The Molecular Basis of Development of the Sword, Asexual Selected Trait

Home , Green swordtail, Xiphophorus, Xiphophorus birchmanni, Xiphophorus nigrensis

The molecular basis of development of the sword, a sexual selected trait in the genus Xiphophorus

Dissertation zur Erlangung des akademischen Grades

des Doktors der Naturwissenschaften (Dr. rer. nat.)

an der

Universität Konstanz, Mathematisch-naturwissenschaftliche Sektion

Fachbereich Biologie

vorgelegt von

Dipl. Biol. Nils Offen

Tag der mündlichen Prüfung: 19.12.2008

Referenten: Prof. Dr. Axel Meyer Prof. Dr. Michael K. Richardson

Acknowledgements

First, I would like to thank my Ph. D. advisor Prof. Axel Meyer who gave me the opportunity to work on this fascinating organism in his lab. I’m grateful to my supervisor PD. Dr. Gerrit Begemann for his advice, various discussion on the the projects and other stuff, inspiration and for the chance to develop and try out own ideas and to improve my scientific and personal skills. I’m thankfull for a quite fascinating time as his PhD student.

I want to thank my current and past collegues of the Meyerlab for various help in the lab and interesting discussions. I’m specially gratefull to (in no particular order) Silke Pittlik, Sebastién Wielgoss, Kai Stölting, Simone Högg, Dominique Leo, Dave Gerrad, Dirk Steincke, Elke Hespeler, Katharina Mebus, Nicola Blum, Matthias Sanetra, Cornelius Eibner, Ylenia Chiari, Maria Buske and Milena Quentin for joining me for a while on my road to Ph. D. and for a memorable time. I also want to thank Helen Gunter and Kathryn Elmer for proof-reading parts of the thesis. And last but not least I’m grateful to Janine Sieling for excellent animal care that made my work so much easier.

Some of those people become more than just colleagues to me and I’m very thankful for several activities and happy hours in and outside the lab. In this context I’d like to mention the Badminton group, which I joined for more than two years.

Especially, I want to thank my parents, my grandparents, my brother and sister and friends outside the university, in particular Martin Eggert, who gave me great support and fresh motivation to continue this work to the end.

I’m also grateful to my better half, Alexandra Schuh, who brightened up my life and supported and motivated me continuously in the last ~two years of my thesis. Last, I like to thank her mother and her stepfather for a place to relax and for delicious food.

Table of contents

General introduction 4 Swords, sexy males and choosy females 5 Sword evolution and the Sensory Exploitation-Preexisting Bias Hypothesis 9 Sword development 11 The male gonopodium 14

Chapter I: Fgfr1 signalling in the development of a sexually selected trait 19 in vertebrates, the sword of swordtail fish 1.1 Abstract 19 1.2 Background 20 1.3 Results 23 1.4 Discussion 35 1.5 Conclusions 41 1.6 Material and Methods 42

Chapter II: A Subtractive Hybridisation approach to identify novel genes 47 involved in the development of the sword in green swordtail (X. helleri) 2.1 Abstract 47 2.2 Introduction 47 2.3 Results 49 2.4 Discussion 61 2.5 Experimental Procedures 67

Chapter III: Retinoic acid is involved in gonopodium formation in the green 72 swordtail, Xiphophorus helleri 3.1 Abstract 72 3.2 Introduction 72 3.3 Material and Methods 76 3.4 Results 82 3.5 Discussion 94

Summary 100 Zusammenfassung 103 Eigenabgrenzung 106 Literature cited 107 Appendix 120

General Introduction Xiphophorus fishes (Fam. Poeciliidae) are small live bearing toothcarps (Cyprinodontiformes) that are endemic to parts of Mexico and Central America [1-3]. The genus itself can be subdivided into four different groups, the northern and southern swordtails and the northern and southern platyfish, depending on their distribution relative to the Trans-Mexican Volcanic Belt in central Veracruz [3]. This classification into northern and southern swordtails, but not that of southern and northern platyfish, is further supported by several molecular phylogenies [4, 5]. However, the platyfish were originally combined in an independent genus ( Platypoecilus ), before Myron Gordon classified the platyfish as a subgroup within the genus Xiphophorus in 1951 [6]. To this day 26 Xiphophorus species have been described [3]. Some species such as X. andersi show a very restricted distribution pattern, whereas others such as the swordtail X. helleri and the two platyfish X. maculatus and X. variatus are the most widely distributed species (Figure 1). The relationship among these 26 species is rather incompletely resolved, since solid phylogenies using both mitochondrial and nuclear markers comprise only 22 of the 26 described species [4, 5]. Interestingly, several Xiphophorus species can be interbred within captivity [7] and hybridization occurs also under natural conditions [5, 8]. Multiple natural hybrid zones were found between X. birchmanni and X. malinche, due to a disruption in chemical communication [8, 9] . In addition, a recent study showed that the swordtail species X. clemenciae resulted from an ancient hybridization event between a swordtail and a platy species [5]. Different populations of Xiphophorus species such as X. helleri or X. maculatus can be amazingly variable in color pattern [10]. Several of these pigment pattern variants showed independent mendelian segregation [11, 12]. The ability to make interspecies crosses and the availability of genetic markers turned a popular aquarium fish into a model for early genetic studies in fish in the first half of the 20th century [11]. Since these days Xiphophorus fishes have become an established model organism for early genetic, as well as for behavioural studies and melanoma formation (reviewed in [13, 14]).

Figure 1: Distribution of Xiphophorus species in Mexico and Central America A: Distribution of: (A) X. meyeri , (B) X. gordoni , (C) X. couchianus , (D) X. xiphidium , (E) X. variatus , (F) X. birchmanni , X. continens, X. cortezi , X. malinche , X. montezumae , X. multilineatus , X. nezahualcoyotl , X. nigrensis and X. pygmaeus , (G) X. evelynae , (H) X. andersi , (I) X. milleri and X. kallmani , (J) three species of X. clemenciae , (K) X. helleri and X. maculatus , (K) X. alvarezi , (M) X. signum , (N) X. mayae , (O) “PMH” type of Xiphophorus [15] The satellite map was obtained from google maps. The distribution of Xiphophorus species was adapted from Kallman and Kazianis [3].

Swords, sexy males and choosy females In the second half of the 20 th century Xiphophorus become a valuable organism for behavioural biologists (reviewed in [14]). Male swordtails perform a complex courtship behaviour that consists of several behavioural sequences such as lateral or sexual display [16, 17]. Interestingly, in several Xiphophorus species not all males perform a complex courtship behaviour, but show an alternative mating strategy (reviewed in [14]). In X. nigrensis , only large males court in front of the female and perform sexual display,

whereas small males perform a sneak and chase behaviour instead [18, 19]. The difference in body length of mature males that determine the mating strategy of the respective individual seems to result from an allelic variation in the pituitary locus ( P). X. nigrensis males with the PL (large) allele show a higher growth rate and mature later than males with the Ps (small) allele [19]. The term P locus was originally termed by Kallman and Schreibman, who proposed that the P locus controls when the pituitary-gonadal axis is activated and sexual maturation is induced [20, 21]. They hypothesised that the P locus acts on the maturation of gonadotropic hormone producing cells. The gonadotropic hormones that are released by the pituitary gland stimulate Leydig-like cells in the testis to secrete testosterone [22, 23]. Both gonadotropic and sex hormones promote spermatogenesis and sexual maturation (reviewed [24]). Interestingly, the PS allele is maintained in populations of X. nigrensis , even though females prefer large, courting males [25]. This might be due to the fact that both PL and Ps males can have equal fitness

[26]. Firstly, Ps males mature and start reproducing earlier that PL males [19]. Secondly, the large PL males need a longer time to reach sexual maturity and are more attractive to predators and therefore likely have a higher mortality rate [19, 27]. However, the genetic nature of the P locus, even though the term was coined by Kallman and Schreibman in 1973 [20] is still unknown, as well as the mechanism how the locus actually times sexual maturation. Besides the attempts to uncover the genetics of behaviour, Xiphophorus is more widely known as a model for sexual selection (reviewed in [14]). Many studies contribute to the understanding of female mate choice and the traits that are involved. The contribution of several traits such as chemical cues [28], colouration [29, 30], vertical bar pattern [31], body size [32] or courtship behaviour itself [33] have been studied by several groups. A first attempt to understand the genetic basis of mate choice was made by M. Cummings and co-workers in a recent study [34]. They identified a couple of genes that are differentially expressed in the brains of X. nigrensis females, when they interact with attractive males. Another sexually selected trait, a colourful extension of the caudal fin called the sword, has been first described by Darwin himself [35]. According to a definition given by Basolo in 1991, the sword is a coloured ventral extension with 0.7-6.0 times the length of

the caudal fin that also exhibits a black ventral margin [36]. However, Meyer and co- workers provided an alternative definition that did not include colouration and a black margin [4, 37]. As a consequence, the colourless ventral extension of X. andersi is considered a sword by this definition [4, 37], whereas after Basolo’s definition it is considered to be a protrusion [38]. In fact, Basolo’s sword definition is supported by several studies that showed the biological relevance of both length and a distinct colour pattern [39-42]. The male sword (Figure 2A) is a morphologically simple trait that is mainly formed by four bony fin rays [43, 44]. The unbranched ventral ray 9 (V9) and the branched ventral ray 8 (V8) are the two major sword rays (Figure 2B and [43]). The melanised ventral rays 10 (V10) and 7 (V7) also contribute to the sword, but grow only to half the length of the sword (Figure 3B and [43]). Besides the length, width and coloration, those rays are not further modified [43, 44]. The fin rays or lepidotrichia are also quite simple structures. A lepidotrichia consists of several segments each made up of two concave, bony units, the hemisegments (Figure 2C and [45, 46]). The different segments are linked to each other by intersegmental joints. Each hemisegment is formed by a calcified matrix, that contains collagen fibrils and glycosamino-glycans such as chondroitin sulfate and is secreted by a monolayer of scleroblasts (Figure 2C and [45- 47]). The two hemisegments enclose a core of soft tissue, such as neurons, blood vessels and fibroblasts (Figure 2C and [45-47]). Behavioural experiments in X. helleri have shown that (1) the total length of the sword is an important criterion during mate choice which is thought to reflect a bias for large body size [33, 39] and (2) the females have a preference for a specific pattern of differently coloured stripes [40, 41]. In the green swordtail this colour pattern consists of a greenish to yellowish pigmentation that is enclosed by a ventral and a dorsal black stripe [40]. In contrast, the sword could also be a negative signal in terms of sexual selection such as in the swordtail species X. birchmanni [48]. X. birchmanni males are swordless according to Basolo’s definition and females evolved a preference against sworded males. The disdain for swords might reflect a general preference against long fins, since females also discriminate males with long dorsal fins against individuals with short ones [49]. Interestingly, the sword could also influence the maturation time of juvenile males and females in a population of green swordtails [50]. If males with either

long or short swords are presented to a group of juvenile fish, maturation of male individuals starts later in response to a long sworded male. Since body size and sword length are positively correlated [51], a long sword might signal to the juveniles that large males are around that will outcompete them in terms of male-male competition and female mate choice, if they mature too early at a small size [32, 51]. This study by Walling and colleagues [50] proposes a more plastic control of sexual maturation than the P locus model of Kallman and Schreibman [20, 21]. In juvenile females the long sword stimulus has the opposite effect and induces earlier maturation [50].

Figure 2: The sword of X. helleri and caudal fin morphology. (A) Male green swordtails ( X. helleri ) develop a brightly ornamented sword, whereas females are swordless. (B) The sword is formed by four caudal fin rays, the ventral rays (V) 7, 8, 9, 10, that are covered either with melanophores or xanthophores. (C) Reconstruction of a teleost caudal fin. The fin rays or lepidotrichia consist of two bony hemirays that are connected by several ligaments. Both hemirays enclose a core of soft tissue containing blood vessels, fibroblasts and nerve bundles. Figures A and B taken from Eibner et al. [52] and Figure C from Becerra et al. [45].

Sword evolution and the Sensory Exploitation-Preexisting Bias Hypothesis The sword not only plays a prominent role in sexual selection, but also has an interesting evolutionary history. One scenario, based on a molecular phylogeny, suggests a sworded common ancestor of all extant Xiphophorus species [4, 37]. However, the ancestral presence of colouration and a black margin could neither be confirmed, nor rejected with confidence [37]. Therefore it remains unclear if the common ancestor had a Basolo-type sword [36] or only a protrusion. Only males of extant swordtail species possess a Basolo-type sword [36], whereas platyfish secondarily lost this trait ([4, 5, 37] and Figure 3). In addition, males of three northern swordtail species, X. pygmaeus , X. continens and X. birchmanni , do not develop a Basolo-type sword and also likely lost it secondarily ([53] and Figure 3). How the sword evolved in the first place and how it got secondary lost in some species is still under debate. One theory that is often employed to explain the evolution of the sword is the Sensory Exploitation-Preexisting Bias hypothesis [38, 42, 54, 55]. Under this theory the (female) preference and the (male) trait evolve out of concert (reviewed in [56]). Due to selection or properties of the neural and cognitive system, the receiver (e.g. the female) shows some bias to a specific signal. If a trait evolves in the male that will accidentally stimulate this pre-existing mechanism in the female, this male will have a reproductive advantage and the trait will spread in the population. Due to the receiver bias theory the female preference for swords was already present before the male sword evolved (Figure 3). This scenario is supported by the fact that the sword is only present in the Xiphophorus lineage, but females of Priapella olmecae and Poecillia latipinna show a preference for sworded males ([38, 54, 55] and Figure 3). This would imply that the preference for swords might already have been present in the last common ancestor of Xiphophorus , Priapella and Poecilia , before the sword evolved (Figure 3). Besides the sword, this receiver bias theory is also used to explain the evolution of the hair tufts on the forelegs of male wolf spiders [57] or call suffixes in Physalaemus frogs [58].

Figure 3: Consensus taxonomy of platyfish and swordtails The consensus taxonomy from Zauner et al. [59] was modified using phylogenetic data from Meyer et al. and Hrbek et al. [5, 60]. For all species mentioned, it is indicated if males exhibit a sword, a protrusion or are swordless based on the definition given by Basolo [36]. The traits sword (S) and female preference for swords (P) are mapped on the cladogram. Swordless species where females show a preference for sworded males [38, 42, 55] have been indicated with a green dot. In addition, the putative origin of the poeciliinae gonopodium is indicated. The gonopodium is not found in other subfamilies within the Poeciliidae such as Aplocheilichthys [61]. The classification of Xiphophorus species in northern and southern swordtails/platyfish was adapted from Kallman and Kazianis [3].

How the sword was lost secondarily also remains elusive. A long sword increases the metabolic cost of swimming and long, ornamented swords make a male individual more conspicuous to predators [27, 62]. It could therefore be that high predation pressure drove the loss of the sword, if its cost was actually higher than its benefit in terms of sexual selection. In fact, several studies in guppies showed that predators can indeed drive the evolution of sexually selected traits, such as coloured spots [63-65]. Interestingly, predators can also modulate the female preference for swords. If green swordtail females are exposed to a video that shows a predation event between a long-sworded male and a cichlid, the females alter their preference for males with long towards males without swords, after the video stimulus [66]. Therefore, changes in female preference could have supported a predator-driven loss of the sword. However, changes in female preference alone might not explain the loss of the sword in some species, since X. maculatus and X. variatus females retained a preference for sworded males, even though males of these two species do not develop a sword [38, 42].

Sword development The fascinating evolutionary history of the sword makes it a valuable system to dissect the molecular events that (1) give raise to the sword in the first place and (2) precede the loss of the sword. To answer these questions, a profound knowledge of the genetic network that promotes sword development is required. Unfortunately, the molecular mechanism of sword development is poorly understood. Zander and Dzwillo performed interspecies crosses between X. helleri and X. cortezi and could show that sword development is controlled by multiple loci [44]. They coined the term “sword genes” (“Schwertgene”), for those genes that confer an ability to produce a sword. Interestingly, sword development can be artificially induced by exogenous testosterone, which implies that androgen signalling might activate different signalling pathways that act together during sword development ([43, 67] and Figure 4). These experiments also showed that the formation of a small ventral protrusion can be induced in some platy species by exogenous testosterone ([44] and Figure 4). Protrusions also occur naturally in males of the platy species X. andersi and X. xiphidium [38, 44]. This implies that the genetic network underlying sword development is not completely lost in those species.

Figure 4: Schematic representations of the caudal fin of several Xiphophorus species The first line shows the silhouettes of the caudal fin of mature males, whereas the second line shows the silhouettes of the caudal fins of methyltestosterone treated individuals of different Xiphophorus species . The platyfish species X. maculatus and X. milleri , swordless under natural conditions, develop a small ventral protrusion under long-time methyltestosterone treatment [43, 44]. Note that in some individuals of X. helleri prolonged methyltestosterone can induce the outgrowth of the dorsal-most fin rays, which results in a so-called “Dorsalschwert” (dorsal sword; [43]). Figure taken from Zander and Dzwillo [44], modified.

In a first attempt to identify genes that are involved in sword development, Zauner and colleagues showed that the homeobox transcriptional repressor msxC (muscle segment homeobox gene C ) is up-regulated in growing sword rays [59]. In a recent study, Eibner and colleagues showed that the major sword rays V9 and V8 acts as local organizer during sword development [52]. Both rays showed the capacity to induce the formation of a second, coloured sword when transplanted dorsally. However, the molecular mechanisms by which V8 and V9 promote sword development remain elusive. To indentify genes that are involved in sword development one can apply a so-called candidate gene approach. The strategy behind this approach is to select genes that are functioning in a related process and to test, if these candidates are also required for sword development. Genes that are known to function either in fin ray growth or colour pattern formation are therefore putative candidate genes. Several studies focussing on fin regeneration in zebrafish revealed multiple genetic pathways, such as Hh, Wnt or Fgf signalling, to control regenerative outgrowth (reviewed in [68, 69]). In Chapter 1 we therefore focussed on Fgf signalling (Fibroblast growth factor) and studied the expression of the fgf receptor 1 and two putative ligands, fgf24 and fgf20a , during sword

development. fgfr1 have been shown to be essential for fin ray growth during fin regeneration [70, 71]. Furthermore, Fgfr1 is an upstream regulator of msxC expression that is up-regulated during sword development [59, 70, 71]. fgf24 and fgf20a are also associated with fin development and/or regeneration and are therefore putative ligands of fgfr1 [70, 72, 73]. fgf24 was shown to be expressed in fin regeneration and is essential for pectoral fin development [70, 72]. fgf20a is involved in early steps of fin regeneration and might therefore also function during sword growth initiation [73]. To test if Fgf signalling is also involved in sword development we analysed the expression pattern of fgfr1, fgf24 and fgf20a in testosterone induced swords. Induction of sword development allows one to produce a sufficient amount of individuals with swords in the same stage of development that can be used for comparative analysis. To obtain a sufficient number of naturally developing swords in comparable stages would be hardly possible, since the age when male swordtails mature can be quite variable. We showed that fgfr1 , but not fgf24 and fgf20a , is specifically up-regulated in developing swords. A similar pattern was also observed in the developing gonopodium, the modified male anal fin that is also induced by exogenous testosterone [74, 75]. fgfr1 is spatial-temporally co-expressed with msxC both in the sword and the gonopodium. Interestingly, in testosterone treated caudal fins of platyfish, fgfr1 and msxC are differently regulated compared to swordtail fins. Both genes are only up-regulated in the ventral caudal fin after prolonged hormone treatment. Finally, we found a strong correlation of fgfr1 and msxC expression levels and fin ray growth rate, by employing the X. maculatus brushtail mutant that exhibits excessive growth of the median caudal fin rays. Candidate gene approaches are useful to identify genes that are involved in sword development. However, this approach has one major limitation. Candidate genes are selected or not selected depending on the knowledge available on the particular gene. If no information is available on the gene or it was not shown to function in a related process it will not be selected. Therefore, most likely not all genes that function in sword development will be analysed. To perturb this problem one can apply alternative techniques where previous knowledge on gene function is not required. This could be techniques that detect differential expression of genes between different tissues or

treatment groups. Microarray technology has proven to be useful in fin regeneration research to identify new genes involved in this process [76, 77]. However, microarrays are not available for any Xiphophorus species. Another approach would be to sequence a large quantity of different clones to analyze the transcriptome of a specific tissue or process, as it was done for fin regeneration in medaka [78]. However, large scale sequencing requires adequate technical and bioinformatical resources. A less resource- intensive method to identify putative “sword genes” is suppression subtractive hybridisation (SSH; [79]). Transcripts from two different tissues or treatment groups are pooled and subtracted against each other. The result is one pool enriched for transcripts that are differentially expressed between the two tissues. In Chapter 2 we took advantage of this method to identify genes that are differently expressed in developing swords and also gonopodia compared to juvenile fins. We constructed a SSH library from induced swords and gonopodia and sequenced 407 clones. After eliminating redundant sequences 128 out of 201 sequences showed significant similarity. For further analysis we focussed on transcription factors and components of signalling pathways and identified four clones with similarity to rack1 , dusp1 , klf2 and tmsb a-like that are specifically up-regulated in developing swords and/or gonopodial rays during outgrowth. Not surprisingly, these genes are also up-regulated in regenerating caudal fins during regenerative outgrowth.

The male gonopodium Not only the sword, but also the male anal fin is modified during sexual maturation ([80, 81] and Figure 5A). The so-called gonopodium is a specialized intromittant organ that enables the male to fertilize the female [82]. Like the sword, the Xiphophorus gonopodium is mainly formed by a small subset of fin rays, the anal fin rays 3, 4 and 5, the so called 3-4-5 complex ([81, 82] and Figure 5B, C). In contrast to the sword, these fin rays are not only modified in terms of length, but also develop complicated terminal structures [81]. The gonopodium of the green swordtail, X. helleri , exhibits several distinct morphological structures [83]. Ray 3 develops a set of segments that carry spines and a terminal hook (Figure 5C). The anterior branch of ray 4 exhibits a terminal ramus, whereas the posterior branch develops spines and serrae (Figure 5C). The terminus of the

anterior branch of ray 5 is formed by a claw-like structure (Figure 5C). The function of the terminal structures remains elusive. It has been suggested that these structures are necessary for successful copulation [83]. This was based on the observation that females were not fertilized by males, where the distal tip of the gonopodium has been surgically removed. Another hypothesis suggests a “lock and key” mechanism for the gonopodium- genital opening-interaction that promotes prezygotic isolation [61]. In fact, the morphology of the gonopodium, in particular the morphology of the terminal structures, varies between species and was used to distinguish the different Xiphophorus species and for morphology-based phylogenetic analysis [84, 85]. However, such a mechanism should prevent hybridization, which occurs in natural population and under laboratory conditions [5, 8, 12].

Figure 5: The male gonopodium of X. helleri The anal fin of male poeciliids like X. helleri is modified during sexual maturation (A, B). The anal fin rays 3, 4 and 5 elongate and develop several modified segments at the distal tip (B, V). These segments exhibit spines or serraes or forming either a terminal claw or hook (C). Figure A taken from Eibner et al. [52], Figure C from Clark et al. [83], modified. (scalebar B: 500 µM)

As for the sword, less is known about the developmental pathways that shape the gonopodium. The poeciliid gonopodium is thought to have evolved only once and therefore all extant species are likely to share a similar developmental program to shape

the gonopodium ([61] and Figure 4). Based on morphological studies of naturally developing and testosterone-induced gonopodia of Gambusia affinis , Turner proposed a two step model for gonopodium development [75, 80]. During the first phase when the testis starts to mature and releases low levels of testosterone, accelerated growth of the 3- 4-5 complex is promoted. The segments that are formed during the first phase are the same size or larger than the “juvenile” segments [80]. When the testis develops further, the testosterone secretion increases and phase 2 is initiated [75, 80]. During this phase so- called differentiation areas arise in a specific temporal sequence at a specific location within the developing gonopodium. These areas add specific structures (e.g. hooks, spines) or segments of smaller size to the rays. The difference in segment length is thought to be a result of local changes in the growth and/or segmentation rate. However, these differentiation areas are a descriptive term and do not describe a kind of molecular organizer. Therefore, the molecular mechanisms that organize the formation of terminal structures are unknown. The two step model is somehow supported by testosterone induction experiments in Gambusia affinis and X. maculatus [74, 75]. Juvenile fish treated with exogenous testosterone develop a gonopodium where outgrowth and formation of terminal segments start almost simultaneously. These gonopodia are much shorter than the natural ones and lack the segments that are formed in phase 1. First molecular studies on induced gonopodia of Gambusia affinis showed that two androgen receptors are expressed during gonopodium development [86]. Inhibition of androgen signalling results in reduced gonopodium growth, which supports a function of androgen signalling in growth control of the 3-4-5 complex. This study also revealed shh and ptc1 to be essential for gonopodium outgrowth [86]. Interestingly, the expression of both genes is controlled by androgen receptors, supporting the role of androgen receptor as a molecular switch that controls gonopodium development. As in developing swords fgfr1 and msxC are up-regulated in developing gonopodia ([59] and Chapter 1), which suggests a role of Fgf signalling in gonopodial growth control. In Chapter 3 we focussed on the role of retinoic acid (RA) signalling during gonopodium development, for two reasons. RA signalling is essential for appendage development in vertebrates [87-89] and it provides positional information along the proximo-distal axis in developing and regenerating limbs [89-92]. Therefore, RA

signalling might either play a general role in gonopodium development or a specific role in establishing the proximo-distal polarity within the gonopodium. RA, a small lipophilic, diffusible molecule is synthesised by retinaldehyde dehydrogenases (Aldh1as) and stimulates gene expression through binding to two types of receptors, retinoic acid receptors (RARs) and retinoic X receptors (RXRs) [93]. In this study we showed that aldh1a2 , a RA synthesising enzyme, and two RA receptors, rar γ-a and rar γ-b, are expressed in developing gonopodia. Inhibiting RA synthesis with DEAB increases the length of newly formed terminal segments, whereas the segment length decreases when RA signalling is over-activated by exogenous RA. In addition, aldh1a2 is co-expressed with one of the androgen receptors (ar β) in developing gonopodia. Interestingly, this expression domain of both genes in the distal tip of the gononopodial rays is not found in developing swords, whereas both rar γ receptors are similarly expressed in developing swords and gonopodia.

