Research Collection

Doctoral Thesis

The Molecular Basis of Embryonic Adaptation to Acid Stress in Amphibians

Author(s): Shu, Longfei

Publication Date: 2014

Permanent Link: https://doi.org/10.3929/ethz-a-010361827

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library

DISS. ETH NO. 22319

The Molecular Basis of Embryonic Adaptation to Acid Stress in Amphibians

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZURICH

(Dr. sc. ETH Zurich)

presented by

LONGFEI SHU

MSc in Biochemistry and Molecular Biology, Shenzhen University, China

Born on 03.02.1987

Citizen of China

Accepted on the recommendation of

Prof. Dr. Jukka Jokela

Prof. Dr. Barbara Tschirren

Dr. Katja Räsänen

Dr. Marc J-F Suter

2014

Table of contents

Summary 4

Zusammenfassung 6

General Introduction 9

Results and discussion 14

Chapter I): Evolution of egg coats 25

Chapter II): Molecular variation of egg coats in amphibians 49

Chapter III): Mechanistic basis of adaptive maternal effects 75

Chapter IV): Genomics resources and maternal effect 98

Chapter V): The roles of embryonic ion channels to acid stress 136

Acknowledgements 157

Curriculum Vitae 158

3

Summary

Understanding the molecular basis of evolution is fundamental to elucidating how biological diversity emerges and is maintained. This understanding will help to better describe how different levels of phenotypes are produced, and in what way they affect natural selection through their contribution to fitness. Environmental stress, such as acidification, can be a powerful evolutionary force and often acts at ecological time scales. However, it is seldom explored from a molecular perspective, except in model taxa.

In my PhD research, I used both common garden and molecular experiments to study the molecular basis of embryonic adaptation to acid stress in amphibians. In Chapter I of my thesis, I reviewed the evolutionary roles of egg coats (i.e., the maternally-derived extracellular structures surrounding the embryo), studied the molecular and functional variation driving egg coat- mediated maternal effects in Chapters II and III, identified candidate maternal effect genes in Chapter IV, and finally, investigated the roles of embryonic ion channels to acid stress in Chapter V.

In Chapter I, I identified several gaps in our current knowledge and highlighted emerging molecular techniques that might increase our understanding of the role of egg coats in the evolution of biological diversity from adaptation to speciation. In Chapter II, I studied the molecular basis of egg coat-mediated maternal effects in embryonic adaptation to acid stress in two amphibian species (Rana arvalis, R. temporaria). Using a proteomics approach, I have shown that there is extensive molecular variation in the gelatinous egg coats (i.e. egg jelly) among and within populations of both amphibian species. My research further indicates that three molecular components of jelly are correlated with embryonic acid tolerance. Using experimental manipulations of jelly and embryos in Chapter III, I showed that acidic pH can cause severe water loss in egg jelly, but that this loss is reduced in an acid-adapted population, indicating that environmental acidification imposes strong selection on this phenotypic trait (i.e., water balance of egg jelly) in amphibians. In this chapter, I have proposed a “jelly water balance” model, whereby intra-specific glycan variability influences the balance between water uptake and water loss of egg jelly under different pH conditions, thus helping to potentially explain the molecular basis of egg coat-mediated adaptive maternal effects. In Chapter IV, I used a transcriptomics approach to identify a set of candidate genes underpinning the glycoprotein variability mediated adaptive maternal effects in amphibian egg coats. This study in Chapter IV indicates that multiple genes and pathways are involved in the biosynthesis of egg jelly and importantly, provides the first set of genomics resources for a non-model amphibian species.

Finally, in Chapter V, using three R. arvalis populations, I studied the role of ion channels and pumps in embryonic acid tolerance by manipulating their functions using ion channel blockers. In this experiment, I found that ion channels mediating Ca2+ influx are essential for embryonic survival under acidic pH, and, intriguingly, that among populations, there is divergence in calcium channel function. Taken together with the findings from my egg coat studies in Chapters

4

II-IV, my results suggest that embryonic adaptation to acidity in amphibians is likely a combination of maternally-derived egg coats and the trait evolution of embryonic ion channels.

In conclusion, my results suggest that rapid adaptation to environmental acidification in amphibians derive from simultaneous changes in two key traits affecting embryonic fitness (egg coats and ion channels), and that both maternal and genetic effects can make an important contribution to adaptive divergence. Moreover, my results highlight that multidisciplinary approaches can increase our understanding on the molecular basis of evolutionary processes, which will help to better understand how biodiversity is created and maintained.

5

Zusammenfassung

Ein Verständnis der molekularen Grundlage von Evolution ist der Schlüssel zu einem besseren Verständnis der Prozesse, die Biodiversität erschaffen und erhalten. Es wird dabei helfen, zu illustrieren, wie verschiedene Ebenen von Phänotypen gebildet werden, und wie diese auf natürliche Selektion reagieren und zur Fitness eines Individuums beitragen. Umweltbedingter Stress, wie zum Beispiel eine Versauerung, kann eine starke evolutionäre Kraft darstellen und wirkt oft in ökologischen Zeitskalen. Dennoch sind die molekularen Grundlagen von umweltbedingtem Stress in Nicht-Modellorganismen schlecht untersucht.

In meiner Doktorarbeit habe ich sowohl Common Garden - als auch molekulare Experimente durchgeführt, um die molekulare Basis der embryonalen Anpassung von Amphibien an Säurestress zu untersuchen. Ich habe mich hauptsächlich mit der evolutionären Rolle der Eihülle beschäftigt (d.h. mit der von der Mutter gebildeten extrazellulären Struktur, die den Embryo umgibt) und darüber einen Bericht verfasst (I); ich habe die molekulare und funktionale Variation untersucht, die Eihülle-abhängigen mütterlichen Effekten zugrunde liegt (II, III); ich habe Kandidatengene für mütterliche Effekte identifiziert (IV); und ich habe genauer betrachtet, welche Rolle embryonale Ionenkanäle bei der Anpassung an Säurestress spielen (V).

Ich habe die wichtigsten Forschungsfelder identifiziert und aufkommende molekulare Techniken hervorgehoben, die unser Verständnis von der Rolle der Eihülle in der Evolution biologischer Diversität von der Anpassung bis zur Artbildung erhöhen könnten. Die molekulare Basis von Eihülle-abhängigen mütterlichen Effekten bei der Anpassung an Säurestress habe ich in zwei Amphibienarten (Rana arvalis und R. temporaria) untersucht. Unter Verwendung eines proteomischen Ansatzes zeige ich, dass die gallertartige Eihülle sowohl innerhalb beider Arten als auch zwischen den Arten eine beträchtliche molekulare Variation aufweist. Meine Arbeit zeigt ausserdem auf, dass drei molekulare Bestandteile der Gallerte mit der embryonalen Säuretoleranz korrelieren. Durch die experimentelle Manipulationen von Gallerte und Embryo lege ich dar, dass ein saurer pH einen schweren Wasserverlust der Gallerte zur Folge haben kann. Dieser Wasserverlust ist vermindert in einer Population, die sich an Säurestress angepasst hat, was darauf hinweist, dass die umweltbedingte Versauerung in Amphibien eine starke Selektion auf den Wasserhaushalt der Gallerte ausübt. Ich schlage vor, dass ein „Gallerte- Wasserhaushalts-Modell“ dabei helfen würde, die molekulare Grundlage von Eihülle- abhängigen mütterlichen Anpassungseffekten zu klären. In einem solchen Modell beeinflusst die innerartliche Variation in Glykanen, ob die Gallerte je nach pH Wasser aufnimmt oder verliert. Ich habe einen transkriptomischen Ansatz angewandt, um eine Gruppe von Kandidatengenen zu identifizieren, die den Glykoprotein-abhängigen mütterlichen Anpassungseffekten in Amphibien zugrunde liegen. Diese Untersuchung legt die Vermutung nahe, dass mehrere und Signalwege in die Biosynthese der gallertartigen Eihülle involviert sind. Zudem stellt meine

6

Analyse die erste Sammlung genomischer Ressourcen für eine Amphibienart dar, die kein Modellorganismus ist.

Abschliessend habe ich in drei Populationen von R. arvalis die Rolle von Ionenkanälen und – pumpen im Kontext der embryonalen Säuretoleranz untersucht, indem ich die Funktion einer Reihe von Ionenkanälen und –pumpen mit Hilfe von Ionenkanalblockern manipuliert habe. In diesem Experiment habe ich herausgefunden, dass Ionenkanäle, die den Ca2+-Zufluss steuern, äusserst wichtig sind für das Überleben des Embryos in saurem pH, und dass sich interessanterweise Populationen in der Funktion dieser Kalziumkanäle stark unterscheiden. Zusammen mit den Ergebnissen der Eihülle-Untersuchungen deuten meine Resultate darauf hin, dass die embryonale Anpassung von Amphibien an Säure wahrscheinlich durch eine Kombination von durch die Mutter produzierten Eihüllen und der adaptiven Evolution embryonaler Ionenkanäle zustande kommt.

Zusammenfassend weisen meine Resultate darauf hin, dass die schnelle Anpassung von Amphibien an eine umweltbedingte Versauerung von der zeitgleichen adaptiven Evolution zweier Schlüsselmerkmale herrührt (der Eihülle und der Ionenkanäle), welche die biologische Fitness von Embryos beeinflussen. Ausserdem legen meine Ergebnisse die Vermutung nahe, dass sowohl mütterliche als auch genetische Faktoren einen wichtigen Beitrag leisten zur adaptiven Auseinanderentwicklung verschiedener Populationen. Überdies zeigen meine Resultate auf, dass multidisziplinäre Ansätze unser Verständnis der molekularen Grundlage von evolutionären Prozessen erhöhen können, was dazu beitragen wird, die Vielfalt des Lebens besser zu verstehen.

7

This thesis is based on the following chapters which have been written as manuscripts, and will henceforth be referred to in the text by their Roman numerals:

I. Shu, L., Suter, MJ-F. and Räsänen, K. Evolution of egg coats: linking molecular

biology and ecology. Molecular Ecology. In revision

II. Shu, L., Suter, MJ-F., Laurila, A. and Räsänen, K. Molecular phenotyping of maternal

egg jelly reveals parallel adaptive divergence to acidity in Rana arvalis and Rana

temporaria. Manuscript

III. Shu, L., Laurila, A., Suter, MJ-F., and Räsänen, K. Mechanistic basis of adaptive

maternal effects: egg jelly water balance mediates embryonic adaptation to acidity in

Rana arvalis. Oecologia. Submitted

IV. Shu, L.. and Räsänen, K. First transcriptome analysis of the moor frog Rana arvalis:

genomics resources and adaptive maternal effects genes. Manuscript

V. Shu, L., Laurila, A., and Räsänen, K. Environmental stress causes adaptive

divergence in ion channel function during embryogenesis. Manuscript

8

General Introduction

Understanding the molecular basis of evolution: necessity and challenge

Understanding the molecular basis of evolution (Fig. 1) is a fundamental challenge in evolutionary biology, which will help us to better comprehend how biodiversity emerges and is maintained (Mitchell-Olds et al. 2007; Houle et al. 2010). It will help to bridge two major gaps in evolutionary studies: 1) linking adaptive traits to their corresponding selective agents, and 2) connecting adaptive traits to their underlying genotypes. Understanding the molecular basis of phenotypes that have fitness consequences will allow us to identify the origins and consequences of natural variation, as well as the mechanisms that control evolutionary change (Linnen et al. 2009; Chan et al. 2010; Jones et al. 2012; Linnen et al. 2013). In addition, by identifying specific adaptive traits, it may also be possible to disentangle the relative contribution of genetic (Mitchell-Olds et al. 2007; Nadeau & Jiggins 2010) and epigenetic effects (Badyaev & Uller 2009; Kappeler & Meaney 2010) to evolutionary processes (Fig. 1).

Figure 1. A schematic presentation on the molecular basis of evolutionary processes. PTM refers to post translational modifications.

However, our ability to understand phenotypic variation is still limited, especially at the molecular level (Houle et al. 2010; Diz et al. 2012). This is particularly so for molecular phenotypes such as mRNA or , which cannot be measured by the traditional approaches used by evolutionary biologists. In addition, given the broad range of phenotypes (from molecular to organismal) (Fig. 1), it will be particularly challenging to measure all of these in the same individuals (Houle et al. 2010). To address these challenges, multidisciplinary approaches such as proteomics (Diz et al. 2012), transcriptomics (Wang et al. 2009) and glycomics (Hart & Copeland 2010) are needed to better understand phenotypic variations at different levels, as well as how they respond to natural selection and contribute to an individual’s fitness (Houle et al. 2010; Diz et al. 2012).

9

Evolution of egg coats

At early life-stages, organisms’ interactions with the environment are often mediated via maternal effects (MEs), here defined as the effects of the mother’s phenotype or environment on offspring phenotype or performance (Mousseau & Fox 1998). MEs can contribute to adaptive divergence of local populations, allow rapid adaptation, and alter the speed and direction of evolution (Mousseau & Fox 1998; Räsänen & Kruuk 2007).

One important, but understudied source of MEs is egg coat (Fig. 2). Egg coats are maternally- derived, extracellular structures that surround the embryonic organism during early life stages, and consist of multiple layers that differ structurally and functionally (Wong & Wessel 2006; Menkhorst & Selwood 2008). These structures are a major component of fitness: they mediate the beginning of life due to their fundamental role in fertilization and protect the embryo from a range of environmental hazards (Monne et al. 2006; Wong & Wessel 2006; Menkhorst & Selwood 2008; Claw & Swanson 2012).

Egg coats show great molecular, structural and functional diversity across species (Monne et al. 2006; Wong & Wessel 2006; Claw & Swanson 2012), but their broad evolutionary and ecological significance, as well as interspecific variation, is currently underappreciated. In addition, despite the crucial functions that egg coats perform, they are seldom studied from a molecular perspective outside model taxa.

Figure 2. A schematic presentation of the basic structure of egg coats in amphibians. The embryo is surrounded by the fertilization envelope (FE), which (in some taxa) is surrounded by gelatinous structures, so called jelly envelopes.

Components of embryonic adaptation to environmental acidity

Environmental stress, defined as a condition that lies outside the optimal conditions for an organism and that impairs Darwinian fitness, can have strong ecological consequences and be a powerful evolutionary force at ecological time scales (Hoffmann & Parsons 1997). Acidic stress arising from environmental acidification can have strong negative impacts on a broad

10

range of taxa (Räsänen & Green 2009; Kroeker et al. 2010; Dam 2013; Kwong et al. 2014;). Whilst often challenging the persistence of natural populations, these negative fitness effects also imply that acidity should cause strong natural selection. In accordance, there is increasing evidence for adaptation to acidity from both freshwater (Derry and Arnott 2007; Hangartner et al. 2011) and marine (Lohbeck et al. 2012; Dam 2013; Evans et al. 2013; Pespeni et al. 2013) taxa.

Amphibians can suffer from acidification in both of their terrestrial and aquatic habitats at local and global scales (Räsänen & Green 2009). When amphibian embryos are exposed to acidic conditions, they typically show a “curling defect” (Dunson & Connell 1982; Pierce 1985), whereby embryos develop, but become tightly curled within the egg coat and, finally, fail to hatch. The curling defect has been suggested to relate to chemical changes in egg coats, which become tight and sticky, shrink in size, and can change color from transparent to opaque at an acidic pH (Dunson & Connell 1982; Pierce 1985; Picker et al. 1993; Räsänen et al. 2003a). However, the molecular underpinnings of the embryonic “curling defect” and, hence, the mechanistic basis of maternally-mediated embryonic adaptation to acid stress remains unexplored.

In addition, although egg coats seem to play significant roles in embryonic adaptation to acid stress, other putative pathways have not been investigated to date. Many environmental stressors, such as salinity and acidification, severely disrupt ionic balance of organisms, thereby challenging the fitness of natural populations (Gonzalez 2012; Stumpp et al. 2012). Although ion channels can have several vital functions during early life-stages (e.g. embryogenesis Leclerc et al. 2006; Hur et al. 2012), it is currently not known how developing embryos maintain proper intracellular conditions when exposed to environmental stress. Moreover, to date, little is known as to what extent environmental stress can drive intra- specific divergence in ion channel functions.

Aim of the thesis

In my thesis, I studied the molecular basis of embryonic adaptation to acid stress in an attempt to answer the following questions:

1. How do egg coats evolve among and within species? I summarized our current understanding of the role that egg coats play in the evolution of biological diversity, ranging from adaptation to speciation (I).

2. Is there intraspecific variation in amphibian egg coats? I compared the molecular phenotypes of egg coat jelly from populations originating from neutral, intermediate and acid pH environments of R. arvalis and R. temporaria (II).

11

3. How is amphibian egg jelly affected by environmental stress? I investigated variation in egg jelly water and jelly charge in three R. arvalis populations originating from neutral, intermediate, and acid pH environments (III).

4. What are the mechanisms of adaptive maternal effects to acidity? I studied the influence of egg coat jelly molecular phenotypes (II), jelly water, and jelly charge (III) on embryonic survival.

5. What are the genes underlying adaptive maternal effects? I investigated the potential maternal effect genes responsible for the biosynthesis of egg jelly, as well as their expression profiles across divergence individuals (IV).

6. What other components influence embryonic adaptation to acid stress? I investigated the potential roles of ion channels and pumps in embryonic acid tolerance in three divergent R. arvalis populations (V).

Study species and populations

I focused on two amphibian species, the moor frog (Rana arvalis, (II, III, IV, V)) and the common frog (R. temporaria, (II)). R. arvalis and R. temporaria are both widely distributed anurans in the western Palearctic (Glandt 2006; Teacher et al. 2009). Both species inhabit a wide range of pHs, but R. temporaria (pH 4.2 - 8.9) is typically more acid-sensitive and is usually absent at sites where pH is below 4.5, whereas R. arvalis (pH 3.5 - 9) is more acid- tolerant (reviewed in Räsänen & Green 2009). R. arvalis is one of the best-characterized study systems for adaptation to acidification and shows adaptive divergence in embryonic survival, larval traits and maternal investment (Räsänen et al. 2003a, b; Räsänen et al. 2008; Räsänen & Green 2009; Hangartner et al. 2011; Egea-Serrano et al. 2014). In contrast, the adaptation of acid-tolerance in R. temporaria is less well studied (Glos et al. 2003).

I studied three R. arvalis (Tottajärn, T, Bergsjö, B and Stubberud, S) (II, III, IV, V) and two R. temporaria populations (Bergsjö, B and Stubberud, S) (II) breeding in permanent ponds in southwestern Sweden (details and maps are provided in (Hangartner et al. 2011)). pH levels in the study ponds included highly acidic (pH 4, site T), intermediate (pH 6, site B) and neutral (pH 7.5, S). Both R. arvalis and R. temporaria occur at sites B and S, but R. temporaria has no viable breeding population at site T. Site T is heavily affected by anthropogenic acidification since the early 1900s (Renberg et al. 1993), whereas site S has remained unaffected by acid rain due to limestone bedrock (Renberg et al. 1993; Hangartner et al. 2011). Site B was heavily acidified in the past (lowest pH measured 4.2), but the lake has been limed regularly since 1989 and current pH fluctuates around ca. 6.

12

Material and methods

I used both common garden and molecular experiments to study the molecular basis of embryonic adaptation to acid stress in amphibians. Males and females in breeding condition (II, IV) or freshly laid eggs (III, V) were collected at each breeding site during spring in 2012 and 2013 and transported to the laboratory at Uppsala University. Jelly and tissue samples were then transported to Eawag, Switzerland. Common garden experiments were conducted at Uppsala University, whereas molecular analyses were conducted at Eawag/ETH Zurich.

Common garden experiment

Embryonic acid tolerance of R. arvalis (II, III, IV, V) and R. temporaria (II) was tested using standard procedures (Räsänen et al. 2003a; Hangartner et al. 2011). In short, embryos were reared at two acidic pH treatments (acid: pH 4.0 and pH 4.2 for R. arvalis and pH 4.2 and pH 4.4 for R. temporaria) and a neutral (pH 7.5 for both species) in a climate-controlled room (16 °C) with 17L: 7D photoperiod. Reconstituted soft water (RSW) was used as the experimental medium (APHA 1985). The pH in the acid treatment was adjusted with 1M H2SO4, whereas the pH in the neutral treatment was not adjusted.

In the ion channels and pumps experiment (V), embryonic acid tolerance (hatching rate) was tested in combination with five inhibitor treatments (Blank control, Amiloride (Ami), Ouabain (Oua), Lanthanum chloride (Lan) and Verapamil (Ver)). The Ami, Oua, Lan and Ver treatments block the flux of H+, Na+ and Ca2+ ions, respectively.

Variation in jelly water content (III) was measured in acid (pH 4.0) and neutral (pH 7.5) treatments, as well as in an alkaline (pH 10) treatment. The pH in the alkaline treatment was adjusted with 1M NaOH, otherwise the general experimental conditions were similar as in above. After exposure of the eggs to respective pH treatments, eggs were manually de-jellied (Hedrick and Hardy 1991), and jelly was collected for subsequent water content measurements.

Molecular experiment

Molecular phenotypes of jelly (II) were measured using a gel-based proteomics approach (SDS-PAGE, Schagger & Vonjagow 1987).

Jelly charge state (zeta potential) of each clutch (III) was measured using the Zetasizer Nano ZSP (Malvern Instruments). Zeta potential was determined using Malvern Zetasizer Series Software (v7.02).

The oviduct transcriptome was sequenced using the Illumina HiSeqTM 2000 ( IV, V) and de novo assembled and annotated. Gene-linked microsatellites (SSRs) and single nucleotide polymorphisms (SNPs) were predicted (IV). Putative maternal effect genes were identified by pathway enrichment (V). Gene expression (V) was normalized using FPKM method (Fragments Per kb per Million reads) and compared using NOIseq method.

13

Results and discussion

Evolution of egg coats

I identified several gaps (I) in our current knowledge and highlight emerging molecular techniques that can increase our understanding of the role of egg coats in the evolution of biological diversity from adaptation to speciation. I summarized current understanding on the inter-specific variation of egg coats, which show great molecular, structural, and functional diversity (Fig. 3). Moreover, I highlighted that intra-specific variability (i.e. variation within and among individuals and populations) and the role of ecology in egg coat evolution have largely been overlooked.

Figure 3. A schematic illustration of the evolutionary loss and gain of ZP genes in major vertebrate groups. ZP gene subfamily names are given on the left and species names at the top. Circle: genes present; red circle: pseudogene; grey circle: gene duplication; blank: no gene exists. Branch colors highlight fish (grey), amphibians (yellow), birds (green) and mammals (red).

Molecular variation in amphibian egg jelly

In general, egg jelly (II) exhibited extensive inter-and intra-specific molecular polymorphism (Fig. 4). In both species, egg jelly consisted of multiple bands, with a molecular weight ranging from 10 kD to 185 kD. Clustering analysis (II) showed that the macromolecular composition of egg jelly is species-specific (Fig. 4).

DAPC (II) identified two phenotypic clusters that were separated along the first axis (Fig. 5): all R. temporaria samples clustered on the left (cluster 1 and cluster 5), whereas all R. arvalis samples clustered on the right (cluster 2, cluster 3 and cluster 4). This again indicated species boundaries in jelly macromolecular-composition. In addition, within both species, jelly clusters showed divergence and were separated in the second axis (Fig. 4), suggesting intra-

14

specific divergence of jelly phenotypes within both species. These data provide the first comprehensive evidence for intra-specific variation in amphibian egg jelly.

Figure 4. Molecular variation of jelly based on SDS page analyses. Molecular weights of each band are presented on the X axis and clutch identity on the Y axis.

Figure 5. Scatterplot of the discriminant analysis of principal components (DAPC) on jelly phenotypes from R. arvalis and R. temporaria (total N = 48 clutches). The first axis is the horizontal

15

axis. Colors represent five phenotypic groups found by the K-means and BIC method (Jombart et al. 2010). At the bottom left, the cumulative variances explained by number of PCA are represented.

Effects of pH on egg jelly

I found that jelly water content (III) depended on the pH that the clutches were exposed to (Fig. 6): jelly contained less water in the acid treatment than in the neutral treatment, indicating that jelly water was lost in the acid treatment. Likewise, jelly contained significantly more water in alkaline treatment, indicating that it absorbed more water in the alkaline treatment than in the neutral treatment (Fig. 6).

In addition, across all experimental clutches (three populations, five clutches per population of R. arvalis) embryonic acid tolerance (survival at pH 4.0) was negatively correlated (III) with water loss (Fig. 7a), suggesting that the negative effects of acidic pH on hatching success are related to jelly water loss. I further showed that all jelly samples were negatively charged (III). As acidic pH is expected to reduce negative charge, my results suggest that low pH conditions reduce the capacity for egg jelly to retain water and causes embryonic mortality. These data provide evidence on how environmental acidity affects egg jelly, and subsequently, affected embryonic fitness. Based on these findings I propose a “pH – jelly water balance” model (Fig. 6).

Figure 6. Schematic presentation of a “pH-jelly water balance” model. According to this model, jelly absorbs less and/or loses more water at acidic pH, while absorbing more water at alkaline pH compared to neutral pH conditions. This is based on the hypothesis that these patterns are mediated via the net electric charge of the jelly: acidic pH (in our case pH 4.0) reduces negative charge, which results in a weaker electrostatic repulsion and, therefore, reduces capacity of retaining water. In contrast, alkaline pH has the opposite effect by causing stronger charge-charge repulsion.

16

Mechanisms of adaptive maternal effects to acidity

In my thesis, I have shown that jelly water, jelly charge (III) and jelly molecular phenotypes (II) are correlated with embryonic survival (II, III) across different populations. In general, embryonic survival was reduced under acidic stress. However, the acid origin population T had higher embryonic survival than the neutral origin population S, whereas B population was intermediate, indicating adaptive divergence of embryonic survival across populations.

In addition, I found that populations differed in jelly water content at different pH’s (III). At pH 4.0, jelly from the acid origin population T had lost least and jelly from the neutral population S most water, while jelly from population B was intermediate in jelly water loss. Jelly charge (zeta potential) also differed among populations, with jelly of population T being more negatively charged than jelly of population S.

Moreover, embryonic acid tolerance (survival at pH 4.0) correlated with both water loss and charge state of the jelly (Fig. 7), indicating that negatively charged glycans influence jelly water balance and contribute to embryonic adaptation to acidity. These results indicate that egg coats can harbour extensive intra-specific variation, facilitated likely in part via strong selection on water balance and glycosylation status of egg jelly.

Figure 7. Scatterplot of clutch means (N = 15) of a) jelly water loss and embryonic acid tolerance (survival at pH 4.0), b) zeta potential and embryonic acid tolerance and c) zeta potential and jelly water loss across three R. arvalis populations (S, B and T).

This was further supported by the molecular phenotyping analysis of jelly (II): within each species, the different clusters (Fig. 5) differed in embryonic acid tolerance. Within R. arvalis, clusters differed in their acid tolerance (Fig. 5), whereby embryos from cluster 3 had the higher survival under acid stress than clusters 4 and clusters 2. Similarly within R. temporaria, embryos from cluster 5 had higher acid tolerance than those embryos from cluster 1 (Fig. 5). These data (II) showed that acid tolerance of the embryos was clearly identified along the second axis, and that this pattern was consistent in both species (Fig. 5).

17

In addition, we found that several molecular components are associated with embryonic acid tolerance (185 kD, 25 kD and 20 kD).

These results suggest that environmental acidification may cause strong positive selection on egg jelly composition. In this scenario, polymorphism in jelly composition (II) has strong fitness consequences in embryonic survival under acid stress. I suggest that the molecular polymorphism of egg jelly may be due to the interaction of balancing selection and positive selection. These findings help shed light on the molecular mechanisms of environmental stress tolerance and adaptive maternal effects.

Maternal effect genes

I used transcriptome analyses on seven female R. arvalis, which bracket the range of most acid tolerant and most acid sensitive phenotypes based on embryonic tolerance estimates (II). I found two groups of candidate ME genes (IV) that are likely involved in the biosynthesis of egg jelly: egg jelly core genes and egg jelly glycosylation genes. The major components of egg jelly core proteins are Mucin and Collagen, and I detected 13 and 11 different types of them, respectively. Of these, Mucin-2, Mucin-5AC and Mucin-5B were very highly expressed, and they were the most abundant transcripts of all unigenes, making them the most likely candidates of egg jelly core protein genes. In addition, I identified five major biosynthesis pathways (IV) likely involved in jelly glycosylation (i.e. Mucin type O−Glycan, Other types of O−glycan, Heparan sulfate, Chondroitin sulfate and Keratan sulfate).

In general, expression of these putative ME genes (IV) was very diverse across the females and extremely variable (Fig. 8), which might explain the high molecular polymorphism of jelly detected in the above session (II). These data suggest that amphibian egg jelly phenotypes are determined by multiple genes and pathways rather than a single gene or pathway, and their expressions are highly variable.

18

Figure 8. Heat map of candidate ME gene expressions. Colors represent high (red), low (green) or average (black) expression values based on Z-score normalized FPKM values for each gene. Egg jelly core protein and jelly glycosylation genes are marked as grey and black, respectively.

Ion channels: novel component in embryonic adaptation to acid stress

Although ion channels can have several vital functions during early life-stages (e.g. embryogenesis), it is currently not known how developing embryos maintain proper intracellular conditions when exposed to environmental stress. Moreover, to date, little is known on to what extent environmental stress can drive intra-specific divergence in ion channels.

I found (V) that Ca2+ ion channels play a significant role during early embryogenesis and embryonic response to acid stress. Intriguingly, I further found among-population differences in Ca2+ ion channel function, indicating that environmental acidification can drive adaptive divergence of ion channel function at early life-stages. I show (V) for the first time that embryonic stress acid tolerance in amphibians is likely dependent on Ca2+ ion flux and that

19

different calcium channels are activated under different pH conditions. Ion channels that mediate Ca2+ influx are essential for embryonic survival under acidic pH.

Figure 9. Effects of two pH (4.0 and 7.5) and five inhibitor (Blank, Ami, Oua, Lan and Ver) treatments on embryonic survival (mean ± SE) in three R. arvalis populations: a neutral origin (S) population, an intermediate pH origin (B) population and an acid origin (T) population.

Summary and general implications

Taken together, my results suggest that environmental acidity can simultaneously drive adaptive divergence in both egg coats (II, III, IV) and embryonic ion channels (V) – in particular the Ca2+ channels (Fig. 10). Below I propose the molecular mechanisms underlying responses to divergent selection via environmental acidification, as well as highlight the necessity to apply multidisciplinary approaches to address questions in evolutionary biology.

In order to develop normally in acidic environments, embryos have to be able to maintain intracellular ionic and pH balance, which imposes selection on Ca2+ channel function (Fig. 10). Meanwhile, adaptive responses in embryonic Ca2+ channel functions themselves may, however, be insufficient to ensure successful hatching, as the jelly envelopes are strongly negatively affected by acidic pH, trapping embryos inside the jelly resulting in failure to hatch. Therefore, I propose that at extremely acidic conditions (such as those the T population is exposed to in nature), there is simultaneous selection on embryonic Ca2+ channels and on the maternally derived egg coats.

20

Figure 10. A schematic presentation of putative mechanisms of pH mediated selection on embryonic acid stress tolerance in amphibians.

From a broader perspective, my study emphasizes the need for inter-disciplinary studies, linking different molecular and evolutionary ecological experimental approaches, to elucidate the molecular basis of evolutionary processes. With such an approach, we will make substantial progress in understanding the underlying mechanisms behind the creation and maintenance of biodiversity.

Future directions

My thesis shows that environmental stress is a strong selective force that can simultaneously drive adaptive divergence in both egg coats and embryonic ion channels in natural populations. However, due to the time constraints imposed by a PhD, several questions raised remain unanswered, and should be addressed in the future.

1. What is the genetic basis of adaptive traits (e.g. egg coats and ion channels)?

This question can be addressed by molecular evolution analysis of the candidate ME genes identified above (IV). ME genes should be sequenced in more populations and individuals to detect indicators of selection such as substitution rate ratios (dN/dS) tests, neutrality tests and possible effects of selection and population history on linkage disequilibrium (LD).

2. What are the relative contributions of genetic and maternal effects to embryonic fitness?

To answer this question requires quantitative measurement of fitness on each adaptive trait. As both traits can be experimentally studied, this is feasible in future studies.

3. Moving towards the next generation of evolutionary study: integrating multiple disciplines

21

I propose that evolutionary biologists should take full advantages of technical advances in other fields and disciplines to better understand the mechanistic basis for evolution. For instance, omics approaches such as proteomics, transcriptomics and glycomics, can be utilized as high-throughput phenomics tools.

References

Altig R, McDiarmid RW (2007) Morphological diversity and evolution of egg and clutch structure in amphibians. Herpetological Monographs 21, 1-32. APHA (1985) Standard methods for the examination of water and wastewater. American Public Health Association, Washington, DC. Badyaev AV, Uller T (2009) Parental effects in ecology and evolution: mechanisms, processes and implications. Philosophical Transactions of the Royal Society B- Biological Sciences 364, 1169-1177. Chan YF, Marks ME, Jones FC, et al. (2010) Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer. Science 327, 302-305. Claw KG, Swanson WJ (2012) Evolution of the egg: new findings and challenges. In: Annual Review of Genomics and Human Genetics, Vol 13 (eds. Chakravarti A, Green E), pp. 109-125. Annual Reviews, Palo Alto. Coyne JA, Orr HA (2004) Speciation. Sinauer, Sunderland, MA. Dam HG (2013) Evolutionary adaptation of marine zooplankton to global change. In: Annual Review of Marine Science, Vol 5 (eds. Carlson CA, Giovannoni SJ), pp. 349-370. Annual Reviews, Palo Alto. Diz AP, Martinez-Fernandez M, Rolan-Alvarez E (2012) Proteomics in evolutionary ecology: linking the genotype with the phenotype. Molecular Ecology 21, 1060-1080. Dunson WA, Connell J (1982) Specific-inhibition of hatching in amphibian embryos by low pH. Journal of Herpetology 16, 314-316. Egea-Serrano A, Hangartner S, Laurila A, Räsänen K (2014) Multifarious selection through environmental change: acidity and predator-mediated adaptive divergence in the moor frog (Rana arvalis). Proceedings of the Royal Society B-Biological Sciences 281, 20133266. Glandt D (2006) Der Moorfrosch. Einheit und Vielfalt einer Braunfroschart, Bielefeld (Laurenti Verlag). Glos J, Grafe TU, Rodel MO, Linsenmair KE (2003) Geographic variation in pH tolerance of two populations of the European common frog, Rana temporaria. Copeia 2003, 650- 656. Gonzalez RJ (2012) The physiology of hyper-salinity tolerance in teleost fish: a review. Journal of Comparative Physiology B-Biochemical Systemic and Environmental Physiology 182, 321-329. Hangartner S, Laurila A, Räsänen K (2011) Adaptive divergence of the moor frog (Rana arvalis) along an acidification gradient. BMC Evolutionary Biology 11, 366.

22

Hart GW, Copeland RJ (2010) Glycomics hits the big time. Cell 143, 672-676. Hoffmann AA, Parsons PA (1997) Extreme environmental change and evolution, Cambridge University Press. Houle D, Govindaraju DR, Omholt S (2010) Phenomics: the next challenge. Nature Reviews Genetics 11, 855-866. Hur CG, Kim EJ, Cho SK, et al. (2012) K+ efflux through two-pore domain K+ channels is required for mouse embryonic development. Reproduction 143, 625-636. Jombart T, Devillard S, Balloux F (2010) Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics 11. Jones FC, Grabherr MG, Chan YF, et al. (2012) The genomic basis of adaptive evolution in threespine sticklebacks. Nature 484, 55-61. Kappeler L, Meaney MJ (2010) Epigenetics and parental effects. Bioessays 32, 818-827. Krapf D, Vidal M, Arranz SE, Cabada MO (2006) Characterization and biological properties of L-HGP, a glycoprotein from the amphibian oviduct with acrosome-stabilizing effects. Biology of the Cell 98, 403-413. Kroeker KJ, Kordas RL, Crim RN, Singh GG (2010) Meta-analysis reveals negative yet variable effects of ocean acidification on marine organisms. Ecology Letters 13, 1419- 1434. Kwong RW, Kumai Y, Perry SF (2014) The physiology of fish at low pH: the zebrafish as a model system. Journal of Experimental Biology 217, 651-662. Leclerc C, Neant I, Webb SE, Miller AL, Moreau M (2006) Calcium transients and calcium signalling during early neurogenesis in the amphibian embryo Xenopus laevis. Biochimica Et Biophysica Acta-Molecular Cell Research 1763, 1184-1191. Linnen CR, Kingsley EP, Jensen JD, Hoekstra HE (2009) On the origin and spread of an adaptive allele in deer mice. Science 325, 1095-1098. Linnen CR, Poh YP, Peterson BK, et al. (2013) Adaptive evolution of multiple traits through multiple mutations at a single gene. Science 339, 1312-1316. Marquis O, Millery A, Guittonneau S, Miaud C (2006) Toxicity of PAHs and jelly protection of eggs in the common frog Rana temporaria. Amphibia-Reptilia 27, 472-475. Menkhorst E, Selwood L (2008) Vertebrate extracellular preovulatory and postovulatory egg coats. Biology of Reproduction 79, 790-797. Monne M, Han L, Jovine L (2006) Tracking down the ZP domain: from the mammalian to the molluscan vitelline envelope. Seminars in Reproductive Medicine 24, 204-216. Mousseau TA, Fox CW (1998) Maternal Effects As Adaptation. Oxford University Press. Mitchell-Olds T, Willis JH, Goldstein DB (2007) Which evolutionary processes influence natural genetic variation for phenotypic traits? Nature Reviews Genetics 8, 845-856. Nadeau NJ, Jiggins CD (2010) A golden age for evolutionary genetics? Genomic studies of adaptation in natural populations. Trends in Genetics 26, 484-492. Nosil P (2012) Ecological speciation. Oxford University Press. Olson JH, Chandler DE (1999) Xenopus laevis egg jelly contains small proteins that are essential to fertilization. Developmental Biology 210, 401-410.

23

Palumbi SR (2009) Speciation and the evolution of gamete recognition genes: pattern and process. Heredity 102, 66-76. Picker MD, Mckenzie CJ, Fielding P (1993) Embryonic tolerance of Xenopus (Anura) to acidic blackwater. Copeia 4, 1072-1081. Pierce BA (1985) Acid tolerance in amphibians. Bioscience 35, 239-243. Räsänen K, Green E (2009) Acidification and its effects on amphibian populations. Amphibian Biology. Conservation and Ecology, Volume 8. Edited by Heatwole H. Surrey Beatty and Sons, Chipping Norton, Australia. Räsänen K, Kruuk LEB (2007) Maternal effects and evolution at ecological time-scales. Functional Ecology 21, 408-421. Räsänen K, Laurila A, Merilä J (2003a) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. I. Local adaptation. Evolution 57, 352-362. Räsänen K, Laurila A, Merilä J (2003b) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. II. Adaptive maternal effects. Evolution 57, 363-371. Räsänen K, Söderman F, Laurila A, Merilä J (2008) Geographic variation in maternal investment: Acidity affects egg size and fecundity in Rana arvalis. Ecology 89, 2553- 2562. Renberg I, Korsman T, Birks HJB (1993) Prehistoric Increases in the pH of acid-sensitive Swedish lakes caused by land-use changes. Nature 362, 824-827. Schagger H, Vonjagow G (1987) Tricine sodium dodecyl-sulfate polyacrylamide-gel electrophoresis for the separation of proteins in the range from 1-kDa to 100-kDa. Analytical Biochemistry 166, 368-379. Stumpp M, Hu MY, Melzner F, et al. (2012) Acidified seawater impacts sea urchin larvae pH regulatory systems relevant for calcification. Proceedings of the National Academy of Sciences of the United States of America 109, 18192-18197. Teacher AGF, Garner TWJ, Nichols RA (2009) European phylogeography of the common frog (Rana temporaria): routes of postglacial colonization into the British Isles, and evidence for an Irish glacial refugium. Heredity 102, 490-496. Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for transcriptomics. Nature Reviews Genetics 10, 57-63. Watanabe A, Onitake K (2002) The urodele egg-coat as the apparatus adapted for the internal fertilization. Zoological Science 19, 1341-1347. Wong JL, Wessel GM (2006) Defending the zygote: Search for the ancestral animal block to polyspermy. Current Topics in Developmental Biology, Vol 72, 1-161.

