phytosiderophore precursors; Mori and Nishizawa, 1987) and ethylene. Alterations in glutamine synthetase (GS) have been reported under drought (Ford et al., 2011; Kang et al., 2012) and cold (Hlavácková et al., ˇ 2013) indicating an important role of nitrogen assimilation and proline biosynthesis in stress acclimation.

Stress affects cellular transport and membrane properties. An enhanced need for ion transport and thus an associated increase in plasma membrane and tonoplast ion transporters such as V-ATPase has been reported not only under salinity (Peng et al., 2009), but also under other stresses such as drought (Ghabooli et al., 2013), heat (Majoul et al., 2004) and osmotic stress (Ye et al., 2013; Zhang et al., 2014). Differential phosphorylation of several transport proteins such as aquaporins, H+-ATPase or monosaccharide sensing protein 2, was also reported in response to stress (Zhang et al., 2014). The effect of several stresses on cell wall remodeling is indicated by alterations in several enzymes involved in lignin metabolism such as caffeoyl-coenzyme A O-methyltransferase CCOMT indicating an important role of cell wall in plant stress response (Sugimoto and Takeda, 2009; Ghabooli et al., 2013).

Long-term and regularly occurring stress factors such as cold during winter also affect plant development. At proteome level, significant changes in the level of small glycine-rich RNA-binding proteins (sGRPs) and in lectins, glycoproteins involved in saccharide signaling, were found in wheat (Rinalducci et al., 2011b; Kosová et al., 2013b). Ricin B lectin 2 was reported to be induced by cold in crowns of both winter barley (Hlavácková et al., ˇ 2013) and winter wheat (Kosová et al., 2013b). Lectin VER2 was reported to accumulate in winter wheat shoot apex until vernalization (Yong et al., 2003; Rinalducci et al., 2011b). Differences in sGRPs and VER2 levels between spring and winter wheat growth habits indicate a differential response to cold within wheat germplasm (Kosová et al., 2013b).

### **PROTEINS REVEALING A DIFFERENTIAL RESPONSE BETWEEN STRESS-TOLERANT AND STRESS-SENSITIVE GENOTYPES**

A differential ability of various wheat and barley genotypes to cope with several stresses is reflected also at protein level. Stresstolerant genotypes do not suffer from a disruption of energy metabolism when exposed to moderate stress levels; moreover, when exposed to stress, they can increase an abundance of key enzymes of energy metabolism to increase ATP production as indicated by a differential response observed in several photosynthesis-related proteins (RubisCO subunits, RubisCO activase), ROS scavenging enzymes as well as respiration (Krebs cycle) enzymes. Quantitative differences in Krebs cycle enzymes such as mitochondrial NAD+-dependent MDH between two differentially frost-tolerant winter wheats (Vítámvás et al., 2012), in aconitase (Budak et al., 2013), thioredoxin *h* and glutathione-Stransferase (GST; Hajheidari et al., 2007; Sarhadi et al., 2010), lipoxygenase 1 and 2 (Alvarez et al., 2014) between differentially drought-tolerant wheats; in Cu/Zn-SOD, Mn-SOD (Ford et al., 2011; Xu et al., 2013), glyoxysomal MDH (gMDH; Ashoub et al., 2013), GST (Rasoulnia et al., 2011), class III peroxidase, catalase and lipoxygenase (Wendelboe-Nelson and Morris, 2012) between differentially drought- and salt-tolerant barleys; in Mn-SOD, MDH and aconitase between salt-treated wheat and wheat × *Lophopyrum elongatum* amphiploid (Jacoby et al., 2013), and a downregulation of MDH and isocitrate dehydrogenase in coldsensitive spring wheat (Rinalducci et al., 2011a) indicate a crucial role of mitochondrial respiration and ROS metabolism in stress acclimation. Along with these data, a differential abundance in storage proteins such as legumin-like protein between two differentially frost-tolerant winter wheats was found by Vítámvás et al. (2012) indicating a higher demand on energy ensured by storage compound degradation in the less-tolerant genotype. Moreover, tolerant genotypes can also afford to accumulate higher amounts of stress-protective proteins such as PR proteins (Witzel et al., 2014) and ABA-responsive proteins (Alvarez et al., 2014). A significant correlation between wheat WCS120 and barley DHN5 dehydrin relative accumulation and acquired frost tolerance (FT) determined as lethal temperature for 50 % of the sample (LT50) was reported for winter genotypes grown under both cold and moderate cold temperatures (Vítámvás et al., 2007, 2010; Kosová et al., 2008c, 2013c). WCS120 and DHN5 can be thus considered promising FT markers.

Stress-tolerant and stress-sensitive genotypes or related plant species also reveal significant differences in proteins involved in regulation of cell cycle and plant development. Factor eIF5A2 does not only regulate translation inititation, but it is also known to participate in the regulation of cell cycle switch between cell proliferation and death (Thompson et al., 2004). Under salinity, a decreased abundance of eIF5A2 with respect to control was found in both salt-sensitive common wheat and salt-tolerant *T. aestivum* × *Thinopyrum ponticum* hybrid (Wang et al., 2008); however, a decrease in *T. aestivum* × *Th. ponticum* was much lower than in *T. aestivum* indicating a higher cell proliferation rate in the salttolerant hybrid. A differential abundance in lectin VER2 between cold-treated spring and winter wheat cultivars corresponds to a differential developmental response with a winter wheat revealing a developmental arrest while a spring wheat revealing a progression to reproductive phase as indicated at proteome and phytohormone levels (Kosová et al., 2012, 2013b).

### **CONCLUSIONS AND FUTURE PERSPECTIVES**

Both abiotic and biotic stress factors induce an active plant stress response including a profound reorganization of plant proteome. Comparative proteomic studies are usually carried out on a limited range of plant material due to their expensiveness and much of sophisticated work. However, they can significantly contribute to identification of novel proteins revealing a differential response in abundance or PTMs between differentially-tolerant genotypes and representing potential protein markers of stress tolerance. The potential markers should be tested on a broad range of genotypes using simple protein quantification methods as ELISA or immunoblots which can be utilized by breeders. As an example, proteomic studies on cold-treated winter wheats resulting in an identification and testing of dehydrin proteins as FT markers can be given (Vítámvás et al., 2007, 2010). Recent publication of draft barley (The International Barley Genome Sequencing Consortium, 2012) and wheat (The International Wheat Genome Sequencing Consortium, 2014) genome sequences will significantly contribute to protein identification, sequentional characterization and preparation of specific antibodies which will stimulate further research and applications in breeding for an improved stress tolerance.

### **AUTHOR CONTRIBUTIONS**

Klára Kosová has outlined the idea and prepared the text. Pavel Vítámvás and Ilja T. Prášil contributed to preparation, drafting, critical reading, and publication of the manuscript.

### **ACKNOWLEDGMENTS**

The work was supported by projects from Ministry of Education, Youth and Sports of the Czech Republic (MEYS CZ) LD14064 and LD14087 as a part of of COST actions FA1204 and FA1208, respectively. The work was also supported by an institutional project MZE RO0414 from Ministry of Agriculture of the Czech Republic (MZe CZ).

### **REFERENCES**


water stress tolerance induced by *Piriformospora indica* in barley. *J. Proteomics* 94, 289–301. doi: 10.1016/j.jprot.2013.09.017


Zhang, M., Lv, D., Ge, P., Bian, Y., Chen, G., Zhu, G., et al. (2014). Phosphoproteome analysis reveals new drought response and defense mechanisms of seedling leaves in bread wheat (*Triticum aestivum* L.). *J. Proteomics* 109, 290–308. doi: 10.1016/j.jprot.2014.07.010

**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 2014; accepted: 26 November 2014; published online: 11 December 2014.*

*Citation: Kosová K, Vítámvás P and Prášil IT (2014) Proteomics of stress responses in wheat and barley—search for potential protein markers of stress tolerance. Front. Plant Sci. 5:711. doi: 10.3389/fpls.2014.00711*

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

*Copyright © 2014 Kosová, Vítámvás and Prášil. 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.*

# Gamma-glutamyl cycle in plants: a bridge connecting the environment to the plant cell?

Antonio Masi <sup>1</sup> \*, Anna R. Trentin<sup>1</sup> , Ganesh K. Agrawal 2, 3 and Randeep Rakwal 2, 3, 4, <sup>5</sup>

*<sup>1</sup> Dipartimento di Agronomia Animali Alimenti Risorse Naturali e Ambiente (DAFNAE), University of Padova, Legnaro, Italy, <sup>2</sup> Research Laboratory for Biotechnology and Biochemistry, Kathmandu, Nepal, <sup>3</sup> GRADE (Global Research Arch for Developing Education) Academy Private Limited, Birgunj, Nepal, <sup>4</sup> Organization for Educational Initiatives, University of Tsukuba, Tsukuba, Japan, <sup>5</sup> Department of Anatomy I, Showa University School of Medicine, Shinagawa, Japan*

Keywords: glutathione, oxidative stress, redox sensing, gamma-glutamyltransferase, plant acclimation

# Apoplast and Redox Components in Plants Acclimation to Environment

The apoplast represents a compartment where an extensive cross-talk occurs among different components, to generate signals that can pass through the plasmalemma and reach the symplast (Agrawal et al., 2010). Both abiotic and biotic stress conditions evoke defensive and adaptive responses. Occurrence of structural and metabolic readjustments is then driven by enzymes (proteins), whose coordinated expressions are regulated by signals and signal transduction pathways (Foyer and Noctor, 2005).

### Edited by:

*Subhra Chakraborty, National Institute of Plant Genome Research, India*

### Reviewed by:

*Dominique Job, Centre National de la Recherche Scientifique, France Norman Peter Andrew Huner, Western University, Canada*

> \*Correspondence: *Antonio Masi, antonio.masi@unipd.it*

### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *28 January 2015* Accepted: *30 March 2015* Published: *16 April 2015*

### Citation:

*Masi A, Trentin AR, Agrawal GK and Rakwal R (2015) Gamma-glutamyl cycle in plants: a bridge connecting the environment to the plant cell? Front. Plant Sci. 6:252. doi: 10.3389/fpls.2015.00252*

External environmental factors initiate extracellular signals. Signals are then transferred to inner compartment via receptors located on the plasma membrane initiating a signal transduction pathway to readjust cell metabolism to the new conditions. This task requires a concerted action of many players: specific genes expression, post-transcriptional and post-translational regulation, hormones, and cell regulators.

When trying to explain the process of plant sensing and acclimation to environment, key questions arise: what are the signals generated by the environment? How can they evoke the response? A widely accepted view is that many unfavorable conditions result in the appearance of reactive oxygen species (ROS) (Pitzschke et al., 2006). ROS are the natural consequence of a life in an oxygen-containing atmosphere, and result from any imbalance in the electron flow in fundamental processes such as photosynthesis and respiration. They are represented by oxygen-containing radical species or hydrogen peroxide, H2O2, having an intrinsic reactivity with the organic molecules which can be consequently either damaged or undergo a redox modification. The ROS increase in apoplast under oxidative conditions has been documented (Mittler et al., 2004; Potters et al., 2010). ROS are involved in cell wall synthesis, remodeling and plant-pathogen interactions (Torres et al., 2006). ROS and redox modifications thus seem to be good candidates in transferring the environment-related information to cell, together with other signaling molecules, such as the extracellular ATP (Cao et al., 2014). Integration of apoplastic and chloroplastic ROS signaling processes has also been studied in different model organisms during stress conditions, suggesting that the extracellular ROS signal is transduced to chloroplasts, thus initiating a secondary and amplified ROS production (reviewed in Shapiguzov et al., 2012).

# Gamma-glutamyl Cycle and Gamma-glutamyl-transferases

In plants, the gamma-glutamyl cycle is a metabolic route of extra-cytosolic (apoplastic and vacuolar) glutathione degradation by gamma-glutamyl-transferase (GGT) and cys-gly dipeptidase, followed by the re-uptake of constituent amino acids, intracellular re-synthesis and extrusion (Ferretti et al., 2009). Ohkama-Ohtsu et al. (2008) demonstrated that an alternative pathway of glutathione degradation by means of gamma-glutamyl cyclotransferase (GGCT) and 5-oxo-prolinase (5Opase) dominates over GGT degradation in plant tissues. However, the two pathways operate in different compartments; GGTs are extracytosolic (apoplastic and vacuolar) whereas GGCT and 5OPase activities are restricted in the cytosol. The two degradation pathways coexist and operate independently of one another, and have therefore distinct physiological significance and regulation. Thus, the gamma-glutamyl cycle involving apoplastic GGTs is functional to the recovery of extracellular glutathione, whereas the alternative GGCT/5OPase pathway participates in controlling cytosolic glutathione homeostasis (Noctor et al., 2011).

In Arabidopsis, a detailed description of the four GGT genes expression was obtained by GUS-staining of transformed lines (Martin et al., 2007). GGT1 and GGT2 have high similarity and sequence identity, and are located to the apoplast (Ferretti et al., 2009). The GGT1 is ionically cell-wall bound and expressed in most vascular tissues (Ferretti et al., 2009), whereas GGT2 seems to be preferentially associated to plasma membranes and expressed in specific tissues in seeds, flowers, and roots. GGT3 is considered a non-functional and truncated sequence, whereas GGT4 is localized to vacuole assisting degradation of the GSconjugates of toxic compounds and xenobiotics (Grzam et al., 2007). The significance of GSH cycling between the extracellular and intracellular space was addressed in the Arabidopsis mutant line lacking the ggt1 isoform by performing comparative quantitative proteomics of the total leaf proteins (Tolin et al., 2013). In that study, it was reported that disrupture of the gammaglutamyl cycle by ggt1 silencing results in increased abundance of an array of antioxidant and defense protein enzymes, which could be collectively described as a "constitutive alert response."

The occurrence of glutathione in apoplast has often been questioned in the past, but several evidences now indicate its existence albeit at low level (Zechmann, 2014). It seems however puzzling that a glutathione degradation activity, occurring outside the cell, can result in redox alteration inside the cell. Due to its low extracellular concentration, it is unlikely that glutathione itself acts as an antioxidant outside the cell. That function might better be fulfilled by abundant ascorbate in apoplast (Pignocchi and Foyer, 2003); where in any case oxidizing conditions are prevalent.

# Extracellular Glutathione and Glutathione Degradation Activity

All this considered, what could be then the function of extracellular glutathione and glutathione degradation activity? Some key elements worth considering are: (i) presence of a redox-sensitive thiol group in the molecule; (ii) apoplastic ROS production as a consequence of adverse conditions; (iii) presence of the plasmamembrane bound receptors; and (iv) redox exchange reactions occurring between the low-molecular-weight thiols and cysteines of plasma-membrane bound proteins, acting as redox switches. In order for a molecule to act as a signal, its concentration should be low and un-buffered, such that perturbations may induce large variations in its pool size. The reversible conversion of reduced to oxidized form may also rapidly modify the GSH pool. The interaction and exchange reactions of low-molecular-weight thiols and cysteines of plasma-membrane receptors and components may secondarily amplify the signal. On the other hand, the possibility that gamma-glutamyl cycling be implicated in the response to oxidative stress might be inferred by some previous reports (Masi et al., 2002; Ferretti et al., 2009).

To better investigate the relationship between oxidizing stress conditions and GGT-driven glutathione degradation, apoplastic fluid proteins were extracted from leaves of the ggt1 mutant following ultraviolet B (UV-B) treatment (Trentin et al., 2015). Comparative quantitative proteomics suggests that while abundance of cell wall remodeling proteins is affected by both UV-B and ggt1 silencing, the mutation itself resulted in reduced expression of a number of plasma-membrane associated genes (cysrich, leucine-rich secretory proteins) involved in signaling and assigned to "response to stimulus" as per the gene ontology. Alteration in expression of ROS components (i.e., superoxide dismutase, glutathione S-transferases or peroxidases) is also observed under stress conditions. But given the presence of parallel alternative pathways, it is hard to predict whether: (i) the level of apoplastic H2O<sup>2</sup> is increased or not; and (ii) H2O<sup>2</sup> is the molecule involved in transferring the signals arising from apoplast.

# Proteomics as a Tool to Understand the Gamma-glutamyl Cycle

Proteomics technology has been very useful in better understanding the gamma-glutamyl cycle by using the GGT mutant plants. It was a proteomics study of the ggt1 mutants that provided evidence on alterations in abundance of protein components involved in antioxidative and defense responses, and that may convey redox information from the extracellular milieu to internal compartments. In the future, proteomics may contribute to pinpoint plasma membrane components that are clearly involved in this process. Proteomics may also help in identifying one missing step in the gamma-glutamyl cycle, i.e., the cysteinyl-glycine dipeptidase, whose occurrence is inferred but not demonstrated yet.

# Conclusions

The significance of the gamma-glutamyl cycle is not fully understood yet. Glutathione cycling between the symplast and apoplast may represent a way to transfer redox information. Functional genomics approaches indicate that disruption of the functional cell-wall bound GGT1 isoform results in a constitutive alert response where anti-oxidative enzymes are up-regulated, probably as an effect of the altered plasma membrane receptors level and the redox state. With the more general aim of understanding how environmental challenges are perceived by plant cells, it seems therefore important to conclusively assign a role for extracellular GGTs and the gamma-glutamyl cycle in controlling

the redox signals generated in apoplast. To this end, further high-throughput and targeted proteomic approaches will be necessary to perform and compare under the diverse stresses as indicated in **Figure 1**.

# Acknowledgments

This work was supported by grants from University of Padova— "MURST ex-60%."

# 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.

Copyright © 2015 Masi, Trentin, Agrawal and Rakwal. 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.

,

# 2-DE proteomics analysis of drought treated seedlings of Quercus ilex supports a root active strategy for metabolic adaptation in response to water shortage

Lyudmila P. Simova-Stoilova1 †, Maria C. Romero-Rodríguez 1 †, Rosa Sánchez-Lucas <sup>1</sup> Rafael M. Navarro-Cerrillo<sup>2</sup> , J. Alberto Medina-Aunon<sup>3</sup> and Jesús V. Jorrín-Novo<sup>1</sup> \*

<sup>1</sup> Agricultural and Plant Biochemistry and Proteomics Research Group, Department of Biochemistry and Molecular Biology, University of Cordoba, Cordoba, Spain, <sup>2</sup> Department of Forestry Engineering, School of Agricultural and Forestry Engineering, University of Coìrdoba, Agrifood Campus of International Excellence, Coìrdoba, Spain, <sup>3</sup> Computational Proteomics, Proteomics Facility, Centro Nacional de Biotecnología – CSIC, Madrid, Spain

Holm oak is a dominant tree in the western Mediterranean region. Despite being well adapted to dry hot climate, drought is the main cause of mortality post-transplanting in reforestation programs. An active response to drought is critical for tree establishment and survival. Applying a gel-based proteomic approach, dynamic changes in root proteins of drought treated Quercus ilex subsp. Ballota [Desf.] Samp. seedlings were followed. Water stress was applied on 20 day-old holm oak plantlets by water limitation for a period of 10 and 20 days, each followed by 10 days of recovery. Stress was monitored by changes in water status, plant growth, and electrolyte leakage. Contrary to leaves, holm oak roots responded readily to water shortage at physiological level by growth inhibition, changes in water status and membrane stability. Root proteins were extracted using trichloroacetate/acetone/phenol protocol and separated by two-dimensional electrophoresis. Coomassie colloidal stained gel images were analyzed and spot intensity data subjected to multivariate statistical analysis. Selected consistent spots in three biological replicas, presenting significant changes under stress, were subjected to MALDI-TOF mass spectrometry (peptide mass fingerprinting and MS/MS). For protein identification, combined search was performed with MASCOT search engine over NCBInr Viridiplantae and Uniprot databases. Data are available via ProteomeXchange with identifier PXD002484. Identified proteins were classified into functional groups: metabolism, protein biosynthesis and proteolysis, defense against biotic stress, cellular protection against abiotic stress, intracellular transport. Several enzymes of the carbohydrate metabolism decreased in abundance in roots under drought stress while some related to ATP synthesis and secondary metabolism increased. Results point at active metabolic adjustment and mobilization of the defense system in roots to actively counteract drought stress.

Keywords: holm oak, roots, drought, recovery, proteomics

### Edited by:

Subhra Chakraborty, National Institute of Plant Genome Research, India

### Reviewed by:

Ján A. Miernyk, University of Missouri, USA Sabine Lüthje, University of Hamburg, Germany Sixue Chen, University of Florida, USA

### \*Correspondence:

Jesús V. Jorrín-Novo, Agricultural and Plant Biochemistry and Proteomics Research Group, Department of Biochemistry and Molecular Biology, University of Cordoba, Agrifood Campus of International Excellence, Campus de Rabanales, Ed. Severo Ochoa, Planta baja, 14071 Cordoba, Spain bf1jonoj@uco.es

### †Present Address:

Lyudmila P. Simova-Stoilova, Plant Molecular Biology Department, Institute of Plant Physiology and Genetics, Bulgarian Academy of Sciences, Sofia, Bulgaria; Maria C. Romero-Rodríguez, Centro Multidisciplinario de Investigaciones Tecnológicas, Universidad Nacional de Asunción, San Lorenzo, Paraguay

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 17 March 2015 Accepted: 29 July 2015 Published: 14 August 2015

# Introduction

Forest trees are of enormous ecological and economic value in global and local scale. They will be among the species most harmfully affected by the predicted climate changes with frequent temperature and precipitation extremes; however for a number of reasons our knowledge of the biochemical mechanisms to counteract inevitable environmental stresses is still very limited, especially concerning trees (Plomion et al., 2006; Abril et al., 2011). Quercus ilex is a dominant tree in western Mediterranean region, one of the main plant species of the so-called savannahtype woodland ecosystems (dehesas) which cover more than 4 million ha in western Mediterranean and northern African countries (David et al., 2007; Corcobado et al., 2013). Holm oak has double importance both from ecological and economic points of view, the latter coming from the fact that its acorns are major constituents in the diet of free range domestic animals; the rich nutrient composition and tannins of acorns give the original and specific taste of local meat products (David et al., 2007). According to climate model simulations, the Mediterranean region is expected to be a hot-spot which will be particularly affected by long term drought and warming episodes (Giorgi and Lionello, 2008).

Holm oak is very well adapted to dry hot climate, due to morphological particularities like a deep and well-structured root system with relatively large surface area and rapid development, which allows efficient capture of water from deepest soil layers, and small evergreen sclerophylous leaves with minimal transpiration and economical water use efficiency (David et al., 2007; Tsakaldimi et al., 2009). More biomass in this species is allocated in roots, forming larger below-ground starch and lipid reserves (Sanz-Pérez et al., 2009). Physiologically, holm oak is considered to be a typical water-spender species which maintains low leaf water potential. Stomatal closure upon water stress does not inhibit carbon assimilation in holm oak contrary to plants with water-saving strategy like pines (Baquedano and Castillo, 2006). It is predicted that the natural habitats of Quercus ilex will not be greatly affected or will even be expanded in the context of future climate changes (David et al., 2007; Bussotti et al., 2014). The overall good adaptation potential and huge economic importance of holm oak make it principal and valuable species for Mediterranean reforestation programs, which fostered research on Quercus ilex variability and stress response, especially at the proteome level (Jorge et al., 2006; Echevarría-Zomeño et al., 2009; Valero Galván et al., 2011, 2013). Main limitation in reforestation practices is the very low field survival rate of planted seedlings. Early seedling establishment is an extremely vulnerable phase of plant development and drought is considered to be the main cause of mortality post-transplanting (Navarro Cerrillo et al., 2005; Tsakaldimi et al., 2009). Besides, prolonged drought weakens tree defense systems against fungal pathogens. The forest decline is attributed to fungal attacks (e.g., Hypoxylon mediterraneum (De. Not.) Mill., Biscogniauxia mediterranea (De Not.) (Kuntze), or Phytophthora cinnamomi Rands.) after severe drought combined with high temperatures (Corcobado et al., 2013; Sghaier-Hammami et al., 2013). An active response to drought is critical for tree establishment and survival, however,

Drought is one of the most deleterious abiotic stresses with respect to plant survival (Wang et al., 2003). Plants counteract drought using a combination of survival strategies like drought escape (through adjustment in the developmental program and ending reproductive cycle before severe drought development), drought avoidance (maintaining the internal water balance under drought conditions mainly through morphological and physiological adjustments), and drought tolerance (coping with water limitation mainly at cellular and biochemical level), which are subordinated to the natural climatic variations (Dolferus, 2014). In the genus Quercus for example, the rapid spring growth and allocation of high quantity of carbon reserves into roots could be regarded as elements of drought avoidance strategy—to develop robust root system and gain access to water in deeper soil layers (Sanz-Pérez et al., 2009). Survival under prolonged drought is more linked to tolerance strategies at the cellular level. Among them should be mentioned: synthesis of compatible solutes, enhanced protection against oxidative damage, protection of membranes, and proteins from denaturation through dehydrins and chaperones, degradation of unnecessary proteins and reusing the building blocks, and others (Wang et al., 2003; Dolferus, 2014). Roots are the plant part directly and primarily exposed to soil drought, so roots play a primordial role in water stress sensing and response. Besides the general anchoring and supportive function, root as a tissue provide water and mineral nutrients to all plant parts, and produce hormones for stress signaling (Ghosh and Xu, 2014). Root functions are indeed impeded upon severe water stress. Root tissue is very actively engaged in the adaptation to drought; for example water scarcity could stimulate root growth contrary to the aboveground plant part where growth is usually inhibited (Yamaguchi and Sharp, 2010). Contrasting metabolic changes have been reported in roots compared to shoots under drought stress—shoots have been metabolically inactivated and have had lower concentrations of sugars, amino acids, nucleosides, whereas roots have been metabolically activated, and more primary metabolites have been allocated to/synthesized in roots under water stress (Gargallo-Garriga et al., 2014). Increasing number of studies is accumulating now concerning root drought response at physiological, molecular biology, biochemical and proteomic levels (reviewed in Ghosh and Xu, 2014). As proteins are the main and direct executors of cellular functions, more pronounced impact of proteomic studies is recently observed in stress research on roots (Sengupta et al., 2011; Swigonska and Weidner, 2013; Oh and Komatsu, 2015). Proteomics along with metabolomics could be very useful tools in the so-called next-generation phenotyping methods of screening for stress tolerance (Dolferus, 2014). Proteomic studies on Quercus face the common problem with orphan species with non-sequenced genome which could be solved by homology-driven cross-species identifications based on high similarity to proteins from other plant species (Valero Galván et al., 2011). Preliminary studies of drought response in 1-year old plantlets of Quercus ilex using a proteomics approach have been performed in our group, analysing the changes in the leaf protein profile (Jorge et al., 2006; Echevarría-Zomeño et al., 2009).

The aim of the present study was to follow the dynamic changes in root proteins of drought treated Quercus ilex seedlings at very early developmental stage, applying a gel-based proteomic approach, and to relate proteome changes to stress severity and to recovery from stress, thus, to build a picture on the functional meaning of the observed protein changes.

# Material and Methods

### Plant Material and Stress Treatment

Holm oak (Quercus ilex subsp. Ballota [Desf.] Samp.) acorns were harvested on place in December 2011 from a region near Cerro Muriano (20 km apart from Cordoba). Mature fruits were dropped down directly from the trees, minimizing contact with soil. Acorns were immediately transported to the laboratory in tightly closed plastic bags, washed, surface-sterilized for 10 min in 10% sodium hypochlorite solution, extensively washed again, and inspected for worm damage. Healthy acorns were blotted dry, put in clean plastic bags and stored at 4–8◦C. Spontaneously germinated acorns after 3 months of storage in cold were used in this study. Plants were grown in individual containers (330 ml volume) in perlite at growth chamber conditions: 12/12 h photoperiod, fixed light (360 µE.m−2. s −1 ), 24/20◦C day/night temperature and 70% air humidity, at optimal water supply (120 ml tap water per 30 g dry perlite). Water stress was applied on 20 days-old holm oak seedlings with developed 9 ± 3 leaves on randomly selected sets of 12 plants, by carefully placing plants into new containers filled with perlite which was wetted with limited water quantity (40 ml water per 30 g dry perlite). Water limitation treatment was for a period of 10 and 20 days, followed by a 10 days recovery phase from stress by restoring the optimal water supply. Sufficient or limited water quantity was maintained by gravimetric measurements and daily compensation for the loss of water due to evapotranspiration. Sampling was performed at the appropriate time points—roots from drought treated plants for 10 (D10) and 20 (D20) days, and after 10 days recovery from 10 or 20-days water limitation period (R10, R20), with the respective age controls (C0—at the treatment beginning; C10 for D10, C20—for R10 and D20, and C30—for R20). Water status and electrolyte leakage were monitored on fresh plant material. For proteomic analyses, roots were quickly but thoroughly rinsed with distilled water, blotted dry, quick-frozen in liquid nitrogen, ground to fine powder and stored at −80◦C.

## Stress Estimation Parameters—Plant Growth, Water Status, Electrolyte Leakage

Stress was monitored by changes in water status, growth, and electrolyte leakage. These parameters were assessed as previously described (Simova-Stoilova et al., 2008). Time course changes were followed in the water content of roots and leaves along with observations for any visible changes in plants. Biomass reduction and recovery growth were registered gravimetrically. Root and shoot fresh weight was taken at sampling from all the sets of plants (n = 12). Water content was measured in 3 individual plants per treatment, calculated according to the formula (FW-DW)/FW where FW is fresh weight, DW—dry weight of the same sample by drying it at 70◦C to constant weight for 48 h, and was expressed in percentage. Water deficit in roots and leaves was estimated as (TW-FW)/TW where TW is the turgid weight after floating the tissues for 4 h at room temperature in deionized water, and expressed in percentage. Electrolyte leakage was estimated by measuring the conductivity of the effusate solution from intact tissue, kept for 4 h at room temperature in deionized water, relative to conductivity of the effusate from the same tissue after boiling it for 10 min and cooling down.

## Protein Extraction and 2-DE Separation

Root samples after 10 and 20 days of drought (D10 and D20) with the respective age controls (C10, C20) and 10-days recovery after 10 days of drought (R10) were analyzed by gel-based proteomics. Proteins were extracted from 3 biological replicas (each replica from 3 individual plants) according to the protocol of Wang et al. (2006) using TCA/acetone-phenol-methanol. Protein content in samples was estimated by the method of Bradford (1976) with bovine serum albumin as a standard. Samples were isoelectrofocused in the range of pI 5–8 using a Protean IEF Cell system (Bio-Rad, Hercules, CA, USA), 17 cm IPG strips, at 400µg protein load per strip, active rehydration for 16 h at 50 V, rapid voltage ramp to 10,000 V, 50,000 Volt-hours in total, 500 V maintaining focused state. The second dimension was run at 12% SDS (PROTEAN <sup>R</sup> Plus Dodeca Cell, Bio-Rad, Hercules, CA, USA) and gels were double stained with colloidal Coomassie. Broad range molecular weight standards (Bio-Rad, Hercules, CA, USA) run by side in the same gel were used for estimation of MW.

### Image Analysis and Selection of Spots of Interest

Images of the gels were captured with a GS-800 densitometer (Bio-Rad, Hercules, CA, USA) and analyzed applying PDQuest software (Bio-Rad, Hercules, CA, USA). Ten-fold over background criterion was used to assess presence/absence of a spot. Data on normalized spot volumes were exported and subjected to multivariate statistical analysis including sample clustering, ANOVA and Principal component analysis using a free online-based software (NIA arrays analysis tools, http:// lgsun.grc.nia.nih.gov/ANOVA/index.html). Prior to statistical analysis the normalized values were log transformed to reduce dependence between abundance and standard deviation. Data were statistically analyzed by ANOVA using the following settings: error model max (average, actual), 0.01 proportions of highest variance values to be removed before variance averaging, 10◦ of freedom for the Bayesian error model, 0.05 FDR threshold, and zero permutations. The principal component analysis (PCA) settings were: covariance matrix type, 4 principal components, 1.5–fold change threshold for clusters, and 0.5 correlation threshold for clusters. PCA results were represented as a biplot, with consistent proteins in those experimental situations located in the same area of the graph. Selected variable spots of interest (90 in total), well defined and presenting statistically significant changes (drought and/or recovery compared to the respective age controls), with at least 1.5-fold difference

in abundance ratio, were manually cut for subsequent MS analysis.

# Mass Spectrometry Analysis, Protein Identification, and Functional Annotation

The MALDI-TOF/TOF Mass Spectrometry and Protein identification analysis was carried out in the UCO-SCAI proteomics facility, a member of Carlos III Networked Proteomics Platform, ProteoRed-ISCIII. The excised spots of interest were digested with porcine trypsin (sequencing grade) and loaded onto a MALDI plate, by using a ProPrep II station (Digilab Genomic Solutions Inc., Cambridgeshire, UK). The gel specimens were destained twice over 30 min at 37◦C with 200 mM ammonium bicarbonate/40% acetonitrile. Gel pieces were then subjected to three consecutive dehydratation/rehydratation cycles with pure acetonitrile and 25 mM ammonium bicarbonate in 50% acetonitrile, respectively, and finally dehydrated for 5 min with pure acetonitrile and dried out over 4 h at room temperature. Then, 20µl trypsin, at a concentration of 12.5 ng/µl in 25 mM ammonium bicarbonate was added to the dry gel pieces and the digestion proceeded at 37◦C for 12 h. Peptides were extracted from gel plugs by adding 1µl of 10% (v/v) trifluoracetic acid (TFA) and incubating for 15 min. Then, extracted peptides were desalted and concentrated by using µC-18 ZipTip columns (Millipore, Billerica, MA, USA) and were directly loaded onto the MALDI plate using α-cyano hydroxycinnamic acid as a matrix. Mass analysis of peptides (MS) of each sample was performed with a MALDI-TOF/TOF 4800 Proteomics Analyzer (Applied Biosystems, Foster City, CA, USA) mass spectrometer in the m/z range 800–4000, with an accelerating voltage of 20 kV. Spectra were internally calibrated with peptides from trypsin autolysis (M+H<sup>+</sup> = 842.509, M+H<sup>+</sup> = 2211.104). The most abundant peptide ions were then subjected to fragmentation analysis (MS/MS), providing information that can be used to determine the peptide sequence. Proteins were assigned identification by peptide mass fingerprinting and confirmed by MS/MS analysis. Mascot 2.0 search engine (Matrix Science Ltd., London, UK; http://www.matrixscience.com) was used for protein identification running on GPS ExplorerTM software v3.5 (Applied Biosystems, Foster City, CA, USA) over non-redundant NCBI protein and Uniprot databases. The following parameters were allowed: taxonomy restrictions to Viridiplantae, one missed cleavage, 100 ppm mass tolerance in MS and 0.5 Da for MS/MS data, cysteine carbamidomethylation as a fixed modification, and methionine oxidation as a variable modification. The confidence in the peptide mass fingerprinting matches (p < 0.05) was based on the MOWSE score, and confirmed by the accurate overlapping of the matched peptides with the major peaks of the mass spectrum. Proteins with statistically significant (p < 0.05) hits were positively assigned identification after considering Mr and pI values. Annotation of their biological function was consistent with Bevan et al. (1998). To predict the most probable intracellular localization of proteins, results from different software were compared— WolfPSORT (Horton et al., 2007, http://www.genscript.com/ psort/wolf\_psort.html), TargetP (Emanuelsson et al., 2000, http://www.cbs.dtu.dk/services/TargetP/), PredSL (http://aias. biol.uoa.gr/PredSL/input.html, Petsalaki et al., 2006), MultiLoc (http://abi.inf.uni-tuebingen.de/Services/MultiLoc/, Höglund et al., 2006). To construct a heat map the identified protein intensity, values were log10 transformed (base-10 logarithm of 1 + mean intensity values) and mean-centered to rescale them. Hierarchical clustering of samples and proteins were performed using MeV4.8 (http://www.tm4.org/mev.html) with agglomeration method set to average and the distances were calculated based on Pearson's correlation.

# Data Submission

Starting from each individual search (Mascot.dat file), the identified spots were translated to PRIDE XML using the PRIDE Converter 2.0 software (Côté et al., 2012). A total of 90 PRIDE XML files together with the corresponding raw MS files, mzXML peak lists and the Mascot search results (.dat file) were submitted to the ProteomeXchange repository (Vizcaíno et al., 2014) following the ProteomeXchange submission guidelines. Data are available with identifier PXD002484.

# Results

# Holm Oak Morphological and Physiological Response to Drought

In this study water stress was imposed on 20 days-old holm oak plantlets with well-developed root system consisting of small white tip part and larger brown lignified part, and well developed 9 ± 3 leaves. At this time the tissue linking cotyledon and plantlet was practically desiccated, and even lost somewhere, thus, cotyledons were not considered as influencing the plantlet. Water limitation was maintained for a period of 10 and 20 days, each followed by 10 days recovery. Appropriate age controls were used for comparison of stress treatments and recovery— C20 was for R10 and D20, C30 was for R20. The experimental design is presented schematically in Supplementary Material Figure S1. Leaf number was not significantly increased during the whole experimental period both in control and in treated plants. Stressed plants had longer and thinner roots with diminished white root part, which was more evident after 20 days of treatment and was not restored in recovery from the longer stress. Photos of control, stressed and recovered plants are presented in Supplementary Material Figure S2. Stress development was monitored by changes in plant growth parameters (**Figure 1**, dark columns 'part—root FW, white columns' part—shoot FW), in water status [relative water content **Figure 2**—separate columns for roots (dark) and leaves (white columns), and water deficit— **Table 1** left part], and in electrolyte leakage (**Table 1** right part) as an indicator of membrane damage. Biomass was significantly reduced only after 20 days of stress, more in roots (by 37.5%) than in shoots (by 20.9%). Water content in control plants was in the limits: root white part—87–78% (slightly diminishing with plant age), leaves—38–45%. Drought stress induced change in relative water content only in roots (diminution by about 10–15%), not in leaves. In roots, this parameter did not decrease further with stress prolongation and was completely restored in recovery. Water deficit increased only in roots after drought treatment, along with increase in membrane instability. A 2-fold rising in relative electrolyte leakage (EL) was registered in roots under 10

days of drought stress. In leaves EL remained very low, possibly linked to establishment of xeromorphic leaf structure. Based on these data, an active strategy for metabolic adaptation to drought is expected to be found in root tissue at the proteome level.

### Quercus Ilex Root Proteome Changes after Drought and Recovery

The protein extraction with modified TCA/acetone/phenol protocol resulted in protein yield of about 370–740µg protein per g FW of root tissue (**Table 2**). Protein extraction was also made of samples recovered from D20, but in R20 the quantity of extracted protein dropped substantially. Five variants of samples were analyzed applying gel based proteomics—C10, D10, R10, C20, D20, in biological triplicates. Due to the insufficient protein yield, R20 was not studied. The 2-DE gel images of each of the variants (pI 5–8, and 12% SDS PAGE), 400 µg protein load, Coomassie colloidal staining) are shown in Figure S3, Supplementary. Images were analyzed with PDQuest software. Approximately 359 ± 9 consistent protein spots were clearly resolved on the gels. Concerning variability in abundance on the basis of spot volume ratio (treated to age control variants), relatively more spots were found to be decreased in abundance than increased, and more variability was found in recovery compared to drought treatment (**Table 2**). Sample clustering and PC analysis data (**Figure 3**) clearly separated the five sample variants – C10, D10, R10, C20, and D20.

## Identification of Differently Abundant Spots and Patterns of Protein Changes

Selected well defined spots, presenting significant changes under stress (at least 1.5-fold change compared to controls) and consistent in the three biological replicas, were subjected to

FIGURE 2 | Relative water content. Separate columns for roots (dark) and leaves (white). Mean values (n = 3) are given. Vertical bars—standard deviations. C-control plants, C0, C10, C20, C30—the respective age controls (of days of treatment). D—plants subjected to water limitation treatment for 10 days (D10) or 20 days (D20). R—recovery by resuming optimal water supply for 10 days after 10 days (R10) or 20 days (R20) of water stress. Different letters above columns denote statistically significant differences.



Mean ± standard deviations from three independent replicates are shown; n.s.-nonsignificant changes. C0-control at the beginning of the treatment, D10—water limitation treatment for 10 days, C10—the respective age control.

MALDI-TOF PMF and MS/MS. For protein identification, combined search was performed with MASCOT search engine over NCBInr Viridiplantae and Uniprot databases. Detailed tables with identified protein species and their most probable subcellular location are presented in the Supplementary Material (Supplementary Tables S1, S2). There were five cases of low score unreliable identification and three cases of 2–3 different proteins identified in the same spot, which were excluded from these tables. In spite of the fact that Quercus ilex is an orphan species and identification was mainly based on homology, the majority of reliable hits for one given spot were for the same protein in different plant species. A few of the proteins were detected in more than one spot, like actin 2, enolase (EC 4.2.1.11), betaine aldehyde dehydrogenase (EC 1.2.1.8), cysteine synthase (EC 2.5.1.47), pyruvate decarboxylase (EC 4.1.1.1), but with some exceptions they presented similar trends of changes. Differences in protein abundance are clearly distinguished in the heat map built on the basis of normalized spot abundancy ratios of treatment to the respective age control (**Figure 4**). Different patterns of changes were found when looking at the identification



Variable spots (>2-fold/>1.5-fold change) are related to the respective age controls. C-control plants, C10, C20—the respective age controls (of days of treatment). D—plants subjected to water limitation treatment for 10 days (D10) or 20 days (D20). R–recovery by resuming optimal water supply for 10 days after 10 days of drought (R10), the age control in this case is C20.

results; however, the majority of spot abundance changes presented the same trend at drought stress irrespective treatment duration—D10 or D20. The following types of dynamic changes were observed—changes in abundance under stress unrelated to how long was the treatment (43 protein spots—24 up and 19 down), increase or decrease in abundance only at D10 (14 protein spots—8 up and 6 down in abundance), decrease only at D20 (2 protein spots), spot abundance changes in different directions comparing D10 and D20 (4 protein spots), prominent abundance changes in recovery, mainly diminution (8 spots), and changes beyond detection level, so-called qualitative changes (11 protein spots). Among the identified proteins with earlier and reversible increase in abundance under drought followed by recovery were found many enzymes related to secondary metabolism such as: caffeoyl CoA 3-O-methyl transferase (EC 2.1.1.104), chalcone synthase (2 spots, EC 2.3.1.74), shikimate dehydrogenase (EC 1.1.1.25), quinone oxidoreductase (EC 1.6.5.5), as well as ubiquitin activating enzyme E1 (EC 6.3.2.19), DEAD box RNA helicase (EC 3.6.4.13), betaine aldehyde dehydrogenase (EC 1.2.1.8). Down-accumulated in D10 were some inducible and hormone responsive proteins, glycyl-tRNA synthetase (EC 6.1.1.14), D-3-phosphoglycerate dehydrogenase (L-serine biosynthesis, (EC 1.1.1.95). There were not up-accumulated proteins typical only for prolonged drought treatment but 2 spots

### FIGURE 4 | Heat map built on the basis of normalized spot abundancy ratios—treatment to the respective age control. D—plants subjected to water limitation treatment for 10 days (D10) or

20 days (D20). R—recovery by resuming optimal water supply for 10 days after 10 days of drought (R10); C10, C20—control plants (the respective age controls of days of treatment).

presented decrease in abundancy only at D20—methylmalonatesemialdehyde dehydrogenase (EC 1.2.1.27), and methylene tetrahydrofolate reductase (EC 1.5.1.20), a cytoplasmic enzyme member of one-carbon metabolism. A few proteins had opposite trends of changes comparing D10 and D20—peroxidase (EC 1.11.1.7)—D10 decrease, D20 increase, S-adenosylmethionine synthase (one-carbon metabolism, 2 spots, (EC 2.5.1.6)— D10 increase, D20 decrease, and the mitochondrial NADHubiquinone oxidoreductase (EC 1.6.5.3)—D10 increase, D20 decrease. The following proteins presented prominent changes in abundance after recovery, mainly diminution: Pyruvate dehydrogenase E1 (EC 1.2.4.1) subunit beta, component 3 of pyruvate dehydrogenase complex, enolase (EC 4.2.1.11), protein disulfide-isomerase (EC 5.3.4.1), STI like HSP (2 spots). The only protein spot with increase in recovery was proteasome subunit beta (EC 3.4.25.1). The most interesting qualitative differences were in one cysteine synthase isoform (EC 2.5.1.47), which spot was below detection under drought; however, there was another spot identified also as cysteine synthase with increased abundance under drought. Identified proteins were classified into functional groups according to Bevan et al. (1998): primary and secondary metabolism, protein biosynthesis and proteolysis, defense against biotic stress, cellular protection against abiotic stress, intracellular transport. In **Table 3** are summarized the main functional groups and subgroups and the tendencies of changes in individual proteins within groups. It is seen that several enzymes of the carbohydrate metabolism were decreased in abundance in roots under drought stress while some related to ATP synthesis and secondary metabolism were increased. Results point at active metabolic adjustment and mobilization of the defense system in roots to actively counteract stress. A summary diagram of molecular mechanisms involved in Quercus ilex root response to water limitation is presented in **Figure 5**.

# Discussion

## Physiological Changes in Quercus ilex Roots Under Water Limitation

Holm oak is regarded as relatively drought tolerant plant species, well adapted to Mediterranean type of climate. Its tolerance is due to the large well developed root system, the evergreen sclerophylous leaf anatomy, the very economic use of water resources, as well as some particularities in photosynthesis (David et al., 2007; Tsakaldimi et al., 2009). However, seedling establishment is one of the most stress vulnerable phases in plant development (Tsakaldimi et al., 2009). In our experimental system, much more plant biomass is allocated in roots than in shoots (2.5 to 3-fold more) at early seedling stage. Significant growth inhibition was detected after relatively long period of water limitation (20 days), both for roots and for shoots, which confirms the expected drought resilience of this tree species. Some reports emphasize the active root growth vs. shoot growth inhibition as an adaptive strategy for drought adaptation, which is reflected in changes in root to shoot biomass ratio (Yamaguchi and Sharp, 2010), while in other cases root growth is inhibited in response to progressive water stress (Sengupta et al., 2011), probably linked to stress duration and species peculiarities in stress tolerance. In the case of Quercus ilex seedlings, however, the total root biomass was more negatively affected by prolonged drought than the shoot biomass was, resulting in diminution in root to shoot ratio compared to the respective age controls. In the time course of the experiment, the root to shoot ratio in the age controls was constantly increasing, which supports the more active root growth compared to shoots. Besides root biomass diminution, prolonged drought leaded to changes in the root aspect—longer and thinner roots with less white tip part at D20, which may be linked to increased lignification as a response to water deficit. Increased degree of lignification in the basal part of root elongation zone has been reported in response to drought for other plant species (Yamaguchi et al., 2010). The difference in leaf and root response to water shortage was further supported by the observed changes in water status and membrane stability after drought stress, expressed mainly for the root system. Pronounced change in water status of roots subjected to drought stress, compared to leaves, is also documented for other plant species (Yoshimura et al., 2008; Wendelboe-Nelson and Morris, 2012).

### Secondary Metabolism is Activated in Roots Under Drought Treatment

Drought adaptation of plants requires complex rearrangements of the metabolism with interactions between several metabolic pathways. One of the most striking observations in our comparative study was the relatively early increase of several cytoplasmic enzymes engaged in secondary metabolism in roots under water stress, like: shikimate dehydrogenase (EC 1.1.1.25), naringenin-chalcone synthase (EC 2.3.1.74), flavanone 3-hydroxylase (EC 1.14.11.9), dihydroflavonol 4-reductase (EC 1.1.1.219), (+)-neomenthol dehydrogenase (EC 1.1.1.208), and caffeoyl CoA 3-O-methyltransferase (EC 2.1.1.104). Shikimate dehydrogenase, an enzyme of the shikimic acid pathway leading to biosynthesis of aromatic amino acids and simple phenolics, catalyzes the reversible NADP+-dependent reaction of 3-dehydroshikimate to shikimate. Chalcone synthase (or naringenin-chalcone synthase) is a plant enzyme in the initial step and central hub for the pathway of flavonoid biosynthesis, leading to production of flavanoids, isoflavonoidtype phytoalexins, and other metabolites with stress protective functions for plants. Other enzymes of the flavonoid biosynthesis pathway which are found to be up-accumulated in concert with chalcone synthase under drought were: flavanone 3-hydroxylase which catalyzes the stereospecific conversion of flavanones to dihydroflavonols, and dihydroflavonol reductase, which catalyzes the reduction of dihydroflavonols to leucoanthocyanins (Dao et al., 2011). Chalcone synthase is induced under different abiotic and biotic stresses like UV, wounding, herbivory, and microbial pathogens, resulting in the production of compounds with antimicrobial, insecticidial, and antioxidant activity (Selmar and Kleinwächter, 2013). Flavonoids interfere with hormone signaling by inhibiting polar auxin transport (Dao et al., 2011). Increasing evidence suggests that plants exposed to drought accumulate secondary metabolites, and a plausible explanation could be to protect cells from oxidative stress by consuming NADPH<sup>+</sup> H<sup>+</sup> for the synthesis of highly reduced precursors TABLE 3 | Main functional groups and subgroups of identified proteins and the tendencies of changes in individual proteins within groups.





Spot volume ratio changes of variants: D—plants subjected to water limitation treatment for 10 days (D10) or 20 days (D20). R—recovery for 10 days after 10 days of drought (R10); C10, C20—control plants (the respective age controls of days of treatment). Statistically significant differences between control and treated variants or between age controls (p < 0.05) are indicated by asterix.

like aromatic amino acids, monoterpens, alkaloids (Selmar and Kleinwächter, 2013). Increased abundance under drought of the enzyme (+)-neomenthol dehydrogenase, which participates in monoterpenoid biosynthesis, observed in this study, could be linked to possible protective function of monoterpens. Caffeoyl-CoA 3-O-methyltransferase is engaged in the pathway of lignin biosynthesis. The accumulation of this enzyme under drought could be related to increased lignification of the cell wall—a modification in order to avoid water loss. Similar up regulation of Caffeoyl-CoA 3-O-methyltransferase in roots subjected to water stress is reported by several authors (Alam et al., 2010; Yamaguchi et al., 2010; Fulda et al., 2011). In concert with the changes in root growth (longer and thinner roots with less biomass) we observed an increase in abundance of actin 2 (component of microphilaments) as well as early decrease and late increase in peroxidase abundance—a cell wall cross-linking enzyme participating in cell wall lignification, defense against pathogen attack, and activated oxygen consumer. Increased content of peroxidase III in roots of wild watermelon under drought has been reported (Yoshimura et al., 2008).

# Carbon Metabolism and Energy Production—Glycolysis is Down-regulated in Roots in Water Deficit Conditions While ATP Synthesis is Stimulated

In this study we observed concerted decrease in abundance of glycolytic enzymes—glucose-6-phosphate isomerase (EC 5.3.1.9), pyrophosphate-dependent phosphofructokinase (EC 2.7.1.90), 2,3-bisphospho glycerate-independent phosphoglycerate mutase (EC 5.4.2.12), as well as of the enzyme pyruvate dehydrogenase (EC 1.2.4.1)—mitochondrial enzyme which links the glycolysis metabolic pathway to the citric acid cycle. Decrease in the amount of the cytoplasmic aconitate hydratase (EC 4.2.1.3), an enzyme of the glyoxylate bypass in plants for utilization of fatty acids as a carbon source, was detected. As for enzymes of the tricarboxylic acid cycle, an increase in abundance of isocitrate dehydrogenase which catalyzes the rate-limiting step of the cycle, was detected (EC 1.1.1.42). Two protein spots related to ATP production were found to be increased in abundance under drought—ATP synthase (EC 3.6.3.14) subunit beta and NADH-ubiquinone oxidoreductase (EC 1.6.5.3). The general down-regulation of carbohydrate degrading glycolytic enzymes could be linked to reduced root biomass accumulation, and could be regarded as a mechanism to accumulate and store sugars for rapid growth in recovery. Similar decrease of glycolysis-related enzymes in roots under drought stress is reported for soybean (Alam et al., 2010). In the roots of other species—the xerophyte wild watermelon, an up-regulation of glycolysis and tricarboxylic acid cycle was found (Yoshimura et al., 2008). Tricarboxylic acid cycle is embedded into a larger metabolic network, constantly sharing substrates and products with other pathways (Sweetlove et al., 2010). Besides carbohydrates, the TCA cycle may be fuelled by products derived from protein and other macromolecules degradation, in order to produce sufficient ATP to meet the energetic needs under stress. The up-regulation of ATP-synthesis related enzymes could be explained by the need of energy for stress protection and maintaining tissue functional state under water limiting conditions. ATP energy is necessary for many cellular processes, including secondary metabolism and protein quality control.

## Dynamic Changes Are Observed in Enzymes Related to Amino Acid and One Carbon Metabolism

We have found increased abundance under drought of some enzymes related to one-carbon and amino acid metabolism—S-adenosyl methionine synthase (EC 2.5.1.6) and formate dehydrogenase (EC 1.2.1.2); glutamine synthetase (EC 6.3.1.2) and cysteine synthase (E.C.2.5.1.47). Decrease in content was detected after prolonged drought for methylene tetrahydrofolate reductase (EC 1.5.1.20), a cytoplasmic enzyme member of one-carbon metabolism. The most interesting qualitative differences were in one cysteine synthase isoform which spot was below detection under drought; however, there was another spot identified also as cysteine synthase with increased abundance under drought. The amino acid cysteine is incorporated into proteins and glutathione (GSH); moreover, it is considered to be the bottleneck for GSH production. The cysteine synthase complex is considered to be the rate-limiting step of cysteine biosynthesis (Chan et al., 2013). Besides, cysteine acts as sulfur donor for methionine (Met) for S-adenosylmethionine and S-methylmethionine synthesis (Ravanel et al., 1998). Glutamine synthetase catalyzes the condensation of glutamate and ammonia to form glutamine, thus playing an essential role in nitrogen metabolism and ammonia assimilation. Ammonia is produced by nitrate reduction or amino acid degradation; on the other hand the amide group of glutamate serves as a readily mobilized nitrogen source for incorporation of amino group in various metabolites. Glutamine synthetase isoforms are reported to be highly responsive to drought (Yoshimura et al., 2008; Alam et al., 2010; Singh and Ghosh, 2013). S-adenosyl methionine synthase catalyzes the conversion, at the expense of ATP, of L-methionine into S-adenosylmethionine—AdoMet or SAM, the major methyl donor for proteins, nucleic acids, carbohydrates, lipids, and small molecules for lignin and many other biosynthesis, precursor for polyamine ant ethylene biosynthesis. Different trends of change in the enzymes related to one-carbon metabolism were reported in drought-stressed roots from different plant species and stress treatment (Yoshimura et al., 2008; Mohammadi et al., 2012; Grebosz et al., 2014). In our study, the accumulation of S-adenosyl methionine synthase and related enzymes in holm oak roots under drought could be linked to enhanced secondary metabolism and lignification which utilize activated one carbon particles.

# Protein Synthesis, Folding/processing and Degradation Processes Are Highly Responsive to the Applied Stress

Changes in the protein complement of cells are indispensable for adaptation to stress. Within the large group of proteins related to protein synthesis, folding/processing and degradation, some spots with opposite trends of changes under stress were found, which may reflect the dynamic changes in cell protein profile. A DEAD box RNA helicase was detected among the identified proteins with earlier and reversible increase in abundance under drought. RNA helicases are involved in various aspects of RNA metabolism, including nuclear transcription, pre mRNA splicing, ribosome biogenesis, nucleocytoplasmic transport, translation, RNA decay, and organellar gene expression (Aubourg et al., 1999). Similar increase of DEAD box RNA helicase was reported in osmotically stressed triticosecale roots (Grebosz et al., 2014). On the other hand, glycyl-tRNA synthetase abundancy was found to decrease under drought and in recovery which is not in favor of up-regulation of translation under drought stress and may rather reflect changes in the composition of newly synthesized proteins. Chaperones and proteases are elements of the protein quality control machinery. Maintaining proteins in their functional conformation, preventing aggregation of nonnative proteins, refolding of denatured proteins and removal of non-functional and potentially harmful polypeptides are very important for cell survival under drought stress (Vaseva et al., 2012). Some molecular chaperones, which assist protein folding or renaturation, presented complex changes: chaperonin 60 kD, protein disulfide isomerase and chaperone of TCP-1 family decreased in abundance, while GroES chaperonin increased following water stress. TCP-1 family chaperones are related to cpn60/groEL chaperonin family and assist the folding of cytoskeletal proteins in the cytoplasm (Wang et al., 2004). Cpn60 is homologous to E. Coli GroEL found in chloroplasts and mitochondria. It acts in cooperation with GroES chaperonin in assisting proper folding of newly synthesized and membranetranslocated proteins into their native conformation. The observed different abundance changes in GroEL/GroES may reflect changes in composition within the chaperonin family. Under stress the folding capacity of cpn60 is suppressed while its binding affinity toward unfolded proteins is increased, thus protecting proteins from unfavorable conditions by sequestration (Vaseva et al., 2012). Protein disulfide isomerase is an enzyme located in the endoplasmic reticulum with the main function—correct arrangement of disulfide bonds in proteins. Decrease in abundance in this enzyme could reflect inhibition of protein synthesis at the endoplasmic reticulum. The proteasome plays a crucial role in the turnover of regulatory proteins, cellular house-keeping and stress tolerance (Kurepa and Smalle, 2008). Relatively early increase in abundance under stress was found for an ubiquitin activating enzyme E1 which may reflect the importance of stress signaling. The 6B regulatory subunit of 26S proteasome, involved in ATPdependent protein degradation in the cytosol and nucleus, decreased under drought, while the catalytic proteasome subunit beta type increased. Some evidence suggests that the control of proteasome function may be at the level of subunit composition rather than total increase in proteasome abundance (Kurepa and Smalle, 2008). The proteasome could function in two forms—26S and 20S, the latter containing only the catalytic core subunits and operating without ATP in degradation of oxidized proteins. Thus, 20S may play an important role in tolerance to the secondary oxidative stress developing under drought stress (Vaseva et al., 2012). Up-regulation of the 20S proteasome subunit was found in drought-treated

# References


poplar trees (Plomion et al., 2006), alfalfa plants (Aranjuelo et al., 2011), and rapeseed roots (Mohammadi et al., 2012). Opposite changes were observed for some aminopeptidases in roots under applied stress. Leucine aminopeptidase (EC 3.4.11.1) abundancy diminished under drought and in recovery, while Xaa-pro aminopeptidase (EC 3.4.11.9) increased. Aminopeptidases liberate free aminoacids from the N-terminus of the polypeptide chains. Amino acid metabolism was found to be among the top biological processes affected by drought (Kang et al., 2011).

In conclusion, differently abundant identified protein species in holm oak roots subjected to water limitation treatment point at early activation of secondary metabolism, down-regulation of glycolysis and stimulation of ATP synthesis, accumulation of some enzymes related to aminoacid and one-carbon metabolism, and complex changes in protein synthesis, folding/processing and degradation processes, which emphasize the active metabolic adjustment and mobilization of the defense system in roots to actively counteract stress.

# Acknowledgments

Financial support of this study by MC IEF project No 271714 QUIPROD funded by the EU, 7th FP, is greatly acknowledged. Protein identification was carried out at the UCO-SCAI proteomics facility, a member of Carlos III Networked Proteomics Platform, ProteoRed-ISCIII, Cordoba, Spain. MR was supported by a grant from the "Fundación Parque Tecnológico Itaipu-Binacional" and "Centro Multidisciplinario de Investigaciones Tecnológicas—Dirección General de Investigación Científica y Tecnológica-Universidad Nacional de Asunción—Paraguay."

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00627

and water-spenders Quercus coccifera and Quercus ilex. Trees 20, 689–700. doi: 10.1007/s00468-006-0084-0


loss caused by Phytophthora cinnamomi. Agricult. Forest Meteorol. 169, 92–99. doi: 10.1016/j.agrformet.2012.09.017


**Conflict of Interest Statement:** The reviewer Sabine Lüthje declares that, despite having co-hosted the Research Topic INPPO World Congress 2014 with the author Jesús V. Jorrín-Novo, the review process was handled objectively. 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.

Citation: Simova-Stoilova LP, Romero-Rodríguez MC, Sánchez-Lucas R, Navarro-Cerrillo RM, Medina-Aunon JA and Jorrín-Novo JV (2015) 2-DE proteomics analysis of drought treated seedlings of Quercus ilex supports a root active strategy for metabolic adaptation in response to water shortage. Front. Plant Sci. 6:627. doi: 10.3389/fpls.2015.00627

Copyright © 2015 Simova-Stoilova, Romero-Rodríguez, Sánchez-Lucas, Navarro-Cerrillo, Medina-Aunon and Jorrín-Novo. 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.

# **Phosphoproteomic analysis of the response of maize leaves to drought, heat and their combination stress**

*Xiuli Hu1 †, Liuji Wu1 †, Feiyun Zhao1, Dayong Zhang2, Nana Li 1, Guohui Zhu3, Chaohao Li <sup>1</sup> and Wei Wang1 \**

*<sup>1</sup> State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou, China, <sup>2</sup> Jiangsu Academy of Agricultural Sciences Institute of Biotechnology, Nanjing, China, <sup>3</sup> Guangdong Provincial Key Laboratory of Protein Function and Regulation in Agricultural Organisms, College of Life Sciences, South China Agricultural University, Guangzhou, China*

### *Edited by:*

*Jesus V. Jorrin Novo, University of Cordoba, Spain*

### *Reviewed by:*

*Ning LI, The Hong Kong University of Science and Technology, China Natalia V. Bykova, Agriculture and Agri-Food Canada, Canada*

### *\*Correspondence:*

*Wei Wang, State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, 63 Nongye Road, Zhengzhou 450002, China wangwei@henau.edu.cn*

> *† These authors have contributed equally to this work.*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 23 January 2015 Accepted: 14 April 2015 Published: 05 May 2015*

### *Citation:*

*Hu X, Wu L, Zhao F, Zhang D, Li N, Zhu G, Li C and Wang W (2015) Phosphoproteomic analysis of the response of maize leaves to drought, heat and their combination stress. Front. Plant Sci. 6:298. doi: 10.3389/fpls.2015.00298*

Frontiers in Plant Science | www.frontiersin.org May 2015 | Volume 6 | Article 298 |

Drought and heat stress, especially their combination, greatly affect crop production. Many studies have described transcriptome, proteome and phosphoproteome changes in response of plants to drought or heat stress. However, the study about the phosphoproteomic changes in response of crops to the combination stress is scare. To understand the mechanism of maize responses to the drought and heat combination stress, phosphoproteomic analysis was performed on maize leaves by using multiplex iTRAQ-based quantitative proteomic and LC-MS/MS methods. Five-leaf-stage maize was subjected to drought, heat or their combination, and the leaves were collected. Globally, heat, drought and the combined stress significantly changed the phosphorylation levels of 172, 149, and 144 phosphopeptides, respectively. These phosphopeptides corresponded to 282 proteins. Among them, 23 only responded to the combined stress and could not be predicted from their responses to single stressors; 30 and 75 only responded to drought and heat, respectively. Notably, 19 proteins were phosphorylated on different sites in response to the single and combination stresses. Of the seven significantly enriched phosphorylation motifs identified, two were common for all stresses, two were common for heat and the combined stress, and one was specific to the combined stress. The signaling pathways in which the phosphoproteins were involved clearly differed among the three stresses. Functional characterization of the phosphoproteins and the pathways identified here could lead to new targets for the enhancement of crop stress tolerance, which will be particularly important in the face of climate change and the increasing prevalence of abiotic stressors.

**Keywords: phosphoproteins, phosphoproteomics, iTRAQ labeling, drought and heat, maize**

### **Introduction**

In the past decades, progress on increasing crop yields using semi-empiric breeding and genetics has reached a plateau, which is largely linked to increasingly adverse environmental conditions, especially drought and heat stress (Parent and Tardieu, 2014). In the field, it is often the simultaneous occurrence of several abiotic stressors, which are most lethal to crops. Heat, drought and their combination are the severer stressors for crops and are responsible for most of production losses (Lobell et al., 2011; Suzuki et al., 2014). Moreover, global climate change may increase the occurrence and distribution of these stressors, causing a further reduction of productivity (Rasul et al., 2011).

Recent studies have demonstrated that plant responses to the combinations of two or more stressors are unique and cannot be directly extrapolated from the response to a single stress. In Arabidopsis response to combination stress, some 61% of the transcriptome changes are not predictable from the response to single stress, and plants prioritized between potentially antagonistic responses for only 5–10% of the responding transcripts (Rasmussen et al., 2013). In wheat response to combination of drought and heat stress, proteomic analysis indicates that few common proteins are observed responding to single and multiple high-temperature events (Yang et al., 2011); we obtained similar results in maize (Hu et al., 2010). So, the simultaneous occurrence of several stress results in highly complex responses of plants; extrapolated the response to combined stresses is largely controlled by different, and sometimes opposing, signaling pathways that may interact with and inhibit each other (Vile et al., 2012; Suzuki et al., 2014).

Plants respond to stress with a wide range of modifications that cause to changes at the morphological, cellular, physiological, biochemical, and molecular levels (Lopes and Reynolds, 2011; Aprile et al., 2013). Overall, protein phosphorylation plays a critical role in regulating many biological functions, including stress responses by signal transduction. Phosphorylation and dephosphorylation can switch many regulatory proteins and enzymes on and off, thus control a wide range of cellular processes and signal relays (Yang et al., 2010). In recent years, global quantitative analysis of protein expression and phosphorylation has been performed using iTRAQ-based quantitative proteomic and LC-MS/MS methods (Alvarez et al., 2014; Han et al., 2014), which facilitate the simultaneous detection of changes in protein expression and phosphorylation levels under control and stressed conditions.

Large-scale phosphoproteomic analyses have been conducted on crops, especially crops response to stress. In wheat seedlings leaves response to drought stress, some phosphoproteins related to drought tolerance and osmotic regulation exhibit significant phosphorylation changes; there are commonalities and differences of phosphoproteins in different cultivars of bread wheat under drought stress (Lv et al., 2014; Zhang et al., 2014). In maize leaves response to drought stress, a total of 138 phosphopeptides display highly significant changes and most phosphorylation changes do not reflect protein abundance variation. These proteins influenced epigenetic control, gene expression, cell cycle-dependent processes and phytohormonemediated responses. Bonhomme et al. (2012). These results provide a series of phosphoproteins and phosphorylation sites and a potential network of phosphorylation signaling cascades in wheat seedling leaves. However, little information is available regarding the changes of phosphoproteins and the phosphorylation sites of many stress-responsive protein kinases under combined drought and heat stress. Therefore, characterizing posttranslational modifications of these proteins in crops response to combined stress are very important for understanding the associated signaling pathways and mechanisms for tolerance to stress.

In many regions of the world, maize is an important cereal crop grown mainly in semi-arid environments that are characterized by water scarcity and high temperature, two conditions that usually occur simultaneously in the field. Exposing plants to a combined stress may lead to agonistic or antagonistic responses or cause to some responses that are potentially unrelated to those response to the corresponding single stress. So, to analyze such responses, we performed iTRAQ-based phosphoproteomic analyses in maize exposed to heat, drought and their combination. Furthermore, bioinformatics analyses were conducted to confirm the constitutive nature of the different stress-responsive phosphorylated proteins. This work would provide a basis for further elucidating maize endurance to drought, heat and their combination stress.

# **Materials and Methods**

## **Plant Material and Stress Treatments**

Maize seeds (Zhengdan 958) were used in the experiments. Zhengdan 958 is a high-yield maize hybrid that is grown in China. The seeds were surface-sterilized for 10 min in 2% hypochlorite, washed in distilled water and germinated on moistened filter paper. Maize seedlings were grown in Hoagland's nutrient solution in a light chamber under 400µmol m−<sup>2</sup> s−<sup>1</sup> photosynthetically active radiation, a 14/10-h day/night cycle, a day/night temperature of 28/22◦C, and a relative humidity of 75%. When the fifth leaves were fully expanded, the seedlings were subjected to various treatments.

Drought stress was imposed by placing the seedlings in PEG solution (−0.7 MPa) for 8 h at 28◦C and 40% relative humidity. Heat stress was applied by raising the temperature from 28 to 42◦C at an interval of 2◦C/h and then kept at 42◦C for 1 h, for a total of 8 h. Therefore, each stress treatment lasted 8 h. The combined stress consisted of simultaneous treatment with PEG and heat stress. The control seedlings were kept at 28◦C and 75% relative humidity. Then, the expanding leaves (the fifth from the bottom) of the treated and untreated seedlings were sampled, immediately frozen in liquid N2, and stored at −80◦C until analysis. Three biological replicates were performed for each treatment.

# **Protein Extraction**

Total proteins from the fifth newly-expanded leaves of the maize seedlings were extracted according to the method reported by Wang et al. (2013) and Zhang et al. (2014). Approximately 0.5 g of fresh leaves from each biological replicate were ground into a fine power in liquid N2 in a mortal and further ground in a 4 ml SDS buffer (30% sucrose, 2% SDS, 100 mM Tris–HCl, pH 8.0, 50 mM EDTA-Na2, 20 mM DTT) and 4 ml phenol (Tris-buffered, pH 8.0) in a 10 ml tube, followed by the addition of 1 mM phenylmethanesulfonyl fluoride (PMSF) and PhosSTOP Phosphatase Inhibitor Cocktail (one tablet/10 ml; Roche, Basel, Switzerland) to inhibit protease and phosphatase activity. The mixture was thoroughly vortexed for 30 s and the phenol phase was separated by centrifugation at 14,000 × g and 4◦C for 15 min. The upper phenol phase was pipetted into fresh 10 ml tubes and 4-fold volumes of cold methanol plus 100 mM ammonium acetate were added. After centrifugation at 14,000 × g and 4◦C for 15 min, the supernatant was carefully discarded and the precipitated proteins were washed twice with cold acetone. Finally, the protein mixtures were harvested by centrifugation. Using a 2-D Quant Kit (Amersham Bioscience, America) containing bovine serum albumin (BSA) (2 mg/ml) as the standard, we carried out the measurement of protein content. To enhance the quantitative accuracy, extracted proteins from every biological replicate were adjusted to the same concentration for the subsequent analysis.

### **Protein Digestion and iTRAQ Labeling**

Protein digestion was performed according to the FASP procedure described by Wisniewki et al. (2009) and Lv et al. (2014), and the resulting peptide mixture was labeled using the 4-plex iTRAQ reagent according to the manufacturer's instructions (Applied Biosystems). Briefly, 200µg of protein from each sample was mixed with 30µl of STD buffer (4% SDS, 100 mM DTT, 150 mM Tris-HCl pH 8.0). The detergent, DTT and the other low-molecular-weight components were removed using UA buffer (8 M urea, 150 mM Tris-HCl pH 8.0) with repeated ultrafiltration (Microcon units, 30 kD). Then, 100µl of 0.05 M iodoacetamide in UA buffer was added to block reduced cysteine residues, and the samples were incubated for 20 min in darkness. The filters were washed three times with 100µl of UA buffer, then twice with 100µl of DS buffer (50 mM triethylammoniumbicarbonate at pH 8.5). Finally, the protein suspensions were digested with 2µg of trypsin (Promega) in 40µl of DS buffer overnight at 37◦C, and the resulting peptides were collected as a filtrate. The peptide content was estimated via UV absorption at 280 nm using an extinction coefficient of 1.1 per 0.1% (g/l) solution, which was calculated based on the proportion of tryptophan and tyrosine residues in vertebrate proteins.

For labeling, each iTRAQ reagent was dissolved in 70µl of ethanol and added to the respective peptide mixture. The samples were called control (under no stress), Drought, Heat and DH (combined drought and heat stress) and were labeled with reagent and vacuum dried.

# **Enrichment of Phosphorylated Peptides using TiO2 Beads**

The labeled peptides were mixed, concentrated using a vacuum concentrator and resuspended in 500µl of loading buffer (2% glutamic acid/65% ACN/2% TFA). Then, TiO2 beads were added and then the sample was agitated for 40 min. The sample was centrifuged for 1 min at 5000 g, yielding the first set of beads. The supernatant from the first centrifugation was mixed with more TiO2 beads, which were treated as before, yielding the second set of beads. Both sets of beads were combined and washed three times with 50µl of wash buffer I (30% ACN/3%TFA) and then three times with 50µl of wash buffer II (80% ACN/0.3% TFA) to remove the remaining non-adsorbed material. Finally, the phosphopeptides were eluted with 50µl of elution buffer (40% ACN/15% NH4OH), lyophilized and subjected to MS analysis.

### **Mass Spectrometry**

For nanoLC-MS/MS analysis, 5µl of the phosphopeptide solution was mixed with 15µl of 0.1% (v/v) trifluoroacetic acid, then 10µl of the mixture was injected into a Q Exactive MS (Thermo Scientific) equipped with an Easy-nLC (Proxeon Biosystems, now Thermo Scientific). The peptide mixture was loaded onto a C18 reversed phase column (15 cm long, 75µm inner diameter, RP-C18 3µm, packed in-house) in buffer A (0.1% formic acid) and separated using a linear gradient of buffer B (80% acetonitrile and 0.1% formic acid) over 240 min at a flow rate of 250 nl/min, which was controlled by IntelliFlow technology. The peptides were eluted with a gradient of 0–60% buffer B from 0 to 200 min, 60 to 100% buffer B from 200 to 216 min, and 100% buffer B from 216 to 240 min.

For MS analysis, the peptides were analyzed in positive ion mode. MS spectra were acquired using a data-dependent, top-10 method by dynamically choosing the most abundant precursor ions from the survey scan (300–1800 m/z) for HCD fragmentation. Determination of the target value was based on predictive automatic gain control (pAGC). The duration of dynamic exclusion was 40.0 s. Survey scans were acquired at a resolution of 70,000 at m/z 200 and the resolution for HCD spectra was set to 17,500 at m/z 200. The normalized collision energy was 27 eV, and the under fill ratio, which specifies the minimum percentage of the target value that is likely to be reached at the maximum fill time, was defined as 0.1%. The instrument was run with peptide recognition mode enabled.

### **Data Analysis**

MS/MS spectra were searched against the Uniprot\_Zea\_mays database (62977 sequences, downloaded June 14th, 2013) and a decoy database using Mascot 2.2 (Matrix Science), which was embedded in Proteome Discoverer 1.4. For protein identification, the following options were used: peptide mass tolerance = 20 ppm; MS/MS tolerance = 0.1 Da; enzyme = trypsin; missed cleavage = 2; fixed modification: carbamidomethyl (C); iTRAQ4plex (K); iTRAQ4plex (N-term); variable modification: oxidation (M), phosphorylation (S/T/Y). The score threshold for peptide identification was set at a 5% false discovery rate (FDR). The PhosphoRS site probability was estimated on the probability (0–100%) of each phosphorylation site being truly phosphorylated. The PhosphoRS site probabilities above 75% indicate that a site is truly phosphorylated.

### **Bioinformatics**

The molecular functions of the identified proteins were classified according to their gene ontology annotations and their biological functions. The subcellular localization of the unique proteins identified in this study was predicted using the publicly available program WolfPsort (http://wolfpsort. org). Protein-protein interaction networks were analyzed using the publicly available program STRING (http://string-db.org/). STRING is a database of known and predicted protein-protein interactions. The interactions include direct (physical) and indirect (functional) associations, and they are derived from four sources: the genomic context, high-throughput experiments, coexpression and previous knowledge. STRING quantitatively integrates the interaction data from these sources for a large number of organisms and, where applicable, transfers information between these organisms.

Motif-X online software (http://motif-x.med.harvard.edu/ motif-x.html) was used to find phosphorylation site motifs in the identified maize proteins and to predict the specificity of these motifs based on the identified phosphopeptide sequences. The parameters were set to peptide length = 21, occurrence = 5, and statistical significance for *p*-values of less than 0.000001.

### **Statistical Analysis**

The phosphoproteins assays were the mean of three replicates. Means were compared by one-way analysis of variance and Duncan's multiple range test at 5% level of significance.

### **Results**

### **Phosphopeptide Identification Under drought, Heat and Combined Stress**

Maize plants at the five-leaf stage were subjected to drought, heat and their combination stress. Multiplex iTRAQ-based quantitative proteomic and LC-MS/MS methods were performed on the total proteins extracted from the fifth newly-expanded leaves, resulting in the identification of 1367 unique phosphopeptides (**Figure 1**). These phosphopeptides corresponded to 1039 proteins and contained 2171 nonredundant phosphorylation sites, of which 1313 (60.48%) were serine (S) residues, 649 (29.89%) were threonine (T) residues and 209 (9.63%) were tyrosine (Y) residues at an estimated false discovery rate of 5%. In detail, based on a significant linear regression (*p <* 0*.*05) and a threshold of ≥1.5-fold or ≤0.66-fold change ratio of stress-induced phosphorylation levels compared with that of the control, the phosphorylation level of 172 phosphopeptides had a significant change under heat, of which 77 were up-regulated and 95 were down-regulated; the phosphorylation level of 149 phosphopeptides had a significant change under drought, of which 69 were up-regulated and 80 were down-regulated; and the phosphorylation level of 144 phosphopeptides had a significant change under the combination stress, of which 70 were up-regulated and 74 were down-regulated. The phosphopeptides with significant changes in phosphorylation levels corresponded to 282 proteins, of which 46 proteins were common under three stress conditions (**Table 1**), 69 proteins were common under heat stress and the combined stress (Table S1), 24 proteins were common under drought stress and the combined stress (Table S2), 15 proteins were common under drought and heat stress (Table S3), 75 proteins were only identified under drought stress (Table S4), 30 proteins were only identified under heat stress (Table S5), and 23 proteins were only identified under the combined stress (Table S6).