Chapter I

Fgfr1 signalling in the development of a sexually selected trait in vertebrates, the sword of swordtail fish

Nils Offen, Nicola Blum, Axel Meyer and Gerrit Begemann BMC Developmental Biology (2008: 8)

1.1 Abstract

Background: One of Darwin’s chosen examples for his idea of sexual selection through female choice was the “sword”, a colourful extension of the caudal fin of male swordtails of the genus Xiphophorus . Platyfish, also members of the genus Xiphophorus, are thought to have arisen from within the swordtails, but have secondarily lost the ability to develop a sword. The sustained increase of testosterone during sexual maturation initiates sword development in male swordtails. Addition of testosterone also induces sword-like fin extensions in some platyfish species, suggesting that the genetic interactions required for sword development may be dormant, rather than lost, within platyfish. Despite considerable interest in the evolution of the sword from a behavioural or evolutionary view point, little is known about the developmental changes that resulted in the gain and secondary loss of the sword. Up-regulation of msxC had been shown to characterize the development of both swords and the gonopodium, a modified anal fin that serves as an intromittent organ, and prompted investigations of the regulatory mechanisms that control msxC and sword growth.

Results: By comparing both development and regeneration of caudal fins in swordtails and platyfish, we show that fgfr1 is strongly up-regulated in developing and regenerating sword and gonopodial rays. Characterization of the fin overgrowth mutant brushtail in a platyfish background confirmed that fin regeneration rates are correlated with the expression levels of fgfr1 and msxC . Moreover, brushtail re-awakens the dormant mechanisms of sword development in platyfish and activates fgfr1/msxC -signalling. Although both genes are co-expressed in scleroblasts, expression of msxC in the distal blastema may be independent of fgfr1 . Known regulators of Fgf-signalling in teleost fins, fgf20a and fgf24 , are transiently expressed only during regeneration and thus not likely to be required in developing swords.

Conclusion: Our data suggest that Fgf-signalling is involved upstream of msxC in the development of the sword and gonopodium in male swordtails. Activation of a gene regulatory network that includes fgfr1 and msxC is positively correlated with fin ray

growth rates and can be re-activated in platyfish to form small sword-like fin extensions. These findings point towards a disruption between the fgfr1/msxC network and its regulation by testosterone as a likely developmental cause for sword-loss in platyfish.

1.2 Background

Charles Darwin conceived not only the theory of natural selection, but also recognized that a theory of sexual selection is necessary to explain the presence of conspicuous traits in male animals that could not have arisen by natural selection [35]. A number of studies provided evidence that sexual selection increases taxonomic diversity, although it remains somewhat controversial if and how sexual selection alone can cause speciation (reviewed in [94, 95]). The body of theory about sexual selection has been extended through several new insights that explain the evolution of sexually selected traits and mating behaviour. Fishes of the genus Xiphophorus are a popular model in which various aspects of sexual selection have been studied extensively (reviewed in [14]). The most prominent sexually selected trait in male swordtail fish of the genus is the sword, a conspicuously pigmented elongation of the ventral caudal fin. The sword consists of several components, i.e. a ventral fin elongation and a characteristic pigmentation pattern [37, 96]. In the green swordtail X. helleri , it consists of centrally located yellow-orange or green coloured rays, that are flanked dorsally and ventrally by rays with strong melanisation (Figure 1.1A, B and [40]). Both length and colouration are important for mating success [32, 40]. The evolutionary history of the sword is of particular interest. One scenario, supported by molecular phylogenies, suggests that all extant Xiphophorus species, swordtails and sword-less platyfish, descended from a common, sworded ancestor [4, 5, 37]. Moreover, short extensions of the ventral portion of the caudal fin are also phylogenetically widespread and, for example, are found in Poecilia petenensis [97]. Platyfish (Figure 1.1C), a common name that is used to describe several swordless species that belong to a monophyletic clade within the genus Xiphophorus , however, secondarily lost their sword during evolution, possibly because the costs in terms of natural selection were higher than the gain in terms of sexual selection. Nonetheless, females of some platy species in which this has been tested still prefer sworded males over the swordless males of their own

species [38, 42]. The preference for elongated caudal fins seems to be much older than the trait itself, since it is also present in at least one species of the sister genus Priapella [54]. Therefore, the sword is thought to have evolved in response to a pre-existing female bias, such as a general preference for the apparent size [32, 42]. Due to this interesting evolutionary history, the sword presents a valuable model to study how evolution acts at the molecular level to generate or abolish a sexually selected trait. This objective has also driven research in other animals, e.g. the colour morphs in males of the livebearing fish Poecilia parae [98], the exaggerated hypercephaly in stalk-eyed flies [99], or the horns of dung beetles [100]. All of these are examples of model systems in which the basis of change in male exaggerated traits under sexual selection is amenable to genetic dissection.

Figure 1.1: Xiphophorus species and strains used. Adult morphologies of X. helleri (A) and X. maculatus (C) used in this study as representatives of the swordtail and platyfish lineages . Overview and nomenclature of adult fin rays (B) in the sword [43] of X. helleri and in the gonopodium [85] that is formed by swordtails and platyfish males. The caudal fin ray overgrowth mutant brushtail (D). Note that C and D show different strains of X. maculatus that exhibit dissimilar body colouration independent of the brushtail mutation.

One way to address this question in swordtails is to dissect the genetic pathways that might be involved in the development of the sword and to characterize these within a phylogenetic framework of the entire genus that involves swordtails and platyfish. So far, only little is known about the molecular basis of sword development. Hybridisation experiments between X. helleri and X. cortezi revealed that multiple genes control sword development, which were collectively termed ‘‘sword genes’’ (“Schwertgene”), i.e. genes or alleles that confer an ability to produce a sword in hybrids of platyfish and swordtails [44]. In addition, fin ray transplantation experiments have shown that sword rays are characterized by the possession of an organizing activity that induces neighbouring fin rays to contribute to the sword [52]. Sword-induction experiments with juvenile swordtails, treated with exogenous testosterone, revealed that testosterone is a sufficient and essential factor that induces sword development [43, 67]. Exogenous testosterone also induces the development of the gonopodium (Figure 1.1B), a modified anal fin used as a copulation organ that is common to all fish in the family Poeciliidae (the livebearing toothcarps). This might suggest that androgen signalling regulates a molecular pathway that induces both sword and gonopodium development. Interestingly, some platy species develop a small ventral extension of the caudal fin through testosterone treatment [43, 67, 101], suggesting that the genetic machinery underlying sword development is still partly intact even in normally swordless platyfishes. Most likely, this machinery has never been lost completely, even though it might have been inactive for more than a million years [4, 5]. Genes that regulate growth-dependent processes like fin regeneration are good candidates for genes involved in sword development. A candidate gene approach revealed msxC (muscle segment homeobox gene C ), a gene known to act in fin regeneration, to be specifically up-regulated in developing swords and gonopodia [59]. By combining available genetic and phylogenetic data, it was hypothesized that genes and pathways that shape the evolutionarily older gonopodium have been partly adapted for sword development [59]. Other putative candidate genes for sword development are upstream regulators of msxC , such as components of the Fgf (Fibroblast growth factor) signalling pathway. Fgf signalling controls epithelial-mesenchymal interactions in the external genital anlagen of

mammalian embryos [102]. Fibroblast growth factor receptor 1 (Fgfr1) appears to regulate msxC and msxB expression during caudal fin regeneration in zebrafish and is required for regenerative outgrowth of fin rays [70, 71]. Furthermore, Fgf ligands such as those encoded by the fgf24 and fgf20a genes have been shown to play a role in caudal fin regeneration or pectoral fin development [72, 73]. To test a putative role of Fgf-signalling in sword development we cloned the fgf receptor 1 and two fgf orthologs, fgf24 and fgf20a , from the swordtail X. helleri and analysed their expression pattern in developing swords and gonopodia as well as regenerating swords. From a developmental point of view, we asked whether regulation of fgf genes expression is associated with growth of the sword and gonopodium during development and sword regeneration. From an evolutionary standpoint, we were interested in evaluating whether potential differences in fgf gene expression between swordtails and platy species contribute to the understanding of the molecular changes that led to the loss of the sword during evolution. Furthermore, we analysed the expression of fgfr1 and msxC in regenerating caudal fins in the platyfish X. maculatus fin overgrowth mutant brushtail , where medial rays of the caudal fin continue to grow throughout the entire life of the animal (Figure 1.1D). We show that genes are regulated similarly in regenerating sword rays and elongated brush rays, although sword regeneration proceeds differently from regeneration in brushtail .

1.3 Results

Cloning and analysis of fgf genes Since sword development in X. helleri requires the growth of caudal fin rays, genes that act to regulate growth in the regenerating zebrafish caudal fin appeared to be suitable candidate genes that may also be involved in sword development. Up-regulation of fgfr1 and msx gene expression has originally been observed in the blastema during zebrafish fin regeneration [70], and it had subsequently been shown that inhibition of Fgf signaling during ongoing fin regeneration prevents further outgrowth and down-regulates the established expression of blastemal msx genes [70, 71, 103, 104]. fgf24 and fgf20a encode putative Fgfr1 ligands that are expressed or fulfill important functions in zebrafish fin regeneration [70, 73]. To clone the Xiphophorus orthologs from the green swordtail, Xiphophorus helleri , we used a RT-PCR strategy and caudal fin blastemata as source for

mRNA. The amplified fragment of the fibroblast growth factor receptor 1 (EU340805) covers 1248 bp of the protein’s open reading frame, including parts of the IG Domain II, the complete IG Domain III and parts of the tyrosine kinase domain (Figures S1.1A,B), found in vertebrate Fgfr1 [105]. Phylogenetic reconstruction of the fgf receptor family, using coding sequence, confirmed that we cloned a partial sequence of the X. helleri Fgfr1 ortholog (Figure 1.2A). The two cloned cDNA fragments of fgf24 include the complete coding sequence of fgf24 , parts of the 5’UTR and the whole 3’UTR sequence. The 633 bp ORF (EU340806) of X. helleri fgf24 codes for a 210 amino acid protein with a heparin-binding growth factors/fibroblast growth factor (HBGF/FGF) family signature (Figures S1.2A, B). Phylogenetic analysis of the 633 bp cDNA sequence verified the sequence to be the X. helleri fgf24 ortholog (Figure 1.2B). In addition we cloned two cDNA fragments of fgf20a , that together cover most of the coding and the complete 3’UTR sequence. The partial protein sequence, coded by 663 bp (EU340807), shows a conserved HBGF/FGF motif (Figures S1.2A, B). Interestingly, there is a QH-rich (aa 22-55) motif close the N-terminus of the sequence (Figure S1.2A). This motif could not be found in Fgf20a sequences of other vertebrate species. The phylogenetic analysis of the coding sequence confirmed it to be the X. helleri fgf20a ortholog (Figure 1.2C).

Figure 1.2 : Phylogeny of fgf genes. Phylogenetic analysis of vertebrate fgf receptors (A), fgf8/17/18/24 (B) and fgf9/16/20 (C) families using PhyML (upper values) and Mr. Bayes (lower values). For analysis the coding regions of fgf genes cDNAs were used. The position of the X. helleri orthologs of fgfr1 , fgf24 and fgf20a within the three phylogenies is highlighted (red box).

fgfr1 and msxC are differently regulated in caudal fins of maturing swordails and platyfish In order to test whether Fgf-signalling is involved in sword development of the green swordtail, X. helleri (Figure 1.1A), we treated 4-5 month old juvenile fish with 17-α- methyltestosterone to artificially induce this process. To allow both for the simultaneous generation of large numbers of experimental animals and for timed induction of sword development we prematurely induced swords in juvenile fish. Importantly, hormonally induced swords in immature juveniles do not show any sex-related morphological differences [43, 44]. Even adult females develop a sword under testosterone treatment that is indistinguishable from the male sword both in length and pigmentation [44], therefore the sex of the individual should not bias the downstream analysis. In developing swords, fgfr1 expression was first observed after 4 days of hormone treatment (dt), when black pigmentation along the dorsal border of the sword also becomes visible (data not shown). After 5 dt, when the outgrowth of sword rays had started, fgfr1 was mainly up-regulated in the distal tip of the main ventral sword-forming fin rays V7-V10 (Figure 1.1B and [43]) compared to median or dorsal rays (Figure 1.3A). However, fgfr1 was expressed much more strongly in V7-V9 than in V10 (Figure 1.3A). A slight up-regulation of fgfr1 was also detected in ray V6 (Figure 1.3A). Importantly, this pattern is comparable to that of msxC a gene that is strongly up-regulated in developing swords (Figure 1.3B and [59]). Up-regulation of fgfr1 was not observed in control fins (Figure 1.3C). This overlap in expression pattern of fgfr1 and msxC persists during later stages of sword outgrowth (compare Figures 1.3D-F and 1.3G-I). These finding suggest that high levels of fgfr1 expression correlate with the development of ventral caudal fin rays into swords. Furthermore, the spatio-temporal overlap of both fgfr1 and msxC expression patterns indicate a likely interaction of these genes during sword development.

Figure 1.3: Expression of fgfr1 and msxC in the developing sword. X. helleri fgfr1 is up-regulated during sword development. When maturation is induced by exogenous testosterone, fgfr1 is up-regulated in the ventral-most caudal fin rays in developing swords at 5 and 10 days of treatment (dt)(A, F). Weaker expression of fgfr1 can also be detected in non-sword rays (A, D, E). fgfr1 is not up-regulated in untreated control fins (C). fgfr1 expression overlaps with msxC , which is also up- regulated in developing sword rays ([59], B and I). In later stages of treatment, up-regulation of both genes in the distal part of the dorsal-most rays is observed in some individuals (G), which may develop a small “upper sword” ([67]). Like fgfr1, msxC expression is also detected in non-sword rays (G, H). When maturation is induced by exogenous testosterone in the platyfish, X. maculatus fgfr1 and msxC are similarly expressed in all caudal fin rays after 5 dt (J and K). The expression levels are comparable to untreated fish (L). After 10 dt fgfr1 is more strongly expressed in the ventral-most fin-rays (O) compared to other rays (M, N), which may correspond to the formation of a small ventral swordlet [67, 101]. White arrowheads indicate gene expression. ( X. helleri : n= 10 for every stage and probe; X. maculatus : 5 dt: n= 5; 10 dt and controls: n= 3; scale bars: 200 µm)

To test if changes in the regulation of fgfr1 and msxC are linked to the absence of the sword in platyfish, we assayed the expression of both genes in the caudal fin of the platyfish X. maculatus (Figure 1.1C) after 5 and 10 days of testosterone treatment. The expression of both genes in caudal fins at 5 dt differs clearly between X. helleri and X. maculatus . In X. maculatus both genes are uniformly expressed with no differences

between sword and non-sword rays (Figures 1.3J and 1.3K). In addition, the expression patterns of both genes in testosterone treated fins are quite similar to those of control fins (Figure 1.3L). At 10 dt however, both genes are up-regulated in a subset of ventral fin rays (Figures 1.3M-O and not shown). This expression pattern is likely to mark the fin rays that will form a small caudal extension under high exogenous levels of testosterone [67, 101]. Based on the expression data, we conclude that loss of sword ray specific regulation of fgfr1 or msxC could have been involved in the secondary loss of this trait in platyfish. fgfr1 and msxC regulation in maturing anal fins is conserved between swordtails and platyfish Males of both platyfish and swordtails as juveniles possess typical anal fins, that during sexual maturation transform into a gonopodium, an intromittent organ for internal fertilisation [61]. Despite some differences in gonopodium morphology between species, all gonopodia are formed by anal fin rays 3-5 that develop into a structure that can deliver sperm into the females genital tract as well as scrape out sperm of other males through hooks that are formed by modification of fin ray elements (Figure 1.1B and [61, 85]). We expected that gonopodia of both swordtails and platyfish show a similar spatio-temporal pattern of fgfr1 and msxC expression. Because it was impractical to identify sufficient numbers of normally developing male juvenile fish at the desired stages, we analysed the expression patterns of both genes in artificially induced gonopodia of the swordtail X. helleri and the platyfish X. maculatus . At 5 days of testosterone treatment, strong expression of fgfr1 was found in the distal part of the main gonopodial rays 3, 4 and 5, the so-called 3-4-5 complex [82] of X. helleri (Figure 1.4A). Because gene expression in deeper layers of fin rays may be shielded from detection during whole mount in situ hybridisation [106], we performed in situ hybridisation on longitudinal sections which reveal strongest expression of fgfr1 in mesenchymal cells at the tip of growing gonopodial rays (Figure 1.4B) . This pattern persists during later stages of gonopodium development (Figure 1.4C). In addition, fgfr1 is up-regulated in the interray tissue (Figures 1.4A and C).

Figure 1.4: Expression of fgfr1 and msxC in the developing gonopodia of X. helleri and X. maculatus. fgfr1 and msxC are both expressed in developing gonopodia of X. helleri and X. maculatus. In X. helleri fgfr1 is up- regulated at 5 days (A) and 10 days (C) of treatment in mesenchymal cells (B) of the main gonopodium-forming rays 3-5 compared to control fins (G). In addition fgfr1 is strongly expressed in the interray tissue of those rays (A, C). As in developing swords, fgfr1 expression overlaps with msxC expression (D-F) . In early stages of gonopodium development (5 dt) of the platyfish X. maculatus , the expression patterns of fgfr1 (H) and msxC (J) resemble that of X. helleri . Both genes are up-regulated in the same set of fin rays compared to untreated controls (L). Expression of both genes (I, K) at 10 dt is comparable to that of X. helleri with species-specific differences in the shape of growing rays. Black arrowheads indicate the expression in the distal part of the fin rays, white arrowheads indicate inter-ray expression. (X. helleri : n= 10 for every stage and probe; X. maculatus : 5 dt: n= 5; 10 dt and controls: n= 3; scale bars: A, C, D, F-L: 200 µm; B and E: 100 µm)

As in developing swords, the spatio-temporal expression pattern of fgfr1 is similar to that of msxC , which is up-regulated in the mesenchyme of gonopodial rays 3 to 5 and in interray tissue (Figures 1.4D-F). Both genes are not up-regulated in untreated fins (Figure 1.4G).

The spatio-temporal expression pattern of fgfr1 and msxC in developing gonopodia of the platyfish X. maculatus approximately resembles the pattern found in X. helleri . Both genes are up-regulated in the distal part of the gonopodial rays 3, 4 and 5 and in the interray tissue (Figures 1.4H, I, J, K) compared to control fins (Figure 1.4L). The different shapes of the distal fin ray tips in X. maculatus and X. helleri are due to species- specific differences between the gonopodia [85]. fgfr1 and msxC show similar expression profiles in regenerating swords High levels of msxC transcription are also associated with regenerating sword rays after amputation [59]. It is assumed that the general mechanisms of growth control that act during early development are re-established during regeneration [107, 108]. To test whether fgfr1 is similarly regulated in regenerating and in developing sword rays, we assayed gene expression in caudal fin blastemata. The regeneration kinetics of X. helleri roughly equals that of zebrafish at 25°C, where the regenerative outgrowth starts at ~4 dpa [109]. fgfr1 is expressed in the basal layer of the epidermis and in a proximal region, which are likely to be scleroblasts (Figure 1.5A). msxC and fgfr1 expression overlap in these cells. Furthermore, msxC is not expressed in the basal epidermal layer, but transcription is high in the distal blastema (Figure 1.5B). Sword rays and non-sword rays show similar levels of fgfr1 and msxC at different stages of regenerative outgrowth (Figures 1.5C-F and Figures 1.5G, H). Both genes stay highly up-regulated in growing blastemata until 7 dpa (Figures 1.5E-H). At 11 dpa, when the sword region has begun to overgrow the rest of the regenerate, fgfr1 and msxC become differently regulated in sword rays compared to other rays. Both fgfr1 (Figure 1.5I) and msxC (Figure 1.5J) are more strongly expressed in sword rays than in non-sword rays, even though this difference in expression was more clearly observed for msxC. Judging from these data, it is apparent that both genes are similarly regulated in developing as well as regenerating swords. Furthermore, it is likely that due to the lack of fgfr1 expression in the distal blastema, msxC expression in this domain is regulated by factors other than Fgfr1.

Figure 1.5: Expression of fgfr1 and msxC during caudal fin regeneration. fgfr1 is expressed in the regenerating caudal fin blastema. In situ hybridisation on longitudinal sections at 4 days post amputation (dpa) reveal fgfr1 expression in the basal layer of the epidermis and in scleroblasts (A). msxC expression overlaps with that of fgfr1 in scleroblasts (B) and shows additional expression in the distal blastema (B, G, H). There is no overall clearly visible difference in expression of fgfr1 (C-F) and msxC (G, H) between sword and non-sword regenerates until 7 dpa. At 11 dpa fgfr1 (I) and msxC (J) show higher levels of expression in regenerating sword rays than in non-sword rays, though this difference is more obvious for msxC . White arrowheads indicate expression in scleroblasts, black arrowheads the msxC expression domain in the distal blastema and white arrows the plain of amputation . bl= basal epidermal layer; db= distal blastema; e= epidermis; l= lepidotrichia; m= mesenchyme (4 dpa fgfr1 : n= 8; 7 dpa fgfr1 : n= 5; 7 dpa msxC : n= 5; 11 dpa fgfr1 : n= 5; 11 dpa msxC : n= 4; scale bars: A and B: 100 µm, C-J: 200 µm)

fgf24 and fgf20a are expressed in regenerating, but not developing swords To further analyse the regulation of sword development and regeneration upstream of fgfr1 , we cloned two putative ligands of Fgfr1, fgf24 and fgf20a , which are known to be involved in fin regeneration and development [72, 73, 110]. To this end we examined the expression patterns of both genes in developing and regenerating swords. We detected strong expression of fgf24 and fgf20a in caudal fin regenerates up to 3 dpa and ~1 dpa, respectively, before the transcription rate of both genes decreased (Figures 1.6A-D and data not shown). Therefore both genes are unlikely to play a role in the regulation of Fgf-

signalling or msxC in later stages of sword regeneration, when gene regulation becomes different between sword and non-sword rays. In addition, as neither fgf24 nor fgf20a were expressed in developing swords or gonopodia (data not shown), it is unlikely that they act as ligands for Fgfr1 during these processes.

Figure 1.6: Expression of X. helleri fgf24 during fin regeneration. fgf24 is expressed in the wound epidermis at 3 dpa (A, B). Expression diminishes after 3dpa and is almost absent by 4 dpa (C, D). fgf24 is not differentially expressed in sword rays (B) compared to non-sword rays (A). White arrowheads indicate the expression in wound epidermis and white arrows the level of amputation. (3dpa : fgf24 : n= 6; 4 dpa fgf24 : n= 5; scale bars: 200 µm)

Expression levels of fgfr1 and msxC are correlated with growth rates of regenerating fin rays To address the question whether enhanced fgfr1 and msxC expression are generally associated with extended growth of fin rays, we analysed gene expression in regenerating caudal fins of X. maculatus brushtail mutants (Figure 1.1D). Individuals carrying the dominant brushtail mutation are characterized by a life-long overgrowth of medial fin rays in the caudal fin (compare Figures 1.7A and 1.7B), which is independent of sex or sexual maturity [10]. The mutation causing this phenotype is not known. Mature male brushtail mutants also grow a swordlet, a small ventral fin extension (Figure 1.7B), similar to the ventral caudal fin extension that naturally occurs in two species of platyfish, X. andersi and X. xiphidium , and similar to that which can be artificially produced by high levels of exogenous testosterone in some species of platyfish such as X. maculatus [67, 101]. However, it lacks the pigmentation pattern typical of swords in swordtails. Since brushtail mutants are already born with a brush [10] and developing embryos are not viable when extracted from their mothers, we asked whether fgfr1 and msxC are differently expressed in regenerating brush rays, compared to more dorsal or ventral caudal fin rays. Expression of fgfr1 and msxC is strongest in the median fin rays (Figures 1.7C-J), which becomes particularly obvious after 4 dpa. Both genes show a graded expression pattern with a decrease of expression levels towards the dorsal and

ventral fin margins (Figures 1.7D-F, H-J). At later stages of regeneration fgfr1 and msxC are also stronger expressed in the ventral-most caudal fin rays of males that form the swordlet, but was absent in females (compare Figure 1.7F to 1.7J, and not shown).

