# **PHYSIOLOGICAL AND MOLECULAR ECOLOGY OF AQUATIC CYANOBACTERIA**

# **Topic Editors George S. Bullerjahn and Anton F. Post**

#### *FRONTIERS COPYRIGHT STATEMENT*

© Copyright 2007-2015 Frontiers Media SA. All rights reserved.

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

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

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

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

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

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

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

**ISSN** 1664-8714 **ISBN** 978-2-88919-318-9 **DOI** 10.3389/978-2-88919-318-9

## *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

## **PHYSIOLOGICAL AND MOLECULAR ECOLOGY OF AQUATIC CYANOBACTERIA**

Topic Editors:

**George S. Bullerjahn,** Bowling Green State University, USA **Anton F. Post,** Marine Biological Laboratory, USA

Surface water, western Lake Erie (Laurentian Great Lakes, USA) during the 2010 Microcystis bloom event.

Photo credit: Jerry Wicks, Bowling Green State University

The cyanobacteria inhabit every illuminated environment on Earth, from polar lakes to desert crusts and through their phototrophic metabolism play essential roles in global geochemical cycles. With the discovery of marine Synechococcus and Prochlorococcus almost 30 years ago, cyanobacteria have now earned their place as dominant primary producers contributing over 25 percent of global photosynthesis. Their global abundance is now explained from the coexistence of ecotypes that occupy different niches along spatial and temporal gradients. New ecotypes of Synechococcus have been identified as abundant components of microbial communities in freshwater

environments and marginal seas. Extensive comparative genomics of marine and freshwater picocyanobacteria have begun to unmask adaptations to light and nutrient (N, P, Fe) limitation that these diverse environments present. Novel types of cyanobacterial diazotrophy input new N and structure microbial communities in the open sea. Current challenges include the understanding of the interactions between marine cyanobacteria and other microbes in their immediate community.

In contrast, mesotrophic and eutrophic environments such as the Laurentian Great Lakes have been increasingly affected by nuisance and toxic cyanobacterial blooms that have yielded severe declines in water quality. Factors promoting bloom formation and the functional roles of toxins are important issues being addressed today.

 This collection of papers is designed to highlight the ecological roles played by cyanobacteria, focusing primarily on their adaptations that allow their dominance and niche differentiation in diverse aquatic environments. These issues are particularly important as we face global environmental alterations that influence aquatic habitats through climate change, ocean acidification and nutrient inputs leading to toxic bloom events.

# Table of Contents


Jill A. Sohm, Bethanie R. Edwards, Brian G. Wilson and Eric A. Webb


## Physiology and molecular biology of aquatic cyanobacteria

#### *George S. Bullerjahn1 \* and Anton F. Post <sup>2</sup>*

*<sup>1</sup> Department of Biological Sciences, Bowling Green State University, Bowling Green, OH, USA*

*<sup>2</sup> Marine Biological Laboratory Woods Hole, Woods Hole, MA, USA*

*\*Correspondence: bullerj@bgsu.edu*

#### *Edited and reviewed by:*

*Jonathan P. Zehr, University of California, Santa Cruz, USA*

**Keywords: cyanobacteria, HABs, Prochlorococcus, Synechococcus, nitrogen fixation**

Cyanobacteria thrive in every illuminated aquatic environment known, contributing at least 25% of primary productivity worldwide. Given their importance in carbon and nutrient cycles, cyanobacteria are essential geochemical agents that have shaped the composition of the Earth's crust, oceans and atmosphere for billions of years. The high diversity of cyanobacteria is reflected in the panoply of unique physiological adaptations across the phylum, including different strategies to optimize light harvesting or sustain nitrogen fixation, but also different lifestyles like psychrotrophy, and oligotrophy. Some cyanobacteria produce secondary metabolites of cryptic function, many of which are toxic to eukaryotes. Consequently, bloom-forming toxic cyanobacteria are global hazards that are of increasing concern in surface waters affected by anthropogenic nutrient loads and climate change. While focusing on cyanobacteria in aquatic environments, the collection of papers herein touches on this broad range of topics.

Regarding the role of cyanobacteria in the ocean, the importance of the unicellular picocyanobacteria is paramount. Indeed, *Synechococcus* and *Prochlorococcus* are the abundant genera in the oligotrophic open sea. Given that cyanobacteria are under regular assault imposed by photooxidative stresses, and that such stresses constrain photosynthetic performance, the paper by Mella-Flores et al. (2012) describes different strategies employed by model strains of *Synechococcus* and *Prochlorococcus* to high light and UV. Nonetheless, in the ocean, *Synechococcus* is composed of several clades that likely define particular ecotypes. Detection and enumeration of distinct *Synechococcus* clades is essential in understanding how these organisms co-exist with one another. To this end, Ahlgren and Rocap (2012) developed a suite of qPCR tools enabling a thorough analysis of *Synechococcus* community structure and dynamics.

Nitrogen fixation by the unicellular *Crocosphaera watsonii* is now recognized to be a significant source of new N to marine environments. Two papers describe physiological properties of *Crocosphaera* relevant to their ecological importance. Sohm et al. (2011) demonstrate that a strain of *C. watsonii* produces high amounts of extracellular polysaccharides. The high level of organic carbon production suggests that this strain may directly support significant heterotrophy in the marine C cycle. Given that arsenate is a toxic anion that can interfere with the transport and metabolism of phosphate, Dyhrman and Haley (2011) report mechanisms of arsenate resistance in *Crocosphaera*, whose habitat includes P-limited ocean gyres. Such regions often have high arsenate:phosphate ratios, thus detoxification mechanisms are an important strategy for N fixation in these oligotrophic regions.

The structure and physiological performance of arctic cyanobacterial mats is of particular interest, given their dominant contribution to biomass at high latitudes, and the stresses imposed by global warming on such fragile environments. Lionard et al. (2012) examine the composition of an artic microbial mat, and demonstrate the robust adaptation of the cyanobacteria to osmotic stress, a likely outcome of climate change.

The characterization of chlorophyll-*b* containing *Prochlorothrix* spp. by Pinevich et al. (2012) indicate that despite the scarcity of this genus in aquatic environments, molecular methods suggest it is more widely distributed. The authors compare the physiological characteristics of the two known species and discuss the evolution of their light harvesting apparatus.

Four papers explore the physiology of bloom-forming and toxic cyanobacteria. Kaplan et al. (2012) address the enigmatic function of the microcystins and cylindrospermopsins, describing experimental evidence for their role in cell-cell communication among bloom-formers. Sukenik et al. (2012) explore the spread of *Aphanizomenon* and *Cylindrospermopsis* to temperate freshwater environments. The appearance of such blooms is a likely consequence of global warming along with anthropogenic nutrient load, leading to a sustained invasion of into the middle latitudes. Given that management of phosphorus inputs is a strategy for limiting algal blooms in freshwaters, Saxton et al. (2012) describe the phosphorus quotas (intracellular and total) of the toxic colonial cyanobacterium, *Microcystis aeruginosa*. In response to increased phosphorus, *Microcystis* exhibited a higher cell P quota, yet a stable growth rate compared to lower phosphorus treatments. Such work has implications in modeling *Microcystis* blooms from P loadings into freshwaters. Gagnon and Pick (2012) contribute a study showing that the potent neurotoxin anatoxin-a is associated with nitrogen availability in *Aphanizomenon*. Understanding environmental cues for toxin production is essential in ultimately mitigating bloom toxigenicity.

Together, these important contributions cover many of the issues tackled in the field today: the role of cyanobacteria in the global carbon and nutrient cycles, cyanobacterial evolution, the adaptation of cyanobacteria to climate change and the consequences of toxigenicity. The studies described in this collection provide an excellent introduction to these topics, while providing important clues regarding the future directions that studies of cyanobacteria will take.

## **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: 17 June 2014; accepted: 26 June 2014; published online: 16 July 2014. Citation: Bullerjahn GS and Post AF (2014) Physiology and molecular biology of aquatic cyanobacteria. Front. Microbiol. 5:359. doi: 10.3389/fmicb.2014.00359 This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology.*

*Copyright © 2014 Bullerjahn and Post. 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.*

## Prochlorococcus and Synechococcus have evolved different adaptive mechanisms to cope with light and UV stress

**Daniella Mella-Flores 1,2,3, Christophe Six 1,2, Morgane Ratin1,2, Frédéric Partensky 1,2, Christophe Boutte1,2 , Gildas Le Corguillé1,4, Dominique Marie1,2, Nicolas Blot 1,2,5,6, Priscillia Gourvil 1,2, Christian Kolowrat 1,2,7 and Laurence Garczarek 1,2\***

<sup>1</sup> Station Biologique, UPMC-Université Paris VI, Roscoff, France


<sup>6</sup> Laboratoire Microorganismes: Génome et Environnement, CNRS, UMR 6023, Aubière, France

<sup>7</sup> Center for Doctoral Studies, University of Vienna, Vienna, Austria

#### **Edited by:**

George S. Bullerjahn, Bowling Green State University, USA

#### **Reviewed by:**

Eric A. Webb, University of Southern California, USA Tom Bibby, University of Southampton, UK

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

Laurence Garczarek, Station Biologique, UPMC-Université Paris VI, Places Georges Teissier, 29680 Roscoff, France. e-mail: laurence.garczarek@ sb-roscoff.fr

Prochlorococcus and Synechococcus, which numerically dominate vast oceanic areas, are the two most abundant oxygenic phototrophs on Earth. Although they require solar energy for photosynthesis, excess light and associated high UV radiations can induce high levels of oxidative stress that may have deleterious effects on their growth and productivity. Here, we compared the photophysiologies of the model strains Prochlorococcus marinus PCC 9511 and Synechococcus sp. WH7803 grown under a bell-shaped light/dark cycle of high visible light supplemented or not with UV. Prochlorococcus exhibited a higher sensitivity to photoinactivation than Synechococcus under both conditions, as shown by a larger drop of photosystem II (PSII) quantum yield at noon and different diel patterns of the D1 protein pool. In the presence of UV, the PSII repair rate was significantly depressed at noon in Prochlorococcus compared to Synechococcus. Additionally, Prochlorococcus was more sensitive than Synechococcus to oxidative stress, as shown by the different degrees of PSII photoinactivation after addition of hydrogen peroxide. A transcriptional analysis also revealed dramatic discrepancies between the two organisms in the diel expression patterns of several genes involved notably in the biosynthesis and/or repair of photosystems, light-harvesting complexes, CO<sup>2</sup> fixation as well as protection mechanisms against light, UV, and oxidative stress, which likely translate profound differences in their light-controlled regulation. Altogether our results suggest that while Synechococcus has developed efficient ways to cope with light and UV stress, Prochlorococcus cells seemingly survive stressful hours of the day by launching a minimal set of protection mechanisms and by temporarily bringing down several key metabolic processes. This study provides unprecedented insights into understanding the distinct depth distributions and dynamics of these two picocyanobacteria in the field.

**Keywords: marine cyanobacteria, Synechococcus, Prochlorococcus, light/dark cycle, light stress, UV radiations, oxidative stress, photophysiology**

## **INTRODUCTION**

Phytoplanktonic cells, and in particular cyanobacteria, experience dramatic daily fluctuations of solar radiations, which can become suboptimal for photosynthetic processes around midday. Photosystem II (PSII) is particularly sensitive to these changes in photon fluxes and under unfavorable or stressful conditions its activity can decline more rapidly than most other physiological processes (Berry and Björkman, 1980; Demmig-Adams and Adams, 1992; Aro et al., 1993; Andersson and Aro, 2001). Photodamages to PSII are thought to start by the inactivation of the oxygen-evolving complex, which is caused by the dissociation of the Mn4Ca2<sup>+</sup> cluster. This process leads to the production of long-lived P<sup>680</sup> +, the oxidized form of the reaction center chlorophyll (Chl) pair, a particularly strong oxidant which in turn provokes the destruction of the PSII core protein D1 (Hakala et al., 2006; Nishiyama, 2006). At low irradiances, the rate of photosynthetic electron transport is proportional to the photon flux density and damaged D1 polypeptides can be removed from the PSII reaction center and rapidly replaced by newly synthesized D1 proteins (Park et al., 1995; Tyystjarvi and Aro, 1996; Nixon et al., 2005; Ohnishi et al., 2005). However, at higher irradiances, the rate at which the PSII reaction center is damaged can exceed its repair rate, which results in an

increase of inactivated PSII centers and a subsequent decline of the quantum yield of photosynthesis, resulting from photoinhibitory fluorescence quenching (Powles, 1984; Prásil et al., 1992;Aro et al., 1993; Andersson and Aro, 2001).

Although the visible part of the solar spectrum (400–700 nm), also called photosynthetically active radiations (PAR), is responsible for most photoinhibitory effects, the contribution of UV-B (280–315 nm) and, to a least extent, UV-A (315–400 nm) is also notable in the uppermost layer of the ocean (Dring et al., 2001; van de Poll et al., 2001; He and Häder, 2002b). UV-B can indeed damage the photosynthetic apparatus about 100-fold more efficiently than visible light and these radiations might directly affect PSII proteins and the Mn4Ca2<sup>+</sup> cluster (Sarvikas et al., 2006; Caldwell et al., 2007). UV and high visible radiations can also cause indirect photoinhibitory effects via the production of reactive oxygen species (ROS; He and Häder, 2002a,b; Rastogi et al., 2010), mainly formed within reaction centers (Asada, 1999) and light-harvesting complexes (Knox and Dodge, 1985; Zolla and Rinalducci, 2002). ROS are powerful oxidizing agents which can react with DNA, lipids, and proteins. Although these compounds are inevitably produced by cell metabolism, even under optimal growth conditions, their production is drastically enhanced when cells are exposed to a variety of stresses, including excess visible light and UV radiations (UVR; Latifi et al., 2005; Ross et al., 2006; Houot et al., 2007; Allakhverdiev and Murata, 2008). The effect of ROS on PSII photoinhibition is thought to act primarily by inhibiting the *de novo* synthesis of proteins, including those required for the repair of PSII (Nishiyama et al., 2004; Nishiyama, 2006; Takahashi and Murata, 2008). A direct effect of ROS on the inactivation of PSII reaction center has also been suggested through triggering D1 degradation (Vass et al., 1992; Aro et al., 1993; Miyao et al., 1995; Keren et al., 1997; Lupinkova and Komenda, 2004). In any case, ROS clearly have a major role in light-mediated photoinhibition as well as in other environmental stresses (Nishiyama, 2006; Allakhverdiev and Murata, 2008; Latifi et al., 2009). Thus, survival of phototrophic organisms depends upon the amount of ROS produced and their efficiency in scavenging these oxygen species.

In this context, marine picocyanobacteria belonging to the genera *Synechococcus* and *Prochlorococcus* constitute two relevant and complementary models to study acclimation processes to high light and UVR and their interrelationships with oxidative stress. In oceanic ecosystems, these two organisms numerically dominate the phytoplanktonic community (Partensky et al., 1999a; Scanlan, 2003) and are considered to be the two most abundant photosynthetic organisms on Earth, with a substantial contribution to Chl biomass and primary production (Liu et al., 1997; Partensky et al., 1999a; Agawin et al., 2000; Garcia-Pichel et al., 2003). Members of the marine *Synechococcus* genus are ubiquitously distributed and are most abundant in coastal regions and mesotrophic open ocean surface waters (Partensky et al., 1999a; Zwirglmaier et al., 2008), whereas *Prochlorococcus* preferentially thrives in warm, stratified, oligotrophic tropical, and subtropical marine areas (Partensky et al., 1999b; Zubkov et al., 2000; Johnson et al., 2006). In the field, these organisms experience large variations in irradiance, linked to the combination of the light/dark (L/D) cycle, water mixing, and a variable cloudiness (MacIntyre et al., 2000). Moreover, their tiny size (0.5–0.8 and 0.8–1.2µm

diameter for *Prochlorococcus* and *Synechococcus*, respectively) confers them a high surface to volume ratio, optimizing their photon capture, and making them particularly sensitive to UVR (Llabres and Agusti, 2006, 2010).

Like other photosynthetic organisms, marine cyanobacteria have evolved a variety of protection mechanisms to ensure their growth and survival in highly illuminated habitats. These mechanisms include thermal dissipation of excess light excitation, structural changes of the photosynthetic machinery as well as enzymatic and non-enzymatic scavenging systems to eliminate ROS, in particular those produced in photosynthetic membranes (for reviews, see Bailey and Grossman, 2008; Latifi et al., 2009). However, several pieces of evidence suggest that *Prochlorococcus* and *Synechococcus* lineages could deal differently with light stress. Indeed, two *P. marinus* strains (PCC 9511 and SS120, a high light- and a low-light-adapted ecotype, respectively) were found to be more sensitive to a transient exposure to high irradiances than three *Synechococcus* spp. strains representative of various trophic environments and exhibiting different pigmentation (RS9917, RCC307, and WH8102; Six et al., 2007b). Similarly, measurements of cell abundances and/or mortality rates of field populations of picocyanobacteria exposed to different levels of natural solar radiations showed that *Prochlorococcus* exhibited a lower resistance to UVR than *Synechococcus* in surface waters of the central Atlantic Ocean (Llabres and Agusti, 2006; Agusti and Llabres, 2007) and the Mediterranean Sea (Sommaruga et al., 2005; Llabres and Agusti, 2010).

In order to reveal potential differences in circadian metabolic rhythms between these two genera, the photophysiology of the model strains *P. marinus* PCC 9511 and *Synechococcus* sp. WH7803 was examined at different times of a modulated L/D cycle of visible light (hereafter VL) with or without UV. Additionally, the diel variability of the sensitivity of *Prochlorococcus* and *Synechococcus* to oxidative stress, as triggered by different H2O<sup>2</sup> concentrations was investigated. Expression of key genes involved in photosynthesis, light, and oxidative stress response and a number of other processes were also monitored in order to get insights about the molecular bases of the observed physiological differences.

## **MATERIALS AND METHODS**

## **STRAINS AND CULTURE CONDITIONS**

The two model strains *P. marinus* PCC 9511 [a strain genetically very close, if not identical, to MED4 (Rippka et al., 2000; Rocap et al., 2003)] and *Synechococcus* sp. WH7803 (Scanlan et al., 2009) used in this study were grown at 22˚C in PCR-S11 medium (Rippka et al., 2000), supplemented with 1 mM NaNO3. All experiments were performed under modulated L/D conditions using a computer-controlled illumination device (cyclostat), as detailed elsewhere (Holtzendorff et al., 2001; Kolowrat et al., 2010). This system allows the simulation of a bell-shaped 12/12 h L/D cycle, which induces a good synchronization of cell division, as observed for field populations (Vaulot et al., 1995; Jacquet et al., 1998). The maximal VL irradiance (at virtual noon) was set at 870µmol photons m−<sup>2</sup> s −1 , corresponding to the reference light condition. In order to test the specific effects of UVR, the same experiments were repeated but with supplementing the VL

condition with modulated UV-A and UV-B radiations (hereafter VL + UV), provided by UV-A-340 fluorescent lamps (Q-Panel Lab products, Cleveland, OH, USA). UVR reached 7.59W m−<sup>2</sup> UV-A (320–340 nm) and 0.57W m−<sup>2</sup> UV-B (280–320 nm) at virtual noon, corresponding to levels representative of natural doses measured in the upper layer of nutrient-poor, oceanic areas (Helbling et al., 1992). Two replicate cultures were acclimated to L/D cycles for at least 2 weeks prior to start monitoring the different parameters. During the experiment, two replicate cultures were grown with a continuous input of fresh medium,in order to maintain cells in exponential growth throughout the whole sampling period. One microliter aliquots were taken every hour, fixed for 10 min with grade I glutaraldehyde (0.25% final concentration; Sigma Aldrich, Saint-Louis, MO, USA), then frozen at −80˚C for delayed analyses of cell abundance and cell cycle using a FACS Canto flow cytometer (Becton Dickinson Biosciences, San Jose, CA, USA), as previously described (Marie et al., 1997, 1999). Because of the continuous dilution, growth rate of these cultures was indirectly assessed from cell cycle data (µcc) using the method described by Carpenter and Chang (1988), as detailed earlier (Kolowrat et al., 2010). To study the kinetics of response to light fluctuations, cultures were sampled at 6, 9, 12, 15, 18, 20, 22, and 2 h over 3 days for measuring a variety of parameters described below.

### **PIGMENT AND FLUORESCENCE MEASUREMENTS**

Photosynthetic pigments were extracted in 95% methanol and analyzed by HPLC, as previously described (Everroad et al., 2006). Whole cell fluorescence emission spectra with excitation at 530 nm were recorded for *Synechococcus* using a LS-50B spectrofluorometer (Perkin Elmer, Waltham, MA, USA). The PSII quantum yield (*F*V/*F*M) was measured using a Pulse Amplitude Modulated fluorimeter (PhytoPAM, Walz, Effeltrich, Germany) connected to a LabPro chart recorder allowing the direct visualization of fluorescence traces (Vernier, Beaverton, OR, USA). A 2 mL aliquot was dark acclimated for 5 min in a quartz cuvette with a mirrored facet facing the photomultiplier to enhance the signal. The modulated light was then turned on to measure the basal fluorescence level F<sup>0</sup> with modulated excitation at 440 nm for *Prochlorococcus* cells and 520 nm for *Synechococcus*. The maximal fluorescence *F*<sup>M</sup> was determined by applying a light saturating pulse in the presence of 50µM of the PSII inhibitor 3-(30,4-dichlorophenyl)-1,1 dimethylurea (DCMU) under ca. 2,000µmol photons m−<sup>2</sup> s −1 . The PSII quantum yield was calculated as:

$$F\_{\sf V}/F\_{\sf M} = (F\_{\sf M} - F\_{\sf O}) / F\_{\sf M} \tag{1}$$

where *F*<sup>V</sup> is the variable fluorescence.

#### **PHOTOSYSTEM II REPAIR**

At each sampling time point, a 20 mL volume of each replicate culture was sampled and split into two 100 mL quartz Erlenmeyer flasks, and one of them was supplemented with 500µg mL−<sup>1</sup> lincomycin, an inhibitor of protein translation (Six et al., 2007a). The two flasks were immediately brought back into the culture system and 2 mL aliquots were collected at 0, 15, 30, and 60 min after the sampling to measure the PSII quantum yield, as described above. The PSII quantum yield was then plotted over time for both the control and lincomycin-treated sub-cultures and plots were fitted with an exponential decay function. The PSII repair rate was estimated as the difference between the exponential decay rates in the absence and presence of lincomycin (Six et al., 2007a, 2009; Campbell and Tyystjarvi, 2012).

## **IMMUNOBLOTTING**

Cell pellets were resuspended in extraction buffer (140 mM Tris base, 105 mM Tris-HCl, 0.5 mM ethylenediaminetetraacetic acid, 2% lithium dodecyl sulfate, 10% glycerol, and 0.1 mg mL−<sup>1</sup> Pefa-Bloc protease inhibitor (Roche, Basel, Switzerland) and total protein concentration was determined using a Lowry protein assay kit (Bio-Rad, Hercules, CA, USA) and bovine serum albumin as protein standards. Samples were then denatured with 50 mM dithiothreitol and heated for 2 min at 80˚C and 2µg total protein was loaded on a 4–12% gradient acrylamide precast NuPAGE Bis-Tris mini-gel (Invitrogen, Carlsbad, CA, USA) along with recombinant standards of D1 or D2 proteins (Agrisera, Vännäs, Sweden) to establish a standard curve. Gels were electrophoresed and the proteins were transferred onto a polyvinylidene fluoride (PVDF) membrane, then immediately immersed into Tris Buffer Saline-Tween (TBS-T) buffer, pH 7.6 (0.1% Tween 20, 350 mM sodium chloride, 20 mM Trizma base) containing 2% (w:v) ECL Advance blocking agent (Amersham Biosciences, Piscataway, NJ, USA). Aliquots of primary antibodies against D1 or D2 proteins (Agrisera) were diluted at 1:50 000 in TBS-T in the presence of 2% blocking agent and membranes were soaked into this solution for 1 h with agitation. After extensive washing of the membrane with TBS-T buffer, anti-rabbit secondary antibodies were applied with the same procedure as for primary antibodies. Membranes were developed by chemoluminescence using the ECL Advance reagent kit (Amersham Biosciences) and visualized with a LAS4000 imager equipped with a CCD camera (GE Healthcare, Waukesha, WI, USA). Signals were quantified using the ImageQuant software and the recombinant protein standard curve. To ease comparisons of the diel trends between strains and conditions, data were normalized at 6:00 am.

#### **OXIDATIVE STRESS ASSAYS**

Samples collected from dawn to dusk (6, 9, 12, 15, and 18 h time points) were split into aliquots of 2.5 mL and subjected to a series of H2O<sup>2</sup> concentrations ranging from 0.1 to 1,800µM. These culture aliquots were then incubated in the culturing system for 50 min and the PSII quantum yield was measured as described above. To quantitatively estimate the PSII resistance to such oxidative stress at each time point, the decay of *F*V/*F*<sup>M</sup> as a function of increasing H2O<sup>2</sup> concentration was plotted as a percentage of *F*V/*F*<sup>M</sup> of the control cultures. The decay curves were then fitted with a two-parameter hyperbolic decay function:

$$y = ab/\left(b+\mathbf{x}\right)\tag{2}$$

The *b*-value proved to be informative because it integrates both the minimal H2O<sup>2</sup> concentration needed to cause a decrease of *F*V/*F*<sup>M</sup> and the decrease rate of the PSII quantum yield. Thus, this parameter can be used as a proxy to assess the global resistance of PSII to oxidative stress induced by H2O2.

## **RNA EXTRACTION AND REAL TIME QUANTITATIVE PCR**

Samples for RNA extraction were harvested and extracted as previously described in Kolowrat et al. (2010) for eight data points per L/D cycle and for two different days, corresponding to biological replicates. Briefly, cell pellets, resuspended in Trizol (Invitrogen, Carlsbad, CA, USA), were extracted using the miRNeasy kit as recommended by the manufacturers (Qiagen, Valencia, CA, USA), followed by two successive DNase treatments performed on the miRNeasy columns using the Qiagen RNase-free DNase Set (Qiagen).

Real time quantitative PCR (hereafter qPCR) was performed on a set of *P. marinus* PCC 9511 and *Synechococcus* sp.WH7803 genes, representative of key metabolic processes (Table S1 in Supplementary Material). Design and optimization of gene specific primers were performed as previously described (Six et al., 2007b) using PrimerExpress™ software v2.0 (Applied Biosystems) and by checking for every set of primers, the linearity of the CT (cycle at threshold) vs. cDNA content within a dilution range of cDNA. Reverse transcription was carried out on 100 ng RNA using SuperScriptII reverse transcriptase (Gibco-BRL, Gaithersburg,MD, USA). qPCR was performed in triplicate on the cDNA obtained after dilution, using the DNA Engine/Chromo4 Real Time PCR-Detector (Bio-Rad, Hercules, CA, USA) and the absolute SYBR Green ROX Mix (Abgene, Epsom, UK), as previously described (Garczarek et al., 2008). Gene expression profiles monitored during L/D cycles were expressed as the ratio of gene expression vs. expression of the aperiodic gene *rnpB* (Mary and Vaulot, 2003; Zinser et al., 2009; Kolowrat et al., 2010) and normalized to the 6:00 time point sampled under VL conditions, using the 2−∆∆CT method (Schmittgen and Livak, 2008).

## **RESULTS**

#### **CELL CYCLE**

The cell cycle of *Synechococcus* sp. WH7803 was strongly synchronized by the alternation of light and darkness, with a peak of DNA replicating cells (S cells) occurring at ca. 16–17:00, i.e., 1–2 h before the light-to-dark transition (LDT; 18:00), in modulated VL only and about 3 h later for cells acclimated to modulated VL + UV (**Figure 1**). A comparable delay of the S phase was previously reported for *P. marinus* PCC 9511 in the presence of UVR, but the peak of S cells occurred at the LDT in VL and about at the same time as for *Synechococcus* in VL + UV (Kolowrat et al., 2010). Another notable difference with *Prochlorococcus* is the occurrence of a second minor S peak in the early morning followed by a small bump of G<sup>2</sup> cell abundance around noontime (mainly visible during the first and third cycle), suggesting some ultradian growth (Shalapyonok et al., 1998). Mean growth rates of *Synechococcus* cultures were assessed from the percentages of cells in S and G<sup>2</sup> (µcc) using the method described by Carpenter and Chang (1988). They were not statistically distinct between the two conditions (average over 3 days and two replicates: µcc = 1.00 ± 0.21 day−<sup>1</sup> in VL; µcc = 1.23 ± 0.11 day−<sup>1</sup> in VL + UV) but were higher than in *Prochlorococcus* (µcc = 0.67 ± 0.05 day−<sup>1</sup> in VL; µcc = 0.68 ± 0.03 day−<sup>1</sup> in VL + UV; Kolowrat et al., 2010). This difference is attributable in part to the shorter delay observed between the maxima of S and G<sup>2</sup> cells in *Synechococcus* (tG2 − t<sup>S</sup> ∼ 2.5 h) than in *Prochlorococcus*

(tG2 − t<sup>S</sup> ∼ 4 h), as this parameter is used for the calculation of growth rates in the µcc method (Carpenter and Chang, 1988).

## **PIGMENT RATIOS**

Comparative diel variations of molar ratios of the main pigments in the two marine picocyanobacteria under both light conditions are reported in **Figure 2**. For *Synechococcus* grown under VL, the zeaxanthin (Zea) to Chl *a* ratio (**Figure 2A**) was maximal between noon and the early afternoon, and then decreased sharply till the LDT and at a lower rate during the night. In VL + UV, the pattern was globally similar, except that (i) the values were higher in VL + UV than in VL, (ii) the midday peak had a lower average amplitude (15.4 vs. 25.4% increase under VL + UV and VL, respectively), and (iii) the ratio remained stable during the night. By comparison, the β-carotene (β-Car) to Chl *a* ratio of *Synechococcus* grown in VL systematically peaked at 9:00, decreased till 15:00 then increased again for the rest of the diel cycle (**Figure 2C**). The pattern was very similar under UV, except that night values were slightly lower during the first two L/D cycles. The pattern of pigment ratios observed in *Prochlorococcus* exhibited a number of differences compared to *Synechococcus*. In VL, the Zea to divinyl-(DV-) Chl *a* ratio started to increase at mid-morning and was maximal around 20:00, then decreased during the rest of the night (**Figure 2B**), corresponding to the cell division phase. The same diel pattern was observed under VL + UV. Although the variation range was moderate for the α-Car to DV-Chl *a* ratio in VL, its diel pattern was comparable to that of the Zea to DV-Chl *a*, except that it increased immediately after dawn (**Figure 2D**). There were almost no diel oscillations of this ratio under VL + UV. The DV-Chl *b* to *a* ratio exhibited no clear diel pattern in either light condition (data not shown).

## **PHOTOSYSTEM II FUNCTION AND REPAIR**

In order to investigate how the daily variations of VL and UVR could affect the PSII activity of cells, we followed the PSII quantum yield in *Synechococcus* sp. WH7803 and *P. marinus* PCC 9511 cultures during three consecutive L/D cycles (**Figure 3**). For both strains and under both light conditions, the *F*V/*F*<sup>M</sup> ratio showed a cyclic evolution, reaching a maximum at night and a minimum at virtual noon. This midday drop, originating either from PSII photoinactivation or from dissipative non-photochemical quenching (NPQ) of fluorescence (or both), was larger for *Prochlorococcus* than for *Synechococcus* under both light conditions. Indeed, there was a relative decrease of the PSII quantum yield of about 40% for *Prochlorococcus*, while in *Synechococcus* it never exceeded 20% (**Figure 3**). The diel patterns of PSII activities showed also some differences between VL and VL + UV, with slightly higher late night and morning yields for *Synechococcus* and lower yields at 6:00, 12:00, and 15:00 for *Prochlorococcus* cells in the latter condition.

Variations in the PSII repair activity of the cells, as assessed by comparing the PSII quantum yield of sub-cultures incubated with or without lincomycin, an inhibitor of protein synthesis, also showed a cyclic pattern, with no measurable PSII repair occurring during the night (**Figure 4**). Moreover, cells acclimated to VL and VL + UV conditions exhibited a very different repair rate,

especially in *Synechococcus*. While VL exposure only led to a moderate induction of the PSII repair rate with regard to night levels in *Synechococcus*, a sharp increase of this rate (from three- to six-fold) was observed for cells acclimated to UV-supplemented light (**Figure 4A**). Such an UV-induced increase of the daily PSII repair rate was also observed in *Prochlorococcus* cells, although much more limited (less than twofold), and the maximal rate were already reached at 9:00 (**Figure 4B**).

## **PHOTOSYSTEM II CORE PROTEIN POOLS**

In *Synechococcus* cells grown under VL, both the D1 and D2 protein pools showed no significant diel oscillations during the L/D cycle (**Figure 5**). Under VL + UV, while D2 also remained stable, cells progressively accumulated D1 proteins during daytime and the pool reached a maximum at the LDT (40% increase compared to the value at 6:00). *Prochlorococcus* cells showed a quite different diel pattern of D1 and D2 protein pools. Under VL, both proteins showed minimal contents during the day (30–40% decrease),when irradiance was maximal,with a subsequent full recovery during the

night. A similar pattern was observed under VL + UV, but with an extended period of low PSII core protein content.

## **RESISTANCE OF PSII TO OXIDATIVE STRESS**

*Synechococcus* and *Prochlorococcus* grown under VL and VL + UV conditions exhibited quite different PSII responses to oxidative stress (**Figure 6**). *Synechococcus* sub-cultures sampled at 6:00 (i.e., at the dark-to-light transition) showed no significant change in their PSII quantum yield until H2O<sup>2</sup> concentrations as high as 300 and 1,800µM in VL + UV and VL, respectively. Sub-cultures sampled at 12:00 were affected by concentrations as low as 10 and 50µM H2O2, the photoinhibitory effect increasing with the concentration of oxidizing agent until a total shutdown of PSII activity at 300 and 900µM H2O<sup>2</sup> for VL + UV and VL, respectively (**Figure 6A**). PSII inactivation occurred at lower doses in *Prochlorococcus* at both time points, starting at about 100µM H2O<sup>2</sup> for cultures collected at 6:00 and reaching 60 and 50% of photoinhibition at 1,800µM H2O<sup>2</sup> in VL and VL + UV, respectively (**Figure 6B**), compared to less than 20% for *Synechococcus*

cells at the same time point. Furthermore, as for *Synechococcus*, *Prochlorococcus* sub-cultures sampled at noon were sensitive to lower H2O<sup>2</sup> concentrations than cultures sampled at 6:00.

In order to compare the rates of decay of the PSII quantum yield obtained at various times of the day, *F*V/*F*<sup>M</sup> vs. H2O<sup>2</sup> concentration curves for all time points (only those obtained for 6:00 and 12:00 are shown in **Figures 6A,B**) were fitted with a twoparameter hyperbolic decay function. The daily variations of the *b*-value, characterizing the global resistance of PSII to oxidative stress induced by H2O<sup>2</sup> (see Materials and Methods), are shown in **Figures 6C,D**. As expected from results at 6:00 and 12:00 (**Figures 6A,B**), PSII proved to be much more sensitive to an artificially induced oxidative stress during the day than during the night in both picocyanobacteria. Still, *Synechococcus* PSII was seemingly much more resistant than *Prochlorococcus* to this stress, when applied in the dark. It is also worth noting that the sensitivity to oxidative stress was enhanced for UV-acclimated cells, especially in *Prochlorococcus*, which displayed a dramatic drop of the *b*-value at 09:00 compared to those grown in VL only. However, after this sharp mid-morning drop, the *b*-value rose again during the day in both strains.

#### **TRANSCRIPTOMIC RESPONSE**

#### **Photosynthesis**

The diel expression pattern of a selection of genes involved in photosynthesis and a number of other processes related to (or affected by) light or UV stress were analyzed by qPCR in *P. marinus* PCC 9511 and *Synechococcus* sp.WH7803 strains, grown under bothVL and VL + UV conditions (**Figure 7**). As expected, photosynthetic genes were among those showing the strongest diel oscillations. In both strains, the daily variations of the total pools of *psbA* and *psbD* transcripts, encoding the two major subunits of the reaction center II (D1 and D2, respectively), closely matched the modulated fluctuations of growth irradiance, with a slight increase in the relative amplitude of the peak of *psbA* transcripts under UVR. It is worth noting that in the case of *Synechococcus*, which possesses four *psbA* genes (compared to only one in *P. marinus* MED4/PCC9511; Garczarek et al., 2008), this global expression pattern translates the temporal succession of the single copy gene encoding the D1.1 isoform (*SynWH7803\_0784*) and the three copies encoding the D1.2 isoforms (*SynWH7803\_0790*, *0366*, and *2084*), the former being repressed at noon and maximum at the LDT, while the three latter exhibited an opposite behavior (note that the expression levels of *SynWH7803\_0366* and *2084* cannot be measured separately). Accordingly, the expression pattern of *PMM0743* and *SynWH7803\_1216*, encoding the FtsH2 protease that is involved in the clearance of damaged D1 proteins from inactivated PSII (Komenda et al., 2006, 2010), showed a broad maximum during the period of highest photon fluxes (9:00–15:00) in both species. However, in *Prochlorococcus*, this gene showed a minimum at dusk, then a continuous increase during the dark period and the early morning, while in *Synechococcus* its relative expression levels were minimal all over the dark period from 18:00 to 6:00.

In contrast to reaction center II genes, the diel expression pattern of *psaB*, encoding one of the two major subunits of the PSI core, was very different between the two cyanobacteria. While in *Synechococcus* its expression level peaked between 15:00 and 18:00 and remained low for the rest of the L/D cycle, *Prochlorococcus*

*psaB* transcript level was at its lowest around noontime and was maximal during the night period. Genes encoding light-harvesting systems also behaved quite differently between *Prochlorococcus* and *Synechococcus*. In the former organism, *pcbA* transcripts displayed a comparable pattern during the day to that of *psaB* transcripts with a significant downregulation in both light conditions, though the relative level of expression was even lower at noon for UVirradiated cells. In contrast, for *Synechococcus* in VL, all genes encoding the different phycobiliprotein α-subunits (*apcA*, *rpcA*, *cpeA*, and *mpeA*) and phycoerythrin (PE) linkers (*cpeE*, *mpeD*, *mpeE*) were highly upregulated during the light period, with maximal expression at 15:00. While the expression patterns of genes coding for allophycocyanin, phycocyanin, and PEI subunits and their linkers were little affected by UVR, the expression of PEII genes showed a drop at noon. Interestingly, the *mpeE* gene, encoding the linker binding the distal PEII disk in this strain (Six et al., 2007c), even exhibited a 3-h delay of its maximal expression level in VL + UV compared to VL only.

Pigment biosynthesis genes also showed large variations of relative expression over the day under both light conditions. The *chlG* gene, encoding the chlorophyll *a* synthase, also behaved in an opposite way between *Prochlorococcus* and *Synechococcus*, with minimal and maximal expression levels at noon, respectively, in both cases more marked for UV-acclimated cells. Accordingly, the diel expression pattern of *PcCao*, a *Prochlorococcus*-specific gene encoding the chlorophyll *b* synthase (Satoh and Tanaka, 2006), also somewhat mirrored diel variations of growth irradiances. In contrast, the *ctrR* gene, encoding the β-Car hydroxylase that catalyzes the last biosynthesis step of zeaxanthin, usually considered as a photoprotective pigment, was maximally expressed during daytime in both organisms, but with a different global pattern, since in *Prochlorococcus* there was a strong surge in expression levels at 9:00 am followed by a progressive decline for the rest of the day, whereas in *Synechococcus* the diel pattern again closely matched the growth irradiance oscillations.

Most marine *Synechococcus* strains, including WH7803, contain the *ocp* gene, which encodes the orange carotenoid protein (OCP), a pigment-protein complex likely located between the phycobilisomes (PBS) and the PSII reaction center and involved in the dissipation of excess energy as heat (Boulay et al., 2008; Kirilovsky and Kerfeld, 2012). In VL, WH7803 *ocp* was maximally expressed during the day with a peak at 9:00, whereas the expression maximum was considerably higher and broadened toward noontime in UV-acclimated cells, suggesting a key role of this gene in the protection against UVR. Although all *Prochlorococcus* strains lack *ocp*, due to the absence of functional PBS, high light-adapted *Prochlorococcus* strains (including MED4/PCC9511), low-light-adapted ecotype LLI (Scanlan et al., 2009) as well as a few marine *Synechococcus* strains (though not WH7803) possess another potentially important gene for photoprotection, *ptox*, which encodes a plastid terminal oxidase. This enzyme was suggested to extract electrons from the electron transport chain between PSII and PSI and to re-oxide the plastoquinone pool reduced by PSII (Bailey et al., 2008; Berg et al., 2011).

Interestingly,*ptox* exhibited very large variations in relative expression with a maximum at 15:00 (**Figure 7**), though contrary to *ocp*, its diel expression pattern was virtually identical between VL and VL + UV.

The ATPase genes *atpD* et *atpH* showed similar expression patterns in *Synechococcus* with high values from 9:00 to 15:00 in VL and even higher underVL + UV,whereas expression of these genes decreased all over the day in *Prochlorococcus* followed by a night recovery (**Figure 7**; see also Figure S3 in Kolowrat et al., 2010). An opposite behavior between the two strains was also observed for the *rbcL* gene, coding for the large subunit of RuBisCO (ribulose-1,5-bisphosphate carboxylase/oxygenase), the major enzyme of the carbon fixation process, except that in *Prochlorococcus*, there was a much sharper drop of expression levels than for *atp* genes at 15:00 in VL, and at 18:00 under VL + UV (**Figure 7**).

The strikingly different diel expression patterns observed between *Prochlorococcus* and *Synechococcus* for genes involved in light-harvesting, ATP formation, as well as the light-independent reactions of photosynthesis (Calvin–Benson cycle) suggest that

most photosynthetic processes are probably controlled by distinct regulation networks in these two cyanobacteria, despite their close phylogenetic relatedness.

### **Redox and ROS detoxification**

Most genes involved in the regulation of redox state and ROS detoxification pathways that are shared by *Synechococcus* sp. WH7803 and *P. marinus* PCC 9511 were differentially regulated during the L/D cycle in both picocyanobacteria and their diel expression patterns were comparable in VL and VL + UV conditions. However, the daily amplitudes of variations of ROS genes were often much larger in *Synechococcus* than in *Prochlorococcus*. Three of the most differentially expressed genes in the former organism, i.e., those coding for rubredoxin (*rub*), glutathione peroxidase (*SynWH7803\_0151*), and a 2-cys peroxiredoxin (*2-Cys prx*), had their maximal daily expression around noontime, again suggesting a tight regulation by light, and their relative expression level significantly increased under UV. In contrast, in *Prochlorococcus*, orthologs of these three genes (*PMM0295*, *PMM1006*, and *PMM0856*, respectively) showed no clear diel pattern. Interestingly, for each cyanobacterium, the different peroxiredoxin gene copies exhibited diel patterns very distinct from one another.

*Synechococcus* sp. WH7803 and *P. marinus* PCC 9511 also possess specific sets of genes involved in ROS-scavenging systems. Among these strain-specific genes, it is worth noting that the catalase-peroxidase gene *katG* of *Synechococcus* was notably upregulated during the day, with a peak at 15:00, while *Prochlorococcus PMM0211*, a NADPH-dependent thioredoxindisulfide reductase (NTR system) was by far the most differentially expressed ROS-scavenging gene of this organism, with a maximal expression at 18:00 in VL. At last, while the only *sod* gene present in *Prochlorococcus*, *sodN*, encoding a Ni-binding superoxide dismutase (SOD), was slightly downregulated at noon relative to nighttime in VL and VL + UV, *Synechococcus sodB*, encoding a Fe-binding SOD, was in contrast strongly upregulated during the light period under both conditions. The second *sod* gene of *Synechococcus*, *sodC* that encodes a Cu/Zn-binding SOD, showed only faint diel variations of its expression in VL, but was slightly downregulated at 20:00 in VL + UV.

#### **Cell cycle, DNA repair, and circadian clock**

In *Synechococcus* sp. WH7803, the DNA replication initiation factor gene *dnaA* was upregulated during the day with a broad peak of expression during the 9:00–15:00 time period in both light conditions. Subtle differences can however be noted between the two conditions, since the expression maximum seemingly occurred at 9:00 in VL and at noon in VL + UV (**Figure 7**). The *ftsZ* gene, which controls the biosynthesis of the Z-ring formed in the middle of the cell prior to cell division, showed very similar diel expression patterns to *dnaA*, except for systematically lower amplitudes of variations over the L/D cycle. The last cell cycle gene that was examined, *sepF*, which codes for a protein interacting with FtsZ during septum formation, showed a bell-shaped diel expression pattern peaking at 15:00 in both light conditions, though relative mRNA levels were again slightly higher in UV-acclimated cells.

In *P. marinus* PCC 9511 grown in VL, the *ftsZ* gene expression level peaked at 15:00, whereas *dnaA* and *sepF* both reached a maximum 3 h later. The relative expression levels of all three genes were considerably lower during the day when cells were grown under VL + UV, as previously reported (Kolowrat et al., 2010).

Since UVR are known to have deleterious effects on DNA structure, we also examined the diel expression patterns of a few

genes involved in DNA repair pathways. The *recA* gene, which encodes an ATPase involved in the repair of double-strand breaks (DSBs) by homologous recombination (Chen et al., 2008), showed a broad maxima during daytime in *Synechococcus* cells grown in VL, while the relative expression of this gene was enhanced under VL + UV and remain high until 20:00 (**Figure 7**). In contrast, in *Prochlorococcus*, the *recA* expression peak occurred at the LDT in VL and was delayed by 2 h in the presence of UVR. The *uvrA* gene, encoding one subunit of the excinuclease UvrABC, an enzyme of the nucleotide excision DNA repair (NER) pathway, exhibited a sharp increase of its expression level during the lit period in both picocyanobacteria, with a peak at noon under both VL and VL + UV conditions. However, the maximal *uvrA* expression level was higher in UV-acclimated *Synechococcus* cells than in VL, whereas in *Prochlorococcus*, patterns were comparable between the two conditions (except at 9:00; **Figure 7**; Kolowrat et al., 2010).

Both picocyanobacteria also possess a family of genes related to DNA photolyases. In *Synechococcus*, the pattern of the photolyase gene *phrA* (Ng et al., 2000) closely matched the L/D cycle, with a peak at noon, which was threefold higher in VL + UV with regard to VL only (**Figure 7**). On the contrary, in *Prochlorococcus*, the *phrA* expression peak heights were similar in the two

light conditions, even though the relative *phrA* mRNA level was higher at mid-morning in VL than VL + UV (Kolowrat et al., 2010). Like other marine *Synechococcus* but unlike *Prochlorococcus*, WH7803 possesses a second member of the DNA photolyase family (PhrB), which was recently suggested to be a cryptochrome (Goosen and Moolenaar, 2008). Interestingly, its diel expression pattern was quite similar to *phrA*, but with a reduced amplitude of variation of the relative mRNA amounts during daytime, even if there was again a very strong relative increase of the expression maximum under UV. The same was observed for a gene encoding an uncharacterized photolyase relative in *Synechococcus* (*SynWH7803*\_*1271*), while its ortholog in *Prochlorococcus* (*PMM0425*) showed no clear diel pattern. Other, shorter members of the Phr family were also analyzed. Indeed, *Synechococcus* possesses a cluster of two genes (*SynWH7803\_0266* and *0267*), each of which corresponds to one of the two domains of photolyases (photolyase and FAD-binding domains, respectively). So, once translated, they might potentially form an additional heterodimeric photolyase complex, as suggested by their diel pattern similar to *phrA* and high relative expression level (**Figure 7**). Interestingly, *Prochlorococcus* also possesses a gene (*PMM1360*) encoding a FAD-binding domain-containing protein, but surprisingly

no gene coding for the other half of a photolyase (Scanlan et al., 2009; Partensky and Garczarek, 2010). Again, this gene had a very similar diel pattern as *phrA* from the same organism but with a much lower expression level.

The diel expression of genes involved into the circadian clock machinery was also studied given their central role in controlling global rhythmic transcriptional activity of the cells. While all marine *Synechococcus* possess the three *kai* genes coding for the core circadian oscillator, *Prochlorococcus* lacks *kaiA* and its circadian clock rather behaves like an "hourglass" that is reset every morning (Holtzendorff et al., 2008; Axmann et al., 2009). Interestingly, *kai* genes had a completely different behavior between the two cyanobacteria. In VL, all three *kai* genes of *Synechococcus* sp. WH7803 exhibited large diel variations with a broad maximum relative expression level at the end of the light period (15:00–18:00), while in *Prochlorococcus*, the expression of *kaiB* was minimal at noon, then increased continuously until dawn and there was no significant diel oscillation of *kaiC* mRNA. Presence of UVR caused a strong decrease of the level of expression of *kaiABC* genes at 15:00 in *Synechococcus* leading to a sharpening of their expression peaks, whereas in *Prochlorococcus*, the *kaiB* gene was further downregulated at noon without time shift. We also looked at the expression of *rpaA*, which codes for a DNA-binding protein acting as the response regulator of the KaiC-interacting kinase *SasA*, a two-component system that mediates the diel oscillations generated by KaiC phosphorylation to global transcription rhythms (Takai et al., 2006). RpaA is also known to be involved in the regulation of energy transfer from PBS to PSI and to interact with ferredoxin (Ashby and Mullineaux, 1999; Hanke et al., 2011). In both organisms, *rpaA* was maximally expressed around noontime and UVR had no significant effect on its diel transcription pattern.

Two genes encoding type II σ factors (*rpoD4* and *rpoD8*), that have been shown to modulate gene expression under different conditions and to respond to L/D stimuli (Imamura et al., 2003; Summerfield and Sherman, 2007), were also analyzed to assess their potential role in controlling global gene expression during L/D transitions. In *Synechococcus*, *rpoD4* exhibited a comparable diel expression pattern in VL and VL + UV, with a continuous increase during the day, a peak at the LDT then a progressive night decrease. In contrast, *rpoD8* showed only mild variations in VL while its diel pattern was prominent in UV-acclimated cells, with a strong upregulation during the day and a downregulation during the night. In *Prochlorococcus*, the pattern of these genes has been previously described (Kolowrat et al., 2010). Briefly, *rpoD8* has a low maximal expression in the mid-morning to noon period, whereas *rpoD4* peaked at the end of the light period (see also **Figure 7**).

At last, we looked at the expression of *nblS*, encoding a membrane-bound, sensor histidine kinase involved in the control by light of the expression of several photosynthesis-related genes, including all three *psbA* genes, the *cpcBA* operon, and some *hli* genes (van Waasbergen et al., 2002; Kappell et al., 2006). In VL, this gene was upregulated during the day with a peak in the morning and its maximum expression level increased almost twofold in the presence of UVR (**Figure 7**).