### **Heat Shock Proteins**

In this study, the phosphorylation levels of seven heat shock proteins (HSPs), including five small HSPs (sHSPs) and two HSP70s, changed significantly under drought, heat

or combined both stresses. The phosphorylation levels of three sHSPs (B4G250, BF976 B6T649) were greatly upregulated under heat and the combined stress. However, under drought, the phosphorylation levels of the three sHSPs did not change significantly. It was worth mentioning that the two phosphopeptides of B4G250 differed significantly in phosphorylation level. In contrast to the three sHSPs, the phosphorylation level of the sHSP C0P2N6 was down-regulated by heat and combined stresses but was not significantly affected by drought stress; the phosphorylation level of sHSP B4FR07 was down-regulated by drought stress. Regarding the two HSP70s, the phosphorylation level of B6SZ69 was significantly upregulated by drought stress and down-regulated by heat stress but had no significant change under the combined stress. The phosphorylation level of K7WBH2 was significantly up-regulated by drought stress, down-regulated by heat stress and had no significant change under the combined stress.

### **Responses of Kinases and Phosphatases to the Three Stresses**

It was also worth mentioning the responses of enzymes, including kinases and phosphatases, to stresses. Under heat (**Table 2**), 31 phosphopeptides, including 24 different enzymes, were identified, of which five were kinases and three were phosphatases. Under drought stress (**Table 2**), 29 phosphopeptides, including 22 unique enzymes, were identified, of which seven were kinases and five were phosphatases. Under the combined stress (**Table 2**), 29 phosphopeptides corresponding to 21 unique proteins, including five kinases


 **and DH.**



Frontiers in Plant Science | www.frontiersin.org May 2015 | Volume 6 | Article 298 | **203**

*CK, control; D, drought stress; H, heat stress; DH, combined drought and heat stress. Each ratio was the average of three replicates.*



*Indicates that these enzymes were common three stress treatments.*

and five phosphatases, were identified. In addition, fructosebisphosphate aldolase, phosphoenolpyruvate carboxykinase, phospholipase c, probable cellulose synthase catalytic subunit 2, probable protein phosphatase 2c 30-like protein, protochlorophyllide reductase b, serine-threonine protein kinase WNK4-like protein, and ubiquitin-protein ligase had significant change of phosphorylation level under the three stresses; 2-aminoethanethiol dioxygenase-like protein, calcium-dependent protein kinase, ferric-chelate reductase 1-like protein, geranylgeranyl pyrophosphate synthase 4, glycerol 3-phosphate permease, lipid phosphate phosphatase 3 and ubiquitin carboxyl-terminal hydrolase 6-like protein changed significantly in phosphorylation level under drought; and histone-lysine N-methyltransferase family protein, plasma membrane H+-transporting ATPase-like protein, and pyruvate orthophosphate dikinase changed significantly in phosphorylation level under the combined stress (**Table 2**).

To elucidate the interactions of the protein kinases/phosphatases and HSPs with other proteins under the three stress treatments, protein-protein interaction analysis of significantly changed phosphoproteins was conducted using STRING software (**Figures 2**–**4**). Under drought and the combined stress, AGC\_PKAPKG\_like.1-ACG kinases (4335426, including homologs of PKA, PKG and PKC) had interactions with transcription factor HY5 (4327123). Protein phosphatase

2C (4341433, 4332374) showed an interactions with initiation factor 2 subunit family domain containing protein (4348536). Under drought, AGC\_PKAPKG\_like.1-ACG kinases also had interactions with heat shock protein DnaJ (OsJ\_01978), initiation factor 2 subunit family domain containing protein (4348536), CBS domain-containing protein (43327739), phospholipase C (4352524) and transcription initiation factor IIF (4348228); protein phosphatase 2C also showed an interaction with heat shock protein DnaJ (OsJ\_01978). Under drought and heat stress, a dnaK family protein (HSP70, 4332413) exhibited an interaction with oxygen-evolving enhancer protein 1 (LOC\_Os01g31690.1). Under three stress treatments, phospholipase C (4352524) exhibited an interaction with an ethylene-responsive element-binding protein (4349277), and chaperone protein dnaJ 10 (sHSP40, 4346080) interacted with a MYB family transcription factor (4335542). These findings indicated that the AGC\_PKA/PKG\_like.1-ACG kinases, protein phosphatase 2C and phospholipase C may play an important role in protein substrate phosphorylation/dephosphorylation and that HSPs may play an important protective role as chaperone proteins under drought stress, heat stress and combined both stresses (maize protein query sequences corresponded to rice protein query sequences; see Supplementary Table S7 for drought stress, Table S8 for heat stress, and Table S9 for the combined stress).

## **The Characteristics of the Different Phosphorylation States of One Protein in Response to Three Stresses**

In this study, 19 phosphoproteins (**Table 3**) were found to have different phosphorylated states. Importantly, these peptides had specific phosphorylation characteristics in response to drought, heat and the combination of both stresses. In particular, phosphoenolpyruvate carboxykinase (C0P3W9) had 11 different phosphorylation sites, and the phosphorylation of sAPStPkR, sAPSTPk, sAPStPkRsAPTtPIk, and gEAAAQGAPstPR changed significantly under the three stresses; sAPttPIk, sAPTTPIk, gEAAAQGAPStPR, and eVDYADNsVTENTR changed significantly under heat and the combined stress conditions; rsAPTtPIk and sAPStPkR changed significantly under drought and the combined drought and heat stress conditions; and gGAHsPFAVAISEEER changed significantly only in response to drought stress. The phosphorylation sites of the other 18 proteins were similar to these. This result shows the diversity of the phosphorylation sites and their specificity in response to different stress treatments.

**TABLE 3 | Characteristic of several phosphopeptides belonging to one protein in response to D, H, and DH.**




*Y stands for significant changes; N stands for no significant changes. "–"indicates that the phosphorylation level was not detected.*

# **Identification of Phosphorylation Motifs in the Phosphopeptides**

In the present study, 160, 159, and 161 proteins were identified as having significantly changed phosphorylation levels under drought, heat and their combination, respectively. To determine whether the phosphorylated versions of these proteins shared common phosphorylation site motifs, Motif-X online software (http://motif-x.med.harvard.edu/motif-x.html) was used to predict the motif specificity of these proteins based on the identified phosphorylation sites. An analysis of serine (S), threonine (T) and tyrosine (Y) residues (**Table 4**) showed that in response to all the three stress treatments, some peptides had the motifs RRx**S** and x**S**Px in common, of which the residues adjacent to the phosphorylated S were enriched for arginine (R) and proline (P); In response to heat stress and the combined stress, the motifs Px**T**P and R**T** were common, of which the residues adjacent to the phosphorylated T were enriched for arginine (P) and proline (R). These results are the first to demonstrate a high sensitivity and specificity for threonine sites under heat and combined drought and heat stress.

## **The Signaling Pathways Associated with Phosphorylated Proteins Under Various Stress Conditions**

All identified phosphoproteins were classified using gene ontology (GO) annotation software and were further categorized into three functional groups: molecular function, biological process and cellular component. The results of the GO analyses for drought, heat and the combined stresses are shown in **Figures 5**–**7**, respectively. The most common molecular functions were binding activity and catalytic activity, and the most common biological processes were cellular process and metabolic process. Moreover, the most common biological processes and molecular functions that the proteins performed were predicted to occur in the organelles, the cytoplasm and the nucleus.

According to biological process analysis using the BLAST2GO program, among these phosphoproteins with significantly changed phosphorylation state, for drought stress, 16 were categorized as "response to stimulus," 12 were involved in transport and 11 were identified as DNA binding proteins that function as transcription factors (**Figure 5B**, **Table 5**); for heat stress, 12 were categorized as "response to stimulus," nine were transporter proteins, and 14 were DNA binding proteins that function as transcription factors (**Figure 6B**, **Table 5**); for the combined stress, nine were categorized as "response to stimulus," 10 were transporter proteins, and eight were DNA binding proteins that function as transcription factors (**Figure 7B**, **Table 5**). In addition, the phosphorylation states of a cell division cycle 5-like, phospholipase C, MYB-CC-type transcription factor, BEL1-type homeodomain protein, bZIP transcription factor superfamily protein, and glycine-rich protein 2 were common for the three stresses; a dnaJ 15-like chaperone protein, disease resistance protein rpp13, and rac GTPase-activating protein 1 were specific to drought; the phosphorylation states of a probable receptor-like protein kinase At5g56460 like, translocase of chloroplast chloroplastic-like isoform x1, uncharacterized LOC100501590, carbohydrate transporter sugar porter transporter, probable metal-nicotianamine transporter ysl12-like, uncharacterized LOC100501590, probable receptor-like protein kinase At5g56460-like, protein stichel-like, hypothetical protein ZEAMMB73\_938746 and an expressed methyl binding domain-containing protein were specific to heat stress. 11 proteins were specific to the combined stress (**Table 5**).

According to KEGG analysis, for drought stress, these phosphoproteins with significantly changed phosphorylation state involved mainly in the protein processing, photosynthesis and carbon metabolism pathways (**Figure 5D**). There were 11 proteins (group accessions: Q8W149, C0P9H7, B6SL90,


### **TABLE 4 | Phosphorylation motif was analyzed with significant phosphorylation sites in response to D, H, and DH.**


*"x" represents any amino acid.*

C0HIN5, B6SZ69, B4FNM4, B6T1H0, B6UIM2, B4FQU2, K7V1I2, B4G250) involved in protein processing, and seven proteins (group accessions: B4FAW3, B4FRF2, P27789, C0P3W9, E9NQE1, C0PD30, and B4FJG1) involved in photosynthesis and carbon metabolism. For heat stress, these phosphoproteins with significantly changed phosphorylation state involved mainly in signaling pathways related to protein processing, inositol phosphate-glycerophospholipid metabolism and photosynthesis (**Figure 6D**). Specifically, there were 10 proteins (group accessions: B6SZ69, BF4F976, B6T649, B4G250, BF4F7Z5, Q8W149, B4FQK5, C4J9U8, B6T1H0, and B4FQU2) involved in protein processing, three proteins (group accessions: B6U8P0, B8A1A6, and B4FY17) involved in inositol phosphateglycerophospholipid metabolism, and five proteins (group accessions: B6SQV5, B4FRF2, C0P3W9, C0PD30, B4FKM0) involved in photosynthesis and carbon metabolism. For drought and heat combined stress, these phosphoproteins with significantly changed phosphorylation state involved mainly in signaling pathways related to photosynthesis and carbon metabolism and protein processing (**Figure 7D**). In detail, there were eight proteins (group accessions: C0P3W9, B7ZYP6, C0PD30, E9NQE1, B6SKI1, B6SQV5, B7ZYP6, and B4FKM0) involved in photosynthesis and carbon metabolism, and there were eight proteins (group accessions: B6T346, K7TLV1, K7TUM2, K7V1I2, B4FQK5 B6T1H0, B6TG30, K7V792) involved in protein processing. These results indicated that the signaling pathways related to protein processing and to photosynthesis and carbon metabolism played an important role under all three stress conditions. In addition, the differentially phosphorylated proteins that were related to phosphateglycerophospholipid metabolism were primarily involved in the signaling pathways induced by heat stress.

# **Changes in Receptor Protein Phosphorylation**

Receptors are proteins that are either embedded in the plasma membrane or localized to the cytoplasm or nucleus of a cell. Receptors enable the body to detect changes in the internal or external environment. In this study, the phosphorylation levels of seven receptor proteins significantly changed under the three stresses. The phosphorylation level of a vacuolarsorting receptor 3-like protein (B6U4K3) was reduced by the three stress treatments; that of a probable receptor-like protein kinase At5g56460-like (C0PHB9) was reduced by heat and the combined drought and heat stress; that of the TPA: leucine-rich repeat receptor-like protein kinase family protein (B7ZYR5) was reduced by the individual drought and heat stress treatments, that of the receptor-like protein kinase HERK1-like (C0PND4) was increased by all three stress treatments; and that of the calciumsensing receptor (B6TVL4) and the gibberellin receptor GID1l2 (B6TY90) were increased by the combined stress.

### **Discussion**

Reversible protein phosphorylation is a ubiquitous regulatory mechanism that plays critical roles in transducing stress signals to bring about coordinated intracellular responses. In this study, we report the comprehensive analysis of the phosphorylation changes in maize leaves response to drought, heat and their combination using iTRAQ-based quantitative proteomic and LC-MS/MS methods.

### **Phosphorylation Regulatory Network in Maize leaves Response to Three Stress Treatments**

The interplay between phosphatases and kinases strictly controls biological processes such as metabolism, transcription, cell

cycle progression, differentiation, cytoskeletal arrangement and cell movement, apoptosis, intercellular communication, and immunological functions (Johnson, 2009; Pjechová et al., 2014). In this study, five kinases and three phosphatases were identified under heat stress, three kinases and two phosphatases were under drought stress, and three kinases and three phosphatases were under combined heat and drought stress. Nevertheless, according to the analysis of protein-protein interactions among significantly changed phosphoproteins, only AGC\_PKAPKG\_like.1-ACG kinases, phospholipase C and protein phosphatase 2C were found to have interactions with some phosphoproteins, which involved in gene expression, protein synthesis, and stress tolerance. Our analysis suggested potential multiple phosphorylation regulatory mechanisms of these phosphoproteins for further experimental validation. Furthermore, phospholipase C and protein phosphatase 2C were showed to have a significant phosphoryltion changes under three stress conditions. Previous study also show that the phosphorylation levels of phospholipase C and protein phosphatase 2C is obviously affected (Li et al., 2013; Hwang et al., 2014; Wei et al., 2014; Zhang et al., 2014), which indicated that the signal pathway related to phospholipase C and protein phosphatase 2C played an important role in plants response to stress.

Besides, we found phosphorylation events in other important kinase. For example, serine threonine-protein kinase wnk4-like (K7TZQ1) changed significantly under three stress conditions. Ser/Thr phosphorylation plays key roles in the regulation of plant growth and development. In wheat response to drought stress, phosphoproteome analysis also reveals two serine threonineprotein kinases (Zhang et al., 2014). Calcium-dependent protein kinase (C4J038, CDPK) changed under drought stress. In plants, calcium is a ubiquitous second messenger in signal transduction cascades. Most of the known Arabidopsis calcium-stimulated protein kinase activities are related to CDPKs (Cheng et al., 2002). In wheat, among nine phosphorylated CDPKs identified, CDPK7 was phosphorylated in response to drought stress (Lv et al., 2014).

## **Enzymes Involved in the Gluconeogenesis Pathway**

Phosphoenolpyruvate carboxykinase (PEPCK, C0P3W9) plays an important role in organic acid metabolism. In the cytoplasm, PEPCK and malate dehydrogenase can synthesize malate from glycolytically derived phosphoenolpyruvate (Fortes et al., 2011). In this study, PEPCK had 7 significantly phosphorylated peptides under heat stress, 5 under drought stress and 8 under the combined stress. Regardless of drought, heat and their combination, the phosphorylation level of these peptides increased or decreased under the same stress. Importantly, although the phosphorylation level changed across the different treatments, the protein expression level did not. Similarly, in drought-stressed *Pinus halepensis*, PEPCK activity increased without an increase in its transcription or translation (Fontaine et al., 2003; Hýsková et al., 2014). These results demonstrated that PEPC phosphorylation plays an important role in plants' responses to abiotic stress.

Fructose-bisphosphate aldolase is a key enzyme in the pathways of gluconeogenesis. In drought-tolerant tomato response to drought, the gene encoding this enzyme is down-regulated by drought stress (Gong et al., 2010). In the present study, the three stresses all increased the phosphorylation level of this enzyme, and heat and combined drought and heat stress decreased the protein expression. Gluconeogenesis consumes plenty of energy thus down-regulation of the fructosebisphosphate aldolase could inhibit gluconeogenesis for keeping energy in stressed plants.

## **Protein Degradation by Ubiquitination**

Protein degradation via the ubiquitin/26S proteasome system is the main protein degradation pathway; it plays a crucial role in removing misfolded or damaged proteins and in controlling the abundance of certain regulatory proteins during abiotic stress (Vierstra, 2003; Stone, 2014). Ubiquitination is a multistep process involving the sequential action of three enzymes: E1 (ubiquitin activating enzyme), E2 (ubiquitin conjugating enzyme), and E3 (ubiquitin ligase). It has been shown that in plants, certain E3 ubiquitin ligases are involved in transcriptiondependent resistance to high temperature and drought stress (Kim and Kim, 2013; Liu et al., 2014). *Saccharomyces* bearing a mutation in RSP5, which encodes an essential E3 ubiquitin

ligase, are hypersensitive to heat stress (Uesugi et al., 2014). In humans, DNA damage induces the phosphorylation-dependent degradation of the E3 ubiquitin ligase TRIM24 in the nucleus, which disrupts the interaction between TRIM24 and p53 and activates p53 (Jain et al., 2014). However, in the present study, the E3 ubiquitin ligases rglg2-like isoform x1 and upl4-like were dephosphorylated under heat stress and the combined drought and heat stress, but the expression level of rglg2-like changed only minimally (upl4-like was not identified in the protein expression analysis). These results indicate that decreasing the phosphorylation level of the 26S proteasome protein complex can promote the eventual degradation of regulatory proteins that are involved in heat stress.

### **Phosphoproteins Involved in Water, Sugar and H+ Transport**

Transporter proteins are important for turgor pressure maintenance and water potential regulation, which are crucial for the growth and survival of plants under biotic and abiotic stress. For example, plasma membrane intrinsic proteins (PIPs) are primary channels that mediate water uptake in plant cells, and they are regulated by phosphorylation. In this study, two aquaporin isoforms were differentially phosphorylated under heat and the combined stress: phosphorylation of the aquaporin PIP2-7 and the integral membrane protein nod26-like were reduced and increased, respectively. Other results also demonstrated that the aquaporin CsPIP2-1, which is involved in phosphorylation-dependent salt- and drought-stress responses in Camelina (Jang et al., 2014), and PIPs from maize shoots are phosphorylated on serine residues by a calcium-dependent kinase both *in vitro* and *in vivo* (Van Wilder et al., 2008). Together, these results indicate that phosphorylation plays an important role in the activity of PIPs and, consequently, of water channels.

H+-ATPase, which belongs to the cation transport ATPase (P-type) family, is related to H+ electrochemical gradient across the plasma membrane, which regulates cell growth and response to stress (Schaller and Oecking, 1999; Bobik et al., 2010). In *Nicotiana tabacum*, the phosphorylation of the H+-ATPases leads to the increase of this enzyme activity (Bobik et al., 2010). In wheat, the H+-ATPase identified in two cultivars showed up-regulated phosphorylation levels (Zhang et al., 2014).

### **TABLE 5 | Phosphoproteins related to response to stimulus, transport and DNA binding under D, H, and DH respectively.**


Peroxisome biogenesis protein 6-like/K7TNN3 Vacuolar amino acid transporter 1-like/ RAN-binding protein 1/B6T8F4

At-hook protein 1/B6TI42 At-hook protein 1/B6TI42 At-hook protein 1/B6TI42

Transcription initiation factor alpha subunit/C0PBP2 Probable receptor-like protein kinase

ZF-HD homeobox protein/B4FQM0 Protein stichel-like/K7UDH3

bel1-type homeodomain protein/B6SXN6 bel1-type homeodomain protein/B6SXN6 Bel1-type homeodomain protein/B6SXN6 Bzip transcription factor superfamily protein/F1DJV0 Bzip transcription factor superfamily protein/F1DJV0 Bzip transcription factor superfamily protein/F1DJV0 Cell division cycle 5-like/Q8W149 Cell division cycle 5-like/Q8W149 Eukaryotic translation initiation factor isoform

SPF1-like DNA-binding protein/B7ZZ27 SPF1-like DNA-binding protein/B7ZZ27 G-box binding factor 1/Q41735 Glycine-rich protein 2/C4JBR4 Glycine-rich protein 2/C4JBR4 Glycine-rich protein 2/C4JBR4 MYB-CC type transfactor/BF4F7W7 Hypothetical protein ZEAMMB73\_938746/K7V8I9 MYB-CC type transfactor/BF4F7W7 PHD-finger family homeodomain protein/Q41812 Methyl- binding domain containing expressed/Q94IQ8 Probable receptor-like protein kinase

at5g56460-like/C0PHB9 SPF1-like dna-binding protein/B7ZZ27 Transcription initiation factor alpha subunit/C0PBP2 Transposon protein/B6SRN0

### **DNA binding DNA binding DNA binding**

4g-1-like/K7TUM2 at5g56460-like/C0PHB9 SPF1-like dna-binding protein/B7ZZ27 MYB-CC type transfactor/BF4F7W7 RNA polymerase-associated protein rtf1 homolog/C0P8E4 Transposon protein/B6SRN0

In our study, the significant phosphorylation (B8A326) was found only under the drought and heat combined stress. Taken together, these results show that H+-ATPase may enhance the activity itself and participate in H+ homeostasis related to osmotic regulation. Besides, glycerol 3-phosphate permease (B8A1D5) and hexose transporter (B6U6U2) showed downregulated and up-regulated phosphorylation level, respectively under drought stress. Carbohydrate transporter sugar porter transporter (B6U8S7) showed up-regulated phosphorylation level under heat stress. Taken together, these results showed that the phosphorylation and dephosphorylation of transporters might help cell to maintain cell solute and ion stability which might play an active role in plant adaptation to abiotic stress.

# **The Sites of Phosphorylation Differ Across the Three Stress Treatments**

Analysis of phosphorylation site patterns showed a striking distinction among the three stress treatments, indicating that drought, heat, and the combined drought and heat stress resulted in different sites being phosphorylated (**Table 1**, Tables S1– S6). Under the three stress treatments, the phosphopeptides of each protein had different phosphorylation levels and regulatory patterns under the three stress treatments (**Table 3**).

Further, analysis of these motifs for significant phosphorylation phosphoproteins showed that [SP] and [RxxS] shared in common under three stress treatments, which were showed. [GxxxT] and [RSA] were specific to drought stress. [PxTP] and [RT] shared in common under heat stress and combined double stresses. [AxS] were specific to combined double stresses. These results provide substantial novel insight into phosphorylation and dephosphorylation and signal transduction; this is the first study to dissect the differences between drought, heat and combined drought and heat stress. Importantly, these results reveal a previously unappreciated quantity and diversity of phosphorylated protein isoforms, and uncover novel signaling pathways.

# **Phosphorylation Changes of Heat Shock Proteins**

Heat shock proteins (HSPs) have "chaperone-like" activities and accumulate in response to various stresses (Eisenhardt, 2013). At least five families of chaperones are found in higher plants: the HSP100 (Clp) family, the HSP90 family, the HSP70 (DnaK) family, chaperonins (GroEL and HSP60), and the small (20–40 kD) HSP (sHSP) family. sHSPs play important and extensive roles in plant defenses against abiotic stress (Mu et al., 2013). The phosphorylation of sHSPs has been demonstrated in maize mitochondria (Lund et al., 2001) and in barley (*Hordeum vulgare*) seed endosperm (Slocombe et al., 2004). In this study, the phosphorylation level of seven HSPs, including five sHSPs (B4G250, B4F976, B6T649, C0P2N6, and B4FR07) and two HSP70s, changed significantly under drought, heat or combined drought and heat stress. Notably, under heat stress and combined drought and heat stress, the phosphorylation level of the identified sHSPs (except B4FR07) was up-regulated. Taken together, these results indicate that the phosphorylation of sHSPs appears to be important for the regulation of sHSP function in plant responses to heat and to combined drought and heat stress. These results are the first to demonstrate the phosphorylation of plant sHSPs under the combination of drought and heat stress.

### **Receptor Proteins**

Most proteins that are synthesized on the rough endoplasmic reticulum are delivered to various cellular destinations, including the vacuoles and lysosomes. Such sorting involves the recognition of targeting signals on the proteins by receptors. The vacuolarsorting receptor (VSR) is involved in sorting clathrin-coated vesicles from the Golgi apparatus to the vacuoles. In pumpkin, a putative vacuolar sorting receptor, PV72, undergoes a Ca2+ dependent conformational change (Watanabe et al., 2002). Here, the phosphorylation level of vacuolar sorting receptor 3-like (B6U4K3) was significantly reduced under the three stress treatments, suggesting that stress affects vacuolar sorting in higher plants. The calcium-sensing receptor modulates the cytoplasmic Ca2<sup>+</sup> concentration and is crucial for proper stomatal regulation in response to elevated levels of external Ca2+. In this study, a calcium-sensing receptor (B6TVL4) located in the chloroplast thylakoid membrane had significantly increased phosphorylation levels under the combined drought and heat stress, but little is known about the role of calciumsensing receptors in plant responses to stress. Gibberellin (GA) perception is mediated by GID1 (GA-INSENSITIVE DWARF1), a receptor that shows similarity to hormone-sensitive lipases. The discovery of GID1 has yielded new insight into how GA is perceived (Hirano et al., 2008). In this study, the phosphorylation level of the gibberellin receptor GID1L2 (B6TY90) clearly increased under combined drought and heat stress. However, at present, the molecular mechanism of GID1L2's function in GA signaling, and especially in GA-dependent regulation of plant responses to stress, is unclear. The probable receptorlike, At5g56460-like protein kinase; a TPA: leucine-rich repeat receptor-like protein kinase family protein; and the receptor-like protein kinase HERK1-like were also identified, although their mechanisms of action require further study.

# **Conclusions**

Compared to single drought or heat stress, their combination had a different effect on phosphorylation sites and level of phosphoproteins. Of the 282 phosphoproteins identified in the present study, only 46 were found to show similar changes under all stress conditions. Of the 12 phosphorylation motifs identified, only two were common under the three stresses. In particular, these phosphoproteins involved in signaling pathways were different among drought, heat and their combination. Therefore, our results could not only provide new information for understanding the mechanisms of crop tolerance to the combined stress, but also contribute to the identification of crop cultivars with increased tolerance to increasing climate variability.

# **Author Contributions**

XH and WW conceived and designed the research. FZ and NL performed the experiments, DZ, GZ and XH analyzed the data. LW and CL contributed reagents/materials/analysis tools. XH and DZ wrote the paper. All authors read and approved the manuscript.

# **Acknowledgments**

This work was supported by the National Natural Science Foundation of China (grant no. 31171470) and the Program for Science and Technology Innovation Talents in Universities of

# **References**


Henan Province (grant no. 13HASTIT001). We thank Miss Gong Fangping and Miss Yang Le for their assistance in manuscript preparation.

## **Supplementary Material**

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00298/abstract


phosphorylates class I heat shock protein. *Plant Physiol. Biochem.* 42, 111–116. doi: 10.1016/j.plaphy.2003.11.009


**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.

*Copyright © 2015 Hu, Wu, Zhao, Zhang, Li, Zhu, Li and Wang. This is an openaccess 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.*

# Physiological and proteomic analyses on artificially aged *Brassica napus* seed

### *Xiaojian Yin1, Dongli He1, Ravi Gupta2 and Pingfang Yang1 \**

*<sup>1</sup> Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, China <sup>2</sup> Department of Plant Bioscience, College of Natural Resources and Life Science, Pusan National University, Miryang, South Korea*

### *Edited by:*

*Ganesh Kumar Agrawal, Research Laboratory for Biotechnology and Biochemistry, Nepal*

### *Reviewed by:*

*Joshua L. Heazlewood, The University of Melbourne, Australia Arkadiusz Kosmala, Institute of Plant Genetics of the Polish Academy of Sciences, Poland*

### *\*Correspondence:*

*Pingfang Yang, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuchang Moshan, Wuhan 430074, China e-mail: yangpf@wbgcas.cn*

Plant seeds lose their viability when they are exposed to long term storage or controlled deterioration treatments, by a process known as seed aging. Based on previous studies, artificially aging treatments have been developed to accelerate the process of seed aging in order to understand its underlying mechanisms. In this study, we used *Brassica napus* seeds to investigate the mechanisms of aging initiation. *B. napus* seeds were exposed to artificially aging treatment (40◦C and 90% relative humidity) and their physio-biochemical characteristics were analyzed. Although the treatment delayed germination, it did not increase the concentration of cellular reactive oxygen species (ROS). Comparative proteomic analysis was conducted among the control and treated seeds at different stages of germination. The proteins responded to the treatment were mainly involved in metabolism, protein modification and destination, stress response, development, and miscellaneous enzymes. Except for peroxiredoxin, no changes were observed in the accumulation of other antioxidant enzymes in the artificially aged seeds. Increased content of abscisic acid (ABA) was observed in the artificially treated seeds which might be involved in the inhibition of germination. Taken together, our results highlight the involvement of ABA in the initiation of seed aging in addition to the ROS which was previously reported to mediate the seed aging process.

**Keywords:** *Brassica napus***, seed aging, controlled deterioration treatments, proteomics**

### **INTRODUCTION**

Successful germination of seeds is a prerequisite for plants to initiate their life cycle and to distribute their progeny, and is largely determined by the seed vigor (Holdsworth et al., 2008; Rajjou et al., 2012). Seed aging is a process which results in delayed germination, reduction in germination rate, and sometimes even a total loss of seed viability (Priestley, 1986). Prolonged storage of seeds induces seed aging which is a major complication in plant germplasm conservation (Garza-Caligaris et al., 2012). In agriculture, the aged crop seeds germinate poorly and negatively affect the seedlings growth and eventually the yield (Ellis, 1992). When stored in uncontrolled conditions, most of the agricultural crops have 1–5 years of seed viability, which is much less when compared to the wild plants. Optimized storage conditions have been proved fruitful in slowing down the rate of seed aging and eventually increasing the seed life span. For orthodox seeds, low temperature and moisture content are helpful (Walters et al., 2005) while high temperature and humidity have been shown to induce and accelerate the seed aging process (El-Maarouf-Bouteau et al., 2011).

Seed aging leads to various cellular and metabolic alterations including loss of membrane integrity, degradation of DNA, reduced primary metabolism and so on (Corbineau et al., 2002; Kibinza et al., 2006; El-Maarouf-Bouteau et al., 2011). Although the mature seeds are physiological quiescent, these could not prevent the production of reactive oxygen species (ROS). The over-accumulation of ROS and its attack on lipids and proteins are supposed to be the major cause of seed aging (Bailly, 2004). ROS result in the peroxidation and degradation of lipids, which eventually damage the integrity of cellular membranes (Lee et al., 2012; Parkhey et al., 2012). Generally, ROS are regarded as the main factor that leads to seed aging during storage (Priestley, 1986). It has been shown that accumulation of hydrogen peroxide is related to the loss of seed viability in sunflower (Bailly et al., 1996; Kibinza et al., 2006). Under stress conditions, ROS promote program cell death (PCD) in both plants and animals (Grant and Loake, 2000; Neill et al., 2002). However, it is still unknown whether the seed aging is also induced by the ROS through the triggering of PCD or not.

Seed aging is highly associated with the storage conditions, however, recent studies have shown that seeds of different plant species show varied rate of seed aging under the same storage conditions (Walters et al., 2005). It is believed that the viability of the seed is determined by its genetic background as well as the storage conditions (Bewley, 1997; Miura et al., 2002; Clerkx et al., 2004). In Arabidopsis, genes involved in flavonoid and tocopherol biosynthesis can contribute to its seed longevity (Debeaujon et al., 2000; Sattler et al., 2004). Furthermore, a dormancy-related gene *delayed of germination 1* (*DOG1*) (Bentsink et al., 2006) and a heat stress responsive transcription factor (Prieto-Dapena et al., 2006) were also found to improve the resistance to aging. Very recent studies verified that methionine sulfoxide reductases from *Medicago truncatula* and PROTEIN L-ISOASPARTYL METHYLTRANSFERASE (PIMT) from Arabidopsis and lotus (*Nelumbo nucifera*) could enhance the seed vigor and longevity (Oge et al., 2008; Chatelain et al., 2013; Verma et al., 2013). With the advancement in genomics and other large scale "omics" techniques, number of transcriptomics and proteomics studies have been conducted in the last decade in order to identify and characterize the potential biomarkers for the seed aging (Nakabayashi et al., 2005; Rajjou et al., 2008).

*Brassica napus* is one of the major sources of edible oil all over the world. However, *B. napus* seeds are harvested in late spring, and their storage go through the summer season, which leads to the loss of their viability. Thus, it is of great practical importance to prevent the loss of seed vigor, which needs to obtain a comprehensive understanding of the mechanisms underlying the seed aging. Unfortunately, study on this topic in *B. napus* is rather weak. In this study, we exposed the *B. napus* seeds to high temperature and humidity, and conducted a comparative proteomic analysis of control and artificial aged seeds in order to understand the underlying mechanisms. A lot of differentially accumulated proteins were identified, which were different with previous studies in other plants. Our results provide some new insights on seed aging mechanisms in *B. napus*.

### **MATERIALS AND METHODS**

### **PLANT GROWTH, AGING TREATMENTS AND GERMINATION ASSAYS**

*B. napus* (zhongshuang11) plants were grown in green house under natural light condition in Wuhan, China. The nondormant seeds were harvested in May of each year and used as experimental materials. Freshly harvested *B. napus* seeds were treated with high temperature and humidity according to Rajjou et al. (2008) with slight modifications. Briefly, seeds were exposed to 40◦C and 90% air humidity for different time points (0, 12, 24, and 48 h). Seeds stored at room temperature in sealed plastic bag, at dry conditions, for 1 year were used as natural aged seeds.

The untreated and treated seeds were dried in oven at 40◦C overnight, and then dipped in distilled water at 26◦C in darkness for germination. The germination rate for each sample was calculated after every 6 h until there is no more seed germination. For abscisic acid (ABA) treatment and gibberellic acid (GA) recovery experiments, seeds were imbibed with 10−<sup>8</sup> M ABA and 10−<sup>7</sup> M GA3 solutions during germination, respectively. Three biological replicates were performed for each treatment as well as germination assay with 50 seeds in each set of the replicate. The schematic flowchart of the whole experiment is shown in Figure S1.

### **MEASUREMENT OF ION LEAKAGE, MALONDIALDEHYDE AND HYDROGEN PEROXIDE CONTENT**

The ion leakage was calculated as described previously (Shi et al., 2012), by measuring the relative conductivity of the samples. Briefly, 0.1 g of seeds at 0 h of germination for both samples were incubated in 6 mL of distilled water, for 4 h at room temperature with constant shaking. After the incubation, the initial conductivity (C1) of the solution was measured. Final conductivity (C2) of the solution was measured after boiling the seeds for 30 min and cooling down the solution to room temperature. REL was calculated as the percentage of conductivity before and after boiling [(C1/C2) × 100] using a conductivity meter (Leici-DDS-307A, Shanghai precision scientific instrument company, Shanghai, China).

The malondialdehyde (MDA) content was measured using a commercial kit (S0131, Beyotime, Nanjing, China) according to the manufacturer's protocol, which is based on the reaction between MDA and thiobarbituric acid to produce a red compound. In brief, 0.2 g of seeds were homogenized with 2 mL of ice-cold phosphate buffer and centrifuged at 1600 × *g* for 10 min at 4◦C. The supernatant was then mixed with an equal volume of 0.5% thiobarbituric acid solution. The mixture was boiled for 10 min. After being cooled down to room temperature with water, the mixture was centrifuged at 3000 × *g* for 15 min at room temperature. The absorbance of the supernatant was determined at 530 nm. The concentration of MDA was calculated according to standard curve which was generated with known concentrations of MDA.

Measurement of H2O2 was carried out according to the method described before (Jaw-Neng Lin, 1998). In brief, 0.1 g of seeds were homogenized in PBS buffer and centrifuged at 12000 × *g* for 20 min at 4◦C. The supernatant was mixed with equal volume of 0.1% titanium sulfate in 20% H2SO4 (v/v), and then centrifuged again at 6000 × *g* for 15 min at room temperature. The absorbance of the supernatant was measured at 410 nm. The concentration of H2O2 was calculated based on the standard curve which was made with a series of H2O2 solutions with known concentration. All the measurements were conducted for three biological replicates.

### **MEASUREMENT OF ABSCISIC ACID CONTENT**

ABA concentration was measured using a derivatization approach coupled with nano-LC-ESI-Q-TOF-MS (Bruker Daltonics, Bremen, Germany) as described previously (Chen et al., 2012). Briefly, 0.1 mg of seeds were homogenized in liquid nitrogen, and then transferring the powder to a 2 mL centrifuge tube, followed by extraction with 500µL modified Bieleski solvent (methanol/water/formic acid, 15/4/1, v/v/v) was added to it and the mixture was incubated at 4◦C for 12 h. The stable isotope labeled ABA ([2H6] ABA, 50 ng/g) was added to each of the samples to serve as internal standards for the quantification. Then, the supernatants were sequentially passed through the tandem solid phase extraction (SPE) cartridges containing C18 adsorbent (50 mg) and SAX adsorbent (200 mg). Before SPE extraction, the tandem cartridges were pre-conditioned with 8 mL H2O, 8 mL methanol, and 8 mL modified Bieleski solvent. After sample loading, the C18 cartridge was removed and the SAX cartridge was rinsed with 2 mL methanol/H2O (20/80, v/v). After that, 3 mL acetonitrile (CAN) with 1% Hydrofluoric acid (FA) (v/v) was applied to elute the targeted ABA and the eluent was evaporated under mild nitrogen stream at 35◦C followed by re-dissolving in 100µL H2O. The resulting solution (100µL) was then acidified with 10µL FA, and extracted with ether (2 × 1 mL). The ether phase was combined,dried under nitrogen gas and reconstituted in 100µL ACN. To the resulting solution, 10µL triethylamine (TEA) (20µmol/mL) and 10µL 3- Bromoactonyltrimethylammonium bromide (BTA) (20µmol/mL) were added. The reaction solution was vortexed for 30 min at 35◦C and evaporated under nitrogen gas followed by re-dissolving in 200µL H2O/ACN (90/10, v/v) for instrumental analysis. The calibration curve was constructed by comparing peak area ratio (analyte/IS) to concentrations. The content of ABA was calculated according to the calibration curve. Three biological replicates were conducted.

### **SUPEROXIDE DISMUTASE AND CATALASE ACTIVITY ASSAYS**

Total superoxide dismutase (SOD) activity was measured using a commercial WST-1 kit (S0102, Beyotime) following manufacturer's protocol. Briefly, seeds were powdered using liquid nitrogen and then homogenized in phosphate balanced solution buffer (PBS, pH7.5). The mixture was centrifuged at 12,000 × *g* at 4◦C for 15 min. The supernatant thus obtained was used for the SOD activity measurement. The principle of this method lied on the coupling of 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2, 16 4-disulfophenyl)-2H-tetrazolium (WST-1) with xanthine oxidase (XO) to generate O2<sup>−</sup> and formazan dye, which can be inhibited by SOD through catalyzing O2<sup>−</sup> into H2O2 and O2. SOD activity can be calculated by determining the absorbance of formazan dye at 450 nm.

Catalase (CAT) activity was assayed using commercial kit (S0051, Beyotime) as described previously (Shi et al., 2012). Briefly, 10µL of 250 mM H2O2 was mixed with 5µL of protein supernatant. The H2O2 was decomposed by CAT for 5 min, and the remaining H2O2 coupled with a substrate was treated with peroxidase (POD) to generate N-4-antipyryl-3-chloro-5 sulfonate-p-benzoquinonemonoimine. CAT activity was determined by calculating the decomposition rate of H2O2at 520 nm. Both enzymes were assayed for three replicates.

### **PROTEIN EXTRACTION AND TWO-DIMENSIONAL ELECTROPHORESIS**

Proteins were extracted from *B. napus* seeds at 0 and 18 h after imbibition according to the method described previously (Chi et al., 2010). Briefly, 0.2 g of seeds were homogenized in ice-cold buffer containing 20 mM Tris-HCl (pH 7.5), 250 mM sucrose, 10 mM ethylenebis(oxyethylenenitrilo) tetraacetic acid (EGTA), 1 mM phenylmethanesulfonyl fluoride (PMSF), 1 mM DL-dithiothreitol (DTT), and 1% Triton X-100. The homogenate was centrifuged at 12000 × *g* for 30 min at 4◦C. The supernatant was mixed with isometric Tris-Phenol (pH7.8) and vortexed for 20 min. Mixture was then centrifuged at 12000 × *g* for 15 min at 4◦C. After centrifugation, the supernatant phenol phase and intermediate denatured protein layer were collected. Phenol phase containing proteins was then mixed with 5 volumes of 0.1 M ammonium acetate in methanol and incubated at −20◦C overnight. Pellet, obtained after centrifugation at 12000 × *g* for 30 min, was washed 4 times with ice-cold acetone and vacuum dried. For two-dimensional electrophoresis (2-DE), the dried proteins pellets were dissolved in rehydration buffer containing 7 M urea, 2 M thiourea, 4% 3-[(3-Cholamidopropyl)dimethylammonio]propanesulfonate

(CHAPS), 0.2% carrier ampholyte, and 65 mM DTT, and quantified according to Bradford's method (Bradford, 1976; Kruger, 1994). A total of 600µg of proteins of each sample were loaded on 17 cm IPG strip by rehydration loading for 12 h at room temperature. Based on primary screening, the IPG strips with pH 5–8 (linear) were selected in this study. Isoelectric focusing (IEF) was carried out at 200, 500, and 8000 V for 1, 1.5, and 10.5 h, respectively (He et al., 2011). After IEF, the strips were incubated in equilibration buffer containing 0.05 M Tris-HCl pH 6.8, 2.5% SDS, 10% (v/v) glycerol and 2% DTT and shaken for 15 min, and then for another 15 min with the iodoacetamide replaced DTT equilibration buffer. The second-dimensional separation of the proteins was carried out on 12% SDS-PAGE (Li et al., 2012).

### **GEL STAINING, SCANNING AND ANALYSIS**

The gels were stained with Coomassie brilliant blue-red (CBB-R) 250 for 40 min, and then destained with 20% ethanol containing 10% acetic acid. The destained gels were scanned using Epson Perfection™ V700 Photo scanner (Epson, China Co., Ltd.) at 800 dots per inch (dpi) resolution. The transparency mode was used to obtain a gray scale image. The images were digitized and analyzed with PDQuest™ 2-DE Analysis Software 8.0 (BIO-RAD, CA, USA). The relative volume of each spot, which is defined as is the ratio between the volume of the given spot and the total volume of all the spots displaying on the gel, was used to represent the corresponding protein abundance. To obtain reproducible result, three biological repeat experiments were conducted. All the 12 gels were digitized separately. Automatic matching between different gels was conducted with four spots as internal standards. After that, manual adjustment was carried out to avoid mismatching. Spots which were detected in all the three replicates were used for comparative analysis. The spots showing more than 2-fold changes in abundance were defined as differentially displayed protein spots.

### **IDENTIFICATION OF PROTEINS THROUGH MALDI -TOF/ TOF MS**

Differentially displayed protein spots were excised from the gels and distained using 50 mM NH4HCO3 in 50% (v/v) ACN. After complete de-staining, excised spots were first dehydrated using 50µL 100% ACN and then rehydrated with 10 pmol trypsin in 25 mM NH4HCO3 at 4◦C for 1 h. Trypsin digestion was carried out at 37◦C overnight. After digestion, the peptides were extracted according to the method described before (Yang et al., 2007). The collected peptides were desalted and analyzed with an ultrafleXtreme Matrix-Assisted Laser Desorption/ Ionization tandem Time of Flight (MALDI-TOF/TOF) mass spectrometer (Bruker, Germany) in a MS-MS mode. All the parameters were set to default. Briefly, 0.5 kHz laser with 40 k resolution was applied. The flexAnalysis software (Bruker) was used to generate the peak lists and process the MS and MS/MS spectra, which were searched against NCBInr (containing 15823071 sequences and 5433757279 residues) and Swiss-Prot databases (containing 138011 sequences from *B. napus*) using MASCOT as a search engine (Mascot Wizard 1.2.0, Matrix Science Ltd.) through BioTools (version 3.2) interface. The search parameters were set as follows: taxonomy, Viridiplantae; fixed modifications, carbamidomethylation; variable modification, methionine oxidation; MS tolerance, 50 ppm; MS/MS tolerance, 0.5 Da, peptide mass, monoisotopic. Only the significant hits (*p <* 0*.*01) with peptide scores *>*45 were accepted.

### **STATISTICAL ANALYSIS**

The statistical analyses were carried out using Student's *t*-test when only two groups were compared, or with One-Way ANOVA followed by Tukey's multiple comparisons test for all other comparisons based on the three independent replicates.

### **RESULTS**

### **DETERMINATION OF SUITABLE ARTIFICIAL AGING TREATMENTS**

Mature seeds of *B. napus* gradually lose their viability during long term storage, which is defined as natural aging. To better understand the process of natural aging, germination percentages of 1 year aged seeds and freshly harvested seeds were calculated. Freshly harvested seeds started germination just after 6 h of imbibition and took less than 12 h for 50% percent of the seeds to germinate (**Figure 1**). In contrast, germination of the aged seeds was clearly delayed (**Figure 1**). It took about 30 h for the 1 year aged seeds to achieve 50% germination. In-spite of delayed germination of the aged seeds, most of the seeds were still able to germinate. Based on these results, we concluded that these seeds were at the early stage of natural aging and thus could be used as criterion to determine the suitable artificial aging treatments.

The fresh harvested seeds were exposed to 40◦C and 90% relative humidity for 0, 12, 24, and 48 h and viability of each seed sample was checked using germination assays (**Figure 1**). The treatment apparently declined the seed germination rate with increasing exposure time (**Figure 1**). For the sake of comparison, sample with similar germination rate with the natural aged control was selected for further study. Based on this criterion, 24 h of treatment (**Figure 1**) was determined as suitable artificial aging treatment as its germination rate was almost similar to that of natural aged seeds. Hereafter, treatment with 40◦C and 90% relative humidity for 0 and 24 h is defined as control (CK) and CDT treatment, respectively in the rest of the manuscript.

### **EFFECTS OF ARTIFICIAL AGING ON PHYSIO-BIOCHEMICAL STATUS OF** *B. NAPUS* **SEED**

As mentioned above, plasma membrane could be damaged during aging (Lee et al., 2012; Parkhey et al., 2012), therefore, in

order to test the integrity of the membranes, relative ion leakage of CK and CDT samples at 0 h of imbibition were measured. The CDT samples showed approximately 1.8 fold higher ion leakage than the CK samples, indicating higher membrane damage in the former sample (**Figure 2**). Moreover, MDA content measurements of CK and CDT seeds also showed an increased MDA in the CDT seeds (**Figure 2**), further suggesting the increased membrane damage in the CDT seeds. Besides ion-leakage and MDA content measurements, we also measured the concentrations of H2O2 and O−, which are considered the two major ROS in plants. Surprisingly, the concentrations of both H2O2and O<sup>−</sup> were found to be similar in CK and CDT samples, indicating that increased ROS concentration might not be a necessary event during seed aging of *B. napus* seeds (**Figure 3A**). A gradual decrease in the concentration of ROS was observed during seed germination (**Figure 3A**). However, the decrease of ROS in the artificial aged seeds was much slower as compared to the control sample (**Figure 3A**), which might explain why the germination in aged seeds was delayed. Consistent with the changes in the ROS concentrations, the activities of the antioxidant enzymes, SOD, and CAT, were also dramatically declined in the artificially aged seeds before germination (**Figure 3B**).

### **CHANGES IN THE PROTEOME OF** *B. NAPUS* **SEEDS EXPOSED TO ARTIFICIAL AGING TREATMENT**

To further explore the molecular mechanisms underlying the loss of seed vigor due to artificial aging treatments; comparative proteomic analysis was carried out. Based on the germination assay result, CK and CDT seeds showed obvious difference in germination rate at 18 h of imbibitions (**Figure 1**), so proteins were isolated from the *B. napus* seeds at 0 h and 18 h after imbibition respectively. The proteins will then be resolved on 2-D PAGE and stained with CBB R-250. Three independent replicates were performed for each sample (**Figures 4**, S2). After staining, the gels were scanned and digitized for the comparison. Analysis of the 2-DE gels was conducted with PDQuest™ 2-DE Analysis software (version 8.0). Combined the biological replicates showed that there were about 600 reproducible protein spots (550–600) on each gel. A comparison of the 2-DE gels showed a total of 81 differentially accumulated spots (more then 2-fold changes in abundance) between CK and CDT samples, of which 36 and 48 were from 0 to 18 h of imbibition respectively. Three spots were common at both time points of imbibition.

To know if any of these differentially accumulated spots were involved in the seed aging process, these were excised from the gels, digested with trypsin and subjected to MALDI-TOF/TOF MS analysis. Based on the criteria described in materials and methods, 49 protein spots were successfully identified (Supplemental **Table S1**). Among these, seven spots were identified as two proteins, which resulted in the identification of a total of 54 distinct proteins (**Table 1**, Figure S3). Of these identified proteins, 15 proteins were only with the *B. napus* EST accession number. The sequences of these proteins were blasted against Arabidopsis database to gain a better understanding of their functions (**Table 2**). For many identified proteins, there were differences between their experimental and theoretical pI and molecular weight. Several possibilities may help to explain

this result. First, the genome of *B. napus* is not fully sequenced and annotated, which might result in deviation; second, some of the proteins might be subjected to post-translational modifications or cleavage. MapMan ontology analysis (Thimm et al., 2004) sorted all the identified proteins into 10 functional groups including metabolism, protein destination, stress response, redox homeostasis, development, hormones related, cell structure, miscellaneous enzymes, storage proteins, and function unassigned proteins (**Table 1**, **Figure 5**).

### **EFFECTS OF ABA AND GA ON THE SEED AGING**

As no major differences in the ROS concentrations and antioxidant enzymes activities were observed after CDT treatment, we speculated that oxidative stress is not the major factor involved in the inhibition of germination after CDT treatment. Therefore, involvement of some other factors, which resulted in delayed germination after CDT treatment, was expected. Among all the external and internal factors, ABA seems to be the most important one which inhibits seed germination. Therefore, we measured the ABA content of the CK and CDT seeds at both 0 and 18 h after imbibition. Interestingly, the ABA content of the CDT seeds was much higher than that observed in the CK seeds at both the time points, and showed a sharp decrease upon imbibition in both the samples (**Figure 6**), suggesting involvement of ABA in seed aging. To further confirm this hypothesis, the CK seeds were germinated in the presence of ABA, and showed a delayed germination (**Figure 7**). It is known that GA and ABA play antagonistic roles in regulating seed germination, so we also germinated the CDT seed in the presence of GA to see if this treatment can recover its germination phenotype or not. Consistently, GA treated CDT seeds showed partially recovered germination (**Figure 7**).

### **DISCUSSION**

Temperature and moisture content (relative humidity) are two important environmental factors that influence seed aging. High temperature and moisture accelerate the process of seed aging,

leading to the rapid loss of seed vigor. The accelerated aging has been applied as an indicator of crop seed storability (Priestley, 1986). To study the mechanism of seed aging, researchers have developed the accelerated aging protocol in laboratory by exposing the seeds to high temperature and humidity conditions (Job, 2005; El-Maarouf-Bouteau et al., 2011). ROS have been presumed as the main factors which lead to the seed aging during storage (Bailly et al., 1996; Bailly, 2004; Kibinza et al., 2006; Parkhey et al., 2012). In order to understand the mechanisms underlying *B. napus* seed aging, especially at the initiation stage, we subjected the *B. napus* seeds with CDT, which obviously delayed the germination and slightly decreased the germination rate which was similar to the 1 year naturally aged seeds (**Figure 1**). These results suggested that this treatment was suitable for further analysis as it could mimic the physiology of 1 year naturally aged seeds.

The CDT resulted in the increased ion leakage (**Figure 2**), which reflects the damages of cellular membranes. This damage, in turn, resulted in the accumulation of MDA in the CDT seeds (**Figure 2**). Previously, it was considered that attack of ROS on membranes might be one of the reasons that lead to the ion leakage (Leymarie et al., 2012). Furthermore, previous study also showed that high temperature and humidity drastically increased the extent of protein oxidation in Arabidopsis seeds (Rajjou et al., 2008). However, after the CDT in the *B. napus* seeds, we did not detect any over accumulation of ROS (**Figure 3A**), and any differentially accumulated antioxidant enzymes except for peroxiredoxin (**Table 1**). Since only about 500 proteins were detected, it could not be absolutely excluded that there were some antioxidant enzymes changed. In spite of this, CDT did decrease the rate of ROS detoxification during germination (**Figure 3A**), suggesting that artificial aging might disturb the ROS scavenging system. This result was also supported by the decline of SOD and CAT activities (**Figure 3B**). The CDT might decrease the activities of the antioxidant enzymes because of the high temperature, since most of the enzymes have their highest activities at around 35◦C under physiological conditions.

Functional classification of the identified proteins showed that proteins related to protein modification and destination, development and cell structure were differentially modulated after CDT. However, previous studies on pea (Barba-Espin et al., 2011) and wheat seeds (Bykova et al., 2011) have shown that H2O2 treatment preferentially regulates proteins majorly involved in redox

### **Table 1 | Identification of the differentially displayed proteins in rapeseeds exposed to artificial aging treatment.**



*\* SC stands for sequence coverage.*

*\*\* Fold changes are Means* <sup>±</sup> *SE from three replicates; –*<sup>∞</sup> *stands for the disappeared spots in the related samples. D and U in protein ID list stand for downregulated protein and up-regulated protein respectively.*

# *Indicates these proteins were statistically not but manually checked to be differentially expressed.*

### **Table 2 | Blast of the identified proteins through searching against** *Brassica napus* **EST database.**


homeostasis, metabolism, hormones related and RNA related, and it could invoke the changes of proteins with similar functions in different tissues (Wan and Liu, 2008; Zhou et al., 2011). The differences in the differentially accumulated proteins in this study with the previously published reports on pea and wheat, suggest that the seed aging mechanism in the *B. napus* seeds is quite different with that indicating in case of pea and wheat seeds. Our results suggest that CDT treatment might initiate the aging of *B. napus*seeds through enhancing the biosynthesis of germination inhibitors and lately through the accumulation of ROS. Although there were 81 differentially accumulated proteins between CK and CDT seeds, only 49 spots were successfully identified. The identification rate is just about 60%, which might be explained by the un-sequenced genome of *B. napus*. Moreover, there seven spots were identified as two different proteins, which should result in difficulties to evaluate a real abundance of each protein within them. This is one of the limitations in 2-D gel based proteomic techniques. Future studies with gel-free systems might help to provide more information.

Since the CDT treatment delayed the germination not primarily because of accumulation of ROS, we speculated that there should be some other mechanisms which mediate the seed aging process in *B. napus* seeds. Among the identified proteins, most of the proteins involved in metabolism and protein destination were

decreased in the CDT seeds. These results indicate that the basic biological activities might be inhibited by the CDT treatment, which would result in the delayed germination. Interestingly, rubisco large subunit was increased in the CDT treated seed at 18 h after germination, which indicates the existence of feedback regulation. In contrast to the metabolism and protein destination related proteins, many cell structural proteins and miscellaneous enzymes, such as actin, mannose-binding lectin superfamily protein, glycosyltransferase, beta-glucosidase, were increased. All these proteins are related to the cell and cell wall structures. It might be interesting to know how these proteins affect seed aging.

Among all the internal factors, ABA has been reported to be one of the main factor that inhibits the seed germination (Gubler et al., 2005; Finch-Savage and Leubner-Metzger, 2006; Penfield et al., 2006). Measurements of the ABA content in *B. napus* seeds showed that there was a sharp increase of ABA content in the CDT treated seeds (**Figure 6**). Although the ABA was degraded during the germination, its concentration was still higher in the CDT treated seeds in comparison with the CK seeds (**Figure 6**). Furthermore, imbibition of the CK seeds with ABA solution showed delayed germination (**Figure 7**). Germination of the CDT treated seeds was partially recovered after the GA treatment,

further confirming the involvement of ABA in the inhibition of seed germination during aging. However, how this ABA concentration is increased during aging treatment in the *B. napus* seeds, is still elusive. Unfortunately, we did not detect any changes of the enzymes involved in the ABA biosynthesis and degradation. It is well known that ABA is synthesized in the seeds during desiccation which allow the survival of seeds in dry state (Tan et al., 1997; Nakabayashi et al., 2005). During this process, genes involved in ABA biosynthesis were found to be highly expressed, and corresponding enzymes were abundantly accumulated. Based on these results, we suggest that exposure of seeds with high humidity and temperature might partly recover the activities of the ABA biosynthesis enzymes which results in the enhanced production of ABA in the *B. napus* seeds during aging.

### **AUTHOR CONTRIBUTIONS**

YX did the experiments; HD prepared the seeds and analyzed some of the data; YP designed the experiments and wrote the manuscript.

### **ACKNOWLEDGMENTS**

The authors are grateful to Ms Tingting Li from Institute of Hydrobiology, Chinese Academy of Sciences for the assistance in MALDI-TOF/TOF MS analysis. This work was supported by the National Natural Science Foundation of China (NSFC, No. 31271805), 100 talents program of Chinese Academy of Sciences, and Sino-Africa Joint Research Project (SAJC201324).

### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at: http://www*.*frontiersin*.*org/journal/10*.*3389/fpls*.*2015*.* 00112/abstract

### **Table S1 | Identification of the differentially expressed proteins.**

### **REFERENCES**

Bailly, C. (2004). Active oxygen species and antioxidants in seed biology. *Seed Sci. Res.* 14, 93–107. doi: 10.1079/SSR2004159

Bailly, C., Benamar, A.,Corbineau, F., and Come, D. (1996). Changes in malondialdehyde content and in superoxide dismutase, catalase and glutathione reductase activities in sunflower seeds as related to deterioration during accelerated aging. *Physiol. Plant.* 97, 104–110. doi: 10.1111/j.1399-3054.1996. tb00485.x


**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 September 2014; accepted: 11 February 2015; published online: 25 February 2015.*

*Citation: Yin X, He D, Gupta R and Yang P (2015) Physiological and proteomic analyses on artificially aged Brassica napus seed. Front. Plant Sci. 6:112. doi: 10.3389/fpls. 2015.00112*

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

*Copyright © 2015 Yin, He, Gupta and Yang. 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.*

# Proteomic analysis reveals differential accumulation of small heat shock proteins and late embryogenesis abundant proteins between ABA-deficient mutant *vp5* seeds and wild-type *Vp5* seeds in maize

## *Xiaolin Wu†, Fangping Gong†, Le Yang , Xiuli Hu , Fuju Tai and Wei Wang\**

*State Key Laboratory of Wheat and Maize Crop Science, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, Zhengzhou, China*

### *Edited by:*

*Silvia Mazzuca, Università della Calabria, Italy*

### *Reviewed by:*

*Myriam Ferro, Commisariat à l'Energie Atomique et aux Energies Alternatives, France Loïc Rajjou, AgroParisTech - Paris Institute of Technology for Life, Food and Environmental Sciences, France*

### *\*Correspondence:*

*Wei Wang, Collaborative Innovation Center of Henan Grain Crops, College of Life Science, Henan Agricultural University, 63 Nongye Road, Zhengzhou 450002, China e-mail: wangwei@henau.edu.cn*

*†These authors have contributed equally to this work.*

ABA is a major plant hormone that plays important roles during many phases of plant life cycle, including seed development, maturity and dormancy, and especially the acquisition of desiccation tolerance. Understanding of the molecular basis of ABA-mediated plant response to stress is of interest not only in basic research on plant adaptation but also in applied research on plant productivity. Maize mutant *viviparous*-5 (*vp5*), deficient in ABA biosynthesis in seeds, is a useful material for studying ABA-mediated response in maize. Due to carotenoid deficiency, *vp5* endosperm is white, compared to yellow *Vp5* endosperm. However, the background difference at proteome level between *vp5* and *Vp5* seeds is unclear. This study aimed to characterize proteome alterations of maize *vp5* seeds and to identify ABA-dependent proteins during seed maturation. We compared the embryo and endosperm proteomes of *vp5* and *Vp5* seeds by gel-based proteomics. Up to 46 protein spots, most in embryos, were found to be differentially accumulated between *vp5* and *Vp5*. The identified proteins included small heat shock proteins (sHSPs), late embryogenesis abundant (LEA) proteins, stress proteins, storage proteins and enzymes among others. However, EMB564, the most abundant LEA protein in maize embryo, accumulated in comparable levels between *vp5* and *Vp5* embryos, which contrasted to previously characterized, greatly lowered expression of *emb564* mRNA in *vp5* embryos. Moreover, LEA proteins and sHSPs displayed differential accumulations in *vp5* embryos: six out of eight identified LEA proteins decreased while nine sHSPs increased in abundance. Finally, we discussed the possible causes of global proteome alterations, especially the observed differential accumulation of identified LEA proteins and sHSPs in *vp5* embryos. The data derived from this study provides new insight into ABA-dependent proteins and ABA-mediated response during maize seed maturation.

**Keywords: abscisic acid (ABA), late embryogenesis abundant proteins (LEA proteins), maize ABA-deficient mutant** *vp5***, mass spectrometry, protein abundance, proteome profiling, small heat shock proteins (sHSPs), 2-D electrophoresis (2-DE)**

### **INTRODUCTION**

Abscisic acid (ABA) is a major hormone that regulates a broad range of plant traits and is especially important for plant adaptation to environmental conditions. In seeds, ABA is thought to play a central role in many developmental stages, such as seed maturation and dormancy, the accumulation of nutritive reserves and the acquisition of desiccation tolerance (Quatrano, 1986). ABA-mediated plant response to stress has been extensively studied in different species ranging from Arabidopsis to crops, especially regarding ABA sensing, signaling, metabolism and transport (Umezawa et al., 2010). Knowledge about the complexity of ABA-mediated plant response to stress is still full of gaps, but the recent identification of ABA receptors (Ma et al., 2009; Santiago et al., 2009) and the key factors of the first step of ABA signal transduction (Park et al., 2009; Nishimura et al., 2010) in Arabidopsis provided an important insight into this mechanism.

Biosynthesis of ABA has been well characterized in Arabidopsis (Zeevaart and Creelman, 1988) and some data is available for other species, such as maize (Tan et al., 1997). Maize, *viviparous-5* (*vp5*) is deficient in ABA biosynthesis with the first step catalyzed by phytoene desaturase being blocked, which results in the precursor phytoene accumulation and carotenoid deficiency (Robichaud et al., 1980; Hable et al., 1998). Previous studies reported that ABA content in *vp5* embryos and endosperms was substantially reduced to 10 and 42% of the corresponding wild-type, respectively (Neill et al., 1986). The *vp5* seeds exhibit a visible phenotypic difference: the endosperm of mutant *vp5* seeds was white, while that of wild-type *Vp5* seeds was yellow. Therefore, *vp5* mutant is particularly useful not only for studies on the regulation of ABA-dependent maize genes, both in embryo and vegetative tissues, but also for studies of embryo development, seed germination and dormancy (Pla et al., 1989; Durantini et al., 2008).

Up to date, the expression of many individual gene/proteins has been studied using ABA-deficient mutant maize *vp5* and wild-type *Vp5* (Pla et al., 1989; Thomann et al., 1992). Williams and Tsang (1991) found *emb564* mRNA is expressed at low level in *vp5* embryos. The level of 3-hydroxy-3-methylglutaryl coenzyme A reductase activity, a rate-limiting enzyme of isoprenoid biosynthesis, is higher in *vp5* endosperm (Moore and Oishi, 1994). Recently, we compared root and leaf proteome differences between *vp5* and *Vp5* seedlings with 2-D gel electrophoresis (2-DE) combined mass spectrometry (MS/MS) and found that many proteins accumulation in roots or leaves are differentially regulated by drought and light in an ABA-dependent way (Hu et al., 2011, 2012). However, protein accumulation alterations caused by ABA-deficient mutation in *vp5* seeds are unclear at a proteome scale. Therefore, the characterization of seed proteome difference between *vp5* and *Vp5* is necessary for dissection of ABA-mediated maize response in the studies involved *vp5* mutants.

2-DE-based proteomics approach provides a powerful tool to analyze the expression levels of proteins, distinguish varieties and genotypes and even to identify single mutations with multiple effects (Lehesranta et al., 2005). This study aimed to characterize proteome alterations due to ABA-deficient mutation and further to identify ABA-dependent protein accumulation during seed maturation. We found significant proteome differences between *vp5* and *Vp5* seeds, where 46 differentially accumulated proteins were successfully identified. Most notably, six out of eight late embryogenesis abundant (LEA) proteins and nine small heat shock proteins (sHSPs) were found to differentially accumulate in ABA-deficient *vp5* embryos: six identified LEA proteins were repressed while nine sHSPs were induced.

# **MATERIALS AND METHODS**

### **PLANT MATERIALS**

Maize (*Zea mays* L.) ABA-deficient mutant *vp5* mutant was provided by Maize Genetics Cooperation Stock Center (Urbana,








*aSubcellular location of proteins was annotated in UniProtKB/Swiss-Prot (http:// www.expasy.org/ ).*

*bSubcellular location of proteins was predicted using the online Plant-mPLoc server (http:// www.csbio.sjtu.edu.cn/ bioinf/ plant-multi/ ).*

IL). The mutant *vp5* was propagated in primarily W64 genetic backgrounds. The *vp5* mutant was maintained as a heterozygote. Heterozygous seeds (*Vp5/vp5*) were planted under natural conditions at the farm of Henan Agricultural University (Zhengzhou, China). The homozygous *vp5* kernels were identified on segregating ears by their lack of carotenoid pigments (Robichaud et al., 1980). The mutant *vp5* kernels appear white, while *Vp5* kernels are yellow. Mature *vp5* and *Vp5* seeds from the same ear were sampled (Presentation 1 in Supplementary Material) used in this study. Dry maize seeds were soaked in water for 2 h to soften starchy endosperm. For each biological replicate, the embryos and the endosperm of 20 maize seeds were manually separated and used for protein extraction, respectively.

### **PROTEIN ISOLATION**

Embryos or endosperms were powdered in liquid N2 and further ground in a buffer containing 0.25 M Tris-HCl (pH 7.5), 1% SDS, 14 mM DTT and a cocktail of protease inhibitors. This slurry was heated to 65◦C for 5 min, vortexed, and heated at 95◦C for 2 min, vortexed again, and then centrifuged at 12,000 g for 10 min to remove cellular debris. The supernatant was recovered and subjected to phenol extraction as described (Wu et al., 2014). The protein pellet was dissolved in 2-DE rehydration buffer containing 7 M urea, 2 M thiourea, 2% (w/v) CHAPS, 20 mM DTT, 0.5% (v/v) IPG buffer (pH 4–7 or 7–10, GE Healthcare). The protein content was determined by Bradford microassay (Bio-Rad) with BSA standards.

### **SDS-PAGE AND IMMUNOBLOT**

SDS-PAGE was performed in a Laemmli gel system (5% stacking gel and 12.5% resolving gel). After electrophoresis, proteins in gels were visualized with colloidal CBB R350 or electroblotted onto polyvinylidene difluoride membrane (Hybond-P, GE healthcare) in a transfer buffer (20% v/v methanol, 50 mM Tris, 40 mM glycine). For immunoblot analysis, protein blots were soaked in TBST buffer (50 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 0.1% Tween-20) containing 5% low fat milk powder and gently shaken for 2 h at room temperature (RT). The blot was then incubated with anti-EMB564 polyclonal antibody (Wu et al., 2013a, 1: 5000 dilution) for 1 h. After washing with TBST, the blot was incubated in peroxidase-conjugated goat anti-rabbit IgG (1: 2000 dilution) at RT for 1 h. The blot was visualized with 0.08% 3,3 -diaminobenzidine tetrahydrochloride, 0.05% H2O2, 0.1 M Tris-HCl, pH 7.5.

### **2-DE, IMAGE AND DATA ANALYSIS**

Isoelectric focusing (IEF) was performed using 11-cm linear pH 4–7 IPG strips with the Ettan III system (GE Healthcare, USA). About 600µg proteins were loaded into the strip by passive rehydration overnight at RT. The IEF voltage was set at 250 V for 1 h, 1000 V for 4 h, finally increasing to 8000 V for 4 h, and holding for 10 h (20◦C). Focused strips were equilibrated in Buffer I (0.1 M Tris-HCl, pH 8.8, 2% SDS, 6 M urea, 30% glycerol, 0.1 M DTT) and then in Buffer II (same as Buffer I, but with 0.25 M iodoacetamide instead of DTT) for 15 min each. SDS-PAGE was run on a 13.5% gel with 0.1% SDS in the gel and the running buffer. The gels were stained with 0.1% CBB G-250 overnight and destained in 7% acetic acid until a clear background.

Protein gels were placed on a white plastic plate with transmission fluorescent lighting, and photographed using a DSLR camera (Nikon D7000) at an automatic mode. The digital images of the gels were analyzed by using PDQUEST 8.0 software (Bio-Rad). Protein spots were detected on scanned gels using the default spot detection setting. Gels of three biological replicates per genotype were analyzed. The spot intensities were normalized according to total intensity of valid spots to minimize possible errors due to differences in the amount of protein and staining intensity. Analysis of variance (ANOVA) was used to deal with protein spots quantification to identify individual protein spots with significantly different expression levels. Only those proteins with at least 1.5-fold quantitative variations in abundance were selected for mass spectrometry (MS) analysis.

### **MS/MS**

Protein spots of interest were manually excised from the gels and digested using trypsin. Proteins were reduced (10 mM DTT), alkylated (50 mM iodoacetic acid) and then digested with 10 mg/ml trypsin for 16 h at 37◦C in 50 mM ammonium bicarbonate. The supernatants were vacuum-dried and dissolved in 10µl 0.1% trifluoroacetic acid and 0.5µl added onto a matrix consisting of 0.5µl of 5 mg/ml 2, 5-dihydroxybenzoic acid in water: acetonitrile (2:1). The digested fragments were analyzed using a MALDI-TOF/TOF analyzer (ultraflex III, Bruker, Germany). MALDI-TOF/TOF spectra were acquired in the positive ion mode and automatically submitted to Mascot 2.2 (http://www*.*matrixscience*.*com, Matrix Science) for peptide mass finger printings against the NCBInr 20131226 database (35149712 sequences; 12374887350 residues, http://www*.*ncbi*.* nlm*.*nih*.*gov/). The taxonomy was Viridiplantae (green plants, 1669695 sequences). The search parameters were as follows: type of search: MALDI-TOF ion search; enzyme: trypsin; fixed modifications: carbamidomethyl (C); variable modifications: acetyl

(protein N-terminal) and oxidation (M); mass values: monoisotopic; protein mass: unrestricted; peptide mass tolerance: ±50 ppm; fragment mass tolerance: ±0.2 Da; max missed cleavages: 1; instrument type: MALDI-TOF. Only significant scores defined by Mascot probability analysis greater than "identity" were considered for assigning protein identity. All of the positive protein identification scores were significant (*P <* 0*.*05, score *>* 49).

### **BIOINFORMATICS**

To identify the sequences of all putative uncharacterized proteins, BLAST searches (http://www*.*expasy*.*org/tools/blast/)

with these protein sequences were performed on the UniProt Knowledgebase (UniProtKB, http://www*.*uniprot*.*org/uniprot) to find their homologs. Functional categorization of the identified proteins was based on annotations in UniProtKB and previous studies on their homologs. Subcellular locations of the identified proteins were determined according to the annotation in UniProtKB or predicted at Plant-mPLoc server (http://www*.*csbio*.*sjtu*.*edu*.*cn/bioinf/plant-multi/). Theoretical Mr and isoelectric point of proteins were predicted at http://web.expasy.org/compute\_pi/. ABA responsive element (ABRE) and dehydration responsive element (DRE) were analyzed using Plantcare (http://bioinformatics*.*psb*.*ugent*.*be/ webtools/plantcare/html/) and PLACE (http://www*.*dna*.*affrc*.*go*.* jp/PLACE/signalscan*.*html).

### **RESULTS**

Maize seed consists of an embryo (a miniature plant), an endosperm (a nutrition provider for seed germination) and a seed coat (protecting structure). To reveal maize seeds proteome alterations due to ABA-deficient mutation and to identify ABA-dependent proteins during seed maturation, embryos and endosperms of the mutant *vp5* and wild-type *Vp5* were used for comparative proteomic analysis. In order to improve protein resolution, two kinds of IPG strips with a pH range of 4–7 and 7–10 were used in 2-DE.

### **PROTEOMIC DIFFERENCE BETWEEN** *vp5* **AND** *Vp5* **EMBRYOS**

The embryo protein profiles between *vp5* and *Vp5* were compared by 2-DE. Approximately 780 ± 10 and 130 ± 5 CBB-stained protein spots were reproducibly detected using pH 4–7 and 7– 10 IPG gels, respectively (**Figure 1**). PDQUEST analysis indicated that 96% of total protein spots were matched, unchanged in abundance between *vp5* and *Vp5* embryos across all the gels (**Figure 1**, Presentations 2, 3 in Supplementary Material). Spotto-spot comparison revealed that 31 protein spots, i.e., 23 in pH 4–7 gels (**Figure 1A** and Presentation 2 in Supplementary Material) and 8 in pH 7–10 gel (**Figure 1B** and Presentation





**Table 4 | The differentially accumulated proteins identified in maize** *vp5* **and** *Vp5* **endosperms associated to putative functions.**

*aSubcellular location of proteins was annotated in UniProtKB/Swiss-Prot (http:// www.expasy.org/ ).*

*bSubcellular location of proteins was predicted using the online Plant-mPLoc server (http:// www.csbio.sjtu.edu.cn/ bioinf/ plant-multi/ ).*

3 in Supplementary Material), showed a minimum of a 1.5 fold difference in spot volume (**Tables 1**, **2**). Those differentially accumulated embryo protein spots are mainly in the range of 10–35 kDa, with a consistent change in three biological replicates, except for three spots in two biological replicates (spot 5, 17, and 22) (Presentation 2 in Supplementary Material). Sixteen embryo protein spots (spots 1–12 and 24–27) accumulated in higher abundance in *vp5* compared to *Vp5* and 15 embryo protein spots (spots 13–23 and 28–31) accumulated in lower abundance in *vp5*. In particular, the abundance of spots 1, 10, and 27 were 13.71, 7.31, and 9.28 folds higher in *vp5* embryos than in *Vp5* embryos, respectively, whereas the abundance of spots 19 and 23 was 6.76 and 6.52 folds higher in *Vp5* than in vp5 embryos, respectively.


### **Table 5 | Cis-acting elements of LEA and sHSPs proteins analyzed by Plantcare.**

### **Table 6 | Cis-acting elements of LEA and sHSPs proteins analyzed by Place.**


The 31 embryo proteins were successfully identified by MS/MS analysis, representing 26 distinct proteins in NCBI or SWISS-PROT protein databases (**Tables 1**, **2**). In several cases, proteins were identified as uncharacterized proteins. We searched their homologous proteins in other plant species by BLAST, such as spot 19 homologous to seed maturation protein of *Glycine latifolia* (identity of 40%) and spots 25, 26, and 28 homologous to lipoprotein-like of *Oryza sativa* (rice, identity of 81%), or were based on family and domain databases, such as spot 23 belonging to Bowman-Birk serine protease inhibitor family, and spot 27 belonging to group 3 LEA protein. These differentially accumulated protein spots were assigned to various LEA family proteins (spots 1, 13–16, 22, 27, and 29), HSP20 family proteins (spots 2–9 and 24), 1-Cys peroxiredoxin PER1 (spot 10), short-chain dehydrogenase/reductase SDR family protein (spot 11, 17, and 18), UDP-glucose 6-dehydrogenase (spot 12), seed maturation protein (spot 19), thioredoxin (spot 20), glyoxalase family protein (spot 21), Bowman-Birk serine protease inhibitor family protein (spot 23), lipoprotein-like (spot 25, 26, and 28), and globulin-1 S allele (spots 30 and 31). Nine sHSPs accumulated in higher abundance in *vp5* embryos, while six out of eight LEA proteins (except for spots 1 and 27) accumulated in higher abundance in *Vp5* embryos (**Figure 2**).

In the present study, two identified proteins belong to group 3 LEA protein (spot 27, K7VM99, spot 29 Q42376), but they exhibit opposite accumulation between *vp5* and *Vp5* embryos. In *vp5* embryos, Q42376, like other five identified LEA proteins, decreased in abundance, while K7VM99, like nine identified sHSPs, increased in abundance. This discrepancy may be explained by protein and gene sequence differences. K7VM99 and Q42376 share only a 35% identity in protein sequence alignment by BLAST.

In addition, we found EMB564 existed in two isoforms: a weak spot 1 and a most predominant spot (indicated in **Figure 1A**). These two isoforms displayed a contrast accumulation in *vp5* compared to *Vp5*: the weak isoform (spot 1) increased greatly whereas the other decreased a little. However, the total abundance of EMB564 was comparable between *vp5* and *Vp5*. This result contrasted to previously characterized, greatly lowered expression of *emb564* mRNA in *vp5* embryos (Williams and Tsang, 1991). In order to further confirm the accumulation level of EMB564 in the two genotypes, we examined the abundance of EMB564 using immunoblot analysis (**Figure 3**). The specificity of the antibody has been recently characterized, and it specifically reacts with EMB564 (Wu et al., 2013a). Obviously, EMB564 existed in comparable levels between *vp5* and *Vp5* embryos, with a slightly reduced level in *vp5*.

### **PROTEOMIC DIFFERENCE BETWEEN** *vp5* **AND** *Vp5* **ENDOSPERMS**

In preliminary 2-DE experiments, none obviously differentially accumulated endosperms protein spots were observed to exist when pH 7–10 IPG gels were used (data not shown). Therefore, endosperm proteome analysis was performed only using pH 4–7 IPG gels.

The endosperm protein profiles between *vp5* and *Vp5* were compared by 2-DE. Approximately 380 ± 10 CBB-stained protein spots were detected in endosperms, which was much less compared to embryos and most of proteins existed in low abundance (**Figure 4**). PDQUEST analysis indicated that 15 protein spots, especially spot 40, showed an obvious difference between *vp5* and *Vp5* endosperms, with a consistent change in three biological replicates (**Figure 4**, Presentation 4 in Supplementary Material). Seven spots (32–38) existed in higher abundance and eight spots (39–46) in lower abundance in *vp5*. These differentially accumulated protein spots were successfully identified by MS/MS (**Tables 3**, **4**), representing 11 unique protein species.

### **DISCUSSION**

### **GLOBAL PROTEOME ALTERATIONS IN EMBRYO AND ENDOSPERM OF MAIZE** *vp5* **AND** *Vp5*

Maize viviparous mutants, germinating directly on the ear (McCarty et al., 1991), are widely used for studying maize seed maturation, dormancy, and germination. Various viviparous mutants have been identified, such as *vp1 vp2 vp5 vp7 vp8* and *vp9*. Among them, *vp5* mutant is deficient in ABA biosynthesis with the first step catalyzed by phytoene desaturase being blocked, resulting in the precursor phytoene accumulation and carotenoid deficiency (Robichaud et al., 1980). However, the background difference at proteome level between *vp5* and *Vp5* seeds is still unclear.