Figure 1.7: Expression of fgfr1 and msxC in regenerating caudal fins of brushtail mutants. Compared to a wildtype platyfish (A), X. maculatus brushtail mutants possess elongated median caudal fin rays (B). Male brushtail mutants also develop a small ventral extension of the caudal fin (swordlet). fgfr1 and msxC show a graded expression pattern in regenerating caudal fins of brushtail mutants at different stages of regeneration with strongest expression in the median fin rays (C-J). fgfr1 (C-F) and msxC (G-J) are expressed in a similar pattern as in X. helleri regenerating caudal fins. At later stages of regeneration, fgfr1 (F) and msxC show stronger expression in the ventral-most caudal fin rays of males compared to females (J). White arrowheads indicate expression in scleroblasts, black arrowheads the msxC expression domain in the distal blastema and white arrows the plain of amputation. (n= 3 for every stage and probe; scale bars: A and B: 1 mm; C-J: 200 µm)

The graded expression patterns suggest that fgfr1 and msxC correlate with different growth rates of median fin rays compared to more ventral or dorsal rays. In order to test this hypothesis, we amputated the caudal fins of adult brushtail mutants and compared the growth rates of regenerating fin rays at different positions within the caudal fin. We did this by calculating the average length difference between the regenerate of the median fin ray 1 and more dorsal fin rays 4, 6 and 8 at 4 dpa and 8 dpa (Figure 1.8A). We found that the individual fin rays show significantly different regeneration rates (as determined by a t-test), depending on their position within the caudal fin, with the median-most ray 1 showing the fastest regeneration rate (Figure 1.8B). The regeneration rate decreases the more closely a fin ray is located to the dorsal edge of the fin, with regenerating ray 8 showing the slowest regeneration rate (Figure 1.8B). Differences in regeneration rates between fin rays according to their position in the caudal fin are more pronounced at 8 dpa (Figure 1.8B). The correlation between higher fgfr1 and msxC expression levels and enhanced regenerative outgrowth suggests that both genes are involved in modulating the growth rate of individual fin rays. Figure 1.8. Different regeneration rate of brushtail fin rays, depending on their position in the caudal fin. The regenerate’s length of four different fin rays, highlighted in the schematic drawing of an adult brushtail fin (A) were measured at 4 days post amputation (dpa) and 8 dpa. The regenerate’s length of the dorsal fin rays 4, 6 and 8 were then compared to that of the median fin ray 1. Dorsal fin rays regenerate more slowly than the median fin ray 1, shown as average length difference between fin ray regenerates (B). The difference in regeneration rate increases the closer a fin ray is located to the dorsal edge of the fin. The position dependence of regeneration rates is more obvious at 8dpa (n= 11; *P<0.00001, t-test).

1.4 Discussion

The sword is a sexually selected trait in swordtail fish that is thought to have evolved in the common ancestor of all extant Xiphophorus fishes, and was lost secondarily in the lineage leading to the platyfish [4, 5, 37]. Questions about the evolution and the subsequent loss of the sword can therefore be cogently formulated within this phylogenetic framework. Zander and Dzwillo coined the term “sword genes” for those genes that confer an ability to produce a sword in hybrids of platyfish and swordtails [44]. We employed a candidate gene approach to more broadly identify genes involved in sword development. We reasoned that genes that act in growth-dependent processes like fin regeneration might also be suitable candidates for regulating growth in growing fin rays. We therefore evaluated changes to the expression of genes involved in Fgf- signalling, a signalling pathway known to control outgrowth, and msxC expression in regenerating caudal fins of zebrafish [70, 73, 104]. fgfr1 expression in developing and regenerating swords Our results show that fgfr1 is specifically up-regulated in growing fin rays of the sword and the gonopodial rays of the 3-4-5 complex as a response to the induction of male sexual trait formation by exogenously supplied testosterone. Several studies suggest that Fgfr1 plays essential roles during appendage formation and regeneration: expression of fgfr1 is re-established during regeneration of limbs and caudal fins [70, 111], and loss of fgfr1 function in these tissues blocks blastema formation and regenerative outgrowth [70, 71, 112]. Moreover, Fgfr1 is involved in supporting outgrowth of the mouse limb bud by maintaining mesenchymal cell survival and influencing the development and identity of digits [113, 114]. Interestingly, the spatio-temporal expression pattern of fgfr1 generally overlaps with that of the transcription factor msxC in developing swords and gonopodia [59]. msx genes are known to keep cells in an undifferentiated state by promoting cell proliferation [115, 116]. In zebrafish, msxB has been shown to regulate cell proliferation in the distal blastema of regenerating caudal fins [71, 117]. Accordingly, knock-down and chemical inhibition of Fgfr1-signalling abolishes the expression of msxB and msxC [70, 71]. Therefore, as judged by the up-regulation in growing sword and gonopodial rays, we propose that an increase in fgfr1-mediated signalling regulates cell-proliferation,

similar to the processes in fin regeneration, through the activation of target genes like msxC . Are there positional values that bias the ventral caudal fin for sword development? In a set of transplantation experiments it has been shown that the two major sword rays, V8 and V9, act to organize flanking rays into the developing sword [52]. This is achieved through a non-autonomus signalling process in which fin rays are induced to grow more strongly the closer they develop to V8 and V9. The development of sword rays is initiated through androgen-signalling in maturing juvenile fish, which leads to mainly autonomous growth of rays V8 and V9 and inductive signalling to flanking rays V7 and V10, which together comprise the mature sword. Moreover, the potential to develop organizing activity in V8 and V9 is established during embryonic or larval development [52]. In zebrafish ventral tail fin tissue has been fate-mapped to late gastrula stage embryos [118, 119]. It is therefore likely that positional information within the Xiphophorus caudal fin is also attained during embryonic stages. Genes of the TGF β gene family, such as alk8 /lost-a-fin or tolloid/mini fin , whose mutant phenotypes include defects of the ventral caudal fin [120, 121], could therefore provide all or part of the positional information that is used to specify sword ray fate. To test whether fgfr1 and msxC are also differently regulated in regenerating swords, and to further characterize these interactions at the cellular level, we compared expression in regenerating sword and non-sword rays. An overlap of fgfr1 and msxC expression was detected in scleroblasts. In addition, fgfr1 is expressed in the basal layer of the epidermis, where msxC was not found, suggesting that signalling through fgfr1 alone is not sufficient to activate msxC expression in the basal epidermal layer. Likewise, msxC is temporally activated in the distal blastema, although fgfr1 is absent from this domain, suggesting that msxC is regulated by signals other than fgfr1 in the distal blastema. Interestingly, such a distal expression domain of fgfr1 is described in zebrafish [70], indicating species-specific differences in fgfr1 - and, probably, in msxC -regulation between X. helleri and D. rerio . This may not be all that surprising since these two fish species are evolutionary rather distantly related [122, 123]. In addition, functional differences between zebrafish and medaka fgfr1 have recently been demonstrated [124]. We also noted that, although expression levels of fgfr1 are clearly elevated in sword rays,

when compared to non-sword rays, the difference was never as strongly as that observed for msxC . A disproportionally higher activation, like for msxC, might be expected in a scenario where msxC expression is up-regulated as part of the downstream effects of signalling through Fgfr1. fgfr1 and msxC are part of a network that regulates sword development We previously proposed that the gene regulatory network underlying sword development did not evolve anew, but was co-opted from the gonopodium [59]. The 3-4- 5 complex of the gonopodium is likely to have evolved only once in the common ancestors of all poeciliid fish and is thus evolutionary older than the sword, which originated in its fully developed form in the lineage leading to the genus Xiphophorus [37, 61, 82]. In support of this view, the regulatory dynamics of fgfr1 and msxC show very similar expression patterns during gonopodium growth and caudal fin regeneration that are conserved between swordtails and platyfish, whereas differences in later stages of gonopodium development are likely to reflect differences in species-specific morphology [85]. The similarities in gene expression between swords and gonopodia supports the hypothesis that sword development evolved by co-option of an androgen-regulated genetic network from the evolutionary older gonopodium. fgfr1 is most likely only one of the regulatory nodes within an intricate network of signalling pathways that converge on the activation of msx genes. There is compelling evidence that msx genes are also regulated by Bmp signalling during development and regeneration of vertebrate appendages [125-127]. For example, inhibition of BMP signalling in the developing mouse limb leads to down-regulation of Msx2 [127], while over-activation of the pathway results in up-regulation of Msx1 and Msx2 [125]. Activation may be direct, as Bmp4 induces the interaction of Smads with Msx1 regulatory sequences [128]. Bmp-regulation of Msx genes seems to be widely conserved, since inhibition of Bmp signalling also leads to down-regulation of msxb in regenerating fins [126] and to a loss of Msx1 expression in the regenerating tail of Xenopus tadpoles [129] . Wnt signalling is a third component of this regulatory network. Manipulation of its activity in regenerating limbs, fins and tadpole tails showed that Wnt signalling regulates the expression of msx and fgf genes and is important for regenerative

outgrowth [130-132]. Recent work in the poeciliid fish Gambusia affinis had shown that induction of sonic hedgehog (shh) expression by androgens is required for gonopodium formation [86]. Given that Shh controls dermal bone development in the zebrafish caudal fin [117], it is likely that also Shh signalling may be required for sword development. Functional tests will be required to dissect the molecular network that controls msxC expression to better understand how these different signalling pathways act together to shape the male sword. The similarities in gene expression between swords and gonopodia supports the hypothesis that sword development evolved by co-option of a modular androgen-regulated fgfr1/msxC network from the evolutionary older gonopodium. Juveniles of X. maculatus fail to up-regulate fgfr1 and msxC in the ventral caudal fin rays at testosterone-levels that otherwise cause sword development in swordtails after 4 days of treatment. It is only after prolonged exposure that both genes start to be expressed more strongly in the ventral-most rays. This expression marks the development of a “swordlet”, a small colourless caudal fin extension, that develops in some platy species under excessive exposure to testosterone [67, 101]. A swordlet is also found naturally in the species X. andersi and X. xiphidium and outside the genus Xiphophorus in other species of poeciliid fishes, e.g. in Poecilia petenensis . The failure to up-regulate the fgfr1/msxC network in response to endogenous androgens may have caused the loss of swords in the lineage leading to the platyfish. Since artificially high levels of testosterone can overcome this inhibition ([67, 101] and Figure 1.3), the evolutionary changes that led to sword-loss may have been quantitative ones that could have altered the strength of genetic interactions within the fgfr1/msxC network or its upstream regulation. One attractive possibility for the evolutionary loss of swords therefore would be an acquired loss in the sensitivity of the fgfr1/msxC network to be activated by androgen-signalling (Figure 1.9), possibly at the level of Fgf regulation.

Figure 1.9. Growth control in prospective sword rays in the ventral caudal fin. An signalling network that includes activation of fgfr1 and msxC expression regulates growth of ventral rays in the caudal fin and is activated by endogenous levels of testosterone in swordtails (A); in platyfish an evolutionary weakening (interrupted arrow) in the ability of testosterone to activate the network results in insufficient signalling and the absence of sword development (B); high exogenous levels of testosterone overcome the inhibition upstream of the network and induce a swordlet in some platyfish species (C); the platyfish mutant brushtail raises overall signalling levels of the network in all rays, allowing endogenous testosterone to suffice for the induction of a swordlet in male fish (D). Width and shading of arrows indicate the strength of fgfr1/msxC signalling activation in sword rays (wide black: induction of a sword; narrow grey: maintenance of basal fin growth; narrow black: induction of a swordlet). brushtail enhances sword development in platyfish We analysed fin ray regeneration in X. maculatus brushtail , a mutant that spontaneously arose in platy strains within the pet trade (Figure 1.1D). The brush resembles the sword with regard to size, but occurs in both sexes, consists of medial fin rays, and lacks conspicuous pigmentation [10]. In contrast to swords, brushes are formed during embryonic or larval development and show allometric growth throughout the whole life of an individual. In adult male platyfish, brushtail often also induces a swordlet on the ventral caudal fin margin (M. Schartl, personal communication). The X. maculatus brushtail fish used in our studies are not congenic and may contain alleles derived from hybrids between swordtails and X. maculatus . It could thus be argued that brushtail hybrids are more inclined to develop swords due to introgression of sword alleles. However, we note that formation of the swordlet in males is always linked to the brushtail phenotype and does not occur in male brushless siblings that derive from a cross between brushtail parents. Therefore brushtail enhances a dormant program of sword development [133], similar to prolonged exposure to testosterone in X. maculatus . We thus reasoned that the mechanisms of fin growth in swords and brushes may be similar. Our analysis of fin ray regeneration revealed that (1) regeneration rates in brushtail and swordtail caudal fins differed in that sword rays regenerate at the same rate

as the rest of the fin (Figures 1.5 and 1.7 and [59]), whereas rays within the brush exhibit accelerated regeneration rates the closer they are to the middle of the fin, and (2) expression levels of fgfr1 and msxC show a positive correlation with regeneration rates. Therefore, brushtail mutants lend support to the hypothesis that the strength of fgfr1/msxC expression is correlated with the rate of regenerative growth, which in brushtail mirrors the life-long overgrowth of the brush. In contrast, sword rays initially exhibit regeneration rates and strong fgfr1/msxC expression indistinguishable from non- sword rays. Growth and gene expression ceases once rays have reconstituted the previous fin length, while fgfr1 /msxC transcription remains high in the sword. The correspondence of gene expression levels in swords, gonopodia, and in brushtail suggests that fgfr1 and msxC expression are a general readout of proliferation in fin rays. Pharmacological interference with fgfr1 signalling is not practicable in Xiphophorus fish, as treatments using the standard zebrafish assay, where many fish are treated in small volumes of water, cause a high mortality, while treatments in volumes used in our testosterone experiments would require excessive amounts of the drug. Moreover, electroporation of morpholinos has failed due to persistent problems with the initial injection, which has to be performed into growing fin rays, rather than into a regeneration blastema (not shown). Therefore we had to resolve to associative studies of gene expression in a phylogenetic framework and in mutants to study the involvement of candidate genes in sword development. As an explanation of the development of a swordlet in adult male brushtail platyfish, our findings suggest that fgfr1/msxC -mediated signalling is constitutively activated throughout the brushtail caudal fin and already in young larvae of both sexes, with strong activation in medial rays and lower, yet elevated levels at the margins. Although endogenous levels of testosterone in platyfish are usually not sufficient to induce or maintain a sword, growth of a swordlet could nevertheless ensue if in maturing male platyfish both testosterone- and brushtail -activated signalling pathways converged on the activation of the fgfr1/msxC regulatory network. brushtail simulates the effect of high exogenous doses of testosterone on swordlet formation in platyfish, in agreement with the idea that the fgfr1/msxC network is a major downstream target of androgen signalling during sword development (Figure 1.9). It is interesting to note that brushtail re-awakens

a dormant program for sword development that had been evolutionary lost in platyfish. Therefore, swordlets likely employ at least parts of the developmental program controlling sword growth [133]. An examination of the molecular mechanisms that activate fgfr1/msxC signalling in brushtail will be valuable for exploring the regulation of this gene network in controlling the growth of ventral caudal fin rays.

The ligands required for Fgfr1 activation during fin development remain elusive To examine the regulation of fin growth upstream of fgfr1 , we isolated fgf20a and fgf24 as potential ligands in caudal fins that develop a sword under testosterone treatment . In zebrafish, fgf20a plays an essential role in blastema formation, while fgf24 is expressed in regenerating caudal fins and is required for pectoral fin bud initiation [70, 72, 73, 110]. Therefore, both genes are putative candidates to control early phases of sword and gonopodium development such as the re-initiation of fin growth under testosterone. We show that both genes are transiently expressed in regenerating fin rays and are down-regulated after 3 dpa, suggesting that fgf20a and fgf24 fulfill similar functions in blastema formation in zebrafish and X. helleri . Since neither gene was expressed at any detectable level in developing swords or gonopodia, other fgfs remain attractive candidates that might mediate the activation of sword development downstream of testosterone-activated androgen receptors. A number of other Fgfs are known with confirmed roles during appendage formation and regeneration, such as Fgf2, Fgf4, Fgf8 and Fgf10, which are able to induce regeneration in Xenopus limbs or chick limb buds [134-136], and Fgf16 as well as Fgf8 , Fgf4 and Fgf10, which are involved in zebrafish pectoral fin development (reviewed in [137]). Further investigation of these candidate genes should uncover the ligand of Fgfr1 that controls development of the sword and gonopodium.

1.5 Conclusions We have shown a correspondence of fgfr1 and msxC expression levels in developing and regenerating swords, gonopodia, and in the fin-overgrowth mutant brushtail, which suggests that fgfr1 and msxC are part of a genetic network that regulates the rates of growth and regeneration in male-specific modifications of adult fins in swordtails and

platyfish. Through inter-species comparisons with swordless platyfish, we have shown that high levels of fgfr1 expression are associated with developing sword rays. In line with this assertion brushtail mutant caudal fin rays exhibit elevated expression of this network, leading both to a regionalised fin overgrowth phenotype and to an enhancement of an otherwise dormant program of sword development in platyfish. Taken together, we propose that changes in the regulation of a genetic network that includes fgfr1 and msxC contribute to the loss of the sword in the platyfish lineage. Our characterization of the two known regulators of Fgf signalling in teleost fin regeneration, fgf20a and fgf24 , rules out an involvement in sword development, but other Fgfs remain candidates that might mediate the activation of sword development downstream of testosterone-activated androgen receptors. Finally, the similarities in gene expression between swords and gonopodia supports the hypothesis that sword development evolved by co-option of an androgen-regulated fgfr1/msxC network from the evolutionary older gonopodium.

1.6 Materials and Methods

Fish stocks and maintenance Xiphophorus helleri and X. maculatus were taken from stocks kept at the animal research facility at the University of Konstanz. The X. maculatus brushtail mutant, a commercial breed, was obtained from a local pet shop. Fish were maintained on a 12:12h light:dark cycle at 24°C in 110-litre densely planted aquaria and were fed TetraMin flakes and Artemia.

Testosterone treatment and fin regeneration Up to six juvenile individuals of X. helleri , aged between 3 and 6 months, were placed in a 30-litre tank. 17-α-Methyltestosterone (1mg/mlstock solution in ethanol; Sigma- Aldrich, Munich, Germany) was added to the water twice a week to a final concentration of 10 µg/l. After 5 or 10 days of testosterone treatment approximately 1/3 of the distal part of the caudal fin and approximately 2/3 of the anal fin was amputated using a sterile razor blade. For fin amputation fish were anesthetized by incubation in a solution of 0.08 mg/ml tricaine (3-aminobenzoicacid-ethylester-methanesulfonate; Sigma-Aldrich, Munich, Germany).

For regeneration experiments adult X. helleri and X. maculatus individuals were anesthetized and 1/3 of the caudal fin was amputated . Subsequently, fins were allowed to regenerate at 24°C for variable time periods, without addition of 17-α- Methyltestosterone, depending on the experiment. Fish were anesthetized again and the blastema was removed. Fins and blastemata used for in situ hybridisation were fixed in 4% paraformaldehyde (Sigma-Aldrich, Munich, Germany) in phosphate buffered saline (PBS) overnight, transferred to methanol and stored at -20°C until use.

Cloning fgfr1 , fgf24 and fgf20a Total RNA was isolated from caudal fin blastemata 24 hpa (for fgf20a ), 48 hpa (for fgf24 ) and 72 hpa (for fgfr1 ) and used for cDNA synthesis as described [59]. Degenerate Primers were designed based on an alignment of cDNAs from Danio rerio , Tetraodon nigroviridis and Takifugu rubripes to amplify cDNA fragments of the desired genes. A 1299 kb fgfr1 fragment was amplified by PCR using the Primer pair fgfr-fw1: 5’- CCIGAIAARATGGARAARAARYTGCAYGC-3’ and fgfr-rev1: 5’- CIGGIACYTGGTMIGGRTTRTARCA-3’. A 605 bp fgf24 fragment was amplified by PCR using the Primers fgf24-fw1: 5’-GAKAGIGCARCRIGYIYGGAIRC-3’ fgf24-rev1: 5’-GTCCICYYIKCCYTTKGGYTGGCGC-3’ and fgf24-rev2: 5’- CCAGTATAAATAAMAYRACAGACAC-3’. A 497 bp fgf20a fragment was amplified by PCR using the Primer pair fgf20a-fw2: 5'-GGSTCTCATTTCGTYCTCAC-3' and fgf20-rev0: 5’-GTRTTRTACCARTTYTCYTC-3’. To obtain fgf24 /fgf20a fragments of appropriate size for RNA probe generation 3’RACE reactions were performed using a 3’RACE kit (Roche). A ~1,3 kb fgf24 fragment was amplified using the gene-specific primers fgf24R-fw1: 5’- CTACAGCAGGACCACGGGCAAAC-3’ and fgf24R-fw2: 5’- CAAGAAAGGCTCACGCACCACGC-3’. A ~1,2 kb fgf20a fragment was amplified using the gene-specific primers fgf20a-fw1: 5’-CATCAGAGGAGTGGACAGCGGC-3’ and fgf20a-fw2: 5’-CGCCATGAACAGCAAGGGGGAG-3’. The PCR products were gel-purified using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany) and cloned into the pCRII-TOPO vector (Invitrogen, Karlsruhe, Germany) for sequencing.

Whole-mount in situ hybridisation Antisense and sense RNA probes were generated using a digoxigenin labelling kit (Roche, Mannheim, Germany). Probes for fgfr1, fgf24 and fgf20a were generated from the cDNA fragments listed above. msxC probes were generated from a 635 bp cDNA fragment [59]. Because of the high sequence conservation within the genus, X. helleri antisense probes could also be used in the platyfish X. maculatus . In situ hybridisation of Xiphophorus fins and blastemata were performed as described [70] with several modifications. Prehybridisation was done 4h at 68°C in formamide solution (50% formamide, 5x SSC, 0,1% Tween-20, pH to 6 with 1 M citric acid). Post- hybridisation washing steps were initiated at 68°C with formamide solution. To block unspecific binding sites 0,5% blocking reagent (Roche, Mannheim, Germany) in PBST (PBS/0.1% Tween-20) was used. Antibody incubation was done at 4°C overnight. After fixation of stained fins/blastemata, the tissue was washed twice for 20 min in PBST, 20 min in ethanol/PBST (70:30) and 20 min in 100% ethanol and stored at 4°C. The specificity anti-sense probes were verified with sense probe experiments.

In situ hybridisation on longitudinal sections 4 day-old caudal fin blastemata and anal fins from individuals treated with 17- α− Methyltestosterone for 5 days were fixed in 4% Paraformaldehyde (Sigma-Aldrich, Munich, Germany). Longitudinal sections of 10 µm thickness were created using a Reichert-Jung Autocut 2040 Microtome and in situ hybridisation was performed as described [138].

Microscopy and image editing Fin explants and brushtail caudal fins were analysed using a Zeiss Stemi SV11 Apo. Pictures were taken using the AxioVision software v3.1 (Zeiss, Jena, Germany) and the digital camera Zeiss AxioCam MRc. The pictures were processed using Adobe Photoshop.

Fin ray measurement and calculation Pictures from regenerating caudal fins of males and females were photographed at 4 and 8 days post amputation. The regenerate length of the median fin ray 1 and more dorsal fin rays 4, 6 and 8 was measured using the software ImageJ [139]. In case of bifurcation both semi-rays’ regenerates were measured and the average was calculated. To eliminate variation in regeneration speed between the individuals, the difference in length between the regenerate of the median ray 1 and the three other ray regenerates were calculated for each fish. Last, the average was calculated for each difference and graphically presented using Microsoft Excel. A paired, one-tail t-test was used to determine whether the calculated average distances differ significantly from each other.

Phylogenetic analysis and motif prediction Phylogenetic trees of fgf receptors , the fgf8/17/18/24 and the fgf9/16/20 subfamily were constructed using Maximum likelihood (ML) and Bayesian methods of phylogeny inference [140]. ML analyses were performed using PHYML [141]. The best fitting models of sequence evolution for ML were obtained by ModelTest 3.7 [142]. ML tree topologies were evaluated by a bootstrap analysis with 500 replicates [143]. To confirm obtained Tree topologies Bayesian analyses were initiated with random seed trees and were run for 100,000 generations for fgf receptors and fgf8/17/18/24 and 1000,000 generations for fgf9 /16 /20 . The Markov chains were sampled at intervals of 100 generations with a burn in of 100 trees for fgf receptors and fgf8/17/18/24 and 500 trees for fgf9/16 /20 . Bayesian phylogenetic analyses were conducted with MrBayes 3.0b4 [144] using the general time reversible model GTR+I+G [145] for fgf receptors and fgf8/17/18/24 and the Tamura-Nei model TrN+I+G [146] for fgf9/16/20 . The sequences used for the analysis are listed in Table S1.1 (Additional material). Sequences that could not be aligned with confidence were excluded from the analysis. ScanProsite [147] was used to predict conserved motifs in the translated amino acid sequences.

Chapter II

A Subtractive Hybridisation approach to identify novel genes involved in the development of the sword in green swordtail (X. helleri )

Nils Offen, Amanda Duckworth , Axel Meyer and Gerrit Begemann (Manusscript)

2.1 Abstract

The sword, a colourful extension of the ventral caudal fin of male swordtails, plays a prominent role during female mate choice. Sword development is initiated by the sustained increase of testosterone in males during sexual maturation, but can be prematurely induced by exogenous testosterone. This offers the opportunity to study which genes are active at a specific stage of sword development. Here, we employed suppression subtractive hybridisation (SSH), to identify genes specifically up-regulated in induced swords and gonopodium at different stages of development. We identified 128 different sequences with significant similarity to known genes. We showed that four of these sequences with similarity to rack1 , dusp1 , klf2 and tmsb are specifically up- regulated in induced swords and/or gonopodia. In parallel, we also showed that these genes are strongly expressed during fin regeneration. Therefore these four genes are interesting candidates to further analyse their role in both sword development and fin regeneration.