## **DISCUSSION**

## **DIFFERENCES IN PHOTOSYSTEM ACTIVITY AND REGULATION BETWEEN PROCHLOROCOCCUS AND SYNECHOCOCCUS**

The alternation of light and darkness is one of the most predictive events that cyanobacteria have to deal with in the field. Strong variations of PAR occur over a daily timescale and are associated, in near surface waters, with concomitant changes of UVR fluxes. Here, we compare PCC9511, a strain representative of the *Prochlorococcus* HLI clade found in near surface oligotrophic waters, to WH7803, a *Synechococcus* strain characteristic of mesotrophic areas. Although these model strains do not represent the whole physiological diversity existing within these two genera, our data clearly show that both *Prochlorococcus* and *Synechococcus* cells are able to tune their photosynthetic apparatus to diurnal irradiance fluctuations. However, *Prochlorococcus* proved to be more sensitive than *Synechococcus* to photoinhibition by high photon fluxes, as suggested by a marked drop of the cellular pool of PSII core proteins (**Figure 5**) and a larger decrease of the PSII quantum yield (**Figure 3**) around noontime. It is noteworthy however that the latter phenomenon might also partly be due to NPQ of PSII fluorescence associated with photoprotective dissipation of light energy as heat (Bailey et al., 2005; Boulay et al., 2008). The diel changes of the photosynthetic activity observed here for *P. marinus* PCC 9511 (**Figure 3**) are quite comparable with those previously described for this strain grown in similar light conditions (Bruyant et al., 2005), except that our cultures exhibited a higher *F*V/*F*<sup>M</sup> during the night (∼0.7 vs. ∼0.55), likely translating a slightly better physiological status. However, they contrast with those obtained on the closely related MED4 strain by Zinser et al. (2009), who did not observed any significant diel variation of the *F*V/*F*M, likely because it was grown at lower irradiance (∼230 vs. 870µmol photon m−<sup>2</sup> s <sup>−</sup><sup>1</sup> here) and over a different L/D cycle (14/10 vs. 12/12 h). Another noticeable observation from our study was that UVR did not cause

any appreciably stronger photoinhibitory effect than VL in both genera, likely due to an increased repair rate under VL + UV compared to VL. However, this UV-induced repair was much more important for *Synechococcus* than *Prochlorococcus* (∼twofold vs. ∼fivefold at noon, respectively). Accordingly, in the former organism, the relative D1 content was enhanced in the presence of UVR, while in *Prochlorococcus* the noontime drop of D1 was somewhat extended for UV-acclimated cells (see also transcriptomic analyses of the *psbA* genes below). These results are consistent with those obtained by Six et al. (2007a), who observed that in response to a transient high light exposure, *P. marinus* PCC 9511 exhibited a lower PSII repair rate (0.9 PSII gained per second) than a range of marine *Synechococcus* strains (1.1–1.6 PSII s−<sup>1</sup> ). In both strains in VL, this rate was more or less proportional to the instantaneous irradiance. However for cells grown under VL + UV, while the repair rate was comparable between the two strains at 9:00 and 15:00, it was significantly depressed at noon in *Prochlorococcus* compared to *Synechococcus*. This suggests that the PSII repair capacity of *Prochlorococcus* was already maximum around 400µmol photons m−<sup>2</sup> s −1 (**Figure 4B**).

The occurrence of different D1 encoding gene copies in *Prochlorococcus* and *Synechococcus* could, at least partially, explain such a distinct behavior. Indeed, while *Prochlorococcus* strains have one to three identical *psbA* gene copies (one in PCC 9511), coding for a single D1:1-like isoform (Hess et al., 1995; Partensky and Garczarek, 2003), *Synechococcus* strains possess three to six *psbA* genes,with only one copy coding for a D1:1 isoform and two to five copies (three in WH7803), coding for D1:2 isoforms (Garczarek et al., 2008). The respective role of these isoforms has been widely studied in the literature, both in freshwater cyanobacteria (Bustos et al., 1990; Clarke et al., 1993; Campbell et al., 1995, 1998a; Sass et al., 1997; Kos et al., 2008) and in marine picocyanobacteria (Garcia-Fernandez et al., 1998; Garczarek et al., 2008). Although some variations among cyanobacteria have been observed, it is generally accepted that the D1:1 isoform would confer a higher PSII activity (Campbell et al., 1996), while D1:2 would provide a lower quantum yield but a higher PSII resistance to photoinhibition (Krupa et al., 1991;Campbell et al., 1995, 1998a; Tichy et al., 2003). A variety of environmental cues, including UV exposure, can induce the exchange of these isoforms (Sicora et al., 2006, 2008; Garczarek et al., 2008; for a review, see Bouchard et al., 2006). Here, we indeed noticed in *Synechococcus* sp.WH7803 an opposite expression pattern of D1:1 (*SynWH7803\_*0784) and D1:2 isoforms encoding genes (*SynWH7803\_0790*, *0366*, and *2084*) during daytime (**Figure 7**). It is worth noting that while there were only slight discrepancies in the expression levels of the different D1:2 encoding genes between VL and VL + UV, the D1:1 encoding gene was about fivefold more repressed at noon under the latter condition, suggesting a complete replacement of the D1:1 isoform by D1:2 isoform(s) that may contribute to the slight midday decrease in PSII quantum yield (**Figure 3**). Interestingly, although the single isoform in *Prochlorococcus* is phylogenetically a D1:1 isoform, as confirmed by the occurrence of a Gln residue at position 130 of the amino acid sequence (instead of Glu in D1:2; Clarke et al., 1993; Giorgi et al., 1996), its transcriptomic pattern was clearly closer from that of D1:2 isoforms, at least in our culture conditions. It is likely however that it presents a lower resistance to PSII

photoinactivation compared to a true D1:2, as suggested by the larger drop of *F*V/*F*<sup>M</sup> at noon (**Figure 3**) and the concomitant lower PSII repair rate (**Figure 4**), compared to *Synechococcus*.

Interestingly, the D2 subunit of the PSII core is also encoded by a single *psbD* gene in all *Prochlorococcus* genomes, whereas all *Synechococcus* sequenced so far possess two nearly identical *psbD* genes, as in most other cyanobacteria (such as e.g., *Synechococcus* sp. PCC 7942; Golden et al., 1989), with one co-transcribed with *psbC* that encodes the internal PSII antenna protein CP43 (Garczarek et al., 2001) and the other isolated in the genome. It is likely that as for *psbA*, the two copies are differently regulated in response to light and/or UV stress as previously reported in freshwater model cyanobacteria (Bustos and Golden, 1992; Kos et al., 2008), although this was not checked in the present study. Alternatively, this may simply contribute to a higher expression level of this key photosynthetic gene, possibly enabling a higher turnover of the corresponding protein.

Another strategy used by cyanobacteria to cope with excess light energy is to decrease the relative amount of PSI reaction center complexes, an adjustment that was shown to decelerate the rate of photosynthetic electron transport (Murakami and Fujita, 1991; Hihara et al., 1998; Muramatsu and Hihara, 2003). Indeed, experiments on *Synechocystis* sp. PCC 6803 mutants impaired in their ability to modulate photosystem stoichiometry showed that this capacity is indispensable for growth under continuous high irradiance (Hihara et al., 1998; Fujimori et al., 2005). Accordingly, in *Synechococcus* sp. WH7803 the relative *psaB* levels were low before dawn (**Figure 7**), likely reflecting a lower PSI cell content at this time of the day, as previously observed in *Crocosphaera watsonii* (Saito et al., 2011). In contrast, this was not the case in *Prochlorococcus*, in which PSI core transcripts were maximal over most of the dark period.

## **DIFFERENTIAL REGULATION OF LIGHT-HARVESTING SYSTEMS IN RESPONSE TO HIGH LIGHT AND UV RADIATIONS**

The striking structural differences between the major PSII antenna complexes of *Prochlorococcus* and *Synechococcus* may also be partially responsible for the different sensitivity of these picocyanobacteria to UV stress. Indeed, whilst *Synechococcus*, as most cyanobacteria, possess a large membrane-extrinsic antenna, the PBS (Sidler, 1994; Six et al., 2007c), *Prochlorococcus*, like the other two green oxyphotobacteria, *Prochloron* and *Prochlorothrix*, use a transmembrane Chl *a/b*-binding Pcb antenna (LaRoche et al., 1996;Garczarek et al.,2003; Partensky and Garczarek, 2003). These dissimilar antenna structures may have important consequences on the way excitation energy is funneled downhill toward the reaction centers, on the regulation of this process as well as on the amount and nature of damages caused by UVR on *Prochlorococcus* and *Synechococcus* cells, since both organisms show some absorption capacities in the 300- to 400-nm band, which are likely related to their antenna (see e.g., Ong and Glazer, 1991; Claustre et al., 2002). Comparing and monitoring UV cross-sections for both picocyanobacteria would help answering this question.

Here, we indeed observed an opposite expression pattern between the genes encoding antenna systems from these two organisms, with a strong downregulation of the *pcb* gene in *Prochlorococcus* during day time under VL (see also Garczarek et al., 2001) that was even more dramatic under VL + UV, while all examined PBS genes were upregulated in the afternoon in both light conditions (**Figure 8**). A similar result was previously obtained in L/D-entrained *Cyanothece* sp. ATCC 51142 cultures, though in the latter case maximum expression was centered at midday (Toepel et al., 2008). These observations suggest that the biosynthesis of antenna complexes occurs at different times of the day in the two picocyanobacteria and is under different light regulation controls. The fact that PBS genes remained upregulated under VL + UV suggests that the buildup of PBS complexes was only moderately affected by these radiations. This contrasts with a previous study where cultures of *Synechococcus* sp.WH8102 grown in continuous low-light were subjected to a sudden shift to UV (Six et al., 2007b). This stress provoked a strong decrease in the relative expression of all PBS genes, associated with a disconnection of the PBS complexes from the thylakoid membrane as well as a dissociation of the distal PEII disks of PBS rods. In the present study, where cells were acclimated to either VL or VL + UV for several weeks, the PE to PC fluorescence emission ratio did not exhibit a peak at noon in either light condition, as would be expected if terminal PE subunits (i.e., PEII) were disrupted (**Figure A1** in Appendix). The higher values of this ratio in UV-acclimated *Synechococcus* cells are likely not related to a higher PE content of the PBS, but rather to permanently decoupled PEII subunits, which would dissipate incident light as fluorescence. Consistently, UV also induced a slight drop of the expression of most PEII genes at noon and a 3-h delay in the timing of the expression peak of *mpeE*, which encodes the linker binding the terminal PEII disk in *Synechococcus* sp. WH7803 (Six et al., 2007c).

#### **DIFFERENTIAL PHOTOPROTECTION MECHANISMS**

*Prochlorococcus* and *Synechococcus* have also developed specific mechanisms to dissipate excess light energy. Most *Synechococcus* strains possess one gene encoding the orange carotenoid protein (OCP; for a recent review, see Kirilovsky and Kerfeld, 2012), which is thought to mediate energy dissipation as heat through interaction with the PBS core, thus inducing a NPQ of PSII fluorescence. Here, we observed that the *ocp* gene was upregulated during daytime in VL and that UVR dramatically enhanced its expression in the early light period that may trigger a temporary increase of the OCP cellular pool and/or activity. To our knowledge, this is a first time that UV is shown to control the expression of this gene since it is usually believed that OCP is only blue light sensitive (Wilson et al., 2008; Kirilovsky and Kerfeld, 2012).

While all *Prochlorococcus* strains lack OCP, all high lightadapted strains (including PCC 9511) and some low-light-adapted strains possess an homolog of *PTOX*, which is thought to extract electrons between PSII and PSI and to combine them with protons and oxygen to generate water (Bailey et al., 2005, 2008). The *ptox* gene was (with *ftsZ*) one of the two most differentially expressed genes among those analyzed here in PCC 9511, with a maximal relative fold change over the day of ∼73-fold [or log2(FC) = 6.2] at 15:00. These high values are consistent with previous results on the closely related strain *P. marinus* MED4 (Zinser et al., 2009;Berg et al., 2011). So the alternative electron flow to oxygen triggered by PTOX might be an important mechanism used by *Prochlorococcus* to struggle against excess light energy arising to PSII (Bailey

**modulated 12/12 h L/D cycle of high visible light with or without UV**

time intervals during which they were significantly upregulated [|log<sup>2</sup> (Fold Change)| > 1.0], with regard to the expression level measured at 6:00 in VL.

et al., 2008; Berg et al., 2011), although high light induced proteins (HLIPs), which were not analyzed in the present study, are also likely important contributors in this process (He et al., 2001).

Other key actors in photoprotection mechanisms are carotenoids. While the cellular localization and precise action mechanism of Zea is unclear yet, Car are known to be mostly bound to PS reaction centers, two of them in close vicinity of Chls, and help mitigate oxidative damages, mainly by quenching the <sup>1</sup>O<sup>2</sup> species resulting from the de-excitation of Chl triplets (Telfer, 2005). Both Zea:Chl *a* and Car:Chl *a* ratios appeared to be tightly coupled to the L/D cycle in *Synechococcus*, but less so in *Prochlorococcus*, for which we observed a continuous increase of both ratios during most of the day and a symmetrical drop during the night concomitant with cell division, as previously noticed by Claustre et al. (2002). In *Synechococcus*, the sharp drop of the β-Car:Chl *a* ratio from noon to dusk, which was more pronounced in UV-exposed cells (**Figure 2C**), likely reflects the progressive destruction of β-Car molecules by oxidative stress (Telfer, 2005). β-Car was then seemingly regenerated at low rate during the night, then much more rapidly during the first hours of the day, suggesting a light-dependency of the β-Car biosynthesis process, as previously reported in other cyanobacteria (Steiger et al., 2005; Ryu et al., 2010). Similarly, the light-modulated variations of the Zea:Chl *a* ratio during the day in *Synechococcus* in both light conditions could be due either to a higher degradation rate of Chl *a* relative to Zea or to a higher synthesis rate of the latter pigment, a likely photoprotective mechanism against high midday photon fluxes. The observed strong upregulation of *crtR* at high irradiances tend to favor the second hypothesis (**Figure 7**). A further increase of the relative *crtR* transcript levels at noon in VL + UV, which was not translated into a significantly higher midday increase in the Zea:Chl *a* ratio than inVL (**Figure 3A**), suggests that Zea msolecules might have a particularly high turnover under UV. In both genera, the systematically higher values of Zea:Chl *a* ratio under VL + UV than VL only suggest that UV exposure

induces a comparable response as a long-term acclimation to high irradiance, as previously reported in other marine cyanobacterial strains (Kana and Glibert, 1987; Moore et al., 1995; Six et al., 2004). Altogether, our results suggest the occurrence of a light-controlled anabolism/catabolism cycle of both photoprotectants (Car and Zea) in *Synechococcus*, while diel changes in the pigment content of *Prochlorococcus* cells were rather related to the cell division cycle or other factors not analyzed here such as PS stoichiometry and/or antenna size.

## **DIFFERENTIAL PROTECTION MECHANISMS AGAINST REACTIVE OXYGEN SPECIES-INDUCED DAMAGES**

Light-harvesting complexes are not only sources but also major targets of ROS (Sies and Menck, 1992; Krieger-Liszkay, 2005). Because of their localization within the thylakoid membrane around PSII complexes (Bibby et al., 2003), the divinyl-Chl *a*/*b*binding Pcb antennae of *P. marinus* PCC 9511 likely produce ROS, particularly noxious for this photosystem. Indeed, it has been shown in other Chl *b*-containing organisms that excited singlet Chls lead to the synthesis of Chl triplet states that can react with <sup>3</sup>O<sup>2</sup> to produce the very reactive species <sup>1</sup>O<sup>2</sup> (Sies and Menck, 1992; Krieger-Liszkay, 2005). Assays performed here by directly adding H2O<sup>2</sup> to sub-cultures at different time points of the modulated L/D cycle strongly suggest that the effect of ROS increased in a light-dependent manner for both organisms. Consistently, a synergistic effect of light and oxidative stress on PSII photoinactivation was previously demonstrated by Blot et al. (2011) who showed in *Synechococcus* sp. WH7803 that ROS may induce PSII inactivation through both direct damages to the reaction center II and inhibition of the PSII repair cycle. The latter phenomenon resulted in faster PSII inactivation in high light- than in low-lightacclimated cultures, due to their higher D1 turnover. Furthermore, *Prochlorococcus* was also more sensitive to H2O2-triggered stress than *Synechococcus* during the night, which might be related to a lower intrinsic resistance of its PSII.

The difference in ROS sensitivity between *Prochlorococcus* and *Synechococcus* strains might be explained, at least partially, by a more complete set of genes involved in ROS protection and detoxification in *Synechococcus*. Indeed, most *Prochlorococcus* lineages have experienced an extensive genome streamlining during evolution, resulting in the loss of a large number of non-essential but potentially useful genes in this context (**Figure 7**; Dufresne et al., 2005; Scanlan et al., 2009; Partensky and Garczarek, 2010). For instance, while *Synechococcus* sp. WH7803 synthesizes one catalase/peroxidase (KatG), *Prochlorococcus* has neither catalases nor peroxidases. *Prochlorococcus* also lack the *ftrC* and *ftrV* genes, encoding the two subunits of ferredoxin-thioredoxin reductase (FTR), as well as one ferredoxin and one thioredoxin, potentially associated to this complex (Dufresne et al., 2008). Furthermore, while *Synechococcus* possesses two superoxide dismutase (SOD), one Fe-type and one Cu-Zn-type, *Prochlorococcus* has only one, Ni-type SOD (SodN; Scanlan et al., 2009). In the present study, the expression levels of most ROS genes showed strikingly higher amplitudes of variation over the day in *Synechococcus* than in *Prochlorococcus*. Even though transcriptomic data are not necessarily synchronized with the activity of the corresponding enzymes, the lower resistance of *Prochlorococcus* to high VL and UVR could at least partially be due to a higher sensitivity to light-driven oxidative stress.

## **CONCLUSION**

The comparison of *Synechococcus* and *Prochlorococcus* cultures acclimated to VL supplemented or not with UVR revealed relatively few physiological responses specific to UV. This notably includes a shift of the DNA synthesis phase (**Figure 1**; see also Kolowrat et al., 2010), an increase of the Zea:Chl *a* ratio (**Figure 2**) and an enhanced PSII repair rate (**Figure 4**). Accordingly, UVR seemingly also had limited effects at the transcriptomic level, as shown by the globally similar diel expression patterns between VL and VL + UV in both strains (**Figures 7** and **8**). It is worth noting however that the relative expression of a few genes was either enhanced (e.g., *psbD* in both strains, D1:2 encoding genes and *crtR* in *Synechococcus* only) or reduced (e.g., *Synechococcus* D1:1 encoding gene or *Prochlorococcus pcbA*) in UV-acclimated cultures. A handful of genes, including *rpoD4*/8 in *Prochlorococcus* and *kaiABC* in *Synechococcus*, also exhibited a delayed expression peak by about 3 h. However, this surprisingly did not translate into any conspicuous changes in the diel patterns of most of the other genes examined here, as could have been expected from the known regulatory role of the circadian clock and sigma factors on gene transcription (see e.g., Summerfield and Sherman, 2007; Ito et al., 2009).

The most striking result of the present study is likely the markedly distinct response to diurnal light variations between *Prochlorococcus* and *Synechococcus*, despite their close phylogenetic relatedness (Scanlan et al., 2009). These two picocyanobacteria indeed exhibited very different degrees of PSII photoinactivation at noon that can be partly explained by their distinct (i) PSII repair capacity (**Figure 4**; see also Six et al., 2007a), (ii) ability to modulate photoprotective pigments (Zea and Car, **Figure 2**), and (iii) resistance capacity against oxidative stress (**Figure 6**). Comparative transcriptomic analyses also revealed that, in *Synechococcus*, genes coding for a number of protective systems were maximally expressed during hours of highest irradiance, i.e., when these mechanisms are most critical to cope with transitory stressful conditions. This includes several genes involved in ROS detoxification enzymes, DNA repair genes as well as genes involved in photoprotection and/or dissipation of excess energy, such as *crtR* and *ocp* (**Figure 7**). In contrast, in *Prochlorococcus*, very few genes of these pathways, including *ptox* and *psbA*, were maximally expressed around midday, while others, such as *phrA* and *crtR*, had already reached their maximal (saturating) expression at mid-morning. Furthermore, many photosynthetic genes that were upregulated during the day in *Synechococcus* were in contrast downregulated in *Prochlorococcus*, including the PSI core gene *psaB* as well as genes involved in light-harvesting, ATP synthase and CO<sup>2</sup> fixation (**Figure 8**). Similarly, *glgA* (encoding glycogen synthase) mRNA abundance was recently shown to exhibit a maximum diel expression at midday in *Synechococcus* sp.WH8103 grown under a 16/8-h L/D cycle, while it peaked in concert with *rbcLS* in *Prochlorococcus marinus* MED4 during the LDT (Wyman and Thom, 2012). Thus altogether, it seems that while *Synechococcus*, as many other cyanobacteria (Kucho et al., 2004, 2005; Stockel et al., 2008; Toepel et al., 2008, 2009; Shi et al., 2010), efficiently copes with the diurnal changes in photon fluxes, *Prochlorococcus* rather displays a stresslike response at midday. The latter response is most likely related to the high irradiance level used in the present study (reaching 870µmol photons m−<sup>2</sup> s −1 at noon), which are typically found in the upper mixed layer of tropical oligotrophic oceans (Holtzendorff et al., 2001). Indeed, several genes that were found here to be downregulated during the light period (e.g., *pcb*, *rbcL*, *chlG*) were in contrast upregulated in *Prochlorococcus* cultures grown at lower irradiances, provided as continuous (Berg et al., 2011) or cyclic light (Zinser et al., 2009). Despite the very atypical transcriptomic response observed here, *Prochlorococcus* was able to recover high PSII quantum yield at night and to maintain an optimal growth rate under these conditions. Thus, we hypothesize that, in contrast to *Synechococcus* sp. WH7803, *P. marinus* PCC 9511 cells manage to cope with harmful light conditions by bringing down temporarily some of the main metabolic processes and by launching a minimal set of protection mechanisms during stressful hours. Whether the absence of a true circadian clock in *Prochlorococcus* (Holtzendorff et al., 2008; Axmann et al., 2009) is involved in this differential management of excess light compared to *Synechococcus* still remains to be investigated.

Altogether, our study reinforces previous studies depicting *Prochlorococcus* as a very specialized organism restricted to a narrow environmental niche, while *Synechococcus* has adopted a generalist strategy enabling it to cope with more variable environmental conditions, a difference consistent with the distinct habitats in which these two organisms predominate (Scanlan, 2003; Kettler et al., 2007; Dufresne et al., 2008; Scanlan et al., 2009).

## **ACKNOWLEDGMENTS**

This work was supported by the French programs ANR PELICAN (PCS-09-GENM-200) and EMBRC France (INFRA-2010-2.2.5), the European Union programs MicroB3 (UE-contract-287589) MaCuMBA (FP7-KBBE-2012-6-311975). Daniella Mella-Flores was supported by the National Commission of Scientific and

Technological Investigation of Chile (CONICYT).We thank Roseline Edern, Fabienne Jalabert, and the Roscoff Culture Collection for maintaining the *Synechococcus* and *Prochlorococcus* strains used in this study.

## **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at http://www.frontiersin.org/Aquatic\_Microbiology/10.3389/fmicb. 2012.00285/abstract

**Table S1 | List of primers used for real time PCR reactions.** Most primers are targeting a specific gene. Exceptions include the psbD pool and psbA pool primers, which were designed to amplify the two psbD gene copies and all four

## **REFERENCES**


non-photochemical quenching of chlorophyll a fluorescence in cyanobacteria. *FEBS Lett.* 579, 275–280.


psbA copies present in Synechococcus sp. WH7803, respectively. Furthermore, one primer set targets two psbA genes (SynWH7803\_0366 and 2084), which have nearly identical coding sequence and 5<sup>0</sup> -UTR. Genes are classified by functional categories, as indicated in the first column. The Cyanorak database of picocyanobacteria protein families is publicly accessible at http://www.sb-roscoff.fr/Phyto/cyanorak/

**Table S2 | Daily variations of the expression of selected genes, as measured by real time quantitative PCR, for picocyanobacterial cells acclimated to a modulated 12/12 h L/D cycle of VL with or without UV radiations.** All data are expressed as log<sup>2</sup> (Fold Change) ± mean deviation for two biological replicates. For each gene, transcript levels are normalized to the reference time point 6:00 in VL. Genes are classified by functional categories, as indicated in the first column. This table is a detailed version of **Figure 7** of the main manuscript.


of primary charge separation in the photosystem II reaction center complex isolated from wild-type and D1-130 mutants of the cyanobacterium *Synechocystis* PCC 6803. *J. Biol. Chem.* 271, 2093–2101.


W. R. (2001). Diel expression of cell cycle-related genes in synchronized cultures of *Prochlorococcus* sp strain PCC 9511. *J. Bacteriol.* 183, 915–920.


Mechanism of photosystem II photoinactivation and D1 protein degradation at low light: the role of back electron flow. *Proc. Natl. Acad. Sci. U.S.A.* 94, 1579–1584.


*Synechococcus*. *Physiol. Plant.* 82, 1–8.


populations of marine picoplankton by flow cytometry using the nucleic acid stain SYBR Green I. *Appl. Environ. Microbiol.* 63, 186–193.


marine *Synechococcus* spp.: a comparative genomics study. *Genome Biol.* 8, R259.


coding region and sensitivity to visible and UV-B radiation. *Biochim. Biophys. Acta* 1605, 55–66.


strain WH8103. *Appl. Environ. Microbiol.* 78, 4744–4747.


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

*Received: 10 April 2012; paper pending published: 12 May 2012; accepted: 19 July 2012; published online: 08 August 2012. Citation: Mella-Flores D, Six C, Ratin M, Partensky F, Boutte C, Le Corguillé G, Marie D, Blot N, Gourvil P, Kolowrat C and Garczarek L (2012) Prochlorococcus and Synechococcus have evolved different adaptive mechanisms to cope with light and UV stress. Front. Microbio. 3:285. doi: 10.3389/fmicb.2012.00285 This article was submitted to Frontiers in Aquatic Microbiology, a specialty of*

*Frontiers in Microbiology. Copyright © 2012 Mella-Flores, Six, Ratin, Partensky, Boutte, Le Corguillé, Marie, Blot, Gourvil, Kolowrat and Garczarek. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, providedthe original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

## **APPENDIX**

**FIGURE A1 | Phycoerythrin to phycocyanin (PE:PC) fluorescence emission ratio as measured by spectrofluorimetry (at 572 and 651 nm, respectively) in Synechococcus sp.WH7803 cells grown under a L/D**

**cycle with and without UV.** This ratio showed significantly higher values in VL + UV than in VL-acclimated cells. In both light conditions, it increased during the day by ∼25% compared to the value at dawn then decreased during the night. The amplitude of these oscillations (ca 0.4 unit) was very low compared to those reported for Synechococcus sp. WH8102 acclimated to a large range

of irradiance (2.5 units; Six et al., 2004). Further investigation is required to determine whether the diel pattern of PE:PC fluorescence emission ratio rather results from a decrease in the efficiency of energy transfer in the PBS rod and/or a progressive accumulation of free phycoerythrins during the day. These light-induced oscillations might be indirectly related with state transitions (Campbell et al., 1998b; Mullineaux and Emlyn-Jones, 2005), a light-dependent process which usually exhibits large amplitudes in marine Synechococcus (Six et al., personal communication).

## Diversity and distribution of marine Synechococcus: multiple gene phylogenies for consensus classification and development of qPCR assays for sensitive measurement of clades in the ocean

## *Nathan A. Ahlgren and Gabrielle Rocap\**

*School of Oceanography, Center for Environmental Genomics, University of Washington, Seattle, WA, USA*

#### *Edited by:*

*Anton F. Post, Marine Biological Laboratory, USA*

#### *Reviewed by:*

*David J. Scanlan, University of Warwick, UK Kathleen Scott, University of South Florida, USA*

#### *\*Correspondence:*

*Gabrielle Rocap, School of Oceanography, University of Washington, Box 357940, Seattle, WA 98195, USA. e-mail: rocap@ocean.washington.edu*

Marine *Synechococcus* is a globally significant genus of cyanobacteria that is comprised of multiple genetic lineages or clades. These clades are thought to represent ecologically distinct units, or ecotypes. Because multiple clades often co-occur together in the oceans, *Synechococcus* are ideal microbes to explore how closely related bacterial taxa within the same functional guild of organisms co-exist and partition marine habitats. Here we sequenced multiple gene loci from cultured strains to confirm the congruency of clade classifications between the 16S–23S rDNA internally transcribed spacer (ITS), 16S rDNA, *narB*, *ntcA*, and *rpoC1* loci commonly used in *Synechococcus* diversity studies. We designed quantitative PCR (qPCR) assays that target the ITS for 10 *Synechococcus* clades, including four clades, XV, XVI, CRD1, and CRD2, not covered by previous assays employing other loci. Our new qPCR assays are very sensitive and specific, detecting down to tens of cells per ml. Application of these qPCR assays to field samples from the northwest Atlantic showed clear shifts in *Synechococcus* community composition across a coastal to open-ocean transect. Consistent with previous studies, clades I and IV dominated cold, coastal *Synechococcus* communities. Clades II and X were abundant at the two warmer, off-shore stations, and at all stations multiple *Synechococcus* clades co-occurred. qPCR assays developed here provide valuable tools to further explore the dynamics of microbial community structure and the mechanisms of co-existence.

**Keywords: microbial ecology, cyanobacteria,** *Synechococcus***, microbial diversity, quantitative PCR, multiple gene locus phylogeny, biogeography, ecotype**

## **INTRODUCTION**

Marine *Synechococcus* is an abundant and environmentally significant group of unicellular cyanobacteria that provides an ideal system to explore the ecology and evolution of closely related bacterial taxa and their niche partitioning of the global oceans (Scanlan and West, 2002). These small cyanobacteria (0.6–1.7μm) belong to the functional class of picophytoplankton, the smallest (<2μm) guild of photoautotrophs and contribute significantly to primary production (Waterbury et al., 1986; Li, 1994; Liu et al., 1997; Partensky et al., 1999; Jardillier et al., 2010). Molecular phylogeny has revealed a large genetic diversity amongst *Synechococcus* found in marine and estuarine waters. Based on 16S rDNA phylogeny, these strains have been classified into three major groups, labeled subclusters 5.1, 5.2, and 5.3 (Herdman et al., 2001; Dufresne et al., 2008; Scanlan et al., 2009). Phylogenies based on several additional loci reveal that these subclusters contain more than 20 distinct genetic clades. Genes used to study the diversity of marine *Synechococcus* include the 16S–23S internally transcribed spacer (ITS; Rocap et al., 2002; Ahlgren and Rocap, 2006; Chen et al., 2006; Choi and Noh, 2009; Cai et al., 2010; Huang et al., 2012), *narB* (nitrate reductase; Jenkins et al., 2006; Paerl et al., 2008, 2011), *ntcA* (a nitrogen regulatory gene; Penno et al., 2006;

Post et al., 2011), *rpoC1* (subunit of RNA polymerase; Palenik, 1994; Toledo and Palenik, 1997; Ferris and Palenik, 1998; Muhling et al., 2005), *rbcL* (Rubisco large subunit; Chen et al., 2006; Paerl et al., 2012), *psbA* (photosystem I gene; Zeidner et al., 2003), *cpeB* (phycoerythrin gene; Steglich et al., 2003), and *petB* (encoding the cytochrome *b*<sup>6</sup> subunit of the cytochrome *b*6f complex; Mazard et al., 2012).

These distinct clades are thought to represent physiologically and ecologically distinct lineages, or ecotypes, as has been seen in *Prochlorococcus*, the sister genus to marine *Synechococcus*. Genetically distinct clades of *Prochlorococcus* contain strains with ecologically relevant differences in light adaptation and nutrient utilization (Moore et al., 1998, 2002; Moore and Chisholm, 1999; Rocap et al., 2002), and these clades occupy distinct oceanic habitats that reflect their specific adaptations (West and Scanlan, 1999; Ahlgren et al., 2006;Johnson et al., 2006). Several *Synechococcus* clades similarly possess distinguishing physiologies that potentially impact their ecology (**Figure 1**). For example clade III strains possess a unique form of motility that is absent in most other clades and may allow them to find and scavenge nutrients in oligotrophic waters (Waterbury et al., 1985; Toledo et al., 1999; Chen et al., 2004). Other clades differ in their accessory light harvesting pigments


**FIGURE 1 | Schematic phylogenetic tree summarizing the genetic and ecological diversity of marine** *Synechococcus* **and molecular assays that have been developed to measure particular clades in the oceans.** The depicted tree is approximately based on a four loci concatenated tree (**Figure A6** in Appendix). Examples of commonly studied strains ("type"

strains) in each clade are shown. Strains in bold were selected for multiple gene loci analysis, and strains with sequenced genomes are underlined. Branch tips without taxa names indicate clades that are only represented by environmental sequences (e.g., clade CRD2). Nodes with bootstrap support *(Continued)*

#### **FIGURE 1 | Continued**

of ≥55% are denoted with circles (closed circles indicate >75% support and open circles indicate 55–75% support). Polytomies indicate cases where the placement of clades is uncertain because of low bootstrap support or the absence of sequences available for multiple loci. Marine *Synechococcus* are classified into three subclusters 5.1, 5.2, and 5.3 (black vertical boxes on left), and phylogeny based on 1,129 core genes from 11 strains with sequenced genomes supports the division of strains in subcluster 5.1 into two groups 5.1A (blue bar) and 5.1B (solid red bar) (Dufresne et al., 2008). The four loci concatenated phylogeny in **Figure A6** in Appendix also supports the grouping of 5.1A strains, including clade XV. Strains in clades VII, CRD1, and XVI cluster with 5.1B strains in clades V and VI with high bootstrap support, however it is still unknown if these additional new strains without sequenced genomes form a single monophyletic cluster with all 5.1B strains (dashed red line). Information about specific physiologies, ecology, or biogeography of clades is summarized with relevant references. Chromatic adaptation (CA) is the ability to shift the ratio of accessory pigments phycourobilin (PUB) and phycoerythrobilin (PEB) according to the spectral quality of light (Palenik, 2001; Six et al., 2007). Circles (or ovals, see footnotes) indicate clades for which quantitative molecular assays have been designed to measure their

abundance in the environment. [ITS, this study; *rpoC1* (Tai and Palenik, 2009; Tai et al., 2011); 16S rDNA (Fuller et al., 2003); *narB* (Paerl et al., 2011)]. *narB* sequences have been classified to groups "A" to "G" rather than the convention of clades used by other loci (Paerl et al., 2011; see **Figure A3** in Appendix). Circles with letters indicate the clade(s) for which *narB* qPCR primers designed for that *narB* group have perfect matches to at least one of the sequences in the clade (see **Figure A3** in Appendix for more info on primer coverage). <sup>a</sup> WH8101 was formerly classified as belonging to cluster 5.2 (Herdman et al., 2001), but this and other phylogenies suggest it belongs in cluster 5.1. bClade II includes clade WPC2 following the convention of Huang et al. (2012). <sup>c</sup> Clade XIII is congruent with and has been subsumed into clade VII (see **Figure A4** in Appendix). dITS qPCR primers were designed to perfectly match clade X strains, now classified as clade 5.3-I. These primers have perfect matches to sequences from 5.3-I through 5.3-III (but not 5.3-IV through 5.3-VI; see **Figure A2** in Appendix). Similarly, the clade X 16S rDNA probe was designed before the discovery of clades 5.3-II–5.3-VI. Because there are no corresponding 16S rDNA sequences for clades 5.3-II–5.3-VI, it is unknown if the clade X probe detects these other subcluster 5.3 clades as indicated by the dashed oval. <sup>e</sup> This 16S rDNA probe detects clades V, VI, VII together.

(Waterbury et al., 1986; Rocap et al., 2002; Six et al., 2007; Choi and Noh, 2009) or the ability to change their light harvesting apparatus in response to the spectral quality of light they experience (chromatic adaptation; Palenik, 2001; Ahlgren and Rocap, 2006; Six et al., 2007). While *Synechococcus* clades are hypothesized to represent ecotypes, physiological characterization of isolates from many clades unfortunately still remains sparse, making it difficult to predict what distinct niches they may occupy (**Figure 1**).

Quantitative molecular assays measuring the abundance of specific *Synechococcus* clades have been used to characterize their biogeography and confirm that these clades occupy different niches and thus represent ecotypes. Assays include a suite of eight oligonucleotide probes that target 10 16S rDNA clades (clades I– X), three of which are detected together by a single probe (V, VI, and VII; Fuller et al., 2003); quantitative PCR (qPCR) assays that target four clades (I–IV) using the *rpoC1* gene (Tai and Palenik, 2009); and qPCR assays that target seven groups of *narB* sequences that each contain one or more clades (Paerl et al., 2011; **Figure 1**). The use of coding genes for quantitative assays can be problematic because the third codon position often necessitates the use of degenerate primers, as is the case for the *rpoC1* and *narB* assays. Furthermore, strains in some clades lack the *narB* gene and cannot utilize nitrate, such as RR9917 in clade IX (Scanlan et al., 2009) and MITS9220 in clade CRD1 (Moore et al., 2002). More recently, deep pyrosequencing of the 16S rDNA has been used to track clade abundances over time in the Red Sea (Post et al., 2011), non-degenerate *petB* primers have been developed for construction of *Synechococcus* specific clone libraries (Mazard et al., 2012), and DNA probes attached to fluorescent beads have been used to follow the abundances of subpopulations within clades I–IV (Tai et al., 2011). While each of the above methods targets a different locus, they reveal some cohesive patterns in the distribution of some of the major clades in the oceans. Clades I and IV are most abundant in colder, nutrient rich coastal waters while clade II and the combined group of clades V/VI/VII are widely distributed in tropical and sub-tropical waters (Zwirglmaier et al., 2007, 2008; **Figure 1**). Clade III is most prevalent in warm, oligotrophic waters, and appears to be specialized to these habitats (Fuller et al., 2003; Zwirglmaier et al., 2007, 2008; Mella-Flores et al., 2011; Post et al., 2011). Clade VIII may be specifically adapted to hypersaline waters (Dufresne et al., 2008; Huang et al., 2012) while other clades such as IX and X have been rarely seen and only at low abundance (Zwirglmaier et al., 2007, 2008).

The growing number of *Synechococcus* diversity studies and molecular assays to study their biogeography are valuable in describing the ecology of putative *Synechococcus* ecotypes, but the use of several different loci has made classification confusing and comparison of results from the different assays difficult. The various loci differ in their relative level of divergence, thus some clades may be defined too broadly or too narrowly. For example, the highly conserved 16S rDNA is less divergent than other loci and does not allow the development of specific oligonucleotides that can resolve clades V, VI, and VII unlike the ITS and other coding genes (Fuller et al., 2003; Ahlgren and Rocap, 2006; Huang et al., 2012). The use of different loci also makes it difficult to determine if clades defined by independent loci correspond to one another. This is especially the case for clades that contain only sequences obtained from environmental clone libraries (e.g., see Saito et al., 2005; Penno et al., 2006). Sequencing multiple loci in cultured isolates is necessary to verify potential overlap of clades identified by different loci. Recently, Mazard et al. (2012) conducted a classical multiple-locus sequence analysis (MLSA) study that established congruence among clades by using core genes including the *rpoC1* and *petB* genes. They also examined the 16S rDNA and ITS, although these genes could not be included in MLSA as most *Synechococcus* genomes contain two identical copies of the 16S rDNA and ITS. Their analysis however did not include the commonly used *ntcA* or *narB* sequences (Ahlgren and Rocap,2006; Jenkins et al., 2006; Penno et al., 2006; Paerl et al., 2008; Post et al., 2011), and the portion of the *rpoC1* gene they sequenced did not overlap with the portion used in the large dataset of available *rpoC1* sequences (Toledo and Palenik, 1997; Ferris and Palenik, 1998).

In order to further the consensus in the classification of *Synechococcus* clades, we sequenced five loci commonly used in field studies, 16S rDNA, ITS, *narB*, *ntcA*, and *rpoC1*, in cultured isolates distributed across 14 clades and compared the phylogenies of the different loci. After confirming that strains formed congruent clades across the loci examined, we developed an expanded set of qPCR assays that target ITS sequences for 10 genetic clades of marine *Synechococcus*, including newly described clades not detected by previous assays. The ITS was chosen because it is divergent enough to allow clear resolution of clades but still contains conserved regions for design of non-degenerate primers. Furthermore, because the ITS has been widely used it is the locus with the largest number of sequences available for primer design. These qPCR assays represent the most comprehensive set of quantitative assays for this genus and provide a valuable tool to further characterize the ecology of *Synechococcus* clades. Application of qPCR assays to a transect in the North Atlantic off of the eastern United States shows co-existence of multiple clades and large shifts in *Synechococcus* community composition between open-ocean and coastal environments.

## **MATERIALS AND METHODS**

## **ISOLATION OF** *SYNECHOCOCCUS* **STRAINS AND ENVIRONMENTAL SEQUENCES**

Three new *Synechococcus* isolates were obtained following the procedures outlined in (Ahlgren and Rocap, 2006). UW1 was isolated from surface waters of Hood Canal (47˚ 36.4 N 122˚ 56.4 W) within Puget Sound (Washington state) on March 27, 2002. Strains UW179 and UW180 were isolated from the Equatorial Pacific south of Hawaii (12˚ 42.7 N 157˚ 59.8 W) from 125 m on February 2, 2004. A single colony was picked for each strain from pour plates made from cells grown in enrichment cultures. The ITS sequence CHA22 was obtained from a clone library constructed from surface water sampled off of San Juan Channel (48˚ 32.5 N, 122˚ 59.4 W) off of San Juan Island, Washington on April 30, 2004. The clone library was constructed following the methods outlined in Ahlgren and Rocap (2006).

## **SEQUENCING OF ISOLATES**

Between 1 and 5 ml of exponential phase culture was centrifuged at 18,000 × *g* for 10 min and the DNA was extracted from the cell pellet using either a DNeasy extraction kit (Qiagen, Valencia, CA, USA) or by quick heat lysis extraction: 95˚C for 10 min in 50–100μl of 10 mM Tris, 1 mM EDTA pH 8.0. PCR primers and reaction conditions used to amplify genes are listed in **Table A1** in Appendix. PCR reaction volumes were 20 or 25 μl and contained 0.2 mM each dNTPs and 1–2μl of template DNA. All PCR reactions started with an initial step of 95˚C for 5–10 to denature the template DNA or activate the polymerase, and all reactions ended with a 6- to 10-min extension step at 72˚C. 16S rDNA and 16S–23S rDNA ITS sequences were obtained by sequencing purified PCR products obtained with Taq polymerase (Promega, Madison, WI, USA). Quadruplicate 20μl reactions were pooled and purified using a Qiaquick PCR Purification Kit (Qiagen) before sequencing reactions with DYEnamic ET dye terminator kit (Amersham Biosciences Corp., Piscataway, NJ, USA) and sequencing on a MegaBACE 1000 automated sequencer (Amersham Biosciences). For *narB*, *ntcA*, and *rpoC1*, PCR products were generated using Accuzyme (high fidelity polymerase PCR mix (Bioline USA Inc., Taunton, MA, USA). Products were gel purified (Qiagen gel extraction kit), cloned into sequencing

vectors (TOPO TA Kit, Invitrogen Corp., Carlsbad, CA, USA), and sequenced with M13 vector primers. Raw sequences were aligned and visually inspected using Sequencher (Gene Codes Inc., Ann Arbor, MI, USA). Sequences have been deposited in GenBank under the following accession numbers: 16S rDNA, JQ421027– JQ421036; ITS, JQ421037–JQ421042, *narB*, JQ421043–JQ421045, *ntcA*, JQ421046–JQ421052; *rpoC1*, JQ421053–JQ421055. Additional details are listed in **Table A2** in Appendix.

## **PHYLOGENETIC ANALYSES**

Phylogenetic trees were constructed using sequences obtained in this study, from previous studies and complete genome sequences (Dufresne et al., 2008). ITS sequences from previous studies (Beja et al., 2000; Suzuki et al., 2001; Ahlgren and Rocap, 2006; Chen et al., 2006; Choi and Noh, 2009; Cai et al., 2010; Choi et al., 2011; Mella-Flores et al., 2011; Huang et al., 2012; Mazard et al., 2012) were aligned to our existing database of sequences using automatic alignment (PT\_server alignment tool) in ARB (Ludwig et al., 2004) and manual adjustment. For phylogenetic analyses, the two tRNAs in the ITS and the loop portions of stem-loop structures (predicted by MFOLD; Zuker, 2003) were excluded from alignments. Other 16S rDNA, *narB*, *ntcA*, and *rpoC1* sequences were taken from previous studies (Giovannoni et al., 1990; Beja et al., 2000; Suzuki et al., 2001; Fuller et al., 2003; Muhling et al., 2005; Everroad et al., 2006; Jenkins et al., 2006; Penno et al., 2006; Dufresne et al., 2008; Kan et al., 2008; Choi and Noh, 2009; Mella-Flores et al., 2011; Mazard et al., 2012) These sequences were aligned either in ARB (16S rDNA) or using clustal (*narB*, *ntcA*, and *rpoC1*) with amino acid sequences.

Trees were constructed in PAUP∗ (Swofford, 1999) or RAxML (Stamatakis,2006). The concatenated gene tree employed four loci, the 16S rDNA gene, the ITS, *ntcA*, and *rpoC1*. For the 16S rDNA, *narB*, *ntcA*, *rpoC1* genes and the four concatenated loci, trees were constructed using minimum evolution as the criterion and the HKY85 model in PAUP∗ with gamma distributed rates of substitution and a proportion of invariable sites (HKY + i + g) estimated by maximum likelihood. Trees found with the HKY + i + g model were very similar to trees constructed using substitutions models suggested by MODELTEST (Posada and Crandall, 1998; data not shown). Bootstrap analysis was performed using neighbor joining. For the ITS, best tree analysis was done in PAUP∗ using neighbor joining and the HKY + i + g substitution model and bootstrap analyses were performed in PAUP∗ (neighbor joining trees) and RAxML with the GTRGAMMA model. Additional bootstrap analysis was done for the concatenated loci using minimum evolution and the Jukes–Cantor model in PAUP∗ and maximum likelihood in RAxML with the GTRGAMMA model.

## **qPCR ASSAY DESIGN**

Primers specific for 10 clades of marine *Synechococcus* were designed for use with qPCR assays (**Table 1**). Design of primers was guided by use of the probe design option in ARB (Ludwig et al., 2004), which searches for regions that are identical within a specified group of sequences and have several mismatches to other sequences. Primers targeted the 16S–23S rDNA ITS region and were designed to be identical to all sequences within each specified clade when possible but with significant mismatches to all other


**Table 1 | Primers, plasmid standards, and optimized reaction and cycling conditions used for qPCR assays.**

*a qPCR assays were not developed to quantify these clades, but plasmid standards were generated for representative strains to check the specificity of the other qPCR assays.*

*b These plasmids are cloned environmental sequences from Ahlgren and Rocap, 2006 (oc5m66 and oc175m12), Saito et al., 2005 (CRD1 and CRD8) or this study (CHA22).*

*c For clade X, lower limits of detection were adjusted to two times above the level at which this assay cross-amplified with clade IV, and similarly the clade CRD1 limit was adjusted to two times the level of cross-amplification with WH 7803 (seeTable A3 in Appendix).*

*Synechococcus* ITS sequences. Primers were designed so that mismatches to non-specific sequences were concentrated at the 3 end. Because at the time of primer design only a few sequences were available for clades V, VI, VII, and IX, reliable primer sets could not be designed for these clades.

Quantitative PCR assays were optimized on an ICycler thermocycler (BioRad Inc., Hercules, CA, USA). For all reactions, 5μl of template DNA was added to a reaction cocktail of 20μl. Reactions contained Immomix PCR mix (Bioline USA Inc., Taunton, MA, USA), 10 nM fluorescein (for normalization of well-to-well variability), 0.4× Sybr Green I (Molecular Probes, Eugene, OR, USA), and 0.1% dimethyl sulfoxide. Primers were used at a final concentration of 500 nM. Magnesium chloride concentrations and thermocycling conditions (**Table 1**) were optimized to maximize amplification efficiency, reduce amplification of no template negative controls, and reduce cross-amplification with non-specific ITS sequences (see below). Cycling programs consisted of an initial 10 min step at 95˚C to activate the polymerase followed by 45 cycles of 30 s at 95˚C, 30 s at an optimized annealing temperature (given in **Table 1**), and 30 s at 72˚C. A melting curve was done at the end of each assay to verify the specificity of amplicons.

Plasmid standards were created by cloning ITS PCR products into *Escherichia coli* using a TOPO TA cloning kit (Invitrogen Corp., Carlsbad, CA, USA). ITS PCR products were generated using the cyanobacterial specific PCR primers 16S-1247f and 23S-241r using reaction conditions in Rocap et al. (2002). Purified plasmids were linearized with restriction enzyme *Pst*I or *Sph*I using conditions recommended by the manufacturer (New England Biolabs, Ipswich, MA, USA). Linear plasmids were run on a 1.5% agarose electrophoresis gel and bands of correct length were excised and purified from the gel using a Qiagen Gel Extraction Kit. Plasmids were diluted serially in 10 mM Tris, 0.1 mM EDTA, pH 8.0, and quantified with Picogreen DNA stain (Molecular Bioprobes, Eugene, OR, USA). Strains used for generating standards for each primer set are listed in **Table 1**. For some clades we used plasmids from previously generated ITS environmental clone libraries. Plasmid standards were also constructed for strains from clades V, VI, VII for use in testing the specificity of qPCR assays (**Table 1**). Plasmid standards generated from cultured strains were sequenced to verify that the regions to which the qPCR primers anneal were identical to the sequence originally determined for the strain.

## **CALIBRATION OF qPCR PLASMID STANDARDS**

To report results from qPCR assays in terms of cells/ml for quantification of field samples, plasmid standards were calibrated to cell concentrations by making DNA filters with known amounts of cultured cells. Cultures were grown in artificial sea water with 800μM ammonium, 50μM phosphate, and trace metal mix (Moore et al., 2002) under a 12-h dark: 12 h light cycle at a light level of approximately 50μmol Q m−<sup>2</sup> s <sup>−</sup>1. Cells were harvested in exponential phase. For DNA extraction 20 or 25 ml of culture was collected on filters with gentle filtration (<15 mm Hg of vacuum). DNA was extracted from the filters with a simple heat lysis protocol (Zinser et al., 2006). Six-hundred fifty microliters of 10 mM Tris 0.1 mM EDTA pH 8.0 was added to filters, and filters were vigorously vortexed using a bead beater (BioSpec Products, Bartlesville, OK, USA) for 2 min at maximum speed to shred the filter and resuspend cells. Five-hundred microliters of the cell mixture was removed from the shredded filter, added to 1.5 ml centrifuge tubes, and incubated at 95˚C for 15 min. To determine cell concentrations diluted cell cultures were preserved with 0.125% glutaraldehyde and frozen and stored in liquid N2 for later analysis using a Cytopeia InFlux flow cytometer (Cytopeia Inc., Seattle, WA, USA) following standard procedures (Marie et al., 2004). Calibration was done for clades for which we had cultured strains: clades I, II, III, VIII, XV, and XVI.

## **FIELD SAMPLING**

Water samples were collected aboard the R/V Oceanus (cruise OC375) along a transect between Massachusetts and the northwest Atlantic in March 2002. Latitudes and longitudes for sample stations were as follows: Station 8 (Sargasso Sea), 34˚ 59.88 N, 70˚ 13.92 W; station 16 (Gulf Stream), 37˚ 36.36 N, 70˚ 44.09 W; station 17 (coastal), 38˚ 35.79 N, 70˚ 45.69 W; and station 19 (coastal), 40˚ 38.06 N, 70˚ 50.59 W. Water was sampled at discrete depths with 20 l Niskin bottles attached to a Sea-Bird conductivity, temperature, depth (CTD) package (Sea-Bird Electronics Inc., Bellevue, WA, USA). One-hundred ml of 20μm pre-filtered sea water was filtered onto 25 mm 0.2μm polycarbonate filters to collect DNA for analysis with qPCR. Cells were collected on filters with gentle filtration (<15 mm Hg of vacuum) and chased with Preservation Solution [0.5 M NaCl, 10 mM Tris (pH 8.0), 100 mM EDTA (pH 8.0)] following the protocols described inAhlgren et al. (2006). Filters were flash frozen in liquid N2 and stored at −80˚C until extraction and processing.