In the present study, comparative proteomics was used to determine the variation of protein expression at the proteome level between maize mutant *vp5* and its wild type *Vp5*. There are great differences in the structure, composition and function between embryo and endosperm of maize seeds. Compared to endosperm, embryo is more active in nucleic acid, protein, and lipid metabolism. In total, 46 seed protein spots were found to be exhibited a differential change in abundance between these two genotypes, 31 spots in embryo and 15 spots in endosperm. Obviously, proteome alterations caused by ABA-deficient mutation are more significant in embryos than in endosperms in *vp5* seeds. This may be explained by two possible causes: Firstly, ABA deficiency mainly takes place in *vp5* embryos. Due to deficient in ABA biosynthesis, maize *vp5* embryos contains a low ABA content (about 10% of *Vp5*) throughout seed maturation, whereas 42% ABA in *vp5* endosperms (Neill et al., 1986). ABA in endosperm can be from maternal organs (Ober and Setter, 1992). Secondly, maize endosperms act mainly as starch storage tissue and contain fewer proteins in low amounts in mature seeds. Although ABA-deficient mutation affects the phenotype of *vp5* endosperm (carotenoid deficiency), proteins or enzymes involved in the related pathways were not detected, probably due to their low abundance and low sensitivity of CBB staining.

In the present study, four proteins, i.e., 1-Cys peroxiredoxin PER1 (spot 10 in embryo, spot 40 in endosperm), globulin-1 s allele (spot 30 and 31 in embryo, spot 42 in endosperm), glyoxalase family protein (spots 21 in embryo and 32 in endosperm) and short-chain dehydrogenase/reductase SDR family protein (spot 11, 17, and 18 in embryo, spot 39 in endosperm) were the same as identified embryo proteins. Among them, globulin-1 s allele (spot 30, 31, and 42) and short-chain dehydrogenase/reductase SDR family protein (spot 17, 18, and 39) decreased both in *vp5* embryo and endosperm compared to *Vp5*.

It is worth to note that EMB564, the most abundant LEA protein in maize embryos, was found to exist in comparable levels between *vp5* and *Vp5* embryos. EMB564 consists of two isoforms with obvious differences in apparent and theoretical values of sizes and isoelectric points, indicating a post-translational modification of this protein (e.g., phosphorylation, Amara et al., 2012). Williams and Tsang (1991) first cloned and characterized an embryo-specific recombinant, termed *emb564* [recently renamed as *embryo specific protein 1* (*esp1*) in NCBI database]. The *emb564* mRNA is expressed at low level in ABA-deficient (e.g., *vp5*) but not in ABA-insensitive (e.g., *vp1*) embryos during embryogenesis, and exogenous ABA has little effect on the accumulation of *emb564* mRNA in more mature embryos (Williams and Tsang, 1991). Therefore, there is a discrepancy between EMB564 protein and mRNA levels.

The transcript/protein discordance has been well documented in mammals, yeasts and plants (mainly Arabidopsis and maize). It is largely of biological origin ("true discordance") and of post-transcriptional regulation (Vélez-Bermúdez and Schmidt, 2014). For example, in *Arabidopsis* roots, a lack of correlation between down-regulated transcripts and the amount of their corresponding proteins was observed in response to phosphate deficiency, whereas for induced genes changes in the levels of mRNAs and proteins were reasonably well correlated (Lan et al., 2012). In the ABA-deficient *vp5* embryos, EMB564 can accumulate to a similar level as its wild type. Thus, this transcript/protein discordance is possibly caused by post-transcriptional regulation (via unknown factors) in an ABA-independent way.

Recently, we found that EMB564 is associated with maize seed vigor (Wu et al., 2011) and is highly thermal stable (Wang et al., 2012). By immunoelctron microscopy, we showed that EMB564 locates preferentially in the nucleus of maize embryonic cells (Wu et al., 2013a). Likewise, Amara et al. (2012) demonstrated that EMB564-green fluorescent protein fusions are expressed in the cytosol and nucleus in the agroinfiltrated leaves of *Nicotiana bentamiana*. Based on bioinformatics analysis, we proposed that EMB564 may function within the nucleus by binding DNA (Wu et al., 2013b).

In addition, we also noticed that some spots are not consistent in abundance between the biological replicates. This inconsistence results mainly from inherit drawbacks of 2-DE (e.g., poor reproducibility between gels) and from minor changes in manipulation during protein extraction among independent biological experiments. However, only those reproducibly, differentially accumulated proteins between *vp5* and *Vp5* are subjected to MS/MS analysis.

### **LEA PROTEINS AND sHSPs DISPLAYED DIFFERENTIAL ACCUMULATIONS IN ABA-DEFICIENT EMBRYOS**

The most interesting finding in this study is that most LEA proteins and sHSPs displayed differential accumulations in ABAdeficient *vp5* embryos: six out of eight identified LEA proteins decreased while nine identified sHSPs increased in abundance, compared to *Vp5*.

LEA proteins are characterized by high hydrophilicity and accumulate to high levels during the last stage of seed maturation (Dure et al., 1989; Espelund et al., 1992). Although the roles of LEA proteins remain speculative, there is evidence supporting their participation in acclimation and/or in the adaptive response to dehydration, low temperature, salinity, or exogenous ABA treatment stress (Battaglia et al., 2008). sHSPs are produced in seeds during maturation and under various stress conditions. The synthesis of sHSPs during seed maturation indicates their probable role in cell component protection mechanisms. Mutants sensitive to desiccation contain smaller amounts of sHSPs during maturation (Wehmeyer and Vierling, 2000).

We tried to explain the possible causes of the observed differential accumulation of six identified LEA proteins and nine identified sHSPs between *vp5* and *Vp5* seeds. Firstly, the threshold content of ABA inducing the accumulation of LEA proteins and sHSPs may differ: some LEA proteins may require higher ABA content than sHSPs during seed maturation. ABA-deficiency in *vp5* embryos greatly inhibited the accumulation of most LEA proteins, but promoted the accumulation of sHSPs. In another words, compared to developmentally specific accumulation of LEA proteins, sHSPs seemed to be less strictly ABA-dependent. In the ABA-deficient *vp5* embryos, EMB564 accumulates to a similar level as its wild type, implying an ABA-independent accumulation of EMB564 in maize embryos. This deduction will be examined by measuring the ABA content and the time-course of accumulation of LEA proteins and sHSPs during *vp5* seed development in future. Previously, differential regulation of ABA-induced 23– 25 kDa proteins in embryo and vegetative tissues of maize *vp* mutants was reported (Pla et al., 1989). In a recent study, we identified several proteins in maize *Vp5* and *vp5* seedlings whose syntheses are ABA-independent, such as ADP-dependent malic enzyme and fructose-bisphosphate aldolase (Hu et al., 2012). Maize seeds undergo dehydration process during maturation. ABA mutant *vp5* seeds contain low ABA content and lack obvious dehydration process; therefore, the accumulation of strict ABAdependent protein (e.g., LEA proteins in this study) may reduce, and not strict ABA-dependent proteins (e.g., sHSPs in this study) may increase to enhance seed dehydration/drought tolerance.

We checked whether there was any difference in the promoter region, especially regarding ABRE and DRE, between differentially accumulated proteins identified here (**Tables 5**, **6**). Most LEA genes are sensitive to ABA because of the presence of ABRE, i.e., regulatory elements in promoter regions, which contain the ACGT sequence called cassette G. Sensitivity to ABA also depends on the presence of MYC elements that include the sequence CACCTG, and MYB elements that include the TAACTG motive (Kalemba and Pukacka, 2007). Dehydration responsive element, DRE, is 9-bp consensus sequence, TACCGACAT, was first identified in the promoter of Arabidopsis rd29A/lti78 and shown to be essential for drought induction in the absence of ABA (Yamaguchi-Shinozaki and Shinozaki, 1994).

In addition, the accumulation pattern of group 3 LEA protein (spot 27, K7VM99) in *vp5* embryo is different with other five identified LEA proteins. The group 3 LEA proteins are characterized by a repeating motif of 11 amino acids (TAQAAKEKAGE) (Dure et al., 1989), which are quite diverse in sequence structure compared with other LEA groups (Battaglia et al., 2008). In maize *vp5* seeds, group 3 LEA protein accumulation is dependent upon ABA (Thomann et al., 1992), whereas corresponding mRNA has no response to exogenous ABA (White and Rivin, 2000). In maize leaves, *group 3 LEA gene* can be induced by ABA (Liu et al., 2013). Cis-acting elements analysis showed that group 3 LEA protein (K7VM99) shares more identical ABRE and/or DRE with sHSPs (e.g., HSP 17.0) than other LEA proteins (**Tables 5, 6**). This may explain why group 3 LEA protein (K7VM99) differentially accumulates in *vp5* embryos, like sHSPs but unlike other LEA proteins.

In conclusion, comparative gel-based proteomics revealed significant proteome difference between *vp5* and *Vp5* seeds. Most notably, LEA proteins and sHSPs displayed a differential accumulation pattern in ABA-deficient *vp5* embryos. The data derived from the present study is highly applicable to other crops. The characterization of proteome difference between *vp5* and *Vp5* seeds is necessary for dissection of ABA-mediated maize response in the studies involved *vp5* mutants. The data derived from this study provides insight into ABA-dependent proteins and ABA-mediated response during seed maturation in maize.

### **ACKNOWLEDGMENTS**

This research was supported by National Natural Science Foundation of China (grant 31371543 and 31171470), Plan for Scientific Innovation Talent of Henan Province (grant 144200510012) and State Key Laboratory of Crop Biology (2012KF01) at Shandong Agricultural University, China.

### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at: http://www*.*frontiersin*.*org/journal/10*.*3389/fpls*.*2014*.*00801/ abstract

### **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: 20 October 2014; paper pending published: 27 November 2014; accepted: 22 December 2014; published online: 20 January 2015.*

*Citation: Wu X, Gong F, Yang L, Hu X, Tai F and Wang W (2015) Proteomic analysis reveals differential accumulation of small heat shock proteins and late embryogenesis abundant proteins between ABA-deficient mutant vp5 seeds and wild-type Vp5 seeds in maize. Front. Plant Sci. 5:801. doi: 10.3389/fpls.2014.00801*

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

*Copyright © 2015 Wu, Gong, Yang, Hu, Tai and Wang. 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.*

# Low temperature conditioning of garlic (*Allium sativum L.*) "seed" cloves induces alterations in sprouts proteome

### Miguel D. Dufoo-Hurtado<sup>1</sup> , José Á. Huerta-Ocampo2 †, Alberto Barrera-Pacheco<sup>2</sup> , Ana P. Barba de la Rosa<sup>2</sup> and Edmundo M. Mercado-Silva<sup>1</sup> \*

<sup>1</sup> Laboratorio de Fisiología y Bioquímica Poscosecha de Frutas y Hortalizas, Departamento de Investigación y Posgrado, Facultad de Química, Universidad Autónoma de Querétaro, Querétaro, Mexico, <sup>2</sup> Laboratorio de Proteómica y Biomedicina Molecular, División de Biología Molecular, Instituto Potosino de Investigación Científica y Tecnológica A.C., San Luis Potosí, Mexico

Low-temperature conditioning of garlic "seed" cloves substitutes the initial climatic requirements of the crop and accelerates the cycle. We have reported that "seed" bulbs from "Coreano" variety conditioned at 5◦C for 5 weeks reduces growth and plant weight as well as the crop yields and increases the synthesis of phenolic compounds and anthocyanins. Therefore, this treatment suggests a cold stress. Plant acclimation to stress is associated with deep changes in proteome composition. Since proteins are directly involved in plant stress response, proteomics studies can significantly contribute to unravel the possible relationships between protein abundance and plant stress acclimation. The aim of this work was to study the changes in the protein profiles of garlic "seed" cloves subjected to conditioning at low-temperature using proteomics approach. Two sets of garlic bulbs were used, one set was stored at room temperature (23◦C), and the other was conditioned at low temperature (5◦C) for 5 weeks. Total soluble proteins were extracted from sprouts of cloves and separated by two-dimensional gel electrophoresis. Protein spots showing statistically significant changes in abundance were analyzed by LC-ESI-MS/MS and identified by database search analysis using the Mascot search engine. The results revealed that low-temperature conditioning of garlic "seed" cloves causes alterations in the accumulation of proteins involved in different physiological processes such as cellular growth, antioxidative/oxidative state, macromolecules transport, protein folding and transcription regulation process. The metabolic pathways affected include protein biosynthesis and quality control system, photosynthesis, photorespiration, energy production, and carbohydrate and nucleotide metabolism. These processes can work cooperatively to establish a new cellular homeostasis that might be related with the physiological and biochemical changes observed in previous studies.

Keywords: *Allium sativum*, sprouts, cold conditioning, two-dimensional electrophoresis, LC-ESI-MS/MS

*Edited by:*

Sabine Lüthje, University of Hamburg, Germany

### *Reviewed by:*

Harriet T. Parsons, University of Copenhagen, Denmark Giridara Kumar Surabhi, Regional Plant Resource Centre, India

### *\*Correspondence:*

Edmundo M. Mercado-Silva, Laboratorio de Fisiología y Bioquímica Poscosecha de Frutas y Hortalizas, Departamento de Investigación y Posgrado, Facultad de Química, Universidad Autónoma de Querétaro, Cerro de las Campanas s/n. Col. Las Campanas, Querétaro, Querétaro 76010, Mexico mercado501120@gmail.com; mercasilva20@yahoo.com.mx

### †*Present Address:*

José Á. Huerta-Ocampo, Programa Cátedras CONACYT, Laboratorio de Bioquímica de Proteínas y Glicanos, Coordinación de Ciencia de los Alimentos, Centro de Investigación en Alimentación y Desarrollo A.C., Hermosillo, Mexico

### *Specialty section:*

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

*Received:* 31 December 2014 *Accepted:* 27 April 2015 *Published:* 13 May 2015

### *Citation:*

Dufoo-Hurtado MD, Huerta-Ocampo JÁ, Barrera-Pacheco A, Barba de la Rosa AP and Mercado-Silva EM (2015) Low temperature conditioning of garlic (Allium sativum L.) "seed" cloves induces alterations in sprouts proteome. Front. Plant Sci. 6:332. doi: 10.3389/fpls.2015.00332

# Introduction

Garlic (Allium sativum) is one of the most economically important Allium species, and has been widely cultivated for more than 5000 years. Garlic bulbs have been used as condiments, spices, seasonings, or flavoring as well as for its medicinal value, while garlic leaves are consumed as green vegetables (Ade-Ademilua et al., 2009). The garlic bulb consists of a discshaped stem that supports the cloves and it is surrounded by the dried basal sheaths of the foliage leaves. Cloves are considered organs of propagation of the specie and contain a set of leaves, including the storage leaf and the vegetative bud that contains a predominant sprout leaf and several primordia leaves that surround the apical meristem, which activates their growth under certain environmental conditions to generate a new plant.

Growth period and bulb production differ greatly from year to year, planting date and location, because there is a strong genotype-environment interaction. Cold storage, growth temperature, and photoperiod are the main environmental factors affecting the ontogeny in this crop, but the response to these factors depend on phenological stage (Takagi, 1990). The emergence of the sprout is controlled mainly by temperature, in absent of dormancy, whereas bulb initiation is promoted by exposure of cloves to low temperature (environmental or cold conditioning), growth temperature and photoperiod.

Kamenetsky et al. (2004) also showed that the temperature and photoperiod, strongly affect garlic morphology and plant development as well as the leaf elongation, clove formation, and dormancy induction, indicating that the environmental regulatory effect is obligatory and yet quantitative in certain varieties. However, some varieties may show different cold requirements, and the same variety can show differences in different growing areas affecting the commercial production of this crop. Which indicates that the length of storage period as well as the temperature used will have different effects on the responses of garlic plant. In addition to their immediate effects, environmental factors also have long-term effect in each of the development stages.

Maintaining the bulbs before planting at temperatures from 0 to 10◦C for a period of 2 months accelerates the cycle and substitutes the initial climatic requirements of the crop. The exposure of garlic "seed" bulbs to low temperatures modifies the hormonal balance, which leads to an early plant development. "Seed" bulbs without cold storage develop only when they receive suitable light period and temperature conditions for their requirements, whereas cold treated seed bulbs begin to develop more quickly at high temperatures than at low temperatures.

The most striking effect of cold conditioning is the increase in earliness, especially in cultivars with greater low temperature requirement for development. However, there are many contradictions about the effects of cold conditioning on the garlic yield; some researchers have observed increases while other reported no significant differences and yet others have reported depressive effects. We have observed that in "seed" bulbs from "Coreano" variety that were conditioned for 16 days at 5◦C, there were no effects on the yield and bulbs quality. However, the conditioning during 5 weeks at 5◦C accelerated the crop cycle, decreased plant height and increased the synthesis of phenolic compounds and anthocyanins in the outer scale leaves of the bulbs at harvest time compared to plants from "seed" bulbs stored at room temperature. That indicates a possible stress for low temperature (Dufoo-Hurtado et al., 2013; Guevara-Figueroa et al., 2015). Nevertheless, knowledge about what happens during cold conditioning at functional and biochemical levels in garlic is very limited.

Low temperature is a major environmental constraint associated with many structural, physiological, and biochemical changes within plant cells, as well as with altered gene expression patterns (Kosmala et al., 2009). While genomic and transcript-profiling studies have provided a wealth of information about the process of cold acclimation, there is a growing recognition that the abundance of mRNA transcripts is not always a representative of cognate protein levels and that mechanisms of posttranslational regulation must also play an important role (Renaut et al., 2006). Consequently, proteomics provides a complementary approach between the classical physiological approach and molecular tools, especially for research using non-model plant species, where no genomic sequencing data is available or it is limited. Results from proteomic analysis revealed different proteins associated with the plant response to low-temperature environment by being newly synthesized, accumulating or decreasing. Among other pathways, the differentially expressed proteins are involved in signaling, translation, host-defense mechanisms, carbohydrate metabolism and amino acid metabolism, including both welldocumented stress-responsive proteins and some novel coldresponsive proteins (Renaut et al., 2006; Kosová et al., 2011). These results have demonstrated the power of the proteomic approach in studies of plant response to low-temperature conditions.

Identification of the changes in the garlic proteome will contribute to the elucidation of the physiological and plant developmental modifications induced by cold conditioning of garlic "seed" cloves. Therefore, the aim of this work was to apply the proteomics tools (two-dimensional gel electrophoresis coupled with tandem mass spectrometry) to study the changes in the protein profiles of garlic "seed" cloves subjected to lowtemperature conditioning.

**Abbreviations:** ADK2, Adenosine kinase 2; ANX2, Annexin D2-like; CBS, Cystathionine β-synthase; CP, Coat protein; DHAR, Dehydroascorbate reductase; FBPA, Fructose-bisphosphate aldolase; FTS, Ferredoxin-thioredoxins system; GarCLV, Garlic common latent virus; GR-RBP, Glycine-rich RNA-binding protein; GST, Glutathione S-transferase; hnRNP, Heterogeneous nuclear ribonucleoprotein; HSP, Heat shock protein; iPGAM, 2,3-bisphosphoglycerateindependent phosphoglycerate mutase; LEA, Late embryogenesis abundant; MDH, Malate dehydrogenase; MIF, Macrophage migration inhibitory factor; NAC, Nascent polypeptide-associated complex; NTF, Nuclear transport factors; PAI1, Plasminogen activator inhibitor 1 RNA-binding protein-like protein; PAPB, Polyadenylate-binding protein; PPIase, Peptidyl-prolyl cis-trans isomerase; RRM, RNA-recognition motif; RT, Conditioning at room temperature; RuBisCO, Ribulose-1,5-bisphosphate carboxylase/oxygenase; SAM, S-adenosyl-Lmethionine; SAMS, S-adenosylmethionine synthase; SnRK1, SNF1-related kinase; SR, Serine/arginine-rich splicing factor; SSU, RuBisCO small subunit; TCA, Tricarboxylic acid cycle.

# Material and Methods

# Plant Materials and Low-temperature Conditioning

"Seed" bulbs of garlic (A. sativum, L.) cv. "Coreano" were provided by the Garlic Producer Association from Aguascalientes and cultivated at Cosio, Aguascalientes, Mexico, during the crop cycle 2011–2012. One hundred and fifty bulbs of garlic cv. "Coreano" harvested in the 2012-2013 season were stored during 4 months at room temperature. These bulbs were separated into two sets of 75 bulbs; one set was maintained at room temperature (RT) (23◦C), and the other was conditioned during 5 weeks at low temperature (5◦C), both treatments were under dark conditions. At the end of this period, three sets of 25 bulbs (three biological replicates) were separated and threshed and the cloves of medium size (5–6 g) were selected. The garlic sprouts (leaf primordia) of fifteen selected cloves, of each treatment and replicate were collected and pooled and used for the analysis. Each set was frozen in liquid nitrogen and stored at −70◦C until their analysis. From each biological replicate, two technical replicates were carried out and processed as independent samples.

# Protein Sample Preparation

Sprout proteins were extracted from three independent biological replicates, each sprout pool was milled in a coffee grinder (Braun, Naucalpan, Mexico) to a fine powder. Three grams of powder were mixed with 20 mL of cold acetone. The suspension was centrifuged 10 min at 7650× g and 4◦C (Super T21; Sorvall, Kendro Laboratory Products, Newtown, CT, USA.). The pellet was washed with 20 mL of cold acetone and centrifuged at the same conditions. The pellet was suspended in 6 mL of extraction buffer (8M urea, 2% w/v 3- [(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate, 20 mM dithiothreitol, 2 mM phenylmethylsulfonyl fluoride, 0.002% w/v bromophenol blue). The suspension was mixed in a vortex for 1 min, sonicated during 150 s at 35% of amplitude (GE-505, Ultrasonic Processor, Sonics & Materials, Inc., Newtown, CT, USA); the suspension was centrifuged 20 min at 20,220× g at 4◦C. The supernatant was filtered with Miracloth (Calbiochem, Darmstadt, Germany) and centrifuged under the same conditions. The supernatant was subjected to protein clean-up according to manufacturer's instructions (ReadyPrep™ 2-D Cleanup Kit, Bio-Rad, Hercules, CA, USA.). The protein pellet was suspended in rehydration buffer (8 M urea, 2% w/v 3- [(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate, 20 mM dithiothreitol, 0.002% w/v bromophenol blue), mixed in vortex for 30 s and sonicated for 80 s. The suspension was centrifuged under the previous conditions and the supernatant was recovered. Protein concentration was determined by using protein assay (Bio-Rad) with bovine serum albumin (BSA) used as standard.

# Two-dimensional Polyacrylamide Gel Electrophoresis and Image Analysis

Sprout soluble proteins (1.5 mg) were suspended in 450µL of rehydration buffer, containing 0.5% v/v IPG buffer pH 4-7 (Bio-Rad), and were loaded onto 24 cm strips, linear immobilized pH gradient 4–7 (IPG) strips (Bio-Rad). Passive rehydration was carried out at room temperature during 14– 16 h. The IEF was conducted at 50 mA per IPG strip and 20 ◦C in an Ettan IPGphor system (GE Healthcare, Piscataway, NJ, USA). The IEF conditions were: (I) 200 V gradient for 2 h; (II) 400 V gradient for 2 h; (III) 1500 V gradient for 2 h; (IV) 4500 V gradient for 3 h; (V) 8000 V gradient for 3 h, and (VI) holding at 8000 V for 10 h. After isoelectric focusing, the IPG strips were stored at −20◦C or immediately equilibrated for 15 min in equilibration buffer (50 mM Tris– HCl pH 8.8, 6 M Urea, 30% v/v glycerol, 2% w/v SDS, 0.002% w/v bromophenol blue, 65 mM dithiothreitol). Strips were transferred to a vertical SDS-polyacrylamide gel. The second dimension was performed on 13% polyacrylamide-SDS gels using an Ettan DALT six electrophoresis unit (GE Healthcare), by SDS electrophoresis buffer (25 mM Tris pH 8.8, 192 mM glycine, and 0.1% w/v SDS) and resolved at 20 mA/gel until the dye (bromophenol blue) reached the bottom of the gels. Two technical replicates per each biological replicates were run for both treatments. After SDS–PAGE, gels were stained with PhastGel Blue R (GE Healthcare) and scanned at 100µm resolution with a Pharos FX plus Molecular Imager (Bio-Rad). Images were analyzed with Melanie v7.0 software (GeneBio, Geneva, Switzerland). Gel image analysis included spot detection, spot measurement, background subtraction, and spot matching. To correct the variability and to reflect quantitative variation, the spot volumes were normalized as the percentage of the total volume of all spots in the gel. The molecular masses of proteins in gels were determined by co-electrophoresis of molecular weight standards (BenchMark Protein Ladder, Invitrogen. Carlsbad, CA, USA), and the isoelectric point of proteins was determined by migration of protein spots on 24 cm IPG linear gradient strips (pH 4– 7). Protein spots were considered as differentially accumulated when their normalized volumes displayed a fold change =1.5 when controls and treatments (RT and 5◦C, respectively) were compared. Significant changes were determined using t-test (p < 0.01).

# In-gel Digestion and Tandem Mass Spectrometry Analysis (LC-ESI-MS/MS)

Selected protein spots were excised from the gels and distained, reduced with 10 mM dithiothreitol in 25 mM ammonium bicarbonate followed by protein alkylation with 55 mM iodoacetamide. Protein digestion was carried out overnight at 37◦C with sequencing grade trypsin (Promega, Madison, WI, USA). Nanoscale LC separation of tryptic peptides was performed with a nanoACQUITY UPLC System (Waters, Milford, MA, USA) and tandem mass spectrometry analysis carried out in a SYNAPT HDMS (Waters) as previously reported (Huerta-Ocampo et al., 2014) with a brief modification: Accurate mass data were collected in an alternating data dependent acquisition mode (DDA). In low energy mode, data were collected at constant collision energy of 3 eV. In elevated-energy mode, the collision energy was ramped from 15 to 45 eV during 5 s of integration.

### Protein Identification and Classification

MS/MS spectra data sets were used to generate PKL files using Protein Lynx Global Server v2.4 (PLGS, Waters). Proteins were then identified using PKL files and the Mascot search engine v2.3 (Matrix Science, London, UK). Searches were conducted against the Viridiplantae subset of the NCBInr protein database (2 391 213 sequences, October 2013) and against the A. sativum subset of the NCBInr EST database (21,694 sequences, October 2013). Trypsin was used as the specific protease, and one missed cleavage was allowed. The mass tolerance for precursor and fragment ions was set to 10 ppm and 0.1 Da, respectively. Carbamidomethyl cysteine was set as fixed modification and oxidation of methionine was specified as variable modification. Significant Mascot scores (>39 for the Viridiplantae subset of the NCBInr protein database or >33 for A. sativum subset of the NCBInr EST database) indicating the identity or extensive homology at p < 0.05 and the presence of at least two peptides were considered necessary for reliable identification. Identified proteins were classified into different categories of biological processes in which they were involved according to Gene Ontology (http://www.geneontology. org/).

# Results and Discussion

### Identification of Low-temperature Conditioning Responsive Proteins in Garlic Sprouts

Four hundred eighty-five protein spots were reproducibly detected in RT samples and conditioned at 5◦C. Sixty two spots showed statistically significant differences at ratios over 1.5-fold in relation with cold conditioning. **Figure 1** shows the position of the 62 differentially accumulated protein spots and close-ups of some of the changes in protein spot abundance between treatments. The reproducibility among the technical and biological replicates of the 2-DE gels is shown in magnified images (Supplementary Figures S1–S3)

Among the differentially accumulated protein spots, 22 increased, whereas 37 decreased significantly in abundance in response to low-temperature conditioning. Interestingly three spots appeared only after low-temperature conditioning. All differentially accumulated protein spots were excised from 2-DE gels and subsequently subjected to LC-ESI-MS/MS analysis. Fifty spots (81%) were successfully identified (p < 0.05) while in 12 cases (19%) the identification was not possible (spots 1, 2, 28, 29, 30, 40, 48, 50, 52, 56, 58, and 60). In four cases, the same spot

matched to more than one protein (**Table 1**, Supplementary Table S1).

Considering the number of proteins identified and according to Gene Ontology these were grouped into nine different categories according to the biological processes in which they are involved (**Table 1**, **Figure 2**): cellular response to stress, carbohydrate binding, regulation of transcription, transport of macromolecules, protein folding, photosynthesis, carbohydrate metabolism, nucleotide metabolism, and miscellaneous. Many of the low-temperature conditioning-responsive proteins were isoforms with a change in pI and/or molecular weight strongly suggesting that the low-temperature conditioning-induced posttranslational modification and translation from alternatively spliced mRNAs of the candidate proteins, potentially including isoforms of multigenic families of proteins.

### Cellular Response to Stress Related-proteins

Thirteen of the differentially accumulated protein spots were classified as related to cellular response to stress (**Table 1**). Spot 6, which decreased under low-temperature conditioning, was identified as the macrophage migration inhibitory factor (MIF) homolog isoform X3. In animal tissues MIF has been assigned different functions including cytokine activity, hormonal peptide, growth stimulant and extracellular oxidoreductase showing phenylpyruvate tautomerase, and dopachrome tautomerase activities (Kudrin and Ray, 2008). Whereas in plant tissues, MIF has been associated with defense mechanisms (Reumann et al., 2007) and as cytokine. Cytokines are secreted by stem cells and regulate plant immunity through cell-cell communication and reprogramming of the expression of immunity related genes (Luo, 2012). Cytokines not only enhance the innate immunity but also regulate division and differentiation of shoot apical meristem cells (Lee et al., 2011). However, the functional role of MIF could be more complex. Previous studies have indicated that MIF binds other proteins as Jab1 (Jun-activation-domain binding Protein) and CSN (COP9 signalosome), which are important in signaling pathways, as well as positive regulators of the cell cycle and protein degradation (Kleemann et al., 2000 and Bech-Otschir et al., 2002). The low accumulation of this protein could be related with the lower cell number and higher cell size in garlic bulbs from plants obtained from "seed" cloves stored 30 days at 4◦C (Rahim and Fordham, 1988) as well as the lower height observed for our group when plants were conditioned at 5◦C (Dufoo-Hurtado et al., 2013; Guevara-Figueroa et al., 2015).

Spots 17 and 18, with similar mass and pI, were identified as cystathionine β-synthase domain-containing protein (CBS); which showed opposite accumulation patterns that could indicate the same protein with some posttranslational modifications. Kushwaha et al. (2009) identified 34 CBS domaincontaining proteins (CDCPs) in Arabidopsis and 59 in Oryza indicating that the CBS domain coexists with other functional domains. In addition, it was reported that CBS are differentially accumulated under cold stress conditions. Yoo et al. (2011) showed in Arabidopsis thaliana that proteins consisting of a single CBS domain pair stabilize cellular redox homeostasis and modulate plant development via regulation of ferredoxinthioredoxin system (FTS). These CDCPs activate the thioredoxin in the FTS in chloroplast or in NADP-thioredoxin system in the mitochondria and thereby controls H2O<sup>2</sup> levels. These authors showed that mutants with 35S:CBSX1 gene overexpressed had defective lignin deposition indicating that these CBS proteins affect the plant growth. Taking into account these observations, the over accumulation of one CBS protein in 5◦C sprouts could indicate alterations in lignin deposition which will modify the plant growth that was reported by our group (Dufoo-Hurtado et al., 2013; Guevara-Figueroa et al., 2015).

Plant glutathione S-transferases (GSTs) have also long been associated with responses to biotic and abiotic stress, plant development and metabolism. Six spots (spots 21–26) that were down-accumulated in garlic sprouts in response to lowtemperature conditioning were identified as GSTs. GST superfamily is categorized into evolutionarily distinct classes. One of them is dehydroascorbate reductase (DHAR) class (spots 21– 24), which is responsible for regenerating the ascorbic acid from an oxidized state maintaining the cellular ascorbic redox state allowing a greater tolerance to reactive oxygen species and better cell performance. The down-accumulation of this DHAR in samples conditioned at 5◦C suggests an early oxidative stress, which could modify the plant development on the field. Chen and Gallie (2006) showed that suppression of DHAR expression in tobacco plants induced lower plant height and number of leaves, reduction of photosynthetic function, and premature leaf aging in comparison with control plants. The low leaves number and low width leaves as well as the lower height and weight of garlic plants from bulbs conditioned at 5◦C observed for our group (Dufoo-Hurtado et al., 2013; Guevara-Figueroa et al., 2015) could be partially explained by the lower accumulation of these proteins. GST proteins also are involved in the synthesis of sulfur-containing secondary metabolites such as volatiles and glucosinolates, as well as in the conjugation, transport and storage of reactive oxylipins, phenolics, and flavonoids (Dixon and Edwards, 2010). The largest proportion of GSTs proteins in RT samples may also indicate more thiosulfinates synthesis (garlic aroma compounds) compared to samples from conditioning at 5◦C. On the other hand, GST proteins also are required as cytoplasmic carrier proteins binding to flavonoids in order to deliver them into vacuoles and maintain the flavonoid pool under stress conditions (Kitamura, 2006). GSTs proteins in the samples from low-temperature conditioning could be involved in an active transport of anthocyanin and phenolic compounds into vacuoles, which increased in garlic bulbs obtained from cloves conditioned at low-temperature (Dufoo-Hurtado et al., 2013).

Increased accumulation of S-adenosylmethionine synthase (SAMS; spots 38, 45, and 61) in garlic sprouts under cold conditioning was observed. SAMS catalyzes the production of S-adenosyl-L-methionine (SAM) from L-methionine and ATP. SAM is a precursor of ethylene and polyamines whose levels significantly rise upon cold (Kosová et al., 2011) in addition, SAM is also a methyl donor potentially regulating recovery-related DNA/protein methylation and derivatives of the phenylpropanoid pathway (Cui et al., 2005; Chen et al., 2013). Enhanced accumulation of SAMS has been reported in response to cold stress (Cui et al., 2005; Amme et al., 2006; Yan et al., 2006; Chen et al., 2013). The enhanced accumulation of SAMS protein

### TABLE 1 | Identification of differentially accumulated proteins in garlic (*Allium sativum* L.) sprouts subjected to low-temperature conditioning.





<sup>a</sup> Spot numbers as indicated in *Figure 1* ; spot numbers in parentheses indicate that the same spot matched to distinct proteins. <sup>b</sup> Plant species represent the most likely orthologous organisms reported in the NCBInr protein database. <sup>c</sup> Accession numbers according to NCBInr databse. When the best matches were against the Allium sativum subset of the NCBInr EST database, the most likely orthologous obtained after BLASTX against the NCBInr protein database are reported. <sup>d</sup> Experimental mass (kDa) and pI of identified protein spots. <sup>e</sup> Theoretical mass (kDa) and pI of identified proteins retrieved from NCBInr database or after calculation using the compute pI/Mw tool (http://web.expasy.org/compute\_pi/). <sup>f</sup> Mascot score reported after database search. Individual ion scores >23 (for the Allium sativum subset of the NCBInr EST database) or >39 (for the Viridiplantae subset of the NCBInr protein database) are statistically significant (p < 0.05). <sup>g</sup> Number of peptides matched. <sup>h</sup> Sequence coverage. <sup>i</sup> Database. As, Allium sativum subset of the NCBInr EST database; NCBI, Viridiplantae subset of the NCBInr protein database. <sup>j</sup> Fold change is expressed as the ratio of the volume % between low-temperature/room-temperature conditioned sprouts, and each value represents the mean value of three biologically independent replicates analyzed by duplicate. For some spots, fold change cannot be accurately calculated because of a complete absence of the spot in room temeprature conditioned samples; this is noted by the symbols <sup>√</sup> , indicating the presence of the spot only in low-temparature conditioned samples. <sup>k</sup> Protein spot accumulation changes in Allium sativum sprouts proteins conditioned at room (23◦C) or low-temperature (5◦C) for 5 weeks. Each column represents the mean value of three biologically independent replicates analyzed by duplicate. Error bars indicate ± standard error (SE).

induced by the cold conditioning could be linked to an enhanced expression of the phenylpropanoid and flavonoid pathways genes in garlic tissues, which has also been previously reported by our group (Dufoo-Hurtado et al., 2013).

Spot 53 was identified as probable protein Pop3 showing an increased accumulation in response to low-temperature conditioning. Pop3 proteins represent a new class of proteins involved in the plant response to abiotic stress. They are hydrophilic like late embryogenesis abundant (LEA)-type proteins, and exhibit the oligomeric structure of heat shock proteins (HSPs; Wang et al., 2002). It has been shown that cold stress induced the accumulation of Pop3 (Renaut et al., 2005).

### Carbohydrate-binding Proteins

Lectins are carbohydrate-binding proteins that specifically recognize and reversibly bind to specific free sugars or glycans present on glycoproteins and glycolipids without altering the structure of the carbohydrate and mediate a variety of biological processes (Lannoo and Van Damme, 2010). It has also been described that garlic mannose-specific lectins are the second most abundant proteins present in garlic cloves (Rabinkov et al., 1995). Ten spots (spots 3–5, 7–11, 51, and 54) were identified as garlic lectins, only one spot (spot 9) increased under cold conditioning, whereas the other nine spots decreased significantly. These results may indicate that cold conditioning of "seed" cloves could alter the metabolism of glycoproteins and glycolipids and their lectins. On the other hand, spot 9 could be associated with a lectin stress as indicated by Lannoo and Van Damme (2010) and Komath et al. (2006). In the meanwhile, since a number of lectins have been isolated from storage tissues in plants, it has been speculated that lectins might serve as plant storage proteins being important sources of nitrogen. These proteins could be degraded during the development process or bound to some non-polar molecules such as growth factors promoting the regulation of cell division (Komath et al., 2006). Therefore, a lower accumulation of these proteins during cold conditioning of "seed" garlic cloves at 5◦C could be related with the less height, less number and width of leaves reported by Dufoo-Hurtado et al. (2013) and Guevara-Figueroa et al. (2015).

### Proteins Related to Regulation of Transcription

Nine of the differentially accumulated protein spots were classified as related to regulation of transcription (**Table 1**). Among them, only one spot (spot 39) decreased under cold conditioning and was identified as plasminogen activator inhibitor 1 RNA-binding protein-like protein (PAI1). On the contrary, eight spots were significantly increased in response to cold conditioning in garlic sprouts. Five spots (spots 13–16, and 55) were identified as glycine-rich RNA-binding proteins (GR-RBPs). One spot (spot 33) was identified as serine/arginine-rich

splicing factor (SR) involved in splicing of pre-mRNA and stress responses (Reddy and Shad Ali, 2011). Spot 36 was identified as heterogeneous nuclear ribonucleoprotein (hnRNP) that could be involved in mRNA transport (Kawamura et al., 2002). Spot 59 was identified as polyadenylate-binding protein (PAPB), which plays critical roles in eukaryotic translation initiation and mRNA stabilization/degradation (Deo et al., 1999). These proteins (GR-RBPs, SR, hnRNP, and PAPB) are characterized by the presence of RNA-recognition motifs (RRMs) and are described as RNA-binding proteins (RBPs). Nowadays there is evidence that suggests that RNA-binding proteins are involved in a series of abiotic stresses responses, including cold adaptation (Kim et al., 2007). The RNA-binding activity of those proteins has been biochemically demonstrated, suggesting that they might be involved in the regulation of pre-mRNA alternative splicing, mRNA export, mRNA translation, mRNA decay, long term storage of "masked messages," and perhaps even mRNA localization. These functions surely facilitate a more rapid and effective response to a severe biotic and abiotic stress (Mangeon et al., 2010; Ciuzan et al., 2015). It is important to note that the changes in these regulatory proteins are expressed in meristematic tissue of "seed" cloves during the cold conditioning and those changes will have a significant effect during plant development.

### Proteins Related to Transport of Macromolecules

Nascent polypeptide-associated complex subunit (NAC, spot 27) was significantly accumulated after cold conditioning. NAC proteins bind to newly synthesized polypeptide chains from ribosome to protect them from proteolysis and to facilitate their folding. Through interaction with signal recognition particles, NAC also participates in the transport to endoplasmic reticulum of newly synthetized proteins, but there also exist evidence that this protein shifts from the soluble to the insoluble aggregated protein fraction during cell aging (Park et al., 2012). This protein acts as a modulator of protein synthesis and it is a key regulator of protein homeostasis (proteostasis) during the response to abiotic stress (Kirstein-Miles et al., 2013; Kogan and Gvozdev, 2014). On the other hand, it has also been reported that silencing of the NAC gene induced tolerance to freezing and drought stress exposures in wheat (Kang et al., 2013), which seems to be different in garlic sprouts.

Spot 34 decreased in abundance in cold conditioning samples and it was identified as annexin D2-like (ANX2). Annexins are soluble proteins and are defined as multifunctional Ca2<sup>+</sup> and lipid-binding proteins. Plant annexins are expressed throughout the life cycle and have been detected in all organs. It is estimated that annexins can comprise 0.1% of plant cell protein. Plant annexin expression and localization appear linked to growth and development (Laohavisit and Davies, 2011). It has been reported that cold stress causes increased accumulation of annexin in poplar leaves (Renaut et al., 2006) but in this study the accumulation of annexin protein was lower in cold conditioning samples than RT samples, which could indicate other functions in these tissues. Mortimer et al. (2008) indicated that they could be central regulators or effectors of plant growth and stress signaling. Potentially, they could relocate to membranes as response to reactive oxygen species and cytosolic free calcium to cope with adverse conditions such as cold. Furthermore, annexins not only participate in the regulation of membrane organization and vesicle trafficking, they also seem to be involved in Ca2+-regulated exocytosis or some endocytic events. Lannoo and Van Damme (2010) reported that annexins are involved in RNA-binding, sensing or transducing Ca2<sup>+</sup> signals or in the regulation of [Ca2+]cyt during signaling process.

Transport of molecules with high molecular weight between the cytoplasm and the nucleoplasm requires an active mechanism facilitated by soluble nuclear transport factors or NTFs (Zhou et al., 2013). Spot 54 identified as NTF2 significantly decreased in abundance after cold conditioning. NTF2 is one of several essential components of Ran cycle involved in nuclear trafficking. Its main role is to transport RanGDP from cytoplasm to nucleus and replenish the nuclear pools of RanGTP (Bian et al., 2011). It has been observed that overexpression of NTF2 inhibits nuclear import of transcription factors, some of which are involved in stress-dependent gene induction (Zhao et al., 2006; Bian et al., 2011; Binder and Parniske, 2013).

### Proteins Related to Protein Folding

The risk of protein misfolding increases at low temperatures, resulting in non-functional proteins. For that reason, a higher accumulation of proteins with chaperone functions has been reported under low temperature conditions (Renaut et al., 2004, 2005; Cui et al., 2005; Yan et al., 2006; Rinalducci et al., 2011; Chen et al., 2013). Five spots of proteins differentially accumulated were identified as proteins with chaperone functions. Three spots significantly increased in response to cold conditioning; spot 41 was identified as chaperonin CPN60-2, spot 43 was identified as heat shock 70 kDa protein (HSP70) and the spot 62 was identified as peptidyl-prolyl cis-trans isomerase FKBP12 (PPIase). According to Mühlhaus et al. (2011) the presence of HSPs in protein aggregates facilitates the access of ATP-dependent chaperones ClpB and Hsp70/Hsp40, which catalyze the refolding of the complex, denatured proteins back to its functional state. However, the increased accumulation of chaperones may play a pivotal role in preventing aggregation and in facilitating the refolding of denatured proteins under normal and low-temperature conditions. In contrast, two spots decreased under cold conditioning were identified as a chaperone (spot 19) and 18.1 kDa class I heat shock protein (HSP18.1; spot 20) but the specific role of these proteins in folding or refolding proteins is not known.

### Photosynthesis-related Proteins

In garlic sprouts, we found two spots identified as small subunits of chloroplast ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO SSU; spots 12 and 62) that showed an interestingly contrasting accumulation. Low temperature affects significantly photosynthesis, but plants may adjust photosynthesis via gene regulation to adapt to cold environment and some proteins involved in Calvin cycle and electron transport. It has been shown that RuBisCO SSUs are important for catalysis by enhancing the catalytic rate and specificity through inducing conformational changes in RuBisCO large subunits (Spreitzer, 2003). Yoom et al. (2001) found that the relative SSU gene expression differed significantly between plants grown at different temperatures showing that some genes were over expressed in cold conditions (5 and 10◦C) and others increased at high temperature. In leaves of Thellungiella exposure to cold conditions (24 h at 4◦C) increased the accumulation of two SSU proteins, while other SSU decreased, whereas the RuBisCO large subunit decreased (Gao et al., 2009). These data suggested that garlic plants might modify their photosynthetic efficiency and could explain the slower growth rate of the plants observed at the end of the crop cycle. Although other explanation could be, as in the case of the photosystem II that was being photo-inhibited because the photorespiration pathway (a protective mechanism of photosynthetic systems in C<sup>3</sup> plants) could be enhanced by low temperature, which also would reduce the CO<sup>2</sup> fixation (Zhang et al., 2013) causing lower plant growth.

## Carbohydrate Metabolism-related Proteins

Different studies have indicated changes in abundance of enzymes involved in carbohydrate metabolism under low temperature conditions. In general, up-regulation of catabolic pathways and down-regulation of anabolic pathways has been observed under cold stress (Kosová et al., 2011). Accumulation of tricarboxylic acid (TCA) cycle enzymes could also suggest an efficient recycling of amino acids as energy source and their subsequent recruitment as substrates in other cellular pathways under low temperature exposure (Rinalducci et al., 2011). Our data showed significant increased accumulation of carbohydrate metabolism-related proteins in six protein spots from garlic sprouts under cold conditioning (**Figure 1**, **Table 1**). Fructose-bisphosphate aldolase (FBPA; spot 37) and 2,3-bisphosphoglycerate-independent phosphoglycerate mutase (iPGAM; spots 42 and 46). FBPA catalyzes the conversion of Dfructose 1,6-bisphosphate into dihydroxyacetone phosphate and D-glyceraldehyde 3-phosphate. This enzyme plays an important role in carbohydrate metabolism and in the production of triose phosphate derivatives, which are important in signal transduction (Schaeffer et al., 1997). iPGAM besides to its role in glycolysis, also plays an important role in the biosynthesis of aromatic amino acids and aromatic compounds (Zhao and Assmann, 2011). Whereas malate dehydrogenase (MDH), detected in spots 35, 37, and 44, is involved in the TCA cycle as well as in glyoxylate bypass, amino acid synthesis, gluconeogenesis and as a shuttle between cytoplasm and subcellular organelles (Musrati et al., 1998). The up-regulation of carbohydrate metabolism might help in more energy production or intermediate compounds needed in cell adaptation to the stress condition.

### Nucleotide Metabolism-related Proteins

Nucleotide metabolism was altered under cold conditioning as revealed by the increased accumulation of adenosine kinase 2 (ADK2; spot 33). ADK is a purine kinase that transfer γ-phosphate from ATP to adenosine to generate 5′ -AMP (Mohannath et al., 2014). This enzyme plays an important role in the adenine salvage pathway and thereby contributes to the maintenance of cellular energy and in the synthesis of different biomolecules including nucleotide cofactors and nucleic acids. ADK has also other roles in plant metabolism; it is associated to maintain methyl transferase activities by sustaining the methyl cycle that generates SAM. Second, is involved in the mechanism for regulating the level of active cytokinin in plants (Moffatt et al., 2002). It has been established that adenosine kinase forms a complex with SNF1-related kinase (SnRK1) in plants and that these complexes may play important roles in energy homeostasis and cellular responses to biotic and abiotic stress (Mohannath et al., 2014).

### Miscellaneous Proteins

Six different spots (**Table 1**) were identified as the coat protein (CP) of garlic common latent virus (GarCLV). Five proteins were decreased in abundance under cold conditioning (spots 31, 32, 47, 49 and 57) and another was increased in abundance (spot 33). GarCLV has been reported in different countries around the world and it is always affecting garlic crop. However, this virus is symptomless on garlic but can induce severe yellowing and mosaic when it occurs in mixture with other viruses (Pramesh and Baranwal, 2013). The decrease in the accumulation of virus coat proteins may be a response of the virus related to environmental conditions (Parrano et al., 2012).

# Concluding Remarks

Based on the previous discussion, a hypothetical model is proposed to illustrate cellular events potentially associated with the effects of low-temperature conditioning (**Figure 3**). We have shown that cold conditioning induces downaccumulation of ANX2, due to a temporary relocation to membranes in response to cold. This situation could change the [Ca2+]cyt inducing a different signaling process. The signal is transmitted to the cellular machinery by signal transduction and transcription factors that regulate the gene expression and

fructose-bisphosphate aldolase; GR-RBP, glycine-rich RNA-binding protein;

protein accumulation to establish a new cellular homeostasis. In this sense, the decrease of NTF2 could lead to an increase in the nuclear import of transcription factors, some of which are involved in stress-dependent gene induction. These responses could explain the accumulation of proteins related with the regulation of transcription, RNA processing, translation, and protein processing, such as PAI1, GR-RBPs, SR, hnRNP, PAPB, and NAC. In addition, protein-folding proteins increased in garlic sprouts in response to the cold conditioning, such as CPN60-2, HSP70, PPIase and Pop3 that stabilize proteins and

Frontiers in Plant Science | www.frontiersin.org May 2015 | Volume 6 | Article 332 |

factor SR34A-like; SSU, RuBisCO small subunit; TCA, tricarboxylic acid; TF, transcription factor; TRX, thioredoxin.

help to restore the cellular proteostasis. Several key enzymes regulating carbohydrate metabolism (and ATP production) accumulation in cold conditioned samples (iPGAM, FBPA, MDH) might help in energy production or intermediate compounds needed in cell adaptation to the low temperature environment. We have found the accumulation of ADK2 in cold conditioned samples, which contributes to energy homeostasis and in the synthesis of nucleotide cofactors and nucleic acids. ADK2 is associated with the methyl cycle that generates SAM, a methyl donor involved in secondary metabolism; in addition, increased accumulation of SAMS in cold conditioned samples was observed. In another hand, the decrease in the abundance of proteins such as MIF, CBS, DHAR and lectins, which have key roles in the redox homeostasis, lignin deposition, division and differentiation of cells, could explain the differences in plant growth. Our results also showed differences in the abundance of RuBisCO SSU related with photorespiration that reflects the importance of this pathway as a protective mechanism of photosynthetic systems in C<sup>3</sup> plants, which also reduces the CO<sup>2</sup> fixation causing lower plant growth. Reduced accumulation of GST proteins in 5◦C samples could indicate an increased accumulation of H2O2. Redox homeostasis and glycolysis metabolites can also act as signals and promote the regulation of gene expression. Conceivably, further studies will provide additional information for this model.

The data presented in this research indicate that the lowtemperature conditioning of "seed" garlic cloves during 5 weeks at 5 ◦C affected different metabolic pathways and physiological processes. The physiological processes affected include cellular growth, antioxidative/oxidative state, macromolecules transport, protein folding, and transcription regulation process. The metabolic pathways affected including protein biosynthesis and quality control system, photosynthesis, photorespiration, energy production, and carbohydrate and nucleotide metabolism. These processes can work cooperatively to establish a new cellular homeostasis that might be related with the physiological and biochemical changes observed in previous studies (Dufoo-Hurtado et al., 2013; Guevara-Figueroa et al., 2015). This is the first work that reports the changes in the protein profiles of garlic "seed" cloves to low-temperature conditioning. The identification of cold-responsive proteins in garlic provides not only new insights into cold conditioning responses but also a good starting point for further dissection of their functions

# References


during the development of the crop using genetic and other approaches.

# Author Contributions

MD, EM, JH, and AB, conceived the study and designed the 2-DE experiments. MD and JH performed the experiments. AB performed the LC-MS/MS analysis. MD and JH carried out the analysis of the data, made the identification of the proteins and drafted the manuscript. EM and AB contributed in the preparation of the final draft of the manuscript. AB provided reagents, materials and analysis tools. All authors (MD, JH, AB, EM, and AB) read and approved the final manuscript.

# Acknowledgments

This research was supported by FOFI-UAQ FCQ201203 Project and by the Garlic Producer Association of Aguascalientes, Mexico. MDDH gives thanks to PhD fellow 339652 from CONACYT-Mexico. We thank to Hostilio Torres-Robles, for providing the "seed" bulbs of garlic. We also thank to CONACYT-MEXICO GRANTS 56787 (Laboratory for Nanoscience and Nanotechnology Research-LINAN) and 204373 (Infrastructure project, INFRA-2013-01). We also thank Antonio Castellanos (CNS-IPICyT) for his valuable help.

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00332/abstract


2, the nuclear import factor of Ran. Plant Physiol. 140, 869–878. doi: 10.1104/pp.105.075499


**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.

Copyright © 2015 Dufoo-Hurtado, Huerta-Ocampo, Barrera-Pacheco, Barba de la Rosa and Mercado-Silva. 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.

# The modulation of leaf metabolism plays a role in salt tolerance of *Cymodocea nodosa* exposed to hypersaline stress in mesocosms

### *Edited by:*

*Dominique Job, Centre National de la Recherche Scientifique, France*

### *Reviewed by:*

*Tiago Santana Balbuena, São Paulo State University, Brazil Ana Paulina Barba De La Rosa, Instituto Potosino de Investigación Científica y Tecnológica, Mexico*

### *\*Correspondence:*

*Silvia Mazzuca, Laboratorio di Biologia e Proteomica Vegetale, Dipartimento di Chimica e Tecnologie Chimiche, Università della Calabria, Ponte Bucci 12C, 87036 Rende, Italy silvia.mazzuca@unical.it*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 13 March 2015 Accepted: 11 June 2015 Published: 26 June 2015*

### *Citation:*

*Piro A, Marín-Guirao L, Serra IA, Spadafora A, Sandoval-Gil JM, Bernardeau-Esteller J, Fernandez JMR and Mazzuca S (2015) The modulation of leaf metabolism plays a role in salt tolerance of Cymodocea nodosa exposed to hypersaline stress in mesocosms. Front. Plant Sci. 6:464. doi: 10.3389/fpls.2015.00464* *Amalia Piro1, Lázaro Marín-Guirao2, Ilia A. Serra1, Antonia Spadafora1, José M. Sandoval-Gil2, Jaime Bernardeau-Esteller2, Juan M. R. Fernandez2 and Silvia Mazzuca1\**

*<sup>1</sup> Laboratorio di Biologia e Proteomica Vegetale, Dipartimento di Chimica e Tecnologie Chimiche, Università della Calabria, Rende, Italy, <sup>2</sup> Spanish Institute of Oceanography, Oceanographic Centre of Murcia, Murcia, Spain*

Applying proteomics, we tested the physiological responses of the euryhaline seagrass *Cymodocea nodosa* to deliberate manipulation of salinity in a mesocosm system. Plants were subjected to a chronic hypersaline condition (43 psu) to compare protein expression and plant photochemistry responses after 15 and 30 days of exposure with those of plants cultured under normal/ambient saline conditions (37 psu). Results showed a general decline in the expression level of leaf proteins in hypersaline stressed plants, with more intense reductions after long-lasting exposure. Specifically, the carbon-fixing enzyme RuBisCo displayed a lower accumulation level in stressed plants relative to controls. In contrast, the key enzymes involved in the regulation of glycolysis, cytosolic glyceraldehyde-3-phosphate dehydrogenase, enolase 2 and triose-phosphate isomerase, showed significantly higher accumulation levels. These responses suggested a shift in carbon metabolism in stressed plants. Hypersaline stress also induced a significant alteration of the photosynthetic physiology of *C. nodosa* by means of a downregulation in structural proteins and enzymes of both PSII and PSI. However we found an over-expression of the cytochrome b559 alpha subunit of the PSII initial complex, which is a receptor for the PSII core proteins involved in biogenesis or repair processes and therefore potentially involved in the absence of effects at the photochemical level of stressed plants. As expected hypersalinity also affects vacuolar metabolism by increasing the leaf cell turgor pressure and enhancing the up-take of Na+ by overaccumulating the tonoplast specific intrinsic protein pyrophosphate-energized inorganic pyrophosphatase (H(+)-PPase) coupled to the Na+/H+-antiporter. The modulation of carbon metabolism and the enhancement of vacuole capacity in Na+ sequestration and osmolarity changes are discussed in relation to salt tolerance of *C. nodosa*.

Keywords: seagrasses, leaf proteomics, hypersaline, mesocosm

# Introduction

Seagrasses are marine plants that have successfully colonized the infralittoral bottoms of tropical and temperate coasts around the world representing one of the most ecologically relevant marine coastal habitats (Kuo and Den Hartog, 2000). Although this ecological group of plants is particularly adapted to live completely submersed in a saline medium, their abundance and distribution are determined by critical environmental factors including salinity (Montague and Ley, 1993; Adams and Bate, 1994). The physiological capacity of seagrasses to tolerate salinity changes is species specific and closely related to the salinity regime of the environments in which they grow (Touchette, 2007). This suggests that particular adaptations acquired along their evolution to live submerged in the marine environment are closely related with their current ecological and biological attributes, thus determining their ability to cope with salinity fluctuations. In this regard, C*ymodocea nodosa,* which is a seagrass species inhabiting open coastal waters with stable saline regimes but also hypersaline lagoons and estuaries with fluctuant salinities, has been considered as an euryhaline species with a higher capacity to tolerate salinity changes than other seagrass species naturally living under a narrower range of salinities (e.g., *Posidonia oceanica*; Tyerman, 1989; Kuo and Den Hartog, 2000; Procaccini et al., 2003; Boudouresque et al., 2009; Sandoval-Gil et al., 2012).

Despite its central role in seagrass ecology, basic knowledge about salinity adaptation and about the specific tolerance mechanisms to increments in seawater salinity is relatively scarce in relation to other issues of seagrass biology and physiology (Touchette, 2007). A key need of physiological studies on seagrasses is the maintenance of plants under controlled mesocosm systems able to simulate the effect of the stress factor of interest (e.g., salinity) and isolate it from the variability of other key factors than can confound or mask the specific responses to the selected stress factor (Coors et al., 2006). In fact, recent ecophysiological studies performed on laboratory mesocosm systems have produced significant new knowledge about the physiological adaptations of Mediterranean seagrass species (mainly *P. oceanica* and *C. nodosa*) to hypersaline stress. Thus, these studies have confirmed that *C. nodosa* is more tolerant to salinity increases than the stenohaline *P. oceanica*. Indeed, several biological attributes of *C. nodosa*, for instance high leaf osmolyte content that confers osmoprotection and high photosynthetic plasticity that allow to balance the plant carbon budget (Marín-Guirao et al., 2011; Sandoval-Gil et al., 2012), have been interpreted as adaptive advantages in its higher resistance to chronic salinity increments. Besides, *C. nodosa* has shown complex responses at the metabolic, physiological and morphological levels that have allowed the species to counterbalance the effects and alterations generated by hypersalinity. These effects include photosynthetic and respiratory alterations, modifications of plant water relations and changes to the metabolite contents of the leaf (Sandoval-Gil et al., 2012, 2014).

Despite this growing physiological background, the application of proteomics is a relatively new tool to be applied to salt response in seagrasses (Serra et al., 2013). Molecular approaches have been recently applied to seagrasses to understand the molecular bases of stress responses, resilience and acclimation to low light (Mazzuca et al., 2009, 2013; Serra and Mazzuca, 2011; Serra et al., 2012; Dattolo et al., 2013, 2014). From these approaches it is possible to recognize new tools that might deserve the designation of "early warning" markers for environmental stresses. A combination of physiology, genomics and proteomics has also recently provided experimental evidence about the induction of specific proteins related to the osmotic stability of *P. oceanica* leaves exposed to salt stress (Serra et al., 2013). Therefore, the combination of molecular techniques and physiological analyses in experiments with plants maintained under controlled mesocosm conditions will represent an effective approach to achieve significant progress in understanding the intrinsic mechanisms of different species of marine plants to cope with hypersaline stress.

On that basis, in this study we used a mesocosm system to expose *C. nodosa* plants to a deliberate increase in seawater salt concentration in order to evaluate how the species modulate their protein expression during the course of the exposure, and how this modulation is linked with their physiological status. To this end, we combined physiological measurements, at the level of plant water relations and plant photochemistry, with a proteomic approach to highlight the molecular adaptation of *C. nodosa* to saline increments. We also took advantage of recent physiological studies on the tolerance of *C. nodosa* to salinity increases to shed light on the possible molecular mechanisms underlying the physiological responses and alterations previously observed in the species.

# Materials and Methods

## Field Plant Sampling and Experimental Design

*Cymodocea nodosa* cuttings (i.e., rhizome apical segments composed of at least 20 connected vertical shoots) were collected by SCUBA divers in a shallow bed (5–6 m deep) located in Isla Plana (Murcia, Spain). Plants were transported under controlled temperature to the laboratory in less than 4 h and rapidly transplanted in the mesocosms. Plant cuttings were fixed onto a grid and mounted in plastic baskets (22 cm × 40 cm base and 10 cm height) filled with pre-washed sediments to a final plant density of 50–60 shoots basket−1; each basket represents a transplantation unit (t.u.) and three of them were arranged in each tank of the mesocosm system (**Figures 1A,B**).

The mesocosm system consists of six independent 500-L tanks each with its own source of illumination provided by a 400 W lamp (Aqua Medic Aqualight −400). The sea water used to fill the mesocosm was collected in a nearby pristine open water area. A detailed description of the system can be found in Marín-Guirao et al. (2011) and Sandoval-Gil et al. (2012). This system is able to maintain healthy plants with survival rates at 100% for several months, long enough to achieve the objectives of the

experiment (e.g., Marín-Guirao et al., 2013; Sandoval-Gil et al., 2014).

The t.u. were maintained during 1-week acclimation period at 22◦C, 37 psu and 300 μmol quanta m−<sup>2</sup> s1 measured on the leaf tips on a 12 h/12 h light/dark cycle (i.e., 12.96 mol quanta m−<sup>2</sup> day−1) according to mean environmental conditions experienced by the donor population during the experimental period (**Figure 1C**). After acclimation and before the application of the experimental treatments (T0), 2 g of fresh leaves were randomly sampled from t.u. collecting shoots from one t.u. of each tank (**Figure 1D**). Then, salinity was increased up to 43 psu in three of the tanks by adding high quality marine salt (Seachem-) as previously described (Marín-Guirao et al., 2011), with all other parameters remaining unchanged. Plants were sampled again from both the control and the hypersaline treatments after 15 and 30 days of hypersaline exposure (**Figures 1E,F**). At each sampling time three different replicates were processed each composed of the material from one t.u. of each independent tank. Mature non-damaged leaves were selected for the analysis to avoid old and necrotic tissues. Collected leaf tissue were washed in sea water, gently cleaned with a razor to remove epiphytes, washed again in distilled water to remove salt from the leaf epidermis, frozen in liquid nitrogen and stored at −80◦C for further proteomic analysis.

### Physiological Measurements

Chlorophyll *a* fluorescence emissions were performed using a diving-PAM portable fluorometer (Walz, Germany). Measurements were done on plants in the mesocosm adapted to darkness overnight (i.e., before switching on the illumination system) to ensure full oxidation of the reaction centers and primary electron acceptors. This allowed us to calculate the maximum quantum yield of the photosystem II (PSII; Fv/F), which represents a measure of the maximum photochemical efficiency of the PSII (Schreiber, 2004). A more detailed description of the measurements can be found in Sandoval-Gil et al. (2012). Measurements were taken before the application of the hypersaline treatment and after 7, 15, and 30 days of hypersaline exposure. In each sampling time measurements were performed on three randomly selected shoots from each t.u. and averaged per tank to have a final number of three replicates (i.e., one replicate tank−1) per treatment.

Leaf-water relation variables (i.e., water potential w, osmotic potential p and turgor pressure P) of *C. nodosa* plants were analyzed before the application of the hypersaline treatment and after 15 and 30 days of hypersaline exposure. Three different shoots from each tank and sampling time were employed for the measurements of leaf-tissue osmolality (mmol kg−<sup>1</sup> FW) using a Wescor Vapor Pressure Osmometer 5520 (Logan, Utah). Measurements were averaged per tank to have three replicates per experimental condition and sampling time. Osmolality was measured both in fresh and frozen blotted leaf segments to obtain w and p for each shoot, and expressed in megapascals (MPa), using the van't Hoff relation (Tyerman, 1989). The leaf-tissue turgor pressure (P) was then calculated as the absolute difference between w and p. Ambient seawater osmolality was also determined by measurements of seawater in 6.5 mm sample disks, following the standard protocol (Wescor Inc.).

# Extraction and Purification of Total Protein from Leaves

Leaf tissues of marine plants are considered to be recalcitrant to the common protocols based on the aqueous buffers extraction because they are rich in secondary metabolites, such as phenols, disaccharides, lipids, that severely interfere with protein extraction and purification (Wang et al., 2006; Spadafora et al., 2008). Here we applied a multistep procedures that precipitated proteins prior to their extraction in a phenol phase. For each extraction 1.4 g of leaves were crushed in a mortar in liquid nitrogen until a fine powder was obtained. This powder was divided into 2 ml microfuge tubes; a volume of 10% TCA in acetone was added and centrifuged at 13000 rpm for 5 min at 4◦C. Subsequently, four washes were performed in 80% acetone in water. TCA is a strong acid and precipitates the protein when it is still in the tissue powder; at the same time, phenols, sugars, and other soluble molecules are washed out by the TCA solution while the hydrophobic molecules are dissolved in the acetone. After centrifugation the pellet containing the precipitated proteins was dried at room temperature. The powder was collected in microfuge tubes and kept at −80◦C for subsequent analysis or immediately processed for phenolic phase extraction of proteins.

Approximately 0.1 g of powdered tissue was dissolved in 0.8 ml of phenol (buffered with Tris-HCL, pH 8.0, Sigma, St. Louis, MO, USA) and 0.8 ml of SDS buffer (30% sucrose, 2% SDS, 0.1 M Tris -HCl, pH 8.0, 5% 2-mercaptoetanol) in a 2 ml microfuge tube. The samples were vortexed for 30 s and centrifuged at 13000 rpm for 5 min to allow proteins to solubilise in the phenol phase. The phenol phase was mixed with five volumes of 0.1 M ammonium acetate in cold methanol, and the mixture was stored at −20◦C for 30 min to precipitate proteins. Proteins were collected by centrifugation at 13000 rpm for 5 min. Two washes were performed with 0.1 M ammonium acetate in cold methanol, and two with cold 80% acetone, and centrifuged at 13000 rpm for 7 min. The final pellet containing purified protein was dried and dissolved in Laemmli 1 DE separation buffer over-night. Proteins were then quantified by the Bradford assay. Protein yield was measured as mg of protein per g fresh tissue weight in five independent biological replicates at each time and treatment.

# Electrophoresis of Leaf Proteins, Protein In-Gel Digestion, and Mass Spectrometry Analyses

A gel was prepared at a concentration of 10% acrylamide/bisacrylamide, according to the method of Laemmli (1970). The ratio of acrylamide/bisacrylamide was 12.5% in the running gel and 6% in the stacking gel. The samples were heated for 5 min at 100◦C before being loaded on the gel. The electrophoretic run was carried out at 60 mA for the stacking gel and 120 mA in the running gel at constant power of 200 V. The electrophoresis ran for an average of 1 h and 15 min. The gels were stained with Coomassie Blue over-night and subsequently destained with several changes of destaining solution (45% methanol, 10% acetic acid).

Digitalized images of the stained SDS-PAGEs were analyzed by the Quantity One 1-D Analysis Software (Bio-Rad) to measure the optical densities at each lane of all biological replicates among the treatments. The amount of protein at bands of 55, 25, and 10 kDa was done using the marker reference bands at 75, 50, and 25 kDa that contained 150, 750, and 750 ng of proteins respectively (**Figures 2A–C**). Each lane of the same SDS-PAGE were divided in six slices from 200 to 10 kDa and manually excised from the gel, cut in small pieces, *S*-alkylated and digested overnight at 37◦C with trypsin (Wilm et al., 1996). Digested peptides were extracted from the gel slices with 25 mM NH4HCO3/ACN 1:1 (v/v) and the peptide mixtures were concentrated by evaporation in a vacuum centrifuge. The gel slices were then treated with 5% (v/v) formic acid in acetonitrile (ACN). After drying, the tryptic peptides were analyzed by tandem mass spectrometry by means of liquid chromatography(LC-MS/MS) using a high resolution mass spectrometer LTQ- Orbitrap XL (Thermo Fisher Scientific). The chromatographic separations were carried out on a Waters XBridgeC18 column (300 μM ID × 100 mm in length and 3.5 μm per particle size) using a linear gradient of 5–90% ACN containing 0.1% formic acid with a flow of 4 μL/min, including the regeneration phase, a run lasted about 70 min; microflow conditions were employed in order to obtain more reproducible semi-quantitative data. Full scan MS high resolution spectra (resolution of 30,000) were acquired. Data were acquired in data dependent scan acquisition (DDA) conditions and the MS/MS spectra were acquired in an Ion Trap in low resolution mode to decrease the acquisition scan duty cycle. The most abundant peak was fragmented under dynamic exclusion conditions. In particular, the peak was fragmented two times and maintained in the dynamic exclusion list for 90 s. The acquisitions were undertaken in scanning mode data-dependent MS/MS (with full scan range of 250– 1800 m/z).

30 days hypersaline condition. (B) Examples of chromatograms of the optical densities (OD) from the lines of protein standard, control sample and hypersaline treatments in a same SDS-PAGEs as a function of molecular weights (kDa). (C) Protein concentration at the bands of 55, 25, and 10 kDa in the control samples and after 15 and 30 days hypersaline treatments.

# Bioinformatics Analysis and Identification of Proteins of *Cymodocea nodosa*

Spectra acquired by LC-MS/MS were used to identify peptide sequences using the open-source system global proteome machine (GPM) engine against the GPM public UniGene (NCBI) on-line plant database1 . Since the GPM plant database lacks seagrass sequences and considers only few species that belong to *Liliopsida*, a search can lead to reduced peptide matches after mass spectrometry. Thus, spectra acquired by LC-MS/MS were also used to match peptide sequences using X!Tandem software (Fenyö et al., 2010) against a customized database built with a collection of protein sequences from multiple databases. This comprised sequences from seagrasses and other species belonging to *Liliopsida* available in the UniProtKB database and included sequences from *P. oceanica, Zostera marina* and from five EST libraries (Pooc\_A, Pooc\_B, Zoma\_A, Zoma\_B, and Zoma\_C) collected from the *Dr.Zompo* database (Wissler et al., 20092 ). In the latter instance, it has been necessary to first create a protein database from the nucleotide sequences as described in Dattolo et al. (2013). The use of all possible reading frames enabled the optimization of peptide identifications. The GPM and !XTandem searches were done in parallel and then results combined to obtain the final outcome. Molecular function and biological processes of each identified protein were obtained from the Gene Ontology (GO) website to assign metabolic pathways for each identified protein.

### Semi-Quantitative Analysis of Identified Proteins

Quantitation of the identified proteins was performed by spectral counting (Zhang et al., 2010). Differences in spectral counts are identified by applying the Normalized Spectral Abundance Factor (Zybailov et al., 2006). The false discovery rate (FDR) was calculated from the matched spectra against a reverse peptide

1http://plant.thegpm.org/tandem/thegpm\_tandem.html 2http://drzompo.uni-muenster.de/

database according to Elias and Gygi (2007); the threshold for protein FDR corresponded to 1%.

Each m/z ratio was subjected to a statistical discriminate analysis using the XCMS algorithm. Basically it can be summarized as a *t-*test peak per peak. Spectral count were undertaken only on statistically validated spectra to increase its accuracy. Consequently, it was used for quantitation comparisons. A peptide with less than two matches was discarded. Three biological replicates for each treatment was used for quantitative analyses. The missing values were considered to be undetectable and assumed they were under the limit of detection, but present. Thus when they were undetectable, a zero value was attributed and they were considered in the statistical calculation. Data from repeated measurements are shown as mean ± SD or ± SE. Comparison of differences among the groups was carried out using a Student's *t-*test. Significance was defined as *p* ≤ 0.05.

# Results

## Physiological Analyses

All along the course of the experiment, plants subjected to hypersaline showed similar (*p* > 0.05) levels of maximum photochemical efficiency of PSII (Fv/Fm) to that of control plants, with mean values within the range 0.763–0.777. At the end of the experimental period, the potential photochemical efficiency of plants under cultured conditions in the mesocosm system (i.e., 0.775 ± 0.004) was similar to that measured in the field, natural population (0.780 ± 0.02). After 7 days acclimation in the mesocosm and before the application of the hypersaline treatment, water relation parameters did not change in either tank (**Figure 3**). Once under hypersaline conditions, *C. nodosa* plants reduced the water potential (w) of their leaves as a response to the reduction in the osmotic potential of seawater caused by the addition of salts. After 30 days of hypersaline exposure the significant (*p* < 0.01) reduction in w was attained through a significant (*p* < 0.05) reduction in the osmotic

potential (i.e., π) of leaves indicating the accumulation of osmolytes, while the osmotic pressure remained unaltered. After an exposure of longer than 30 days, the osmotic potential of leaves further decreased, indicating a higher accumulation of osmolytes. It was at this sampling time that an increase in turgor pressure was observed in leaves of plants exposed to hypersalinity; observed values were on average 90% higher those of control plants.

### Protein Expression

The total amount of protein extracted from *C. nodosa* leaf tissues along the course of the experiment from both the control and hypersaline treatments are shown in **Table 1**. The protein yields decrease about 20 and 30% in hypersaline stressed plants after 15 and 30 days of exposure, respectively. Highly purified proteins from *C. nodosa* leaves are well resolved by SDS-PAGE. As shown in **Figure 2**, the SDS-PAGE pattern of control plants grown for 7 days in the mesocosm revealed a high number of polypeptide bands demonstrating the efficient protein extraction and purity by means of a multistep protocol. A considerable reduction of proteins at 55 and 25 kDa, corresponding to the RuBisCo large subunit and LHCP respectively, was detected in plants exposed to hypersalinity for 15 days with respect to the control (**Figures 2B,C**); overall a reduction in the number of major resolved protein bands occurred after 30 days, with a further decrease of protein at 55 and 25 kDa, while proteins with small molecular weights (10 kDa) increased (**Figures 2B,C**).

For mass spectrometry analyses, gel pieces were associated in pairs along a molecular weight gradient to compare the protein expression among different sample treatments and durations. Samples collected after 7 days acclimation and before the application of the experimental treatments were used as the control (**Figure 2** lanes 1, 2, 3). For each identified protein the expression level has been calculated using a mean of the spectral count and values were compared among control and hypersalinetreated plants at different times; the results of the analysis are reported in the **Figure 4**. Statistical parameters of each identified protein, number of MS spectra for each peptide assigned to each protein in the control and hypersaline-treated plants are reported in Supplementary Table S1; single peptide sequence assigned to each identified protein and their related accession numbers are reported in Supplementary Table S2.

Despite the LC-MS/MS analysis resulting in an array of MS/MS spectra, many could not be attributed to peptide sequences, even though their fragmentation pattern was typical to that of a peptide (data not shown). Many LC-MS spectra did

TABLE 1 | Protein yield from leaf tissues of *Cymodocea nodosa* after 7, 15, and 30 days culture in the mesocosm (control) and after 15 and 30 days of hypersaline treatments.


*Values are the main of five independent replicates (*±*SE).*

not match, even after *de novo* sequencing and BLAST searches. Thus it can be concluded that many sequence were likely absent from our databases and this resulted in the lack of associated with a known protein.

Among the identified proteins, a set of 32 differentially accumulated proteins was identified in controls vs. hypersalinetreated plants, including a group of proteins whose levels are lower in the hypersaline-treated plants, the PSII subunit PsbS, the RuBisCo large subunit and malate dehydrogenase all appeared down-regulated (**Figure 4**). This group of proteins progressively decreased over the duration of 30 days of hypersaline exposure. After this time, the long lasting exposure resulted also in the chloroplastic ATP synthase subunit gamma and the mitonchondrial ATP synthase alpha subunit being down-regulated in leaves of stressed plants. A second group of proteins showed a high number of spectral counts in the samples under hypersaline treatment; among these are the cytosolic glyceraldehyde-3-phosphate dehydrogenase, the cytochrome b559 subunit alpha (PSII subunit V), enolase 2, peroxidase 1 and the tonoplastic intrinsic protein pyrophosphate-energized inorganic pyrophosphatase (H(+)-PPase). Few proteins displayed lower or no significant variation in their spectral counts according to the treatments (**Figure 4**). Proteins were grouped according to their functional pathways assigned to each protein using GO. In **Figure 5**, the percentage of spectral counts assigned to each identified protein was grouped according to their function. In control plants the proteins involved in Calvin-Benson cycle and the respiratory chain accounted for 60% of total identified proteins; hypersaline treatments effected both these functional classes with a reduction of overall protein content that showed a strong decrease after the long-lasting treatment (30 days). The proteins allocated to photosynthetic metabolism were also drastically reduced by hypersaline conditions after 15 days, while enzymes involved in the glycolytic pathway were unchanged. Differences were not significantly detected in other metabolic pathways.

# Discussion

This research combined, for the first time, proteomic and physiological approaches to assess the effects of hypersalinity on *C. nodosa*. Even if a low number of differentially accumulated proteins were identified because of poor genomic resources in marine plants, results obtained support some of the current and previous physiological responses observed in mesocosm studies and analyses, shedding light on the possible molecular mechanisms underlying the physiological responses and alterations caused by hypersalinity on this seagrass species.