24

Chapter I

Evolution of egg coats: linking molecular biology and ecology

Shu, Longfei1, Suter, Marc J-F2 and Räsänen, Katja1

Affiliations:

1Eawag, Department of Aquatic Ecology, Switzerland and ETH Zurich, Institute of

Integrative Biology, Switzerland

2Eawag, Department of Environmental Toxicology, Switzerland and ETH Zurich, Institute of

Biogeochemistry and Pollutant Dynamics, Switzerland

25

Abstract

One central goal of evolutionary biology is to explain how biological diversity emerges and is maintained in nature. Given the complexity of the phenotype and the multifaceted nature of inheritance, modern evolutionary ecological studies rely heavily on the use of molecular tools. Here we show how molecular tools help to gain insight into the role of egg coats (i.e. the extra-cellular matrices surrounding eggs and embryos) in evolutionary diversification. Egg coats are maternally produced extra-cellular structures that surround most animals during the embryonic stage. They have many biological functions from mediating fertilization to protecting the embryo from environmental hazards and show great molecular, structural and functional diversity across species. Yet, intra-specific variability and the role of ecology in egg coat evolution have largely been overlooked. Given that most of the variation that influences egg coat function is ultimately determined by their molecular phenotype, cutting edge molecular tools (e.g. proteomics, glycomics and transcriptomics), combined with functional assays, are needed for rigorous inferences on the evolutionary ecology of egg coats. Here we identify key research areas and highlight emerging molecular techniques that can increase our understanding of the role of egg coats in the evolution of biological diversity, from adaptation to speciation.

Keywords: diversification, egg coats, natural selection, proteomics, glycomics

26

Introduction

One central goal of evolutionary biology is to explain how biological diversity emerges and is maintained in nature. However, due to the multifaceted levels of organismal diversity (from DNA sequences to complete phenotypes), and the complexity of mechanisms of inheritance (from direct genetic to epigenetic and parental effects) (Danchin et al. 2011), modern evolutionary ecological studies are increasingly reliant on molecular approaches, such as genomics (Hawkins et al. 2010) and proteomics (Diz et al. 2012). In this review, we highlight how bridging the gap between molecular and ecological studies can help to understand the variability in and the evolutionary role of egg coats.

Egg coats are maternally derived, extracellular structures that surround the organism during early life stages, and consist of multiple functionally and structurally different layers (Box 1). These structures are a major component of fitness: they mediate the beginning of life due to their fundamental role in fertilization (Monne et al. 2006; Wong & Wessel 2006; Menkhorst & Selwood 2008; Claw & Swanson 2012) and protect the embryo from a range of environmental hazards (Table 1). Egg coats show great molecular, structural and functional diversity across species (Monne et al. 2006; Wong & Wessel 2006; Menkhorst & Selwood 2008; Box 1), but - as we argue in this review – their broad evolutionary and ecological significance, as well as intra-specific variation, is currently underappreciated. In particular, given their key role in both sperm-egg interactions (fertilization) and interactions with the external environment during the embryonic development, egg coats can strongly influence reproductive fitness, be an important source of adaptive maternal effects (Mousseau & Fox 1998; Räsänen et al. 2003b) and influence the evolution of biological diversity from adaptation to speciation (Turner & Hoekstra 2008a; Palumbi 2009).

Recent progress in molecular techniques (i.e. proteomics, glycomics and X-ray crystallography) makes it now possible to identify and quantify large biomolecules, such as proteins and glycans, and, hence, gain knowledge about the functional consequences of molecular variation in egg coats (Claw & Swanson 2012). As we show in this review, the effective use of molecular tools is crucial as most of the functional consequences of egg coats arise directly from their molecular phenotype (i.e. variation in their chemical and structural composition). Therefore, these emerging molecular tools allow to better understand the link between the genotype and environment determining egg coat composition and the functional role of egg coats as part of the early phenotype of an organism – providing a good example of the merits of a phenomics approach (Houle et al. 2010).

Many reviews exist on the role of egg coats in sperm-egg interactions and their inter-specific variability (e.g. Jovine et al. 2005; Wong & Wessel 2006; Hedrick 2008; Menkhorst & Selwood 2008; Claw & Swanson 2012). What has largely been neglected to date is that egg coats mediate natural selection via embryonic performance. Here we emphasize the need for

27

an integrative view that incorporates the role of ecology and intra-specific variation (from the molecular to the functional level) of egg coats to gain insight into their role in evolutionary diversification. We identify important research gaps and challenges and illustrate some emerging molecular techniques that can increase our inferential ability when addressing ecological and evolutionary questions about egg coat variation. The review is structured as follows: 1) Why egg coats matter, 2) Inter-specific variation of egg coats, 3) The missing component: intra-specific variation, 4) Research gaps and challenges, and 5) Strategies and molecular tools for inter-disciplinary studies.

Table 1. Selected set of examples for the diverse functional roles of egg coats. Different egg coat layers have partially different functions.

Function References

Fertilization Increased sperm target size (Podolsky 2002) Sperm binding (Runft et al. 2002; Pang et al. 2011)

Acrosome reaction (Gunaratne 2007)

Block against polyspermy (Wong & Wessel 2006)

Species barrier (Turner & Hoekstra 2008a; Palumbi 2009; Hart et al. 2014)

Protection Adhesion (Pechenik 1979)

Oxygen transfer (Salthe 1963; Pinder & Friet 1994; Seymour 1994)

Thermoregulation (Salthe 1963)

Dehydration (Holmstrup & Westh 1995; Podrabsky et al. 2001)

Interactions with chemicals (Villalobos et al. 2000; Räsänen et al. 2003b; Edginton et al. 2006; Marquis et al. 2006) Pathogen resistance (Gomez-Mestre et al. 2006)

Acquirement of beneficial micro- (Pinder & Friet 1994; Kerney et al. 2011) organisms

Predation (Rawlings 1993; Roche et al. 2011)

Development Morphogenesis (Tsang et al. 2010)

Nutrition (Salthe 1963; Altig & McDiarmid 2007)

Maternal signaling (Tadros & Lipshitz 2009)

Implantation (Marco-Jimenez et al. 2012)

Hatching (Dunson & Connell 1982; Gomez-Mestre et al. 2006; Touchon et al. 2006)

28

Box 1. Basic structure and terminology of egg coats

At their simplest, egg coats can be divided into the vitelline envelope (VE), called fertilization envelope (FE) upon fertilization, and various kinds of outer layers. The structure of the outer layers varies strongly across diverse taxa, from thin or thick gelatinous structures (so called jelly envelopes) to the egg capsules of marine invertebrates and the egg shells of birds (Menkhorst & Selwood 2008). In some taxa, the outer layers are gelatinous – highly glycosylated jelly envelopes (the main focus of our review). The composition and number of the different egg coat layers differ among taxa, and so does the site of production: the FEs originate during oogenesis from the oocyte, follicle cells or liver cells, while the jelly layers are produced by the mother in the oviduct, liver cells or uterus, depending on the taxa (Menkhorst & Selwood 2008). With regard to the relative functional roles of different egg coat components (VE/FE, jelly coats and egg capsules or egg shells), it is important to note that VE and jelly coats are produced prior to fertilization, whereas egg capsules and egg shells are produced after fertilization (Menkhorst & Selwood 2008). Likewise, the relative length of time that offspring develop within these different structures vary from a few days to a few months.

It is important to note that - although egg coats play similar roles across animal taxa and consist of evolutionarily conserved glycoproteins (Wong & Wessel 2006) - the layers are named differently in different taxa in part for historic reasons. In particular, prior to fertilization, the innermost layer is called vitelline layer in echinoderms, vitelline membrane in dipterans, chorion in fish, vitelline envelope or zona radiata in amphibians, vitelline envelope in birds and zona pellucida in mammals (Wong & Wessel 2006; Hedrick 2008). The nomenclature of egg coat glycoproteins is also confusing. They are named differently depending on the methods by which they were identified. When analyzed using SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis), the glycoproteins are identified and generally named by their molecular weight (e.g. gp37, Kubo et al. 2000), whereas when they are analyzed using cDNA cloning they are named by gene name (e.g. ZPD, Lindsay et al. 2002).

For consistency, in our review we designate the whole extracellular matrix structure as “egg coats”, and divide them into the two core structures, the inner “fertilization envelope” (FE) and the outer layers. As we focus on gelatinous layers, we call them “jelly envelopes”, and follow Hedrick’s nomenclature for egg coat glycoproteins (Hedrick 2008). A good description of different types of egg coat structures in a range of taxa is provided in Menkhorst & Selwood (2008). For classification of ZP subfamily genes, we follow the recent nomenclature of six subfamilies: ZPA/ZP2, ZPB/ZP4, ZPC/ZP3, ZP1, ZPAX and ZPD (Wong & Wessel 2006; Goudet et al. 2008).

29

Figure Box 1. A schematic presentation of the basic structure of egg coats as used in this review. The embryo is surrounded by the fertilization envelope (FE), which (in some taxa) is surrounded by gelatinous structures, so called jelly envelopes. FEs are relatively similar across taxa, but the jelly layers can be highly variable in number of layers, internal structure and composition across taxa (e.g. Altig & McDiarmid 2007). For a thorough overview of variation in type of structures across taxa see Menkhorst & Selwood (2008).

Why egg coats matter

Egg coats are indispensable structures that have multiple roles during an animal’s early life stages. First, egg coats mediate fertilization – and hence the beginning of life itself - via sperm recognition, sperm binding, and blocking of polyspermy (Wong & Wessel 2006; Hedrick 2008). Second, egg coats have multiple roles during embryonic development. These range from providing a floating media (in aquatic taxa) and media for attaching eggs to the surroundings, to protecting the embryo from a range of abiotic (e.g. dehydration, UV radiation and pollutants) and biotic (e.g. predators and pathogens) environmental hazards (Table 1). As several good reviews exist on the function of egg coats in a range of taxa (e.g. eutherians, Denker 2000; Herrler & Beier 2000; anurans, Salthe 1963; Wake & Dickie 1998; Altig & McDiarmid 2007; fish, Berois et al. 2011) and vertebrates in general (Wong & Wessel 2006; Menkhorst & Selwood 2008; Claw & Swanson 2012), we will not discuss them in detail here, but instead focus on key aspects that have been neglected to date.

One such aspect, with potentially important evolutionary consequences, is that egg coats are maternally derived (Box 1) and can therefore be an important source of maternal effects (Räsänen et al. 2003b; Räsänen & Kruuk 2007). Whereas egg size and egg content related maternal effects are frequently addressed in animal studies (reviewed in Mousseau and Fox 1998), egg coats are rarely explored explicitly in the context of adaptive maternal effects. Egg coats often influence embryonic survival under abiotic and biotic environmental stress (e.g. acidity, Räsänen et al. 2003a; dehydration, Podrabsky et al. 2001; pollution, Edginton et al. 2006; Marquis et al. 2006; pathogens, Gomez-Mestre et al. 2006), and therefore can give rise to maternal effects that can determine responses of natural populations to environmental challenges, strongly influence individual reproductive success and, possibly, the evolutionary

30

trajectories of natural populations (e.g. Räsänen & Kruuk 2007). Finally, because of their joint role in sperm-egg interactions and their ecological importance, egg coats may act both as species barriers (e.g. Wong & Wessel 2006; Palumbi 2009) and be under natural selection (Podolsky 2002; Räsänen et al. 2003b) and could, thereby, contribute to both non-ecological and ecological reproductive isolation (i.e. speciation, Coyne & Orr 2004; Palumbi 2009; Nosil 2012), a topic to which we will return below.

Egg coats are still studied from a relatively narrow point of view and separately in different fields (molecular biology versus ecology). Whereas molecular biologist are interested in identifying molecules involved in fertilization (Pang et al. 2011), biochemists in identifying the composition of egg coats (Aviles et al. 2009) and medical researchers in linking their variation to pregnancy (Host et al. 2002), ecologist are typically interested in their functional role and fitness consequences (Salthe 1963; Pechenik 1979; Podolsky 2002). As these diverse studies seldom cross disciplines, it is currently difficult to fully appreciate the extent and evolutionary role of egg coat variability. We here want to emphasize the need for integrative studies to gain insight into their multifarious role and, in particular, the need to understand intra-specific variation (i.e. variation within and among populations of a given species) of egg coats. Therefore we briefly summarize the current status of egg coat studies, and identify some key research questions and challenges. Finally, we suggest possible tools and strategies for integrative future research.

Inter-specific variation of egg coats

Structure and function: conserved roles yet great diversity

Egg coats consist of a complex extracellular matrix, which typically has multiple layers, the number and type depending on species (Hedrick 2008; Menkhorst & Selwood 2008) (Box 1). These structures exist in all sexually reproducing animals, as well as many asexual metazoans (Wassarman 2008), and can vary in size from a few microns to over 20 centimeters (Ebert & Davis 2007). The core structures of egg coats can be divided into the vitelline envelope (VE) - or fertilization envelope (FE) after fertilization - and various outer layers (Box 1). The structure of the outer layers varies strongly across diverse taxa, from thin or thick gelatinous structures (so called jelly envelopes) to the egg capsules of marine invertebrates and the egg shells of birds (Menkhorst & Selwood 2008) (Box 1). As the VE and jelly layers are produced by the female prior to fertilization and structures such as the egg capsules in some marine invertebrates and egg shells of birds are produced by the female after fertilization, VE and jelly can function during both fertilization and early development of the offspring (see further below), whereas egg capsules and shells have primarily ecological functions (i.e. affecting offspring performance).

In this review, we focus on FE and the jelly, the latter being a common feature in a broad range of taxa (such as many invertebrate, gastropod, fish and amphibian species) in which

31

embryos develop outside the mother. We consider taxa with jelly envelopes both to direct attention to the role of ecology in driving egg coat diversification and because focus on jelly envelopes highlights the need to understand the composition of glycans (Varki et al. 2009), in addition to that of proteins (Diz et al. 2012; Box 2, 3). Although egg capsules and shells can have long-lasting fitness consequences and certainly are under strong natural selection, we do not specifically consider them here to allow us to dwell in more depth into the role of glycans. Glycans are a fundamental component of organismal cells and have many important biological functions, from adhesion to the development of cells and interactions with pathogens and chemicals (Hart & Copeland 2010). However, they are also much more complex to analyse (Varki et al. 2009; Hart & Copeland 2010), which renders the understanding of molecular underpinnings of egg coat evolution challenging.

We will use the term FE for the inner layer and the term jelly for the outer layers. All sexually reproducing animals have FEs and their basic functions are relatively similar across taxa (i.e. key role in sperm-egg interactions and basic protective layer to embryo). FEs consist of fibrous structures called zona pellucida (ZP) glycoproteins that are well studied across animal taxa (e.g. gastropods, Monne et al. 2006; fish, Berois et al. 2011; amphibians, Hedrick 2008; birds, Smith et al. 2005 and mammals, Litscher & Wassarman 2007; Wassarman et al. 2009, Claw & Swanson 2012). ZP glycoproteins share a common structural motif, known as the ZP domain (Jovine et al. 2005) and consist largely of proteins, with some attached glycans. ZP domain proteins have been found in all vertebrate taxa and many non-vertebrate taxa, indicating that the basic structure of egg coats has been conserved during evolutionary history (Monne et al. 2006; Claw & Swanson 2012). Despite the generally conserved role of ZP glycoproteins, they also represent rapidly evolving reproductive proteins, highlighting the potential for rapid evolutionary responses via egg coats (Turner & Hoekstra 2008a; Palumbi 2009; Claw & Swanson 2012).

In some taxa (e.g. gastropod molluscs, sea urchins, some fish and amphibians), the FEs are surrounded by thick jelly envelopes (e.g. Salthe 1963; Segall & Lennarz 1979; Hawkins & Hutchinson 1988; Menkhorst & Selwood 2008), which can be highly variable in structure, composition and function. Also the jelly envelopes consist primarily of glycoproteins, but the proportion of oligosaccharides is higher than in the FEs (Bonnell et al. 1994, 1996). For instance, in the African clawed frog (Xenopus laevis), one of the model systems for egg coat studies, more than 60% of the jelly (Yurewicz et al. 1975) - but only approximately 10% of FEs - consist of oligosaccharides (Hedrick 2008). These mucin type oligosaccharides of the jelly are highly species specific. In two closely related toad species (Bombina bombina and B. variegata), for example, glycan chains extracted from the oviducal mucins can be considered phenotypic markers of the two species (Strecker et al. 2003). Importantly, the jelly provides an adhesive medium, interacts with the physical environment and provides protection against diverse environmental hazards (e.g. dehydration, Podrabsky et al. 2001; predators, Roche et al. 2011; pathogens, Gomez-Mestre et al. 2006; UV radiation, Marquis & Miaud 2008). Therefore, jelly coats should be prime targets for natural selection. Their profiles, as well as

32

the detailed functions and the underlying genes mediating variation in jelly glycans, are still largely unknown in most taxa. This is an important gap given the multiple key biological functions of glycans (Varki et al. 2009; Hart & Copeland 2010).

Box 2. How to perform a proteomics study of egg coats in non-model species

Proteomics has only recently rigorously entered the domain of evolutionary ecology (Diz et al. 2012). As a comprehensive database is essential for a successful proteomics project (Diz et al. 2012), the most detailed studies on egg coat proteins to date have been performed in model species. However, once a corresponding database is available, it is also possible to perform proteomics in non-model species. The most common approach is to use a reference genome(s) of one or more model species (Forne et al. 2010). Unfortunately, this method only works well on homologous proteins and depends on the phylogenetic distance between the reference and the target study species (Diz et al. 2012). Another approach is to use de novo sequencing. This approach can overcome the “lack of database” problem, because it can infer the polypeptide sequences needed to identify and characterize proteins directly from the MS⁄MS spectra without the help of sequence database (Dancik et al. 1999; Savitski et al. 2005). Finally, RNA-seq based approaches are becoming available. These approaches can identify and quantify the transcriptome from both model and non-model species (Hawkins et al. 2010; De Wit et al. 2012), and the full length cDNA library can subsequently be translated into protein sequences, and used as a database in the MS data alignment (Wang et al. 2009). Knowledge on the site of egg coat production (e.g. liver versus oviduct) will then allow targeted (i.e. tissue specific) transcriptomics analyses.

Figure Box 2. A schematic presentation of the basic steps of the proteomics workflow for analysis of egg coats of non-model species. Egg coat proteins can be separated by gel-based (electrophoresis in polyacrylamide gels) or gel free approaches (separation of peptides by liquid chromatography) (Diz et al. 2012). Subsequently the proteins (when necessary upon removal of glycans; Box 3) are analyzed using mass spectrometry (MS/MS) and sequences finally blasted against a reference genome or analyzed via de novo sequencing. FEs are typically analyzed using proteomics tools as their glycoproteins consist mostly of proteins.

33

Evolution of egg coat genes: a progressive loss and gain across taxa?

Due to their many functions, egg coats are highly conserved (Jovine et al. 2005; Claw & Swanson 2012), but they can evolve rapidly (Aagaard et al. 2006, 2013; Turner & Hoekstra 2008a; Palumbi 2009). The genes coding for egg coats (VE and FE in particular) have been intensively studied in some taxa. The genetically best characterized egg coat structures are the FEs of the house mouse (Mus musculus). In the mouse, FEs are coded by three ZP genes: mZP1, mZP2 (also called ZPA) and mZP3 (also called ZPC) (reviewed in Wassarman 2008; Claw & Swanson 2012). The mouse ZP is a three dimensional fibrous matrix, in which mZP2 and mZP3 form polymers that are cross-linked by mZP1 (Wassarman 2008). Classically, the mouse mZP1 and mZP2 were deemed to be responsible for blocking polyspermy, and the mZP3 is a sperm receptor and inducer of the acrosome reaction, in addition to being a structural protein. However, more recent work challenges the ZP3 model, and indicates a role for ZP2 for sperm-binding in Xenopus (Tian et al. 1999) and mammals (Avella et al. 2013; 2014).

Although studies in other taxa are less detailed, the ZP gene family is found in a wide range of taxa (Litscher & Wassarman 2007; Meslin et al. 2012): ZP (or ZP like) genes have thus far been reported in at least 74 species (GENEBANK - http://www.ncbi.nlm.nih.gov/ ). Phylogenetic analyses indicate that ZP genes can be classified into six sub-families and that they typically evolve through gene duplication and pseudogenisation (Goudet et al. 2008; Claw & Swanson 2012; Meslin et al. 2012) (Figure 1). Taxa can differ in the number of ZP genes and the ZP3 is the only universal ZP gene (Figure 1). In contrast to the three ZP genes in the mouse, chicken has six, X. laevis five and humans four ZP genes, respectively (Goudet et al. 2008; Meslin et al. 2012). The universal presence of the ZP3 gene is likely due to the fundamental role of FEs in fertilization (i.e. the sperm receptor is an indispensable part in all animal taxa with sexual reproduction; Monne et al. 2006), but it is also generally recognized as the ancestral gene of all other ZP gene families. The presence of pseudogenes indicates that several ZP genes have been lost during evolution (Goudet et al. 2008; Meslin et al. 2012). For example, the ZP4 occurs as a pseudogene in the mouse, while the ZP1 occurs as a pseudogene in dog, pig, cat and cow (Figure 1). Most notably: the ZPD and ZPAX genes - which are present in Xenopus and chicken – have been pseudogenified or lost in all mammals (Goudet et al. 2008; Meslin et al. 2012) (Figure 1). For fish, the phylogeny of ZP genes is less well resolved because of the high level of gene duplications (Goudet et al. 2008; Meslin et al. 2012).

The reason of the loss of some ZP genes in mammals is currently not clear (Goudet et al. 2008), but may be in the different requirements posed by the selective environments that

34

Figure 1. A schematic illustration of the evolutionary loss and gain of ZP genes in major vertebrate groups. ZP gene subfamily names are given on the left and species names at the top (following Goudet et al. 2008) (see also Box 1). Circle: genes present; red circle: pseudogene (Goudet et al. 2008; Meslin et al. 2012); grey circle: gene duplication; blank: no gene exists (Supplementary Table 1). Branch colors highlight fish (grey), amphibians (yellow), birds (green) and mammals (red). The blue line in the evolution of ZP gene family indicates the unclear evolutionary origins of the ancestral ZPC/ZP3 gene (Goudet et al. 2008; Claw & Swanson 2012). The genes used here originate from GENEBANK (see Supplementary Table 1).

35

different taxa are exposed to. Alternatively, it is possible that some ZP genes do not play a significant role in matrix formation and sperm-egg interactions in mammals (e.g. Gahlay et al. 2010, Avella et al. 2014). It has been proposed that the loss of genes would relate to shifts from external to internal fertilization, although phylogenetic studies do not provide unambiguous support for this hypothesis (Goudet et al. 2008). It is important to keep in mind that, beyond ZP genes coding for components of the egg coats that mediate fertilization, egg coats have many other ecological relevant functions. Therefore, the evolution of ZP genes may also be influenced by natural selection acting via embryonic performance – although this seems to be largely ignored in current empirical work. For instance, in organisms with external development, such as most fish, amphibians and birds, embryos develop in risky environments and, hence, often require more additional functions from egg coats than do taxa with internal development (see Table 1). It may be that in mammals the genes coding for these additional structures/functions are present only as pseudogenes – but further studies are needed to test this hypothesis.

The missing component: Intra-specific variation

One of the core points of our review is to highlight the importance of quantifying molecular variation in egg coats, the functional consequences of this variation and how little is known about intra-specific variation in these structures. Why do we care about intra-specific variation in egg coats? As for any other trait, intra-specific phenotype variation is the raw material for selection to act upon, reflects the selective history and the potential of natural populations to evolve in response to environmental change, and contribute to the evolution of reproductive isolation.

Most of the data to date on intra-specific variation of egg coats comes from DNA sequence based analyses of selection (e.g. abalones, Aagaard et al. 2013; sea urchins, Pujolar & Pogson 2011; Vacquier & Swanson 2011; mammals, Turner & Hoekstra 2008b). One of the best characterized examples is the VE receptor for lysine (VERL) in abalones, which is responsible for gamete interaction and essential during fertilization (Aagaard et al. 2006, 2009, 2013). With regard to quantifying intra-specific variation in the molecular composition of jelly envelopes, one of the very few studies was done on jelly envelope mucins in X. laevis (Guerardel et al. 2000). In this study, a comparison of clutches indicated intra-specific polymorphism in O-glycans equivalent to that of human blood groups. However, this intra- specific polymorphism had no consequences for fertilization success (Guerardel et al. 2000), further indicating that – as in mammals – glycans may play a minor role in sperm-egg intractions (Gahlay et al. 2010; Avella et al. 2014). Given the multitude of functional roles of glycans, several alternative hypotheses for the functional significance of this jelly variation remain to be tested.

36

A role of egg coats in adaptive divergence and speciation?

Few studies have directly quantified intra-specific variation of egg coats at the phenotypic and functional level – and the consequences of this variation to diversification of natural populations. As stated above, this is a clear gap because egg coats are a prime source of adaptive maternal effects, with potential to allow evolutionary responses at ecological time scales (Räsänen & Kruuk 2007). An example of egg coat mediated adaptive divergence comes from our work on a ranid frog (moor frog, Rana arvalis) adapted to environmental acidification (Räsänen et al. 2003b; Hangartner et al. 2012). In this species, jelly manipulation experiments, combined with reciprocal quantitative genetic crosses among populations, showed that strong adaptive divergence in embryonic acid tolerance (survival) is mediated via the jelly envelopes (Räsänen et al. 2003b). This finding is similar to mechanisms suggested to mediate differences in embryonic acid tolerance between two species of Xenopus (X. laevis and X. gilli), inhabiting neutral/alkaline versus extremely acidic environments, respectively (Picker et al. 1993). The molecular composition and genetic basis underlying these adaptive maternal effects in R. arvalis are currently being explored (Shu et al., unpublished data). Similar studies in other taxa, and in relation to other putative selective factors, would shed light on the role of egg coat composition in adaptation.

Furthermore, egg coats can evolve rapidly and are an essential component of reproductive isolation in a range of taxa (Palumbi 2009). Their role in fertilization, gamete interactions and block to polyspermy has long been extensively studied as a species barrier (Wong & Wessel 2006) and, subsequently, in speciation (Coyne & Orr 2004; Palumbi 2009). Yet the integration of ecology in this process has been little considered to date. Here the dual role of egg coats in ecologically relevant functions (i.e. environment dependent effects on embryonic performance) and in sperm-egg interactions is of key importance.

First, if there is strong divergent natural selection on egg coats (e.g. different selective environments favoring a different molecular composition of egg coats), this might facilitate the evolution of reproductive isolation via adaptive divergence (i.e. ecological speciation, Turner & Hoekstra 2008a, Nosil 2012) between populations inhabiting different environments. Gene flow among populations could then be reduced either via direct viability selection against immigrants (Nosil et al. 2005) - mediated by differential embryonic performance - or due to the disruption of locally adapted sperm and egg genotype combinations. On the other hand, the fundamental role of egg coats in fertilization may impose constrains on their continued evolution under natural selection if sperm-egg interactions are influenced by different selective forces acting on sperm and on the egg coats (Palumbi 2009; Aagaard et al. 2013).

A rare example linking intra-specific variation in egg coats (variation in FEs from the molecular to the functional level) with speciation comes from a recent study on the sea star (Patiria miniata) where a combination of molecular genetics tools (e.g. RNA seq based

37

analyses) with fertilization experiments showed that sperm-egg interactions have facilitated speciation between two clades (Hart et al. 2014). The putative driver of FE diversification in this system was suggested to be sexual conflict, although ecologically mediated divergence (i.e. via population density related sperm competition) remained an alternative explanation. Future studies incorporating the ecological functions of egg coats would increase our understanding to what degree egg coats contribute to adaptation and the evolution of ecological reproductive isolation. We argue that a full understanding of the evolution of egg coats needs studies incorporating both sperm-egg interactions and ecological effects on embryonic performance.

Research gaps and challenges

Although much work has been done in identifying the structural and molecular genetic basis of egg coat variation, there are important open questions to be addressed. Some of the core questions, as we see them, are: 1) How much intra-specific variation do egg coats harbor? 2) What is the functional role of the molecular variation of egg coats? 3) What role do egg coats play in ecological and evolutionary processes of natural populations?

Given the fundamental role of egg coats in reproductive success, the above questions are important in evolutionary ecology. However, they are difficult to answer without the use of modern molecular tools and more integrative approaches in order to understand the links between intra-specific variability, structure and function of egg coats. First, intra-specific variability of egg coats has to date been almost exclusively studied at the DNA sequence level - rather than at the phenotypic level. Yet, it is the phenotype (including the genetic and plastic components) that expresses the function and is the direct target of natural selection, at least in the short term. Therefore, DNA sequence variation alone is not sufficient to understand the evolutionary processes (Danchin et al. 2011; Houle et al. 2010) influencing egg coat variation. The extent of phenotype variation and the fitness consequences of this variation need to be understood as well. This has resulted in the variability of egg capsules across individuals, populations and different environments often being ignored.

In other words, evolutionary ecologists find the rabbit (above questions), but cannot catch it without the right tools (emerging molecular techniques), while molecular biologist have the tools, but they just simply cannot see the rabbit. We argue that this gap needs to be bridged in order to achieve a comprehensive understanding about egg coat variability within and among species. We next suggest evolutionary ecologists some insight into available tools and to introduce some emerging strategies bridging molecular biology with evolutionary ecology that can be applied in the study of egg coat variation.

Strategies and molecular tools for inter-disciplinary studies

Modern molecular techniques make it possible to identify and quantify different types of egg coat variations. However, as most population level data on molecular variation of egg coats is

38

at the DNA level, we are still often missing the link between the genes, the environment and the phenotype (Diz et al. 2012). This is important since variation in gene expression can directly cause variations in proteins and, subsequently, in the phenotype and function (Danchin et al. 2011). Second, extensive post-translational modifications (PTMs), such as glycosylation and phosphorylation (Carr 1997; Diz et al. 2012), may cause alterations in the proteins coded by egg coat genes and make it often almost impossible to infer functional consequences based solely on DNA sequences (Danchin et al. 2011). This is particularly important for the highly glycosylated egg coat glycoproteins. Moreover, from an evolutionary ecological point of view, phenotypic plasticity (in case of egg coats likely mediated as transgenerational plasticity via the mother; Agrawal et al. 1999) often affects organismal responses to environmental heterogeneity (Ghalambor et al. 2007). However, to our knowledge, little is known about environment-dependent egg coat variability. Therefore, techniques which can directly study molecular variation of egg coats at the phenotypic level are highly desired. This is particularly true for egg coats as the functional consequences most often directly arise from variation in their molecular phenotype.

Recent advances in mass spectrometry (MS) and molecular techniques, such as proteomics and glycomics (Lazar et al. 2011; Diz et al. 2012; Jensen et al. 2012), allow identifying egg coat glycoproteins, as well as quantifying their variation. For example, with a “bottom-up” approach (from egg coat to amino acid sequence rather than amino acid sequence to egg coat), the proteome structures of FEs have been identified in the hamster (Aviles et al. 2009), rabbit (Stetson et al. 2012) and chicken (Mann 2008). Such quantification of protein variation, and identification of the underlying genes linked to this variation, is relatively well established and straightforward (Box 2). However, a large part of the molecular phenotype of egg coats – especially the jelly - consists of sugars, which are much more complex to analyze (Varki et al. 2009; Hart & Copeland 2010). Therefore, compared to the relatively well-studied FE proteins, the composition and structure of egg coat glycans still largely remains unknown in most taxa, let alone the extent of intra-specific variation or the genetic basis of this variation.

Fortunately, glycan profiles can now be identified and compared with the aid of MS (Box 3), as has been done for jelly envelopes in several amphibian species (Guerardel et al. 2000; Delplace et al. 2002; Strecker et al. 2003; Li et al. 2011). In some cases other techniques, such as nuclear magnetic resonance (NMR), are also needed to acquire comprehensive structural information of glycans. Progress in high resolution X-ray crystallography, which presents the atomic architecture of proteins, makes it possible to disentangle how glycans bind with amino acid sites, as well as their functions in egg coats (Monne et al. 2008; Han et al. 2010). However, to make rigorous inferences about the evolutionary ecological processes influencing egg coat diversification, a combination of various molecular techniques coupled with functional performance tests are needed (Turner & Hoekstra 2008).

39

Box 3. Structural analysis of egg coat glycans

Glycans perform key biological functions in organismal cells (Varki et al. 2009; Hart & Copeland 2010) and represent dynamic and complex structures with variable amounts and types of saccharides, arranged in multiple branches (antenna). However, due to the large variability in the glycan structures, and variation in their biosynthesis pathways, they are challenging to analyze (Varki et al. 2009; Hart & Copeland 2010). As the analytical details are beyond the focus of this review, we here shortly summarize the key points and we advice the reader to consult recent empirical work on egg coats (Li et al. 2011) and generic reviews for glycan analyses (Varki et al. 2009; Jensen et al. 2012).

The two major types of oligosaccharides are named N-linked and O-linked glycans, according to their point of attachment to proteins. Typically, N linked glycans are a common feature of FEs, wheresa O-linked glycans are a major component of jelly envelopes (Wong & Wessel 2006; Hedrick 2008). N-linked glycans can be detached from their protein backbone by specific enzymes, such as PNGase F (Jensen et al. 2012). In contrast, there is no universal enzyme for O-glycan release, and therefore chemical methods (e.g. β-elimination) are often used when analyzing O-linked glycans (Packer et al. 2010; Jensen et al. 2012). Following a chromatographic separation, structural analysis of glycans is usually done by NMR or MS. NMR spectroscopy is a powerful tool for de novo structural characterization (Lundborg & Widmalm 2011), while MS based methods can perform high-throughput glycomic profiling (Packer et al. 2010; Jensen et al. 2012). Once detailed structural information of glycans is acquired (Figure Box 3), the underlying genes and biosynthesis pathways can be infered, thus allowing inference of the genetic basis of glycan variation (Nairn et al. 2008; Varki et al. 2009).

Figure Box 3. Schematic representation of the basic steps in the workflow of the structural analysis of egg coat glycans. In the initial step, glycans are separated enzymatically or chemically (depending on the type of chemical linkage), from their protein backbone. This is then followed by MS or NMR analyses, which allow to identify the particular glycans. This part is particularly relevant to jelly envelopes, given its high amount of glycans.

40

Evolutionary and functional analysis of egg coats

To detect selection on egg coat genes (i.e. genes coding for egg coats and variation in them), comparisons of the dN/dS ratio of the underlying DNA sequences have been applied. For example, to infer selection on ZP genes in cetaceans (Amaral et al. 2011), sea urchins (Palumbi 2009), rodents (Turner & Hoekstra 2008b) and humans (Männikkö et al. 2005). Traditional methods for the functional analyses of egg coats have focused on biochemical approaches to identify the role of egg coats during fertilization (e.g. Segall & Lennarz 1979; Bonnell et al. 1996). When combined with experiments that allow establishing fitness consequences of egg coat variation in different ecologically relevant contexts, molecular techniques can allow establishing the links among molecular variation, function and fitness within and between species. Such experiments should include rearing of embryos from different populations under a range of ecological relevant standardized conditions (e.g. Räsänen et al. 2003a; Hangartner et al. 2012), testing for performance of different egg coat variants in the wild (e.g. Linnen et al. 2013) and/or experimental evolution approaches (e.g. Zhou et al. 2012). Ultimately, it may be possible to infer how egg coat evolution is influenced by both sperm-egg interactions and natural selection acting via embryonic performance.

Currently, one of the big remaining challenges in applying –omics approaches is the availability of databases. This is particularly true when working on non-model species for which genomes have not yet been sequenced – as is the case for most taxa of interest in evolutionary ecology. However, the field is developing fast and there are now several methods to overcome this challenge (Box 2). One of the powerful approaches is RNA sequencing (RNA-seq) (De Wit et al. 2012). Tissue specific transcriptomics (i.e. whereby tissue to be analyzed is selected depending on the site of egg coat production; Box 1) would allow to identify and quantify transcriptomes from both model and non-model species and discover novel genes and pathways involved in the biosynthesis of egg coats. Finally, in some model systems, modern molecular genetics techniques, such as gene knock-in and knock-out experiments, provide powerful tools establishing the functional role of observed genetic variation in egg coats. In this line, knock-out studies in mice have helped to understand the role of ZP2 and ZP3 genes (Rankin et al. 2003; Gahlay et al. 2010) but have yet to be applied in this context to other taxa. Generally, recent work using transcription activator-like effector nucleases (TALENs; reviewed in Young and Sanders 2012) in taxa such as Xenopus (Lei et al. 2012) and medaka (Ansai et al. 2013) indicates promising avenues for genetic manipulations.

Concluding remarks

Above we summarized the current progress on egg coat studies, as well as research gaps and challenges. We emphasize that inter-disciplinary studies, linking different molecular and evolutionary ecological experimental approaches, are needed to quantify intra-specific variation in egg coats and the consequences of this variation for organismal fitness. In doing

41

so we can make substantial progress in understanding the underlying factors behind the immense variability in egg coat structures and molecular composition across and within taxa and their role in evolution of biological diversity - in particular as it relates to maternal effects, adaptation and speciation.

Acknowledgements

This work was supported by a grant (31003A_133042) from the Swiss National Science foundation (to KR).

Author contributions

KR had the initial idea for the manuscript, LS wrote the first draft of the manuscript, prepared figures and tables and extracted gene data from GENEBANK, and MJFS provided expertises on molecular analyses. All authors contributed to the writing of the manuscript.

Glossary

De novo sequencing: peptide sequencing performed without prior knowledge of the amino acid sequence. Often performed in non-model species. dN/dS ratio: the ratio of non-synonymous coding sequence substitutions at non-synonymous sites (dN) to synonymous (silent) coding sequence substitutions at synonymous sites (dS) (Hurst 2002). A ratio greater than one indicates positive selection; less than one implies purifying selection; and a ratio of one indicates neutral selection. Egg coat: the complete extracellular structure surrounding the embryo. These structures are produced by the mother and typically consist of glycoproteins. (See Box 1 for a detailed description). Fertilization: the fusion of gametes to produce a new organism. This process can be either internal (within the body of the female) or external (outside the body of the female). Egg coats play a fundamental role in sperm-egg interactions. Fertilization envelope (FE): a glycoprotein membrane surrounding the plasma membrane of a zygote. Glycome: the entire set of sugars of a given cell or organism. Glycomics: the comprehensive study of all glycan structures of a given glycome. Glycoprotein: A protein that contains covalently attached oligosaccharide chains. Glycosylation: An enzymatic process that attaches glycans to proteins, lipids, or other organic molecules. Jelly envelope (JE): A thick sticky gelatinous structure surrounding the FE in some taxa, such as fish, sea urchins and amphibians. JEs are highly glycosylated (i.e. consist primarily of oligosaccarides).

42

Knock-in and knock-out techniques: A genetic engineering method that allows inserting a protein coding cDNA sequence into a particular locus of a species (knock-in) or allows making a given gene inoperative (knock-out). Mass spectrometry (MS): An analytical technique that measures the mass-to-charge ratio of charged molecules and its fragments (MS/MS). It is widely used to identify chemical structures and peptide sequences and to quantify organic molecules. Maternal effects: the effect of a mother’s phenotype and environment on offspring phenotype and performance (Mousseau & Fox 1998). Maternal effects can either be environmentally induced or have a genetic basis (Räsänen & Kruuk 2007). Proteome: the entire set of proteins of a given cell or organism. Proteomics: The large-scale study of proteins, including their structures, functions and interactions. Pseudogenes: dysfunctional genes that have lost their protein-coding ability or are no longer expressed during evolution. PTM (post-translational modification): the chemical modification (e.g., glycosylation, phosphorylation) of a protein after its translation. Vitelline envelope (VE): a glycoprotein membrane surrounding the plasma membrane of an unfertilized egg. X-ray crystallography: a method of determining the atomic and molecular structure of a crystal. Zona pellucida (ZP) domain: a family of evolutionarily related proteins. ZP glycoproteins share a common structural motif, known as the ZP domain.