For flow cytometric analyses whole seawater samples were preserved with 0.125% glutaraldehyde, flash frozen in liquid N2, and stored at −80˚C for later analysis. For stations 16, 17, and 19 cyanobacterial concentrations were determined using a Cytopeia InFlux flow cytometer (Cytopeia Inc., Seattle, WA, USA) following standard procedures (Marie et al., 2004). *Synechococcus* and *Prochlorococcus* concentrations were previously reported for station 8 on this cruise using a different flow cytometer (Ahlgren and Rocap, 2006). A partial depth profile from station 8 was rerun on the InFlux cytometer with slightly different results, likely due to inaccuracy in the flow rate assumed by the facility where samples were initially run. Thus the *Prochlorococcus* and *Synechococcus* abundances we report for station 8 were adjusted by a correction factor to account for discrepancies between the two measurements.

For measurement of clade abundances DNA filters were extracted with the heat lysis protocol described above. Triplicate PCR reactions were performed for each clade-specific primer set using 5μl of template DNA in a reaction cocktail of 20 μl and other conditions described above and in **Table 1**. *Synechococcus* clades were quantified on an ICycler thermocycler (BioRad Inc.,Hercules, CA, USA) using a five place plasmid standard curve (also amplified in triplicate) and cell concentrations were calculated using the calibration curves of cells versus plasmids standards.

## **RESULTS**

## **PHYLOGENIES OF MULTIPLE GENE LOCI**

Because multiple loci (ITS, 16S rDNA, *narB*, *ntcA*, and *rpoC1*) have been used to describe the diversity of marine *Synechococcus*, we compared phylogenies of 24 strains inferred using these five loci to establish congruency in clades defined with individual loci. Where possible at least two strains from each clade were sequenced at all loci (**Table A2** in Appendix). Comparison of the different trees reveals that the same monophyletic clades supported by high bootstrap values were seen across all loci (**Figures A1**– **A5** in Appendix). We therefore created a phylogenetic tree using concatenated sequences of four of these loci (**Figure A6** in Appendix; **Figure 1**) to create consensus assignments of strains to clades (**TableA2** in Appendix). The *narB* gene was excluded from the concatenated analysis because it is absent in some strains (MITS9220 and RS9917).

Sequencing of multiple loci in key strains allowed us to resolve several issues in *Synechococcus* classification. Our phylogenies confirm that strains previously classified to clade VII based on 16S rDNA (Fuller et al., 2003) actually comprise two distinct clades, VII and CRD1 (Ahlgren and Rocap, 2006; Mazard et al., 2012; **Figures A2**–**A5** in Appendix). Strains UW92 and RS9920 accordingly retain their classification to clade VII while strains CC9305, CC9317, CC9318, Eum14, MITS9220 are assigned to clade CRD1 (**Table A2** in Appendix). Second, the framework established here allows placement of newly isolated or previously unclassified strains into clades. We isolated two new strains, UW179 and UW180, from the Equatorial Pacific which fall into clade CRD1. Strain CC9803 (obtained from B. Palenik) is a member of clade XV. Strain M16.17 (Everroad et al., 2006) clearly belongs to clade XVI along with strains UW140 and UW105 (**Figure A1** in Appendix). On the other hand, strain M11.1 (Everroad et al., 2006) is more divergent from other strains than the typical divergence between established clades (**Figure A1** in Appendix), and may represent a novel clade.

Finally, our phylogenies also resolve the identity of a few groups of sequences previously represented only by environmental clones. For example, by sequencing multiple loci in strains UW140 and UW105, we found that a previously obtained environmental sequence from the Red Sea was closely related to clade XVI (**Figure A5** in Appendix) suggesting that this clade may be found in waters other than the Atlantic (Muhling et al., 2006). Similarly, clade XIII, previously defined only by environmental *ntcA* sequences, forms a well supported monophyletic group with clade VII strain UW92 (**FigureA4** in Appendix). We conclude that clades VII and XIII are one and the same and we propose that the label XIII no longer be used. Despite this consolidation of sequences into previously defined clades, several clades represented only by environmental sequences still remain (e.g., clades CRD2, XVII, and XVIII for the ITS; XI and XII for *ntcA*; see **Figure 1**; **Figures A2**, **A4**, and **A5** in Appendix).

Although single gene trees provide strong bootstrap support for individual clades, they did not always give strong support for the branching order of clades relative to one another (**Figures A1**– **A5** in Appendix). However, a few clades do consistently branch together, often with high bootstrap support. For example clades V and VI appear as sister clades in all trees, and clades II and XV also frequently branch together. Our concatenated phylogeny of 24 strains using four loci also supports the existence of two larger subclusters of clades (5.1A and 5.1B) inferred using 11 strains and a phylogeny of >1,000 concatenated core genes (Dufresne et al., 2008). That study assigns clades II, III, IV to subcluster 5.1A and clades I, V, VI, VIII, and IX to subcluster 5.1B. Our four loci concatenated phylogeny supports the 5.1A clustering of clades II, III, and IV and adds clade XV (**Figure 1**; **Figure A6** in Appendix). This phylogeny also suggests that clades I, V, VI, VII, XVI, and CRD1 form a second cluster, but without high bootstrap support (**Figure A6** in Appendix). One difference between these phylogenies is that the core gene phylogeny places clade VIII strain RS9917 within subcluster 5.1B, but in our tree clade VIII strains appear to form a separate lineage, although there is not strong bootstrap support for this placement (**Figure A6** in Appendix).

## **DESIGN, SPECIFICITY AND CALIBRATION OF qPCR ASSAYS**

Quantitative PCR primer sets targeting the 16S–23S rDNA ITS region were designed to quantify 10 clades of *Synechococcus* (**Figure A2** in Appendix; **Table 1**). Primers were designed to be identical to all sequences within each specified clade when possible and have significant mismatches to all other *Synechococcus* ITS sequences.

After the design of these primers, additional *Synechococcus* ITS sequences have been published, and we reevaluated specificity of our primers to all publically available sequences. The degree to which primers match sequences (with 0, 1–2, or >2 mismatches) in each clade are shown in **Figure A2** in Appendix. In most cases, our primers perfectly match the majority of sequences within each clade, and of the 758 total ITS sequences publically available for the 10 clades that our qPCR primers target, 598 or 79% of ITS sequences perfectly match our primers across the respective clades.

It is also important to note that Huang et al. (2012) identified several new sequences similar to clade X and reclassified this group of sequences into clades 5.3-I through 5.3-VI within subcluster 5.3 (see **Figure 1**), such that clade X sequences (e.g., strain RCC307) fall within clade 5.3-I. Our clade X qPCR primers are identical to most sequences within clades 5.3-I through 5.3-III but have two or three mismatches to sequences belonging to subclades 5.3-IV, 5.3-V, and 5.3-VI (**Figure A2** in Appendix; Huang et al., 2012).

After optimizing qPCR reactions for maximal PCR efficiency by adjusting MgCl2 concentration and annealing temperature (**Table 1**), initial lower limits of detection for each assay were set by determining the lowest concentration of plasmid standard at which the *C*<sup>T</sup> values of triplicate reactions still had an error of ≤3%. The specificity of each assay was tested by challenging clade-specific qPCR assays with plasmid standards of other clades and the lower limits of detection were adjusted as necessary. Non-specific plasmids were tested at concentrations of 1.5 <sup>×</sup> 105 copies/μl, equivalent to cell densities of approximately <sup>3</sup> <sup>×</sup> 105 cells/ml. Except for primer sets for clade X, XVI, and CRD1, there was no non-specific amplification above the lowest set of plasmid standards. For clade X, the clade IV standard plasmid amplified nominally and the lower limit of detection was adjusted to two times that value (678 cells/ml, **Table 1**; **Table A3** in Appendix). The CRD1 primer set amplified plasmids for strains UW92 (clade VII), WH 7803 (clade V), and clone CRD8 (clade CRD2) to starting concentrations <239 copies/μl and the lower limit of detection was adjusted to two times 239 μl/copies or the equivalent of 1830 cells/ml (**Table 1**;**TableA3** in Appendix). Crossamplification of the CRD1 set with plasmid oc175m12 (clade XVI) was significant, so a range of oc175m12 plasmid concentrations were quantified (from 1.54 <sup>×</sup> 105 to 3.42 <sup>×</sup> 102 copies/μl). The clade XVI plasmid cross-amplified at a ratio of 1:25 (concentration of non-specific plasmid to concentration quantified, **Table A3** in Appendix). Melt curves of clade XVI plasmids were very similar to those of the clade CRD1 standards (data not shown). The detection limit reported for clade CRD1 is two times the amplification of the WH 7803 plasmid standard (**Table 1**;**Table A3** in Appendix), but because of the high cross-amplification of clade XVI plasmid, clade CRD1 qPCR concentrations must be evaluated judiciously if clade XVI sequences are present at high levels in the same sample.

Plasmid standards used for real-time PCR were calibrated to DNA extractions of known quantities of cell cultures in order to calculate and report concentrations of *Synechococcus* clades in environmental samples in terms of cells per ml. DNA was extracted from filters that were generated with cultures over a range of concentrations from strains presenting six different clades (I, II, III, VIII, and XV, and XVI). ITS concentrations determined by qPCR were plotted against the number of cells in extractions (**Figure 2**). Log–log regressions all had strong correlation coefficients, and the slopes of regressions were all very close to one (**Table 2**). There was some variability between the y-intercepts of the regressions for the different strains tested. Excluding clade XV strains, the calibration curves for the different *Synechococcus* strains had similar y-intercepts (**Table 2**), and calibration points grouped closely together (**Figure 2**). This degree of variability in y-intercepts was similar to the variability seen in previous calibrations of *Prochlorococcus* assays (range of 0.38 versus 0.46 in Zinser et al., 2006). The variability in y-intercepts translates to roughly twofold differences (100.38 <sup>=</sup> 2.4) between calibration curves and likely represents the limitations of qPCR to reliably detect at least twofold changes in copy number (Bar et al., 2003). Clade XV strains had noticeably higher *y*-intercepts than all other *Synechococcus* strains (**Table 2**; **Figure 2**), perhaps because these strains are more resilient to lysis than other clades resulting in lower yields of extracted genomic DNA.

Regressions were used to convert ITS copies quantified in field samples to cells per ml. For the clade XV assay, the regression of all three calibrations (two replicate calibrations for strain UW69 and one for strain UW106) was used (**Table 2**). For clades I, II, III, VIII, and XVI, the regression values of each individual clade were

used in back-calculating cell concentrations. For clades for which calibrations were not done (clades IV, X, CRD1, and CRD2), it was assumed that these clades have similar calibration curves to the other *Synechococcus* clades (excluding clade XV), and the slope and y-intercept from the regression of clades I, II, III, VIII, and XVI together were used.

using plasmid standards. Triplicate filters were made and extracted for each concentration of cells tested. Points represent averages of quadruplicate qPCR reactions performed on each extraction, and error bars represent the standard deviation. Log–log regression lines are shown for all clade XV strains tested (dashed line, UW69 both replicate filter sets and UW106) and all remaining strains together (solid line).

#### **Table 2 | qPCR assay calibration curve statistics (see Figure 2).**


*a Log-log regressions of cells extracted versus ITS copies/*μ*l detected by qPCR (see Figure 2).*

*b95% confidence interval of slope is reported.*

#### **APPLICATION OF qPCR ASSAYS IN THE SARGASSO SEA**

Samples were collected on a transect off of the eastern United States in March of 2002 which traversed from the Sargasso Sea (station 8) through the Gulf Stream (station 16) and into coastal waters south of Massachusetts (stations 17, 19; **Figure 3**). The water column in the Sargasso Sea was well mixed to at least 175 m (**Figure 4**), as is typical for spring (DuRand et al., 2001). The Gulf Stream exhibited a warm surface layer (24˚C) and a small chlorophyll maximum just below the surface mixed layer at 40 m. Station 17 was fairly uniform in temperature, but showed a subsurface peak in chlorophyll fluorescence twice as large as the maximum seen at other stations. Station 19 was located over the continental shelf and the water column was well mixed with chlorophyll fluorescence levels slightly higher than the Gulf Stream and Sargasso Sea. Water temperature at station 17 (14˚C) was significantly colder than at the off-shore stations, and the drop in temperature from station 17 to station 19 (6˚C) was just as large as the shift from off-shore to near-shore stations.

## *SYNECHOCOCCUS* **AND** *PROCHLOROCOCCUS* **POPULATION STRUCTURE**

Total *Synechococcus* concentrations measured by flow cytometry decreased along the transect from the Sargasso Sea to the coast (**Figure 4**). Typical for this time of year, *Synechococcus* concentrations in the Sargasso Sea (14,000–36,000 cells/ml) outnumbered *Prochlorococcus* by 1.5- to 2-fold (DuRand et al., 2001). In contrast, *Prochlorococcus* concentrations in the Gulf Stream

**FIGURE 3 | Map depicting sampling locations (black dots) along a transect in the northwest Atlantic, south of Massachusetts in March 2002.** Bathymetric contours (black lines, depth in m) are shown, and the location of Bermuda is marked with a star. Station 8 was located in the Sargasso Sea and station 16 was located in the Gulf Stream (based on satellite sea surface temperature, data not shown). Stations 17 and 19 represent coastal stations.

were <sup>&</sup>gt;<sup>3</sup> <sup>×</sup> 104 cells/ml, 5- to 10-fold more abundant than *Synechococcus*. Neither genus was seen below 75 m in the Gulf Stream. *Synechococcus* concentrations at station 17 decreased slightly with depth ranging from 5,400 cells/ml at the surface to 934 cells/ml at 150 m. *Synechococcus* concentrations were fairly uniform (1,200– 2,000 cells/ml) at the coastal station 19. No *Prochlorococcus* were observed at stations 17 and 19.

*Synechococcus* qPCR assays were applied to samples from these water columns. Amplicons of environmental samples had melt curves very similar to those of the plasmid standards for all clades detected (data not shown), further supporting the specificity of these assays. The abundance of the various clades reveal drastic shifts in community structure across the transect (**Figures 5** and **6**). Most significant was the shift in dominance of clades II and X in the warmer Sargasso Sea and Gulf Stream stations to clades I and IV at the colder coastal stations (stations 17 and 19). Only clades I and IV were detected at the coldest coastal station 19, but at station 17 several other clades (II, X, XV, and XVI) were detected at low abundance. Clades II, X, XV, and XVI were also present in the Sargasso Sea station. The Gulf Stream station was unique in

that clade III was detected there and nowhere else. Clade II, X, and XVI were also detected in the Gulf Stream.

Total *Synechococcus* cell abundances determined by flow cytometry were compared to the sum of clade concentrations determined by qPCR (**Figure 6**). In the Sargasso Sea, the sum of qPCR results were very similar to total *Synechococcus* densities determined by flow cytometry at all depths (ratio of 0.95–1.3), but the sum of qPCR abundances were greater than flow cytometry numbers for the remaining three stations (ratio of 1.1–5.7).

## **DISCUSSION**

A variety of loci (e.g., 16S, ITS,*cpeA*, *narB*, *ntcA*, *petB*,*rbcL*,*rpoC1*) have been used to assess and classify the diversity of marine *Synechococcus*. A MLSA using seven core genes found phylogenetic congruence between loci, reflecting relatively infrequent recombination between clades and general clonality in *Synechococcus* populations (Mazard et al., 2012). Consistent with this, the individual phylogenies determined here show congruence among five loci commonly used in diversity studies. Because our phylogenies included *ntcA* and the 5 portion of *rpoC1* not covered in the prior MLSA study (Mazard et al., 2012), we were able to show that previously unassigned *rpoC1* sequences belong to clade XVI and that clade XIII defined by *ntcA* is synonymous with clade VII. We also assigned previously unclassified strains to appropriate clades (e.g., strain M16.17). However, there are still many clades across the various loci that are defined only by environmental sequences [e.g., clade XI and XII from *ntcA* (**Figure A4** in Appendix) or clade CRD2 in the ITS (**Figure A2** in Appendix)], and it is possible that some of these clades correspond to one another. Obtaining new

isolates and determining relevant sequences is necessary to further resolve if any of the clades defined by different loci are in fact synonymous.

Establishing that commonly used loci define congruent clades enables comparison of clade abundances measured by molecular assays that use different target loci such as the ITS, as here, *rpoC1* (Tai and Palenik, 2009) and 16S rDNA (Fuller et al., 2003) providing a better global picture of the biogeography of *Synechococcus*. One small caveat in these comparisons is that clade VII as defined by 16S rDNA sequences actually comprises two distinct clades (Ahlgren and Rocap, 2006; Mazard et al., 2012). Here we clarify the assignment of strains to clades VII and CRD1 and establish an independent assay for clade CRD1. However, it is not possible to compare quantitative data from assays currently designed for *narB* (Paerl et al., 2011) with the 16S rDNA, *rpoC1*, and ITS based assays. The *narB* qPCR primers detect sequences assigned to phylogenetic groups, named "A" through "G," that do not follow the conventional classification of clades established by previous studies (Paerl et al., 2008, 2011) and in many cases these groups include multiple clades as defined here and elsewhere. The *narB* primers also have mismatches to the majority of sequences within their putative target groups (**Figure A3** in Appendix). We suggest that these assays probably do not measure biologically relevant groups of sequences.

Phylogenomic analyses have divided marine *Synechococcus* clades into two larger phylogenetic groups (subclusters 5.1A and 5.1B; Dufresne et al., 2008). These subclusters were generally supported in our four gene concatenated tree, but individual gene phylogenies could not resolve the deeper evolutionary history of clades, consistent with other analyses of individual core genes (Mazard et al., 2012). The apparent rapid diversification of *Synechococcus* (Urbach et al., 1998) combined with low phylogenetic signal from the relatively short sequences used in single locus trees makes it difficult to resolve the higher branching order of the clades. Thus, although most individual loci are not adequate for resolving evolutionary relationships between clades, they contain enough variation to resolve fine-scale classification of *Synechococcus* diversity (placement of strains into clades), and can direct the selection of particular strains for genome sequencing to further resolve deeper evolutionary histories.

It has been suggested that subclusters 5.1A and 5.1B may reflect ecologically coherent groups that share common lifestyle strategies (Dufresne et al., 2008; Scanlan et al., 2009). Specifically 5.1A has been labeled as "open-ocean/specialists" and 5.1B as "coastal/opportunists". However, the idea of ecological coherence within groups 5.1A and 5.1B is in conflict with the fact that clade I and IV have nearly identical habitat ranges and appear to be specialized for cold, coastal, and mesotrophic habitats (Zwirglmaier et al., 2007, 2008; Tai and Palenik, 2009; and this study) yet belong two different subclusters, 5.1B and 5.1A respectively. Similarly, clade II appears to be more like an "opportunist" or "generalist" type, but it belongs to subcluster 5.1A that is claimed to contain specialist types. Clade II is found in a variety of habitats, in both oligotrophic and upwelling regions as well as coastal and open-ocean habitats (Zwirglmaier et al., 2008; Huang et al., 2012; this study). We suggest an alternative hypothesis wherein clades with similar lifestyle strategies have arisen independently via parallel evolution since the divergence of the 5.1A and 5.1B lineages. Additional data on clade biogeography and phylogenomics are needed to corroborate this hypothesis and elucidate the mechanism by which disparate lineages may arrive at common adaptive solutions to life in the oceans.

Having established congruency in clade phylogeny among multiple loci, we developed a new comprehensive set of ITS qPCR assays that detect 10 clades of marine *Synechococcus* and tested them in the Western Atlantic. We chose the ITS as a sequence target because it is much more variable than the 16S rDNA, allowing better resolution of closely related clades. Because the ITS has been widely used in diversity studies in the past 10 years, it is the locus with the largest number of sequences available for primer design. Despite its overall sequence variability, the ITS retains conserved regions, likely with structural function (Rocap et al., 2002), that enable robust clade-specific primer design. This is in contrast to protein coding genes where third codon position variability requires either the use of degenerate primers (Tai and Palenik, 2009) or acceptance of a high percentage of mismatches to the target sequences (Paerl et al., 2011). The ITS assays quantify most of the clades detected by previous 16S rDNA based probe methods (I–IV, VIII, X; Fuller et al., 2003) and all clades detected by *rpoC1* qPCR assays (I–IV; Tai and Palenik, 2009). Our new qPCR assays also detect newly characterized clades not covered by previous

methods (XV, XVI, CRD1, and CRD2; Saito et al., 2005; Ahlgren and Rocap, 2006). Overall, the assays are extremely sensitive, able to detect, in some cases, concentrations as little as tens of cells per ml (**Table 1**). In all but one case, these assays are highly specific with little or no cross-amplification of other clades. The one exception is the clade CRD1 primer set which exhibits significant cross-amplification with clade XVI (at a ratio of 1:25), but clade CRD1 can still be reliably quantified when clade XVI is present at low concentrations.

High-throughput sequence analysis methods provide new approaches to quantifying genetic types from field samples. Mazard et al. (2012) have recently utilized sequencing and RFLP analysis of many sequences from *petB* clone libraries to characterize communities, and Post et al. (2011) have used high depth pyrosequencing of the 16S rDNA to track the abundance of *Synechococcus* clades over time. qPCR assays can still provide an advantage over sequence analysis methods in their ability to quantitatively measure abundance over several orders of magnitude and in particular to sensitively measure minority clades (e.g., see **Figure 6**). Less abundant types (i.e., those less than 0.1% of the population, as we have detected here, see **Figure 6**) may be missed or misrepresented by PCR or clone library based methods that require enormous depths of sampling to quantitatively measure rare taxa. There is growing interest in the significance of rare types as potential reservoirs of genomic diversity (Sogin et al., 2006), and indeed rare bacteria can contribute significantly to the metabolic function of ecosystems (Musat et al., 2008).

The new *Synechococcus* qPCR assays successfully detected seven clades in Atlantic water samples, including recently described clades XV and XVI. Clades XV and XVI were detected in the same waters from which we previously discovered these two new clades using both culture isolation and cyanobacterial clone libraries (Ahlgren and Rocap, 2006), as were most of the other clades observed in that previous study. Clade VIII was not detected by qPCR in any samples. This is not surprising since clade VIII appears to be most prevalent in hypersaline habitats (Zwirglmaier et al., 2008; Huang et al., 2012). Clades CRD1 and CRD2 also were not seen in the Atlantic, but the ITS qPCR assays can successfully detect these clades in the Pacific (Ahlgren and Rocap, unpub. data), consistent with their detection using clone libraries in Pacific waters (Saito et al., 2005; Huang et al., 2012).

Summed *Synechococcus* concentrations of clades determined by qPCR were often greater than those determined by flow cytometry. Application of the *rpoC1* qPCR assays to field samples has also produced a similar overestimation of total *Synechococcus* populations (Tai and Palenik, 2009). Those authors speculate that this discrepancy is due to the presence of dividing cells that possess multiple genome copies but that are counted as only one cell by flow cytometry. Although most *Synechococcus* genomes contain two identical copies of the ITS, normalization of plasmid standards to cell cultures takes into account this fact. Some *Synechococcus* strains are known to possess multiple genome copies, and in some cases cells in culture maintain multiple genome equivalents at all times over the light:dark cycle even though field populations typically exhibit cyclical changes in ploidy level due to synchronization of cell cycle to the diel cycle (Binder and Chisholm, 1995; Vaulot et al., 1996; Liu et al., 1998;Jacquet et al., 2001; Holtzendorff et al., 2008; Griese et al., 2011). Natural *Synechococcus* populations can contain cells that contain >2 genome equivalents per cell, and the proportion of the population that has multiple genome copies varies between samples (Liu et al., 1998). Thus variability in ploidy level in natural populations across habitats and possible differences between laboratory cultures and natural populations could potentially explain overestimation of total cell number. Because overestimation by qPCR is typically threefold or less, we contend that significant ecological conclusions can still be drawn from the order of magnitude differences in clade abundances seen across the different sites sampled here.

The significant shift in *Synechococcus* community structure that we see in our Atlantic samples (**Figures 5** and **6**) is consistent with major patterns of clade biogeography inferred by previous studies. We find clade II highly abundant in warmer, off-shore stations, as seen in quantitative assays (Zwirglmaier et al., 2007, 2008) and diversity studies (Huang et al., 2012;Mazard et al., 2012). We found clade III only in surface waters in the Gulf Stream, the warmest station sampled (**Figure 5**). Clade III is typically found in warm oligotrophic water masses, such as the Mediterranean and the Red Sea only during summer (Fuller et al., 2005; Zwirglmaier et al., 2008; Post et al., 2011). Clade III has therefore been classified as an "oligotroph" (Zwirglmaier et al., 2008).

Our qPCR assays also confirm that clades I and IV dominate *Synechococcus* populations in cold, coastal waters (Zwirglmaier et al., 2007, 2008; Tai and Palenik, 2009; Tai et al., 2011; Huang et al., 2012; Mazard et al., 2012). The relative ratio of these clades shifts over seasonal cycles in California coastal waters and between different high latitude locales. In our samples, clade I was always more abundant than clade IV (1.7- to 6.5-fold), similar to observations in the spring off of California (Tai and Palenik, 2009). However in the Arctic, South Pacific, and Eastern Atlantic, clade IV is consistently more abundant than or on par with clade I concentrations (Zwirglmaier et al., 2008). These differences probably point to distinct adaptations between these two cold-adapted clades. Further application of qPCR assays to field samples or in lab competition experiments could provide insight into what controls the co-existence and relative abundance of clades I and IV in colder, coastal habitats.

Results from this Atlantic transect also provide insights into *Synechococcus* community structure that contrast from previous work. Here, clade X together with clade II dominated the warmer off-shore stations, but in previous 16S rDNA surveys, clade X was only rarely detected and never exceeded 10% of total cyanobacterial DNA amplified (Zwirglmaier et al., 2007, 2008). The prevalence of sequences from clades 5.3-I/X, 5.3-II, and 5.3- III in clone libraries constructed from open-ocean locations (e.g., sites UTK211, UTK262 from Huang et al., 2012) together with our data strongly suggests that these clades are adapted to open-ocean habitats.

We detected as many as six clades in a single sample (at coastal station 17, **Figures 5** and **6**) confirming the oft observed pattern that multiple clades of *Synechococcus* co-exist. For example, both here and in other studies, although clade I and IV are dominant in colder coastal waters, they are also present in warmer off-shore waters, albeit at low densities (**Figures 5** and **6**; Zwirglmaier et al., 2008). In addition, several clades such as IX, XV, and XVI have still

only been seen to comprise minor portions of total *Synechococcus* populations. These clades may occur at higher concentrations only under particular conditions. For example in the Red Sea, clade III remains at low concentrations at most times of the year but peaks in abundance during summer, stratified conditions (Fuller et al., 2005; Post et al., 2011).

The qPCR assays developed in this study provide valuable tools to further identify the specific adaptations and niches of *Synechococcus* clades. Statistical analysis of clade abundance and concurrently measured environmental parameters suggests factors such as temperature, depth, light level, and macronutrient concentrations only explain a portion of the variance in clade abundance (Johnson et al., 2006; Zwirglmaier et al., 2008). Other factors not measured, such as trace metal availability and topdown pressure from grazers or phage are likely to be important in the adaptation and ecology of cyanobacterial ecotypes (Mann et al., 2002; Sullivan et al., 2003; Zwirglmaier et al., 2009; Apple et al., 2011; Strom et al., 2011). Only one study has simultaneously followed the diversity of *Synechococcus* and their infecting phage, and although clade abundances were not measured, significant covariation of host and phage diversity was observed, strongly suggesting that phage impact community structure (Muhling et al.,

## **REFERENCES**


2005). Coupling qPCR assays with measurement of these other factors will help illuminate the ecology of *Synechococcus* clades. A mechanistic understanding of clade dynamics within this one group of marine cyanobacteria will enrich our broader understanding of the important drivers that control microbial community structure in the oceans and the mechanisms of co-existence for bacteria in general. Relatively small differences in nutrient utilization or adaptation to temperature between *Prochlorococcus* ecotypes can impact ocean basin-scale distributions (Johnson et al., 2006). Therefore understanding the structure of microbial populations at the ecotype scale of diversity is critical to assessing metabolic functioning of the oceans on a global scale (Follows et al., 2007).

## **ACKNOWLEDGMENTS**

We thank the captain and crew of the R/V Oceanus as well as E. V. Armbrust and R. Olson (chief scientists). We also thank D. Scanlan for cell pellets for strains RS9911 and RS9920 and B. Palenik for cell pellets from strain CC9803 and K. Peterson and C. Williams for assistance with sequencing. This work was supported by NSF grants OCE-0220826 and OCE-0352190 to G. Rocap and a NDSEG fellowship to N. Ahlgren.


of marine *Synechococcus*. *Environ. Microbiol.* 14, 372–386.


Diversity and evolution of phycobilisomes in marine Synechococcus spp.: a comparative genomics study. *Genome Biol.* 8, R259.


group. *Appl. Environ. Microbiol.* 65, 5247–5251.


improved quantitative PCR method. *Appl. Environ. Microbiol.* 72, 723–732.


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

*Received: 14 February 2012; accepted: 24 May 2012; published online: 19 June 2012.*

*Citation: Ahlgren NA and Rocap G (2012) Diversity and distribution of marine Synechococcus: multiple gene phylogenies for consensus classification and development of qPCR assays for sensitive measurement of clades in the ocean. Front. Microbio. 3:213. doi: 10.3389/fmicb.2012.00213*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Ahlgren and Rocap. This is an open-access article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits non-commercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*

## **APPENDIX**

## **Table A1 | Primers and PCR conditions used to amplify and sequence five loci from** *Synechococcus* **strains.**




*(Continued)*

#### **Table A2 | Continued**


*(Continued)*

#### **Table A2 | Continued**


*Accession numbers of sequences generated in this study are in bold.*

*g The genomes of these strain have been sequenced.*

*"JCVI" – sequence retrieved from the J. Craig Venter Institute (https://research.venterinstitute.org/moore/).*

### **Table A3 | Level of amplification when qPCR primer sets were challenged with non-specific template.**


*a Numbers indicate the degree of amplification in terms of copies/*μ*l of the specific standards when qPCR primer sets were challenged with high concentrations (1.5* × *105 copies/*μ*l) of the non-specific templates listed inTable 1 for each clade. Non-specific templates that did not amplify are indicated with "n.a." for no amplification.*

*bPrimer set CRD1 produced a high level of non-specific amplification, so lower concentrations of clade XVI plasmid oc175m12 were also tested. Challenging the CRD1 primer set with 3.5* × *104, 3.5* × *103, and 3.5* × *102 copies/*μ*l of plasmid oc175m12 resulted in amplification to 1,280, 141, and 14.7 copies/*μ*l respectively, or on average 25-fold less than the concentration of plasmid tested.*

**FIGURE A2 | Phylogenetic tree based on 322 positions of the 16S–23S rDNA ITS.** The tree shown is the best scoring tree, as determined by maximum likelihood score, amongst 1000 neighbor joining trees constructed using the HKY + i + g substitution model and with randomized addition of taxa. Clade groupings shown in this neighbor joining tree are also supported by bootstrap analysis using maximum likelihood with the GTRGAMMA model in RAxML (1000 replicates) and neighbor joining and the HKY + i + g substitution model (1000 replicates). Nodes display bootstrap values (ML/NJ) where bootstrap support is >50% in at least one method. Clades were collapsed into groups represented by

trapezoids. The number inside or to the right of the trapezoid indicates the number of sequences in that clade or subcluster. Clade designations are indicated with solid vertical lines, and representative strains within the clades are shown to the left of clade labels. Clade XV forms a monophyletic clade highly supported by bootstrap analysis (80/86, not shown on tree) and groups together with clade II strains. For clades for which qPCR assays were designed, numbers to the right of clade names show the number of sequences that have zero, one to two, or greater than two mismatches to both the forward and reverse qPCR primers combined (0 mismatches/1–2 mismatches/>2 mismatches).

**FIGURE A3 | Phylogenetic tree based on 648 nucleotide positions of the** *narB* **gene.** Tree construction, bootstrap analysis, and taxa and clade labels as the same as described in **Figure A1**. Because of the large number of sequences, selected sequences are grouped together and represented by trapezoids. The number inside the trapezoid indicates the number of sequences in that group. Strains within those groups that have been used in the concatenated tree (**Figure A6**) are shown to the right of the trapezoids. The assignment of *narB* sequences to groups according to Paerl et al. (2011) is indicated with dashed lines. Group A, Group B, and Group G encompass

multiple well-established clades as defined by phylogenies of other loci. Numbers below group names indicate the number of total mismatches that sequences within the group have to the primers and probe sets described in Paerl et al. (2011) (0 mismatches/1 or 2 total mismatches/>2 mismatches). For Group G which contains multiple clades as defined by other loci, the number of mismatches is also given for sequences belonging to clades II, III, and XV. Two primer and probe sets were designed to detect subgroups of Group D, D\_C1, and D\_C2 (designated with blue lines), and the number of mismatches that sequences have to their respective primer probe sets are given.

## Constitutive extracellular polysaccharide (EPS) production by specific isolates of Crocosphaera watsonii

## *Jill A. Sohm, Bethanie R. Edwards, Brian G.Wilson and Eric A.Webb\**

*Department of Biological Sciences, University of Southern California, Los Angeles, CA, USA*

#### *Edited by:*

*Rex Malmstrom, DOE Joint Genome Institute, USA*

#### *Reviewed by:*

*Kathleen Scott, University of South Florida, USA Hans Paerl, University of North Carolina–Chapel Hill, USA*

#### *\*Correspondence:*

*Eric A. Webb, Department of Biological Sciences, University of Southern California, 3616 Trousdale Parkway, AHF 107, Los Angeles, CA 90089-0371, USA. e-mail: eawebb@usc.edu*

Unicellular dinitrogen (N2) fixing cyanobacteria have only recently been identified in the ocean and recognized as important contributors to global N2 fixation. The only cultivated representatives of the open ocean unicellular diazotrophs are multiple isolates of *Crocosphaera watsonii*. Although constituents of the genus are nearly genetically identical, isolates have been described in two size classes, large <sup>∼</sup>5μm and small <sup>∼</sup>3μm cell diameters. We show here that the large size class constitutively produces substantial amounts of extracellular polysaccharides (EPS) during exponential growth, up to 10 times more than is seen in the small size class, and does so under both N2 fixing and non-N2 fixing conditions. The EPS production exceeds the amount produced by larger phytoplankton such as diatoms and coccolithophores by one to two orders of magnitude, is <sup>∼</sup>22% of the total particulate organic C in the culture, and is depleted in N compared to cellular material.The large difference in observed EPS production may be accounted for by consistently higher photochemical efficiency of photosystem II in the large (0.5) vs. small (∼0.35) strains. While it is known that *Crocosphaera* plays an important role in driving the biological carbon (C) pump through the input of new nitrogen (N) to the open ocean, we hypothesize that this species may also contribute directly to the C cycle through the constitutive production of EPS. Indeed, at two stations in the North Pacific Subtropical Gyre, <sup>∼</sup>70% of large *Crocosphaera* cells observed were embedded in EPS. The evolutionary advantage of releasing such large amounts of fixed C is still unknown, but in regions where *Crocosphaera* can be abundant (i.e., the warm oligotrophic ocean) this material will likely have important biogeochemical consequences.

**Keywords:** *Crocosphaera***, extracellular polysaccharides, diazotroph**

## **INTRODUCTION**

Dinitrogen (N2) fixation is the process by which diazotrophs can access the largest nitrogen (N) reservoir on earth, N2 gas, and turn it into a biologically usable form. This process is exclusively prokaryotic, carried out by relatively few organisms, and represents an important source of new N to the N-starved open ocean (Capone et al., 2005). In the tropics and subtropics, the filamentous colony-forming cyanobacterium *Trichodesmium* is the most well studied diazotroph; however, it is now known that unicellular cyanobacteria also contribute significantly to N2 fixation in these areas (Montoya et al., 2004). There are currently three groups of unicellular, diazotrophic cyanobacteria defined: UCYN-A, which is most closely related to *Cyanothece* ATCC 51142 but appears to be a photoheterotroph based on genomic research; group B, or *Crocosphaera watsonii*; and UCYN-C, a sister taxa to UCYN-A (Foster et al., 2007; Zehr et al., 2008; Tripp et al., 2010). Of these three marine groups, only *Crocosphaera* is represented in culture collections.

*Crocosphaera* appears to be widespread in the tropical and subtropical oceans; it has been identified by *nifH* abundance (a gene encoding for a subunit of the enzyme responsible for N2 fixation) in large areas of the North Atlantic (Langlois et al., 2008), and the North Pacific (Church et al., 2008) and by flow cytometry and direct counting in the South Pacific (Campbell et al., 2005; Webb et al., 2009). In the North Atlantic, *Crocosphaera* is not always present; however, when it is, the abundance (determined as copies of *nifH*) can be quite high for an N2-fixer, in the range of 10<sup>4</sup> L−<sup>1</sup> (Langlois et al., 2008). In the North Pacific, *Crocosphaera nifH* abundance appears to be closer to 10<sup>3</sup> L−1, but concentrations up to 104 L−<sup>1</sup> have also been seen (Church et al., 2005, 2008), while in the warm pool of the South Pacific, concentrations up to 106 L−<sup>1</sup> were recently shown (Moisander et al., 2010). Despite the potential biogeochemical importance of *Crocosphaera*, we are only just beginning to understand the ecology and physiology of this genus.

The type strain of group B is *C. watsonii* WH8501 and the laboratory isolates of this genus can be divided into two cell size classes, <4 and >4μm, that are essentially phylogenetically indistinguishable (Zehr et al., 2007; Webb et al., 2009). Nevertheless, the bigger size class produces large amounts of extracellular material that becomes so abundant that it makes the media gel-like (Webb et al., 2009). Herein this material is identified as extracellular polysaccharides (EPS) that is similar in composition to transparent exopolymeric particles (TEP). The major difference between TEP and EPS is that TEP is defined as discrete particles of EPS, while EPS remains as a matrix around cells (Passow, 2002b). Past work suggests that the large *Crocosphaera*-like cells produce

this material in the field; cells of the large size class (∼5–6μm) are retained on an 8-μm pore size filter, and are seen in loose groupings (Webb et al., 2009). Presumably, these cells exist in a matrix of EPS and can therefore be caught on a filter with pore size larger than the cell diameter. Fitting with this hypothesis, similar groupings of *Crocosphaera*-like cells in a matrix were seen in a sample from a coral lagoon in the southwest Pacific Ocean (Biegala and Raimbault, 2008). These observations suggest that *Crocosphaera* produces this EPS *in situ* as well, however it is not known if this production is constitutive, confined to a specific growth phase or impacted by nutrient limitation.

The purpose of EPS production in *Crocosphaera* is unknown; however, in other phytoplankton, it appears that photosynthetically fixed carbon (C) is channeled into exopolymers during nutrient limited growth (Corzo et al., 2000). Many cyanobacteria have been reported to produce EPS, and some strains appear to produce this material exclusively under N2-fixing conditions (Otero and Vincenzini, 2003). While TEP is mainly recognized as a diatom product, it is also produced by cyanobacteria, other phytoplankton, and bacteria (Passow, 2002a; Berman-Frank et al., 2007). These exopolymers are an important aspect of the global C cycle, as they comprise the matrix of marine snow particles that are a major contributor to C flux to the deep sea (Passow et al., 2001).

In this study we tracked exopolymer production in both large and small strains of *Crocosphaera* in culture and compared the values obtained with published data from other marine phytoplankton. Furthermore, we also investigated the controls of EPS production (i.e., could it be increased, reduced, or eliminated) by growing *Crocosphaera* in different nutrient and light regimes, and documented EPS production by *Crocosphaera*-like cells in the North Pacific Ocean.

## **MATERIALS AND METHODS**

## **STRAINS AND CULTURE CONDITIONS**

In this experiment, six strains of *C. watsonii* were tested for EPS production under diazotrophic growth conditions: three strains of the small size class (WH8501, WH0002, WH0401) and three strains of the large size class (WH0003, WH0005, WH0402). These cultures were isolated from diverse locations. These cultures are not currently available axenic and thus were not axenic for the experiments described here; however, EPS production was qualitatively observed when each strain was originally isolated and rendered axenic by John Waterbury. WH8501 was isolated in the South Atlantic, WH0002, WH0003, and WH0005 were isolated in the North Pacific and WH0401 and WH0402 were isolated in the North Atlantic (Webb et al., 2009). Thus, the strains used originate from many different areas, with both EPS and non-EPS producing strains from the same basin. All cultures were grown in triplicate batches on N free YBCII media (Chen et al., 1996), an artificial seawater media, in an incubator with a 14-h:10-h light:dark cycle at 27˚C, and kept in constant motion on a shaker table (∼100 RPM). With the exception of the light/N experiment (see below) all cultures were grown at 75–80μmol m−<sup>2</sup> s <sup>−</sup><sup>2</sup> light intensity; the light/N experiment was carried out at 45μmol m−<sup>2</sup> s <sup>−</sup>1. All cultures were tested for EPS production over the growth cycle, although only four strains (two large and two small) reached

stationary phase before the termination of the experiments after about 3 weeks (these are shown in **Figure 2**). A number of additional experiments were also carried out with a representative of the large (WH0005) and small (WH0401) strains in order to answer the following questions: (1) is EPS production in the large strains controlled by light intensity or the presence of fixed N, as is seen in some other N2-fixing cyanobacteria (Otero and Vincenzini, 2003), and (2) given previous work showing that nutrient limitation can induce EPS production in cyanobacteria (see Berman-Frank et al., 2007), can the levels of production seen in the large strains be induced in the small strains grown in nutrient deficient conditions? The following experiments were carried out to address these questions: (1) quantifying EPS production in WH0005 during exponential growth at a light intensity of half that used in the standard growth curves and on media with no N (−N) and with 1 mM NO− <sup>3</sup> (+N) added (named the light/N experiment), (2) measuring EPS production in a small strain (WH0401) grown on media with 1 mM NO− <sup>3</sup> for purposes of comparison to the experiment with WH0005, and (3) monitoring EPS production in WH0401 under low Fe and PO3<sup>−</sup> <sup>4</sup> conditions (20 and 25 times less than standard YBCII, respectively).

Sampling of cultures was carried out at the beginning of the light period. One sample for cell counts, chlorophyll (chl) *a* and EPS concentration was taken from each replicate culture over the exponential and stationary growth phase – values reported are the average of the three samples from the biological replicates. Cell numbers were counted with an epifluorescent microscope using a hemacytometer where replicate fields were counted until whichever came first: 10 fields or 200 cells. Chl *a* was determined in a fluorometer after samples were dissolved in acetone for 24 h at −20˚C (Parsons et al., 1984). During early exponential growth, cell counts were the more accurate (less variable) way to quantify biomass and growth in the large strain cultures; however, in late log and stationary phase, the cells began to form large groupings (see **Figures 1A,B**) that could not be accurately counted with microscopy, as the distribution of cells in the field was not uniform. Therefore, growth of the large strain cultures was tracked with Chl *a* concentration only in experiments extending into stationary phase while both cell density and Chl *a* was used with the small size strains and in the light/N experiment where growth was tracked in early exponential phase only.

## **EPS QUANTIFICATION**

To visualize and quantitatively measure the EPS, methods developed for staining and quantification of TEP were used (Passow and Alldredge, 1994, 1995b). Briefly, samples for microscopic evaluation of the EPS were filtered onto a 0.4-μm filter and stained for 2 s with a 0.02% Alcian Blue solution in 0.06% acetic acid, followed by rinsing with milli-Q water (Passow and Alldredge, 1994). The filter was mounted onto a slide and examined at 40× magnification on a Zeiss Axiostar plus microscope using light microscopy for the exopolymers and epifluorescent microscopy to visualize the location of the cells with relation to the EPS. Whole images were minimally adjusted in Adobe Photoshop CS4 for contrast, brightness, and hue in order to accentuate the borders of the cells and stained EPS. Samples for quantitative measurement of EPS were stained as above, placed dry into a glass test tube and stored

**DAPI longpass filter set (B,D).** Stain and cell images show the same field of view of the large cell type WH0005 **(A,B)** and small cell type WH0401 **(C,D)** under 40× magnification.

in the dark for <sup>∼</sup>3 weeks until processing. For quantification, the filters with stained samples were placed in 3 mL 80% H2SO4 for 2 h to dissolve the EPS, and the absorbance was read on a Bio-Rad SmartSpec 3000 spectrophotometer (Philadelphia, PA, USA) at 787 nm (Passow and Alldredge, 1995b). Xanthan gum standards were made in ethanol according to Claquin et al. (2008) and values are reported in microgram Xanthan gum equivalents per milliliter (abbreviated μg Xeq. mL−1). The volume filtered was adjusted throughout the growth curve to account for the increase in EPS, 3–5 mL in the early stages of growth, down to 0.2–0.5 mL by the time late log or stationary phase were reached. Chl *a* and cell normalized EPS values for each strain are reported as the average of these values taken over the exponential growth period, as this is the period of active production.

## **ELEMENTAL ANALYSES**

To assess the C and N content of the EPS, 30 mL of log phase WH0005 was centrifuged at 4000 × *g* for 10–15 min to separate the cells from the EPS. This strain was chosen, as it appears to be easier to separate from the EPS than the other isolates. The cell pellet and the supernatant containing the EPS were filtered separately onto pre-combusted GF/F filters and dried. These filters were then pressed into a pellet and analyzed on a Carlo Erba elemental CHN analyzer (Currently – CE Elantech, Lakewood, NJ, USA). The number of cells that did not separate from the EPS was determined by microscopy (as described above) using a subsample of the supernatant, and the final C and N values of the EPS reported were corrected for the C and N present as cells. The total particulate organic C or N (POC and PON) is defined as the C or N in EPS plus cells, as all this material is caught on a GF/F filter.

## **PHOTOSYNTHETIC EFFICIENCY**

To test for intrinsic photosynthetic differences between the two size classes of *Crocosphaera*, the quantum yield of photosystem II (*F*v/*F*m) of the six strains was tested with 3-(3,4-dichlorophenyl)- 1,1-dimethylurea (DCMU) as described in Samuelsson and Öquist (1977). Briefly, triplicate 30 mL cultures of each strain were grown in 50 mL glass tubes in YBCII medium at <sup>∼</sup>75μmol m−<sup>2</sup> <sup>s</sup> <sup>−</sup><sup>2</sup> and the growth was followed with fluorescence. During acclimated exponential growth, the Chl *a* fluorescence was measured in dark adapted cells before (*F*o) and after (*F*m) the addition of DCMU to a final concentration of 10μM using a Turner TD-700 fluorometer. The quantum yield of photosystem II was then calculated as (*F*<sup>m</sup> − *F*o)/*F*m.

## **FIELD SAMPLING**

Large *Crocosphaera*-like cells were investigated at two stations (27˚19.19 N, 157˚46.65 W, and 23˚35.59 N, 157˚40.24 W) in the North Pacific Subtropical Gyre (NPSG) in July 2010 on the R/V Kilo Moana, at depths of 5 and 25 m. About 100–150 mL samples were filtered onto 8μm filters, stained with Alcian Blue as described above, and *Crocosphaera*-like cells and EPS were observed under epifluorescence and light microscopy, respectively. For the second station, the number of large cells on the filter was counted, noting the number that were embedded in stainable EPS.

## **RESULTS**

## **CELL STAINING**

Alcian Blue is a stain specific to acid mucopolysaccharides (Passow and Alldredge, 1995b). The large strains produced material that was stainable with Alcian Blue (**Figure 1**), therefore we conclude that it is chiefly composed of polysaccharides and is similar to the Alcian Blue stainable exopolymers released by other phytoplankton (EPS and TEP; Passow, 2002b). Microscopic observations showed that large strains (e.g., WH0005) do not appear to be releasing particles of exopolymers (or TEP) into the media, but rather a matrix loosely bound to the cells (or EPS; **Figures 1A,B**). To be consistent with the literature, the previously designated ECM of *Crocosphaera* (Webb et al., 2009) will be referred to as EPS. Images of a small strain representative (WH0401) show little to no stainable material under microscopic observation during stationary phase (**Figures 1C,D**).

## **EPS PRODUCTION**

To determine the timing and magnitude of EPS production in *Crocosphaera*, three strains each of the large and small size class were grown under N2-fixing conditions (no N added) and EPS production was quantified. Two of the large strains, WH0005 and WH0402, grew at the same rate based on the increase in Chl *a*, 0.29 day−1, while WH0003 grew somewhat slower in the same conditions, at a rate of 0.17 day−1. Regardless of the difference in growth rate, all three large strains showed EPS production during exponential growth that increased exponentially at a rate similar to its growth rate: EPS increased in large cell cultures at a rate of 0.31, 0.26, and 0.20 day−<sup>1</sup> in WH0005, WH0402, WH0003, respectively. The amount of EPS produced by the cells during exponential growth, normalized to Chl *a* content, averaged 374 ± 33μg Xanthan gum equivalents (μg chl *<sup>a</sup>*)−<sup>1</sup> for WH0005, 209 <sup>±</sup> <sup>92</sup>μg Xanthan gum equivalents (μg chl *<sup>a</sup>*)−<sup>1</sup> for WH0402, and 392 <sup>±</sup> <sup>116</sup>μ<sup>g</sup> Xanthan gum equivalents (μg chl *a*)−<sup>1</sup> for WH0003 (**Table 1**), and in the cultures of WH0005 and WH0003, plateaued just after cultures reached stationary phase (**Figure 2**). This pattern was also

seen in two of the small strains, WH0401 and WH0002 (**Figure 2**). This shows that EPS production occurs mainly in exponential phase in both large and small strains and stationary phase production is not a large contributor to EPS production in *Crocosphaera*. While visual observations indicated that the small size class strains were not producing EPS, when measured spectrophotometrically, all three small strains tested (WH8501, WH0002, and WH0401) did produced small amounts of EPS, <sup>∼</sup>10 times less than is produced by the large size class. Accumulation rates were very similar to their growth rates of 0.1 (WH0401) and 0.28 day−<sup>1</sup> (WH0002; **Table 1**; **Figure 2**).

While it is possible that differential cell lysis between large and small size classes could contribute to the order of magnitude difference in EPS production between the two classes, microscopic investigation of large and small strains determined that small, but similar, numbers of lysing cells were identifiable in cultures of both (data not shown). This is consistent with EPS being produced by living *Crocosphaera* cells.

#### **NUTRIENT AND LIGHT EFFECTS ON EPS PRODUCTION**

The addition and depletion of different nutrients in the media and growth on low light levels appeared to have little effect on the amount of EPS produced by representatives of the large and small strains. A comparison of EPS production in cultures of *Crocosphaera* WH0005 grown at low light (half the intensity of the growth curves described above) and with and without fixed N shows that lower light reduces the growth rates of the culture (from 0.29 to 0.20 day−1) but the accumulation of EPS still occurs at a similar rate to growth (0.23 day−1). The total amount of EPS produced per unit Chl *a* was somewhat reduced compared to cells grown on higher light (**Table 1**), but still an order of magnitude greater than produced by small cells. This may be due to higher Chl *a* production in cells grown under lower light conditions, rather than reduced EPS production; however, because cell counts were not possible through the entirety of the growth curve in standard conditions, these data are not available.

The addition of fixed N did not reduce the rate of EPS accumulation compared to the culture grown without fixed N (0.20 vs. 0.21 day−1, respectively, *<sup>p</sup>* <sup>=</sup> 0.80). EPS was also produced in similar amounts in the +N and −N treatments (**Figure 3**, **Table 1**). WH0401, a small strain, also did not alter its EPS production when grown with fixed N, or when grown under Fe or PO3<sup>−</sup> <sup>4</sup> deficient conditions (**Table 1**). Chl *a* and cell normalized


**Table 1 | Exopolymer production by batch cultures of** *Crocosphaera* **in log phase compared to other marine phytoplankton, normalized to both cellular volume and chl** *a* **content.**

*a At maximal growth in semi-continuous culture.*

*bAssuming a cell diameter of 5*μ*m and chl a content of 0.3 pg cell*−*1.*

*c During stationary phase, 15–100*μ*mol m*−*<sup>2</sup> s*−*<sup>1</sup> light intensity. Assuming a cell diameter of 4*μ*m.*

*dAverage during log phase.*

*eLarge* >*4*μ*m cell type.*

*f Small* <*4*μ*m cell type.*

*Error is shown as SD.*

EPS concentrations in WH0401 were not significantly different between the three nutrient treatments and the control (*p* > 0.05).