The maximum efficiency of PSII to drive the light absorbed by the leaf pigment matrix into photochemistry (i.e., Fv/Fm) was not altered in *C. nodosa* plants as a consequence of the hypersaline exposure, which is in agreement with previous studies (e.g., Sandoval-Gil et al., 2012, 2014). Interestingly when these plants are exposed to hypersalinity they display an over-expression of the cytochrome b559 alpha subunit in their leaves, together with the down-regulation of other structural proteins of both the PSII

and the PSI. The cytochrome b559, which accumulates in the thylakoid membrane even in the absence of other PSII subunits, is considered as a prerequisite for PSII assembly (Morais et al., 1998), and together with protein D2, the PSII initial complex that serves as a receptor for other PSII core proteins during the biogenesis or the PSII repair process (Adir et al., 1990; van Wijk et al., 1997; Müller and Eichacker, 1999; Zhang and Aro, 2002). Therefore both processes, the repair of damaged PSII cores and the biogenesis of new PSII, could be enhanced during the hypersaline exposure to maintain PSII basal activity. The increased accumulation of this protein may also partially explain the absence of alterations at the level of PSII repair cycle previously reported for *C. nodosa* plants chronically stressed by hypersalinity, which contrasts with the alterations detected in the less tolerant species *P. oceanica* (Marín-Guirao et al., 2013). On the other hand, hypersaline-stressed *C. nodosa* plants have previously shown an improved capacity to harvest light as a response to counterbalance photosynthetic inhibition and this could probably be related to the genesis of new PSII.

In spite of the absence of alterations at the level of PSII photochemistry, the hypersaline exposure induced a significant alteration of the photosynthetic physiology of *C. nodosa* by means of a marked down-regulation of the key enzyme involved in photosynthesis, the carbon-fixing enzyme RuBisCo. As suggested by Sandoval-Gil et al. (2012), the reduction in this enzyme is more likely to be the cause of the saltinduced photosynthetic inhibition reported in this species instead of the damage to photosynthetic structures that has been reported for other seagrass species under more severe hypersaline conditions (e.g., Kahn and Durako, 2006; Koch et al., 2007). Indeed, the lower accumulation of the carbon-fixing enzyme RuBisCo is in agreement with Beer et al. (1980) who provided some experimental evidence of RuBisCo activity reduction in epidermic cells of *C. nodosa* in an *in vitro* high salt assay media. Interestingly, these authors also found that the activity of PEPcase (phosphoenolpyruvate carboxylase) increases in *C. nodosa* due to salt stress, potentially conferring the species the ability to compensate for the reduction in carbon assimilation as a consequence of the reduction in RuBisCo activity.

It has been demonstrated that the photosynthetic physiology of *C. nodosa* is significantly altered following exposure to chronic salinity increases, although the nature of the observed

photosynthetic inhibition differed depending on the severity of the hypersaline stress, but also on the origin of plants (Sandoval-Gil et al., 2012, 2014). In relation to the intensity of the stress, at salinity levels slightly lower than the one employed here (i.e., 39–41), net photosynthetic rates of stressed plants declined due to a substantial respiratory increase. However, under a similar hypersaline stress level employed here (i.e., 43 psu), the respiration rate of plants was limited, but instead an inhibition of gross photosynthetic rates caused a significant reduction in carbon assimilation of stressed plants.

Regarding the reduction in the respiratory activity, proteomic analyses have revealed an overall down-regulation of both mitochondrial and chloroplastic ATP synthases suggesting a reduction of the oxidative and photoxidative phosphorylation processes that are directly related to respiration and photosynthesis. Interestingly in the same sample, the levels of some chloroplast ATP synthase subunits appeared up-regulated or not affected by treatment; this behavior is also found in other species subjected to long lasting salt stress; Wang et al. (2015) reported that chloroplast ATP synthase levels decrease in abundance under long term salt stress, with beta subunits down-regulated and gamma subunits unaffected with respect to controls, indicating that energy supply was affected. However, the authors limited their discussions of these finding, so there is little evidence to justify such a behavior. Anyway, we speculate that as the beta and gamma subunits are encoded by a chloroplastic and a nuclear gene respectively, timing of transcription and translation might not be synchronous and thus differently regulated (Drapier et al., 2007; Rott et al., 2011). A similar behavior has been found for the mitochondrial ATP synthase subunits under salinity stress leading to the up-regulation of the beta subunit and to the down-regulation of alpha subunit (Srivastava et al., 2009). Overall, these results are consistent with the reduction in ATP production from both photophosphorylation and oxidative phosphorylation, as suggested by the reduced levels of ATP synthase. This could be the reason for respiratory inhibition as a plant response to balance the altered ratio between production and consumption of ATP. This suggests that under NaCl stress conditions the carbon balance switches to favor the inorganic carbon increase in tissue as a response to the decrease in photosynthesis rate. In addition, hypersaline treated-plants showed higher levels of the key enzymes involved in glycolysis, cytosolic glyceraldehyde-3-phosphate dehydrogenase, as well as of other enzymes involved in this metabolic pathway such as enolase 2 and triose-phosphate isomerase. The increased concentration of these compounds in leaf tissues of salt-stressed plants reflect an overall up-regulation of the glucose reduction in leaf cells, and suggest that glycolysis may be balancing the demand for energy by producing ATP in the reduction steps from 1,3- diphosphoglycerate to phosphoglycerate and from phosphoenolpyruvate to pyruvate. This supports the idea that non-structural carbohydrates have a high functional plasticity under hypersaline stress, and together with their role as osmolytes (e.g., Touchette, 2007) they can also be involved in other key metabolic functions, for instance acting as respiratory substrates (Drew, 1978; Bisson and Kirst, 1995). This increased cell respiration, however, can jeopardize plant survival through the consumption of their energetic reserves in the long-term.

*Cymodocea nodosa* is able to maintain its leaf osmotic stability under hypersaline conditions through different dehydration avoidance strategies, including cell-wall hardening and osmoregulatory processes. During osmoregulation, inorganic ions sequestered into vacuoles and cytosolic compatible organic osmolytes (e.g., sugars, proline) have been reported to play a Piro et al. Seagrass proteomics under hypersaline stress

key role in seagrasses (e.g., Koch et al., 2007; Sandoval-Gil et al., 2012, 2014; Marín-Guirao et al., 2013) and must be responsible for the osmotic potential reduction observed in this study. Indeed, the participation of ions in leaf osmotic potential of *C. nodosa* at the early stages (i.e., 1-week) of its acclimation adjustment to hypersalinity has recently been shown to increase, in parallel with increases in salinity (Garrote-Moreno et al., 2014). In this case, after a longer exposure of 1 month we observed a marked increase in pressure of the leaf cell turgor, which is an atypical response previously reported in *P. oceanica* (Sánchez-Lizaso et al., 2008; Ruiz et al., 2009; Marín-Guirao et al., 2013; Sandoval-Gil et al., 2014), as well as in some terrestrial plants and marine algae exposed to salt stress (Behboudian et al., 1986; Kirst, 1989). In agreement, vacuolar metabolism of these hypersaline stressed plants showed signs of being perturbed as reflected by the over-expression of the tonoplast specific intrinsic protein pyrophosphate-energized inorganic pyrophosphatase (H(+)-PPase). This response suggests that vacuoles are engaged in Na+ sequestration accordingly with a high capacity of proton pumping and Na+ uptake via the Na+/H+-antiporter, a response previously observed in seagrasses (Muramatsu et al., 2002), halophytes (Debez et al., 2006) and in other higher plants under hypersaline stress (Fukuda et al., 2004; Epimashko et al., 2006). A possible explanation of these responses is the involvement of the *turgor sensing mechanisms*, by which the disturbance of the plasma membrane by turgor increments led to the activation of downstream signaling cascades involved in the osmoacclimation responses (Zimmerman, 1978; Kirst, 1989; Bisson and Kirst, 1995). Nonetheless, there is no reliable explanation for this behavior and further work is needed to determine which are the specific molecular and physiological mechanisms involved and their role in the acclimation capacity of the species to saline increments.

# Conclusion

We have found severe changes in the leaf primary metabolisms due to hypersalinity both at short and long-lasting treatments. Overall, the proteomic analysis revealed that the physiological tolerance of *C. nodosa* to sudden and chronic increases in external salinity is mediated by its capacity to modulate primary metabolism resulting in a new carbon balance combined with efficient Na+ sequestration against its electrochemical gradient toward the vacuoles of mesophyll cells. As would be required for cytoplasm protection and suitable osmotic

# References


adjustments (Barbourina et al., 2000; Taji et al., 2004). These drastic rearrangements at the level of primary metabolism allowed *C. nodosa* plants to survive for more than 1 month (plant mortality was not detected in the present work) evidence supporting their successful metabolic and physiological adaptation to the selected ionic and osmotic stressful conditions. The absence of mortality at this level of chronic stress have previously been reported for the species under controlled experimentation in mesocosm (Sandoval-Gil et al., 2012) contrary to the population decline observed in field studies associated with constant salinity fluctuations (Garrote-Moreno et al., 2014) or to the synergistic effects of hypersalinity with other products employed in the maintenance of seawater reverse osmosis plants (i.e., metabisulphite; Portillo et al., 2014).

# Acknowledgments

The authors are grateful to the COST Action ES0906 "*Seagrass productivity: from genes to ecosystem management*" to fund the research activities of the Short Term Scientific Mission (COST-STSM-ES0906-06317) awarded to AP. Furthermore, the authors thank Rocío García Muñoz of the Spanish Oceanography Institute, Oceanography Centre of Murcia (Spain) for technical support on the use of the mesocosm system, which was funded by two Spanish Government Ministries: the 'Ministerio de Medioambiente y Medio Rural y Marino' (OSMOGRASS project no. 021/SGTB/2007/1.3) and the 'Ministerio de Ciencia e Innovación' (OSMOGRASS II project no. CTM2009- 08413MAR). It was also funded by two research grants awarded to JS-G and J-E by the Department of Marine Sciences and Applied Biology of the University of Alicante. The authors also thank Simone Cristoni for his valuable support in the statistical analysis of data from mass spectrometry. This article was made possible thanks to the support of the European Commission, the FSE European Social Fund for the Calabria Region. The authors are solely responsible for this article and the European Commission and the Region Calabria disclaim any responsibility for use that may be made of the information contained therein.

# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015.00464


*Posidonia oceanica* in response to simulated salinity increases in a laboratory mesocosm system. *Estuar. Coast. Shelf. S.* 92, 286–296. doi: 10.1016/j.ecss.201 1.01.003


**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.

*Copyright © 2015 Piro, Marín-Guirao, Serra, Spadafora, Sandoval-Gil, Bernardeau-Esteller, Fernandez and Mazzuca. 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.*

# Quantitative proteomics reveals role of sugar in decreasing photosynthetic activity due to Fe deficiency

Sajad M. Zargar <sup>1</sup> \*, Ganesh K. Agrawal 2, 3, Randeep Rakwal 2, 3, 4 and Yoichiro Fukao<sup>5</sup>

*<sup>1</sup> Centre for Plant Biotechnology, Division of Biotechnology, S K University of Agricultural Sciences and Technology of Kashmir, Srinagar, India, <sup>2</sup> Research Laboratory for Biotechnology and Biochemistry, Kathmandu, Nepal, <sup>3</sup> GRADE Academy Private Limited, Birgunj, Nepal, <sup>4</sup> Faculty of Health and Sport Sciences and Tsukuba International Academy for Sport Studies, University of Tsukuba, Tsukuba, Japan, <sup>5</sup> Department of Bioinformatics, Ritsumeikan University, Shiga, Japan*

Keywords: Arabidopsis, proteomics, Fe deficiency, sugar, photosynthesis

# Importance of Iron in Plant

### Edited by:

*Subhra Chakraborty, National Institute of Plant Genome Research, India*

### Reviewed by:

*Michael A. Grusak, United States Department of Agriculture-Agricultural Research Service, Children's Nutrition Research Center, USA*

> \*Correspondence: *Sajad M. Zargar,*

*smzargar@gmail.com*

### Specialty section:

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

Received: *23 April 2015* Accepted: *17 July 2015* Published: *03 August 2015*

### Citation:

*Zargar SM, Agrawal GK, Rakwal R and Fukao Y (2015) Quantitative proteomics reveals role of sugar in decreasing photosynthetic activity due to Fe deficiency. Front. Plant Sci. 6:592. doi: 10.3389/fpls.2015.00592* Iron (Fe) is an essential micronutrient and its deficiency is a serious nutritional problem for all living organisms. This is because Fe is not only a basic requirement in cellular functions such as the redox reactions in photosynthesis and respiration, but is also required in the enzymatic processes like DNA replication, lipid metabolism, and nitrogen fixation in plants (Lan et al., 2011; Briat et al., 2015). As the photosynthetic apparatus contains much Fe, involved in many metabolic reactions in plastids, it becomes an important factor for survival of green plants. In plants, Fe deficiency can be observed by the development of chlorosis, which reduces the photosynthetic activity (Spiller and Terry, 1980; Terry, 1980; Straus, 1994; Briat et al., 2015).

# Proteomics Studies Related to Iron Deficiency

Proteomics is being increasingly used to expand our understanding of plant growth and development under both normal and stressful environmental conditions (Agrawal and Rakwal, 2008). Proteomic technology has also been employed as a powerful tool in the elucidation of metabolic rearrangements caused by Fe deficiency (López-Millán et al., 2013). Recently, quantitative proteomics approach was applied to understand the impact of Fe deficiency on plant metabolism; combined with physiological studies, the impact of Fe deficiency on photosynthesis was discerned (Zargar et al., 2013, 2015). Fe deficiency is known to alter both chloroplast structure and photosynthetic rate in higher plants as it alters the chlorophyll synthesis (Briat et al., 2015). The comparative proteome analysis of chloroplast thylakoids explains the plasticity of thylakoid membranes in response to Fe deficiency (Andaluz et al., 2006). A phosphoproteomic study of the thylakoid membrane proteome, from Fe-sufficient and Fe-deficient plants identified several proteins with post-translational modifications, that included, the doubly phosphorylated form of the photosystem II oxygen evolving complex, PSBH, ascorbate peroxidase, peroxiredoxin Q, and two major LHC IIb proteins (LHCB1 and LHCB2) (Laganowsky et al., 2009). Lan and coworkers have used the iTRAQ method to examine protein regulations involved in Fe homeostasis in Arabidopsis shoots (Lan et al., 2011). The abundance of 45 phosphoproteins was significantly changed upon Fe deficiency, which includes kinase A/calcium calmodulin-dependent kinase II, casein kinase II, and proline-directed kinase, indicating a possible critical function of these kinase classes in Fe homeostasis (Lan et al., 2012).

Recently, we applied the iTRAQ-OFFGEL method for understanding impact of Fe deficiency on photosynthesis and to unravel the proteome underlying the cross-talk between Fe deficiency and excess Zn in Arabidopsis (Zargar et al., 2015). Results revealed that Fe deficiency might lead to disruption of sugar synthesis and utilization.

# Iron Deficiency Influences the Photosynthetic Machinery and Sugar Levels: Proteomic Insights

The impact of Fe deficiency on photosynthesis in Arabidopsis has been very well documented (Zargar et al., 2013). Here we will majorly focus on the role of sugar in decreasing photosynthetic activity due to Fe deficiency. Two sugar transporters, major facilitator super family protein (STP13; AT5G26340) and sugar transporter 4 (STP4; AT3G19930) that have shown higher expression levels under Fe-deficient conditions were identified. STP13 and STP4 protein expressions were increased to 8.179- and 1.968-fold in Fe-deficient condition (Zargar et al., 2015). STP13 is known to be involved in transport of sucrose, glucose, and hexose (Saier et al., 1999; Lemoine, 2000; Norholm et al., 2006), while STP4 is a monosaccharide transporter (Fotopoulos et al., 2003). Further, we observed that the concentration of sucrose, fructose, and glucose were significantly increased in 2-weeks-old shoots of Arabidopsis grown on Fe deficient conditions compared to the control condition (Zargar et al., 2015). Thus, under Fe deficiency, a higher expression of sugar transporters as well as higher sugar concentration in shoots was observed. As such, Fe deficiency leads to accumulation of sugars in shoots, as synthesis and utilization of these sugars were not properly managed.

Past evidences have shown that root glycolytic (Zocchi, 2006; Jelali et al., 2010) and fermentation (Thimm et al., 2001) processes are enhanced under Fe deficiency, leading to sugar accumulation that derives from starch degradation and/or reorientation of photo-assimilate partitioning probably via sorbitol or sucrose (Loescher et al., 1990). Since these two sugar transporters are mainly expressed in roots and vascular bundle in shoots, these transporters may contribute to the

transport of sugars from mesophyll cell to vascular bundle for photosynthesis. Fe deficiency decreases photosynthetic activity, and as such sugar synthesis decreases. Therefore, the plant might need higher sugar levels to maintain fundamental metabolisms; hence sugars might be translocated from roots to shoots. Since STP13 was induced under stress condition, and involved in reabsorption of sugars from roots (Yamada et al., 2011), we presume that higher expression of sugar transporters might have a role in increasing sugar levels in shoots to maintain fundamental processes.

# Sensing the Role of Sugar

The down-regulated proteins due to Fe deficiency mostly include proteins involved in photosynthesis or ribosomal proteins. It has been well known that Fe deficiency largely affects protein synthesis in chloroplasts as compared to the cytoplasm, because chloroplastic mRNA and rRNA levels are significantly reduced (Spiller et al., 1987). In addition, the expression of various genes involved in different metabolic processes including photosynthesis is regulated by the sugar-driven signals (Sheen, 1990; Oswald et al., 2001). The negative correlation between sugar concentration and photosynthetic activity, and photosynthetic genes expression has also been reported earlier (Foyer, 1988; Sheen, 1990; Oswald et al., 2001). Therefore, the lower expression of photosynthetic genes under Fe-deficient conditions may be partly affected by high sugar concentration.

Aforementioned and other key proteins identified in our study were mapped onto metabolic and biological pathways as depicted in (**Figure 1**), and that explains the possible role of sugars in decreasing photosynthetic activity in

# References


Arabidopsis. Based on our results we believe that sugar might have a role in decreasing photosynthetic activity under Fe deficiency conditions. Further, we presume that Fe deficiency in Arabidopsis might lead to reduction in phloem unloading in sink tissues due to which sugars get accumulated in the shoots. Moreover source tissues load solutes into the phloem, but the restricted unloading under Fe deficiency may lead to sugar accumulation, which in turn has a negative effect on the expression levels of proteins involved in photosynthesis. There is also a possibility of sugar signaling involvement in the inhibition of photosynthesis. For example, cells under Fe-deficient conditions lead to decrease in photosynthesis by inducing sugar signaling, which might have role in decreasing expression of proteins involved in photosynthesis. Despite the above evidences and discussion therein, we are of the opinion that further intensive studies will be required linking physiology, biochemical processes with sugar signaling and regulation of genes involved in carbohydrate metabolism, transport, and partitioning.

# Funding

This work was supported by a Grant-in-Aid for Organelle Differentiation as the Strategy for Environmental Adaptation in Plants for Scientific Research of Priority Areas (No. 19039022 to YF) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; a Grant-in-Aid for Scientific Research from Nara Institute of Science and Technology supported by The Ministry of Education, Culture, Sports, Science and Technology, Japan. SZ acknowledges the DBT, New Delhi, India for award of CREST, Overseas fellowship to undertake this research.


**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.

Copyright © 2015 Zargar, Agrawal, Rakwal and Fukao. 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.

# **Protein profile of** *Beta vulgaris* **leaf apoplastic fluid and changes induced by Fe deficiency and Fe resupply**

*Laura Ceballos-Laita1, Elain Gutierrez-Carbonell 1, Giuseppe Lattanzio1, Saul Vázquez 1, Bruno Contreras-Moreira2, 3, Anunciación Abadía1, Javier Abadía1 and Ana-Flor López-Millán1*<sup>∗</sup>

*<sup>1</sup> Plant Stress Physiology Group, Department of Plant Nutrition, Aula Dei Experimental Station, Consejo Superior de Investigaciones Científicas, Zaragoza, Spain, <sup>2</sup> Laboratory of Computational and Structural Biology, Aula Dei Experimental Station, Consejo Superior de Investigaciones Científicas, Zaragoza, Spain, <sup>3</sup> Fundación ARAID, Zaragoza, Spain*

### *Edited by:*

*Jesus V. Jorrin Novo, University of Cordoba, Spain*

### *Reviewed by:*

*Natalia V. Bykova, Agriculture and Agri-Food Canada, Canada Jesus V. Jorrin Novo, University of Cordoba, Spain Arkadiusz Kosmala, Institute of Plant Genetics of the Polish Academy of Sciences, Poland*

### *\*Correspondence:*

*Ana-Flor López-Millán, Aula Dei Experimental Station, Consejo Superior de Investigaciones Científicas, Department of Plant Nutrition, Avenida Montañana 1005, 50059 Zaragoza, Spain anaflor@eead.csic.es*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

> *Received: 30 December 2014 Accepted: 23 February 2015 Published: 18 March 2015*

### *Citation:*

*Ceballos-Laita L, Gutierrez-Carbonell E, Lattanzio G, Vázquez S, Contreras-Moreira B, Abadía A, Abadía J and López-Millán A-F (2015) Protein profile of Beta vulgaris leaf apoplastic fluid and changes induced by Fe deficiency and Fe resupply. Front. Plant Sci. 6:145. doi: 10.3389/fpls.2015.00145* The fluid collected by direct leaf centrifugation has been used to study the proteome of the sugar beet apoplastic fluid as well as the changes induced by Fe deficiency and Fe resupply to Fe-deficient plants in the protein profile. Plants were grown in Fe-sufficient and Fe-deficient conditions, and Fe resupply was carried out with 45μM Fe(III)-EDTA for 24 h. Protein extracts of leaf apoplastic fluid were analyzed by two-dimensional isoelectric focusing-SDS-PAGE electrophoresis. Gel image analysis revealed 203 consistent spots, and proteins in 81% of them (164) were identified by nLC-MS/MS using a custom made reference repository of beet protein sequences. When redundant UniProt entries were deleted, a non-redundant leaf apoplastic proteome consisting of 109 proteins was obtained. TargetP and SecretomeP algorithms predicted that 63% of them were secretory proteins. Functional classification of the non-redundant proteins indicated that stress and defense, protein metabolism, cell wall and C metabolism accounted for approximately 75% of the identified proteome. The effects of Fe-deficiency on the leaf apoplast proteome were limited, with only five spots (2.5%) changing in relative abundance, thus suggesting that protein homeostasis in the leaf apoplast fluid is well-maintained upon Fe shortage. The identification of three chitinase isoforms among proteins increasing in relative abundance with Fe-deficiency suggests that one of the few effects of Fe deficiency in the leaf apoplast proteome includes cell wall modifications. Iron resupply to Fe deficient plants changed the relative abundance of 16 spots when compared to either Fe-sufficient or Fe-deficient samples. Proteins identified in these spots can be broadly classified as those responding to Fe-resupply, which included defense and cell wall related proteins, and non-responsive, which are mainly protein metabolism related proteins and whose changes in relative abundance followed the same trend as with Fe-deficiency.

**Keywords: leaf apoplast, iron deficiency, proteome, sugar beet, two-dimensional electrophoresis**

# **Introduction**

Iron is the fourth most abundant element in the earth's crust and it is an essential micronutrient for all living organisms including plants. However, its low availability in neutral or alkaline soils, which account for approximately 30% of the world's arable soils, causes Fe deficiency (Abadía et al., 2004). Iron deficiency is the most common nutritional disorder in many plants and typical symptoms include chlorosis of young leaves (leaf yellowing) and reduced crop yield and quality, which result in increased orchard management costs (Álvarez-Fernández et al., 2006; Rombolà and Tagliavini, 2006; Abadía et al., 2011).

Plants have developed two main mechanisms to allow Fe uptake from the soil: a strategy based on Fe (III) chelation (Strategy II) used by graminaceous plants, and a strategy used by non-graminaceous plants based on Fe(III) reduction (Strategy I) (Römheld and Marschner, 1986; Curie and Briat, 2003; Abadía et al., 2011). When challenged with Fe shortage, Strategy I plants such as *Beta vulgaris* increase the activity of several enzymes at the root plasma membrane level. These changes are aimed at increasing Fe uptake and include increases in a Fe(III) reductase (FRO, Ferric Reductase Oxidase; Robinson et al., 1999), an Fe transporter (IRT1, Iron Regulated Transporter) which introduces Fe(II) into the root cell (Eide et al., 1996; Fox and Guerinot, 1998) and an H+-ATPase that lowers the pH of the rhizosphere increasing soil Fe solubility (Santi et al., 2005; Santi and Schmidt, 2008, 2009). Also, several changes occur at the metabolic level in order to support the increased demand of energy and reducing power of Fe-deficient Strategy I roots (Zocchi, 2006). These changes include increased activity of the glycolytic pathway and TCA cycle, shifts in the redox state of the cytoplasm and in the mitochondrial electron transport chain (Schmidt, 1999; López-Millán et al., 2000b; Zocchi, 2006; Vigani, 2012).

While it is well-known that Fe is transported to the shoot *via* xylem, complexed by citrate (López-Millán et al., 2000a; Rellán-Álvarez et al., 2010), the mechanisms for Fe loading and unloading from the vasculature system are not yet fully understood. These processes could take place *via* parenchyma cells or by passive diffusion to the apoplastic space driven by transpiration (Abadía et al., 2011). Also, Fe uptake by mesophyll cells is not as well-studied as in roots. An Fe-reductase activity has been detected in leaf cells and protoplasts (Nikolic and Römheld, 1999; González-Vallejo et al., 2000; Jeong and Connolly, 2009) and AtFRO6 has been located in leaf PM-membranes (Mukherjee et al., 2006; Jeong et al., 2008). However, *fro6* mutant plants do not display any Fe-deficiency symptoms (Jeong and Connolly, 2009) therefore suggesting the existence of other reducing mechanisms. Factors such as differences in apoplastic pH and carboxylate concentrations as a result of Fe deficiency may also regulate leaf Fe reductase activity. On the other hand, light has also been proposed to directly photoreduce Fe (III)-citrate complexes in the leaf apoplast (Nikolic and Römheld, 2007).

The apoplast is a free diffusional space outside the plasma membrane that occupies less of 5% of the plant tissue volume in aerial organs (Steudle et al., 1980; Parkhurst, 1982) and the root cortex (Vakhmistrov, 1967). Among other important functions, such as transport and storage of minerals (Starrach and Mayer, 1989; Wolf et al., 1990; Zhang et al., 1991) or signal transmission (Hartung et al., 1992), the apoplast plays a major role in plant defense (Pechanova et al., 2010). Given that the composition of the apoplastic fluid results from the balance between xylem and phloem transport and mesophyll cell uptake processes, small changes in these fluxes could produce large changes in the solute concentrations in the apoplast. Changes in the apoplastic composition have been described in biotic and abiotic stresses such as Fe deficiency, air pollutants, heavy metal toxicity, drought, salinity, and extreme temperature (Griffith et al., 1992; Brune et al., 1994; Covarrubias et al., 1995; Dietz, 1997; López-Millán et al., 2000a; Fecht-Christoffers et al., 2003). For instance, Fe deficiency causes a slight decrease in the pH of the apoplast and has a strong impact on the carboxylate composition, with major increases in the concentrations of citrate and malate (López-Millán et al., 2000a). These Fe-deficiency induced changes in the leaf apoplast chemical environment have been suggested to play a role in Fe homeostasis (López-Millán et al., 2000a).

Apoplastic fluid isolation is always carried out using some degree of force (e.g., vacuum perfusion, leaf centrifugation, or pressure using a Schölander bomb), therefore leading to the presence of some cytosolic components in the samples (Lohaus et al., 2001). This contamination may be originated by the rupture of a certain fraction of the leaf mesophyll cells, or, alternatively, by the rupture of the plasmodesmata that communicate neighboring cells. The degree of contamination is usually assessed using cellular marker enzymes such as cytosolic malate deshidrogenase (c-mdh) or other cytoplasmic or internal organelle components, with values ≤3% considered acceptable (Dannel et al., 1995; Lohaus et al., 2001).

Proteomic approaches are useful to understand the general effect that a given stress exerts on metabolic processes (Jorrín-Novo et al., 2009). These approaches have been previously used to study the effects of Fe deficiency in several plant tissues, including thylakoids and roots (López-Millán et al., 2013). Most of the leaf apoplastic proteomic studies so far have been devoted to the study of the protein profile or the effect of biotic stresses. However, knowledge about the effects of nutritional stresses such as Fe deficiency in the apoplastic fluid protein profile is still very limited. Therefore, the aim of this study was first to obtain an overview of the leaf apoplast proteome in sugar beet plants and second to characterize the changes induced by Fe deficiency and resupply in the protein profile of this compartment, with the final goal of shedding light into the major processes taking place in the apoplast and the effect of Fe deficiency on them.

# **Material and Methods**

# **Plant Material and Growth Conditions**

Sugar beet (*Beta vulgaris* L. cv. Orbis) was grown in a growth chamber with a photosynthetic photon flux density (PPFD) of 350μmol m−<sup>2</sup> s−<sup>1</sup> PAR, 80% relative humidity and a photoperiod of 16 h, 23◦C/8 h, 18◦C day/night regime. Seeds were germinated and grown in vermiculite for 2 weeks. Seedlings were grown for an additional 2 weeks period in half-strength Hoagland nutrient solution with 45μM Fe(III)-EDTA, and then transplanted to 20 L plastic buckets (four plants per bucket) containing half-strength Hoagland nutrient solution with either 0 or 45μM Fe(III)-EDTA. Iron-free nutrient solutions were buffered at pH 7.7 with 1 mM NaOH and 1 g L−<sup>1</sup> of CaCO3. Young leaves from plants grown for 10 d in the presence and absence of Fe were used to collect the apoplastic fluid in all experiments. In the Fe-resupply experiment, 45μM Fe (III)-EDTA was added to the nutrient solution of plants grown for 10 d in the absence of Fe. The apoplastic fluid of these plants was collected 24 h after Fe addition.

### **Experimental Design**

The experiment was repeated four times with independent sets of plants. Each batch of plants consisted of four buckets per treatment with four plants per bucket. Apoplastic fluid was isolated from the four to five youngest leaves of each plant. Cytosolic contamination was assayed in each sample as described below. Samples with less than 3% of cytosolic contamination from a given treatment per batch were pooled together and considered as a biological replicate (*n* = 4).

### **Collection of Leaf Apoplastic Fluid**

Leaf apoplastic fluid was obtained from whole sugar beet leaves by direct centrifugation (López-Millán et al., 2000a). Briefly, leaves were excised at the base of the petiole and each leaf was rolled and placed into a plastic syringe barrel. Leaf-filled syringes were first centrifuged at low speed (2500 g, 4◦C, 15 min) to remove the xylem sap from the main vein and the fluid collected was discarded. A second centrifugation was then carried out at 4◦C, 4000 g for 15 min and the fluid collected was considered as soluble apoplastic fluid. The activity of cytosolic malate dehydrogenase (c-mdh; EC 1.1.1.37) in the collected fluid was measured immediately and used as a cytosolic contamination marker. The activity of c-mdh was measured spectrophotometrically at 340 nm in a final reaction mixture containing 46.5 mM Tris (pH 9.5), 0.1 mM NADH, 0.4 mM oxaloacetate and 5μL of apoplastic fluid and referred to activity measured in a whole leaf extract (López-Millán et al., 2000a). For the activity in whole leaf extracts, approximately 0.1 g of leaf tissue was homogenized with 2 mL of a buffer (pH 8.0) containing 100 mM HEPES, 30 mM sorbitol, 2 mM DTT, 1 mM CaCl2, 1% (w/v) bovine serum albumin and 1% (w/v) polyvinylpyrrolidone. The supernatant was collected and analyzed immediately after a 10 min centrifugation at 10,000 g.

### **Protein Extraction**

Proteins in approximately 1 mL of apoplastic fluid were precipitated by adding five volumes of cold 10% TCA. Samples were incubated for at least 14 h at 4◦C and then centrifuged at 10,000 g for 15 min. The pellet was washed twice with cold methanol, dried with N2 gas and solubilized in a sample rehydration buffer containing 8 M urea, 2% (w/v) CHAPS, 50 mM DTT, 2 mM PMSF and 0.2% (v/v) IPG buffer pH 3–10 (GE Healthcare, Uppsala, Sweden). After rehydration, samples were incubated in a Thermomixer Confort device (Eppendorf AG, Hamburg, Germany) at 29◦C and 1000 rpm during 3 h, then centrifuged at 10,000 ×*g* for 10 min at RT and filtered (0.45μm ultrafree-MC filters, Millipore, Bedford, USA). Protein concentration in the samples was quantified immediately with the Bradford method in microtiter plates using an Asys UVM 340 spectrophotometer (Biochrom Ltd., Cambridge, UK) and BSA as standard.

# **2-DE**

A first dimension IEF separation was carried out on 7 cm ReadyStrip IPG Strips (BioRad, Hercules, CA, USA) with a linear pH gradient 3–10 in a Protean IEF Cell (BioRad). Strips were passively rehydrated for 16 h at 20◦C in 125μL of rehydration buffer containing 80μg of apoplast proteins and a trace of bromophenol blue and then strips were transferred onto a strip electrophoresis tray. IEF was run at 20◦C, for a total of 14,000 V h (20 min with 0–250 V linear gradient; 2 h with 250–4000 V linear gradient and 4000 V until 10,000 V h). After IEF, strips were equilibrated for 10 min in equilibration solution I [6 M urea, 0.375 M Tris-HCl, pH 8.8, 2% (w/v) SDS, 20% (v/v) glycerol, 2% (w/v) DTT] and for another 10 min in equilibration solution II [6 M urea, 0.375 M Tris-HCl pH 8.8, 2% (w/v) SDS, 20% (v/v) glycerol, 2.5% (w/v) iodoacetamide]. For the second dimension, polyacrylamide gel electrophoresis (SDS-PAGE) and equilibrated IPG strips were placed on top of vertical 12% SDS-polyacrylamide gels (8 × 10 × 0*.*1 cm) and sealed with melted 0.5% agarose in 50 mM Tris-HCl (pH 6.8) containing 0.1% SDS. SDS-PAGE was carried out at 20 mA per gel for approximately 1.5 h at 4◦C, until the bromophenol blue reached the plate bottom, in a buffer containing 25 mM Tris Base, 192 mM glycine, and 0.1% SDS. Gels were subsequently stained with Coomassie blue G-250 (Serva, Barcelona, Spain). For each treatment, gels were made from four independent apoplast protein extracts (*n* = 4), each of them obtained by pooling the apoplastic fluid collected from 5 to 6 leaves in a given batch.

### **Gel Image and Statistical Analysis**

Stained gels were scanned with an Epson Perfection 4990 Photo Scanner (Epson Ibérica, Barcelona, Spain), previously calibrated using the SilverFast 6 software (LaserSoft Imaging AG, Kiel, Germany) using an IT8 reference card. Experimental MR values were calculated by mobility comparisons with Precision Plus protein standard markers (BioRad) run in a separate lane on the SDSgel, and pI was determined by using a 3–10 linear scale over the total dimension of the IPG strips. Spot detection, gel matching and interclass analysis were performed with PDQuest 8.0 software (BioRad). Normalized spot volumes based on total intensity of valid spots were calculated for each 2-DE gel and used for statistical calculations of protein abundance; for all spots present in the gels, pI, Mr, and normalized volumes (mean values and SD) were determined. Only spots present in all four replicates from at least one treatment were considered as consistent and used in further analysis. The spots were also manually checked, and a high level of reproducibility between normalized spot volumes was found in all four different replicates.

Univariate and multivariate statistical analyses were carried out. Protein response ratios were defined as the relative abundance in a given treatment divided by the relative abundance in the control; when ratios were lower than one the inverse was taken and the sign changed. Spots changing in relative abundance were defined using a Student *t*-test and a significance level of *p <* 0*.*05. Among these, only protein species with mean response ratios above 2.0 or below −2.0 were considered relevant and are discussed in this study. Principal component analysis (PCA) analysis was also carried out, using Statistical software (v. 10.0) and including only those spots showing differential accumulation as a result of the Fe-deficiency and Fe-resupply treatments.

# **Protein in Gel Digestion**

Consistent spots were excised automatically using a spot cutter EXQuest (BioRad), transferred to 500μL Protein LoBind Eppendorf tubes, distained in 400μL of 40% [v/v] acetonitrile (ACN) and 60% [v/v] 200 mM NH4HCO3 for 30 min and dehydrated in 100% ACN for 10 min. Gel pieces were dried at RT and then *in gel* digested with 15μL Trypsin solution (Sequencing Grade Modified Trypsin V511, Promega, Madison, WI, US; 0.1μg μL−<sup>1</sup> in 40 mM NH4HCO3/9% ACN). After incubation for 5 h at 37◦C, the reaction was stopped by adding 1μL of 1% TFA. The peptide solution was finally analyzed using mass spectrometry (MS).

# **Reference Repository of Beet Protein Sequences**

Proteomes of five sequenced beet accessions (RefBeet, KDHBv, YMoBv, UMSBv and YTiBv) were downloaded from http://bvseq.molgen.mpg.de/Genome/Download, corresponding to gene models annotated as of February 2013. In addition, all *B. vulgaris* protein sequences annotated in Uniprot (www.uniprot.org) were retrieved, and added to the set, which was subsequently filtered to remove redundant sequences with software CD-HIT (http://www.ncbi.nlm.nih.gov/pubmed/ 23060610) with cutoff -c 1.0 and otherwise default parameters. The final non-redundant set contained 82,368 protein sequences.

# **Protein Identification by Nano-Liquid Chromatography-Tandem Mass Spectrometry (nLC-ESI-MS/MS)**

Peptides present in 6μL of sample were pre-concentrated on line onto a 300μm i.d. × 5 mm, 5μm particle size ZORBAX 300SB-C18 trap column (Agilent Technologies, Waldbronn, Germany), using a 100μL min−<sup>1</sup> flow rate of 3% ACN, 0.1% formic acid, in a nano-HPLC system 1200 series (Agilent Technologies). Backflow elution of peptides from the trap column was carried out, and separation was done with a 75μm i.d. × 150 mm, 3.5μm particle size ZORBAX 300SB-C18 column (Agilent Technologies), using a 300 nL min−<sup>1</sup> nano-flow rate and a 55 min linear gradient from solution 97% A (0.1% formic acid) to 90% of solution B (90% ACN, 0.1% formic acid). The nano-HPLC was connected to a HCT Ultra high-capacity ion trap (Bruker Daltoniks, Bremen, Germany) using a PicoTip emitter (50μm i.d., 8μm tip i.d., New Objective, Woburn, MA, USA) and an on line nano-electrospray source. Capillary voltage was ×1.8 kV in positive mode and a dry gas flow rate of 10 L min−<sup>1</sup> was used with a temperature of 180◦C. The scan range used was from 300 to 1500 m/z. The mass window for precursor ion selection was ±0.2 Da and the rest of parameters were those recommended by the manufacturer for MS/MS proteomics work. Peak detection, deconvolution and processing were performed with Data Analysis 3.4 software (Bruker Daltoniks).

Protein identification was carried out using the Mascot search engine v. 2.5.0 (Matrix Science; London, UK) and the non-redundant *B. vulgaris* 20140811 (82,368 sequences; 28,127,547 residues), NCBI 20130310 (23,641,837 sequences; 8,123,359,852 residues), and Plants\_EST EST\_114 20140804 (158,278,518 sequences; 27,948,288,346 residues). Search parameters were: monoisotopic mass accuracy, peptide mass tolerance ±0.2 Da, fragment mass tolerance ±0.6 Da, one allowed missed cleavage, fixed modification carbamidomethylation (Cys), and variable modification oxidation (Met). Positive identification was assigned with Mascot *P*-values below the threshold (*p <* 0*.*05), at least two identified peptides with a score above homology, 10% sequence coverage and similar experimental and theoretical MW and pI values. We used the GO biological process annotation (http://www.geneontology.org) for classification of each individual protein identified. The secretion of apoplastic proteins was predicted using TargetP (www.cbs.dtu.dk/services/TargetP) (Emanuelsson et al., 2007), and SecretomeP (www.cbs.dtu.dk/services/SecretomeP) analysis to predict classical and non-classical secreted proteins, respectively (Bendtsen et al., 2004, 2005).

# **Quantitative RT-PCR**

Expression of chitinase and thaumatin genes was analyzed by qRT-PCR in two batches of plants. Total RNA from *B. vulgaris* leaves was isolated using the RNeasy Plant mini kit from QIAGEN (QIAGEN Inc., Valencia, CA, USA) according to the manufacturer's instructions. Samples were treated with DNAsa (recombinant DNase from Macherey-Nagel, Düren, Germany) to remove contaminating genomic DNA. The concentration of RNA and cDNA was determined with a Nanodrop system (Thermo Fisher Scientific, Waltham, MA, USA) and the structural integrity the RNA was checked using non-denaturing agarose gel stained with SybrSafe (Invitrogen, Carlsbad, CA, USA). One μg of total RNA was reverse transcribed to cDNA by using the SuperScript III reverse transcriptase and 2.5μM poly(dT)20 primer in a final volume of 20μl according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). Quantitative real time polymerase chain reactions were performed in a AB7500 Fast Real-Time PCR system (Applied Biosystems by Life Technologies, Grand Island, New York) with equal amount of cDNA for all samples and 10μl SYBR green master mix (Applied Biosystems, Warrington, UK) using gene specific primers and two different housekeeping genes (actin and tubulin) in a final volume of 20μl. Primer sequences and fragment sizes are listed in Table S1. The qPCR program was 50◦C for 2 min, 95◦C for 10 min, 40 cycles of 95◦C for 15 s and 60◦C for 1 min; and a final dissociation stage of 95◦C for 15 s, 60◦C for 1 min, and 95◦C for 30 s. A previous experiment was performed to assess for primer efficiency with different sets of primers for each target gene. Primer efficiencies of the chosen sets are listed in the Table S1.

# **Low-Temperature Scanning Electron Microscopy**

Leaf pieces were mounted on aluminum stubs with adhesive (Gurr<sup>R</sup> , optimum cutting temperature control; BDH, Poole, UK), cryo-fixed in slush nitrogen (−196◦C), cryo-transferred to a vacuum chamber at −180◦C, and then fractured using a stainless steel spike. Once inside the microscope, the samples underwent superficial etching under vacuum (−90◦C, 120 s, 2 kV), and then were overlaid with gold for observation and microanalysis. This freeze-fracture procedure leads to cell rupture only at the fracture plane, whereas the general internal leaf structure is well-preserved. Fractured samples were observed at low temperature with a digital scanning electron microscope (Zeiss DSM 960, Oberkochen, Germany) using secondary and back-scattered electrons. Secondary electron images (1024 × 960 pixels) were obtained at 133 eV operating at a 35◦ take-off angle, an accelerating voltage of 15 kV, a working distance of 25 mm and a specimen current of 1–5 nA. Microscopy was run in the Institute of Agricultural Sciences-CSIC (ICA-CSIC), Madrid, Spain.

## **Results**

Sugar beet plants showed Fe-deficiency symptoms 5 days after the treatment onset, with a marked yellowing of the younger leaves (Table S2). A freeze-fracture electron microscopy micrograph provided a representative image of a *B. vulgaris* leaf, with the apoplastic space surrounding mesophyll cells, as well as the epidermal cells and the minor vein tissues (**Figure 1**). The micrograph also shows the presence of plasmodesmata that communicate neighboring cells.

The apoplastic fluid collected from these leaves was assayed for c-mdh activity and only samples with contamination levels *<*3% (mean 1.72%, expressed on a total leaf activity basis) were used for 2-DE protein profiling (Table S2). Typical protein extraction yields ranged between 0.4 and 0.8μg protein μL−<sup>1</sup> of apoplastic fluid (Table S2).

### **Protein Profiles of the Apoplastic Fluid**

The protein profile of apoplast extracts from *B. vulgaris* leaves was studied by 2-DE IEF-SDS-PAGE electrophoresis. Experimental conditions allowed for the separation of proteins within pI and MW ranges from 3.5 to 8 and from 18 to 106 kDa, respectively. Typical real scans of 2-DE gels obtained from apoplastic fluid protein extracts of Fe-sufficient (+Fe), Fedeficient (−Fe), and Fe-resupplied Fe-deficient plants (−FeR) are shown in **Figures 2A–C**, respectively. The average number of detected spots (mean ± SD) was 210 ± 12, 216 ± 11, and 211 ± 20 in gels from plants grown in +Fe, −Fe, and −FeR conditions, respectively (Table S3 and **Figure S1**). The total number of spots consistently detected in the whole experiment (present in all four gels of at least one treatment) was 203. A composite averaged virtual map containing all consistent spots is shown in **Figure 2D**. All consistent spots were analyzed by MS, and proteins were unambiguously identified in 78% of the cases (158 spots) (Table S4 and **Figure S1**). A large percentage (97%) of identifications was achieved using the beet custom reference repository. To identify UniProt entries, BLAST searches (*E-*values *<* 1e-30) of the unambiguously identified protein hits were run when needed. This approach revealed a high degree of redundancy in the identified protein species. When duplicates (same UniProt entry) were removed, the 158 identified proteins were reduced to 109 nonredundant proteins and this protein set was considered as the leaf apoplastic protein profile (**Table 1**). However, it should be noted that there may be still certain degree of redundancy left, since some hits correspond to the same protein description but from different plant species (**Table 1**).

The distribution of non-redundant proteins according to the biological process in the sugar beet leaf apoplast indicated that the major functional categories within the apoplastic proteome were C metabolism (25%; 27 protein species), stress and defense

### **TABLE 1 | Non-redundant proteome of the leaf apoplastic fluid of sugar beet plants.**




*aNumber of spots with the same protein description.*

*<sup>b</sup> Protein description.*

*<sup>c</sup> UniProt entries sharing same protein description.*

*<sup>d</sup> TargetP algorithm predictions: C, M, S, and—indicate chloroplast, mitochondrion, secretory pathway and any other location, respectively.*

*<sup>e</sup> SecretomeP algorithm predictions: CS, nCS and—indicate classical secreted, non-classical secreted, and non-secreted proteins, respectively.*

*<sup>f</sup> Description of the GO: P (biological process) term.*

(21%; 23 proteins), and protein related processes (19%; 21 proteins), followed by cell wall related processes (9%, 10 proteins)

(**Table 1**, **Figure 3A**).

From the non-redundant leaf apoplastic proteins, 26% (28 protein species) were predicted to have a signal peptide sequence using the TargetP or SecretomeP softwares, whereas SecretomeP revealed that 38% (41 proteins) were assigned to non-classical secreted proteins lacking a signal peptide (**Table 1**, **Figure 3B**).

### **Effect of Fe-Deficiency and Fe Resupply on the Leaf Apoplastic Fluid Protein Profile**

The statistical analysis (*p <* 0*.*05; *t*-Student) of averaged 2- DE maps indicated that 8% (16 spots) of the total number of consistent spots changed significantly and *>*2-fold in relative abundance in the whole experiment (spots labeled 1–16 in **Figure 2D**). Among them, 88% of the spots (14) matched reliably to known proteins and were identified by database searches (1–14, **Figure 2D**). Their metabolic functions were assessed manually by GO annotation and identified proteins were classified into five functional categories: cell wall related processes (4 spots, 2 proteins), protein metabolism (4 proteins), and defense, amino acid and C-related pathways, with two proteins each (**Table 2**). The principal component analysis of differentially accumulated spots showed a good separation between treatments, with the first and second components explaining approximately 48 and 38% of the variation, respectively (**Figure 4**).

When the protein profile of -Fe samples was compared to that of the +Fe samples, only five spots changed in relative abundance. Four of them increased, including three spots (spots 1–3) identified as chitinase and one (spot 9) identified as a thaumatin-like protein. Only one spot (spot 13) decreased as a result of Fe deficiency and it was identified as carbonic anhydrase (**Figure 2D**, **Table 2**).

In the comparison of −FeR vs. +Fe, eight spots changed in relative abundance, four of them increased whereas the other four decreased (**Figure 2D**, **Table 2**). Two of the spots increasing in relative abundance were identified as chitinase and thaumatin (spots 2 and 9, respectively) whereas the other two could not be identified (spots 15 and 16). The spots decreasing in relative abundance were identified as a cysteine proteinase RD19 like protein (spot 5), a peptidyl-prolyl *cis*-*trans* isomerase (spots 6, 7) and a serine hydroxymethyltransferase (spot 11) (**Table 2**).

When the -FeR samples were compared to the –Fe ones, six spots changed in relative abundance (**Table 2**). Among them, one spot was detected *de novo* (spot 14) and identified as the 23 kDa OEC protein and a second one (spot 15) could not be identified. Four spots showed significant decreases in relative abundance (**Table 2**, **Figure 2D**), and they were identified as <sup>β</sup>xylosidase/alpha-L-arabinofuranosidase (spot 4), a cytosolic heat shock 70 protein (spot 8), lactoylglutathione lyase (spot 10) and glutamine synthetase (spot 12).

### **Relative Transcript Abundance of Target Genes**

The chitinase and thaumatin-like 1 proteins identified in spots 1 and 9 (**Table 2** and Table S4, **Figure 2**), respectively, showed the highest increases in relative abundance as a result of Fedeficiency and were selected to study transcript abundances using q-PCR. Sequences containing the peptides matched during protein identification, KDHBv\_S14175\_58500.t1 for chitinase and BQ584258 for the thaumatin-like 1 protein, were selected to design primers for specific amplification of target genes (Table S1). In Fe-deficient leaves, the relative abundances of chitinase and thaumatin-like 1 protein transcripts increased 3- and 2 fold, respectively, when compared to controls (**Figure 5**), whereas upon Fe resupply the relative abundance of chitinase transcript was higher and that of thaumatin-like 1 protein was not significantly different at *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*05 (**Figure 5**). When compared to the Fe-deficient controls, changes in transcript abundances upon Fe-resupply were not statistically significant at *p <* 0*.*05 (**Figure 5**).

## **Discussion**

### **Leaf Apoplastic Protein Profile**

The 2-DE proteomic approach allowed us to resolve 203 spots, with 158 of them (78%) being identified and 109 accounting for non-redundant proteins. These results are similar to those reported for gel-based leaf apoplastic proteome studies in other plant species, which ranged between 93 and 470 spots, with an average of 200 spots in most of the studies (**Table 3** and references therein). Functional classification of the non-redundant leaf apoplastic proteins of sugar beet indicates that stress and defense, protein metabolism, cell wall and C metabolism account for approximately 75% of the identified proteome (**Figure 3A**).

Stress and defense related proteins accounted for 21% of the non-redundant sugar beet apoplastic proteome. Similar to what has been reported in other plant species (references in **Table 3**), proteins identified were peroxidases, osmotin-like, thaumatinlike, and pathogenesis-related proteins. Our results also indicate the presence in the leaf apoplast of enzymes participating in defense against oxidative stress (two peroxiredoxins, CuZnSOD and two enzymes from the ascorbate-glutathione cycle: ascorbate peroxidase and monodehydroascorbate reductase) and in the detoxification of methylglyoxal (two UniProt entries described as glyoxalase I). These proteins have also been found in the leaf apoplast from poplar (Pechanova et al., 2010) and are also present in fluids from the vascular tissue (Lattanzio et al., 2013; Lucas et al., 2013). The presence of this wide spectrum of defense proteins in non-stressed plants has been attributed to a preformed defense that creates a hostile environment for pathogens (Pechanova et al., 2010; Delaunois et al., 2013).

The contribution of protein metabolism-related proteins to the sugar beet apoplastic proteome (19%, with 14 of the 21 UniProt entries being proteases) is within the range reported in other studies using grapevine, Arabidopsis, and rice (16–20%; Floerl et al., 2012; Delaunois et al., 2013; Kim et al., 2013), but higher than percentages described for other plant species, which range from 6 to 10% (Boudart et al., 2005; Goulet et al., 2010; Pechanova et al., 2010). The presence of such a high number of proteolysis-related proteins in the apoplast has been proposed to be species-dependent (Delaunois et al., 2013). Subtilisinlike, serine-carboxypeptidases, and aspartic proteases found in this study have been consistently described in the apoplast of several plant species (Goulet et al., 2010; Floerl et al., 2012;

### **TABLE 2 | Spots showing differences in relative abundance (Student** *t***-test** *p <* **0***.***05) as a result of Fe deficiency and Fe resupply.**


*aSpot number as in Figure 1.*

*bSpot number as in Table S3.*

*c–eFold change in the Fe-deficient vs. Fe sufficient, Fe resupplied vs. Fe-sufficient and Fe-resupplied vs. Fe-deficient comparisons, respectively, values in bold indicate significant changes and when the ratios were lower than one the inverse was taken and the sign changed.*

*f Protein description. gPlant Species.*

*hUniprot entry (*\* *denotes protein entry in UniParc).*

*i GO:P term description.*

*j SecretomeP algorithm predictions: CS, nCS and—indicate classical secreted, non-classical secreted and non-secreted proteins, respectively.*

Delaunois et al., 2013). In addition, our results indicate that sugar beet apoplast also contains several subunits of the proteasome (six UniProt entries), which is involved in the ubiquitin dependent degradation of damaged and miss-folded proteins (Kurepa and Smalle, 2008). These apoplastic proteases may play a role in plant defense against pathogens and also in signaling (Van Der Hoorn and Jones, 2004; Xia et al., 2004; Pearce et al., 2010).

The cell wall related category accounted for 9% (10 proteins) of the sugar beet leaf proteome and included nine glycoside hydrolases. Glycoside hydrolases modify cell walls by metabolizing carbohydrate compounds from plant cell polysaccharides and by interacting with hemicellulases and pectic enzymes (Numan and Bhosle, 2006; Minic, 2008). The percentage of cell wall related proteins found in sugar beet is similar to those reported in grapevine and tobacco but lower than those found in rice, poplar or maize (**Table 3**). On the other hand, the percentage of proteins participating in C-related processes in the leaf apoplast of sugar beet was slightly higher (25%, 27 proteins) than those reported in other plant species (**Table 3** and references therein). However, these values vary depending on the functional classification of certain proteins, such as peroxidases that are included in defense or cell wall depending on the study, and on the consideration of carbohydrate metabolism and cell wall as one or two functional categories.

The relatively high percentage of C-related proteins found in the sugar beet leaf proteome may have several causes. First, a large number of the spots identified as proteins participating in C metabolism have a low spot intensity and therefore they are over-represented in the functional categorization based on protein number. Second, the sugar beet apoplastic proteome was obtained by direct leaf centrifugation, whereas most of the proteomes from other plant species were obtained using leaf vacuum infiltration followed by centrifugation (**Table 3** and references therein); this later technique may be somewhat better at preventing leakage of proteins from the cytoplasm (Lohaus

et al., 2001; Witzel et al., 2011). Interestingly, the percentage of C-related proteins in the stem apoplast of poplar (18%) as well as the identity (TCA and glycolysis-related) (Pechanova et al., 2010), are similar to those found in sugar beet (25%). This comparison might suggest that the sugar beet apoplastic fluid obtained by direct centrifugation of the leaves could have a high contribution of the fluid contained in the xylem sap vessels of the main vein of the leaf.

The presence of cytoplasmic contamination in our samples was always lower than 3%, with an average value of 1.5%, as assessed by the activity of c-mdh (Table S2). However, *in silico* analysis of the non-redundant proteome using the TargetP and the Secretome P algorithms (for classical and non-classical secretory proteins) predicted 26% (28 proteins) and 38% (41 proteins) of the non-redundant proteome as classical and nonclassical secretory proteins, respectively, whereas 36% of the proteins were predicted to be non-secretory (**Table 1**, **Figure 3B**). This percentage (64%) of secretory proteins is within the range of those reported in the leaf apoplast proteome of other plant species (from 50% in Arabidopsis to 80% in grapevine and poplar; Casasoli et al., 2008; Pechanova et al., 2010; Delaunois et al., 2013). The fact that our samples contain a relative large number of proteins tagged as non-secretory is not surprising, since the c-mdh assay indicates that up to 3% of the leaf cells could have delivered cytosolic components into the isolated apoplastic fluid. Furthermore, the release of cytosolic components may be in part associated to the rupture of the pasmodesmata that exist in these leaves (**Figure 1**), even if mesophyll cells remain intact. On the other hand, RuBisCO was identified in six spots of the total 203 of the leaf apoplastic proteome map (Table S4). In addition there were 10 more spots whose identification yielded proteins related to the Calvin cycle and are, most likely, plastid located (**Table 1**). Therefore, we could assume that at least 16 spots (approximately 8% of the total) are probably a result of cell leakage. These results may suggest


**290**

Frontiers in Plant Science | www.frontiersin.org March 2015 | Volume 6 | Article 145 |



**References Plant specie Technique Leaf material Study Separation Number of spots Analysis Changes Functional classification in control** Petriccione et al., 2014 *Actinidia deliciosa* VIC-(100mM Tris-HCl, pH 7.5, 10 mM KCl, 1mM phenylmethanesul- fonyl fluoride) Leaf without midrib Pseudomonas infection 7 cm 3-10 ca. 220 60 Not reported Kim et al., 2014 *Oryza sativa* CA-VIC-(200 mM CaCl2, 5 mM Na-acetate, pH 4.3) 5 cm leaf cuts Fungus infection Shotgun 470 174 secreted proteins by in silico analyses Not reported *VIC stands for vacuum infiltration followed by centrifugation.*

that the real contamination by cell rupture is likely to be higher than that estimated by the use of c-mdh as a contamination marker.

### **Changes Induced by Fe Deficiency in the Leaf Apoplastic Proteome**

The largest part of the changes caused by Fe-deficiency and Feresupply corresponded to proteins tagged as secretory proteins (10 spots), probably corresponding to true components of the apoplast. Changes also occurred in a relatively small number (four) of the apoplastic fluid proteins tagged as non-secretory ones, and possibly associated to cell or plasmodesmata rupture (these are marked by ∗ in the following paragraphs).

Iron-deficiency caused changes in the relative abundance of five spots (2.5% of the consistent spots of the leaf apoplast proteome), suggesting that protein homeostasis in the leaf apoplast fluid is well-maintained upon Fe shortage. This number of changes is markedly low when compared to those induced by Fe deficiency in other proteomes, including those of roots and thylakoids of sugar beet plants (44 and 53%, respectively; Andaluz et al., 2006; Rellán-Álvarez et al., 2010) and falls within the lower range of the number of changes caused by other abiotic stresses in the leaf apoplast (**Table 3** ; Dani et al., 2005; Casasoli et al., 2008).

Three spots identified as chitinase increased in relative abundance as a result of Fe-deficiency (spots 1–3; **Figure 2**, **Table 2**). These three spots had the same molecular weight and different pIs (8.2, 7.1, and 7.6), indicating that Fe-deficiency alters the isoform pattern of this enzyme, with one of the isoforms identified *de novo* in Fe-deficient samples. Chitinases are hydrolytic enzymes that break down glycosidic bonds, removing xylosyl residues of xyloglucan oligosaccharides in the cell wall (Sampedro et al., 2001). Therefore, these increases in chitinase suggest the existence of Fe deficiency-induced cell wall modifications. This is in agreement with the changes elicited by Fe deficiency in leaf morphology, which include reduction of xylem vessel size (Fernández et al., 2008; Eichert et al., 2010). Furthermore, changes in lignification have been reported in roots of Fe-deficient pear and quince cultivars (Donnini et al., 2009) and cell wall related proteins commonly show changes in abundance in proteomic studies of Fe deficient plants (see López-Millán et al., 2013 for a review).

A spot identified as a thaumatin-like protein 1 (spot 9) also increased in relative abundance as a result of Fe-deficiency. Thaumathins are pathogenesis-related (PR) proteins from the PR5 subfamily, which are induced by biotic and abiotic stresses. Some members of PR5 subfamily have been described to play distinctive roles in the defense system that protects against high-salt stress or osmotic imbalance (Tachi et al., 2009), which is likely to occur in the Fe-deficiency treatment in the presence of CaCO 3 . A PR5b protein also showed increases in abundance in roots and stems of Fe-deficient *M. truncatula* plants grown in the presence of CaCO 3 (Rodríguez-Celma et al., 2011), suggesting that thaumatins are ubiquitously up-regulated by Fe deficiency. Interestingly, both chitinase and the thaumatin-like 1 protein were also affected in the leaf proteome of cowpea submitted to Mn toxicity (Fecht-Christoffers et al., 2003).

**TABLE 3 | Continued**

Only one protein, identified as carbonic anhydrase (CA; spot 13), decreased as a result of Fe deficiency. Carbonic anhydrase interconverts CO2 and bicarbonate to maintain the acid-base balance. A decrease in CA activity could be attributed to the presence of bicarbonate in the nutrient solution that may reach the leaf apoplast *via* xylem (Nikolic and Römheld, 2007). Although some mammalian CA isoforms are extra-cellular and have been described in saliva and milk (Karhumaa et al., 2001; Leinonen et al., 2001), and CA is classified as a non-classical secretory protein by SecretomeP, plant isoforms are distinct from an evolutionary standpoint and have been mainly localized in the chloroplast or the cytosol.

### **Changes Induced by Fe Resupply in the Leaf Apoplastic Proteome**

Iron resupply to Fe deficient plants caused significant changes in the relative abundance of 13 spots when compared to either Fe-sufficient or Fe-deficient plants. These spots can be roughly classified into two major groups. The largest group (seven spots) was composed by those spots increasing in relative abundance (significantly or not) with Fe-deficiency and decreasing significantly with Fe resupply when compared either with the control or with Fe-deficient samples. This group contained two cell wall related proteins [chitinase (spot 2), and β-xylosidase (spot 4)], three stress-related proteins [a heat shock 70 protein (spot 8∗), thaumatin-like 1 protein (spot 9), glyoxalase I (spot 10)], glutamine synthase (spot 12∗) and the unidentified spot 16. These results indicate that Fe resupply causes changes in the short-term (within 24 h) in cell wall and stressrelated processes of the Fe-resupplied plants toward values found in the Fe-sufficient controls. One more protein (spot 14∗, the 23 kDa OEC protein, which nuclear-encoded and synthesized in the cytosol) was also responsive to short term Feresupply but followed a different trend, not-detected in Fedeficient samples but detected upon Fe-resupply. This likely reflects transitory increases in the cytosolic levels of this protein upon short term Fe-resupply, which are necessary for the slight recovery of the photosynthetic system at this resupply stage (Larbi et al., 2004).

On the other hand, a second group of four spots (spots 5–7 and 11∗) decreased in relative abundance with Fe deficiency (although not significantly) and decreased significantly upon Fe-resupply when compared either with controls or Fedeficient samples. This group included two proteins identified as protein-metabolism related [a cysteine protease (spot 5), a peptidyl-prolyl *cis*-trans isomerase (spots 6,7)] and a serine hydroxymethyltransferase (spot 11∗). One more spot (spot 15, unidentified protein) followed the opposite behavior (increased). This group of proteins can be classified as not responsive to short term Fe-resupply, since they may require a time longer than 24 h to reset to control values after resupply.

# **Concluding Remarks**

In summary, this study provides information on the composition of the apoplast proteome in *B. vulgaris* leaves, which appears to be quite similar to that of other previously studied plant species. The study shows that Fe deficiency and Fe resupply cause significant changes in a limited number of proteins in the leaf apoplast, and none of them is expected to play a significant role in metal homeostasis. This is in contrast with the intense changes previously found in the concentrations of metabolites (e.g., carboxylates) that could interact with metals in the same compartment. Data found contribute toward the understanding of metal homeostasis, and in particular on the still poorly known mechanisms of Fe acquisition by plant mesophyll cells. Results presented here open an interesting line of work regarding possible modifications of cell wall that ultimately may affect permeability or transport of Fe across the plasma membrane.

# **Acknowledgments**

Work supported by the Spanish Ministry of Science and Competitiveness (MINECO; projects AGL2012-31988 and AGL2013- 42175-R (co-financed with FEDER), and the Aragón Government (group A03). LCL, EGC, GL, and SV were supported by FPI-MINECO, I3P-CSIC and FPI-MINECO predoctoral contracts and a JAE-CSIC postdoctoral contract (co-financed by the European Social Fund), respectively. This paper is dedicated to the memory of Mr. Fernando Pinto (ICA-CSIC), electron microscopy expert.

# **Supplementary Material**

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fpls.2015. 00145/abstract

**Figure S1 | Virtual composite image containing all consistent spots present in the real gels.** Numbers are as in Table S4 (SSP).

# **References**


Álvarez-Fernández, A., Abadía, J., and Abadía, A. (2006). "Iron nutrition in plants and rhizospheric microorganisms," in *Iron Deficiency, Fruit Yield and Fruit Quality*, eds L. L. Barton and J. Abadía (Dordrecht: Springer), 85–101.


**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.

*Copyright © 2015 Ceballos-Laita, Gutierrez-Carbonell, Lattanzio, Vázquez, Contreras-Moreira, Abadía, Abadía and López-Millán. 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.*

# Comparison of proteome response to saline and zinc stress in lettuce

### *Luigi Lucini\* and Letizia Bernardo*

*Institute of Environmental and Agricultural Chemistry, Università Cattolica del Sacro Cuore, Piacenza, Italy*

Zinc salts occurring in soils can exert an osmotic stress toward plants. However, being zinc a heavy metal, some more specific effects on plant metabolisms can be forecast. In this work, lettuce has been used as a model to investigate salt and zinc stresses at proteome level through a shotgun tandem MS proteomic approach. The effect of zinc stress in lettuce, in comparison with NaCl stress, was evaluated to dissect between osmotic/oxidative stress related effects, from those changes specifically related to zinc. The analysis of proteins exhibiting a fold change of 3 as minimum (on log 2 normalized abundances), revealed the involvement of photosynthesis (via stimulation of chlorophyll synthesis and enhanced role of photosystem I) as well as stimulation of photophosphorylation. Increased glycolytic supply of energy substrates and ammonium assimilation [through formation of glutamine synthetase (GS)] were also induced by zinc in soil. Similarly, protein metabolism (at both transcriptional and ribosomal level), heat shock proteins, and proteolysis were affected. According to their biosynthetic enzymes, hormones appear to be altered by both the treatment and the time point considered: ethylene biosynthesis was enhanced, while production of abscisic acid was upregulated at the earlier time point to decrease markedly and gibberellins were decreased at the later one. Besides aquaporin PIP2 synthesis, other osmotic/oxidative stress related compounds were enhanced under zinc stress, i.e., proline, hydroxycinnamic acids, ascorbate, sesquiterpene lactones, and terpenoids biosynthesis. Although the proteins involved in the response to zinc stress and to salinity were substantially the same, their abundance changed between the two treatments. Lettuce response to zinc was more prominent at the first sampling point, yet showing a faster adaptation than under NaCl stress. Indeed, lettuce plants showed an adaptation after 30 days of stress, in a more pronounced way in the case of zinc.

### Keywords: osmotic stress, oxidative stress, shotgun MS, photosystem I, phytohormone

## Introduction

Salt stress and NaCl in particular, is probably the most common abiotic stress affecting crop production and reducing yields. In response to salinity, plants evolved specific mechanisms to sense and contrast this stress at both metabolite and protein level. The proteins identified in different studies are involved in the major plant metabolic processes, such as photosynthesis, energy metabolism, ROS scavenging, and ion homeostasis, protein synthesis, nitrogen assimilation as well as the secondary oxidative stress (Parida and Das, 2005; Zhang et al., 2012). High levels of

### *Edited by:*

*Jesus V. Jorrin Novo, University of Cordoba, Spain*

### *Reviewed by:*

*Ing-Feng Chang, National Taiwan University, Taiwan Ana-Flor Lopez-Millan, Estacion Experimental de Aula Dei – Consejo Superior de Investigaciones Cientificas, Spain Jesus V. Jorrin Novo, University of Cordoba, Spain Klára Kosová, Crop Research Institute, Czech Republic*

### *\*Correspondence:*

*Luigi Lucini, Institute of Environmental and Agricultural Chemistry, Università Cattolica del Sacro Cuore, Via Emilia Parmense 84, 29122 Piacenza, Italy luigi.lucini@unicatt.it*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

> *Received: 30 December 2014 Accepted: 25 March 2015 Published: 16 April 2015*

### *Citation:*

*Lucini L and Bernardo L (2015) Comparison of proteome response to saline and zinc stress in lettuce. Front. Plant Sci. 6:240. doi: 10.3389/fpls.2015.00240* NaCl can be toxic for plants and cause stunted growth together with reduction in water potential (Borgognone et al., 2013).

Conversely, Zn2<sup>+</sup> is an essential micronutrient for plants, and it is therefore essential for plant growth, development, and many metabolic processes being one of the major cofactors, together with iron and manganese, in numerous enzymes (Andreini et al., 2008). More than 1200 proteins are predicted to contain, bind, or transport Zn2<sup>+</sup> (Hänsch and Mendel, 2009). Furthermore, the zinc-inorganic phosphate (Pi) relationship has been observed in numerous plant species such as tomato, cotton, barley, and *Arabidopsis* (Bouain et al., 2014). Besides being a micronutrient, zinc is also a heavy metal and can have detrimental effects on many vital processes in plant cells. Geological and/or anthropogenic activities can result in zinc concentration in soil above toxic levels for crops, leading to chlorosis, biomass reduction, and necrotic lesions on leaves. Nowadays, few studies have investigated the effects of zinc stress to plants by a proteomic approach. Zinc toxicity effects have been mainly investigated in roots of model plants by means of gel-based proteomic approaches (Gutierrez-Carbonell et al., 2013; Romeo et al., 2014). In *Arabidopsis thaliana* plants exposed to zinc at subtoxic levels, Barkla et al. (2014) found that the enzymes involved in one carbon metabolism and protein synthesis were involved in acclimation to heavy metal stress. At higher concentration, this species showed alteration of proteins related to oxidative stress, proteasome, and energy metabolism (Fukao et al., 2009). Even if a few studies dealing with hyper-accumulating plants have been reported (Barkla et al., 2014; Lefèvre et al., 2014), the knowledge about zinc toxicity in crops is still limited. Most of the work in agricultural species work has been done in sugar beet, demonstrating the imbalance of photosystems, oxidative stress, alteration of carboxylates trafficking, and low photosynthesis rates as a consequence of stomatal and mesophyll conductance to CO2 (Sagardoy et al., 2009, 2010, 2011).

Lettuce (*Lactuca sativa* L.) is one of the most common fresh-cut vegetables in the Mediterranean diet, containing healthpromoting phytochemicals such as phenolic compounds, vitamin C, and carotenoids (Garrido et al., 2014). This species has been proposed as a good model to study the zinc influence on plant growth, showing differential biomass and photosynthesis rate (Bouain et al., 2014). The zinc toxicity in lettuce has been related to carboxylate metabolism (Barrameda-Medina et al., 2014). However, lettuce is also moderately sensitive to salt stress, and exhibits reduced growth under saline conditions. Kim et al. (2008) have investigated the secondary metabolite profile of lettuce under high NaCl concentration, observing that biomass was reduced above 100 mM NaCl soil concentration for 15 days, in comparison to untreated plants.

Although plant response to salinity has been widely investigated in many species at protein level (reviewed by Zhang et al., 2012; Hossain and Komatsu, 2013), very little and fragmentary information is available regarding the response to zinc exposure. In this work, lettuce has been used as a model to investigate and compare salt and zinc stresses, aimed to understand the differences at proteome level and hence the modulation of metabolic pathways. In the study, a gel-free bottom-up proteomic approach was chosen to provide understanding of the molecular mechanisms underlying plant adaptation to zinc-contaminated soils, in comparison to salinity stress. Despite zinc salts in soil can represent a plant stressor related to changes in osmolality and water availability, as well as salinity does, this heavy metal might induce some additional and more specific responses. The assessment of the responses in common between the two stresses, together with the disjoint ones, can provide useful insights on the plant response to either zinc or salt contaminated soil, thus supporting crop production, and promoting a more efficient land use.

# Materials and Methods

## Plant Material and Mineral Analysis

Young plants of lettuce (*L. sativa* L. cultivar longifolia), about 15 cm in length, were purchased at a local nursery and transplanted into large pots (nine plants per pot at 6 cm distance, in a 36 L pot) filled with a commercial topsoil. Overall, six pots were prepared: two serving as control, two for NaCl, and two for zinc treatments, respectively. Plants were grown indoor, under natural light and at room temperature.

After 1 week of acclimation, two pots were irrigated with demineralized water (control), two with a 100 mM NaCl solution, and two with a 100 mM ZnSO4 solution, with the same volume twice a week. After both 15 and 30 days of treatment, a pot per treatment was withdrawn and three plant leaves per pot collected and pooled as biological replications. Therefore, three replicate samples of leaves (each of them taken from three plants) were harvested per treatment, at both 15 and 30 days.

Leaf samples were frozen in liquid N2 and ground into a fine powder using a mortar and pestle. A portion (100 mg) of each sample was mineralized in 0.5 mL of hydrogen peroxide and 2 mL of nitric acid, under heating for 6 h, diluted in 5% nitric acid, and then analyzed by inductively coupled plasma atomic emission spectroscopy. A multi-element source was used for calibration purposes and a certified reference material was analyzed before samples; each sample was read in duplicate.

# Protein Extraction

Tissue powder (125 mg) was suspended in 0.8 mL ice-cold SDS-phenol buffer (0.1 M Tris-HCl pH 8.0, 30% sucrose, 2% SDS, 2% DTT) and equal volume of Tris-buffered phenol pH 8.0, shaken and centrifuged at 15000 *g* for 5 min at 4◦C. The phenolic phase was collected and proteins were precipitated with 5 volumes of 0.1 M ammonium acetate in methanol, overnight at −20◦C. Samples were centrifuged for 15 min to 10000 *g* at 4◦C and then pellets were washed three times with 80% acetone. Finally, the dry pellets were resuspended in a buffer containing 7 M urea and 2 M thiourea. The protein concentration was measured with the Bio-Rad protein assay kit using bovine γ-globulin as standard, following to the manufacturer's instructions. Protein extracts were stored at −20◦C until use. Fifty micrograms of proteins were then reduced with dithiothreitol, alkylated with iodoacetamide, and digested with Trypsin (Promega, Madison, WI, USA) at 37◦C overnight.

### Tandem MS Analysis

Tryptic peptides were analyzed by a shotgun MS/MS approach using a hybrid quadrupole-time-of-flight (Q-TOF) mass spectrometer. With this purpose, an Agilent 6550 IFunnel Q-TOF mass spectrometer, with a nano LC Chip Cube source (Agilent Technologies, Santa Clara, CA, USA), was used. The chip consisted of a 40-nL enrichment column (Zorbax 300SB-C18, 5 μm pore size) and a 150 mm separation column (Zorbax 300SB-C18, 5 μm pore size) coupled to an Agilent Technologies 1200 series nano/capillary LC system and controlled by the MassHunter Workstation Acquisition (version B.04).

Peptides were loaded onto the trapping column at 2.6 μL min−<sup>1</sup> in 2% (v/v) acetonitrile and 0.1% (v/v) formic acid. After enrichment, the chip was switched to separation mode and peptides were backflush eluted into the analytical column, during a 150 min acetonitrile gradient (from 3 to 70% v/v) in 0.1% (v/v) formic acid at 0.3 μl min−1. The mass spectrometer was used in positive ion mode and MS scans were acquired over a range from 300 to 1700 mass-to-charge ratio at 4 spectra s−1. Precursor ions were selected for auto-MS/MS at an absolute threshold of 1000 and a relative threshold of 0.01%, with a maximum of 20 precursors per cycle and active exclusion set at 2 spectra (with release after 0.2 min). Analysis of MS/MS spectra for peptides identification was performed by protein database searching with Spectrum Mill MS Proteomics Workbench (Rev B.04; Agilent Technologies). Auto MS/MS spectra were extracted from raw data accepting a minimum sequence length of three amino acids and merging scans with the same precursor within a mass window of ±0.4 mass-to-charge ratio in a time frame of ±30 s. Search parameters were Scored Peak Intensity (SPI) ≥50%, precursor mass tolerance of ±10 ppm and product ions mass tolerance of ±20 ppm. Carbamidomethylation of cysteine was set as fixed modification and trypsin was selected as enzyme for digestion, accepting two missed cleavages per peptide. The search was conducted against the proteome of *L. sativa* (UniProt, downloaded October 2014); the database was concatenated with the reverse one. Auto thresholds were used for peptide identification in Spectrum Mill to achieve a target 1% false discovery rate. A labelfree quantitation, using the protein summed peptide abundance, was carried out after identification.

The results were directly exported to Mass Profiler Professional B.04 (Agilent Technologies) for statistics and pathway analysis. Protein intensities were log 2 normalized and baselined versus the control, and then ANOVA and fold-change analysis were done. Those proteins showing a normalized fold-change above 3 were considered and finally linked to *A. thaliana* metabolic pathways (source: WikiPathways) in the Pathway Architect add-in of Mass Profiler Professional, using their Uniprot ID annotation.

Differential proteins identified from Swiss-Prot were searched against AgriGO analysis tool to assign functional ontology information. The proteins with associated GOs were then loaded in REVIGO to visualize proteins GOs frequencies into functional processes against all *A. thaliana* GOs.

### Results

The different treatments affected the production of biomass, as plant mean weight was significantly reduced under saline conditions (**Figure 1**), whereas the detrimental effect of zinc on yield was less evident (ANOVA at alpha = 0.05, Student– Newman–Keuls *post hoc* test). However, besides the above mentioned reduction in yield, no symptoms of toxicity could be evidenced at phenotype level. Metals analysis in plant leaves, investigated through inductively coupled plasma atomic emission spectroscopy, revealed that zinc and sodium concentration increased as a consequence of the corresponding treatment. Zinc was accumulated in leaves, raising up to 468 mg/kg at fresh weight (three times and 12 times the level in control, after 15 and 30 days of treatment, respectively), while sodium levels reached 3.5% of leaf fresh weight (four times and 27 times the level in control, after 15 and 30 days of treatment, respectively). Therefore, both Zn2<sup>+</sup> and Na<sup>+</sup> were actually translocated to leaves, supporting metal-related variations at proteome level.