2.2 Introduction The green swordtail, X. helleri , is a popular fish model to study sexual selection (reviewed in [14]. Male swordtails exhibit a colourful caudal fin extension, named the sword, first introduced by Charles Darwin as a sexually selected trait in fishes [35]. The sword is a complex trait consisting of elongated ventral fin rays and a distinct colour pattern of two black and one yellow-orange or green stripe [40]. Both length and coloration are important for mating success [32, 40]. Interestingly, only males of swordtail species develop a sword. Male platyfish, a monophyletic clade within the genus Xiphophorus , lack a sword, even though females of some platy species show a bias for sworded males [38, 42]. Therefore, the evolutionary history of the sword has been the interest of several studies. Molecular data support the hypothesis that all extant Xiphophoru s species descend from a sworded ancestor and where the sword was secondarily lost in the lineage leading to the platyfish [4, 5, 37]. Even though the sword serves an important role in sexual selection via female choice, a high predation pressure might have driven the loss of the sword in platy species. Predators are attracted by the conspicuous ornamentation of the sword and exposure to predators can reverse the female

bias for swords [27, 66]. The evolutionary changes that caused the loss of the sword, however, are poorly understood on the molecular level. One way to address this question in swordtails is to dissect the genetic network that controls sword development and to characterize the evolutionary modifications to this network within a phylogenetic framework of the entire genus that includes swordtails and platyfish. So far, only little is known about the molecular basis of sword development. Hybridisation experiments between X. helleri and X. cortezi suggest that multiple genes, collectively termed “sword genes” (“Schwertgene”), contibute to sword formation [44]. The induction of sword gene activation is thought to be mediated by elevated testosterone levels during sexual maturation, because hormone treatment experiments with juvenile swordtails revealed that testosterone is a sufficient and essential factor to induce sword development [43, 67]. Exogenous testosterone also induces the male-specific modification of the anal fin into an intromittent organ, the gonopodium [74, 101]. Therefore, one might suggest that sword and gonopodium induction is controlled by androgen signalling. Moreover, Zauner and colleagues hypothesised that the genetic network controlling gonopodium development was co-opted to form the sword in the caudal fin and therefore both processes might be controlled by similar genes [59]. During sword development the main sword rays V8 and V9 have been shown to act as signalling centers that promote growth of neighbouring fin rays, yet the molecular nature of the signal remains to be determined [52]. Candidate gene approaches revealed fgfr1 and its putative upstream target msxC to be specifically up-regulated in growing sword and gonopodial rays [59, 148]. Furthermore, both genes are differently regulated in caudal fins of testosterone-treated swordtails and platyfish, suggesting that changes in regulation of both genes might be involved in sword loss. Only a subset of the genes involved in sword development can be targeted by candidate gene approaches, since candidate genes are chosen based on previous knowledge about their function. Genes that have not been functionally characterized so far are discriminated against by this approach. In addition, not all of the chosen candidate genes are involved in sword development, even though they are involved in similar processes, as shown for fgf24 and fgf20a [148]. To bypass these limitations, alternative techniques are available that do not rely on previous information about gene function.

Microarray experiments and large scale sequencing of cDNAs have uncovered several new genes involved in a related morphogenetic process in a teleost fin, the regeneration of the zebrafish caudal fin upon amputation [76-78]. However, microarrays are only available for a limited number of organisms, and more recent advances, such as large scale sequencing, are costly and require suitable bioinformatical resources for sequence analysis. A more economic technique that can be applied as an alternative to identify genes involved in sword development is suppression subtractive hybridisation (SSH; [79]). This method allows one to detect differences in the abundance of individual transcripts, represented by cDNAs, between two transcriptomes (e.g. those of two different tissues or two different developmental stages). This techniques has been successfully applied for various purposes such as to identify genes up-regulated in tumours, genes that react to specific drug treatment and to identify genes differently regulated in caudal fin regeneration [149-151]. In order to identify genes that are involved in sword and gonopodium development, we constructed a SSH library to detect genes that are differentially expressed in developing swords and gonopodia compared to juvenile fins. X. helleri sequences obtained from the SSH library were compared to known sequences in genebank using blast. Furthermore, the expression pattern of selected genes was analysed in developing swords and gonopodia and regenerating swords. We identified four clones with similarity to rack1 , dusp1 , klf2 and tmsb a-like were differentially expressed in the initial stages of androgen- induced metamorphosis towards swords and gonopodia.

2.3 Results

Construction of a suppression subtractive hybridisation library (SSH) from X. helleri Candidate gene approaches can only target a selected subset of genes that might be involved in sword development [59], because previous knowledge about gene function is needed to select appropriate candidates. In order to identify genes that are differentially expressed between developing swords and gonopodia on the one hand, and juvenile fins on the other hand, without ab initio knowledge of their identity or function, we employed

a suppression subtractive hybridisation (SSH) technique to bypass this limitation. To do so, we prematurely induced swords and gonopodium development with 17 α- Methyltestosterone to allow both for the simultaneous generation of large numbers of experimental animals required for tissue isolation and for timed induction of both processes. Several studies suggest that sword and gonopodium development are induced by increasing testosterone levels in both sexes, and that fin morphology after metamorphosis is very similar in the sexes [43, 44, 74, 86]. It is thus likely that exogenous testosterone treatment will induce the same genes that act during normal sword development independent of the sex of the individual used. To produce the SSH library, we pooled cDNAs from caudal and anal fins that were previously treated with 17 α-Methyltestosterone for 1, 2, 4 or 5 days or from untreated fins. We reasoned that direct targets of testosterone should be activated within the first two days, whereas at 4 and 5 days of treatment genes that are indirectly controlled by testosterone signalling (e.g. growth-promoting genes) should be induced. Developing swords and gonopodia are thought to modify the activity of parts of the same gene regulatory networks [59, 148], thus both tissues were used to increase the starting material. We employed a PCR-based SSH approach [79] to enrich for cDNAs that are more abundant in the cDNA pool derived from testosterone-induced swords and gonopodia. These cDNAs are likely to represent genes specifically up-regulated in developing swords and are putative candidates to function in sword and gonopodium development.

Figure 2.1. Subtractive hybridisation efficiency. A: A fragment of the gapdh transcript was amplified from subtracted and unsubtracted cDNA, derived from testosterone treated fins. PCR was performed for 15, 20 and 25 cycles using species-specific oligoncleotide primers. Amplification of the gapdh fragment is detected five cycles later in the subtracted than in the unsubtracted cDNA, suggesting successful reduction of common transcripts in the subtracted pool. B: In total 406 sequences from the SSH were sequenced and analysed. They were reduced to 201 independent contigs/sequences, of which 128 out of 201 sequences showed similarity to known genes from other species. These 128 positive hits were grouped into five categories according to their functions. *: A contig/sequence is independent, when no significant overlap with other contigs/sequences was found. **: A sequence was considered to be similar to a sequence in the database, if the e-value (Blast) was -15 or lower.

To test the efficiency of the subtractive hybridisation method we used a PCR-based assay. A fragment of the gapdh cDNA was amplified from the pool derived from testosterone-induced fins before and after subtractive hybridisation was performed. We checked for gapdh PCR products after different numbers of PCR cycles. A gapdh PCR product was obtained 5 cycles later after subtractive hybridisation was performed, demonstrating that the amount of gapdh transcript, a ubiquitously expressed gene, was successfully reduced by SSH (Figure 2.1A). The subtracted cDNAs were used to construct a SSH library and 406 clones were chosen for sequencing (Figure 2.1B). The insert size of selected clones was between 100 and 700 bp with an average size of ~400 bp. Further analysis of the sequence data showed that the 406 sequences could be reduced to 201 contigs and independent sequences (Figure 2.1B). A contig or sequence was considered to be independent if it showed no significant overlap with other contigs and sequences. 128 out of 201 (~64%) of the contigs/sequences showed reliable similarity to known genes using blast ([152] and Figure 2.1B). A blast hit was considered to be reliable, if the e-value was -15 or lower. The remaining 73 sequences consisted of repetitive elements (6/8%) vector or artificial sequences (5/7%) and sequences without reliable blast hit (62/85%), likely to be UTR sequences. From the 128 sequences with a significant blast hit 27 (53%) showed reliable similarity to ribosomal or other housekeeping genes, 27 (21%) to components of the cytoskeleton and structural genes, 16 (13%) to transcription factors or genes involved in signal transduction, 5 (4%) to known stress and immune response genes and 12 (9%) to genes with other or unknown function (Figure 2.1B and Table 2.1). Some of these sequences, although independent from each other, showed similarity to the same gene and might just represent different parts of it (Table 2.1).

Table 2. 1: Genes with known functions showing sim ilarity to identified X. helleri sequences

Housekeeping genes 59. Ubiquitin 1. 16S ribosomal RNA (2x) 2. 28S ribosomal RNA Structure & Cytoskeleton 3. Abhydrolase domain containing 12 60. Actin -related protein 3 4. ATP synthase, H+ transporting, mitochondrial F0 61. Alpha-tubulin complex, subunit C3 62. Beta actin (6x) 5. ATPase synthase protein 9 63. Tubulin beta-1 chain 6. ATPase, H+ transporting, lysosomal accessory 64. BsR19 keratin protein 2 65. type 1 Collagen alpha 1 (3x) 7. Cytochrome c oxidase subunit I (2x) 66. type 1 Collagen alpha 2 8. Cytochrome c oxidase subunit II 67. Collagen, type X, alpha 1 9. Deoxyhypusine hydroxylase/monooxygenase 68. Cytokeratin 10. Elongation factor 1 α 69. Keratin 15 (2x) 11. Eukaryotic translation elongation factor 2 70. Keratin 5 protein, transcript variant 1 12. Histone 3B 71. Keratin 5 protein, transcript variant 2 13. Integral membrane protein 2B 72. Keratin K10 14. kaa190 solute carrier family 25 member 5 73. MID1 interacting protein 1 15. mitochondrial ATP synthase H+ transporting 74. Osteonectin complex 1 delta subunit 75. type V/XI Collagen pro -alpha 1 16. 40S Ribosomal protein S2 (3x) 76. type I Keratin isoform 1 17. 40S Ribosomal protein S3a 77. type II keratin 18. 40S Ribosomal protein S5 78. type II keratin E3 19. 40S Ribosomal protein S7 20. 40S Ribosomal protein S8 Stress & Immun response 21. 40S Ribosomal protein S9 79. Armet pro tein 22. 40S Ribosomal protein S11 80. Ferritin heavy chain subunit 23. 40S Ribosomal protein S13 81. B2-microglobulin 24. 40S Ribosomal protein S14, transcript variant 1 82. Peroxiredoxin 4 (2x) 25. 40S Ribosomal protein S16 (2x) 26. 40S Ribosomal protein S17 Signalling & Transcription factors 27. 40S Ribosomal protein S19 83. C-fos 28. 40S Ribosomal protein S20 (2x) 84. 14-3-3.a protein 29. 40S Ribosomal protein S27 85. Calmodulin (2x) 30. 40S Ribosomal protein S27a 86. Cystatin B 31. 60S Ribosomal protein L3 87. Dual specificity phosphatase 1 32. 60S Ribosomal protein L4 (2x) 88. Kruppel-like factor 2a 33. 60S Ribosomal protein L5 89. Kruppel-like factor 2b (2x) 34. 60S Ribosomal protein L6 90. M-Calpain 35. 60S Ribosomal protein L7 91. Protein phosphatase 1, catalytic subunit, beta isoform 36. 60S Ribosomal protein L7a (2x) 92. Receptor for activated protein kinase C (RACK1) (2x) 37. 60S Ribosomal protein L9 93. S100-like 38. 60S Ribosomal protein L10 94. Thymosin beta a-like 39. 60S Ribosomal protein L10a (2x) 40. 60S Ribosomal protein L11 Other 41. 60S Ribosomal protein L13 95. Cysteine-rich protein 2 42. 60S Ribosomal protein L13a 96. DEAD (Asp -Glu -Ala -Asp) box polypeptide 5 43. 60S Ribosomal protein L14 97. Hemoglobin beta-A chain 44. 60S Ribosomal protein L18 98. Human DNA sequence from clone RP5-1107C24 on 45. 60S Ribosomal protein L18a chromosome 20 46. 60S Ribosomal protein L23a 99. Protein LOC553453 47. 60S Ribosomal protein L27 100. Selenoprotein W2a 48. 60S Ribosomal protein L31 101. Sperm plasma glycoprotein 120 49. 60S Ribosomal protein L35a 102. Splicing factor 3b, subunit 1 50. 60S Ribosomal protein L35b 103. Translationally controlled tumor protein-like 51. 60S Ribosomal protein L36 104. Tetraodon nigroviridis full-length cDNA 52. 60S Ribosomal protein L37a 105. Tetraodon nigroviridis full-length cDNA 53. 60S Ribosomal protein L38 106. Tetraodon nigroviridis full-length cDNA 54. Succinate-CoA ligase, alpha subunit 55. Tomm40 56. Transglutaminase 2 57. Translation factor sui1-like 58. Translation initiation factor 4E transporter

SSH enriched genes are expressed in developing swords and gonopodia In trying to better understand the molecular mechanism of sword and also gonopodium development we are strongly interested in transcription factors and signalling pathways that might control this process. Therefore, we choose the clones with similarity to 14.3.3a , c-fos , dual specific phosphatase1 (dusp1 ), receptor for activated protein kinase C (rack1 ), krueppel-like factor 2 (klf2 ), m-calpain and thymosin beta a-like (tmsb a-like ) for further analysis. As an assay independent of the SSH procedure, we confirmed the expression of the selected genes in developing gonopodia and swords by non-quantitative reverse-transcribed (RT)-PCR. We used fin tissue from fish treated with testosterone for 2 and 5 days. 2 days of treatment represented the stage of primary response to testosterone (direct targets), whereas 5 days represented the stage of secondary response to testosterone (indirect targets). All genes are expressed in caudal and anal fins after 2 or 5 days of testosterone treatment (dt) as well as in untreated control fins (Figure 2.2). The RT-PCR experiment confirmed that all genes are indeed expressed in developing swords and gonopodia and are not an artefact created by the method itself.

Figure 2. 2. Expression of SSH -derived genes. Seven genes from the “signalling & transcription factor” category were chosen and their expression was confirmed by RT- PCR. All seven genes are expressed in both testosterone treated caudal and anal fins after 2 and 5 days of treatment, as well as in untreated fins. gapdh was used for the positive control.

SSH enriched genes are specifically up-regulated in sword rays Next we aimed to determine whether these genes are differentially expressed (1) in induced swords compared to control fins and (2) in sword rays after testosterone treatment for a range of time periods. Therefore, we induced sword development in juvenile fish with 17 α-methyltestosterone and harvest the caudal fins for expression

analysis using whole mount in situ hybridisation after 2 or 5 days. Given that the isolated cDNA clones are too short for an appropriate expression analysis by in situ hybridisation, we produced a full length cDNA library from induced swords and gonopodia after 1, 2, 4 and 5 dt. This was then used for rapid and efficient isolation of cDNA fragments of sufficient length. After 2 dt X201 (rack1 ) was expressed at similar levels throughout the whole caudal fin with no obvious differences between sword and non-sword rays (Figure 2.3A). The sword is formed by the ventral fin rays V7-V10 using Dzwillo’s nomenclature for the caudal fin rays of X. helleri [43]. On the contrary X201 was strongly up-regulated in distal tip of the sword forming rays V8-V9 compared to non- sword rays in the median or dorsal caudal fin after 5 dt (Figures 2.3B, B’). X201 expression levels in V7 were similar to non-sword rays and no outgrowth of V7 was yet visible (Figures 2.3B, B’). The expression pattern observed after 2 dt is similar to that of control fins (Figure 2.3C). In contrast, X190 (dusp1 ) expression could not be detected in a distinct pattern in caudal fins after 2 or 5 dt (Figures 2.3D, E and E’) or in untreated fins (Figure 2.3F), even if the fins were stained for an extensive time period or different dusp1 RNA probes were used. X118 (klf2 ) is regulated similarly to X201 (rack1 ). After 2 dt X118 expression could be detected at basal levels in caudal fin rays with no obvious differences between sword and non-sword rays (Figure 2.3G). However, after 5 dt X118 is more strongly expressed in ventral sword rays V7-V10 than in the other fin rays (Figures 2.3H, H’). X118 (klf2 ) expression seems to overlap with that of X201 (rack1 ) (compare Figures 2.3B’ and H’). In control fins, X118 is not differentially expressed between sword rays and the remaining of fin rays (Figure 2.3I).

Figure 2.3. Regulation of SSH-derived genes in developing swords. Transcript levels of X201 (rack1 ) (B, B’), X118 (klf2 ) (H, H’) and X179 (tmsb a-like ) (K, K’), but not X190 (dusp1 ) (E, E’) are higher in developing sword rays after 5 days of testosterone treatment (dt) compared to non-sword rays. In caudal fins after 2 dt and in control fins, expression levels of X201 (A, C), X118 (G, I) and X179 (J, L) are indistinguishable between sword and non-sword rays. The dusp1 transcript cannot be detected by in situ hybridisation in caudal fin rays after 2 dt (D), 5 dt (E, E’) or control fins (F). White arrowheads indicate gene expression. (5 dt: n= 5 for X201 , X118 and X179 ; n=10 for X190 ; 2 dt and control fins: n=4; for every probe; scale bars: 200 µm). V: ventral ray

X179 (tmsb a-like ) transcription did not differ between ventral, median and dorsal caudal fin rays after 2 dt (Figure 2.3J). In contrast, after 5 dt X179 is clearly up-regulated in the sword region and seems to enclose the distal tip of the sword rays (Figures 2.3K, K’). X179 is also expressed in non-sword rays, although at lower levels. In control fins only weak expression of X179 (tmsb a-like ) could be detected in some fin rays (Figure 2.3L). The expression patterns of X201 , X190 , X118 and X179 were confirmed using sense probes (Figure S2.1A-D). X65 (14.3.3a ) showed ubiquitous expression in the whole caudal fin with no obvious differences between sword and non-sword rays. For X169 (c- fos ) and X75 (m-calpain ), no distinct expression pattern could be obtained (data not shown). Our data show that X201 (rack1 ), X118 (klf2 ) and X179 (tmsb a-like ) are up- regulated in sword rays during sword outgrowth at 5 dt.

SSH enriched genes also show gonopodial ray specific regulation Furthermore, we analysed the expression pattern of these genes in the developing gonopodium, another modified fin in males of X. helleri . The gonopodium mainly develops from the anal fin rays 3, 4 and 5, the so-called 3-4-5 complex [82]. X201 (rack1 ) is only up-regulated in a subset of fin rays, the 3-4-5 complex, after 2 dt (Figure 2.4A) and 5 dt (Figure 2.4B) compared to the other anal fin rays (Figures 2.4A, B). Thus X201 (rack1 ) is up-regulated earlier in the gonopodial than in the sword rays. In control fins X201 is expressed at basal levels in all anal fin rays (Figure 2.4C). For X190 (dusp1 ) no expression could be detected in anal fins after 2 dt (Figure 2.4D). After 5 dt however, X190 is strongly expressed in the distal-most part of the 3-4-5 complex, but not in other anal fin rays (Figure 2.4E). Thus differential expression of X190 (dusp1 ) could be detected in growing gonopodia, but not in swords (compare Figures 2.3E and 2.4E). X190 expression was not detected in control fins (Figure 2.4F). The spatio-temporal expression pattern of X118 (klf2 ) in developing gonopodia is comparable to that of swords. After 2 dt the X118 transcript could not be detected in the anal fin (Figure 2.4G), but after 5 dt X118 was exclusively up-regulated in the 3-4-5 complex (Figure 2.4H). No X118 (klf2 ) expression could be detected in control fins (Figure 2.4I).

X179 (tmsb a-like ) showed comparable expression in the 3-4-5 complex and the other anal fin rays after 2 dt (Figure 2.4J). As in induced swords, X179 is up-regulated in the 3- 4-5 complex after 5 dt (Figure 2.4K). In some samples X179 (tmsb a-like ) showed slightly stronger expression in rays 3 and 4 (Figure 2.4K). Control fins expressed X179 at a basal level in all fin rays (Figure 2.4L). The expression patterns of X201 , X190 , X118 and X179 were confirmed using sense probes (Figures S2.1E-H). X65 (14.3.3a ) was also ubiquitously expressed as in the sword, and no distinct expression pattern could be detected for X169 (c-fos ) and X75 (m-calpain ) (data not shown). Our data show that X201 (rack1 ), X190 (dusp1 ), X118 (klf2 ) and X179 (tmsb a-like ) are all up-regulated in 3-4-5 complex during fin ray outgrowth (5 dt). X201 however is also up-regulated in the 3-4-5 complex, before outgrowth starts (2dt).

Figure 2.4. Regulation of SSH-derived genes in developing gonopodia. X201 (rack1 ), X190 (dusp1 ), X118 (klf2 ) and X179 (tmsb a-like ) are up-regulated in the 3-4-5 complex of developing gonopodia. The anal fin rays 3, 4 and 5 show higher levels of X201 (B), X179 (E), X118 (H) and X179 (K) transcripts after 5 days of testosterone treatment (dt) than other anal fin rays. After 2 dt rack1 is also clearly up-regulated in the anal fin rays 3, 4 and 5 (A). X179 (tmsb a-like ) transcription levels might be slightly higher in the 3-4-5 complex after 2 dt (J). X190 (dusp1 ) (D) and X118 (klf2 ) (G) expression is not detectable by in situ hybridisation in anal fins after 2 dt. In control fins X201 (C) and X179 (L) are transcribed at basal levels in all fin rays, whereas X190 (F) and X118 (I) expression can not be detected. White arrowheads indicate gene expression. (5 dt: n= 5 for every probe; 2 dt: n=4; for every probe; control fins: n=3 for every probe scale bars: 200 µm). R: anal fin ray

SSH derived genes are also expressed in regenerating caudal fins The expression data of X201 (rack1 ), X190 (dusp1 ), X118 (klf2 ) and X179 (tmsb a-like ) suggest that these genes are involved in regulation of fin ray growth during sword and/or gonopodium development. This would imply that they are also expressed during fin regeneration, another process in which fin rays exhibits accelerated growth. To test this hypothesis we amputated the caudal fins of male swordtails and allowed them to regenerate for 4 days. Gene expression was then analysed by in situ hybridisation on whole fins and longitudinal sections, which enables the analysis of gene expression on a cellular level. We found that all four genes were also expressed during fin regeneration. X201 (rack1 ) was strongly expressed in the blastema of non-sword (Figure 2.5A) and sword rays (Figure 2.5B). The expression domain of X201 in sword rays however is clearly wider, due to the larger size of the sword ray blastemata. X201 (rack1 ) expressing cells are found in the distal-most mesenchyme and in more proximal cells, such as differentiating scleroblasts (Figure 2.5C). X190 (dusp1 ) is expressed in a cap-like pattern in the distal tip of non-sword (Figure 2.5D) and sword rays (Figure 2.5E). Fin sections revealed the distal-most blastema to express X190 where it is co-expressed with X201 (rack1 ) is expressed (Figure 2.5F). X118 (klf2 ) shows a expression pattern in non-sword (Figure 2.5G) and sword rays (Figure 2.5H) that is comparable to that of X201 (rack1 ). However, X118 is only expressed in proximal mesenchymal cells, likely to be scleroblasts, but not in the distal-most blastema (Figure 2.5I). The last candidate, X179 (tmsb a-like ), is expressed at similar levels in normal (Figure 2.5J) and sword rays (Figure 2.5K). X179 expression is restricted to the basal epidermal layer that covers the mesenchymal blastema (Figure 2.5L). The expression patterns of X201 , X190 , X118 and X179 were confirmed using sense probes (Figures S2.1I-L). Judging from these data, it is apparent that all four genes are expressed in developing as well as regenerating swords and therefore possibly fulfil a similar role in promoting fin rays growth in both sword and gonopodium development and caudal fin regeneration.

Figure 2.5. Expression of SSH-derived genes in regenerating caudal fins. X201 (rack1 ), X190 (dusp1 ), X118 (klf2 ) and X179 (tmsb a-like ) are also expressed in regenerating caudal fins. All genes are expressed at similar levels between non-sword (A, D, G, J) and sword rays (B, E, H, K). X201 expression can be found in the distal-most blastema and more proximally located mesenchymal cells (e.g. scleroblasts; C). X190 is expressed in the distal-most blastema, where it overlaps with X201 (F). klf2 expression overlaps with that of X201 in the proximal sclerosblasts (I). X179 is expressed in the epidermis specifically in the basal epidermal layer (L). White arrowheads indicate gene expression. (n= 4 for every probe, except X118 : n=7; scale bars: A, B, D, E, G, H, J, K: 200 µm, C, F, I, L: 100 µm). dpa: days post amputation

2.4 Discussion The molecular mechanism controlling the development of the sword, a sexually selected trait in the genus Xiphophorus , have been targeted by two previous studies [59, 148]. However, these studies employed a candidate gene approach and focused on well- studied genes. As candidate genes are selected on the basis of prior information about their expression or function in other contexts, genes with unexpected functions or novel genes will not be identified by this approach. In this study we employed suppression

subtractive hybridisation (SSH) to bypass this problem [79]. We successfully identified genes that are differentially expressed in developing swords and gonopodia compared to juvenile fins before metamorphosis.

Genes involved in sword development can be identified by Subtractive Hybridisation In total we identified 201 independent sequences or contigs, of which 128 showed a significant similarity to sequences in public databases. To our knowledge this study provides the largest collection of ESTs (expressed sequence tags) from developing swords and gonopodia currently available. A large fraction of sequences (73) showed no significant similarity to sequences in the database. These sequences could represent UTR (untranslated region) sequences or weakly conserved parts of the coding region. A subset of 15 transcripts seem to be represented by multiple (2-6), independent sequences. This is quite likely for abundant transcripts, given that the average insert length of SSH clones is 400 bp. Housekeeping genes, but also genes encoding for structural components inside and outside the cell make up 75% of our EST pool. These transcripts were also quite abundant in an SSH library of regenerating fins [150]. Theoretically, genes required for cellular maintenance should be eliminated by this method. However, transcripts of housekeeping genes are probably more abundant in growing fin rays, due to a higher demand for energy or protein synthesis in growing tissue and will be only partly removed by SSH. Genes encoding for structural components like Keratins and collagens have been shown to be more strongly expressed in regenerating fins of zebrafish and medaka than in uninjured fins [76-78, 150]. They are likely to participate in the structuring of new fin tissue, since Keratins are the major structural proteins in epithelial tissues and collagens are part of lepidotrichia and actinotrichia [153, 154]. To better understand the genetic network that controls sword and gonopodium development, we mainly focused on genes that codes for transcription factors or are involved in cell signalling. Approximately half of the genes from that category were also found to be expressed in regenerating caudal fins of zebrafish and medaka [77, 78, 150]. This is not surprising, since both sword development and fin regeneration are

characterised by elevated outgrowth of fin rays, which seems to be controlled by a conserved genetic network (reviewed in [68]). However, to our knowledge our study is the first to study the expression pattern of some of these genes in developing swords and gonopodia and also regenerating swords.