#### **ELEMENTAL ANALYSES**

To determine the elemental composition of large *Crocosphaera* cells and the EPS they produce, a total of six separate culture replicates of WH0005 grown on YBCII with no fixed N added were analyzed for the C and N content. Cellular content of WH0005 averaged 6.4 <sup>±</sup> 1.5 pg C cell−<sup>1</sup> and 1.3 <sup>±</sup> 0.4 pg N cell−<sup>1</sup> while the EPS averaged 1.8 <sup>±</sup> 0.3 pg EPS-C cell−<sup>1</sup> and 0.12 <sup>±</sup> 0.03 pg EPS-N cell−<sup>1</sup> (**Table 2**). The contribution of EPS to total POC (EPS-C plus cellular C) was 22% during exponential growth. The contribution of EPS to total PON is much less than the contribution to POC (8.5%), as evidenced by an EPS C:N ratio of 25 compared to a cellular ratio of 7.2 (**Table 2**). Despite the possibility that heterotrophic bacteria are contributing N to the EPS value, the EPS C:N ratios determined show that the material is depleted in N relative to the cellular fraction and thus is an incomplete nutrient source.

**Table 2 | Average carbon and nitrogen content of** *Crocosphaera* **WH0005 cells and EPS (normalized to cell number) and the average C:N ratio of cellular material and EPS.**


*a Note that the C:N ratio is the average of the C:N of each sample, rather than the C:N of the average C and N content.*

*Values are shown with SD.*

#### **PHOTOSYNTHETIC EFFICIENCY**

The photochemical efficiency of photosystem II was determined for all six strains of *Crocosphaera* and it was found that the *F*v/*F*<sup>m</sup> of the large strains was <sup>∼</sup>0.5, whereas the small strains were

<sup>∼</sup>0.35 (**Figure 4**). Comparisons among the large and small strains showed that *F*v/*F*<sup>m</sup> within each group was statistically indistinguishable (*p* > 0.3 and 0.1, respectively). Additionally, pairwise *t*-tests between all large and small strains showed that *F*v/*F*<sup>m</sup> was statistically significantly different between these two phenotypes (*p* < 0.05). Hence the large strains show a greater quantum yield of photosystem II than the small strains.

## *CROCOSPHAERA* **AND EPS IN THE FIELD**

Extracellular polysaccharides production by field populations of large size class *Crocosphaera* was investigated using Alcian Blue staining and microscopy. Groupings of large *Crocosphaera*-like cells were observed at two stations in the North Pacific (**Figure 5**). Quantification at the second station showed that 74 and 65% of large *Crocosphaera*-like cells observed were embedded in stainable EPS at depths of 5 and 25 m, respectively. These observations support previous evidence that the large phenotype not only produces EPS in cultures, but in the field as well. These particles of *Crocosphaera*-like cells and EPS were also found in high abundance in a floating sediment trap deployed at 75 m for 24 h.

## **DISCUSSION**

This study shows that *Crocosphaera* strains with cells >4μm produce large amounts of EPS in exponential growth, which is consistent with the qualitative observation previously reported by Webb et al. (2009). However, small amounts of EPS were also

**FIGURE 4 | Quantum yield of photosystem II (***F***v/***F***m) in replicate cultures of six** *Crocosphaera* **strains.** All large strain samples were statistically significantly different than all small strain samples (*p* < 0.05), while among each size class, values were statistically indistinguishable. Error bars show SD.

measured in the three small strains tested. Therefore, the previous conclusion based on visual observations that the small strains produce no EPS, does not appear to be the case. Rather, the phenotypic difference between large and small strains is an order of magnitude difference in EPS production, along with a previously described difference in chl *a*-normalized N2 fixation rates (Webb et al., 2009). This result is remarkable considering that laboratory strains of *Crocosphaera* are nearly identical based on sequence divergence (Zehr et al., 2007; Webb et al., 2009) and that large metagenomic fragments isolated from the subtropical North Pacific showed 97– 99% similarity to the WH8501 genome (Zehr et al., 2007). Thus, despite low apparent genomic diversity, there are two distinct phenotypic groups identified with different potential biogeochemical consequences, and this should be considered when quantifying *Crocosphaera* in the field by microscopy, flow cytometry, or *nifH* sequence abundance.

Early work on dissolved organic C (DOC) production by phytoplankton suggested that it is released in much greater quantities in stationary phase, compared to exponentially growing cultures (Berman and Holmhansen, 1974). However, like *Crocosphaera*, many phytoplankton have since been shown to produce EPS and TEP during the exponential growth phase, for example, the diatoms *Chaetoceros affinis*, *Thalassiosira weissflogii*, and *Nitzchia angularis*, the dinoflagellate *Gonyaulax polyedra* (Passow, 2002a) and the N2-fixing cyanobacterium *Anabaena flos-aquae* (Surosz et al., 2006). Furthermore, Claquin et al. (2008) found that TEP was produced in semi-continuous cultures kept in log phase, and that maximal TEP production occurred at the optimal growth temperature in five out of seven marine phytoplankton species tested, indicating that for some species, TEP production is greatest when growth is optimal, rather than the more commonly held view that it is maximal in unbalanced growth (Nagata, 2000). The production of EPS in *Crocosphaera* fits with this model of extracellular release under optimal conditions and exponential growth. The timing of exopolymer release and response to environmental factors could depend on the function of exopolymers in different species (Passow,2002a). In most cases,the purpose of these materials are not well understood, but some of the proposed functions of exopolymers are the absorption of nutrients (Sutherland, 1988), creation of a protective microenvironment (Reddy et al., 1996), and a sink of excess fixed C during growth imbalance (Otero and Vincenzini, 2004).

While *Crocosphaera* is not atypical in producing EPS during exponential growth, the amount that the large strains WH0005, WH0402, and WH0003 produce, compared to other tested species, is quite large. Conversely, the small strains (WH0002, WH8501, and WH0401) produced EPS quantities comparable to other phytoplankton (**Table 1**). Furthermore, the quantity of EPS produced by WH0005 by the end of the light/N experiment, normalized to cell volume, was 1–2 orders of magnitude greater than Passow (2002a) measured (as TEP) in nine species of marine phytoplankton from four phyla during both exponential and stationary growth (**Table 1**). Chl *a*-normalized EPS production varied somewhat among the three large strains; however WH0003, WH0005, and WH0402 all produced an order of magnitude more than other phytoplankton (**Table 1**). *Phaeocystis antarctica* is the only phytoplankton we found that produces an amount of exopolymeric material equal to or greater than that measured in the three large strains, but it does so in stationary and death phase (Hong et al., 1997), while *Crocosphaera* produces EPS mainly in exponential phase (**Figure 2**). However, the contribution of mucopolysaccharides to total POC of *Phaeocystis* is 2–11% in

exponential phase (Alderkamp et al., 2007), compared to 22% that we found during exponential growth in WH0005. Thus, EPS production appears fundamentally different between these two species with high production values of exopolymers.

A number of strains of *Nostoc*, a freshwater, diazotrophic cyanobacterium,have been shown to produce EPS under N2-fixing conditions, only to reduce or eliminate production when grown on an exogenous N source (Otero and Vincenzini, 2003). It is hypothesized that N2-fixing *Nostoc* are N limited, leading to the fixation of extra CO2 that is shunted to EPS, while growth on fixed N causes CO2 limitation and cessation of EPS production (Otero and Vincenzini, 2004). In contrast, both the large (WH0005) and small (WH0401) strains produced similar amounts of EPS when grown with or without fixed N. This disparity between *Crocosphaera* and *Nostoc* suggests that neither WH0005 nor WH0401 are producing EPS as a sink of overflow production, and instead are producing it constitutively. Furthermore, nutrient depletion for P and Fe did not cause EPS production to increase in a small strain representative (WH0401, **Table 1**), data that stands in contrast to increases in exopolymer production under Fe stress observed in the diazotrophic cyanobacterium *Trichodesmium* (Berman-Frank et al., 2007). It remains to be seen if there are other conditions where EPS production in large or small strain *Crocosphaera* can be altered.

One potential explanation for the difference in EPS produced between the two size classes may be that large strains of *Crocosphaera* have more efficient CO2 fixation mechanisms than small strains, leading them to fix more CO2 than the small strains and release the excess as EPS. While it is not a direct measure of the efficiency of CO2 fixation, the difference in the quantum yield of photosystem II (*F*v/*F*m) between the large and small strains (0.5 vs. 0.35 respectively) shows that the large strains are more efficient at using available photons for photochemistry (Falkowski and Raven, 2007). It is important to note, however, that the connection between quantum yield and EPS is shown here to be correlative, not causative. In fact, recent genomic study has revealed the presence of polysaccharide production genes in the genome of a large strain, but not a small strain (Bench et al., in press). The *F*v/*F*<sup>m</sup> values determined for the large strains were similar to those seen in other phycoerythrin-containing cyanobacterial cultures; *Trichodesmium* IMS101 and three *Synechococcus* strains have *F*v/*F*<sup>m</sup> values <sup>∼</sup>0.55 (Berman-Frank et al., 2007; Six et al., 2007). Considering these data, the large *Crocosphaera* appear to have a "typical" efficiency for phycoerythrin-containing cyanobacteria, while the small strains have a low efficiency by comparison. It remains to be seen, however, why the smaller *Crocosphaera* isolates are less photochemically efficient.

The fact that EPS production has an associated energy cost suggests that it provides some benefit to the large cells. The lower surface area to volume ratio of the large size class could make them more susceptible to nutrient limitation than the small size class. There is evidence that these acidic polysaccharide exudates could act as metal chelators (Geesey et al., 1988), making micronutrients that are scarce in the open ocean more available to the large *Crocosphaera*. Another possibility is that the EPS production by large cells serves as a grazing deterrent to protistan grazers (Liu and Buskey, 2000).

Regardless, this EPS excretion has a number of hypothetical biogeochemical consequences in the field, as we have shown that a majority of large cells produce EPS there. EPS could feed into the microbial loop by providing a C source for heterotrophic bacteria. Additionally, *Crocosphaera* EPS might have some nutritional value as appears to contain some N, but this amount would not likely provide complete N nutrition to organisms feeding on it. Alternatively, particles of EPS with *Crocosphaera* embedded could be grazed directly by larger zooplankton, short-circuiting the microbial loop. This would be a direct input of C and newly fixed N into the classical food chain and up into higher trophic levels. Calanoid copepods collected off coastal California were able to feed on TEP (Ling and Alldredge, 2003), and the cladoceran *Ceriodaphnia cornuta* showed better growth and reproductive performance when feeding on EPS of the freshwater N2-fixing cyanobacterium *Anabaena spiroides* compared to feeding on seston (Choueri et al., 2007), suggesting particles of EPS containing *Crocosphaera* cells could be eaten successfully by larger zooplankton and sustain their growth.

The most well known characteristic of exopolymers is their role in the aggregation of phytoplankton into marine snow (Passow and Alldredge, 1995a). Since the sinking of marine snow

## **REFERENCES**


an important source of new nitrogen to the tropical and subtropical North Atlantic Ocean. *Global Biogeochem. Cycles* 19, GB2024.


is a major pathway of C flux to the deep sea (Passow et al., 2001), *Crocosphaera* produced EPS may provide a substrate for aggregation of smaller particles that could then sink out of the euphotic zone. We have observed large *Crocosphaera*-like cells in EPS in shallow sediment traps in the NPSG, providing evidence for the direct sinking of these particles, and showing that EPS production could have a biogeochemical impact in this way. However, more work on the *in situ* fate of the *Crocosphaera* produced EPS is needed to resolve its absolute biogeochemical impact.

## **ACKNOWLEDGMENTS**

This work was supported by National Science Foundation grants OCE-0623499 and OCE-0825922 to Eric A. Webb, and the NSF USC Research Experience for Undergraduates grant OCE-0552988 for Bethanie R. Edwards. The authors would also like to thank Will Berelson and Lisa Collins for carrying out the elemental analysis, Linda Duguay, and the rest of USC REU program team for support and guidance, Chief Scientist Ben Van Mooy for providing the opportunity to collect field data, and the Captain and crew of the R/V Kilo Moana.


charide production in the marine, aerobic nitrogen-fixing unicellular cyanobacterium *Cyanothece* sp. *World J. Microbiol. Biotechnol.* 12, 311–318.


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

*Received: 20 September 2011; accepted: 28 October 2011; published online: 15 November 2011.*

*Citation: Sohm JA, Edwards BR, Wilson BG and Webb EA (2011) Constitutive extracellular polysaccharide (EPS) production by specific isolates of Crocosphaera watsonii. Front. Microbio. 2:229. doi: 10.3389/fmicb.2011.00229*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2011 Sohm, Edwards, Wilson and Webb. This is an open-access article subject to a non-exclusive license between the authors and Frontiers Media SA, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and other Frontiers conditions are complied with.*

## Arsenate resistance in the unicellular marine diazotroph Crocosphaera watsonii

## *Sonya T. Dyhrman\* and Sheean T. Haley*

*Department of Biology, Woods Hole Oceanographic Institution, Woods Hole, MA, USA*

## *Edited by:*

*George S. Bullerjahn, Bowling Green State University, USA*

#### *Reviewed by:*

*Matthew Church, University of Hawaii at Manoa, USA Lisa Moore, University of Southern Maine, USA Eric A. Webb, University of Southern California, USA*

#### *\*Correspondence:*

*Sonya T. Dyhrman, Department of Biology MS#33, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA. e-mail: sdyhrman@whoi.edu*

The toxic arsenate ion can behave as a phosphate analog, and this can result in arsenate toxicity especially in areas with elevated arsenate to phosphate ratios like the surface waters of the ocean gyres. In these systems, cellular arsenate resistance strategies would allow phytoplankton to ameliorate the effects of arsenate transport into the cell. Despite the potential coupling between arsenate and phosphate cycling in oligotrophic marine waters, relatively little is known about arsenate resistance in the nitrogen-fixing marine cyanobacteria that are key components of the microbial community in low nutrient systems. The unicellular diazotroph, *Crocosphaera watsonii* WH8501, was able to grow at reduced rates with arsenate additions up to 30 nM, and estimated arsenate to phosphate ratios of 6:1. The genome of strain WH8501 contains homologs for *arsA*, *arsH*, *arsB,* and *arsC*, allowing for the reduction of arsenate to arsenite and the pumping of arsenite out of the cell. The short-term addition of arsenate to the growth medium had no effect on nitrogen fixation. However, arsenate addition did result in the up-regulation of the *arsB* gene with increasing arsenate concentrations, indicating the induction of the arsenate detoxification response. The *arsB* gene was also up-regulated by phosphorus stress in concert with a gene encoding the high-affinity phosphate binding protein *pstS*. Both genes were down-regulated when phosphate was re-fed to phosphorus-stressed cells. A field survey of surface water from the low phosphate western North Atlantic detected expression of *C. watsonii arsB*, suggestive of the potential importance of arsenate resistance strategies in this and perhaps other systems.

**Keywords: cyanobacteria, phosphorus, marine, arsenate, diazotroph,** *Crocosphaera*

## **INTRODUCTION**

Arsenic is prevalent in the marine environment, where it can occur as arsenite [As (III)] and arsenate [As (V)]. Both oxidation forms are toxic to the majority of living organisms, with arsenite interfering with enzyme function, and arsenate, the thermodynamically dominant form in most oxygenated surface waters (as AsO3<sup>−</sup> <sup>4</sup> ), behaving as a phosphate analog (Apte et al., 1986) and disrupting phosphate uptake and utilization (Wängberg and Blanck, 1990; Bleeker et al., 2003).

In the surface ocean, arsenic has relatively high concentrations and follows a nutrient-like profile, with less arsenic at the surface than at depth suggestive of biological uptake and transport (Andreae, 1979; Cutter and Cutter, 1995). Due to the potential competitive inhibition between arsenate and phosphate (Smedley and Kinniburgh, 2002; Manomita et al., 2004), arsenate resistance, or detoxification pathways may be particularly important in low phosphate systems like the ocean gyres, including the western North Atlantic where surface phosphate levels are typically around 5 nM and can be sub-nanomolar (Wu et al., 2000; Lomas et al., 2010). Surface inorganic arsenic (arsenate and arsenite) concentrations in this region can be >12 nM (Cutter and Cutter, 1995) with an arsenic to phosphate ratio thus exceeding 2:1.

It has been proposed that under low phosphate conditions, microbial uptake of arsenate and its subsequent reduction to and

excretion as arsenite is rapid (Andreae, 1979). There are relatively few studies of arsenic speciation in the ocean gyres, but of the studies that have been done, the speciation of dissolved arsenic (arsenite, arsenate, and methylated arsenic forms) was subtly controlled by the arsenate to phosphate ratio (Cutter and Cutter, 2006). The microbial detoxification products of arsenic including arsenite and methylated forms of arsenic have surface maxima, and in a study of profiles in the western Atlantic tracked closely with the chlorophyll*a* profile (Cutter and Cutter, 1995). These detoxification products have a short (days–months) residence time and their presence suggests active microbial arsenic detoxification in the surface ocean. Although some arsenic accumulation can occur in microbes (Statham et al., 1987), arsenic distributions suggest that the uptake of arsenic by phytoplankton primarily results in cycling between chemical forms within the euphotic zone and that the flux of arsenic into the deep ocean by particle transport is relatively minor (Andreae, 1979; Sanders and Windom, 1980).

Because of the prevalence of arsenic in the environment and its toxicity, many microbes carry well-described pathways for arsenate detoxification or resistance. The most common arsenate resistance system (*ars*) involves an arsenate reductase (ArsC), an arsenite efflux pump (ArsB or ACR3), and a transcriptional repressor (ArsR; Rosen, 1999; Mukhopadhyay et al., 2002; Busenlehner et al., 2003; López-Maury et al., 2003). Some bacteria also

possess other *ars* genes including *arsA*, an arsenite-stimulated ATPase that results in more efficient arsenite efflux through ArsB (Ordóñez et al., 2005; Silver and Phung, 2005). The presence of arsenate detoxification genes has been reported in a wide variety of picocyanobacteria (López-Maury et al., 2003; Scanlan et al., 2009). Furthermore, *Prochlorococcus* arsenate resistance genes are overrepresented in metagenome sequence from the surface North Atlantic with an elevated arsenate to phosphate ratio relative to the North Pacific (Coleman and Chisholm, 2010). This finding underscores the potential importance of arsenate detoxification, especially in the low phosphorus North Atlantic, yet arsenate resistance strategies have not been investigated in the unicellular marine nitrogen-fixing cyanobacteria.

In the oligotrophic oceans, nitrogen-fixing cyanobacteria are considered "keystone" species (Hewson et al., 2009a), because of their low relative abundance but significance to carbon and nitrogen fixation. Of these nitrogen-fixing cyanobacteria, unicellular diazotrophs are increasingly recognized as critically important to nitrogen cycling (Montoya et al., 2004), but are less well understood relative to larger filamentous nitrogen-fixing cyanobacteria like *Trichodesmium*. Phylogenetic analyses of the nitrogenaseencoding *nifH* sequences obtained from field surveys have shown that there are two distinct groups of the unicellular diazotrophs, one of which (group B) includes *Crocosphaera watsonii* WH8501 (Zehr et al., 2001; Falcón et al., 2002, 2004). *C. watsonii* is widely distributed through the surface waters of the tropical oceans (Church et al., 2005; Zehr et al., 2007), including low phosphate environments like the Sargasso Sea (Hewson et al., 2007). Recent work leveraging the whole genome sequence of *C. watsonii* WH8501 suggests that it has a robust capacity for scavenging phosphorus (e.g., presence of *pstSCAB*, a high-affinity phosphate transport system; Dyhrman and Haley, 2006), and as a diazotroph, it may drive down local phosphorus concentrations via nitrogen fixation. In these low phosphate scenarios there may be coupling between arsenate and phosphate cycling with the possible induction of arsenate detoxification or resistance strategies. Here we used a combination of genomic observations, physiological studies, and field analyses to examine arsenate resistance in *C. watsonii*.

## **MATERIALS AND METHODS CULTURE CONDITIONS**

Axenic *Crocosphaera watsonii* WH8501, previously designated as *Synechocystis* sp. strain WH8501, was obtained from John B. Waterbury at Woods Hole Oceanographic Institution. Cultures were grown at 27.5˚C using a 14:10 h light dark cycle provided by cool white fluorescent bulbs with <sup>∼</sup>65μmol quanta m−<sup>2</sup> <sup>s</sup> <sup>−</sup>1. Unless otherwise noted phosphorus replete (referred to as +P or Replete) cultures were grown in 2 L SO medium (Waterbury et al., 1986), made with a 0.2μm filtered 75% Sargasso seawater base and 45μM K2HPO4. Continued sterility was confirmed by testing for growth of contaminating organisms with a tryptone-fortified medium (Andersen et al., 1991). Growth was monitored by *in vivo* fluorescence on a Turner Designs TD-700 fluorometer.

## **ARSENATE AND PHOSPHATE GROWTH EXPERIMENT**

*Crocosphaera watsonii* WH8501 was grown in triplicate on different concentrations of added arsenate and phosphate in SO medium made as above and amended with no added phosphate (NoP), 500 nM phosphate (LowP), and 45μM phosphate (Replete). Sterile-filtered (0.2μm) arsenate was added as ACSgrade sodium arsenate, Na2HAsO4·7H2O (Chem Service, West Chester, PA, USA) in appropriate quantities to yield the following final concentrations in the NoP treatments: 0, 15, 30, 150, 500, 1000, and 5000 nM. Arsenate was added to the LowP treatments to yield the following final concentrations: 0, 500, 1000, and 5000 nM. No arsenate was added to the Replete cultures as a control. Cells used as the inoculum for all treatments were centrifuged initially for 10 min at 7000 rpm and resuspended in medium without added phosphate to restrict carryover. Where no arsenate or phosphate was added to the cultures, As:P ratios were estimated based on literature values for ambient arsenate (12 nM; Cutter and Cutter, 1995) and phosphate concentrations (5 nM; Jakuba et al., 2008) for the region where the water was collected.

## **ARSENATE ADDITION EXPERIMENTS**

*Crocosphaera watsonii* was grown in 500 mL +P SO medium (as described above) to mid log phase and then equal volumes (25 mL) were dispensed into glass culture tubes. Arsenate was added to triplicate cultures at final concentrations of 0, 15, 30, 150, and 500 nM. The triplicate treatments were pooled and harvested by filtration (0.2μm, 25 mm polycarbonate filters) after a 24 h incubation under the growth conditions detailed above. Samples were flash frozen immediately and later stored in liquid nitrogen for subsequent gene expression analyses. In similar experiments to assay nitrogen fixation, 30 mL of exponentially growing +P *C. watsonii* WH8501 cells were dispensed into 75-mL square polycarbonate culturing bottles. Arsenate was added to triplicate culture bottles at final concentrations of 0, 15, 30, 150, 500, and 1000 nM arsenate.

## **PHOSPHORUS MANIPULATION EXPERIMENTS**

Triplicate +P (45μM) cultures were grown as a scale up as described above and harvested at late log phase by centrifugation (7000 rpm). The cell pellets were pooled and split into two equal volumes, washed twice in either +P or −P (−P; 1μM K2HPO4) medium to restrict carryover and resuspended in 25 mL of the same medium. The resulting +P and −P cell suspensions were divided among 3–500 mL flasks of like medium. Whole flasks were harvested onto polycarbonate filters (47 mm, 0.2μm) at 0, 24, 72, and 120 h after resuspension and flash frozen in liquid nitrogen for subsequent analyses. In an additional experiment, triplicate +P and −P cultures were inoculated each with 2 mL P replete stock that had been centrifuged, split into two equal volumes, washed twice in either +P or −P medium and resuspended in 2 mL of the same medium. Cultures were harvested from mid log phase +P cultures. When growth had ceased in −P cultures, the triplicate cultures were pooled and split into six flasks. Triplicate cultures were re-fed with phosphate to replete P levels (45μM) and the remaining three were unchanged. Flasks were subsampled at 0, 24, 48, 72, 120, and 168 h after phosphate addition, harvested onto polycarbonate filters (25 mm, 0.2μm), and flash frozen in liquid nitrogen for subsequent analyses.

## **NITROGEN FIXATION**

Nitrogen fixation was assayed 24 h after the addition of arsenate (as described above), over the midpoint of the dark period in the incubator by the acetylene reduction assay described by Capone and Montoya (2001). Briefly, all culture bottle caps were replaced with silicone septa. Acetylene was produced by mixing 50 mL of MilliQ water with 15 g calcium carbide and stored in a bladder. Six milliliters of acetylene were injected into the bottles. At times 0, 30, 60, 90, and 120 min after injection, duplicate 0.4 mL samples were removed from the headspace of the bottles and 0.3-mL samples were injected into a Shimadzu GC-8A gas chromatograph and integrated by a Shimadzu CR8A Chromatopac to measure the ethylene peaks. Standards of 9.1 ppm ethylene were used to calibrate the peak heights. Ethylene formed was calculated according to Capone and Montoya (2001) with Bunsen coefficients calculated for a 75% seawater/25% MilliQ water mixture and ambient temperature according to Breitbarth et al. (2004). Nitrogenase activity was calculated as the average rate of ethylene production per volume (nmol C2H4 h−<sup>1</sup> mL−1) with a ratio of 3:1. Growth of cultures was monitored by relative fluorescence with a Turner Designs Aquafluor fluorometer. The nitrogen fixation rate calculated for the control (no arsenate added) was set to one and the nitrogen fixation rates for different arsenate concentrations were compared as a proportion above or below the control.

## **FIELD SAMPLE COLLECTION**

Samples were collected on board the *R/V Bank of Bermuda Atlantic Explorer* in the western North Atlantic (**Figure 1**) in 2006 on the BVAL37 transect from Bermuda to Puerto Rico. At each station,

12 L of near surface (∼5 m) seawater was collected via a Niskin bottle and filtered through 0.2μm Sterivex cartridge filters (Millipore) under low pressure. Samples were flash frozen and stored in liquid nitrogen prior to analyses. Total processing time from water to liquid nitrogen was roughly 20 min. Dissolved inorganic phosphate (DIP) concentration was determined on Niskin-collected water from the same depth and analyzed as described elsewhere with a typical detection limit of 0.5 nM (Lomas et al., 2010).

## **TOTAL RNA EXTRACTION FROM CULTURE AND FIELD SAMPLES**

RNA extractions were performed on cultured*C. watsonii* WH8501 cells with the RNeasy Mini Kit (Qiagen, Valencia, CA, USA) with a small modification to the lysis procedure. Lysis was performed by adding∼500μl zirconium/silica beads (0.5 mm) and vortexing for 10 min at 250 rpm. The RNA was then treated with Turbo DNase (Ambion, Austin, TX, USA) following the rigorous protocol to remove genomic DNA. RNA was quantified spectrophotometrically, and a total of 60 ng of RNA was transcribed into cDNA using random hexamers included in the iScript cDNA Synthesis kit (Bio-Rad, Hercules, CA, USA), according to the instructions. Additional reactions for each sample were set up without reverse transcriptase (RT) to ensure the absence of genomic DNA in no-RT controls.

RNA was extracted from field samples using the RNeasy Mini Kit (Qiagen) with minor modifications in the lysis procedure. Lysis was performed by adding <sup>∼</sup>500μl zirconium/silica beads (0.5 mm), TE/lysozyme buffer (15 mg mL−1), and Buffer RLT (Qiagen) to the Sterivex cartridge. Cartridges were vortexed for 10 min at 250 rpm. The resulting lysate was removed through the Sterivex inlet with a syringe, transferred into a 15-mL conical tube, and processed as per the remaining steps of the RNeasy protocol. Nucleic acid concentrations were quantified using a NanoDrop® ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Field samples typically yielded <sup>∼</sup>35 ng total RNA. For the field samples, RNA was transcribed and amplified in one step (see below) using gene-specific primers (**Table 1**).

## **GENE EXPRESSION**

*Crocosphaera watsonii* WH8501 culture samples were quantitatively assayed by qRT-PCR while field samples were qualitatively assayed by RT-PCR. Primer sets for *arsB*, *pstS*, and *rotA* (**Table 1**)



*a Primers were designed by Ilikchyan et al. (2009).*

*bDecreased by 0.5˚C each cycle until 55˚C was reached, after Ilikchyan et al. (2009).*

were designed based on the sequenced genome of *C. watsonii* WH85011. Primers, designed and utilized by Ilikchyan et al. (2009), were used to amplify the *Synechococcus* spp. *phnD* gene encoding a putative phosphonate binding protein (**Table 1**). All analyses were performed in triplicate in an iCycler iQ Real-time PCR detection system (Bio-Rad) in a total volume of 25μl. The reaction mix contained 12.5μl iQ SYBR Green Supermix (Bio-Rad), 0.2μM primer concentrations, and 1 μl template cDNA (diluted 1:10). Reactions were cycled using a temperature profile of 95˚C for 5 min (1×), 95˚C for 30 s (1×), and 59.5˚C for 30 s (45×). The temperature profile also included a melt curve protocol at the end of the program. In all cases, melt curves were used to confirm single amplification products for the different reactions. In addition, for each sample and primer pair no-RT controls and no template controls were run to test for DNA contamination and specificity of amplification. These negative controls did not typically amplify, but in the rare cases where there was any amplification the measured template concentration was at least two orders of magnitude less than the sample template concentration.

The comparative, or ΔΔCt, method was used to normalize the expression of each gene for the qRT-PCR. The amplification efficiency and validation of the ΔΔCt method were tested for each primer set on cDNA. Amplification efficiency was determined to be acceptable if it fell between 90 and 110% with an *r* <sup>2</sup> of 0.98 or higher. Each primer pair was considered acceptable if the ΔCt of the two slopes was less than 0.1. Samples were normalized to the constitutively expressed *rotA* (ZP\_00514036), encoding a peptidyl–prolyl *cis*-trans isomerase (Hassidim et al., 1992; Thellin et al., 1999). This gene has previously been used as a reference gene with the marine diazotroph *Trichodesmium erythraeum* (Orchard et al., 2009), and the Ct of this reference gene did not vary with plate or treatment in samples normalized to total RNA. Fold changes were calculated using the Relative Expression Software Tool, which accounts for differences in efficiency in its fold change calculation2. For each experiment a single *T*<sup>0</sup> time point sample was used as a calibrator for calculating fold changes. The *p*-values were calculated using a pair-wise fixed reallocation randomization analysis as noted (Pfaffl et al., 2002).

RT-PCR of environmental samples was performed using a OneStep-RT-PCR kit (Qiagen) according to the manufacturer's instructions. Briefly, the reaction mix contained RNase-free water, 5× OneStep-RT-PCR buffer, dNTP mix (containing 10 mM of each dNTP), 0.6μM gene-specific primer concentrations (final), OneStep-RT-PCR enzyme mix, and template RNA (∼10 ng) for a total reaction volume of 20 μl. Reactions were cycled with a temperature profile of 50˚C for 30 min (1×), 95˚C for 15 min (1×), 94˚C for 30 s, 59.5˚C for 30 s, and 72˚C for 1 min (40×), and 72˚C for 10 min. To validate the integrity of the extracted RNA, *Synechococcus* spp. *phnD* amplification was included for each environmental sample. The temperature profile was the same for the reverse transcription and initial PCR activation steps. However, the temperature profile for amplification was modified to

95˚C for 5 min (1×), 95˚C for 1 min, an initial annealing temperature of 65˚C for 1 min decreasing by 0.5˚C each cycle until 55˚C was reached, and 72˚C for 1 min (40×), and extension at 72˚C for 20 min (Ilikchyan et al., 2009). Additionally, reactions were performed without RT to ensure the absence of genomic DNA in the RNA samples. RT-PCR products were resolved on 2% agarose gels. Selected bands were identified, excised, and gel purified using the QIAquick gel extraction kit (Qiagen). Sequencing of PCR products was performed at the University of Maine Sequencing Facility (Orono, ME). All sequence data were analyzed manually using the programs Sequencher (Gene Codes Corporation, Ann Arbor, MI, USA), MacVector (Accelrys, Burlington, MA, USA), and BLASTN3. The sequence of the field amplicon has been deposited in the EMBL Nucleotide Sequence Database<sup>4</sup> with accession number FR873108.

## **RESULTS**

#### **GENOMIC OBSERVATIONS**

The genome of *C. watsonii* WH8501 contains homologs for genes involved in the reduction of arsenate to arsenite and the pumping of arsenite out of the cell (**Figure 2**). A homolog of an arsenite efflux pump, *arsB* (ZP\_00514915) and an arsenate reductase, *arsC* (ZP\_00514912), are present in the *C. watsonii* genome (**Figure 2**)*.* These genes are separated by two genes, *arsH* and *arsH*- , encoding proteins of unknown function (**Figure 2**). The *arsA* (ZP\_00516616) gene, an arsenical pump-driving ATPase, is 87.82% identical to *Synechococcus* PCC 6803. In both *Synechococcus* PCC 6803 and *C. watsonii* WH8501 *arsA* is not contiguous to the other *ars* genes, and in *C. watsonii* WH8501 it is isolated on a relatively small contig. The *C. watsonii* WH8501 genome also contains the *pstSCAB* genes that encode a high-affinity P transport system and an additional putative low affinity phosphate permease (Dyhrman and Haley, 2006).

## **ARSENATE ADDITION EFFECTS ON GROWTH AND NITROGEN FIXATION**

Different concentrations of arsenate were added to *C. watsonii* WH8501 cultures to measure the effect on growth in combination with two different phosphate concentrations. Growth rates were calculated for each biological replicate (e.g., each flask) and then averaged within a treatment. The growth rate calculated for the

3http://www.ncbi.nlm.nih.gov/BLAST 4http://www.ebi.ac.uk/embl/

in this study.

<sup>1</sup>http://img.jgi.doe.gov/pub/main.cgi

<sup>2</sup>http://www.gene-quantification.de/download.html

Replete control (no arsenate added, 45μM phosphate) was set to one and the growth rate for different arsenate and phosphate concentrations were normalized to the Replete growth rate. Significant differences in normalized growth rates were found in different treatments (**Figure 3A**, one-way ANOVA, *p* < 0.0001). *Post hoc* testing at the *p* < 0.05 level (Student Newman–Keuls) found no significant differences in normalized growth rate between the Replete and the cultures grown with no added arsenate, regardless of added phosphate concentration, although yield on day 8 of the experiment was highest in the Replete treatment (**Figure 3B**). All treatments with added arsenate had normalized growth rates significantly lower than the Replete. Four arsenate addition levels were replicated with and without added phosphate; a 2-way ANOVA using these arsenate and phosphate concentrations as factors found no significant effect of phosphate concentration on normalized growth rate (*p* = 0.3341; **Figure 3A**). However, increasing arsenate concentration had a highly significant effect on normalized growth rate (*p* < 0.0001; **Figure 3A**). There was no significant interaction effect between phosphate and arsenate (*p* = 0.9051). For the cultures with NoP (0 nM – typically <sup>∼</sup>5 nM residual phosphate in culture medium), growth rates were depressed in the 15- and 30-nM treatments (**Figure 3A**). The transition from positive growth to negative growth occurred with an arsenate addition between 30 and 150 nM (**Figure 3A**). Short-term (24 h) additions of arsenate in concentrations of 0, 15, 30, 150, 500, and 1000 nM had no affect on *in vivo* fluorescence (data not shown). Further, these same arsenate additions had no acute effect (one-way ANOVA, *p* = 0.9816) on nitrogen fixation rate after 24 h (**Figure 4**).

### **GENE EXPRESSION PATTERNS**

In +P cultures of *C. watsonii* WH8501, expression of *arsB* was higher, and statistically different from the no addition control with the addition of 30 nM (*p* = 0.000; **Figure 5**). In this experiment, fold changes increased as a function of increasing arsenate (**Figure 5**). The expression of *arsB* was higher in −P cultures of *C. watsonii* WH8501 relative to the +P control by 72 h (*p* = 0.000), and further increased by 120 h (**Figure 6**). Consistent with previous work (Dyhrman and Haley, 2006), the expression of *pstS* in the −P cultures was statistically higher than in the +P cultures by 120 h (*p* = 0.000), although overall expression levels of *pstS* were low relative to *arsB* (**Figure 6**). When P was re-supplied to P-stressed cultures, *arsB* (*p* = 0.000), and *pstS* (*p* = 0.012; data not shown) were statistically lower within 48 h and remained low (*arsB*; *p* = 0.013 and *pstS*; *p* = 0.044) over the course of the 168-h experiment (**Figure 7**).

#### **FIELD DETECTION OF** *C. watsonii ARSB*

Three near surface samples were examined for the expression of the *C. watsonii arsB* gene in an oligotrophic region of the western North Atlantic during a transect from Bermuda to Puerto Rico in October 2006 (BVAL 37). The field stations examined had DIP concentrations <10 nM (**Table 2**). A *C. watsonii arsB* transcript was detected at station 12, and confirmed with sequencing of the amplicon, which was 100% identical to the *arsB* WH8501 genome sequence. *C. watsonii arsB* transcripts were not detected at stations 8 and 11. Amplification was not observed in the no-RT controls for the stations tested, and expression of the *Synechococcus* sp. *phnD* transcript was detected at all stations (**Table 2**).

## **DISCUSSION**

Arsenate is common in the ocean with surface water concentrations from approximately 5–20 nM (Andreae, 1979; Cutter and Cutter, 1995). Arsenate and phosphate share the same oxidative state and as a consequence, enzymes, and proteins may bind arsenate and move it into the cell by phosphate transport systems with little selectivity (Tawfik and Viola, 2011). Although some microbes thrive in high arsenate (Brunskill et al., 1980; Thiel, 1988; López-Maury et al., 2003), marine microorganisms, which must take up phosphate to supply their metabolic needs, face a unique problem in oligotrophic regimes, where phosphate concentrations are low and arsenate may equal or exceed the phosphate concentration

**FIGURE 3 |** *C. watsonii* **WH8501 growth responses to arsenate (As). (A)** Growth rate of *C. watsonii* WH8501 normalized to the Replete control with 45μM phosphate (P) and no added As. Treatments consisted of different concentrations of As added to medium with either LowP (500 nM) or NoP. Error bars are the SE of the mean (*n* = 3). Significantly lower

growth rates from the Replete control (*p* < 0.05) are indicated (\*) for each treatment using a 1-way ANOVA. **(B)** *C. watsonii* WH8501 relative fluorescence as a function of day in growth curves for each P and as treatment. Treatment key is ordered by the yield on day 8. Error bars are the SE of the mean (*n* = 3).

(Johnson and Pilson, 1972). Nitrogen-fixing marine cyanobacteria are often present in low phosphate environments, and exhibit evidence of P stress-induced adaptations in some regions like the western North Atlantic (Dyhrman et al., 2002, 2007; Dyhrman and Haley, 2006). It has also been hypothesized that they may draw down phosphate to the extent that they induce phosphate deficiency in local microbial populations (Hewson et al., 2009b). In these systems, cellular arsenate resistance strategies would allow diazotrophs to ameliorate the effects of arsenate transport into the cell.

Despite the potential coupling between arsenate and phosphate cycling in marine waters, relatively little is known about arsenate resistance in the nitrogen-fixing marine cyanobacteria that are common in oligotrophic systems. The unicellular diazotroph, *C. watsonii* WH8501, was able to grow, but with reduced rates, with arsenate additions up to 30 nM, and estimated arsenate to phosphate ratios up to 6:1. Cell abundance declined rapidly in

**FIGURE 6 | Expression patterns for** *arsB* **and** *pstS* **in cultures transferred into either phosphate replete (+P) or no added P (−P) medium.** The expression ratio is the fold change of triplicate measurements between the sample expression and the calibrator expression (e.g., the zero time point). Error bars indicate the SE of triplicate measurements of the expression ratio. Significance (*p* < 0.05) is indicated (\*) for the −P versus the +P culture at a single time point for pair-wise fixed reallocation randomization analysis.

treatments with 500 nM arsenate and phosphate together at a 1:1 ratio, suggesting that *C. watsonii* is sensitive to the absolute concentration of arsenate and not simply the arsenate to phosphate ratio. Several cyanobacterial species, such as *Anabaena variabilis* and *Synechocystis* strain PCC6803 are able to grow in the presence of high concentrations of arsenate (up to 75 mM) and in lowmillimolar concentrations of arsenite, respectively (Thiel, 1988; López-Maury et al., 2003). At this juncture, *C. watsonii* WH8501 appears to be more sensitive to arsenate additions, ceasing net growth with additions of 500 nM arsenate in medium with 500 nM phosphate, and with additions of only 30 nM arsenate in medium made from low phosphate Sargasso Sea seawater with no added P. The selectivity of phosphate transporters for phosphate relative to arsenate can be variable (Budd and Craig, 1981; Takahashi et al., 2001; Tawfik and Viola, 2011) and was not examined herein, but



*ND, not determined – there was not enough template for this sample to perform a no-RT reaction for arsB; DIP, Dissolved inorganic phosphate; BLD, Below detection limit.*

these growth data suggest that arsenate is able to enter the cell and effect net growth rate in both 0 nM added and 500 nM added phosphate treatments. Phosphate may be transported by *C. watsonii* WH8501 using a low affinity phosphate permease system under high phosphate, and via the canonical *pstSCAB* high-affinity system under low phosphate (Dyhrman and Haley, 2006). These growth data are consistent with this hypothesis, given that the arsenate exhibited toxic effects on growth and was likely transported into the cells in both the 500- and 0-nM treatments. This is also consistent with the sensitivity of *C. watsonii* to arsenate compared to other cyanobacteria, as the putative low affinity phosphate permease transport system is not common in the other marine cyanobacteria (Dyhrman and Haley, 2006).

Arsenate metabolism and the related genes are well studied in a diverse array of microbes (Stolz et al., 2006). A conserved strategy for arsenate detoxification among the marine picocyanobacteria examined to date is the reduction of arsenate to arsenite within the cell by means of an arsenate reductase and subsequent removal by an arsenite efflux pump (ArsB; Scanlan et al., 2009). The required genes are broadly distributed in the marine picocyanobacteria *Prochlorococcus* and *Synechococcus* (Scanlan et al., 2009). The arsenate resistance genes are also present in *C. watsonii* WH8501, which has homologs for *arsA* and *arsCHH*- *B*. The *arsA* gene is not contiguous with the other ars genes in the genome. It encodes an arsenical pump-driving ATPase*, arsC* encodes a putative arsenate reductase, and *arsB* encodes the arsenite efflux pump. The ArsH protein has been shown to be required for resistance to arsenite and arsenate in some bacteria, such as *Yersinia enterocolitica* (Neyt et al., 1997), and *arsH* is present in the *arsBHC* operon involved in arsenic resistance in the cyanobacteria *Synechocystis* sp. PCC6803 (López-Maury et al., 2003), but its role in detoxification is unknown.

The short-term addition of arsenate to high phosphate cultures of *C. watsonii* WH8501 resulted in the up-regulation of the *arsB* gene, demonstrating its responsiveness to arsenate. Expression levels were titered to increasing arsenate concentration, indicating the induction of an arsenate detoxification response that is sensitive to absolute concentration of arsenate. Although *arsB* was significantly induced with arsenate additions of 30 nM or greater, arsenate did not result in acute effect on growth or nitrogen fixation over the 24-h incubation. In fact, there may be no acute effect on nitrogen fixation because *C. watsonii* WH8501 is able to detoxify arsenate. Given the growth responses to longer-term arsenate exposure highlighted above, there is likely an effect on nitrogen fixation in scenarios of chronic exposure, or where the arsenate resistance system is unable to compensate. This could be the result of arsenic accumulation in the cell, which has been shown to occur

with chronic exposure in other phytoplankton (Sanders and Windom, 1980). As such, arsenate toxicity could be a controlling factor on nitrogen fixation in areas where the arsenate to phosphate ratio, or the absolute arsenate concentration, is chronically high. Regardless, *C. watsonii* WH8501 clearly induces *arsB* in response to even low arsenate additions, which underscores the sensitivity of the resistance response to arsenate.

The *arsB* expression patterns examined herein demonstrate that the gene is concomitantly expressed during P stress with *pstS*. The differences in the *arsB* and *pstS* expression ratios, are consistent with the heterogeneity observed between P-responsive genes in *T. erythraeum* (Orchard et al., 2009), and likely reflect different transcript abundances. Expression levels of both genes decline upon P addition to phosphate deficient cells within 48 h. This is somewhat slower than what is observed with *Prochlorococcus* strains (Martiny et al., 2006), and the diazotroph *T. erythraeum* (Orchard et al., 2009), which may be related to the relatively slow doubling times (∼40–60 h) reported for *C. watsonii* strains (Webb et al., 2009). Although the arsB gene is clearly induced in concert with pstS, a known phosphate-responsive phosphate binding protein gene, the arsB in this case may be responsive to the changing arsenate uptake dynamics driven by changes in the expression of the high-affinity phosphate transport system, and the arsenate to phosphate ratio in the medium, rather than as a direct response to phosphate supply.

Unicellular diazotrophs such as *C. watsonii* are increasingly recognized for their contribution to oceanic nitrogen fixation (Zehr et al., 2001; Montoya et al., 2004), despite their low abundance. This low abundance makes tracking *C. watsonii* gene expression patterns in the field a challenge relative to more abundant picocyanobacteria. For example, even at its highest cell densities in *C. watsonii* bloom scenarios, relatively few *C. watsonii* transcripts could be detected in metadata sets (Hewson et al., 2009a). Low field abundances have thus, in part, restricted transcriptomic studies of this diazotroph in the field to surveys of *nifH* transcripts and the one metatranscriptomic study in a high density population highlighted above (Hewson et al., 2009a). Herein, we observed expression of *C. watsonii arsB* in the near surface waters of the tropical western North Atlantic. The likely explanation for why two of the three stations did not have a detectable *C. watsonii arsB* amplicon is low cell abundance, since estimated cell density in this general region of the western North Atlantic is both spatially variable and low; ranging from undetectable to about seven *C. watsonii nifH* copies L−<sup>1</sup> (Langolis et al., 2008; Moisander et al., 2010). *C. watsonii* abundance was not examined on the transect, but *Synechococcus* abundances in this region averaged 10<sup>4</sup> cells mL−<sup>1</sup> in surface waters (Casey et al., 2009), much higher than

typical *C. watsonii* abundances (<<10<sup>3</sup> cells mL−1). The fact that the samples amplified for *Synechococcus* sp. *phnD* suggests that the RNA was of good quality, and either *C. watsonii arsB* was not expressed, or *C. watsonii* was not present at abundances where *arsB* transcripts could be detected. Despite these challenges, the *arsB* transcript detected at station 12 was clearly from *C. watsonii*. Without quantitative data we are unable to ascertain if this transcript was induced to levels that would suggest substantial arsenate detoxification, however, the expression of *arsB* is consistent with the detection of the arsenite detoxification product in the North Atlantic (Andreae and Klumpp, 1979; Sanders and Windom, 1980).

The expression of arsenate detoxification genes has not been comprehensively examined in marine microbial populations, although expression of arsenate reductase, *arsA*, and *arsB* were detected in metatranscriptome data from a *Trichodesmium* consortium collected from the South Pacific (Hewson et al., 2009b). In this case, the *arsA* was from *Trichodesmium*, and the other signals were from other heterotrophic bacteria present in the sample (Hewson et al., 2009b). The presence of these transcripts in the relatively high phosphate environment of the South Pacific may be attributed to changes in P cycling driven by a dense population of diazotrophs (Hewson et al., 2009b), and/or differences in the sensitivity of different microbes to arsenate toxicity. Of the very few*C. watsonii* transcripts detected in this study, none were related to arsenate detoxification (Hewson et al., 2009b). The *arsB* transcript was also not detected in the metatranscriptome of a bloom of *C. watsonii* in the South Pacific (Hewson et al., 2009a). The detection of *C. watsonii arsB* in the western North Atlantic and not the South Pacific is consistent with the relatively high phosphate environment of the South Pacific relative to the western North Atlantic. However it is difficult to cross-compare between

## **REFERENCES**


O. (2003). Mechanisms of arsenate tolerance in *Cytisus striatus*. *New Phytol.* 157, 33–38.


PCR and metatranscriptome studies because of differences in the sensitivities of the two methods, and the lack of quantitative data. Regardless, the field detection *arsB* in the western North Atlantic is permissive of the induction of *C. watsonii* arsenate detoxification strategies in this low phosphate environment.

More detailed, quantitative studies that track the expression of *arsB*, and rates of arsenate reduction are required to examine coupling between the cycling of phosphate and arsenate in diazotrophs like *C. watsonii* in the field, and to determine the impact of phosphate and arsenate supply on *C. watsonii* growth and nitrogen fixation *in situ.* However, the data presented herein are a first step toward these more detailed characterization studies. Taken together, these data identify that *C. watsonii* has an arsenate resistance strategy that is sensitive to changes in arsenate, and phosphate concentration. Chronic arsenate exposure may reduce growth rates despite the presence of a detoxification mechanism in this group of diazotrophs, and the expression of *C. watsonii arsB* in the western North Atlantic suggests that the coupling of the arsenate and phosphate cycles should be examined in greater detail in oligotrophic marine systems.

## **ACKNOWLEDGMENTS**

The authors thank Dr. Elizabeth Orchard (WHOI), Dr. Mike Lomas (BIOS), Dr. Andrew Juhl (LDEO), and Dr. Louie Wurch (WHOI) for helpful discussions and assistance. The authors also thank the captain and crew of the *R/V Bank of Bermuda Atlantic Explorer* for assistance at sea, as well as Harriet Alexander (WHOI) for generating the cruise transect map. Special thanks are due to Abigail Heithoff (WHOI) for conducting nitrogen fixation measurements. This research was funded in part by the National Science Foundation #OCE-0451419, and the Center for Microbial Oceanography: Research and Education.


S. T., and Ammerman, J. W. (2010). Sargasso Sea phosphorus biogeochemistry: an important role for dissolved organic phosphorus (DOP). *Biogeosciences* 7, 695–710.


Martiny, A. C., Coleman, M. L., and Chisholm, S. W. (2006). Phosphate acquisition genes in *Prochlorococcus* ecotypes: evidence for genomewide adaptation. *Proc. Natl. Acad. Sci. U.S.A.* 103, 12552–12557.


expression results in real-time PCR. *Nucleic Acids Res.* 30, E36.


the marine unicellular cyanobacterium *Synechococcus*," in *Photosynthetic Picoplankton,* eds T. Platt and W. K. W. Li (Ottawa: Canadian Department of Fisheries and Oceans), 71–120.