Changes at proteome level were investigated through a shotgun MS approach, with label-free quantitation. Only those proteins passing the validation step done using the Spectrum Mill search engine (auto-thresholds, FDR 1%), have been considered. Single peptide identification was conditional to the use of unique peptides, then the following filtering in Mass Profiler Professional let to identify and report those proteins being detected in two out of three replications, as minimum. Overall, 124 and 122 identified proteins were validated at the first (15 days) and second (30 days) sampling point, respectively. This dataset was then subjected to statistics and fold-change analysis. **Figure 2** summarizes the proteins detected in each treatment together with their functional ontology, while a detailed list of all identified proteins with their abundances and score is given as Supplementary Material.

Several studies are present in literature to uncover the plant response to salinity at molecular and physiological level. Salinity leads to a cellular osmotic adjustment to hinder the toxic effects

of Na+ and alters the salt-responsive proteins (involved in photosynthesis, energy metabolism, ROS scavenging, and ion homeostasis; Zhang et al., 2012). The increase in superoxide dismutase (SOD) and phenylalanine ammonia-lyase (PAL) played a key role to cope with the oxidative stress related to salinity. Several proteins involved in protein synthesis, turnover, and degradation were up accumulated in our experiments [i.e., ribosomal proteins, heat shock proteins, maturase K (matK)] together with proteins involved in hormonal signaling and in ethylene metabolism, such as the cysteine-rich RLK protein, 1-aminocyclopropane-1-carboxylic acid (ACC) synthase 1 (ACS1), ACC oxidase (ACO) and the transcription factor MYC2. Although the differential proteins identified from the salt-stress treatment (**Table 1**) were in common with those from zincstress, differences could be noted in abundance profiles and trend across time points. Our results from the NaCl trial are in agreement with previously published studies (Golldack et al., 2014). No novel findings were pointed out in comparison with previous works; therefore, our results are mainly focused on zincrelated changes at proteome level, while less attention has been given to salinity. Results are reported at two time points, 15 and 30 days after stress application, in order to point out an eventual adaptation of lettuce plant to high zinc concentrations in soil.

The initial elaborations using Volcano analysis, using a foldchange cut off value of 3 and a *p* = 0.05, gave limited results. Only three enzymes passed the thresholds after 15 days of stress, namely ATP synthase subunit a, 30S ribosomal protein S4, and glutamine synthetase(GS; Benjamini–Hochberg corrected *<sup>p</sup>* <sup>=</sup> 2.1 <sup>×</sup> <sup>10</sup>−3, 2.1 <sup>×</sup> <sup>10</sup>−3, and 1.3 <sup>×</sup> <sup>10</sup>−3, respectively), probably as a consequence of the high variability related to shotgun data-dependent analysis. Therefore, the proteomic results were discussed considering a fold-change threshold of 3 (on log 2 normalized abundances). This subset of proteins was loaded into the Pathway Architect add-in of Mass Profiler Professional. The list of proteins showing a fold-change of at least 3 in zinc-stressed plants only is reported in **Table 1**. Proteins are listed according to their accumulation trend and as a function of the time point considered. The Venn analysis on those proteins passing the fold-change cut off was next carried out, in order to dissect the changes specifically related to zinc toxicity from the variation induced by osmotic stress, being the former in common with soil salinity.

To strengthen the outcome from fold-change and Venn analysis, the dataset was subjected to multivariate Partial Least Square Discriminant Analysis (PLS-DA) in Mass Profiler Professional. The loadings used to build the class prediction model were plotted according to their weight within the latent vectors, and the most relevant ones in predicting zinc-treated samples (i.e., those having a score of above +0.1 rather than below −0.1) were exported and recorded. In **Figure 3**, the distribution of each replication within the hyperspace of PLS-DA is given together with the plot of the class prediction loadings used to model covariance structures.

Finally, statistical confirmation regarding the involvement of the proteins identified in fold-change has been achieved using the "find minimal entities" naïve Bayesian analysis in Mass Profiler Professional. According to this approach, a target number of 30 proteins able to better explain differences between zinc treated samples from the others (forward selection algorithm; evaluation metric: overall accuracy = 100) was identified.

Interestingly, both covariance-based PLS-DA and Bayesian analysis outcomes included proteins already pointed out in TABLE 1 | Proteins resulted down-accumulated and up-accumulated in lettuce leaf after 15 and 30 days under zinc stress conditions, considering a fold change above three on log 2 normalized intensities.



*Protein fold changes are listed in the table.* <sup>a</sup>*Proteins identified that are positive to PLS-DA statistical analysis.* <sup>b</sup>*Proteins identified that define naïve Bayesian analysis.*

the former fold-change analysis. Information regarding those proteins evidenced by multivariate statistics is given in **Table 1** together with results from fold-change analysis.

The results, grouped according to their ontology and biological meaning, are presented as follows more detailed.

# Photosynthesis and Electron Transport Chain

The Far-red impaired response protein-like protein, PSI chlorophyll *a* apoproteins A1 and A2 (PSI-A1 and PSI-A2)several NAD(P)H-quinone oxidoreductase subunits involved in the electron transport chain in chloroplast, exhibited changes in abundance as a consequence of zinc stress. Additionally, the photosystem I assembly protein YCF3 is required to assembly of PSI by interacting with its subunits (Naver et al., 2001). Also the biosynthesis of plastoquinone (PQ) might be altered under zinc exposure, although with contrasting results. The 37 kDa inner envelope membrane protein, a pivotal component in chloroplast development, and PQ biosynthesis (Motohashi et al., 2003), was down-accumulated at 15 days of treatment; conversely, the MPBQ/MSBQ transferase that methylates 2-methyl-6-phytylbenzoquinol for the production of tocopherols or PQ, was up-accumulated after 30 days of stress.

### Energy Metabolism

Besides these chloroplastic responses, the glycolytic pathway was also affected by the treatments considered. The increase in cytosolic triose phosphate isomerase (TPI) was consistent with the results from Riccardi et al. (1998) who related TPI accumulation to drought stress and water availability. An increase of TPI provides evidence in the formation of glyceraldehyde-3-phosphate, which can positively affect the energy level in plant and thus supports the plant growth (Sharma et al., 2012a). The up-accumulation of the downstream following glycolytic enzymes NAD(P)H-dependent glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and pyruvate kinase, common to salinity and zinc, confirms the increase in catabolic energy production.

### Transcription and Protein Metabolism

The clearest evidence regarding the involvement of nitrogen compounds in response to zinc stress was related to GS. This protein was up-accumulated under both salinity and zinc stress, suggesting an improved assimilation of ammonium.

Additionally, several ribosomal proteins changed their abundance under zinc stress, and two heat shock proteins related to protein folding (HSP70, HSP90) were up-accumulated in response to zinc; these molecular chaperones assist the correct folding and contrast protein degradation in various abiotic stresses. The HSP90 then decreases after 30 days of zinc stress, probably as a consequence of plant adaption.

The protein metabolism was affected by zinc at different levels. MatK, a protein involved in in post-transcriptional processes during chloroplast development, was down-accumulated at 15 days. A zinc finger-homeobox (ZF-HD) protein-like protein, acting as transcription factor in the DNA binding, was not detected at the first time point. Indeed, zinc plays a role in plastidial transcription as a cofactor for RNA polymerase and ZFs-containing nucleic acid-binding proteins, as well as the up-accumulated DNA-directed RNA polymerase.

In addition, the reduced level of plastidial fructokinase-like proteins, the target of plastidial thioredoxin TRX z, regulated the plastid-encoded RNA polymerase dependent transcription in plastids (Arsova et al., 2010).

The down-accumulation of putative ankyrin-like protein (AKR), related to transcriptional regulation and signal transduction, is known to play an important role in protein–protein interaction in biotic and abiotic responses (Voronin and Kiseleva, 2008). In *Arabidopsis*, a reduced expression of AKR has been associated to the regulation of antioxidant metabolism that is shared by both disease resistance and stress responses (Yan et al., 2002).

### Phytohormones

Besides signaling, the key enzymes for the biosynthesis of ethylene were among our differential results. ACS catalyzes the formation of ACC, and then converted in ethylene by ACO. Two different isoforms of ACS have been pointed out in this work: ACS1 was down-represented after 15 days of stress, while ACS2 was up-represented during all the experiment. ACS isoforms have different activities in response to different stimuli: ACS1 may function as regulator of other ACS enzymes (Wang et al., 2002). ACO1, however, was up-accumulated in both salinity and zinc stress. Finally, a putative ethylene receptor (ETR1) was also upaccumulated at the first sampling event in lettuce exposed to zinc. The enhanced biosynthesis of ethylene during the whole experiment is not surprising, taking into account that this hormone has been linked to both ROS and heavy metals stress (Steffens, 2014).

Furthermore, both abscisic acid (ABA) biosynthesis and degradation were strongly affected by zinc treatment. ABA is synthetized in the cytosol from xanthoxal and is related to developmental processes, responses to abiotic stress (drought, cold, salt, and wounding), seed development, and dormancy. The carotenoid cleavage dioxygenases (CCDs) catalyze the selective oxidative cleavage of carotenoids, leading to apocarotenoids and then to ABA. CCDs can exert their role by modulating the phytohormones network (Cazzonelli, 2011). In our experiments, CCD2 presented a reduced level of expression starting from 15 days of treatment, while CCD1 decreased at 30 days. The nine-*cis*epoxycarotenoid dioxygenase (NCED), a subfamily of CCD for ABA biosynthesis exhibiting different specificity for the double bound that it cleaves (Lu and Li, 2008), presented a trend consistent with CCDs. NCED3 (related to environmental stress response) and NCED2 were down accumulated, while NCED4 was up-accumulated; however, this latter seems to be involved in thermal inhibition and no evidences are reported regarding water stress (Huo et al., 2013). The P450 monooxygenase enzyme ABA 8-oxidase (ABA8ox), involved in the oxidative degradation of ABA, was found up-accumulated.

The transcription factor MYC2 is a basic helix-loop-helix type protein that modulates different aspects of jasmonic acid signal transduction by regulating the transcription of its target genes (Zhai et al., 2013), is involved in mediating the cross-talk between jasmonate, ethylene, gibberellin, and light signaling, and it is related to ABA response (Finkelstein, 2013). According to our results, MYC2 decreased during zinc stress consistently to ABA biosynthesis, which is negatively regulated.

In association to ABA, the gibberellins metabolism was engaged in this study consistently with DELLA proteins accumulation. Gibberellins are known to regulate, among others, the response to abiotic stress. The down-accumulated germacrene A (GA) insensitive Dwarf1 A (GID1) protein is a gibberellin nuclear receptor that can interact with DELLA proteins to bind the active GA form. In agreement with the accumulation of DELLA proteins, ent-kaurene oxidase 1 (KO), catalyzing the second step for the plastidial synthesis of GA (Swain et al., 2005), was down-accumulated.

### Membrane Transport and Cell Wall Metabolism

Two enzymes belonging to lignin biosynthesis were differentially measured after zinc exposure: pectin acetylesterase (PAE) was up-accumulated while a putative cellulose synthase (CesA) was down-accumulated.

Besides cell wall, transport across the membrane was also affected by zinc treatment. Plants can cope with osmotic stress by means of the water channels aquaporins; the putative PIP2 aquaporin was identified in reduced abundance, while PIP1 aquaporin was up-accumulated under both zinc and salinity stress. The ABC cassette transporters YCF1 and MDR-like *P*-glycoprotein were also up-accumulated.

### Phenylpropanoids Biosynthesis

The phenylpropanoids pathway was altered as a response to zinc stress, although with differences between phenolic classes. Surprisingly PAL, the upstream enzyme in phenylpropanoids biosynthesis, was measured at reduced amounts in plants although flavonoid accumulation is induced by saline stresses (Colla et al., 2013). The same trend was observed for 4-coumarate-CoA ligase (4CL), the key enzyme in chalcone synthesis, under both salinity and zinc stress. Conversely, the flavanone-3-beta-hydroxylase (F3H), another enzyme of the phenylpropanoid pathway that catalyzes the formation of flavonols, was found up-accumulated at both time points. Finally, the increased amount of cinnamate-4-hydroxylase (C4H) suggests the biosynthesis of hydroxycinnamic acids.

# Other Responses to Osmotic/Oxidative Stress

Several enzymes related to secondary compounds synthesis changed their abundance in zinc-stressed plants, like for tocopherol cyclase (TC) that exhibited low accumulation.

Regarding sesquiterpene lactone biosynthesis, two key enzymes were up-accumulated after exposure to zinc. The increase in GA synthase (GAS) LTC1, an enzyme involved in the biosynthesis of secondary metabolites such as phytoalexins, was observed. The second sesquiterpene lactone biosynthetic enzyme that was found up-accumulated is costunolide synthase (CYP71BL2), supposed to act as a phytoalexin and to modulate oxidative stress. Consistently, the degrading enzyme GA oxidase (GAO), catalyzing the formation of an acid group in GA, was decreased under salinity and zinc stress.

The accumulation of proline can be expected considering the up-accumulation of delta 1-pyrroline-5-carboxylate synthetase (P5CS).

Among plant antioxidants, L-ascorbate has a key role in the mitigation of ROS. The L-galactono-1,4-lactone dehydrogenase (L-GalLDH) increase, converting L-GalL into ascorbate.

The terpenoids biosynthetic pathway was also affected by the stress considered. The enzyme isopentenyl pyrophosphate:dimethyllallyl pyrophosphate isomerase (IDI), a protein that catalyzes the isomerization of isopentenyl pyrophosphate and dimethylallyl pyrophosphate in terpenoid biosynthesis, was increased during the experiments.

Finally, some resistance proteins like the resistance gene candidate (RGC) and the non-specific lipid transfer protein (nsLTP) were also imbalanced in response to zinc exposure.

# Discussion

## Zinc-Related Effects

The comparison of results from zinc-treated and salt-treated plant samples let to identify those changes at proteome level, specifically induced by zinc. The up-accumulation of Far-red impaired response protein-like protein (a positive regulator of the genes for porphobilinogen synthase) and PSI-A1 and PSI-A2 indicate an increase in chlorophyll biosynthesis. Therefore, the plant photosynthetic process was strongly affected, mainly at chlorophyll biosynthesis and at photosystem I level. Coherently, the reduction in plastocyanin and photosystem II reaction center protein H, suggested a modulation of photosynthesis toward PSI. Since PSI is the terminal electron carrier in the chloroplast, coupled to ROS-scavenging and/or photophosphorylation in the chloroplast (Sharma et al., 2012b), these findings can be correlated to the osmotic stress. A confirmation of this hypothesis was found in the concomitant up-accumulation of ATP synthase subunits.

The glycolytic process was stimulated by the treatment with zinc, probably as a supply of energy substrates. Analogously, ammonium assimilation (through stimulation of GS) was induced by zinc in soil. This latter step provides with an essential nitrogen precursor for almost all nitrogenous compounds; indeed, the key role of GS in contrasting saline stress has been pointed out by Teixeira and Fidalgo (2009). Coherently, transcription and protein synthesis, as well as proteolysis, were strongly affected by the treatment with zinc. Proteolysis is a part of the regulatory mechanism for a broad spectrum of cellular processes affecting not only the stability of key metabolic enzymes but also on the removal of terminally damaged polypeptides. Several of these processes are connected to ATP-dependent proteases. In plants, the balance of cystatin/cysteine protease plays an essential role in protein turnover and in the response to biotic/abiotic stresses (Benchabane et al., 2010). In our experiments, both cystatin and cysteine protease were up-accumulated after zinc exposure.

As expected, the hormone network seems to be altered by the treatments. Several proteins involved in plant hormone metabolism and cross-talking were identified in our experiments, including ABA, gibberellin, and ethylene metabolism. According to our results, the crosstalk of plant hormone signaling should have a pivotal role in stress hindering and osmotic adaptation. Besides the specific stimulation of ethylene biosynthesis enzymes, ABA biosynthesis was induced after 15 days of stress, via the up-accumulation of some CCD and NCED proteins. At the same time point GA synthesis did not show changes. Then, ABA synthesis seemed to decrease markedly, while GA followed the same trend but with a less evident decrease. The coordinated change in ABA/GA ratio we expect from biosynthetic enzymes was consistent with the assumption regarding their antagonistic role in plant response to stresses. GIGANTEA (GI), a protein related to GA signaling and involved in oxidative stress tolerance, cold stress responses, and carbohydrate metabolism (Molas et al., 2006), was down accumulated in response to zinc. These findings are consistent with previous literature reporting decreased amounts of GI under salt treatment (Park et al., 2013). The kinase receptor cysteinerich RLK, involved in hormonal signaling in several species, was up-accumulated, suggesting that signal transduction is also involved.

Membrane trafficking was affected at both aquaporins and ABC cassette transporters level. The accumulation of PIP1 aquaporin only, a transporter actually reported as the more efficient water channel (Kaldenhoff and Fischer, 2006), was coherent with the treatment applied. ABC transporters play an important role in growth, plant nutrition, and development, response to abiotic stress and stimuli. It has been reported that they might have a possible role in metal accumulation in the vacuole, as detoxification mechanism (Luque-Garcia et al., 2011).

Although phytochelatins are described to play an important role in heavy metal detoxification in plants, phytochelatin synthase 1 (PCS1) was down-accumulated under both salinity and zinc stress. Interestingly, some authors (Lee et al., 2003) reported that the overexpression of PCS led to hypersensitivity toward cadmium and zinc stress in *Arabidopsis*.

Several other enzymes acting in secondary metabolism at different levels have been regulated in lettuce as a response to zinc exposure. TC is involved in tocopherol synthesis at the inner chloroplast envelope level (Cheng et al., 2003), representing a plant response to deal with the accumulation of ROS then protecting from lipid peroxidation. Surprisingly, this enzyme was down-accumulated throughout the whole experiment, in agreement to the low abundance of the 37 kDa inner envelope membrane protein.

Other enzymes involved in secondary metabolism included sesquiterpene lactones, proline, ascorbate and terpenoids biosynthesis, and therefore are consistent with zinc stress. The phenylpropanoids pathway likely switches from flavonoids to flavonols and hydroxycinnamic acids synthesis, thus accumulating these latter. Sesquiterpene lactone phytoalexins were also induced in lettuce leaf, as well as resistance proteins and proline biosynthesis. Considering that LTC1 converts farnesyl diphosphate to GA, a Mg2<sup>+</sup> ion binding compound, its role in zinc stress might be related to GA metal binding capacity and/or to the phytoalexin activity of sesquiterpene lactones.

The accumulation of proline has the role of protecting proteins and quenching ROS, by acting as organic osmolyte and contributing to osmotic adjustment (Borgognone et al., 2013). The role of ascorbate in stress mitigation can be also related to ROS scavenging, being oxidative stress a consequence of osmotic stress.

Although the role of terpenoids in contrasting osmotic stresses has been already reported (Lucini et al., 2015), IDI is a divalent metal ion-requiring enzyme and seems to bind Mg2<sup>+</sup> and Zn2+. Therefore, its involvement in the response to zinc could be actually related to the stress induced.

Summarizing, the coordinate response to zinc stress included more expected responses, like the imbalance of hormone network rather than the need to cope with osmotic and oxidative stress, as well as some more specific and more surprising changes like the modulation of the photosynthetic activity and the involvement of proteolysis.

### Changes During Time

The proteomic response of lettuce throughout our experiment, suggests that the plant actually exhibits an adaption to stress over time. Overall, 58 proteins (34 of which could be ascribed to zinc only) were up-accumulated after 15 days of stress, to decrease after 30 days (27 proteins up-regulated, 12 of which related to zinc). Regarding the down-regulated ones, 43 proteins were identified at 15 days (23 those related to zinc only) while 62 showed the same trend (25 those from zinc stressed plants only) after 30 days.

Some key processes, like those catalyzed by GS and ATP synthase (both up-accumulated) and PCS (down-accumulated), did not show variation in trend during the experiment. However, the most of the proteins changed their abundance during the time course.

### Early Response to Stress

The involvement of PSI electron transport chain was evident after 15 days of stress, where chlorophyll *a* apoproteins and far-red impaired response protein-like protein were up-accumulated, to decrease then at 30 days. At this latter time point, these proteins abundance in stressed plants was even below the level of control plants. Consistently, plastocyanin followed an opposite trend than chlorophyll apoproteins and far-red impaired response protein-like protein.

Sesquiterpene lactones synthesis was induced at 15 days, as proven by the increase in both GAS (concurrent to a down accumulation of GAO and costunolide synthase, to return at the same level as in the control, at 30 days. Similar trend was observed for resistance proteins, CesA and heat-shock proteins (excepting NBS-LLRs that were down accumulated at both time points and HSP90 that decreased below the reference control at 30 days). Analogously, the alcohol dehydrogenase (ADH) mediated coping of reactive nitrogen species, was observed at the earlier time point. The energy supply via glycolysis was stimulated at 15 days of stress, being TPI and GAPDH up-accumulated; the former was then down accumulated after 30 days.

### Late Response to Stress

The accumulation of some enzymes, known to be implicated in plant coping with stress, was induced after 30 days of stress. The biosynthesis of ascorbate via L-GalLDH and the synthesis of MDR-like ABC cassette are both induced at the later sampling point.

Phenylpropanoids biosynthetic pathway was downaccumulated at both time points, being both 4CL and PAL reduced, although an increase in hydroxycinnamic acids can be forecast at 30 days, where C4H level increased.

## Comparison of Zinc-Related versus Salinity-Related Changes

Although the proteins involved in the response to zinc stress and to salinity were substantially the same, their abundance changed between the two treatments. As a general consideration, lettuce response to zinc was more prominent at the first sampling point, yet showing a faster adaptation than under NaCl stress after 30 days.

Regarding hormone profile, it can be pointed out that the down accumulation of GID1 A protein occurred already after 15 days of stress in zinc, while was later in salt stressed plants. Consistently, DELLA 2, a protein being accumulated under low GA levels, was up-accumulated in zinc-stressed plants only. These evidences suggest that zinc stress has a more prominent effect on GA metabolism than salinity. In addition, ethylene biosynthetic enzymes were more abundant under zinc conditions.

The increased abundance of PIP2 aquaporin and MDR-like glycoprotein ABC cassette in zinc-stressed plants after 15 days, followed by reduced amounts after 30 days, suggest that lettuce better managed to cope with ion-related electrolyte-water imbalance under zinc stress than under saline conditions. Similarly, the induction of sesquiterpene lactones, proline, and ascorbate was more pronounced in zinc stressed plants, being biosynthetic enzymes up-accumulated after 15 days; in all cases, except for the latter one, protein abundances decreased to basal level after 30 days.

The enhanced engagement of PSI was more marked in zinctreated plants after 15 days, as well as ADH, glycolytic enzymes and ATP synthase; again, the plant showed an adaption to stress and in most of the cases enzymes decreased to be less abundant under zinc than NaCl conditions.

Finally, in agreement with the general trend above reported, heat shock proteins were more abundant in zinc-stressed lettuce at 15 days, to decrease at 30 days, when GS became down accumulated.

# Conclusion

The effect of zinc stress in lettuce, in comparison with salt stress, was investigated at proteome level via shotgun data-dependent mass spectrometry. The analysis of proteins exhibiting a fold change of 3 as minimum (on log 2 normalized abundances), revealed the involvement of photosynthesis, via stimulation of chlorophyll synthesis and enhanced role of photosystem I, as well as stimulation of photophosphorylation. Glycolytic supply of energy substrates, together with ammonium assimilation (through formation of GS) were also stimulated by zinc in soil. Similarly, protein metabolism (at both transcriptional and ribosomal level), heat shock proteins, and proteolysis were affected.

# References


The treatment and the time point considered affected plant hormone profile: ethylene biosynthesis was enhanced throughout the whole experiment, while production of ABA was induced at the earlier time point to decrease markedly, and gibberellins were decreased at the later time point.

Besides aquaporin PIP2 synthesis, other osmotic/oxidative stress related compounds were enhanced under zinc stress, i.e., proline, hydroxycinnamic acids, ascorbate, sesquiterpene lactones, and terpenoids biosynthesis.

Most of the proteins pointed out were in common between zinc and NaCl stress, although the Zn-related response was quantitatively different and anticipated in time. Finally, lettuce plants showed an adaptation after 30 days of stress that was more evident in the case of zinc.

# Supplementary Material

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fpls.2015.00240/ abstract


Kim, H. J., Fonseca, J. M., Choi, J. H., Kubota, C., and Kwon, D. Y. (2008). Salt in irrigation water affects the nutritional and visual properties of romaine lettuce (*Lactuca sativa* L.)*. J. Agric. Food Chem.* 56, 3772–3776. doi: 10.1021/jf0733719


crop performance of lettuce grown under saline conditions. *Sci. Hortic.* 182, 124–133. doi: 10.1016/j.scienta.2014.11.022


**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.

*Copyright © 2015 Lucini and Bernardo. 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.*

# Chernobyl seed project. Advances in the identification of differentially abundant proteins in a radio-contaminated environment

*Namik M. Rashydov1 and Martin Hajduch2\**

*<sup>1</sup> Department of Biophysics and Radiobiology, Institute of Cell Biology and Genetic Engineering, National Academy of Sciences of Ukraine, Kiev, Ukraine, <sup>2</sup> Department of Developmental and Reproduction Biology, Institute of Plant Genetics and Biotechnology, Slovak Academy of Sciences, Nitra, Slovakia*

### *Edited by:*

*Joshua L. Heazlewood, The University of Melbourne, Australia*

### *Reviewed by:*

*Harriet T. Parsons, University of Copenhagen, Denmark Neil James Willey, University of the West of England, UK*

### *\*Correspondence:*

*Martin Hajduch, Department of Developmental and Reproduction Biology, Institute of Plant Genetics and Biotechnology, Slovak Academy of Sciences, Akademicka 2, P.O. Box 39A, Nitra, Slovakia hajduch@savba.sk*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 19 December 2014 Accepted: 22 June 2015 Published: 06 July 2015*

### *Citation:*

*Rashydov NM and Hajduch M (2015) Chernobyl seed project. Advances in the identification of differentially abundant proteins in a radio-contaminated environment. Front. Plant Sci. 6:493. doi: 10.3389/fpls.2015.00493* Plants have the ability to grow and successfully reproduce in radio-contaminated environments, which has been highlighted by nuclear accidents at Chernobyl (1986) and Fukushima (2011). The main aim of this article is to summarize the advances of the Chernobyl seed project which has the purpose to provide proteomic characterization of plants grown in the Chernobyl area. We present a summary of comparative proteomic studies on soybean and flax seeds harvested from radio-contaminated Chernobyl areas during two successive generations. Using experimental design developed for radiocontaminated areas, altered abundances of glycine betaine, seed storage proteins, and proteins associated with carbon assimilation into fatty acids were detected. Similar studies in Fukushima radio-contaminated areas might complement these data. The results from these Chernobyl experiments can be viewed in a user-friendly format at a dedicated web-based database freely available at www*.*chernobylproteomics*.*sav*.*sk.

Keywords: soybean, flax, ionizing radiation, ecology, experimental design, seed filling, 2-DE, mass spectrometry

# Introduction

Radioactive minerals have accumulated on the Earth's surface since early Achaean times (3500–4000 million year ago) and probably helped precipitate and concentrate organic carbonrich matter (Parnell, 2004). The first scientific recordings indicating that radioactivity affects living matter dates back to late 19th and early 20th century when Marie Skłodowska-Curie mentioned in her thesis that "The action of radium upon the skin can take place across metal screens, but with weakened effect" (Richards, 1915). Similarly, Henri Becquerel observed negative effects of radioactivity on his own body, after he carried a small tube of impure radium in his pocket for a few hours (Baskerville, 1905). Early experiments on the effect of ionizing radiation (IR) on plants were performed during late 19th and early 20th century (Gager, 1908). It was soon realized that radiation is a powerful mutagen (Nadson and Philippov, 1925), can induce variations within species (Goodspeed and Olson, 1928; Olson and Gilbert, 1928), and can control rates of mutations (Babcock and Collins, 1929).

Plants can easily cope with increased levels of IR. This has been demonstrated in the radiocontaminated Chernobyl (Shkvarnikov, 1990) and Fukushima (Mimura et al., 2014) environments, as well as their successful growth in space (Dubinin et al., 1973). Plant radio-resistance is maybe not surprising since radioactive materials occurred on the Earth's surface when plants

first appeared during the Mid-Ordovician period, about 460– 470 million years ago (Wellman and Gray, 2000; Karam and Leslie, 2005). It has also been proposed that present day areas with high-levels of background natural radiation and the reduced levels of plant migration may have both contributed to plant radio-resistance (Moller and Mousseau, 2013). To investigate the molecular aspects of this process in plants, various analyses have been undertaken (Moller and Mousseau, 2015). Recent meta-analysis of 45 published studies on DNA mutations in Chernobyl showed that plant growth in radiocontaminated environment is associated with increased levels of mutation (Moller and Mousseau, 2015). It appears that DNA methylation and increased extra chromosomal homologous recombination events also contribute to successful plant growth in radio-contaminated environments (Kovalchuk et al., 2003, 2004).

However, transcript expression and protein abundance are found to poorly correlate (Chen et al., 2002; Griffin et al., 2002; Orntoft et al., 2002; Pascal et al., 2008; Hornshoj et al., 2009; Hajduch et al., 2010), including plants growing in radio-contaminated areas. Therefore, the complementation of expression studies with proteomics can provide new insight into molecular mechanisms of plant growth in radio-contaminated environments. Indeed, proteome alterations induced by IR are the subject of increased research interest, especially in mammalian systems (Azimzadeh et al., 2014; Leszczynski, 2014). In plants, this is also appears to be the case, as differential abundances of proteins associated with defense and stress responses were detected in leaves harvested from rice grown in the soil taken around Chernobyl reactor site (Rakwal et al., 2009). Importantly, it has been demonstrated that proteome changes increase with irradiation dose; observations were based on the analysis of X-rays irradiated plantlets of the reference plant *Arabidopsis thaliana* (Gicquel et al., 2011).

# Experimental Design of the Chernobyl Seed Project

Experimental design for ecological field experiments should include several experimental fields to avoid pseudoreplication (Hurlbert, 1984). However, it is often difficult to establish and manage several experimental fields in heavily controlled radio-contaminated areas. Therefore, experimental design for Chernobyl (**Figure 1**) was modified and included (i) two non-radioactive fields (control) and one radio-contaminated experimental field (**Supplementary Figures S1A,B**), (ii) two plant species, and (iii) two successive years. An important aspect of this experimental design (**Figure 1**) is the changed location of the non-radioactive field after the first year. The logic behind this is to exclude alterations related to the differences between experimental fields (soil, pests, weather, etc).

In 2007, local varieties of soybean (*Glycine max* [L.] Merr., variety Soniachna) and flax (*Linum usitatissimum*, L., variety Kyivskyi) were sown in radio-contaminated experimental fields (soil radioactivity 20650 <sup>±</sup> 1050 Bq.kg−<sup>1</sup> of 137Cs, and 5180 ± 550 Bq.kg−<sup>1</sup> of 90Sr) located near the village Chistogalovka approximately 5 km from the Chernobyl Nuclear Power Plant (CNPP). The non-radioactive control experimental field (1350 <sup>±</sup> 75 Bq.kg−<sup>1</sup> for 137Cs and 490 <sup>±</sup> 60 Bq.kg−<sup>1</sup> for 90Sr) was established near Zhukin, a village approximately 100 km from CNPP (**Supplementary Figure S1A**). Soybean and flax seeds were harvested and mature seed proteomes comparatively analyzed in biological triplicate (**Figure 1**). In 2008, soybean and flax seeds harvested from the first generation of plants were sown onto the same radio-contaminated field, but a different non-radioactive field in the Chernobyl area (**Supplementary Figure S1B**), to obtain the second generation of seeds. A new non-radioactive experimental field was established directly in the town of Chernobyl, in an area with soil radioactivity of 1414 <sup>±</sup> 71 Bq.kg−<sup>1</sup> of 137Cs and 550 <sup>±</sup> 55 Bq.kg−<sup>1</sup> of 90Sr (**Supplementary Figure S1B**). The Chernobyl area is characterized by sod-podzolic soil (pH5.6–pH6.6, 12% clay, 2.0% organic compounds) which is a typical soil in the Ukrainian region of Polessia. Generally, in this area, the content of aleurite (silt) and pelitic soil ranges from 20 to 30% (Rashydov and Malinovskiy, 2002).

# Advances in the Establishment of Protein Abundance Profiles and Web-Based Database

In soybeans of the first generation, only 9.2% 2-DE spots, out of 698 quantified, were found differentially abundant between mature seeds harvested from non-radioactive and radiocontaminated Chernobyl areas (Danchenko et al., 2009). Similar to this, the analysis of the first generation of mature flax seeds showed differential abundance only in about 4.9% of resolved features from 720 quantified 2-DE spots (Klubicova et al., 2010). However, the results from these initial soybean and flax generations do not represent a large enough dataset upon which it is possible to base solid conclusions; it appears that growth in radio-contaminated environments has a relatively small effect on the seed proteome. Similar effects of IR have been previously shown on animal proteomes (Park et al., 2006; Guipaud et al., 2007) and support the notion that the exposure to low levels of IR do not significantly alter overall metabolism. Such speculation may be further supported by a study on the roots of the reference plant *Arabidopsis thaliana* under low levels of 137Cs, where only a small percentage of the transcriptome was differentially expressed (Sahr et al., 2005).

In order to provide a more detailed insight into the seed proteome in radio-contaminated environments, protein abundances profiles were established from developing soybean and flax seeds (**Figure 2**) from the second generation which were harvested from both Chernobyl experimental fields (**Supplementary Figure S1B**). Protein abundance profiles are capable of comprehensively characterizing protein abundances during seed filling. The approach has been used successfully in soybean (Hajduch et al., 2005), canola (Hajduch et al., 2006), castor (Houston et al., 2009), and *Arabidopsis* (Hajduch et al., 2010). In these Chernobyl studies, protein abundance

FIGURE 1 | Experimental design in the Chernobyl area during the two-year proteomic survey. In the first year, local varieties of soybean (*Glycine max* [L.] Merr., variety Soniachna) and flax (*Linum usitatissimum*, L., variety Kyivskyi) were planted in radio-contaminated (R) and non-radioactive (C1) experimental fields in the Chernobyl area (Supplementary Figure S1). Seeds were harvested in biological triplicate and subjected to proteomic

analyses. The following year, seeds not used for the analyses were planted into the same radio-contaminated field (R), but different non-radioactive (C2) experimental fields to obtain seeds from the second generation. To exclude alterations in seed proteomes related to field locations, only those differentially abundant proteins commonly observed across the two soybean and flax generations were considered.

developing soybean seeds were harvested at 4, 5, 6 weeks after flowering (WAF) (flax seeds at 2, 4, and 6 WAF) and at a mature stage from plants grown in non-radioactive and radio-contaminated Chernobyl experimental fields (Supplementary Figure S1). Isolated total protein was using ImageMaster 4.9 software. Finally, abundance profiles from both experimental fields were matched and joint abundance profiles, i.e., profiles for the same spot across seed filling in non-radioactive and radio-contaminated experimental fields, were established.

profiles were first established for each experimental field and then matched to obtain joint abundance profiles (**Figure 2**). Using this approach, it was possible to provide a detailed overview of protein abundances during seed filling in soybean (Klubicova et al., 2012a) and flax (Klubicova et al., 2010) across both experimental fields. For instance, it was revealed that β-conclycinin significantly decreased during seed filling in radio-contaminated areas in the second soybean generation (Klubicova et al., 2012a). These analyses also revealed alterations of proteins associated with carbon metabolism in the cytoplasm and plastids and to the carboxylic acid cycle in the mitochondria (Klubicova et al., 2012a). In flax, increased abundance of proteins associated with isocitrate dehydrogenation, L-malate decarboxylation, pyruvate biosynthesis, and ethanol oxidation to acetaldehyde were detected at early stages of seed filling (Klubicova et al., 2013).

The data from these experiments can be viewed in a userfriendly format at dedicated web-based database that is freely available at www*.*chernobylproteomics*.*sav*.*sk. The aim of this online data depository is to allow scientific community (but also general public) to access the data from this project in user-friendly format. At the time of the database establishment (Klubicova et al., 2012b) the database contained the data from first, second soybean and first flax generation. Since then, the data from second flax (Klubicova et al., 2013) and third (Gabrisova et al., in review) generations were uploaded.

# Chernobyl Seed Project Suggested the Identity of Proteins Putatively Associated with Plant Growth in Radio-Contaminated Environments

The aim of these studies was to detect alterations in seed proteomes related to the radio-contaminated environment. However, the alterations in seed proteomes described above might also be associated with the differences between the experimental fields (soil, pests, weather etc). To exclude this possibility, data were further analyzed and alterations common for both plant generations and plant species identified.

Altered abundance of enzymes associated with the glycine betaine biosynthetic pathway was jointly detected in the first generation of soybean (Danchenko et al., 2009) and flax (Klubicova et al., 2010). It is tempting to speculate that glycine betaine is involved at early stages of plant response toward the radio-contaminated environment. Interestingly, the involvement of glycine betaine in protection against IR was shown previously in human blood (Monobe et al., 2005). Since plants with altered levels of glycine betaine have already been produced (Waditee et al., 2007) it should be possible to directly test the putative protective role of glycine betaine in radio-contaminated environments.

The mobilization of seed storage proteins (SSP) and alteration of proteins associated with carbon assimilation and fatty acid metabolism were observed jointly in both generations of soybean and flax. These data support the notion that SSPs are involved in seed defense against various threats, as has been shown previously with their role in defense against Bruchids (Sales et al., 2000). Interestingly, the application of salicylic acid during germination *Arabidopsis thaliana* resulted in mobilization of SSPs (Rajjou et al., 2006). Furthermore, it has been proposed that class 2S albumin SSPs are defensive proteins (Regente and De La Canal, 2001), while salt stress has been shown to alter the abundance of β-conglycinin SSP (Aghaei et al., 2009).

An interesting aspect of these Chernobyl studies are differential abundance of proteins associated with carbon assimilation and fatty acid metabolism in both generations of soybean and flax. As a result of this, the second generation of soybean (Klubicova et al., 2012a) and flax (Klubicova et al., 2013) showed altered total oil content in mature seeds. However, additional studies are needed to determine whether altered seed oil content is the result of genetic mutation or has an epigenetic or posttranslational explanation.

# Studies in Fukushima Radio-Contaminated Environment

Similar to the disaster at the CNPP in 1986, the accident at the Fukushima Daiichi Nuclear Power Plant in 2011 contaminated large areas with radioactivity (Buesseler et al., 2011; Kinoshita et al., 2011; Yasunari et al., 2011). Unfortunately, nuclear accidents provide unexpected justifications for research aimed at understanding plant survival and adaptation in radiocontaminated environments. Indeed, Hayashi at al. (2014) performed a pioneering study in the Fukushima radiocontaminated areas through the investigation of rice seedlings under continuous low-dose radiation. This study provided an overview of the transcriptome response in rice toward low level of gamma radiation and identified large numbers of genes with altered expression patterns (Hayashi et al., 2014).

It will be interesting to compare results from the Chernobyl studies using similar experimental setups in the Fukushima radio-contaminated area. The web based database (chernobylproteomics.sav.sk) might be a good tool for quick data comparison. Ideally, follow-up studies in Fukushima should include several non-radioactive and radio-contaminated experimental fields to avoid pseudoreplication (Hurlbert, 1984). If this is not feasible due to restricted/closed areas, an experimental design for radio-contaminated areas presented in this current study (**Figure 1**) could be applied.

# Conclusion

The outcome of these Chernobyl studies was the identification of several proteins with differentially abundances in soybean and flax seeds harvested during two successive generations. It is tempting to speculate that these proteins are associated with plant growth and adaptation in radio-contaminated environments. However, follow-up studies in both the Chernobyl and Fukushima radio-contaminated areas are required to further develop these hypotheses.

# Acknowledgments

This research was supported by the Slovak Research and Development Agency (APVV-0740-11), by European Community under Project no 26220220180: Building Research Centre "AgroBioTech," and Seventh Framework Program of the European Union (IRSES-GA-2013-612587). This research was also partially supported by the Scientific Grant Agency of the Ministry of the Slovak Republic and Slovak Academy of Sciences (VEGA-2/0016/14).

# References


# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal*.*frontiersin*.*org/article/10*.*3389/fpls*.*2015*.*00493

FIGURE S1 | The location of the Chernobyl experimental fields. The radio-contaminated field (R) was established 5 km from Chernobyl Nuclear Power Plant (C), within the exclusion zone (D). The non-radioactive field (C1) was initially established in 2007 about 100 km from C (A) but since 2008 the non-radioactive field (C2) has been transferred directly to the town of Chernobyl (B).

proteome database*. Plant Physiol.* 137, 1397–1419. doi: 10.1104/pp.104. 056614


rice as a Grass Model. *Int. J. Mol. Sci.* 10, 1215–1225. doi: 10.3390/ijms100 31215


**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.

*Copyright © 2015 Rashydov and Hajduch. 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.*

# Understanding the plant-pathogen interactions in the context of proteomics-generated apoplastic proteins inventory

Ravi Gupta<sup>1</sup> , So Eui Lee<sup>1</sup> , Ganesh K. Agrawal 2, 3, Randeep Rakwal 2, 3, 4, 5 , Sangryeol Park <sup>6</sup> , Yiming Wang<sup>7</sup> \* and Sun T. Kim<sup>1</sup> \*

<sup>1</sup> Department of Plant Bioscience, Life and Industry Convergence Research Institute, Pusan National University, Miryang, South Korea, <sup>2</sup> Research Laboratory for Biotechnology and Biochemistry, Kathmandu, Nepal, <sup>3</sup> Global Research Arch for Developing Education (GRADE), Academy Private Limited, Birgunj, Nepal, <sup>4</sup> Organization for Educational Initiatives, University of Tsukuba, Tsukuba, Japan, <sup>5</sup> Faculty of Health and Sport Sciences, Tsukuba International Academy for Sport Studies, University of Tsukuba, Tsukuba, Japan, <sup>6</sup> Bio-crop Development Division, National Academy of Agricultural Science, Rural

Breeding Research, Cologne, Germany

### Edited by:

Jesus V. Jorrin Novo, University of Cordoba, Spain

### Reviewed by:

Jörg Schumacher, Imperial College London, UK Raquel Gonzalez-Fernandez, Autonomous University of Ciudad Juárez, Mexico

### \*Correspondence:

Sun Tae Kim, Department of Plant Bioscience, Pusan National University, Miryang 627-706, South Korea stkim71@pusan.ac.kr; Yiming Wang, Department of Plant Microbe Interactions, Max Planck Institute for Plant Breeding Research, Carl-von-Linne weg 10, Cologne 50829, Germany ywang@mpipz.mpg.de

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 31 December 2014 Accepted: 03 May 2015 Published: 02 June 2015

### Citation:

Gupta R, Lee SE, Agrawal GK, Rakwal R, Park S, Wang Y and Kim ST (2015) Understanding the plant-pathogen interactions in the context of proteomics-generated apoplastic proteins inventory. Front. Plant Sci. 6:352. doi: 10.3389/fpls.2015.00352 The extracellular space between cell wall and plasma membrane acts as the first battle field between plants and pathogens. Bacteria, fungi, and oomycetes that colonize the living plant tissues are encased in this narrow region in the initial step of infection. Therefore, the apoplastic region is believed to be an interface which mediates the first crosstalk between host and pathogen. The secreted proteins and other metabolites, derived from both host and pathogen, interact in this apoplastic region and govern the final relationship between them. Hence, investigation of protein secretion and apoplastic interaction could provide a better understanding of plant-microbe interaction. Here, we are briefly discussing the methods available for the isolation and normalization of the apoplastic proteins, as well as the current state of secretome studies focused on the in-planta interaction between the host and the pathogen.

Development Administration, Jeonju, South Korea, <sup>7</sup> Department of Plant Microbe Interactions, Max Planck Institute for Plant

Keywords: apoplast, apoplastic proteins, pattern-triggered immunity, effector-triggered immunity, secretome, protein secretion, plant-pathogen interaction

# Introduction

Plant-pathogen interaction is a multifaceted process, mediated by the pathogen- and plantderived molecules which mainly include proteins, sugars and lipopolysaccharides (Boyd et al., 2013). Secreted molecules, derived from the pathogens, are the key factors which determine their pathogenicity and allow their successful colonization inside the host. On the other hand, plant derived molecules are involved in the recognition of these pathogens in order to elicit the defense response. The first interaction between the plants and microbes take place in apoplast and is mediated by the recognition of microbial elicitors by the receptor proteins of the plants (Dodds and Rathjen, 2010). These microbial elicitors, also known as pathogen-associated molecular patterns (PAMPs), are recognized by the membrane-localized pattern recognition receptors (PRRs) of plants (Boyd et al., 2013; Zipfel, 2014). The bacterial flagellin and elongation factor (EF)-Tu peptide surrogates, flg22 and elf18, and chitin, are common examples of PAMPs, which are recognized by the plant PRRs that include the three receptor-like kinases, flagellin-sensitive22 (FLS2), EF-Tu receptor (EFR), and chitin elicitor receptor kinase1 (CERK1) (Liu et al., 2013). The successful recognition of microbial derived PAMPs by PRRs of the plants activates a first line of defense which is known as PAMPtriggered immunity (PTI). To counter-attack the PTI, many pathogens deliver various "effector" proteins inside the host cell, which suppress the components of PTI. These pathogen derived "effector" proteins include various avirulence (Avr) proteins like Slp1 of Magnoporthe oryzae and TALEs of Xanthomonas oryzae (Boyd et al., 2013; Liu et al., 2014a). However, resistance (R) proteins of plants recognize these effector proteins of pathogens and can induce a second line of defense which is known as the effector-triggered immunity (ETI) (Jones and Dangl, 2006). ETI is quantitatively stronger and faster than PTI and can result in a localized cell death (hypersensitive response) to kill both pathogen and pathogen infected plant cells. PTI and ETI together constitute a major innate immune response, enabling plants to recognize and battle against the pathogen attack. However, the components of ETI and PTI in response to interaction with different pathogens remain largely unknown, requiring a large-scale investigation of proteins for better understanding of the plant-pathogen interactions, which would be important to generate the stress tolerant crops. As the first interaction between plant and pathogens occur in apoplast, analyzing the dynamic changes of apoplastic proteins through proteomics approach is necessary for a deep understanding of the components of signal perception and signal transduction during pathogen attack.

The past few years have seen remarkable efforts in solving the mystery of plant-pathogen interaction in the apoplast (reviewed in Krause et al., 2013; Delaunois et al., 2014; Tanveer et al., 2014). For the analysis of secreted proteins in response to pathogen attack, mostly in-vitro interaction systems using suspensioncultured cells were used, due to relatively easy isolation of secreted proteins from them (reviewed in, Agrawal et al., 2010). However, recent comparative studies strongly suggest that the components of the in-vitro and the in-planta secretome can be relatively different, sharing sometimes less than 3% of common proteins. A comparison of the in-vitro and in-planta secreted proteins showed only 6 common proteins out of the total 222 identified proteins in rice (Jung et al., 2008). Moreover, the in-vitro secretome analysis may not illustrate the real state of host-pathogen interaction, thereby necessitating extraction of apoplastic proteins from the in-planta systems. In this minireview, we have summarized the progress made so far in this area to present the current scenario of secretomics during the plant-pathogen interaction.

# Methods to Isolate Apoplastic Proteins

Due to the biochemical and technical advances, it is possible to isolate the proteins directly from the apoplast which can be analyzed by gel-based or gel-free proteomics approaches. However, relatively limited number of studies have been conducted so far, to identify the pathogen-secreted proteins inplanta (**Table 1**). The successful isolation of apoplastic proteins is the most critical step prior to utilizing the samples for proteome analysis. For the isolation of apoplastic proteins, a number of methods including vacuum infiltration (VIC) and gravity extraction methods are available (reviewed in Agrawal et al., 2010). However, only VIC method along with its modified version (termed CA-VIC), has been used for the isolation of apoplastic proteins in response to pathogen infection (Floerl et al., 2008) (**Figure 1**). In the VIC method, leaves are cut into small sections followed by extensive washings of these sections to remove cytoplasmic proteins from the cut ends. The washed leaves sections are then incubated in the extraction buffer which is allowed to infiltrate into the cells through a pressure change induced by vacuum. Finally, apoplastic proteins are recovered by centrifugation at low speed. This method was used to isolate the apoplastic proteins from the leaves of Arabidopsis and tobacco (De-la-Pena et al., 2008; Delannoy et al., 2008). However, this VIC method is less efficient in isolating the apoplastic proteins from the waxy coated leaves, like leaves of rice and maize. Moreover, previous studies in which apoplastic proteins were extracted from Arabidopsis and Brassica leaves by VIC method, showed identification of only few differential proteins in response to Verticillium longisporum infection, indicating the limitation of this method for comparative proteome analysis (Floerl et al., 2008: Floerl et al., 2012; Shenton et al., 2012). Furthermore, this VIC method yields much lower amount of apoplastic proteins which is a major constrain for large scale proteome analysis. Keeping these limitations in mind, the VIC method was modified (CA-VIC) to isolate the apoplastic proteins from rice leaves with increased amount (**Figure 1**). This method involves shaking of the cut segments of the leaves in a calcium based buffer for 1 h on ice, followed by vacuum infiltration, centrifugation, and phenol precipitation (Kim et al., 2013). This method yields higher amount of apoplastic proteins than classical VIC method, may be due to the addition of calcium, which facilitates the extraction of loosely bound cell wall proteins (Watson et al., 2004). A comparative analysis was carried out to select the best buffer for isolation of apoplastic proteins. Among all the extraction buffers tested, sodium phosphate or ascorbic acid with calcium chloride were the most efficient, while extraction with water or Tris showed contamination from vacuole and other organelles (Witzel et al., 2011; Gupta and Deswal, 2012). Therefore, the selection of an appropriate extraction method is crucial for apoplastic protein extraction in different plant species.

# Validation and Normalization of Apoplastic Proteins

After isolation of apoplastic proteins, the next step is to assess the purity of isolated proteins as contamination from cytoplasm and other organelles can lead to the false positive results. To assess the purity of extracted apoplastic proteins, several methods including cytoplasmic marker enzymes activity assays and Western blot analysis of marker proteins, can be performed (**Figure 1**). Glucose-6-phosphate dehydrogenase (G6PDH), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and malate dehydrogenase (MDH) are some of the common cytoplasmic enzymes which are being widely used as biomarkers for cytoplasmic contamination, while RuBisCO antibodies are used for assessing the chloroplastic contamination in the apoplastic proteins. In addition to assessing the contamination,


Western blots with antibodies against apoplastic markers like β-1,3-glucanase (Jung et al., 2008), Duf26 (or OsRMC) (Zhang et al., 2009), AtPR-1 (Wang et al., 2005), thaumatin-like protein and glucanase-2 (Kim et al., 2013), can also be used for sample validation to ensure the successful enrichment of apoplastic proteins. Moreover, this approach can also be employed to measure the contamination ratio with other cellular organelle markers. Previous reports have shown that both the vacuum infiltration and calcium buffer based apoplastic protein extraction methods show none/low ratio of contaminates (Agrawal et al., 2010; Gupta and Deswal, 2012; Kim et al., 2013). As an example, apoplastic proteins, isolated by CA-VIC method showed negligible cytoplasmic contamination as observed by low G6PDH activity. In addition, Western blots of intracellular proteins, OsPR-10 and PBZ1, did not detect any signal, suggesting the low levels of cytoplasmic contamination in isolated apoplastic proteins. Furthermore, the enrichment of apoplastic proteins was also shown by assessing the expression of apoplastic marker proteins (glucanse-2 and thaumatin-like protein) using Western blotting, indicating the efficacy of calcium based buffer in isolation of apoplastic proteins (Kim et al., 2013). Taken together, assessing the enrichment of apoplastic proteins and contaminations are the essential steps in the analysis of secretome.

To examine the differences in global protein secretion upon pathogen infection, sufficient normalization of protein samples is necessary. Based on previous studies, two possible normalization methods were applied to evaluate the protein abundances (**Figure 1**). In the first method, protein abundance is normalized with same amount of isolated proteins (Kaffarnik et al., 2009). As it is possible that the rate of protein secretion would also be affected in addition to changes in which proteins were secreted, normalization with protein concentration would provide absolute changes in protein identities (or proteins that changed dramatically in concentrations). However, as infection of pathogen could enhance the protein secretion in plants (Watanabe et al., 2013), and pathogen effectors could block the secretion of protein from plants (Lee et al., 2012), the disadvantage of normalization with protein amount is that the real protein secretion changes might be concealed. Another choice for sample normalization is on the fresh tissue amount. A significant increase of protein secretion was detected comparing with non-infected tissues (Kim et al., 2013). These results indicated that upon pathogen infection the overall protein secretion might be enhanced. Moreover, the selection of extraction buffer and protein loss during extraction procedures may strongly affect the final proteomics results. Therefore, normalization with fresh tissue amount may illustrate the real case of protein secretion upon pathogen infection. However, as it is difficult to distinguish the cytoplasmic and apoplastic proteins when normalized with the fresh tissue amount, use of cytoplasmic and apoplastic marker is highly recommended in order to check the cytoplasmic contamination. Taken together, the issue of how the amount of apoplastic proteins should be normalized has to be considered prior to the proteomics analysis.

# Proteomics Investigations of In-planta Secreted Proteins During Plant-pathogens Interactions

M. oryzae (a hemi-biotrophic fungus) causes rice blast disease which results in huge loss of productivity of this most important cereal grain worldwide (Talbot, 2003). A gel-based proteomics approach was used to identify the rice-M. oryzae interaction in the apoplast which led to the identification of several rice secreted proteins including three DUF26 domain containing cysteine rich repeat proteins and PR-proteins. In addition, a M. oryzae secreted virulence factor protein, cyclophilin CYP1, was also identified in rice apoplast (Shenton et al., 2012). In another similar study, 732 secreted proteins were identified, of which 40 and 60% were from rice and M. oryzae, respectively (Kim et al., 2013). Furthermore, a higher level of up-regulation of glycosylhydrolase and chitinase proteins was observed in case of incompatible interactions as compared to the compatible, suggesting the involvement of these proteins in the resistance against rice blast fungus. These studies indicated that the pathogenic fungus also secretes numerous proteins into the apoplastic space. Plant secreted proteins were mainly glycosyl hydrolase family proteins, esterases, proteases and peptidases, suggesting that the cell wall and protein modifications are important aspects for the resistance against M. oryzae infection (Kim et al., 2013) (**Figure 2**). Identification of apoplastic proteins during rice-Cochliobolus miyabeanus (a necrotrophic fungus) infection led to the identification of 501 proteins, of which only 31 (6.2%) were secreted from C. miyabeanus, whereas 470 were secreted from rice. These results suggest that the host-secreted proteins are more abundant during the C. miyabeanus infection. Proteins with decreased abundance were mainly related to the Calvin cycle and glycolysis, whereas abundance of the proteins involved in the TCA cycle, amino acids, and ethylene biosynthesis was increased (Kim et al., 2014). V. longisporum is one of the most devastating diseases of the Brassicaceae members. Analysis of plant-V. longisporum interactions, using Arabidopsis and Brassica as hosts, showed up-regulation of various PRproteins including endochitinase, peroxidases, β-1,3-glucanase, PR-4, PRX52, PRX34, P37, serine carboxypeptidase SCPL20, α-galactosidase AGAL2, and germin-like protein (Floerl et al., 2008, 2012). A lectin-like, chitin-inducible protein was downregulated upon V. longisporum infection (Floerl et al., 2012). Surprisingly, no fungal protein was identified in these studies, suggesting that either the amount of fungal secreted proteins are much lower (beyond the detection limits of the MS) or V. longisporum majorly secretes other metabolites like sugars and lipopolysaccharides to interact with its hosts.

In addition to the fungal infections, apoplastic proteins were also analyzed during rice-bacterial infections. X. oryzae is a Gram-negative bacterium that causes the rice bacterial blight disease. Analysis of rice-X. oryzae interaction led to the identification of 109 proteins of which only six were secreted from rice, indicating that the percentage of bacterialsecreted proteins is much higher than its host rice (Wang et al., 2013). It was also shown that the highly conserved proteins secreted from X. oryzae in-vitro and in-planta, were related with the metabolic and nutrient uptake activities. The pathogenicity-related proteins were highly enriched inplanta, but not detected in-vitro, further implying differences in secretory proteins between in-vitro and in-planta systems. Overall, these findings suggest that the nutrient uptake from surrounding environments and sustaining basic metabolism is the primary task of fungus, while modification of host immunity is essential for in-planta survival and spread of the bacterium. Taken together, it can be concluded that the ratio of secretory proteins from host and pathogens (hemibiotrophic, necrotrophic fungus, and Gram-negative bacteria) differs significantly, and likely their functions. However, it is worth mentioning here that the ratio of pathogen secreted proteins also varies in accordance with the compatible or incompatible interaction with the host. During the incompatible interactions, the growth of pathogens is very much reduced with much less secretion of protein while during compatible interactions the growth of pathogen is highly pronounced which

results in secretion of more proteins (Gonzalez-Fernandez and Jorrin-Novo, 2012).

# Leaderless Secretory Proteins

Apoplastic proteins were long thought to be secreted only through the "Golgi-endoplasmic reticulum pathway" due to the presence of an N-terminal signal peptide. However, growing body of evidences suggest that the apoplast also harbors proteins which lack the signal peptide and therefore, these proteins are supposed to be secreted via non-classical protein secretion pathways (leaderless secretory pathways) (Agrawal et al., 2010; Kim et al., 2013; Wang et al., 2013). These proteins are well-known as the leaderless secretory proteins (LSPs) and constitute up to 80% of the total apoplastic proteins, depending upon the tissue and stress conditions (Agrawal et al., 2010). Metabolism-related proteins, which are mainly cytosolic, were commonly identified in the apoplastic region, suggesting that these proteins are pumped out through unknown mechanisms, and might be essential for plant immune responses. It is well documented that fungus secretes mannitol in the plant apoplastic space during infection to quench the reactive oxygen species which could otherwise elicit the plant defense response (Cheng and Williamson, 2010). On counter-attack, plants secret mannitol dehydrogenase (MTD) in the apoplast where it catabolizes the mannitol secreted by the fungus. However, MTD does not contain a signal peptide and therefore would be secreted by the non-classical pathways. Similarly, superoxide dismutase, which also lacks a signal peptide, has been confirmed as a resident of the apoplast (Cheng and Williamson, 2010; Gupta and Deswal, 2012). Many more such proteins have been identified and their inventory is increasing as we investigate more and further—thanks to the proteomics technologies.

In addition to these well characterized LSPs, several other cytoplasmic or non-secretory proteins are also observed in the apoplast during pathogen attack. These other cytoplasmic proteins might be the LSPs which are yet to be characterized. Notwithstanding, it can also be speculated that these proteins are just the contaminants which are released to the apoplast due to cell lysis. It is well known that both plant and pathogen secrete cell wall degrading enzymes which result in lyses of cell wall of opposite partner, thus resulting in the leakage of cytoplasmic proteins in the apoplast. However, there are few reports which neglect this possibility. During fungal infections, plants form extra-haustorial membrane (EHM) and extra-invasive hyphal membrane (EIHM), to separate the plant cytoplasm from the haustorium of oomycetes and invasive hypha of filamentous fungus respectively (Yi and Valent, 2013). These additional membranes maintain the integrity of the host cell and prevent its rapid lysis during pathogen infection, even though the cell wall is perforated. Due to the presence of these membranes, cytoplasmic proteins cannot be simply leaked out from the cytoplasm and therefore their accumulation in the apoplast must be mediated by some protein secretory pathway(s). Moreover, during the cell lysis, leakage of whole set of abundant cytoplasmic proteins is expected in the apoplast however, only limited set of cytoplasmic proteins are detected in the apoplast which further suggest controlled or regulated secretion of these proteins during pathogen attack (Kaffarnik et al., 2009).

# Prospects and Further Application

All the studies conducted so far to investigate the plant-microbe interactions have utilized a gel-based proteomics approach for the identification of in-planta secreted proteins. However, as gel based approaches have several limitations including identification of low-abundance proteins, it is possible that some of the key proteins of both plant as well as microbe, would be missed during the analysis. As an example, analysis of secreted proteins from V. longisporum-Arabidopsis or Brassica did not identify any fungal protein, may be due their low-abundance.

With the advancement in the proteomics technologies, new and more precise gel-free quantitative proteomics approaches with improved sensitivity are being developed which could be applied for protein identification, dynamic regulation, and analyzing the post-translational modifications (Picotti et al., 2009). Utilization of newly developed proteome tools like multiple reaction monitoring MS (MRM-MS), will benefit for the characterization and quantification of protein profile during plant-microbe interactions (Schumacher et al., 2014). Moreover, taking the advantage of genetic modified materials, such as plant or bacterial lacking secretion systems, much more novel proteins could be identified which can illustrate a deeper understanding of apoplastic interaction between host and pathogen (Hemsley et al., 2013; Schumacher et al., 2014). However, it still will be a long way to utilize those proteins for crop modification and further field applications. Fortunately, researchers have started looking inside the role of secreted proteins in plant-microbe interactions even only few publications were released. For instance, secretion of lysozyme-like hydrolase exhibits infections of bacterial pathogen

# References


by increasing the release of peptidoglycans from bacterial cell wall and triggering the PTI in Arabidopsis (Liu et al., 2014b). Those functional investigations of secreted proteins may provide better understanding of their role in plant-microbe interactions which would be helpful in the development of effective crop protection strategies (Gonzalez-Fernandez and Jorrin-Novo, 2012).

# Conclusions

Upon infection, both plants as well as pathogens secrete molecules including proteins which determine the fate of their interaction. While pathogen secreted proteins are involved in infection and pathogenicity, plant secreted proteins play crucial roles in its resistance. Therefore, identification and functional analyses of apoplastic proteins will open new horizons in our understanding of plant-pathogen interactions. Moreover, application of the in-planta protein extraction techniques at multiple times post-infection will reveal the "real" composition and dynamic changes of the apoplast and can result in the identification of more components of PTI. Furthermore, unraveling complete proteome of apoplastic region in multiple plants, pathogens, and their interactions would be highly fruitful for understanding the biology of plant-pathogen interactions, and that will help in designing new strategies for generating the next-generation crops resistance to multiple pathogens and environmental stresses.

# Acknowledgments

This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science, and Technology (2013R1A1A1A05005407) and Next-Generation BioGreen 21 Program (Plant Molecular Breeding Center, PJ011038), Rural Development Administration, Republic of Korea. YW was supported by Max Planck Institute for Plant Breeding Research, and Alexander-Bayer Fellowship from Alexander von Humboldt Foundation, and Bayer Science and Education Foundation.


proteome of the rice–Magnaporthe interaction. J. Plant Res. 125, 311–316. doi: 10.1007/s10265-012-0473-y


**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.

Copyright © 2015 Gupta, Lee, Agrawal, Rakwal, Park, Wang and Kim. 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.

# Proteomic analysis of apoplastic fluid of *Coffea arabica* leaves highlights novel biomarkers for resistance against *Hemileia vastatrix*

Leonor Guerra-Guimarães 1, 2 \*, Rita Tenente<sup>1</sup> , Carla Pinheiro3, 4, Inês Chaves 3, 5 , Maria do Céu Silva1, 2, Fernando M. H. Cardoso<sup>6</sup> , Sébastien Planchon<sup>7</sup> , Danielle R. Barros 1, 8, Jenny Renaut <sup>7</sup> and Cândido P. Ricardo<sup>3</sup>

<sup>1</sup> Centro de Investigação das Ferrugens do Cafeeiro, Instituto de Investigação Científica Tropical, Oeiras, Portugal, <sup>2</sup> Linking Landscape, Environment, Agriculture and Food, Instituto Superior de Agronomia, Universidade de Lisboa, Lisboa, Portugal, 3 Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa (UNL), Oeiras, Portugal, <sup>4</sup> Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Caparica, Portugal, <sup>5</sup> Instituto de Biologia Experimental e Tecnológica, Oeiras, Portugal, <sup>6</sup> Global Health and Tropical Medicine, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisboa, Portugal, <sup>7</sup> Luxembourg Institute of Science and Technology, Belvaux, Luxembourg, <sup>8</sup> Department de Fitossanidade, Universidade Federal de Pelotas, Pelotas, Brasil

A proteomic analysis of the apoplastic fluid (APF) of coffee leaves was conducted to investigate the cellular processes associated with incompatible (resistant) and compatible (susceptible) Coffea arabica-Hemileia vastatrix interactions, during the 24–96 hai period. The APF proteins were extracted by leaf vacuum infiltration and protein profiles were obtained by 2-DE. The comparative analysis of the gels revealed 210 polypeptide spots whose volume changed in abundance between samples (control, resistant and susceptible) during the 24–96 hai period. The proteins identified were involved mainly in protein degradation, cell wall metabolism and stress/defense responses, most of them being hydrolases (around 70%), particularly sugar hydrolases and peptidases/proteases. The changes in the APF proteome along the infection process revealed two distinct phases of defense responses, an initial/basal one (24–48 hai) and a late/specific one (72–96 hai). Compared to susceptibility, resistance was associated with a higher number of proteins, which was more evident in the late/specific phase. Proteins involved in the resistance response were mainly, glycohydrolases of the cell wall, serine proteases and pathogen related-like proteins (PR-proteins), suggesting that some of these proteins could be putative candidates for resistant markers of coffee to H. vastatrix. Antibodies were produced against chitinase, pectin methylesterase, serine carboxypeptidase, reticuline oxidase and subtilase and by an immunodetection assay it was observed an increase of these proteins in the resistant sample. With this methodology we have identified proteins that are candidate markers of resistance and that will be useful in coffee breeding programs to assist in the selection of cultivars with resistance to H. vastatrix.

Keywords: coffee leaf rust (CLR), cytology, MALDI-TOF/TOF MS, 2-DE, antibody production, ELISA assay

### *Edited by:*

Silvia Mazzuca, Università della Calabria, Italy

### *Reviewed by:*

Martin Hajduch, Slovak Academy of Sciences, Slovakia Letizia Bernardo, Università Cattolica del Sacro Cuore, Italy

### *\*Correspondence:*

Leonor Guerra-Guimarães, Centro de Investigação das Ferrugens do Cafeeiro/BioTrop, Instituto de Investigação Científica Tropical, Quinta do Marquês, 2784-505 Oeiras, Portugal leonorguima@gmail.com

### *Specialty section:*

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

*Received:* 16 April 2015 *Accepted:* 15 June 2015 *Published:* 30 June 2015

### *Citation:*

Guerra-Guimarães L, Tenente R, Pinheiro C, Chaves I, Silva MC, Cardoso FMH, Planchon S, Barros DR, Renaut J and Ricardo CP (2015) Proteomic analysis of apoplastic fluid of Coffea arabica leaves highlights novel biomarkers for resistance against Hemileia vastatrix. Front. Plant Sci. 6:478. doi: 10.3389/fpls.2015.00478

# Introduction

Coffee leaf rust (CLR), caused by the fungus Hemileia vastatrix Berkeley and Broome, is the most important disease of Coffea arabica L. Since the first reported outbreak of CLR in 1867, that caused the eradication of coffee cultivation in Sri-Lanka, the disease has spread to all the coffee growing regions (Bettencourt and Rodrigues, 1988; Várzea and Marques, 2005). The current highly intense epidemic of CLR in Colombia and Central America has considerably affected coffee production with yield losses estimated as several 100 million dollars (Avelino et al., 2015). Although application of fungicides can provide adequate control, the use of coffee resistant varieties has been the most appropriate and sustainable strategy against this disease (Várzea and Marques, 2005).

H. vastatrix, like other rust fungi, is a biotrophic fungus entirely dependent on plant living cells for growth and reproduction. Rust fungi interact intimately with the plant host cells (by means of haustoria, highly specialized intracellular hyphae) modifying plant metabolism to serve the fungus nutrient needs for completion of its life cycle. This mode of interaction involves a prolonged and effective suppression of the host immune system and, at the same time, the induction of specific host genes for establishing biotrophy (Schulze-Lefert and Panstruga, 2003; Voegele and Mendgen, 2003). H. vastatrix starts to colonize the plant surface and after developing appressoria penetrates the host tissues through stomata, growing initially in the intercellular space before the formation of the first haustoria inside the subsidiary stomatal cells (Silva et al., 1999). The apoplast (the extracellular space that comprises cell walls and the intercellular fluid) is a metabolically very active cellular compartment, since it serves transport, environmental sensing and defense, as well as the construction and maintenance of cell walls. It is in the apoplast where the pathogen and plant first contact, and the primary defenses are activated (Agrawal et al., 2010; Floerl et al., 2012; Delanois et al., 2014; Guerra-Guimarães et al., 2014).

Plants respond to pathogen infection using a multilayer immune system, consisting of both constitutive and inducible mechanisms. The plant's ability to discriminate between its own molecules and those of the other organisms represents the first essential line of defense of any immune system (Doehlemann and Hemetsberger, 2013). The eliciting pathogen molecules (pathogen-associated molecular patterns - PAMPs) trigger in plants the first level of induced defenses or PAMP-trigger immunity (PTI). Successful pathogens deliver effectors that interfere with PTI, enabling pathogen nutrition and dispersal, and resulting in effector–triggered susceptibility (ETS). As a second defense layer, plants use resistance (R) genes to activate effector-triggered immunity (ETI) upon detection of effectors. ETI is associated with more sustained and robust immune responses including cell death by hypersensitive reaction (HR) (Jones and Dangl, 2006; Doehlemann and Hemetsberger, 2013; Delanois et al., 2014).

Coffee—H. vastatrix rust interactions are governed by the gene-for-gene relationship (Flor, 1942). The resistance of coffee plant is conditioned by nine major dominant genes (SH1– SH9) that have the corresponding virulence genes (v1–v9) in the pathogen (Rodrigues et al., 1975; Bettencourt and Rodrigues, 1988; Várzea and Marques, 2005). There is no evidence of constitutive defenses in coffee against H. vastatrix, but several resistance mechanisms are induced upon fungus infection (Silva et al., 2006 and references therein). Previous cytological studies have shown that for a number of coffee genotypes, the first signs of incompatibility (resistance) to H. vastatrix correspond to HR (Rijo et al., 1991; Silva et al., 2002, 2008). During the last decade, information on the molecular processes of the coffee-CLR interactions have been gathered using different approaches (e.g., suppression subtractive hybridization method, 454pyrosequencing and qRT-PCR) what allowed the identification of several genes putatively involved in host resistance (Fernandez et al., 2004, 2012; Ganesh et al., 2006; Diniz et al., 2012). It was thus found that more than one-quarter of the predicted proteins of the expressed sequence tags (ESTs) are disease resistance proteins, stress- and defenseproteins and components of signal transduction pathways (e.g., chitinases, beta-1,3-glucanases, PR10, lipoxygenase, AP2-type, WRKY transcription factors). Activity of oxidative enzymes (lipoxygenase, peroxidase, superoxide dismutase, and germinlike protein), phenylalanine ammonia-lyase, chitinases, and glucanases were detected in the resistance reaction. In the susceptible reaction some of these enzymes are also expressed but later (or slower) in the infection process and, therefore, are ineffective to arrest the pathogen (Maxemiuc-Naccache et al., 1992; Rojas et al., 1993; Silva et al., 2002, 2008; Guerra-Guimarães et al., 2009a,b, 2013).

Proteomics is a valuable analysis when aiming for an overview of the biochemical pathways involved in the defense response. In fact, it is an untargeted approach that provides insight into protein localization, protein-protein interactions, enzymatic complexes, or post-translational modifications (PTMs) that are essential for a better understanding of plant-pathogen interactions (Abril et al., 2011; Delanois et al., 2014; Pinheiro et al., 2014; Jorrín-Novo et al., 2015).

Based on a cytological characterized study, we conducted a 2-DE proteomic analysis of incompatible (resistant) and compatible (susceptible) C. arabica-H. vastatrix interactions with the main objectives of: further our understanding of proteins that are present in the leaf apoplast, investigate the dynamic nature of the proteins in relation to coffee-fungus interactions and link these proteins with the resistant/susceptible response pathways on the basis of their physiological role. Proteins identified by MS and that were associated with the pathogen resistance response, namely, glycohydrolases, proteases, and PR-proteins, were chosen for antibody (Ab) production. To validate these proteins as potential biomarkers of resistance, the Abs were used in an immunodetection assay. With this methodology we have identified proteins that are potential candidate markers of resistance that will be useful to assist in the selection of coffee cultivars with resistance to H. vastatrix.

# Materials and Methods

# Biological Material

Five-year-old Coffea arabica S4 Agaro, genotype SH4SH5, that resulted from clonally propagated stem cuttings, were grown in 50 L pots in a mixture of soil:peat:sand (1:1:1) under greenhouse conditions as previously stated (Guerra-Guimarães et al., 2014). Two races of the fungus Hemileia vastatrix were used in this study, one that establish a compatible interaction characterized by fungus growth and plant disease (susceptible reaction) and another one that establish an incompatible interaction characterized by a resistance response of the plant that leads to fungus death (Várzea and Marques, 2005). So, when race XV (v4,5) was inoculated the plant showed disease symptoms indicating it was susceptible to this fungal race and it is said that a compatible plant-fungus interaction was established. Inoculation with H. vastatrix race II (v5) showed resistant symptoms to this fungal race and it is said that an incompatible plantfungus interaction occurred (Várzea and Marques, 2005). Fresh uredospores of H. vastatrix (1 mg/pair of leaves) were spread over the lower surface of young coffee leaves, as previously described (Silva et al., 2002). Healthy leaves sprayed with water and kept in the same conditions as inoculated leaves were used as control. For each coffee—rust interaction, inoculations were performed during September/October on at least three separate occasions, using different batches of spores. Leaves were collected 24, 48, 72, and 96 h after inoculation (hai) for experimental purpose.

## Light Microscopy

Cross sections of infected leaf fragments made with a freezing microtome (Leica CM1850) were stained and mounted in cotton blue lactophenol to evaluate fungal growth stages (Silva et al., 2002). To detect autofluorescent cells, cross sections of infected leaf fragments were placed in 0.07 M pH 8.9 phosphate solution (K2HPO4) for 5 min, and mounted in the same solution (Silva et al., 2002). Autofluorescence and/or browning of cell contents, under blue light epifluorescence are thought to indicate plant cell death (Heath, 1998). Autofluorescence can also be used as an indicator of fungal death (Heath, 1984). Observations were made with a Leica DM-2500 microscope equipped with a mercury bulb HB 100W, blue light (excitation filter BP 450–490; barrier filter LP 515). Data were recorded from 75 to 100 infection sites/coffee-rust interaction/observation time/experiment. Since no significant differences were observed between different sets of experiments, data for each coffee-rust interaction were pooled. Arcsine-transformed percentages and Student t-test for statistical analysis were used.

## Plant Protein Extraction

The apoplastic fluid (APF) of the leaves was obtained from samples that represent a pool of 8 pairs of leaves (10 ± 2.5 g fresh weight) from 3 to 4 different plants. The leaves were vacuum infiltrated as previously described (Guerra-Guimarães et al., 2009b). Briefly, square sections of about 2 cm<sup>2</sup> of leaves were vacuum infiltrated, in 100 mM Tris-HCl buffer (pH 7.6) solution, containing 50 mM L-ascorbic acid, 500 mM KCl and 25 mM 2-mercaptoethanol (at 4◦C). The blotted sections were centrifuged at 5000 g, during 15 min at 4◦C, and the collected APF frozen. This fraction was subsequently desalted, concentrated and purified (Guerra-Guimarães et al., 2014). APF protein quantification was made using a modified Bradford assay method (Ramagli, 1999). The purity of the APF was evaluated, prior to protein denaturing, by measuring the relative activity of malate-dehydrogenase, used as a cytosolic marker (Alves et al., 2006; Guerra-Guimarães et al., 2009b).

# 2D Electrophoresis

As previously described (Guerra-Guimarães et al., 2014) IEF was performed in IPG strips with slight alterations. One hundred microgram of protein was loaded to 13 cm IPG strips (linear pH gradient of 4–7; GE Healthcare). The Ettan IPGphor (GE Healthcare) was used under the following conditions: a total of 33,000Vh at 20◦C; Step-n-hold 100V-2h; Step-n-hold 30V-10h; Step-n-hold 250V-250Vh; Step-n-hold 500V-750Vh; Step-n-hold 1000V-1500Vh; Step-n-hold 2500V-2500Vh; Gradient 8000V-4h; Step-n-hold 8000V-40000Vh; maximum current setting of 50µA per strip. After IEF, the proteins in the IPG strip were equilibrated for 15 min on a buffer (100 mM Tris–HCl pH 8.8, 6 M urea, 2% SDS, 30% glycerol, and 0.2 mg/mL bromophenol blue) containing 5 mg/mL DTT (to reduce proteins), followed by another 15 min equilibration in the same buffer but containing 25 mg/mL iodoacetamide (to alkylate proteins) instead of DTT.

The second dimension SDS-PAGE was performed at 20◦C with 12% resolving gels using the Hoefer SE 600 Ruby apparatus (GE Healthcare) at 10 mA per gel, for the first 15 min, and 20 mA per gel for the next 4 h, or until the bromophenol blue dye front had run off the gel. Precision Plus Protein All Blue Standards (Bio-Rad, Hercules, CA) were used for molecular mass determinations.

# Gel Staining and Image Analysis

For informatics analysis gels were first stained with Ruthenium II Tris (bathophenantroline disulfonate) (RuBP) according to Lamanda et al. (2004) and the images acquired in the FLA-5100 Fluorescent Image Analyzer (FujiFilm), with the LPFR filter and at 550 V and 50µm resolution. For spot picking, the same gels were subsequently stained in Colloidal Coomassie Blue (Neuhoff et al., 1985). The image gel analysis was carried out using the Progenesis SameSpots 2D software v. 4.5 (Nonlinear Dynamics Ltd). The spot volumes were normalized using the mean value of the replicates (Grove et al., 2008) (Supplementary Table S1). One-Way ANOVA was performed between the 3 samples analyzed (resistant, susceptible, and control) using a p-value of 0.05. For the proteins with statistically significant changes (and a fold change >1.5) a principal component analysis (PCA) was carried out and a hierarchical clustering was performed applying a Pearson correlation using the MeV 4.9 (Supplementary Table S2).