SSH candidates are differentially expressed in developing swords and gonopodia and regenerating caudal fins A detailed analysis of the expression pattern of 7 clones from the transcription factor/cell signalling category showed that the genes with similarity to rack1 , dusp1 , klf2 and tmsb a-like are differentially expressed in developing swords and/or gonopodia compared to juvenile fins before testosterone-induced metamorphosis. In addition those genes are expressed in a similar spatial pattern in regenerating fins, which suggest that they fulfil a similar function in both processes. X201 (rack1 ) showed the most widespread expression pattern of all four genes (Figure 2.6) and was also found to differ in the temporal regulation between developing swords and gonopodia. X201 transcription is elevated after 2 dt before the outgrowth of the 3-4-4 complex starts. X201 might therefore fulfil a function in growth induction. The fact that no up-regulation is observed in the sword rays at 2 dt, might just reflect the fact, that testosterone induced gonopodia start to develop earlier than induced swords (personal observation). A detailed analysis of the spatial expression of X201 (rack1 ) in longitudinal sections showed two expression domains, one in newly formed scleroblasts and one in the distal-most mesenchyme (Figure 2.6). The expression of X201 in the scleroblasts might indicate that it is involved in dermal bone formation. rack1 might be linked to the network controlling this process via Fgf or Bmp signalling. There are two lines of evidence that supports a putative interaction between rack1 and Fgf signalling. Firstly, we showed in a previous study that fgf receptor1 is up-regulated in growing swords and gonopodia and regenerating caudal fins of X. helleri . In addition, fgfr1 is co-expressed with X201 (rack1 ) in scleroblast cells. Secondly, rack1 expression is regulated by Fgf signalling in developing limb buds of chick embryos [155]. Fgf signalling both activates Protein kinase C (PkC) and increases rack1 levels to enhance Pkc activity [155], since Rack1 binds and stabilizes activated PkC and recruits the kinase to its targets (reviewed

in [156]). It is likely that, Fgf signalling activates both PkC and rack1 transcription during fin ray growth and regeneration to activate up-stream target genes via the PlC-PkC signalling cascade [105]. However, rack1 might also act on bone formation via the Bmp signalling pathway. Rack1 has been shown to be required for Bmp2 induced Smad phosphorylation via the Bmp receptor 2 [157]. In regenerating caudal fins of zebrafish, bmp2b is expressed in the scleroblast and both knockdown of Bmp signalling and over- expression of bmp2b impairs dermal bone formation [107, 117, 126]. Rack1 might also interact with one of these pathways in the distal-most mesenchyme (DMM). Knockdown of both Fgf and Bmp signalling impairs the expression of genes in the DMM such as msxb , which is essential for blastema cell proliferation [70, 71, 126]. However, functional data is required to proof an interaction between Rack1 and one of these pathways and to dissect the functions of rack1 in the DMM and the scleroblasts. A second SHH-derived gene, X118 (klf2 ), was up-regulated after 5 dt and is likely to function exclusively during fin ray outgrowth. An analysis of X118 expression on the cellular level showed that it is co-expressed with X201 (rack1 ) in scleroblasts (Figure 2.6) and might also function during dermal bone formation. Klf2, is a C2/H2 zinc finger transcription factor that might either activate or repress transcription of target genes in this process (reviewed in [158]). However, to our knowledge a function of klf2 in bone formation or an interaction with rack1 has not been reported so far. Due to the fish- specific genome duplication, two klf2 paralogs have been described in zebrafish [159]. Klfa seems to fulfil the ancestral function of Klf2 in blood vessel development and blood pressure control [159-161]. Therefore, the expression of X118 (klf2 ) in growing or regenerating fin rays might point either towards an undescribed function of Klf2 during appendage regeneration and development or towardsa neofunctionalisation event. An adequate annotation of both paralogs using phylogentic methods, followed by a detailed analysis of the expression and function during zebrafish fin regeneration will be helpful to either support or reject one of these hypotheses. In addition a functional analysis of both klf2 and rack1 will show whether both genes act within the same or parallel pathways. The third gene , X190 (dusp1 ), might function exclusively during gonopodium growth and fin regeneration, since it is up-regulated in 3-4-5 complex after 5 dt and in

regenerating fin rays. Like X118 (klf2 ) it is also co-expressed with X201 (rack1 ), but in the distal-most mesenchyme (DM; Figure 2.6). A study by Kinney and colleagues indicates a putative function of dusp1 (or MAP kinase phosphatase-1) in endothelial cell migration [162]. In endothelial cells (EC), dusp1 is activated by Vegf-A and Vegf-E via the Vegf receptor 2 and knockdown of dusp1 perturbs VEGF-induced EC migration. vegfr2 is expressed in the DM of regenerating caudal fins of zebrafish and inhibition of Vegf signalling showed that angiogenesis is essential for regenerative outgrowth of fin rays [163]. dusp1 might also be involved in regulating the migration of ECs in growing and regenerating fin rays to promote the formation of new blood vessels. As a MAP kinase phosphatase, Dusp1 is likely to regulate the activity of MAP kinases [164, 165] that transmit the Vegf signal [162, 166]. Inhibition of Vegf signalling as performed by Bayliss and colleagues [163] or knockdown of dusp1 in the zebrafish system will be helpful to validate or reject this hypothesis. Furthermore, functional data will useful to show whether X190 (dusp1 ) and X201 (rack1 ) fulfil different roles during ray outgrowth or interact with each other. The fact that we could not detect any X190 (dusp1 ) in developing swords might be due to a technical problem rather than due to the absence of X190 expression, since (1) expression was confirmed by RT-PCR and (2) growing gonopodia and regenerating swords showed up-regulation of X190 (dusp1 ). The spatio-temporal expression of X179 (tmsb a-like ) in developing swords and gonopodia and regenerating fins also argues for a function of X179 during ray outgrowth. In contrast, X179 is not expressed in the mesenchyme, but in the basal epidermal layer (Figure 2.6). Besides its main cellular function as a regulator of actin polymerisation in a subset developing neurons [167], ß-thymosins can promote angiogenesis and accelerate dermal wound healing [168, 169]. Application of Tmsb4 (Thymosin β4) to dermal wounds of rats stimulates migration of endothelial cell [168]. Furthermore, Tmsb4 was shown to promote the formation of new blood vessels in subcutaneously injected matrigel plugs soaked with the protein [170]. In addition exogenous Tmsb4 can also stimulate the migration of keratinocyte and collagen deposition [169]. Furthermore ß-thymosins are involved in cornea repair and in heart and brain regeneration after hypoxia (reviewed in [171]). In teleost this might be all mediated by one ß-thymosin, since only one has been described so far, with the exception of two ß-thymosin genes in rainbow trout [172].

Therefore, Tmsb might assist in the formation of new blood vessels and skeletal elements in growing fin rays. Even though X179 is expressed in the basal epidermal layer (Figure 2.6), ß-thymosins have been shown to be secreted into the extracellular compartment [173], which would enable X179 (tmsb ) to diffuse to its target site.

Figure 2.6: Summary of the expression pattern of X201 , X190 , X118 and X179 The figure summarizes the expression pattern of X201 (rack1 ), X190 (dusp1 ), X118 (klf2 ) and X179 (tmsb a-like ) in regenerating fin rays of X. helleri. X201 expression partly overlaps with that of X190 in the distal mesenchyme and with that of X118 in the newly formed scleroblasts. BL: basal epidermal layer, DM: distal-most mesenchyme, E: epidermis, L: lepidotrichia, M: mesenchyme, Sc: scleroblasts

For two other SHH-derived genes with similarity to c-fos and m-calpain, we failed to show any distinct expression in developing swords and gonopodia or regenerating caudal fins. It is likely that both genes are expressed at such a low level that expression could not be detected by in situ hybridisation , as RT-PCR clearly showed that the two genes are transcribed in developing swords and gonopodia. In summary we isolated a set of four new candidate genes using the SSH technique that showed differential expression in induced swords and gonopodia compared to control fins. Our study was the first that also described the expression of these genes in regenerating caudal fins. It will also be interesting to further dissect the molecular function of these genes and to analyse a putative interaction between co-expressed SSH genes. Since a wider spectrum of molecular methods is available and those genes seems

to be regulated quite similarly in developing swords, gonopodia and regenerating fins, we suggest the zebrafish model for further analysis.

2.5 Experimental Procedures

Fish stocks and maintenance Juvenile and adult green swordtails ( X. helleri ) were taken from stocks kept at the “Tierforschungsanlage” at the University of Konstanz. Fish were maintained on a 12:12h light:dark cycle at 24°C in 110-litre densely planted aquaria and were fed TetraMin flakes and Artemia.

Testosterone treatment and fin regeneration For SSH and λ-phage cDNA libraries, 120 juvenile individuals of X. helleri each, aged between 3 and 6 months, were treated with 17 α-methyltestosterone (1 mg/ml stock solution in ethanol; Sigma-Aldrich, Munich, Germany) that was added to the water twice a week to a final concentration of 10 µg/l. The 120 individuals were divided into 4 groups of 30 individuals each and were treated in 110-litre tanks. After 1, 2, 4 and 5 days of treatment, 1/3 of the caudal and anal fin was harvested from individuals of one group with a sterile razor blade. For fin amputations, fish were anesthetized by incubation in a solution of 0.08 mg/ml tricaine (3-aminobenzoicacid-ethylester-methanesulfonate; Sigma-Aldrich, Munich, Germany). For the SSH library additional 120 individuals were mock-treated with ethanol and fin tissue was amputated as described above. Testosterone and ethanol treated tissue was pooled and used for RNA extraction. For the RT PCR 5 to 8 juvenile fish were treated for 2 days or 5 days with testosterone, or 5 days with ethanol, followed by the amputation of 1/3 of the distal part of the caudal fin and approximately 2/3 of the anal fin. Caudal and anal fin tissue from the 3 treatment groups was pooled and used for RNA extraction. For gene expression analysis up to six juvenile individuals were placed in a 30-litre tank and treated with 17 α-methyltestosterone to a final concentration of 10 µg/l. After 2 or 5 days of testosterone treatment fish were anesthetized and approximately 1/3 of the distal part of the caudal fin and approximately 2/3 of the anal fin was amputated.

For regeneration experiments adult X. helleri individuals were anesthetized and 1/3 of the caudal fin was amputated . Subsequently, fins were allowed to regenerate at 24°C for 4 days. Fish were anesthetized again and the blastema was removed. Fins and blastemata used for in situ hybridisation were fixed in 4% paraformaldehyde in PBS (phosphate buffered saline) overnight, transferred to methanol and stored at -20°C until use.

λλλ-phage cDNA library construction Total RNA was isolated from caudal and anal fin tissue as described [59]. PolyA + RNA was purified using the Qiagen Oligotex mRNA Mini kit (Qiagen, Hilden, Germany). 5 µg of PolyA + RNA was used to construct a λ-phage cDNA library with the ZAP-cDNA® Library Construction Kit (Stratagene, Heidelberg, Germany) were used to construct a X. helleri λ-phage cDNA library according to the manufacturer’s instructions. The amplified library was stored in SM buffer (100 mM NaCl, 8 mM MgSO 4, 50 mM Tris-HCl, pH 7.5) with 5% DMSO at -80°C.

Isolating cDNA from recombinant λλλ-phages 750 µl of the amplified cDNA library was treated with 10 U RNAseA and DNAseI (Fermentas, St. Leon-Rot, Germany) prior to phage particle lysis for 10 min at 37°C. Phage particles were lysed by adding 150 µl STEP buffer (0,4 M EDTA, 50 mM Tris- HCl, pH 8, 1% SDS) and 100 µg Proteinase K (Sigma-Aldrich, Munich, Germany) at 65°C for 30 min. DNA was purified by standard methods [174].

Suppression subtractive library construction The SSH library was constructed using the PCR-Select cDNA subtraction kit (Takara Bio/Clontech, Heidelberg, Germany) subtraction, according to the manufacturer’s instructions. 2 µg PolyA + RNA (purified as described above) from testosterone treated fins were used as tester, and 2 µg PolyA + RNA from ethanol treated fins as driver fractions. The driver pool was subtracted from the tester pool and the subtracted cDNAs were cloned into the pCRII vector using the T/A cloning kit and propagated in E. coli INVaF´ (Invitrogen, Karlsruhe, Germany). Subtractive hybridisation efficiency was

tested by amplification of a gapdh cDNA fragment from both the subtracted and unsubtracted cDNA pool according to the manufacturer’s instructions.

SSH clone selection and sequencing DNA was prepared from selected colonies using established procedures [174]. To ensure that as many independent clones as possible were sequenced, inserts were amplified with nested primers supplied with the PCR-Select cDNA subtraction kit (Takara Bio/Clontech, Heidelberg, Germany) and digested with several restriction enzymes with a 4 base-pair recognition site. PCR fragment length and digestion pattern of all clones was compared to each other. If two or more clones showed an identical pattern, only one of these clones was sequenced. In the end, 406 sequences were selected and sequenced using the M13F/M13R primer set or the supplied nested primer set on an ABI3100 automatic DNA sequencer (Applied Biosystems, Darmstadt, Germany). The sequences were then analysed using contig express (vector NTI 10, Invitrogen). Redundant sequences were eliminated and partly overlapping sequences were grouped into contigs. Independent sequences/contigs, which showed no overlap with other sequences and contig consensus sequences, were identified using BLAST [152]. Sequences and BLAST results are provided on request.

RT PCR To detect expression patterns of selected genes, total RNA was isolated from caudal and anal fin tissue as described [59]. 1 µg of total RNA was transcripted into single- stranded cDNA using the Superscript III reverse transcriptase (Invitrogen, Karlsruhe, Germany). DNA contamination was removed by incubating total RNA with DNAseI (1 U/µl; Fementas, St. Leon-Rot, Germany) for 30 min. cDNA fragments of the selected genes were amplified by PCR using gene specific primers (Table S2.1). Primers were designed from SSH clone sequences using Generunner (Hastings Software Inc.). X. helleri gapdh primers were used for the positive control.

RNA probe synthesis To obtain fragments of SSH clones with sizes appropriate for generating RNA antisense probes, the 3’ ends were amplified from the cDNA library clones using PCR

with gene specific primers (Table S2.1).The PCR products were gel-purified using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany) and cloned into the pCRII- TOPO vector (Invitrogen, Karlsruhe, Germany) for sequencing. Antisense and sense RNA probes were generated using either the digoxigenin or labelling kit (Roche, Mannheim, Germany).

Whole-mount in situ hybridization

In situ hybridisation of Xiphophorus fins and blastemata were performed as described [70] with several modifications. Prehybridisation was done 4 h at 68°C in formamide solution (50% formamide, 5x SSC, 0,1% Tween20, pH to 6 with 1 M citric acid). Post- hybridisation washing steps were initiated at 68°C with formamide solution. To block non-specific binding sites 0,5% blocking reagent (Roche, Mannheim, Germany) in PBT (PBS + 0.1% Tween-20, both from Sigma-Aldrich, Munich, Germany) was used. Antibody incubation was done at 4°C overnight. After fixation of stained fins/blastemata, the tissue was washed twice for 20 min in PBT, 20 min in ethanol/PBT (70:30) and 20 min in 100% ethanol and stored at 4°C. The specificity of anti-sense probes was verified with sense probe hybridisations.

In situ hybridisation on longitudinal sections In situ hybridisation was performed on longitudinal sections of 16 µm thickness from fixed caudal fin blastemata as described [138] with one exception: For pre-hybridisation and hybridisation the same solution was used as for whole mount in situ hybridisation. Sections were created with a Reichert-Jung Autocut 2040 Microtome.

Microscopy and image editing Whole mount fins were analysed using a Zeiss Stemi SV11 Apo. Logitudinal sections were analysed using a Zeiss Axiophot 2. Pictures were taken using the AxioVision software v3.1 (Zeiss) and the digital camera Zeiss AxioCam MRc. Images were processed using Adobe Photoshop 7.0.

Chapter III

Retinoic acid is involved in gonopodium formation in the green swordtail, Xiphophorus helleri

Nils Offen, Nicola Blum , Axel Meyer and Gerrit Begemann (Manusscript)

3.1 Abstract

The gonopodium of male poeciliids is a modified anal fin used for internal fertilization. The gonopodium is formed by the 3-4-5 complex, a subset of elongated anal fin rays that develop specific terminal structures like hooks or claws. Gonopodium development is induced by testosterone and it is thought that low levels of testosterone promote ray outgrowth, wheras high levels induce the formation of terminal structures. Shh, androgen and probably Fgf signalling are involved in gonopodium development. Here we presented first evidence, that also retinoic acid (RA) signalling is likely involved in gonopodium development. We showed that aldh1a2 , a RA synthesising enzyme, and two RA receptors , rar γ-a and rar γ-b, are expressed in developing gonopodia. Furthermore, we found that inhibition of RA signalling increased the length of the ray segments added to the 3-4-5 complex, whereas overactivation of RA signalling led to a reduction in segment length. Finally, we showed that androgen receptors β (ar β), a putative regulator upstream of RA signalling is co-expressed with aldh1a2 in gononopodial rays. Interestingly, both genes are expressed in the distal tip of the gononopodial but not of the sword rays, whereas both rar γs are similarly expressed in developing swords and gonopodia.

3.2 Introduction Most of all extant fish species (>95%) belong to the group of ray finned fish [175]. With more than 23.000 species this group represents approximately 50% of all extant vertebrate species. Most of all extant ray-finned fish species (98%) exhibit an oviparous mode of reproduction, but in at least 54 fish species viviparous reproduction can be found [176]. One of those families are the Poeciliid fish (Fam: poeciliidae), which consist of the three subfamilies Aplocheilichthyinae, Procatopodinae and the Poeciliinae [1]. The Poeciliinae is one of three groups within the toothed carps (suborder Cyprinodontoidei) that are thought to have evolved internal fertilization and a specialized intromittant organ independently from each other [61]. The poeciliid intromittant organ is called gonopodium and mainly develops from the anal fin rays 3-5, the so-called 3-4-5 complex, during sexual maturation [2, 82]. These rays are modified in terms of ray length, segment

thickness and different distal structures like blades, claws, spines, hooks and serraes [85]. The morphology of the gonopodium, in particular the morphology of the terminal structures, greatly differs between species and was extensively used for morphology- based phylogenetic analysis [84]. In their 1951 paper, Gordon and Rosen studied this species-specific variability in gonopodium morphology of Xiphophorus species, such as the green swordtail (Figure 3.1A), by comparing the three gonopodial rays and their distal structures between several species [85]. Ray 3 is not bifurcated, whereas Ray 4 and 5 bifurcate into two sister rays, an anterior (a) and a posterior (p) one (Figure 3.1B). Serraes or spines are formed by the rays 4p and 3 and their number is quite variable even within species. In addition to spines, ray 3 exhibits a terminal hook. Ray 5a carries a terminal claw that is absent in some species. Furthermore, a terminal blade with a species-specific shape develops between the rays 3 and 4a. The growth and segmentation rate can differ between the gonopodial rays, which results in segments of different length during gonopodium growth [84]. For Gambusia affinis , another poeciliid species, it has been shown that these differences in growth and segmentation rate can vary with the age of the fish and the number of segments that are already present [80]. In general, the final length of the gonopodium depends on the length of the fish [84]. The genetic network underlying gonopodium development is poorly understood. In 1941, Turner postulated a two step model for gonopodium development [75, 80]. First, when the testis starts to develop low levels of testosterone are released and promote outgrowth of the gonopodial rays. Next, when the testis develops further and releases higher amounts of testosterone, local differentiation areas arise within the growing gonopodium. These areas appear at a specific location and in a specific temporal sequence during gonopodium development and persist until this process is finished. Therefore the shape of these areas and their expansion along the proximo-distal axis is quite variable. Each area adds new bony segments with a specific shape to one or more fin rays. Area II, for example, adds spines to a couple of ray segments in ray 3. Gonopodium induction experiments in Gambusia affinis and X. maculatus supports the role of testosterone as a key factor in gonopodium development, since exogenous testosterone can artificially induce gonopodium development in juvenile fish [74, 75]. The responsiveness to testosterone somehow decreases when juveniles mature into

females. The anal fin of adult females gets only slightly masculinised and develops just a few gonopodial-like structures [74, 75]. Interestingly, if testosterone is applied during regeneration of the female anal fin, a gonopodium-like structure is regenerated that almost resembles a natural gonopodium. However, one important difference between naturally developing and artificially induced gonpodia was found in these experiments. Induced gonopodia basically form all terminal structures but are shorter (Figure 3.1C), because the high levels of testosterone are thought to induce phase one and two about the same time [74, 75].

Figure 3.1. The male gonopodium. Male swordtails modify their anal fin into a gonopodium (A). The male gonopodium is mainly formed by the rays 4-5 and exhibit various specialised structures such as spines, hooks and blades on the distal end (B). In induced gonopodia (24 days of testosterone treatment), these structures develop earlier, resulting in shorter gonopodia (C). Balck arrows indicate present or forming terminal structures. (scale bars: B, C: 500 µm; C: claw, R: ramus, S: spine, TH: terminal hook)

Gene expression data further support a role of androgen signalling in gonopodium development. Ogino and colleagues showed that both androgen receptor α and β are expressed in the developing gonopodium of Gambusia affinis [86]. In addition to androgen signalling, other genes and pathways are likely to be involved. Both shh and its

receptor ptc1 are expressed during gonopodium development of G. affinis and inhibition of Shh signalling blocks gonopodium development [86]. In addition, fgfr1 and msxC are up-regulated in growing gonopodial and are thought to promote gonopodium outgrowth [59]. Experiments by Pickford and Atz also suggest a putative role of thyroid signalling. Treatment of juvenile fish with thyroid hormone resulted in anal fin ray growth [177]. Retinoic acid (RA) is an important signal molecule in embryonic development, since it is involved in many key events during embryonic development, such as somitogenesis, left-right asymmetry formation, heart development and neurogenesis [178-180]. RA, a small lipophilic, diffusible molecule, is synthesised by a group of retinaldehyde dehydrogenases (Aldh1as) and stimulates gene expression through binding to two types of receptors, retinoic acid receptors (RARs) and retinoic X receptors (RXRs) [93]. Furthermore, RA also plays a crucial role in the formation of paired appendages, since it is essential for forelimb bud initiation in either mouse and zebrafish [87-89]. Additionally, it is involved in proximo-distal patterning of skeletal elements in later stages of limb development [89-91]. Interestingly, the mechanism by which RA determine the proximo-distal axis of the developing limb is reactivated during limb regeneration [92]. Since RA signaling is crucial for the development of paired fins [87, 88], we wanted to test if RA also plays a role in the development of the gonopodium, which is an unpaired fin. We therefore cloned aldh1a2 and two rar γ receptors from growing gonopodia and examine their expression pattern during gonopodium development. Furthermore, we show that segment length can be altered by either blocking RA signalling with DEAB or over-activating it by using all-trans retinoic acid. We also found that aldh1a2 is co- expressed with androgen receptor β and both genes are differently regulated in developing swords.

3.3 Materials and Methods

Fish stocks and maintenance Xiphophorus helleri were taken from stocks kept at the “Tierforschungsanlage” at the University of Konstanz. Fish were maintained on a 12:12h light:dark cycle at 24°C in 110-litre densely planted aquaria and were fed TetraMin flakes and Artemia.

RA and DEAB treatment Up to six juvenile individuals of X. helleri , aged between 3 and 6 months, were placed in a 30-litre tank and treated as follows. DEAB experiment groups were treated with 5 µg/l 17-α-Methyltestosterone (1 mg/ml stock solution in ethanol; Sigma-Aldrich, Munich, Germany) and 5 µM DEAB (500 mM stock 4-diethylaminobenzaldehyde in ethanol; Sigma-Aldrich, Munich, Germany). DEAB dramatically decreases the survival rate of experimental animals (~39%) within the first 7 days. Therefore, the experiment duration was limited to seven days because of a limited amount of juvenile fish. RA experiment groups were treated with 5 µg/l 17-α-Methyltestosterone and 10 -8 M atRA (10 -2 M stock all-trans retinoic acid in ethanol; Sigma-Aldrich, Munich, Germany). Control groups were treated with 5 µg/l 17-α-Methyltestosterone only. The testosterone treatment was repeated every forth day, DEAB treatment was done twice a day and RA treatment was repeated every 24 hours at the beginning of the dark cycle. For analysis of morphological changes in early gonopodium development fish were anesthetized by incubation in a solution of 0.08 mg/ml tricaine (3-aminobenzoicacid- ethylester-methanesulfonate; Sigma, Munich, Germany) and anal fins were photographed. Photographs were taken before the treatment was started and after 7 days of treatment. For in situ hybridisation juvenile fish were either treated with 5 µg/l 17-α- Methyltestosterone or EtOH (control) for a variable number of days. The testosterone or EtOH treatment was repeated every forth day. At the end of the treatment fish were anesthetized and 2/3 of the anal and 1/3 of the caudal fin were amputated using a sterile razor blade. The fins were fixed in 4% paraformaldehyde in PBS (phosphate buffered

saline, Sigma, Munich, Germany) overnight, transferred to methanol and stored at -20°C until use.

Isolation of DNA from recombinant λλλ-phages 750 µl of the amplified cDNA library (see Chapter II) was treated with 10U RNAseA and DNAseI (Fermentas, St. Leon-Rot, Germany) prior to phage particle lysis for 10 min at 37°C. Phage particles were lysed by adding 150 µl STEP buffer (0.4 M EDTA, 50 mM Tris-HCL, pH 8.0, 1% SDS) and 100 µg Proteinase K (Sigma-Aldrich, Munich, Germany) at 65°C for 30 min. DNA was purified by standard methods [174].