**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: 09 August 2011; paper pending published: 27 August 2011; accepted: 06 October 2011; published online: 25 October 2011.*

*Citation: Dyhrman ST and Haley ST (2011) Arsenate resistance in the unicellular marine diazotroph Crocosphaera watsonii. Front. Microbio. 2:214. doi: 10.3389/fmicb.2011.00214*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2011 Dyhrman and Haley. This is an open-access article subject to a non-exclusive license between the authors and Frontiers Media SA, which permits use, distribution and reproduction in other forums, providedthe original authors and source are credited and other Frontiers conditions are complied with.*

## Benthic cyanobacterial mats in the high Arctic: multi-layer structure and fluorescence responses to osmotic stress

## *Marie Lionard1,2,3\*, Bérangère Péquin1,2,4,5, Connie Lovejoy 1,2,4,5 andWarwick F. Vincent 1,2,3*

*<sup>1</sup> Département de Biologie, Université Laval, Québec, QC, Canada*

*<sup>2</sup> Takuvik Joint International Laboratory, Université Laval – CNRS, QC, Canada*

*<sup>3</sup> Centre d'Études Nordiques, Université Laval, Québec, QC, Canada*

*<sup>5</sup> Institut de Biologie Intégrative et des Systèmes, Québec, QC, Canada*

#### *Edited by:*

*George S. Bullerjahn, Bowling Green State University, USA*

#### *Reviewed by:*

*Michael R. Twiss, Clarkson University, USA*

*Rachael Marie Morgan-Kiss, Miami University, USA*

#### *\*Correspondence:*

*Marie Lionard, Département de Biologie, Université Laval, Québec, QC, Canada G1V 0A6. e-mail: marie.lionard@bio.ulaval.ca*

Cyanobacterial mats are often a major biological component of extreme aquatic ecosystems, and in polar lakes and streams they may account for the dominant fraction of total ecosystem biomass and productivity. In this study we examined the vertical structure and physiology of Arctic microbial mats relative to the question of how these communities may respond to ongoing environmental change. The mats were sampled from Ward Hunt Lake (83˚5.297- N, 74˚9.985- W) at the northern coast of Arctic Canada, and were composed of three visibly distinct layers. Microsensor profiling showed that there were strong gradients in oxygen within each layer, with an overall decrease from 100% saturation at the mat surface to 0%, at the bottom, accompanied by an increase of 0.6 pH units down the profile. Gene clone libraries (16S rRNA) revealed the presence of Oscillatorian sequences throughout the mat, while *Nostoc* related species dominated the two upper layers, and Nostocales and Synechococcales sequences were common in the bottom layer. High performance liquid chromatography analyses showed a parallel gradient in pigments, from high concentrations of UV-screening scytonemin in the upper layer to increasing zeaxanthin and myxoxanthin in the bottom layer, and an overall shift from photoprotective to photosynthetic carotenoids down the profile. Climate change is likely to be accompanied by lake level fluctuations and evaporative concentration of salts, and thus increased osmotic stress of the littoral mat communities. To assess the cellular capacity to tolerate increasing osmolarity on physiology and cell membrane integrity, mat sections were exposed to a gradient of increasing salinities, and PAM measurements of *in vivo* chlorophyll fluorescence were made to assess changes in maximum quantum yield.The results showed that the mats were tolerant of up to a 46-fold increase in salinity. These features imply that cyanobacterial mats are resilient to ongoing climate change, and that in the absence of major biological perturbations, these vertically structured communities will continue to be a prominent feature of polar aquatic ecosystems.

**Keywords: microbial mats, cyanobacteria, polar lake, pigments, molecular phylogeny**

## **INTRODUCTION**

Cyanobacterial mats are well known for their ecological success in extreme environments (Stal and Krumbein, 1985). Some of the most striking examples occur in the polar regions where cyanobacterial mats often dominate the biomass and productivity of freshwater ecosystems (Vincent, 2000). This ubiquitous and abundant distribution has been attributed to the ability of Arctic and Antarctic cyanobacterial communities to withstand extreme conditions of the polar environment including continuous solar radiation (UV and photosynthetically active radiation, PAR) during summer, persistent cold temperatures, and freeze– thaw cycles (Zakhia et al., 2008). These ambient conditions of the Arctic are currently being modified by the impacts of global climate change. Observations and models show faster climatic changes in the Arctic than at other latitudes, with a rate of atmospheric warming over the last four decades that has been up to

three times greater than the global average (Serreze and Francis, 2006), and this trend is likely to continue in the future (IPCC, 2007).

Two major impacts of warming on Arctic lakes are reduced thickness and duration ice cover, and a shift toward an increasingly negative precipitation minus evaporation balance. The latter is favored by increased evaporation at warmer lake water temperatures, as well as the increasing exposure of the lakes to the atmosphere during a longer ice-free season. For some High Arctic lakes this has led to increased salinities,and in some cases,complete evaporation to dryness (Smol and Douglas, 2007). Cyanobacterial mats living in these shallow water ecosystems will be increasingly subject to the osmotic stresses of solute concentration and desiccation. These effects may be further compounded by salt-exclusion during freeze-up of shallow water lakes (Schmidt et al., 1991) and the littoral zone of deeper lakes. Cyanobacteria have long

*<sup>4</sup> Québec-Océan, Québec, QC, Canada*

been known to be tolerant of salinity fluctuations, but to a variable extent among taxa (Mackay et al., 1984), and osmotic effects have not been assessed in the Arctic where the physiological stress may be exacerbated by cold ambient temperatures. Both evaporation and salt-exclusion contribute to high salinity and the polar microbial mats are exposed to higher salinities in liquid water compared to microbial mats at lower latitudes. Reduced ice cover (and its associated snow; Belzile et al., 2001) also results in greater exposure to solar UV radiation, which is increasing over the Arctic as a result of stratospheric ozone depletion (Manney et al., 2011).

The phylogenetic composition of temperate latitude microbial mats has been extensively studied (e.g.,Balskus et al., 2011; Bolhuis and Stal, 2011) and these communities contain diverse cyanobacteria, other Bacteria, and some eukaryotes. However, despite their ubiquitous distribution throughout the Arctic, in lakes, ponds and streams, the community composition, and structure of microbial mats in northern Polar Regions is much less studied. Initial surveys of bacteria (Bottos et al., 2008), cyanobacteria (Jungblut et al., 2010), and one metagenomic study of mats on Arctic ice shelves, has confirmed the diversity of cyanobacteria along with other Bacteria, notably Proteobacteria, and Archaea (Varin et al., 2010). The ice shelf mats also provide a protected habitat for eukaryotes including diatoms, chlorophytes, flagellates, ciliates, and microinvertebrates (Vincent, 2000).

Most studies to date on high latitude microbial mats have focused on their bulk taxonomic, pigment, and physiological characteristics; however there is also evidence of strong vertical gradients in these properties within such communities (e.g., Quesada et al., 1999; Hawes and Schwarz, 2001). Such gradients are also well known from temperate latitude mats. For example, a study conducted on microbial mats of the intertidal zone showed major vertical changes down the mat profile in coloration, pigment content, and phylogenetic composition (Balskus et al., 2011).

Our aims in the present study were to characterize cyanobacterial mats growing at the northern limit of the terrestrial High Arctic, and to evaluate the hypothesis that these communities have a high tolerance to the osmotic stresses that may be increasingly common in the future. We determined the oxygen and pH gradients in mat samples by way of microsensor profiling in two successive years of sampling. We partitioned the mats into their component layers distinguishable by color, and examined the molecular phylogeny (by 16S rRNA gene analysis) and pigment composition (by high performance liquid chromatography, HPLC) of each stratum to evaluate the vertical structure of these communities that thrive in the extreme High Arctic environment.

## **MATERIALS AND METHODS**

#### **STUDY SITE**

Sampling took place in July 2010 and 2011 at Ward Hunt Lake (WHL; 83˚05.297- N,74˚09.985- W),as part of the program"Northern Ellesmere Island in the Global Environment" (NEIGE). This lake is on Ward Hunt Island off the northern coast of Ellesmere Island in the Canadian High Arctic and within Quttinirpaaq National Park. The total area is of the lake is 0.37 km2, with a maximum recorded depth of 7.5 m. At this extreme latitude, ice covers the lake for at least 9 months of the year and there are 147 days of continuous light during summer (Mueller et al., 2005). Microbial mats occur in the seasonally ice-free littoral zone along the northern and western sides of the lake. Although, phytoplankton in this oligotrophic lake are strongly nutrient limited, high concentrations of nutrients have been recorded in the interstitial waters of the mats (Villeneuve et al., 2001), and in 2-week bioassays the mat phototrophs showed no response to nutrient enrichment (Bonilla et al., 2005).

## **FIELD SAMPLING**

Temperature, oxygen, and conductivity of lake water in the littoral zone were measured on July 10, 2011 using a Hydrolab DS5X profiler (Loveland, CO, USA).

The mats were sampled from 10 to 20 cm water depth in July 2010 and 2011. The samples were placed in a cooler with lake water and taken directly to a field laboratory. Microenvironmental gradients within the mats were assessed within 20–40 min of collection using micromanipulator-controlled, 10 μm diameter Unisense microsensors for oxygen and pH (Unisense, Aarhus, Denmark), with accurate positioning to within 10μm. The microsensor measurements were made at 100μm intervals to obtain vertical profiles through the mats.

The cyanobacterial mats samples were separated at the field laboratory into three layers: surface black colonies, an upper pink layer, and a lower green layer (**Figure 1**). Subsamples for HPLC analysis of the three layers were placed into a dry-shipper (previously cooled with liquid nitrogen) in the field and transferred to −80˚C storage until pigment analysis. Samples for DNA extraction were stored in 1.5 mL of buffer (1.8 mL of 40 mmol L−<sup>1</sup> EDTA; 50 mmol L−<sup>1</sup> Tris pH <sup>=</sup> 8.3; 0.75 mol L−<sup>1</sup> sucrose) and kept frozen in the dry-shipper in the field and transferred to −80˚C storage until analysis.

**FIGURE 1 | Photograph of theWard Hunt Lake microbial mats.** The mats were photographed 30 min after sampling, on July 7, 2010. The mat on the left shows the view from the top, with black colonies distributed over the pink surface. The mat on the right has been turned upside down to show the green layer, which extended down into interstices of the underlying rocky substrate. The black colonies were up to 5 mm in diameter.

## **PIGMENT ANALYSIS**

Within 1 month of collection, samples from the three layers of the microbial mat destined for pigment analysis were freeze-dried using a Labconco Freezone 12 (Labconco, Kansas City, USA). Mat subsamples were then weighed and extracted in the dark by grinding for 2 min followed by two sonication treatments for 30 s each at 17 W in 4 mL 90% acetone:water (vol/vol). The samples were then left overnight at −20˚C under an argon gas atmosphere, and the supernatant recovered after vortexing and centrifugation (4150 rpm for 15 min at 4˚C). This was filtered through a 0.2-μm pore size PTFE Acrodisc filter (PALL Corporation, Ann Arbor, MI, USA) and placed in a 2-mL amber vial with an argon gas atmosphere for immediate analysis by HPLC. Each layer of the mat was extracted three more times to ensure the full recovery of all material. Four vials were thus analyzed for each mat sample, and the results summed.

High performance liquid chromatography analysis was performed using a ProStar HPLC system (Varian, Palo Alto, CA, USA) with a Symmetry C8 column (3.5μm pore size, 4.6 mm × 150 mm; Waters Corporation, Milford, MA, USA) at 25˚C, with a C8 guard column (5μm pore size, 3.9 mm × 20 mm; Waters Corporation). Carotenoids were quantified by their absorbance at 450 nm in a Diode Array Detector (350–750 nm) and chlorophylls were detected by fluorescence (excitation at 400 nm and emission at 650 nm). The HPLC separation method was as in Zapata et al. (2000). Pigments were identified based on retention time and spectral comparisons with standards from DHI (Water & Environments, Horsholm, Denmark).

## **MOLECULAR ANALYSIS**

Mat samples for DNA analysis were extracted using a modified saltbased protocol as in Harding et al. (2011). Cyanobacteria-specific primers 27F1 and 809R (Jungblut et al., 2005, 2010) were used to amplify 16S rRNA gene fragments using the polymerase chain reactions (PCR) with Phusion® High-Fidelity Polymerase, buffer, and dNTPs (New England BioLabs, Pickering, ON, Canada) on the iCyclerTM Thermal Cycler (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Denaturation was for 30 s at 98˚C, followed by 29 cycles of 10 s at 98˚C, 30 s at 55˚C, 30 s at 72˚C, and a final extension time of 5 min at 72˚C.

Amplified PCR products were verified by gel electrophoresis then purified with a QIAquick PCR Purification kit (Qiagen, Mississauga, ON, Canada). Amplicons were cloned using a StrataClone PCR Cloning Kit (Stratagene, Agilent Technologies, Santa Clara, CA, USA). Positive clones were transferred to 96-well plates containing Luria Bertani medium with 7% glycerol. The target cloned fragments were amplified using the vector-specific primers M13F and M13R. PCR products were verified by gel electrophoresis and the results visualized using the Bio-Rad Gel Doc imaging system and Quantity One Software (Bio-Rad Laboratories, Hercules, CA, USA). For each library 30–50 clones were sequenced using the vector-specific T7 universal primer (single read) at the Centre Hospitalier de l'Université Laval (CHUL, QC, Canada) using an ABI 3730xl system (Applied Biosystems, Foster City, CA, USA), after purification.

A total of 108 high quality sequences were retained and manually checked using the BioEdit program (Hall, 1999 – version 7.0.5.3). Suspect sequences were checked manually to detect chimeras by separate nBLAST searches of short fragments. The closest matches for each ribotype based on nBLAST searches (Altschul et al., 1990) of the nr/nt GenBank database were selected as reference sequences, additional sequences of cultured species were also added in an attempt to increase the phylogenetic resolution. If the closest match was to an uncultured clone, the closest isolated strain sequence was retrieved and included in our phylogenetic analysis. Following our microscopy identification (see below) of the morphological genus *Dichothrix* we directly sequenced several filaments from one colony, which were picked under a stereo microscope. Four samples were amplified directly using cyanobacterial specific primers 27F1 and 809R and sequenced in both directions using the same primers. The eight separate sequencing reactions yielded the same sequence (CBS4-C1). The phylogenetic tree was constructed following, initial alignment with multiple alignment using fast Fourier transform (MAFFT, Katoh and Toh, 2008) on the EMBL-EBI portal (http://www.ebi.ac.uk/Tools/msa/mafft/), all sequences were then trimmed to 754 nt and realigned using first MAFFT and then ClustalW2, both provided similar arrangements and Nexus guide trees with the 16S rRNA gene sequence of *E. coli* DP170 as the outgroup. The final tree was constructed and bootstrapped (1000 replicates) using the Maximum Likelihood in RAxML program (Version 7.3.0; Stamatakis, 2006; Stamatakis et al., 2008). The evolutionary model used was GTR +i+ gamma (Tavaré, 1986). The sequences are deposited in GenBank (accession numbers JQ249752–JQ249830).

## **MICROSCOPY ANALYSIS**

The microbial mat samples were preserved with a 10% (final concentration) glutaraldehyde–paraformaldehyde solution (Lovejoy et al., 1993) and maintained in the dark at 4˚C until analysis. Samples were dispersed prior to enumeration and diluted in purified water (milliQ water, Millipore, Billerica, USA) to be sedimented in 15 mL cylindrical counting chambers (Villeneuve et al., 2001); and observed by Fluorescence, Nomarski, Utermöhl (FNU) microscopy (Lovejoy et al., 1993). Samples were examined using an Olympus IX71 inverted microscope (Olympus Canada, Richmond Hill, ON, Canada) at 200× and 400× magnification under UV, green, or blue excitation as well as under bright field (Köhler) illumination.

Another microbial mat sample for taxonomic analysis was kept frozen at −20˚C without fixatives. This sample was thawed and dispersed in water prior to examination with an Olympus IX71 inverted microscope at 200× magnification under visible light.

## **SALINITY TOLERANCE EXPERIMENT**

Twelve 150 mL beakers were filled with 100 mL of water sampled from WHL on July 7, 2010, and these lake water samples were then adjusted to the following conductivities with a NaCl solution: 0.25 (original lake water, no salt addition) 0.5, 1, 2.5, 5, 10, 30, 100, 150, 200, 250, and 300 mS cm−1. NaCl solution was used as sodium and chloride are the most abundant dissolved ions in seawater, and the marine aerosols that reach this coastal lake. The experiment was repeated the next year (July 7, 2011) with the same range of salinities, plus an additional treatment of 50 mS cm−1. A 1.3-cm2

sample of freshly collected microbial mat was placed in each beaker, and the active fluorescence parameters *F*<sup>0</sup> (minimum fluorescence) and *F*<sup>m</sup> (maximum fluorescence) were then measured over the subsequent 24 h using a PAM 2000 (Walz, Germany). Photochemical quantum yield [*F*v/*F*<sup>m</sup> = (*F*<sup>m</sup> − *F*0)/*F*m] was calculated according to Butler (1978) to assess the physiological state of the photosynthetic communities within the microbial mats. The measurements were made placing the fluorescence detector at a distance of 6 mm from black communities of the surface. In the 2010 experiment, the mats were also inverted to measure the response of the bottom community.

## **RESULTS**

#### **MICROSENSOR PROFILES**

On July 10, 2010, the lake water at the mat sampling site had a measured conductivity of 250μS cm−1, the pH was 7.45, the water temperature was 5.2˚C, and the oxygen concentration was at 107% of saturation. The microbial mat was composed of three distinct layers: the black layer from the surface to 2.2 mm depth in 2010 and to 1.5 mm in 2011; the pink layer to 4 mm depth; and the green layer to the bottom at 5.5 mm in 2010 and 4.8 mm depth in 2011 (**Figure 2**). Within the mat, the *in situ* percentage of oxygen saturation decreased from 100% at the surface to 56% within the upper black layer, further decreasing to 19% through the middle pink layer and to 0% at the bottom of the green layer. There was a steep gradient within each layer, as indicated by the high coefficients of variation (**Table 1**). The pH slightly increased down the mat profile from 7.45 to 7.52 in the upper black layer, to 7.80 through the pink layer, and to 7.93 in the bottom green layer, with peaks in pH at the midpoint of each layer (**Figure 2A**; **Table 1**). The oxygen and pH profiles were similar in 2011, but oxygen did not fall completely to zero at the bottom of the profile (**Figure 2B**).

### **PIGMENT ANALYSES**

The major pigments as shown by HPLC analysis were scytonemin and its decomposition product reduced-scytonemin, 4-ketomyxol-2- -methylpentoside (keto-mmp), fucoxanthin, myxoxanthin, alloxanthin, zeaxanthin, lutein, canthaxanthin, echinenone, Chlorophyll *a* (Chl *a*), and β-carotene (**Table 2**). Chl *a* concentrations were similar in the pink and black layers, and increased eightfold in the lower layer, from 55 (pink) to 430μg Chl *a* g−<sup>1</sup> dry weight (green). The black and pink layers had similar ratios of pigments per unit Chl *a*, with high ratios of scytonemin, red-scytonemin, keto-mmp, canthaxanthin, echinenone, and βcarotene. Canthaxanthin, echinenone, and β-carotene were still present in the green layer, along with myxoxanthin and zeaxanthin and trace amounts of reduced-scytonemin, scytonemin, and keto-mmp. The eukaryotic pigments fucoxanthin, alloxanthin, and lutein were also recovered from this bottom layer. The concentrations of photoprotective carotenoids (echinenone, zeaxanthin,lutein,and canthaxanthin) and photosynthetic carotenoids (fucoxanthin and alloxanthin) varied among layers. Myxoxanthin and β-carotene have more than one role in the cell and were therefore excluded from this analysis. Photoprotective carotenoids per unit Chl *a* were twofold higher in the black and pink layers relative to the bottom green layer, while the two photosynthetic carotenoids were only detected in the green layer (**Table 2**).

#### **MOLECULAR ANALYSIS**

A clone library was generated for each layer. We obtained 26, 41, and 41 high quality sequences for the black, pink, and green layer respectively for a final number of 108 sequences The cyanobacteria sequences from the three layers were diverse and while several of the sequences had good matches to cultured species for example *Snowella* spp. and *Synechococcus* sp. PC7502, the majority had best matches to other environmental clones, notably those



*Black, Pink, and Green designate the upper, middle, and bottom layers of the mat.*

**Table 2 | Pigment characteristics of the microbial mat layers as defined in Table 1.**


*The pigment ratios are per unit by weight. The photoprotective carotenoids values are for summation of echinenone, zeaxanthin, lutein, and canthaxanthin. The photosynthetic carotenoids are for fucoxanthin and alloxanthin.The abbreviations used are: keto-mmp, 4-ketomyxol-2*- *-methylpentoside; n.d., not detectable.*

recovered previously form WHL sequences (**Figure 3**) reported in Jungblut et al. (2010). Three environmental clusters accounted for the majority of the clones and these were named after the original WHL sequences (WHL-82, WHL-87, and WHL-69). Nearest cultured matches to other sequences included *Nostoc*, *Leptolyngbya*, *Phormidium*, *Oscillatoria*, *Chamaesiphon*, and *Gloeobacter* (**Figure 3**). Oscillatorian sequences were found in all three layers, but in different proportions. The black layer sequences were mostly *Nostoc*, which accounted for 65% of the sequences. *Synechococcus* and *Leptolyngbya* were the second and third most frequent followed by *Pseudanabaena* and *Gloeobacter*. The pink layer was similar to the black layer with *Nostoc* accounting for 50% of sequences, but the second most common sequences were closest to *Pseudanabaena* with 22%, followed by *Gloeobacter*, *Synechococcus*, *Phormidium*, and *Leptolyngbya* and *Chamaesiphon*. The green layer was quite different with lower representation by *Nostoc* (29% of sequences). *Snowella* sequences accounted for 22% and

*Synechococcus* for 27% of the sequences, and taxa with sequences closest to *Pseudanabaena*, *Leptolyngbya*, *Phormidium*, *Oscillatoria*, and *Gloeobacter* were also present in this layer (**Figure 4**). The sequence of *Dichothrix* morphospecies aligned most closely to *Leptolyngbya* in the Oscillatoriales.

## **MICROSCOPIC ANALYSIS**

Diverse morphospecies were observed within the microbial mat. In the black and pink layers, heterocystous *Dichothrix* sp. was well represented, with its usual associated fine filaments. *Nostoc* sp., mostly in small colonies, were also noted. The pink layer contained various Oscillatoriales of diverse filament diameters, within the morpho-genera *Lyngbya*, *Oscillatoria*, *Phormidium*, *Pseudanabaena*, and *Leptolyngbya*. In the lower green layer, the order Synechococcales was represented, specifically, the colonial genera *Snowella* and *Aphanocapsa*. Other than cyanobacteria, pennate diatoms occurred throughout the microbial mat and included the genera *Achnanthes*, *Caloneis*, *Cymbella*, *Denticula*, *Eunotia*, *Navicula*, and *Nitzschia*. Nematodes were observed in the green layer.

## **SALINITY RESPONSES**

The initial *F*v/*F*<sup>m</sup> of the microbial mat in 2010 ranged from 0.35 to 0.39. For the upper black layer, the *F*v/*F*<sup>m</sup> values immediately after the salt amendments defined as *T*<sup>0</sup> remained above 0.25 up to 150 mS cm−1, and fell to less than 0.1 at conductivities of 200 mS cm−<sup>1</sup> and above. After 8 and 24 h of incubation, high *F*v/*F*<sup>m</sup> values (>0.2) were maintained up to 30 mS cm−<sup>1</sup> (**Figure 5A**). A similar set of responses was recorded for the green layer, with high *F*v/*F*<sup>m</sup> values (>0.2) up to 100 mS cm−<sup>1</sup> at *T*<sup>0</sup> and to 30 mS cm−<sup>1</sup> at 8 and 24 h (**Figure 5B**). This experiment was repeated in 2011 for the upper black layer, and the results were similar to those in 2010. *F*v/*F*<sup>m</sup> values above 0.2 were recorded at conductivities up to 50 mS cm−<sup>1</sup> over 4 h. After 2 days of incubation, *F*v/*F*<sup>m</sup> values >0.29 were maintained up to 30 mS cm−<sup>1</sup> and there was a decline to <0.1 at 50 mS cm−<sup>1</sup> (**Figure 5C**).

To test the capacity of the photosynthetic communities to recover from the salinity increase, all mat sections at the end of the first experiment were placed back into lake water with its natural conductivity of 0.25 mS cm−1. The mats that had been previously exposed to conductivities of less than 50 mS cm−<sup>1</sup> showed ongoing *F*v/*F*<sup>m</sup> ratios above 0.2 and values above 0.3 were recorded for prior exposures of up to 10 mS cm−1. There was some evidence of partial recovery in the two most extreme salinities (150 and 200 mS cm−1), where *F*v/*F*<sup>m</sup> values rose from zero to around 0.2 after 4 h (**Figure 6**).

## **DISCUSSION**

## **COMMUNITY COMPOSITION**

Cyanobacteria dominated the microbial mats from WHL, as in many lakes of the polar regions (e.g., Vincent, 2000; Zakhia et al., 2008). Previous studies on this lake have reported taxa from the Nostocales, Oscillatoriales, and Synechococcales (Villeneuve et al., 2001; Bonilla et al., 2005; Jungblut et al., 2010), consistent with our analyses. *Nostoc* was found throughout the mat, producing black globular colonies at the surface. The sequence from filaments of the "*Dichothrix*" sample was found to be within the Oscillatoriales, rather than the Nostocales as expected. Reliably classified

**FIGURE 3 | Phylogenetic tree from clone library and reference sequences.** Sequences from the black (CBS1), pink (CBS2), and green (CBS31 and CBS32) layers are in black, pink, and green respectively. The different sequence designations from the libraries follow the dash, and the numbers of identical sequences are given in parenthesis. The *Dichothrix* filaments CBS4-C1 direct PCR sequence is light blue. Reference sequences are dark blue. WHL sequences are from Jungblut et al. (2010) and cCLA sequences from Harding et al. (2011). Bootstraps values ≥60% based on maximum likelihood are indicated at the nodes.

*Dichothrix* species are not as yet represented in GenBank and further investigation is required to determine if this *Dichothrix*, which had heterocysts and therefore is morphologically a heterocystous Nostocales, has affinities to the non-heterocystis Oscillatoriales. Alternatively, the thin trichomes that appear on the tips of *Dichothrix* colonies may be a separate oscillatorian taxon, which we amplified from our single trichome. Deeper within the mat matrix there were no macroscopic colonies, suggesting a shift toward the uniseriate, filamentous form of *Nostoc* that is often found in association with *Oscillatoria* (Potts, 2000). The ability of *Nostoc* to fix nitrogen, resist desiccation (Mataloni et al., 2000), and form UV-screening pigments (see below) would make it well suited to life at the mat surface in the shallow littoral waters of the lake. While not observed in the present study,*Nostoc* also produces motile hormogones (Broady, 1979), enabling vertical migration to optimize UV-protection deeper within the mat, light-harvesting toward the mat surface, and access to nutrient resources in the mat interior.

In the green layer of the mat, there was an increased presence of taxa that had their closest 16S rRNA gene sequence matches to *Snowella* and *Synechococcus*. These mostly planktonic taxa have previously been reported as metaphyton (Komárek and Komárková-Legnerova, 1992). *Synechococcus* mats are known from hot spring environments (Stal, 2000), but have been less reported from the polar regions.

*Gloeobacter*-like sequences (WHL-69 cluster, **Figure 3**) were found throughout the mat, and have been reported previously from WHL and a pond mat from Ellesmere Island (Jungblut et al., 2010). The sequences in the present study also closely match a set of Gloeobacterales sequences from an epilithic biofilm of a Spanish high mountain lake (GenBank accession number FR667281 and associated sequences). The Gloeobacterales lineage is divergent from other cyanobacteria in a variety of striking ways, including the absence of thylakoids and unique genes for carotenoid biosynthesis (Tsuchiya et al., 2005). However the GenBank reference *Gloeobacter* in culture (strain PCC 7421) is only 93% similar to our sequences, the Jungblut et al. (2010)sequences, and those from the Spanish alpine lakes, implying that this phylogenetically unusual component of the cold biosphere may prove to be a new genus.

Oscillatorian cyanobacteria were found through the mat profile and are well known from Arctic and Antarctic mats. These organisms in culture have an ability to acclimate via pigment shifts to large variations in ambient irradiance including UV radiation (Quesada and Vincent, 1997). Most are cold-tolerant rather than psychrophiles (Tang et al., 1997). The oscillatorians recorded here had close matches to those previously reported in this region of the High Arctic (Jungblut et al., 2010; Harding et al., 2011) and included those that remain distance from cultured groups (WHL-82 clade and WHL-87 clade). Overall our phylogenies highlighted the poor concordance between morphospecies and 16S rRNA gene sequences as previously noted by Jungblut et al. (2010) and others. The limited morphological repertoire of cyanobacteria likely contributes to this poor phylogeny and suggests that this group requires major careful taxonomic revision. The 16S rRNA gene may also lack resolution at higher taxonomic ranks and additional genes need to be included in such revisions. However, despite these shortcomings, at the level of 99–100% similarity many of the

sequences were previously recovered from the same lake and all of their nearest environmental matches were to other cold regions as earlier reported by Jungblut et al. (2010).

## **PIGMENT COMPOSITION**

The HPLC analyses showed clear differences in pigment concentration and composition through the mat profile, and this vertical structure may reflect both species differences as well as physiological differences among the layers (Quesada et al., 1999). Reduced-scytonemin and scytonemin were major pigments of the black layer of the microbial mat, and are well known in the Nostocales. The decrease in scytonemin and reduced-scytonemin with an increase in keto-mmp was consistent with *Pseudanabaena* and other oscillatorians in the pink layer. The photoprotective carotenoids were in highest concentration, both in absolute terms and per unit Chl *a* in the upper two layers where irradiance and oxygen levels were highest, consistent with their role as quenching agents of reactive oxygen species.

The pigment signature of the green layer was in accordance with the appearance of myxoxanthin and zeaxanthin, which are typical for Synechococcales, and the presence of three pigments that are characteristic of eukaryotic taxa: fucoxanthin (found in diatoms and other chrysophytes), alloxanthin (found in cryptophytes), and lutein (characteristic of green algae). Our observations as well as previous taxonomic analyses of these mats have observed pennate diatoms, notably *Cymbella* spp. and *Achnanthes* spp., as well as chlorophytes, notably *Mougeotia* sp. and *Zygnema* sp. (Villeneuve et al., 2001). Canthaxanthin, echinenone, and β-carotene are found in all cyanobacteria, and were detected in all three layers, as expected.

Scytonemin and its reduced derivative (Garcia-Pichel and Castenholz, 1991) contributed the dominant fraction of the total pigment content of the mat, collectively accounting for 2% of the dry weight of the surface community, 0.6% of the pink layer, and 0.2% of the bottom layer. In cultured organisms, scytonemin has been recorded up to 5% of dry weight, and can similarly accumulate to high levels in natural assemblages (Karsten et al., 1998). This pigment has a broad absorbance maximum that peaks in the UV-A waveband but that also extends into the UV-B and PAR wavebands (Proteau et al., 1993). It can absorb up to 90% of incident radiation (Castenholz and Garcia-Pichel, 2000), and thereby acts as an effective sunscreen for the entire community. The Ward Hunt communities therefore seem well protected from any future increases in UV exposure as a result of stratospheric ozone depletion or increased duration of ice-free conditions.

#### **OXYGEN AND pH PROFILES**

The oxygen content of the mats dropped from 100% saturation, to 0% at the bottom of the green layer in 2010, and to below 40% in 2011. This implies that the mat consortia have a strong respiratory demand for oxygen as a result of the bottom layer phototrophs growing under shaded conditions, as well as the high concentration of bacterial heterotrophs typically found in polar microbial mats (e.g., Varin et al., 2010). It may also reflect the strong oxygen sink conditions immediately beneath the mats, where heterotrophic bacteria would dominate. The low photosynthesis/respiration (P/R) balance is likely to extend upward through the mat later in the season when the lake becomes covered by ice and snow and the day length shortens, eventually to continuous winter darkness. In shallow Antarctic ponds, the bottom waters become anoxic during winter-freeze-up, and the mats experience continuous reducing conditions through much of the year (Schmidt et al., 1991). Similar conditions are likely to prevail in WHL during late Arctic autumn and winter.

The pH profiles of the Ward Hunt mats showed an increase with depth, contrary to the decreasing P/R ratios down the profile as inferred from the oxygen data. However the shifts were relatively small, in accordance with the high dissolved inorganic

salinities, and the quantum yield ratio (*F*v/*F* m) was measured immediately (*T* 0), and after 8 and 24 h in 2010 and immediately (*T* 0) and after 4, 6, 24, and 48 h in 2011. The 2010 data are for the surface black layer **(A)** and the bottom green layer **(B)**, and the 2011 data are for the surface black layer only **(C)**. The horizontal dashed lines are for the mean *F*<sup>v</sup> /*F* <sup>m</sup> values calculated for all time intervals at salinity up to 10 mS cm−<sup>1</sup> (vertical dashed lines).

carbon concentrations, and thus buffering capacity, that have been observed in the interstitial waters of these mats. Up to 141 mg C L−<sup>1</sup> were measured in the mats, while the concentrations in the overlying water were around 16 mg C L−<sup>1</sup> (Villeneuve et al., 2001). The pH peaks in each layer might reflect a depth optimum for oxygenic photosynthesis in each community, with the equivalent peak in oxygen obliterated by its rapid diffusive flux downward as a result of the strong vertical gradient in concentration. Although rare, 16S rRNA gene sequences of green sulfur bacteria have been recorded from these microbial mats (Lionard and Lovejoy, unpublished data). However, no marker pigments (bacteriochlorophylls) for photosynthetic sulfur bacteria were detected in the HPLC analyses, implying that any anoxygenic photosynthesis was likely minimal. The pH values throughout the mat were within the neutral to alkaline range in which cyanobacteria are commonly found (Castenholz, 2009).

## **SALINITY TOLERANCE**

The Ward Hunt mats tolerated increased salinity up to a conductivity of 10 mS cm−1, 46 times ambient salinities of the lake water, with little change in photophysiological state as measured by the active fluorescence parameter *F*v/*F*m. For the mats previously exposed to high salinities and then placed in freshwater, this quantum yield ratio returned within hours to the range 0.35–0.39, indicative of excellent physiological conditions for *Nostoc* (Bowker et al., 2002). The mucilaginous, exopolysaccharide sheath enveloping *Nostoc* colonies, is known to provide a buffer from external variations in salinity (Webb et al., 2003; Hall-Stoodley et al., 2004). However, salinity tolerance varies widely among cyanobacteria, which range from species of relatively narrow tolerance (Stal and Krumbein, 1985) to euryhaline species (Nübel et al., 1999). The broad tolerances measured here for both the surface and bottom communities imply that cyanobacteria species in the mats continue to be metabolically active despite large and rapid fluctuations in salinity, effectively increasing the range of environmental conditions where active photosynthesis can occur, with implications for the energy budget within the microbial mats. These results are consistent with measurements of photosynthesis by microbial mats from Arctic (Mueller et al., 2005) and Antarctic ice shelves (Hawes et al., 1992); and Antarctic lakes (Sabbe et al., 2004), which have shown acclimation to greatly increased salinities.

## **CONCLUSION**

The cyanobacterial mats of WHL were suited to resist the major fluctuations in environmental conditions that occur at present in the High Arctic. The layered community structure, with UVscreening pigments and photoprotective carotenoids at the surface, allows a more diverse community including eukaryotes to grow in the bottom green communities of the mats. These bottom communities must contend with low oxygen tensions in summer, and the entire mat community is likely to experience anoxia during winter-freeze-up. The salinity responses measured here indicate broad tolerances and no loss of photosynthetic function under high and fluctuating osmolarities. Previous work on polar microbial mats has shown their tolerance of desiccation (Hawes et al., 1992), and that cyanobacterial isolates from Arctic and Antarctic mats grow faster at temperatures that are well above the present day ambient conditions (Tang et al., 1997). In combination, these results imply that cyanobacterial communities will continue to dominate polar aquatic ecosystems despite ongoing variations in water balance, increased UV radiation, and

prolongation of ice-free conditions. However, these slow-growing mat communities currently experience little grazing pressure, and their persistence may be less assured if the northward expansion of invasive species accompanying climate change (Vincent et al., 2011 and references therein) includes arthropod grazers that feed on these large stocks of benthic microbial biomass.

## **ACKNOWLEDGMENTS**

We acknowledge financial support from the Canada Research Chair in Aquatic Ecosystem Studies, the Network of Centers of

## **REFERENCES**


high Arctic lake: pigment structure and contrasting responses to nutrient enrichment. *J. Phycol.* 41, 1120–1130.


and M. Potts (Dordrecht: Kluwer Acad. Publ.), 591–611.

on the submitted manuscript.

Excellence program ArcticNet, and Natural Sciences and Engineering Council of Canada. Logistical support was supplied by the Polar Continental Shelf Project (PCSP publication number) and we are also grateful to Parks Canada for access to their facilities in Quttinirpaaq "top of the world" National Park. We also thank Denis Sarrazin and Sébastien Bourget for field assistance, Marie-Josée Martineau for laboratory assistance, and Anne-Dorothée Jungblut for advice on the molecular analyses. We also thank the two referees for their insightful review comments and suggestions


and recovery of Antarctic cyanobacterial mats. *Polar Biol.* 12, 587–594.


ecosystem. *FEMS Microbiol. Ecol.* 53, 73–87.


J. D., and Sharp, M. (2011). Ecological implications of changes in the Arctic cryosphere. *Ambio* 40, 87–99.


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

*Received: 29 January 2012; paper pending published: 17 February 2012; accepted: 24 March 2012; published online: 26 April 2012.*

*Citation: Lionard M, Péquin B, Lovejoy C and Vincent WF (2012) Benthic cyanobacterial mats in the high Arctic: multi-layer structure and fluorescence responses to osmotic stress. Front. Microbio. 3:140. doi: 10.3389/fmicb.2012.00140*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Lionard, Péquin, Lovejoy and Vincent. This is an open-access article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits noncommercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*

## Cyanobacteria of the genus Prochlorothrix†

## *Alexander Pinevich\*, Natalia Velichko and Natalia Ivanikova*

*Microbiology Department, Faculty of Biology and Soil Science, St. Petersburg State University, St. Petersburg, Russia*

#### *Edited by:*

*George S. Bullerjahn, Bowling Green State University, USA*

#### *Reviewed by:*

*George S. Bullerjahn, Bowling Green State University, USA Edward Hall, United States Geological Survey, USA*

#### *\*Correspondence:*

*Alexander Pinevich, Microbiology Department, Faculty of Biology and Soil Science, St. Petersburg State University, 29, Lane 16 Vasilievskii island, St. Petersburg, Russia e-mail: pinevich.a@mail.ru*

Green cyanobacteria differ from the blue–green cyanobacteria by the possession of a chlorophyll-containing light-harvesting antenna. Three genera of the green cyanobacteria namely *Acaryochloris*, *Prochlorococcus,* and *Prochloron* are unicellular and inhabit marine environments. *Prochlorococcus marinus* attracts most attention due to its prominent role in marine primary productivity. The fourth genus *Prochlorothrix* is represented by the filamentous freshwater strains. Unlike the other green cyanobacteria, *Prochlorothrix* strains are remarkably rare: to date, living isolates have been limited to two European locations. Taking into account fluctuating blooms, morphological resemblance to *Planktothrix* and *Pseudanabaena*, and unsuccessful attempts to obtain enrichments of *Prochlorothrix*, the most successful strategy to search for this cyanobacterium involves PCR with environmental DNA and *Prochlorothrix*-specific primers. This approach has revealed a broader distribution of *Prochlorothrix*. Marker genes have been found in at least two additional locations. Despite of the growing evidence for naturally occurring *Prochlorothrix*, there are only a few cultured strains with one of them (PCC 9006) being claimed to be axenic. In multixenic cultures, *Prochlorothrix* is accompanied by heterotrophic bacteria indicating a consortium-type association. The genus *Prochlorothrix* includes two species: *P. hollandica* and *P. scandica* based on distinctions in genomic DNA, cell size, temperature optimum, and fatty acid composition of membrane lipids. In this short review the properties of cyanobacteria of the genus *Prochlorothrix* are described. In addition, the evolutionary scenario for green cyanobacteria is suggested taking into account their possible role in the origin of simple chloroplast.

**Keywords: cyanobacteria,** *Prochlorothrix***, Prochlorophytes**

## **INTRODUCTION**

Oxygenic photosynthetic bacteria are represented by blue–green cyanobacteria and green cyanobacteria. This distinction is analogous to that between green chlorosome-containing and purple non-chlorosome-containing anoxygenic photosynthetic bacteria and is based on the supramolecular organization of the lightharvesting complex rather than cell color (certain blue–green cyanobacteria are blue, green, red, violet, or brownish). Blue–green cyanobacteria are characterized by a constantly expressed phycobiliprotein light-harvesting antenna, or phycobilisome, with the exception of inducible mutants and terminally differentiated wild type cells called heterocysts. In contrast, green cyanobacteria have a constitutive chlorophyll-protein light-harvesting antenna instead of the phycobilisome (the former is expressed in some blue–green cyanobacteria in stress conditions). In terms of pigments, green cyanobacteria possess accessory chlorophyll(s), whereas blue– green cyanobacteria have only *chl a* (Partensky and Garczarek, 2003).

Green cyanobacteria are worth interest primarily for two reasons. First of all, *Prochlorococcus marinus* dominates pelagic environments for large regions of the world's oceans. Secondly, the remaining four species (*Acaryochloris marina*, *Prochloron didemni*, *Prochlorothrix hollandica*, and *Prochlorothrix scandica*) are limited to specific eutrophic habitats, in contrast to ubiquitously distributed *P. marinus* and blue–green cyanobacteria (Partensky et al., 1999).

Cyanobacteria of the genus *Prochlorothrix* are unique among green cyanobacteria being filamentous and freshwater, as opposed to the other three marine unicellular species.

In this review, the properties of cyanobacteria of the genus *Prochlorothrix* are summarized, and the phylogeny of green cyanobacteria is reconsidered in the light of their role in the origin of simple chloroplast.

#### **SCOPE OF OBJECTS**

#### **CIRCUMSTANCES OF DISCOVERY**

Green cyanobacteria were initially discovered by the chromatographic detection of the accessory *chl b* and/or other accessory chlorophylls (*a*2, *b*2, *c*, etc) in collected samples. This approach lead to the discovery of two unicellular species *P. didemni* (Lewin, 1981) and *A. marina* (Miyashita et al., 1996), as well as the filamentous species *P. hollandica* (Burger-Wiersma et al., 1989) and *P. scandica* (Pinevich et al., 1999). The third unicellular species,

<sup>†</sup>The GenBank accession number for the 16S rRNA gene sequence of *Prochlorothrix* sp. NIVA-8/90 is HQ316169. The accession number for the sequences of the single 16S–23S-ITS, *psbO* and *PsaF* genes in *P. hollandica* PCC 9006 are JQ739143, JQ739144, and HQ316172 respectively. The accession numbers for the sequences of two 16S–23S-ITS in *Prochlorothrix* sp. NIVA-8/90 are HQ316170 and HQ316171 respectively. The accession number for the partial 16S rRNA sequence of the uncultured *Prochlorothrix* of the Finnish Bay is JQ739142.

*P. marinus*, was revealed by flow cytometry in oceanic picoplankton (Chisholm et al., 1988). Later on, pigment-based methods were supplemented with or gave way to PCR detection of the 16S rRNA gene with specific primers (Castiglioni et al., 2004). In order to increase the sensitivity of detection, as well as for a better separation of gene clusters, internal transcribed spacers of 16S– 23S-rRNA were also tested (Laloui et al., 2002; Rocap et al., 2002). Screening for green cyanobacteria is also performed using other primers which help amplify genes which code for photosynthetic proteins, e.g., Pcb proteins of *chl a*/*b* antenna distinct from CAB proteins in green algae, euglenae, and higher plants (Geiß et al., 2003).

## **REPRESENTATIVES**

The first member of the genus *Prochlorothrix*, the type species *P. hollandica* (**Figure 1A**) was validly described according to the rules of Bacteriological Code (Burger-Wiersma et al., 1989). In the modern taxonomic system, it is ascribed to Subsection III of Phylum X. *Cyanobacteria* (Oxygenic Photosynthetic Bacteria) or the order "*Oscillatoriales*" (Castenholz, 2001).

The second member of the genus *Prochlorothrix* is *Prochlorothrix* sp. NIVA 8/90 (**Figure 1B**). This strain was named *P. scandica* (Pinevich et al., 1999) legitimized under the rules of Botanical Code (Skulberg, 2008).

*Prochlorothrix* sp. NIVA 8/90 is considered as separate species primarily based on genomic DNA differences with *P. hollandica* (**Table 1**). Noteworthy, the previous demand for 3% dissimilarity of 16S rRNA is now considered to be not obligatory in species delimitation (Wayne et al., 1987). *Prochlorothrix hollandica* and *P. scandica* also differ in 16S–23S-ITS, and have one (1,100 bp) and two (1,100 and 620 bp) 16S–23S-ITS, respectively. The larger ITS in *Prochlorothrix* sp. NIVA 8/90 is 99% identical to the single ITS in *P. hollandica*. At the same time, the two ITS in *Prochlorothrix* sp. NIVA 8/90 demonstrate low mutual similarity (Velichko et al., unpublished). In addition, these two cyanobacteria differ by another genetic criterion: the electrophoretic profile of PCRamplified highly iterated palindromic sequences (HIP-1;Robinson et al., 1995; Smith et al., 1998; Velichko et al., unpublished).

Phenotypic criteria for the species difference between *P. hollandica* and *P. scandica* are cell size, temperature optimum (see below), and fatty acid composition of the membrane lipids (**Table 2**).

Due to the presence of C16:2 and C18:2 fatty acids both species should be categorized as group 5 cyanobacteria (Cohen et al.,

#### **Table 1 | Genetic distinction between** *P. hollandica* **and** *P. scandica.*


**Table 2 | Cellular fatty acid composition (%) of** *P. scandica* **and** *P. hollandica* **grown under identical conditions.**


*\*The position(s) of double-bond(s) is not thoroughly determined; trans-cis isomeric state of unsaturated fatty acids is not determined. The most prominent differences between the two species are in bold.*

1995). In both species, C14:0, C16:0, C14:1, and C16:1 fatty acids are among the major components. The profile of saturated (straightchain and branched) fatty acids is comparable in the two species, except for a double amount of C18:0 stearic acid in *P. scandica*. At the same time, cellular content of unsaturated fatty acids is species-specific. The most prominent differences of *P. scandica* from *P. hollandica* are a twofold increase (8.7% vs 5.0%) in the amount of the C16:1ω5 isomer palmitoleic acid, sevenfold increase (3.4% vs 0.5%) in the amount of C18:1ω9 oleic acid, and a twofold decrease (11.9% vs 21.4%) in the amount of the C16:1ω7 isomer palmitoleic acid. In addition, the trace amounts of dienoic C16:2ω4 and trienoic C16:3ω3 acids are 1.8% vs 1.0% and 1.5% vs 0.7% in *P. scandica* and *P. hollandica*, respectively.

## **DISTRIBUTION**

Evidence for the presence of the genus *Prochlorothrix* has been restricted to eutrophic freshwater or brackish habitats of North Europe (**Figure 2**). *Prochlorothrix hollandica* was accidentally encountered during a routine study of cyanobacterial blooms in the Loosdrecht lake system near Amsterdam (The Netherlands; Burger-Wiersma et al., 1989). These polymictic and highly eutrophic (∼60 mg m−<sup>3</sup> *chl a*) water pools of average depth <sup>∼</sup>2 m originate from a flooded turf pit, with phytoplankton represented mostly by blue–green filamentous cyanobacteria of the group *Limnothrix*/*Pseudanabaena* (Zwart et al., 2005). In addition, "Dutch" green cyanobacteria were also found in a shallow eutrophic Lake Tjeukemeer (van Liere et al., 1989). Finally, according to personal communication by J. van der Does (van Liere et al., 1989), the presence of filamentous green cyanobacteria was registered in other shallow eutrophic Dutch lakes by fluorescence microscopy.

The second representative of the genus *Prochlorothrix*, *P. scandica* was isolated from Lake Mälaren (1,140 km2, medium depth <sup>∼</sup>13 m) near Stockholm, Sweden (Skulberg, 2008).

Based on these discoveries, the representatives of the genus *Prochlorothrix* are common in eutrophic lakes occupied mainly by cyanobacteria (Lewin et al., 1980; van Liere et al., 1989).

In addition to direct evidence for the presence of these cyanobacteria in eutrophic lakes, they have been indirectly detected by the PCR-amplification of 16S rRNA gene and *pcb* genes (Geiß et al., 2003). Natural DNA was obtained from water samples of estuary Darss-Zingst located near the South-West coast of the Baltic Sea between the cities Rostock and Stralsund (Germany). This polymictic and eutrophic water body is represented by a chain of lagoons fed by the mouth of Prerow River, separated from the Baltic Sea by the Peninsula Frischland/Darss, and is characterized by opposite gradients of dissolved organic substances and salinity (Geiß et al., 2004). It is reasonable to expect that dynamic combination of these (a)biotic factors favors the development of cyanobacteria from the genus *Prochlorothrix*.

In summer 2011 *Prochlorothrix*-like 16S rRNA nucleotide sequences were detected in the brackish waters of the Finnish Bay in the environs of St. Petersburg, Russia (Velichko et al., unpublished data). Also, we came across an environmental *Prochlorothrix*-like 16S RNA gene sequence revealed during the metagenomic screening of New Orleans area water systems after Hurricanes Katrina and Rita (Amaral-Zettler et al., 2008).

Phylogenetic reconstruction of all available *Prochlorothrix* 16S rRNA gene sequences demonstrates close clustering of the environmental sequences from different regions of the world within the *Prochlorothrix* group (**Figure 3**). In fact, within the *Prochlorothrix* group sequences of the 16S rRNA gene are 98–99% similar.

In seems that the strains of *Prochlorothrix* detected in brackish coastal waters are merely temporary invaders from the mainland, and they could even sporadically occur in open waters of the Baltic

Sea (Geiß et al., 2003). At the same time, there is no indication of their presence in any oligotrophic lake.

## **FUNCTIONAL ECOLOGY**

Filamentous green cyanobacteria never dominate the phytoplankton of small polymictic lakes. Growth rate of natural populations of *P. hollandica* is usually 30–40% that of *Oscillatoria*-like blue– green cyanobacteria (0.02–0.14 day−1; see Pel et al., 2004). This is possibly the reason why these blue–green cyanobacteria dominate and even suppress green cyanobacteria in natural environment.

### **PRIMARY PRODUCTIVITY**

The biomass of *Prochlorothrix* which reaches its maximum during the bloom season can be used as a principal nutrient by the zooplankton. In particular, the number of *Euchlanis dilatata* (Rotatoria, Nemathelminthes) which is among the main primary suppliers in the Loosdrecht Lakes (Gulati et al., 1993), is increased in late summer and in early spring which coincides with mass development of *Prochlorothrix*. The intensity of *Prochlorothrix* biomass consumption by this invertebrate is commeasurable with that of the blue–green cyanobacterium *Aphanizomenon flos-aquae* (Gulati et al., 1993).

## **PHOTOACCLIMATION**

Genes *pcbABC* in *P. hollandica* constitute an operon (Nikolaitchik and Bullerjahn, 1998; van der Staay et al., 1998). The antenna proteins pcbAB selectively associate with PSII, while the antenna protein PcbC is attached to PSI (Bumba et al., 2005). Photosystems I and II in *P. hollandica* have the peripheral antennae which produce a 18-meric ring around the PSI trimer and a 14-meric ring around the PSII dimer (Boichenko et al., 2007). *Prochlorothrix hollandica* is the only green cyanobacterium in which a shift of photosynthetic States 1↔2 (i.e.,preferential light channeling) was demonstrated. However, the details of the structural/functional association between the *chl a/b* antenna with photosystems remain obscure. According to one of the two conflicting models, in State 1 the main *chl a/b* antenna (which consists of proteins PcbB and PcbC) transiently associates both PS. In State 2, the minor *chl a/b* antenna (which consists of PcbA protein) dissociates from the super complex of the main *chl a/b* antenna and PSII (Post et al., 1992). In another model (van der Staay and Staehelin, 1994), in

State 1 the stacked thylakoids are enriched with the super complex of *chl a/b* antenna and PSII, whereas unstacked thylakoids contain PSI.

Later experimental data suggest a compromise model (Partensky and Garczarek, 2003) based on the spatial redistribution of (de)phosphorylated Pcb proteins.