43

References

Aagaard JE, Springer SA, Soelberg SD, Swanson WJ (2013) Duplicate abalone egg coat proteins bind sperm similarly, but evolve oppositely, consistent with molecular mimicry at fertilization. Plos Genetics 9, e1003287. Aagaard JE, Vacquier VD, MacCoss MJ, Swanson WJ (2009) ZP domain proteins in the abalone egg coat include a paralog of VERL under positive selection that binds lysin and 18-kDa sperm proteins. Molecular Biology and Evolution 27, 193-203. Aagaard JE, Yi X, MacCoss MJ, Swanson WJ (2006) Rapidly evolving zona pellucida domain proteins are a major component of the vitelline envelope of abalone eggs. Proceedings of the National Academy of Sciences 103, 17302-17307. Agrawal AA, Laforsch C, Tollrian R (1999) Transgenerational induction of defenses in animals and plants. Nature 401, 60-63. Altig R, McDiarmid RW (2007) Morphological diversity and evolution of egg and clutch structure in amphibians. Herpetological Monographs 21, 1-32. Amaral AR, Moller LM, Beheregaray LB, Coelho MM (2011) Evolution of 2 reproductive proteins, ZP3 and PKDREJ, in Cetaceans. Journal of Heredity 102, 275-282. Aviles M, Izquierdo-Rico MJ, Jimenez-Movilla M, et al. (2009) Hamster zona pellucida is formed by four glycoproteins: ZP1, ZP2, ZP3, and ZP4. Journal of Proteome Research 8, 926-941. Berois N, Arezo MJ, Papa NG (2011) Gamete interactions in teleost fish: the egg envelope. Basic studies and perspectives as environmental biomonitor. Biological Research 44, 119-124. Bonnell BS, Keller SH, Vacquier VD, Chandler DE (1994) The sea-urchin egg jelly coat consists of globular glycoproteins bound to a fibrous fucan superstructure. Developmental Biology 162, 313- 324. Bonnell BS, Reinhart D, Chandler DE (1996) Xenopus laevis egg jelly coats consist of small diffusible proteins bound to a complex system of structurally stable networks composed of high-molecular- weight glycoconjugates. Developmental Biology 174, 32-42. Carr SA (1997) Post-translational modifications of proteins. Faseb Journal 11, A1121-A1121. Claw KG, Swanson WJ (2012) Evolution of the egg: new findings and challenges. In: Annual Review of Genomics and Human Genetics, Vol 13 (eds. Chakravarti A, Green E), pp. 109-125. Annual Reviews, Palo Alto. Coyne JA, Orr HA (2004) Speciation. Sinauer, Sunderland, MA. Danchin E, Charmantier A, Champagne FA, et al. (2011) Beyond DNA: integrating inclusive inheritance into an extended theory of evolution. Nature Reviews Genetics 12, 475-486. Dancik V, Addona TA, Clauser KR, Vath JE, Pevzner PA (1999) De novo peptide sequencing via tandem mass spectrometry. Journal of Computational Biology 6, 327-342. De Wit P, Pespeni MH, Ladner JT, et al. (2012) The simple fool's guide to population genomics via RNA-Seq: an introduction to high-throughput sequencing data analysis. Molecular Ecology Resources 12, 1058-1067. Delplace F, Maes E, Lemoine J, Strecker G (2002) Species specificity of O-linked carbohydrate chains of the oviducal mucins in amphibians: structural analysis of neutral oligosaccharide alditols released by reductive beta-elimination from the egg-jelly coats of Rana clamitans. Biochemical Journal 363, 457-471. Denker HW (2000) Structural dynamics and function of early embryonic coats. Cells Tissues Organs 166, 180-207.

44

Diz AP, Martínez-Fernández M, Rolán-Alvarez E (2012) Proteomics in evolutionary ecology: linking the genotype with the phenotype. Molecular Ecology 21, 1060-1080. Dunson WA, Connell J (1982) Specific inhibition of hatching in amphibian embryos by low pH. Journal of Herpetology 16, 314-316. Ebert DA, Davis CD (2007) Descriptions of skate egg cases (Chondrichthyes : Rajiformes : Rajoidei) from the eastern North Pacific. Zootaxa 1393, 1-18. Edginton AN, Rouleau C, Stephenson GR, Boermans HJ (2006) 2,4-D butoxyethyl ester kinetics in embryos of Xenopus laevis: The role of the embryonic jelly coat in reducing chemical absorption. Archives of Environmental Contamination and Toxicology 52, 113-120. Forne I, Abian J, Cerda J (2010) Fish proteome analysis: Model organisms and non-sequenced species. Proteomics 10, 858-872. Gahlay G, Gauthier L, Baibakov B, Epifano O, Dean J (2010) Gamete recognition in mice depends on the cleavage status of an egg's zona pellucida protein. Science 329, 216-219. Ghalambor CK, McKay JK, Carroll SP, Reznick DN (2007) Adaptive versus non-adaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Functional Ecology 21, 394-407. Gomez-Mestre I, Touchon JC, Warkentin KM (2006) Amphibian embryo and parental defenses and a larval predator reduce egg mortality from water mold. Ecology 87, 2570-2581. Goudet G, Mugnier S, Callebaut I, Monget P (2008) Phylogenetic analysis and identification of pseudogenes reveal a progressive loss of zona pellucida genes during evolution of vertebrates. Biology of Reproduction 78, 796-806. Guerardel Y, Kol O, Maes E, et al. (2000) O-glycan variability of egg-jelly mucins from Xenopus laevis: characterization of four phenotypes that differ by the terminal glycosylation of their mucins. Biochemical Journal 352, 449-463. Gunaratne HJ (2007) Modifications of acrosome reaction-inducing egg coat glycans. Trends in Glycoscience and Glycotechnology 19, 61-66. Han L, Monne M, Okumura H, et al. (2010) Insights into egg coat assembly and egg-sperm interaction from the x-ray structure of full-length ZP3. Cell 143, 404-415. Hangartner S, Laurila A, Räsänen K (2012) Adaptive divergence in moor frog (Rana arvalis) populations along an acidification gradient: Inferences from Qst-Fst correlations. Evolution 66, 867-881. Hart GW, Copeland RJ (2010). Glycomics Hits the Big Time. Cell 143, 672-676. Hart MW, Sunday JM, Popovic I, Learning KJ, Konrad CM (2014) Incipient speciation of sea start populations by adaptive gamete recognition evolution. Evolution 68, 1294-1305. Hawkins LE, Hutchinson S (1988) Egg capsule structure and hatching mechanism of Ocenebra erinacea (L) (Prosobranchia, Muricidae). Journal of Experimental Marine Biology and Ecology 119, 269- 283. Hawkins RD, Hon GC, Ren B (2010) Next-generation genomics: an integrative approach. Nature Reviews Genetics 11, 476-486. Hedrick JL (2008) Anuran and pig egg zona pellucida glycoproteins in fertilization and early development. International Journal of Developmental Biology 52, 683-701. Herrler A, Beier HM (2000) Early embryonic coats: Morphology, function, practical applications - An overview. Cells Tissues Organs 166, 233-246. Holmstrup M, Westh P (1995) Effects of dehydration on water relations and survival of lumbricid earthworm egg capsules. Journal of Comparative Physiology B-Biochemical Systemic and Environmental Physiology 165, 377-383.

45

Host E, Gabrielsen A, Lindenberg S, Smidt-Jensen S (2002) Apoptosis in human cumulus cells in relation to zona pellucida thickness variation, maturation stage, and cleavage of the corresponding oocyte after intracytoplasmic sperm injection. Fertility and Sterility 77, 511-515. Houle D, Govindaraju DR, Omholt S (2010) Phenomics: the next challenge. Nature Reviews Genetics 11, 855-866. Hurst LD (2002) The Ka/Ks ratio: diagnosing the form of sequence evolution. Trends in Genetics 18, 486-487. Jensen PH, Karlsson NG, Kolarich D, Packer NH (2012) Structural analysis of N- and O-glycans released from glycoproteins. Nature Protocols 7, 1299-1310. Jovine L, Darie CC, Litscher ES, Wassarman PM (2005) Zona pellucida domain proteins. Annual Review of Biochemistry 74, 83-114. Kerney R, Kim E, Hangarter RP, Heiss AA, Bishop CD, Hall BK. (2011). Intracellular invasion of green algae in a salamander host. Proceedings of the National Academy of Sciences 108, 6497-6502. Kubo H, Kawano T, Tsubuki S, et al. (2000) Egg envelope glycoprotein gp37 as a Xenopus homolog of mammalian ZP1, based on cDNA cloning. Development Growth & Differentiation 42, 419-427. Lazar IM, Lazar AC, Cortes DF, Kabulski JL (2011) Recent advances in the MS analysis of glycoproteins: Theoretical considerations. Electrophoresis 32, 3-13. Li BS, Russell SC, Zhang JH, Hedrick JL, Lebrilla CB (2011) Structure determination by MALDI- IRMPD mass spectrometry and exoglycosidase digestions of O-linked oligosaccharides from Xenopus borealis egg jelly. Glycobiology 21, 877-894. Lindsay LL, Yang JC, Hedrick JL (2002) Identification and characterization of a unique Xenopus laevis egg envelope component, ZPD. Development Growth & Differentiation 44, 205-212. Linnen CR, Poh YP, Peterson BK, et al. (2013) Adaptive evolution of multiple traits through multiple mutations at a single gene. Science 339, 1312-1316. Litscher ES, Wassarman PM (2007) Egg extracellular coat proteins: From fish to mammals. Histology and Histopathology 22, 337-347. Lundborg M, Widmalm G (2011) Structural analysis of glycans by NMR chemical shift prediction. Analytical Chemistry 83, 1514-1517. Mann K (2008) Proteomic analysis of the chicken egg vitelline membrane. Proteomics 8, 2322-2332. Männikkö M, Törmälä RM, Tuuri T, et al. (2005) Association between sequence variations in genes encoding human zona pellucida glycoproteins and fertilization failure in IVF. Human Reproduction 20, 1578-1585. Marco-Jimenez F, Naturil-Alfonso C, Jimenez-Trigos E, Lavara R, Vicente JS (2012) Influence of zona pellucida thickness on fertilization, embryo implantation and birth. Animal Reproduction Science 132, 96-100. Marquis O, Miaud C (2008) Variation in UV sensitivity among common frog Rana temporaria populations along an altitudinal gradient. Zoology 111, 309-317. Marquis O, Millery A, Guittonneau S, Miaud C (2006) Toxicity of PAHs and jelly protection of eggs in the common frog Rana temporaria. Amphibia-Reptilia 27, 472-475. Menkhorst E, Selwood L (2008) Vertebrate extracellular preovulatory and postovulatory egg coats. Biology of Reproduction 79, 790-797. Meslin C, Mugnier S, Callebaut I, et al. (2012) Evolution of genes involved in gamete interaction: evidence for positive selection, duplications and losses in vertebrates. Plos One 7. Monne M, Han L, Jovine L (2006) Tracking down the ZP domain: from the mammalian zona pellucida to the molluscan vitelline envelope. Seminars in Reproductive Medicine 24, 204-216.

46

Monne M, Han L, Schwend T, Burendahl S, Jovine L (2008) Crystal structure of the ZP-N domain of ZP3 reveals the core fold of animal egg coats. Nature 456, 653-682. Mousseau TA, Fox CW (1998) Maternal effects as adaptations. Oxford University Press. Nairn AV, York WS, Harris K, Hall EM, Pierce JM, Moremen KW (2008). Regulation of glycan structures in animal tissues: transcript profiling of glycan-related genes. The Journal of Biological Chemistry 283, 17298-17313. Nosil P (2012) Ecological speciation Oxford University Press, Oxford, UK. Nosil P, Vines TH, Funk DJ (2005). Perspective: Reproductive isolation caused by natural selection against immigrants from divergent habitats. Evolution 59, 705-719. Packer NH, Jensen PH, Kolarich D (2010) Mucin-type O-glycosylation - putting the pieces together. Febs Journal 277, 81-94. Palumbi SR (2009) Speciation and the evolution of gamete recognition genes: pattern and process. Heredity 102, 66-76. Pang PC, Chiu PCN, Lee CL, et al. (2011) Human sperm binding is mediated by the sialyl-lewisx oligosaccharide on the zona pellucida. Science 333, 1761-1764. Pechenik JA (1979) Role of encapsulation in invertebrate life histories. The American Naturalist 114, 859-870. Picker MD, Mckenzie CJ, Fielding P (1993) Embryonic tolerance of Xenopus (Anura) to acidic blackwater. Copeia 1993, 1072-1081. Pinder AW, Friet SC (1994) Oxygen transport in egg masses of the amphibians Rana sylvatica and Ambystoma maculatum - Convection, diffusion and oxygen Production by algae. Journal of Experimental Biology 197, 17-30. Podolsky RD (2002) Fertilization ecology of egg coats: physical versus chemical contributions to fertilization success of free-spawned eggs. The Journal of Experimental Biology 205, 1657-1668. Podrabsky JE, Carpenter JF, Hand SC (2001) Survival of water stress in annual fish embryos: dehydration avoidance and egg envelope amyloid fibers. American Journal of Physiology- Regulatory Integrative and Comparative Physiology 280, R123-R131. Pujolar JM, Pogson GH (2011) Positive Darwinian selection in gamete recognition proteins of Strongylocentrotus sea urchins. Molecular Ecology 20, 4968-4982. Rankin TL, Coleman JS, Epifano O, et al. (2003) Fertility and taxon-specific sperm binding persist after replacement of mouse sperm receptors with human homologs. Developmental Cell 5, 33-43. Räsänen K, Kruuk LEB (2007) Maternal effects and evolution at ecological time-scales. Functional Ecology 21, 408-421. Räsänen K, Laurila A, Merilä J (2003a) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. I. Local adaptation. Evolution 57, 352-362. Räsänen K, Laurila A, Merilä J (2003b) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. II. Adaptive maternal effects. Evolution 57, 363-371. Rawlings TA (1993) Encapsulation of eggs by marine gastropods: Effect of variation in capsule form on vulnerability of embryos to predation. American Zoologist 33, 110A. Roche A, Maggioni M, Narvarte M (2011) Predation on egg capsules of Zidona dufresnei (Volutidae): ecological implications. Marine Biology 158, 2787-2793. Runft LL, Jaffe LA, Mehlmann LM (2002) Egg activation at fertilization: Where it all begins. Developmental Biology 245, 237-254. Salthe SN (1963) The Egg Capsules in the Amphibia. J Morphol 113, 161-171. Savitski MM, Nielsen ML, Kjeldsen F, Zubarev RA (2005) Proteomics-grade de novo sequencing approach. Journal of Proteome Research 4, 2348-2354.

47

Segall GK, Lennarz WJ (1979) Chemical characterization of the component of the jelly coat from sea urchin eggs responsible for induction of the acrosome reaction. Developmental Biology 71, 33- 48. Seymour RS (1994) Oxygen diffusion through the jelly capsules of amphibian eggs. Israel Journal of Zoology 40, 493-506. Smith J, Paton IR, Hughes DC, Burt DW (2005) Isolation and mapping the chicken zona pellucida genes: An insight into the evolution of orthologous genes in different species. Molecular Reproduction and Development 70, 133-145. Stetson I, Izquierdo-Rico MJ, Moros C, et al. (2012) Rabbit zona pellucida composition: A molecular, proteomic and phylogenetic approach. Journal of Proteomics 75, 5920-5935. Strecker G, Coppin A, Florea D, Maes E, Cogalniceanu D (2003) Comparative study of carbohydrate chains released from the oviducal mucins of the two very closely related amphibian species Bombina bombina and Bombina variegata. Biochimie 85, 53-64. Tadros W, Lipshitz HD (2009) The maternal-to-zygotic transition: a play in two acts. Development 136, 3033-3042. Touchon JC, Gomez-Mestre I, Warkentin KM (2006) Hatching plasticity in two temperate anurans: responses to a pathogen and predation cues. Canadian Journal of Zoology 84, 556-563. Tsang KY, Cheung MCH, Chan D, Cheah KSE (2010) The developmental roles of the extracellular matrix: beyond structure to regulation. Cell and Tissue Research 339, 93-110. Turner LM, Hoekstra HE (2008a) Causes and consequences of the evolution of reproductive proteins. International Journal of Developmental Biology 52, 769-780. Turner LM, Hoekstra HE (2008b) Reproductive protein evolution within and between species: maintenance of divergent ZP3 alleles in Peromyscus. Molecular Ecology 17, 2616-2628. Vacquier VD, Swanson WJ (2011) Selection in the rapid evolution of gamete recognition proteins in marine invertebrates. Cold Spring Harbor Perspectives in Biology 3. Varki A, Cummings RD, Esko JD, Hudson HF, Stanley P, Bertozzi CR, Hart GW, Etzler ME (Eds) (2009). Essentials of Glycobiology, Cold Spring Harbort Laboratory Press. Villalobos SA, Hamm JT, Teh SJ, Hinton DE (2000) Thiobencarb-induced embryotoxicity in medaka (Oryzias latipes): stage-specific toxicity and the protective role of the chorion. Aquatic Toxicology 48, 309-326. Wake MH, Dickie R (1998). Oviduct structure and function and reproductive modes in amphibians. The Journal of Experimental Zoology 282, 477-506. Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for transcriptomics. Nature Reviews Genetics 10, 57-63. Wassarman PM (2008) Zona pellucida glycoproteins. Journal of Biological Chemistry 283, 24285- 24289. Wassarman PM, Litscher ES, Williams Z (2009) Zona pellucida glycoprotein ZP3 and fertilization in mammals. Molecular Reproduction and Development 76, 933-941. Wong JL, Wessel GM (2006) Defending the zygote: Search for the ancestral animal block to polyspermy. Current Topics in Developmental Biology, Vol 72 72, 1-151. Yurewicz EC, Oliphant G, Hedrick JL (1975) Macromolecular composition of Xenopus laevis egg jelly coat. Biochemistry 14, 3101-3107. Zhou, S, Campbell TG, Stone EA, Mackay TDC, Anholt RRH (2012) Phenotypic plasticity of the Drosophila transcriptome. PLOS Genetics 8, e10025934.

48

Chapter II

Molecular phenotyping of maternally mediated parallel adaptive divergence

within Rana arvalis and Rana temporaria

Longfei Shu1, Anssi Laurila2, Marc J-F Suter3 and Katja Räsänen1

Affiliations:

1Eawag, Department of Aquatic Ecology, Switzerland and ETH Zürich, Institute of Integrative

Biology, Switzerland

2Animal Ecology/Department of Ecology and Genetics, Evolutionary Biology Center, Uppsala

University, Sweden

3Eawag, Department of Environmental Toxicology, Switzerland and ETH Zürich, Department of

Environmental Systems Science, Switzerland

49

Abstract

Environmental stress can be a powerful evolutionary force and often acts at ecological time scales. When similar selection acts on the same traits in multiple species or populations, parallel evolution can result in similar phenotypic changes but the underlying genetic architecture of parallel phenotypic divergence can be variable. Maternal effects can have strong effects on evolution at ecological time scales and facilitate rapid local adaptation; however, their contribution to parallel adaptive divergence is unclear. In this study, we i) reared embryos in a common garden experiment and ii) used a molecular phenotyping approach of egg coats, a putative driver of maternally mediated adaptive divergence in response to acidification, to investigate the molecular basis of parallel adaptive divergence to environmental acidification in two amphibian species (Rana arvalis and R. temporaria). Our results on three R. arvalis and two R. temporaria populations show adaptive divergence in embryonic acid tolerance mediated via maternally derived egg coats in both species. In addition, here we show for the first time that egg jelly coats exhibit extensive molecular polymorphism within species and identify several macromolecular components associated with embryonic acid tolerance. These data provide evidence for parallel mechanisms of adaptive divergence in two species. We highlight the importance of studying intraspecific variations in egg coats and molecular phenotyping in evolutionary ecology.

Keywords

Acid tolerance; Amphibian, Egg jelly; Maternal effects; Molecular phenotyping; Parallel evolution

50

Introduction

When similar selection acts on the same traits in multiple species or populations, parallel evolution can result in similar phenotypic changes (Pearse & Pogson 2000; Lee et al. 2011; Macqueen et al. 2011). However, the genetic architecture of parallel phenotypic changes may be similar or different (Arendt & Reznick 2008; Pearse et al. 2014). Parallel phenotypic evolution is usually assumed to arise through direct genetic effects (i.e. genetic variation of the organisms themselves) (Tennessen & Akey 2011; Hohenlohe et al. 2012; Pearse et al. 2014). However, parallel evolution may also take place without the aid of an individual’s own genotype (Kidd et al. 2012). This is the case, in particular, when evolutionary responses are mediated via parental effects (Uller 2008; Badyaev & Uller 2009), most commonly maternal effects (MEs). Maternal effects, defined as the effects of the mother’s phenotype or environment on offspring phenotype or performance (Mousseau & Fox 1998), can alter the speed and direction of evolution, allow rapid adaptation, and contribute to adaptive divergence of local populations (Mousseau & Fox 1998; Räsänen& Kruuk 2007). However, despite the important roles of MEs in rapid adaptation (Mousseau & Fox 1998; Räsänen & Kruuk 2007) and potential consequences for symmetry in gene flow among locally adapted populations (Wade 1998; Hangartner et al. 2012), their contribution to parallel evolution has been less well studied.

One important source of MEs during the early life-stages in many taxa are egg coats (i.e. the maternally derived extracellular structures surrounding the embryo; reviewed in Menkhorst & Selwood 2008; I). These structures are highly variable across taxa, and have important functional roles in sperm-egg interactions as well as pre-hatching and hatching related performance of offspring (Wong & Wessel 2006; Menkhorst & Selwood 2008). Egg coats can also act as a species barrier and influence speciation (Palumbi 2009; Hart et al. 2014), but only a handful of studies on egg coats have investigated intraspecific adaptive divergence in them (Räsänen et al. 2003b; Aagaard et al. 2006; Aagaard et al. 2013; reviewed in I). In many taxa, such as fish, echinoderms and amphibians, the outer layers of egg coats consist of a thick jelly envelope (Bonnell et al. 1994; Menkhorst & Selwood 2008). This egg jelly is typically highly variable in structure, composition and function among taxa (Segall & Lennarz 1979; Altig & McDiarmid 2007; Menkhorst & Selwood 2008). It can have major impact on fitness by providing adhesive medium for the embryo, interacting with the physical environment and providing protection against various biotic and abiotic environmental hazards (Podrabsky et al. 2001; Räsänen et al. 2003b; Gomez-Mestre et al. 2006; Marquis et al. 2006; Roche et al. 2011). However, despite these crucial roles of egg jelly, intra-specific variability and its role as a potential target of diversifying natural selection have largely been overlooked (Shu et al. I).

One potentially strong source of natural selection on egg jelly is environmental acidification. Acid stress has strong negative impacts on a broad range of taxa (Räsänen & Green 2009;

51

Kroeker et al. 2010; Stumpp et al. 2012). These negative fitness effects can cause strong selection on natural populations in many taxa (Lohbeck et al. 2012; Dam 2013; Pespeni et al. 2013), including amphibians (Räsänen et al. 2003a; Merilä et al. 2004; Räsänen & Green 2009; Hangartner et al. 2012; Egea-Serrano et al. 2014). In amphibians, embryos are generally surrounded by a variable number of jelly layers (Hedrick & Nishihara 1991; Altig & McDiarmid 2007; Hedrick 2008). When amphibian embryos are exposed to acidity, they typically show a “curling defect”, in which embryos may develop normally but become curled within the egg coats and finally fail to hatch (Connell 1978; Dunson & Connell 1982; Pierce 1985). This curling defect has been suggested to arise from alterations in the chemical composition of the egg coats under acidic conditions, which is apparent in egg coats becoming tight and sometimes changing color from transparent to opaque (Dunson & Connell 1982; Picker et al. 1993; Räsänen et al. 2003b). In two species of Xenopus (X. laevis and X. gilli) showing adaptive divergence in acid tolerance (Picker et al. 1993), jelly removal experiments indicated that inter-specific differences in embryonic acid tolerance are egg jelly related. Similarly, in the moor frog (Rana arvalis) (Räsänen et al. 2003b; Merilä et al. 2004; Persson et al. 2007), adaptive divergence among populations has been suggested to arise via egg jelly mediated maternal effects. Taken together, these findings suggests egg jelly mediated parallel adaptation to acid stress in amphibians. In a previous study (Brunold 2009), a combination of jelly removal experiments and reciprocal among population crosses showed that egg coat mediated maternal effects underlie parallel adaptive divergence in embryonic acid stress tolerance in R. arvalis as well as in the closely related but more acid sensitive common frog (R. temporaria). As the molecular basis of these egg jelly mediated adaptive maternal effects have not been previously investigated, we here return to these systems to explore the macromolecular underpinnings of adaptive maternal effects – and to investigate for the first time intra-specific variation in egg jelly.

Surprisingly little is known about intra-specific variation and the functional consequences of jelly composition (Shu et al. I). As egg jelly may often be a target of natural selection at early life-stages, and its fitness effects often are likely mediated via its molecular composition (Shu et al. I), understanding the evolutionary ecology of egg coats necessitates studying their intra- specific variation and the putative agents of selection influencing egg coat variability. Many past studies have mainly relied on de-jellying (i.e. removal of the jelly whilst retaining the innermost fertilization envelope intact) to investigate the ecological roles of egg jelly (Picker et al. 1993; Räsänen et al. 2003b; Edginton et al. 2006; Marquis et al. 2006). However, as jelly composition is usually visually indistinguishable among populations and even among species (Salthe 1963; Altig & McDiarmid 2007;), an alternative phenotyping method is needed to detect such variation in such a putatively “cryptic” phenotype.

Quantifying phenotypic variation is important because it is the raw material for selection to act upon (Houle et al. 2010). Therefore, phenotypic data are the most powerful predictors for fitness, the key interests of biological sciences (Houle et al. 2010). However, our ability to characterize phenotypes is surprisingly limited, especially at the molecular level (“cryptic” phenotype; Houle

52

et al. 2010), as compared to the ever-growing genomics sequencing data. While phenomics (large-scale phenotyping) approaches are at the moment highly demanded, additional molecular techniques are needed for describing “cryptic” phenotypes (Houle et al. 2010; Diz et al. 2012). We here use a proteomics approach (Diz et al. 2012) to quantify molecular variation in egg jelly of R. arvalis and R. temporaria. Previous work on Xenopus and other taxa indicates that amphibian egg coats mainly consist of glycoproteins (Yurewicz et al. 1975; Olson & Chandler 1999; Scillitani et al. 2011), and that egg jelly coat mainly consists of highly variable oligosaccharides attached to a protein backbone (Coppin et al. 1999; Strecker et al. 1999). This makes molecular phenotyping of egg jelly well suited for polyacrylamide gel electrophoresis (SDS-PAGE) (Laemmli 1970), in which macromolecular-composition of egg jelly can be analyzed by separating molecules according to their molecular weights.

In this study, we investigated the macromolecular basis of parallel adaptive divergence to acidity in three R. arvalis and two R. temporaria populations, breeding in ponds with contrasting pH, in southwestern Sweden. First, we test for adaptive divergence in embryonic acid tolerance in both species in a common garden laboratory experiment and, second, we perform molecular phenotyping (SDS-PAGE, Sodium dodecyl sulfate polyacrylamide gel electrophoresis) on all experimental clutches to quantify “cryptic” phenotypic variation in egg jelly and to gain first insight to the role of molecular variation in maternally mediated parallel divergence in embryonic acid tolerance. We predicted that 1) if environmental acidity drives parallel adaptive divergence in embryonic acid tolerance in these species, populations inhabiting acidic ponds should show higher embryonic acid tolerance within both species; 2) if there is intra-specific molecular variation in the jelly within species, we should see molecular variation in the “cryptic” egg jelly phenotypes among clutches using a molecular phenotyping approach and, finally, 3) if the macromolecular composition of egg jelly has parallel functional consequences (i.e. affect embryonic acid tolerance) in both species we should see an association between the molecular phenotype and embryonic survival in both species.

Materials and Methods

Study system

Both R. arvalis and R. temporaria are widely distributed anurans in the western Palearctic (Glandt 2006; Teacher et al. 2009). Both species inhabit a wide range of pHs, but R. temporaria is typically more acid sensitive and usually absent at sites where pH is below 4.5, while R. arvalis is more acid tolerant and can inhabit pH’s as low as pH 4 (reviewed in Räsänen & Green 2009). R. arvalis is one of the best-characterized study systems for adaptation to acidification and shows adaptive divergence in embryonic survival, larval traits, as well as maternal investment (Räsänen et al. 2003a, b; Räsänen et al. 2008; Hangartner et al. 2011; Egea-Serrano et al. 2014), while R. temporaria is less well studied (Glos et al. 2003; Brunold 2009). Three R. arvalis (Tottajärn, T, Bergsjö, B and Stubberud, S) and two R. temporaria populations (Bergsjö,

53

B and Stubberud, S) breeding in permanent ponds in southwestern Sweden were used in this study (details for study sites are provided in Hangartner et al. 2011). pH in the study ponds ranged from highly acidic (pH 4, site T) to intermediate (pH 6, site B) and to neutral (pH 7.5, S). Both R. arvalis and R. temporaria occur at sites B and S, but R. temporaria has no viable breeding population at site T. Site T is heavily affected by anthropogenic acidification since the early 1900s (Renberg et al. 1993), whereas site S has remained unaffected by acid rain due to limestone bedrock (Hangartner et al. 2011). Site B was heavily acidified until 1987 (lowest pH measured pH 4.2), but is since being limed regularly. It has a somewhat fluctuating pH around pH 6 (Annica Karlsson, pers. comm., Västragötaland county board).

During the breeding season of 2012, nine to eleven females and five males in breeding condition were collected from each of the three R. arvalis and two R. temporaria populations and transported to the laboratory at Uppsala University, Uppsala, Sweden (59°50`N, 17°50`E). The adults were kept in containers in moist Sphagnum moss in a climate chamber at 2- 4°C until artificial crosses were made a few days later.

Artificial crosses

As our focus was on maternal effects, we aimed to reduce variation due to the male contribution (whilst maximizing number of females) in estimates embryonic survival. We therefore aimed to cross two females per each of the males, creating paternal half-sibs and maternal full sibs within each of the populations. In three cases (B population of R. temporaria) one male was crossed with three females.

Artificial crosses were performed at +16°C according to standard procedures (see Räsänen et al. 2003a for details). In short, males and females were first anesthetized with tricaine methanosulfonate (MS222). The males testes were removed and sperm was collected by crushing the testes in 10% Amphibian Ringer Solution. Eggs, which were stripped from each female, were subsequently mixed with the sperm solution and treated using standard procedures, whereby eggs are allowed to swell in pH 7 artificial soft water RSW for two hours (Räsänen et al. 2003a; Hangartner et al. 2011). Fertilized eggs were subsequently divided into the treatments before the first cell division. Artificial mating minimized any bias due to variation of early environment.

Embryonic acid tolerance test

Embryonic acid tolerance of the two species was tested using standard procedures (Räsänen et al. 2003a; Hangartner et al. 2011). In short, embryos were reared in a walk-in climate room (16 °C) with 17L:7D photoperiod at three pH treatments (acid: pH 4.0 and pH 4.2 for R. arvalis and pH 4.2 and pH 4.4 for R. temporaria, neutral: pH 7.5 for both species). Reconstituted soft water (RSW), prepared in 120L tanks two days prior to use, was used as the experimental medium

54

(APHA 1985). The pH in the acid treatment was adjusted with 1M H2SO4, whereas the pH in the neutral treatment was not adjusted (RSW has a nominal pH of 7.2-7.6).

The embryonic acid tolerance experiments were performed at the same time for both species and as a nested randomized block design. As our focus is on within species divergence, and the treatments partially had to differ due to the innate differences in pH tolerance of these two species (Brunold 2009), we considered each species separately. For R. arvalis, the experimental design was 3 × 3 with three populations (T, B and S), three pH treatments (pH 4.0, 4.2 and 7.5) and seven to nine clutches (i.e. maternal full-sib, paternal half-sib families) per population. For R. temporaria, the experimental design was 2 × 3 with two populations (B and S), three pH treatments (pH 4.2, 4.4 and 7.5) and eight to ten clutches (i.e. maternal full-sib, paternal half-sib families) per population. Each family – pH treatment combination was replicated three times, resulting in a total of 414 experimental units. The replicates were fully randomized and blocked across three blocks on a shelf system. Each experimental unit consisted of 20 to 40 embryos from a given family placed in a plastic vial (0.9 L), containing 0.5 L of treatment water. Embryos were reared from fertilization to day 12 (when all surviving embryos should have hatched). Unfertilized eggs (i.e. eggs were considered unfertilized if no cell division was apparent) were determined at day 3 and excluded from the analyses of hatching success. Water was changed every third day to maintain appropriate pH and pH was measured with each water exchange (mean ± S.D. at pH 4.0: pH 3.97 ± 0.04, pH 4.2: 4.23 ± 0.03 and pH 4.4: 4.43 ± 0.03). The number of dead embryos was recorded visually at each water exchange, but any dead individuals were left untouched. Only final survival (day 12) was used in the statistical analyses. Survival was calculated as number of hatchlings/total number of fertilized eggs for each experimental unit.

Jelly sampling

Approximately 200 eggs from each female clutch in both species were collected for molecular phenotyping of egg jelly. Jelly was first manually removed using watchmaker’s forceps and subsequently solubilized in pH 8.9 DeBoers solution containing 45 mM β-mercaptoethanol (Hedrick & Hardy 1991). The innermost jelly layer was collected by placing a single layer of de- jellied eggs in a flat-bottomed dish and treating it with β-mercaptoethanol (Hedrick& Hardy 1991). The pooled solubilized jelly solution was then adjusted to pH 7 and stored at + 4 °C for later molecular analyses (Hedrick & Hardy 1991).

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE)

SDS-PAGE (Schagger & Vonjagow 1987) was used to quantify the macromolecular composition of the egg jelly. Ultracentrifugation (Vivacell 250, Sartorius) with a molecular weight cut-off of 5 kDa was used to increase the concentration of the jelly solution. Jelly sample protein concentration was estimated according to Bradford (1976) with bovine serum albumin as a standard. All jelly samples were diluted 1:2 in Laemmli sample buffer (62.5 mM Tris-HCl, pH

55

6.8, 10% glycerol, 2% SDS, 5% 2-mercaptoethanol, 0.05% bromophenol blue) and boiled for 5 minutes. Gels were run under denaturing conditions in 5% stacking gel and 10% separation gel using standard procedures, during which the molecules were separated according to their molecular weight (Laemmli 1970). Gels were first run at 90V when the proteins passed through the stacking gel. The voltage was then increased to 200 V until the bromophenol blue dye had migrated to the bottom of the gel (Laemmli 1970). Three technical replicates were run for each clutch and all samples were randomly distributed in the gels to avoid any bias from gel effects.

Imaging and phenotyping

After electrophoresis, the gels were stained with Coomassie blue R250 (0.1% R250, 10% acetic acid, 50% methanol and 40% H2O) for 3 hours, de-stained by soaking in de-staining solution (10% acetic acid, 50% methanol and 40% H2O) until the background was nearly clear and scanned using a Bio-Rad GS-800 densitometer. The scans measured the optical densities (O.D.) of each band, which were further processed by Quantity-One software (Bio-rad). O.D. represents the presence/absence of a band as well as abundance of a given glycoprotein. The molecular mass of each band was calculated with a standard protein marker using Quantity-One software (Bio-rad). In total, 132 jelly samples were phenotyped, including 24 R. arvalis (S: 9, B: 7 and T: 8) and 20 R. temporaria clutches (S: 10 and B: 10).

Statistical analyses

Embryonic survival

Embryonic survival was analyzed with a generalized linear mixed model (GLMM) with binomial errors and logit link function in the GLIMMIX procedure of SAS 9.3 (SAS Institute, Inc.). In these analyses, pH treatment, population and their interaction were used as fixed factors and family (nested within population), and its interaction with pH treatment as random factors. Although our data consisted of (for the most part) maternal full-sib, paternal half-sib clutches, we analyzed the data on the female identity level to reduce complexity of the otherwise too complex models (In addition, males had no significant effects to the full models, results not shown). Block was used as a fixed effect to control for a known temperature gradient across the experimental shelves. Relevant pairwise comparisons were conducted using Tukey's Post Hoc Tests.

Jelly phenotype

In the jelly phenotype analyses, the mean of the three jelly replicates per female was used as the observational unit. Jelly phenotype was presented as O.D. reads from each band for a given jelly sample. To avoid any bias due to variation of sample amount loading on gels, O.D. read of each band on a given gel was standardized by dividing by the total O.D. of all bands on a given gel, therefore each band was presented as a proportion of the total. To visualize the molecular

56

variation of egg jelly, one-way unsupervised hierarchical clustering was performed using gplots and heatmap.2 packages in R (R Core Team 2014).

Discriminant analysis of principal components (DAPC) (Jombart et al. 2010), a multivariate method implemented in the R package adegenet (Jombart 2008), was first used to cluster phenotypically similar clutches according to the macromolecular composition (O.D. read) of the jelly. This method first transforms the data using principal component analysis, which ensures that the variables are not correlated and that the number of variables is smaller than the number of individuals. A sequential K-means procedure is used to infer the optimal number of groups, minimizing the within-group variability (Jombart et al. 2010). The Bayesian information criterion (BIC) was used to assess the optimal number of groups. A discriminant analysis (DA) was then applied on these first results, summarizing the differentiation among groups and producing a visual assessment of this differentiation (Jombart & Ahmed 2011; Jombart et al. 2010). Because DAPC does not assume any underlying population genetic model and is not restrained to genetic data, it is applicable to the quantitative data on jelly phenotypes (Jombart et al. 2010).

Embryonic acid tolerance – jelly phenotype association

To visualize the relationships between the jelly phenotypes and embryonic acid tolerance, family means of embryonic acid tolerance were compared among jelly phenotypic clusters (as identified based on the DAPC analysis). Because survival of R. arvalis and R. temporaria showed strongest within species divergence at pH 4.0 and pH 4.2, respectively, we compared differences among the clusters in embryonic acid tolerance within these two pH treatments with generalized linear models, followed by Tukey's post-hoc tests. To further investigate which specific bands contributed to embryonic acid tolerance, variable contributions were interpreted using the “loadingplot” function in the R package adegenet (Jombart & Ahmed 2011).

Results

Embryonic acid tolerance

In general, the acidic treatments reduced embryonic survival in both species by 20 to more than 90 % (Fig. 1), whereas embryonic survival in all populations was generally high at pH 7.5. However, the negative effects of pH differed between species and populations (Fig.1, Table 1 & 2). In R. arvalis, T embryos had ca. 20% survival at pH 4.0 and 80 % at pH 4.2, whilst the B and S embryos had 85 % and 55 % survival at pH 4.2, respectively, and almost no survivors at pH 4.0 (Fig. 1). In R. arvalis, however, the pH treatment × population interaction was not significant (Table 1). In R. temporaria, a significant pH treatment × population interaction (Table 2) indicated that populations differed in the responses to acid stress: embryos of B population had ca. 90 % survival at pH 4.4 and 85 % at pH 4.2, whereas embryos of S population had 85 % and 60 % survival, respectively (Fig. 1).

57

Strong within-population variation in acid tolerance was indicated by the significant family × pH treatment interactions in both species (Table 1 & 2). This was also evident in the variability in family level reaction norms across the different pHs, whereby the relative proportion of relatively acid tolerant and relatively acid sensitive clutches differed among populations (Fig. S1 & S2). In R. arvalis, in the T population embryonic survival ranged from 10% up to 50% at pH 4.0 while the other two populations had near zero to 18% survival in this treatment (Fig. S2). In R. temporaria, at pH 4.2 in the B population embryonic survival ranged from ca. 65 to 95 %, while in the S population variation was larger, ranging from almost zero to over 85 % at this pH (Fig. S1).

Figure 1 Embryonic survival (mean ± S.E.) of A) three R. arvalis populations (T, B and S) under three pH treatments (pH 4.0, pH 4.2 and pH 7.5) and B) two R. temporaria populations (B and S) under three pH treatment (pH 4.2, pH 4.4 and pH 7.5).

58

Table 1 Generalized linear mixed model of embryonic survival in R. arvalis. Results are shown for three populations (B, S and T) under three pH treatment (pH 4.0, pH 4.2 and pH 7.5). Significant effects are highlighted in bold.