Cloning aldh1a2 , rar-γγγa, rar-γγγb and androgen receptors

cDNA fragments of raldh2 , rar −γ a and rar −γ b were isolated from recombinant phage DNA, derived from the X. helleri λ-phage cDNA library (see Chapter II), by PCR using degenerate Primers. A 767 bp aldh1a2 fragment was amplified by PCR using the Primers raldh2-fw1: 5’-GGI TAY GCI GAY AAR ATH CAY GG-3’and raldh2-rev1: 5’-ACR TTI GAR AAI ACI GTI GGY TC-3’. A 602 bp rar −γ a and a 603 bp rar −γ b fragment were amplified by PCR using the Primers RAR-fw2: 5’-TGY GAR GGI TGY AAR GGI TT-3’ and RAR-rev2: 5’-GGI CCR AAI CCI GCR TTR TG-3’. To obtain appropriate size rar −γ a/rar −γ b fragments for RNA probe generation, the 3’ ends of the cDNAs were amplified from the cDNA library using PCR with the primer pairs RAR1-fw1: 5'-GGAGAGCTTGAAGAACTGGTC-3'/ lib-univ: 5'- CACTATAGGGCGAATTGGCTACCG-3' for rar −γ a and RAR2-fw1: 5'-GAA CTG GAG GAG CTT GTG AAC-3'/ lib-univ for rar −γ b. For rar −γ a, a ~1,3 kb and for rar −γ b a ~1,5 kb fragment was amplified. The PCR products were gel-purified using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany) and cloned into the pCRII- TOPO vector (Invitrogen, Karlsruhe, Germany) for sequencing. To obtain a fragment of both androgen receptors to generate an RNA probe giving a reliable signal, the phage λ-phage cDNA library was screened with DIG labelled RNA 6 probe derived from X. helleri ar α and ar β cDNA fragments (Offen, unpublished). 10 recombinant phages were grown, transferred to nitrocellulose membranes (Nitropure 45 µm, Osmonics, Minnetonka, USA) and prepared for screening according to the ZAP-

cDNA® Library Construction Kit manual (Stratagene, Heidelberg, Germany). The membranes were treated with Proteinase K (2 mg/ml) in PBS for 30 min at 37 °C, washed with ddH 20 and prehybridised in hybridisation buffer (50% Formamide, 5x dehardt solution, 5x SSC, 0,1% SSC, 250 µg/ml sheared herring sperm DNA) for 1 h at 50°C. RNA probe was added and allowed to hybridise to the complementary cDNA for more than 16h at 50°C. Afterwards, membranes were washed five times for 10 min in 2x SSC with 0,1% SDS, two times at RT and three times at 42°C. After blocking unreacted binding sites on the membrane with 1% blocking agent (Roche, Mannheim, Germany) in maleic acid buffer (100 mM maleic acid, 150 mM NaCl) for 1 h, immunolabelling of hybridised probe was performed using a alkaline phosphatase coupled DIG antibody (1:2000 in maleic acid buffer; Roche, Mannheim, Germany) for 2h. After washing several times in maleic acid buffer, the antibody detection was performed as described for in situ hybridisation [70]. The pBluescript phagemid containing the cDNA insert was excised from the λ-phage genome as described in the ZAP-cDNA® Library Construction Kit manual (Stratagene, Heidelberg, Germany).

RNA probe synthesis and whole-mount in situ hybridisation Antisense and sense RNA probes were generated using a digoxigenin labelling kit (Roche, Mannheim, Germany). Probes for aldh1a2 , rar −γ a and rar −γ b, ar α and ar β were generated from the cDNA fragments listed above. In situ hybridisation on Xiphophorus fins were performed as described [70] with several modifications. Prehybridisation was done 4h at 68°C in formamide solution (50% formamide, 5x SSC, 0.1% Tween20, pH to 6 with 1 M citric acid). Post-hybridisation washing steps were initiated at 68°C with formamide solution. To block unspecific binding sites 0.5% blocking reagent (Roche, Mannheim, Germany) in PBT was used. Antibody incubation was done at 4°C overnight. After fixation of stained fins/blastemata, the tissue was washed twice 20 min in PBT, 20 min in ethanol/PBT (70:30) and 20 min in 100% ethanol and stored at 4°C. The specificity of anti-sense probes was verified with sense probe experiments.

In situ hybridisation on longitudinal sections Anal fins from individuals treated with 17 α-Methyltestosterone for 7 days were fixed in 4% Paraformaldehyde (Sigma). Longitudinal sections of 10 µm thickness were created using a Reichert-Jung Autocut 2040 Microtome and in situ hybridisation was performed as described [138] with one exception. The same hybridisation buffer was used as for whole mount in situ hybridisation.

Microscopy and image editing Fin explants and anal fins were analysed using a Zeiss Stemi SV11 Apo. Pictures were taken using the AxioVision software v3.1 (Zeiss) and the digital camera Zeiss AxioCam MRc. The pictures were processed using Adobe Photoshop.

Segment measurement Pictures from anal fins of juvenile X. helleri were taken after 0 and 7 days of DEAB, RA or control treatment. The pictures were blinded and the number of segments that were present before (old) and after the treatment (new) was counted for the rays 3, 4a, 4p, 5a and 5p. In addition, the length of all segments that are formed during the treatment was measured using the software ImageJ [139]. The length of the last four segments present before the treatment was started was also measured. Whereas the boundaries between the segments are clearly visible, there is often no sharp boundary between one or both lateral edges of the segment and the adjacent tissue. This might be due to the curved structure of the hemirays. Therefore measurement of the segment’s width could not be done accurately for all segments. The width of the segments was therefore not included in the analysis.

Analysis of segment measurement The average number and length of new segments per ray was calculated for one treatment group and graphically presented using Microsoft Excel. For the average length calculation only the segments present in most of the fish of a treatment group were used (ray 3: new segment 2-5; ray 4a: new segment 2-5; ray 4p: new segment 2-5; ray5a: new segment 2-3). The first new segment was excluded, because its formation might have

already begun before the treatment was started. Ray 5a produces on average only 2 segments during the experiment and was therefore also excluded. In addition the average length of the old segments was also calculated. The dataset was checked for normal distribution using graphical methods (normality plot) and statistical tests (Shapiro-Wilk, Anderson-Darling). To test if segment length differs significantly between the treatment groups, a one tailed t-test was used. The same comparison was done for the average segment number using a Mann-Whitney test. To test if the length of the new segments depends on the length or number of old segments, a correlation analysis (Pearson or Spearman correlation) was performed.

Phylogenetic analysis cDNA sequences of retinoic acid receptors , aldh1a enzymes and androgen receptors were sampled from GenBank and Ensembl using the Blast algorithm [152] and aligned using ClustalW [181]. For aldh1as the full cDNA alignment (excluding the third position) was used for the phylogenetic analysis. For the retinoic acid and androgen receptors a cDNA fragment coding for the C4 zinc finger and the hormone binding domain was used to build the tree. Sequences that could not be aligned with confidence were excluded from the analysis. Based on the alignments, phylogenetic trees were constructed using maximum likelihood (ML) and Bayesian methods of phylogeny inference [140]. ML analyses were performed using PHYML 2.4 [141]. The best fitting models of sequence evolution for ML were obtained by ModelTest 3.7 [142]. For retinoic acid receptors the Tamura-Nei model TrN+I+G (alpha =0.7916, pinv =0.4111; [146], for aldh1a enzymes the general time reversible model GTR+I+G (alpha = 1.1961, pinv = 0.2136; [145] and for androgen receptors the Hasegawa-Kishino-Yano model HKY+G (alpha =0.5019, TRatio=1.1257; [182] was used. ML tree topologies were evaluated by a bootstrap analysis with 500 replicates [143]. To confirm obtained tree topologies Bayesian analyses were initiated with random seed trees and were run for 1,000,000 generations. The Markov chains were sampled at intervals of 100 generations with a burn in of 1000 generations. Bayesian phylogenetic analyses were conducted with MrBayes 3.0b4 [144].

The following sequences were used for the phylogentic analysis: aldh1a1: Homo sapiens (NM_000689), Mus musculus (NM_013467), Gallus gallus (NM_204577), Xenopus laevis (NM_001087772) aldh1a2: Homo sapiens (NM_003888), Mus musculus (NM_009022), Gallus gallus (NM_204995), Xenopus laevis (NM_001090776), Danio rerio (AF315691), Gasterosteus aculeatus (ENSGACT00000020927), Oryzias latipes (ENSORLT00000010445), Takifugu rubripes (NM_001033639), Tetraodon nigroviridis (CAAE01013867) aldh1a3: Homo sapiens (NM_000693), Mus musculus (NM_053080), Gallus gallus (NM_204669), Xenopus laevis (NM_001095605), Danio rerio (DQ300198), Gasterosteus aculeatus (ENSGACT00000018580), Tetraodon nigroviridis (GSTENT00012805001), Takifugu rubripes (NEWSINFRUT00000155714) aldh1a1/2/3 : Ciona intestinalis a (ENSCINT00000016285), Ciona intestinalis b (ENSCINT00000016054), Ciona intestinalis c (ENSCINT00000016069), Ciona intestinalis d (ci0100136702) rar α: Homo sapiens (NM_000964), Mus musculus (NM_009024), Gallus gallus (X73972), Notophthalmus viridescens (X17585 ) rar αa: Danio rerio (NM_131406), Tetraodon nigroviridis (GSTENT00024106001), Takifugu rubripes (GENSCAN00000028342) rar αb: Danio rerio (NM_131399), Gasterosteus aculeatus (ENSGACT00000007038), Takifugu rubripes (GENSCAN00000013561), Tetraodon nigroviridis (GWSHT00007447001) rar β: Homo sapiens (NM_000965), Mus musculus (NM_011243), Gallus gallus (NM_205326), Notophthalmus viridescens (AY847515) rar γ: Homo sapiens (NM_000966), Mus musculus (NM_011244), Mesocricetus auratus (AY046945) rar γa: Danio rerio (S74156), Takifugu rubripes (GENSCAN00000021740), Tetraodon nigroviridis (GSTENT00028047001), Gasterosteus aculeatus (ENSGACT00000012380) rar γb: Danio rerio (NM_001083310), Gasterosteus aculeatus (ENSGACT00000000789), Takifugu rubripes (GENSCAN00000014750)

ar: Homo sapiens (NM_000044), Mus musculus (NM_013476), Gallus gallus (NM_001040090), Xenopus laevis (NM_001090884) arβ: Gasterosteus aculeatus (AY247207), (AY247206), Oryzias latipes (NM_001122911), Tetraodon nigroviridis (CAAE01014703), Takifugu rubripes (GENSCAN00000027349), Oreochromis niloticus (AB045212), Gambusia affinis (AB182329) arα: Gasterosteus aculeatus (GENSCAN00000022206), Oryzias latipes (NM_001104681), Tetraodon nigroviridis (CAAE01014998), Takifugu rubripes (GENSCAN00000026438), Oreochromis niloticus (AB045211), Gambusia affinis (AB174849) pgr : Homo sapiens (NM_000926), Mus musculus (NM_008829), Gallus gallus (NM_205262)

3.4 Results

Isolation of aldh1a2, rar-γγγa, rar-γγγb and two androgen receptors from the green swordtail Retinoic acid (RA) is a key signalling molecule in several developmental pathways, including fin and limb development, where it is essential for initiation and outgrowth of the fin/limb bud [87-89]. We therefore reasoned that RA might also be involved in the development of the gonopodium. To test this we first screened a X. helleri cDNA library, constructed from developing swords and gonopodial tissue, for orthologs of RA synthesizing enzymes ( aldh1as ) and retinoic acid receptors (rars ). We successfully cloned cDNA fragments of one aldh1a enzyme and two rars. The amplified 721 bp fragment of the aldh1a2 ortholog (FJ372848) codes for a 240 aa fragment of the protein. Phylogenetic reconstruction of aldh1a enzymes, using coding sequences, confirmed that we cloned a partial sequence of the X. helleri aldh1a2 ortholog (Figure 3.2A). In addition, we cloned four cDNA fragments that cover part of the open reading frame and the complete 3’UTR sequence of two rar-γ orthologs. Phylogenetic reconstruction of retinoic acid receptors , using coding sequence, confirmed that we cloned a partial sequence of X. helleri rar-γa (FJ372849) and rar-γb (FJ372850), respectively (Figure

3.2B). The cloned cDNAs of X. helleri rar-γa and rar-γb code for parts of the protein sequence including most of the C4 zinc finger DNA binding domain and the complete nuclear hormone receptor ligand-binding domain. Gonopodium development in Xiphophorus fish is controlled by androgen signalling and can be artificially induced by exogenous testosterone [74]. In addition, androgen receptors have been shown to be expressed in the developing gonopodium of Gambusia affinis [86]. Therefore we isolated two androgen receptor cDNAs from the λ-phage library by filter screening. A 2418 bp cDNA clone codes for 596 aa of the Androgen receptor α, including the C4 zinc finger DNA binding domain and the complete nuclear hormone receptor ligand-binding domain and the 3’UTR. A second 3867 bp cDNA clone was identified as androgen receptor β. The cDNA clone covers the complete coding region and 3’UTR and parts of the 5’UTR sequence. Pfam detects the same two conserved domains in the 756 aa protein sequence as in the first clone. Phylogenetic reconstruction using the amino acid sequence translations from cDNA fragments coding for the zinc finger and the ligand binding domains confirmed the cDNAs as the androgen receptor α (FJ372851) and androgen receptor β (FJ372852) orthologs, respectively, of X. helleri (Figure 3.3).

Figure 3.2. Phylogenetic reconstruction of aldh1a enzymes and rars . Phylogenetic analysis of vertebrate aldh1a enzymes (A) and retinoic acid receptors using PhyML (upper values) and Mr. Bayes (lower values). For analysis the coding regions of aldh1a enzymes and rars cDNAs were used. The position of the X. helleri orthologs of aldh1a2 , rar-γa and rar-γb within the two phylogenies is highlighted (grey box).

Figure 3.3. Phylogenetic reconstruction of androgen receptors . Phylogenetic analysis of vertebrate androgen receptors using PhyML (upper values) and Mr. Bayes (lower values). For analysis the coding regions of androgen receptor (ra) cDNAs were used. The position of the X. helleri ar α and ar β orthologs within the phylogeny is highlighted (grey box).

Components of the RA signalling pathway are expressed in developing gonopodia As a first step towards uncovering whether RA signalling plays a role in gonopodium development, we analysed the expression of aldh1a2 and the two rar-γ paralogs. To obtain the required amount of gonopodia in comparable stages of development, we artificially induced gonopodium development in juvenile fish by treating them with 17 α- methyltestosterone. To obtain a sufficient amount of naturally developing gonopodia in the same developmental stage is not feasible, because male maturation can start between 3 month and more than a year [183]. The induced gonopodia are shorter than normal gonopodia (compare Figures 3.1A and 3.1C), because terminalisation and formation of terminal structures is induced earlier due to higher testosterone levels [74, 75]. Therefore, in a strict sense, we test for aldh1a2 , rar-γa and rar-γb expression during terminal structure formation. After 5 days of treatment (dt) aldh1a2 (Figure 3.4A), rar-γa (Figure 3.4B) and rar-γb (Figure 3.4C) are up-regulated in the 3-4-5 complex that will give rise to most of the gonopodial structures. In addition, the two rars (Figure 3.4B, C) and aldh1a2 are also

expressed in rays 6 and 7 that flank the 3-4-5 complex (compare Figures 3.4A and 3.4B, C). aldh1a2 expression seems to be even stronger in ray 5 and 6 than in the 3-4-5 complex. At 7 dt, when the elongation of the 3-4-5 complex is clearly visible, aldh1a2 is expressed in a continuous distal stripe that encloses the distal tip of the 3-4-5 complex and several other gonopodial rays (Figure 3.4D). The expression pattern of the two receptors at 7 dt appears to be unchanged compared to 5 dt (Figures 3.4E, F). In situ hybridisation on fin sections revealed aldh1a2 to be expressed in the distal-most mesenchyme of the gonopodial rays (Figure 3.4G), whereas the rar-γs are expressed in the distal-most and more proximal located mesenchymal cells (Figures 3.4H, I). aldh1a2 remains up-regulated even at later stages of gonopodium development when the first terminal structures, the spines, start to form (Figures 3.4J and 3.1C). rar-γa and rar-γb are slightly down-regulated compared to 7 dt (Figures 3.4K, L). Untreated control fins do not express aldh1a2 (Figure 3.4M), whereas the two rar-γs are expressed at a basal level (Figures 3.4N, O). The observed changes in gene expression profiles during gonopodial development suggest that transcriptional activation of aldh1a2 and the two rar-γ paralogs correlates with the transformation of the male anal fin into an intromittent organ.

Figure 3.4. Expression of aldh1a2 , rar-γa and rar-γb during gonopodium development. aldh1a2 and both rar γ paralogs are expressed in developing gonopodia of X. helleri . At 5 (A-C), 7 (D-F) and 18 (J-L) days of testosterone treatment (dt). aldh1a2 is expressed in row of cells in the distal tip of the main gonopodial rays and also ray 6 and 7 (A, D, J), more precisely aldh1a2 expression is located in the distal-most mesenchyme (G). aldh1a2 expression was absent in the control fins (M). rar γa and rar γb are

expressed at similar pattern. At 5 (B, C) and 7dt (E, F) strong expression of both genes could be detected in the gonopodial rays 3-5 and also partly in ray 6 and 7. At 18 dt (K, L) both genes seemed to be slightly down-regulated. The expression domain of both genes covers the distal-most and more proximal mesenchymal cells (H, I). No up-regulation of both receptors was detected in the control fins (N, O). White arrowheads indicate gene expression. (n= 6 for 5 and 7dt, n= 4 for 18 dt and n= 5 for controls for every probe; scale bars: A-F and J-O: 200 µm; G-I: 100 µm) aldh1a2 , but not rar-γγγa and rar-γγγb, is differently regulated in developing swords A subset of fish in the genus Xiphophorus , the swordtails, develop an additional, male- specific modification of the caudal fin, the sword [42]. The sword is formed by the ventral-most fin rays of the caudal fin which elongate and become brightly ornamented in mature males [44]. The sword can also be artificially induced by testosterone treatment of juvenile fish [43, 67]. Furthermore, it has been proposed by Zauner and colleagues that the developmental program shaping the gonopodium was partly co-opted to develop a sword [59]. We therefore tested whether components of RA signalling are also expressed in the growing sword of testosterone-treated juvenile X. helleri . Compared to the anal fin after 5 days of testosterone treatment (dt), aldh1a2 expression could not be detected in the distal tip of either sword nor non-sword rays (Figure 3.5A). In contrast, rar-γa (Figure 3.5B) and rar-γb (Figure 3.5C) are clearly up-regulated in the distal tip of sword-forming rays compared to more dorsally located rays. At 7 dt, when the sword had started to grow out, aldh1a2 seemed to be expressed in cells of unknown type scattered randomly all over the fin (Figure 3.5D). The amount of stained cells varied strongly between individual samples, ranging from only a few to many stained cells. Fin samples hybridised to a sense probe did not show any comparable pattern (data not shown), indicating that those cells indeed express aldh1a2 . The expression pattern of the two rar γ paralogs after 7 dt is similar to that of 5 dt (Figures 3.5E, F). At later stages of sword development (18 dt) rar-γa and rar-γb are slightly down-regulated as observed in developing gonopodia (data not shown). In control fins neither aldh1a2 (Figure 3.5G), nor rar-γa (Figure 3.5H) and rar-γb (Figure 3.5I) were up-regulated in any caudal fin ray. In summary, the localized expression of aldh1a2 in the gonopodium suggests that RA signalling might play an important role in gonopodial development. In contrast,

expression in the sword does not follow the same pattern, but is rather scattered. The two RA receptors however are regulated similarly in developing gonopodia and swords.

Figure 3.5. Expression of aldh1a2 , rar-γa and rar-γb in the developing sword aldh1a2 is differently expressed in the developing sword compared to the gonopodium. At 5 days of testosterone treatment (dt), when the sword has started to grow, no aldh1a2 can be detected in the sword rays (A), which is similar to the situation in control fins (G). At 7 dt aldh1a2 is expressed in several cells scattered across the whole fin (D). The two rar γ paralogs are similarly expressed as in developing gonopodia. At 5 (B, C) and 7 dt (E, F) both genes are up-regulated in the growing sword rays compared to the controls (H, I). White arrowheads indicate gene expression. (n= 6 for 5 and 7dt, n= 4 for 18 dt and n= 5 for controls for every probe; scale bars: 200 µm)

RA signalling is involved in patterning of the gonopodium

The expression of aldh1a2 and the two rar-γ paralogs let us assume that both a source of retinoic acid (RA) and receptors to convert the signal into gene expression are present in developing gonopodia, while developing swords appear to lack a localized source of RA. We therefore tested for a putative function of RA signalling in gonopodium development. We treated juvenile fish with 17 α-methyltestosterone to simultaneously

induce gonopodium development in a number of experimental animals. At the same time we disrupted RA synthesis using the Aldh1a2 inhibitor DEAB. We noticed that DEAB treatment was toxic and only 39% of all animals survived. A subset of the surviving individuals adds only one to two additional segments to their anal fin rays 3 and 4. Anal fins of these juveniles are comparable to those of testosterone-treated adult females that do not develop a fully grown gonopodium, but only add a few gonopodium-like ray elements [74, 75], and were therefore ignored for further analysis. Superficially, DEAB did not affect gonopodium morphology (Figure 3.6A) compared to the control group (Figure 3.6B). We also performed the reverse experiment and artificially enhanced RA signalling by applying exogenous retinoic acid (all-trans RA) to developing gonopodia. All-trans RA (atRA) treatment did not have any influence on fish survival. Similar as in the DEAB treatment, not all individuals induce gonopodium development and were excluded from the analysis. At concentrations of 10-8 M all-trans RA (atRA), no obvious defects in gonopodium morphology could be seen (Figure 3.6C), whereas at concentration higher than 10 -7, the gonopodial rays started to grow but patterning of the segments was massively affected (Figure 3.6D). Especially segment boundary formation seemed to be impaired. This might imply that RA signalling might be involved in patterning of the segments. To test this we analysed the morphology of segments in more detail that were added during experimental treatments. The newly added fin ray segments of DEAB treated animals were on average longer than the corresponding segments in the fin rays of control animals (Figure 3.6E). This trend was found for all analysed anal fin rays. However, statistical analysis of the data showed that this trend is only significant for the rays 4a and 4p (one tail t-test, p<0.05). If induced gonopodia were treated with 10 -8 M atRA we found that segment size in all rays analysed was reduced compared to the control group (Figure 3.6E). This trend was significant for the rays 3, 4a and 5a (one tail t-test, p<0.05). Neither DEAB nor atRA treatment had a significant influence (Mann- Whitney test, p<0.05) on the number of segments added to any of the fin rays compared to the control fins (Figure 3.6F).

Figure 3.6. DEAB and atRA treatment influences segment patterning Induced gonopodia seem to develop normally when treated with either 5 µM DEAB (A) or 10 -8 M atRA (C) compared to controls (B). If induced gonopodia are treated with >10 -7 M atRA segmental patterning is affected (D). Further analysis showed that the average segment length (E), but not the average number of new segments (F) is altered by DEAB or atRA treatment. (n= 11 for DEAB and control and n=8 for atRA treatment group; scale bars: 200 µm; *p<0.03, **p<0.01, one-tail t-test).

Even though the observed trends are consistent and partly significant, one could think of an alternative explanation for this variance. To reject the alternative explanation that the length of the new segments might depend on either the length or number of the segments made prior to the treatment, we assayed for possible correlations between (1) average length of the segments that developed before and during the treatment and (2) average length of segments and the number of segments present prior to the treatment. As a result, neither the average length of segments (Table S3.1) nor the segment number before the treatment (Table S3.2) was significantly correlated with the average size of new segments under experimental conditions. We conclude that these findings constitute

first evidence that RA signalling is involved in patterning of the segments that are formed during the metamorphosis of the anal fin into the gonopodium.

The spatial expression of androgen receptor βββ differs between gonopodia and swords Both the gonopodium and the sword can be induced by exogenous testosterone [43, 67, 74, 75], suggesting that androgen signalling is located up-stream of a signalling cascade that results in the activation of downstream effectors, such as Fgf-signalling or msxC , to shape the gonopodium and the sword [59, 148]. Due to an ancient, fish-specific genome duplication event, teleost fish harbour two androgen receptor paralogs [184]. To analyse the expression of the two androgen receptors of the green swordtail, we treated juvenile individuals with testosterone and analysed gene expression by in situ hybridisation at different stages of gonopodium and sword development. androgen receptor α showed a diffuse expression in fin rays of developing gonopodia and swords with no clearly restricted pattern (data not shown). In contrast, after 5 days of testosterone treatment (dt), androgen receptor-β (ar β) is strongly up-regulated in the distal tip of the 3-4-5 complex and partly in the interray tissue (Figure 3.7A). A similar patter can be observed after 7 dt, when the 3-4-5 complex is clearly elongated compared to the fin rays that do not participate in gonopodium formation (Figure 3.7B). In situ hybridisation on sections of 7 dt gonopodia revealed that ar β is expressed in a layer of mesenchymal cells underlying the epidermis (Figure 3.7C, C’). The expression domain covers the distal-most mesynchme, where also aldh1a2 is expressed, and expands proximally, overlapping with the expression of the two rar-γ paralogs (compare Figures 3.7C, C’ and 3.4G-I). Up- regulation of ar β in the 3-4-5 complex persists to later stages of gonopodium development (18 dt) when the first distal structures have formed (Figure 3.7D). ar β expression could also be detected in the 3-4-5 complex of untreated control fins, but at lower levels than in treated fin (compare Figure 3.7E to 3.7A). In addition, expression could be detected in the segment borders when samples were stained for a longer time period (Figure 3.7E).