### MS-based Spot Identification

Polypeptide spots (n = 169) whose abundance changed significantly between samples (p-value of 0.05 and fold change > 1.5) and were visually detected in Colloidal Coomassie Blue stained gels were excised from the gels and processed using the Tecan freedom EVO200 (Tecan, Männedorf, CH). Briefly, each sample was washed initially in a 50 mM ammonium bicarbonate solution containing 50% (v/v) methanol and dehydrated using a 75% (v/v) acetonitrile (ACN) solution and dried at 37◦C. Proteins were then digested in 8µL of trypsin Gold (Promega), 5 ng/µL trypsin in 20 mM ammonium bicarbonate. After extraction with 50% (v/v) ACN containing 0.1% (v/v) trifluoroacetic acid (TFA), the peptides were dried at 50◦C and spotted on MALDI-TOF target plates. A volume of 0.7µL of 7 mg/mL α-cyano-4 hydroxycinnamic acid in 50% (v/v) ACN containing 0.1% (v/v) TFA was added. A MALDI peptide mass spectrum was acquired using the AB Sciex 5800 TOF/TOF (AB Sciex, Foster City, CA, USA), and the 10 most abundant peaks, excluding known contaminants, were selected and fragmented.

The ProteinPilot™ software 4.0.8085 was used for database searches with an in-house MASCOT platform (version 2.3, Matrix Science, www.matrixscience.com, London, UK). All proteins were identified by search against 2 databases: an EST database of coffee containing 1527276 sequences and downloaded on September 29, 2014; a NCBInr database with the taxonomy Viridiplantae (http://www.ncbi.nlm.nih.gov) containing 40910947 sequences and downloaded on October 30, 2014. All searches (combined MS and 10 MS/MS spectra) were carried out using a mass window of 100 ppm for the precursor and 0.5 Da for the fragments. During the different searches the following parameters were defined: two missed cleavages, fixed carbamidomethylation of cysteine, variable oxidation of methionine or tryptophan, and tryptophan to kynurenine or double oxidation to N-formylkynurenine. The proteins identified without clear annotation have been used for BLAST analysis and the protein with the highest homology (when significant) added in Supplementary Table S3.

All identifications were manually validated and extra precursors were selected for fragmentation if the obtained data were judged as insufficient. When high quality spectra were not matched to sequences, a sequence was determined manually and in the current data set could be linked to the identified protein by allowing for more missed cleavages, semitryptic peptides, or specific modifications. Only spots considered for discussion were the ones that have an unique and significant protein identification. The spots which contained more than one protein were not considered in the study, since we don't know which protein increased/decreased.

### Further Data Processing

For the polypeptide spots that only gave one identified protein a subsequent bioinformatic analysis was performed. The basic information was obtained using the InterProt, UniProt, and NCBI databases. The conserved domains of each protein as well as the superfamily were determined using the NCBI tools (http://www.ncbi.nlm.nih.gov). The subcellular location assignment of the proteins were performed using TargetP 1.1, SignalP 4.1 and SecretomeP 2.0 servers (http:// www.cbs.dtu.dk/services/), and the LocTree3 (https://rostlab. org/services/loctree2/) (Emanuelsson et al., 2007; Bendtsen et al., 2004; Petersen et al., 2011; Goldberg et al., 2014). The evaluation of the Transmembrane domains was carried out using Transmembrane Hidden Markov Model analysis on TMHMM server v2.0 (http://www.cbs.dtu.dk/services/TMHMM-2.0/) and the presence of a glycosylphosphatidylinositol (GPI) anchor was carried out using GPI-anchor Predictor (http:// gpcr.biocomp.unibo.it/predgpi/pred.htm) and big-PI Plant Predictor (http://mendel.imp.ac.at/gpi/plant\_server.html) (Krogh et al., 2001; Eisenhaber et al., 2003; Pierleoni et al., 2008). Assignment for functional annotation of the identified proteins was based on MapMan "Bin" ontology (http://mapman. gabipd.org/web/guest/mapman) using Mercator Automated Sequence Annotation Pipeline (http://mapman.gabipd.org/web/ guest/app/mercator) (Lohse et al., 2014) and Gene Ontology Annotation (GO; http://www.geneontology.org) using Blast2GO software (version 2.8.2, http://www.blast2go.de/) (Conesa and Gotz, 2008). The default parameters were used for all the programs.

# Immunodetection Assays Peptide Selection

In order to produce antisera against the coffee apoplastic protein sequences, peptides with minimal homology (to reduce the chance of non-specific antibody binding) were selected after BLASTp search. With the overall aim to identify protein regions that are most likely accessible on its surface, the hydrophobic status was determined by the software BioEdiT. A hydrophilicity plot (calculated using the Kyte-Doolittle or the Hopp-Woods algorithm) indicates which parts of the protein are probably exposed. Structure predictions were done with Chou-Fasman plots. We selected two potential peptide candidates with typical lengths from 12 to 13 amino acids for each protein (Supplementary Table S4). Peptides were purchased from Thermo Fisher Scientific Inc. (NYSE: TMO).

## Peptide Conjugation

To increase the immunogenicity of the peptides they were carrier conjugated to ovalbumin (OVA) or bovine serum albumin (BSA). Coupling was performed using one step glutaraldehyde conjugation (Hermanson, 2013), using a 5:1 ratio peptide/protein. The BSA-peptide conjugates were used in the immunization protocol and the OVA-peptide conjugates were used in the ELISA assay.

## Animals

CD1 male mice were obtained from the Breeding Laboratory of IHMT/UNL and were housed in cages and fed autoclaved chow and water ad libitum.

## Immunization Protocol

The pre-immune serum was collected by sub-mandibular bleeding and then the mice were immunized with five doses. Doses were administered via the intra-peritoneal (doses 1 and 2), intra-dermal and subcutaneous (doses 3–5) with 10 to 15 days intervals between doses. Complete Freund's adjuvant was used in dose 1 and incomplete Freund's adjuvant plus peptide adjuvant (MDP, muramyl dipeptide, 10µg per mice) and synthetic dsRNA (Double-stranded homopolymer Poly (I:C), 10µg per mice, Sigma-Aldrich), was used in the other doses. No adjuvant was used in the fifth dose.

### Elisa Procedure

Wells of microtiter plates (Greiner) were coated with plant extract samples (10–100µg/ml) in 50µl of extraction buffer (0.1 M Tris-HCl, 0.5 M KCl, 0.1 mM PMSF, and 0.1% sodium sulphite, pH 7.4) or with peptides conjugated with OVA (10µg/ml) for 1 h at 37◦C. The plates were then blocked with 100µl blocking buffer (PBS with 1% PVA, pH 7.4) for 1 h at room temperature (22◦C). Polyclonal antibodies in gelatin buffer (PBS, pH 7.4, containing 0.1% gelatin) were then added at 1:500 concentration, and plates were incubated for 1 h. Secondary antibodies (anti-mouse IgM or IgG Alkaline phosphate conjugated, Sigma-Aldrich) in washing buffer [PBS, pH 7.4, containing tween 0.05% (v/v)] were added at a dilution of 1:10000 and incubated for 1 h at room temperature (22◦C). The plates were incubated with chromogen/substrate [nitrophenyl phosphate (4-NPP), in 10 mM ethanolamine buffer, pH 9.6, containing 0.5 mM MgCl2]. The absorbance at 405 nm was checked with an ELISA microplate reader. The volume was 50µl/well except for the blocking buffer (100µl/well). For each antigen, the cut-off value, which differentiates positive from negative results, was set by defining the cut-off as the mean value of the normal serum group plus three standard deviations.

### Ethics Statement

Animal studies were carried out in strict accordance with the Guidelines for Proper Conduct of Animal Experiments by DGAV (Portugal) and approved (ref 0421/000/000/2013). The animal experiments were conducted in strict compliance with animal husbandry and welfare regulations. Regular veterinary care and monitoring, balanced nutrition, and environmental enrichment were provided by the IHMT-UNL.

# Results

# Fungal Growth and Hypersensitive Host Cell Death

During H. vastatrix growth, after the differentiation of germ tubes and appressoria over stomata, the fungus infected both susceptible and resistant leaf tissues in a similar way, reaching in succession the stages of penetration hypha, anchor, and haustorial mother cell (HMC). The stomatal subsidiary cells were the first plant cells to be invaded by the haustoria. These specialized intracellular hyphae (responsible for fungus nutrients absorption) started to be formed between 24 and 48 hai. In the leaves of the resistant samples, the penetration hypha (**Figure 1A**) was the fungal growth stage observed with higher frequency during the all time-course of the experiment reaching about 55% at 24 hai and 50% at 96 hai, while HMC with haustorium (HMC/h) only reached 15% of infection sites at 72 hai, and did not exceed 22% at 96 hai (**Figure 1D**); at this stage the fungus stop growth and died. In the leaves of the susceptible samples, the penetration hypha was also the most representative stage at 24 hai (61%) and 48 hai (41%) but, later on, the HMC/h greatly increased in frequency (40% at 72 hai and 44% at 96 hai), being responsible for the successful fungal growth (**Figures 1B,D**). The death of the fungus was experimentally assessed by the autofluorescence of the fungal structures that, at 96 hai, reached

The first cytological response induced by the fungus in the resistant and susceptible samples is the hypersensitivelike reaction (HR) observed initially in the stomata guard and subsidiary cells and later in mesophyll cells. At 24 hai, HR occurred for both resistant and susceptible samples reaching, respectively, 33 and 20% of infection sites, where the fungus stopped growth (at the stages of appressorium or penetration hypha). HR was always significantly higher in the resistant than in the susceptible samples at all time-points (**Figures 1C,E**). Only in the resistant samples was the HR observed in subsidiary stomatal cells and mesophyll cells invaded by haustoria, from 72 hai (65%) onwards (71% at 96 h).

### APF Protein Expression upon Infection

The APF was obtained from resistant, susceptible and control leaves (mock-inoculated) along the H. vastatrix infection process (24–96 hai). Proteins were separated by 2-DE and statistical analysis of the gel patterns was performed to reveal the polypeptide spots whose volume significantly changed in abundance (p-value ≤ 0.05 and fold change > 1.5) between samples for each of the time-points. The number of spots that changed were 35, 37, 84, and 54, respectively, at 24, 48, 72, and 96 hai. MALDI—TOF/TOF MS analysis of the excised polypeptide spots (n = 169) revealed 116 spots that have only one protein identification (**Figure 2**, **Table 1** and Supplementary Table S3). According to their conserved domains (NCBI database), these proteins belong to 23 diverse superfamilies (**Table 1**). Bioinformatic tools suitable for predicting secreted proteins were used in order to confirm the extracellular localization of the identified proteins. No trans-membrane domain (TMD) (TMHMM2.0) or glycosylphosphatidylinositol (GPI)-anchor (GPI-anchor Predictor and big-PI Plant Predictor) were detected in the sequences of the 116 proteins analyzed. Making use of a set of several secrete protein predictor programs (SignalP4.1, TargetP1.1, LocTree3, and SecretomeP) all the 116 proteins were indicated to be of secreted nature; from which 110 proteins had the N-terminal signal peptide typical of the classical secretory pathway (SignalP4.0 and/or TargetP1.1) and the remaining 6 proteins (that lack the classical terminal signal peptide) were recognized as leaderless secretory proteins (SecretomeP program) (Supplementary Table S5). Overall, the results confirm the high quality of the APF samples, since no or little cytoplasmic contamination was detected (APF activity of malate dehydrogenase was always less than 5% of the activity of total leaf homogenates). The functional categorization of the identified proteins was performed, gathering information from different annotation tools (GO ontology and MapMan "Bin"). Annotation revealed that the identified proteins were involved in: protein degradation (36%), cell wall metabolism (23%), stress/defense (23%), miscellaneous enzyme families (11%), minor carbohydrates (CHO) metabolism (4%), secondary metabolism (2%), and redox (1%) (**Table 1** and **Figure 3A**). Seventy three percent of these proteins have EC numbers (Blast2GO analysis) which mainly represent hydrolases, particularly sugar hydrolases and peptidases/proteases

(**Figure 3B**). Analyzing the changes along the infection process it is more evident that the % of proteins involved in proteolysis decreased after an initial increase (53% at 24 hai and 12% at 96 hai) while the % of proteins involved in stress/defense increased along the infection process (9% at 24 hai and 40% at 96 hai). A few proteins are present at all time-points, namely, xylosidases, mannosidases, chitinases, subtilases, and aspartic proteases.

# APF Proteins Associated with Resistance and Susceptibility

A Principal Component Analysis (PCA) was performed for the spots whose volume significantly changed in abundance during the infection. This analysis revealed a clear separation of the three samples (resistant, susceptible, and control) for each of the four time-points, the two first axes always representing more than 70% of the total variance (**Figure 4** and Supplementary Table S2). To visualize the relative accumulation of the spots in the resistant (R) and the susceptible (S) samples, a hierarchical cluster analysis was performed (**Figure 5**). At 24 hai, the protein patterns for the two infected samples showed differences mainly concerning an increase in proteolysis and in stress/defense in the R samples (e.g., cysteine proteinases, subtilases, berberine bridge enzyme, cysteine-rich repeat secretory protein, osmotin, and chitinase). Changes in a calcineurin-like phosphoesterase and a GDSL-motif lipase/hydrolase (miscellaneous enzyme families) are also of significance. At 48 hai, it is remarkable that the two infected samples did not markedly differ from each other, both showing a strong decrease in abundance for the same proteins, e.g., beta-D-xylosidase, chitinases, and aspartic proteases. It is at 72 hai, that the main differences between R and S samples started to be evident. Most of the proteins that increase in the R sample at 72 hai are involved in proteolysis (e.g., subtilases and serine carboxypeptidases) and in cell wall degradation/modification (e.g., betaxylosidase/alpha-arabinofuranosidases, chitinase, glucanase and pectin methylesterase, purple acid phosphatase, reticuline oxidase). However, a strong increase in stress/defense proteins (e.g., PR-1, osmotin, chitinases, thaumatin-like, NtPRp27

protein, and beta-1,3-glucanase) and beta-galactosidases were observed mostly in the S sample, at 96 hai. There is a noticeable increase in alpha-L-fucosidase proteins in the R sample at 96 hai.

### Immunodetection Assay

Some of the identified proteins, referred above, were selected as antigen for the production of antibodies, such as, chitinase, pectin methylesterase, serine carboxypeptidase, reticuline oxidase, and subtilase. Peptides corresponding to these proteins were synthesized and after conjugation with BSA and OVA allowed the production of specific antibodies. The results obtained show a higher level of detection of those proteins in the R than in the S or control samples (**Figure 6**).

## Discussion

We have been studying the APF coffee leaf proteins in response to H. vastatrix infection (Guerra-Guimarães et al., 2009b, 2013), and recently, we have characterized the proteome of APF healthy coffee leaves (Guerra-Guimarães et al., 2014). With the present study we complement the knowledge on the importance of the proteins present in this sub-cellular compartment, particularly in relation to pathogen defense. In addition to the proteins previously found, a further seven protein superfamilies were now identified in the APF of coffee leaves (control sample), making a total of 29 protein superfamilies. The new identified protein superfamilies are mainly PR proteins, phosphatases and oxi-reductases, highlighting the existence of an important constitutive defense mechanism in C. arabica leaves against pathogens. We further studied the C. arabica-H. vastatrix pathosystem aiming to discover changes in the leaf APF proteome during the evolution of the infection process (24 hai-96 hai) in both incompatible and compatible interactions (R and S samples). The results obtained support the existence of two phases of defense responses, an initial/basal response and a later/specific response. The number of proteins involved in the initial/basal phase (24–48 hai), is half of the number involved in the late/specific phase (72–96 hai), grouped in 23 protein superfamilies of which four are present only in the initial phase, nine in the late phase and the remaining 10 in both phases.


### TABLE 1 | Annotation of the coffee leaf apoplastic proteins that changed in abundance along the infection process.



<sup>a</sup> Functional characterization of the proteins based on MapMan "Bin" and GO ontology.

<sup>b</sup> The number that identified protein spots on 2-D apoplastic gel.

<sup>c</sup> The peptide identification based on homology to proteins characterized in different species by BLASTp. search on NCBI Viridiplantae and ESTcoffee databases.

<sup>d</sup> The accession number from GenBank assigned to the polypeptide after MS/MS analysis.

<sup>e</sup> Superfamily according to NCBI classification. GH, Glycoside Hydrolase; SGNH\_hydrolase, diverse family of lipases and esterases; FAD\_binding, flavodoxin binding oxiredutase; GluZincin, thermolysin-like peptidases including several zinc-dependent metallopeptidases.

<sup>f</sup> hours after inoculation with H. vastatrix.

<sup>g</sup> Samples R (resistant), S (susceptible) that change in abundance relatively to control, ↑ (increase), ↓ (decrease).

### Initial/Basal Defense Responses

The identification of GDSL-motif lipase/hydrolase (spot #3136) and calcineurin-like phosphoesterase (spot #1658) at 24 hai, suggests the potential involvement of these proteins in pathogen perception and signal transduction cascades. GDSL esterases/lipases are proteins with multifunctional properties, described as having a role in the regulation of plant development, morphogenesis, synthesis of secondary metabolites, and defense response (Chepyshko et al., 2012). In Arabidopsis thaliana a GDSL LIPASE1 protein seems to protect plants from Alternaria brassicicola attack in two distinct ways: by directly disrupting fungal spore integrity, and by activating defense signaling in the plants (Oh et al., 2005). In our study we have detected a decrease in the accumulation of the protein GDSL-motif lipase/hydrolase at 24 hai, in both infected tissues. According to Lee et al. (2009) such a decrease can be either a negative regulation of proteins to inhibit fungal infection/growth or, in addition, the effect of fungal interacting with the plant cell (by means of effector proteins) by suppressing the host immune system. On the other hand, the increased accumulation of calcineurin-like phosphoesterase (a calcium–dependent phosphatase) can be important in the regulation of various cellular processes with emphasizes in signal transduction as has already been shown (Kudla et al., 1999; Luan, 2003). It is known that upon perception of microbial signals, kinases and phosphatases target specific proteins, often modifying complex signaling cascades that allow for rapid defense responses (Delanois et al., 2014). The presence of phosphatases in the extracellular proteome of Arabidopsis infected with Pseudomonas syringae suggests that potential phosphorylation/dephosphorylation reversible regulation could occur in the apoplast (Kaffarnik et al., 2009). Moreover, Ndimba et al. (2003) have shown that chitosan treatment of Arabidopsis cell-suspentions induced phosphorylation of a receptor-like kinase, and other proteins like chitinases and glucanases (proteins that we have also found to be accumulated at 24 h, particularly in the R samples).

The increased accumulation of PR proteins (chitinases, osmotin and a cysteine-rich repeat secretory protein) in both infected tissues (more markedly in the R samples) also indicates the induction of the basal defense responses, possibly salicylic acid (SA) regulated. Molecular studies on Coffea spp.—H. vastatrix incompatible interaction did show the activation of genes (ex.pr1b and gt) known to be involved in the SA mediating signaling pathway around 21–24 hai (Diniz et al., 2012). Furthermore, SA quantification by HPLC/ESI-MS/MS showed an increase in this signaling compound at 24 hai in Coffea spp.—H. vastatrix incompatible interaction, suggesting again the involvement of an SA-dependent pathway in coffee resistance to CLR (Sá et al., 2014).

The accumulation of berberine bridge enzyme (a reticulinelike oxidase) in the R sample at 24 hai and a copperzinc superoxide dismutase (SOD) in S sample at 48 hai, suggests that these "PR-like" proteins may co-regulate basal defenses. Extracellular oxidases have been suggested to catalyze the generation of reactive oxygen species (ROS), such as superoxide anions and hydrogen peroxide during the "oxidative burst" (Martinez et al., 1998; Mika et al., 2004). Indeed, previous cytochemical data in an incompatible C. arabica— H. vastatrix interaction, revealed hydrogen peroxide in the interface between the cuticle and the fungal pre-penetration structures at the infection sites (Silva et al., 2008). Furthermore, the increase in the activity of peroxidases, SOD and oxalate oxidases (germin–like proteins) have already been reported during the resistant response of coffee to CRL (Silva et al., 2006, 2008; Guerra-Guimarães et al., 2009a, 2013). The oxireductase activity observed during infection by pathogens indicates that plants were either initiating the production of ROS to fight directly the pathogen or responding to oxidative intermediates produced as a result of cell wall or membrane damage leading to cell death during HR response (Lee et al., 2009).

### Late/Specific Defense Responses

Although the HR was already observed at 24 hai, it continued to increase with time and at 72 hai it was much higher in the R than in the S samples. Simultaneously, in the R sample the number of proteins changing in volume increased dramatically, suggesting their contribution to a second and stronger line of defense responses. On the contrary, in the S sample the fungus

FIGURE 4 | Principal Component Analysis (PCA) performed for the spots whose volume significantly changed in abundance (*p*-value < 0.05), for each time-point of the infection (24–96 hai). Distinct groups were obtained per sample: control (C), resistant (R), and susceptible (S).

continued growing with no apparent inhibition, the HR stabilized and protein levels did not change much more than in the control. Most of the proteins that increased in the R sample at 72 hai have hydrolytic activity, being either involved in the cell wall metabolism (beta-xylosidase/alpha-arabinofuranosidases, chitinases and glucanase, pectin methylesterase, purple acid phosphatase, and reticuline oxidase) or in proteolysis (subtilases and serine carboxypeptidases).

It is known that plant glycohydrolases (GH) can play various important functions such as, cell wall expansion, modification during development, defense, and signaling. Since plant cell wall polysaccharides are very heterogeneous and complex polymers, GH activities must be very diverse (Jamet et al., 2008) and with our proteomic approach we identified in the apoplast a total of eight GHs superfamilies (3, 17, 18, 20, 31, 35, 38, 64). According to the carbohydrate-active enzymes database (CAZy; www.cazy.org) (Lombard et al., 2014), the GHs families GH3, GH31 and GH35 comprise enzymes that are mainly involved in the reorganization of cell wall carbohydrates. The other GHs families seem to be involved in glycoprotein post-translational modifications (PTMs), such as alpha-L-arabinofuranosidases (GH3), chitinases (GH18), beta-D-galactosidases (GH35), and alpha-D-mannosidases (GH38) (Jamet et al., 2008).

Alpha-L-arabinofuranosidases are particularly interesting since they accumulate only in the R sample at 72 hai. They are plant enzymes capable of releasing terminal arabinofuranosyl residues from cell wall matrix polymers (Saha, 2000), functioning as a candidate for a role in softening-related depolymerization of the cell wall during the HR response (Cantu et al., 2007). Several other apoplastic proteins identified, are also GHs, and appear to contribute to plant defense. Chitinases (GH18) and beta-1,3-glucanases (GH17) that are PR proteins possess antifungal activity limiting pathogen progression, and their expression is often triggered by pathogen infection (Silva et al., 2006; Guerra-Guimarães et al., 2009b). Leah et al. (1991) and Mauch et al. (1988) showed that the antifungal proprieties of plant chitinases are enhanced when beta-1,3-glucanases were added in combination with them. In transgenic tobacco plants, susceptibility to fungal attack decreased when chitinase and glucanase genes were both over-expressed (Zhu et al., 1994). Other PR-proteins such as PR-1 and PR-5 also increased in the resistant sample from 72 hai onwards.

Also relevant was the detection of pectin methylesterases (PMEs) and purple acid phosphatases (PAP) exclusively at 72 hai in the resistant sample. The activities of PMEs from both plants and pathogens and the degree and pattern of pectin

methyl esterification are critical for the outcome of plant– pathogen infections. The cell walls containing highly methyl esterified pectin are somehow protected against the action of pathogens (Lionetti et al., 2012). Concerning the PAP, it was shown that a PAP5 is required for maintaining basal resistance against Pseudomonas syringae in Arabidopsis, suggesting a role for PAP5 in pathogen triggered immunity (Ravichandran et al., 2013).

Proteolytic enzymes that are thought to be involved in maturation of enzymes, signaling, protein turnover, and defense against pathogens (Jamet et al., 2008) were the proteins that mostly changed in abundance between the R and S samples, at 72 hai. They represent 36% of the total proteins identified and belong to 4 different superfamilies; subtilisinlike protease, serine carboxypeptidase, aspartic protease, and cysteine proteinase. Serine proteases (subtilases and serine carboxypeptidases) were the most relevant as they increased abundantly in the resistant sample, particularly at 72 hai. Several subtilases are specifically induced following pathogen infection and an Arabidopsis subtilase (SBT3.3) was very recently hypothesized to function as a receptor located in the plasma membrane that activates downstream immune signaling processes (Ramirez et al., 2013). When comparing grapevine genotypes resistant and susceptible to Plasmopara viticola, a subtilisin-like protein sharing sequence similarity with the tomato P69 (a PR protein specifically induced following pathogen infection) was shown to be constitutively expressed in the resistant genotype; and its expression was induced after pathogen infection (Vartapetian et al., 2011; Monteiro et al., 2013; Figueiredo et al., 2014).

In addition to the already referred functions of oxidases in the defense responses, it should be discussed the later increase in reticulin oxidase and germin-like proteins (oxalate oxidase-like) at 96 hai. These proteins can have a role in the oxidative crosslinking of cell wall proteins around the site of infection (Bradley et al., 1992; Silva et al., 2008). Crosslinks between phenolic compounds, the plant cell wall polysaccharides and proteins enhance the protection of the cell wall to digestion by microbial degrading enzymes and, thus, increase the global resistance to fungi (Bily et al., 2003). Deposition of chlorogenic acids and lignin has, indeed, been associated with the resistance of coffee to H. vastatrix (Silva et al., 2002, 2006; Leitão et al., 2011).

Overall, the protein changes occurring in the APF of coffee leaves upon H. vastatrix infection indicate that cell wall reorganization, accumulation of PR proteins and excretion of hydrolytic enzymes are likely to be important defense mechanisms of coffee. The use of antibodies produced against chitinase, pectin methylesterase, serine carboxypeptidase, reticuline oxidase, and subtilase showed an increased detection of these proteins in the incompatible interaction what strengthens their involvement in the resistant response of coffee against H. vastatix.

## Conclusions

Important constitutive defense proteins were revealed in the APF of C. arabica leaves. Upon infection by H. vastatrix, APF

## References


proteins were modulated establishing two distinct phases of defense responses, an initial/basal one (at 24–48 hai) and a late/specific one (at 72–96 hai). The number of proteins detected for the initial/basal phase is essentially half of the number of the proteins for the late/specific phase. When comparing the susceptible and resistant sample it was found that the increase in proteins was always greater in the resistant samples and more markedly in the late/specific phase. The resistant response involves the participation of several important groups of proteins, namely: GH of the cell wall, serine proteases (subtilases and carboxypeptidases) and PR proteins. The GHs confer great plasticity to cell wall polysaccharides, the proteases (together with phosphatases) lead to a complex regulation of cell wall proteins through PTMs and PR proteins are directly involved in antifungal activity. These results suggest that some glycohydrolases, proteases, and PR-proteins are putative candidates for resistant markers of coffee to CLR. The production of antibodies against chitinase, pectin methylesterase, serine carboxypeptidase, reticuline oxidase, and subtilase enabled the validation of the importance of these proteins in the coffee resistance response by immunodetection assay. Reliability of these putative resistant markers will be subsequently tested in several well-known coffee cultivars with commercial value. The genes corresponding to the protein biomarkers can be integrated in marker-assisted breeding programs aiming to assist in the selection of appropriate coffee genotypes with resistance to H. vastatrix.

# Acknowledgments

This work was supported by Portuguese Funds through FCT (Fundação para a Ciência e a Tecnologia), under the project PTDC/AGR-GPL/109990/2009 (at CIFC/IICT, ITQB/UNL, and IHMT/UNL) and PEst-OE/EQB/LA0004/2011 (at ITQB/UNL) and GHTM – UID/Multi/04413/2013 (at IHMT/UNL). DRB was on a postdoctoral grant from CNPq (Brazil) and RT received a STSM grant from COST action FA1306. The authors wish to thank Doctor Colin E. McVey (Principal Investigator and Head of Structural Virology Lab, ITQB/UNL) for critically reviewing the manuscript.

## Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00478

proteins. Proteomics 10, 799–827. doi: 10.1002/pmic.2009 00514


(2008–2013): impacts, plausible causes and proposed solutions. Food Sec. 7, 303–321. doi: 10.1007/s12571-015-0446-9


quantitative RT-PCR. Plant Sci. 170, 1045–1051. doi: 10.1016/j.plantsci.2005. 12.009


coffee leaf rust resistance," in Proceedings of the 24th International Conference on Coffee Science (ASIC2010) (Bali).


**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.

Copyright © 2015 Guerra-Guimarães, Tenente, Pinheiro, Chaves, Silva, Cardoso, Planchon, Barros, Renaut and Ricardo. 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.

# **The quest for tolerant varieties: the importance of integrating "omics" techniques to phenotyping**

*Michel Zivy <sup>1</sup> , Stefanie Wienkoop <sup>2</sup> , Jenny Renaut <sup>3</sup> , Carla Pinheiro 4,5, Estelle Goulas <sup>6</sup> and Sebastien Carpentier 7,8 \**

*<sup>1</sup> Department Génétique Quantitative et Évolution, Le Moulon INRA, CNRS, AgroParisTech, Plateforme PAPPSO, Université Paris-Sud, Gif-sur-Yvette, France, <sup>2</sup> Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria, <sup>3</sup> Department of Environmental Research and Innovation, Luxembourg Institute of Science and Technology, Belvaux, Luxembourg, <sup>4</sup> Instituto de Tecnologia Química e Biológica, New University of Lisbon, Oeiras, Portugal, <sup>5</sup> Faculdade de Ciências e Tecnologia, New University of Lisbon, Caparica, Portugal, <sup>6</sup> Department of Sciences et Technologies, CNRS/Université Lille, Villeneuve d'Ascq, France, <sup>7</sup> Department of Biosystems, University of Leuven, Leuven, Belgium, <sup>8</sup> SYBIOMA, University of Leuven, Leuven, Belgium*

### *Edited by:*

*Jesus V. J. Novo, University of Cordoba, Spain*

### *Reviewed by:*

*Ján A. Miernyk, University of Missouri, USA Borjana Arsova, Université de Liège, Belgium*

### *\*Correspondence:*

*Sebastien Carpentier, SYBIOMA, University of Leuven, Willem de Croylaan 42, Box 2455, 3001 Leuven, Belgium sebastien.carpentier@biw.kuleuven.be*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 13 March 2015 Accepted: 31 May 2015 Published: 09 July 2015*

### *Citation:*

*Zivy M, Wienkoop S, Renaut J, Pinheiro C, Goulas E and Carpentier S (2015) The quest for tolerant varieties: the importance of integrating "omics" techniques to phenotyping. Front. Plant Sci. 6:448. doi: 10.3389/fpls.2015.00448* The primary objective of crop breeding is to improve yield and/or harvest quality while minimizing inputs. Global climate change and the increase in world population are significant challenges for agriculture and call for further improvements to crops and the development of new tools for research. Significant progress has been made in the molecular and genetic analysis of model plants. However, is science generating false expectations? Are 'omic techniques generating valuable information that can be translated into the field? The exploration of crop biodiversity and the correlation of cellular responses to stress tolerance at the plant level is currently a challenge. This viewpoint reviews concisely the problems one encounters when working on a crop and provides an outline of possible workflows when initiating cellular phenotyping via "-omic" techniques (transcriptomics, proteomics, metabolomics).

**Keywords: proteomics, data integration and computational methods, phenotype, omics-technologies, crop improvement**

# **Introduction**

The need for higher yields with lower inputs is widely recognized as necessary to meet the challenge of feeding 9 billion people in 2050 (Godfray et al., 2010). Agricultural management is being challenged by erratic climates and the occurrence of extreme stress events that have the potential to destroy crop production in many geographical regions. Stress is complex and involves timing, duration and severity (Blum, 2014). Moreover, plant stress in agriculture is a phenomenon that is correlated with the genotype, environment and management (G *×* E *×* M). Breeding toward stress tolerance is limited in many crops and the current commercially grown varieties have mainly been selected for production and excellent post-harvest qualities, with less attention to other features (e.g., drought tolerance, nutrient use efficiency, durable pest and disease resistance, environmental repercussions etc.). The effects of these factors have been/are mitigated by the use of treatments such as irrigation, pesticides and fertilizers. These management processes have reduced the sense of urgency and resulted in the use of existing plant genetic resources in overcoming crop limitations. Consequently, several ancient varieties and landraces or even wild crop relatives containing useful sources of resistance or tolerance are now underutilized. However, the use of pesticides, fertilizer and water must be reduced and agriculture must become more sustainable. Recently, many governments have commenced initiatives to promote plant germplasm collections which increase the range of material that can be explored in search of genotypes less affected by stress1,2. Reliable identification of tolerant varieties and the understanding of their genetic diversity are urgently needed. The knowledge gap is a strong propeller for the generation of biological knowledge (both fundamental and applied) and provides the plant biology community with the opportunity to establish different experimental models (genomics, transcriptomics, proteomics, metabolomics, phenomics) for different crops. Phenotyping is an emerging field that characterizes plant behavior and quantify features, such as growth and yield, in a way that allows linking to genetic control. However, the evaluation of genetic biodiversity is a research bottleneck and there is still a significant gap between the lab and the field. Is science generating false expectations? Are 'omic techniques, such as proteomics and metabolomics, generating valuable information that can be translated into practice? Is proteomics better than metabolomics, transcriptomics or genomics? Recently, a European network was created to help tackle this issue and develop new workflows to integrate the different 'omic techniques: COST action FA1306 "The quest for tolerant varieties—Phenotyping at plant and cellular level<sup>3</sup> ." The aim of the Action is the improvement and exchange of scientific knowledge in plant phenotyping through the creation of a network between European interdisciplinary scientists and to use this network to: map valuable gene bank collections and breeding programs in Europe, train breeders and physiologists in screening techniques and data interpretation, get insight into the genetic basis of tolerance, to characterize current biodiversity and rank it according to tolerance levels and to apply the knowledge for agricultural management. The Action started in May 2014 and 28 countries have currently joined. This viewpoint embodies the vision of this COST action and describes concisely the problems one encounters when phenotyping the diversity found in crops. It specifically provides an outline of the problems encountered when initiating "cellular phenotyping" through 'omics techniques, highlighting proteomics.

# **Understanding Gene function**

Understanding gene function can be approached via several techniques: genomics, transcriptomics (messenger, structural and regulatory RNA's), proteomics (proteins and their putative posttranslational modifications (PTM) and peptides) and metabolomics (primary and secondary metabolites). In prokaryotes, gene finding is essentially a matter of identifying open reading frames. As genomes get larger, it becomes increasingly complicated. Several sophisticated software algorithms have been designed to handle gene prediction in eukaryotic genomes. Despite considerable progress, gene prediction entirely based on DNA analysis is cumbersome and requires support from "functional genomics," i.e., transcriptomics, proteomics, and metabolomics. Indeed, genomics focuses on the static aspects of genome information. Gene prediction and annotation in a reference variety is the initial step for every crop, but this is not sufficient to get complete insight into the phenotypic plasticity and the agricultural potential of the biodiversity. Functional genomics deals with dynamic aspects, reflecting environmental adaptations and allows the description of gene functions as well as the interactions between gene products that may provide a view of the agricultural potential of a variety/genotype.

# **Transcriptomics**

Probably the easiest way to study changes on a genome-wide scale is through transcriptomics. The structure of RNA is homogenous and relatively simple and therefore the analysis is the most straightforward when compared to protein and metabolite analyses.

Serial analysis of gene expression (SAGE), developed by Velculescu et al. (1995), is based on the generation of 15 bp tags from a defined position in each transcript, which are then concatenated, cloned into a plasmid vector and ultimately sequenced (Velculescu et al., 1995). Massively parallel signature sequencing is a more advanced technique based on sequencing of tags. It generates 17 bp tags that are sequenced using a fluorescence-based signature sequencing method on microbeads and was applied on multiple model organisms (Brenner et al., 2000; Reinartz et al., 2002). To analyze the abundance of a transcript, one simply calculates the number of times that a certain tag was found. Though no sequence data needs to be identified *a priori*, the tag needs to be identified as belonging to a gene to convey its biological meaning and this step can be difficult in plants with limited genetic resources. DNA sequences are not as well conserved as amino acid sequences and therefore a cross-species identification based on a short tag is problematic. Matsumura et al. (2003) developed superSAGE in rice, which utilizes longer tags (26 bp). However, the generation of longer tags still resulted in the SAGE-approach for unsequenced non-model crops challenging, as illustrated for banana (Coemans et al., 2005; Carpentier et al., 2008a). At the same time microarrays, which were significantly cheaper and was a more high-throughput technology, were also developed for transcriptomic studies. Microarrays use known probes that will hybridize with the labeled sample and based on the intensity of these dyes, transcript levels are estimated. This however, implies that sequence information exists before generation of the microarray and this is a serious limitation when applied to nonmodel crops. The limited sequence availability in non-models can be overcome by the use of microarrays of closely related species or by the generation of a species-specific microarray based on known expressed sequence tag (EST) data for instance, but these analyses will be less informative (Davey et al., 2009; Pariset et al., 2009). Microarray analysis is hampered by a high background noise due to cross-hybridization as well as saturation of signals. Microarrays therefore have a limited sensitivity and dynamic range. Furthermore, microarrays are closed platforms as unknown transcripts cannot be detected. An alternative for the standard gene expression microarray is the tiling microarray. These are high-density arrays composed of oligonucleotide probes that span the entire genome of an organism (Yazaki et al., 2007).

<sup>1</sup>http://www.croptrust.org/

<sup>2</sup>http://www.cgiar.org/

<sup>3</sup>http://costfa1306.eu/

Whole-genome tiling arrays may provide part of the solution toward the detection of new gene transcripts in a sequenced organism but are more expensive and still suffer from the general drawbacks of a microarray approach (Valdés et al., 2013). With the availability of next generation sequencing (NGS) technologies, the possibility to directly sequence mRNA at relatively reduced costs became available. This technique, termed RNA-seq, has clear advantages over the other transcriptomics methods: a higher sensitivity and dynamic range can be achieved (Wang et al., 2009) and no previous sequence knowledge is *per se* required. Reads can be mapped to a known reference genome or *de novo* assembled. *De novo* assembly of reads into contigs increases the use of this technique for crops whose genome has not been sequenced, as was demonstrated for wheat, agave and horse gram (Bhardwaj et al., 2013; Gross et al., 2013; Oono et al., 2013). However, a reference genome is highly recommended to assure the correct assembly of the reads and to deal with paralogs and allelic variants. Moreover, most read-mapping software has been written to analyze diploid genomes (Langmead and Salzberg, 2012) and is unsuited for polyploid organisms (Page et al., 2013a). Read mapping is a fundamental part of next-generation genomic research but is complicated by genome duplication in many plants. When a reference genome is already available, RNA-seq can provide additional information necessary to identify previously unknown gene coding sequences. Categorizing DNA sequence reads into their respective genomes enables current methods to analyze polyploid genomes as if they were diploid. Page et al. (2013a,b) developed software for SNP detection in cotton, which is an allotetraploid. Using SNP-tolerant mapping, the software uses the SNPs between genomes to categorize reads according to their respective genomes. Furthermore RNA-seq data can also be used to improve existing annotations both in identifying actual intronexon structures as well as in identifying different splice variants as was shown in maize (Kakumanu et al., 2012) or identifying homeologs. Genome-wide quantification of homeolog expression ratios was technically hindered because of the high homology between homeologous gene pairs. Additionally, in contrast to the high background noise caused by cross-hybridization in microarrays, most RNA-seq reads can be unambiguously mapped to a region of the reference genome. This makes RNA-seq in combination with reference genomes an excellent tool to differentiate between isoforms of a gene family, which are a widespread phenomenon in complex crop genomes. On the other hand, the alignment of short sequence reads that are shared between several loci and therefore align to several locations on the genome is still complicated. One solution is to assign these reads proportionally to the number of unique splice reads at these loci (Mortazavi et al., 2008). Moreover, aside from being relatively unbiased toward previous sequence knowledge, RNAseq is also more sensitive. This sensitivity comes at a price. To detect rare transcripts, coverage and therefore sequencing depth is required, which increases the sequencing cost. Lastly, the dynamic range of RNA-seq is also substantially higher, at about five orders of magnitude compared to several hundred-fold for microarrays (Wang et al., 2009; Zhao et al., 2014).

With the introduction of NGS, RNA-seq appears to be the transcriptomic tool for the future, especially in crops. At the moment, the costs associated with RNA-seq prevent large scale analysis of many varieties in different conditions and multiple biological replicates. However, as the NGS technique keeps evolving, costs are likely to drop and may no longer be a limiting factor in the future. As more and more genomes are sequenced, alignments to reference genomes should become standard practice, which will also significantly reduce the analysis time required for *de novo* assembly.

# **Proteomics**

In contrast to genomics and transcriptomics, proteomics is often regarded as a slow and cumbersome art. The discovery of soft ionization techniques for mass spectrometry (MS) by Nobel Prize winners Fenn and Tanaka, the coupling of MS to liquid chromatography and the genomic and computational advances, have made the high throughput large scale analysis of proteins feasible (Karas and Hillenkamp, 1988; Fenn et al., 1989; Henzel et al., 1993; McCormack et al., 1997). Thus, after a significant lag phase, high throughput proteomics has become an important research tool for model organisms and is currently finding its way to crop species.

Two approaches are generally distinguished in the field of proteome analysis: a protein based approach (in general, referred to as gel based) and a peptide based approach (in general referred to as gel free or shotgun). In the gel based approach, proteins are separated and quantified via gel electrophoresis. The proteins of interest are then picked from the gel, digested and the resulting peptides identified via MS by comparing experimental *versus* theoretical masses present in various databases. This technique has the advantage that protein separation and analysis via (two-dimensional) electrophoresis prior to MS analysis ensures physical connectivity between the peptides and the protein and significantly reduces complexity (Carpentier et al., 2008b). Currently, it is still the most widely used approach in crop proteomics. Unfortunately the technique has some major drawbacks, i.e., it has a very poor performance when analyzing hydrophobic and basic proteins and can be quite limited with respect to throughput.

In the gel free approach, protein digestion precedes the separation and quantification of peptides. Gel free differential proteomics provides a broader coverage of the proteome and also enables the identification of membrane proteins. However, a major disadvantage of this approach lies in the disconnection between the protein and its peptides (Carpentier and America, 2014). In general, most approaches use a bottom-up strategy where proteins are first digested with a proteolytic enzyme. Yates and colleagues were one of the early pioneers to explore the use of liquid chromatography coupled to electrospray ionization tandem mass spectrometry (LC/MS/MS) and realize the potential of automated high throughput proteomics (McCormack et al., 1997; Ducret et al., 1998; Link et al., 1999). However, proteolytic digests of a higher eukaryotic proteomes, like crops, exceed the analytical capacity of most MS. During recent years, MS have been developed with high mass accuracies, resolving power, sensitivity, scan speed, reproducibility and lower detection limits (Domon and Aebersold, 2006; Mann and Kelleher, 2008). For example, the use of Fourier transform ion cyclotron resonance based spectrometry (Yang and Yen, 2002), hybrid Linear Trap Quadrupole-Orbitrap devices (Makarov et al., 2006; Olsen et al., 2009), high energy C-trap dissociation (Olsen et al., 2007), parallel reaction monitoring (Peterson et al., 2012), the coupling of a quadrupole mass filter to an Orbitrap analyser (Michalski et al., 2011; Kelstrup et al., 2012), Ultra Performance Liquid Chromatography (UPLC) combined with moist static energy (MSE; Plumb et al., 2006), combining quadrupole, Orbitrap and ion trap mass analysis (Lebedev et al., 2014), and hybrid quadrupole time-of-flight MS (Andrews et al., 2011), have all contributed to improvements in proteomic experiments, and in particular toward better peptide identifications and quantification. Despite the development of new MS, a protein sample from a crop species is still challenging to analyze and contains several thousand proteins. This might lead to both identification and quantification problems, especially in the case of crops with complex polyploid genomes and large protein families. Peptides shared between several proteins do not contribute to the conclusive identification of a particular protein. This is the so-called protein inference problem (Nesvizhskii and Aebersold, 2005). Unique peptides need to be measured and identified for final protein identification and quantification. So a gel free approach is only applicable for crops, in practice, once a reference genome is available or when substantial EST libraries become available (Vertommen et al., 2011b). Typically, a gel free analysis starts with an MS survey scan where peptide precursor masses are measured, followed by an MS/MS scan for fragmentation of the selected precursor ion. This is called datadependent acquisition (DDA). The serial nature of the MS and MS/MS cycles and the complexity of the proteome in crops remain a challenge. There is a bias toward the more abundant peptides and no MS scan can be obtained while fragmentation is being performed in the MS/MS scan in most current MS. Moreover, coeluting peaks can lead to chimeric spectra, reduced reproducibility and loss of information about less abundant peptides. To increase the chance of identifying specific tryptic peptides, it is important to ensure a good peptide separation and to keep the mixture of co-ionizing peptides as simple as possible even in fast modern MS. Vertommen et al. (2011a) proposed a workflow for banana samples where this was achieved by using a two-dimensional RP-RP chromatography system coupled to a high accuracy MS. To identify the peptides in a gel free approach for a crop, it is important to build a species specific in-house database and to search this database in consecutive steps: a non-error tolerant manner, subsequently an error tolerant and finally *de novo* sequencing. This *de novo* approach is a crucial step, since it allows the identification of variety specific peptides via a homology search of sequences instead of a search based on m/z values (Vertommen et al., 2011a).

To identify and quantify peptides in a rapid, consistent, reproducible, accurate and sensitive way, data independent acquisition (DIA) protocols have been developed for label-free shotgun proteomics as an alternative to DDA (Plumb et al., 2006). To increase the identification rate of label free DIA experiments for samples from the apple variety Braeburn, a new workflow was developed by Buts et al. (2014) where a DDA database was constructed and linked to the DIA data. A ten-fold increase in peptides was identified from a single DIA run and proteins correlated to the storage quality of apples were found.

While in sequenced crops, two-dimensional electrophoresis (2DE) may currently no longer be the tool of choice for highthroughput differential proteomics, it is still very effective in identifying and quantifying protein species as a result of genetic variations, alternative splicing and/or PTM. As an example, by using combined 2DE and 2D DIGE with *de novo* MS/MS sequencing, Carpentier et al. (2011) were able to identify interand intra-cultivar protein polymorphisms in banana correlated to drought tolerance. Using an 2D-DIGE LC MS/MS approach Vanhove et al. (2015) were able to characterize the complex protein family of HSP70s and identify an osmotic stress specific isoform. Likewise, the molecular mechanisms for rapid metabolic responses to stress remain largely unknown and to fill this gap, the role of PTMs needs to be investigated. As an example, in response to cold stress, which involves quick adjustments to the photosynthetic machinery, many cold-acclimation-related proteins are putatively regulated by PTMs, as has been recently highlighted in pea by using 2D-DIGE analysis (Grimaud et al., 2013).

# **Metabolomics**

In the 'omics field, metabolomics generates large datasets for the identification and quantification of small molecules. Usually, the approach is undertaken for the high throughput detection of secondary (flavonoids, sugar-phosphates, phytohormones, phytoalexins, etc.) and primary metabolites (sugars, organicand amino acids, etc.). Complementary MS-based LC and GC approaches are required to adequately profile this metabolic diversity (Scherling et al., 2010). Although it is possible to gather tens of thousands of metabolic variables, their accurate identification remains the major bottleneck in metabolomics studies to date. Like for proteins, due to the large dynamic range and the difference in physico-chemical properties, detecting the entire "metabolome" is not possible. In contrast to proteomics, for metabolomics analyses, functional identification is not dependent on the availability of genome sequence information. However, the availability of purified standard substances and/or spectral libraries/databases is necessary. Of the various MS methods, two profiling strategies can be typically distinguished: a targeted and a non-targeted approach (Wienkoop et al., 2008, 2010). The non-targeted technique allows for the quantitative evaluation of unknown, yet unidentified, variables. Nevertheless, metabolomics databases and the availability of MS-derived metabolite fragment information are increasing. Furthermore, novel strategies for pathway and structural assignments of untargeted high-throughput metabolomics data are being developed. For example, an algorithm termed mzGroupAnalyzer was developed to investigate metabolite transformations caused by biochemical or chemical modifications, and the approach led to the identification of novel molecule structures (Doerfler et al., 2014). Specifically, it resulted in the detection of 15 unknown, putative cold and light stress regulated metabolites of the flavonoid-pathway in *Arabidopsis thaliana*. Metabolomics is also playing an increased role in stress marker detection and contributing to improved stress tolerance in crops, as previously reviewed (Hirayama and Shinozaki, 2010; Weckwerth, 2011; Martinez-Gomez et al., 2012; Rasmussen et al., 2012). Metabolomics has also contributed to the detection of putative marker(s) induced by the pathogen *Rhizoctonia solani* in potatoes (Aliferis and Jabaji, 2012) and soybean (Aliferis et al., 2014) as well as bacterial blight-resistance in rice (Wu et al., 2012).

# **QTL Analysis**

Most traits of interest for crop breeding are controlled by multiple quantitative trait loci (QTL), and the major objective of using 'omics in this context is the characterization of these QTLs. Analyzing the genetic variations of transcripts, proteins and metabolites at a large-scale allows the search for causal relationships between molecular and phenotypic variations with a global approach and without *a priori* knowledge. The study of natural genetic variations is not only interesting for breeding purposes, but also to decipher the biological processes involved in the genotype to phenotype relationship. Indeed, it is not uncommon that loss of function mutations have no clear phenotype, and the analysis of small disturbances caused by QTLs may allow a better understanding of metabolic pathways and regulatory networks involved in the variations of the phenotypic trait (Sulpice et al., 2010).

Genotyping is the information on which all breeding programs are based to map the QTLs, to measure kinship between genotypes or populations, to analyze changes in allele frequency during selection, etc. In recent years, considerable advances have been made in genotyping and hundreds of thousands of SNP markers (single nucleotide polymorphism) are available today, at relatively low cost for crops whose genome has been sequenced. This information is subsequently used to identify candidate genes or proteins. Overall, the strategy consists of mapping QTLs of transcript expression (eQTLs) or abundance proteins (PQLs or pQTLs), and looking for co-localization between them and QTLs of the traits of interest. The underlying idea is that QTL/(eQTL or PQL) co-localizations may be due to a single polymorphism that causes quantitative or qualitative (amino acid polymorphism) variation of the transcript and/or protein, that in turn would be the cause of the variation of the trait of interest. When these colocations also co-localize with the gene encoding the transcript or protein, then the causal polymorphism is likely located within the gene, including the promoter region (*cis*-QTLs). This gives breeders the opportunity to select the favorable allele itself, which is far more efficient than using QTL flanking markers.

When QTL/(eQTL or PQL) co-localizations are found outside the region of the gene (*trans*-QTLs), the QTL can be any sequence that influences the protein or transcript abundance. In this case, it is not precisely identified; nevertheless the co-localization indicates a possible involvement of the gene/protein in the genetic variation of the phenotypic trait. This information is interesting both for the breeder and for understanding the mechanisms involved in the variation of the trait. As co-localizations can also be found due to chance, candidate genes or proteins are selected according to *a priori* knowledge on gene function or regulation that support the involvement of the gene in the studied trait. After validation, the information brought by these co-localizations can be used in breeding programs by selecting alleles that influence the abundance of the transcript or protein, or by transgenesis to over- or under-express these genes.

As can be deduced from the discussions above, each 'omics approach offers advantages and drawbacks. Transcripts are direct gene products, and thus the link between genomic information and expression data is quite simple. On the other hand, the link between transcript abundance and the phenotype is loose, because of the multiple steps from transcripts to protein, from protein abundance to activity and metabolite concentrations, and from metabolites to cellular, physiological and plant phenotypes. Quite a number of eQTL analyses have been performed in plants, including in crops. For example, Li et al. (2013b) mapped a total of 30,774 eQTLs for 22,242 genes by RNA sequencing a maize population of intermated recombinant lines. Thirty-seven percent were *cis*-eQTLs, while the other 63% were *trans*-eQTLs. The latter were often grouped in hotspots. In many of these hotspots, the effect of alleles from the same parent affected gene expression in the same direction. The genes controlled by hotspots were often enriched in a particular functional category. The last two observations suggested that hotspots contain upstream regulators controlling cellular processes. *Cis*- and *trans*-eQTLs have been observed in various proportions according to the species and the study, and hotspots showing some kind of functional specialization are also often observed (Kliebenstein, 2009). Several eQTL studies have been performed with the aim of identifying candidate genes. For example, Li et al. (2013a) identified in 2013 74 loci highly associated with maize oil concentrations or composition by genome-wide association study (GWAS). The expression of all 41 genes at these loci was controlled by *cis*-eQTLs and for 32 it was correlated to the targeted or related traits. Sequencing five of them in a collection of genotypes allowed the identification of polymorphisms in their UTR or promoter regions, which was likely the cause of the variation of their expression and for the variation of kernel oil content and composition. Most eQTL studies have been carried out by analyzing segregating populations, but when the objective was to target a particular QTL, near isogenic lines (NILs) were also used (Bolon et al., 2010; Kugler et al., 2013). Bulk segregant analysis, where different genotypes are mixed according to the genotype in the QTL region, were also used (Chen et al., 2011).

Metabolite QTLs (mQTLs) have also been mapped both in crops and in model organisms (Kliebenstein, 2009). Metabolites are products of biochemical reactions catalyzed by proteins (enzymes), and as such they are closer to the phenotype than transcripts. On the other hand, the relation to genomic sequences is tenuous, because metabolite amounts may depend on many other biochemical reactions than those directly involved in their synthesis or degradation: many enzymes can be directly or indirectly responsible for the variation of a single metabolite. Many of the mQTL studies or analyses of the natural genetic variability of metabolites were performed with the goal of linking metabolite variations to other biochemical, physiological or phenotypic traits. Kerwin et al. (2011) compared *Arabidopsis* mQTLs to eQTLs of genes and showed a single peak of expression per day. The results, combined with the analysis of mutants, allowed them to conclude that variations in glucosinolate content can influence the internal circadian clock. Sulpice et al. (2013) analyzed the relationships between metabolism and biomass in a panel of 97 *Arabidopsis* accessions and observed that correlationbased networks were very specific to growth conditions. Desnoues et al. (2014) analyzed the correlations between sugars and enzymatic activities in peach fruits in a progeny of 106 genotypes. Interestingly the variations in sugar content was only poorly explained by the variations of enzymatic capacities of the enzymes directly involved in their synthesis or degradation, suggesting that their variations could be related to changes in other components of sugar metabolism.

In several mQTL studies, parallel mapping of eQTL or analysis of gene expression were performed to identify candidate genes. For example, Brotman et al. (2011) identified a gene encoding a putative fumarase in the region of a fumarate mQTL in *Arabidopsis*. They showed that the fumarate content was greatly reduced in mutants and that the expression of the candidate gene was 16 times more highly expressed in the parent that showed the highest level of fumarate. These results suggest that there is a causal relationship between the genetic variation of the expression of this gene at this locus and the natural variation in fumarate content. A similar strategy was followed by Zorrilla-Fontanesi et al. (2012) to identify a candidate gene for the production of mesifurane by the strawberry fruit.

Some proteins are in-between transcripts and metabolites in the genome-to-phenotype relationship, since they are effectors of biological processes and other proteins are real end products influencing the phenotype directly. In proteomics studies, the link between genes and proteins can be ambiguous in particular because of peptides shared between members of gene families. On the other hand, as proteins are the results of transcription, transcript turnover and translatability, post-translational events and protein turnover, their amount represents the integration of many processes that lead to the cellular and *in fine* plant phenotype. Consequently, correlations between protein and transcripts amounts are relatively low (Gygi et al., 1999; Carpentier et al., 2008a; Schwanhausser et al., 2011) which supports the necessity to analyze proteome variations. Wilhelm et al. (2014) have calculated the protein/mRNA ratio for many protein and transcripts in a certain tissues and claim that knowing the translation rate constant, it becomes possible to predict protein abundances with good accuracy from the measured mRNA abundance. This is because the translation rate constant is a dominant factor in determining protein abundance (Wilhelm et al., 2014). Some further confirmatory experiments will need to be undertaken to confirm whether this hypothesis is adequately supported.

The first PQLs were mapped in 1994 (Damerval et al., 1994) and the candidate protein strategy was developed in 1999 (de Vienne et al., 1999). It allowed the identification of the ZmASR1 candidate protein for tolerance to drought, whose effect on yield was then confirmed by over-expression (Virlouvet et al., 2011). The ASR protein was also found to be relevant in other species, e.g., tomato, grape, lily and banana (Maskin et al., 2001; Cakir et al., 2003; Wang et al., 2005; Henry et al., 2011). Segregating populations were also used to map seed and leaf protein PQLs in pea and wheat (Amiour et al., 2003; Bourgeois et al., 2011; Legrand et al., 2013). More often, proteomics has been used to search for candidate proteins in relation to a particular QTL, by using NILs (Hajduch et al., 2007; Torabi et al., 2009; Bernardo et al., 2012; Gunnaiah et al., 2012; Lesage et al., 2012; Li et al., 2014). The analysis of bulked genotypes grouped according to their phenotype can also be useful. It has helped in identifying candidate genes in potato as a complement to a study based on association genetics (Fischer et al., 2013).

To our knowledge, no study on the genetic diversity of PTMs has been performed, although it is well known that protein phosphorylation can play an important role in plant response to biotic and abiotic stresses (Bonhomme et al., 2012; Rampitsch and Bykova, 2012). Although the quantification of PTMs is more difficult than the analysis of protein abundance, and that even identifying a correlation between a modifying enzyme and its target is challenging, it is likely that "candidate PTMs" would be of great interest for QTL characterizations. The analysis of large numbers of genotypes is necessary for PQL mapping or genome wise genetics association studies. The constant progression of the performances of MS-based quantitative proteomics will allow those types of analysis to be performed in the near future, to achieve in plants what has already begun in yeast (Foss et al., 2007; Picotti et al., 2013; Skelly et al., 2013) and in humans (Wu et al., 2013).

# **Data Integration**

Prior to the development of high throughput methods for extensive metabolome, proteome, transcriptome, genome and recently phenome research, scientists dreamed about the integration of different datasets to gain a deeper insight. Data integration, as a tool to enable in-depth insights into processes, can be performed at two levels. On the one hand, one can integrate homogeneous data derived from measurements on the same entity made in the same experimental set-up. On the other hand, meta-analysis can be applied to integrate heterogeneous data derived from different experiments (Dupae et al., 2014). The latter approach poses a challenge as the integration of data performed using different standards and methods are difficult. Thus, we will focus on the integration of homogenous data derived from the same experimental set-up that has been used to get a clearer view on plant (dys)function. Over the past few years, questions on how to unravel the existing links between the metabolome, proteome, transcriptome, genome and the phenome from a particular entity have been discussed. As well as how this knowledge can provide better insights in mechanisms like abiotic stress tolerance or biotic stress resistance. Statistical multivariate methods have been refined to uncover those links and are now used to integrate two or even more datasets based on predefined models of one-by-one dataset relationships (Le Cao et al., 2011; Gunther et al., 2014; Tenenhaus et al., 2014). This not only makes it possible to integrate multiple 'omic levels, but even enables scientists to "supervise" and direct such analyses toward meaningful relationships between the 'omics datasets (Gunther et al., 2014).

A major advantage in integrating metabolomics with proteomics is gaining spatial and temporal information on the end products upon environmental constrains. Since specific protein isoforms can target a specific protein to particular tissues and/or compartment, the use of integrative subcellular fractionation and localization strategies will allow the detection of dynamic distributions within the cell. The temporal plasticity of metabolism constitutes various phases of adjustment. In order to capture the interaction between the metabolome and the proteome, it is necessary to investigate the system along a period of time. Correlative network- as well as Granger-Causality (time-series correlation) analyses have been demonstrated to be effective tools in obtaining information on pathway interplay and reprogramming cues (Doerfler et al., 2014). Linking 'omics data with such mathematical approaches facilitates the interpretation of time dependent chronological processes and the identification of variables being controlled by time-lagged values of other variables.

The next step is to link the transcriptome/proteome/ metabolome to plant performance (phenome), i.e., linking cellular phenotyping to plant phenotyping. The detailed characterisation (physiological, cytological, biochemical, gene expression, protein and metabolite profiles) of different plant species represent a real and difficult challenge. The necessary resources and or knowledge are most likely not present in one research group or country. This is where a European network such as the COST project can make the difference. An integration of these data into regulation/interaction networks will represent new and important advances for understanding plant responses. A multivariate analysis approach becomes of interest because the algorithms are designed to explain as much of the variability at the plant level as well as the variability at the cellular level. This method is especially useful to assess polygenic traits such as drought tolerance, because one can specifically look for cellular and plant level variations that explain the variation in genotypes and treatments used. In this way candidate genes related to important tolerance mechanisms at the plant level can be retained for further analysis, a procedure called feature selection. Retained candidate genes can further be related to the kyoto encyclopedia of genes and genomes (KEGG), gene ontology (GO), protein domain, protein family, protein function, and more categorical databases, as a form of heterogeneous data integration (Gomez-Cabrero et al., 2014). Identification of key genes, pathways, regulation networks of metabolism and stress responses, in association with physiological stress-related data, represent crucial information that have to be integrated and presented in web databases. This way the physiological context of the genes of interest can be compared to the literature to confirm the obtained results.

# **'Omics and Breeding: From Lab to Field**

The genetic gain obtained via breeding programs supports the yield gain of new crop varieties. In this way it is relevant to increase the genetic base available for selection, i.e., the genetic diversity, and also to characterize the molecular and phenotypic consequences of such diversity. Germplasm characterization through 'omics allows one to perform the molecular characterization of genotypes, providing a list of candidate genes/gene products that are highly valuable for breeding and/or engineering stress-tolerant crops with novel traits (Mir et al., 2012; Pinheiro et al., 2014). The characterization of genetic diversity in germplasm collections is typically performed through DNA based markers. However, the use of non-DNA markers (as transcripts, proteins and metabolites) has the advantage to provide information on the molecular interactions and networks operating in a given genotype; these have the potential to help evaluate the potential of the different genotypes. As indicated above, each technique focuses on a subset of the biological interaction network and each technique has its strong and weak points. At present, blind high throughput proteomics and metabolomics studies are often regarded as being descriptive. This might be due to the fact that not many of the putative markers that have been proposed have been evaluated and transferred to the productive sector. The reason might be that high throughput crop proteomics and metabolomics are still emerging 'omic approaches with limited resources and high associated costs. Due to cost, setup, training, requirement for labs /greenhouses etc, most studies are limited in size and number of biological replicates. In many cases these numbers are far too low. Currently the number of studies in crops exploring diversity via different 'omic techniques is limited. Germplasm

and metabolomics) and integrates it to the phenotypic (morphological and

agronomic) data for target traits.

screening and/or discrimination between genotypes of *Phaseolus vulgaris* (Mensack et al., 2010), *Oryza sativa* (Heuberger et al., 2010), *Miscanthus* (Straub et al., 2013), and *Hordeum vulgare* (Heuberger et al., 2014) are some examples. Technologies keep evolving and become more powerful and cheaper. Proteomics and metabolomics are powerful tools when combined with a good experimental design and when the candidates are validated under realistic conditions. It is essential to identify potential traits under lab conditions that are responsible for superior yields under realistic field conditions in different environments.

The current breeder's toolbox makes use of genetic molecular markers, QTLs, gene expression and biochemistry and phenotypic (morphological and agronomic) data for target traits. The use of these tools is dependent on the breeding objective (phenology, yield, yield components, quality, disease, adaptation) and on the crop. While available tools can differ from crop to crop, the crop growing habit (cool vs warm season crops) and the agricultural system also needs to be considered. For each crop, and for each climatic zone, major constraint(s) need to be identified and goals established. The strategy of "one size fits all" is not well suitable, and regional and local goals need to be addressed (climate, soil, social). Another important issue is the availability and the costs of each technology. Langridge and Fleury (2011) present a summary of 'omic resources available for some crops.

# **References**


In the 21st century, the research and agricultural community face a challenge to deliver new stress tolerant and productive crops. Does science and 'omic techniques generate false expectations? No, 'omics technology is powerful and promising when combined with relevant experimental design and subsequently validated in a realistic environment. The use of the 'omics generated knowledge is a promising tool in the breeders tool box (**Figure 1**). A dialog and consultation between breeders, farmers, researchers from natural and social sciences and politicians is required. This is where the COST project "The quest for tolerant varieties—Phenotyping at plant and cellular level" needs to contribute. Through a strong dialog between stakeholders and their cooperative activity it will be possible to deliver better agricultural products that utilize less inputs, have lower environmental costs and provide higher levels of social well-being.

# **Acknowledgments**

This paper has been written in the framework of the COST Action FA1306 "The quest for tolerant varieties—Phenotyping at plant and cellular level." Financial support from the Action to organize meetings and disseminate our viewpoint is gratefully acknowledged.


sequencing. *Anal. Bioanal. Chem.* 400, 1967–1978. doi: 10.1007/s00216- 011-4948-9


infrared spectroscopy study. *Plant Physiol.* 130, 1032–1042. doi: 10.1104/ pp.004325


**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.

*Copyright © 2015 Zivy, Wienkoop, Renaut, Pinheiro, Goulas and Carpentier. 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.*

# Proteomic analyses reveal differences in cold acclimation mechanisms in freezing-tolerant and freezing-sensitive cultivars of alfalfa

## *Jing Chen1, Guiqing Han1,2 \*, Chen Shang2 , Jikai Li <sup>2</sup> , Hailing Zhang2 , Fengqi Liu2 , Jianli Wang2 , Huiying Liu2 and Yuexue Zhang2*

<sup>1</sup> College of Life Sciences and Technology, Harbin Normal University, Harbin, China

<sup>2</sup> Institute of Grass Research, Heilongjiang Academy of Agricultural Sciences, Harbin, China

### *Edited by:*

Silvia Mazzuca, Università della Calabria, Italy

### *Reviewed by:*

Jill Christine Preston, University of Vermont, USA Pingfang Yang, Wuhan Botanical Garden – Chinese Academy of Sciences, China

### *\*Correspondence:*

Guiqing Han, College of Life Sciences and Technology, Harbin Normal University, South of Shida Road No. 1, Harbin 150050, China e-mail: ccyj15@163.com

Cold acclimation in alfalfa (Medicago sativa L.) plays a crucial role in cold tolerance to harsh winters. To examine the cold acclimation mechanisms in freezing-tolerant alfalfa (ZD) and freezing-sensitive alfalfa (W5), holoproteins, and low-abundance proteins (after the removal of RuBisCO) from leaves were extracted to analyze differences at the protein level. A total of 84 spots were selected, and 67 spots were identified. Of these, the abundance of 49 spots and 24 spots in ZD and W5, respectively, were altered during adaptation to chilling stress. Proteomic results revealed that proteins involved in photosynthesis, protein metabolism, energy metabolism, stress and redox and other proteins were mobilized in adaptation to chilling stress. In ZD, a greater number of changes were observed in proteins, and autologous metabolism and biosynthesis were slowed in response to chilling stress, thereby reducing consumption, allowing for homeostasis. The capability for protein folding and protein biosynthesis in W5 was enhanced, which allows protection against chilling stress. The ability to perceive low temperatures was more sensitive in freezing-tolerant alfalfa compared to freezing-sensitive alfalfa. This proteomics study provides new insights into the cold acclimation mechanism in alfalfa.

**Keywords: proteomics, cold acclimation, RuBisCO, metabolism, homeostasis, alfalfa**

### **INTRODUCTION**

A lack of cold tolerance is a major limiting factor for crop production, especially in northern areas. Plants exposed to low but non-freezing temperatures have developed strategies for adapting to stress and enhance cold tolerance via many complex physiological and biochemical changes. Such adaptations are primarily attributed to changes in the expression of functional and regulatory genes (Gomat et al., 2011). Numerous studies have demonstrated that cold acclimation is associated with structural and compositional modifications of compatible solutes in various subcellular compartments and changes in the transcriptome and metabolome (Knaupp et al., 2011; Zuther et al., 2012; Vaclavik et al., 2013). Although cold acclimation mechanisms have been studied for many years, the molecular and genetic foundations of this adaptation in plants remain largely unknown.

Proteomics bridges the gap between gene expression and metabolism (Heazlewood, 2011; Carrol et al., 2013), promoting the study of adaptive mechanisms to chilling stress in many plants (Cui et al., 2005; Amme et al., 2006; Yan et al., 2006; Hashimoto and Komatsu, 2007; Degand et al., 2009; Gao et al., 2009; Lee et al., 2009; Balbuena et al., 2011; Rinalducci et al., 2011; Sánchez-Bel et al., 2012a,b; Uváˇcková et al., 2012; Takahashi et al., 2013). Changes in proteins play a vital role in cold adaptation due to their direct action on metabolism and biosynthesis pathways. Comparative proteomics were recently applied to analyze changes in cold-sensitive proteins in different cold-tolerant cultivars, such as

meadow fescue (leaf), pea (leaf and chloroplast), perennial ryegrass (leaf), strawberry (crown), and winter wheat (leaf; Kosmala et al., 2009; Bocian et al., 2011; Dumont et al., 2011; Koehler et al., 2012; Grimaud et al., 2013; Xu et al., 2013). Proteins involved in energy metabolism, photosynthesis, reactive oxygen species (ROS) scavenging, storage, protection from stress, regulation of the cell cycle and plant development in wheat and barley showed differential abundance between stress-tolerant and stresssensitive genotypes (Kosová et al., 2014). Proteomics can provide new insights into cold acclimation and improve our understanding of the genetic differences underlying cold tolerance in plants.

High-abundance proteins, such as ribulose-1, 5-bisphosphate carboxylase/oxygenase (RuBisCO), can mask low-abundance proteins of interest. High-abundance proteins reduce the dynamic resolution of proteins and affect the analysis of some functional proteins in proteomic studies. In protein extracts from *Arabidopsis* and rice leaf, ∼12 and 35.3% of the spots, respectively, have been identified as RuBisCO and its derivatives (Giavalisco et al., 2005; Yan et al., 2006). RuBisCO accounts for a high percentage in abundance in green leaf tissue due to protein load capacity limitations (Xi et al., 2006); co-migration with RuBisCO also masks neighboring species (Corthalis et al., 2000; Krishnan and Natarajan, 2009). To overcome this problem, a polyethylene glycol (PEG) mediated pre-fractionation method was used to remove RuBisCO; this protocol depletes most RuBisCO and their derivatives (Kim et al., 2001).

As leguminous forage, alfalfa contributes to biological nitrogen fixation and is widely planted throughout the world. In northern climates, fall dormant cultivars have a greater capability to resist harsh winters than non-dormant varieties (Timmons and Salmon, 1932; Sprague and Fuelleman, 1941; Smith et al., 1979). It has been established that freezing-tolerant and freezingsensitive cultivars of alfalfa share similar cold regulated (COR) gene complements; differences exist in the rate and extent of the expression of these genes in response to cold. COR genes are induced by low temperature and their translation products can mechanistically protect plants against environmental stresses (Seki et al., 2004). Marked differences were found between propagated clones from genotypes by comparing their gene products, which confirmed that variations in molecular changes that occur at low temperature are under some level of genetic control (Castonguay et al., 2006). Castonguay et al. (2006) hypothesized that some of COR genes fail to be expressed in a timely manner in response to environmental cues in freezing-sensitive alfalfa. The ability to perceive and transduce external signals into a series of molecular events leading to physiological responses differs widely among genotypes within a species (Kaur and Gupta, 2005). However, differences in the cold acclimation mechanisms between freezing-tolerant and freezing-sensitive alfalfa remain unknown. In our study, holoproteins and low-abundance proteins were extracted from alfalfa, and changes in protein categories and relative abundances during chilling were analyzed. Differences in the cold acclimation mechanisms in alfalfa would be first discussed at the protein level using proteomics in our study.

### **MATERIALS AND METHODS**

### **PLANT MATERIAL AND CULTURE CONDITIONS**

Two alfalfa cultivars with different freezing tolerances were studied: freezing-tolerant cultivar ZhangDong (ZD), which is fall dormant alfalfa (winter resistant cultivar), and freezing-sensitive cultivar WL525HQ (W5), which is non-dormant alfalfa (winter irresistant cultivar). ZD andW5 seeds were germinated under controlled environmental conditions as follows: 280 μmol/m2/s; 16 h light/8 h dark; and a day/night temperature regime of 25/20◦C. For the cold treatment, 50 days old seedlings were transferred to a cold chamber set to 4◦C for 7 days under other conditions as described above.

### **PROTEIN PREPARATION**

Leaves (1 g) from both cultivars were sampled at 0 (control), 12 h, and 7 days under cold treatment and were immediately frozen in liquid nitrogen.

### *Holoproteins extraction*

For holoproteins extraction, frozen samples were pulverized in a pre-cooled mortar with liquid nitrogen. Samples were then suspended in three volumes of pre-cooled 10% (w/v) trichloroacetic acid (TCA)/acetone with 0.07% (v/v) β-mercaptoethanol and kept at –20◦C overnight. The homogenate was centrifuged at 40,000 × *g* for 1 h at 4◦C. The supernatant was removed, and the precipitate was resuspended in three volumes of cold acetone containing 0.07% (v/v) β-mercaptoethanol. The mixture was incubated at –20◦C for at least 1 h and centrifuged at 40,000 × *g* for 1 h at 4◦C. The aforementioned steps were repeated until the supernatant was colorless. The vacuum-dried pellet was dissolved in 2 ml of a lysis solution containing 7 M urea, 2 M thiourea, 4% (w/v) 3-[(3-Cholamidopropyl) dimethylammonio]- 1-propanesulfonate (CHAPS), 40 mM dithiothreitol (DTT), and 1% (v/v) protease inhibitors (PMSF). The sample was shaken at 4◦C for more than 1 h. The mixture was centrifuged at 100,000 × *g* for 1 h at 4◦C to remove any solids. The protein concentration was quantified using a 2D-Quant kit, and the protein solution was stored at –80◦C.

### *Low-abundance proteins extraction*

Low-abundance proteins were extracted with Mg/NP-40 buffer and fractionated by PEG 4000 (PEG4000), as described by Kim et al. (2001) with slight modifications. Frozen leaves were ground in a pre-cooled mortar with liquid nitrogen. Five milliliters Mg/NP-40 buffer containing 0.5 M Tris-HCl (pH 8.3), 2% (v/v) NP-40, 20 mM MgCl2, 2% (v/v) β-mercaptoethanol, 1 mM PMSF, 1 mM ethylene diamine tetraacetic acid (EDTA) and 1% (w/v) polyvinylpolypyrrolidone (PVPP) was added, and the sample was ground on ice for 15 min. After centrifugation at 12,000 × *g* for 15 min at 4◦C, a 50% (w/v) PEG4000 stock solution was added to the supernatant to a final PEG concentration of 17.5% (w/v). The mixture was incubated on ice for 30 min and centrifuged at 12,000 × *g* for 15 min at 4◦C. Three volumes of pre-cooled 10% (w/v) TCA/acetone was added to the supernatant and kept at –20◦C overnight. The remaining steps were performed according to the holoproteins extraction method described above.

The quantified holoprotein and low-abundance protein samples were loaded into an SDS-PAGE gel [12.5% (w/v)] and run for ∼2.5 h. The gel was used to determine the quality of proteins and remove RuBisCO.

### **TWO-DIMENSIONAL GEL ELECTROPHORESIS**

Quantified protein solutions were diluted in rehydration buffer containing 8 M urea, 2% (w/v) CHAPS, 0.3% (w/v) DTT, and 1% (v/v) IPG buffer. IEF strips (24 cm, linear pH 4–7) were rehydrated in 450 μl of rehydration protein solution at 20◦C for 12 h in an Ettan IPGphor 3 electrophoresis system (GE Healthcare). IEF was run using the following parameters: 200 V for 2 h, 500 V for 2 h, 1000 V for 2 h, 8000 V for 3 h, and 8000 V for 65000 VH. After IEF, the strips were equilibrated twice in equilibration buffer [6 M urea, 1.5 M Tris-HCl (pH 8.8), 30% (v/v) glycerol, and 2% (w/v) SDS] with 1% (w/v) DTT for 15 min and then with 4% (w/v) iodoacetamide for 15 min. For twodimensional electrophoresis, the strips were placed on a 12.5% (w/v) SDS-PAGE gel and run at 1 w per gel for 1 h, followed by 13 w per gel for 4.5–5 h. An Ettan DALTSix system (GE Healthcare) was used for two-dimensional electrophoresis. Subsequently, the gels were fixed in 40% (v/v) alcohol and 10% (v/v) acetic acid for 30 min, swollen in 10% (v/v) acetic acid for 20 min, and stained with Coomassie brilliant blue G-250. The gels were rinsed with 10% (v/v) acetic acid until the protein spots were distinct from the background. The gels were then stored in deionized water.

### **IMAGE ACQUISITION, DATA ANALYSIS, AND PROTEIN IDENTIFICATION**

Gels were scanned using an ImageScanner III (GE Healthcare), and images were analyzed using ImageMaster 2D Platinum v7.0 software (GE Healthcare). Gels of three independent biological replicates per treatment were analyzed. Spots were automatically detected and matched, and mismatched and unmatched spots were artificially modified through manual editing. The spot intensities were normalized according to total intensity of valid spots to reduce the differences in the protein loading and gel staining. Tukey's test (*P* < 0.05) was applied to test the abundance change of the spots. Only spots with volume ratios over 1.5-fold (*P* < 0.05) were selected for MS identification.

Each marked protein was cut from the gel and cleaved with trypsin. Peptide identification was performed using a 5800 MALDI-TOF/TOF mass spectrometer (AB SCIEX) according to the protocol described by Kosová et al. (2013). The obtained peak list was used to search the databases NCBI *Medicago* (3967), downloaded on June 12, 2014, NCBI Viridiplantae (973373), downloaded on September 13, 2013, and Uniprot (540732), downloaded on September 3, 2013, using MASCOT V2.2 software. Database searches were conducted using the following parameters: peptide mass tolerance ± 100 ppm; fragment mass tolerance ± 0.4 Da; a maximum of one missed cleavage; cysteine carbamidomethylation allowed as a fixed modification; and oxidation of methionine allowed as a dynamical modification. Only significant hits, as defined by the MASCOT probability analysis (*P* < 0.05) with a protein score CI % greater than 95 and a protein score above 50, were accepted. A functional classification of proteins was performed based on the Gene Ontology database<sup>1</sup> and the Uniprot database2.