Random effects Variance ± SE Z P Family (Population) 0.39 ± 0.22 1.73 0.041 pH Treatment × Family (Population) 0.66 ± 0.19 3.36 0.000 Fixed effects ndf ddf F P pH Treatment 2 42 185.43 0.000 Population 2 21 7.82 0.002 pH Treatment × Population 4 42 1.12 0.360 Block 2 137 0.94 0.394 Contrasts Variance ± SE t P pH Treatment (4.0 vs. 4.2) -4.05 ± 0.28 -14.10 0.000 pH Treatment (4.0 vs. 7.5) -5.68 ± 0.30 -18.75 0.000 pH Treatment (4.2 vs. 7.5) -1.62 ± 0.27 -5.84 0.000 Population (T vs. S) -1.57 ± 0.39 -3.95 0.002 Population (T vs. B) -0.78 ± 0.44 -1.76 0.208 Population (B vs. S) 0.79 ± 0.44 1.79 0.198

Table 2 Generalized linear mixed model of embryonic survival in R. temporaria. Results are shown for two populations (S and B) and under three pH treatment (pH 4.2, pH 4.4 and pH 7.5). Significant effects are highlighted in bold.

Random effects Variance ± SE Z P Family (Population) 0.46 ± 0.19 2.37 0.000 pH Treatment × Family (Population) 0.23 ± 0.08 2.73 0.003 Fixed effects ndf ddf F P pH Treatment 2 36 9.89 0.000 Population 1 18 2.62 0.123 pH Treatment × Population 2 36 7.63 0.001 Block 2 119 0.63 0.537 Contrasts Variance ± SE t P pH Treatment (4.2 vs. 4.4) -0.64 ± 0.19 -3.38 0.005 pH Treatment (4.4 vs. 7.5) -0.15 ± 0.20 -0.76 0.726 pH Treatment (4.2 vs. 7.5) -0.79 ± 0.19 -4.15 0.000

59

Molecular variation of egg jelly

With regard to banding patterns (Fig. 2), SDS-PAGE analyses indicated that, in both species, egg jelly consisted of multiple bands, with a molecular weight ranging from 10 kD to 185 kD. In R. arvalis, eight major bands (20 kD, 25 kD, 34 kD, 48 kD, 60 kD, 80 kD, 134 kD and 185 kD) were detected, with two high molecular weight bands (134 kD and 185 kD) and one low molecular weight band (25 kD) being highly expressed (Fig. 2). In R. temporaria, ten major bands (10 kD, 20 kD, 25 kD, 34 kD, 48 kD, 60 kD, 80 kD, 116 kD, 134 kD and 185 kD) were detected, of which only the 34 kD band was highly expressed (Fig. 2). In addition, in R. temporaria two bands (10 kD and 116 kD) were detected that were absent in R. arvalis, although the 116 kD band was also absent in several R. temporaria clutches (Fig. 2). Three clutches (one R. arvalis: S6_A; two R. temporaria: B3_T and B4_T) showed a distinct pattern compared to the other families (Fig. 2) by having strong bands in the low molecular weight region (25 kD to 60 kD) and a lack of bands in the high molecular region.

Figure 2 Molecular variation of jelly. Molecular weights of each band are shown on the X axis and family IDs on the Y axis. One-way unsupervised hierarchical clustering of phenotypic data from 44 jelly samples shows that all jelly samples could be classified into two major groups (R. arvalis and R. temporaria) which match with the species identity (Fig. 3).

60

Multivariate analysis of jelly phenotype

For the complete data set, the sequential K-means procedure (associated with Bayesian Information Criterion (BIC)), indicated seven phenotypic clusters (Fig. S3). This multivariate analyses further indicated that the three distinct clutches (see above) clustered together (cluster 5, Fig. S3), and were strongly separated from the rest of the jelly samples (Fig. S3). As these clutches represent putative outliers and were strongly deviating, we excluded these three clutches from the following analyses.

In the reduced data set, five phenotypic clusters were identified using BIC and the sequential K- means procedure. DAPC identified two phenotypic clusters that were separated along the first axis (Fig. 3): all R. temporaria samples clustered on the left side of the axis (clusters 1 and 5), while almost all R. arvalis samples clustered on the right (clusters 2, 3 and 4). This indicated that axis 1 reflects clearly species boundaries in jelly macromolecular-composition. Interestingly, within both species, jelly clusters (R. temporaria clusters 1 and 5; R. arvalis clusters 2, 3 and 4) showed divergence along the second axis (Fig. 3), suggesting intra-specific divergence of jelly phenotypes within both species.

Figure 3 Scatterplot of the discriminant analysis of principal components (DAPC) on jelly phenotypes. The first axis is the horizontal axis. Colors represent the 5 phenotypic groups found by the K-means and BIC method. At the bottom left, the cumulated variances explained by number of PCA are represented.

61

Association between jelly composition and embryonic acid tolerance

Within each species, the different clusters differed in embryonic acid tolerance. In R. arvalis, clusters differed in their acid tolerance (F2,21 = 5.78, P = 0.011, Fig. 4), such that embryos from cluster 3 had higher survival under acid stress than cluster 4 (Tukey test, P = 0.014), but there was no difference between cluster 3 and cluster 2 (Tukey test, P = 0.071). Similarly within R. temporaria, embryos from cluster 5 had higher acid tolerance than those from cluster 1 (F1,17 = 4.96, P = 0.022, Fig. 4). This shows that differences in acid tolerance of the embryos are clearly identified along the second axis of the molecular phenotype, and that this pattern is consistent in both species (Fig. 3 & 4).

These results indicate that the total variation among the clusters could be explained by two main axes: species boundaries on the first axis and embryonic acid tolerance on the second axis (Fig. 3). “Loadingplot” analyses further indicated that the first axis, which represents species differences, is mainly explained by bands of 34 kD, 20 kD and 10 kD (Fig. 5), while the second axis, which is associated with embryonic acid tolerance, is explained by bands of 185 kD, 25 kD and 20 kD (Fig. 5).

Figure 4 Embryonic acid tolerance (i.e. mean ± S.E. survival of embryos under acidic pH treatment; pH 4.2 for R. temporaria and pH 4.0 for R. arvalis) for each jelly cluster in both species.

62

Figure 5 Loading plot of relative contributions of the different molecular bands to A) species boundary; and B) embryonic acid tolerance. Molecular weights are presented in the X axis while their loadings are presented in the Y axis.

63

Discussion

MEs can play an important role in rapid evolution and local adaptation (Wade 1998; Mousseau & Fox 1998; Räsänen & Kruuk 2007). However, their contribution to parallel evolution is rarely investigated. Our study provides rare evidence for parallel divergence via maternal effects at the molecular level in general, and in embryonic acid tolerance of R. arvalis and R. temporaria more specifically. Molecular phenotyping of egg jelly further indicated that this parallel divergence arises from the variation in macromolecular-composition of egg jelly and provides first rigorous evidence for intra-specific variation in egg jelly coats (Shu et al., I).

Maternal effects mediated parallel evolution

Intriguingly, for R. temporaria, our results are in line with previous work on R. arvalis that populations breeding in acidic sites have higher embryonic acid tolerance. This provides further evidence that environmental acidification acts as a strong selection force and can drive parallel adaptive divergence in natural populations (e.g. Räsänen et al. 2003a; Hangartner et al. 2011). Although for R. arvalis, our results for embryonic acid tolerance where somewhat contradictory as we found no statistical support for among population differences in acid tolerance (i.e. no significant pH x population interaction), our data was to a similar direction as several experiments show previously (i.e. T population having, on average, higher acid tolerance than more neutral populations), which together support the classical view that environmental stress is a powerful evolutionary force in generating local adaptations and rapid evolution (Hoffmann & Parsons 1997; Bijlsma & Loeschcke 2005).

MEs have been suggested to play an important role in rapid adaptation (Wade 1998; Mousseau & Fox 1998; Räsänen & Kruuk 2007). Interestingly, here we found that parallel adaptive divergence in embryonic acid tolerance within two amphibian species likely arises from variation in the macromolecular composition of the egg jelly, indicating parallel adaptive maternal effects in these species. Moreover, molecular phenotyping indicated that the macromolecules that contribute to acid tolerance were similar in both species (i.e. indicating by similar variation loading on the second axis in the DAPC analysis, Fig. 3). Our study here is the very first study indicating that MEs can mediate parallel divergence in macromolecular composition of egg coats in response to environmental stress.

MEs are often environmentally induced, but can have a genetic basis (Mousseau & Fox 1998; Räsänen & Kruuk 2007). With regard to egg jelly variation, little is known about the genetic or environmental basis. The specific genes underlying egg jelly production and composition are currently unknown, but as egg jelly coats are produced by the mother when eggs pass through the oviduct (Hedrick & Nishihara 1991; Wake & Dickie 1998; Altig & McDiarmid 2007), variation in egg jelly phenotypes should be determined by the MEs genes expressed in the

64

mother’s oviduct. Previous work in Xenopus indicates that the egg jelly mainly consists of highly variable oligosaccharides attached to a protein backbone (Coppin et al. 1999; Hedrick& Nishihara 1991; Strecker et al. 1999; Yurewicz et al. 1975). Furthermore, our own work strongly suggesting that glycosylation related genes expressed in the oviduct are likely candidate MEs genes (Shu et al. IV).

Molecular polymorphism of egg jelly

We found extensive molecular polymorphism in egg jelly at both inter- and intra-species levels (Fig. 2). The strong species-specific differences in the jelly macromolecular composition are in accordance with previous work in R. arvalis and R. temporaria (Coppin et al. 1999; Strecker et al. 1999), as well as in other taxa. For instance, early studies on X. laevis and X. tropicalis suggested that these two species have different jelly macromolecular-composition (Lindsay 2003). At the coarse level, jelly macromolecular-composition is relatively more similar within the genus Rana (this study) and within genus Xenopus (Lindsay 2003) than between these two genera, indicating that macromolecular composition of the jelly depends on the phylogenetic relatedness. Until now, however, phylogenetic studies on composition of egg coats have been conducted only on higher taxonomic levels and focusing on the genes coding for the innermost vitelline envelope (i.e. ZP genes; Spargo and Hope 2003; Goudet et al. 2008). Thus variation at finer taxonomic levels and in the jelly envelope is unexplored to date. To investigate the phylogenetic relationships of compositional variation in egg jelly coats, more amphibian species need to be studied.

We found extensive intra-specific polymorphism in macromolecular composition of egg jelly in both Rana species. This is, to our knowledge, the first study explicitly exploring intra-specific variation in egg envelope composition. The only other study aiming to look at intra-specific variation in egg jelly macromolecular composition was conducted on six laboratory-raised X. laevis females (Guerardel et al. 2000). We hypothesize that polymorphism in the macromolecular composition of the egg jelly may be universal and suggest that this may reflect a combination of directional and balancing selection.

First, strong positive selection on egg jelly composition may be caused by environmental stressors, such as acidification, that alters the jelly chemically with severe consequences for embryonic fitness (Shu et al. III and this manuscript). In this scenario, jelly composition may have strong fitness consequences in embryonic survival under acid stress. In particular, our results suggest that strong acid mediated selection may act on bands of 185 kD, 25 kD and 20 kD (Fig. 5), though the specific components that these bands reflect, and their detailed functional consequences, are yet to be established. Future work should investigate how environmental stress (such as acidity) induced positive selection affects standing molecular polymorphism of jelly.

On the other hand, balancing selection maintains allelic polymorphism in many processes, such as the self-incompatibility (sperm-egg interactions) system in plants and the major

65

histocompatibility complex (MHC, cell surface immune defense) loci (Piertney & Oliver 2006; Delph & Kelly 2014). Egg jelly plays highly diverse roles, ranging from sperm-egg interactions (Wong & Wessel 2006; Menkhorst& Selwood 2008) to protection against various biotic and abiotic environmental hazards (Olson & Chandler 1999; Watanabe & Onitake 2002; Räsänen et al. 2003b; Krapf et al. 2006; Marquis et al. 2006; Altig & McDiarmid 2007). Of particular relevance here are pathogens, which are highly spatially and temporally variable in nature and typically interact with glycans on cell surfaces (Varki 2011), possibly leading to selective interactions between pathogens and protein and glycan variants in the jelly. We suggest that jelly polymorphism may be maintained by a combination of positive and balancing selection.

Utility of emerging molecular tools in detecting “cryptic” phenotype

In this study, we quantified the molecular variation of egg jelly using a 1D polyacrylamide gel electrophoresis based approach. We detected a high level of intra-specific variation in egg jelly, and identified macromolecular components that likely have functional consequences. Our results highlight the necessity of quantifying variation in apparently “cryptic phenotypes”, such as egg jelly, which could be overlooked in a traditional approach. In this study, we were not yet able to characterize the detailed structure of the egg jelly glycans, and hence the actual components of importance in acid tolerance, but we emphasize the necessity to utilize emerging techniques such as glycomics (Dobson 2012; Hart & Copeland 2010) and transcriptomics (Hawkins et al. 2010) in future studies. Whilst glycomics approaches will allow characterizing the detailed glycan identity and variation, transcriptomics has the potential to identify the genes underlying variation in egg jelly composition – i.e. candidate MEs genes (Shu et al. IV), as well as investigate how gene expression patterns in the oviduct are affected by environmental acidity.

Phenotypic variation is the raw material for selection to act upon and reflects the selective history and the potential of natural populations to evolve in response to environmental change. These aspects were recently highlighted in the road map for molecular ecology (Andrew et al. 2013). We believe that acquiring detailed phenotypic data is as crucial as genetic data when trying to fully understand evolutionary processes. Multidisciplinary approaches, such as proteomics (Diz et al. 2012), glycomics (Hart & Copeland 2010) and transcriptomics (Hawkins et al. 2010), should be applied widely in future evolutionary studies for detecting variation in “cryptic phenotypes” as well as the functional consequences of this variation (Andrew et al. 2013; Houle et al. 2010).

Acknowledgements

We thank Beatrice Lindgren and Corinne Schenkel for invaluable help with the field and laboratory work. The experiments were conducted under permissions from the Västra Götaland county board and the Ethical committee for animal experiments in Uppsala County. This study was supported by the Swiss National Science foundation (to KR).

66

Author Contributions

LS and KR conceived and planned the study. LS performed the experiments. AL and MS participated in the conceptual development and discussion of the experiments. All authors discussed the results and commented on the manuscript. All authors read and approved the final manuscript.

References

Aagaard JE, Springer SA, Soelberg SD, Swanson WJ (2013) Duplicate abalone egg coat proteins bind sperm lysin similarly, but evolve oppositely, consistent with molecular mimicry at fertilization. Plos Genetics 9, e1003287. Aagaard JE, Yi X, MacCoss MJ, Swanson WJ (2006) Rapidly evolving zona pellucida domain proteins are a major component of the vitelline envelope of abalone eggs. Proceedings of the National Academy of Sciences 103, 17302-17307. Altig R, McDiarmid RW (2007) Morphological diversity and evolution of egg and clutch structure in amphibians. Herpetological Monographs 21, 1-32. Andrew RL, Bernatchez L, Bonin A, et al. (2013) A road map for molecular ecology. Molecular Ecology 22, 2605-2626. APHA (1985) Standard methods for the examination of water and wastewater. American Public Health Association, Washington, DC. Arendt J, Reznick D (2008) Convergence and parallelism reconsidered: what have we learned about the genetics of adaptation? Trends in Ecology & Evolution 23, 26-32. Badyaev AV, Uller T (2009) Parental effects in ecology and evolution: mechanisms, processes and implications. Philosophical Transactions of the Royal Society B-Biological Sciences 364, 1169-1177. Bijlsma R, Loeschcke V (2005) Environmental stress, adaptation and evolution: an overview. Journal of Evolutionary Biology 18, 744-749. Bonnell BS, Keller SH, Vacquier VD, Chandler DE (1994) The sea-urchin egg jelly coat consists of globular glycoproteins bound to a fibrous fucan superstructure. Developmental Biology 162, 313-324. Bradford MM (1976) Rapid and sensitive method for quantitation of microgram quantitites of protein utilizing principle of protein-dye binding. Analytical Biochemistry 72, 248-254. Browne RK, Seratt J, Vance C, Kouba A (2006) Hormonal priming, induction of ovulation and in-vitro fertilization of the endangered Wyoming toad (Bufo baxteri). Reproductive Biology and Endocrinology 4. Brunold C (2009) Acid stress tolerance and adaptive maternal effects of Rana arvalis and Rana temporaria. M. Sc thesis, ETH zurich. Coppin A, Maes E, Flahaut C, Coddeville B, Strecker G (1999) Acquisition of species-specific O-linked carbohydrate chains from oviducal mucins in Rana arvalis - A case study. European Journal of Biochemistry 266, 370-382.

67

Dam HG (2013) Evolutionary adaptation of marine zooplankton to global change. In: Annual Review of Marine Science, Vol 5 (eds. Carlson CA, Giovannoni SJ), pp. 349-370. Annual Reviews, Palo Alto. de Nadal E, Ammerer G, Posas F (2011) Controlling gene expression in response to stress. Nature Reviews Genetics 12, 833-845. Delph LF, Kelly JK (2014) On the importance of balancing selection in plants. New Phytologist 201, 45-56. Diz AP, Martinez-Fernandez M, Rolan-Alvarez E (2012) Proteomics in evolutionary ecology: linking the genotype with the phenotype. Molecular Ecology 21, 1060-1080. Dobson C (2012) High-throughput O-Glycomics. Glycobiology 22, 1608-1608. Dunson WA, Connell J (1982) Specific-inhibition of hatching in amphibian embryos by low pH. Journal of Herpetology 16, 314-316. Edginton AN, Rouleau C, Stephenson GR, Boermans HJ (2006) 2,4-D butoxyethyl ester kinetics in embryos of Xenopus laevis: the role of the embryonic jelly coat in reducing chemical absorption. Archives of Environmental Contamination and Toxicology 52, 113-120. Egea-Serrano A, Hangartner S, Laurila A, Räsanen K (2014) Multifarious selection through environmental change: acidity and predator-mediated adaptive divergence in the moor frog (Rana arvalis). Proceedings of the Royal Society B-Biological Sciences 281, 20133266. Glandt D (2006) Der Moorfrosch. Einheit und Vielfalt einer Braunfroschart. Bielefeld (Laurenti Verlag). Glos J, Grafe TU, Rodel MO, Linsenmair KE (2003) Geographic variation in pH tolerance of two populations of the European common frog, Rana temporaria. Copeia 3, 650-656. Gomez-Mestre I, Touchon JC, Warkentin KM (2006) Amphibian embryo and parental defenses and a larval predator reduce egg mortality from water mold. Ecology 87, 2570-2581. Goudet G, Mugnier S, Callebaut I, Monget P (2008) Phylogenetic analysis and identification of pseudogenes reveal a progressive loss of zona pellucida genes during evolution of vertebrates. Biology of Reproduction 78, 796-806. Guerardel Y, Kol O, Maes E, et al. (2000) O-glycan variability of egg-jelly mucins from Xenopus laevis: characterization of four phenotypes that differ by the terminal glycosylation of their mucins. Biochemical Journal 352, 449-463. Hangartner S, Laurila A, Räsänen K (2011) Adaptive divergence of the moor frog (Rana arvalis) along an acidification gradient. Bmc Evolutionary Biology 11, 366. Hangartner S, Laurila A, Räsänen K (2012) The quantitative genetic basis of adaptive divergence in the moor frog (Rana arvalis) and its implications for gene flow. J Evol Biol. 25, 1587- 1599. Hart GW, Copeland RJ (2010) Glycomics hits the big time. Cell 143, 672-676. Hart MW, Sunday JM, Popovic I, Learning KJ, Konrad CM (2014) Incipient speciation of sea star populations by adaptive gamete recognition coevolution. Evolution 68, 1294-1305.

68

Hawkins RD, Hon GC, Ren B (2010) Next-generation genomics: an integrative approach. Nature Reviews Genetics 11, 476-486. Hedrick J, Hardy D (1991) Chapter 12 Isolation of extracellular matrix structures from Xenopus laevis oocytes, eggs, and embryos. Proceedings of the XIX International Grassland Congress 36, 231-247. Hedrick JL (2008) Anuran and pig egg zona pellucida glycoproteins in fertilization and early development. International Journal of Developmental Biology 52, 683-701. Hedrick JL, Nishihara T (1991) Structure and function of the extracellular-matrix of anuran eggs. Journal of Electron Microscopy Technique 17, 319-335. Hoffmann AA, Parsons PA (1997) Extreme environmental change and evolution Cambridge University Press. Hohenlohe PA, Bassham S, Currey M, Cresko WA (2012) Extensive linkage disequilibrium and parallel adaptive divergence across threespine stickleback genomes. Philosophical Transactions of the Royal Society B-Biological Sciences 367, 395-408. Houle D, Govindaraju DR, Omholt S (2010) Phenomics: the next challenge. Nature Reviews Genetics 11, 855-866. Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics 24, 1403-1405. Jombart T, Ahmed I (2011) adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics 27, 3070-3071. Jombart T, Devillard S, Balloux F (2010) Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics 11. Kidd MR, Duftner N, Koblmuller S, Sturmbauer C, Hofmann HA (2012) Repeated parallel evolution of parental care strategies within Xenotilapia, a genus of cichlid fishes from lake Tanganyika. Plos One 7. Krapf D, Vidal M, Arranz SE, Cabada MO (2006) Characterization and biological properties of L-HGP, a glycoprotein from the amphibian oviduct with acrosome-stabilizing effects. Biology of the Cell 98, 403-413. Kroeker KJ, Kordas RL, Crim RN, Singh GG (2010) Meta-analysis reveals negative yet variable effects of ocean acidification on marine organisms. Ecology Letters 13, 1419-1434. Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680-685. Lee CE, Kiergaard M, Gelembiuk GW, Eads BD, Posavi M (2011) Pumping ions: rapid parallel evolution of ionic regulation following habitat invasions. Evolution 65, 2229-2244. Lindsay L (2003) Cross-fertilization and structural comparison of egg extracellular matrix glycoproteins from Xenopus laevis and Xenopus tropicalis. Comparative Biochemistry and Physiology - Part A: Molecular & Integrative Physiology 136, 343-352. Lohbeck KT, Riebesell U, Reusch TBH (2012) Adaptive evolution of a key phytoplankton species to ocean acidification. Nature Geoscience 5, 346-351.

69

Macqueen DJ, Kristjansson BK, Paxton CGM, Vieira VLA, Johnston IA (2011) The parallel evolution of dwarfism in Arctic charr is accompanied by adaptive divergence in mTOR- pathway gene expression. Molecular Ecology 20, 3167-3184. Marquis O, Millery A, Guittonneau S, Miaud C (2006) Toxicity of PAHs and jelly protection of eggs in the Common frog Rana temporaria. Amphibia-Reptilia 27, 472-475. Menkhorst E, Selwood L (2008) Vertebrate extracellular preovulatory and postovulatory egg coats. Biology of Reproduction 79, 790-797. Merilä J, Soderman F, O'Hara R, Räsanen K, Laurila A (2004) Local adaptation and genetics of acid-stress tolerance in the moor frog, Rana arvalis. Conservation Genetics 5, 513-527. Mousseau TA, Fox CW (1998) Maternal Effects As Adaptation. Oxford University Press. Olson JH, Chandler DE (1999) Xenopus laevis egg jelly contains small proteins that are essential to fertilization. Developmental Biology 210, 401-410. Palumbi SR (2009) Speciation and the evolution of gamete recognition genes: pattern and process. Heredity 102, 66-76. Pearse DE, Miller MR, Abadia-Cardoso A, Garza JC (2014) Rapid parallel evolution of standing variation in a single, complex, genomic region is associated with life history in steelhead/rainbow trout. Proceedings of the Royal Society B-Biological Sciences 281. Pearse DE, Pogson GH (2000) Parallel evolution of the melanic form of the California legless lizard, Anniella pulchra, inferred from mitochondrial DNA sequence variation. Evolution 54, 1041-1046. Persson M, Räsänen K, Laurila A, Merilä J (2007) Maternally determined adaptation to acidity in Rana arvalis: Are laboratory and field estimates of embryonic stress tolerance congruent? Canadian Journal of Zoology-Revue Canadienne De Zoologie 85, 832-838. Pespeni MH, Sanford E, Gaylord B, et al. (2013) Evolutionary change during experimental ocean acidification. Proceedings of the National Academy of Sciences of the United States of America 110, 6937-6942. Picker MD, Mckenzie CJ, Fielding P (1993) Embryonic tolerance of Xenopus (Anura) to acidic blackwater. Copeia, 1072-1081. Pierce BA (1985) Acid tolerance in amphibians. Bioscience 35, 239-243. Piertney SB, Oliver MK (2006) The evolutionary ecology of the major histocompatibility complex. Heredity 96, 7-21. Podrabsky JE, Carpenter JF, Hand SC (2001) Survival of water stress in annual fish embryos: dehydration avoidance and egg envelope amyloid fibers. American Journal of Physiology-Regulatory Integrative and Comparative Physiology 280, R123-R131. R Core Team (2014) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Räsänen K, Green E (2009) Acidification and its effects on amphibian populations. In Amphibian Biology, Volume 8. Decline: Diseases, Parasites, Maladies and Pollution. H. Heatwole (ed.), Surrey Beatty and Sons, Chipping Norton, Australia. pp. 3244-3267.

70

Räsänen K, Kruuk LEB (2007) Maternal effects and evolution at ecological time-scales. Functional Ecology 21, 408-421. Räsänen K, Laurila A, Merilä J (2003a) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. I. Local adaptation. Evolution 57, 352-362. Räsänen K, Laurila A, Merilä J (2003b) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. II. Adaptive maternal effects. Evolution 57, 363-371. Räsänen K, Söderman F, Laurila A, Merilä J (2008) Geographic variation in maternal investment: Acidity affects egg size and fecundity in Rana arvalis. Ecology 89, 2553- 2562. Renberg I, Korsman T, Birks HJB (1993) Prehistoric increases in the pH of acid-sensitive swedish lakes caused by land-use changes. Nature 362, 824-827. Roche A, Maggioni M, Narvarte M (2011) Predation on egg capsules of Zidona dufresnei (Volutidae): ecological implications. Marine Biology 158, 2787-2793. Rugh R (1962) Experimental embryology Burgess Publishing, Minneapolis, MN. Salthe SN (1963) The egg capsules in the amphibia. J Morphol 113, 161-171. Schagger H, Vonjagow G (1987) Tricine sodium dodecyl-sulfate polyacrylamide-gel electrophoresis for the separation of proteins in the range from 1-Kda to 100-Kda. Analytical Biochemistry 166, 368-379. Scillitani G, Moramarco AM, Rossi R, Mastrodonato M (2011) Glycopattern analysis and structure of the egg extra-cellular matrix in the Apennine yellow-bellied toad, Bombina pachypus (Anura: Bombinatoridae). Folia Histochemica Et Cytobiologica 49, 306-316. Segall GK, Lennarz WJ (1979) Chemical characterization of the component of the jelly coat from sea-urchin eggs responsible for induction of the acrosome reaction. Developmental Biology 71, 33-48. Spargo SC, Hope RM (2003) Evolution and nomenclature of the Zona pellucida gene family. Biology of Reproduction 68, 358-362. Strecker G, Coppin A, Maes E, Morelle W (1999) Structural analysis of 13 neutral oligosaccharide-alditols released by reductive beta-elimination from oviducal mucins of Rana temporaria. European Journal of Biochemistry 266, 94-104. Stumpp M, Hu MY, Melzner F, et al. (2012) Acidified seawater impacts sea urchin larvae pH regulatory systems relevant for calcification. Proceedings of the National Academy of Sciences of the United States of America 109, 18192-18197. Teacher AGF, Garner TWJ, Nichols RA (2009) European phylogeography of the common frog (Rana temporaria): routes of postglacial colonization into the British Isles, and evidence for an Irish glacial refugium. Heredity 102, 490-496. Tennessen JA, Akey JM (2011) Parallel adaptive divergence among geographically diverse human populations. Plos Genetics 7. Uller T (2008) Developmental plasticity and the evolution of parental effects. Trends in Ecology & Evolution 23, 432-438.

71

Varki A (2011) Evolutionary forces shaping the golgi glycosylation machinery: why cell surface glycans are universal to living cells. Cold Spring Harbor Perspectives in Biology 3 a005462. Wade MJ (1998) The evolutionary genetics of maternal effects. In Maternal Effects as Adaptations. Mousseau TA and Fox CW (ed.), Oxford University Press, pp. 5-21. Wake MH, Dickie R (1998) Oviduct structure and function and reproductive modes in amphibians. Journal of Experimental Zoology 282, 477-506. Watanabe A, Onitake K (2002) The urodele egg-coat as the apparatus adapted for the internal fertilization. Zoological Science 19, 1341-1347. Wong JL, Wessel GM (2006) Defending the zygote: Search for the ancestral animal block to polyspermy. Current Topics in Developmental Biology, Vol 72, 1-161. Yurewicz EC, Oliphant G, Hedrick JL (1975) Macromolecular-composition of Xenopus laevis egg jelly coat. Biochemistry 14, 3101-3107.

72

SI Tables and Figures

Fig. S1 Reaction norms of embryonic survival under pH 4.0, 4.2 and 7.5 within three R. arvalis populations (S, B and T). The values represent family means.

Fig. S2 Reaction norms of embryonic survival under pH 4.2, 4.4 and 7.5 within two R. temporaria populations (S and B). The values represent family means.

73

Fig. S3 Scatterplot of the discriminant analysis of principal components (DAPC) on jelly phenotypes of all clutches. The first axis is the horizontal axis. Colors represent the 7 phenotypic groups found by the K-means and BIC method. At the bottom left, the cumulated variance explained by number of PCA are represented.

Fig. S4 The “optimal” value of retaining PCA components was five according to the a-score optimization procedure proposed in adegenet.

74

Chapter III

Mechanistic basis of adaptive maternal effects: egg jelly water balance

mediates embryonic adaptation to acidity in Rana arvalis

Longfei Shu1, Marc J-F Suter2 Anssi Laurila3 and Katja Räsänen1

Affiliations:

1Eawag, Department of Aquatic Ecology, Switzerland and ETH Zürich, Institute of Integrative

Biology, Switzerland

2Eawag, Department of Environmental Toxicology, Switzerland and ETH Zürich, Department of

Environmental Systems Science, Switzerland

3Animal Ecology/Department of Ecology and Genetics, Evolutionary Biology Center, Uppsala

University, Sweden

75

Abstract

Environmental stress, such as acidification, can challenge persistence of natural populations and act as a powerful evolutionary force at ecological time scales. The ecological and evolutionary responses of natural populations to environmental stress are often mediated via maternal effects. During early life-stages, maternal effects commonly arise from egg capsules (the extracellular structures surrounding the embryo), but the role of egg capsules have rarely been studied in the context of adaptation to environmental stress. Previous studies on the moor frog Rana arvalis found egg capsule mediated adaptive divergence along an acidification gradient in embryonic acid stress tolerance. However, the exact mechanisms underlying these adaptive maternal effects remain unknown. Here we investigated the role of water balance and charge state (zeta potential) of egg jelly in embryonic adaptation to acid stress in three populations of R. arvalis. We found that acidic pH causes severe water loss in egg jelly, but that jelly from an acid adapted population retained more water than jelly from populations not adapted to acidity. Moreover, embryonic acid tolerance (survival at pH 4.0) correlated with both water loss and charge state of the jelly, indicating that negatively charged glycans influence jelly water balance and contribute to embryonic adaptation to acidity. These results indicate that egg capsules can harbor extensive intra-specific variation, facilitated likely in part via strong selection on water balance and glycosylation status of egg jelly. These findings shed light on the molecular mechanisms of environmental stress tolerance and adaptive maternal effects.

Keywords: adaptive divergence, amphibians, environmental stress, jelly glycan, water balance, zeta potential

76

Introduction

Environmental stress, defined as a condition that lies outside the optimal conditions for an organism and impairs Darwinian fitness, can have strong ecological consequences and be a powerful evolutionary force at ecological time scales (Hoffmann and Parsons 1997). Given the large spatial heterogeneity in environmental conditions, and the drastic ongoing environmental changes on global and local scales (Hoffmann and Sgrò 2011), it is of key interest how organisms are affected by environmental stress and how they can adapt to them. At early life-stages, organismal responses to stress are often mediated via maternal effects (MEs), defined as the effects of the mother’s phenotype or environment on offspring phenotype or performance (Mousseau and Fox 1998), which also can provide a powerful source of rapid adaptation (Mousseau and Fox 1998; Räsänen and Kruuk 2007). MEs can contribute to adaptive divergence of local populations, allow rapid adaptation, and alter the speed and direction of evolution (Mousseau and Fox 1998; Räsänen and Kruuk 2007). One important - but understudied - source of MEs in numerous taxa, from simple sexually reproducing multicellular organisms to amphibians and mammals, are egg capsules, which are maternally derived extracellular structures that surround the organism during early life stages. These structures can have major impacts on fitness, as they mediate the beginning of life due to their fundamental role in fertilization and protect the embryo from a range of environmental hazards, and can even facilitate speciation (e.g. Wong and Wessel 2006; Hedrick 2008; Menkhorst and Selwood 2008; Palumbi 2009; Berois et al. 2011). Despite these crucial roles of egg capsules, however, their role as a target of diversifying natural selection, and the extent of intra-specific variability (i.e. variation within and among individuals and populations) in them, has largely been overlooked.

Environmental stress arising from natural and anthropogenic acidification influences a broad range of aquatic and terrestrial taxa (e.g. Collier et al. 1990; Räsänen and Green 2009; Kroeker et al. 2010; Azevedo et al. 2013). Anthropogenic acidification became an urgent environmental issue upon the industrial revolution (e.g. Seip and Tollan 1978), and rising carbon dioxide (CO2) contents in the atmosphere have alerted more recent concern about ocean acidification (Doney et al. 2009; Honisch et al. 2012). There is an extensive literature on how organisms are affected by acidification via negative effects on, for example, survival (e.g. Findlay et al. 2008; Munday et al. 2009), reproduction (e.g. Räsänen et al. 2008; Byrne et al. 2010), calcification (Anthony et al. 2008) and other physiological processes (Iglesias-Rodriguez et al. 2008; Jokiel et al. 2008; Pespeni et al. 2013). Whilst often challenging the persistence of natural populations, these negative fitness effects also imply that acidity should cause strong natural selection. In accordance, there is increasing evidence for adaptation to acidity from both freshwater (e.g. Derry and Arnott 2007; Hangartner et al. 2011) and marine (e.g. Lohbeck et al. 2012; Dam 2013; Evans et al. 2013; Pespeni et al. 2013) taxa. The moor frog, Rana arvalis, is one of the best- characterized study systems for adaptation to acidification. R. arvalis shows adaptive divergence in embryonic survival, larval traits, as well as maternal investment (Räsänen et al. 2008; Räsänen

77

and Green 2009; Hangartner et al. 2011; Egea-Serrano et al. 2014), providing a well-suited model system to study mechanisms of adaptation to acid stress.

Adaptive divergence in embryonic acid stress tolerance of R. arvalis has been previously indicated to arise maternal effects (Merilä et al. 2004; Persson et al. 2007; Hangartner et al. 2012b), likely mediated via egg capsules that surround the developing embryos (Räsänen et al. 2003b). A similar mechanism was proposed for two species of Xenopus (X. laevis and X. gilli) differing in acid tolerance (Picker et al. 1993), possibly suggesting a general mechanism of adaptation to acid stress in amphibians. Amphibian embryos are surrounded by egg capsules, which consist of a fertilization envelope and variable numbers of gelatinous outer egg capsules (henceforth called egg jelly) (Hedrick 2008). When amphibian embryos are exposed to acidic conditions, they typically show a “curling defect” (Dunson and Connell 1982; Pierce 1985), whereby embryos develop, but become tightly curled within the egg capsule and, finally, fail to hatch. The curling defect has been suggested to relate to chemical changes in egg capsules, which become tight and sticky, shrink in size and can change color from transparent to opaque at acidic pH (Dunson and Connell 1982; Pierce 1985; Picker et al. 1993; Räsänen et al. 2003a). However, the molecular underpinnings of the embryonic curling defect and, hence, the mechanistic basis of maternally mediated embryonic adaptation to acid stress remain unexplored. Here we test the role of chemical alterations of the egg jelly as a mechanism of curling defect and adaptive maternal effects. We hypothesize that the apparent shrinkage of egg capsules under acidic conditions is caused by water loss of the egg jelly and examine to what extent this may be related to the glycosylation status of the jelly.

Glycosylation is one of the most important post translational modification of proteins (Varki et al. 2011). Amphibian egg jelly consists of highly glycosylated glycoproteins (Lee 1967; Yurewicz et al. 1975; Carroll 1991; Arranz et al. 1997). For example, in the African clawed frog, X. laevis, the jelly consists of more than 60% of glycans, and many of these are negatively charged (Hedrick and Nishihara 1991; Guerardel et al. 2000). These thick glycan coats are highly hydrated, giving the jelly a considerable capacity to hold water, and this process is correlated with the charge state of glycan molecules (Bansil et al. 1995). Because the charge state is pH dependent (Mullet et al. 1997), we predict that environmental pH can affect the charge state of jelly glycans and, consequently, alter jelly water content - resulting in the observed shrinkage of the egg capsules and the “curling defect” (Dunson and Connell 1982). In addition, we predict that if jelly charge status and water retention capacity underlie the maternal effects in embryonic adaptation to acid stress, intra-specific variation in water retention and jelly charge should correlate with embryonic acid tolerance.

To test these hypotheses we combined three approaches. First, to quantify variation in embryonic acid tolerance among and within populations, we conducted a common garden laboratory experiment using three R. arvalis populations (Table 1) known to differ in embryonic acid tolerance (Hangartner et al. 2011). Second, to test how pH influences water balance of egg jelly, we measured jelly water retention at pH 4.0, 7.5 and 10. Third, to test how water balance 78

correlates with charge state of the egg jelly, we measured its zeta potential (Kirby and Hasselbrink 2004a; Kirby and Hasselbrink 2004b). Finally, to test whether jelly water balance and zeta potential differs among and within populations and how it correlates with embryonic acid tolerance, we conducted these measurements on replicate clutches within each of the three populations. This design allowed us to investigate the molecular basis of maternally mediated adaptation to acidity and its consequences for embryonic fitness.

Materials and Methods

Study system

R. arvalis is a widely distributed anuran in the western Palearctic and inhabits a wide range of pH’s (Glandt 2006). Three populations (Table 1) breeding in permanent ponds in forested areas in southwestern Sweden, and known to differ in embryonic acid tolerance (details are provided in Hangartner et al. 2011), were used in this study. The pH in these ponds ranged from roughly pH 4 in the most acid tolerant population (Tottajärn, T) to pH 6 in the intermediately tolerant (Bergsjö, B) and pH 7 in the highly acid-sensitive population (Stubberud, S; Table 1). Site T is situated centrally within a geographic area that has been heavily affected by anthropogenic acidification since the early 1900s (Renberg et al. 1993), whereas site S is situated centrally within a geographic area that has remained unaffected by acid rain due to limestone bedrock (Brunberg and Blomqvist 2001; Hangartner et al. 2011). At both these sites, pH is relatively stable across years. The site B is acid sensitive and is being limed annually since 1989 to counteract acidification. Site B has an average pH around 6, but pH in this lake is highly variable across years (pH 4.5-7.5; Räsänen, pers. obs.). These patterns indicate selection for increased acid tolerance prior to liming, but potentially fluctuating selection during the last two decades.

During the breeding season of 2013, five clutches were collected in each population from the breeding ponds within ca. 30 min of egg laying (each site was continuously checked during the sampling night and only freshly laid clutches were collected, see Electronic Supplementary Material (ESM) Fig. S1). The eggs were immediately transferred to reconstituted soft water

(RSW, 48 mg/l NaHCO3, 30 mg/l CaSO4·2H2O, 61.4 mg/l MgSO4·7H2O and 2 mg/l KCl, pH 7.2-7.6; APHA 1985), maintained at cool temperature to slow down embryonic development, and transported to the laboratory at Uppsala University within one day of collection.

79

Table 1. Descriptive information on three study populations of R. arvalis used in this study. A) Number of full-sib families used per population, B) coordinates (N, E), C) mean ± SD pond pH, D) likely acidification history and E) pond size (m3). Pond pH is based on averages of three sites within each pond in April 2008 and April, May and June in 2009 (Hangartner et al. 2011). For a map and further details see Hangartner et al. (2011).