## **TEMPERATURE**

The culture optimum growth of *P. hollandica* PCC 9006 and *P. scandica* NIVA-8/90 is 20 and 24˚C respectively (Pinevich et al., 1999; Skulberg, 2008). However, in nature they were shown to grow at lower temperatures. Compared to other green cyanobacteria, *Prochlorothrix* occupies a narrow geographic range characterized by significant seasonal changes in temperature, as well as by the episodic formation of ice cover in winter time. For instance in July 1984 when *P. hollandica* PCC 9006 was discovered in Lake Loosdrecht the water temperature was 18˚C (Burger-Wiersma et al., 1989), whereas in April 1998 it was only 8˚C (Zwart et al., 2005). As a rule, mass development of *Prochlorothrix* in this lake occurs in the end of summer when water temperature is 15–18˚C (Post and Bullerjahn, 1994).

Water sampling from Lake Mälaren, where *P. scandica* NIVA-8/90 was detected was performed in October 1990. During this period, the temperature is 10–13˚C throughout the water column due to the active mixing (Skulberg, 2008).

#### **SALINITY RESPONSE**

Cyanobacteria of the genus *Prochlorothrix* are traditionally considered to inhabit freshwater pools, and they are sensitive even to a small increase in salinity. *Prochlorothrix hollandica* PCC 9006 demonstrated poor growth in the presence of 25 mM NaCl, and the growth was fully suppressed at 100 mM or after the addition of sea water equivalent (Burger-Wiersma et al., 1989). Our data on the comparative salt resistance in *P. hollandica* PCC 9006 and *P. scandica* NIVA-8/90 confirm the inhibitory effect of 50–100 mM NaCl (Pinevich et al., unpublished data). Increased salinity provoked morphological changes in both species: cells were elongated, and the filaments became zigzag. At >200 mM NaCl, the filaments of both species degraded after a week growth.

Although high sensitivity to salt can be easily explained by the inability to synthesize osmotic protectors, salt adaptation may depend on the local ion climate. In particular, *P. hollandica* was resistant to 170 mM NaCl when grown in a semi-synthetic medium supplied with water from the estuary Darss-Zingst. Under these conditions, an increase in the cellular content of sucrose was registered (Bergmann et al., 2008). This data are in agreement with the detection of *Prochlorothrix*-like marker genes in the phytoplankton of the Baltic Sea open waters (Geiß et al., 2003).

## **INVOLVEMENT IN SYMBIOSES**

In contrast to the rest of green cyanobacteria, the strains of *Prochlorothrix* are easily cultured on liquid mineral media (see below). However, most attempts to obtain pure cultures on solid medium proved unsuccessful. Cultures ceased to grow, when attempts to get rid of heterotrophic contaminants were made by the authors. FISH hybridization with taxon-specific probes EUB338II, EUB338III, ALF968, BET42a, GAM42a, and DELTA495a revealed a vast diversity of heterotrophic bacteria associated with *P. hollandica* PCC 9006 and *P. scandica* NIVA-8/90. In case of *P. hollandica* PCC 9006, a positive signal was obtained with probes specific to alpha-, beta-, and gammaproteobacteria. On the other hand, heterotrophic epiphytes of *P. scandica* NIVA-8/90 belong to alpha- and betaproteobacteria (Velichko et al., unpublished data).

These data, along with the fact that bacteria-free filaments of *P. hollandica* PCC 9006 demonstrated negligible growth in liquid culture (Schyns et al., 1997), indicate a mutually favorable consortium-type association between the filamentous green cyanobacteria and their heterotrophic satellites.

## **CULTIVATION**

#### **MEDIA AND STRAINS**

Cyanobacteria of the genus *Prochlorothrix* are grown in mineral media FPG, BG-11, and Z8 which contain relatively low amounts of inorganic salts and are routinely used for the cultivation of freshwater blue–green cyanobacteria (**Table 3**). Double-distilled water is necessary. Otherwise, unidentified contaminating chemicals inhibit growth. For slant cultivation, semisolid agar (≤0.5%) is recommended, since ≥1% inhibits growth. In addition to doubledistilled water and mineral salts of high quality, purified agar (e.g., Difco) is needed. Cell growth is suppressed by elevated concentrations of NaCl, and the degree of salt resistance depends on ion content of the medium (Burger-Wiersma et al., 1989; Bergmann et al., 2008). Since *Prochlorothrix* sp. is unable to fix nitrogen even in microaerobic conditions, an addition of combined nitrogen is necessary.

In collections *Prochlorothrix* sp. is represented by several cultured strains. One of them, isolated in the Netherlands and validly described as *P. hollandica* (Burger-Wiersma et al., 1989), is maintained under different strain identifiers (CALU 1027, CCAP 1490/1, NIVA-5/89, PCC 9006, and SAG 10.89). Several multixenic strains isolated in Sweden are stored in the Norwegian Institute for Water Research (Oslo, Norway) collection, namely NIVA 8/90, 9/90, and 16/90 (Skulberg, 2008). Strain *P. hollandica* PCC 9006 (CALU 1027) and strain *P. scandica* NIVA-8/90 (CALU 1205) are stored at the collection of St. Petersburg University (St. Petersburg, Russia). Various methods of axenization were attempted including streaking, repeated washing, and micromanipulation, as well as

**Table 3 | Nutrition media for the cultivation of** *Prochlorothrix***.**


*\*Extracted from Burger-Wiersma et al. (1989); \*\*Rippka et al. (1979); \*\*\*Kotai (1972). Double-distilled water is generally used.*

differential sensitivity to antibiotics (Burger-Wiersma et al., 1989; Pinevich et al., 1999; Skulberg, 2008). All of them proved unsuccessful because of the loss of viability in both liquid and solid medium. In the latter case, filaments grow not as colonies, but rather as films surrounded by dim layer of colorless heterotrophic bacteria.

## **GROWTH PARAMETERS**

Batch cultures of *P. hollandica* PCC 9006 in FPG medium at 40μE m−<sup>2</sup> s <sup>−</sup><sup>1</sup> and 20˚C grew at a maximum rate of 0.02 h−<sup>1</sup> (Burger-Wiersma et al., 1989). When the BG-11 medium was used, grow rate varied from 0.007 h−<sup>1</sup> (Burger-Wiersma et al., 1989) to 0.026 h−<sup>1</sup> (Bergmann et al., 2008). Optimum growth temperature for *P. hollandica* PCC 9006 is 20˚C (Burger-Wiersma et al., 1989; Pinevich et al., 1999), while that for *P. scandica* is 24˚C (Pinevich et al., 1999). Stirring or air bubbling failed to stimulate growth and even resulted in decreased viability due to mechanical damage of the filaments.

## **MORPHOLOGY**

Immotile, non-branching filaments of *Prochlorothrix* sp. grow by equal binary fission of intercalary cells and multiply by fortuitous fragmentation. Hormogonia or other types of differentiated structures are never observed.

Filaments of *P. hollandica* PCC 9006 and *P. scandica* NIVA-8/90 (**Figure 1**) are ≤250μm long. Mean cell size of *P. hollandica* PCC 9006 is 11.8 ± 0.9 × 1.6 ± 0.1μm (length vs width, 7.4:1), while that of *P. scandica* NIVA-8/90 being 7.4 ± 0.7 × 2.1 ± 0.1μm (3.5:1). In both cases, the width of cell-to-cell junction is 1/8–1/5 of filament diameter. In this respect, the filaments of *Prochlorothrix* more closely resemble those of *Limnothrix* than those of *Pseudanabaena*. In the latter case, the width of cell-to-cell junction is 1/2 of filament diameter (Castenholz, 2001).

## **SELECTED FEATURES OF ULTRASTRUCTURE**

## **THYLAKOID STACKING**

In addition to unstacked thylakoids which are typical of blue– green cyanobacteria, *P. hollandica* PCC 9006 contains stacked thylakoids vaguely resembling grana thylakoids in chloroplasts (Giddings et al., 1980; Schuster et al., 1984; Miller et al., 1988; Golecki and Jürgens, 1989). In both cases, membrane adhesion is maintained by the PSII proteins and *chl a*/*b* antenna proteins (van der Staay and Staehelin, 1994).

Thylakoids in freeze-fractured cells of *P. hollandica* PCC 9006 were shown to be laterally heterogeneous (van der Staay and Staehelin, 1994). Similarly to chloroplasts, stacked regions are enriched with globular particles which correspond to the super complexes of PSII with *chl a*/*b* antenna. By contrast, PSI and ATP synthase are preferentially localized in the unstacked regions (Schuster et al., 1984; van der Staay and Staehelin, 1994).

Membranes of stacked thylakoids in *P. scandica* NIVA-8/90 are covered with particles (viridosomes) which are 10.0 ± 2.3 nm diameter and are located with the periodicity of 20.0 ± 5.6 nm. These are suggested to correspond to the epimembrane domain of *chl a*/*b* antenna (Pinevich et al., 1999). The analogous particles (xanthosomes) are present in both the chromophyte algae (e.g., *Bacillariophyceae* and *Phaeophyceae*) and *Dinozoa* (Katoh et al., 1993; Green and Durnford, 1996).

Thylakoid stacking in *Prochlorothrix* may help redistribute excitation energy between the adjoining thylakoids (Dekker and Boekema, 2005). Alternatively, it may be a functionless result of segregation of PSII centers in thylakoid membranes (Bullerjahn and Post, 1993; see Functional Ecology and Photoacclimation).

## **THYLAKOID CENTERS**

An intriguing detail of ultrastructure in *P. hollandica* PCC 9006 and *P. scandica* NIVA-8/90, also rarely observed in blue–green cyanobacteria, is the thylakoid center. The latter corresponds to a <sup>∼</sup>200-nm electron dense zone between cell membrane and two to four radiating thylakoids (Pinevich et al., 1996). These metastable structures possibly channel the molecules of *chl-ide a*GG (the precursor of *chl* a) from the sites of synthesis in cell membrane to the functional sites in thylakoids (Nierzwicki-Bauer et al., 1983; Bryant, 1986; Hinterstoisser et al., 1993).

## **GAS VESICLES**

Similarly to many planktonic filamentous blue–green cyanobacteria, *P. hollandica* PCC 9006 contains honeycomb-like aggregates of gas vesicles. An individual vesicle is a cylinder of 50 nm diameter with conical ends. Its 125–265 nm length mirrors different stages of self-assembly (Golecki and Jürgens, 1989; Pinevich et al., 1999).

## **CARBOXYSOMES**

Similarly to blue–green cyanobacteria, the carboxysomes of *P. hollandica* PCC 9006 are enveloped in a non-unit membrane. However, the carboxysomes of *P. scandica* NIVA-8/90 lack this shell resembling the pyrenoid of green algae and higher plants (McKay and Gibbs, 1990; Morden and Golden, 1991; Pinevich et al., 1996). According to the immune electron microscopy data, all RuBisCO of *P. hollandica* PCC 9006 is associated with carboxysomes. Whereas, the RuBisCO in *P. scandica* NIVA-8/90 exists in both the immobilized and soluble form (Pinevich et al., 1996). Carboxysomes in *Prochlorothrix* were shown to contact thylakoids (Pinevich et al., 1999), which is not the case in blue–green cyanobacteria. This association may promote phosphorylation of ribulose-5-phosphate, as well as the activity of carbonic anhydrase (Pinevich et al., 1999).

## **GENETICS**

Genome size of *Prochlorothrix* is <sup>∼</sup>5.5 Mb (Post and Bullerjahn, 1994; Schyns et al., 1997) which is within the range of most cyanobacteria. The G + C Mol% of genomic DNA is 51–53 (Herdman,1981;Coleman and Lewin,1983;Burger-Wiersma et al.,1989; Dufresne et al., 2003).

## **GENOME SEQUENCING**

Genome sequencing of *P. hollandica* PCC 9006 is currently performed by the authors. Automatic annotation using The RAST Server (Rapid Annotations using Subsystems Technology; Aziz et al., 2008) was performed on the draft sequence, and the manual curation of automatically annotated genes is being done. We are focusing primarily on photosynthetic genes. Phylogenetic affiliation for two of them, namely *psbO* and *psaF*, are given below (see Phylogeny).

## **ENERGY ASSIMILATION**

In contrast to unicellular green cyanobacteria, *Prochlorothrix* has not been tested for dark chemotrophy; the only studied mode of energy assimilation in this case is phototrophy.

## **PHOTOSYNTHETIC PIGMENTS**

*Prochlorothrix hollandica* PCC 9006 contains accessory *chl b* (Burger-Wiersma et al., 1989). It is suggested to be synthesized from *chl b*<sup>2</sup> in one stage using 4-vinyl-reductase (Helfrich et al., 1999). The alternative way involves *chl a* oxygenase which was the enzyme also found in *Prochloron didemni* and higher plants (Tanaka et al., 1998; Tomitani et al., 1999; Nagata et al., 2005).

Among the carotenoids in *P. hollandica* PCC 9006 are βcarotene and zeaxanthin (van der Staay et al., 1992). The content of β-carotene in *P. hollandica* PCC 9006 is four times that of zeaxanthin. The former preferentially associates with PSI, while the latter complexes with 56/58 kDa proteins integrated in the cell membrane (Omata et al., 1985; Engle et al., 1991).

## **LIGHT-HARVESTING ANTENNA**

In contrast to blue–green cyanobacteria, *P. hollandica* PCC 9006 and *P. scandica* NIVA 8/90 possess the light-harvesting antenna containing both *chl a* and *chl b* (Bryant, 1986; Burger-Wiersma and Post, 1989; Hess et al., 1996; Marquardt et al., 1998). Although it was originally thought that Pcb proteins of this antenna share the origin with CAB proteins in chloroplasts, in fact they have no similarity (La Roche et al., 1996; van der Staay et al., 1998). In particular, Pcb proteins of *P. hollandica* PCC 9006 do not react with antibodies raised against CAB proteins of green algae and higher plants (Hiller and Larkum, 1985; Bullerjahn et al., 1990; van der Staay and Staehelin, 1994). At the same time, Pcb proteins of *P. hollandica* PCC 9006 are similar to PsbB and PsbC (CP43) proteins of PSII core antenna. In addition, they share similarity with *IsiA* (CP43 ) protein of PSI stress antenna which is formed in blue–green cyanobacteria under iron starvation (Bullerjahn et al., 1987, 1990; van der Staay and Staehelin, 1994; Nikolaitchik and Bullerjahn, 1998; van der Staay et al., 1998; Partensky et al., 1999). Hexadomain proteins Pcb/Cp43/*IsiA* vary in the length of the loop exposed at the thylakoid exoplasmic surface between helices V and VI (Herbstová et al., 2010).

Genes *pcbABC* of *P. hollandica* PCC 9006 form an operon (Nikolaitchik and Bullerjahn, 1998; van der Staay et al., 1998). The genes *pcbA* and *pcbB* genes belong to the evolutionary lineage of *IsiA* and code for polypeptides of 32 and 33 kDa. The third gene, *pcbC*, belongs to a separate lineage and codes for a polypeptide of 38 kDa (van der Staay and Staehelin, 1994; van der Staay et al., 1998; Garczarek et al., 2000).

Pcb proteins aggregate in multimeric super complexes (Bibby et al., 2001, 2003; Chen et al., 2002, 2005; Partensky and Garczarek, 2003) which surround the oligomers of PSI and PSII (Bibby et al., 2003; Boichenko et al., 2007). PcbA/PcbB proteins build up the mobile antenna of both photosystems, while the PcbC proteins associate with a portion of antenna tightly bound to PSI (Herbstová et al., 2010).

## **PHOTOSYSTEMS**

PSI polypeptides of *P. hollandica* PCC 9006 are the same as in blue–green cyanobacteria (Schuster et al., 1984; Bullerjahn et al., 1987; van der Staay et al., 1998; Garczarek et al., 2000). They include major PsaA/PsaB proteins (66 kDa) and 6 minor proteins PsaC–F, I–M (20 kDa; Burger-Wiersma and Post, 1989; van der Staay et al., 1993). PSII proteins of *P. hollandica* PCC 9006 include CP43/CP47, D1/D2, PsbH, and PsbO, as well as low molecular weight polypeptides PsbJ and PsbK (Greer, 1991; Greer and Golden, 1992; Mor et al., 1993; Post and Bullerjahn, 1994). Interestingly, *psbA* gene has a 3 -terminal 21-nucleotide deletion; the same deletion is present in the chloroplast ortholog (Morden and Golden, 1991).

The oxygen evolving complex (OEC) in *P. hollandica* PCC 9006 lacks proteins PsbP (17 kDa) and PsbQ (23 kDa) typical of chloroplasts (Mor et al., 1993). The interaction of OEC with Cl− and Ca2+ is provided by the hydrophobic Mn-stabilizing protein PsbO (37 kDa) different from the more amphiphilic analog in blue–green cyanobacteria (26.5 kDa; Mor et al., 1993; Barber and Kühlbrandt, 1999).

## **FACULTATIVE ANOXYGENIC PHOTOSYNTHESIS**

Among green cyanobacteria, P. *hollandica* PCC 9006 is uniquely capable of photosynthesis with sulfide as electron donor and, respectively, without dioxygen production (Post and Arieli, 1997). OEC of this cyanobacterium retains activity at ≤1 mM sulfide. High resistance to sulfide stress is provided by the PsbO protein which protects the catalytic site of OEC (Mor et al., 1993; Post and Arieli, 1997). The transition to anoxygenic photosynthesis occurs at >1 mM sulfide when this substance inhibits both the biogenesis and activity of PSII, especially that of OEC.

A relative resistance to sulfide allows *P. hollandica* dominate the phytoplankton of water bodies with sulfide accumulating in warm seasons (Post and Arieli, 1997). Given the trophic state of shallow waters harboring *Prochlorothrix*, sulfide production in the anoxic sediments may mix with surface waters. Thus, sulfide resistance may be an essential adaptation in this habitat.

## **SUBSTRATE ASSIMILATION AND ENZYMOLOGY CARBON ASSIMILATION**

Data on the ability to use organic substrates (photomixotrophy or photoheterotrophy, in the light; chemoheterotrophy, in darkness) in *Prochlorothrix* is absent. Similarly to most blue–green cyanobacteria, *Prochlorothrix* demonstrates photoautoptrophy based on Calvin cycle with RuBisCO as key enzyme. *Prochlorothrix hollandica* PCC 9006 contains IB-subform of RuBisCO which clusters together with RuBisCO of blue–green β-cyanobacteria from the so-called "green" phylogenetic branch of the enzyme (Morden and Golden, 1991; Shimada et al., 1995; Chen et al., 2004; Badger et al., 2006).

RuBisCO of *P. hollandica* PCC 9006 is encoded by the dicistronic operon *rbcLS* which consists of genes for the large (L, 57 kDa) and small (S, 13 kDa) subunits, respectively. The holoenzyme (560 kDa) belongs to the L8S8-type (Hawthornthwaite et al., 1990; Morden and Golden, 1991). Enzymatic activity is maximum at 30˚C and pH 8; Mg2<sup>+</sup> is used as a cofactor (Hawthornthwaite et al., 1990).

## **NITROGEN ASSIMILATION**

There are no data on the ability of *P. hollandica* PCC 9006 for aerobic nitrogen fixation. Initial attempts to induce nitrogenase under the conditions of nitrogen starvation proved unsuccessful (Burger-Wiersma et al., 1989). Furthermore, there are no *nif* genes sequences in the partially reconstructed genome of this cyanobacterium (Pinevich et al., unpublished data).

At the same time, *P. hollandica* PCC 9006 assimilates both ammonium and nitrate. Maximum growth rate in culture conditions is observed at 0.75 mM NH4 +, whereas growth is inhibited at 3 mM. Growth is also absent if 1.5 mM urea substitutes for nitrate, indicating lack of the urease enzyme (Burger-Wiersma et al., 1989).

#### **PHOSPHORUS ASSIMILATION**

According to field experiments, the most extensive growth of *P. hollandica* PCC 9006 takes place in those parts of water bodies where phytoplankton is limited by phosphorus deficiency (Burger-Wiersma et al., 1989). *Prochlorothrix hollandica* PCC 9006 is characterized by higher maximum rate of phosphorus uptake and by lower saturation constant for this substrate as compared to other filamentous cyanobacteria, in particular *Planktothrix* sp. (Ducobu et al., 2002; Dignum et al., 2005).

## *Special enzymology: RNAse P*

This enzyme performs the maturation of 5 -end of tRNA and is ubiquitous in prokaryotes including plastids and mitochondria. *Prochlorothrix hollandica* PCC 9006 contains a rare type of this RNAse characterized by long domain P15/16, as well as by a GGU motif which binds 3 -CCA of pre-tRNA and is absent from the ribozymes of other cyanobacteria (Fingerhut and Schön, 1998).

## **PHYLOGENY**

Tracing phylogeny of green cyanobacteria raises the questions regarding: (i) relations within the green cyanobacteria group; (ii) relations between green and blue–green cyanobacteria; (iii) relations between green cyanobacteria and simple plastids; (iv) the ancestor of green cyanobacteria; and (v) the origin of chlorophyll light-harvesting antenna.

Here, two phylogenies are involved. One of them is the general evolution of cyanobacteria reconstructed by the comparative analysis of the major informational (in particular, 16S rNNA gene and *rpoS*) and photosynthetic genes (in particular, *psbA* and *petE*). Secondly, it is the evolution of green cyanobacteria traced by the analysis of the *pcb* genes coding for the chlorophyll light-harvesting antenna.

## **16S rRNA GENE**

According to this evolutionary criterion, green cyanobacteria are polyphyletic (Urbach et al., 1992). Paradoxically,*P. hollandica* PCC 9006 clusters close to the unicellular strain *Synechococcus* PCC 6301 (Pinevich et al., 2010). Detailed dendrograms showing the distribution of green cyanobacteria between separate clades of cyanobacterial tree, as well as mutual subordination of strains from the same genus, are presented elsewhere (Turner, 1997; Moore et al., 1998; Partensky et al., 1999; Miyashita et al., 2003; Partensky and Garczarek, 2003; Rocap et al., 2003; de los Rios et al., 2007; Münchhoff et al., 2007; Scanlan et al., 2009).

## **RpoC GENE**

Another informational gene which helps reconstruct the evolutionary history of cyanobacteria is *rpoC1* coding for a subunit of the DNA dependent RNA polymerase. Here, again, green cyanobacteria were shown to be polyphyletic, and *P. hollandica* PCC 9006 groups with *Synechococcus* PCC 6301 (Palenik and Haselkorn, 1992).

The phylogenies of the major photosynthesis genes [*psaF*, coding for the conservative PsaF protein, a 21-kDa three-domain component of the reaction center complex; *psbA,* coding for the 32 kDa core PSII protein; *psbO*, coding for the Mn-stabilizing protein PsbO; and *petE,* coding for the copper-containing soluble protein plastocyanin (∼11 kDa), which donates electrons to PSI] repeat that of the 16S rRNA gene with green cyanobacteria being polyphyletic and *P. hollandica* PCC 9006 clustering with *Synechococcus* PCC 6301 (**Figures 4** and **5**; Pinevich et al., 2010).

#### **Pcb GENES**

The *pcbABC* genes of *P. hollandica* PCC 9006 which code for the antenna proteins of family Pcb/*IsiA*/CP43 cluster together with their counterparts in unicellular green cyanobacteria from the genera *Acaryochloris* and *Prochloron* (Pinevich et al., 2010). The highly conserved proteins of Pcb/*IsiA*/CP43 family are thought to be inherited from a common ancestor of cyanobacteria. They might evolve according to two scenarios.

According to the first scenario, these proteins may have appeared in *Synechococcus*-like ancestor of *P. marinus* which inhabited oceanic systems poor in iron (Ting et al., 2002). In these stress conditions, they constitutively expressed both the *IsiA* (CP43 ) protein and the phycobilisome. Among the descendants of this hypothetical cyanobacterium were phycoerythrinrich strains of *Synechococcus*, as well as phycobiliprotein free or phycobiliprotein-poor (phycobilisome free) strains of *P. marinus*.

By the second scenario, *pcb* and *isiA* genes evolved independently by duplication and subsequent specialization of the copied *psbC*-like gene in a protocyanobacterium (Chen et al., 2005). This scenario is more plausible since it explains the existence of a set of *chl*-binding proteins in blue–green and green cyanobacteria.

Both scenarios assume a transient coexistence of the *chl*-based antenna with the phycobiliprotein-based antenna (Shimada et al., 2003).

## **GENES CODING FOR ENZYMES OF THE CHROMOPHORE BIOSYNTHESIS**

The evolution of these genes is distinct from that of the *pcb* genes. In fact, presence of the conservative His residues in Pcb proteins is only a prerequisite of *chl* binding. In other words, the formation of *chl*-containing antennae, in particular the *chl a*/*b* antenna, needs biosynthetic enzymes for the respective chromatophores. Since the ancestor of Pcb proteins is thought to be *chl a*-binding protein (see above), the incorporation of *chl b*, *a*2, *b*2, and *d* in antennae of different green cyanobacteria suggests the formation of these pigments as byproducts in the metabolic pathway of *chl a*. In fact, monovinyl *chls* are synthesized from divinyl precursors by reductases, whereas *chl b* and *chl d* are formed by oxygenases (Beale, 2008).

The *cao* genes coding for *chl a* oxygenase in *P. hollandica* PCC 9006 and *P. didemni* are highly similar (Tomitani et al., 1999). Correspondingly, these green cyanobacteria did not acquire this gene independently, but rather the ability to synthesize *chl b* was a trait of common ancestor of cyanobactetria.

The evolutionary loss of phycobilins resulted in the appearance of green cyanobacteria like *P. hollandica* PCC 9006 and *P. didemni*, while the loss of *chl b* resulted in the lineage of blue– green cyanobacteria, which allows to consider both these green cyanobacteria as a model for the protocyanobacterium (Tomitani et al., 1999).

## **RELATION OF GREEN CYANOBACTERIA TO THE ORIGIN OF THE SIMPLE CHLOROPLAST**

Although cyanobacterial origin of the simple chloroplast is beyond doubt (Morden et al., 1992; McFadden, 1996), the specific nature of ancestry is debatable.

The *Prochloron*-type green cyanobacteria were rejected as a candidate, since Pcb proteins are dissimilar with the members of CAB family (La Roche et al., 1996; van der Staay et al., 1998). This conclusion prematurely deprived green cyanobacteria of any role in the origin of the simple chloroplast.

However, it should be noted that Pcb proteins cluster with PsbC/*IsiA*(CP43 ) proteins (Bullerjahn et al., 1987, 1990; van der Staay and Staehelin, 1994; Nikolaitchik and Bullerjahn, 1998; van der Staay et al., 1998; Partensky et al., 1999). This finding, as well as data on the relationships between green cyanobacteria and their blue–green relatives (see above) help restore the role of the former as both the ancestor of the chloroplast, as well as the ancestor of oxygenic phototrophic bacteria.

#### **CHLOROPLAST ORIGINS**

The ancestor of chloroplasts is hypothesized to possess both the *pcb* genes, coding for *chl*-binding antenna proteins, and the *pbp* genes coding for phycobilisome (Tomitani et al., 1999). This cyanobacterium, engulfed by a non-photosynthetic protist, became a precursor of three lineages of the simple chloroplasts (**Figure 6**).

One lineage is represented by the chloroplast of *Chlorophyceae*/*Charophyceae*/*Plantae* which lost the ancestral *pcb* and *pbp* genes. The loss of "old" *chl a*/*b*-binding Pcb proteins was compensated by the acquisition of convergent "new" *chl a*/*b*-binding CAB proteins.

Two other lineages are represented by the chloroplasts of *Glaucocystophyceae* and *Rhodophyceae* which lost the ancestral *pcb* genes but retained the ancestral *pbp* genes.

#### **CYANOBACTERIAL ORIGINS**

Cyanobacteria are hypothesized to share the ancestor with chloroplasts, i.e., the primordial cyanobacterium had both the *pcb* genes, coding for *chl*-binding antenna proteins, and the *pbp* genes coding for phycobilisome (Tomitani et al., 1999). This cyanobacterium became a precursor of two lineages of cyanobacteria (**Figure 6**).

One lineage, resulting from the loss of ancestral *pcb* genes is represented by blue–green cyanobacteria. The other lineage originating from the complete or partial loss of ancestral *pbp* genes is represented by green cyanobacteria. Noteworthy, the latter were not chloroplast ancestors.

#### **REFERENCES**


A., Formsma, K., Gerdes, S., Glass, E. M., Kubal, M., Meyer, F., Olsen, G. J., Olson, R., Osterman, A. L., Overbeek, R. A., McNeil, L. K., Paarmann, D., Paczian, T., Parrello, B., Pusch, G. D., Reich, C., Stevens, R., Vassieva, O., Vonstein, V., Wilke, A., and Zagnitko, O. (2008). The RAST Server: rapid annotations using subsystems technology. *BMC Genomics* 9, 75.

Interestingly, the genes responsible for *chl b* synthesis were divergently inherited by green cyanobacteria and chloroplasts.

#### **GENERAL CONCLUSION**

The primordial cyanobacterium which possessed both the *pcb* and *pbp* genes is by definition a green cyanobacterium Therefore, the ancient green bacteria should be reconsidered as key players of the evolutionary process which resulted in the biodiversity of cyanobacteria and plastids.

## **CONCLUDING REMARKS**

Cyanobacteria of the genus *Prochlorothrix*, in contrast to the other cyanobacteria, are both very rare and largely incapable of axenic growth. They combine trivial features (primarily, oxygenic photosynthesis) with specific, yet obscure aspects of homeostasis and competitive potential. The analysis of their environmental and symbiotic ecology can help better understand this uniqueness. Thereby, the complete annotation of *P. hollandica* PCC 9006 genome is important, as well as search and isolation of new strains of the genus.

## **ACKNOWLEDGMENTS**

We are grateful to Dr. Svetlana Averina and Dr. Olga Gavrilova for fruitful collaboration. Our thanks are due toYaroslav Borisov who participated in the annotation of *P. hollandica* PCC 9006 genes.


hollandica strain SAG 10.89. *Microbiol. Ecol.* 55, 685–696.


Balloni, and Materassi 1986, with emendation of the ordinal description. *Int. J. Syst. Bacteriol.* 39, 250–257.


structure of stacked and unstacked regions of the photosynthetic membranes of Prochloron sp., a prokaryote. *Proc. Natl. Acad. Sci. U.S.A.* 77, 352–356.


Prochlorophyta). *Biochim. Biophys. Acta* 806, 107–115.


hollandica. *FEMS Microbiol. Lett.* 68, 187–194.


chlorophyll a/b antenna apoprotein in the photosynthetic prokaryote Prochlorothrix hollandica. *Biochim. Biophys. Acta* 1110, 75–82.


sequence of ribulose-1,5 bisphosphate carboxylase/oxygenase and the physiology of Prochloron and Prochlorococcus (Prochlorales). *J. Mol. Evol.* 40, 671–677.


F., Hoogveld, H. L., and Gons, H. J. (2005). Molecular characterization of cyanobacterial diversity in a shallow eutrophic lake. *Environ. Microbiol.* 7, 365–377.

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

*Received: 30 January 2012; accepted: 19 April 2012; published online: 21 May 2012.*

*Citation: Pinevich A, Velichko N and Ivanikova N (2012) Cyanobacteria of the genus Prochlorothrix. Front. Microbio. 3:173. doi: 10.3389/fmicb.2012.00173*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Pinevich, Velichko and Ivanikova. This is an open-access article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits non-commercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*

## The languages spoken in the water body (or the biological role of cyanobacterial toxins)

## *Aaron Kaplan1\*, Moshe Harel 1, Ruth N. Kaplan-Levy 2, Ora Hadas 2, Assaf Sukenik <sup>2</sup> and Elke Dittmann3*

*<sup>1</sup> Department of Plant and Environmental Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel*

*<sup>2</sup> The Yigal Allon Kinneret Limnological Laboratory, Israel Oceanographic and Limnological Research, Migdal, Israel*

*<sup>3</sup> Institut für Biochemie und Biologie, Universität Potsdam, Potsdam, Germany*

#### *Edited by:*

*George S. Bullerjahn, Bowling Green State University, USA*

#### *Reviewed by:*

*Michael R. Twiss, Clarkson University, USA Greg Boyer, State University of New York, USA*

#### *\*Correspondence:*

*Aaron Kaplan, Department of Plant and Environmental Sciences, The Hebrew University of Jerusalem, Edmond J. Safra Campus, Givat Ram, Jerusalem 91904, Israel. e-mail: aaron.kaplan@mail.huji.ac.il*

## **INTRODUCTION**

"What determines who is there?" is a long standing but nevertheless intriguing biological/environmental question. Numerous field and laboratory studies, and modeling of the data thereof, were performed in an attempt to reveal the effects of biotic and abiotic parameters on the temporal and spatial dynamics of phytoplankton populations (Ogawa, 1988; Walsby, 1997; Abraham, 1998). There is a growing body of evidence that secondary metabolites play an important part in the shaping of the community composition through inter- and intra-species communication but, for the most part, their biological role, mode of action, and regulation of their production are poorly understood. One example is the evolutionary drivers for the richness and large diversity of secondary metabolites produced by various cyanobacteria, a phenomenon that we know little about. It has been proposed that at least part of the richness observed in the case of a *Lyngbya* sp. is due to improper systematic analysis and that several strains are in fact involved (Engene et al., 2011) but, if any, this argument strengthens the message that a large diversity of secondary metabolites is formed in this genus (Liu and Rein, 2010). A comprehensive chemotyping, i.e., analysis of the array of secondary metabolites within a given organism (Sivonen, 2008), combined with the application of genetic approaches using various genomic tools are required and to some extent emerging (Mejean et al., 2010; Jones et al., 2011). In the long run it will provide us with a complete array of metabolites produced and enable recognition of the genetic potential and its expression particularly under changing environmental conditions.

Cyanobacteria are known to produce a wide variety of secondary metabolites including non-ribosomally made peptides, polyketides, and alkaloids. Naturally most of the research on cyanobacterial secondary metabolites focused on those that were

Although intensification of toxic cyanobacterial blooms over the last decade is a matter of growing concern due to bloom impact on water quality, the biological role of most of the toxins produced is not known. In this critical review we focus primarily on the biological role of two toxins, microcystins and cylindrospermopsin, in inter- and intra-species communication and in nutrient acquisition. We examine the experimental evidence supporting some of the dogmas in the field and raise several open questions to be dealt with in future research. We do not discuss the health and environmental implications of toxin presence in the water body.

**Keywords: aoa, cylindrospermopsin, microcystin, cyr, mcy**

classified as toxic to eukaryotes, mainly microcystin, nodularin, saxitoxin, and cylindrospermopsin (Pearson et al., 2010), since their presence may constitute a serious threat to water quality in many freshwater bodies worldwide (Carmichael, 1997; Griffiths and Saker, 2003; Zegura et al., 2011). There is a growing body of evidence that regional and global climatic change may benefit various species of harmful cyanobacteria by increasing their growth rates, dominance, persistence, geographic distributions, and activity (see Paerl and Huisman, 2009; Sukenik et al., this issue).

Numerous publications have described the implications of cyanobacterial toxin production on environmental and health aspects but this is beyond the scope of the present paper. Here we discuss the nature and biological role of two cyanobacterial toxins, microcystin and cylindrospermopsin, for which experimentally based information has recently been obtained.

## **MICROCYSTIN**

## **OCCURRENCE AND TOXICITY**

Microcystins are the most frequently reported cyanobacterial toxins in freshwater environments and are regarded as a major health hazard and challenge to the quality of drinking water (Dittmann and Wiegand, 2006). Microcystins are predominantly produced by freshwater strains including *Microcystis*, *Planktothrix*, and *Anabaena* (Welker and von Dohren, 2006) but were also incidentally reported from terrestrial environments. Toxic and non-toxic strains usually co-exist in natural bloom samples due to the sporadic distribution of toxin biosynthesis genes in the producing genera (Via-Ordorika et al., 2004). Microcystin penetrates the liver where it inhibits eukaryotic protein phosphatases type 1 and 2A, eventually leading to liver hemorrhage and death (Runnegar et al., 1995). The hepatotoxin was implicated in a number of animal and human intoxications, the

most fatal case being reported from a hemodialysis center in Brazil where patients died with symptoms of hepatotoxicity after exposure to water from a nearby reservoir contaminated with cyanobacteria (Jochimsen et al., 1998). The cyclic heptapeptide displays the general structure *cyclo*(d-Ala-X-d-MeAsp-Z-Addad-Glu-MdhA) in which X and Z represent variable l amino acids. d-MeAsp is d-erythro-β-methyl-aspartic acid, MdhA is *N*-methyldehydroalanine and Adda is (2*S*,3*S*,8*S*,9*S*)-3-amino-9 methoxy-2,6,8-trimethyl-10-phenyldeca-4,6 dienoic acid (Botes et al., 1984). Many variants have been described in the scientific literature, mostly differing in amino acid positions X and Z (Sivonen and Jones, 1999; Welker and von Dohren, 2006). The related nodularins are exclusively produced by *Nodularia* strains in brackish water environments and possess the general structure *cyclo*(d-MeAsp-l-arginine-Adda-d-Glu-Mdhb), in which Mdhb is 2-(methylamino)-2-dehydrobutyric acid (Annila et al., 1996).

## **PHYSIOLOGICAL STUDIES ON MICROCYSTIN PRODUCTION**

Research toward an understanding of the biological role of microcystin started in the 1980s when several research groups began to study the impact of environmental factors on the cellular toxin content. Generally, the production of microcystin by a single strain seems to be rather consistent with variations usually not exceeding a factor of four (Sivonen and Jones, 1999). This led to the general notion that microcystin production is directly correlated to growth, perhaps showing a linear correlation between cell division and toxin production (Orr and Jones, 1998). This was further examined by the development of a model for the prediction of cellular microcystin contents from growth rates in N-limited cultures (Long et al., 2001). This model is in agreement with data obtained in several studies that have reported high microcystin production rates in cultures with high nitrogen and high phosphorus concentrations that are not limited in growth (Sivonen et al., 1990; Vezie et al., 2002). From their data, Orr and Jones (1998) concluded that microcystin is not a secondary metabolite but rather displays attributes of an essential intracellular compound. The fact that microcystin and many other cyanobacterial "secondary metabolites" are produced from the beginning of the logarithmic phase indeed distinguishes them from secondary metabolites of many other microorganisms.

However, a number of other studies have reported effects of specific environmental factors on microcystin production. Lukac and Aegerter (1993) correlated low iron concentrations with an increased toxin production, whereas Utkilen and Gjolme (1992), detected higher microcystin production at high iron conditions. Other authors have suggested the impact of light, temperature, and pH (Vanderwesthuizen and Eloff, 1985; Rapala et al., 1997; Wiedner et al., 2003). Interestingly, some of the environmental factors had a differential effect on various isoforms of the toxin (Rapala et al., 1997). Many of the observed effects on microcystin formation might be due to a limiting response of growth to the respective parameter. The largest problem for the interpretation and comparison of the different studies, however, is the observed strain-specific difference in toxin production under standard growth conditions. The environmentally induced changes are far less pronounced than the differences in toxin contents measured in individual strains. Physiological studies could thus

provide an overview about the influence of abiotic factors on microcystin production but could not provide a clue about the role of the toxin in the producing cells.

## **BIOSYNTHESIS, REGULATION, AND EVOLUTION OF MICROCYSTIN**

A new era of research toward understanding the biological role of microcystin became possible with the identification of the *mcy* genes encoding the non-ribosomal complex for microcystin biosynthesis. The *mcy* clusters were first described for two strains of *Microcystis aeruginosa* (Nishizawa et al., 1999, 2000; Tillett et al., 2000). The genomic locus in *Microcystis* spans 55 kb and comprises 10 bidirectionally organized genes encoding non-ribosomal peptide synthetases (NRPS), polyketide-synthases (PKS), tailoring enzymes, and an ABC transporter (McyA-J; Nishizawa et al., 1999, 2000; Tillett et al., 2000). The *mcy* gene cluster was later also characterized for *Planktothrix* and *Anabaena* (Christiansen et al., 2003; Rouhiainen et al., 2004), a closely related gene cluster that could be assigned to the biosynthesis of nodularin in *Nodularia* (Moffitt and Neilan, 2004).

Knowledge about the biosynthesis gene cluster ultimately permitted analysis of the impact of environmental stimuli on the transcription of the *mcy* genes and a possible interaction of the promoter region with transcription factors. Several studies revealed a clear impact of light and iron-limiting conditions on *mcy* transcript abundance. Kaebernick and co-workers used the RNAse protection assay to quantify *mcy* transcripts under various light intensities and qualities. They observed that *mcy* transcripts are elevated above a critical threshold of light of about 30μmol photons m−<sup>2</sup> s <sup>−</sup><sup>1</sup> (Kaebernick et al., 2000). In a followup study, the authors could show alternative transcriptional start sites for the *mcyA-C* and *mcyD-J* operons in *Microcystis* under low light and high light conditions, respectively (Kaebernick et al., 2002). An increased *mcy* transcript accumulation under high light conditions was also found in *Planktothrix agardhii* (Tonk et al., 2005). A significant rise in the *mcy* transcript accumulation and an increased toxin production was observed following iron starvation (Sevilla et al., 2008). In contrast, the effect of light on transcription could not be correlated with higher toxin content per cell. Studies on the transcription of *mcy* genes were thus not only answering questions but also raising new ones. For example, differential effects on the transcription of various *mcy* genes were observed following various stresses such as oxidative stress. The *mcy*B was only 20% up-regulated whereas *mcy*A, *mcy*H, and *mcy*D transcripts increased by 270, 330, and 370%, respectively, in agreement with published data (Straub et al., 2011).

Screening of the 750-bp long bidirectional *mcy* promoter region between *mcy*A and *mcy*D revealed the presence of transcription factor binding sites for the ferric uptake regulator Fur and the global nitrogen regulator NtcA (Kaebernick et al., 2002). For both factors, binding to the *mcy* promoter region could be confirmed *in vitro* (Martin-Luna et al., 2006; Ginn et al., 2010; **Figure 1**). The Fur binding suggests a response of *mcy* genes to iron availability and redox status of the cell, which could explain the effects of high light and iron-limiting conditions on transcript abundance. On the other hand, NtcA binding typically mediates a response of cyanobacterial cells to nitrogen-limiting conditions

*M. aeruginosa***.** Microcystin biosynthesis is up-regulated under high light and iron deficiency. Two transcriptional regulators, NtcA and Fur, were shown to bind to the bidirectional promoter region of the *mcy* operon (left). A fraction of the toxin is released from lysing cells and via the ABC transporter McyH, thereby acting as an infochemical for the neighboring cells (e.g., leading to an autoinduction of microcystin biosynthesis) and influencing extracellular

proteins such as the lectin MVN and the glycoprotein MrpC. The microcystin receptor is still unknown (middle). Inside cells, microcystin is partly bound to proteins. Binding partners include proteins of the phycobilisomes (PBA) and the large subunit of Rubisco (RbcL). The covalent binding to cysteines is preferentially formed under highlight and oxidative stress conditions. Microcystin can thus interfere with the known redox control of proteins and affect their stability and/or activity (right). TL, thylakoids; CS, carboxysomes.

but the expected transcriptional changes (under different nitrogen concentrations) could not be confirmed (Sevilla et al., 2010). Considering the fact that the cellular redox state influences NtcA expression (Alfonso et al., 2001) and that NtcA and Fur regulons have been previously shown to overlap in cyanobacteria (Lopez-Gomollon et al., 2007), NtcA binding to the *mcy* promoter region may reflect the dependence of *mcy* transcription on the redox status of the cell rather than on nitrogen availability.

Elucidation of *mcy* genes in different genera also provided a basis for the reconstruction of the evolutionary history of these genes. Whereas the sporadic distribution of genes suggested an impact of horizontal gene transfer on *mcy* evolution, the comparison of *mcy* genes from various cyanobacterial genera with house-keeping genes could clearly rule out the possibility of a lateral transfer between different cyanobacterial genera. An evolutionary scenario has been proposed in which the ancestral *mcy* genes were already present in the last common ancestor of all present-day cyanobacteria (Rantala et al., 2004). The sporadic distribution was explained as a result of frequent gene losses. This hypothesis is supported by the observation of ongoing deletion events of partial or complete *mcy* gene clusters in natural and laboratory populations (Schatz et al., 2005; Christiansen et al., 2008). Prior to the phylogenetic analysis and related data showing dynamic changes in the mcy genes, a major role of microcystin as a feeding deterrent was widely accepted. However, since the last ancestor of all extant cyanobacteria preceded the appearance of the metazoans, a primary defense role of microcystin against eukaryote grazers is questionable. It was recently demonstrated that microcystin binds to thiol groups of various proteins (Zilliges et al., 2011; see below) similarly to its binding to certain eukaryotic protein phosphatases. It seems likely that the toxic effect of microcystin (due to inhibition of these phosphatases) reflects a recent evolution of these proteins rather than its primary role. In this respect it is worth mentioning that a recent bioinformatic study suggested that the *sxt* genes encoding saxitoxin, a cyanobacterial neurotoxin, are also ancient and that potassium rather than sodium channels was the original target of this toxin (Murray et al., 2011).

## **COMPARATIVE ANALYSIS OF** *M. AERUGINOSA* **AND MICROCYSTIN-DEFICIENT MUTANTS**

As mentioned above, strain-specific differences in the rate of microcystin production are a major problem for the interpretation and comparison of studies on the impact of environmental stimuli. Moreover, comparative growth or feeding assays with toxic and non-toxic strains are biased by other strain-specific differences in addition to toxicity. The elucidation of microcystin biosynthesis genes provided the possibility to construct mutants that are deficient in microcystin production, but otherwise identical to the corresponding wild-type strains (Dittmann et al., 1997). The use of these mutants in comparative studies has significantly advanced our understanding about the biological role of microcystin.

The microcystin-deficient mutant Δ*mcy*B and the corresponding wild-type strain were used in feeding studies with *Daphnia*. These studies revealed that different *Daphnia* sp. could not distinguish between the wild-type and the microcystin-deficient mutant; both strains were ingested in similar rates (Rohrlack et al., 1999, 2001). These finding argued against a primary role as feeding deterrent as was widely accepted at the time. Wild-type and mutant were also compared for their growth and pigmentation under different light regimes (Hesse et al., 2001). The growth rates of the mutant and the WT were similar when grown at light intensities of 4–110 μmol photons m−<sup>2</sup> s <sup>−</sup><sup>1</sup> but the absorption spectra in the range of photosynthetic active radiance varied significantly. The mutant cells possessed lower contents of chlorophyll a and of several carotenoids under these conditions (Hesse et al., 2001). Furthermore, a lower capability of mutant Δ*mcy*B to acclimate to low inorganic carbon concentrations was proposed (Jahnichen et al., 2007). These results were supported by growth competition studies using *M. aeruginosa* PCC7806 and mutant Δ*mcyB*. The microcystin-producing wild-type had a strong selective advantage under low CO2 conditions (Van de Waal et al., 2011). A spontaneous *Microcystis* mutant, strain MG-J, bearing a deletion in mcy and unable to produce microcystins was isolated from a toxic wild-type strain MG-K (Schatz et al., 2005). Laboratory and field experiments comparing the wild-type with this spontaneous mutant indicated an advantage of the toxic strain over its non-toxic successor. When grown separated by a membrane, which allowed passage of the media but not the cells, MG-K severely inhibited the growth of MG-J. In addition, when MG strains were placed in dialysis bags in Lake Kinneret (Sea of Galilee), Israel, during the season in which *Microcystis* is often observed, cells of MG-J lysed but not MG-K (Schatz et al., 2005). Taken together, the data support the notion that microcystins play an important physiological role in the producing strains.

This, however, is difficult to accommodate with the observation that under natural conditions non-toxic colonies are present, sometimes in large proportion, together with toxic ones. What determines the proportion of microcystins-producing-cells in the natural *Microcystis* population is not clear. A thorough direct analysis of single colonies revealed that most of the colonies in the examined lake did not produce microcystins (Fastner et al., 2001).

## **TOWARD CLARIFICATION OF THE MECHANISTIC ROLE OF MICROCYSTIN-EXTRACELLULAR ROLE**

The availability of the microcystin-deficient mutant also offered the possibility for a more detailed analysis of the role of microcystin at the molecular level. Proteomic comparisons between wild-type and mutant strains revealed the differential accumulation of a prominent protein spot. The protein identified by a reverse genetic approach was designated MrpA (microcystinrelated protein A) and shows some similarity to the RhiA protein of the symbiotic strain *Rhizobium leguminosarum* biovar viciae (Dittmann et al., 2001). Moreover, the *mrpA* gene was cotranscribed with a second gene, *mrpB* that codes for a protein with high similarity to the RhiB protein encoded as part of the same *rhiABC* operon (Dittmann et al., 2001). The Rhi proteins were implicated in symbiotic nodule formation. Interestingly, the operon is under the control of extracellular acylhomoserine lactone (AHL) type signals in *Rhizobium* (Rodelas et al., 1999). This raised the possibility of a role of microcystin as an intercellular signal. Indeed, microcystin is released from *Microcystis* cells, in particular above a critical light threshold (Wiedner et al., 2003). Transport of microcystin through the inner membrane might be mediated by the activity of McyH, which shows homology to ABC transporter (**Figure 1**). A knock-out of the corresponding gene *mcyH*, however, leads to a complete loss of toxin production (Pearson et al., 2004) and thus the mutant could not be used to study the impact of McyH on internal microcystin level or assess a specific role in sensing of extracellular microcystin.

A possible extracellular role could nevertheless be assessed by an external addition of microcystin. Schatz et al. (2007) could show that addition of supernatants from cells lysed either mechanically or because of stress conditions leads to a strong upregulation of microcystin biosynthesis genes in the remaining cells. A less pronounced autoinduction effect could also be seen after addition of pure microcystin or other peptides, micropeptin, and microginin, produced by *Microcystis* which, together with microcystin, formed the stronger effect of the cell-free supernatant. Noticeably, the response of the remaining cells is strongly affected by the age of the culture but was hardly detected in older cultures. Analysis of the abundance of *mcy* transcripts clearly indicated a large accumulation at stationary phase thereby reducing the response to externally provided microcystin. On the other hand, *mcy* transcripts were hardly detected in cultures maintained in early log phase for several generations, but these cells responded strongly to the addition of microcystins or the cell-free supernatant by upregulating the expression of *mcy* genes. It was therefore suggested that microcystins serve as infochemicals; their release to the media by lysing cells due to various stressors is sensed by the rest of the *Microcystis* cells which upregulate *mcy* transcription (**Figure 1**; Schatz et al., 2007). Larger production of microcystin may promote the fitness of the *Microcystis* cells and their competitiveness with other phytoplankton species (reviewed by Pflugmacher, 2002; Babica et al., 2006).

Furthermore, several studies have supported a microcystin role in *Microcystis* colony formation. The expression pattern of two extracellular proteins, the lectin MVN (Kehr et al., 2006) and the glycoprotein MrpC (Zilliges et al., 2008) were strongly altered in mutant strains lacking microcystin. MrpC is the dominant extracellular protein in *Microcystis* and directly implicated in cell–cell contacts between *Microcystis* cells (Zilliges et al., 2008). Both MVN and MrpC seem to be specific for certain *Microcystis* ecotypes. It is not clear yet, if microcystin is directly involved in morphotype formation and the mechanism whereby it influences the cell surface proteins (**Figure 1**).