### **STATISTIC ANALYSIS**

The data shown in image analyses (holoproteins and lowabundance) and the differential abundance of proteins in the three samples under evaluation (control, 12 h and 7 days) were both compared by analysis of variance (ANOVA) followed by Tukey's multiple comparison test (*P* < 0.05). Cluster analysis was used to show identified proteins visually changed in the relative abundance using the software PermutMatrix v1.9.3 (Caraux and Pinloche, 2005). The average values of three biological replicates were used to compare the protein changes among different treatments.

### **RESULTS**

### **IMAGE ANALYSES OF HOLOPROTEINS AND LOW-ABUNDANCE PROTEINS OF alfalfa**

In our study, both holoproteins and low-abundance proteins of alfalfa were extracted to analyze protein changes during chilling stress. The SDS-PAGE result (**Figure 1**) revealed that 17.5% PEG could remove most RuBisCO large subunits; new electrophoretic bands containing low-abundance proteins were clearly detected. Image analysis revealed approximately 498 ZD holoprotein spots and 516 W5 holoprotein spots (**Figure 2A**). Approximately 423 spots were detected in the ZD low-abundance proteome, of which 308 spots were unique to low-abundance versus holoproteins; the

**FIGURE 1 | SDS-PAGE of contrast of holoproteins and low-abundant proteins. (A)** Are holoproteins extracted by TCA/acetone. **(B)** Are low-abundant proteins extracted by PEG. 12.5% SDS-PAGE gel showing RuBisCO depletion. RuBisCO large subunit (LSU) and small subunit (SSU) are marked.

ratio of low-abundance protein spots to holoprotein spots was ∼61.85%. In the W5 proteome, this ratio was 64.13%; approximately 320 of 499 spots were unique to low-abundance versus holoproteins (**Figure 2B**). PEG fractionation was useful for the resolution of low-abundance proteins. The detection of new lowabundance proteins aided our analysis of protein variations and adaptive mechanisms during chilling.

### **PROTEOMIC DIFFERENCES IN alfalfa IN RESPONSE TO LOW TEMPERATURE**

To detect differences in cold acclimation between freezing-tolerant alfalfa and freezing-sensitive alfalfa, the variety and relative abundance of proteins were analyzed in two alfalfa cultivars (ZD and W5) in three different growth conditions (control, 4◦C for 12 h, and 4 ◦(C for 7days). Holoproteins and low-abundance proteins were extracted from leaves. Two-dimensional gel electrophoresis (2-DE) revealed that a total of 84 spots were differentially expressed at ratios over 1.5-fold in relation to cold acclimation. Of these, 67 spots were successfully identified by MALDI\_TOF/TOF (**Table 1**).

<sup>1</sup>www.geneontology.org

<sup>2</sup>www.uniprot.org


**Table 1 |The list of 67 indentified**

 **spots by** 

**MALDI-TOF/TOF**

 **analysis.**





**Table 1 |**

**Continued**


**www.frontiersin.org** February 2015


**Frontiers in Plant Science** | Plant Proteomics February 2015

**Table 1 |**

**Continued** Clustering analysis (**Figure 3**) was used to visually describe changes of 67 spots in relative abundance during chilling using Permut-Matrix software v1.9.3. Compared with W5, more spots were identified in ZD as responding to low temperature: the relative abundance of 49 spots changed in ZD, whereas 24 spots changed in W5. Of these, 15 spots changed in both cultivars in response to low temperature. Compared with controls, chilling (12 h) caused a significant up-accumulation of 13 spots in ZD and six spots in W5, and 34 spots were down-accumulated in ZD (spots 17 and 28 unchanged) and 14 spots in W5 (spots 17, 28, L55 and L56 unchanged). Compared with chilling for 12 h, chilling for 7 days caused an up-accumulation of seven spots in ZD (spots 4, 18, 26, 33, L25, L28, and L46) and five spots (spots 4, 18, 26, L46 and L68), 13 spots (spots 17, 28, L3∼L4, L12∼L15, L19, L29, L33, and L55∼L56) in ZD and eight spots (spots 48∼51, L43 and L48∼L49) in W5 were down-regulated. In additional, nine unchanged spots were found in W5 with a greater relative abundance than in ZD. Chilling affected the relative abundance and variety of proteins. The number of changed proteins were more at 4◦C for 12 h than for 7 days.

Protein functional analysis has been carried out according to Gene Ontology database and the Uniprot database. Based on their functional features, the 67 differentially expressed proteins were classified into five categories, as follows: photosynthesis, protein metabolism, energy metabolism, stress and redox, and other functions (**Figure 4** and **Table 1**). Among these protein spots, 11 changed in a similar manner in both ZD and W5 under cold stress. Compared with control, the relative abundance of uncharacterized protein LOC101499502 (spot 11), unnamed protein product (spot 12), and RuBisCO activase (spots 22, 23, and 25) declined at 12 h and that of RuBisCO activase (spots 17 and 28) declined at 7 days, the relative abundance of RuBisCO small subunit (spot 4), RuBisCO activase (spot 18) and eukaryotic translation initiation factor 5A-2 (eIF-5A, spot L46) continuously increased at 12 h and 7 days, and glutamate 1-semialdehyde aminotransferase (spot 26) was down-regulated at 12 h and up-regulated at 7 days. The expression patterns of RuBisCO large subunit-binding protein subunit beta (spot 33), 5-methyltetrahydropteroyltriglutamatehomocysteinemethyltransferase (spots L55 and L56), and oxygenevolving enhancer protein (spot L68) were altered in different ways in ZD and W5 (**Figure 3**). During chilling, the relative abundance of the remaining 32 spots changed in ZD and remained constant in W5. Of these, RuBisCO large subunit (spots 5 and 9), RuBisCO small subunit (spots 6 and 8) and oxygen-evolving enhancer protein (spots L64∼L66) were up-regulated at 12 h, and triosephosphate isomerase (spots L3 and L4), peptide methionine sulfoxide reductase (spots L12 and L13), thioredoxin-like protein (spots L14 and L15), cinnamoyl-CoA reductase (spot L19) and *S*-adenosyl-L-methionine synthetase (spot L33) were continuously down-regulated at 12 h and 7 days. Monodehydroascorbate reductase (spots 27 and L32), GTPase obg (spot L5), RuBisCO small subunit (spot L6), cytochrome b6-f complex iron-sulfur subunit (spot L8), glutathione peroxidase (spot L11), malate dehydrogenase precursor (spot L20), aldo-keto reductase (spot L21), RuBisCO activase (spot L23), phosphoribulokinase (spot L24), sedoheptulose-1, 7-bisphosphatase (spot L27), glutamate 1-semialdehyde aminotransferase (spot L30), 1-deoxy-D-xylulose

reveals changes of the relative abundance of identified protein by software PermutMatrix v1.9. Normalized rows (Z scores) and Ward's algorithm have been used to analyze data. In the figure, ZC is control of ZD, Z12 is treatment of ZD at 4◦C for 12 h, Z7 is treatment of ZD at 4◦C for 7 days, WC is control of W5, W12 is treatment of W5 at 4◦C for 12 h, and W7 is treatment of W5 at 4◦C for 7 days. Digits in the figure represent spots No.

5-phosphate reductoisomerase (spot L31) and aconitate hydratase 2 (spots L41 and L42) maintained the same lower relative abundance at 12 h and 7 days compared with the control. The relative abundance of adenosine kinase 2 (spot L25) and unnamed protein product (spot L28) decreased at 12 h and increased at 7 days. Two spots were expressed exclusively in ZD: the relative (three spots).

abundance of glutamate 1-semialdehyde aminotransferase (spot L29) decreased and that of oxygen-evolving enhancer protein (spot L67) increased. In W5, only nine proteins spots were differentially expressed and remained unchanged in ZD. RuBisCO large subunit (spot L43) and peptidyl-prolyl *cis*-*trans* isomerase (PPlase, spot L44) were up-regulated, transketolase (spots 48∼51) and 2- Cys peroxiredoxin (spots L48 and 49) were down-regulated at 12 h and 7 days, and RuBisCO large subunit (spot 64) was only downregulated at 12 h. In additionally, RuBisCO small and large subunit (spots 34, 36, 39, and 41), chlorophyll A/B binding protein (spots 42 and 43) and chaperone protein ClpC (spots 45∼47) were only highly expressed in W5 under chilling stress.

Cold acclimation in plant is a complex progress and involved in various proteins. In our study, photosynthesis related proteins were the largest category of differentially expressed in both ZD and W5. A series of changes between RuBisCO activase (spots 17, 18, 22, 23, 25, and L23) and RuBisCO large and small subunits (spots 4, 5, 6, 8, 9, 64, L6, and L43) were found in ZD and W5 in response to low temperature. Compared to freezing-sensitive alfalfa, a greater number of proteins were changed in freezing-tolerant alfalfa under chilling stress. In ZD, protein metabolism, energy metabolism and stress and redox related proteins were influenced, such as *S*-adenosyl-L-methionine synthetase (spot L33), GTPase obg (spot L5), glutathione peroxidase (spot L11), malate dehydrogenase precursor (spot L20), and so on. Autologous metabolism and biosynthesis were slowed to response low temperature in ZD. Increased PPIase and eIF-5A is consistent with an increased capability for protein folding and protein biosynthesis, suggesting increased protection against chilling stress in W5.

### **DISCUSSION**

### **THE REMOVAL OF "OVERABUNDANT" PROTEINS**

RuBisCO exists in most green plants and comprises 30∼60% of the total protein. RuBisCO is an important photosynthesis enzyme that fixes CO2 in the Calvin cycle (Ellis, 1979; Voet and Voet, 1995; Whitney and Andrews, 2001; Parry et al., 2003; Giavalisco et al., 2005). As an "overabundant" protein, RuBisCO influences the detection of some low-abundance proteins. PEG, at a proper concentration, effectively removes most large subunits of RuBisCO and improves the dynamic resolution of low-abundance proteins in different plant species. In *Arabidopsis thaliana*, 16% PEG was applied to extract low-abundance proteins using a fractionation method; 80% more spots were revealed compared to a holoprotein sample extracted using a TCA/acetone method (Xi et al.,2006). Lee et al. (2007)suggested that 15% PEG could be used for the effective extraction of low-abundance protein to analyze protein changes in rice leaves under cold stress. In our study, 17.5% PEG significantly reduced the masking effect of RuBisCO, and new proteins were detected in both ZD and W5. Thirty-five out of 67 spots were low-abundance proteins in response to chilling stress (**Table 1**). These data could help us to better understand cold adaptation mechanisms in alfalfa.

### **PROTEINS CHANGES IN PROTEOME** *Differentially expressed photosynthetic proteins*

Proteomic analysis revealed that different cold acclimation mechanisms exist in ZD and W5. In some cases, the same protein exhibited differential accumulation patterns in alfalfa cultivars with different tolerance to cold stress. In our study, the largest category of proteins that responded to chilling in both ZD and W5 is photosynthesis-related proteins. It has been reported the interaction between RuBisCO activase and RuBisCO large and small subunits (Portis, 1990; Spreitzer and Salvucci, 2002). When *RdreB1BI* transgenic strawberries are exposed to low temperature, RuBisCO activase reactivates RuBisCO to fix any remaining CO2 and protect the dissipation of energy from the photo respiratory oxygenase reaction (Haupt-Herting et al., 2001; Gua et al., 2013). In alfalfa, RuBisCO activase (spot 18) increased in both ZD and W5; the expression of this protein was higher in ZD than that inW5 at 4◦C for 7 days. A greater number of up-regulation of spots corresponding to the large and small subunits of RuBisCO (spots 4, 5, 6, 8, and 9) were found in ZD at 12 h, compared toW5 (spots 4 and L43). The abundance of the RuBisCO large subunit-binding protein subunit β (spot 33) decreased for the correct assembly of the RuBisCO holoenzyme in W5; a similar result was observed in pea (Barraclough and Ellis, 1980; Grimaud et al., 2013). Chilling stress also affected the relative abundance of RuBisCO activase (spots 17, 22, 23, 25, and L23). The relative abundance of the RuBisCO small subunit (spot L6) decreased in ZD, and RuBisCO large subunit (spot 64) decreased in W5. The disassembly of RuBisCO may reduce the photosynthetic rate under cold stress (Yan et al., 2006; An et al., 2011; Zhang et al., 2012). The series of changes between RuBisCO activase and RuBisCO large and small subunits could make the active site more accessible for carbamylation, thereby enhancing CO2 fixation to resist cold stress (Portis et al., 2008). The relative abundance of the cytochrome b6-f complex iron-sulfur subunit (spot L8) declined in ZD as in *Thellungiella* rosette and in a cold-tolerant Champagne cultivar of pea, which affected electron transport from PSII to PSI during cold acclimation (Gao et al., 2009; Grimaud et al., 2013); whereas its expression remained unchanged in W5. Oxygen-evolving enhancer protein 1 (spots L64, L65, L67, and L68) increased in ZD, which stabilizes

the tetranuclear manganese (Mn) center that is the location of photoinhibition in PSII (Sarvikas et al., 2006). In addition to L68, abundance of the other three oxygen-tetranuclear Mn center proteins (L64, L65, and L67) was unchanged in W5. As in a freezetolerant genotype of *Festuca pratensis*, oxygen-evolving enhancer protein 1 (L68) in W5 was degraded at 12 h and accumulated at 7 days. Kosmala et al. (2009) postulated that this change may contribute to additional metabolic disturbances rather than lower susceptibility to photoinhibition.

Compared to W5, more proteins were mobilized in ZD during adaptation to low-temperature stress. The relative abundance of photosynthesis-related proteins in ZD was altered in a complex manner to relieve the influence of chilling stress on photosynthesis. However, the abundance of these proteins remained relatively stable in W5. The RuBisCO small subunit (spot 34) and large subunit (spots 36, 39, and 41) were only highly expressed in W5. The greater relative abundance of chlorophyll A/B binding protein (spots 42 and 43), a component of the light-harvesting complex, allows for the absorption of more light by chlorophyll excitation, and the transfer of energy to the photochemical reaction centers allows for adaptation to chilling stress in W5 (Green and Durnford, 1996).

### *Differentially expressed metabolism proteins*

The category of protein metabolism involved proteins related folding and disassembling, biosynthesis and amino acid metabolism. Chilling induced eIF-5A (spot L 46) accumulation in both ZD and W5 as in wheat and *Thellungiella* rosette (Gao et al., 2009; Vítamìvasì et al., 2012; Kosová et al., 2013). EIF-5A is mainly involved in RNA metabolism, protein translation and the regulation of the cell cycle (Schatz et al., 1998; Thompson et al., 2004; Jao and Chen, 2006; Feng et al., 2007). Kosová et al. (2013) indicated that these changes were associated with protein expression and protein regulation and reflect more profound changes at the regulatory level in cold-treated plants. Cold affected the *S*-adenosylmethionine (SAM) synthetic pathway in both ZD and W5. In this pathway, methionine production is catalyzed by methionine synthase with assistance of the coenzyme 5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase (spots L55 and L56). *S*-adenosyl-L-methionine synthetase (spots 28 and L33) then transfers adenosine to catalyze SAM. Proteomic studies have demonstrated that SAM plays an important role in cold stress resistance (Cui et al., 2005; Kosová et al., 2013). The observed reduction in the relative abundance of spots L55 and L56 began at 12 h in ZD and at 7 days in W5. These findings suggest that cold stress affects the activity of 5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase, as reported in rice seedlings (Hashimoto and Komatsu, 2007). *S*-adenosyl-L-methionine synthetase was consumed to produce SAM, preventing cold injury in alfalfa, although the reduction in the coenzyme led to an insufficient supply of substrate. Chilling had a clear influence on amino acid metabolism in alfalfa, leading to a reduction in amino acid synthesis for energy conservation during low temperature adaptation.

GTPase obg (spot L5) is a conserved GTPase binding protein mainly involved in cell proliferation, cell development, signal transduction, and protein translation (Bourne et al., 1990; Kaziro et al., 1991). Through an analysis of *obgc* mutants in *Arabidopsis* and rice, Bang et al. (2012) found that plastid rRNA processing is defective, indicating that *obgc* functions primarily in plastid ribosome biogenesis during chloroplast development. In ZD, it was speculated that the decreased relative abundance of GTPase obg during chilling might impede the synthesis of a series of related proteins. Biosynthesis-related enzymes were also significantly influenced by cold stress. Glutamate 1-semialdehyde aminotransferase (spots 26, L29, and L30) catalyzes the conversion of glutamate-1-semialdehyde to aminolevulinic acid, which is a step in the assembly of chlorophyll, coenzyme B12, heme, and other tetrapyrrolic proteins (Jordan and Shemin, 1972; Gough et al., 2001). Cinnamoyl-CoA reductase (CCR, spot L19) catalyzes the first step in monolignol biosynthesis and plays a key role in synthesis of lignin (Zhou et al., 2010). The protein 1-deoxy-D-xylulose 5-phosphate reductoisomerase (spot L31) is a key rate-limiting enzyme and the regulatory site of terpenoid synthesis (Takahashi et al., 1998). The relative abundance of these spots decreased in ZD, and with the exception of spot 26, the expression of these spots remained unchanged in W5. Chilling stress was clearly more severe in ZD than in W5, as three biosynthetic pathways were disrupted in ZD.

In W5, the expression of PPlase (spot L44) increased under cold stress. This is similar to a finding in *Populus cathayana* males (Zhang et al., 2012), indicating that these proteins are involved in protein folding to overcome stress (Budiman et al., 2011). Additionally, high expression of the chaperone protein ClpC (spots 45, 46, and 47) was only observed in W5, which regulated protein metabolism and kept homeostasis. As a protein chaperone, ClpC is mainly involved in regulating the structure and function of many polypeptides and hydrolyzes irreversibly damaged proteins to prevent the accumulation of potentially cytotoxic polypeptides (Parsell and Lindquist, 1993).

### *Differentially expressed energy metabolism proteins*

The third functional category of proteins that respond to low temperatures in alfalfa involve energy metabolism such as the Calvin cycle, Kreb's cycle, and glycolysis pathways. Chilling had a large effect on carbon metabolism (phosphoribulokinase, spot L24, and sedoheptulose-1, 7-bisphosphatase, spot L27), ATP production (malate dehydrogenase precursor, spot L20, and aconitate hydratase 2, spots L41 and L42) and glycolysis (triosephosphate isomerase, spots L3 and L4) in ZD. Transketolase (spots 48∼51), which is involved in the Calvin cycle were down-regulated in W5. Unlike cold-tolerant pea, transketolase was more highly expressed in W5 compared to ZD, whereas this protein was constitutively expressed in ZD under cold stress (Grimaud et al., 2013). In addition, a reduction in the abundance of adenosine kinase 2 (spot L25) was shown to disturb the regulation of energy metabolism and the balance of ATP, ADT, and AMP at the beginning of chilling in ZD (Veuthey and Stucki, 1987; Zeleznikar et al., 1995). The increase in adenosine kinase 2 expression allowed for recovered activity and the maintenance of the energy balance in the cell at 7 days. We deduced that a decrease in the abundance of several enzymes leads to slowed energy metabolism as energy is stored for maintaining body balance in alfalfa.

### *Differentially expressed stress and redox proteins*

In plants, cellular redox homeostasis can be disturbed by cold stress. Such disturbances result in the production of ROS, which stimulate oxidative damage in the organism. Therefore, certain ROS-scavenging enzymes play an important role in the scavenging of ROS and maintaining iron balance. Glutathione peroxidase (spot L11; Liu et al., 2014), thioredoxin-like protein (spots L14 and L15; Zhang et al., 2012; Kosová et al., 2013), probable aldoketo reductase 2 (spot L21; Gao et al., 2009; Koehler et al., 2012) and 2-Cys peroxiredoxin (spots L48, L49, and L66; Yan et al., 2006; Gao et al., 2009) were up-regulated under cold stress. Their functions are mainly involved in detoxication, protection from oxidative damage, preventing the membrane lipid from peroxidation and enhancing tolerance to ROS (Bartels, 2001; Kim et al., 2011). Monodehydroascorbate reductase (spot 27 and L32) is an important enzyme involved in the regeneration of ascorbic acid for scavenging hydrogen peroxide (Arrigoni et al., 1981), and peptide methionine sulfoxide reductase A3 (spots L12 and L13) may repair oxidatively damaged proteins *in vivo* (Brot et al., 1982a,b). To resist cold stress, ZD consumes abundant ROSscavenging enzymes to maintain cellular redox homeostasis and protect the body. The relative abundance of 2-Cys peroxiredoxin (spot L66) increased; this protein catalyzes the reduction of various hydroperoxide to the corresponding alcohol or water and detoxifies alkyl hydroperoxides and peroxinitrite (Dietz et al., 2006; Kim et al., 2011). The relative abundance of 2-Cys peroxiredoxin (spots L48 and L49) declined in W5, whereas other ROS-scavenging enzymes were not significantly affected by low temperature. It may be inferred that ZD is more sensitive to chilling stress than W5 in spite of its freeze-tolerant genotype. When chilling (4◦C) occurs over a short period (such as 12 h), ZD consumes enzymes to resist stress. The enzyme expression profile then changed to maintain homeostasis, as indicated by a low relative abundance observed at 7 days. The observed response to chilling in ZD was more drastic and complex than in W5.

As a freezing-tolerant genotype, ZD has stronger cold tolerance than the freezing-sensitive cultivar W5. However, ZD is more sensitive to chilling than W5, and a greater number of changes were observed in proteins involved in photosynthesis, energy metabolism, stress and redox, biosynthesis metabolism and amino acid metabolism in ZD under chilling stress. On one hand, more enzymes were consumed to produce proteins for regulating metabolism and maintaining homeostasis in ZD. On the other hand, low temperature influenced enzymatic activity and altered the metabolic and synthesis pathways. In W5, the expression of many proteins remained unchanged. However, protein metabolism-related proteins were more active in W5 compared to ZD. In conclusion, ZD mobilizes a large number of proteins to adapt low temperature, and autologous metabolism and biosynthesis are slowed to reduce consumption for homeostasis. W5 enhances its capability for protein folding and protein biosynthesis to overcome chilling stress. The perception of low temperature is more sensitive in freezing-tolerant alfalfa than in freezing-sensitive alfalfa. Proteomics provides new insight into cold acclimation mechanisms in alfalfa.

### **ACKNOWLEDGMENTS**

The research was supported by special postdoctoral funding from Heilongjiang (LBH-TZ1209), the National Natural Science Foundation of Heilongjiang (C201313), and the 12th 5-year-plan scheme (2011BAD17B04-2).

### **REFERENCES**


Voet, D., and Voet, J. G. (1995). *Biochemistry.* New York: John Wiley & Sons, Inc.


electrophoresis analysis of plant proteome. *Phytochemistry* 67, 2341–2348. doi: 10.1016/j.phytochem.2006.08.005


**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 December 2014; paper pending published: 16 January 2015; accepted: 09 February 2015; published online: 27 February 2015.*

*Citation: Chen J, Han G, Shang C, Li J, Zhang H, Liu F, Wang J, Liu H and Zhang Y (2015) Proteomic analyses reveal differences in cold acclimation mechanisms in freezing-tolerant and freezing-sensitive cultivars of alfalfa. Front. Plant Sci. 6:105. doi: 10.3389/fpls.2015.00105*

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

*Copyright © 2015 Chen, Han, Shang, Li, Zhang, Liu, Wang, Liu and Zhang. 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.*

# Current perspectives in proteomic analysis of abiotic stress in Grapevines

### *Iniga S. George and Paul A. Haynes\**

*Department of Chemistry and Biomolecular Sciences, Macquarie University, North Ryde, NSW, Australia*

### *Edited by:*

*Dominique Job, Centre National de la Recherche Scientifique, France*

### *Reviewed by:*

*Delphine Vincent, Department of Environment and Primary Industries, Australia Jerome Grimplet, Instituto de*

*Ciencias de la Vid y del Vino, Spain*

### *\*Correspondence:*

*Paul A. Haynes, Department of Chemistry and Biomolecular Sciences, Macquarie University, F7B 331, North Ryde, NSW 2109, Australia e-mail: paul.haynes@mq.edu.au*

**INTRODUCTION**

Grapes are a valuable fruit crop and wine production is a globally important industry with 265 million hectoliters of wine produced in 2011 (www*.*oiv*.*int) (2012). Grapevine production can be hampered by influential abiotic stresses like drought, climate fluctuations, and salinity. These factors pose a direct threat to viticulture practices. Global warming reports estimate an increase in temperature by 2–5◦C by the end of the twentyfirst century (Salinger, 2005), along with higher probability of stronger, more powerful, and more frequent climate fluctuations. The future beholds a warmer and more arid planet with sudden temperature fluctuations, caused by either natural or anthropogenic impacts. Global warming can lead to desertification, drought and intense soil salinity, all of which adversely affect grapevine quality and quantity. Studies have reported that abiotic stresses can impact wine grape production by decreasing yield and lowering quality of grapes produced (Jones et al., 2005; Cramer, 2010; Hannah et al., 2013). There is a forecast estimated decrease of up to 73% of surface land area suitable for viticulture in the main wine producing regions of the world by 2050 (Hannah et al., 2013). Although vineyards in some areas are adjusting to climate acclimation (Van Leeuwen et al., 2013), there is a need to develop environmentally sustainable crops without compromising on productivity and quality. This has driven much research into studies on plant responses to abiotic stresses. Proteomics using state of the art mass spectrometry is a powerful and promising tool to study molecular mechanisms and biological traits in plants. Grapevine responses to abiotic stresses like salt stress (Vincent et al., 2007; Jellouli et al., 2008), drought (Vincent et al., 2007; Grimplet et al., 2009b; Cramer et al., 2013), and temperature (Liu et al., 2014) have been effectively investigated. This short article will discuss the developments in grapevine proteomics, consider its current role in unraveling insights in molecular responses to abiotic stresses,

Grapes are an important crop plant which forms the basis of a globally important industry. Grape and wine production is particularly vulnerable to environmental and climatic fluctuations, which makes it essential for us to develop a greater understanding of the molecular level responses of grape plants to various abiotic stresses. The completion of the initial grape genome sequence in 2007 has led to a significant increase in research on grapes using proteomics approaches. In this article, we discuss some of the current research on abiotic stress in grapevines, in the context of abiotic stress research in other plant species. We also highlight some of the current limitations in grapevine proteomics and identify areas with promising scope for potential future research.

**Keywords: grape, proteomics, grapevine, abiotic stress**

and briefly discuss the current limitations of proteomic studies in grapevines.

### **DEVELOPMENTS IN PROTEOMICS**

Proteomic analysis techniques are constantly developing, with continuing improvements in sensitivity, resolution, accuracy, and speed of analysis. Advances in sample preparation techniques, mass spectrometry instrumentation and bioinformatics tools have paved the way for high throughput analysis. There have been great advancements in this field over the past two decades and these developments continue to expand, thus enhancing our understanding of molecular systems. In the past, sample preparation techniques using both in-gel digestion and in-solution digestion have been employed in proteomics studies on grape. Proteomic responses have been studied in tissues of grape berry (Sarry et al., 2004; Vincent et al., 2006; Grimplet et al., 2009b; Giribaldi et al., 2010; Martinez-Esteso et al., 2011b), leaf (Sauvage et al., 2007; Jellouli et al., 2010b; Delaunois et al., 2013; Nilo-Poyanco et al., 2013; Liu et al., 2014), stem (Jellouli et al., 2008), root (Castro et al., 2005; Jellouli et al., 2010b), shoot (Vincent et al., 2007; Cramer et al., 2013), and cell cultures (Martinez-Esteso et al., 2009, 2011c). Previously, two-dimensional gel electrophoresis techniques were mainly used (Vincent et al., 2007; Jellouli et al., 2008, 2010a; Grimplet et al., 2009b; Giribaldi and Giuffrida, 2010), but these are now being replaced by shotgun proteomics techniques including iTRAQ and TMT (Martinez-Esteso et al., 2011a; Liu et al., 2014), or label-free quantitation methods (Cramer et al., 2013), using ever more sophisticated mass spectrometers. Mass spectrometry instrumentation has evolved over the years from basic time-of-flight tandem mass spectrometers to multiplexed hybrid mass spectrometers. Instruments have become faster and more sensitive, with concomitant increases in resolution, thus generating far more data at higher accuracy. To keep up with these advancements, and the tremendous amount of data acquired, statistical software used to analyse mass spectrometry results, including from grapevine studies, has also been the subject of intense development. Statisticians, mathematicians and computer scientists have made efforts to create new and user friendly databases and algorithms to help understand molecular mechanisms. Consequently, the sequencing of the grape genome in 2007 (Jaillon et al., 2007; Velasco et al., 2007) represented a major breakthrough transition in grapevine proteomic research. The use of the *Vitis vinifera* genome sequence, containing approximately 30,000 genes, in database searches provided more reliable results than could be produced previously. Previous studies on grape have generally used the NCBI non-redundant database or EST contigs for identifying proteins (Marsh et al., 2010; Martinez-Esteso et al., 2011a) which works reasonably well but produces data which often represents an incomplete picture.

### **ABIOTIC STRESS STUDIES IN DIFFERENT SPECIES**

It is essential to produce sustainable plant varieties that adapt to climate variability, and to develop a broad spectrum of abiotic stress tolerant crops. Environmental factors influence dynamic changes in plants, often caused by either single or joint effects of numerous abiotic stress responsive pathways, that can be well characterized at the global level using high-throughput proteomic approaches. Proteomics has been successfully used to study abiotic stress responses in a wide range of plants like Arabidopsis (Rocco et al., 2013; Vialaret et al., 2014), rice (Neilson et al., 2013; Mirzaei et al., 2014), maize (Benesova et al., 2012), and poplar (Zhang et al., 2010), among many others, all of which have genomes that have been sequenced. This approach has also been employed for biomarker discovery in plant species with incomplete genome sequences, like peanut (Kottapalli et al., 2013), mango (Renuse et al., 2012), and even rare species like Pachycladon (Mirzaei et al., 2011), an Alpine species endemic to New Zealand. Among *Vitis vinifera* cultivars, proteomic studies prior to the sequencing of the grape genome relied on searching mass spectra against NCBI non-redundant protein databases or ESTs (Sarry et al., 2004; Vincent et al., 2006). The process of using mass spectrometers to identify proteins by cross species peptide identification is difficult, but it has become easier with the development of more high accuracy mass spectrometers.

Abiotic stress responses have been investigated in grapevine varieties of Chardonnay (Castro et al., 2005; Vincent et al., 2007), Tunisian Razegui (Jellouli et al., 2008), Cabernet Sauvignon (Vincent et al., 2007; Grimplet et al., 2009b; Cramer et al., 2013; Liu et al., 2014), and Pinot Noir (Negri et al., 2011). Grapevines have developed several adaptive approaches at the cellular and metabolic levels to mitigate, and recuperate from, the destructive effects of hostile environmental conditions. General responses include differential regulation of sugar metabolism, signaling, growth, protein synthesis, and hormone metabolism. As an example, we have observed changes in phenylpropanoid biosynthesis in Cabernet Sauvignon cells exposed to temperature stress (George et al., 2014). Osmotic stress response is the most common response to harsh environments (Cramer, 2010). Proteomics has aided in the study of differential expression of single proteins, global expression patterns, and association with regulatory pathways, and has also been used to substantiate and complement transcriptomic and metabolomic studies (Cramer et al., 2007; Zamboni et al., 2010). For example, a strong correlation was observed between transcriptomic and proteomic data in the investigation of biotic stress response to trunk diseases in green stems of Chardonnay (Spagnolo et al., 2012).

In order to better understand the metabolic changes involved in stress responses in both vegetative and reproductive parts of grapevines, and how dynamic the adaptative responses are in such situations, better experiments are needed. Ideally, if funding permitted, one would design experiments including sampling of all tissues—roots, shoots, leaves, and berries—at different developmental stages, including berry growth, veraison and ripening, and under various environmental conditions at different locations. Such an exhaustive study would be an invaluable resource for the grapevine research community, especially if it was expanded to include transcriptomic and metabolomic analysis in addition to the proteomic data sets.

## **CURRENT LIMITATIONS OF GRAPEVINE PROTEOMICS AND FUTURE SCOPE**

With the genome sequences of various plants being completed and released regularly, it is becoming easier to examine the biological pathways that trigger plant protein responses. Studies in grape have increased exponentially since 2007, after the release of the grape genome sequence data. A vast amount of knowledge has been obtained from these studies on various defense mechanisms and biological pathways triggered by external factors, including both biotic, and abiotic stresses. Investigations have been performed on different varieties ranging from the widely recognized *Vitis vinifera* cultivars like Cabernet Sauvignon and Chardonnay (Vincent et al., 2007), to other species like *Vitis riparia* (Victor et al., 2010) and *Vitis rotundifolia* Michx Muscadine (Kambiranda et al., 2014). Despite these advances, detailed understanding of proteins and protein families which are essential for stress responses are still limited. The available grape genome sequence was based on analyses of only one *Vitis vinifera* variety, Pinot Noir PN40024. Most studies so far have used only the Pinot Noir genome sequence as the reference genome. It is well known that a significant amount of transcript and protein sequences are either species specific or cultivar specific and hence may not be well represented within the Pinot Noir genome. This may lead to incompleteness in protein identification when studying other grape varieties or species. Thus, there is a clear need for sequencing of more cultivars, such as the commercially important Cabernet Sauvignon, and more related species such as *V. rotundifolia* and *V. riparia*. There are now hundreds of genome sequences available for different ecotypes of Arabidopsis (Weigel and Mott, 2009), and with the continued rapid developments in gene sequencing technologies we would hope that in the near future we will also see the publication of complete genome sequence information data for many different varieties of grapevine.

A critical challenge in grapevine proteomics is to infer biological meaning from the huge amount of mass spectrometry data acquired. The general procedure for the study of plant-environment interactions includes protein identification, protein characterization (including function annotation), construction of identified proteins into a biological network, characterization of differential protein expression under stress conditions, and assimilation of all the above into a linking framework. The initial step for this type of workflow is to identify and annotate proteins, and integrate them into the biological context.

In order to illustrate some of the current difficulties in this process, we surveyed grapevine protein sequences in NCBI and Uniprot, using the simple keyword "*Vitis vinifera.*" We found 161926 sequences in NCBI and 65548 sequences in Uniprot, which is indicative of a high level of redundancy and repetition, particularly in the NCBI database. We examined the entries in the Uniprot database in more detail. **Table 1** shows the number of protein entries for different grapevine species found in the Uniprot database, along with an indication of how many of these are still uncharacterized. Most protein entries in the database are unreviewed, which means that no additional supporting information has been presented for them. Moreover, 78% of the protein entries in the *Vitis vinifera* database are listed as "putative uncharacterized proteins." Hence, proteomic study is severely limited by the lack of better quality annotations.

In previous studies, since the roles of many individual proteins were not well defined, their biological functions were inferred from homologous proteins from other species. This task is tedious and time consuming, and produces less than complete protein identification data. Although grapevine does not have a wellestablished database like PPDB (Sun et al., 2009) (which is dedicated exclusively for *Ar*a*bidopsis thaliana* and *Zea mays* research), there is a basic database developed uniquely for grapevine molecular network study called VitisNet (Grimplet et al., 2009a). VitisNet was developed from the combination of *Vitis vinifera* (cv. Pinot Noir PN40024) genome sequencing project data, and ESTs from the *Vitis* genus, and is very useful for annotating grapevine proteins (Grimplet et al., 2012). PlantPReS (not yet published) (http://proteome*.*ir/PlantStress*.*aspx) is a freely available plant stress protein database which integrates different plant proteomic responses to stress studies. It currently has 83 plant species and is inclusive of *Vitis vinifera*. This database is still under construction, but the data that have been incorporated so far have proved



useful in annotating grapevine proteins identified in proteomics experiments.

There is a pressing need to enable the integration of large datasets, streamline biological functional processing, and improve the understanding of dynamic processes in systems biology experiments in grapevines. At the moment, software packages available for this purpose are mainly designed to work with mammalian systems. It is to be hoped that in the future more software is available that is specifically designed to function with plant protein and genome sequences, including grapevines.

### **CONCLUSION**

Proteomics is a powerful tool for molecular level discovery of biological networks in grapevine. Plant species with completely sequenced genomes, smaller genome sizes and well annotated libraries are easier to study and understand; grapevines remain a challenge. Recent advancements in mass spectrometry and proteomic techniques, coupled with the availability of complete genome sequences and improvements in bioinformatics tools, are continually strengthening this field of study. Research in this area, however, needs to be further accelerated by sequencing more grapevine varieties and different cultivars. Protein sequences in database repositories need much more functional annotation, which will help obtain better results and a more comprehensive understanding of biological responses. Proteomics has an important role to play in the future in helping to understand at the molecular level how grapevines respond to the many challenges they face.

### **ACKNOWLEDGMENTS**

Paul A. Haynes acknowledges support from the Australian Research Council for the Food Omics Research Centre and thanks Scott Ireland for continued support and encouragement. Iniga S. George acknowledges support from Macquarie University in the form of an iMQRES scholarship, and travel grants from the International Plant Proteomics Organization (INPPO) and the Australian Grape and Wine Authority, (AGWA).

### **REFERENCES**


large resveratrol production by grapevine (*Vitis vinifera* cv. *Gamay)* cell cultures in response to methyl-beta-cyclodextrin and methyl jasmonate elicitors*. J. Proteomics* 74, 1421–1436. doi: 10.1016/j.jprot.2011.02.035


Zamboni, A., Di Carli, M., Guzzo, F., Stocchero, M., Zenoni, S., Ferrarini, A., et al. (2010). Identification of putative stage-specific grapevine berry biomarkers and omics data integration into networks. *Plant Physiol.* 154, 1439–1459. doi: 10.1104/pp.110.160275

Zhang, S., Chen, F., Peng, S., Ma, W., Korpelainen, H., and Li, C. (2010). Comparative physiological, ultrastructural and proteomic analyses reveal sexual differences in the responses of *Populus cathayana* under drought stress. *Proteomics* 10, 2661–2677. doi: 10.1002/pmic.200900650

**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 October 2014; accepted: 18 November 2014; published online: 08 December 2014.*

*Citation: George IS and Haynes PA (2014) Current perspectives in proteomic analysis of abiotic stress in Grapevines. Front. Plant Sci. 5:686. doi: 10.3389/fpls.2014.00686 This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science.*

*Copyright © 2014 George and Haynes. 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.*

# Proteomic and metabolic traits of grape exocarp to explain different anthocyanin concentrations of the cultivars

### Alfredo S. Negri, Bhakti Prinsi, Osvaldo Failla, Attilio Scienza and Luca Espen\*

Dipartimento di Scienze Agrarie e Ambientali, Produzione, Territorio, Agroenergia, Università degli Studi di Milano, Milano, Italy

### Edited by:

Jesús V. Jorrín-Novo, University of Cordoba, Spain

### Reviewed by:

Georgia Tanou, Aristotle University of Thessaloniki, Greece Reinhard Turetschek, University of Vienna, Austria

### \*Correspondence:

Luca Espen, Dipartimento di Scienze Agrarie e Ambientali, Produzione, Territorio, Agroenergia, Università degli Studi di Milano, via Celoria n.2, Milano 20133, Italy luca.espen@unimi.it

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 29 April 2015 Accepted: 21 July 2015 Published: 04 August 2015

### Citation:

Negri AS, Prinsi B, Failla O, Scienza A and Espen L (2015) Proteomic and metabolic traits of grape exocarp to explain different anthocyanin concentrations of the cultivars. Front. Plant Sci. 6:603. doi: 10.3389/fpls.2015.00603 The role of grape berry skin as a protective barrier against damage by physical injuries and pathogen attacks requires a metabolism able to sustain biosynthetic activities such as those relating to secondary compounds (i.e., flavonoids). In order to draw the attention on these biochemical processes, a proteomic and metabolomic comparative analysis was performed among Riesling Italico, Pinot Gris, Pinot Noir, and Croatina cultivars, which are known to accumulate anthocyanins to a different extent. The application of multivariate statistics on the dataset pointed out that the cultivars were distinguishable from each other and the order in which they were grouped mainly reflected their relative anthocyanin contents. Sorting the spots according to their significance 100 proteins were characterized by LC-ESI-MS/MS. Through GC-MS, performed in Selected Ion Monitoring (SIM) mode, 57 primary metabolites were analyzed and the differences in abundance of 16 of them resulted statistically significant to ANOVA test. Considering the functional distribution, the identified proteins were involved in many physiological processes such as stress, defense, carbon metabolism, energy conversion and secondary metabolism. The trends of some metabolites were related to those of the protein data. Taken together, the results permitted to highlight the relationships between the secondary compound pathways and the main metabolism (e.g., glycolysis and TCA cycle). Moreover, the trend of accumulation of many proteins involved in stress responses, reinforced the idea that they could play a role in the cultivar specific developmental plan.

Keywords: anthocyanins, exocarp grape berry, metabolomics, proteomics, stress response, Vitis vinifera

# Introduction

Grape berry of the perennial and deciduous woody vines of the genus Vitis is one of the economically most important fruit crop in the world. As recounted in the 2012 report of Food and Agriculture Organization (FAO, http://faostat.fao.org/site/567/ DesktopDefault.aspx?PageID=567#ancor), 69.7694 hectares are dedicated to the cultivation

**Abbreviations:** 2-DE, two-dimensional electrophoresis; ACN, acetonitrile; cCBB, colloidal Coomassie Brilliant Blue G-250; C, Croatina; FS-LDA, Forward Stepwise Linear Discriminant Analysis; PCA, Principal Component Analysis; PG, Pinot Gris; PN, Pinot Noir; R, Riesling Italico; SIM, Selected Ion Monitoring.

of grapevine (Vitis vinifera L), with an estimation of a production of about 67 million tons per year.

During recent years, there was a burst of genomic information about the development and ripening in grape berries. After the first pioneer gene-specific molecular approaches, the appearance of large collections of grapevine ESTs (Ablett et al., 2000) and the construction of grapevine nucleotide microarrays (Terrier et al., 2005; Waters et al., 2005) opened new horizons in the study of grape berry ripening (Grimplet et al., 2007; Zamboni et al., 2010; Fortes et al., 2011). Moreover, the work of grape genome sequencing by Jaillon et al. (2007) has contributed to provide the necessary genomic information (on 2th July 2015, 94,556 protein sequences were available in the NCBI database, http://www.ncbi.nlm.nih.gov/sites/entrezz). These studies paved the way for investigating berry proteome (Deytieux et al., 2007; Giribaldi et al., 2007; Negri et al., 2008a,b, 2011; Grimplet et al., 2009; Giribaldi and Giuffrida, 2010; Zamboni et al., 2010; Martínez-Esteso et al., 2011; Niu et al., 2013; Fraige et al., 2015).

Grape is a non-climacteric fruit formed by three major tissues: exocarp (i.e., skin), mesocarp (pulp) and endocarp which surrounds seeds. Exocarp represents a physical barrier between the external environment and the inner tissues, protecting them by physical damage and pathogen attack (Grimplet et al., 2007). It is metabolically active during all developing phases being the site of the synthesis of exclusive compounds, such as aroma and some phenolic classes. Aromas arise from volatile molecules, such as terpenes, norisoprenoids and thiols that are usually stored as amino acid and sugar conjugates in the vacuole (Lund and Bohlman, 2006). Among phenolic compounds synthetized in the exocarp cells there are anthocyanins. Their color promotes seed dispersal thanks to the high contrast between background foliage and fruits (Burns and Dalen, 2002). Moreover, these compounds are involved in the protection from UV light exposure (Solovchenko and Schmitz-Eiberger, 2003). The biosynthesis of these compounds begins at véraison and continues throughout the ripening phase. The levels of anthocyanins are influenced by many factors, such as genetic background, pedo-climatic conditions and vineyard management while their profiles are relatively stable for each variety (Mattivi et al., 2006; Castellarin et al., 2012; Zheng et al., 2013). The studies on the regulation of anthyocyanin pathway revealed that the synthesis of these compounds requires the UDP-glucose flavonoid glycosyl transferase (UFGT) expression. Northern blot analysis conducted on the exocarp of a range of white and red cultivars showed, in fact, that the transcript of this enzyme was detectable only in colored grapes (Boss et al., 1996a,b). Moreover, by the isolation of several myb-related genes from berries, it was shown that the lack of expression of VvmybA1 in white cultivars results from the insertion of a retrotransposon in its promoter region absent in red cultivars, thus suggesting that the expression of this transcription factor may be the trigger of color set in grape (Kobayashi et al., 2004). Nevertheless, Ageorges et al. (2006) found that in addition to UFGT, at least three other isogenes related to the anthocyanin pathway, such as chalcone synthase 3 (CHS3) and the downstream elements glutathione Stransferase (GST) and caffeoyl methyl transferase (CaoMT) can be clearly associated to color in grape berries.

In the last years, studies performed on separate tissues pointed out several peculiar traits in gene expression, according to their functional role. As showed by the comparison among the main mature berry tissues of cultivar Cabernet Sauvignon, through the use of Affymetrix GeneChip <sup>R</sup> technology, more than 28% of the genes with significant differential expression showed differences in a particular tissue (Grimplet et al., 2007). The most expressed transcripts in the exocarp were those relating to secondary metabolism, amino acid and lipid metabolism. Moreover, some transcripts that related to some enzymes involved in carbon metabolism were over-represented in this tissue as well as some pathogenesis-related proteins were more abundant in the exocarp at the maturity (Deytieux et al., 2007; Grimplet et al., 2007). A further proteomic investigation performed on exocarp of cultivar Barbera showed that during the final ripening stage an increase in abundance of enzymes involved in primary metabolism, such as the glycolytic pathway, occurred (Negri et al., 2008b).

As expected, some -omic studies revealed that the environmental conditions, such as water availability or sunlight exclusion, deeply affected the metabolism of skin tissue (Grimplet et al., 2007, 2009; Niu et al., 2013; Zheng et al., 2013). Furthermore, a study, in which metabolite and transcript profiling of berry skin of Cabernet Sauvignon and Shiraz were compared, found that there were peculiar metabolic differences between the two cultivars (Degu et al., 2014).

The aim of the present work was to investigate possible relationships among the main biochemical pathways characterizing grape berry ripening and anthocyanin accumulation. For this purpose, four grape cultivars with increasing anthocyanin contents [Riesling Italico (synonymous: Welschriesling) (R), Pinot Gris (PG), Pinot Noir (PN), and Croatina (C)] were compared at the mature berry stage. To mitigate the seasonal effects, the study was conducted analyzing samples harvested in two different years and was performed combining a proteomic analysis (i.e., 2-DE gels/ LC-ESI-MS/MS) with a metabolomic one (i.e., GC-MS). Using multivariate statistical analysis (i.e., FS-LDA) on spot volume dataset, it was possible to discriminate the differences among the four cultivars and to sort the matches according to their discriminating power.

Through the integration of proteomic and metabolomic analyses, this work provided new insights on the ripening process in the skin and on the grape heterogeneity. The results clearly showed that the process of grape ripening in the skin differs among cultivars in some central metabolic traits. These variations appeared to be linked to the different trends of accumulation of secondary metabolites, but they appeared also related to the ripening plan.

# Materials and Methods

## Plant Material

Exocarps were harvested during the 2005 and 2006 seasons from Vitis vinifera L. cv. R, PG, PN, and C RS38 berries (full-ripe berries according to modified E–L system, Coombe, 1995). The plants were grown at the Experimental Station of the "Ente Regionale per i Servizi all'Agricoltura e alle Foreste" (ERSAF) of Regione Lombardia (Montebello della Battaglia, PV, Italy).

About 50 g of isolated skins of at least five different randomly chosen plants of one cultivar were detached immediately by squishing the berries in order to remove the seeds and the bulk of the mesocarp. By pressing and smearing the inner part of the skin on two layers of cheesecloth the residual pulp was completely taken away. This operation was repeated three times and each pool, representing a biological replicate (4 cultivars × 2 years × 3 replicas = 24 pools), was frozen in liquid nitrogen, ground and stored at −80◦C until use. From each powder pool both proteomic (5 g/extraction) and metabolomic (150 g/extraction) technical replicates were obtained.

### Evaluation of Anthocyanin Contents

Anthocyanins were extracted and measured as previously described by Fumagalli et al. (2006) and Negri et al. (2008b), respectively.

### Protein Extraction and Two-dimensional Polyacrylamide Gel Electrophoresis Analysis

Protein fraction was extracted using the method previously described by Negri et al. (2008b) with two modifications. In detail: (i) the cold acetone powder was firstly resuspended in phenol and then incubated for 30 min at 4◦C, afterwards an equal volume of extraction buffer was added to proceed with repartition step (Hurkman and Tanaka, 1986); (ii) proteins were resuspended in the IEF pH 4–7 buffer (GE Healthcare).

Protein concentration was determined by 2-D Quant Kit (GE Healthcare). Five-hundred µg of protein sample was used for each 2-DE analysis that was performed using the pH 4–7, 24 cm IPG strips (GE Healthcare) as previously described in Negri et al. (2008b).

The gels were stained according to colloidal Coomassie Brilliant Blue G-250 (cCBB) procedure, (Neuhoff et al., 1988). The gels were then scanned by an Epson Expression 1680 Pro Scanner. For each of the 6 biological replicates of a sample (3 per year), two 2-DE gels were obtained (n = 12).

### Gel and Proteomic Data Analysis

Gels were analyzed with ImageMaster 2-D Platinum Software v. 6.0 (GE Healthcare) matching the gels using the landmarkassisted procedure with one of the 48 gels as reference. Since the relevant number of gels and the possibility of spreading the matching errors, an incremental checking method was set up. Briefly, a sub-reference gel was chosen for the group of maps relating to the skins of the same cultivar and year. The spots of the sub-reference gel were ordered according to their decreasing %Vol and only the matching of the 1000 largest spots were checked. The procedure was repeated for all the 8 distinct samples taking into account only their relative 6 gels. In the second step, the analysis comprised the 12 maps of a single cultivar obtained from the skins of both years. Also in this case, only the matches of the 1000 largest spots of the sub-reference gels were evaluated. Similarly, the gels of the different pairs of cultivars were compared. Only in the final step the check was performed considering all the 48 maps. In addition to the cutting off spots of really low abundance, matches grouping less than 3 spots in a sample were discarded. Through this procedure it was possible to eliminate a relevant part of noise in the dataset and to rescue significant matches that are often lost during automatic matching.

The molecular weight and pI of the spots were estimated as previously described by Negri et al. (2008b).

The differences among the four cultivars were assessed analyzing the %Vol dataset through the application of Principal Component Analysis (PCA) and Forward Stepwise Linear Discriminant Analysis (FS-LDA) on the first 10 PCA scores as previously described by Negri et al. (2011).

Significant differences relative to identified proteins were analyzed through the one-way hierarchical clustering methodology using the software PermutMatrix (Caraux and Pinloche, 2005; Meunier et al., 2007). The 2-DE data were converted into a binary matrix replacing the missing values by zero. The row by row normalization of data was performed using the classical zero-mean and unit-standard deviation technique. Pearson's distance and Ward's algorithm were used for the analysis.

# Protein In-gel Digestion and LC-ESI-MS/MS Analysis

Spots excised from gels stained with cCBB were digested as described by Prinsi et al. (2011).

The LC-ESI-MS/MS experiments were conducted using a Surveyor (MS pump Plus) HPLC system directly connected to the ESI source of a Finnigan LCQ DECA XP MAX ion trap mass spectrometer (ThermoFisher Scientific Inc). Chromatography separations were performed by using an Inerstil WP300 C18 column (200µm I.D × 150 mm length, 5µm particle size) and a gradient from 5% to 80% solvent B [solvent A: 0.1% (v/v) formic acid; solvent B: ACN containing 0.1% (v/v) formic acid] as described by Niessen and co-workers (Niessen et al., 2006), with a flow of 2µl min−<sup>1</sup> . ESI was performed in positive ionization mode with the following parameters: (i) spray voltage: 2.5 kV (ii) capillary temperature: 220◦C. Data were collected in the full-scan and data dependent MS/MS mode with collision energy of 35% and a dynamic exclusion window of 3 min.

Protein identification was performed by Spectrum Mill MS Proteomics Workbench (Rev B.04.00.127; Agilent Technologies). Cysteine carbamidomethylation and methionine oxidation were set as fixed and variable modifications, respectively, accepting two missed cleavages per peptide. The search was conducted against the subset of Vitis protein sequences (ID 3603; June 2015, 94558 entries) downloaded from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/) and concatenated with the reverse one. The threshold used for peptide identification was Spectrum Mill score >10, Score Peak Intensity ≥70%, mass tolerance of ±2 Da for parent ion and ±1 Da fragment ions, and Database Fwd-Rev Score ≥2. Physical properties of the proteins were predicted by in silico tools at ExPASy (http://web.expasy.org/compute\_pi/).

The identified proteins were sorted in metabolic functional classes according to the MapMan BIN ontology. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Vizcaíno et al., 2014) via the PRIDE partner repository with the dataset identifier PXD002539.

### Metabolite Extraction and Derivatization

In order to extract the metabolites of the polar fraction, the protocol by Lisec et al. (2006) was used with some modifications. One hundred-fifty milligrams (150 mg) of the frozen powder relative to the 6 biological replicates (3 per year) were resuspended in 1.4 ml of −20◦C cooled methanol and 60µl of 2 mg ml−<sup>1</sup> ribitol were added as internal standard. The samples were incubated at 70◦C for 15 min in continuous agitation (1200 rpm) with a thermo mixer and subsequently centrifuged for 10 min at 11,000 g at room temperature. After recovering the surnatant, 750µl of chloroform and 1.5 ml of distilled water were added. The samples were vortexed and then centrifuged for 15 min at 2200 g at 4◦C in order to separate the phases with different polarity. Aliquots of 150µl of the water/methanol supernatant were then transferred to a clean eppendorf tube and dried on a Speedvac (RVC 2– 18 CDplus, CHRIST) without heating for 16 h. The dried residues were redissolved in 40µl of methoxyamination reagent (20 mg ml−<sup>1</sup> of metoxyamine hydrochloride in pyridine) at 30◦C and 700 rpm for 2 h and derivatized in 60µl of Nmethyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA) in the same conditions for 6 h. A retention time standard (10µl), obtained diluting a saturated C7–C40 Alkane Mixture (Supelco, 1000µg/mL for each component) 1:20 in MSTFA, was added to each sample. Before further analysis, the samples were transferred in glass vials. For each biological replicate, the analysis was repeated twice.

The employed standard substances were dissolved in distilled water or methanol at the concentration of 10 mg ml−<sup>1</sup> and 1µl was dried in vacuum and derivatized as described above to get spectral information.

### GC-MS Analysis and Data Processing

Metabolite quantification was conducted using the instrument GC-MSD comprising the gas chromatograph 7890 and the single-quadrupole spectrometer 5975 (Agilent Technologies). The employed method was the one defined by Golm Metabolome Database (http://csbdb.mpimp-golm.mpg.de/cgi-bin/madb2ml. cgi?org=msri&c=ml&o=ht&typ=met&inp=m%5B2%5D) and here briefly summarized. One microliter of sample was injected at 230◦C in splitless mode through the autosampler CTC PAL (CTC Analytics AG). The analysis was conducted using a 30 m × 0.25 mm ID × 0.25µm film thickness DB-5 column (Agilent Technologies). Helium BIP™ (Sapio) was used as carrier gas with a constant flux of 1 ml min−<sup>1</sup> . The oven ramp was so set up: 1 min at 70◦C, 6 min ramp to 76◦C, 45 min ramp to 350◦C, 1 min at 350◦C, 10 min at 330◦C. The spectra and the retention times of the standard substances were acquired in a m/z range between 40 and 600. Metabolite analysis was performed in SIM mode following a maximum of 7 ions per time-subgroup and setting a dwell time of 20 ms. The MS source and quad were maintained at 230◦C and 150◦C, respectively, using the ionization for electronic impact at −70 eV.

Spectral integration was carried out through the software MetaQuant 1.3 (Bunk et al., 2006).

Integrated peaks were normalized by the peak area of the ion with m/z = 219 of ribitol. Data were Box-Cox transformed and the differences were evaluated though ANOVA with p ≤ 0.05 using the software STATISTICA v. 7.1 (StatSoft Inc., Tulsa, OK, USA).

# Results

### Anthocyanin Contents

The anthocyanin content of the exocarp berry of four cultivars (**Figure 1**) moved from the undetectable levels of R to the high levels of C (2.25 mg g−<sup>1</sup> ) passing through the pale PG (0.21 mg g −1 ) and PN (0.83 mg g−<sup>1</sup> ).

## 2-DE Analysis

The proteomic analysis was achieved comparing 2-DE gels relating to exocarp of the four cultivars harvested in 2005 and 2006 vintages (n = 12). The number of the detected spots were comparable in the four samples and resulted to be of about 1300 spots per gels. **Figure 2A** shows the representative 2-DE maps of total protein fraction from exocarp berries of R, PG, PN, and C. Although the gel analysis pointed out a similar pattern, some differences in spot abundance were detected among the four cultivars. After automatic matching and filtering, the correspondence among the spots was assessed by manual checking. We thus focused our attention on the resulting 732 matches. In **Figure 2B**, some spots that resulted to be present with different abundance in the analyzed genotypes were reported.

Through PCA it was possible to provide a first description of the relationship among the proteomes. PC1 accounted for the 18% of explained variance and showed a tendency to move apart

FIGURE 1 | Anthocyanin contents in grape berry exocarp of Riesling Italico, Pinot Gris, Pinot Noir, and Croatina. R, Riesling Italico; PG, Pinot Gris; PN, Pinot Noir; C, Croatina. Data are the means ± ES, n = 6. Samples indicated with the different letters significantly differ according to Tukey's test (p < 0.01).

the different cultivars. PC2 (9% of explained variance), on the other hand, put PG and PN at positive values while R and C were placed in the 3rd and 4th quadrant, respectively (**Figure 3** and Supplementary Table S1). Overall, the score plot of the PCA clearly isolated C from the other 3 cultivars, while it showed the great affinity between PG and PN.

In order to distinguish the cultivars and to isolate the spots that were responsible of the observed differences, a linear discriminant analysis (LDA) was applied on the first 10 PCs calculated. The analysis was conducted by means of the forward stepwise (FS) algorithm (Fto\_Enter = 13) that selected only 3 PCs (PC1, PC2, PC3) for the classification.

The evaluation of matrix classification showed that the 12 gels of each cultivar were correctly attributed to the class of their belonging (**Table 1**). Principal components and spot volumes are linked by a linear combination: this evidence permits to calculate each discrimination model according to the spots present on the gels. After the calculation of the 4 discrimination models, it was possible to order the spots in terms of their coefficients that is according of the weight by which they contribute in classifying

FIGURE 3 | Principal Component Analysis (PCA). The score plot showed in the figure was performed on the overall dataset considering the first two PCs. The samples under investigation are coded by a twelve-circle symbol colored in yellow, gray, red and violet for Riesling Italico, Pinot Gris, Pinot noir, and Croatina, respectively.

TABLE 1 | Classification matrix of Forward Stepwise—Linear Discriminant Analysis (FS-LDA).


The analyses was performed on the principal components (PCs) calculated, allowing the discrimination of the classes of samples and sorting the variables according to their relevance.

the samples (**Table 2**). For each of the four cultivars the 30 most variables showing largest positive and negative coefficients were considered for identification. Since some of the selected spots overlapped among the four discrimination models the potential number of analyzed spots (n = 240) decrease to 144.

### Protein Identification and Functional Distribution

One hundred spots were identified by LC-ESI-MS/MS with a high degree of confidence among the group of 144 selected from FS-LDA (**Table 3** and Supplementary Table S3). The missed identifications referred both to spots present in lower abundance that were not pickable from the gels (14% of the spots) and to the excised spots for which the analysis gave results under the set level of significance (19% of analyzed spots). Some spots, such as ACO (spots 1095, 1096, 1097, and 1114), NADP-ME (spots 1315 and 1325), ENO (spots 1505, 1531, and 2619), ENO-1 (spots 1471 and 1517), IFL-5 (spots 1924, 1935, and 2618), MET, (spots 1154, 1157, and 1159), VHA-B2-X1 (spots 1411 and 1450), PPO (spots 1821 and 2386) and many spots classified in stress group



Squared Mahalanobis distances from group centroids were obtained from PC1, PC2, and PC3.

(HSP70-2, Thaumatin-like protein, TLP, PRP-10, and PRP-4) were identified as the same protein, indicating the presence of different forms with peculiar pI and/or M<sup>r</sup> . Nevertheless, different isoforms of some proteins were found (i.e., ENO and IFR).

According to their function, the identified proteins were grouped in 12 main classes (**Table 3** and **Figure 4A**). In detail, they were involved in Photosynthesis/Cell Wall (4%), C-compound/carbohydrate/energy metabolism (23%), N and amino acid metabolism (6%), Redox/Cell organization/Signal/Transport (8%), Secondary metabolism (11%), Hormone metabolism (6%), Protein (11%), Other functions (11%) and Stress (20%).

The quantitative differences among the four genotypes are showed in the Supplementary Figure S1, in which spot volume percentages were reported for all identified proteins. As a whole, the results showed that the greater differences occurred between R and C genotypes. In this view, we created two functional distribution pie charts in which R and C genotypes were compared. **Figures 4B,C** showed the proteins having higher abundance in C and in R, respectively. It is interesting to observe that in C the proteins with higher abundance belonged to the Ccompound/carbohydrate/energy metabolism, N and amino acid metabolism functional classes, while in R Secondary metabolism and Stress ones prevailed.

Using PermutMatrix software, a hierarchical clustering of the different functional groups was created to depict in detail the differences in protein abundances as well as to appreciate the gap among the proteomes of the four cultivars (**Figure 5**). The most striking evidence was that many proteins showed a direct or inverse relation with anthocyanin content. R and C showed the most divergent proteomes, while PG and PN looked more similar.

### Metabolite Analysis

GC-MS analysis permitted the identification and quantification of 56 metabolites (Supplementary Table S2). The choice of performing these analyses through a SIM approach was linked to the fact that, especially in a ripe fruit, the metabolome is dominated by large amount of accumulated sugars. Because of this, the peaks of ions relating to primary metabolites could be reduced or not easily integrated. In order to solve these analytical problems, we thus decided to limit the acquisition to the ions of interest, improving the signal and the quality of ion shape (data not shown). The ions used for the quantification were chosen on the basis of the spectra and the retention


(Continued)

TABLE 3 | List of spots identified by

LC-ESI-MS/MS.



cAmino acid coverage (%).

dInformation

 obtained by blastp (protein-protein

ePartial sequence.

 BLAST) algorithm.

times obtained by the isolated injection of standard substances under the same chromatographic and MS conditions. Among the quantified metabolites, there were sugars (16), amino acids (15), organic acids (13), and phenolic acids (5). Moreover, among the identified compounds, 5 were glycolytic intermediates and 7 were intermediates of TCA cycle. Sixteen metabolites resulted significantly different to ANOVA test (p ≤ 0.05) and some of these showed a trend that was related with the anthocyanin one (Supplementary Figure S2).

# Discussion

## The Multivariate Statistical Analysis Permitted to Isolate Spots Useful to Discriminate the Four Cultivar

Through multivariate analyses it was possible to overview the relationship among the proteomes of the 4 cultivars, clearly distinguishing the samples according to the cultivar and sorting the most relevant observed differences in spot abundance.

As expected, the biological dataset resulted to be quite complex since the first 5 PCs accounted only the 43.37% of the explained variance (Supplementary Table S1). In every case, it is interesting to note that in PC1 the cultivars were roughly ordered following their anthocyanin content, moving from R at negative values to C proteomes at positive values. Nevertheless, the placement on PC2, in which there was substantial distinction between Pinot and non-Pinot cultivars, suggested a contribution of genetic factors (**Figure 3**). At the same time, it was surprising that on the first 2 PCs differences linked to seasonal variation did not emerge. In this view, it must be considered that the technique used in this study permitted to consider only a part of whole proteome and this could be inadequate to find the plasticity of the grapevine berry previously revealed by transcriptomic study (Dal Santo et al., 2013).

The relation between the four proteomes was further investigated through FS-LDA performed on the first 10 PCs calculated. The trick of using PCs instead original variables led to dimensionality reduction and to noise elimination by excluding the less significant ones as performed in Negri et al. (2011). The linear discriminant analysis reflected and reinforced PCA results since, as witnessed by the classification matrix (**Table 1**), it correctly distinguished the gels according to the cultivar of belonging. Moreover, the squared Mahalanobis distances from group centroids (**Table 2**) demonstrated that the objects (i.e., the gels) could be ordered following the anthocyanin content of the 4 cultivars: R gels, for instance, are thus really distant from the centroid of C and rather close to the PG one. At the same time, it was interesting to note that although they are quite identical at the genetic level and overlapped in PCA score plot (**Figure 3**), the gels of the two Pinot cultivars were clearly separated by FS-LDA, showing a clear proximity as inferable by the small values of the relative squared Mahalanobis distances (**Table 2**).

As described, an interesting number of the variables with the highest or the lowest coefficients of discrimination showed to have a high significance in the models of 3 or even all the 4 cultivars, suggesting that some of the main factors that differentiate them, as for anthocyanin content, could pass through the modulation of some specific proteins.

# Proteomic and Metabolomic Data Highlight New Peculiar Traits of the Metabolism Operating in Grape Berry Exocarp

Previous proteomic studies revealed that, differently to the mesocarp, in the exocarp tissue of grape berry the glycolysis and the hexose-monophosphate shunt pathways remained active also at full ripe stage (Sarry et al., 2004; Negri et al., 2008b, 2011). This metabolic state was attributed to the demand of C-skeletons and energy to sustain the synthesis of secondary compounds (mainly anthocyanins) and/or other activities linked to defensive mechanisms.

Focusing the attention on primary carbon metabolism (i.e., glycolysis and TCA cycle) and respiration chain, the accumulation trend of the spots referring to PGMc (spot 1326),

their functional class. Two-way hierarchical clustering analysis of proteins that resulted to be significantly different in their relative spot volumes and identified by LC-ESI-MS/MS (Table 3) was performed with PermutMatrix. Pearson's distance and Ward's algorithm were used for the analysis. Each colored cell represents the average of the relative

Phothosynthesis. (B) C-compound/carbohydrate/energy metabolism. (C) Cell wall. (D) Nitrogen metabolism and amino acid metabolism. (E) Secondary metabolism. (F) Hormone metabolism. (G) Redox. (H) Protein. (I) Cell organization/Signal. (L) Transport. (M) Other functions. (N) Stress.

PGK (spot 1767), ENO-1 and ENO (spot 1517 and 1531), ACLA-2 (spot 1576), DLD (spot 1437), and NDUFS-1 (spot 1197) showed a good association with anthocyanin content also in this study (**Figures 1**, **2**, **5**; **Table 3**; Supplementary Figure S1). According to a greater activation of glycolysis and TCA cycle, some intermediates of these pathways, such as fructose-1,6 phosphate, pyruvate and succinate, showed the same tendency (Supplementary Figure S2).

The greater activation of the carbon metabolism was also suggested by a coherent increase in the red cultivar of aconitate hydratase, an enzyme having a central role in citric acid metabolism (Terol et al., 2010). Three spots identified as ACO showed high similarity with the same accession (XP\_002278138.1), so suggesting that the spots could refer to different phosphorylated forms, according to the PTMs previously identified for this enzyme (Bycova et al., 2003; Millar et al., 2005).

The results regarding enolase highlighted the multifunctional role of these glycolytic enzymes. Of the five spots referring to enolase, three revealed to be the same protein (accession XP\_002267091.2), while the others shared the greatest similarity with the accessions XP\_002283632.1 suggesting the presence of different isoforms (**Table 3**). Although, the current knowledge on the pattern of Vitis vinifera enolase did not permit further consideration from the classification point of view, by using iPSORT software (http://ipsort.hgc.jp/index.html) we verified the absence of N-terminal signal peptide, suggesting that all identified forms are cytosolic enolase. These data supported the suggestion that also in grape berry exocarp, as previously observed in others plant tissues (Voll et al., 2009), the plastidic isoform could be lacking. As recently emerged, in the absence of a complete glycolysis pathway in the plastids, cytosolic enolase plays a central role to modulate the synthesis of aromatic amino acids and secondary phenyilpropanoid compounds (Voll et al., 2009; Eremina et al., 2015). Moreover, the trend of spots 1517 and 1531 could suggest that they are the forms that modulate shikimate pathway.

On the base of the above conclusion that cytosolic phosphoenolpyruvate is requested for both phenylpropanoid biosynthesis and respiratory pathway, the higher level of pyruvate in red cultivars could appear unexpected (Supplementary Figure S2). In this context, it could be observed that among the spots showing an increase in abundance in red cultivars, two were identified as a NADP-dependent malic enzyme (NADP-ME, spots 1315 and 1325), supporting the idea that this enzyme in the skin tissue could play an important role to sustain pyruvate request. Nevertheless, the level of malic acid was higher in the red cultivars (Supplementary Figure S2), suggesting that a concomitant supply of this organic acid occurred. Previously, Iland and Coombe (1988) showed that during ripening the levels of this organic acid decreased in the mesocarp, whilst it did not change in the exocarp. Moreover, these authors found that the leaching of malate from the mesocarp tissue increased during ripening. Although a direct evidence is not available, an interesting hypothesis is that during ripening malic acid could move from mesocarp to skin, sustaining the carbon demand occurring in this tissue.

Two spots identified as a mitochondrial formate dehydrogenase (spots 1715 and 1726) showed greater abundance in C. Their reciprocal position on the gel appeared ascribable to PTMs, such as a phosphorylation (Bycova et al., 2003; Millar et al., 2005). Although a complete understanding of the functional role of this enzyme is awaited, its accumulation could be interpreted as a further request of reducing power (Plaxton and Podestà, 2015) that occurs in the cultivar with the highest level of anthocyanins.

Previously, a role of glycolytic enzymes, such as aldolase and enolase in the activation of vacuolar H+-ATPase through an association with a subunit VHA-B has been described (Barkla et al., 2009 and references therein). Moreover, an activation of vacuolar proton pumps during grape berry was reported (Terrier et al., 2001; Grimplet et al., 2009; He et al., 2010). In red cultivars an upsurge of both the FbPA (spots 1779) and the subunit VHA-B2-X1 (spots 1411) took place (**Figures 5B,L**). Although the anthocyanin transport into vacuole in exocarp tissue involves different mechanisms (Coon et al., 2008; Gomez et al., 2009; Sweetman et al., 2009; Francisco et al., 2013), these results suggest that on the whole the transport of these compounds could require an increase of the tonoplastic H+-pump activity.

Some identified spots, such as CHI-1, ANS and UFGT (spots 2117, 1653, and 1616, respectively), were enzymes operating in the anthocyanin synthesis pathway. Their trends were well related to the anthocyanin levels (**Figures 1**, **5E**). These results appear in accordance to previous transcriptional analysis in which the expression of the genes codifying for these proteins were mainly (CHI-1, ANS) or only (UFGT) detected in the red cultivars (Boss et al., 1996b). In this context, the greater abundance in PN or C of two spots that were identified as glutamine synthetase (Spots 1731, 1720) was in agreement with interlinks between anthocyanin metabolism and nitrogen recycling (Singh et al., 1998; Cantón et al., 2005). Moreover, we found that the abundances of three spots corresponding to methionine synthase (spots 1154, 1157, and 1159) related with the anthocyanin contents (**Figure 5D**). Nevertheless the levels of methionine were not different among the four cultivars, suggesting that a simultaneous transformation of this amino acid occurred in red cultivars (Supplementary Figure S2). Although a more direct evidence is needed, an intriguing hypothesis is that this amino acid could be required for the biosynthesis of ethylene, according to an involvement of this hormone in the anthocyanin production (El-Kereamy et al., 2003; Böttcher and Davies, 2012).

From the proteomic analysis differences in protein metabolism emerged (**Figure 5H**). In C the Elongation factor 2-like isoform 1 (EF-2, spot 1088), a probable Xaa-Pro aminopeptidase P (AMPP, spot 1150) and the subunit zeta T-Complex protein 1 (TCP-1, spot 1390) resulted more abundant, while some proteins involved in mitochondrialproteolytic system (MPP-α and MPP-β, spots 1463 and 1447) or in protein degradation (PSMD-7 and PMSD-14, spots 1803 and 1917, respectively) were inversely related to the anthocyanin content. Taken together, these results pointed out that in C prevailed activities involved in protein synthesis, whilst in the other cultivars a more evident protein catabolism seemed to take place. In this view, it is interesting to observe that in C CDC48, TUB-3 and ACT-7 (spots 1070, 1514, and 1661, respectively) resulted most abundant, so sustaining that in this cultivar was operating a more active cellular metabolism (**Figures 5H,I**).

## Many of the Identified Proteins are Involved in the Stress-related Processes

This work brought to the identification of many proteins that are known to be involved in defense and stress responses. It is interesting to observed that many studies performed on grape berry reported the presence of proteins, such as thaumatinlike and chitinases, even in the absence of pathogen infections (Deytieux et al., 2007; da Silva et al., 2005; Negri et al., 2008b, 2011; Fraige et al., 2015), suggesting that their accumulation could be linked to a preventive defensive plan closely related to the genetic background of the cultivar. Nevertheless, it was suggested that these proteins could also play a multifaceted role in the ripening process (van Hengel et al., 2001, 2002; Kasprzewska, 2003). In our work we find the presence of proteins belonging to the stress class also in healthy fruit as well as we confirm that they resulted to be linked to the genetic background (**Figure 5N** and **Table 3**).

Previously, Pilati et al. (2007) described in the red cultivar PN that the start of ripening phase was characterized by an oxidative burst and that this depended on specific changes in gene expression. Recently, these authors found that this process occurs mainly in the skin tissue and reported evidences on possible role of ROS as cellular signal (Pilati et al., 2014). In our work the majority of the proteins classified in the stress group showed an abundance that was both directly or inversely related to the anthocyanin contents. Considering their functions, we found a similar oxidative stress condition, but the data suggested that the analysed cultivars used different strategies in responding to it. This was particular evident comparing C with R. In this last cultivar some proteins that are known to be involved in the oxidative stress responses (i.e., PPO and SOD, spots 1818, 1821, and 2269 respectively) were present in higher abundance (**Figures 5E,G**). Considering the Mr it was possible to conclude that these PPO spots corresponded to the active form of the enzyme (Dry and Robinson, 1994). According to the possibility that a suffering condition was occurring in the R, five HSPs were more accumulated (spots 1198, 2342, 2344, 2415, 2446). Moreover, the level of ascorbic acid resulted significantly lower in R respect to other cultivars (Supplementary Figure S2). This result could be linked to genetic characteristic as well as to be a symptom of a suffering status.

It is interesting to observe that in this cultivar a 9,10[9′ ,10′ ]carotenoid cleavage dioxygenase (CCD, spot 1391) was found to be present in the highest amount, an enzyme involved in the carotenoid cleavage to produce apocarotenoid from which important signaling molecules, such as abscisic acid (ABA) and stringolactones derive (Harrison and Bugg, 2014). Recently, Ye et al. (2011) reported interesting evidences on possible role of ABA in the control of catalase gene expression, an enzyme that was described to play a central role in the ROS detoxification also in grape berry skin (Pilati et al., 2014). Although the pH range used in IEF excluded the possibility to detect the catalase, taken together, also these results further sustained the idea that in the white cultivar R biochemical mechanisms to counteract an oxidative stress condition were activated.

This study led to the identification of some forms of isoflavone reductase-like (IFL) protein (spots, 1916, 1924, 1935, and 2618). Three of these (1924, 1935, and 2618) showed high similarity with the same accession (CAI56334.1). On the basis of experimental pI, it could be suggested that the three spots refer to different phosphorylation state, a PTM that is not yet described for this enzyme. Since the stereospecific reduction of isoflavones by isoflavone reductase (IFR) is restricted primarily to legumes, a distinct reductase reaction was postulated for IFL. In fact, this enzyme was related to oxidative stress occurring during the somatic embryogenesis of Vitis vinifera callus (Zhang et al., 2009). Considering that three IFL spots showed to be more abundant in R, our data suggested a similar protective role in grape berry skin.

Differently from the white cultivar R, in the red cultivars many of these responses resulted less active. In this context, it could be underlined that flavonoid biosynthesis produces some compounds, such as quercetin, myrecetin, kaempferol, having high antioxidant activity (Pietta, 2000). The metabolic profiling on berry skin performed on a large number of cultivars, revealed that these compounds are higher in PN and C than in R (Mattivi et al., 2006). Hence, our work revealed that the oxidative burst occurring in grape berry skin ripening stimulates typical antioxidant responses in which the phenolic compounds have a central role. In the red cultivars, in fact, the high levels of phenolic compounds probably compensate for the induction of the other antioxidant mechanisms observed in the white cultivar.

# Concluding Remarks

This work highlighted new information about the biochemical and physiological events occurring in the skin tissue during grape berry ripening. Considering the functional distribution of the identified proteins and the trends of some metabolites, the results showed how many physiological processes, such as carbon metabolism (e.g., glycolysis and TCA cycle), energy conversion, secondary metabolism and oxidative stress, are involved in the protective role against damage by physical injuries and pathogen attacks. Nevertheless, some biochemical responses appeared requested to counteract oxidative burst, an event that characterizes the ripening step of grape berry skin. In this view, the strategy used strictly depended on flavonoid biosynthesis.

# Author Contributions

AN contributed to the conception of the experimental design, carried out protein extraction, 2-DE, gel analysis, metabolomic analysis and statistical analysis. BP contributed to the conception of the experimental design, carried out protein characterization by LC-ESI-MS/MS, analyzed the MS data. OF and AS participated to the manuscript revision. LE conceived the study, coordinated the experiments, wrote and edited the manuscript. All authors read and approved the final manuscript.