Population A) B) C) D) E)

Stubberud 58°46N, Buffered against acidification due to 5 7.3 ± 0.2 34128 (S) 13°76E limestone bedrock 58°20N, Limed since 1989. pH ca. 4.4 prior Bergsjön (B) 5 6.1 ± 0.3 7221305 13°48E to acidification 57°60N, Tottajärn (T) 5 4.0 ± 0.2 Natural & human acidification 462683 12°60E

Embryonic acid tolerance test (Experiment 1)

The embryonic acid tolerance of each clutch was tested using standard procedures (Hangartner et al. 2011; Räsänen et al. 2003a). In short, embryos were reared in a walk-in climate room (~ 16 °C) with 17L: 7D photoperiod at two pH treatments (acid: pH 4.0, neutral: pH 7.5). RSW was used as the experimental medium (APHA 1985) and it was prepared in 120L tanks two days prior to use. The pH in the acid treatment was adjusted with 1M H2SO4, whereas the pH in the neutral treatment was not adjusted (nominal pH of RSW is 7.2-7.6 when organisms are in the water (APHA 1985)). Embryos were placed in the experimental treatments within three hours after arrival to the laboratory at Uppsala University and before their first cell division (Gosner 1960).

Experiment 1 was performed as a 2 × 3 × 5 nested randomized design, with two pH treatments (pH 4.0 and 7.5), three populations (T, B and S), and five clutches (i.e. full-sib families) per population. Each family – pH treatment combination was replicated three times, resulting in a total of 90 experimental units. Due to logistic constraints, we chose to use the most acid tolerant, the most acid sensitive and one intermediate population along the acidification study gradient (Hangartner et al. 2011; Hangartner et al. 2012) to bracket the typical type of populations and full range of acid tolerances in this area. The replicates were fully randomized over the experiment shelves. Each experimental unit consisted of 20 to 40 embryos from a given family placed in a plastic vial (0.9 L), containing 0.5 L of treatment water. Embryos were reared from fertilization to day 12 (when all surviving embryos should have hatched). Unfertilized eggs (i.e. if no cell division was apparent at eggs were assumed to be unfertilized) were determined at day 3 and excluded from the analyses of survival. Fertilization rate was very high (almost all fertilized). 80

Water was changed every third day to maintain appropriate pH and pH measured in conjunction of each water change (mean ± S.D. at acid: pH 4.01 ± 0.04). The number of hatchlings and the number of obviously dead embryos or hatchlings were recorded visually at each water change, but any dead individuals were left untouched. Only final survival (day 12) was used in the statistical analyses. Survival was calculated as number of live hatchlings /total number of fertilized eggs for each experimental unit.

Variation in jelly water content (Experiment 2)

Variation among clutches and pH treatments in jelly water content was investigated in Experiment 2 on a separate subset of eggs from the same clutches used in Experiment 1. Variation in jelly water content was measured in acid (pH 4.0) and neutral (pH 7.5) treatments, as well as in an alkaline (pH 10) treatment. These treatments were chosen to test, on one hand, whether jelly loses water under acidic pH and, on the other hand, whether jelly absorbs more water in an alkaline environment. These patterns would be expected as at acidic conditions, jelly glycans will be less negatively charged and consequently their ability to retain water will be reduced, whereas at alkaline conditions jelly should encounter a converse process because of the opposite effect of the higher charge–charge repulsions caused by alkaline pH.

Experiment 2 was performed as a 3 × 3 × 5 nested fully randomized design, with three pH treatments (pH 4.0, 7.5, 10), three populations (T, B, S), and five families per population. Each family - pH treatment combination was replicated three times, resulting in a total of 135 experiment units. Each replicate vial had 10 eggs. The pH in the alkaline treatment was adjusted with 1M NaOH. Because pH was less stable in the alkaline treatment (mean ± SD: pH 9.94 ± 0.11), water in this treatment was changed daily. Otherwise the general experimental conditions were similar as in Experiment 1. Experiment 2 ended at day 3, when the eggs were manually de- jellied using watchmaker’s forceps (Hedrick and Hardy 1991) and jelly was collected for water content measurements.

To estimate jelly water content, the fully hydrated jellies from each of the three replicates (per family and treatment), as well as individually marked filter papers (Whatman No. 42), were weighed to the nearest 0.0001g (Lab Scales). The jelly samples (i.e. pooled jelly sample of 10 eggs per replicate vial) were then blotted on individually assigned filter paper and oven dried at +50 °C for 4 h (Lab BenchTop, STATUS). To estimate the amount of dry jelly material (i.e. jelly content without water), jelly dry mass was determined after drying (g) (see electronic supplementary material for jelly dry mass data). Absolute jelly water content was calculated as the difference in jelly mass (g) before and after drying (Berner and Ingermann 1990). As jelly water content might be influenced by the amount of dry jelly mass, relative jelly water content (jelly water content per unit of jelly) was calculated as absolute jelly water content / dry jelly mass (g). Given that all clutches were initially maintained at pH 7.0 (i.e. prior to experimental set up), jelly water loss or gain was calculated as the difference in jelly mass (g) between the neutral and acid (or the neutral and alkaline) treatments.

81

Zeta potential analyses

The Zeta potential (mV) is the electric potential across the double layer of a charged particle or molecule in solution (McNaught et al. 1997). The zeta potential is directly affected by the net electrical charge of the particle and thus it is widely used for quantification of the magnitude of the charge (Kosmulski 2009). In this study, we used the zeta potential as a measure of charge state (negative, neutral or positive) of egg jelly glycans. Negative zeta potential of the jelly sample would indicate negatively charged and thereby acidic glycans in the jelly, whereas a neutral or positive zeta potential would indicate neutral or alkaline glycans.

To determine the zeta potential, a separate subsample of 50 eggs per clutch from a neutral RSW treatment was collected manually using watchmaker’s forceps and subsequently solubilized in pH 8.9 DeBoers solution containing 45 mM mercaptoethanol (Hedrick and Hardy 1991). This jelly solution was then adjusted to pH 7 and maintained at 4°C for later analyses (Hedrick and Hardy 1991). The zeta potential of each jelly sample was measured using the Zetasizer Nano ZSP (Malvern Instruments). The instrument was calibrated and optimized before each run. Three technical replicates were run on each jelly sample, and the mean of the three replicates used in statistical analyses. Zeta potential was determined using Malvern Zetasizer Series Software (v7.02).

Statistical analyses

The response variables were survival, absolute and relative jelly water content (g) and zeta potential (mV). Survival (number hatched/total embryos/replicate) was analyzed with a generalized linear mixed model (GLMMs) with binomial errors and logit link function in the GLIMMIX procedure of SAS 9.3 (SAS Institute, Inc.). In the analyses of survival, pH treatment (4.0, 7.5), population (S, B, and T) and their interaction were used as fixed factors and family (nested within population) and family – treatment interactions as random factors. In subsequent analyses of survival within each of the two pH treatments, population (S, B, T) was used a fixed factor and family (nested within population) as a random factor.

After checking normality and variances, absolute or relative jelly water content was analyzed with type III general linear mixed models in SPSS 20, with pH treatment (4.0, 7.5, 10), population (S, B, T) and their interaction as fixed factors, and family (nested within population) and its interaction with pH treatment as random factors. Relative jelly water content was log- transformed to homogenize variances. As the pH treatment - family interaction was not significant for either absolute or relative jelly water content, it was excluded from the final models for these two variables. Variation in zeta potential was analyzed based on clutch means (total N = 15) with a one way analysis of variance (ANOVA) with population as the fixed factor in SPSS 20. Significance of relevant pairwise differences in least square means was tested using Tukey tests.

82

To gain correlative evidence for relationships among embryonic acid tolerance (survival at pH 4.0), jelly water change (i.e. the absolute difference in jelly water content between neutral and acid treatment for a given clutch) and zeta potential, we calculated Pearson correlations in SPSS 20 among clutch means (N = 15) of survival, jelly water change and jelly zeta potential.

Results

Embryonic acid tolerance

In general, embryonic survival was reduced by up to 60% at pH 4.0 compared to pH 7.5, as indicated by the significant pH treatment main effect (Fig. 1, Table 2a), but pH treatment × population interaction was not significant (Table 2a) in this study. However, a significant pH treatment × family interaction indicated strong family level variation within populations in pH tolerance (Table 2a). To investigate the nature of the pH × family interactions, we subsequently analyzed the data within each of the pH treatments. At pH 7.5, survival was very high (ranging from 92.8% to 100%) in all clutches and there was no significant family or population level variation (Table 2). At pH 4.0, a significant family effect indicated family level variation in embryonic acid tolerance within populations (Table 2b). In addition, a significant population main effect (Table 2) arose as embryos from population T had significantly higher survival than embryos from population S (Tukey test, P = 0.005), while survival of embryos from the population B was intermediate and did not differ significantly from either T or S population (Tukey test, both P > 0.1) (Fig. 1, Table 2b).

Table 2 Generalized linear mixed models of embryonic survival in three R. arvalis populations (acid, intermediate and neutral origin) under pH 4.0 and 7.5 treatments in the a) full model and b) by pH treatment. Significant effects (P < 0.05) are highlighted in bold. a)

Random effects Variance ± SE Z P Family (Population) 0 - - pH Treatment × Family (Population) 0.67 ± 0.33 2.04 0.021 Fixed effects ndf ddf F P pH Treatment 1 12 43.31 0.000 Population 2 12 2.54 0.121 pH Treatment × Population 2 12 1.53 0.256

83

b)

pH 4.0 treatment pH 7.5 treatment Random effects Variance ± SE Z P Variance ± SE Z P Family (Population) 0.65 ± 0.34 1.90 0.029 0.78 ± 0.91 0.86 0.195 Fixed effects ndf ddf F P ndf ddf F P Population 2 12 7.62 0.007 2 12 0.38 0.693

84

Figure 1 Survival rate (mean ± S.E.) of R. arvalis embryos in three populations (S, B, T) at two pH treatments (pH 4.0 and 7.5).

Population and pH treatment variation in jelly water content

Absolute jelly water content differed significantly among the three pH treatments (Fig. 2a, Table 3): jelly contained less water in the acid treatment than in the neutral treatment (pairwise Tukey test: P < 0.001, Fig. 2a), indicating that jelly water was lost in the acid treatment. Likewise, in absolute terms, jelly contained significantly more water in the alkaline treatment, indicating that it absorbed more water in the alkaline treatment than in the neutral treatment (pairwise Tukey test: P < 0.001, Fig. 2a). These patterns were paralleled by visual inspection that indicated that the jelly became compact and less pliable in the acid treatment, while it expanded in the alkaline environment to the extent of showing more flabby structure (L. Shu, pers. obs.). Moreover, a significant pH treatment × population interaction indicated that populations differed in water content at different pH’s. At pH 4.0, jelly from population T had lost least and jelly from population S most water, while jelly from population B was intermediate in jelly water loss (Fig. 2a, Table 3).

85

Table 3 General linear mixed models of a) absolute and b) relative (corrected for jelly dry mass of 10 eggs/replicate) jelly water content in three R. arvalis populations (acid, intermediate and neutral origin) under pH 4.0, 7.5 and 10 treatments. Significant effects (P < 0.05) are highlighted in bold.

a) Absolute jelly water content b) Relative jelly water content Type III Type III Mean Mean Sum of df F P Sum of df F P Square Square Squares Squares

Random effects

Family 1.49 12 0.12 1.05 0.412 0.48 12 0.04 0.80 0.654 (Population)

Fixed effects

pH Treatment 47.46 2 23.73 199.61 0.000 3.48 2 1.74 34.69 0.000 Population 0.40 2 0.20 1.61 0.241 17.55 2 8.77 219.86 0.000 Population × pH 4.63 4 1.16 9.73 0.000 1.50 4 0.38 7.48 0.000 Treatment

Also in terms of relative jelly water content (Fig. 2b), acid treatment caused jelly water loss compared to the neutral treatment (pairwise Tukey tests, P < 0.001, Fig. 2b). However, the difference between the alkaline treatment and neutral treatment was not significant (Tukey test, P = 0.920, Fig. 2b). A significant population main effect arose as the jelly from population T had significantly higher relative jelly water content than jelly in the other two populations at all pH treatments (pairwise Tukey tests, both P < 0.001) and jelly from population B had a lower relative water content than jelly from population S (Tukey test, P < 0.001, Fig. 2b, Table 3). This indicated that populations overall differed in overall ability to absorb water. In accordance with the absolute jelly water content analysis, a significant pH treatment × population interaction showed that populations differed in relative water loss in the acid treatment, whereby T population lost least water at pH 4.0 (Fig. 2b, Table 3).

86

Figure 2 Jelly water content in three R. arvalis populations (S, B, T) at three pH treatments (pH 4.0, 7.5 and 10). The figures represent mean ± S.Es of a) absolute jelly water content; b) relative jelly water content (corrected for jelly dry mass).

Variation in zeta potential of jelly

The zeta potential of all clutches was negative (range -12.600 mV to -32.533 mV; Fig. 3), indicating that the jelly consisted of negatively charged components. Moreover, the distribution of zeta potential in all clutches had a unique peak (Fig. 3a), indicating that egg jelly glycans assemble to form a stable homogenized structure with a negative surface charge. Zeta potential differed among populations (F2,12 = 4.43, P = 0.036, Fig. 3b), with jelly of population T being more negatively charged than jelly of population S (Tukey test, P = 0.031, Fig. 3b). There was no significant difference between the intermediate population B and the other two populations ( T: P = 0.534, and S: P = 0.196, Fig. 3b).

87

a)

b)

Figure 3 Zeta potential of jelly as a) distribution from one representative jelly sample and b) boxplot of egg jelly zeta potential in each of three R. arvalis populations (S, B and T). All samples had a single peak (details not shown), which indicates that egg jelly glycans assemble to form a stable homogenized structure with a negative surface charge. In b), the boxes indicate the inter-quartile range (25% and 75%) and the thick line within the box represents the median of five clutches/populations. The whiskers represent the minimum or maximum values.

88

Figure 4 Scatterplot of clutch means (N = 15) of a) jelly water loss and embryonic acid tolerance (survival at pH 4.0), b) zeta potential and embryonic acid tolerance and c) zeta potential and jelly water loss across three R. arvalis populations (S, B and T).

89

Correlations between jelly water content, acid tolerance and zeta potential

In general, embryonic acid tolerance was negatively correlated with acidity induced jelly water loss (Fig. 4a, r = - 0.807, P < 0.0001; N = 15), suggesting that the ability to retain water in the jelly at acidic conditions has adaptive value. Moreover, zeta potential was negatively correlated with embryonic acid tolerance (Fig. 4b, r = - 0.653, P < 0.009, N = 15), and positively correlated with acidity induced jelly water loss (Fig. 4c, r = 0.664, P < 0.009, N = 15), suggesting that negatively charged glycans influence jelly water content and, hence, contribute to embryonic acid stress tolerance.

Discussion

Maternal effects can be an important means of adaptation to environmental stress (Räsänen and Kruuk 2007), and at early life-stages often may arise via egg capsules. We here found that embryonic survival (hatching rate) correlated with the extent of water loss from the gelatinous egg capsules (jelly) at acidic conditions, and that jelly from an acid adapted population retained more water at pH 4.0 than jelly from the two other populations. Moreover, the significant correlations between embryonic survival, jelly water loss and zeta potential (indicative of the surface charge) of the jelly suggest that maternal effects derived from glycosylation status of the egg capsules may play an essential role in embryonic adaptation to acidity. These results shed light on the underlying sources of variation in embryonic stress tolerance both among and within populations (Räsänen et al. 2003a; Hangartner et al. 2011). Our study is the first to show that acidity causes severe water loss of egg jelly in amphibians and provides first evidence for intra- specific adaptive divergence in jelly water balance. In the following, we discuss the possible role of environmental acidification as a selective force on jelly water content, the molecular basis of this process and propose a “water balance model” for jelly responses to pH.

Acidification as an environmental stressor and selective force on jelly water content

Many amphibians, as well as for example gastropod molluscs, sea urchins and many fish species (Menkhorst and Selwood 2008), have thick gelatinous egg capsules (i.e. jelly). These jelly capsules consist of variable numbers of thin gelatinous layers, which take up water and swell upon their first contact with water (Salthe 1963). The high capacity of jelly to absorb water is of crucial importance for the reproducing female and the eggs: it allows for smaller space (and hence more eggs) whilst the eggs are still inside the mother, yet enables the subsequent swelling and ecological functions, such as protecting the embryo from a range of environmental hazards (e.g. dehydration, pathogens, pollution or UV-radiation; Menkhorst and Selwood 2008). Hence, any environmental factor that substantially disrupts jelly water balance, and as a consequence, embryonic fitness, may cause strong selection based on jelly water balance – either by influencing jelly water uptake capacity and/or by affecting jelly water loss.

90

We here show that acidic pH reduces embryonic survival and induces water loss from the jelly of R. arvalis and that jelly of the acid tolerant (T) population lost less water under acidic conditions than jelly of the acid sensitive (S) population, whereas the extent of water loss was intermediate in the population (B) with intermediate acid tolerance. Moreover, across all clutches, we found correlative evidence that embryonic acid stress tolerance increased as jelly water loss decreased.

Our results have the following main implications in relation to acid stress from the ecological and evolutionary point of views. From an ecological point of view, the results suggest that the commonly observed “curling defect”, which has been put forward as the main cause behind reduced embryonic survival under acidic conditions in amphibians (reviewed in Räsänen and Green 2009), arises at least in part from jelly water loss: as acidic pH induces water loss, jelly becomes more tight and sticky, and, consequently, traps the embryo inside. Jelly water loss may also disturb the water balance and gas exchange of embryos, further reducing embryonic survival (Seymour 1994; Herrler and Beier 2000).

Intriguingly, from an evolutionary point of view, our results suggest that adaptation to acidification in R. arvalis may occur via jelly water balance. Adjustments of jelly water balance may therefore provide a key adaptive mechanism behind the egg capsule related maternal effects, facilitating adaptive divergence in embryonic acid tolerance in amphibians (Räsänen et al. 2003a; reviewed in Räsänen and Green 2009; Hangartner et al. 2011). However, for logistical reason, we used only the two most extreme and one intermediate population known to differ in embryonic acid tolerance in this study and to further validate the generality of our findings, more populations across pH gradients should be examined. In addition, such adaptive mechanisms are yet to be explored in other taxa, such as those exposed to ocean acidification (Kroeker et al. 2010; Pespeni et al. 2013).

Role of extracellular glycan diversity in adaptive evolution

All cells are covered by glycan coats, which are essential for numerous biological processes, such as cell communication and pathogen defense (Gagneux and Varki 1999; Hart and Copeland 2010). These extracellular glycans show great diversity and are highly dynamic across taxa (Varki 2011). However, although glycans often are a key component of egg capsules, and egg capsules have fundamental biological functions from fertilization to protection from environmental hazards (Menkhorst and Selwood 2008; Shu et al. submitted), to date next to nothing is known about intra-specific glycan variation of egg capsules, or the functional and evolutionary consequences of this diversity. Most studies to date have focused on quantifying glycan compositional variation among taxa (e.g. Coppin et al. 1999). In a rare study, intra- specific polymorphism, equivalent of human blood groups, was found in the macromolecular composition of the egg jelly in X. laevis – but the functional consequences remained unclear (Guerardel et al. 2000).

91

Our study is among the first to quantify a molecular component of intra-specific jelly variation (here related to zeta potential). In doing so, we have taken a step closer to link molecular variation with its fitness consequences. In addition to the clear differences in jelly water content at different pHs and in different populations, we also found that R. arvalis jelly is negatively charged, indicating that it primarily consists of acidic glycans. Intriguingly, our analyses of zeta potential further indicate that the acid tolerant population (T) have a higher proportion of acidic glycans in their jelly (as indicated by the more negative charge status of jelly), which correlates with the a higher ability to retain more water per jelly unit compared to the more acid sensitive S and B populations. Across all experimental clutches, the the jelly charge status correlated with higher embryonic fitness (survival) under acidic conditions, indicating that intra-specific extracellular glycan diversity (e.g. amount or type of acidic glycans) may undergo adaptive evolution in response to environmental stress, such as acidification. Several different pathways may influence glycan variation, but to what extent this variation reflects environmentally induced plasticity or variation in gene expression or strict genetic responses (Gagneux and Varki 1999; Varki 2011), is currently unclear. Hence, future studies should quantify glycan composition and the genetic basis of this variation to gain insight into the molecular basis of egg capsule mediated maternal effects.

General “pH - jelly water balance” model

Although the swelling of egg jelly in water is a well-known phenomenon, surprisingly few studies have explored the underlying mechanism of water uptake and how jelly water balance is influenced by environmental factors. With regard to amphibians, a study on the salamander Ambystoma macrodactylum indicated that sialic acid may play a role in jelly water retention (Berner and Ingermann 1990). However, mass spectrometric studies found no evidence of sialic acid in jelly of R. arvalis (Coppin et al. 1999), which we confirmed for our study populations using a sialic acid detection kit (data not shown). Thus, other mechanisms are called for variation in R. arvalis water balance in relation to pH.

With regard to pH, we propose a general “pH - jelly water balance” model, whereby jelly shrinks at acidic pH due to water loss and expands at neutral or alkaline pHs due to water uptake (Fig. 5). We suggest that this effect arises because the surface charge of the glycans in the egg jelly is affected by water pH (Mullet et al. 1997): compared to neutral conditions, acidic pH (in our case pH 4.0) reduces net negative charge by protonating acidic moieties, which results in a weaker electrostatic repulsion, and therefore reduces capacity of retaining water. Alkaline pH on the other hand has the opposite effect by increasing the net negative charge and thus, causing a stronger charge–charge repulsion (Mullet et al. 1997; Patil et al. 2007). These hypotheses were supported by our jelly water content measurement under different pH, and the zeta potential measurements. We believe this model is extendable to other aquatic species that have gelatinous structures surrounding their embryos, such as sea urchins, fish, and other amphibians (Menkhorst and Selwood 2008). Finally, given that jelly water balance can have strong effects on embryonic

92

fitness (as shown here), environmental stressors that affect jelly water balance (e.g. acidification, salinity or temporally drying breeding sites), may induce strong selection on jelly water balance – and hence glycan diversity - in aquatic species that have gelatinous egg capsules.

Figure 5 Schematic presentation of a “pH-jelly water balance” model. According to this model, jelly absorbs less and/or loses more water at acidic pH, while absorbing more water at alkaline pH compared to neutral pH conditions. Based on the hypothesis laid out here these patterns are mediated via the net electric charge of the jelly: acidic pH (in our case pH 4.0) reduces negative charge, which results in a weaker electrostatic repulsion, and therefore reduces capacity of retaining water. In contrast, alkaline pH has the opposite effect by causing a stronger charge - charge repulsion.

Conclusions

Our study revealed that acid stress causes severe water loss of egg jelly and likely imposes strong selection on water balance of egg jelly in amphibians due to subsequently reduced embryonic hatching success. We suggest that a “pH- jelly water balance” model, mediated via intra-specific glycan variability, may help explaining the molecular basis of embryonic adaptation to acid stress and egg capsule mediated adaptive maternal effects.

Acknowledgements

We thank Beatrice Lindgren for invaluable help with the field and laboratory work, Baptiste Pasteur for help in collecting the frogs, and Yang Yue for help with the zeta potential measurement. The experiments were conducted under permissions from the Västra Götaland

93

county board and the Ethical committee for animal experiments in Uppsala County. This study was supported by Swiss National Science foundation (to KR).

References Anthony KRN, Kline DI, Diaz-Pulido G, Dove S, Hoegh-Guldberg O (2008) Ocean acidification causes bleaching and productivity loss in coral reef builders. Proceedings of the National Academy of Sciences of the United States of America 105:17442-17446 APHA (1985) Standard methods for the examination of water and wastewater, vol. 16. American Public Health Association, Washington, DC Arranz SE, Albertali IE, Cabada MO (1997) Bufo arenarum egg jelly coat: Purification and characterization of two highly glycosylated proteins. Biochemical Journal 323:307-312 Azevedo LB, van Zelm R, Hendriks AJ, Bobbink R, Huijbregts MA (2013) Global assessment of the effects of terrestrial acidification on plant species richness. Environ Pollution 174:10- 15 Bansil R, Stanley HE, Lamont JT (1995) Mucin Biophysics. Annual Review of Physiology 57:635-657 Berner NJ, Ingermann RL (1990) Role of sialic acid in exogenous protein accumulation and water retention by the egg jelly of the Salamander Ambystoma macrcbdactylum. The Journal of Experimental Zoology 256:38-43 Berois N, Arezo MJ, Papa NG (2011) Gamete interactions in teleost fish: the egg envelope. Basic studies and perspectives as environmental biomonitor. Biological Research 44:119- 124 Brunberg AK, Blomqvist P (2001) Quantification of anthropogenic threats to lakes in a lowland county of central Sweden. Ambio 30:127-134 Byrne M, Soars N, Selvakumaraswamy P, Dworjanyn SA, Davis AR (2010) Sea urchin fertilization in a warm, acidified and high pCO(2) ocean across a range of sperm densities. Marine Environmental Research 69:234-239 Carroll EJ (1991) Structure and macromolecular-composition of the egg and embryo jelly coats of the anuran Lepidobatrachus laevis. Development Growth & Differentiation 33:37-43 Collier KJ, Ball OJ, Graesser AK, Main MR, Winterbourn MJ (1990) Do organic and anthropogenic acidity have similar effects on aquatic fauna? Oikos 59:33-38 Coppin A, Maes E, Flahaut C, Coddeville B, Strecker G (1999) Acquisition of species-specific O-linked carbohydrate chains from oviducal mucins in Rana arvalis - A case study. European Journal of Biochemistry 266:370-382 Dam HG (2013) Evolutionary Adaptation of Marine Zooplankton to Global Change. In: Carlson CA, Giovannoni SJ (eds) Annual Review of Marine Science, Vol 5, vol 5. Annual Reviews, Palo Alto, pp 349-370 Derry AM, Arnott SE (2007) Adaptive reversals in acid tolerance in copepods from lakes recovering from historical stress. Ecological Applications 17:1116-1126

94

Doney SC, Fabry VJ, Feely RA, Kleypas JA (2009) Ocean Acidification: The Other CO2 Problem. Annual Review of Marine Science 1:169-192 Dunson WA, Connell J (1982) Specific-Inhibition of Hatching in Amphibian Embryos by Low Ph. Journal of Herpetology 16:314-316 Egea-Serrano A, Hangartner S, Laurila A, Räsänen K (2014) Multifarious selection through environmental change: acidity and predator-mediated adaptive divergence in the moor frog (Rana arvalis). Proceedings of the Royal Society B-Biological Sciences 281: 20133266 Evans TG, Chan F, Menge BA, Hofmann GE (2013) Transcriptomic responses to ocean acidification in larval sea urchins from a naturally variable pH environment. Molecular Ecology 22:1609-1625 Findlay HS, Kendall MA, Spicer JI, Turley C, Widdicombe S (2008) Novel microcosm system for investigating the effects of elevated carbon dioxide and temperature on intertidal organisms. Aquatic Biology 3:51-62 Gagneux P, Varki A (1999) Evolutionary considerations in relating oligosaccharide diversity to biological function. Glycobiology 9:747-755 Glandt D (2006) Der Moorfrosch. Einheit und Vielfalt einer Braunfroschart. Bielefeld (Laurenti Verlag) Gosner KL (1960) A simplified table for staging anuran embryos and larvae with notes on identification. Copeia:183–190 Guerardel Y, Kol O, Maes E, Lefebvre T, Boilly B, Davril M, Strecker G (2000) O-glycan variability of egg-jelly mucins from Xenopus laevis: characterization of four phenotypes that differ by the terminal glycosylation of their mucins. Biochemical Journal 352:449- 463 Hangartner S, Laurila A, Räsänen K (2011) Adaptive divergence of the moor frog (Rana arvalis) along an acidification gradient. BMC Evolutionary Biology 11:366 Hangartner S, Laurila A, Räsänen K (2012a) Adaptive divergence in moor frog (Rana Arvalis) populations along an acidification gradient: inferences from Qst-Fst correlations. Evolution 66:867-881 Hangartner S, Laurila A, Räsänen K (2012b) The quantitative genetic basis of adaptive divergence in the moor frog (Rana arvalis) and its implications for gene flow. Journal of Evolutionary Biology 25:1587-1599 Hedrick J, Hardy D (1991) Chapter 12 Isolation of extracellular matrix structures from Xenopus laevis oocytes, eggs, and embryos. In Xenopus laevis: practical uses in cell and molecular biology, Volume 36. Edited by Kay BK, Peng HB. London: Acedemic press 36:231-247 Hedrick JL (2008) Anuran and pig egg zona pellucida glycoproteins in fertilization and early development. International Journal of Developmental Biology 52:683-701 Hedrick JL, Nishihara T (1991) Structure and function of the extracellular matrix of anuran eggs. Journal of Electron Microscopy Technique 17:319-335

95

Herrler A, Beier HM (2000) Early embryonic coats: Morphology, function, practical applications - An overview. Cells Tissues Organs 166:233-246 Hoffmann AA, Parsons PA (1997) Extreme environmental change and evolution. Cambridge University Press, Cambridge Hoffmann AA and Sgrò CM (2011) Climate change and evolutionary adaptation. Nature 470:479-485 Honisch B, Ridgwell A, Schmidt DN, Thomas E Gibbs SJ et al. (2012) The Geological Record of Ocean Acidification. Science 335:1058-1063 Iglesias-Rodriguez MD, Halloran PR, Rickaby RE, Hall IR et al. (2008) Phytoplankton

calcification in a high CO2 world. Science 320:336-340 Jokiel PL, Rodgers KS, Kuffner IB, Andersson AJ, Cox EF, Mackenzie FT (2008) Ocean acidification and calcifying reef organisms: a mesocosm investigation. Coral Reefs 27:473-483 Kirby BJ, Hasselbrink EF (2004a) Zeta potential of microfluidic substrates: 1. Theory, experimental techniques, and effects on separations. Electrophoresis 25:187-202 Kirby BJ, Hasselbrink EF (2004b) Zeta potential of microfluidic substrates: 2. Data for polymers. Electrophoresis 25:203-213 Kosmulski M (2009) Surface charging and points of zero charge. CRC Press, Boca Raton Kroeker KJ, Kordas RL, Crim RN, Singh GG (2010) Meta-analysis reveals negative yet variable effects of ocean acidification on marine organisms. Ecology Letters 13:1419-1434 Lee PA (1967) Studies of frog oviducal jelly secretion 2. Cytology of secretory cycle. Journal of Experimental Zoology 166:107-119 Lohbeck KT, Riebesell U, Reusch TBH (2012) Adaptive evolution of a key phytoplankton species to ocean acidification. Nature Geoscience 5:346-351 McNaught AD, Wilkinson A (1997) International Union of Pure and Applied Chemistry. Compendium of chemical terminology : IUPAC recommendations. 2nd edn. Blackwell Science, Oxford England; Malden, MA, USA Menkhorst E, Selwood L (2008) Vertebrate extracellular preovulatory and postovulatory egg coats. Biology of Reproduction 79:790-797 Merilä J, Söderman F, O'Hara R, Räsänen K, Laurila A (2004) Local adaptation and genetics of acid-stress tolerance in the moor frog, Rana arvalis. Conservation Genetics 5:513-527 Mousseau TA, Fox CW (1998) Maternal effects as adaptations. Oxford University Press, USA Mullet M, Fievet P, Reggiani JC, Pagetti J (1997) Surface electrochemical properties of mixed oxide ceramic membranes: Zeta-potential and surface charge density. Journal of Membrane Science 123:255-265 Munday PL, Dixson DL, Donelson JM, Jones GP, Pratchett MS, et al. (2009) Ocean acidification impairs olfactory discrimination and homing ability of a marine fish. Proceedings of the National Academy of Sciences of the United States of America 106:1848-1852

96

Palumbi SR (2009) Speciation and the evolution of gamete recognition genes: pattern and process. Heredity 102:66-76 Patil S, Sandberg A, Heckert E, Self W, Seal S (2007) Protein adsorption and cellular uptake of cerium oxide nanoparticles as a function of zeta potential. Biomaterials 28:4600-4607 Persson M, Räsänen K, Laurila A, Merilä J (2007) Maternally determined adaptation to acidity in Rana arvalis: Are laboratory and field estimates of embryonic stress tolerance congruent? Canadian Journal of Zoology 85:832-838 Pespeni MH, Sanford E, Gaylord B et al. (2013) Evolutionary change during experimental ocean acidification. Proceedings of the National Academy of Sciences of the United States of America 110:6937-6942 Picker MD, Mckenzie CJ, Fielding P (1993) Embryonic tolerance of Xenopus (Anura) to acidic blackwater. Copeia:1072-1081 Pierce BA (1985) Acid tolerance in amphibians. Bioscience 35:239-243 Räsänen K, Green E (2009) Acidification and its effects on amphibian populations. Amphibian Biology. Conservation and Ecology, Volume 8. Edited by Heatwole H. Surrey Beatty and Sons, Chipping Norton, Australia. Räsänen K, Kruuk LEB (2007) Maternal effects and evolution at ecological time-scales. Functional Ecology 21:408-421 Räsänen K, Laurila A, Merilä J (2003a) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. I. Local adaptation. Evolution 57:352-362 Räsänen K, Laurila A, Merilä J (2003b) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. II. Adaptive maternal effects. Evolution 57:363-371 Räsänen K, Söderman F, Laurila A, Merilä J (2008) Geographic variation in maternal investment: Acidity affects egg size and fecundity in Rana arvalis. Ecology 89:2553- 2562 Renberg I, Korsman T, Anderson NJ (1993) A temporal perspective of lake acidification in Sweden. Ambio 22:264-271 Salthe SN (1963) The egg capsules in the amphibia. Journal of Morphology 113:161-171 Seip HM, Tollan A (1978) Acid precipitation and other possible sources for acidification of rivers and lakes. Science of the Total Environment 10:253-270 Seymour RS (1994) Oxygen diffusion through the jelly capsules of amphibian eggs. Israel Journal of Zoology 40:493-506 Varki A (2011) Evolutionary forces shaping the golgi glycosylation machinery: Why cell surface glycans are universal to living cells. Cold Spring Harbor Perspectives in Biology 3 Wong JL, Wessel GM (2006) Defending the zygote: Search for the ancestral animal block to polyspermy. Current Topics in Developmental Biology, Vol 72 72:1-151 Yurewicz EC, Oliphant G, Hedrick JL (1975) Macromolecular composition of Xenopus laevis egg jelly coat. Biochemistry 14:3101-3107

97

Chapter IV

First transcriptome analysis of the moor frog Rana arvalis: genomics

resources and candidate maternal effect genes

Longfei Shu and Katja Räsänen

Affiliations:

Eawag, Department of Aquatic Ecology, Switzerland and ETH Zürich, Institute of Integrative

Biology, Switzerland

98

Abstract

Evolutionary studies of non-model species are often hampered by the lack of genomic resources. We here provide the first set of genomics resources for the moor frog (Rana arvalis), an emerging model system for the studies of ecology and evolution in natural populations. Egg coat mediated maternal effects are an important component of reproductive fitness in a broad range of taxa, including R. arvalis – yet the genetic architecture of maternal effects is poorly understood. We here aim to identify candidate genes for egg coat mediated adaptive maternal effects in R. arvalis using tissue specific transcriptome analyses (RNAseq on oviduct samples).

In R. arvalis, adaptive divergence among populations in embryonic acid stress tolerance is mediated via the gelatinous egg coats (i.e. egg jelly). We collected oviduct samples from seven R. arvalis females, representing the full range of within and among population variation in embryonic acid stress tolerance previously observed in this species. After de novo assembly, 124 071 unigenes were detected, with an N50 of 1 212 bp and a total length of 90.3 Mb. Functional annotation analysis identified a total of 57 839 annotated unigenes with 45 071, 39 138, 37 262, 31 405, 13 501, 24 452 unigenes representing the Nr, Nt, Swiss-Prot, KEGG, COG, GO databases, respectively. We further provide 26 711 gene-linked microsatellite (SSRs) and 231 274 single nucleotide polymorphism (SNPs) markers. In addition, we identified two types of candidate maternal effect (ME) genes likely underlying variation in the hyper-variable egg jelly: jelly core protein genes and five groups of jelly glycosylation genes. These new genomic resources will aid population genomic and molecular ecological studies of amphibians and provide general insight to the genetic architecture of egg coat mediated maternal effects.

Keywords: Amphibian; Maternal effect genes; RNA Seq; Transcriptome; SNP; Rana arvalis

99

Introduction

Evolutionary studies of non-model species are typically hampered by the lack of genomic resources (Nadeau & Jiggins 2010), yet understanding the evolutionary processes in natural populations necessitates a good understanding of the genetic architecture of trait variation in an ecologically relevant context (e.g. Mitchell-Olds et al. 2007; Houle et al. 2010; Nadeau & Jiggins 2010). The emergence of next generation sequencing tools and the decline in associated costs is now, however, allowing to bridge this gap. In particular, RNA-seq has proved to be an effective way to characterize the transcriptome (Wang et al. 2009) and to develop molecular tools in non-model species (Ekblom & Galindo 2011). We here provide the first transcriptomics data set for an ecologically well-studied ranid amphibian, the moor frog, Rana arvalis.

Amphibians are common model systems for a range of ecological and evolutionary studies, ranging from understanding evolutionary responses of natural populations to different selective factors (e.g. Van Buskirk et al. 1997; Van Buskirk & Arioli 2005; Egea-Serrano et al. 2014), the spatial context of adaptation (Richter-Boix et al. 2011), the role of developmental plasticity (e.g. Gomez-Mestre & Buchholz 2006; Touchon et al. 2006) and adaptive maternal effects (Mousseau & Fox 1998; Räsänen et al. 2003). Despite the increasing numbers of genomes being sequenced (Altshuler et al. 2012; Jones et al. 2012), the genomic resources for amphibians remain depauperate. Thus far, the genome of only a single species, the western clawed frog, Xenopus tropicalis has been sequenced (Hellsten et al. 2010). To increase our inferential ability on evolution in natural populations of amphibians, a broader range of genomic resources are needed.

R. arvalis is an important study species for a broad range of ecological and evolutionary questions, including developmental biology (Laurila et al. 2002; Loman 2002), population biology (e.g. Babik et al. 2004; Laurila et al. 2006; Knopp & Merilä 2009a, b), environmental toxicology (e.g. Greulich & Pflugmacher 2003; Hangartner & Laurila 2012), road ecology (e.g. Hels & Buchwald 2001; Balkenhol & Waits 2009; Beebee 2013), behavioral ecology (Knopp et al. 2008), adaptive maternal effects (Räsänen et al. 2003b; Räsänen et al. 2005; Räsänen et al. 2008), and adaptive divergence in general (e.g. Richter-Boix et al. 2011; Hangartner et al. 2011, 2012; Egea-Serrano et al. 2014).

R. arvalis is one of the best-characterized vertebrate systems for studying adaptation of natural populations to environmental stress, in particular acidification (Merilä et al. 2004; Persson et al. 2007; Hangartner et al. 2011, 2012; Egea-Serrano et al. 2014). In Sweden, R. arvalis shows adaptive divergence along an environmental acidification gradient in embryonic survival, larval predator defense and life-history traits, as well as maternal investment (e.g. Egea-Serrano et al. 2014; Hangartner et al. 2011; Räsänen et al. 2008), providing a well-suited model system to study mechanisms of adaptation to environmental stress. Furthermore, quantitative genetic approaches indicate that adaptive divergence in embryonic acid stress tolerance of R. arvalis

100

primarily arises through maternal effects (Hangartner et al. 2012; Merilä et al. 2004; Persson et al. 2007; Räsänen et al. 2003b), likely mediated via egg coats that surround the developing embryos (Räsänen et al. 2003b).

In amphibians, and many other taxa (reviewed in Shu et al., I), embryos are surrounded by egg coats – these are maternally derived, extracellular structures that consist of multiple functionally and structurally different layers (Altig & McDiarmid 2007; Menkhorst & Selwood 2008). These structures have many key functions, mediating the beginning of life due to their fundamental role in fertilization and protecting the embryo from a range of environmental hazards (Claw & Swanson 2012; Menkhorst & Selwood 2008; Wong & Wessel 2006; reviewed in I). These egg coats consist of glycoproteins, with a complex glycan structure attached to a protein backbone (Hedrick & Nishihara 1991; Maack et al. 1985) and are highly species specific (e.g. Coppin et al. 1999; Delplace et al. 2002; Strecker et al. 1999). Our own work on R. arvalis indicates that egg coat mediated maternal effects on embryonic acid stress tolerance is most strongly related to the gelatinous layers (i.e. egg jelly), which are highly variable in composition among and within populations (Shu et al. II). However, the genetic architecture of egg coat variation and, hence, adaptive maternal effects, is currently unknown.