## **EXPERIMENTAL SUPPORT FOR AN INTRACELLULAR ROLE OF MICROCYSTIN**

As indicated, physiological experiments using wild-type and mutants suggested that the ability to produce microcystin is beneficial for the *Microcystis* cells. Elucidation of the *M. aeruginosa* PCC7806 genome enabled a more comprehensive analysis of molecular differences between wild-type and microcystin-deficient mutant such as via proteomic comparisons in cells maintained under various conditions such as high light and darkness (Zilliges et al., 2011). Remarkably, under all the conditions examined loss of microcystin led to considerable alteration in the protein pattern. Noticeably, a large number of spots could be assigned to Calvin– Benson cycle proteins, the phycobilisome, to redox-dependent reductases, and unidentified proteins (Zilliges et al., 2011). In particular, the differences in protein isoform formation were striking. Generally, the wild-type showed more protein isoforms than the microcystin-deficient mutant (Zilliges et al., 2011). In addition, it was shown that microcystin binds covalently to many of the differentially accumulating proteins *in vitro* and *in vivo* including phycobilisome antennae proteins and RbcL, the large subunit of Rubisco. This may at least partly explain the additional isoforms for proteins in the microcystin-producing wild-type (**Figure 1**;Zilliges et al., 2011) and the change in pigmentation during growth experiments (Hesse et al., 2001). Microcystin attachment to proteins was strongly enhanced under high light and oxidative stress conditions, raising the possibility of a consequent effect on their activity and stability. Many of the differentially accumulating proteins are indeed "cysteine-sensitive," i.e., subject to degradation upon oxidation of their cysteines following oxidative stress. A stabilizing effect of the microcystin binding was confirmed in the case of RbcL that was clearly more resistant to protease digestion in the wild-type (Zilliges et al., 2011). Moreover, growth experiments performed under oxidative stress or very high light conditions revealed a clear advantage for the microcystin-producing wildtype (Zilliges et al., 2011) and electron microscopy showed the localization of microcystin in the thylakoid and carboxysome areas (Gerbersdorf, 2006). These data may explain the differences in pigmentation and adaptation to low inorganic carbon concentrations discussed above and explain some of the controversial issues in earlier physiological studies. The fact that high light conditions lead to an increase in *mcy* transcription but not microcystin production is likely due to strong bias in microcystin detection due to covalent binding to proteins.

Another proteomic study revealed similar differences in the protein patterns between six different natural toxic and non-toxic strains, particularly in nine spots differentially accumulated in all toxic strains. Those were assigned to carbon–nitrogen metabolism and redox maintenance (Alexova et al., 2011). Differences in the acclimation to changing redox or light or carbon conditions may therefore be the main biotic parameters distinguishing toxic from non-toxic strains. This is at least partly supported by growth experiments that revealed an advantage of natural toxic over non-toxic strains under low (Schatz et al., 2005) but not under high carbon concentrations (Van de Waal et al., 2011), and an advantage of toxic strains under elevated temperatures (Dziallas and Grossart, 2011).

## **CYLINDROSPERMOPSIN OCCURRENCE AND TOXICITY**

Cylindrospermopsin (CYN) is an alkaloid toxin produced by several filamentous cyanobacteria including *Aphanizomenon ovalisporum* (hereafter *Aphanizomenon*), *Cylindrospermopsis raciborskii*, *Oscillatoria* sp. PCC 6506, *Anabaena bergii*, *Umezkia nathans*, *Aphanizomenon flos-aquae*, and others (Banker et al., 1997; Kaebernick and Neilan, 2001; Schembri et al., 2001; Preussel et al., 2006; Mazmouz et al., 2010). CYN was shown to accumulate in the environment (Kinnear, 2010) and inhibit protein and glutathione synthesis leading to cell death (Runnegar and Lu, 1994; Runnegar et al., 1995, 2002; Froscio et al., 2001; Metcalf et al., 2004). Effects of genotoxicity and on cholesterol levels have also been reported (Humpage et al., 2000, 2005; Shen et al., 2002; Reisner et al., 2004). Activity of cytochrome P450 enzyme system is apparently essential for the development of CYN toxicity (Runnegar et al., 1995). Damage to DNA (Straser et al., 2011) and carcinogenic activity were also suggested (Humpage et al., 2000; Murphy and Thomas, 2001). The role of specific residues essential for CYN toxicity is emerging in that the uracil moiety and the hydroxyl at position C7 are essential (Banker et al., 2001).

Due to its practical importance, emphasis was placed on abiotic conditions under which blooms of CYN-forming organisms may occur (Hadas et al., 1999, 2002; Burford and Davis, 2011; De Figueiredo et al., 2011) and to assessment of the effect of abiotic parameters on CYN production under field and laboratory conditions such as S and P limiting conditions (Bacsi et al., 2006). A comprehensive description of the many excellent studies on bloom conditions in various water bodies is beyond the scope of this paper. One example is the case of *Aphanizomenon* population in Lake Kinneret that showed close agreement between the optimal conditions for its growth in the laboratory and those which occurred during its bloom in 1994 (Hadas et al., 2002). An important aspect is the size of the inoculum; formation of akinetes; and their germination thereby raising the new population is poorly understood and requires further research (Sukenik et al., 2007). Maximal CYN formation in *C. raciborskii* is observed under light intensity higher than required for optimal growth (Dyble et al., 2006), suggesting that the redox state of the cells is involved. Apparently the status of the *C. raciborskii* culture, logarithmic, or stationary, is also critical to the rate of CYN formation. In both diazotrophic and nitrate supplied log phase cultures, CYN accumulation is linear with the growth rate but decoupling between the two parameters is observed at stationary phase (Hawkins et al., 2001). Similarly, differential effects of temperature on growth and CYN production in various *C. raciborskii* strains were observed (Saker and Griffiths, 2000). Taken together, the bulk of the data indicate that little is known about the cellular mechanisms involved in the regulation of CYN formation

and their interaction with the abiotic conditions experienced by the cells. Furthermore, as in the case of microcystin discussed above, comparison of published data is rather difficult because of lack of information on the exact conditions and the very different conditions in field and laboratory experiments.

## **THE** *aoa***/***cyr* **GENE CLUSTERS AND CYN BIOSYNTHESIS**

The mechanism of CYN biosynthesis was partly resolved by following production in *C. raciborskii* cultures that were supplemented with stable 13C, 18O, 15N, and 3H isotopes, suggesting that five molecules of acetate are catalytically condensed by a polyketide-synthase (PKS) to form the carbon skeleton of CYN. Two other carbons and one of the nitrogen atoms of the guanidine moiety are probably donated by glycine. Guanidinoacetic acid was proposed as a source for two nitrogen atoms of the guanidino moiety used as a starter unit that initiates the biosynthetic process. The guanidinoacetic acid is probably formed by amidination of glycine by amidinotransferase but the exact donor of the other N atoms in the guanidino moiety, most likely arginine, has not been resolved (Burgoyne et al., 2000; Murphy and Thomas, 2001; Shalev-Alon et al., 2002; Pearson et al., 2010). Following the recognition that PKS and amidinotransferase are involved,Shalev-Alon et al. (2002) identified a genomic region in *Aphanizomenon* likely involved in CYN biosynthesis; recognition of a similar genomic region in *C. raciborskii* (Mihali et al., 2008) and *Oscillatoria* sp. strain PCC 6506 (Mazmouz et al., 2010) followed (**Figure 2**). It should be noted, however, that we are missing unequivocal evidence that the *aoa*/*cyr* gene clusters in these organisms are engaged in CYN biosynthesis. Although genes essential for CYN biosynthesis are present in these clusters, since these organisms are, so far, not amenable to genetic manipulations it was not possible to inactivate these genes and thereby provide a critical evidence that we have been seeking.

Support to the notion that the aoa/cyr gene cluster is indeed involved in CYN biosynthesis rests on the observation that a sulfotransferase-encoding gene essential for CYN synthesis, *cyrJ*, is present only in CYN-producing *C. raciborskii* strains (Mihali et al., 2008; Alster et al., 2010). Similarly, *cyrJ* is detected in four CYN-producing *Oscillatoria* strains but lacking in a non-CYN-producing strain (Mazmouz et al., 2010). Finally, non-toxic *Aphanizomenon* isolated from Lake Kinneret possess *aoaA* and *aoaC* but is missing *aoaJ* (Ballot et al., 2011).

Following the identification of the gene clusters putatively involved in CYN biosynthesis (**Figure 2** and below), an attractive (and most likely) scheme of the cascade of events leading to CYN formation was proposed (Mihali et al., 2008; Pearson et al., 2010; the reader is referred to these publications for a comprehensive discussion of the proposed scheme). However, decisive evidence that this is indeed the case must await biochemical confirmation. Furthermore, the proposed scheme does not account for the formation of two CYN analogs, 7-deoxycylindrospermopsin (7 deoxy-CYN) and 7-epicylindrospermopsin (7-epi-CYN) whose relative abundance varies in different cyanobacterial species (see Mazmouz et al., 2011), suggesting alterations in the activity and regulation of various steps in CYN production. It was suggested that CyrI, a 2-oxoglutarate-dependent iron oxygenase performs the last step of CYN biosynthesis from 7-deoxy-CYN (Mazmouz et al., 2011). These authors raised the possibility that a relaxed

stereospecificity of CyrI produces both stereoisomers at position 7. Alternatively, an unidentified hydroxylase produces the 7-epi-CYN from 7-deoxy-CYN. The levels of CyrI or its activity in various CYN-producing cyanobacteria should be examined to account for the variability in the abundance of 7-deoxy-CYN compared to CYN; 7-deoxy-CYN is a minor product in *C. raciborskii* and *Raphidiopsis curvata* (Li et al., 2001) but a major one in *Lyngbya wollei* (Seifert et al., 2007). On the other hand 7-epi-CYN was detected only in *Aphanizomenon* in small quantities.

role of this cluster.

**Figure 2** shows the genomic organization of the *aoa*/*cyr* in four CYN-producing cyanobacteria. In the absence of a detailed phylogenetic analysis it is hard to assess the development of these clusters but high similarity between the genes does indicate that they are evolutionarily related, possibly from a common ancestor, and that an horizontal gene transfer was involved (Stuken and Jakobsen, 2010). The differences in the gene organization clearly indicate divergence from this ancestor; shuffling of the aoa/*cyr* genes is substantial. The two*Aphanizomenon* clusters are quite similar but with two main exceptions. Two *aoaO* genes can be identified in strain ILC146 (draft, yet unpublished sequence) but we could not detect it in strain 10E6 (Accession Number, WOS, 000281239600017;

Stuken and Jakobsen, 2010). Altogether, the role of *aoaO*/*cyrO* in CYN production, if any, is not clear due to its low homology to other genes in the data bases. Again, a genetic approach could resolve this question. Compared with strain ILC146, the region of *aoaH-K* underwent reorganization in strain 10E6 with the *aoaK* encoded on the opposite strand and precedes *aoaH*. The presence of several transposase encoding genes is noted and they may be involved in the reorganizations.

## **REGULATION OF TRANSCRIPTION**

Following the identification of the genomic regions putatively involved in CYN biosynthesis in *Aphanizomenon* (Shalev-Alon et al., 2002), *C. raciborskii* (see Neilan et al., 2008), and *Oscillatoria* strain PCC 6506 (Mazmouz et al., 2010), it became possible to study their regulation. In both *Aphanizomenon* species but not in *C. raciborskii* and *Oscillatoria* sp. (**Figure 2**) the amidinotransferase and polyketide-synthase encoding genes, *aoaA* and *aoaC*, respectively, are transcribed on the reverse strands. Analysis of the genomic region between them identified two transcription start points for each of these genes, differentially expressed under nitrogen and light-stress conditions (Shalev-Alon et al., 2002) and by phosphate availability (Bar-Yosef et al., 2010). Gel shift assays and DNA affinity columns isolated a protein that specifically binds to a 150-bp DNA fragment from the region between *aoaA* and *aoaC* in *Aphanizomenon* ILC146 and MS/MS analyses identified similarity to AbrB in other cyanobacteria and in *Bacillus* sp. Comparison of the native AbrB isolated from *A. ovalisporum* with that obtained after cloning and over-expression of *abrB* in *E. coli* identified specific post-translational modifications in the native cyanobacterial AbrB. This includes N-acetylation of residue T2 and methylation of specific residues,which are all missing in the protein expressed in *E. coli*. A gene encoding a putative acetylase was identified in the draft genomic sequence of *Aphanizomenon* ILC146 (yet unpublished) but, as mentioned with respect to the role of *aoa* genes, failure to genetically modify or inactivate genes in this organism does not enable recognition of the significance of these modifications to *aoaA*/*C* expression. The exact role of these post-translational modifications to the regulation of CYN formation is yet to be established. It is worth mentioning that AbrB-like proteins were thereafter recognized in various cyanobacteria. In *Synechocystis* sp. PCC6803 they were implicated in the regulation of several genes belonging to inorganic carbon uptake (Lieman-Hurwitz et al., 2009), nitrogen metabolism and C–N regulation (Yamauchi et al., 2011), the bidirectional hydrogenase (Oliveira and Lindblad, 2008), and in iron superoxide dismutase in *Nostoc* sp. strain PCC 7120 (Agervald et al., 2010). However, as much as information gained from these fields may contribute to our understanding of the involvement of these AbrB-like (designated CyAbrB) in CYN formation, a comprehensive analysis is beyond the scope of the present review.

## **THE BIOLOGICAL ROLE OF CYN**

Laboratory and field experiments revealed the biological role and function of CYN (Bar-Yosef et al., 2010). Many microorganisms, from bacteria to yeast, respond to inorganic phosphate (Pi) limitation by expressing their PHO regulon and excretion of alkaline phosphatase (APase) thereby enhancing the supply of Pi from organic compounds and improving their competitiveness for resources and domination. Analysis of the abundance of *Aphanizomenon* and APase activity in Lake Kinneret during the summers of 1994–2009 showed strong correlation between these parameters, suggesting that APase emerged from Pi-limited filaments, but closer analysis clearly indicated this was not the case (Bar-Yosef et al., 2010). Application of enzyme labeled fluorescence (ELF) on lake samples showed APase in various organisms but not in *Aphanizomenon*, and staining by DAPI revealed high level of polyphosphate bodies (PPBs) in *Aphanizomenon*. ELF-APase signals were detected in *Aphanizomenon* only after it exploited the PPBs under prolonged Pi deprivation in laboratory cultures or toward the end of its natural bloom, in the autumn (Bar-Yosef et al., 2010). *Aphanizomenon* cultures respond to removal of external Pi by fast induction of CYN production and only part of its PHO regulon, including the high affinity Pi uptake system, PstS, and *phoD*, encoding an internal APase. In contrast, transcripts encoding extracellular APase and its activity were observed only after 5–7 days during which it consumes the internal Pi.

Addition of spent media from *Aphanizomenon* or CYN to various phytoplankton species (*Chlamydomonas reinhardtii* used as a model organism) led to induction of genes typically up-regulated by Pi-limitation (Bar-Yosef et al., 2010) and a large rise in the extracellular APase activity despite the presence of ample Pi in the medium. Using ELF-APase to stain mixed cultures of *Aphanizomenon* with *Chlamydomonas* or *Debarya* sp. showed APase in the latter two but not in *Aphanizomenon*. These data suggested that CYN induces the excretion of APase by *other* phytoplanktons in the water body. *Aphanizomenon* can then compete for the Pi released consequent to APase activity from other organisms. Apparently *Aphanizomenon* does it quite successfully, grows rapidly, and dominates in the seasonal bloom. CYN is used to "enslave" other organisms to fulfill the Pi demands of the "toxic" strains. It explains the observed rise in APase activity in the water body during the early phase of *Aphanizomenon* bloom and reveals a mechanism whereby it can form massive blooms despite limiting Pi conditions.

Noticeably, the response of target organisms to *Aphanizomenon* spent medium was greater than observed following application of purified CYN, although its concentration in the used media was considerably smaller. This suggests a synergistic involvement of other secondary metabolites in the used media, the nature of which is not known. It resembles the stronger induction of *mcy* genes in *Microcystis* sp. by its own spent medium rather than by isolated microcystin, discussed above.

Naturally we expect other CYN producers such as *C. raciborskii* to induce APase in other organisms. Lack of ELF-APase staining in lake samples of *Anabaena* sp., *Planktothrix* sp., and *Planktolyngbya* sp. (Rengefors et al., 2003) and delayed APase secretion in Pi-deprived *Trichodesmium* sp. (Orchard et al., 2009) may indicate similar capabilities in other cyanobacterial lineages that are not known to produce CYN.

Finally, there are numerous reports over the last decade showing a significant rise in the abundance of toxic filamentous N2 fixing cyanobacteria in freshwater bodies (See Sukenik et al., this issue). The reason is not known, particularly as it occurs simultaneously with intensive efforts of lake management authorities to reduce nutrient inflow from the watershed. The rise in cyanobacterial abundance has been attributed to various causes, including global warming, but in view of the role attributed to CYN we raise an alternative possibility that superior competitiveness for Pi enables these organisms to flourish in such lakes. In terms of competitiveness, it is worth mentioning that N allocation to produce CYN is cheaper for the cells than APase secretion (Bar-Yosef et al., 2010).

#### **OUTLOOK: THE MAIN OPEN QUESTIONS**

Here we focused on the biological role of two cyanobacterial toxins,microcystin and cylindrospermopsin, but many open questions need to be resolved. We proposed external and internal roles of microcystin and presented experimental evidence to support our view. Given the phylogenetic analysis that revealed that *mcy* (and *sxt*) are ancient but were lost in most of the present-day cyanobacterial strains, we raise the question what, if any, fulfills these biological roles in the other cyanobacterial strains lacking microcystin. What were the evolutional drivers that led to the deletion of *mcy* genes from most strains? We are missing an experiment whereby spent media, and/or supernatant from broken cells of one of the strains for which genetic information and tools (such as microarray) are available, will be applied to its own population. This will be followed by an analysis of the transcript abundances. Perhaps some other secondary metabolites perform these roles, as do micropeptin and microginin for the infochemical task in *Microcystis*. Furthermore, given the advantage of toxic over non-toxic *Microcystis* strains in growth competition experiments, why do we often find them inhabiting the same habitat, indoor as well as outdoor? Is it possible that while saving the resources necessary to produce microcystin, the non-toxic strains use microcystin produced by the toxic ones? What is in fact the role of McyH, which shows homology to ABC transporters but its inactivation blocks microcystin biosynthesis? Does it, like some

## **REFERENCES**


of a non-toxic invasive *Cylindrospermopsis raciborskii* from a Mediterranean Lake. *Hydrobiologia* 639, 115–128.


other transporters, serve as the sensor for microcystin presence in the environment?

For CYN, with the emergence of more *aoa*/*cyr* sequences, it is becoming possible to perform a phylogenetic analysis for these gene clusters as well but the main task is to develop the means to make at least one of the CYN-producing strains amenable to genetic manipulations. We are still missing experimental evidence that this gene cluster is involved in CYN production. What are the mechanisms whereby presence of CYN induces the PHO regulon in other phytoplankton species even in the presence of ample Pi externally? What are the mechanisms enabling the differential expression of some of the PHO regulon genes in *Aphanizomenon* whereby external alkaline phosphatase is blocked as long as internal P is available in the form of polyphosphate bodies? What is the nature of the other components in the spent medium from *Aphanizomenon* that act synergistically with CYN? Is this ability to improve its own P supply by enslaving other organisms unique to *Aphanizomenon*? Can a similar mechanism be found in other organisms for P or other nutrients? These are just some of the open questions ensuring an exciting research period ahead of us.

## **ACKNOWLEDGMENTS**

We thank Dr. R. Reinhardt and Dr. M. Kube from Max Planck Institute for Molecular Genetics, Berlin, for their guidance and support with the molecular analysis of *A. ovalisporum* genome. Research in our laboratories is supported by grants from the Israel Science Foundation to Aaron Kaplan; Israeli Science and Technology Ministry (MOST) to Aaron Kaplan and Assaf Sukenik; the German Research Foundation (Di910/3-1) to Elke Dittmann; the Hebrew University of Jerusalem to Aaron Kaplan; the Israel Water Authority to Assaf Sukenik and Aaron Kaplan; the Yohai Ben-Noon Foundation to Moshe Harel; and by the EU-NEST program to Assaf Sukenik, Ora Hadas, and Ruth N. Kaplan-Levy.

C. (2011). Occurrence of noncylindrospermopsin-producing *Aphanizomenon ovalisporum* and *Anabaena bergii* in Lake Kinneret (Israel). *J. Plankton Res.* 33, 1736–1746.


*aeruginosa* PCC 7806. *Appl. Environ. Microbiol.* 73, 6994–7002.


apparent photosynthetic affinity for CO2 by modulating low-CO2 induced gene expression. *Environ. Microbiol.* 11, 927–936.


gene cluster responsible for cylindrospermopsin biosynthesis. *Appl. Environ. Microbiol.* 74, 716–722.


analysis. *PLoS ONE* 6, e16208. doi:10.1371/journal.pone.0016208


a water column. *New Phytol.* 136, 189–209.


T., and Dittmann, E. (2008). An extracellular glycoprotein is implicated in cell-cell contacts in the toxic cyanobacterium *Microcystis aeruginosa* PCC 7806. *J. Bacteriol.* 190, 2871–2879.

**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: 21 January 2012; accepted: 23 March 2012; published online: 18 April 2012.*

*Citation: Kaplan A, Harel M, Kaplan-Levy RN, Hadas O, Sukenik A and Dittmann E (2012) The languages spoken in the water body (or the biological role of cyanobacterial toxins). Front. Microbio. 3:138. doi: 10.3389/fmicb.2012.00138*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Kaplan, Harel, Kaplan-Levy, Hadas, Sukenik and Dittmann. This is an open-access article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits non-commercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*

## Invasion of Nostocales (cyanobacteria) to subtropical and temperate freshwater lakes – physiological, regional, and global driving forces

## *Assaf Sukenik 1\*, Ora Hadas 1, Aaron Kaplan2 and Antonio Quesada3*

*<sup>1</sup> The Yigal Allon Kinneret Limnological Laboratory, Israel Oceanographic and Limnological Research, Migdal, Israel*

*<sup>2</sup> Department of Plant and Environmental Sciences, The Hebrew University, Jerusalem, Israel*

*<sup>3</sup> Departamento de Biología, Universidad Autónoma de Madrid, Madrid, Spain*

### *Edited by:*

*Anton F. Post, H. Steinitz Marine Biology Laboratory, USA*

#### *Reviewed by:*

*Robert M. McKay, Bowling Green State University, USA Rachael Marie Morgan-Kiss, Miami University, USA*

#### *\*Correspondence:*

*Assaf Sukenik, The Yigal Allon Kinneret Limnological Laboratory, Israel Oceanographic and Limnological Research, P.O. Box 447, Migdal 14950, Israel. e-mail: assaf@ocean.org.il*

Similar to the increased number of studies on invasive plants and animals in terrestrial and aquatic ecosystems, many reports were recently published on the invasion of Nostocales (cyanobacteria) to freshwater environments worldwide. Invasion and proliferation of Nostocales in new habitats have the potential to significantly alter the structure of the native community and to modify ecosystem functioning. But most importantly, they influence the water quality due to a variety of toxic compounds that some species produce. Therefore a special attention was given to the invasion and persistence of toxic cyanobacteria in many aquatic ecosystems. Here we summarize the currently published records on the invasion of two Nostocales genera, *Cylindrospermopsis* and *Aphanizomenon*, to lakes and water reservoirs in subtropical and temperate zones.These invading species possess traits thought to be common to many invasive organisms: high growth rate, high resource utilization efficiency and overall superior competitive abilities over native species when local conditions vary. Assuming that dispersion routes of cyanobacteria have not been changed much in recent decades, their recent establishment and proliferation in new habitats indicate changes in the environment under which they can exploit their physiological advantage over the native phytoplankton population. In many cases, global warming was identified as the major driving force for the invasion of Nostocales. Due to this uncontrollable trend, invasive Nostocales species are expected to maintain their presence in new habitats and further expand to new environments. In other cases, regional changes in nutrient loads and in biotic conditions were attributed to the invasion events.

**Keywords: cyanobacteria, Nostocales,** *Cylindrospermopsis***,** *Aphanizomenon***, invasive species, climate change, eutrophication**

## **INTRODUCTION**

The entry, establishment, and spread of non – native species to a new ecosystem, is frequently described as biological invasion (Vitousek et al., 1997; Ricciardi and Cohen, 2007). In many cases, biological invasions are caused by human – induced environmental interference and have substantial economic impact, serious ecological consequences, and human health hazards (Pimentel et al., 2001). They also may threaten global biodiversity by altering the structure and function of ecosystems and disrupting key biological interactions (Levine et al., 2003; Traveset and Richardson, 2006). The invasion of freshwater and marine ecosystems by non-native species of various taxonomic groups, mostly crustaceans, mussels, fish, and aquatic macrophytes is well documented (Meinesz et al., 2001; Richardson, 2011). The invasion of free-living microorganisms to new aquatic habitats is rather cryptic and difficult to detect therefore invasions of these "invisible invaders" have been rarely reported (Litchman, 2010). Furthermore, information on prior and current composition of the microbial communities, which is required to determine if a given microorganism is an invader, is often missing. Algae and cyanobacteria are exceptions in that sense

as they have visible characteristic of spectral signature and microscopic morphological features. This property partly contributed to the increased number of reports on cyanobacteria invasion to lakes and water reservoirs across Europe, North, and South America and the Middle East that were recently published (Dyble et al., 2002; Wiedner et al., 2007; Hadas et al., submitted).

Here we summarize recently published records on the invasion of cyanobacteria to subtropical and temperate lakes and water reservoirs and try to forecast different scenarios of ecological and management consequences. The invasion of two Nostocales genera, *Cylindrospermopsis* and *Aphanizomenon* is presented and analyzed based on the main concepts of invasion ecology. The following issues are critically addressed: recent variations in geographic distribution of invasive Nostocales; How does a Nostocales species become an invader? What are the special traits of the invading Nostocales? What are the level and the role of genetic variability in the invasion of Nostocales? Is invasion of Nostocales to new location further expected and under what conditions? What are the expected ecological effects of the invasion?

## **THE CONCEPT OF INVADING CYANOBACTERIA**

Biological invasion is considered as an extension of normal colonization processes such as succession (Elton, 1958) therefore it was proposed that the term "invasive species" will be restricted to describe non-native species that expanded their geographic range, became abundant and have environmental and economic impact (Colautti and Macisaac, 2004). Because the terminology associated with biological invasion is rather confusing and sometimes misleading (Colautti and Macisaac, 2004), here we define an invading or a non-native species as a species that spreads beyond its native range or a species that enters a habitat where it had not resided before. However, it should be kept in mind that many cyanobacteria species (as well as other microorganisms) can inhabit an aquatic ecosystem unnoticed, since it may remain at a rather low biomass concentration and it does not form a conspicuous fraction of the community. Furthermore, the lack of detailed water quality monitoring program or an insufficiently comprehensive surveillance do not pick up the presence of these species (Kling et al., 2011). Typically it is only when a non-native population is widespread and abundant, and causes ecological or economic harms, that it is named "invasive." It is possible for a species to be invasive in one ecosystem, but non-invasive in another. This can be due to a variety of factors, such as the presence of less-than-ideal habitat conditions (Lockwood et al., 2007). We further suggest evaluating the invasion of free-living microorganisms within an appropriate timeframe, since an invader may become a common resident of a dynamically changing environment. Therefore native and non-native species may be viewed as part of a continuum with respect to their time of residency and the extent of evolutionary and ecological interactions they have within a given environment.

## **THE GLOBAL EXPANSION OF NOSTOCALES**

During the last decades, the number of reports on cyanobacterial domination and persistence in freshwater and marine ecosystems significantly increased covering spatial and temporal distribution of persistent perennial blooms. Many of these bloom events are of toxic *Microcystis* species and strains that proliferate under current environmental alterations, including nutrient enrichment, global warming, and regional hydrologic changes (Paerl and Paul, 2011). However, not all blooms are associated with an invasion process as many species are of broad geographic distribution and rapidly respond to current environmental changes.

The most important freshwater invasive Nostocales are the tropical species *C. raciborskii* and to a lesser extent also *Aph. ovalisporum*. The map in **Figure 1** shows the current global distribution of *C. raciborskii* and *Aph. ovalisporum* in their native tropical region and in subtropical and temperate areas to which they expanded in the last two decades. *C. raciborskii* is a tropical species which was first described from Indonesia (Java) and was subsequently discovered in India and Thailand, Japan, the Philippines, Egypt, Senegal, Brazil, Cuba, Chile, Uruguay, Australia, New Zealand, and Florida (Bouvy et al., 2000; Molica et al., 2002; Dobberfuhl, 2003; Komárek and Komárkova, 2003; Wood and Stirling, 2003; Briand et al., 2004; Chonudomkul et al., 2004; Vidal and Kruk, 2008). Recently Kaštovský et al. (2011) summarized the invasion of freshwater cyanobacteria and algae to aquatic ecosystems in the Czech Republic and portrayed the invasion of *C. raciborskii* from its native tropical area to subtropical and temperate zones. The expansion of *C. raciborskii* and *Aph. ovalisporum* to new geographic regions is described below.

## **CENTRAL EUROPE (GERMANY, FRANCE) AND THE IBERIAN PENINSULA**

The invasion of cyanobacteria strains in Central Europe as high biomass of the typically tropical species *C. raciborskii* was first reported in the 1970s (Romo and Miracle, 1994; Padisák, 1997). Long term phytoplankton records in some European lakes demonstrate a clear increase of some cyanobacterial species that had been absent or overlooked before (Padisák, 1997). Although *C.*

*raciborskii* was known in Europe since the 1930s, it was marginal to the southern European countries (e.g., Greece; see references in Mehnert et al., 2010). However, in the last two decades, this species has appeared in northern European regions and is currently widespread at central latitudes (Kokoci ´nski et al., 2009). Other Nostocales such as *Aph. aphanizomenoides* (currently designated as *Sphaerospermum*) and *Aph. ovalisporum*, are also considered invasive species in Europe and have been reported at some locations as newcomers. *Aph. ovalisporum* is especially relevant because it is known to produce cylindrospermopsin, although several nontoxic strains were reported (Ballot et al., 2011). *Aph. ovalisporum* was described in different ecosystems of southern Europe (Bazzichelli and Abdelahad, 1994) and in the Iberian Peninsula where it produced massive blooms for the first time in 2005 (Quesada et al., 2006). Later, it has been found at least in seven water bodies in central Spain (Cirés, 2012) and in several water bodies in the southern areas of the country (C. de Hoyos, Personal Communication). *Aph. ovalisporum* has been found in the Iberian Peninsula, both in deep stratified reservoirs and in very shallow (less than 1 m) ponds of urban gardens. The spread of *Aph. ovalisporum* throughout the peninsula is considered invasion since a solid reference for its presence and bloom prior to the current reports is missing.

While field studies on the adaptation of the European species and their invasive potential are very scarce, laboratory studies with isolated strains have demonstrated that some invasive species have strong competitive abilities under changing scenarios. Mehnert et al. (2010) compared several "native" strains with "invasive" species and found that under low temperature, native species showed a higher growth rate than the invasive ones. However, under higher temperatures, the trend was the opposite and invasive strains performed better. Using these data, the authors, in a modeling effort, demonstrated that under a scenario of climate change with an increase of 4˚C in the water temperature, the invasive species *C. raciborskii* would outcompete the native species (*Aph. gracile*) and that the potentially invasive *Aph. ovalisporum* would be an important component in the community (Mehnert et al., 2010). Although akinete production and overwintering/seeding play an important role in the life cycle of Nostocales, it is not necessarily the only mechanism for seeding; for example *Aphanizomenon* can seed from vegetative cells and filaments overwintering under ice (Grossart HP, personal communication). Temperature-dependent release of cylindrospermopsin was reported for *Aph. ovalisporum*, implicating management complications associated with global change scenarios (Cirés et al., 2011).

#### **MIDDLE EAST**

An exceptional bloom of *Aph. ovalisporum* first appeared in Lake Kinneret (Sea of Galilee), Israel in July 1994 and dominated the summer phytoplankton population for a few weeks (Pollingher et al., 1998). This toxic species (a producer of cylindrospermopsin) remained a regular member of the summer phytoplankton community since then and variably contributes between 10 and 40% to the summer phytoplankton biomass (Zohary, 2004). The reason for *Aph. ovalisporum's* first appearance and its later establishment in Lake Kinneret was a subject for debate (Berman, 1997, 2001; Berman and Shteinman, 1998; Gophen et al., 1999) but it is clear now that this species took advantage of a new set of conditions that were imposed by global and regional changes (Hadas et al., submitted). The origin of the *Aph. ovalisporum* strain that established in Lake Kinneret remained unclear. A long and detailed record of the phytoplankton population in Lake Kinneret (since 1964) clearly shows its absence, although its presence at a level which is below the detection limit cannot be excluded. Interestingly, the *Aph. ovalisporum* strain isolated from Lake Kinneret was genotypically related to strains of *Anabaena bergii* isolated from constructed man-made lakes in Queensland,Australia (Shaw et al., 1999). Nevertheless, *Anabaena bergii* was not reported to bloom in the lake. The invasion process was repeated once again in Lake Kinneret as *C. rasiborskii* was first reported in 2000 as a minor component of the summer phytoplankton community but dominated the community in summer 2005 with biomass as high as 68 g (ww) m−<sup>2</sup> (Hadas et al., submitted). Again, it is rather difficult to trace the origin of Lake Kinneret *C. raciborskii* but the fact that it is a non-cylindrospermopsin producing strain suggests that it may be related to European strains. These isolates do not produce cylindrospermopsin as opposed to Australian strains known to produce cylindrospermopsin and to Brazilian strains that have been reported to produce paralytic shellfish poisoning toxins (Neilan et al., 2003).

Other reports on Nostocales species in the Middle East include the Caspian Sea and Egypt. *C. raciborskii* was reported as a dominant phytoplankton species in the open and freshwater basin in the west part of the Anzali Lagoon (Caspian Sea, Iran). It was also dominant for short periods after flooding events in rivers flowing into the lagoon (Ramezanpoor, 2004). Toxic *C. raciborskii* and *Raphidiopsis mediterranea* were reported in an Egyptian shallow (4 m depth) freshwater El-Dowyrat fish pond for the first time in 2002. Human activities at the end of 2001, including removal of trees and elimination of macrophytes around the fish pond have led to an increase in temperature at the surface waters of the pond (25–30˚C). Both species showed seasonal variations with highest densities recorded in August each year (Mohamed, 2007). The loss of macrophytes was reported as one of the variables promoting the proliferation of *C. raciborskii* also in New Zealand lakes (Ryan et al., 2003). Thus local maintenance activities may be involved in an increase in temperature of the water column of temperate lakes and expand the distribution of *C. raciborskii* (Briand et al., 2004).

## **NORTH AND SOUTH AMERICA**

The record of *C. raciboskii*-like Nostocales in North America goes back to 1955 when *Anabaenopsis seriata* Prescott, was reported from Kansas. Other reports of morphologically similar species originated from Minneapolis and Lake Erie (Kling, 2009). It was only in the 1990s when*C. raciborskii* was identified in many Florida lakes and rivers (Chapman and Schelske, 1997). Since then, *C. raciborskii* has been found in many water bodies of the central and eastern United States (Kling, 2009) including in many lakes and reservoirs in Indiana (Jones and Sauter, 2005). These sites tended to be shallow and turbid with high epilimnetic phosphorus and high chlorophyll *a* values (Jones and Sauter, 2005). In 2004, *C. raciborskii* was identified in water samples from the Assiniboine River, Manitoba Canada. Variable morphological features

were reported in accordance with earlier reports from Europe, and predictions based on the adaptability of that species to eutrophic conditions, elevated temperatures and turbid waters suggested its expansion in Lake Winnipeg and replacing the original phytoplankton assemblage (Leavitt et al., 2006; Patoine et al., 2006). Indeed summer blooms development and phytoplankton succession was recently reported. Lake Winnipeg shows reduced taxonomic diversity and an increased predominance of nitrogen-fixing cyanobacteria (Kling et al., 2011).

In South America *C. raciborskii* is well known in tropical waters of Brazil (Huszar et al., 2000), although the information about its distribution in southern latitudes is disperse and incomplete (Komárek, 2002). Vidal and Kruk (2008) analyzed spatial distribution and relative frequency of *C. raciborskii* in southern Uruguay, probably its southernmost distribution. *C. raciborskii* was found in several polymictic shallow eutrophic lakes as a dominant summer population and variable morphology. It was suggested higher occurrence of *C. raciborskii* in that climatic zone was associated with shallow, well-mixed eutrophic systems. Furthermore, morphological variations in *C. raciborskii* were not associated with latitudinal, continental, or even country distribution and independent of climatic origin.

## **THE INVASION PROCESS**

Successful biological invasions involve complex interactions between the invading species and the physical and biological characteristics of the recipient environment. Often, invasive cyanobacteria species owe their success in colonizing new ecosystems to unique physiological characteristics (**Figure 2**). The invasion may be initiated with dispersion to new zones and is facilitated by

human activities by migrating animals or birds or by winds. Obviously, aeolian transport is an efficient means to transfer free-living microorganisms, prokaryotes, and protists (Smith et al., 2011). But reaching a new environment is not sufficient and the invader needs a variety of traits that support its establishment and proliferation. Two distinctive traits of Nostocalesa are: (1) The ability to form dormant cells (akinetes) that may survive long and extreme dispersion routes, on one hand and serve as overwintering form that survive unfavorable conditions and assure perennial germination and proliferation, on the other hand. (2) The ability to fix atmospheric nitrogen in the absence of combined inorganic sources. This last trait opens invasion opportunities for Nostocales in oligotrophic aquatic ecosystems, besides ecosystems temporarily limited in combined nitrogen and extends the spectrum of ecosystems to which they can invade.

Invasive cyanobacteria have growth rates that are relatively high compared with those of native species (Isvánovics et al., 2000). They may have high resource use efficiency or possess traits allowing them to gain access to resources unavailable to other species, such as the capacity to fix molecular nitrogen. *C. raciborskii* has high competitiveness under fluctuating nitrogen availability due to efficient uptake capacity for ammonium (Figueredo et al., 2007) and it also can fix atmospheric nitrogen, which allows it to escape N limitation (Briand et al., 2004; Moisander et al., 2011). This species may also utilize other limiting resources, such as phosphorus, more efficiently than other cyanobacteria due to high affinity and P storage capacity (Isvánovics et al., 2000;Wu et al., 2011) and, thus, be a superior nutrient competitor. Padisák (1997) suggested that *C. raciborskii*, tends to invade lakes with very high nutrient concentrations and low phytoplankton diversity, in addition to its ability to form akinetes allowing easy dispersal and environmental resistance. An important feature of *Cylindrospermopsis* is its wide thermal tolerance (Briand et al., 2004), which is essential to maintain the populations during cold winters. Detailed information on the effect of temperature on Nostocales life cycle processes such as akinete formation, their survival and germination, and the recruitment of vegetative filaments is currently scarce, but Cirés (2012) indicated that akinetes production of *Aph. ovalisporum* is rather restricted to a relative narrow temperature range. Padisák (1997) reported that akinetes of *C. raciborskii* germinate at temperatures <24˚C. Further elucidation of life cycle processes responding to temperature is required in order to determine the invasiveness capabilities and limitations of various Nostocales species.

In many cases, floating Nostocales populations create a thick scum that blocks light penetration through the water column and inhibits the growth of other species, thus under high cyanobacterial growth, the populations remain established for longer periods minimizing competition by other algal groups. A scum formation is a dynamic process that depends on a delicate balance between flotation, cellular ballast and water turbulence (Calandrino and Paerl, 2011). Filaments floating at the water surface gain advantage over sub-surface phytoplankton populations. They directly intercept CO2 diffusing into the water from the atmosphere, thus minimizing inorganic carbon limitation of photosynthetic growth. Buoyant species contain UV-absorbing compounds such as mycosporine-like amino acids (MAAs) and scytonemin that ensure survival under extremely high irradiance conditions (Paerl

and Paul, 2011). However, in some occasions scum formation represents a trap for the organisms that lose their dynamic buoyancy behavior and they are exposed to full sunshine (including high UV radiation in summer), as well as to nutrient deprivation. Frequently they die and remain on the water body shores with extremely low metabolic activities (i.e., C and N uptake, Oliver and Ganf, 2000 and references therein).

Another trait commonly found among Nostocales but also in other cyanobacteria (i.e., *Microcystis*) is the ability to synthesize a suite of secondary metabolites some of which were identified toxic to humans and animals (Stewart et al., 2011) and others may affect community structure functioning as allelo-chemicals that inhibit other phytoplankton species or deter and reduce grazing (Fastner et al., 2007; Paerl et al., 2011). Allelopathy was suggested as a beneficial trait of *C. raciborskii* that contributes to its stable dominance and geographic expansion (Figueredo et al., 2007) but also for other cyanobacteria species (Kaplan et al., in review). A unique feature of *Aph. ovalisporum* to acquire phosphate under oligotrophic conditions was recently reported by Bar-Yosef et al. (2010). Excreted chemicals including the toxin cylindrospermopsin cause cells of some other phytoplankton species to make extracellular alkaline phosphatase whose inorganic phosphate product can be used by *Aph. ovalisporum* due to its high affinity to phosphate, thus contributing to its proliferation and domination. *Aph. ovalisporum* has been also found to outcompete other cyanobacteria at high temperatures due to higher growth rates than other Nostocales under these circumstances (Mehnert et al., 2010; Cirés et al., 2011).

In an attempt to follow the invasion route of *C. raciborskii* to new ecosystems Neilan et al. (2003) studied genetic variations between strains isolated from freshwater rivers and reservoirs in Australia, Brazil, Germany, Hungary, Portugal, and the USA by characterizing their 16S rRNA gene nucleotide sequences and by analyzing cyanobacterium-specific short tandem repeat sequence (HIP1). Three distinct groups of *C. raciborskii* strains were clustered: (1) a group consisting of strains from the USA and Brazil; (2) a group comprised of European strains (Germany, Hungary, and Portugal); (3) a group of strains from Australia. A subsequent study that examined other isolates of *C. raciborskii* revealed the same continental cluster distribution based on the 16S–23S internally transcribed spacer (ITS1) sequences (Gugger et al., 2005). This study suggests that the current expansion of *C. raciborskii* in Europe and in Central – and North America did not result from recent invasion and colonization by African or Australian strains (Padisák, 1997) but rather represent local strains that maintained "cryptic" populations over time and only recently proliferated due to climate change and variations in other environmental conditions. In a recent study, Piccini et al. (2011) proposed that phenotypic and genetic variability of *C. raciborskii* populations is linked to the existence of different ecotypes whose success is subject to the local environmental conditions.

Nevertheless, analysis of the 16S rRNA gene nucleotide sequences revealed 99.1% similarity between the strains of *C. raciborskii* collected from a range of global locations and continental clustering was not supported by statistically significant bootstrap values (Neilan et al., 2003). Therefore it cannot be excluded that *C.* *raciborskii* was transferred from the Australasian region to Europe unintentionally by humans or by migrating birds, thus providing the source of this cyanobacterium which is now reported in many previously unaffected temperate and subtropical water bodies. It is further speculated that some physiological characteristics unique to *C. raciborskii* support its proliferation in newly invaded ecosystems as they are currently exposed to higher temperature and environmental perturbations.

## **GLOBAL VERSUS REGIONAL DRIVING FORCES**

In order to exploit its wide spectrum of physiological traits and to successfully outcompete native species, the conditions in a new habitat have to fit the invader's needs. Increased water temperatures, mostly during the summer, in subtropical, and temperate zones, is considered one of the major factors supported the invasion of *C. raciborskii* and other Nostocales (**Figure 2**). (Padisák, 1997; Briand et al., 2004; Rücker et al., 2009). Both *C. raciborskii* and *Aph. ovalisporum* are known to proliferate at temperatures higher than 20˚C (Hadas et al., 1999, 2002; Mehnert et al., 2010), thus moderate elevation in water temperatures may open new ecological habitats for them. The persistence of Nostocales in their new locations is further maintained by their ability to form dormant cells, akinetes, which stay dormant but viable in the sediment for extended periods of time (Hense and Beckmann, 2006;Kaplan-Levy et al., 2010). This seed bank may reside in the sediment and wait for a warm summer to germinate. Consequently, bloom events may follow local climatic patterns. It is important to note that freshly settled akinetes are more likely to germinate during the next summer relative to those settled in earlier blooms as the latter are already covered deeper in the sediments and probably have lower germination potency.

Nostocales species are recently reported to reside in, and dominate eutrophic as well as oligotrophic aquatic systems (**Figure 3**). Eutrophic conditions provide ample nutrients that can be easily and competitively utilized by opportunistic invaders (Piehler et al., 2009). Such eutrophic systems maintain high internal nutrient recycling processes that fortify primary production and biomass accumulation. Although N2 fixation is commonly recorded during Nostocales blooms in eutrophic aquatic systems (mainly indicated by the presence of heterocysts, specialized cells formed under N limiting conditions), this process is responsible for up to 2% of N acquisition by the phytoplankton community (Ferber et al., 2004; Ariosa et al., 2006). The ability of many Nostocales to efficiently use low light, allows their growth in turbid waters as well as at high biomass concentrations.

Regional inappropriate management of watersheds and high pollution impact induce eutrophication, and are certainly involved in the Nostocales geographic expansion. But these invading species can survive and propagate also under oligotrophic conditions (**Figure 3**). This is achieved due to their various efficient phosphate acquisition capabilities (Isvánovics et al., 2000; Spr˝ober et al., 2003; Posselt et al., 2009; Bar-Yosef et al., 2010) and nitrogen fixation which seems to be more relevant under oligotrophic than under eutrophic conditions. Thus, in oligotrophic systems, nitrogen cannot be considered as a limiting factor. Phosphate availability controls the development of the N2 fixing population and its growth. Consequently, management efforts to control

eutrophication by reducing N loads, apparently effective in reducing blooms of *Microcystis* and other Chroococcales species, may actually support the growth of Nostocales which easily outcompete native species in such N limited ecosystems (Schindler et al., 2008).

## **CONSEQUENCES OF NOSTOCALES INVASIONS**

The immediate consequence of Nostocales invasion, besides the ecological impairment, is the expansion of toxic blooms as various strains produce harmful substances. Many *Aph. ovalisporum* and *C. raciborskii* strains produce the cytotoxic compound cylindrospermopsin whereas several Brazilian *C. raciborskii* strains have been reported to produce paralytic shellfish poisoning toxins: neosaxitoxin, saxitoxin, and gonyautoxins (Lagos et al., 1999). Interestingly, many strains isolated from temperate lakes in Europe and Northern America did not produce any known toxin, although they were found toxic in various bioassay experiments (Saker et al., 2004). Recent studies indicated that a given population may be composed of toxic and non-toxic strains (Ballot et al., 2011). The bloom of toxic Nostocales presents an immediate threat to human and animals that use the water for drinking, bathing and recreational activities (Carmichael, 2001). Local and governmental authorities are requested to provide solutions to minimize and control the bloom events by taking scientifically based management measures that should include consideration of the invasiveness of this cyanobacterial group under a scenario of global change (Jones and Sauter, 2005; Schindler et al., 2008).

The domination of recently invaded Nostocales affects the entire ecosystem as native phytoplankton species become rare and displaced and higher trophic levels may be affected via their toxins or due to loss of essential components of their food chain. *C. raciborskii* is known to affect the growth, reproduction and feeding of zooplankton (Leonard and Paerl, 2005) and cylindrospermopsin was shown to be accumulated in fish and in freshwater mussels (Saker and Eaglesham, 1999; Saker et al., 2004).

## **PROSPECTIVE AND PREDICTIONS**

The current trend of increasing bloom events of cyanobacteria, including Nostocales, will probably be further enhanced as global warming continues. Any further temperature increase would promote the growth and development of Nostocales species in general, and that of the invasive species in particular, and would enable further expansion in the temperate zone (Mehnert et al., 2010). This trend was further confirmed by a coupled biological–physical model that predicted that high temperatures favor cyanobacteria over other phytoplankton taxa through increased growth rates that boost the development of blooms (Juehnk et al., 2011). A similar conclusion was reached by Wiedner et al. (2007) who evaluated a case of an earlier rise in water temperature associated with climate change. They suggested that such a scenario will prompt further spread of *C. raciborskii* to the temperate zone as earlier warming permits earlier germination, thereby shifting the pelagic populations to conditions which advance population establishment and growth. However, these models do not consider the ecological variations expected after this earlier germination, as different grazing pressure due to different grazer population (e.g., Winder and Schindler, 2004) or nutrient depletion that may follow an early bloom. These ecological variations are hard to predict and should be further investigated to improve models prediction.

While the possibilities to control and reduce the current trend of global climate change are rather limited, the management of eutrophication processes is feasible. Since a synergistic effect of nutrients and climate was frequently indicated in many sites invaded by Nostocales, it is important that nutrient concentrations in many temperate and subtropical lakes be reduced substantially from present values if cyanobacterial dominance is to be controlled (Kosten et al., 2012). Based on long term experimental manipulation, Schindler et al. (2008) concluded that N-fixing cyanobacteria cannot be limited by a shortage of dissolved N and instead are competitively favored. Thus reducing N inputs could actually intensify the dominance of N-fixing cyanobacteria thus enhancing the expansion of invasive Nostocales. The alternative approach to control and reduce blooms of invasive Nostocales species and their further expansion is to control and reduce external and internal sources of phosphorus, a complex task by itself.

## **REFERENCES**


*raciborskii* (cyanobacteria) invasion at mid-latitudes: selection, wide physiological tolerance, or global warming? *J. Phycol.* 40, 231–238.


## **ACKNOWLEDGMENTS**

This work was partly supported by the German Ministry of Research and Technology (BMBF) and Israel Ministry of Science and Technology (MOST) under contracts FKZ 02WT0985 and WR803. Assaf Sukenik and Ora Hadas acknowledge the continuous support of the Israel Water Authority provided to the Kinneret Limnological Laboratory, IOLR.

(Cyanobacteria) isolates from diverse geographic origins based on nifH and cpcBA-IGS nucleotide sequence analysis. *Appl. Environ. Microbiol.* 68, 2567–2571.


(2011). Bloom development and phytoplankton succession in Lake Winnipeg: a comparison of historical records with recent data. *Aquat. Ecosystems Health Manag.* 14, 219–224.


in aquatic and terrestrial ecosystems. *Ecol. Lett.* 13, 1560–1572.


of its ecology. *Archiv für Hydrobiologie* 107, 563–593.


does not predict its impact. *Biol. Invasions* 9, 309–315.


conditions. *Hydrobiologia* 506–509, 169–174.


(Cyanobacteria) extends its distribution to Latitude 34(53- S: taxonomical and ecological features in Uruguayan eutrophic lakes. *Pan Am. J. Aquat. Sci.* 3, 142–151.


cylindrospermopsin-producing cyanobacterium *Cylindrospermopsis raciborskii* in New Zealand. *N. Z. J. Mar. Freshwater Res.* 37, 821–828.


*Received: 11 January 2012; accepted: 20 February 2012; published online: 09 March 2012.*

*Citation: Sukenik A, Hadas O, Kaplan A and Quesada A (2012) Invasion of Nostocales (cyanobacteria) to subtropical and temperate freshwater lakes – physiological, regional, and global driving forces. Front. Microbio. 3:86. doi: 10.3389/fmicb.2012.00086*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Sukenik, Hadas, Kaplan and Quesada. This is an openaccess article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits non-commercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*

## Plasticity of total and intracellular phosphorus quotas in Microcystis aeruginosa cultures and Lake Erie algal assemblages

## *Matthew A. Saxton1†, Robert J. Arnold1, Richard A. Bourbonniere2, Robert Michael L. McKay <sup>3</sup> and StevenW.Wilhelm1\**

*<sup>1</sup> Department of Microbiology, University of Tennessee, Knoxville, TN, USA*

*<sup>2</sup> National Water Research Institute, Environment Canada, Burlington, ON, USA*

*<sup>3</sup> Department of Biological Sciences, Bowling Green State University, Bowling Green, OH, USA*

#### *Edited by:*

*George S. Bullerjahn, Bowling Green State University, USA*

#### *Reviewed by:*

*Assaf Sukenik, Israel Oceanographic and Limnological Research, Israel Ilana R. Berman-Frank, Bar Ilan University, Israel*

#### *\*Correspondence:*

*Steven W. Wilhelm, Department of Microbiology, The University of Tennessee, M409 Walters Life Sciences, Knoxville, TN 37996-0845, USA.*

*e-mail: wilhelm@utk.edu*

#### *†Present address:*

*Matthew A. Saxton, Keck Environmental Laboratory, Environmental Science and Policy Program, College of William and Mary, 200 Wake Drive, Williamsburg, VA 25185, USA.*

## **INTRODUCTION**

Widespread nutrient loading has led to the eutrophication of aquatic ecosystems throughout the developed world. A primary concern linked to eutrophication is the appearance of harmful algal blooms (HABs). HAB events are well-documented, ecosystem-wide problems that can impact human health, kill livestock, foul potable water supplies, and compromise the integrity of both recreational and commercial fisheries (Carmichael, 2001). HABs have also been implicated in contributing to the formation of hypoxic zones (Paerl et al., 1998).