Figure 3.7. Expression of androgen receptor β in the developing gonopodium and sword androgen receptor β (ar β) is expressed in developing gonopodia and sword of X. helleri . At 5 days of testosterone treatment (dt) ar β is strongly up-regulated in distal tip of the gonopodial rays 3, 4 and 5 and clearly weaker in the rays 6-7 (A). Expression could also be detected in the interray tissue. This expression pattern persist at 7 (B) and 18 dt (D). Longitudinal sections of anal fins after 7 dt revealed ar β to be expressed both in the distal and proximal mesenchyme (C, C’). In control fins ar β is expressed at basal levels (E). In developing swords after 5 (F), 7 dt (H) and control fins (I) ar β is expressed in a stripe-like expression domain in sword and non-sword rays. This expression overlaps with the segment boundaries of two joined segments (G). Expression at the segment boundaries could also be found in gonopodia that are stained for a longer time period (E). White arrowheads indicate gene expression. (n> 10 for every stage and probe; scale bars: A, B, D-F and H, I: 200 µm; C, C’, G: 100 µm)

The expression of ar β in developing swords clearly differs from that in developing gonopodia. In the caudal fin after 5 dt, arß expression is detected in a stripe-like expression domain (Figure 3.7F) that covers the segment boundaries in both sword and non-sword rays (Figure 3.7G). The distal-most domains are therefore likely to mark new segment boundaries. In addition, arß expression cannot be detected in all segment borders at the same time, which might suggest a dynamic expression pattern. No difference in expression was found between sword and non-sword rays. A similar pattern can also be observed after 7 dt (Figure 3.7H) and in control fins (Figure 3.7I). Experiments with sense probes on 7 dt caudal fins confirmed the obtained results (data

not shown). Our data therefore suggests that androgen receptor β is differently regulated in developing gonopodia compared to swords.

3.5 Discussion

Maturating male fish of the family poeciliidae modify their anal fin into an intromittant organ, the gonopodium [61]. The gonopodium is a morphological complex structure that is mainly formed by the anal fin rays 3-5, the so-called 3-4-5 complex [2, 82]. The molecular mechanisms of gonopodium development are thought to be widely conserved within the whole group, even though some variation in morphology (e.g. terminal structure) occurs between species [75, 85]. During its development the gonopodium acquires a distinct proximo-distal polarity (Figure 3.1B). The terminal segments of the 3- 4-5 complex decrease in length and partly exhibit specific structures like serraes and spines, or are transformed into hooks and claws [80, 82, 83, 85]. Turner proposed a two- phase model of gonopodium development [75, 80]. In the first phase, under low levels of the male sexual hormone testosterone growth of additional ray segments is initiated in the anal fin. The added segments are of similar length or even longer than the old segments. In the second phase of gonopodium development the so-called differentiation areas arise in response to a steady increase in testosterone levels. According to Turner these differentiation areas add new bony segments or structures with a defined shape such as spines or serraes to the fin rays. The term “differentiation areas” was coined on the basis of empirical observations half a century ago and so far no molecular data confirmed the presents of different areas within the gonopodium that regulated the formation of specific structures. The term “differentiation areas” might therefore just be a nomenclature that groups the gonopodium into specific areas where bony structures of different shape develops. However, a two-phase model could explain why gonopodia induced with high levels of exogenous testosterone are shortened compared to normal gonopodia (compare Figure 3.1B to 3.1C). Due to the high concentration of testosterone an induced gonopodium will lack the first phase of fin ray growth, because terminalisation of the gonopodium is induced much too early. Therefore fewer segments are formed because the terminal segments that contribute to the distal structures develop almost immediately after gonopodial ray growth is induced. This view is consistent with Turners and

Grobsteins [74, 75] as well as with our own observations. The genetic nature of terminal differentiation of developing gonopodia has so far been completely unknown. Mature gonopodia exhibit a strong proximo-distal polarity, with segments in the distal tip of the gonopodium that are much shorter and more differentiated than more proximal segments. Signals that can mediate positional information along the proximo-distal axis might therefore be involved in gonopodium development.

Retinoic acid signalling is involved in patterning of the gonopodium Retinoic acid (RA) signalling has been shown to mediate positional information along the proximo-distal axis in developing and regenerating limbs [90, 91, 185]. In this study we showed that both aldh1a2 and two rar-γ paralogs are expressed throughout different stages of gonopodium development. We demonstrated that RA signalling is involved in patterning of the gonopodium, since the length of newly made segments negatively correlates with RA levels (Figure 3.6E). Inhibition of RA signalling increases segment size, while exogenous RA reduces it. The effect of DEAB and atRA is most likely to be specific, since we did not find evidence that other factors influenced segment length, such as segment number or length prior to the time of the experiment. However, this increase or decrease in segment size is not significant for every ray, which might indicate that RA signalling is not efficiently blocked or over-activated during the whole experiment. Firstly, DEAB might be metabolised by microorganism in the aquarium water, since the experiments cannot carried out in a sterile environment. Indeed, when we reared zebrafish embryos as a control in the DEAB treated aquarium water, no DEAB induced developmental defect could be observed when the last DEAB treatment had been more than 12 h ago (unpublished data). Secondly, the experiment fish experienced the normal day:night cycle of 12 h each. atRA is sensitive to light [186] and was therefore likely inactivated at some time during the day cycle. The most compelling explanation for the effects of retinoic acid in the light of Turner’s proposed model for gonopodium development [75, 80] would be that RA signalling is involved in formation of the terminal segments in phase 2 of gonopodium development. RA signalling might directly or indirectly modulate the growth rate of the segments. Alternatively, RA signalling could induce the premature formation of a new segment

boundary, which would explain why extensive RA treatment creates a segment pattern phenotype. Both mechanisms would result in shorter segments as observed in the distal tip of the gonopodium. Various studies have shown that RA signalling can inhibit cell growth and trigger cell differentiation in vitro and in vivo [187, 188]. However, a role of RA in growth regulation of fin ray segments has been shown for the first time in our experiments. The best studied RA-dependend mechanism that regulates segment size is acting during somitogenesis [189]. Studies in mouse and Xenopus showed that somite length is regulated by two antagonizing gradients of RA and Ffg8 [190, 191]. However, a disruption of the RA gradient results in smaller somites, whereas reduction in RA synthesis in anal fins results in larger segments. It is therefore not likely that RA signalling controls gonopodial segment size in a similar way than in the somites. One could also think of an alternative mechanism of how RA signalling could be involved in terminalisation of the gonopodium. RA signalling might provide positional information along the proximo-distal axis that is used to specify the distal part of the gonopodium and possibly the individual terminal segments. RA might form a distal- proximal gradient, that is converted into gene expression by rar-γs in a concentration depended matter. This localized gene expression might contribute to the determination of individual segments. A function of RA as a mediator for positional information along the proximo-distal axis is most extensively studied in the developing forelimb of tetrapods [89-91, 192]. In the developing limb proximo-distal polarity is thought to be established by an RA activity gradient. RA is synthesized by aldh1a2 in the lateral mesoderm [87-91, 192, 193]. RA diffuses distally and activates target genes such as meis homebox transcription factors in the proximal limb bud mesenchyme, which determine these cell to a proximal fate [89-91]. RA activity therefore has to be restricted to the proximal limb bud, which is achieved by degradation of RA through Cyp26B1 in the distal limb bud and possibly by antagonizing activity of Fgf8 [91, 192]. There is good evidence that this mechanism for positional information is also reused during limb regeneration [92, 185]. In addition, in mouse and newt RA receptor γ was found to be involved in converting the RA signal into gene activation [194-197]. One rar γ ortholog was also found to be expressed in caudal fin regeneration [198]. Furthermore some studies suggest that RA signalling may control the position of bifurcation via regulation of genes like shh and

therefore mediate proximo-distal information in a similar way as in paired appendages [107, 199]. Even though aldh1a2 and rar-γ are expressed in both developing/regenerating limbs and developing gonopodia, two things argue against this scenario. Firstly, RA signalling in limbs seems to determine cells to a proximal fate, because treatment with exogenous RA leads to proximalisation of distal cells [90, 185]. However, over-activation of RA-signalling seems to distalise the gonopodium, if we assume RA signalling to specify the proximo-distal axis. Secondly, this model would assume that shorter segment length is a result of distalisation. This is challenged by the fact that terminal segments closer to the distal end are not necessary shorter than more proximally located terminal segments (Figure 3.6 and data not shown). Additional experimental data is needed to elucidate the function of RA signalling in gonopodium development and to see if some of these mechanisms are involved. One important goal would be to overcome the experimental limitations of the system. At the moment, we are evaluating the possibility to use morpholinos to knockdown aldh1a2 in the developing gonopodium. If this is feasible, MO should enable us to inhibit RA signalling more reliable than treatment with DEAB and produce a less variable phenotype to analyse. In addition, new atRA analogs such as DTAB (IUPAC: 3-[(4,6-diphenoxy- 1,3,5-triazin-2-yl)amino]benzoic acid), specific to Rarb and Rarg [200], might be more stable under experimental conditions and could be used instead of atRA. DTAB could be used in a setup with modified testosterone concentration, to test if RA signalling is necessary to induce phase two of gonopodium development. Futhermore, expression analysis of RA targets, such as meis and cyp 26 genes, should give further information about how gonopodial patterning is regulated by RA signalling. aldh1a2 and androgen receptor βββ might interact during gonopodium development Gonopodium and also sword development is thought to be activated by elevated levels of testosterone in poeciliid fish [43, 67, 75, 80]. Several molecular studies support this view: (1) The expression of genes involved in fin ray growth, such as msxC , fgfr1 and shh , is induced by testosterone [59, 86, 148], (2) androgen receptors (ars ) are expressed in the developing gonopodium [86] and (3) inhibition of androgen signalling down- regulates gene expression and perturbs gonopodium growth [86]. We have shown that

aldh1a2 expression is induced by testosterone in the anal fin, suggesting that aldh1a2 induction depends on androgen signalling. Ar β might thus be a putative regulator of aldh1a2 activity in the gonopodium. Firstly, ar β is expressed in a distinct pattern that overlaps with that of aldh1a2 in the distal-most mesenchyme of the 3-4-5 complex, but also in the anal fin rays 6 and 7. Secondly, both genes are expressed differently in developing swords compared to developing gonopodia. Changes in ar β regulation might therefore account for the loss of aldh1a2 expression in the distal mesenchyme of growing sword rays. In vitro experiments have shown that aldh1a3 , another RA producing enzyme, is regulated by androgen signalling in specific cell lines [201]. However, more evidence is needed to support that aldh1a2 expression depends on Ar β activity. Unfortunately, inhibition of androgen signalling with flutamide, an androgen antagonist, did not result in a significant decrease of aldh1a2 expression (unpublished data). However, this could simply be due to technical difficulties since aldh1a2 staining also varied between different samples. Despite the loss of expression domains in the sword compared to the gonopodium, ar β is still expressed in the segment boundaries of two adjoined segments in both organs, which only becomes visible after prolonged staining (compare Figures. 3.7E to 3.7F, H, I). The function of ar β in the segment boundaries is unknown. Inhibition of androgen signalling with flutamide in regenerating caudal fins of males, where ar β is also express in the segment boundaries, only slowed down regeneration but did not lead to an obvious segment patterning phenotype (unpublished data). In addition, a role of ar α in aldh1a2 regulation cannot be excluded, since ar α is also expressed in the developing gonopodium, though in a diffuse pattern. On the other hand, ar α is not differently regulated between gonopodia and swords, whereas ar β and aldh1a2 are. This question awaits improvement of molecular manipulation techniques to be answered. ar β expression also overlaps with that of the two rar γ paralogs. Experiments in rats confirmed a regulation of rar γ by Androgen signalling in several tissues [202]. In developing swords however, rar γ expression is likely to be independent of Ar β, since the regulation of rar-γa and rar-γb is comparable between gonopodia and swords, even though ar β is regulated differently between these two structures.

Only parts of the developmental program that shapes the gonopodium was co-opted to form the sword Sword and gonopodium development show striking similarities. Both are male-specific structures that are formed by a specific subset of fin rays in response to testosterone [43, 67, 75, 80]. Moreover, recent studies provided first evidence that the molecular mechanisms promoting sword and gonopodium growth might be similar [59, 148]. Zauner and colleagues argued that the genetic network underlying gonopodium development was partly adopted to form the sword [59]. Despite those similarities in development, the mature sword and gonopodium are obviously different from each other in respect to their morphology [43, 85]. On the one hand, the sword morphology is less complex than that of the gonopodium. On the other hand the sword exhibits a complex colour pattern that is absent in the gonopodium. Therefore it is likely that the genetic repertoire of sword and gonopodium development differs in some aspects. In this study we showed that components of RA and Androgen signalling are differently regulated between swords and gonopodia and therefore first evidence that not all gonopodial modules were adopted for sword development. Both retinoic acid receptor (rar ) paralogs are regulated similarly in developing swords and gonopodia, but aldh1a2 regulation differs clearly between these two structures. In contrast to gonopodial rays a localised expression of aldh1a2 as a source for RA was not found in sword rays. Therefore, sword rays might be capable of receiving and transmitting a RA signal, but a signalling source is likely to be missing since we found no other RA producing enzyme expressed in the sword. A sword is morphologically more simple and mainly its length and colouration is important to attract females [32, 40]. Therefore, evolution might have acted mainly on these two parameters. If RA signalling is involved in creating complex distal structures the maintaining of a functional RA signalling pathway might have not been necessary. On the contrary, these distal structures are thought to be involved in the mating process [83] and selection most likely maintained the molecular mechanisms that shape this structures in the gonopodium. Both structures can therefore be used in further comparative approaches to identify genetic modules that are unique to either the sword or the gonopodium.

Summary

The sword, a colourful extension of the ventral caudal fin of male swordtails of the genus Xiphophorus was one of Darwin’s chosen examples for his idea of sexual selection. Experiments in X. helleri have shown that (1) the total length of the sword is an important criterion during mate choice and (2) the females have a preference for a specific pattern of differently coloured stripes. Besides its role in the process of sexual selection the sword has an interesting evolutionary history. Only males of swordtail species develop a sword, whereas males of platyfish, another group within the genus Xiphophorus , are swordless. One scenario suggests a sworded common ancestor of all Xiphophorus species and a secondary loss of the sword in platyfish during evolution. It remains elusive which molecular events preceded the loss of the sword in platyfish, since the genetic network that controls sword development is poorly understood. Data from interspecies crosses suggests that multiple loci control sword development. In addition, testosterone was identified as sufficient factor to induce sword development in immature fish, which indicates that sword development is controlled by androgen signalling. Recent work also identified the homeobox transcription factor msxC as another potential candidate, since it was shown to be up-regulated in growing sword rays. Up-regulation of msxC has also been found in the developing gonopodium, the modified male anal fin that is also induced by exogenous testosterone. The gonopodium is evolutionary older than the sword and it was assumed that the genetic network controlling gonopodium development was partly co-opted for the sword. In chapter I we focussed on Fgf signalling that has been shown to regulate msxC expression during caudal fin regeneration. Both sword development and fin regeneration are characterized by elevated outgrowth of fin rays, which is likely controlled by a conserved genetic network. We showed that fgfr1 is specifically up-regulated in developing swords, which presents first evidence that fgfr1 is involved in sword development. A similar pattern was also observed in the developing gonopodium. fgfr1 is spatial-temporally co-expressed with msxC both in the sword and the gonopodium, which might indicate a putative interaction between both genes. Interestingly, in the ventral caudal fin rays of testosterone treated platyfish, fgfr1 and msxC are only up-regulated after prolonged hormone treatment. This points towards a disruption between the

100

fgfr1/msxC network and its regulation by testosterone as a likely developmental cause for sword-loss in platyfish. Finally, we demonstrated that fgfr1 and msxC activation is correlated with fin ray growth rates by employing the X. maculatus brushtail mutant that exhibits excessive growth of the median caudal fin rays. Only a subset of the genes involved in sword development can be targeted by candidate gene approaches (e.g. as performed in chapter I), because prior knowledge about gene function is needed to select appropriate candidates. In chapter II we employed the suppression subtractive hybridisation (SSH) technique to bypass this limitation, because this method can be applied to isolate genes that are differentially expressed in swords and gonopodia compared to juvenile fins without ab initio knowledge of gene identity or function. In this study we identified 128 different sequences with significant similarity to known genes. We showed that four of these sequences with similarity to rack1 , dusp1 , klf2 and tmsb a-like are specifically up-regulated in induced swords and/or gonopodia. In parallel, we also showed that these genes are strongly expressed during fin regeneration. Therefore these four genes are interesting candidates to further analyse their role in both sword development and fin regeneration. The anal fin of male Xiphophorus fishes is modified into an intromittant organ, the gonopodium. The gonopodium is formed during sexual maturation by a subset of three anal fin rays, the 3-4-5 complex. These three rays are modified in terms of ray length, segment thickness and different distal structures like blades, claws, spines, hooks and serraes. Therefore, the mature gonopodium exhibits a strong proximo-distal polarity due to the smaller terminal segments and terminal structures. Gonopodium development is thought to proceed in two phases. During the first phase, low levels of testosterone promote ray outgrowth, whereas high levels of testosterone induce the formation of terminal structures during the second phase. Shh, androgen and probably Fgf signalling are involved in gonopodium development. In chapter 3 we tested the role of retinoic acid (RA) signalling during gonopodium development, for two reasons. RA signalling is essential for paired appendage development in vertebrates and it provides positional information along the proximo- distal axis in developing and regenerating limbs. Therefore, RA signalling might either play a general role in gonopodium development or specific role in establishing the

101

proximo-distal polarity within the gonopodium. RA, a small lipophilic, diffusible molecule is synthesised by retinaldehyde dehydrogenases (Aldh1as) and stimulates gene expression through binding to two types of receptors, retinoic acid receptors (RARs) and retinoic X receptors (RXRs). In this study we showed that aldh1a2 , a RA synthesising enzyme, and two RA receptors, rar γ-a and rar γ-b, are expressed in developing gonopodia. Inhibiting RA synthesis with DEAB increases the length of newly formed terminal segments, whereas the segment length decreases when RA signalling is over- activated by exogenous all-trans RA. Both the expression and the functional data present first evidence that RA signalling is involved in gonopodium development. Finally, we showed that androgen receptors β (ar β), a putative regulator up-stream of RA signalling is co-expressed with aldh1a2 in the distal mesenchyme of the gononopodial rays. Interestingly, developing swords lack the distal expression domain of both aldh1a2 and ar β, whereas the two rars are similarly expressed in developing swords and gonopodia. This might point towards an interaction between these two genes.

102

Zusammenfassung

Das Schwert, eine farbige Verlängerung der Kaudalflosse männlicher Schwertträger der Gattung Xiphophorus , war eines von Charles Darwins Beispielen für seine Idee der sexuellen Selektion. Experimente mit X. helleri zeigten zum einen, dass die Länge des Schwertes eine wichtige Rolle bei der Partnerwahl spielt und zum anderen, dass Weibchen eine Präferenz für ein spezielles Farbmuster des Schwertes mit verschiedenfarbigen Streifen haben. Neben seiner Funktion als sexuell selektiertes Merkmal, besitzt das Schwert eine interessante Evolutionsgeschichte. Nur die Männchen der Schwerträgerarten entwickeln ein Schwert, während die Männchen von Platyarten, die eine zweite Gruppe innerhalb der Gattung Xiphophorus darstellen, kein Schwert besitzen. In einem Szenario stammen beide Gruppen von einem gemeinsamen Vorfahren ab, dessen Männchen ein Schwert ausbildeten. Das Schwert ging daher folglich sekundär im Laufe der Evolution in Platys verloren. Es bleibt schwer zu fassen, welche Ereignisse auf der molekularen Ebene zum Verlust des Schwertes führten, da das genetische Netzwerk, das die Schwertentwicklung kontrolliert sehr schlecht untersucht ist Kreuzungsexperimente mit unterschiedlichen Arten zeigten, dass mehrere genetische Loci die Schwertentwicklung kontrollieren. Zusätzlich konnte Testosteron als ausreichender Faktor für die Induktion der Schwertentwicklung in Fischen vor der Geschlechtsreife identifiziert werden. Dies deutet darauf hin, dass die Schwertentwicklung durch Androgene kontrolliert wird. Eine kürzlich verfasste Arbeit zeigte zudem, dass der Homeobox-Transkriptionsfaktor msxC in wachsenden Schwertstrahlen hochreguliert wird und somit vermutlich in der Schwertentwicklung eine Rolle spielt. msxC wird zudem in Flossenstrahlen des wachsenden Gonopodiums hochreguliert, eine Analflosse die wie die Kaudalflosse im Männchen unter dem Einfluss von Testosteron modifiziert wird. Das Gonopodium entstand im Laufe der Evolution vor dem Schwert und es wird angenommen, dass ein Teil des genetischen Netzwerks, das die Gonopodiumentwicklung kontrolliert, für die Entwicklung des Schwertes adaptiert wurde. In Kapitel I konzentrierten wir uns auf den Fgf Signalweg, der die Expression von msxC während der Regeneration der Kaudalflosse reguliert. Sowohl die Schwertentwicklung als auch die Flossenregeneration zeichnen sich durch ein schnelles

103

Wachstum der Flossenstrahlen aus und es wird angenommen, dass dieses Wachstum von einem konservierten genetischen Netzwerk kontrolliert wird. Wir konnten zeigen, dass fgfr1 in wachsenden Schwertern und Gonopodien hochreguliert wird, was ein erstes Anzeichen dafür ist, dass fgfr1 eine Rolle in diesen Prozessen spielt. fgfr1 und msxC wurden in einem vergleichbaren zeitlichen und räumlichen Muster exprimiertund interagieren daher wahrscheinlich miteinander. Hingegen wurden beide Gene im ventralen Teil der Kaudalflossen von Testosteron-behandelten Platys erst nach extensiver Behandlung mit dem Hormon hochreguliert. Dies spricht dafür, dass es im Laufe der Evolution zu einer Störung der Regulation des fgfr1 /msxC -Netzwerks durch Testosteron kam, die in Platys zum Verlust der Fähigkeit führte, ein Schwert zu entwickeln. Zum Abschluss zeigten wir, in X. maculatus brushtail Mutanten, dass die Stärke der Aktivierung von fgfr1 und msxC in den Flossenstrahlen positiv mit deren Wachstumsrate korreliert. Die medianen Flossenstrahlen dieser Mutante zeichnen sich durch starkes Wachstum aus. Nur ein kleiner Teil der Gene die die Schwertentwicklung kontrollieren, können mit einem sogenannten Kandidatengenverfahren (siehe Kapitel I) identifiziert werden, da nur Gene ausgewählt werden, wenn Informationen über diese Gene und/oder ihre Funktion vorhanden sind. Um diese Limitierung zu umgehen verwendeten wir in Kapitel 2 die „suppression subtractive hybridisation” Technik (SSH), da es diese Methode ermöglicht Gene zu identifizieren, die in sich entwickelnden Schwertern und Gonopodien im Vergleich zu Flossen von Jungtieren hochreguliert sind. Die Methode erfordert dabei keine vorherige Kenntnis über bestimmte Gene oder deren Funktion. In dieser Studie wurden von uns 128 Sequenzen mit einer hohen Ähnlichkeit zu Genen in Datenbanken isoliert. Wir zeigten, dass vier dieser Gene mit Ähnlichkeit zu rack1 , dusp1 , klf2 and tmsb a-like spezifisch in sich entwickelnden Schwertern und Gonopodien hochreguliert wurden. Zusätzlich konnten wir zeigen, dass diese Gene auch während der Flossenregeneration hochreguliert wurden. Somit sind diese Gene interessante Kandidaten für weitere Analysen bezüglich ihrer Rolle in der Schwertentwicklung und Flossenregeneration. Die Analflosse von Männchen aller Xiphophorus -Arten wird in ein Begattungsorgan, das sogenannte Gonopodium umgewandelt. Das Gonopodium wird von drei

104

Flossenstrahlen, dem 3-4-5 Komplex gebildet. Diese drei Flossenstrahlen sind stark verlängert und weisen unterschiedliche terminale Strukturen, wie Hacken, Klauen, Dornen und Serraes auf. Daher weist das fertig entwickelte Gonopodium eine starke proximal-distale Polarität auf, die durch die terminalen Strukturen und kürzere terminale Segmente bedingt wird. Die Entwicklung des Gonopodiums erfolgt in zwei Phasen. Während der ersten Phase regt ein geringer Testosteronspiegel das Wachstum des 3-4-5 Komplexes an, während ein hoher Testosteronspiegel in Phase zwei die Bildung von terminalen Strukturen auslöst. Der Shh-, androgene und vermutlich auch der Fgf- Signalweg spielen bei der Gonopodiumentwicklung eine Rolle. In Kapitel 3 testeten wir aus zwei Gründen die Rolle des Retinsäure-Signalwegs (RA) während der Gonopodiumentwicklung. Zum einen ist der RA-Signalweg essentiell für die Entwicklung von paarigen Extremitäten in Wirbeltieren, zum anderen vermittelt er Positionsinformationen entlang der proximal-distalen Achse in wachsenden und regenerierenden Beinen. Daher könnte der RA-Signalweg eine eher generelle Rolle in der Entwicklung des Gonopodiums spielen oder speziell in die Etablierung der proximal- distalen Polarität involviert sein. Retinsäure (RA) ist ein kleines, lipophiles, diffundierendes Molekül, das von Retinaldehyddehydrogenasen (Aldh1as) synthetisiert wird und über zwei Typen von Retinsäurerezeptoren (RARs and RXRs) seine Zielgene aktiviert. In dieser Studie konnten wir zeigen, dass aldh1a2 , ein RA-Syntheseenzym, und zwei RA Rezeptoren ( rar γ-a and rar γ-b) in sich entwickelnden Gonopodien hochreguliert werden. Eine Hemmung der RA Synthese mit DEAB erhöhte die Länge von neu gebildeten Segmenten, während eine überstarke Aktivierung des RA-Signalwegs mit exogener Retinsäure zu einer Verkürzung der Segmente führte. Sowohl die Expressionsdaten als auch die funktionellen Daten weisen auf eine Funktion von RA während der Gonopodiumentwicklung und speziell bei der Segmentierung hin. Abschließend zeigten wir, dass der androgene Rezeptor β (ar β), ein möglicher Regulator des RA-Signalwegs, mit aldh1a2 im distalen Mesenchym von wachsenden Gonopodien exprimiertwird. In wachsenden Schwertern fehlt diese Expressionsdomäne beider Gene, während beide Rezeptoren in Schwertern und Gonopodien ähnlich reguliert werden. Dies deutet darauf hin, dass beide Gene miteinander interagieren könnten.