Like in many other ecological model species, genomics resources are almost absent in R. arvalis. The genome of R. arvalis has not been sequenced, and no transcriptome or expressed sequence tags (ESTs) are available in this species, making gene-based inferences difficult. Until recently, studies investigating genetic variation in this species relied mainly on a limited number of microsatellite markers, which are normally adapted from other species (Hangartner et al. 2012; Knopp & Merila 2009a; Richter-Boix et al. 2011) and no genome wide SNP markers (Morin et al. 2004) are available in this species. Although the glycan composition of R. arvalis egg coats has been analyzed to some degree (Coppin et al. 1999), until now it is not clear which genes and pathways are involved in the biosynthesis of egg coats in R. arvalis. Hence, the lack of genomic resources has limited our understanding on the diversity and evolution in this species in general, and the genetic architecture underlying adaptive maternal effects more specifically.

In this study, we performed the first deep sequencing of the R. arvalis transcriptome using tissue specific RNA sequencing (RNA-seq). As our specific interest is in identifying genes underlying egg jelly coat variation, we collected samples from the oviduct (the specific tissue where egg jelly is produced; Hedrick & Nishihara 1991)). A selected set of seven R. arvalis individuals bracketing the full range of embryonic acid tolerance variation (Shu et al. II) among and within our study populations were used to provide the first transcriptomic and genome wide molecular markers (SNPs and SSRs, Supplemental material) and to identify candidates for maternal effect genes related to jelly coats.

Materials and Methods

Study system

101

R. arvalis is an anuran amphibian, broadly distributed across the western Palearctic (Glandt 2006). Individuals from three R. arvalis populations (Tottajärn, T, Bergsjö, B and Stubberud, S) breeding in permanent ponds in southwestern Sweden were used in this study (Hangartner et al. 2011; II). pH in the study ponds ranges from highly acidic (pH 4, site T) to intermediate (pH 6, site B) to neutral (pH 7.5, S). These populations were chosen as they represented the extreme ends of adaptive divergence along an acidification gradient in embryonic acid stress tolerance (T most acid tolerant, S most acid sensitive, with B being intermediate; Hangartner et al. 2012; Shu et al. II). During the breeding season of 2012, females and males in breeding condition were collected and transported to the laboratory at Uppsala University (59°50`N, 17°50`E). Artificial crosses and standard rearing were used to establish acid tolerance of each clutch in a common garden laboratory experiment on 7 to 10 clutches per population (Shu et al. II). These same clutches were also studied for variation in macromolecular composition of egg jelly, which indicated high intraspecific variation in egg jelly that surrounds the developing embryos (II, III). All adults were maintained in similar conditions in containers with moist Sphagnum moss (antibacterial medium) in a climate chamber at 2- 4°C until artificial crosses and RNA sampling were conducted a few days later.

Sampling and RNA extraction

To bracket a broad range of gene expression, the individuals chosen for RNA analyses were selected to represent the most acid tolerant and most acid sensitive clutch within each of the three populations (total N = 6). In addition, we included a sample from one female that had not yet fully ovulated at sampling point, hence providing a reference for gene expression at an earlier stage of egg coat production. For each female, total RNA was sampled from the oviduct, which is the specific tissue where egg jelly is produced (Hedrick & Nishihara 1991). Total RNA extraction was conducted using TRIzol (Life Technologies) according to the manufacturer's protocol, followed by DNase (Qiagen) treatment to eliminate potential genomic DNA contamination. Both the concentration and integrity of the RNA samples for transcriptomic analysis were evaluated with the Angilent 2100 Bioanalyzer. Samples with an RNA integrity value (RIN) greater than 8 were used to construct the cDNA libraries. cDNA library construction and sequencing cDNA libraries were generated using the TruSeq RNA-Seq Sample Prep kit according to the manufacturer's protocol (Illumina Inc., San Diego, CA, USA). Briefly, magnetic beads with Oligo(dT) were used to isolate the poly(A)+ mRNA. Fragmentation buffer was added in the presence of divalent cations to break the mRNA into short fragments of approximately 200 bp. Short fragments were purified with QiaQuick PCR extraction kit, followed by end reparation, adding poly(A) and sequencing adapters. The suitable fragments were selected for the PCR amplification as templates. In total, seven cDNA libraries were constructed and sequenced using

102

the Illumina HiSeqTM 2000. The sequencing reactions were conducted at the Beijing Genomics Institute (BGI), Shenzhen.

Assembly and annotation

Raw reads were filtered to remove reads containing adaptors, reads with unknown nucleotides greater than 5% and low-quality reads with more than 20% bases of quality value ≤ 10. Only clean reads were used in the following analyses. Transcriptome de novo assembly was carried out using the assembly program Trinity (Grabherr et al. 2011). Briefly, the software first combined reads of certain lengths of overlap to form longer fragments called contigs. Subsequently, the reads were mapped back to the contigs, which were connected until extension proceeded on neither end (Grabherr et al. 2011). After removing any redundancy, the resulted sequences were defined as unigenes.

Unigene annotation provides the functional information of that gene. Unigenes were first annotated using blastx against the database Nr (http://www.ncbi.nlm.nih.gov/), with a cut-off E- value of 1e-5. To acquire a more comprehensive annotation, Unigene sequences were also aligned to the protein databases Swiss-Prot, KEGG and GO (1e-5) by blastx. In order to predict and classify the possible functions of the unigenes, they were also annotated to Cluster of Orthologous Group (COG), which classifies orthologous gene products (Tatusov et al. 2003).

Protein Coding Sequence Prediction (CDS)

Unigenes were first aligned by blastx (e value < 0.00001) to protein databases in the priority order of NR, Swiss-Prot, KEGG and COG. Unigenes aligned to a higher priority database were not aligned to lower priority database. Proteins with highest ranks in the blast results were used to decide the coding region sequences of the Unigenes, then the coding region sequences were translated into amino acid sequences with the standard codon table. This produced both the nucleotide sequences (5'->3') and amino acid sequences of the Unigene coding region. Unigenes that could not be aligned to any database were scanned by ESTScan (Iseli et al. 1999), producing nucleotide sequence (5’->3’) direction and amino sequence of the predicted coding region.

Simple sequence repeats (SSRs) and Single nucleotide polymorphisms (SNPs)

We further identified gene linked microsatellite markers and SNPs (Rozen & Skaletsky 2000; Li et al. 2009). The details of these are provided in supplementary material.

Metabolic pathway analysis

To investigate the functions of the unigenes in metabolic process, we acquired pathway annotation by mapping the unigenes to (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Kanehisa et al. 2008). Gene Ontology (GO) database is a relational database, which has three ontologies: molecular function, cellular component and

103

biological process. The basic unit of GO is GO-term and each belongs to a type of ontology. KEGG is a database for understanding high-level functions and utilities of the biological system. It is able to analyze gene products during metabolism and cellular processes and allowed us to get pathway annotation for unigenes. We used the Blast2GO program with default setting to acquire GO and KEGG annotation of Unigenes (Conesa et al. 2005).

Identification of maternal effect genes

As candidate ME genes related to R. arvalis egg jelly, we considered unigenes that i) were expressed in all individuals, ii) were mapped to the category of “extracellular matrix” in GO and KEGG annotations and iii) were involved in the glycosylation process of the egg jelly, including O−glycan biosynthesis and proteoglycans, which were considered as likely candidates based on prior work on R. arvalis and other amphibians (Coppin et al. 1999; Strecker et al. 1999; Guerardel et al. 2000;).

Differential expression analysis

Unigene expression was calculated using the FPKM (Fragments Per kb per Million reads), which can eliminate the influence of different gene length and sequencing level on the calculation of gene expression (Mortazavi et al. 2008). Group comparison was performed using the R Bioconductor package NOISeq, which is a novel data-adaptive and nonparametric method (Tarazona et al. 2011). KEGG enrichment analysis was performed with a Fisher’s exact test in Blast2GO (Conesa et al. 2005). Pathway enrichment analysis identifies significantly enriched metabolic pathways or signal transduction pathways in differential expressed genes (DEGs). After correction for multiple testing, we chose pathways with Q value ≤ 0.05 as significantly enriched in DEGs (Conesa et al. 2005; Gotz et al. 2008).

Results

Sequencing and assembly

In total, 53 330 025 420 bp bases were generated from the R. arvalis oviduct transcriptome. The raw reads of seven cDNA libraries will be deposited in the NCBI Sequence Read Archive (SRA) database. Total clean reads of the seven cDNA libraries ranged from 81 166 804 to 87 485 924 (Table 1), with an average GC content of 44.95%. In the final assembly, 69 987 to 112 136 unigenes were detected in the seven cDNA libraries (Table 2). Pooled together, 124 071 unigenes were detected in all cDNA libraries, with an N50 of 1, 212 bp and a total length of 90.3 Mb. The average length of unigenes was 728 bp, and the size distribution of unigenes is shown in Fig. 1.

104

Figure 1. The length distribution of the unigenes identified based on seven R. arvalis oviduct transcriptomes. The X axis shows the length distribution (bp) of sequenced unigenes and Y axis indicates number of unigenes for a given length.

105

Table 1. Results of RNA sequencing from seven R. arvalis oviduct. Total reads and total nucleotides are given after adaptor trimming and quality filtering. Q20 percentage is the proportion of nucleotides with quality value larger than 20; GC percentage is proportion of guanidine and cytosine nucleotides among total nucleotides. The sample ID indicates the seven different females, originating from three populations (T = acid origin, S = neutral origin and B = intermediate pH origin). The B3 individual represents an individual which had not yet fully ovulated at the time of sampling (see material and methods section for further details). 1 represents most acid sensitive while 2 represents the most acid tolerant maternal genotypes (based on screening of embryonic acid tolerance in a laboratory experiment, Shu et al. II).

Sample Total Raw Total Clean Total Clean Q20 GC ID Reads Reads Nucleotides percentage percentage B1 92,353,906 81,166,804 7,305,012,360 98.07% 44.92%

B2 96,108,078 85,373,136 7,683,582,240 98.05% 44.83%

B3 92,534,944 83,077,054 7,476,934,860 98.12% 44.74%

S1 99,177,178 86,475,930 7,782,833,700 98.09% 44.68%

S2 99,884,602 87,485,924 7,873,733,160 98.04% 45.11%

T1 95,886,666 84,428,864 7,598,597,760 98.10% 44.74%

T2 108,026,198 84,548,126 7,609,331,340 98.27% 45.68%

106

Table 2. Results of assembly. N50 represents the contig or unigenes lengths. Total Consensus Sequences represents all assembled unigenes. Distinct Clusters represents the cluster unigenes and the same cluster contains highly similar (more than 70%) unigenes which may come from the same gene or a homologous gene. Distinct Singletons represents a unigene that comes from a single gene. The sample ID indicates the seven different females, originating from three populations (T = acid origin, S = neutral origin and B = intermediate pH origin). The B3 individual represents an individual which had not yet fully ovulated at the time of sampling (see material and methods section for further details).

Sample Total Mean Total Consensus Distinct Distinct Total Length N50 ID Number Length Sequences Clusters Singletons

Contig B1 174,223 49,240,955 283 392 - - -

B2 183,977 51,106,261 278 381 - - -

B3 136,122 40,773,868 300 449 - - -

S1 194,272 53,890,758 277 362 - - -

S2 217,356 59,657,739 274 366 - - -

T1 160,756 43,939,551 273 342 - - -

T2 153,751 46,501,333 302 436 - - -

Unigene B1 90,424 46,895,863 519 830 90,424 13,740 76,684

B2 91,647 47,344,659 517 787 91,647 14,419 77,228

B3 69,987 38,855,076 555 900 69,987 9,972 60,015

S1 104,705 50,126,396 479 693 104,705 15,515 89,190

S2 112,136 55,372,134 494 729 112,136 16,008 96,128

T1 91,528 40,874,115 447 632 91,528 11,288 80,240

T2 87,401 45,945,512 526 839 87,401 12,775 74,626

All 124,071 90,322,330 728 1212 124,071 28,452 95,619

Transcriptome annotation

The E-value distributions of the Unigenes showed that approximately 60% of the Unigenes had very strong homogeny (< 1e-30) with the Nr database, while the rest ranged from 1e-5 to 1e-30 (Fig. 2A). Thirty percent of the Unigene sequences had over 80% similarity with the Nr database, while the remaining 70% of the sequences had a similarity ranging from 17% to 80% (Fig. 2B). Across species used in the analyses, our R. arvalis sequences matched best with Xenopus (Silurana) tropicalis (43.4%), while only 13.1% matched to Xenopus laevis (Fig. 2C). The next best match was a 12.1% matched sequences to the liver fluke Clonorchis sinensis (Figure 2C).

107

Unigenes were also aligned to Cluster of Orthologous Group (COG) of protein database. Functional classes were successfully annotated in 13 501 unigenes using COG (Fig. 3). To get a more comprehensive annotation, unigenes were further annotated by BLASTX against Swiss-Prot, KEGG, NT and GO database, which resulted in 37 262, 31 405, 39 138 and 24 452 hits, respectively (Fig. 4). Altogether, 57 839 R. arvalis unigenes (46.6% of all 124 071 unigenes) had significant matches with existing databases (Fig. 4). Among the 124 071 unigenes, 48 850 (39.4%) were predicted as Protein coding sequences (CDS). Of these, 44 809 unigenes were aligned using existing databases, while 4 041 unigenes which could not be annotated with any database were predicted by EST Scan. The length distribution of the CDS protein sequences is available in the Supplementary Material.

Figure 2 Annotation of R. arvalis unigenes against the Nr database: (A) E-value distribution of the top BLAST hits for each unique sequence; (B) Similarity distribution of the top BLAST hits for each unique sequence; (C) Species distribution of the top BLAST hits for all homologous sequences.

108

Figure 3. COG functional classification of R. arvalis unigenes. X axis shows the different functional classes, Y axis the number of genes annotated into a given class.

70,000 57,839 60,000

50,000 45,071 39,138 37,262 40,000 31,405 30,000 24,452 20,000 13,501 10,000 0

Unigenes annotated Unigenes NR NT Swiss-Prot KEGG COG GO ALL

Public database

Figure 4. Number of unigenes annotated based on different public databases (see methods for details on databases).

109

Functional pathway annotation

For GO annotation, 24 452 unigenes were successfully categorized into 60 GO functional groups (Fig. 5), which could be classified to three major categories: biological processes (23), cellular components (18) and molecular functions (19) (Fig. 5). A total of 31 405 unigenes were annotated in KEGG, which could be assigned to 259 known KEGG pathways (Table S2). The highly enriched pathways included: metabolic pathways (3796, 12.09%), purine metabolism (1910, 6.08%), regulation of actin cytoskeleton (1202, 3.83%), focal adhesion (1161, 3.7%) and calcium signaling pathway (1159, 3.69%).

Figure 5. GO categories of unigenes identified from the transcriptome of seven R. arvalis oviduct samples. The unigenes were annotated in three categories as represented on the x axis: biological processes (23), cellular components (18) and molecular functions (19). The X axis indicates the GO term, while the Y axis indicates the number and percentage of unigenes for each GO term.

Candidate ME genes

We found two groups of candidate ME genes related to egg jelly: 1) egg jelly core protein genes (Table 3) and 2) egg jelly glycosylation genes (Table 4). The major components of egg jelly core protein (extracellular matrix fiber, ECM) are mucin and collagen, of which 13 and 11 different types, respectively, were detected in our dataset (Table 3). Within the Mucin gene family, Mucin-2, Mucin-5AC and Mucin-5B were very highly expressed (Table 3). They were the most abundant transcripts of all unigenes, making them the most likely candidates of egg jelly core protein genes. Several other minor components were also identified, including two types of Lectins (Attractin , Collectin-12), three excellular matrix

110

proteins (Dermatopontin , Fibulin-5 and Fibrinogen-like protein 1 ), one Proteoglycan (Decorin) and four ECM components (EMILIN-1, Fibrillin-1, Fibronectin, Laminin) (Table 3).

Table 3. Candidate ME genes related to egg jelly core protein. Identified from the oviduct of seven R. arvalis females. Asterisk (*) indicates highly expressed genes. ECM = extracellular matrix.

Component Gene Function Mucin Mucin-1 ECM fibers Mucin-2* ECM fibers Mucin-4 ECM fibers Mucin-5AC* ECM fibers Mucin-5B * ECM fibers Mucin-6 ECM fibers Mucin-7 ECM fibers Mucin-15 ECM fibers Collagen Collagen alpha-1(I) ECM fibers Collagen alpha-1(III) ECM fibers Collagen alpha-1(V) ECM fibers Collagen alpha-1(XI) ECM fibers Collagen alpha-1(XII) ECM fibers Collagen alpha-1(XVIII) ECM fibers Collagen alpha-1(XXVII) ECM fibers Collagen alpha-2(I) chain ECM fibers Collagen alpha-2(IV) ECM fibers Collagen alpha-2(V) ECM fibers Collagen alpha-2(VI) ECM fibers Collagen alpha-5(IV) ECM fibers Collagen alpha-6(IV) chain ECM fibers Others Attractin Lectin Collectin-12 Lectin Decorin Proteoglycan Dermatopontin ECM protein EMILIN-1 ECM fibers EMILIN-2 ECM fibers Fibrillin-1 ECM fibers Fibrinogen-like protein 1 ECM protein Fibronectin ECM fibers Fibulin ECM protein Laminin ECM fibers

111

Table 4. Candidate ME genes related to egg jelly glycosylation as identified from the oviduct of seven R. arvalis females. The genes in bold indicate highly expressed enzymes.

Glycan Glycan type Gene Pathway Mucin type beta-1,3-N-acetylglucosaminyltransferase O−Glycan alpha-N-acetylgalactosaminide alpha-2,6-sialyltransferase beta-1,6-N-acetylglucosaminyltransferase beta-1,4-galactosyltransferase 5 glycoprotein-N-acetylgalactosamine 3-beta-galactosyltransferase N-acetylglucosaminyltransferase 3, mucin type polypeptide N-acetylgalactosaminyltransferase sialyltransferase 4A sialyltransferase 7A Other type O-linked Protein O-GlcNAc transferase of O−glycan GlcNAc type O-linked Man beta-1,2-N-acetylglucosaminyltransferase type beta-1,4-galactosyltransferase 1 carbohydrate 3-sulfotransferase 10 dolichyl-phosphate-mannose-protein mannosyltransferase glucuronosyltransferase sialyltransferase 6 4-galactosyl-N-acetylglucosaminide 3-alpha-L-fucosyltransferase O-linked Fuc beta-1,4-galactosyltransferase 1 type peptide-O-fucosyltransferase sialyltransferase 6 O-linked Glc protein glucosyltransferase type UDP-xylose:glucoside alpha-1,3-xylosyltransferase O-linked Gal collagen beta-1,O-galactosyltransferase type lysyl hydroxylase/galactosyltransferase/glucosyltransferase Heparan alpha-1,4-N-acetylglucosaminyltransferase EXTL3 sulfate alpha-1,4-N-acetylglucosaminyltransferase EXTL2 glucuronyl/N-acetylglucosaminyl transferase EXT1 N-deacetylase/N-sulfotransferase (heparan glucosaminyl) 2 Heparan sulfate glucosamine 3-O-sulfotransferase 1 Chondroitin chondroitin sulfate N-acetylgalactosaminyltransferase 1/2 sulfate chondroitin sulfate synthase galactosylxylosylprotein 3-beta-galactosyltransferase galactosylgalactosylxylosylprotein 3-beta-glucuronosyltransferase 1 protein xylosyltransferase xylosylprotein 4-beta-galactosyltransferase Keratan beta-1,4-galactosyltransferase 1

112

sulfate beta-1,4-galactosyltransferase 4 beta-1,3-N-acetylglucosaminyltransferase 7 carbohydrate 6-sulfotransferase 2 N-acetyllactosaminide beta-1,3-N-acetylglucosaminyltransferase sialyltransferase 4A

In addition to the egg jelly core protein genes, five major biosynthesis pathways involved in jelly glycosylation were identified based on KEGG: Mucin type O−Glycans, Other types of O−glycans, Heparan sulfate, Chondroitin sulfate and Keratan sulfate (Table 4). Of these, the Mucin type O−Glycan gene is the most likely candidate involved in jelly glycosylation (Coppin et al. 1999; Lang et al. 2007) (Fig. 6). However, interestingly, of the eight types of Mucin type O−Glycan, only part of these genes were expressed in R. arvalis: genes related to cores 1, 2, 3, 4 and 6 Mucin type O−Glycan biosynthesis were detected while cores 5, 7 and 8 were absent (Fig. 6).

Figure 6. Biosynthesis pathway (KEGG) of the Mucin type O-glycans. Red squares indicate the genes expressed in the oviduct of R. arvalis.

Differential expression analysis

Differences in gene expression were examined, and differentially expressed genes (i.e. DEGs) were identified by pairwise comparisons between each population pair (S vs T, S vs B, T vs B). Overall, 4457, 4198 and 3691 differentially expressed genes were identified between S vs T, S vs B and T vs B, respectively. In general, T and B individuals had much

113

lower gene expression levels compared to S individuals. For instance, 3397 and 3388 genes were down-regulated in the T and B individuals compared to the S individuals, while only 1060 and 810 genes were up-regulated compared to the S individuals. The smallest number of differentially-expressed genes occurred in the T vs B population comparison, in which 2256 were up-regulated and 1435 were down-regulated.

Enrichment of KEGG pathway in the differentially expressed unigenes was assessed using a Fisher’s exact test (FDR < 0.05). The FDR analyses identified 37 and 49 significantly enriched KEGG pathways in T (Table S4) and B (Table S3) compared to S individuals, whereas only two were identified between T and B (Table S5). Again, the suggested expression profiles of T and B individuals were more divergent compared to that of S individuals, while differentiation between the T vs. B was very small. Of these pathways, Ribosome (KO03010) and Oxidative phosphorylation (KO00190) were the most enriched pathways in both S vs. T and S vs. B comparisons (Table S3, S4). This indicated that T and B females had, in general, lower rates of energy production as well as protein biosynthesis.

In general, the genes that were identified as candidate ME genes showed a different expression pattern compared to the global profile (Fig. 7), where there were more genes up- regulated in the T and B individuals than in the S individuals. For instance, Mucin-5AC, Mucin-5B and Mucin-6 (i.e. the major component of jelly core proteins), were relatively highly expressed in the T individuals. However, in general, expression of the candidate ME genes were very diverse across the seven individuals in our data set (Fig. 7), indicating that the jelly phenotypes (i.e. products of these candidate ME genes) would also be extreme variable.

114

Figure 7. Heat map of jelly coat gene expression in oviducts of six R. arvalis females. On the left bar, egg jelly core protein and jelly glycosylation genes are marked as grey and black, respectively. The colours represent high (red), low (green) or average (black) gene expression based on Z-score normalized FPKM values for each gene. The individual females identity from the three study populations (T, S and B) is indicated below. Within each population, the number indicates the individual female, whereby the acid most tolerant clutch belongs to female 2 and acid most sensitive clutch to female 1 (acid tolerance was estimated in Shu et al. II). B3 female was left out from this analysis because it had not fully ovulated at the time of sampling and hence was not directly comparable to the other individuals in terms of expected gene expression patterns.

Discussion

115

Next generation sequencing is an effective way to develop molecular tools in non-model species (Ekblom & Galindo 2011). In this study, we used tissue specific RNA-seq analyses to identify candidate genes for maternal effects (MEs) related to the gelatinous egg coats (i.e. egg jelly) and provide the first transcriptome genomics resources for a non-model amphibian, R. arvalis. Given the role of this species in a broad range of evolutionary ecological questions, we believe this transcriptomic dataset will benefit future studies of molecular ecology and evolution in natural populations.

General genomics aspects

We characterized 124 071 unigenes from R. arvalis transcriptome, and successfully annotated 57 839 of them (46.6% of all 124 071 unigenes). These unigenes mostly related to Xenopus (Silurana) tropicalis (43.4%), while only 13.1% matched to X. laevis (Fig. 2C). We also found that 12.1% of the sequences matched to the liver fluke Clonorchis sinensis (Figure 2C), despite the fact that this species of fluke is not able to infect R arvalis. This could be due to the fact that, in general, amphibian genomes are less well annotated (as only one amphibian genome sequenced thus far), or that genes expressed in the amphibian oviduct may in part orthologous with that of liver fluke.

However, more than half of the unigenes in our database could not be annotated with any existing databases. This mostly reflects the lack of genomics resources in R. arvalis or any closely related frog species, but also in amphibians as a whole. Although numerous genomes are now available (Altshuler et al. 2012; Jones et al. 2012), surprisingly, only one amphibian genome (X. tropicalis) has been sequenced thus far (Hellsten et al. 2010). We believe that more work on amphibian genomics would be highly valuable as 1) amphibians are the most ancient class of land-dwelling vertebrates and their genomic resources are essential for understanding vertebrate development and evolution; 2) the understanding of evolutionary processes in amphibian populations would greatly be facilitated by studies on the genomic architecture of trait variation, and 3) because amphibians are under serious global decline, their genomic diversity is also disappearing (Stuart et al. 2004), and amphibian genomic resources are, hence, urgently needed for amphibian conservation (Calboli et al. 2011). We believe that the transcriptome of R. arvalis sequenced in this study will not only provide valuable information for this species, but also aid to develop amphibian genomics resources. Furthermore, the large number of potentially amplifiable SSRs and SNP markers detected in this study (Supplementary material) represent an important resource for applications in population genetics and for the detection of interesting functional genetic variants (Li et al. 2009; Morin et al. 2004; Schunter et al. 2014).

Candidate ME genes

Maternal effects (MEs), the effects of a mothers phenotype or environment, rather than offsprings own genes, can contribute to adaptive divergence of local populations, allow rapid adaptation, and alter the speed and direction of evolution (Wade 1998; Mousseau & Fox 1998; Räsänen & Kruuk 2007). However, despite these crucial roles of MEs, the genetic architecture of MEs is generally unclear. This is in particularly crucial as MEs are often

116

strongly influenced by environmental variation (e.g. Rossiter 1998; Vivas et al. 2013), which traditionally resulted in MEs being considered a nuisance in evolutionary studies. Although recently the adaptive value and potential for a genetic basis of MEs has been highlighted (e.g. Räsänen & Kruuk 2007), most studies aiming to actually identify “ME genes” have focused on the role of early embryonic development per se (e.g. Tong et al. 2000; Wu et al. 2003). To what extent MEs genes contribute to adaptive divergence of local populations and response to natural selection at early life stages is therefore still largely unknown.

We identified two groups of candidate ME genes based on their role in egg jelly biosynthesis: egg jelly core protein genes and egg jelly glycosylation genes. Of these, particularly the mucin type core protein and the mucin type O-glycan genes are likely candidates underlying jelly coat mediated maternal effects, given that previous work found that amphibian egg jelly mostly consists of mucin type glycoprotein (Coppin et al. 1999; Guerardel et al. 2000; Strecker et al. 1999). We suggest that the candidate genes identified here for egg jelly coat mediated MEs may contribute to adaptive divergence of local populations. However, we also found multiple genes related to jelly coat variation, indicating that the genetic basis of jelly coat mediated adaptive maternal effects could be potentially complex. This parallels previous work on R. arvalis where 19 different jelly glycans were identified (Coppin et al. 1999) and our own work that used a proteomics approach and found polymorphims among individual R. arvalis females in the macromolecular composition of the jelly (Shu et al. II). The potential complexity of the jelly phenotype and its function is also highlighted in the high degree of variation in expression of the putative ME genes across different individuals in our study (Fig. 7). Given the many important roles that the egg jelly coats play in sperm-egg interactions, pathogen defense and responses to various environmental stressors, such as acidity (reviewed in Menkhorst & Selwood 2008; Shu et al. I), this complexity perhaps is not surprising. Future studies should continue to investigate molecular evolution of these “MEs genes”, as well as to study how variation in these genes contribute to adaptive divergence of local populations and responses to natural selection at early life stages.

In conclusion, the transcriptome and molecular markers (SSRs and SNPs; Supplementary material) identified here for R. arvalis will provide valuable genomics resources for a broad range of evolutionary biological studies on R. arvalis, as well as amphibians in general. In particular, we hope that the egg jelly related genes identified here will contribute to disentangle the genetic architecture of egg coat evolution and adaptive maternal effects.

Acknowledgements

We thank Beatrice Lindgren for invaluable help in the field and Dr. Alice Dennis for comments on earlier drafts of this manuscript and. The experiments were conducted under permissions from the Västra Götaland county board and the Ethical committee for animal experiments in Uppsala County. This study was supported by Swiss National Science foundation (to KR).

117

Data Accessibility

Data will be deposited in Dryad. Accession numbers and DOIs will be added after acceptance.

Author Contributions

LS and KR conceived and planned the study. LS performed the experiments. Both authors discussed the results and commented on the manuscript. Both authors read and approved the final manuscript.

References

Altig R, McDiarmid RW (2007) Morphological diversity and evolution of egg and clutch structure in amphibians. Herpetological Monographs 21, 1-32. Altshuler DM, Durbin RM, Abecasis GR, et al. (2012) An integrated map of genetic variation from 1,092 human genomes. Nature 491, 56-65. Babik W, Branicki W, Sandera M, et al. (2004) Mitochondrial phylogeography of the moor frog, Rana arvalis. Molecular Ecology 13, 1469-1480. Balkenhol N, Waits LP (2009) Molecular road ecology: exploring the potential of genetics for investigating transportation impacts on wildlife. Molecular Ecology 18, 4151- 4164. Beebee TJC (2013) Effects of road mortality and mitigation measures on amphibian populations. Conservation Biology 27, 657-668. Calboli FCF, Fisher MC, Garner TWJ, Jehle R (2011) The need for jumpstarting amphibian genome projects. Trends in Ecology & Evolution 26, 378-379. Claw KG, Swanson WJ (2012) Evolution of the Egg: New Findings and Challenges. In: Annual Review of Genomics and Human Genetics, Vol 13 (eds. Chakravarti A, Green E), pp. 109-125. Annual Reviews, Palo Alto. Conesa A, Gotz S, Garcia-Gomez JM, et al. (2005) Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics 21, 3674- 3676. Coppin A, Maes E, Flahaut C, Coddeville B, Strecker G (1999) Acquisition of species- specific O-linked carbohydrate chains from oviducal mucins in Rana arvalis - A case study. European Journal of Biochemistry 266, 370-382. Delplace F, Maes E, Lemoine J, Strecker G (2002) Species specificity of O-linked carbohydrate chains of the oviducal mucins in amphibians: structural analysis of neutral oligosaccharide alditols released by reductive beta-elimination from the egg- jelly coats of Rana clamitans. Biochemical Journal 363, 457-471. Egea-Serrano A, Hangartner S, Laurila A, Räsänen K (2014) Multifarious selection through environmental change: acidity and predator-mediated adaptive divergence in the moor frog (Rana arvalis). Proceedings of the Royal Society B-Biological Sciences 281, 20133266. Ekblom R, Galindo J (2011) Applications of next generation sequencing in molecular ecology of non-model organisms. Heredity 107, 1-15.

118

Glandt D (2006) Der Moorfrosch. Einheit und Vielfalt einer Braunfroschart. Bielefeld (Laurenti Verlag). Gomez-Mestre I, Buchholz DR (2006) Developmental plasticity mirrors differences among taxa in spadefoot toads linking plasticity and diversity. Proceedings of the National Academy of Sciences of the United States of America 103, 19021-19026. Gomez-Mestre I, Touchon JC, Warkentin KM (2006) Amphibian embryo and parental defenses and a larval predator reduce egg mortality from water mold. Ecology 87, 2570-2581. Gotz S, Garcia-Gomez JM, Terol J, et al. (2008) High-throughput functional annotation and data mining with the Blast2GO suite. Nucleic Acids Research 36, 3420-3435. Grabherr MG, Haas BJ, Yassour M, et al. (2011) Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nature Biotechnology 29, 644-U130. Greulich K, Pflugmacher S (2003) Differences in susceptibility of various life stages of amphibians to pesticide exposure. Aquatic Toxicology 65, 329-336. Guerardel Y, Kol O, Maes E, et al. (2000) O-glycan variability of egg-jelly mucins from Xenopus laevis: characterization of four phenotypes that differ by the terminal glycosylation of their mucins. Biochemical Journal 352, 449-463. Hangartner S, Laurila A (2012) Effects of the disinfectant Virkon S on early life-stages of the moor frog (Rana arvalis). Amphibia-Reptilia 33, 349-353. Hangartner S, Laurila A, Räsänen K (2011) Adaptive divergence of the moor frog (Rana arvalis) along an acidification gradient. BMC Evolutionary Biology 11, 366. Hangartner S, Laurila A, Räsänen K (2012) The quantitative genetic basis of adaptive divergence in the moor frog (Rana arvalis) and its implications for gene flow. Journal of Evolutionary Biology 25, 1587-99. Hedrick JL, Nishihara T (1991) Structure and function of the extracellular-matrix of anuran eggs. Journal of Electron Microscopy Technique 17, 319-335. Hellsten U, Harland RM, Gilchrist MJ, et al. (2010) The genome of the western clawed frog Xenopus tropicalis. Science 328, 633-636. Hels T, Buchwald E (2001) The effect of road kills on amphibian populations. Biological Conservation 99, 331-340. Houle D, Govindaraju DR, Omholt S (2010) Phenomics: the next challenge. Nature Reviews Genetics 11, 855-866. Iseli C, Jongeneel CV, Bucher P (1999) ESTScan: a program for detecting, evaluating, and reconstructing potential coding regions in EST sequences. Proc Int Conf Intell Syst Mol Biol, 138-148. Jones FC, Grabherr MG, Chan YF, et al. (2012) The genomic basis of adaptive evolution in threespine sticklebacks. Nature 484, 55-61. Kanehisa M, Araki M, Goto S, et al. (2008) KEGG for linking genomes to life and the environment. Nucleic Acids Research 36, D480-D484. Knopp T, Heimovirta M, Kokko H, Merila J (2008) Do male moor frogs (Rana arvalis) lek with kin? Molecular Ecology 17, 2522-2530. Knopp T, Merilä J (2009a) Microsatellite variation and population structure of the moor frog (Rana arvalis) in Scandinavia. Molecular Ecology 18, 2996-3005.

119

Knopp T, Merilä J (2009b) The postglacial recolonization of Northern Europe by Rana arvalis as revealed by microsatellite and mitochondrial DNA analyses. Heredity 102, 174-181. Lang TA, Hansson GC, Samuelsson T (2007) Gel-forming mucins appeared early in metazoan evolution. Proceedings of the National Academy of Sciences of the United States of America 104, 16209-16214. Laurila A, Pakkasmaa S, Crochet PA, Merilä J (2002) Predator-induced plasticity in early life history and morphology in two anuran amphibians. Oecologia 132, 524-530. Laurila A, Pakkasmaa S, Merilä J (2006) Population divergence in growth rate and antipredator defences in Rana arvalis. Oecologia 147, 585-595. Li RQ, Li YR, Fang XD, et al. (2009) SNP detection for massively parallel whole-genome resequencing. Genome Research 19, 1124-1132. Loman J (2002) Temperature, genetic and hydroperiod effects on metamorphosis of brown frogs Rana arvalis and R. temporaria in the field. Journal of Zoology 258, 115-129. Maack CA, James TC, Champion J, Hunter IR, Tata JR (1985) Xenopus Egg Jelly Coat Proteins .1. Identification and Characterization of Proteins in Individual Coats in Eggs and Oviduct. Comparative Biochemistry and Physiology B-Biochemistry & Molecular Biology 80, 77-87. Menkhorst E, Selwood L (2008) Vertebrate extracellular preovulatory and postovulatory egg coats. Biology of Reproduction 79, 790-797. Merilä J, Söderman F, O'Hara R, Räsänen K, Laurila A (2004) Local adaptation and genetics of acid stress tolerance in the moor frog, Rana arvalis. Conservation Genetics 5, 513- 527. Mitchell-Olds T, Willis JH, Goldstein DB (2007) Which evolutionary processes influence natural genetic variation for phenotypic traits? Nature Reviews Genetics 8, 845-856. Morin PA, Luikart G, Wayne RK, Grp SW (2004) SNPs in ecology, evolution and conservation. Trends in Ecology & Evolution 19, 208-216. Mortazavi A, Williams BA, Mccue K, Schaeffer L, Wold B (2008) Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nature Methods 5, 621-628. Mousseau TA, Fox CW (1998) Maternal Effects As Adaptation. Oxford University Press. Nadeau NJ, Jiggins CD (2010) A golden age for evolutionary genetics? Genomic studies of adaptation in natural populations. Trends in Genetics 26, 484-492. Pakkasmaa S, Merilä J, O'Hara RB (2003) Genetic and maternal effect influences on viability of common frog tadpoles under different environmental conditions. Heredity 91, 117- 124. Persson M, Räsänen K, Laurila A, Merilä J (2007) Maternally determined adaptation to acidity in Rana arvalis: Are laboratory and field estimates of embryonic stress tolerance congruent? Canadian Journal of Zoology-Revue Canadienne De Zoologie 85, 832-838. Räsänen K, Green E (2009) Acidification and its effects on amphibian populations. Amphibian Biology. Conservation and Ecology, Volume 8. Edited by Heatwole H. Surrey Beatty and Sons, Chipping Norton, Australia. Räsänen K, Kruuk LEB (2007) Maternal effects and evolution at ecological time-scales. Functional Ecology 21, 408-421.

120

Räsänen K, Laurila A, Merilä J (2003a) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. I. Local adaptation. Evolution 57, 352-362. Räsänen K, Laurila A, Merilä J (2003b) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. II. Adaptive maternal effects. Evolution 57, 363-371. Räsänen K, Laurila A, Merilä J (2005) Maternal investment in egg size: environment- and population-specific effects on offspring performance. Oecologia 142, 546-553. Räsänen K, Söderman F, Laurila A, Merilä J (2008) Geographic variation in maternal investment: Acidity affects egg size and fecundity in Rana arvalis. Ecology 89, 2553- 2562. Richter-Boix A, Quintela M, Segelbacher G, Laurila A (2011) Genetic analysis of differentiation among breeding ponds reveals a candidate gene for local adaptation in Rana arvalis. Molecular Ecology 20, 1582-1600. Schunter C, Garza JC, Macpherson E, Pascual M (2014) SNP development from RNA-seq data in a nonmodel fish: how many individuals are needed for accurate allele frequency prediction? Molecular Ecology Resources 14, 157-165. Strecker G, Coppin A, Maes E, Morelle W (1999) Structural analysis of 13 neutral oligosaccharide-alditols released by reductive beta-elimination from oviducal mucins of Rana temporaria. European Journal of Biochemistry 266, 94-104. Stuart SN, Chanson JS, Cox NA, et al. (2004) Status and trends of amphibian declines and extinctions worldwide. Science 306, 1783-1786. Tarazona S, Garcia-Alcalde F, Dopazo J, Ferrer A, Conesa A (2011) Differential expression in RNA-seq: a matter of depth. Genome Research 21, 2213-2223. Tatusov RL, Fedorova ND, Jackson JD, et al. (2003) The COG database: an updated version includes eukaryotes. Bmc Bioinformatics 4, 41. Tong ZB, Gold L, Pfeifer KE, et al. (2000) Mater, a maternal effect gene required for early embryonic development in mice. Nature Genetics 26, 267-268. Touchon JC, Gomez-Mestre I, Warkentin KM (2006) Hatching plasticity in two temperate anurans: responses to a pathogen and predation cues. Canadian Journal of Zoology- Revue Canadienne De Zoologie 84, 556-563. Van Buskirk J, Arioli M (2005) Habitat specialization and adaptive phenotypic divergence of anuran populations. Journal of Evolutionary Biology 18, 596-608. Van Buskirk J, McCollum SA, Werner EE (1997) Natural selection for environmentally induced phenotypes in tadpoles. Evolution 51, 1983-1992. Van Buskirk J, Relyea RA (1998) Selection for phenotypic plasticity in Rana sylvatica tadpoles. Biological Journal of the Linnean Society 65, 301-328. Vivas M, Zas R, Sampedro L, Solla A (2013) Environmental maternal effects mediate the resistance of maritime pine to biotic stress. Plos One 8, doi:10.1371/ journal.pone. 0070148 Wade MJ (1998) The evolutionary genetics of maternal effects. In Maternal Effects as Adaptations. Mousseau TA and Fox CW (ed.), Oxford University Press, pp. 5-21. Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for transcriptomics. Nature Reviews Genetics 10, 57-63. Wong JL, Wessel GM (2006) Defending the zygote: search for the ancestral animal block to polyspermy. Current Topics in Developmental Biology, Vol 72 72, 1-161.