Efforts to limit algal bloom formation have historically focused on reducing nutrient loading. Phosphorus (P) has been suggested to be the nutrient that limits phytoplankton growth in most freshwater ecosystems (Schindler, 1977; Schindler et al., 2008). Regulating P-loading has led to successes in controlling algal bloom formation in some eutrophic systems (Makarewicz and Bertram, 1991). This developing understanding of eutrophication has been critical to the continued development of mathematical environmental models which have become substantially more sophisticated with advances such as coupling hydrodynamic, ecological, and watershed models (Jørgensen, 2010).

*Microcystis* spp. are among the most common bloom-forming cyanobacteria in freshwaters and have a global distribution

Blooms of the potentially toxic cyanobacterium *Microcystis* are common events globally, and as a result significant resources continue to be dedicated to monitoring and controlling these events. Recent studies have shown that a significant proportion of total cell-associated phosphorus (P) in marine phytoplankton can be surface adsorbed; as a result studies completed to date do not accurately report the P demands of these organisms. In this study we measure the total cell-associated and intracellular P as well as growth rates of two toxic strains of *Microcystis aeruginosa* Kütz grown under a range of P concentrations*.* The results show that the intracellular P pool in *Microcystis* represents a percentage of total cell-associated P (50–90%) similar to what has been reported for actively growing algae in marine systems. Intracellular P concentrations (39–147 fg cell−1) generally increased with increasing P concentrations in the growth medium, but growth rate and the ratio of total cell-associated to intracellular P remained generally stable. Intracellular P quotas and growth rates in cells grown under the different P treatments illustrate the ability of this organism to successfully respond to changes in ambient P loads, and thus have implications for ecosystem scale productivity models employing P concentrations to predict algal bloom events.

**Keywords: cyanobacteria, phosphorus,** *Microcystis*

(Chorus and Bartram, 1999). The effects of P limitation on Puptake in *Microcystis* have been extensively studied. Like other cyanobacteria, *Microcystis* possesses the ability to store excess P intracellularly, as polyphosphate granules, and P-starved cells have higher rates of P-uptake than P-sufficient cells (Jacobson and Halmann, 1982). *Microcystis* has been shown to respond to P limitation by rapid P-uptake with a notable increase in *V* max observed under P-limiting conditions (Kromkamp et al., 1989). The recently sequenced *Microcystis* genomes indicate the presence of ATP driven P-uptake systems similar to what has been described in *Synechococcus* (Ritchie et al., 1997, 2001; Kaneko et al., 2007; Frangeul et al., 2008).

Recently, an oxalate rinse developed to remove surface adsorbed iron (Tovar-Sanchez et al., 2003) was shown to also effectively remove surface adsorbed P from marine plankton (Sãnudo-Wilhelmy et al., 2004; Fu et al., 2005). Subsequently, these studies have shown a significant proportion (15–45%) of total cellassociated P to be surface adsorbed in marine cyanobacteria, and that this P is likely associated with surface adsorbed manganese. These studies have also suggested surface adsorption of P may be part of a two-step process in which surface adsorption is followed by internalization. Arguably, this phenomenon, if universal in nature, is predicted to be of even greater relevance to freshwater ecosystems given their more universal P-limited nature. A consequence of the presence of adsorbed P on the surface of phytoplankton is that previous estimates of cyanobacterial P requirements may overstate the need for this nutrient. Alternatively, the ability of cyanobacteria to store P on their exterior surface may act as a novel strategy to allow persistence during periods of prolonged P deficiency (or to survive "boom-and-bust" cycles of nutrient availability). As such, a necessary first step is to determine the ability of cells to accumulate extracellular P and, further, to determine the contribution of this pool to estimated total P quotas.

In this study we determined the total cell-associated and intracellular P quotas and growth rates of two toxic *Microcystis* strains grown under a series of P concentrations. We similarly examined the ratio of total cell associated to intracellular P of natural communities in the western basin of Lake Erie where *Microcystis* forms an important part of the endemic plankton assemblage. Establishing accurate P quotas (and their plasticity) for this potentially toxic bloom-forming cyanobacterium is a necessary step toward the development of realistic environmental phytoplankton growth models and management strategies for many important freshwater resources.

## **MATERIALS AND METHODS**

#### **CYANOBACTERIAL CULTURE CONDITIONS**

Experiments were performed with *M. aeruginosa* PCC 7806 obtained from the Pasteur Culture Collection. While *M. aeruginosa* PCC 7806 is a globally used type strain, it is not representative of much of the observed *Microcystis* diversity in Lake Erie (Ouellette and Wilhelm, 2003; Ouellette et al., 2006; Ye et al., 2009). *M. aeruginosa* LE3, an isolate from Lake Erie (Brittain et al., 2000), was also included as it has been shown to be representative of Lake Erie *Microcystis* (e.g., Rinta-Kanto and Wilhelm, 2006) and has been used as a standard in numerous studies (e.g., Rinta-Kanto et al., 2009b). Cultures were maintained in BG-11 medium

(Rippka et al., 1979) at pH 7.5 with P concentrations ranging from 1.75 to 175μM K2HPO4. Cultures were grown in 50 mL glass culture tubes (Kimble, Vineland, NJ, USA) at 24˚C under continuous illumination of <sup>∼</sup>30μmol photons m−<sup>2</sup> <sup>s</sup> <sup>−</sup>1. Prior to use, all culture tubes were soaked overnight in 1% HCl and rinsed multiple times with Milli-Q® water (Millipore Corp., Billerica, MA, USA). Samples were grown and transferred at experimental P concentrations two times prior to data collection. Multiple transfers were performed to ensure the stabilization of cellular P stores and the physiological equilibration of cell cultures to experimental P levels *vis a vis* Wilhelm et al. (1996). Cyanobacterial growth was monitored daily using a Turner Designs TD-700 fluorometer (Sunnyvale, CA, USA) equipped with an *in vivo* chl*a* filter set (excitation λ = 340–500 nm; emission λ = >665 nm). Growth rates were determined through the measurement of the slope of the growth curve during the exponential growth phase plotted as the natural log. All experiments were performed in triplicate unless otherwise noted.

## **FIELD SAMPLE COLLECTION**

Field samples were collected at sites across Lake Erie (**Table 1**) between the dates of August 3 and 7, 2010 aboard the *CCGS Limnos*. Surface water was collected at each site using 10 L Niskin sampling bottles and subsequently stored in 3 L opaque amber polycarbonate bottles until filtration (with the time between collection and storage less than 1 h).

#### **PHYTOPLANKTON ENUMERATION**

Phytoplankton enumeration was performed using samples preserved in Lugol's Iodine as previously described (Rodhe et al., 1958; Anderson, 2005). Fifty milliliter samples from each incubation were collected and immediately distributed into amber bottles with predispensed Lugol's Iodine resulting in a final concentration of 2% (vol/vol) fixative. Fixed samples were enumerated using a

**Table 1 | Hydrological and biological parameters of surface water sampled from Lake Erie August 2010.**


*\*Indicates n* = *2. <sup>a</sup> Nitrate* + *nitrite as N. bBDL, below detectable limits (*<*0.005 mgN L*−*1).* Sedgwick-Rafter counting slide (Wildco) on a Micromaster light microscope (ThermoFisher). Samples were allowed to settle in the counting chamber for 3–5 min prior to enumeration. Four horizontal passes were made across each slide encompassing the full height and width of the slide; no fewer than 40 fields of view were randomly selected across each slide. Each field of view was imaged and saved using Micron imaging software (Westover Scientific); the use of this software allowed for the measurement of the field of view.

## **QUANTIFICATION OF TOTAL CELLULAR AND INTRACELLULAR P**

Total cellular and intracellular P concentrations were independently measured. Surface bound P was removed using an oxalate wash reagent initially designed for removal of iron (Tovar-Sanchez et al., 2003) and subsequently demonstrated to be applicable for removal of surface-sorbed P (Sãnudo-Wilhelmy et al., 2004). Aliquots of 5 mL from individual cultures and 50–200 mL from field samples were harvested onto pre-combusted (450˚C for 4 h) 25 mm Whatman (Kent, UK) GF/F glass fiber filters for the determination of total P in samples. The range in sample volumes collected at the individual sites was to account for different phytoplankton concentrations at each station. For the determination of intracellular P from *Microcystis* isolates, aliquots were first mixed with an equal volume of oxalate wash and allowed to incubate for 5 min prior to harvest. Field samples were filtered until approximately 5 mL of sample water remained in the filter unit; 5 mL of oxalate rinse was then applied and incubated for 5 min. A top-up solution containing equal parts oxalate rinse and filtered lakewater was applied as needed to counter losses due to residual vacuum. Following this incubation, the filter was rinsed three times with 5 mL of P-free growth media or filtered lake water. To control for the removal of surface associated P during the washing process, control samples were rinsed with P-free medium or filtered lake water four times in the same manner as experimental samples. Unused growth medium of each of the experimental P concentrations was passed through filters and rinsed with P-free media and oxalate rinse solution as blanks to control for media associated P binding to the filter. Filtered station water was used as blanks for field samples. All treatments were performed in triplicate and blanks were performed in duplicate. Following the oxalate rinse procedure, filters were stored at −20˚C until analysis.

## **P MEASUREMENT**

Particulate organic P was collected on filters and converted to inorganic P *via* persulfate oxidation (Menzel and Corwin, 1965;Wetzel and Likens, 2000). Freshly made 5% potassium persulfate solution was added to samples that were then placed in a boiling water bath for 1 h. Total P concentrations were measured spectrophotometrically (885 nm) using the ammonium molybdate method (Wetzel and Likens, 2000). All measurements for field samples were performed on a Thermo Spectronic Genesys 20 spectrophotometer (ThermoFisher Waltham, MA, USA), whereas cultured samples were processed using a Biomate 5 spectrophotometer (ThermoFisher Waltham, MA, USA).

Field samples were prepared by first digesting with 5% potassium persulfate, followed by autoclaving for 30 min. Total P was spectrophotometrically analyzed (880 nm) with the molybdate ascorbic acid method (Strickland and Parsons, 1972) using a 10-cm quartz cell and a Cary 50 UV–vis spectrophotometer (Varian Palo Alto, CA, USA). A calibration was conducted using phosphate reference standards of known concentration processed in the same manner as sample material. Individual replicates were omitted from field sample analysis if one sample was more than 100% different than the average of the two other samples. Variability among replicates was likely the result of the incomplete rinsing of cell aggregates.

Subsamples of cultured samples (1.5 mL) were also collected from each tube for direct counts of cell concentration. *Microcystis* cell densities were determined by epifluorescence microscopy with a Leica (Wetzlar, Germany) DMXRA epifluorescence microscope using the Texas Red filter set targeting accessory pigment (phycobillin) autofluorescence as previously described (Wilhelm et al., 2006). This was performed to normalize P measurements to *Microcystis* cell abundance.

## **WATER CHEMISTRY**

Samples for nutrients, including total dissolved phosphorus (TDP), soluble reactive phosphorus (SRP), total dissolved nitrogen (TDN), NH+ <sup>4</sup> - N, NO<sup>−</sup> <sup>3</sup> - N + NO<sup>−</sup> <sup>2</sup> − *N*, silicate (SiO2), sulfate (SO2<sup>−</sup> <sup>4</sup> ), particulate organic nitrogen (PON), and particulate organic carbon (POC), were collected and stored at 4˚C until analysis. Analysis was conducted at the National Laboratory for Environmental Testing (Environment Canada) using their standard techniques (NLET, 1994).

## **EXTRACTED CHLOROPHYLL** *a*

Chlorophyll *a* (chl*a*) concentration at field stations was determined as a proxy for total phytoplankton biomass. Seston was filtered through 0.2-μm nominal pore-size polycarbonate membranes. Chl*a* concentrations were determined using the nonacidification method (Welschmeyer, 1994) following extraction in 90% acetone overnight at −20˚C using a Turner Designs 10-AU fluorometer.

## **RESULTS**

## **P-CONSTRAINTS ON MICROCYSTIS GROWTH**

Cultured *Microcystis* growth rates across a range of P concentrations are shown in **Table 2**. Growth rates in both strains were only slightly variable across the tested experimental P concentrations. *M. aeruginosa* PCC 7806 growth rates ranged between 0.079 and 0.162 divisions day−<sup>1</sup> with *M. aeruginosa* LE3 growth rates ranging from 0.058 to 0.142 divisions day−1. Growth rates increased with a slope of 0.0002 in *M. aeruginosa* PCC 7806 while *M. aeruginosa* LE3 rates decreased with a slope of −0.0001. Growth rates were especially steady among those cultures maintained at higher P concentrations. Growth rates were more variable in *M. aeruginosa* LE3 than in *M. aeruginosa* PCC 7806. Although growth rates were stable across concentrations of P, biomass was notably lower in cultures maintained under the lower P concentrations (results not shown). It should also be noted that *Microcystis* PCC 7806 did not grow consistently at the lowest P concentrations in spite of repeated attempts. We recognize that the reported growth rates are lower than published values. These lower growth rates are the result of the cultures being grown in conditions not optimized to

**Table 2 |** *Microcystis aeruginosa* **strains PCC 7806 and LE3 growth rates.**


*Growth rates measured in triplicates except where \*n* = *1 as replicates would not grow at these concentrations.*

obtain maximum growth rates, but rather they best reflect natural conditions (lower than optimal temperature and light). It is also possible cultures may have became CO− <sup>2</sup> limited as a result of small culture volume.

## **TOTAL CELL-ASSOCIATED AND INTRACELLULAR P QUOTAS IN CULTURED** *MICROCYSTIS*

To obtain accurate intracellular P quotas for *Microcystis aeruginosa*, we utilized the oxalate rinse technique to differentiate between the total cell-associated and intracellular P pools (**Figure 1**). The range of total cell-associated P concentrations reported here (39–173 fg cell−1) are within the range of previously reported values (29–433 fg cell−1; Rhee and Gotham, 1980; Sbiyyaa et al., 1998; Tsukada et al., 2006; Shen and Song, 2007). As was anticipated, intracellular P quotas (34–146 fg cell−1) were significantly smaller (ANOVA *p* = 0.001) than the cell-associated values (39–173 fg cell−1). As had been previously reported for marine cyanobacteria (Fu et al., 2005), our results show that surface adsorbed P represents a significant portion (between 10 and 50%) of total cell-associated P in *Microcystis*. We also found that this percentage is stable across growth conditions in each of the strains tested (**Figure 1C**).

## **TOTAL CELL-ASSOCIATED AND INTRACELLULAR P IN NATURAL SAMPLES**

The ratio of total cell-associated P to intracellular P was also investigated in natural phytoplankton communities. Samples were collected across Lake Erie with specific emphasis on the Western Basin. This sampling focus reflects the recurrence of *Microcystis* blooms in this region (Rinta-Kanto et al., 2005, 2009a; Moorhead et al., 2008)*. Microcystis* spp. was present at all tested stations, though a diverse assemblage of cyanobacteria and eukaryotic algae were observed (**Table A1** in Appendix). Hydrological and biological parameters measured as part of this study were similar to those previously reported across Lake Erie (**Table 2**; DeBruyn et al., 2004). Intracellular P made up an average of 89% (±11%) with a range of 0–65% of total particulate P in these samples (**Table 1**).

as mean ± SD.

Pearson correlation analysis of Western Basin samples showed weakly significant correlations (*p* < 0.1) between the proportion of surface associated extracellular P and several environmental variables, including SRP, POC, PON, and SiO2 (**Table 3**). Notably, no significant correlation was observed between the proportion of extracellular P and either TP or TDP (**Table 3**; **Figure 2**).

**Table 3 | Pearson correlation analysis results comparing the extracellular percentage of total particulate P to hydrological and biological parameters.**


*a Nitrate* + *nitrite as N.*

#### **DISCUSSION**

A general tenet of limnology is that P is the nutrient limiting phytoplankton biomass in many freshwater systems and that, by reducing P loading to eutrophic bodies of water, we can begin to control formation of algal blooms and biomass accumulation (Schindler et al., 2008). While the previous statement is a commonly held assumption (although not always true, e.g., Scott and McCarthy, 2010; Wilhelm et al., 2011), it is important to note that individual HAB organisms have specific nutrient needs and differ in their abilities to assimilate nutrients from different chemical species. For example, it is now well established that some freshwater cyanobacteria possess the capacity to assimilate P from complex organic sources, such as phosphonates (Ilikchyan et al., 2009; Saxton et al., 2011). Moreover, the addition of organic sources of nutrient have been shown to influence phytoplankton community structure in bioassays performed using natural assemblages (Wang et al., 2010). To this end, we examined the P requirements of one of the most prominent bloom-forming cyanobacteria, *M. aeruginosa*, so as to develop a dynamic cellular quota that can be used to inform credible data reporting. To do so, we applied recently developed techniques that allow for the individual analysis of total cell-associated vs*.* intracellular P pools in phytoplankton.

Our results show plasticity in both cellular and surface absorbed P quotas. The ability to adjust intracellular P levels while maintaining a stable growth rate across a range of P concentrations is important for bloom-forming cyanobacteria as it allows for persistence in conditions that do not allow for the accumulation of significant biomass. This nutritional flexibility is important for microbes in environments that experience large nutrient variability, such as large lakes and estuarine systems. The plasticity of cellular P quotas is in contrast to the stability of the total cellassociated P to intracellular P ratio. Our data show this ratio to be stable, demonstrating an ability of *Microcystis* to adjust its P needs. These observations imply that surface associated P is a function of the external concentration in the environment. One potential route for P adjustment has been described in marine cyanobacteria that substitute phospholipids for nitrogen and sulfur containing

lipids in response to P limitation (Van Mooy et al., 2009), though this ability has not yet been described in *Microcystis.* Observed variability in P quota values in the tested strains is likely attributable to variance between strains. This variance is also observed in genetic differences between the sequenced *Microcystis* strains (Kaneko et al., 2007; Frangeul et al., 2008).

In contrast to our observations in cultured *Microcystis,*the ratio of intracellular to extracellular P was not constant between Lake Erie stations.

Whereas this variability is correlated to differences in ambient SRP concentrations, dissimilarity in plankton communities between these sites may also contribute. The high proportion of intracellular P as compared to the total P pool in Lake Erie is indicative of an actively growing phytoplankton community (Sãnudo-Wilhelmy et al., 2004). Another factor influencing the intracellular-total P ratio in natural *Microcystis* populations may be colony morphology. If exposure to dissolved phosphate is a driver of the presence of extracellular P, as is suggested by the positive correlation between SRP and the percentage of extracellular P, then the cellular surfaces not exposed to the environment will not have externally bound P.

Unfortunately the fate of stored P on the cell surface remains elusive; it is possible that when cells are shifted from P-replete to P-limiting conditions, this surface associated material can be assimilated by cells as opposed to the alternative of this P being lost back to the environment as has been previously hypothesized (Sãnudo-Wilhelmy et al., 2004). While the hypothesis that surface adsorbed P is available to phytoplankton is intriguing, we observed no evidence to suggest that it is the case in *Microcystis.* Were this true we would expect the ratio of surface-intracellular P to shift toward the intracellular fraction under conditions of P stress as the cells scavenged P from the surface. However, we observed this ratio to remain constant across the growth conditions tested in this study. This observation paired with the positive correlation between SRP and surface adsorbed P in our Lake Erie samples suggests that the presence of P on the surface of these cells is a by-product of living in an environment where dissolved P is present.

Models of eutrophication and phytoplankton growth remain a critical tool for systems managers attempting to control frequent bloom formation and have been applied to Lake Erie since the middle of the twentieth century (Di Toro et al., 1975; Léon et al., 2005; Zhang et al., 2008; Schwab et al., 2009). *Microcystis* growth has been specifically targeted in many models because it is a problematic, bloom-forming cyanobacterium. Physiological parameters, such as cellular nutrient quotas, nutrient uptake rates and cellular growth rates, are integral to generating accurate predictions of cellular growth and toxin production under specific environmental conditions. Cellular P quota is a variable used (or is integral to variables used) in environmental models that have been applied to both *Microcystis* and Lake Erie (Robson and Hamilton, 2004; Léon et al., 2011; Burger et al., 2008; Trolle et al., 2008). This use of P quota as a modeling variable coupled with the fact that

## **REFERENCES**


the relative importance of internal and external nutrient loads on water column nutrient concentrations and phytoplankton biomass in a shallow polymictic lake. *Ecol. Modell.* 211, 411–423.


literature P quotas in *Microcystis* do not take into account the presence of externally bound P hinders the ability of these models to accurately project the occurrence of blooms and underestimates the algal biomass that can be supported by current P inputs. For example, the minimum internal *p* value applied to the CAEDYM model by Robson and Hamilton (2004) of 0.4 mg P (mg chl*a*)−<sup>1</sup> represents approximately a 40× overestimation of the P needs of this organism compared to the values reported here. The accurate prediction of natural conditions by environmental models is dependent on the use of precisely measured variables, so it is critical that the modeling community use the most appropriate values. Moreover, the significant metabolic plasticity with respect to P quotas demonstrated by *Microcystis* in the present study is also not taken into account in current models. The current mathematical models use one or two values, often the minimum and maximum, to describe biological parameters such as growth rate or internal P quota (to limit the complexity of already complex models). *Microcystis*, however, has the ability to adapt to environmental challenge. As such, the ability of new hydrodynamic models, such as ELCOM, to more effectively describe changing physical conditions and nutrient concentrations in the water column suggest biological models would benefit from incorporating a more complex view of cyanobacterial growth over nutrient and temperature gradients.

The importance of controlling the growth of HAB cyanobacteria cannot be understated because of the water quality effects in both undeveloped and industrial countries. Previous studies examining P quotas in *Microcystis* measured only the total cell-associated P pool. In this study we differentiate between the total cell-associated P pools, including surface bound P, and the intracellular P pool; thus the P quotas reported here are a more representative examination of the P needs of *Microcystis* than has been previously reported. These results will be useful to resource managers attempting to model and control this nuisance alga.

## **ACKNOWLEDGMENTS**

We thank Michael Schultz for processing particulate organic phosphorus samples and Jennifer DeBruyn for aid with statistical analyses. We would also like to thank the captain and crew of the *CCGS Limnos*. This research was supported by NOAA Oceans and Human Health external grant NA05NO54781251, Ohio Sea Grant College Program award R/ER-73, and NSF-IOS #0841918 and NSF–CBET#0826838).


Hutchins, D. A. (2005). The biological and biogeochemical consequences of phosphate scavenging onto phytoplankton cell surfaces. *Limnol. Oceanogr.* 50, 1459–1472.


uptake in the cyanobacterium *Synechococcus* R-2 PCC 7942. *Plant Cell Physiol.* 38, 1232–1241.


The relationship between nutrients, cyanobacterial toxin and the microbial community in Lake Tai (Taihu), China. *Harmful Algae* 10, 207–215.


cyanobacteria in Chinas third largest lake, Lake Taihu. *Harmful Algae* 8, 637–644.

Zhang, H., Culver, D. A., and Boegman, L. (2008). A two-dimensional ecological model of Lake Erie: Application to estimate dreissenid impacts on large lake plankton populations. *Ecol. Modell.* 214, 219–241.

**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 October 2011; paper pending published: 23 November 2011; accepted: 03 January 2012; published online: 18 January 2012.*

*Citation: Saxton MA, Arnold RJ, Bourbonniere RA, McKay RML and Wilhelm SW (2012) Plasticity of total and intracellular phosphorus quotas in Microcystis aeruginosa cultures and Lake Erie algal* *assemblages. Front. Microbio. 3:3. doi: 10.3389/fmicb.2012.00003*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Saxton, Arnold, Bourbonniere, McKay andWilhelm. This is an open-access article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits non-commercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*

## **APPENDIX**

**Table A1 | Abundance of selected phytoplankton genus determine** *via* **microscopic examination of samples collected from Lake Erie in August 2010.**


## Effect of nitrogen on cellular production and release of the neurotoxin anatoxin-a in a nitrogen-fixing cyanobacterium

## *Alexis Gagnon\* and Frances R. Pick*

*Center for Advanced Research in Environmental Genomics, University of Ottawa, Ottawa, ON, Canada*

#### *Edited by:*

*George S. Bullerjahn, Bowling Green State University, USA*

#### *Reviewed by:*

*Xosé Anxelu G. Morán, Instituto Español de Oceanografía, Spain Shawn R. Campagna, University of Tennessee Knoxville, USA*

#### *\*Correspondence:*

*Alexis Gagnon, Center for Advanced Research in Environmental Genomics, University of Ottawa, Ottawa, ON, Canada K1N 6N5. e-mail: agagn042@gmail.com*

Anatoxin-a (ANTX) is a neurotoxin produced by several freshwater cyanobacteria and implicated in lethal poisonings of domesticated animals and wildlife. The factors leading to its production in nature and in culture are not well understood. Resource availability may influence its cellular production as suggested by the carbon-nutrient hypothesis, which links the amount of secondary metabolites produced by plants or microbes to the relative abundance of nutrients. We tested the effects of nitrogen supply (as 1, 5, and 100% N of standard cyanobacterial medium corresponding to 15, 75, and 1500 mg L−<sup>1</sup> of NaNO3 respectively) on ANTX production and release in a toxic strain of the planktonic cyanobacterium *Aphanizomenon issatschenkoi* (Nostocales). We hypothesized that nitrogen deficiency might constrain the production of ANTX. However, the total concentration and more significantly the cellular content of anatoxin-a peaked (max. 146μg/L and 1683μg g−<sup>1</sup> dry weight) at intermediate levels of nitrogen supply when N-deficiency was evident based on phycocyanin to chlorophyll *a* and carbon to nitrogen ratios. The results suggest that the cellular production of anatoxin-a may be stimulated by moderate nitrogen stress. Maximal cellular contents of other cyanotoxins have recently been reported under severe stress conditions in another Nostocales species.

**Keywords: cyanobacteria, anatoxin-a, nitrogen deficiency, carbon-nutrient hypothesis**

## **INTRODUCTION**

Many cyanobacteria are capable of producing compounds that are toxic to other organisms. The cyanotoxins have been classified into four main groups according to their mode of action. In addition to the hepatoxins, cytotoxins, and gastrointestinal toxins, there are several types of neurotoxins (Codd et al., 2005). A potent neurotoxin synthesized by freshwater cyanobacteria is anatoxin-a (ANTX), likely one of the smallest toxic alkaloids described to date (LD50 of 0.2 mg/kg IP in mice, Carmichael et al., 1979). Prior to its chemical identification, ANTX was labeled the Very Fast Death Factor due to its ability to induce death within 4 min following intraperitonial injection in mice. ANTX has now been detected in lakes and rivers throughout the world (e.g., Codd et al., 2005; Cadel-Six et al., 2007; Wood et al., 2007; Aráoz et al., 2008), and, in some cases, at concentrations leading to lethal poisonings of livestock, domestic animals, and wildlife.

The factors regulating ANTX production, whether in the field or in the laboratory, are not well understood (Osswald et al., 2007). The toxin has been studied less that the more commonly encountered microcystins associated with freshwater cyanobacterial blooms (a Web of Science search since 1995 yields about a quarter of the number of papers on ANTX relative to the number of papers on microcystins). ANTX production is strain specific with cellular contents ranging up to 13,000μg/g dry weight in *Anabaena* (reviewed by Osswald et al., 2007). The optimal temperature for production appears to lie between 19.8 and 22.0˚C and is maximal during the exponential phase of growth in batch cultures.

Several hypotheses have been proposed to explain why toxins more generally are produced by cyanobacteria. They may be produced as a defense mechanism against grazers or other organisms in order to gain ecological advantage. Or they may be involved in cell signaling processes (Wiegand and Pflugmacher, 2005). Cyanotoxins including ANTX have been labeled as secondary metabolites as they do not appear to be involved in primary metabolism (Carmichael, 1992). In this regard, an increase in demand for resources at the cellular level would in theory lead to a decrease in ANTX production.

The availability of nutrients may play a role in the production of secondary metabolites, including nitrogen containing alkaloids such as ANTX, as suggested by the carbon-nutrient balance hypothesis (Hamilton et al., 2001). This hypothesis links the amount of secondary metabolites produced by plants to the relative abundance of nutrients. Recently, the production and composition of microcystins (MC) was investigated in relation to nitrogen to carbon supply ratios (N:C) in the freshwater cyanobacterium *Microcystis aeruginosa* (van de Waal et al., 2009). Consistent with the carbon-nutrient balance hypothesis, a high N:C ratio in the external media and within the cells led to higher concentrations of microcystins (and of the nitrogen-rich MC variant MC-RR) in comparison to lower N:C ratios. Despite the fact that less than 1% of the total cellular nitrogen was invested in microcystins, the availability of inorganic nitrogen appeared to influence secondary metabolite concentrations and composition. Nitrogen availability may also play an analogous role in the production of other nitrogen containing secondary metabolites such as ANTX.

In the following study, the relationship between nitrogen availability and ANTX production was investigated in *Aphanizomenon issatschenkoi* (Ussaczew) Proschkina-Lavrenko using an ANTX producing strain isolated from a hypereutrophic New Zealand lake (CAWBG02,Wood et al., 2007).*A. issatschenkoi* is a colonial planktonic species, which can produce heterocysts (Ballot et al., 2010), the primary site for atmospheric nitrogen (N2) fixation in the Nostocales in the absence of available inorganic nitrogen. The process of nitrogen fixation being energy demanding, this study hypothesized that nitrogen fixation might lead to further reductions in the production of secondary metabolites containing nitrogen when the supply of inorganic N was low. Both intracellular and extracellular production was estimated: given its small molecular weight ANTX can be found in significant but variable amounts in solution (Rapala et al., 1993).

## **MATERIALS AND METHODS**

## **CULTURE CONDITIONS AND SAMPLING**

A culture of *A. issatschenkoi* was obtained from the Cawthron Institute, New Zealand (strain CAWBG02). Batch cultures were grown at a light intensity of 85μE m−<sup>2</sup> s <sup>−</sup><sup>1</sup> on a 12:12 light: dark cycle at 20 ± 1˚C in a Conviron growth chamber (E-15) in 500 mL Erlenmeyer flasks. Three different growth media were used with sodium nitrate (NaNO3) at 100, 5, and 1% of full strength BG11 (Andersen, 2005) corresponding to 1500, 75, and 15 mg L−<sup>1</sup> of NaNO3 respectively. Total culture volumes were 250 mL and inocula for all the experimental cultures came from the same parent culture in exponential phase acclimated to the same temperature and light regime. Under these conditions, *A. issatschenkoi* grew as single trichomes and minimal clumping occurred which facilitated subsampling. Two milliliters from each flask were sub-sampled every 3–4 days to measure optical density using a Pye-Unicam SP-100 UV-spectrophotometer at 750 nm. The optical density at 750 nm provides a measure of turbidity of the culture and is a function of both cell density and cell size. Growth rates were obtained by plotting the natural logarithm of absorbance readings at 750 nm during the exponential growth and using linear analysis (Guillard, 1973). Optical densities at 627 and 438 nm were also measured as an estimate of phycocyanin to chlorophyll *a* changes (van de Waal et al., 2009). Phycocyanin is a protein and accessory pigment found in all cyanobacteria and involved in light harvesting (Whitton and Potts, 2000).

For toxin extraction from cells, 10 mL from each flask was filtered on days 10, 20, 30, and 40 of growth using pre-ashed and pre-weighed Whatman GF/C filters. The first three time points corresponded to early, mid, and late exponential phases whereas the last corresponded to stationary phase at least in the 1 and 5% N cultures. A subsequent 10 mL was filtered from each flask using the same filter type for the analysis of particulate organic carbon (C) and nitrogen (N). Filters were oven dried overnight at 60˚C, re-weighed to obtain the dry weight (mg mL−1) and stored frozen at −20˚C. During filtrations, one 20 mL of filtrate was collected from each culture in solvent rinsed (acetone and hexane) scintillation vials wrapped in foil. Filtrates were spiked with 50 μL formic acid to stabilize ANTX and frozen immediately at −20˚C. For chlorophyll *a* analysis, 5 mL were sub-sampled and filtered using 934-AH glass microfiber filters with approximately the same

pore size as GF/C. Filters were inserted in clean 15 mL plastic test tubes and stored at −20˚C for subsequent analysis. The experiment was terminated when the 1 and 5% N cultures had clearly reached stationary phase.

## **EXTRACTION AND ANALYSIS OF ANATOXIN-A, PARTICULATE ORGANIC CARBON AND NITROGEN, AND CHLOROPHYLL** *a*

The extraction of ANTX from cyanobacterial cells was performed using an ASE 200 device Dionex Corporation (Bannockburn, IL, USA). Frozen GF-C filters with cyanobacterial biomass were thawed and inserted in stainless-steel ASE cells (11 mL) packed with pre-washed Hydromatrix. The ASE parameters were optimized for anatoxin-a specifically (**Table 1**) based on a previous ASE protocol optimized for microcystins (Aranda-Rodriguez et al., 2005). Extracts were collected in solvent (acetone and hexane) rinsed amber vials and then spiked with 50μL formic acid (pH <sup>∼</sup>3). Extracts were then evaporated down to dryness at 59˚C under a gentle nitrogen flow using a Zymark Turbovap II and then resuspended in 1 mL of 50% methanol in water. Filtrates were freeze-dried and then resuspended in 1 mL of 50% methanol in water. All extracts, both particulate and liquid phase, were filtered with pre-conditioned 0.2μm Acrodisc filters and stored at −20˚C until analysis.

Toxin analyses were performed on a Sciex QTRAP 3200 LC-MS/MS (ABSciex, Toronto, Canada). The system consisted of a 1200 series Agilent liquid chromatograph with a high performance autosampler (model G1376B), a binary pump (model G1312A), a column thermostat (model G1316A), and a triple quadrupole linear ion trap mass spectrometer equipped with a turbospray ion source, specifically electrospray ionization (ESI) in positive mode with a voltage of 4500 eV and setpoint temperature of 350˚C. The acquisition of data was performed with Analyst software (version 1.4.1). Chromatographic separations were achieved with a Zobrax SB-C18 Rapid Resolution column (50 × 2.1 mm I.D., 1.8μm particle size column) and a guard column (12.5 × 2.1 mm I.D, 5μm; Agilent Technologies, Canada) at 40˚C. The optimal mobile phase conditions were: water and MeOH with 0.1% formic acid and ammonium formate 20 mM with a constant flow rate of 0.3 mL/min. Injection volume for samples was set to 1 μL. The needle was washed with 50% methanol, 50% acetonitrile, and 0.01% formic acid at the flush port (3×) after each injection to minimize carry over. The method successfully separated anatoxin-a from phenylalanine which often interferes with anatoxin analyses



as the compounds are isobaric (Furey et al., 2005). ANTX and PHE standards were individually infused in the MS with a 4.6 mm I.D Harvard syringe at a flow rate of 10μL/min. Multiple reaction monitoring (MRM) scans were performed for each standard by selecting the protonated molecular ion [M <sup>+</sup> H]+<sup>1</sup> (parent ion) with the first mass filter quadrupole (Q1). The product ions, yielded by collision-induced dissociation, were scanned by the third quadrupole (Q3). Unambiguous Q1/Q3 pair transitions were obtained by manually optimizing the declustering potential to get characteristic product ions of ANTX and PHE (**Table 2**). Standard curves were produced with anatoxin-a fumarate obtained from Tocris Bioscience (USA) and analyzed in line with unknowns. All samples were analyzed on the same run to minimize any day to day instrument variability. The instrument limit of detection was 1 pg of ANTX and the limit of quantification 3.5 pg on column.

Particulate organic C and N were measured using a Costech elemental analyzer (ECS 4010; Costech,Valencia,CA,USA) following Frost et al. (2009). Chlorophyll *a* was extracted by incubating the frozen 934-AH filters with culture material in 13 mL of 95% ethanol at 4˚C for about 24 h (Jespersen and Christoffersen, 1987). Following centrifugation of the extracts, chlorophyll *a* concentrations were then estimated based on optical densities at 750, 665, and 649 nm (Bergmann and Peters, 1980).

#### **STATISTICAL ANALYSES**

In order to compare the effects of N treatments, one-way analyses of variance (ANOVA) were performed on the following end-points: total and cellular toxin concentrations, growth rates, chlorophyll *a*, phycocyanin to chlorophyll *a* ratios, and particulate organic carbon to nitrogen (C:N) ratios at time intervals corresponding to the early to late exponential phase as well as the early stationary phase. A one-way ANOVA was chosen over a repeated measures ANOVA in order to focus analyses on specific and predictable growth phases. In the case of significant treatment effects (*p* < 0.05), Tukey's pairwise comparison test was subsequently used to determine differences between treatments. All statistical analyses were carried out using SigmaStat (version 3.1) software.

## **RESULTS**

#### **GROWTH AND NITROGEN STRESS**

All cultures entered the exponential phase of growth after about 18 days and reached the stationary phase after approximately 34 days for the 1% N treatment and 40 days for both 5 and 100% N treatments (**Figure 1**). The growth curves based on optical density were clearly different between the three nitrogen treatments (**Figure 1**). By day 30, the mean optical density of the 100% N treatment was significantly higher (*p* < 0.001) than that at the lower N concentrations. On days 34 and 40, the treatments were all significantly different from one another, with the lowest optical densities corresponding to the 1% N treatment. The average biomass as dry weight (**Table 3**) of both the 1 and 5% N treatments was also significantly lower than that of 100% N treatment on all four sampling dates (10, 20, 30, and 40). Population growth rates ranged from 0.03 to 0.07 day−<sup>1</sup> and were significantly higher in the 100% N cultures in comparison to the 1% N (**Table 3**).

Chlorophyll *a* concentrations of the cultures showed an analogous pattern to absorbance (data not presented). By day 20 and on **Table 2 | MRM transitions of each analyte obtained after infusion into turboV electrospray source of 3200 QTRAP and their respective voltages.**


three subsequent sampling dates all three treatments were significantly different from one another (*p* < 0.001). The highest chlorophyll *a* concentrations were measured in the 100% N treatment, followed by the 5 and 1% N treatments: maximum values were attained on day 30 (584 <sup>±</sup> 26, 382 <sup>±</sup> 8, and 192 <sup>±</sup> <sup>16</sup>μg L−<sup>1</sup> respectively). In contrast to the optical density data (**Figure 1**), by day 40 chlorophyll *a* concentrations began to decline across all treatments.

With respect to the ratio of phycocyanin to chlorophyll *a*, the 100% N batch cultures had clearly the highest mean ratio and the 1% N the lowest (**Figure 2**). Significant differences between either all treatments or the 1 and 100% N treatments were observed from day 7 to day 34. On day 40, the mean ratio for the 1% N was significantly different (*p* < 0.001) from both higher N treatments. With respect to cellular nutrient composition, the particulate organic carbon to nitrogen ratios (**Table 3**) began to differ from day 20 onward. Batch cultures grown in 100% N medium were significantly different (*p* < 0.001) from the other two treatments with a lower mean ratio on both day 20 and 30. At day 40 however, no difference between the C:N mean ratios of batch cultures grown in 100 and 1% N was detected using Tukey's all pair wise test. This was due to the high standard error between triplicates of the


**Table 3 | Average growth rate based on optical density changes (**± **standard error), carbon to nitrogen (C:N) molar ratios and yield as dry weight biomass (mg/mL) for** *Aphanizomenon issatschenkoi* **grown in different nitrogen concentrations (shown as % of full strength BG11 media).**

*Within the same column (parameter) values with different letters are significantly different from each other as determined by a one-way ANOVA and Tukey's post hoc pairwise comparisons (n* = *3, p* < *0.05). The different letters signify significant differences between the means in each column at p<0.05.*

1% N treatment. However, a Duncan all pair wise test showed a significant difference (*p* = 0.045) between the 100% N treatment and the other two treatments.

#### **ANATOXIN-a**

The total ANTX concentrations ranged from 4 to 146μg L−<sup>1</sup> across time and treatments with the highest concentrations observed during the exponential phase (day 30) for all three nitrogen concentrations (**Figure 3**). On day 10, the highest total ANTX mean concentration corresponded to the 5% N treatment and the lowest to the 1% N. By day 30, the 5% N cultures had the highest total ANTX concentrations with a mean concentration of <sup>111</sup> <sup>±</sup> <sup>18</sup>μg L−1; the 100% N cultures had the lowest concentrations ranging from 51 to 74μg L−1. However, with the exception of day 10, the total ANTX culture concentrations were not statistically different between the three treatments on a given day, with the exception of day 10, despite highly significant differences in culture biomass.

The ANTX cellular content ranged from 6 to 1683μg g−<sup>1</sup> dry weight with the highest values in each treatment corresponding to 30 days of growth or the late exponential phase (**Figure 4**). ANTX content was significantly different among treatments on all dates, with higher content for the 5% N treatment compared to the other two treatments on day 20 (*p* = 0.018) and 30 (*p* = 0.003). The highest ANTX cellular content was measured in the 5% N treatment on day 30 with a mean of 1408 <sup>±</sup> <sup>181</sup>μg g−1. The 5 and 1% N treatment had higher ANTX cell content than the batch cultures grown in 100% N with the 5% N treatment resulting in the highest amounts except on day 40, which corresponded to the onset of the stationary phase.

The extracellular ANTX concentrations ranged from below detection (<0.1μg L−1) to 13μg L−<sup>1</sup> and were considerably lower than particulate concentrations (**Figure 3**) while representing between 3 and 47% of the total production. Cultures grown in 100% N medium had significantly (*p* < 0.001) lower ANTX extracellular concentrations and lower percent extracellular release than the other two treatments on days 20 and 30. The highest mean extracellular concentrations were found in the 1% N treatments on days 30 and 40, reaching 9 <sup>±</sup> 0.12μg L−1. Two of the 1% cultures also had higher concentrations of ANTX in the extracellular phase compared to the intracellular phase one on day 10 and one on day 20.

## **DISCUSSION**

The total amount of ANTX produced by*A. issatchenkoi* CAWBG02 was lower in this study than that reported for this strain when first isolated. For a 23 day culture Selwood et al. (2007) reported a maximum concentration of 1105μg L−1, whereas a maximum of 146μg L−1was estimated here. This could stem in part from differences in growing conditions related to differences in the composition of media, temperature, and light as well as aeration since the present cultures were not bubble aerated which Selwood et al. (2007) mentioned could double the toxin production. Furthermore differences in both extraction and analytical procedures for ANTX could also lead to differences: there are

at present no common standard methods for most algal toxins or certified reference materials in use. The cellular content reported is within the range of that reported for a recently isolated strain of *A. issatschenkoi* from northeastern Germany (Ballot et al., 2010). Collectively the species appears less toxic than the

are standard errors of mean (*n* = 3).

benthic cyanobacterium *Phormidium favosum* which produces 8000μg g−<sup>1</sup> per dry weight (Gugger et al., 2005) or previous reports for*Anabaena* strains (summarized in Osswald et al., 2007), although the problem of analytical interference from phenylalanine may have inflated these earlier estimates of ANTX which did not clearly separate these isobaric compounds.

In this study, nitrogen limited the biomass yield of *A. issatschenkoi* in the low N cultures as evidenced by lower dry weights and overall final yields under the 1 and 5% relative to 100% N (**Table 3**). The growth rate was also reduced in the 1% cultures and, to a lesser extent, in 5% cultures. Nitrogen deficiency was also reflected in cellular constituents: in particular lower chlorophyll *a* concentrations and overall lower ratios of phycocyanin to chlorophyll *a* (**Figure 2**). The phycobiliproteins, being nitrogen-rich accessory pigments, are typically reduced relative to chlorophyll *a* under nitrogen stress in cyanobacteria (Turpin, 1991).

When soluble inorganic nitrogen becomes limiting in the environment, many cyanobacteria belonging to the *Nostocales*, such as *Aphanizomenon*, are capable of fixing atmospheric nitrogen to compensate for their needs for that specific nutrient. Hence, the particulate organic nitrogen content should have been more or less constant throughout the 40 days in all three treatments if N2 fixation had occurred to compensate for nitrogen limitation. On days 20 and 30, the C:N ratios were significantly higher for cultures grown in 1 and 5% N compared to 100% N treatment. This further confirms that nitrogen was limiting and that N2 fixation did not compensate for the N-deficiency, particularly in the 1% N treatment. In fact, heterocysts were not evident in the cultures and were either non-existent or too small to differentiate from vegetative cells. This may be consistent for CAWBG02 as it did not produce heterocysts in culture with or without nitrogen supplementation in the original report (Wood et al., 2007). With respect to another non-fixing cyanobacterium, van de Waal et al. (2009) reported molar C:N ratios ranging from 6.25 to 7.14 for *M. aeruginosa* when neither N nor C was limiting. *A. issatschenkoi* in the present study had lower C:N ratios overall. Nevertheless, significantly higher ratios of C:N were reached in the cultures grown under low nitrogen concentrations (**Table 3**).

According to the carbon-nutrient balance hypothesis, nutrients should be allocated to the production of secondary metabolites once growth is assured (Hamilton et al., 2001). Because cyanotoxins are generally considered secondary metabolites (Carmichael, 1992), their production should increase whenever the availability of essential nutrients increases. A higher production of anatoxina, an alkaloid that contains a nitrogen atom, was expected under higher nitrogen availability. In this study, the highest nitrogen supply corresponded to the highest growth rate and yield (**Table 3**) of CAWBG02 but not necessarily to higher total ANTX concentrations (**Figure 3**). In fact, the highest concentrations of total ANTX were produced by cultures grown in 5% N while the lowest mean concentrations of total ANTX were found in the 100% N treatment although the differences were not highly significant. Highly significant differences were observed in ANTX cellular contents (**Figure 4**): the 5% N cultures produced the highest cellular content and the 100% N treatment the lowest. The 100% N cultures had the highest growth rate, the highest biomass yield, the highest levels of particulate organic nitrogen and the highest phycocyanin to chlorophyll *a* ratios (**Table 3**; **Figure 2**), yet they produced the lowest ANTX cellular content. Additional concentrations of nitrogen should be tested to confirm these results based on only three contrasting treatments. If ANTX is indeed a type of defense compound then, its production might be a mechanism that confers ecological advantage particularly under low nutrient conditions that are common in aquatic systems during summer stratification when cyanobacteria tend to thrive and zooplankton grazing is typically high (Sommer et al., 1986).

The mean extracellular to total toxin concentration ratios were higher in 1% N treatment than in the other two treatments, the extracellular fraction almost exceeding the particulate fraction in stationary phase (**Figure 3**). Stress conditions in cyanobacteria can lead to permeability changes of the cell membrane that result in leakage (Whitton and Potts, 2000). However, it should be noted that the extracellular ANTX concentrations presented in this study were likely underestimated, as is likely the case for previous studies (Rapala et al., 1993). ANTX is quite labile under light and may degrade once released into the medium (Stevens and Krieger, 1988). Short-term radio-labeling experiments would be required to estimate more accurately the extracellular release. Furthermore, although the cultures were initially bacteria free, prepared and sub-sampled under sterile conditions, it is possible that some biodegradation arose.

This study suggests that toxin production may be higher when cyanobacterial cells are under moderate nutrient stress. Recently, Kurmayer (2011) reported that another nitrogen-fixing cyanobacterium *Nostoc* produced higher levels of microcystin cell content under stress conditions, particularly under low phosphorous (P-PO4) and low light irradiance, even though growth rates were reduced up to 100-fold compared to controls. As a result, microcystin content per cell was negatively correlated to P-PO4 and irradiance. Interestingly nitrogen reduction did not have the same effect, but, in this case, the number of heterocysts increased suggesting some alleviation of N-deficiency. Even if *Nostoc* produced higher levels of microcystin per cell when grown under stress conditions, the total toxin production was still lower than when grown under optimal conditions. In contrast, in the present study, both cellular and total ANTX concentrations in *A*. *issatschenkoi* were higher in cultures grown at intermediate N concentrations (**Figures 3** and **4**). Similarly, Rapala et al. (1993) observed that ANTX cellular contents were higher in both *A. flos-aquae* and *Anabaena flos-aquae* when grown in nitrogen-free media as opposed to nitrogen-rich control media. In this study, mean ANTX cell contents of CAWBG02 were both significantly lower in 100 and 1% N media than in 5% N. Stress conditions may increase ANTX cellular production to a certain point where cyanobacterial cells are still capable of producing the toxin without compromising primary metabolism. Mean cellular ANTX concentrations for the 1% N were slightly higher than those of the 100% N, but the difference was not significant. Based on cellular composition, the 1% N cultures were more severely stressed and the N concentration may have been too low to sustain much toxin production.

The results presented in this study suggest that ANTX production may increase in *A. issatschenkoi* under moderate nitrogen limitation. According to the carbon-nutrient balance hypothesis and our current understanding of secondary metabolism, the opposite was expected. One possible ecological advantage of such a strategy could be that when nitrogen is limiting and population density declines, additional population losses through grazing would be minimized via higher cell specific toxin concentrations. In the present experiments batch cultures were used as they best represent algal bloom dynamics typical of nuisance cyanobacterial growth in nature. However, it is possible that factors that are not well controlled in batch cultures may have influenced the results. Growth rates were slightly different among the treatments and might have had an effect on ANTX concentrations. Although it has not been specifically demonstrated for anatoxin-a, in the case of microcystins the cellular growth rate is the primary factor regulating production (Orr and Jones, 1998). However, in this case the higher growth rate observed in the 100% N culture did not lead to the higher anatoxin concentration or cellular content. Further studies encompassing a wider range of nutrient concentrations and using other toxin producing cyanobacterial taxa are necessary to confirm these results and to determine if physiological stress maximizes toxin production more generally.

## **ACKNOWLEDGMENTS**

We thank A. Saleem (University of Ottawa), R. Aranda-Rodriguez, and Z. Jin (Health Canada, Ottawa), along with M. Quilliam (National Research Council of Canada, Halifax) for advice with LS/MS/MS instrumentation and analyses. We thank L. Kimpe for her assistance with the Accelerated Solvent Extraction (ASE) and S. Bhatti with algal culturing. This work was supported by an Ontario Ministry of Environment Best in Science to Frances R. Pick and M. Quilliam.


stress conditions. *J. Phycol.* 47, 200–207.


carbon-nutrient balance hypothesis. *Ecol. Lett.* 12, 1326–1335.


**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: 06 February 2012; accepted: 24 May 2012; published online: June 2012. 12*

*Citation: Gagnon A and Pick FR (2012) Effect of nitrogen on cellular production and release of the neurotoxin anatoxin-a in a nitrogen-fixing cyanobacterium. Front. Microbio. 3:211. doi: 10.3389/fmicb.2012.00211*

*This article was submitted to Frontiers in Aquatic Microbiology, a specialty of Frontiers in Microbiology.*

*Copyright © 2012 Gagnon and Pick. This is an open-access article distributed under the terms of the Creative Commons Attribution Non Commercial License, which permits non-commercial use, distribution, and reproduction in other forums, provided the original authors and source are credited.*