# References


# Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00603

sequence tags from multiple vitis species and development of a compendium of gene expression during berry development. Plant Physiol.139, 574–559. doi: 10.1104/pp.105.065748


**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.

Copyright © 2015 Negri, Prinsi, Failla, Scienza and Espen. 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.

# **Quantitative analysis of proteome extracted from barley crowns grown under different drought conditions**

*Pavel Vítámvás <sup>1</sup> \*, Milan O. Urban1, Zbynek Škodácek ˇ 1, Klára Kosová1, Iva Pitelková1, Jan Vítámvás 1, 2, Jenny Renaut <sup>3</sup> and Ilja T. Prášil <sup>1</sup>*

*<sup>1</sup> Division of Crop Genetics and Breeding, Plant Stress Biology and Biotechnology, Crop Research Institute, Prague, Czech Republic, <sup>2</sup> Faculty of Forestry and Wood Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic, <sup>3</sup> Department of Environmental Research and Innovation, Luxembourg Institute of Science and Technology, Belvaux, Luxembourg*

### *Edited by:*

*Ganesh Kumar Agrawal, Research Laboratory for Biotechnology and Biochemistry, Nepal*

### *Reviewed by:*

*Tiago Santana Balbuena, State University of São Paulo, Brazil Niranjan Chakraborty, National Institute of Plant Genome Research, India*

### *\*Correspondence:*

*Pavel Vítámvás, Division of Crop Genetics and Breeding, Plant Stress Biology and Biotechnology, Crop Research Institute, Drnovská 507/73, 161 06 Prague 6, Czech Republic vitamvas@vurv.cz*

### *Specialty section:*

*This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science*

*Received: 27 February 2015 Accepted: 15 June 2015 Published: 30 June 2015*

### *Citation:*

*Vítámvás P, Urban MO, Škodácek Z, ˇ Kosová K, Pitelková I, Vítámvás J, Renaut J and Prášil IT (2015) Quantitative analysis of proteome extracted from barley crowns grown under different drought conditions. Front. Plant Sci. 6:479. doi: 10.3389/fpls.2015.00479* Barley cultivar Amulet was used to study the quantitative proteome changes through different drought conditions utilizing two-dimensional difference gel electrophoresis (2D-DIGE). Plants were cultivated for 10 days under different drought conditions. To obtain control and differentially drought-treated plants, the soil water content was kept at 65, 35, and 30% of soil water capacity (SWC), respectively. Osmotic potential, water saturation deficit, 13C discrimination, and dehydrin accumulation were monitored during sampling of the crowns for proteome analysis. Analysis of the 2D-DIGE gels revealed 105 differentially abundant spots; most were differentially abundant between the controls and drought-treated plants, and 25 spots displayed changes between both drought conditions. Seventy-six protein spots were successfully identified by tandem mass spectrometry. The most frequent functional categories of the identified proteins can be put into the groups of: stress-associated proteins, amino acid metabolism, carbohydrate metabolism, as well as DNA and RNA regulation and processing. Their possible role in the response of barley to drought stress is discussed. Our study has shown that under drought conditions barley cv. Amulet decreased its growth and developmental rates, displayed a shift from aerobic to anaerobic metabolism, and exhibited increased levels of several protective proteins. Comparison of the two drought treatments revealed plant acclimation to milder drought (35% SWC); but plant damage under more severe drought treatment (30% SWC). The results obtained revealed that cv. Amulet is sensitive to drought stress. Additionally, four spots revealing a continuous and significant increase with decreasing SWC (UDP-glucose 6-dehydrogenase, glutathione peroxidase, and two non-identified) could be good candidates for testing of their protein phenotyping capacity together with proteins that were significantly distinguished in both drought treatments.

**Keywords:** *Hordeum vulgare***, crown, drought, proteomics, phenotyping candidate**

# **Introduction**

Drought, which significantly reduces agricultural production, represents the most severe abiotic stress worldwide. There are several definitions of drought based on different views and constraints such as meteorological drought, physiological drought, etc. (Lawlor, 2013). Physiological drought represents a discrepancy between plant water uptake and water release, resulting in water deficit and cellular dehydration. Cellular dehydration induces profound alterations in plant cell structure and metabolism, aimed at minimizing the harmful effects of the drought. Drought induces several processes in plant cells including: increased levels of abscisic acid, the levels of some metabolites such as proline, induction of stress-regulated genes, and changes in the activity of some proteins (Kosová et al., 2011). Proteins play an important role in plant adjustment to water deficit since they are directly involved in plant cell structure and metabolism. Stress-induced proteins include regulatory proteins (e.g., transcription factors, protein kinases, protein phosphatases, signaling proteins), as well as effector proteins directly involved in stress tolerance acquisition (such as chaperones), late embryogenesis abundant (LEA) proteins (such as dehydrins), mRNA-binding proteins, water channel proteins, osmolyte synthesis enzymes, components of protein biosynthesis and degradation, cytoskeletal proteins, and detoxification enzymes (Kosová et al., 2011). Dehydrins belong to the LEA protein family, and are induced by low-temperature, drought, and salinity stress in plants (Kosová et al., 2014). Moreover, in our previous studies, the accumulation of dehydrins was used to correlate cereal genotypes with different tolerance levels to abiotic stresses (Vítámvás et al., 2007, 2010; Ganeshan et al., 2008; Kosová et al., 2008, 2010, 2012, 2013; Vítámvás and Prášil, 2008; Holková et al., 2009). However, the resulting level of abiotic stress tolerance also depends on components of plant stress response other than dehydrin accumulation; therefore, detailed knowledge about the stress-dependent proteome changes is necessary.

Plant response to drought can be very diverse, depending on the severity of stress and stress timing with respect to the plant's developmental phase. Plant stress response represents a dynamic process where several phases with unique proteome compositions can be distinguished (Levitt, 1980; Larcher, 2003; Kosová et al., 2011). The initial phases of stress response (alarm and acclimation phases) usually reveal more profound differences in proteome composition (with respect to the controls), compared to later phases of stress (tolerance phase) when a novel homeostasis between plant and environment has already been established.

Barley (*Hordeum vulgare*) is a relatively drought- and salttolerant cereal crop having originated in semi-arid regions of the Middle East. Recent publication of the complete barley genome sequence (The International Barley Genome Sequencing Consortium, 2012) has significantly improved the accuracy of protein sequence identification, thus enhancing the reliability of the proteomic results. Therefore, barley represents an ideal model for investigation of crop proteome response to several stress factors. Several studies have compared barley's response to drought at the transcript level (Ueda et al., 2004; Talame et al., 2007; Tommasini et al., 2008; Guo et al., 2009). However, the accumulation of transcripts only gives a rough estimation of the protein accumulation due to translational regulations and post-translational modification or degradation of the protein (Kosová et al., 2011). Barley proteome response of barley organs to drought has only been studied by a few researchers (leaf and root—Wendelboe-Nelson and Morris, 2012; leaf—Ashoub et al., 2013; leaf—Ghabooli et al., 2013; shoot—Kausar et al., 2013). Moreover, until now, no study of changes in barley crown proteome under drought was performed. In cereals, it has been shown that survival of a plant depends on the survival of its crown tissues (e.g., Tanino and McKersie, 1985), since crowns contain both root and shoot meristems, and thus are crucial for both root and shoot regeneration after stress treatment. Due to a lack of RuBisCO, many more protein spots could be detected and analyzed by the gel-based proteomic approach than from leaf samples (e.g., Hlavácková et al., 2013 ˇ ).

The aim of the study was to investigate the response of spring barley Amulet to drought at two different intensities of drought stress, characterized by different levels of soil water capacity (SWC): 35% SWC—mild drought—D1; 30% SWC severe drought—D2; and 65% SWC—control—C. Therefore, our study had the following partial goals: (1) To investigate barley cv. Amulet response to two differential levels of drought characterized by plant water relationships (water saturation deficit, osmotic potential), the effect of drought on photosynthesis and water use efficiency characterized by 13C discrimination, and total proteome analysis by two-dimensional difference gel electrophoresis (2D-DIGE). (2) To compare the effects of two intensities of drought stress on barley plants with respect to the severity of stress impacts on the plant characteristics described above. Detection and identification of barley crown proteins revealing a differential abundance between control and drought, as well as between the two drought treatments associated with a determination of basic plant water characteristics; enabling us to distinguish common processes underlying barley's response to drought as well as specific processes differentiating the two drought intensities.

# **Materials and Methods**

## **Plant Materials and Growth Conditions**

The experiments were performed on spring barley cv. Amulet (*Hordeum vulgare* L.) obtained from the Gene Bank of the Crop Research Institute in Prague (Czech Republic). Amulet is an important spring malting barley cultivar grown in the Czech Republic (see detailed characterization and pedigree at http:// genbank.vurv.cz/genetic/resources/asp2/default\_a.htm). The seeds were germinated at 20◦C for 3 days in darkness. After germination, the seedlings (9 plants per 8 L pot) were grown in soil (a mixture of Alfisol with manure and sand, 6:2:1) under controlled conditions in a greenhouse (20◦C with 16/8 h of light/dark provided by a high-pressure sodium lamp with an irradiation intensity of 450µmol·m−2·s−1). The humidity of the soil was maintained at 65% of soil water capacity (SWC), with watering of the pots each day to maintain a constant weight (5500 g). Under these conditions, the plants were grown to the stage of the full development of the 2nd leaf. Next, one third of the plants were kept under this optimal watering (C); while the other plants had water withheld until the SWC reached 35% in the second third of pots, and 30% in the last third (5 and 6 days, respectively) under the same growth conditions. For the next 10 days (9 days for D2) the plants were watering at these SWS levels to reach plants grown at the three levels of SWS: C (65%), D1 (35%), and D2 (30%). Next, the youngest but fully-developed leaf was sampled for water-related parameters and for content of dehydrins; and the crowns for 2D-DIGE analysis. At least three biological and technical replicates of the leaves and crowns were harvested for these analyses. Samples were taken during the fourth hour of the light period. Samples for dehydrin content and 2D-DIGE analysis were immediately frozen in liquid nitrogen and kept at −80◦C.

### **Water Saturation Deficit (WSD)**

Immediately after sampling the leaves were cut into 1 cm long segments. After measurement of their weight (initial weight; Wi) on an analytical scale, the segments were fully water saturated in polyurethane cells for 3 h and weighed (weight after saturation; Ws). Afterwards, the segments were dried overnight at 95◦C and the dry weight (DW) was measured. WSD (%) was calculated as WSD = 100 × (Ws-Wi)/(Ws-DW).

## **Osmotic Potential (OP)**

The leaves were inserted into a sterile syringe and isolated by Parafilm PM-992 (Bemis). Syringes were kept at −80◦C in a freezer. Afterwards, the sample was defrosted at room temperature before the measurement of OP. The liquid needed for OP measurement on a HR 33T Dew Point Micrometer (Wescor) was obtained by pressure on the leaves in the syringe.

## **Carbon Isotope 13C discrimination**

Leaves for isotope analysis were dried (80◦C until of constant weight), ground in a Micro ball mill MM 301 (Retsch). Discrimination of 13C was measured by an IsoPrime High Performance Stable Isotope Ratio Mass Spectrometer (GV Instruments), connected with an Euro EA 3200 analyser (Eurovector), according to the manufacturer's instructions.

### **Accumulation of Dehydrins**

Dehydrin accumulation was investigated by immunoblotting of protein soluble upon boiling with anti-dehydrin antibody (Enzo Life Sciences) described by Vítámvás et al. (2010). In short, proteins soluble upon boiling were extracted by Tris buffer [0.1 M Tris-HCl, pH 8.8, containing complete EDTA-free protease inhibitor cocktail (Roche)], from frozen leaves ground in a mortar and pestle under liquid nitrogen. After the boiling step (15 min), the proteins were precipitated under acetone with 1% ß-mercaptoethanol. The protein concentrations were determined utilizing a 2-D Quant kit (GE-Healthcare). About 2.2µg of the extracted proteins were loaded into each well of 10% SDS-PAGE (Laemmli, 1970). The proteins were electrophoretically transferred to nitrocellulose (0.45µm; Pharmacia Biotech). The anti-dehydrin antibody, bound to the protein bands, was visualized by BCIP/NBT staining (Bio-Rad). A GS-800 calibrated densitometer (Bio-Rad) was used for image capture of the visualized dehydrin bands. Densitometric quantification of the dehydrin bands was done by Quantity One version 4.6.2 software (Bio-Rad).

### **2D-DIGE Analysis**

Protein extraction from the frozen crowns was carried out as described by Wang et al. (2006) with some modifications. Briefly, 200 mg of crowns (i.e., 9–12 plants) were ground in liquid nitrogen to a fine powder in 1 mL cold TCA in acetone with 0.07% DTT. After 30 min of incubation in a freezer (−20◦C), the homogenate was centrifuged (10,000 × g; 5 min; 4◦C); next, the supernatant was decanted and the pellet was washed twice (10,000 × g; 3 min; 4◦C). After overnight drying of the pellet at room temperature, the pellet was re-suspended in 0.7 mL phenol (Tris-buffered, pH 8.0) and 0.7 mL SDS buffer (30% sucrose, 2% SDS, 0.1 M Tris-HCl, pH 8.0, 5% 2-mercaptoethanol), and then thoroughly vortexed and centrifuged (10,000 × g; 3 min). The upper phenol phase was added to cold methanol with 0.1 M ammonium acetate (1:5 volume ratio), kept 30 min at −20◦C, and then centrifuged (10,000 × g; 5 min; 4◦C). The supernatant was discarded. The pellet was washed twice with cold methanol with 0.1 M ammonium acetate, and twice washed with 80% acetone (10,000 × g; 3 min; 4◦C). The pellet was dried and dissolved in lysis buffer (30 mM Tris pH 8.0, 7 M urea, 2 M thiourea, 4% w/v CHAPS). The pH of the lysate was adjusted to 8.5 by the careful addition of 50 mM NaOH, and the protein concentration was quantified by use of a 2-D Quant kit (GE Healthcare). Protein extracts were labeled (with dye switching between repetitions) prior to electrophoresis with the CyDyes™ (GE Healthcare) according to the manufacturer's instructions. Ninety micrograms of the proteins (30µg of each sample plus 30µg of internal standard) were loaded on each gel and separated by 2-DE (O'Farrell, 1975). Isoelectric focusing was run on ReadyStrip™ IPG strips (pH 4–7, 24 cm; Bio-Rad) on a PROTEAN IEF cell (Bio-Rad) according to the manufacturer's instructions until 90,000 V h were reached. The rehydration buffer contained 9.8 M urea and 4% CHAPS. After equilibration of IPG strips in equilibration buffer with DTT, and then with iodoacetamide, the focused proteins were separated in the second dimension by 12.5% SDS-PAGE (Laemmli, 1970). SDS-PAGE was performed in Ettan DALT six (GE Healthcare). Image capture of the gels was done using the PharosFX Plus (Bio-Rad) at a resolution of 100µm.

Densitometric analysis of the scanned images was carried out using PDQuest Advanced 8.0.1 (Bio-Rad). Protein spot normalization was carried out using the local regression model, and the spot manual editing was carried out using the group consensus tool. The differentially abundant protein spots (at least a two-fold change; *p* = 0.05) were chosen for spot excision (ExQuest Spot Cutter; Bio-Rad), and identification from preparative gels (2-DE of 750µg of internal standard sample) were stained by Bio-Safe Coomassie G-250 stain (Bio-Rad).

Protein identification was carried out by MALDI-TOF/TOF. After washing and desalting in ammonium bicarbonate 50 mM/50% methanol v/v, followed by 75% ACN v/v, the spots were then digested *in situ* with Trypsin Gold (mass spectrometry grade, Promega, 10 mg/mL in 20 mM ammonium bicarbonate) using an Ettan Digester robot (GE Healthcare) from the same workstation. Automated spotting of the samples was carried out with the spotter of the Ettan Spot Handling Workstation (GE Healthcare). Peptides dissolved in a 50% ACN containing 0.5% TFA (0.7 mL) were spotted on MALDI-TOF disposable target plates (Applied Biosystems) before the deposit of 0.7 mL of CHCA (10 mg/mL ACN 50% v/v/TFA 0.1% v/v, Sigma Aldrich). Peptide mass determinations were carried out using an Applied Biosystems 5800 Proteomics Analyzer (Applied Biosystems). Both the peptide mass fingerprinting and tandem mass spectrometry (MS/MS) analyses in reflection mode were carried out with the samples. Calibrations were carried out with the peptide mass calibration kit for 4700 (Applied Biosystems). Proteins were identified by searching against the SWISS-PROT, TREMBL, NCBI, and a wheat expressed sequence tag database generated from the databases using MASCOT [Matrix Science, http://www.matrixscience.com; NCBInr downloaded on June 6, 2014 (40,910,947 sequences; 14,639,572,021 residues); EST\_monocots downloaded on December 16, 2013 (45,575,892 sequences; 7,829,773,678 residues)]. All searches were carried out using a mass window of 100 ppm and a fragment mass window tolerance of 0.5 Da, and with "Viridiplantae (Green Plants)" as taxonomy for the NCBI database (1,717,798 sequences). The search parameters allowed for the carboxyamidomethylation of cysteine, dioxidation of tryptophan, and oxidation of methionine. Homology identification was retained with the probability set at 95%. All identifications were manually validated. Protein spots containing more than one significantly identified protein were excluded from further analysis. In the case of protein sequences identified as "predicted protein" with an unknown function, protein BLAST search (BLASTP) was carried out against the NCBInr database [NCBInr 20150217 (61,023,628 sequences) and UNIPROT database (UniProtKB Protein) generated for BLAST on Feb 2, 2015 (91,447,086 sequences)] to find an identified protein revealing a significant sequence similarity. Theoretical pI and MW values were calculated from the identified sequence in NCBInr using ExPASy tool (www.expasy.org). The protein functions were assigned using a protein function databases Inter-Pro (http://www.ebi.ac.uk/interpro/), Pfam (http://pfam. janelia.org/) and Gene Ontology (http://www.geneontology. org/).

### **Statistical Analyses**

For each treatment, the statistical analysis was carried out with at least three biological replicates for proteomics, three biological and three technical replicates for WSD, OP, discrimination of 13C, and dehydrin accumulation analysis. By analyzing protein abundance values, only statistically significant results were considered (One-Way and Two-Way analysis of variance (ANOVA), *p* < 0.05), and differentially abundant proteins with a ratio of at least 2.0 in absolute value, observed in at least one condition, were selected. A principal component analysis (PCA) was run on the protein spots matched on the different spot maps for qualitative appreciation of the proteomic results. Two-Way ANOVA and PCA were performed on Statistica version 10 software (StatSoft). For detailed analysis of proteome changes, the protein ratio was calculated between treatments and statistically analyzed by Student's *T*-test (*p* < 0.05). All spot densitometric data from samples grown under D1 and D2 were also used as one data set to obtain common plant responses to the drought condition (D). Cluster analysis of the protein spot

# **Results**

### **Physiological Characterization of Barley Plants**

In barley leaves, the following physiological characteristics were determined at each sampling (in each experimental variant): water saturation deficit (WSD), osmotic potential (OP), 13C discrimination (-13C; **Figure 1**). Drought led to an increase in WSD and a decrease in OP, with D2 leading to higher dehydration than D1 (milder drought). Regarding dehydrin DHN5 relative accumulation and -13C, there was a significant increase in DHN5 and a significant decrease in -13C upon drought with respect to the control; however, there were no significant differences between the two drought treatments. Plants grown under drought conditions revealed slower growth and development than plants under optimal watering (C) conditions.

### **Proteomic Analysis**

Total proteome analysis of barley crowns using the 2D-DIGE approach has led to detection of 1004 distinct protein spots thorough all gels in experiments (matched and normalized) in the pI range 4–7 (**Figure 2**, details in Supplementary Data). Quantitative analysis of protein spot density (protein spot relative accumulation) has led to detection of 105 protein spots (spots of interest), revealing significant quantitative differences between the treatments (more than two-fold change at 0.05 level); the spots were selected for MALDI-TOF/TOF protein identification. Eighty-two spots of interest were successfully identified. However, 6 spots (118, 1104, 4007, 6402, 7702, and 8001) showed double identification in the same spots; therefore they were excluded from quantitative analyses. Cluster analysis of the spots of interest revealed eight different patterns of quantitative changes between the three treatments (**Figure 3**). Principal component analysis (PCA) of all the matched protein spots revealed a clear distinction between the three treatments (C, D1, D2), with a prominent difference between the C and drought treatments (**Figure 4A**). Protein spots of interest are placed in the distant parts of the PCA protein spot area of protein relative accumulation (**Figure 4B**). The sum of standard deviations of density of the protein spots under C, D1, and D2 conditions showed different variabilities in the spot density of spots of interest (0.19, 0.17, 0.24, respectively), and in the density of all matched and normalized spots (2.06, 0.68, 1.05, respectively).

Cluster analysis revealed the presence of 8 clusters based on the differential pattern of protein abundance with respect to the individual treatments (C, D1, D2). Cluster 1 includes proteins with the highest abundance at D1 with respect to the C and D2 treatments; clusters 2 and 3 encompass proteins revealing an increase under drought with respect to the C; cluster 4 includes proteins revealing an enhanced abundance at D2 with respect to D1 and the C; cluster 5 encompasses proteins revealing a

decreased abundance at D1 with respect to D2 and C conditions; cluster 6 includes proteins revealing a decrease under both drought treatments with respect to the C; and clusters 7 and 8 encompass proteins revealing a decrease at D2 with respect to D1 and C conditions. A Venn diagram shows that: 8 proteins reveal an increase, and 14 proteins reveal a decrease, specifically in ratio D1/C; further, that 24 proteins reveal an increase, and 19 proteins reveal a decrease, specifically in ratio D2/C; while 15 proteins and 13 proteins are increased and decreased, respectively, under both drought treatments with respect to the C (**Figure 5**).

Significant differences between D2 and D1 in 27 spots of interest were also found (**Table 1**, details in Supplementary Data). An increase in accumulation was shown in 15 spots of interest (9 were identified). A decrease in accumulation was revealed in 12 spots of interest (9 were identified). Four protein spots (4614—UDP-glucose 6-dehydrogenase, 7006 glutathione peroxidase, and non-identified spots 7001 and 8104) revealed a continuous significant increase between C, D1, and D2 treatments (in all ratios). No continuous significant decrease was observed in the data set obtained. Seventy-six distinct protein spots were identified as 68 distinct proteins, distributed in 15 major functional categories regarding biological processes (**Figure 6**, **Table 1**) including: signaling and regulatory proteins (6 spots), proteins involved in regulation of DNA and RNA activity and processing (9 spots), cytoskeleton and transport proteins (3 spots), proteins involved in energy metabolism including carbohydrate metabolism (9 spots), ATP metabolism (4 spots), respiration (5 spots), and photosynthesis (2 spots), proteins involved in amino acid metabolism (10 spots), protein metabolism (8 spots), S-adenosylmethionine (SAM) metabolism (1 spot), flavonoid metabolism (3 spots), phospholipid metabolism (1 spot), phytohormone metabolism (1 spot), stress and defense responses (14 spots). Four proteins were identified in multiple proteins spots (putative 32.7 kDa jasmonate-induced protein—4201, 5202; UDP-glucose 6-dehydrogenase—4611, 4614; 2,3-bisphosphoglycerateindependent phosphoglycerate mutase-like—4708, 5701; methionine synthase—2813, 7802, 8806, 8808, 8809, 9801). A complete GO annotation (Gene Ontology database) regarding the three GO criteria (cellular localization, molecular function, and biological process) of the identified protein spots and detailed MS/MS analysis is provided in the Supplementary Data.

# **Discussion**

## **The Effect of Drought Compared to Control**

Two different drought intensities (35 and 30% of SWC; D1 and D2, respectively) were studied in the experiment. Drought induces profound alterations in plant metabolism directed toward an adjustment of plant cells to dehydration. Determination of physiological parameters WSD, OP, and -13C revealed significant dehydration and limited stomatal openness (limited CO2 availability) in drought-treated barley plants vs. controls. Similarly, an enhanced accumulation of dehydrin protein DHN5 on the immunoblots indicates cellular dehydration. Interestingly, standard deviation in the

physiological parameters and spot density indicated some trends in the samples. The least difference in variability of the data found in D1 samples could indicate a functional stress response of plants (slower growth and development, accumulation of stress, and defense proteins and metabolites). The higher variability in the control condition could be the result of faster growth and development of plants under optimal watering, while the population of individual plants could be slightly differentiated in the same biological repetition. The higher variability in D2 could be related to more severe water condition of the plants, where the damage could influence the physiological and quantitative proteomics data. What could happen with the sample if in some part of the plant a higher ratio of a senescence process or cell death occurred? For example, density analysis of spot 7109 (GST6, cluster 2; Supplementary Data) revealed that after a high increase in D1, the accumulation was decreased in D2. However, in a detailed view, half of the D2 samples accumulated at similar levels as in D1; and in the other half at a similar level as in the C. Therefore, the variability in the data set could represent an additional explanation of plant status (stress response and plant damage) and the obtained proteome results, which are discussed in the following parts of the text.

The qualitative analysis of proteome not only significantly distinguished spots between control and drought conditions, but also differences (27 spots) between both drought treatments were

drought 2 (D2) conditions. Euclidean distance and Ward's minimum criteria

found. Some spots identified as one protein (e.g., 5 spots of methionine synthase in clusters 3, 4, and 6) revealed different abundance between treatments and thereby were placed to different clusters. This could indicate different functionality of the isoforms and/or post-translational modifications under varying drought conditions and demonstrates one of the main advantages of the gel-based method compared to the gel-free approach. For example, the isoform analysis was shown by Erban and Hubert (2015) for zymogens and active-enzyme forms of house dust mite fecal allergens.

In comparison to previous proteome studies on barley's response to drought treatment, in this study, more proteins were found and identified (even for more robust protein accumulation between treatments = two-fold change). Wendelboe-Nelson and Morris (2012) identified 24 leaf and 45 root differentially accumulated proteins (however, the roots were cultivated under different conditions than the leaves) between the drought sensitive European malting barley Golden Promise (GP), and the Iraqi Basrah barley adapted to hot and dry conditions. Only four proteins from the leaf tissue (HSP70, OEE1 and 2, and methionine synthase) were the same as in our study. GP showed a lower expression and/or accumulation of constitutively present proteins, which could be connected to the slower response of GP to stress, compared to Basrah. According to the results of Wendelboe-Nelson and Morris (2012), our cv. Amulet showed a similar accumulation pattern as a sensitive GP. However, the authors did not show any detailed information about soil water content between genotypes (e.g., caused by different rates of transpiration), and did not carefully take into account their possible different response in biomass allocations (and thus their real drought adaptability). Ashoub et al. (2013) found about 22 accumulated and 6 down-accumulated proteins between tolerant (#15,141) and sensitive (#15,163) cultivars. After 5 days, pot soil field capacity drops to 10%; this evokes very sandy soil, quick, and deep stress with reduced genotype-based acclimation ability. Compared to Ashoub et al. (2013), only 3 proteins (methionine synthase, HSP90, and HSP70) were identified also in our study. Ghabooli et al. (2013) found 62 protein spots (only 45 was identified) with significant differences between *Piriformospora indica*-colonized GP plants compared with non-inoculated plants in response to drought stress (14 days; 25% field capacity). Compared to our study, only OEE1, peptidyl-prolyl cis-trans isomerase, and 60S ribosomal protein were found in Ghabooli et al. (2013). Kausar et al. (2013) identified 24 protein spots extracted from shoots in drought-sensitive Pakistani genotype 004186 and 19 spots in drought-tolerant Pakistani genotype 004223 after only 3 days of treatment. The identifications shared with our study included only protein spots identified as malate dehydrogenase, HSP70, OEE1, OEE2, methionine synthase, and glutathione transferase (GST). We found an analogous protein accumulation to patterns found in sensitive genotypes in all studies mentioned above (for details, see the text below).

In the paragraphs below, the proteins identified are briefly discussed with respect to their biological functions:

Proteins involved in signaling and regulatory processes, phospholipid metabolism:

were applied for the analysis.

The accumulations of all of proteins in this functional group decreased under drought conditions, and belong in cluster 6. Generally, the decrease could be explained by a reduction of plant growth and development under drought conditions.

Putative phospholipase D, alpha 1 (ssp 1905) catalyzes the cleavage of phospholipids, leading to formation of phosphatidic acid (PA) and other small molecules that can act as signals.

Proteins (ssp 4201, 4203, and 5202) were identified as jasmonate-regulated lectins. Several lectins have been reported to accumulate in cereal crown tissues, where the shoot apex is located, and to affect shoot apex development. For example, an accumulation of lectin VER2 was reported in cold-treated wheat crown tissue until vernalization (Rinalducci et al., 2011; Kosová et al., 2013). Therefore, in our study, the observed results (i.e., a decrease under drought) in these proteins indicated that the proteins belong into the several lectins with some stimulating role in plant development contrary to VER2 lectin in plants under cold treatment found in Rinalducci et al. (2011) and Kosová et al. (2013).

Gibberellin receptor GID1L2 (ssp 2303) is a part of the gibberellins (GAs) perception process (Ueguchi-Tanaka et al., 2007). A decrease in ssp 2303 relative abundance under both drought treatments, with respect to the control, corresponds well with the adverse effects of stress on plant growth and development.

Protein cdc48 (ssp3810) is involved in the cell division process, and is known to be downregulated in differentiated cell types. Up to now, no evidence of such protein identification was found in studies on plant abiotic stress response. However, Skadsen et al. (2000) found also decrease of cdc48 mRNA after inoculation of barley spikes with *Fusarium graminearum*.

Proteins involved in DNA and RNA regulatory processes:

Generally, the proteins involved in DNA and RNA regulatory processes have a role in plant development or growth and are decreased under drought.

MutT/nudix protein (ssp 906; cluster 6) belongs to the family of nucleoside diphosphate hydrolases, which are involved in the repair of DNA during replication. A revealed decrease in accumulation could indicate a reduced speed of replication (i.e., slower plant growth and development).

Histone H2B.1 (ssp 9004; cluster 7) belongs within the histone group. Several nucleosomal histones (histone H2A.1, histone H2B.10, histone H3.2) were found to be altered (increased or decreased) in germinating durum wheat seedlings upon salt stress (Fercha et al., 2013).

Changes in glycine-rich RNA binding proteins (ssp 12, cluster 5; 2004, cluster 1) were reported also in wheat upon cold (Rinalducci et al., 2011; Kosová et al., 2013). The members of the glycine-rich RNA-binding protein family are known to


**TABLE 1 | A list of 99** 

**differentially**

 **abundant protein spots selected for protein** 

**identification.**



*(Continued)*

**TABLE 1 | Continued**


*from the sequence, experimental pI and MW values determined from the protein spot position on 2D-DIGE gel, cluster number determined by cluster analysis, significant ratio of the changes in protein spot relative abundance betweencontrol and drought-treated samples (see below), and basic parameters related to protein spot MALDI-TOF/TOF identification. Regarding protein identification,* \**means that the protein GI sequence points to a predicted protein withunknown function and that the protein possible identity was determined by a PBLAST search against NCBInr database.*↑*, significant increase in ratio upon drought than in control;* ↓*, significant decrease in ratio upon drought thanin control; 1, D1/Control; 2, D2/Control; 3, D2/D1; 4, Drought/Control; exp, experimental; E-Value, Expectation value; NP, number of sequenced peptides by MS/MS fragmentation; QM, matched peptides; Score, MS Score; theor,theoretical.*

*(*\**) in the case of uncharacterized predicted proteins the protein name came from the best BLAST result against NCBI database—www.ncbi.nlm.nih.gov, for detail see Supplementary Data.*

regulate RNA processing, transport, and to reveal regulatory functions.

Transcription factor Pur-alpha-1 (ssp 3306, cluster 6) is involved in the initiation of nuclear DNA replication. Its decrease during drought may indicate a reduced rate of cell division upon stress conditions. It is in according with a decrease in protein containing DNA polymerase III domain (ssp 3409, cluster 7) accumulation. DNA polymerase III is a prokaryotic DNA polymerase involved in the replication of circular DNA; its homologs are found in plants in mitochondria and plastids (mitochondrial and plastidic DNA polymerases).

DEAD-box ATP-dependent RNA helicase (ssp 8709, cluster 7) was described to regulate mRNA export from nucleus to cytoplasm (i.e., to function as a RNA chaperone). In *Arabidopsis thaliana*, a mutation in the locus encoding DEAD-box RNA helicase has led to enhanced cold induction of CBF2 and its downstream genes including Cor/Lea genes (Gong et al., 2005). According to Wendelboe-Nelson and Morris (2012) , who found increase of DEAD box RNA helicase in droughtstressed roots of drought-tolerant Basrah compared to no change in sensitive genotype GP. Taken together, a higher trend in dehydrin accumulation and revealed decrease of this protein is also supporting our idea about Amulet as a sensitive genotype.

# **Energy Metabolism—ATP Metabolism, Carbohydrate Metabolism, Photosynthesis, Respiration**

Stress factors profoundly affect energy metabolism, since plant adjustment to an altered environment generally means an enhanced need for immediately available energy. Changes in several enzymes involved in ATP metabolism, especially the Vítámvás et al. Barley crowns DIGE—drought

cleavage of phosphate bonds, were found in our study (adenosine kinase 2—ssp 1310, cluster 8; inorganic pyrophosphatase—ssp 2106, cluster 1, and ssp 3103, cluster 7; nucleoside diphosphate kinase—ssp 8007, cluster 5). An enhanced need for ATP as a universal energy source has been reported in many proteomic studies aimed at plant stress responses, as indicated by the reports on increases in ATP synthase subunits (Vítámvás et al., 2012; Kausar et al., 2013). The major sources of novel ATP molecules represent both processes of anaerobic and aerobic respiration as well as photosynthesis. The anaerobic portion of respiration includes glycolysis. An increased relative abundance of glycolytic enzymes was found in several proteomic studies on stress-treated plants (Vítámvás et al., 2012; Kosová et al., 2013). However, in the present study, a decrease in some glycolytic enzymes: cytosolic triosephosphate isomerase (ssp 3107, cluster 6), and chloroplast fructose bisphosphate aldolase (ssp 2304, cluster 7); and an increase in others: 2,3-bisphosphoglycerate-independent phosphoglycerate mutase-like (ssp 4708, 5701, cluster 4), and pyruvate kinase (ssp 8607, cluster 3) were found under drought, with respect to the controls. A possible explanation of the observed difference could lie in the fact that the samples for proteome analysis were taken from plants exposed to long-term drought treatment, and the plants were fully acclimated to altered conditions without need for extra energy.

Regarding anaerobic respiration, an increase in alcohol dehydrogenase (ssp 7403, cluster 2) was found under drought. Regarding aerobic respiration, a drought-induced decrease in Krebs cycle enzyme ATP-citrate synthase (ssp 4501, cluster 7) and in complex I of respiratory electron transport chain (NADH dehydrogenase [ubiquinone] flavoprotein 2—ssp 3114, cluster 7) was found. In contrast, the levels of other Krebs cycle enzymes—succinate dehydrogenase (ssp 17, cluster 3), and malate dehydrogenase (ssp 1403, cluster 2) were increased upon drought with respect to the controls. These data indicate stress-induced imbalances in aerobic metabolism (imbalances between primary electron transport reactions and secondary enzymatic reactions) and an enhanced risk of ROS formation, which results in the downregulation of aerobic electron transport reactions, and a relative upregulation of alternative anaerobic pathways. Similar results (increase in alcohol dehydrogenase, formate dehydrogenase, aldehyde dehydrogenase) were obtained by Fercha et al. (2014) in salt-treated germinating wheat seedlings indicating the severe impact of drought on the aerobic portion of energy metabolism in our study.

Photosynthesis is known to be very sensitive to several stresses including cold, drought, and salinity. In our study, changes in two components of the oxygen-evolving complex (OEC) were found: proteins OEE1 (PsbO; ssp 205, cluster 1), and OEE2 (PsbP; ssp 3008, cluster 6). These dynamics indicated an increase under the milder drought (D1) with respect to the controls; however, a decrease under the more severe drought (D2). Changes in OEC proteins were found in drought-treated wheat genotypes (intolerant Kukri; tolerant Excalibur, and RAC875; Ford et al., 2011). Changes in OEE1 and OEE2 proteins were frequently found in salt-treated barley (Rasoulnia et al., 2011; Fatehi et al., 2012) and durum wheat (Caruso et al., 2008). Moreover, increase of OEC proteins were found in our previous studies on cold-acclimated wheat (Vítámvás et al., 2012) or barley (Hlavácková et al., 2013 ˇ ). Additionally, an increase in OEE1 protein was observed in drought-treated barley infected by *Piriformospora indica* (Ghabooli et al., 2013). Wendelboe-Nelson and Morris (2012) have demonstrated a decrease of OEE1 in stressed leaves of the sensitive barley GP, compared to the increase of OEE2 in tolerant Basrah. Kausar et al. (2013) showed an increase in OEE proteins under milder drought and in tolerant plant materials; while also showing a decrease under severe drought or in sensitive plant materials. These findings are in accordance with our findings, and we can postulate Amulet as a sensitive genotype to drought. However, the lack of other photosynthetic proteins corresponded with the material used (crowns are a non-photosynthetic tissue; see (Hlavácková et al., ˇ 2013) for a comparison of crown and leaf proteome); therefore, only two photosynthetic proteins with a difference in protein accumulation were found.

Regarding carbohydrate anabolism, an increased relative abundance of UDP-glucose 6-dehydrogenase (ssp 4611, 4614, cluster 3), and sucrose-UDP-glucosyltransferase (ssp 5821, cluster 3) was found under drought. An increase in UDP-glucose 6-dehydrogenase may indicate enhanced synthesis of pectins and hemicelluloses, as well as the remodeling of cell walls in response to stress. Generally, in our previous studies on coldacclimated wheat and barley (Vítámvás et al., 2012; Hlavácková ˇ et al., 2013), the decrease of accumulation of the sucrose-UDP-glucosyltransferase and UDP-glucose 6-dehydrogenase was observed. It implicates specific plant response to different abiotic stresses.

## **Protein Metabolism, Amino Acid Metabolism, SAM Metabolism**

The stress acclimation process is also associated with significant alterations in protein metabolism, regarding both protein biosynthesis and degradation. Alterations in protein biosynthesis are reflected in the changes of 60S ribosomal proteins L4-1-like (ssp 8403, cluster 7), L9-like (ssp 9009, cluster 3), as well as in eukaryotic translation initiation factor 5A3 (ssp 4009, cluster 3). Alterations in ribosomal proteins are described in several studies that focused on stressed wheat and barley plants (Patterson et al., 2007; Vítámvás et al., 2012; Ghabooli et al., 2013; Fercha et al., 2014; Gharechahi et al., 2014), which indicated an enhanced need for novel proteins during stress acclimation. An increase in eukaryotic translation initiation factor eIF5A3 (ssp 4009) was found under drought with respect to the C. It has been found that eIF5A not only functions as an initiation translation factor, but it can also undergo a post-translational modification of lysine residue to hypusine, and that the stoichiometry of different hypusinated forms of eIF5A can affect a switch between cell proliferation and cell death (Thompson et al., 2004). An increase in eIF5A2 in spring wheat Sandra under cold (with respect to the C), and a relatively enhanced level of eIF5A2 in spring wheat Sandra (with respect to winter wheat Samanta) was found by Kosová et al. (2013).

Protein conformation is regulated by peptidyl-prolyl *cis-trans* isomerase (ssp 9006, cluster 2). Alterations in two isoforms of peptidyl-prolyl *cis-trans* isomerase were also found in barley cv. GP colonized by *Piriformospora indica* when subjected to drought (Ghabooli et al., 2013). An increased rate of protein degradation, associated with alterations in the metabolism of stressed plants, is indicated by a drought-increased level of the proteasome subunit beta type (ssp 119, cluster 4). However, some identified proteasome subunits proteins revealed a decrease in accumulation—i.e., 26S protease regulatory subunit S10B homolog B-like (ssp 8411, cluster 6), and proteasome subunit alpha type (ssp 2104, cluster 6). Proteasomes are involved in the degradation of ubiquitin-tagged proteins. Alterations in proteasome subunits were found in several proteomic studies dealing with abiotic stresses (Rampitsch et al., 2006; Rinalducci et al., 2011; Fercha et al., 2013; Ghabooli et al., 2013).

Aminopeptidases catalyze protein degradation by the hydrolysis of N-terminal amino acid. Methionine aminopeptidase (ssp 7502, cluster 6) was found to be decreased upon both drought treatments with respect to the control. However, leucine aminopeptidase RNAs, proteins, and activities have been found to be increased following drought and wound stress signal systems, such as methyl jasmonate and abscisic acid in tomato (Chao et al., 1999).

Significant alterations were also found in several enzymes involved in amino acid metabolism. It should be noted that amino acids not only form peptides and proteins, but they are also involved in the metabolism of carbon and nitrogen, as well as in the metabolism of several stress-related compounds (e.g., S-adenosylmethionine metabolism, metabolism of phenolic compounds).

Delta-aminolevulinic acid dehydratase (ssp 408) catalyzes the first and rate-limiting step of the conversion of non-protein amino acid delta-aminolevulinic acid to porphyrin molecules, namely chlorophyll. In our results, the cv. Amulet showed an increase of this protein in D2 and D, which is connected to the greater accumulation of Heme-binding protein (ssp 114), and an increase in glutathione S-transferase (ssp 3116, 6108, and 7109). An increase in delta-aminolevulinic acid causes porphyria (Vanhee et al., 2011). All 6 of the identified methionine synthases, except one, showed increased accumulation upon drought. Wendelboe-Nelson and Morris (2012) found an increase in stressed leaves of the tolerant Basrah genotype; while Ashoub et al. (2013) found a non-significant difference between tolerant and sensitive genotypes. On the basis of published results, there are still some questions about the ability of methionine synthase to distinguish tolerant or sensitive genotypes. Methionine synthase catalyzes biosynthesis of methionine, which is not only a protein amino acid, but also a precursor of S-adenosylmethionine (SAM). S-adenosylmethionine, and the S-methylated form of methionine, not only represents a universal methyl donor in plant cells, it also functions as a precursor of several stressrelated compounds including polyamines (spermine, spermidine, putrescine), ethylene, vitamin H (biotin), and phytosiderophores (polymers derived from non-protein amino acids deoxymugineic acid and mugineic acid involved in Fe uptake). Possible alterations in phytosiderophore biosynthesis are indicated by alterations in one enzyme of the Yang cycle, methylthioribose kinase like-1 (ssp 6513, cluster 2). Changes in methionine synthase and SAM synthase were reported in several proteomic studies dealing with abiotic stress responses (Yan et al., 2006; Vítámvás et al., 2012; Kosová et al., 2013). Alterations in methylthioribose kinase were already described by Patterson et al. (2007) in barley roots exposed to elevated boron, and by Fercha et al. (2013) in germinating wheat seedlings exposed to salinity. It is known that free metal ions can act as catalyzers of ROS formation in plant cells. Phytochelatins bind metal ions, thus preventing ROS formation.

### **Stress- and Defense-Related Proteins**

A total of 14 proteins including proteins with chaperone and protective functions, as well as proteins directly involved in detoxification of ROS and xenobiotics, were identified in drought-treated barley crowns. Increased levels of the formation of xenobiotics is indirectly indicated by the enhanced accumulation of several glutathione-S-transferase (GST) isoforms (ssp 6108, cluster 1—glutathione-S-transferase I, subunit, ssp 7109, cluster 2—glutathione-S-transferase 6, chloroplastic, ssp 3116, cluster 1—glutathione-S-transferase F8, chloroplastic-like), which are known to conjugate xenobiotics with glutathione, resulting in the degradation of several xenobiotics. Increases in various GST classes has been reported by several proteomic studies dealing with stress (Kawamura and Uemura, 2003; Cui et al., 2005; Vítámvás et al., 2012; Budak et al., 2013; etc.). Moreover, several other roles for GST in protein regulation via S-glutathionylation as a posttranslational modification have been reported in plants (Sappl et al., 2004; Dixon et al., 2010). Additionally, GST-catalyzed S-glutathionylation has also been reported for intermediates of several plant secondary metabolites such as tetrapyrroles, quercetin, glucosinolates, etc. (Dixon et al., 2010). Glutathione peroxidase (GPX; ssp 7006, cluster 4) catalyzes the reduction of peroxides and has cytoplasmic and membrane-associated forms. GPX belongs to the ROS scavenging enzymes; an increase in several ROS scavenging enzymes has been reported in most all the proteomic studies dealing with plant stress response, since imbalances in energy metabolism during stress treatments are associated with the enhanced risk of oxidative stress (Kosová et al., 2011; Vítámvás et al., 2012).

Protein spot 114 (cluster 3) was identified as heme-binding protein 2 (SOUL protein superfamily) involved in tetrapyrrole metabolism. In *Arabidopsis thaliana*, heme-binding protein TSPO is known to bind tetrapyrroles, and its dynamics of degradation seems to be affected by the level of deltaaminolevulinic acid and by abscisic acid. TSPO was found to attenuate plant cell porphyria by delta-aminolevulinic acid levels and the accumulation of tetrapyrroles (Vanhee et al., 2011). Therefore, our results also indicate the attenuation of porphyria in Amulet due to increase of heme-binding protein 2, deltaaminolevulinic acid dehydratase, and GSTs.

Cellular dehydration caused by decreased SWC induces the accumulation of several proteins with protective functions; these include hydrophilic LEA proteins (ssp 401, cluster 4—Late embryogenesis abundant protein Lea14-A), and proteins related to the heat shock protein (HSP) family (ssp 803, cluster 3 cytosolic HSP90) with a chaperone function. The increased accumulation of hydrophilic LEA proteins, chaperones, and HSP was reported in several proteomic studies (Caruso et al., 2008; Sarhadi et al., 2010; Kang et al., 2012; Budak et al., 2013; Kosová et al., 2013; Xu et al., 2013). However, some HSPs were also found to be decreased under abiotic stress treatment (e.g., HSP90 in cold-treated samples in Vítámvás et al., 2012). Wendelboe-Nelson and Morris (2012) and Ashoub et al. (2013) found a higher accumulation of HSP70 in tolerant genotypes of barley. Moreover, the opposite trends of obtained results compared to previous results on cold-acclimated cereals (Vítámvás et al., 2012; Hlavácková et al., 2013; Kosová et al., 2013 ˇ ) in HSP90 (ssp 803; increase vs. decrease) and HSP70 (ssp 3804; decrease vs. increase), respectively, together with a decrease in the accumulation of cold shock protein (ssp 3006) could indicate a specific plant response to different abiotic stresses.

Protein spot 1119 (cluster 3) was identified as chitinase II, and revealed an increase under D1 with respect to the control. Chitinases belong to several classes of pathogenesis-related (PR) proteins including PR-3, 4, 8, and 11 classes (Edreva, 2005). Not only was an increase in chitinase accumulation found in cereals exposed to fungal pathogens (Yang et al., 2010; Eggert et al., 2011), but also exposed to several abiotic stresses such as cold (Sarhadi et al., 2010), salinity (Witzel et al., 2014), and others. Protein spot 8006 was identified as a PR17c precursor. An interaction with effector proteins secreted by fungal pathogens such as barley powdery mildew has been reported for PR17c in barley; however, the molecular function of PR17 proteins still remains to be well characterized (Zhang et al., 2012). Recently, PR17 was found to be increased in salt-treated barley (Witzel et al., 2014), which underlines our finding that this protein is also responsive to abiotic stress.

Protein spot 4303 (cluster 3) was identified as ricin B lectin 2, and it revealed an increase upon drought with respect to the control. Increased accumulation of ricin B lectin 2 was also found in the crowns of winter barley (Hlavácková et al., 2013 ˇ ) and winter wheat (Kosová et al., 2013) exposed to cold.

Increased protein accumulation of r40c1 (ssp 8201, cluster 3) is supported by other results in drought-treated barley cultivars with contrasting drought tolerances. Protein r40c1 was found to have constant level in the drought-tolerant barley cv. Basrah, while being stress increased in the roots of the susceptible cv. GP (Wendelboe-Nelson and Morris, 2012). Therefore, the trend obtained in the accumulation of r40c1 supports the hypothesis that Amulet could be ranked as a genotype sensitive to drought. Moreover, an increased dephosphorylation was found in the putative r40c1 protein in drought-treated rice (Ke et al., 2009).

### **Phytohormone Metabolism**

IAA-amino acid hydrolase ILR1-like protein 1 (ssp 3401, cluster 6) catalyzes a reversible IAA inactivation. Certain IAA-amino acid conjugates inhibit root elongation and therefore, the decrease of the protein revealed upon drought indicates an increase in root development (LeClere et al., 2002).

### **Flavonoid Metabolism**

Flavonoids represent a group of secondary plant metabolites containing at least two phenolic rings in their molecules. Flavonoids display several antioxidant and antimicrobial functions; thus, playing an important role in the plant stress response. However, in our study, the enzymes of flavonoid metabolism showed a decrease after drought treatments but flavoprotein wrbA-like isoform 1 (ssp 7110, cluster 2). Flavonoid biosynthesis in plants is realized from malonyl-CoA via the phenylpropanoid pathway to yield tricetin. Tricetin is then sequentially O-methylated by tricin synthase (ssp 112, cluster 1), using SAM as a methyl donor to yield tricin. The decrease of isoflavone reductase (ssp 1213, cluster 6) under drought treatments is quite interesting since isoflavone reductase is a NADPH-dependent enzyme involved in the biosynthesis of defense-related isoflavonoid phytoalexins (Oommen et al., 1994). To our knowledge, no proteome study of drought-treated barley that revealed differential accumulation of flavonoid metabolism enzymes was published. However, some studies revealed differential changes in these proteins under salt treatment in plants. Flavone-O-methyltransferase was reported to be decreased upon salinity, with respect to the control, in germinating wheat seedlings by Fercha et al. (2014). An increase in isoflavone reductase was reported in salt-treated pea (Kav et al., 2004).

### **Transport and Cytoskeleton-Related Proteins**

Annexin (ssp 8202, cluster 1) is a soluble protein that interacts with plasma membrane phospholipids. Monomeric annexins can form oligomeric channels enabling ion transport through plasma membrane, and they are also involved in vesicular trafficking and calcium signaling via MAPK cascade (for a review, see Laohavisit and Davies, 2011). An increase in annexin abundance was found also in salt-treated potato (Aghaei et al., 2008) and tomato (Manaa et al., 2011) plants, indicating their role in abiotic stress signaling.

Protein spot ssp 8708 (cluster 7) is identified as dynaminrelated protein 5A-like protein, which belongs to the dynamin M family. Plant dynamins are GTPases, which are involved in clathrin-mediated endocytosis process, as well as in vesicle transport between TGN and the plasma membrane. They also form a ring in the plant plastid division process; thus, the results could indicate reduced plant cell division (Bednarek and Backues, 2010). However, up to now, no differential protein changes in cereal dynamin were published in drought treated plants.

Chloroplastic protein TOC75 (ssp 8801, cluster 6) is a part of the TOC transmembrane channel in the outer chloroplast membrane. TOC75 is directly involved in protein-protein interaction and transport (Andrès et al., 2010). In our study, the protein level of TOC75 was found to decrease upon drought with respect to the control, which corresponds to the decrease in OEE proteins as components of photosystem II when observed under stress.

### **Quantitative Changes between Drought Conditions**

Not only were differences found in physiological parameters and in the density of protein spot accumulations between drought and control conditions, but also between both drought conditions. The drought treatments were clearly distinguished at the level of cellular dehydration (WSD and OP values); however, -13C and DHN5 relative accumulation did not reveal significant differences between the two treatments. Nonetheless, the analyses of each protein spot density should reveal detailed information about plant response to abiotic stress conditions. Based on revealed differential changes in identified proteins, it could be hypothesized that the two drought treatments differed in their intensity, which has been mirrored with some components of energy metabolism (glycolytic enzymes, ATP metabolism) and protein degradation (proteasome subunits). However, with regards to the D2 treatment, several protein spots have shown a relatively high variability in spot density between the four biological replicates, indicating that the D2 treatment may represent a threshold between plant stress acclimation and stress damage with regard to the intensity of stress (e.g., an increase in spot 119 identified as proteasome subunit beta; or a decrease in spot 205 identified as chloroplast OEE1 protein, under D2 with respect to D1 treatments). Thus, the different biological processes under severe stress conditions related to plant damage or exhaustion could influence the proteome profile of D2 compared to D1. Therefore, different trends in the accumulation of a few protein spots were obtained (e.g., 2,3-bisphosphoglycerateindependent phosphoglycerate mutase-like, ssp 4708). However, the fact that most of protein spots showed the same trends of protein accumulation compared to control conditions, with higher significant differences in D2 than in D1 compared to the C (71 and 50, respectively), indicated that for plant survival careful regulation of the same biological processes and pathways are needed in both stress conditions.

# **Conclusions**

Due to precise quantification of proteome changes by analysis of 2D-DIGE gels, we were able to determine 105 differently accumulated spots, and 76 of these were successfully identified. Until now, no other barley-drought proteome study had analyzed such a large number of protein spots. The study on drought response in the spring barley cv. Amulet has revealed that both drought treatments profoundly affected plant growth and development (changes in glycine-rich RNAbinding proteins, cell division cycle protein 48-like, gibberellin receptor GID1L2, translation initiation factor eIF5A) as well as plant energy metabolism. An enhanced need for available energy resources during the acclimation to stress conditions is indicated by profound changes in ATP metabolism as a resource of macroergic phosphate bonds. However, an enhanced risk of oxidative stress, as a consequence of imbalances in energy metabolism, leads to a downregulation of aerobic metabolism (photosynthesis, Krebs (TCA) cycle, mitochondrial electron transport chain) with respect to anaerobic metabolism (glycolysis, alcoholic fermentation). An increased risk of protein damage leads to an increase in several subunits of the proteasome complex and several protective proteins (cold shock protein, LEA-14A). Moreover, the metabolism of several stress-related metabolites (SAM metabolism—SAM as a precursor of polyamines, ethylene, phytosiderophores; flavonoid metabolism—flavonoids as protective pigments (anthocyanins) and cofactors of electron transport chain components) were significantly affected. In addition, the abundances of several proteins involved in cytoskeleton organization, protein and ion transport, etc., were affected by drought. Analysis of the obtained proteome changes demonstrated the possibility of the proteomics method used (2D-DIGE) for the evaluation of plant sensitivity or tolerance to abiotic stresses (i.e., for protein phenotyping of drought plant response). The enhanced severity of the D2 treatment was also observed at the proteome level as indicated by the differential abundance of several proteins involved in energy metabolism (glycolytic enzymes, ATP metabolism) and protein degradation (proteasome subunits); this was also validated on the physiological level (WSD and OP). Moreover, the high variability in the relative protein abundance (e.g., ssp 7109, GST6) between the four biological replicates in D2 treatment indicates an increased imbalance in cellular homeostasis in the D2 treatment, indicating a threshold between drought acclimation and damage under D2 conditions. Therefore, the wider comparison of protein abundances between other studies and ours (especially ssp 205, 401, 803, 2303, 3008, 4501, 7403, 8201, and 8201) focused on barley drought-induced proteome changes (Wendelboe-Nelson and Morris, 2012; Ashoub et al., 2013; Ghabooli et al., 2013; Kausar et al., 2013) can prove Amulet sensitivity to drought solely on the results of proteomic analysis.

For future protein phenotyping for drought plant response, the repeating and significant trends in protein spot accumulation under both drought conditions should be interesting to test. From the four protein spots that revealed a continuous significant increase under C, D1, and D2 treatments, only two spots (ssp 4614—UDP-glucose 6-dehydrogenase, cluster 3; and ssp 7006 glutathione peroxidase, cluster 3) were identified. However, all four (i.e., also ssp 7001 and 8104) could be good candidates for testing of their protein phenotyping capacity together with proteins that were significantly distinguished in both drought treatments.

# **Acknowledgments**

This work was supported by the Ministry of Agriculture of the Czech Republic (QJ1310055 and MZE RO0415) and by the Ministry of Education, Youth and Sports (LD14064 and LD14087 as parts of COST Actions FA1204 and FA1208, respectively). We thank Sébastien Planchon for his technical support in protein identifications. We thank Dr. Peter Lemkin, PhD. for English revision of the manuscript.

# **Supplementary Material**

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2015. 00479

# **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.

*Copyright © 2015 Vítámvás, Urban, Škodáˇcek, Kosová, Pitelková, Vítámvás, Renaut and Prášil. 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.*

# The conundrum of discordant protein and mRNA expression. Are plants special?

### *Isabel Cristina Vélez-Bermúdez and Wolfgang Schmidt\**

*Integrative Root Development Lab, Institute of Plant and Microbial Biology, Academia Sinica, Taipei, Taiwan \*Correspondence: wosh@gate.sinica.edu.tw*

### *Edited by:*

*Sabine Lüthje, University of Hamburg, Germany*

### *Reviewed by:*

*Katja Baerenfaller, Swiss Federal Institute of Technology Zurich, Switzerland Christine Vogel, New York University, USA*

**Keywords: omics, systems biology, post-transcriptional regulation, ribosomes, alternative splicing, gene expression**

Rapid progress in transcriptional and proteomic profiling methodologies, such as RNA-sequencing and high-resolution tandem mass spectrometry in combination with nanoflow chromatography, allows for a more accurate comparisons of disparate omics data sets at high resolution. Despite the increased fidelity of such surveys, several studies in mammals, yeast and plants have shown that transcript levels are not always a good proxy for protein abundance. While to some extent the modest concordance observed in studies across platforms may result from fundamentally different protocols that are used for the detection of proteins and mRNAs or caused by differences in the sensitivity of detection and data analysis ("false discordance"), it is becoming obvious that transcript/protein discordance is largely of biological origin ("true discordance") and represents a critical layer of regulatory processes at the post-transcriptional level that is often neglected.

### **THE MORE THE BETTER? THE CONCEPT OF POTENTIATION**

Gene activity is the result of complex dynamics between the transcription rate of the DNA template, the stability of the mRNA, the translation efficiency of the transcript, and degradation of the protein. Interestingly, the concordance between the abundance of orthologous proteins among related species is higher than that between proteins and their cognate mRNAs within a species (Kwon et al., 2013). This suggests that a certain set point of protein abundance is established to ensure optimal function, although this set point is not necessarily controlled at the transcriptional level. Rather, protein synthesis and feedback inhibition of the synthesis rate appear to dictate protein expression (Kristensen et al., 2013).

Contrary to expectations, direct coupling of transcription and translation, which occurs in prokaryotes, results in lower mRNA/protein concordance when compared to eukaryotes where the two processes are spatially separated (Vogel and Marcotte, 2012). Interestingly, concordance values are highest in single cell eukaryotes and lowest in humans (overview in De Sousa Abreu et al., 2009), indicating that cellular diversity is a large contributor to the difference between mRNA and protein abundance. In particular, decreased mRNA abundance is not conclusive. In a comprehensive survey of changes in protein and transcript profiles in *Arabidopsis* roots in response to phosphate deficiency, we observed a complete lack of correlation between downregulated transcripts and the amount of their corresponding proteins, whereas for induced genes changes in the levels of mRNAs and proteins were reasonably well correlated (Lan et al., 2012). A similar observation was reported for yeast cells subjected to salt treatment, in which transcript reduction produced only minor changes in the abundance of the corresponding proteins (Lee et al., 2011). Strong induction of transcription is a much better predictor of changes in protein levels than decreased expression. This might be related to the importance of stressassociated proteins that are required to recalibrate cellular metabolism. In fact, highly and lowly transcribed genes differ in their translational fitness, i.e., their association with polysomes (Preiss et al., 2003). Relatively few studies compare mRNA and protein expression in plants using (mainly Arabidopsis and maize), but the general picture that emerges from these studies is a variable correlation of protein–transcript pairs (Baerenfaller et al., 2012; Walley et al., 2013; Ponnala et al., 2014) that is comparable to what has been reported for other organisms. Thus, although technical improvements and the use of techniques measuring *in vivo* translation such as ribosome profiling (Ingolia, 2014) will correct the observed correlations, it appears that at least a substantial proportion of this lack of concordance is of biological origin and thus reflect post-transcriptional regulation rather than technical constrains.

Preferential translation of transcripts derived from highly induced genes, referred to as homo-directional coregulatory response or potentiation (Preiss et al., 2003), amplifies transcriptional changes at the translational level, leading to a fast and robust change in gene activity. In plants, translation efficiency can change dramatically in response to abiotic stress, leading to a massive bias in the pool of mRNAs that are actively translated (Mustroph et al., 2009; Juntawong et al., 2014). Potentiation of gene expression can be encoded in both DNA and mRNA; translation of *Arabidopsis* mRNAs upon light treatment is dependent on the presence of the sequence motifs TAGGGTTT or AAAACCCT in their 5- UTR (Liu et al., 2012). These elements are also required for transcription (Tremousaygue et al., 1999; Tatematsu et al., 2005), suggesting that such co-regulatory responses have evolved to switch the expression of genes into a fast-forward mode when the demand for the encoded proteins is high. One may speculate that, while gene induction is amplified by preferred translation of transcripts derived from these genes to secure fast acclimation, down-regulation of gene expression is a weaker signal and the full scale of regulatory post-translational mechanisms results in pronounced discordance between transcripts and proteins.

### **STOP MAKING SENSE: PRODUCTION OF NON-FUNCTIONAL TRANSCRIPTS TO TUNE PROTEIN ABUNDANCE**

In principle, the mechanisms that control RNA turnover, translation, and protein stability are similar among eukaryotes. However, lacking behavioral recourses in coping with unfavorable conditions, plants can adjust their developmental, metabolic and physiological programs in a much broader way than, for example, mammals. The phenotypic plasticity of plants results from sophisticated sensing and signaling circuits that integrate disparate environmental cues and modulate gene activity to adjust the phenotypic readout to the prevailing conditions.

The ability of plants to adapt rapidly to a wide range of conditions is reflected by numerous plant-specific peculiarities in the control of gene activity. This high level of transcriptional regulation may correspond to an equally pronounced abundance of post-transcriptional regulatory processes. An interesting example of a process that is seemingly similar in plants and animals, but that has significantly distinct consequences, is the splicing of pre-mRNAs during transcription. In human cells, the vast majority of intron-containing genes (∼95%) are alternatively spliced, leading to the generation of multiple, distinct mRNAs from a single gene. The predominant form of alternative splicing in mammals is the exclusion of cassette exons together with its flanking introns, a process referred to as exon skipping, which leaves the open reading frame uninterrupted (Kornblihtt et al., 2013). Exon skipping is thought to contribute to proteome diversity by producing protein isoforms that are structurally and functionally distinct. Other forms of alternative splicing, i.e., introns retention or alternative 5 and 3 splice sites, produce transcripts that mostly contain premature stop codons (PTCs), targeting these transcripts for degradation via the nonsensemediated decay RNA surveillance pathway, a mechanism that prevents the production of truncated proteins by eliminating PTC-containing mRNAs after the pioneer round of translation.

In plants, intron retention and alternative 5 and 3 splice sites comprise the majority of alternative splicing events (Filichkin et al., 2010; Marquez et al., 2012; Li et al., 2013; Wu et al., 2014). Splicing patterns appear to be more complex during acclimation to nutrient deficiencies or light exposure than under constant conditions, resulting in significantly induced intron retention features (Li et al., 2013; Wu et al., 2014). The reasons for the different forms and consequences of alternative splicing in animals and plants are currently unknown. Two plausible scenarios would explain the high number of intron- (and, in most cases, PTC-) containing transcripts in plants. Firstly, non-functional mRNAs could be stored as ribonucleoproteins and processed when needed. This would allow for a fast increase in populations of mature mRNAs that code for proteins that are required upon stress exposure or during development. In an alternative, but not mutually exclusive scenario, non-functional transcripts are produced to tune the abundance of proteins with critical functions in response to a fluctuating environment. A quick shift between the production of functional and non-functional transcripts also aids in rapid re-adjustment of protein levels after the stress is relieved. Production of non-functional transcripts would provide an alternative mechanism for such an adjustment, circumventing changes in transcription rates, which involve recruitment/dismissal of transcription factors and changes in chromatin structure.

Changes in splicing patterns in plants as a means to calibrate protein abundance would imply a feedback mechanism that communicates the demand for a given protein to the splicing machinery, and a switch to adjust the ratio of functional to non-functional transcripts. Shifting between the production of functional and PTC-containing transcripts could be achieved by post-translational modifications of proteins from the splicing machinery such as serine/arginine (SR) rich splicing factors to facilitate or repress interaction of the spliceosome with a subset of mRNAs.

Another possible mechanism relies on the presence of information in the differentially retained introns or in the flanking exons, such as *cis*-acting intronic splicing silencers that are differentially recognized by *trans*-acting mRNA-binding proteins. In this scenario, post-translational modifications of mRNA-binding proteins could also modulate the probability of splicing events near a given site. In any case, the production of non-functional mRNA isoforms contributes massively to the apparent transcriptome/proteome discordance in plants, but to a much lesser extent in animals where the vast majority of alternative splicing events yield functional products. Quantitative distinction between functional and non-functional transcripts, presently still a challenging task for genome-wide transcriptional surveys, would most probably change the correlation between transcript and protein levels toward a higher concordance correlation coefficients in plants.

### **BUILD TO ORDER: DO PLANTS PRODUCE SPECIALIZED RIBOSOMES?**

Translation is mediated by ribosomes, intricate molecular machines composed of ribosomal RNA and ribosomal proteins (r-proteins) that translate the genetic code encrypted in the DNA into proteins. Because the function of ribosomes is highly conserved in both prokaryotes and eukaryotes, r-proteins are traditionally classified as housekeeping. However, eukaryotic ribosomes contain more proteins than their bacterial counterparts and possess diverse r-RNA modifications that are not found in prokaryotic ribosomes, indicating more sophisticated molecular functions of eukaryotic ribosomes.

In contrast to animals, in which rproteins are mostly encoded by a single gene, plant r-proteins are encoded by paralogous families comprising several members that generate diverse, functional proteins. For example, the 81 r-proteins of *Arabidopsis* are encoded by more than 200 genes, with each r-protein family consisting of 2–7 members. This does not only complicate coordinate expression of equiamounts of r-proteins to secure ribosomal function, but also allows for a nearly infinite number of differently composed ribosomes, the heterogeneity of which can be further increased by numerous post-translational modifications. In humans and Drosophila, defects in rprotein expression have been associated with diseases (Kongsuwan et al., 1985; Uechi et al., 2001), suggesting functions beyond translation. Similarly, in plants several r-protein mutants are affected in cell division and/or cell expansion resulting in deformed leaves (Rosado et al., 2012), indicating specific, extra-ribosomal functions of some r-proteins in developmental processes. It should be noted that in plants accurate detection of protein concentrations of paralogous proteins by mass spectrometry is rendered difficult as these proteins often have identical sequence parts that cannot be distinguished.

### **A PLANT-SPECIFIC RIBOSOME CODE?**

The large number of r-protein paralogs in plants invites speculation as to whether populations of structurally diverse ribosomes produced during development or in response to environmental signals can capture mRNAs differentially and prioritize the translation of specific subsets of mRNAs. The relative incorporation of r-protein paralogs is altered by growth conditions at the protein level (Hummel et al., 2012), and transcripts encoding r-protein accumulate differentially upon iron and phosphate deficiency (Rodríguez-Celma et al., 2013; Wang et al., 2013), suggesting that the translational machinery is remodeled in response to environmental signals. Heterogeneous ribosomal populations would contribute markedly to discordant changes in transcript and protein profiles. Translatome profiling studies support a regulatory intervention of protein abundance at the translational level (Mustroph et al., 2009; Juntawong et al., 2014). A transcriptomic comparison of steady-state and polysome-bound mRNAs revealed that translational control is independent of mRNA abundance (Liu et al., 2012). The authors of this study concluded that translational control has a greater effect on gene activity than the high steady state mRNA levels. Dynamic changes in r-protein composition would offer a plausible explanation for the observed differential translational efficiency of mRNAs in response to changing environmental conditions.

### **CONCLUSIONS**

Based on the above considerations, we propose that the uncoupling of transcript and protein abundances in plants is driven by additional mechanisms that are not prominent or not present at all in other organisms. This proposal is supported by the predominant forms of alternative splicing in plants (i.e., intron retention and alternative donor or acceptor splice sites) and a highly dynamic ribosome composition that aids in tuning protein profiles to cellular demands during development or stress. Although ribosomal specificity might also exist in mammals (O'Leary et al., 2013), the large number of r-protein genes suggests that heterogeneity of ribosomes in plants is much more pronounced than in other eukaryotes. Also, recruitment of ribosomes to mRNA may have plant-specific dynamics that affect translation efficiency (Lan and Schmidt, 2011). An underappreciated factor is the cell type-specific variation in splicing patterns (Lan et al., 2013), r-protein composition (Whittle and Krochko, 2009), and other, not yet explored processes that may differ among cell types such as mRNA export, and protein stability, which may contribute to the high mRNA/protein discordance in multicellular organisms. Another factor that has not yet been explored at the whole-genome scale is the impact of microRNAs on translation and thus on mRNA/protein abundance correlation in plants. The points brought up here are not only of academic interest. "True" discordance mirrors an underappreciated regulatory layer for determining the final concentrations of proteins. Attempts to engineer the genetics of crop plants to improve plant performance under stress conditions traditionally aim at the control of gene expression. Knowledge regarding post-transcriptional regulation that contributes to transcript/protein discordance may aid in generating stress-resistant germplasm.

### **ACKNOWLEDGMENTS**

We thank Marjori Matzke (IPMB, Taiwan) for critical comments on the manuscript. Work in the Schmidt lab is supported by MoST and Academia Sinica.

# **REFERENCES**


in transcriptome composition and mRNA translation induced by rapamycin and heat shock. *Nat. Struct. Biol.* 10, 1039–1047. doi: 10.1038/nsb1015


**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 August 2014; accepted: 21 October 2014; published online: 07 November 2014.*

*Citation: Vélez-Bermúdez IC and Schmidt W (2014) The conundrum of discordant protein and mRNA expression. Are plants special? Front. Plant Sci. 5:619. doi: 10.3389/fpls.2014.00619*

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

*Copyright © 2014 Vélez-Bermúdez and Schmidt. 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.*

# Regulation of mRNA translation controls seed germination and is critical for seedling vigor

Marc Galland1, 2 † and Loïc Rajjou1, 2 \*

1 INRA, Institut Jean-Pierre Bourgin, UMR 1318 INRA/AgroParisTech, ERL Centre National de la Recherche Scientifique 3559, Laboratory of Excellence "Saclay Plant Sciences" (LabEx SPS), Versailles, France, <sup>2</sup> Chair of Plant Physiology, AgroParisTech, Paris, France

Keywords: seed, plant, dormancy, germination, longevity, translation, mRNA, proteins

### Edited by:

Ganesh Kumar Agrawal, Research Laboratory for Biotechnology and Biochemistry, Nepal

### Reviewed by:

Ján A. Miernyk, University of Missouri, USA Pingfang Yang, Chinese Academy of Sciences, China

> \*Correspondence: Loïc Rajjou, loic.rajjou@agroparistech.fr

†Present Address: Marc Galland, Department of Plant Physiology, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, Netherlands

### Specialty section:

This article was submitted to Plant Proteomics, a section of the journal Frontiers in Plant Science

Received: 28 February 2015 Accepted: 09 April 2015 Published: 28 April 2015

### Citation:

Galland M and Rajjou L (2015) Regulation of mRNA translation controls seed germination and is critical for seedling vigor. Front. Plant Sci. 6:284. doi: 10.3389/fpls.2015.00284 The control of seed germination capacity is a multi-level molecular process including, epigenetic, transcriptional, post-transcriptional, translational, and post-translational regulation (Rajjou et al., 2012). Since the beginning of the twenty-first century, a wide range of genetic, genomic, and post-genomic approaches have been used to decipher the underlying molecular and biochemical bases of dormancy, vigor, and longevity. In particular, the regulation of stored mRNA translation appears as an essential determinant of seed quality. Indeed, proteomic approaches unveiled the main importance of protein synthesis during seed germination (Rajjou et al., 2004; Kimura and Nambara, 2010), as well as factors involved in seed longevity (Rajjou et al., 2008). In aged seeds, the protein synthesis capacity decline together with the loss of germination potential. In contrast, both dormant and non-dormant Arabidopsis seeds display equal translational activity 1 day after imbibition although distinct protein pools are synthetized (Chibani et al., 2006). In sunflower, seed dormancy release by after-ripening would not be related to transcriptomic changes but associated with oxidation of specific subsets of stored mRNAs, thus impairing their translation (Bazin et al., 2011; Meimoun et al., 2014). In the aim to get a comprehensive view of translational control of seed dormancy and germination in sunflower, a microarray-based translatome analysis was performed, highlighting differential accumulation of polysome-associated mRNAs between dormant and non-dormant imbibed seeds (Layat et al., 2014). However, multiple ribosomes are not necessarily translationally active. Indeed, it has been observed that both active and stalled ribosomes are able to co-sediment during isolation of polysome complexes (Sivan et al., 2007). As a result, polysome profiling does not fully discriminate translationally active from repressed mRNAs. This concern should be particularly true in the case of dry seeds where polysomes would not be functional. Indeed, it has been observed that the ribosomes are condensed into regions consisting of closely packed particles in the dry seed related with a latent potential for protein synthesis (Chapman and Rieber, 1967). A rapid polysome formation occurs during early germination related with the transition from a dry and quiescent state to a fully imbibed and metabolically active state. In non-dormant Arabidopsis seeds, the comparison between the transcript changes and the protein changes from dry to 1d-imbibed seeds showed strong discordance (Galland et al., 2012). This is in accordance with previous work in plants reporting that the abundance of a transcript does not necessarily reflect its translation (Bailey-Serres et al., 2009). It is likely that when conditions are favorable for the maintenance of seed dormancy, translational selectivity will promote the translation of stored mRNA associated with maturation program (Arc et al., 2012). The time course of seed germination is related to both sequential and selective mRNA translation emphasizing a fine regulation of the translational machinery (Galland et al., 2014). Indeed, the temporal profiling of protein synthesis highlights that Arabidopsis seed germination consists of a series of sequential events overlapping with the three canonical phases of this process namely, water uptake (Phase I), lag phase (Phase II), and

radicle growth (Phase III). Germination sensu stricto (i.e., prior to radicle emergence) refers to phases I and II while phase III consists of seedling growth resumption accompanied by both cell elongation and cell division. In the early step of water uptake, germination begins with a resumption of maturation program through the translation of mRNA associated with storage proteins and tolerance desiccation. This relates to an important checkpoint where, in a favorable environment, germination of non-dormant seeds is accompanied by a radical change in their translational program. Indeed, in the lag phase a sequential translation of mRNA related with antioxidant mechanisms, cell detoxification, protein fate, energy, and amino acids metabolism occurs. At the end of this lag phase, proteins involved in protein degradation and nitrogen remobilization are neosynthetized in preparation for the seedling growth. Therefore, it can be assumed that mRNA translation and protein post-translational modifications constitute the main levels of control for germination completion (Arc et al., 2011; Rajjou et al., 2012). These processes are highly regulated in plants and represent rapid and efficient way to cope with environmental variations. The regulation of mRNA translation is extremely complex and not explored enough in seed biology. Still the seed would be an excellent model for studying translation and selectivity mechanisms due to the presence of different mRNA populations in the dry mature seed. It seems relevant to conduct a comprehensive investigation about the mRNA recruitment by the nuclear cap-binding complex (CBC) and by the cytoplasmic mRNA CBC (eIF4F) since several phenotypes were observed in mutant seeds affected in genes involved in these mechanisms. The eIF4F and polyA-binding protein (PABP) promotes the transcript stabilization and the ribosome–mRNA interactions (Gingras et al., 1999; Hinnebusch and Lorsch, 2012). The exon junction complex (EJC) would be involved in plant translational selectivity since it links the different aspects of mRNA biogenesis, such as transcription, splicing, export,

## References


surveillance, and nonsense-mediated mRNA decay (NMD) (Nyikó et al., 2013). Particular attention should be paid to DEAD-box RNA helicases presumably involved in translation by assisting ribosome maturation (Cordin et al., 2006). Indeed, the Plant RNA helicase75 (PRH75) have been shown to be a target of the Protein Isoaspartyl Methyltransferase 1 (PIMT1) that repair non-functional isoAsp residues upon seed deterioration (Nayak et al., 2013). This is probably an explanatory element of the impairment of seed vigor and longevity in PIMT-deficient genotypes (Ogé et al., 2008; Verma et al., 2013). In addition, the cap-independent process through direct mRNA recruitment by ribosomal subunits on an internal ribosome entry sites (IRES) would be possible since mature seeds have proteins named ITAFs (IRES-specific cellular trans-acting factors) involved in this process (Catusse et al., 2008). To date, mRNA decay and translational repression by small RNAs remain non-addressed in seed biology but may be a determinant way for translational selectivity. The impact on translation of the plant TOR (target of rapamycin) protein kinases pathway is associated with abscisic acid (ABA) and growth processes in plants (Deprost et al., 2007). Further investigation of TOR-dependent phosphorylation signaling in seed dormancy, germination, and longevity appears required. The involvement of different ribosomal subunits and their post-translational regulation also remains unexplored in the control of seed germination. Thus, through this reasoning about the central role of translational regulation in the control of germination, future work on this issue should provide a better understanding of the mechanisms underlying seed physiology and provide robust markers for seed vigor.

### Acknowledgments

Our heartfelt gratitude to the French Ministry of Industry (FUI, NUTRICE) and the European Commission's Seventh Framework Programme (KBBE, EcoSeed) for funding our research.


treatment on the resumption of transcription during imbibition. Plant Mol. Biol. 73, 119–129. doi: 10.1007/s11103-010-9603-x


proteome highlights the distinct roles of stored and neosynthesized mRNAs during germination. Plant Physiol. 134, 1598–1613. doi: 10.1104/pp.103. 036293


**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.

Copyright © 2015 Galland and Rajjou. 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.