121

Wu XM, Viveiros MM, Eppig JJ, et al. (2003) Zygote arrest 1 (Zar1) is a novel maternal- effect gene critical for the oocyte-to-embryo transition. Nature Genetics 33, 187-191.

Supplementary Information

SI methods

Simple sequence repeats (SSRs) analysis

Simple sequence repeats (SSRs) were identified in the final assembly with the software MicroSAtellite (MISA, http://pgrc.ipk-gatersleben.de/misa/) using unigenes as the reference. Primers were designed using the primer3 (Rozen & Skaletsky 2000). Primer design parameters were set as follows: length range = 18–28 nucleotides; PCR product size range from 80 to 300 bp; average annealing temperature = 60 °C; GC content 40–60%.

Single nucleotide polymorphisms (SNPs) analysis

Assembled contigs were scanned for SNPs with SNP detection software SOAPsnp (Li et al. 2009ab). The program can assemble consensus sequence for the genome of a newly sequenced individual based on the alignment of the raw sequencing reads on the unigenes. The SNPs can then be identified on the consensus sequence through the comparison with the unigenes. The program calculated the likelihood of each genotype at each site based on the alignment of short reads to a unigenes set together with the corresponding sequencing quality scores. It then inferred the genotype with highest posterior probability at each site based on Bayes' theorem (the reverse probability model). Therefore, the intrinsic bias or errors that are common in Illumina GA sequencing data have been taken into account and the quality values for use in inferring consensus sequence have been recalibrated.

SI result and discussion

SSRs and SNPs

A total of 26 711 SSRs were identified in all unigenes, in which Mono-nucleotide repeats (19 215), Di- nucleotide repeats (5 050) and Tri-nucleotide repeats (2 002) were the most abundant SSR motif classes, and Quad-nucleotide repeat (345), Penta-nucleotide repeats (79) and Hexa-nucleotide repeats (20) were detected at much lower frequencies (Fig. S1, Table S2). Among Mono-nucleotide repeats, (A/T)n was the most abundant motif. Among Di- nucleotide repeats, the most abundant repeats were (AC/GT)n, (AG/CT)n and (AT/AT)n. Among Tri-nucleotide motifs, the most abundant repeats were (AAT/ATT)n and (AGG/CCT)n. Among Quad-nucleotide motifs, the most abundant repeats were (AGAT/ATCT)n and (AAAG/CTTT)n. Among Penta-nucleotide motifs, the most abundant repeats were (AGAGG/CCTCT)n and (AAAAT/ATTTT)n. Among Hexa-nucleotide motifs, the most abundant repeats were (AAAAAT/ATTTTT)n and (AGGGGC/CCCCTG)n. After

122

filtration, 8 073 high-quality primer pairs amplifying a single product were obtained (see Supporting information for details), which could be used for a wide range of applications in molecular ecology or population genetic studies.

We identified a total of 231 274 SNPs in all unigenes. The number of SNPs in seven individual cDNA library ranged from 63 354 to 86 608 (Fig. S2). The average Ts/Tv ratio (the numbers of transitions and transversions at the SNP sites) of all SNPs was ca. 1.75.

Table S1. Distribution of SSRs in the R. arvalis oviduct transcriptome based on seven individuals.

Quad- Penta- Number of Mono-nucleotide Di-nucleotide Tri-nucleotide Hexa-nucleotide nucleotide nucleotide repeats repeats repeats repeats repeats repeat repeats

4 0 0 0 0 69 20 5 0 0 1,180 292 10 0 6 0 2,048 472 50 0 0 7 0 1,068 286 1 0 0 8 0 645 60 1 0 0 9 0 467 3 1 0 0 10 0 456 1 0 0 0 11 0 348 0 0 0 0 12 5,141 16 0 0 0 0 13 3,858 1 0 0 0 0 14 2,742 0 0 0 0 0 15 2,105 0 0 0 0 0 16 1,488 0 0 0 0 0 17 977 0 0 0 0 0 18 826 0 0 0 0 0 19 750 0 0 0 0 0 20 702 0 0 0 0 0 21 397 1 0 0 0 0 22 172 0 0 0 0 0 23 53 0 0 0 0 0

123

24 4 0 0 0 0 0 Sub Total 19,215 5,050 2,002 345 79 20

Figure S1 Statistics of SSR classification from R. arvalis transcriptome based on oviduct samples from seven individuals.

124

Figure S2 Statistics of SNP numbers in seven cDNA libraries. The X-axis indicates the SNP types, the Y-axis the number of SNPs for a given type. The colors indicate individual females from different populations (acid origin population, T, neutral origin population, S, and intermediate pH population, B). The numbers 1 to 3 indicate the individual identity of the females.

Table S2. KEGG annonation of R. arvalis oviduct transcriptome.

All genes with Pathway Pathway pathway annotation Level 1 ID (31405) Metabolic pathways 3796 (12.09%) ko01100 Metabolism Purine metabolism 1910 (6.08%) ko00230 Metabolism Huntington's disease 1612 (5.13%) ko05016 Human Diseases Pathways in cancer 1405 (4.47%) ko05200 Human Diseases Regulation of actin cytoskeleton 1202 (3.83%) ko04810 Cellular Processes Focal adhesion 1161 (3.7%) ko04510 Cellular Processes Environmental Information Calcium signaling pathway 1159 (3.69%) ko04020 Processing Influenza A 1121 (3.57%) ko05164 Human Diseases Lysine degradation 1120 (3.57%) ko00310 Metabolism

125

Genetic Information Ubiquitin mediated proteolysis 1097 (3.49%) ko04120 Processing Morphine addiction 1081 (3.44%) ko05032 Human Diseases Herpes simplex infection 1077 (3.43%) ko05168 Human Diseases Epstein-Barr virus infection 1058 (3.37%) ko05169 Human Diseases Fc gamma R-mediated phagocytosis 1051 (3.35%) ko04666 Organismal Systems T cell receptor signaling pathway 967 (3.08%) ko04660 Organismal Systems Protein processing in endoplasmic Genetic Information 950 (3.02%) ko04141 reticulum Processing Genetic Information Spliceosome 910 (2.9%) ko03040 Processing Environmental Information Cell adhesion molecules (CAMs) 886 (2.82%) ko04514 Processing Olfactory transduction 886 (2.82%) ko04740 Organismal Systems Tight junction 859 (2.74%) ko04530 Cellular Processes Endocytosis 852 (2.71%) ko04144 Cellular Processes Environmental Information MAPK signaling pathway 797 (2.54%) ko04010 Processing Genetic Information RNA transport 792 (2.52%) ko03013 Processing Taste transduction 790 (2.52%) ko04742 Organismal Systems Pyrimidine metabolism 773 (2.46%) ko00240 Metabolism Vascular smooth muscle contraction 763 (2.43%) ko04270 Organismal Systems Alzheimer's disease 753 (2.4%) ko05010 Human Diseases Primary immunodeficiency 729 (2.32%) ko05340 Human Diseases HTLV-I infection 683 (2.17%) ko05166 Human Diseases Hepatitis C 646 (2.06%) ko05160 Human Diseases Amoebiasis 639 (2.03%) ko05146 Human Diseases Genetic Information Ribosome biogenesis in eukaryotes 636 (2.03%) ko03008 Processing Insulin signaling pathway 634 (2.02%) ko04910 Organismal Systems Measles 629 (2%) ko05162 Human Diseases Transcriptional misregulation in 615 (1.96%) ko05202 Human Diseases cancer Salmonella infection 614 (1.96%) ko05132 Human Diseases Phagosome 575 (1.83%) ko04145 Cellular Processes Chemokine signaling pathway 571 (1.82%) ko04062 Organismal Systems Environmental Information NF-kappa B signaling pathway 570 (1.81%) ko04064 Processing Dilated cardiomyopathy 566 (1.8%) ko05414 Human Diseases Tuberculosis 563 (1.79%) ko05152 Human Diseases Adherens junction 561 (1.79%) ko04520 Cellular Processes Genetic Information RNA polymerase 546 (1.74%) ko03020 Processing Environmental Information Wnt signaling pathway 544 (1.73%) ko04310 Processing Hypertrophic cardiomyopathy 528 (1.68%) ko05410 Human Diseases (HCM) Bile secretion 498 (1.59%) ko04976 Organismal Systems Lysosome 493 (1.57%) ko04142 Cellular Processes

126

Environmental Information ECM-receptor interaction 484 (1.54%) ko04512 Processing Neurotrophin signaling pathway 476 (1.52%) ko04722 Organismal Systems Viral myocarditis 475 (1.51%) ko05416 Human Diseases Genetic Information mRNA surveillance pathway 461 (1.47%) ko03015 Processing Pathogenic Escherichia coli 458 (1.46%) ko05130 Human Diseases infection Axon guidance 456 (1.45%) ko04360 Organismal Systems Osteoclast differentiation 440 (1.4%) ko04380 Organismal Systems Leukocyte transendothelial 438 (1.39%) ko04670 Organismal Systems migration Vibrio cholerae infection 438 (1.39%) ko05110 Human Diseases Environmental Information Hedgehog signaling pathway 434 (1.38%) ko04340 Processing Basal cell carcinoma 430 (1.37%) ko05217 Human Diseases B cell receptor signaling pathway 429 (1.37%) ko04662 Organismal Systems Environmental Information ABC transporters 425 (1.35%) ko02010 Processing Prostate cancer 414 (1.32%) ko05215 Human Diseases Steroid hormone biosynthesis 412 (1.31%) ko00140 Metabolism Bacterial invasion of epithelial cells 411 (1.31%) ko05100 Human Diseases RIG-I-like receptor signaling 405 (1.29%) ko04622 Organismal Systems pathway Cardiac muscle contraction 400 (1.27%) ko04260 Organismal Systems Toxoplasmosis 391 (1.25%) ko05145 Human Diseases Dopaminergic synapse 389 (1.24%) ko04728 Organismal Systems Cell cycle 388 (1.24%) ko04110 Cellular Processes Oocyte meiosis 381 (1.21%) ko04114 Cellular Processes Alcoholism 380 (1.21%) ko05034 Human Diseases Primary bile acid biosynthesis 368 (1.17%) ko00120 Metabolism Shigellosis 362 (1.15%) ko05131 Human Diseases Small cell lung cancer 359 (1.14%) ko05222 Human Diseases Natural killer cell mediated 354 (1.13%) ko04650 Organismal Systems cytotoxicity Long-term potentiation 353 (1.12%) ko04720 Organismal Systems Genetic Information Basal transcription factors 352 (1.12%) ko03022 Processing Neuroactive ligand-receptor Environmental Information 350 (1.11%) ko04080 interaction Processing Progesterone-mediated oocyte 350 (1.11%) ko04914 Organismal Systems maturation Serotonergic synapse 348 (1.11%) ko04726 Organismal Systems Environmental Information ErbB signaling pathway 329 (1.05%) ko04012 Processing Protein digestion and absorption 327 (1.04%) ko04974 Organismal Systems Salivary secretion 327 (1.04%) ko04970 Organismal Systems Gastric acid secretion 315 (1%) ko04971 Organismal Systems Oxidative phosphorylation 311 (0.99%) ko00190 Metabolism Parkinson's disease 310 (0.99%) ko05012 Human Diseases

127

Vasopressin-regulated water 310 (0.99%) ko04962 Organismal Systems reabsorption Melanogenesis 308 (0.98%) ko04916 Organismal Systems Chronic myeloid leukemia 308 (0.98%) ko05220 Human Diseases Amyotrophic lateral sclerosis (ALS) 305 (0.97%) ko05014 Human Diseases Starch and sucrose metabolism 304 (0.97%) ko00500 Metabolism Renal cell carcinoma 302 (0.96%) ko05211 Human Diseases GnRH signaling pathway 299 (0.95%) ko04912 Organismal Systems Phosphatidylinositol signaling Environmental Information 298 (0.95%) ko04070 system Processing Environmental Information TGF-beta signaling pathway 292 (0.93%) ko04350 Processing NOD-like receptor signaling 291 (0.93%) ko04621 Organismal Systems pathway Hematopoietic cell lineage 288 (0.92%) ko04640 Organismal Systems Genetic Information RNA degradation 286 (0.91%) ko03018 Processing Environmental Information mTOR signaling pathway 285 (0.91%) ko04150 Processing Endometrial cancer 285 (0.91%) ko05213 Human Diseases Gap junction 284 (0.9%) ko04540 Cellular Processes Amino sugar and nucleotide sugar 283 (0.9%) ko00520 Metabolism metabolism Maturity onset diabetes of the 281 (0.89%) ko04950 Human Diseases young Chagas disease (American 280 (0.89%) ko05142 Human Diseases trypanosomiasis) Environmental Information Jak-STAT signaling pathway 275 (0.88%) ko04630 Processing Pancreatic cancer 271 (0.86%) ko05212 Human Diseases Glioma 269 (0.86%) ko05214 Human Diseases Pancreatic secretion 260 (0.83%) ko04972 Organismal Systems Apoptosis 260 (0.83%) ko04210 Cellular Processes Colorectal cancer 258 (0.82%) ko05210 Human Diseases Acute myeloid leukemia 255 (0.81%) ko05221 Human Diseases Arrhythmogenic right ventricular 255 (0.81%) ko05412 Human Diseases cardiomyopathy (ARVC) Genetic Information Ribosome 254 (0.81%) ko03010 Processing Pertussis 244 (0.78%) ko05133 Human Diseases Fc epsilon RI signaling pathway 244 (0.78%) ko04664 Organismal Systems Environmental Information Notch signaling pathway 243 (0.77%) ko04330 Processing Peroxisome 241 (0.77%) ko04146 Cellular Processes Environmental Information VEGF signaling pathway 240 (0.76%) ko04370 Processing Cholinergic synapse 236 (0.75%) ko04725 Organismal Systems Cytokine-cytokine receptor Environmental Information 236 (0.75%) ko04060 interaction Processing Non-small cell lung cancer 235 (0.75%) ko05223 Human Diseases Long-term depression 232 (0.74%) ko04730 Organismal Systems

128

Glutamatergic synapse 231 (0.74%) ko04724 Organismal Systems PPAR signaling pathway 230 (0.73%) ko03320 Organismal Systems Glycolysis / Gluconeogenesis 229 (0.73%) ko00010 Metabolism Inositol phosphate metabolism 224 (0.71%) ko00562 Metabolism Adipocytokine signaling pathway 224 (0.71%) ko04920 Organismal Systems Glycerophospholipid metabolism 216 (0.69%) ko00564 Metabolism Complement and coagulation 214 (0.68%) ko04610 Organismal Systems cascades Carbohydrate digestion and 207 (0.66%) ko04973 Organismal Systems absorption Synaptic vesicle cycle 205 (0.65%) ko04721 Organismal Systems Melanoma 204 (0.65%) ko05218 Human Diseases Legionellosis 203 (0.65%) ko05134 Human Diseases Toll-like receptor signaling pathway 202 (0.64%) ko04620 Organismal Systems Amphetamine addiction 201 (0.64%) ko05031 Human Diseases Antigen processing and presentation 194 (0.62%) ko04612 Organismal Systems Epithelial cell signaling in 187 (0.6%) ko05120 Human Diseases Helicobacter pylori infection Thyroid cancer 182 (0.58%) ko05216 Human Diseases Metabolism of xenobiotics by 181 (0.58%) ko00980 Metabolism cytochrome P450 p53 signaling pathway 179 (0.57%) ko04115 Cellular Processes Genetic Information Fanconi anemia pathway 179 (0.57%) ko03460 Processing Valine, leucine and isoleucine 177 (0.56%) ko00280 Metabolism degradation Retinol metabolism 172 (0.55%) ko00830 Metabolism Pentose and glucuronate 170 (0.54%) ko00040 Metabolism interconversions Retrograde endocannabinoid 168 (0.53%) ko04723 Organismal Systems signaling Arachidonic acid metabolism 166 (0.53%) ko00590 Metabolism Staphylococcus aureus infection 165 (0.53%) ko05150 Human Diseases Galactose metabolism 163 (0.52%) ko00052 Metabolism N-Glycan biosynthesis 163 (0.52%) ko00510 Metabolism Dorso-ventral axis formation 162 (0.52%) ko04320 Organismal Systems Prion diseases 162 (0.52%) ko05020 Human Diseases Drug metabolism - cytochrome 161 (0.51%) ko00982 Metabolism P450 Bladder cancer 161 (0.51%) ko05219 Human Diseases Arginine and proline metabolism 160 (0.51%) ko00330 Metabolism Type II diabetes mellitus 159 (0.51%) ko04930 Human Diseases Fatty acid metabolism 158 (0.5%) ko00071 Metabolism Leishmaniasis 158 (0.5%) ko05140 Human Diseases GABAergic synapse 157 (0.5%) ko04727 Organismal Systems Endocrine and other factor- 154 (0.49%) ko04961 Organismal Systems regulated calcium reabsorption Rheumatoid arthritis 154 (0.49%) ko05323 Human Diseases Aminoacyl-tRNA biosynthesis 154 (0.49%) ko00970 Genetic Information

129

Processing Cytosolic DNA-sensing pathway 152 (0.48%) ko04623 Organismal Systems Genetic Information Nucleotide excision repair 151 (0.48%) ko03420 Processing Pyruvate metabolism 149 (0.47%) ko00620 Metabolism Systemic lupus erythematosus 143 (0.46%) ko05322 Human Diseases Fructose and mannose metabolism 139 (0.44%) ko00051 Metabolism Mineral absorption 138 (0.44%) ko04978 Organismal Systems Drug metabolism - other enzymes 138 (0.44%) ko00983 Metabolism Citrate cycle (TCA cycle) 134 (0.43%) ko00020 Metabolism Propanoate metabolism 132 (0.42%) ko00640 Metabolism Cysteine and methionine 130 (0.41%) ko00270 Metabolism metabolism Ascorbate and aldarate metabolism 130 (0.41%) ko00053 Metabolism Aldosterone-regulated sodium 127 (0.4%) ko04960 Organismal Systems reabsorption Phototransduction - fly 125 (0.4%) ko04745 Organismal Systems Phototransduction 124 (0.39%) ko04744 Organismal Systems Glycerolipid metabolism 122 (0.39%) ko00561 Metabolism Glutathione metabolism 120 (0.38%) ko00480 Metabolism Cocaine addiction 114 (0.36%) ko05030 Human Diseases Sphingolipid metabolism 114 (0.36%) ko00600 Metabolism Genetic Information DNA replication 110 (0.35%) ko03030 Processing Tryptophan metabolism 109 (0.35%) ko00380 Metabolism Genetic Information Proteasome 109 (0.35%) ko03050 Processing Other types of O-glycan 108 (0.34%) ko00514 Metabolism biosynthesis Glycine, serine and threonine 107 (0.34%) ko00260 Metabolism metabolism Ether lipid metabolism 107 (0.34%) ko00565 Metabolism Tyrosine metabolism 106 (0.34%) ko00350 Metabolism Linoleic acid metabolism 105 (0.33%) ko00591 Metabolism Genetic Information Base excision repair 102 (0.32%) ko03410 Processing Malaria 97 (0.31%) ko05144 Human Diseases beta-Alanine metabolism 96 (0.31%) ko00410 Metabolism Pentose phosphate pathway 94 (0.3%) ko00030 Metabolism African trypanosomiasis 92 (0.29%) ko05143 Human Diseases Nicotinate and nicotinamide 92 (0.29%) ko00760 Metabolism metabolism Alanine, aspartate and glutamate 91 (0.29%) ko00250 Metabolism metabolism Glycosaminoglycan biosynthesis - 89 (0.28%) ko00534 Metabolism heparan sulfate Genetic Information Homologous recombination 88 (0.28%) ko03440 Processing Glyoxylate and dicarboxylate 86 (0.27%) ko00630 Metabolism metabolism

130

Circadian rhythm - mammal 84 (0.27%) ko04710 Organismal Systems Porphyrin and chlorophyll 82 (0.26%) ko00860 Metabolism metabolism Butanoate metabolism 82 (0.26%) ko00650 Metabolism Autoimmune thyroid disease 81 (0.26%) ko05320 Human Diseases Environmental Information MAPK signaling pathway - fly 78 (0.25%) ko04013 Processing Histidine metabolism 78 (0.25%) ko00340 Metabolism Allograft rejection 77 (0.25%) ko05330 Human Diseases SNARE interactions in vesicular Genetic Information 76 (0.24%) ko04130 transport Processing Fat digestion and absorption 75 (0.24%) ko04975 Organismal Systems Genetic Information Mismatch repair 74 (0.24%) ko03430 Processing Vitamin digestion and absorption 73 (0.23%) ko04977 Organismal Systems Collecting duct acid secretion 73 (0.23%) ko04966 Organismal Systems Glycosylphosphatidylinositol(GPI)- 73 (0.23%) ko00563 Metabolism anchor biosynthesis Genetic Information Protein export 72 (0.23%) ko03060 Processing Graft-versus-host disease 69 (0.22%) ko05332 Human Diseases Type I diabetes mellitus 69 (0.22%) ko04940 Human Diseases Terpenoid backbone biosynthesis 66 (0.21%) ko00900 Metabolism Biosynthesis of unsaturated fatty 65 (0.21%) ko01040 Metabolism acids Regulation of autophagy 63 (0.2%) ko04140 Cellular Processes Fatty acid elongation 63 (0.2%) ko00062 Metabolism Mucin type O-Glycan biosynthesis 58 (0.18%) ko00512 Metabolism Proximal tubule bicarbonate 58 (0.18%) ko04964 Organismal Systems reclamation Other glycan degradation 57 (0.18%) ko00511 Metabolism Glycosaminoglycan degradation 55 (0.18%) ko00531 Metabolism Phenylalanine metabolism 54 (0.17%) ko00360 Metabolism Intestinal immune network for IgA 54 (0.17%) ko04672 Organismal Systems production Glycosphingolipid biosynthesis - 52 (0.17%) ko00601 Metabolism lacto and neolacto series Selenocompound metabolism 49 (0.16%) ko00450 Metabolism Ubiquinone and other terpenoid- 46 (0.15%) ko00130 Metabolism quinone biosynthesis Genetic Information Non-homologous end-joining 46 (0.15%) ko03450 Processing Steroid biosynthesis 45 (0.14%) ko00100 Metabolism Glycosphingolipid biosynthesis - 44 (0.14%) ko00604 Metabolism ganglio series Pantothenate and CoA biosynthesis 43 (0.14%) ko00770 Metabolism Circadian rhythm - fly 42 (0.13%) ko04711 Organismal Systems alpha-Linolenic acid metabolism 40 (0.13%) ko00592 Metabolism Butirosin and neomycin 39 (0.12%) ko00524 Metabolism biosynthesis

131

Glycosaminoglycan biosynthesis - 38 (0.12%) ko00533 Metabolism keratan sulfate Genetic Information Sulfur relay system 38 (0.12%) ko04122 Processing One carbon pool by folate 37 (0.12%) ko00670 Metabolism Glycosaminoglycan biosynthesis - 35 (0.11%) ko00532 Metabolism chondroitin sulfate Riboflavin metabolism 34 (0.11%) ko00740 Metabolism Glycosphingolipid biosynthesis - 34 (0.11%) ko00603 Metabolism globo series Caffeine metabolism 32 (0.1%) ko00232 Metabolism Asthma 32 (0.1%) ko05310 Human Diseases Renin-angiotensin system 31 (0.1%) ko04614 Organismal Systems Folate biosynthesis 29 (0.09%) ko00790 Metabolism Sulfur metabolism 29 (0.09%) ko00920 Metabolism Fatty acid biosynthesis 24 (0.08%) ko00061 Metabolism Synthesis and degradation of ketone 23 (0.07%) ko00072 Metabolism bodies Cyanoamino acid metabolism 17 (0.05%) ko00460 Metabolism Taurine and hypotaurine 16 (0.05%) ko00430 Metabolism metabolism Vitamin B6 metabolism 14 (0.04%) ko00750 Metabolism Nicotine addiction 13 (0.04%) ko05033 Human Diseases Phenylalanine, tyrosine and 13 (0.04%) ko00400 Metabolism tryptophan biosynthesis D-Arginine and D-ornithine 11 (0.04%) ko00472 Metabolism metabolism Valine, leucine and isoleucine 11 (0.04%) ko00290 Metabolism biosynthesis Thiamine metabolism 10 (0.03%) ko00730 Metabolism Lipoic acid metabolism 6 (0.02%) ko00785 Metabolism Biotin metabolism 6 (0.02%) ko00780 Metabolism D-Glutamine and D-glutamate 5 (0.02%) ko00471 Metabolism metabolism Insect hormone biosynthesis 4 (0.01%) ko00981 Metabolism Polyketide sugar unit biosynthesis 2 (0.01%) ko00523 Metabolism Lysine biosynthesis 2 (0.01%) ko00300 Metabolism

Table S3 Enrichment of KEGG pathway in the differentially expressed unigenes between S and B populations.

All genes with DEGs with pathway Pathway Pathway pathway Pvalue Qvalue annotation ID annotation (1717) (31405) Ribosome 87 (5.07%) 254 (0.81%) 0.000 0.000 ko03010

132

Oxidative phosphorylation 82 (4.78%) 311 (0.99%) 0.000 0.000 ko00190 Parkinson's disease 74 (4.31%) 310 (0.99%) 0.000 0.000 ko05012 Proteasome 44 (2.56%) 109 (0.35%) 0.000 0.000 ko03050 Metabolic pathways 313 (18.23%) 3796 (12.09%) 0.000 0.000 ko01100 Alzheimer's disease 88 (5.13%) 753 (2.4%) 0.000 0.000 ko05010 Lysosome 66 (3.84%) 493 (1.57%) 0.000 0.000 ko04142 Phototransduction - fly 28 (1.63%) 125 (0.4%) 0.000 0.000 ko04745 Phagosome 70 (4.08%) 575 (1.83%) 0.000 0.000 ko04145 Vibrio cholerae infection 57 (3.32%) 438 (1.39%) 0.000 0.000 ko05110 Antigen processing and 34 (1.98%) 194 (0.62%) 0.000 0.000 ko04612 presentation Pathogenic Escherichia coli 56 (3.26%) 458 (1.46%) 0.000 0.000 ko05130 infection Glycolysis / Gluconeogenesis 34 (1.98%) 229 (0.73%) 0.000 0.000 ko00010 Citrate cycle (TCA cycle) 24 (1.4%) 134 (0.43%) 0.000 0.000 ko00020 Protein processing in 87 (5.07%) 950 (3.02%) 0.000 0.000 ko04141 endoplasmic reticulum Protein export 15 (0.87%) 72 (0.23%) 0.000 0.000 ko03060 Gastric acid secretion 36 (2.1%) 315 (1%) 0.000 0.000 ko04971 Collecting duct acid secretion 14 (0.82%) 73 (0.23%) 0.000 0.000 ko04966 Amphetamine addiction 26 (1.51%) 201 (0.64%) 0.000 0.001 ko05031 Alcoholism 40 (2.33%) 380 (1.21%) 0.000 0.001 ko05034 N-Glycan biosynthesis 22 (1.28%) 163 (0.52%) 0.000 0.001 ko00510 Spliceosome 76 (4.43%) 910 (2.9%) 0.000 0.002 ko03040 Pentose phosphate pathway 15 (0.87%) 94 (0.3%) 0.000 0.002 ko00030 Pyruvate metabolism 20 (1.16%) 149 (0.47%) 0.000 0.002 ko00620 Oocyte meiosis 38 (2.21%) 381 (1.21%) 0.000 0.003 ko04114 Rheumatoid arthritis 20 (1.16%) 154 (0.49%) 0.000 0.003 ko05323 Systemic lupus erythematosus 19 (1.11%) 143 (0.46%) 0.000 0.003 ko05322 Huntington's disease 118 (6.87%) 1612 (5.13%) 0.001 0.007 ko05016 Aminoacyl-tRNA biosynthesis 19 (1.11%) 154 (0.49%) 0.001 0.007 ko00970 Other types of O-glycan 15 (0.87%) 108 (0.34%) 0.001 0.007 ko00514 biosynthesis Synaptic vesicle cycle 23 (1.34%) 205 (0.65%) 0.001 0.007 ko04721 Galactose metabolism 19 (1.11%) 163 (0.52%) 0.002 0.012 ko00052 PPAR signaling pathway 24 (1.4%) 230 (0.73%) 0.002 0.014 ko03320 Gap junction 28 (1.63%) 284 (0.9%) 0.002 0.014 ko04540 Glutathione metabolism 15 (0.87%) 120 (0.38%) 0.002 0.016 ko00480 Propanoate metabolism 16 (0.93%) 132 (0.42%) 0.002 0.016 ko00640 Fat digestion and absorption 11 (0.64%) 75 (0.24%) 0.002 0.016 ko04975 Cardiac muscle contraction 36 (2.1%) 400 (1.27%) 0.002 0.016 ko04260 Insulin signaling pathway 52 (3.03%) 634 (2.02%) 0.002 0.016 ko04910 Phototransduction 15 (0.87%) 124 (0.39%) 0.003 0.020 ko04744 Vasopressin-regulated water 29 (1.69%) 310 (0.99%) 0.004 0.022 ko04962 reabsorption Tryptophan metabolism 13 (0.76%) 109 (0.35%) 0.007 0.039 ko00380

133

Long-term potentiation 31 (1.81%) 353 (1.12%) 0.007 0.039 ko04720 Arginine and proline 17 (0.99%) 160 (0.51%) 0.007 0.039 ko00330 metabolism Legionellosis 20 (1.16%) 203 (0.65%) 0.008 0.045 ko05134 Valine, leucine and isoleucine 18 (1.05%) 177 (0.56%) 0.008 0.046 ko00280 degradation Pancreatic secretion 24 (1.4%) 260 (0.83%) 0.009 0.047 ko04972 Glycosaminoglycan 11 (0.64%) 89 (0.28%) 0.009 0.048 ko00534 biosynthesis - heparan sulfate Starch and sucrose metabolism 27 (1.57%) 304 (0.97%) 0.009 0.048 ko00500

Table S4 Enrichment of KEGG pathway in the differentially expressed unigenes between S and T populations.

All genes with DEGs with pathway Pathway Pathway pathway Pvalue Qvalue annotation ID annotation (1529) (31405) Ribosome 87 (5.69%) 254 (0.81%) 0.000 0.000 ko03010 Oxidative phosphorylation 73 (4.77%) 311 (0.99%) 0.000 0.000 ko00190 Parkinson's disease 68 (4.45%) 310 (0.99%) 0.000 0.000 ko05012 Proteasome 41 (2.68%) 109 (0.35%) 0.000 0.000 ko03050 Alzheimer's disease 82 (5.36%) 753 (2.4%) 0.000 0.000 ko05010 Antigen processing and 35 (2.29%) 194 (0.62%) 0.000 0.000 ko04612 presentation Lysosome 60 (3.92%) 493 (1.57%) 0.000 0.000 ko04142 Pathogenic Escherichia coli 54 (3.53%) 458 (1.46%) 0.000 0.000 ko05130 infection Vibrio cholerae infection 52 (3.4%) 438 (1.39%) 0.000 0.000 ko05110 Metabolic pathways 260 (17%) 3796 (12.09%) 0.000 0.000 ko01100 Phagosome 61 (3.99%) 575 (1.83%) 0.000 0.000 ko04145 Phototransduction - fly 21 (1.37%) 125 (0.4%) 0.000 0.000 ko04745 Glycolysis / Gluconeogenesis 27 (1.77%) 229 (0.73%) 0.000 0.000 ko00010 Citrate cycle (TCA cycle) 19 (1.24%) 134 (0.43%) 0.000 0.000 ko00020 Protein processing in 75 (4.91%) 950 (3.02%) 0.000 0.000 ko04141 endoplasmic reticulum Protein export 13 (0.85%) 72 (0.23%) 0.000 0.001 ko03060 Collecting duct acid 13 (0.85%) 73 (0.23%) 0.000 0.001 ko04966 secretion Huntington's disease 110 (7.19%) 1612 (5.13%) 0.000 0.003 ko05016 Glutathione metabolism 16 (1.05%) 120 (0.38%) 0.000 0.003 ko00480 Alcoholism 35 (2.29%) 380 (1.21%) 0.000 0.003 ko05034 Fat digestion and absorption 12 (0.78%) 75 (0.24%) 0.000 0.003 ko04975 Amphetamine addiction 22 (1.44%) 201 (0.64%) 0.000 0.004 ko05031

134

N-Glycan biosynthesis 18 (1.18%) 163 (0.52%) 0.001 0.011 ko00510 Cardiac muscle contraction 34 (2.22%) 400 (1.27%) 0.001 0.013 ko04260 Spliceosome 65 (4.25%) 910 (2.9%) 0.001 0.014 ko03040 Proximal tubule bicarbonate 9 (0.59%) 58 (0.18%) 0.002 0.018 ko04964 reclamation Arginine and proline 17 (1.11%) 160 (0.51%) 0.002 0.019 ko00330 metabolism Pyruvate metabolism 16 (1.05%) 149 (0.47%) 0.003 0.022 ko00620 Phototransduction 14 (0.92%) 124 (0.39%) 0.003 0.025 ko04744 Gastric acid secretion 27 (1.77%) 315 (1%) 0.003 0.027 ko04971 Rheumatoid arthritis 16 (1.05%) 154 (0.49%) 0.004 0.028 ko05323 Glycosaminoglycan 11 (0.72%) 89 (0.28%) 0.004 0.030 ko00534 biosynthesis - heparan sulfate Systemic lupus 15 (0.98%) 143 (0.46%) 0.004 0.032 ko05322 erythematosus Insulin signaling pathway 46 (3.01%) 634 (2.02%) 0.005 0.035 ko04910 Legionellosis 19 (1.24%) 203 (0.65%) 0.005 0.035 ko05134 Synaptic vesicle cycle 19 (1.24%) 205 (0.65%) 0.006 0.038 ko04721 Arachidonic acid metabolism 16 (1.05%) 166 (0.53%) 0.007 0.048 ko00590

Table S5 Enrichment of KEGG pathway in the differentially expressed unigenes between T and B populations.

DEGs with All genes with Pathway Pathway pathway pathway Pvalue Qvalue ID annotation (615) annotation (31405)

Morphine addiction 39 (6.34%) 1081 (3.44%) 0.000 0.030 ko05032 Salivary secretion 17 (2.76%) 327 (1.04%) 0.000 0.030 ko04970

135

Chapter V

Environmental stress mediated adaptive divergence in ion channel function

during embryogenesis in Rana arvalis

Longfei Shu1, Anssi Laurila2 and Katja Räsänen1

1Eawag, Department of Aquatic Ecology, Switzerland and ETH Zurich, Institute of

Integrative Biology, Switzerland

2Animal Ecology/Department of Ecology and Genetics, Evolutionary Biology Center,

Uppsala University, Sweden

136

Abstract

Ion channels and pumps are responsible for ion flux in cells, and are key mechanisms mediating cellular function. Many environmental stressors, such as salinity and acidification, are known to severely disrupt ionic balance of organisms thereby challenging fitness of natural populations. Although ion channels can have several vital functions during early life- stages (e.g. embryogenesis), it is currently not known i) how developing embryos maintain proper intracellular conditions when exposed to environmental stress and ii) to what extent environmental stress can drive intra-specific divergence in ion channels. Here we studied the moor frog, Rana arvalis, from three divergent populations to investigate the role of different ion channels and pumps for embryonic survival under acid stress (pH 4 vs 7.5) and whether populations adapted to contrasting acidities differ in ion channel/pump functioning. We found that ion channels that mediate Ca2+ influx are essential for embryonic survival under acidic pH, and, intriguingly, that populations show differences in calcium channel function. Our results suggest that adaptive divergence in embryonic acid stress tolerance of amphibians may in part be mediated by Ca2+ balance. We suggest that ion channel function may mediate adaptive divergence of natural populations at early life-stages in the face of environmental stress.

Keywords: Amphibian; Adaptive divergence; Environmental stress; Embryogensis; Ion channels

137

Introduction

Eukaryotic cells are highly compartmentalized, wherein protons and other ions (e.g. Na+, K+ and Ca2+ ) play a crucial role in signaling, maintaining the structure and function of proteins, and storing energy as an electrochemical gradient across the membrane (Casey et al. 2009). Ion channels and pumps, which are responsible for ion flux, have been extensively investigated in relation to nervous systems (Noskov 2011; Roux et al. 2011). However, relatively little is known on the function of specific channels and pumps during embryogenesis. Studies in model systems, such as mouse, chicken and Xenopus, indicate dynamic expression of ion channels and pumps during embryogenesis (Rutenberg et al. 2002; Leclerc et al. 2006; Hur et al. 2012) – whereas studies on natural populations facing variable ecological conditions are essentially missing.

This gap is particularly important as embryos of many aquatic taxa develop in direct contact with the external environment and are highly influenced by environmental stressors, such as salinity, temperature and pH (Parker et al. 2009; Gonzalez 2012; Stumpp et al. 2012), that mediate their negative effects via the disruption of ion balance and thereby affect the reproductive success and viability of natural populations. Moreover, environmental stress can have strong ecological and evolutionary consequences on natural populations and cause strong selection at short time scales (Hoffmann & Parsons 1997). It is therefore of key interest to investigate how developing embryos maintain proper intracellular conditions under stressful conditions, and whether environmental stress can cause selection on ion channels and pumps during embryogenesis. Most studies so far have focused on the constitutive role of ion channels and pumps during early embryogenesis, such as neural induction, cavitation or gap junctions (Tosti 2010). To our knowledge, no studies have investigated the role of environmental stress on ion channels and pumps during embryogenesis. Moreover, very few studies have investigated intra-specific variation (e.g. variation within and among populations) in ion balance in natural populations – in particular in relation to early life stages (see McCairns & Bernatchez 2010; Lee et al. 2011; in adults).

One potential source of environmental stress, that may impose strong selection in ion channel function, is environmental acidification. Environmental acidification is a global environmental problem in both freshwater and marine ecosystems (e.g. Räsänen & Green 2009; Kroeker et al. 2010). The negative effects of acid stress are, in part, mediated via the disrupted ion balance, as shown in fish (Kwong et al. 2014), larval amphibians (Freda & Dunson 1984; Freda & Dunson 1985), sea urchins (Stumpp et al. 2012) and marine phytoplankton (Taylor et al. 2011). However, how acid stress affects functioning of ion channels and pumps during embryogenesis has not been studied to date. This is important because in fish and amphibians acid stress typically causes reduced reproductive success due to high embryonic mortality (e.g. Räsänen & Green 2009; Kwong et al. 2014). Amphibians can be strongly negatively affected by acidity at all life-stages (reviewed in Räsänen & Green 2009). The strongly reduced embryonic survival under acidic conditions has traditionally been explained by a “curling defect”, whereby embryos develop, but become tightly curled within the egg coat and, finally, fail to hatch under acidic conditions (Dunson & Connell

138

1982; Pierce 1985). Based on egg capsule manipulation experiments, this effect has been suggested to be mediated via chemical alterations in the maternally derived egg capsules (Dunson & Connell 1982; Picker et al. 1993). Recent evidence further indicates that adaptive divergence among and within species arises via egg coat mediated maternal effects (Picker et al. 1993; Räsänen et al. 2003b). However, the potential role of disrupted embryonic ion channel function in acidity induced mortality of embryos, and potential adaptive divergence in ion channels and pumps has not been studied to date.

Here, we combined a common garden laboratory experiment (rearing of embryos at different pHs) with a pharmaceutical approach (ion channel manipulation) to study variation in ion channel and pump function at benign (pH 7.5) and stressful (pH 4) conditions among three populations known to differ in embryonic acid tolerance (Hangartner et al. 2011). Specifically, we used ion channel blockers to manipulate four ion channels and pumps that are expected to be responsible for maintaining H+, Na+ and Ca2+ balance. We focused on these channels and pumps for the following reasons: H+ because “pH” directly indicates the concentration of H+ ions, whereby lower pH reflects higher H+ concentration; Na+ because acidic pH can inhibit Na+ uptake and cause passive Na+ loss in fish and amphibian (Freda & Dunson 1984; Hwang et al. 2011; Kwong et al. 2014), subsequently having sub-lethal and lethal effects and, finally, Ca2+ because water hardness (primarily reflecting Ca2+ ion concentration) affects acid stress tolerance in fish and amphibians and acidified surface waters are typically soft (low Ca2+ concentration, Hendry & Brezonik 1984; Dale et al. 1985; Freda & Dunson 1985). We predicted that i) if these ion channels are generally important during embryogenesis, blocking the relevant ion channels should increase mortality of embryos, but that these effects should be pH dependent if the ion channels are important under acidic conditions, and ii) if populations adapted to different pH conditions have been exposed to divergent selection on ion channels and pumps function, embryos from different populations should show divergent responses to our blocking treatments.

Results

Variation in acid stress tolerance and the role of egg jelly

We initially tested embryonic survival in both presence (henceforth, jellied) and absence (henceforth, de-jellied) of the gelatinous egg jelly coats that surround R. arvalis embryos, because i) under acidic conditions jelly envelopes typically prevent the embryos from hatching (the curling defect) and jelly removal may either increase or decrease embryonic survival (Picker et al. 1993; Räsänen et al. 2003b); ii) jelly envelopes also serve as a protective barrier of embryos from environmental hazards, therefore could potentially affect our inhibitor treatments.

Overall, acid treatment (pH 4) strongly reduced embryonic survival (compared to the neutral, pH 7.5 treatment), but the populations differed in their pH tolerance, whereby population T had over two fold higher survival at pH 4 than the S population and B population was intermediate (when jelly was intact; Table S1, Fig S1). In general, jelly removal had significant effects on hatching success, as indicated by several strong jelly treatment

139

interactions (Fig. S1, Table S1). Furthermore, jelly removal removed differences in acid stress tolerance between the S and B population, whereas the T population had higher survival at pH 4 even after jelly removal (Fig. S1). This result indicates that part of the among population differences in embryonic acid stress tolerance at extreme acidic pHs are independent of jelly coats.

There was a significant inhibitor × jelly × population interaction (P = 0.025; Table S1). Furthermore, jelly clearly affected effectiveness of the inhibitor treatment in many cases (i.e. hatching success was reduced more strongly in the inhibitor treatments when jelly was removed compared to intact jelly treatment; Fig. S1). As our focus here is on ion channel function, we therefore, in the following, focus mainly on the dejellied treatments.

Ion channels and pumps

To investigate the potential role of ionic and acid balance at neutral and acidic pH for embryonic survival, we blocked acid-sensing ion channels (ASICs) by a Amiloride (Ami) treatment (Ugawa et al. 2002), Na+/K+-ATPase (Sodium pumps) by a Ouabain (Oua) treatment (Workman et al. 2003), and Ca2+ ion channels by Verapamil (Ver, L-type voltage- dependent calcium channels specific inhibitor; Amaral et al. 2004) and Lanthanum (Lan, a general calcium channel inhibitor; Marin et al. 2010) treatments.

Acidic pH strongly reduced embryonic survival in all populations also in the dejellied treatment but, as indicated by a significant pH × population interaction (Table 1), this effect was much weaker in the acid origin T population (Fig. 1). We found strong Inhibitor main effects as well as significant pH × Inhibitor and Population × Inhibitor interactions (Table 1). It was apparent that Ver treatment had strong negative effects at pH 7.5 in all populations, whereas the Lan treatment had strong negative effects at pH 4 in all populations (Fig. 1). To establish the nature of the interactions, we next investigated the potential role of the inhibitors in embryonic performance within each pH treatment separately (Tables 2 and 3).

Within pH 7.5 (Table 2), there was a highly significant Inhibitor main effect and Inhibitor × Population interaction, indicating that differences among inhibitor treatments and among population divergence in Inhibitor function was expressed at neutral pH. In contrast, at pH 4 Population main effect was significant, but there was no significant Inhibitor main effect or Inhibitor × Population interaction (Table 2).

Taken together, these results indicate that the T population had consistently higher acid tolerance, but that in terms of inhibitor effects, the populations responded similarly to acidic pH (Fig. 1). Moreover, it is clear that inhibitor effects are strongly pH dependent. To gain insight to the nature of the interactive effects, we next compared the impacts of the different inhibitor effects by Population and pH treatment (Table 3; Fig. 1)

140

Figure 1 Effects of two pH (4 and 7.5) and five inhibitor (Blank, Ami, Oua, Lan and Ver) treatments on embryonic survival (mean ± SE) for three R. arvalis populations: the neutral origin (S, white) population, the intermediate pH origin (B, grey) population and the acid origin (T, black) population.

141

Table 1. Generalized linear model of embryonic survival in response to inhibitor treatments. Results are shown for three R. arvalis populations under two pH (pH 4 and pH 7.5) and the four inhibitor (Blank control and Ami, Oua and Ver) treatments. The treatment Lan had was excluded from the model because the model did not converge due to complete mortality at pH 7.5. Significant effects are highlighted in bold.

Fixed effects df χ2 P Population 2 0.65 0.723 pH 1 0.00 0.999 Inhibitor 3 282.27 <0.001 pH × Population 2 13.96 <0.001 pH × Inhibitor 3 193.78 <0.001 Population × Inhibitor 6 14.00 0.030 pH × Population × Inhibitor 6 10.52 0.104

Table 2 Generalized linear model of embryonic survival by pH treatment. At pH 7.5, results are shown for three R. arvalis populations under all five inhibitor treatments. At pH 4, the complete mortality of embryos in the Lan treatment resulted in problems with model convergence and Lan was, hence, excluded from the pH 4 analysis. Significant effects are highlighted in bold.

A) pH 7.5 (all inhibitors) B) pH 4 (without Lan) Fixed effects df χ2 P df χ2 P Population 2 0.34 0.842 2 14.55 <0.001 Inhibitor 4 260.00 < 0.001 3 2.77 0.428 Population × Inhibitor 8 24.16 0.002 6 0.58 0.748

142

Table 3. Dunnett’s tests of pairwise differences in least square means from a generalized linear model of survival of R. arvalis embryos between a given inhibitor treatment and the control (blank) for each population within each pH. The inhibitor effect tests exclude the Lan treatment, which showed near complete mortality in all populations.

A) pH 7.5 B) pH 4 Contrast Population z Adj P z Adj P

S -0.01 1.000 0.82 0.750

Amiloride vs. Blank B -0.26 0.995 -0.42 0.955 T -1.59 0.270 -0.77 0.764

S -1.55 0.289 -1.06 0.586

Ouabain vs. Blank B -1.53 0.314 -1.53 0.293 T -1.15 0.535 -2.426 0.060

S -6.38 <0.001 -0.37 0.966

Verapamil vs. Blank B -5.71 <0.001 -0.86 0.730

T -6.56 <0.001 -0.95 0.641

S -4.09 <0.001 n.a. (due to complete Lanthanum vs. Blank B -1.53 0.310 mortality in Lan at pH 4 in all populations) T -0.46 0.914

143

General calcium flux mediating channels

The Lan treatment had different effects under different pHs (Fig. 1, Table 2). Strikingly, at pH 4, in all populations almost all embryos in the Lan treatment failed to hatch (Fig. 1; Table 3) and most of the embryos died before day 7 (Shu, pers. obs). At pH 7, populations differed in sensitivity to the Lan treatment: embryonic survival was reduced by the Lan treatment only in the neutral origin S population, while T and B populations were not affected (Fig. 1; Table 3).

These results indicate that Ca2+ influx is crucial for embryonic acid stress tolerance and suggest among population divergence of calcium channel function in R. arvalis. However, as Lanthanum chloride is a general calcium channel inhibitor, we could not specify which type(s) of calcium channels may have been selected by acidic stress and contributed to embryonic fitness.

Acid-sensing ion channels (ASICs)

In general, we found no significant effects of the Ami treatment (compared to the blank) on survival at either of the pH treatments (Fig.1, Table 3). ASICs are generally considered to be activated under acidic pH by extracellular protons and play crucial role in acid sensing (Jasti 2007), but our data suggests that ASICs did not play an essential role during early embryonic development when facing acidic stress in R. arvalis.

Na+/K+-ATPase (Sodium pumps)

There was a slight reduction in survival in the Oua treatments (Fig. 1). Although there was no significant Population × Inhibitor interaction at pH 4 (Table 2), the pairwise comparisons indicated that the Oua treatment reduced survival (though marginally) only in the T population at pH 4 (Fig. 1, Table 3). This subtle divergence indicates possible divergence in sodium pump function in response to acidic stress, although its contribution to embryonic fitness seems relatively weak at this point.

Intriguingly, however, all larvae (in all populations and both pH treatments) in the Oua treatment died at Gosner stage 22/23 (the stage prior to gill absorption; Gosner 1960; Shu pers. obs., data not shown). This indicates that Na+ balance becomes important at early larval stages in amphibians, in accordance with previous studies in fishes (Hwang et al. 2011; Kwong et al. 2014).

L-type voltage-dependent calcium channels

In contrast to the Lan treatment, the Ver treatment dramatically reduced hatching at pH 7.5, but had no significant effect at pH 4, and this effect was seen in all three populations (Fig. 1, Table 3). This indicated that L-type voltage-dependent calcium channels are not involved in embryonic acid stress tolerance, but have a crucial function during embryonic development at neutral pH. In addition, the fact that our artificial blockage by the Ver treatment did not have any significant effects on embryonic fitness under pH 4 suggests that this channel may be

144

naturally blocked when challenged by acidic stress. Previous studies on X. laevis suggested that L-type calcium channels play a role in neural induction of amphibian embryos (Drean et al. 1995). Therefore, it would be interesting to further investigate that if L-type calcium channels indeed can be blocked by acidic stress, how could the embryos still successfully develop without the aid of this channel.

Discussion

We found that Na+ and, in particular, Ca+ ion channels and pumps play diverse roles during early embryogenesis and embryonic responses to environmental stress. Intriguingly, our results further show among population differences in Ca+ ion channel function (in particular, effect of Lan treatment on S population at pH 7.5), indicating that environmental acidification may drive adaptive divergence of ion channel function at early life-stages. We show, for the first time, that embryonic stress acid tolerance in amphibians is dependent on Ca2+ ion flux and that different calcium channels are activated under different pH conditions. We next discuss the results and their general implications.

Ca2+ influx is essential for embryonic survival under acidic pH

Maintaining an optimal intracellular pH is crucial for homeostasis of organisms, but this can be affected by various environmental conditions, such as salinity and environmental acidity (Casey et al. 2009). Although it is not clear to date how environmental stress affects functioning of ion channels and pumps during embryogenesis, current evidence from various systems (Freda & Dunson 1984; Taylor et al. 2011; Stumpp et al. 2012; Kwong et al. 2014) suggests that the negative effects of acid stress are in part mediated via the disrupted ion balance of H+ and Na+.

Therefore, a reasonable assumption would be that H+ and Na+ channels are important also for embryonic acid stress tolerance in amphibians and other taxa directly exposed to environmental pH. However, we found no support for the above hypothesis. Instead, we showed that embryonic acid stress tolerance in R. arvalis was Ca2+ flux dependent: blocking of Lanthanum chloride sensitive calcium channels caused almost 100 % mortality under acidic conditions, whereas Ca2+ flux seemed to be less essential under neutral conditions (Fig. 1). This suggests that under acid stress, embryos need to uptake Ca2+ from the environment to maintain ionic balance and stay alive.

Our findings provide an interesting perspective to explain several previous findings. First, presence of Ca2+ in acidic environments should reduce the negative effects of acid stress. This is in accordance with previous experimental studies that manipulated calcium content in experimental settings, as well as the (sometimes) positive consequences of liming that have been used to counteract acidification. For example, in Salmo trutta increased external concentrations of calcium increased egg survival from 10% to 90% (Brown & Lynam 1981). Increased calcium levels have been found to increase survival at acidic pH also in several amphibians (e.g. Freda & Dunson 1984, 1985). Likewise, liming of acidified ponds increased

145

embryonic survival of R. arvalis and R. temporaria (Bellemakers & Van Dam, 1992; Beattie et al., 1993).

Taken together, our results show, to our knowledge for the first time, that Ca2+ influx is essential for embryonic survival under acidic pH. Given that Ca2+ has a dual role in both signaling and ion balance during embryogenesis (Ducibella et al. 2002), we suggest that the role of Ca2+ dynamics during embryogenesis should be considered more comprehensively, in particular in the context of environmental stress.

New insight to components of embryonic acid stress tolerance

Prior work in amphibians identified the “curling defect” as the main mechanism drastically reducing embryonic survival at acidic pH (Dunson & Connell 1982; Pierce 1985; Räsänen & Green 2009). Moreover, studies on both between and within species in embryonic acid stress tolerance suggested that adaptive divergence is driven primarily by egg coat mediated maternal effects (Picker et al. 1993; Räsänen et al. 2003b). The present study adds a new aspect to understanding adaptive responses to environmental acidification and suggests that the ion channels influencing embryonic performance may also be under divergent natural selection by environmental pH.

Figure 2. Putative key components of embryonic acid stress tolerance in amphibians. A) A schematic presentation of an amphibian embryo. Embryos are surrounded by a perivitelline space, and the egg coats, which can be divided into the innermost fertilization envelopes (FE) and the outer gelatinous layers (jelly envelopes); B) Effects of environmental acidity on embryos, with healthy embryos at the left, and dying embryos exposed to acidity on the right; C) A schematic presentation of the main mechanisms for pH mediate selection on embryonic acid stress tolerance in amphibians.

In particular, we found that Lanthanum chloride sensitive Ca2+ channels influenced embryonic acid stress tolerance. This indicates that maintaining a certain level of intracellular Ca2+ concentration is crucial for embryonic fitness under acidic stress. Moreover, we found a clear pattern of population divergence at neutral conditions (Fig.1, Table 1): embryos from the acid origin (T) and the intermediate pH origin (B) population were not affected by the Lan treatment at pH 7.5, but the embryos from the neutral origin (S) population had strongly

146

reduced survival in this treatment combination (Fig.1, Table 1, 3). This suggests a significant functional divergence of Ca2+ channel across populations. However, adaptive divergence in Lan treatment was not manifested at pH 4. This may be because (for logistic reasons) only one - relatively high - concentration of the ion channel blocker was used and adaptive divergence may have been evident at lower concentrations of Lanthanum chloride.

Nevertheless, taken together with the evidence for divergence via egg coat mediated adaptive maternal effects (Picker et al. 1993; Räsänen et al. 2003b), we propose that environmental acidity can simultaneously drive adaptive divergence in both egg coats and embryonic ion channels – in particular Ca2+ channels (Fig. 2). First, in order to develop normally in acidic environments, embryos have to be able to maintain intracellular ionic and acid balance, which imposes selection on Ca2+ channel function (Fig. 2). Second, as the jelly envelopes are strongly negatively affected by acidic pH trapping embryos inside the jelly resulting in failure to hatch, there is selection on the egg coats – possibly in their composition (Shu et al. II, III). Therefore, we propose that at extreme acidic conditions – such as those the T population is exposed to - there is simultaneous selection on Ca2+ channel and egg coats.

Materials and Methods

R. arvalis is a widely distributed anuran in the western Palearctic and inhabits a wide range of pH’s (Glandt 2006). Three populations breeding in permanent ponds (pond T pH 4.0, B pH 6.1 and S pH 7.5) in southwestern Sweden, and known to differ in embryonic acid tolerance (Hangartner et al. 2011, 2012), were used in this study (Details are provided in Supplementary material SI). The whole experiment was performed as a fully factorial 2 × 2 × 5 × 3 × 5 nested randomized design, with two pH treatments (pH 4.0, and 7.5), two jelly treatments (jellied and dejellied), five inhibitor treatments (Blank control, Amiloride (Ami), Ouabain (Oua), Lanthanum chloride (Lan) and Verapamil (Ver)), three populations (T, B and S) and five clutches (i.e. full-sib families) per population. The Ami, Oua, Lan and Ver treatments block the flux of H+, Na+ and Ca2+ ions, respectively. The embryos were reared in a standard media (reconstituted soft water, RSW), for which pH was adjusted in the pH 4.0 treatment with sulphuric acid H2SO4.

Embryos were reared individually and each treatment – family combination was replicated 10 times, resulting in a total of 3000 experimental units (details are provided in SI). To reduce model complexity, the data was grouped by family for statistical analyses (resulting in five replicates/population treatment combination). The response variable was survival to day 12 (hatched/total embryos). The data was analyzed with generalized linear effects models (GLM) with binomial error structure and logit link function with the GENMOD procedure in SAS 9.3 (SAS Institute, Inc.).

Three main sets of data analyses were conducted (See Supplementary material): 1) a full model including jelly treatments, but excluding the Lan treatments (Table S1), 2) a model within the dejellied treatment, excluding the Lan treatments (Table 1) and 3) models within

147

each pH treatment, whereby also Lan was included in the pH 7.5 treatment, but excluded in the pH 4.0 treatment (Table 2). These modifications were necessary due to the near complete mortality of embryos at pH 4, which affected model convergence. The modification are indicated at the given result section. The effects of inhibitors (compared to control) were tested using planned contrasts on LSmeans (Blank vs. a given inhibitor) with a Dunnett’s test for each population within a given pH treatment (Table 3).

Acknowledgements

We thank Beatrice Lindgren for invaluable help with the field and laboratory work, and Baptiste Pasteur for help in setting up the experiment. The experiments were conducted under permissions from the Västra Götaland county board and the Ethical committee for animal experiments in Uppsala County. This study was supported by Swiss National Science foundation (SNF) grant (to KR)

148

SI Materials and Methods

Study system

R. arvalis is a widely distributed anuran in the western Palearctic and inhabits a wide range of pH’s (Glandt 2006). The species breeds in early spring, after snow melt, and females produce a single clutch of eggs per year, laid directly in water. The clutch size ranges from about 500–2000 eggs in our study region (Räsänen et al. 2008). Embryos are surrounded by maternally derived egg coats, which can be divided into the fertilization envelope (FE) and the gelatinous outer layers (so called Jelly envelopes) (Fig. 3A). The species can inhabit a wide range of pHs (from 4 to 9; reviewed in Räsänen& Green 2009) and has become a model system for studies on adaptation to acidification (Räsänen et al. 2003a, b; Merilä et al. 2004; Persson et al. 2007; Räsänen et al. 2008; Hangartner et al. 2011, 2012; Egea-Serrano et al. 2014).

Three populations breeding in permanent ponds in southwestern Sweden, and known to differ in embryonic acid tolerance to varying degree, were used in this study (Figure S1, details are provided in Hangartner et al. 2011). The pH in these ponds ranged from 4.0 in the most acid tolerant population (Tottajärn, T, 57°60N, 12°60E) to 6.1 in an intermediately tolerant (Bergsjö, B, 58°20N, 13°48E) to 7.3 in a highly acid-sensitive population (Stubberud, S, 58°46N, 13°76E). Site T is situated centrally within an area that has been heavily affected by anthropogenic acidification since the early 1900s (Renberg et al. 1993), whereas site S has remained unaffected by acid rain due to limestone bedrock (Brunberg and Blomqvist 2001; Hangartner et al. 2011). Site B was heavily acidified until 1987 (lowest pH measured pH 4.2), but is since being limed regularly. It has a somewhat fluctuating pH around pH 6 (Annica Karlsson, pers. comm., Västragötaland county board).

In each population, five freshly fertilized clutches were collected in the breeding ponds within ca. 30 min of egg laying (i.e. prior to first cell division and when eggs or egg coats have not yet absorbed substantial amounts of water). Each site was continuously checked during the sampling night. Only freshly laid clutches (within 30 min of egg laying) were collected and immediately transferred to reconstituted soft water (RSW, pH 7.2-7.6; APHA 1985). The embryos were maintained cool to slow down embryonic development, and transported to the laboratory at Uppsala University within one day from collection.

Experimental procedures

Jelly removal

Jelly can serve as a protective barrier to environmental hazards (Altig& McDiarmid 2007; Menkhorst& Selwood 2008) and therefore could also potentially influence successful application of the inhibitors which consist of relatively large molecules. Therefore, half (i.e. 100 eggs) of the eggs in each clutch were de-jellied manually using watchmaker’s forceps (Hedrick and Hardy

149

1991). Each egg was inspected under a stereomicroscope to assure that the jelly removal left the FE apparently intact. Any damaged or unfertilized eggs were disregarded prior to the experiment. This resulted in a total of 1, 500 de-jellied eggs.

Inhibitor preparation

Amiloride hydrochloride (Amiloride hydrochloride hydrate 95%, 2016-88-8), Ouabain octahydrate (Ouabain octahydrate 95% 11018-89-6), Lanthanum chloride (Lanthanum(III) chloride anhydrous, beads, −10 mesh, 99.9% trace metals basis, 10099-58-8) and Verapamil hydrochloride (Verapamil hydrochloride ≥99% (titration), powder, 152-11-4), purchased from Sigma-Aldrich Co. LLC. Sweden, were used as ion channel blockers. Deionized distilled water (Milli Q water purification system, Millipore) was used for preparation of stock solutions (10 mM) and experimental solutions. For logistic reasons, only one concentration (0.1 mM) was used within each inhibitor treatment. This concentration was chosen based on those commonly used in the literature in ion channel studies (Ugawa et al. 2002; Workman et al. 2003; Amaral et al. 2004; Marin et al. 2010). Inhibitor stock solutions were prepared fresh, immediately prior to the experiment. 160 eggs from each family were used for each inhibitor treatment, resulting in a total of 3 000 eggs for the following experiment.

Rearing of embryos

The acid tolerance test of each clutch within the three populations were performed in a walk-in climate room (~ 16 °C) with 17L: 7D photoperiod and embryos were reared at two pH treatments (acid: pH 4.0, neutral: pH 7.5). Reconstituted soft water (RSW; APHA 1985) was used as the experimental medium, similar to our former work (Räsänen et al. 2003a, Hangartner et al. 2011). The pH in the neutral treatment was not adjusted (nominal pH of RSW is 7.2-7.6 when organisms are in the water), whereas in the acid treatment it was adjusted with 1M H2SO4 in 200L containers at least two days prior to use. Embryos were placed in the experimental treatments within three hours after arrival. Embryos were reared singly in PP plastic vials (0.25 L), containing 0.1 L of treatment water. No water change was conducted during the experiment. Embryos were reared from fertilization to day 12 (when all surviving embryos should have hatched).

Experimental setup

The whole experiment was performed as a factorial 2 × 3 × 5 × 5 × 2 nested randomized design, with two pH treatments (pH 4.0, and 7.5), three populations (T, B and S), five clutches (i.e. full- sib families) per population, five inhibitor treatments (Blank control, Amiloride, Ouabain, Lanthanum chloride and Verapamil) and two jelly treatments (jellied and dejellied). Each family treatment combination was replicated ten times, resulting in a total of 3, 000 experimental units. The replicates were fully randomized over the experiment shelves. Any unfertilized eggs (i.e. if no cell division was apparent eggs were assumed to be unfertilized) were determined at day 3

150

and excluded from the analyses of hatching success. However, fertilization success was very high (near 100%). Hatching was recorded visually at each day, but only final survival (day 12) was used in the statistical analyses.

Statistical analyses

To reduce model complexity, the data was grouped by family for statistical analyses. Hence, family (instead of individual vials) was the unit of analysis (resulting in five replicates/population treatment combination). The response variable was survival to day 12 (hatched/total embryos). The data was analyzed with generalized linear effects models (GLM) with binomial error structure and logit link function with the GENMOD procedure in SAS 9.3 (SAS Institute, Inc.). The full model (across the whole experiment) included jelly, pH treatment, population, inhibitors and all possible interactions as fixed factors. However, as the jelly clearly affected the efficiency of the inhibitor treatments (Fig. S1), the analyses testing the effects of inhibitors were conducted within the dejellied treatment.

The submodels within the dejellied treatment included pH treatment, population, inhibitors and all possible interactions as fixed factors. Some modifications had to be made due to near complete mortality in the Lan treatment at pH 4 for all populations, and hence complete separation of treatment responses. The models including this treatment combination failed to converge. Hence, this treatment was excluded in some of the analyses and only subsets of data were analyzed. For comparative interpretation of the data in the Lan-pH 4 treatment, we use visual interpretation (see Fig.1, Figure S1). Three main sets of data analyses were conducted: 1) a full model including jelly treatments, but excluding the Lan treatments (Table S1), 2) a model within the dejellied treatment, excluding the Lan treatments (Table 1) and 3) models within each pH treatment, whereby also Lan was included in the pH 7.5 treatment, but excluded in the pH 4 treatment (Table 2). The effects of inhibitors were rested using planned contrasts on LSmeans (control vs. a given inhibitor) with a Dunnett’s test for each population within a given pH treatment (Table 3).

151

SI Table and figures

SI Table 1. Generalized linear models of embryonic survival in three R. arvalis populations in response to four inhibitor (Blank control, Ami, Oua and Ver), two jelly (jelly intact, de-jellied) and two pH (pH 4.0 and pH 7.5) treatments. As the complete mortality of embryos in the Lan treatments at pH 7.5 resulted in lack of convergence of the model, model was ran without the Lan treatment. Significant effects (P < 0.05) are highlighted in bold.

Fixed effect df χ2 P pH treatment 1 0.00 1.000 Jelly treatment 1 44.09 <0.0001 Inhibitor treatment 3 357.11 <0.0001 Population 2 1.78 0.410 pH × Jelly 1 1.90 0.168 pH × Inhibitor 3 6.45 <0.0001 Jelly × Inhibitor 3 1.81 0.612 pH × Population 2 18.64 <0.0001 Jelly × Population 2 8.03 0.018 Inhibitor × Population 6 6.45 0.375 pH × Jelly × Inhibitor 3 0.29 0.962 pH × Jelly × Population 2 2.10 0.350 pH × Inhibitor × Population 6 10.01 0.124 Jelly × Inhibitor × Population 6 14.44 0.025 pH × Jelly × Inhibitor × Population 6 9.10 0.168

152

SI Figure 1. Survival (mean ± SE) of embryos at two jelly (jelly intact, de-jellied), five inhibitor (Blank control, Ami, Lan, Oua and Ver) and two pH (pH 4.0 and pH 7.5) treatments in three R. arvalis populations (S, B and T).

153

References

Altig R, McDiarmid RW (2007) Morphological diversity and evolution of egg and clutch structure in amphibians. Herpetological Monographs 21, 1-32. Amaral L, Martins M, Viveiros M, et al. (2004) Verapamil and reserpin promote the killing of intracellular bacteria. Anticancer Research 24, 3422-3422. APHA (1985) Standard methods for the examination of water and wastewater. American Public Health Association, Washington, DC. Bolker BM, Brooks ME, Clark CJ, et al. (2009) Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution 24, 127-135. Brown DJA, Lynam S (1981) The effect of sodium and calcium concentrations on the hatching of eggs and the survival of the yolk-sac fry of brown trout, Salmo trutta L at low pH. Journal of Fish Biology 19, 205-211. Casey JR, Grinstein S, Orlowski J (2009) Sensors and regulators of intracellular pH. Nature Reviews Molecular Cell Biology 11, 50-61. Dale JM, Freedman B, Kerekes J (1985) Acidity and associated water chemistry of amphibian habitats in Nova-Scotia. Canadian Journal of Zoology-Revue Canadienne De Zoologie 63, 97-105. Drean G, Leclerc C, Duprat AM, Moreau M (1995) Expression of L-type Ca2+ channel during early embryogenesis in Xenopus laevis. International Journal of Developmental Biology 39, 1027-1032. Ducibella T, Huneau D, Angelichio E, et al. (2002) Egg-to-embryo transition is driven by differential responses to Ca2+ oscillation number. Developmental Biology 250, 280-291. Dunson WA, Connell J (1982) Specific inhibition of hatching in amphibian embryos by low pH. Journal of Herpetology 16, 314-316. Egea-Serrano A, Hangartner S, Laurila A, Räsänen K (2014) Multifarious selection through environmental change: acidity and predator-mediated adaptive divergence in the moor frog (Rana arvalis). Proceedings of the Royal Society B-Biological Sciences 281, 20133266. Freda J, Dunson WA (1984) Sodium balance of amphibian larvae exposed to low environmental pH. Physiological Zoology 57, 435-443. Freda J, Dunson WA (1985) The influence of external cation concentration on the hatching of amphibian embryos in water of low pH. Canadian Journal of Zoology-Revue Canadienne De Zoologie 63, 2649-2656. Glandt D (2006) Der Moorfrosch. Einheit und Vielfalt einer Braunfroschart Bielefeld (Laurenti Verlag). Gonzalez RJ (2012) The physiology of hyper-salinity tolerance in teleost fish: a review. Journal of Comparative Physiology B-Biochemical Systemic and Environmental Physiology 182, 321-329.

154

Gosner KL (1960) A simplified table for staging anuran embryos and larvae with notes on identification. Copeia 1960, 183–190. Hangartner S, Laurila A, Räsänen K (2011) Adaptive divergence of the moor frog (Rana arvalis) along an acidification gradient. BMC Evolutionary Biology 11, 366. Hangartner S, Laurila A, Räsänen K (2012) The quantitative genetic basis of adaptive divergence in the moor frog (Rana arvalis) and its implications for gene flow. Journal of Evolutionary Biology 25, 1587-99. Hendry CD, Brezonik PL (1984) Chemical-composition of softwater Florida lakes and their sensitivity to acid precipitation. Water Resources Bulletin 20, 75-86. Hoffmann AA, Parsons PA (1997) Extreme environmental change and evolution Cambridge University Press. Hur CG, Kim EJ, Cho SK, et al. (2012) K+ efflux through two-pore domain K+ channels is required for mouse embryonic development. Reproduction 143, 625-636. Hwang PP, Lee TH, Lin LY (2011) Ion regulation in fish gills: recent progress in the cellular and molecular mechanisms. American Journal of Physiology-Regulatory Integrative and Comparative Physiology 301, R28-R47. Kroeker KJ, Kordas RL, Crim RN, Singh GG (2010) Meta-analysis reveals negative yet variable effects of ocean acidification on marine organisms. Ecology Letters 13, 1419-1434. Kwong RW, Kumai Y, Perry SF (2014) The physiology of fish at low pH: the zebrafish as a model system. Journal of Experimental Biology 217, 651-662. Leclerc C, Neant I, Webb SE, Miller AL, Moreau M (2006) Calcium transients and calcium signalling during early neurogenesis in the amphibian embryo Xenopus laevis. Biochimica Et Biophysica Acta-Molecular Cell Research 1763, 1184-1191. Lee CE, Kiergaard M, Gelembiuk GW, Eads BD, Posavi M (2011) Pumping ions: rapid parallel evolution of ionic regulation following habitat invasions. Evolution 65, 2229-2244. Marin M, Sellier C, Paul-Antoine AF, et al. (2010) Calcium dynamics during physiological acidification in Xenopus oocyte. Journal of Membrane Biology 236, 233-245. McCairns RJS, Bernatchez L (2010) Adaptive divergence between freshwater and marine sticklebacks: insights into the role of phenotypic plasticity from an integrated analysis of candidate gene expression. Evolution 64, 1029-1047. Menkhorst E, Selwood L (2008) Vertebrate extracellular preovulatory and postovulatory egg coats. Biology of Reproduction 79, 790-797. Merilä J, Söderman F, O'Hara R, Räsänen K, Laurila A (2004) Local adaptation and genetics of acid-stress tolerance in the moor frog, Rana arvalis. Conservation Genetics 5, 513-527. Noskov S (2011) Molecular mechanisms of ion selectivity in membrane proteins: ion channels and transporters. Biophysical Journal 100, 6-6. Parker LM, Ross PM, O'Connor WA (2009) The effect of ocean acidification and temperature on the fertilization and embryonic development of the Sydney rock oyster Saccostrea glomerata (Gould 1850). Global Change Biology 15, 2123-2136.

155

Persson M, Räsänen K, Laurila A, Merilä J (2007) Maternally determined adaptation to acidity in Rana arvalis: Are laboratory and field estimates of embryonic stress tolerance congruent? Canadian Journal of Zoology-Revue Canadienne De Zoologie 85, 832-838. Picker MD, Mckenzie CJ, Fielding P (1993) Embryonic tolerance of Xenopus (Anura) to acidic blackwater. Copeia, 1072-1081. Pierce BA (1985) Acid tolerance in amphibians. Bioscience 35, 239-243. Räsänen K, Green E (2009) Acidification and its effects on amphibian populations. Amphibian Biology. Conservation and Ecology, Volume 8. Edited by Heatwole H. Surrey Beatty and Sons, Chipping Norton, Australia. Räsänen K, Laurila A, Merilä J (2003a) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. I. Local adaptation. Evolution 57, 352-362. Räsänen K, Laurila A, Merilä J (2003b) Geographic variation in acid stress tolerance of the moor frog, Rana arvalis. II. Adaptive maternal effects. Evolution 57, 363-371. Räsänen K, Soderman F, Laurila A, Merilä J (2008) Geographic variation in maternal investment: Acidity affects egg size and fecundity in Rana arvalis. Ecology 89, 2553- 2562. Roux B, Berneche S, Egwolf B, et al. (2011) Ion selectivity in channels and transporters. Journal of General Physiology 137, 415-426. Rutenberg J, Cheng SM, Levin M (2002) Early embryonic expression of ion channels and pumps in chick and Xenopus development. Developmental Dynamics 225, 469-484. Stumpp M, Hu MY, Melzner F, et al. (2012) Acidified seawater impacts sea urchin larvae pH regulatory systems relevant for calcification. Proceedings of the National Academy of Sciences of the United States of America 109, 18192-18197. Taylor AR, Chrachri A, Wheeler G, Goddard H, Brownlee C (2011) A voltage-gated H+ channel underlying pH homeostasis in calcifying coccolithophores. Plos Biology 9. Tosti E (2010) Dynamic roles of ion currents in early development. Molecular Reproduction and Development 77, 856-867. Ugawa S, Ueda T, Ishida Y, et al. (2002) Amiloride-blockable acid-sensing ion channels are leading acid sensors expressed in human nociceptors. Journal of Clinical Investigation 110, 1185-1190. Workman AJ, Kane KA, Rankin AC (2003) Characterisation of the Na, K pump current in atrial cells from patients with and without chronic atrial fibrillation. Cardiovascular Research 59, 593-602.

156

Acknowledgements

Doing this PhD is a great adventure for me: first time in a foreign country; first time studying evolution; first time in a field trip; first time trying a cheese fondue… There are hard times, but also a lot of fun.

Firstly, I would like to thank Katja for giving me the opportunity to get involved in this project, and in evolutionary biology. Thank you for always helping me in all aspects, from mentoring to cooking in the field! And I do learn a lot! Also thank you for always being open to new approaches and allowing me to develop my own ideas.

Thank Marc and Anssi for all the important inputs and help on my thesis. Thanks to Jukka for being my Doktorvater and your help when needed. Thank my committee member Thomas for your inputs and comments. And thanks to Barbara for being my external examiner.

I want to thank all the persons that helped me and made me feel welcome in Sweden: Bea, Baptiste, Corrine, Tamara, Gunilla, Emma. Thank all Eco members for making it such a great place to work! And special thanks to Anja for your Zusammenfassung! And thanks to the Suter group for all of your help.

I also want to thank all of my basketball crews for allowing me to keep my hobby in a place where almost nobody is playing... Thanks to the poker game groups for numerous funny weekends. And thanks to all of the persons who invite me for dinner whenever I am missing Chinese food… specially to Xiaomei, Yigang and Yang.

Finally, I want to thank my family. You are always being there and supportive to whatever I like to do. I can’t be here without you. And I want to thank my fiancee Yijing, for making this whole adventure special and memorable.

157

CURRICULUM VITAE Longfei Shu Inst. of Integrative Biology, ETH Zürich & Dept. of Aquatic Ecology, Eawag Zürich, Switzerland [email protected]

EDUCATION ETH Zürich, Switzerland 09/2011 – 10/2014 Ph.D. Evolutionary Ecology (defended 24.10.2014, to be conferred January 2015) Shenzhen University, China 09/2008 – 07/2011 M.S. Biochemistry and Molecular Biology 09/2004 – 07/2008 B.S. Biotechnology

RESEARCH EXPERIENCES 09/2011 – 10/2014, PhD research, supervised by Dr. Katja Räsänen, Dr. Marc J-F Suter and Prof. Jukka Jokela. I investigated the molecular basis of embryonic acid adaptation in amphibian populations in Sweden, using common garden experiments and different molecular approaches, including pharmaceutical ion channel manipulations, proteomics, glycomics and transcriptomics. 09/2008 – 07/2011, Master research, supervised by Prof. Zhangli Hu. I studied the impact and application of sulfur deprivation on microbial algae using Proteomics and microRNA profiling analysis. 06/2007 – 07/2008, Undergraduate research, supervised by Prof. Zhigang Liu. I performed a transgenic approach to express of cecropin CBM1 from Chinese Silkworm in Escherichia coli.

FELLOWSHIP AWARDS Best Graduate of Shenzhen University - RMB 5, 000 (2011)

158

Dean's fellowship awards - RMB 3, 000 (2011) Academic Leadership Award - RMB 3, 000 (2010)

GRANTS As PI 2013 - 2014: Popularizing Evolution in China, European Society for Evolutionary Biology (ESEB). As member 2010 - 2013: Study of Chlamydomonas reinhardtii microRNAs involved in sulfur-deprived and

H2-producing, National Natural Science Foundation of China (NSFC). 2012 - 2014: Artificial microRNAs based bio-hydrogen production by Chlamydomonas reinhardtii, National Natural Science Foundation of China (NSFC).

PUBLICATIONS Peer Reviewed Publications

Published

1. Longfei Shu and Zhangli Hu. 2012. Characterization and differential expression of microRNAs elicited by sulfur deprivation in Chlamydomonas reinhardtii. BMC Genomics 13:108 2. Zhangli Hu, Longfei Shu and Deming Gou. 2011. MicroRNAs quantification and related target genes for response to sulfur deprivation in Chlamydomonas reinhardtii. Journal of Shenzhen University (Science & Engineering) 3: 237 - 242 3. Longfei Shu, Zhangli Hu. 2010. Small silencing RNAs in Chlamydomonas reinhardtii. Minerva Biotecnologica, 22: 29-37

159

Submitted

4. Longfei Shu, Anssi Laurila, Marc J-F Suter and Katja Räsänen. Mechanistic basis of adaptive maternal effects: egg jelly water balance mediates embryonic adaptation to acidity in Rana arvalis. Oecologia (in review) 5. Shu, Longfei, Marc J-F Suter and Räsänen, Katja. Evolution of egg capsules: linking molecular biology and ecology. Molecular Ecology (in revision)

Manuscripts in preparation

6. Shu, L., Suter, MJ-F., Laurila, A. and Räsänen, K. Molecular phenotyping of maternal egg jelly reveals parallel adaptive divergence to acidity in Rana arvalis and Rana temporaria. Manuscript 7. Shu, L.. and Räsänen, K. First transcriptome analysis of the moor frog Rana arvalis: genomics resources and adaptive maternal effects genes. Manuscript 8. Shu, L., Laurila, A., and Räsänen, K. Environmental stress causes adaptive divergence in ion channel function during embryogenesis. Manuscript

Non Peer Reviewed Publications

Longfei Shu. NextGen VOICES. Science, 2012, 338 (6103): 40-43

REVIEWER FOR Aquatic Ecology

CONFERENCES AND SEMINARS 2014 Poster at SGE 2014, Switzerland “The molecular basis of adaptive maternal effects in amphibians” 2014 Spotlight poster at SETAC Europe 24th Annual Meeting “Ecological and evolutionary impact of environmental acidification on amphibians” 2014 Talk at 4th ECO PhD Symposium, Switzerland “Single environmental stress drives simultaneously adaptive evolution at multiple traits”

160

2013 Talk at 31th SGMS (Swiss Group for Mass Spectrometry), Switzerland “Linking proteomics with evolution: necessity and challenge” 2013 Talk at ZIS (Zurich interaction seminar), Switzerland “What can you do when born in a harsh environment?” 2013 Poster at XIV Congress of the European Society for Evolutionary Biology, Portugal “Defending the embryo from environmental stress: a frog mother’s strategy” 2012 Talk at 18th European Meeting of PhD Students in Evolutionary Biology, Finland “The molecular basis of adaptive maternal effects in amphibians”

MEMBER OF SOCIETIES European Society for Evolutionary Biology (ESEB) Swiss Group for Mass Spectrometry (SGMS)

161