105

Results produced by collaborators

Unless otherwise mentioned, all results in this thesis were performed by myself or under my direct supervision.

Chapter 1: Gerrit Begemann (University of Konstanz) designed Figure 1.9. Nicola Blum (University of Konstanz) performed some of the brushtail in situ hybridisations.

Chapter 2: Amanda Duckworth performed some of the in situ hybridisations as part of her Vertiefungskurs project.

Chapter 3: Nicola Blum (University of Konstanz) performed some of the arß in situ hybridisations.

106

Literature cited

1. Ghedotti MJ: Phylogenetic analysis and taxonomy of the poecilioid fishes (Teleostei; Cyprinodontiformes). Zoological journal of the Linnean Society 2000, 130 . 2. Parenti LR: A Phylogenetic and Biogeographic Analysis of Cyprinodontiform Fishes (Teleostei, Atherinomorpha). Bull Amer Mus Nat Hist 1981, 168: 335- 557. 3. Kallman KD, Kazianis S: The Genus Xiphophorus in Mexico and Central America. Zebrafish 2006, 3: 271-285. 4. Meyer A, Morrissey JM, Schartl M: Recurrent origin of a sexually selected trait in Xiphophorus fishes inferred from a molecular phylogeny. Nature 1994, 368: 539-542. 5. Meyer A, Salzburger W, Schartl M: Hybrid origin of a swordtail species (Teleostei: Xiphophorus clemenciae ) driven by sexual selection. Mol Ecol 2006, 15: 721-730. 6. Gordon M: Platypoecilus now becomes Xiphophorus . The Aquarium 1951, 20: 277-279. 7. Atz JW: Effects of hybridization on pigmentation in fishes of the genus Xiphophorus . Zoologica: NY Zoolog Soc 1962, 47 . 8. Rosenthal GG, De La Rosa Reyna XF, Kazianis S, Stephens MJ, Morizot DC, Ryan MJ, De Leon FJG: Dissolution of sexual signal complexes in a hybrid zone between the swordtails Xiphophorus birchmanni and Xiphophorus malinche (Poeciliidae). Copeia 2003 :299-307. 9. Fisher HS, Wong BB, Rosenthal GG: Alteration of the chemical environment disrupts communication in a freshwater fish. Proc Biol Sci 2006, 273: 1187- 1193. 10. Axelrod HR, Wischnath L: Swordtails and Platies. Neptune, Nj: TFH Publications; 1991. 11. Gordon M: Heritable color variations in the Mexican swordtail-fish. J Hered 1937, 28: 223-230. 12. Zander CD: Uber die Entstehung und Veränderung von Farbmustern in der Gattung Xiphophorus (Pisces). Mitt Hamburg Zool Mus Inst 1969, 66: 241-271. 13. Meierjohann S, Schartl M: From Mendelian to molecular genetics: the Xiphophorus melanoma model. Trends Genet 2006, 22: 654-661. 14. Rosenthal GG, Garcıà de Leon FJ: Sexual Behavior, Genes, and Evolution in Xiphophorus . Zebrafish 2006, 3: 85-90. 15. Rosen DE: Fishes from the uplands and intermontane basins of Guatemala: revisionary studies and comparative geography. Bull Amer Mus Nat Hist 1979, 162: 269-375. 16. Franck D: Weitere Untersuchungen zur vergleichenden Ethologie der Gattung Xiphophorus . Behaviour 1968, 30: 75-95. 17. Franck D: Vergleichende Verhaltensstudien an lebendgebärenden Zahnkarpfen der Gattung Xiphophorus . Zool Jb Physiol 1964, 71: 117-170.

107

18. Ryan MJ, Causey BA: "Alternative" mating behavior in the swordtails Xiphophorus nigrensis and Xiphophorus pygmaeus (Pisces: Poeciliidae). Behav Ecol Sociobiol 1989, 24: 341-348. 19. Zimmerer EJ, Kallman KD: Genetic Basis for Alternative Reproductive Tactics in the Pygmy Swordtail, Xiphophorus nigrensis . Evolution 1989, 43: 1298-1307. 20. Kallman KD, Schreibman MP: A sex-linked gene controlling gonadotrop differentiation and its significance in determining the age of sexual maturation and size of the platyfish, Xiphophorus maculatus . Gen Comp Endocrinol 1973, 21: 287-304. 21. Schreibman MP, Kallman KD: The genetic control of the pituitary-gonadal axis in the platyfish, Xiphophorus maculatus . J Exp Zool 1977, 200: 277-293. 22. Chang YS, Huang FL: The mode of action of carp gonadotropin on the stimulation of androgen production by carp testis in vitro. Gen Comp Endocrinol 1982, 48: 147-153. 23. Yazawa T, Uesaka M, Inaoka Y, Mizutani T, Sekiguchi T, Kajitani T, Kitano T, Umezawa A, Miyamoto K: Cyp11b1 is induced in the murine gonad by luteinizing hormone/human chorionic gonadotropin and involved in the production of 11-ketotestosterone, a major fish androgen: conservation and evolution of the androgen metabolic pathway. Endocrinology 2008, 149: 1786- 1792. 24. Okuzawa K: Puberty in teleosts. Fish Physiol Biochem 2002, 26: 31-41. 25. Ryan MJ, Hews DK, Wagner WE: Sexual Selection on Alleles That Determine Body Size in the Swordtail Xiphophorus-Nigrensis . Behavioral Ecology and Sociobiology 1990, 26: 231-237. 26. Ryan MJ, Pease CM, Morris MR: A Genetic-Polymorphism in the Swordtail Xiphophorus-Nigrensis - Testing the Prediction of Equal Fitnesses. Am Nat 1992, 139: 21-31. 27. Rosenthal GG, Martinez TYF, Garcıá de Leon FJ, Ryan MJ: Shared Preferences by Predators and Females for Male Ornaments in Swordtails. Am Nat 2001, 158: 146-154. 28. McLennan DA, Ryan MJ: Responses to conspecific and heterospecific olfactory cues in the swordtail Xiphophorus cortezi. Animal Behaviour 1997, 54: 1077-1088. 29. Cummings ME, Rosenthal GG, Ryan MJ: A private ultraviolet channel in visual communication. P Roy Soc Lond B Bio 2003, 270: 897-904. 30. Kingston JJ, Rosenthal GG, Ryan MJ: The role of sexual selection in maintaining a colour polymorphism in the pygmy swordtail, Xiphophorus pygmaeus . Animal Behaviour 2003, 65: 735-743. 31. Morris MR, Mussel M, Ryan MJ: Vertical Bars on Male Xiphophorus- Multilineatus - a Signal That Deters Rival Males and Attracts Females. Behavioral Ecology 1995, 6: 274-279. 32. Rosenthal GG, Evans CS: Female preference for swords in Xiphophorus helleri reflects a bias for large apparent size. Proc Natl Acad Sci U S A 1998, 95: 4431- 4436.

108

33. Rosenthal GG, Evans CS, Miller WL: Female preference for dynamic traits in the green swordtail, Xiphophorus helleri . Animal Behaviour 1996, 51: 811-820. 34. Cummings ME, Larkins-Ford J, Reilly CR, Wong RY, Ramsey M, Hofmann HA: Sexual and social stimuli elicit rapid and contrasting genomic responses. Proc Biol Sci 2008, 275: 393-402. 35. Darwin C: The descent of man, and selection in relation to sex. London: J. Murray; 1871. 36. Basolo AL: Male swords and female preferences: response. Science 1991, 253 . 37. Meyer A: The evolution of sexually selected traits in male swordtail fishes (Xiphophorus : Poeciliidae). Heredity 1997, 79: 329-337. 38. Basolo AL: A further examination of a pre-existing bias favouring a sword in the genus Xiphophorus . Anim Behav 1995, 50: 365-375. 39. Basolo AL: Female preference for male sword length in the green swordtail, Xiphophorus helleri (Pisces: Poeciliidae). Anim Behav 1990, 40: 332–338. 40. Basolo AL, Trainor BC: The conformation of a female preference for a composite male trait in green swordtails. Anim Behav 2002, 63: 469-474. 41. Trainor BC, Basolo AL: Location, location, location: stripe position effects on female sword preference. Anim behav 2006, 71: 135-140. 42. Basolo AL: Female Preference Predates the Evolution of the Sword in Swordtail Fish. Science 1990, 250: 808-810. 43. Dzwillo M: Sekundäre Geschlechtsmerkmale einiger Xiphophorini unter dem Einfluss von Methyl-testosteron. Kosswig-Festschrift 1964 :15-22. 44. Zander CD, Dzwillo M: Untersuchungen zur Entwicklung und Vererbung des Caudalfortsatzes der Xiphophorus-Arten (Pisces). Zeitschrift für Wissenschaftliche Zoologie 1969, 178: 276-315. 45. Becerra J, Montes GS, Bexiga SR, Junqueira LC: Structure of the tail fin in teleosts. Cell Tissue Res 1983, 230: 127-137. 46. Montes GS, Becerra J, Toledo OM, Gordilho MA, Junqueira LC: Fine structure and histochemistry of the tail fin ray in teleosts. Histochemistry 1982, 75: 363- 376. 47. Santamaria JA, Mari-Beffa M, Becerra J: Interactions of the lepidotrichial matrix components during tail fin regeneration in teleosts. Differentiation 1992, 49: 143-150. 48. Wong BB, Rosenthal GG: Female disdain for swords in a swordtail fish. Am Nat 2006, 167: 136-140. 49. Fisher HS, Rosenthal GG: Male swordtails court with an audience in mind. Biol Lett 2007, 3: 5-7. 50. Walling CA, Royle NJ, Metcalfe NB, Lindstrom J: Green swordtails alter their age at maturation in response to the population level of male ornamentation. Biol Lett 2007. 51. Benson KE, Basolo AL: Male-male competition and the sword in male swordtails, Xiphophorus helleri . Animal Behaviour 2006, 71: 129-134. 52. Eibner C, Pittlik S, Meyer A, Begemann G: The developmental basis of evolutionary innovation: Inductive signals coordinate the development of a sexually selected trait in swordtail fish. Evol Dev 2008, 10: 403-412.

109

53. Rauchenberger M, Kallman KD, Morizot DC: Monophyly and geography of the Río Pánuco Basin swordtails (genus Xiphophorus ) with descriptions of four new species. Am Mus Novitates 1990, 2975: 1-41. 54. Basolo AL: Phylogenetic evidence for the role of a pre-existing bias in sexual selection. Proc Biol Sci 1995, 259: 307-311. 55. Basolo AL: Congruence between the sexes in preexisting receiver responses. Behav Ecol 2002, 13: 832-837. 56. Ryan MJ: Sexual selection, receiver biases, and the evolution of sex differences. Science 1998, 281: 1999-2003. 57. McClintock WJ, Uetz GW: Female choice and pre-existing bias: Visual cues during courtship in two Schizocosa wolf spiders (Araneae: Lycosidae). Animal Behaviour 1996, 52: 167-181. 58. Ryan MJ, Rand AS: Sexual Selection and Signal Evolution - the Ghost of Biases Past. Philos T Roy Soc B 1993, 340: 187-195. 59. Zauner H, Begemann G, Mari-Beffa M, Meyer A: Differential regulation of msx genes in the development of the gonopodium, an intromittent organ, and of the "sword," a sexually selected trait of swordtail fishes ( Xiphophorus ). Evol Dev 2003, 5: 466-477. 60. Hrbek T, Seckinger J, Meyer A: A phylogenetic and biogeographic perspective on the evolution of poeciliid fishes. Mol Phylogenet Evol 2007, 43: 986-998. 61. Meyer A, Lydeard C: The evolution of copulatory organs, internal fertilization, placentae and viviparity in killifishes (Cyprinodontiformes) inferred from a DNA phylogeny of the tyrosine kinase gene X-src . Proc Biol Sci 1993, 254: 153-162. 62. Basolo AL, Alcaraz G: The turn of the sword: length increases male swimming costs in swordtails. Proc Biol Sci 2003, 270: 1631-1636. 63. Endler JA: Natural Selection on Color Patterns in Poecilia reticulata. Evolution 1980, 34: 76-91. 64. Karim N, Gordon SP, Schwartz AK, Hendry AP: This is not deja vu all over again: male guppy colour in a new experimental introduction. J Evol Biol 2007, 20: 1339-1350. 65. Millar NP, Reznick DN, Kinnison MT, Hendry AP: Disentangling the selective factors that act on male colour in wild guppies. Oikos 2006, 113: 1-12. 66. Johnson JB, Basolo AL: Predator exposure alters female mate choice in the green swordtail. Behav Ecol 2003, 14: 619-625. 67. Dzwillo M: Einfluss von Methyltestosteron auf die Aktivierung sekundärer Geschlechtsmerkmale über den arttypischen Zustand hinaus (Untersuchungen an xiphophorinen Zahnkarpfen). Verh Deutsch Zool Gesellsch Wien 1962 :151–159. 68. Iovine MK: Conserved mechanisms regulate outgrowth in zebrafish fins. Nat Chem Biol 2007, 3: 613-618. 69. Stoick-Cooper CL, Moon RT, Weidinger G: Advances in signaling in vertebrate regeneration as a prelude to regenerative medicine. Genes & Dev 2007, 21: 1292-1315.

110

70. Poss KD, Shen J, Nechiporuk A, McMahon G, Thisse B, Thisse C, Keating MT: Roles for Fgf signaling during zebrafish fin regeneration. Dev Biol 2000, 222: 347-358. 71. Thummel R, Bai S, Sarras MP, Jr., Song P, McDermott J, Brewer J, Perry M, Zhang X, Hyde DR, Godwin AR: Inhibition of zebrafish fin regeneration using in vivo electroporation of morpholinos against fgfr1 and msxb . Dev Dyn 2006, 235: 336-346. 72. Fischer S, Draper BW, Neumann CJ: The zebrafish fgf24 mutant identifies an additional level of Fgf signaling involved in vertebrate forelimb initiation. Development 2003, 130: 3515-3524. 73. Whitehead GG, Makino S, Lien CL, Keating MT: fgf20 is essential for initiating zebrafish fin regeneration. Science 2005, 310: 1957-1960. 74. Grobstein C: Endocrine and developmental studies of gonopod differentiation in certain Poeciliid fishes. II. Effect of testosterone propionate on the normal and regenerating anal fin of adult Platypoecilus maculatus females. J Exp Zool 1942, 89: 305 - 328. 75. Turner CL: Gonopodial characteristics produced in the anal fins of females of Gambusia affinis affinis by treatment with ethinyl testosterone. Biol Bull 1941, 80: 371-383. 76. Nishidate M, Nakatani Y, Kudo A, Kawakami A: Identification of novel markers expressed during fin regeneration by microarray analysis in medaka fish. Dev Dyn 2007, 236: 2685-2693. 77. Schebesta M, Lien CL, Engel FB, Keating MT: Transcriptional profiling of caudal fin regeneration in zebrafish. ScientificWorldJournal 2006, 6: 38-54. 78. Katogi R, Nakatani Y, Shin-i T, Kohara Y, Inohaya K, Kudo A: Large-scale analysis of the genes involved in fin regeneration and blastema formation in the medaka, Oryzias latipes . Mech Dev 2004, 121: 861-872. 79. Diatchenko L, Lau YF, Campbell AP, Chenchik A, Moqadam F, Huang B, Lukyanov S, Lukyanov K, Gurskaya N, Sverdlov ED, Siebert PD: Suppression subtractive hybridization: a method for generating differentially regulated or tissue-specific cDNA probes and libraries. Proc Natl Acad Sci U S A 1996, 93: 6025-6030. 80. Turner CL: Morphogenesis of the gonopodium in Gambusia affinis affinis . Ibid 1941, 69: 161-185. 81. Grobstein C: Endocrine and developmental studies of gonopod differentiation in certain poeciliid fishes. I. The structure and development of the gonopod in Platypoecilus maculatus . Univ Calif Publ Zool 1940, 47: 1-22. 82. Langer WF: Beiträge zur Morphologie der viviparen Cyprinodontiden. Gegenbauer's Morph Jahrb 1913, 47: 193-307. 83. Clark E, Aronson LR, Gordon M: Mating behavior patterns in two sympatric species of xiphophorin fishes: their inheritance and significance in sexual isolation. Bull Amer Mus Nat Hist 1954, 103: 139-225. 84. Rosen DE, Bailey RM: The poeciliid fishes (Cyprinodontiformes), their structure, zoogeography, and systematics. Bull Amer Mus Nat Hist 1963, 126: 1-176.

111

85. Gordon M, Rosen DE: Genetics of species differences in the morphology of the male genitalia of xiphophorin fishes. Bull Amer Mus Nat Hist 1951, 95: 413- 464. 86. Ogino Y, Katoh H, Yamada G: Androgen dependent development of a modified anal fin, gonopodium, as a model to understand the mechanism of secondary sexual character expression in vertebrates. FEBS Lett 2004, 575: 119-126. 87. Begemann G, Schilling TF, Rauch GJ, Geisler R, Ingham PW: The zebrafish neckless mutation reveals a requirement for raldh2 in mesodermal signals that pattern the hindbrain. Development 2001, 128: 3081-3094. 88. Gibert Y, Gajewski A, Meyer A, Begemann G: Induction and prepatterning of the zebrafish pectoral fin bud requires axial retinoic acid signaling. Development 2006, 133: 2649-2659. 89. Mic FA, Sirbu IO, Duester G: Retinoic acid synthesis controlled by Raldh2 is required early for limb bud initiation and then later as a proximodistal signal during apical ectodermal ridge formation. Journal of Biological Chemistry 2004, 279: 26698-26706. 90. Mercader N, Leonardo E, Azpiazu N, Serrano A, Morata G, Martinez C, Torres M: Conserved regulation of proximodistal limb axis development by Meis1/Hth. Nature 1999, 402: 425-429. 91. Mercader N, Leonardo E, Piedra ME, Martinez AC, Ros MA, Torres M: Opposing RA and FGF signals control proximodistal vertebrate limb development through regulation of Meis genes. Development 2000, 127: 3961- 3970. 92. Mercader N, Tanaka EM, Torres M: Proximodistal identity during vertebrate limb regeneration is regulated by Meis homeodomain proteins. Development 2005, 132: 4131-4142. 93. Morriss-Kay GM, Ward SJ: Retinoids and mammalian development. Int Rev Cytol 1999, 188: 73-131. 94. Panhuis TM, Butlin R, Zuk M, Tregenza T: Sexual selection and speciation. Trends Ecol Evol 2001, 16: 364-371. 95. Ritchie MG: Sexual Selection and Speciation. Annu Rev Ecol Evol Syst 2007, 38: 79-102. 96. Wiens JJ, Morris MR: Character definition, sexual selection, and the Evolution of swordtails. Am Nat 1996, 147: 866-869. 97. Parzefall J: Zur vergleichenden Ethologie verschiedener Mollienesia-Arten einschliesslich einer Höhlenform von M. sphenops . Behaviour 1969, 33: 1-37. 98. Lindholm AK, Brooks R, Breden F: Extreme polymorphism in a Y-linked sexually selected trait. Heredity 2004, 92: 156-162. 99. Warren I, Smith H: Stalk-eyed flies (Diopsidae): modelling the evolution and development of an exaggerated sexual trait. Bioessays 2007, 29: 300-307. 100. Emlen DJ, Corley Lavine L, Ewen-Campen B: On the origin and evolutionary diversification of beetle horns. Proc Natl Acad Sci U S A 2007, 104 Suppl 1: 8661-8668.

112

101. Gordon M, Cohen H, Nigrelli RF: A hormone-produced taxonomic character in Platypoecilius maculatus diagnostic of wild P. xiphidium . Am Nat 1942, 77: 569-572. 102. Satoh Y, Haraguchi R, Wright TJ, Mansour SL, Partanen J, Hajihosseini MK, Eswarakumar VP, Lonai P, Yamada G: Regulation of external genitalia development by concerted actions of FGF ligands and FGF receptors. Anat Embryol (Berl) 2004, 208: 479-486. 103. Tawk M, Tuil D, Torrente Y, Vriz S, Paulin D: High-efficiency gene transfer into adult fish: a new tool to study fin regeneration. Genesis 2002, 32: 27-31. 104. Lee Y, Grill S, Sanchez A, Murphy-Ryan M, Poss KD: Fgf signaling instructs position-dependent growth rate during zebrafish fin regeneration. Development 2005, 132: 5173-5183. 105. Groth C, Lardelli M: The structure and function of vertebrate f ibroblast growth factor receptor 1 . Int J Dev Biol 2002, 46: 393-400. 106. Smith A, Zhang J, Guay D, Quint E, Johnson A, Akimenko MA: Gene expression analysis on sections of zebrafish regenerating fins reveals limitations in the whole-mount in situ hybridization method. Dev Dyn 2008, 237: 417-425. 107. Laforest L, Brown CW, Poleo G, Geraudie J, Tada M, Ekker M, Akimenko MA: Involvement of the sonic hedgehog , patched 1 and bmp2 genes in patterning of the zebrafish dermal fin rays. Development 1998, 125: 4175-4184. 108. Borday V, Thaeron C, Avaron F, Brulfert A, Casane D, Laurenti P, Geraudie J: evx1 transcription in bony fin rays segment boundaries leads to a reiterated pattern during zebrafish fin development and regeneration. Dev Dyn 2001, 220: 91-98. 109. Johnson SL, Weston JA: Temperature-sensitive mutations that cause stage- specific defects in Zebrafish fin regeneration. Genetics 1995, 141: 1583-1595. 110. Draper BW, Stock DW, Kimmel CB: Zebrafish fgf24 functions with fgf8 to promote posterior mesodermal development. Development 2003, 130: 4639- 4654. 111. Poulin ML, Patrie KM, Botelho MJ, Tassava RA, Chiu IM: Heterogeneity in the expression of fibroblast growth factor receptors during limb regeneration in newts ( Notophthalmus viridescens ). Development 1993, 119: 353-361. 112. D'Jamoos CA, McMahon G, Tsonis PA: Fibroblast growth factor receptors regulate the ability for hindlimb regeneration in Xenopus laevis . Wound Repair Regen 1998, 6: 388-397. 113. Li C, Xu X, Nelson DK, Williams T, Kuehn MR, Deng CX: FGFR1 function at the earliest stages of mouse limb development plays an indispensable role in subsequent autopod morphogenesis. Development 2005, 132: 4755-4764. 114. Verheyden JM, Lewandoski M, Deng C, Harfe BD, Sun X: Conditional inactivation of Fgfr1 in mouse defines its role in limb bud establishment, outgrowth and digit patterning. Development 2005, 132: 4235-4245. 115. Lallemand Y, Nicola MA, Ramos C, Bach A, Cloment CS, Robert B: Analysis of Msx1 ; Msx2 double mutants reveals multiple roles for Msx genes in limb development. Development 2005, 132: 3003-3014.

113

116. Zhang H, Hu G, Wang H, Sciavolino P, Iler N, Shen MM, Abate-Shen C: Heterodimerization of Msx and Dlx homeoproteins results in functional antagonism. Mol Cell Biol 1997, 17: 2920-2932. 117. Quint E, Smith A, Avaron F, Laforest L, Miles J, Gaffield W, Akimenko MA: Bone patterning is altered in the regenerating zebrafish caudal fin after ectopic expression of sonic hedgehog and bmp2b or exposure to cyclopamine. Proc Natl Acad Sci U S A 2002, 99: 8713-8718. 118. Lee RK, Stainier DY, Weinstein BM, Fishman MC: Cardiovascular development in the zebrafish. II. Endocardial progenitors are sequestered within the heart field. Development 1994, 120: 3361-3366. 119. Woo K, Fraser SE: Specification of the hindbrain fate in the zebrafish. Dev Biol 1998, 197: 283-296. 120. Connors SA, Trout J, Ekker M, Mullins MC: The role of tolloid /mini fin in dorsoventral pattern formation of the zebrafish embryo. Development 1999, 126: 3119-3130. 121. Mintzer KA, Lee MA, Runke G, Trout J, Whitman M, Mullins MC: Lost-a-fin encodes a type I BMP receptor, Alk8, acting maternally and zygotically in dorsoventral pattern formation. Development 2001, 128: 859-869. 122. Chen W-J, Orti G, Meyer A: Novel evolutionary relationship among four fish model systems. Trends Genet 2004, 20: 424-431. 123. Steinke D, Salzburger W, Braasch I, Meyer A: Many genes in fish have species- specific asymmetric rates of molecular evolution. BMC Genomics 2006, 7: 20. 124. Yokoi H, Shimada A, Carl M, Takashima S, Kobayashi D, Narita T, Jindo T, Kimura T, Kitagawa T, Kage T, et al: Mutant analyses reveal different functions of fgfr1 in medaka and zebrafish despite conserved ligand-receptor relationships. Dev Biol 2007, 304: 326-337. 125. Khokha MK, Hsu D, Brunet LJ, Dionne MS, Harland RM: Gremlin is the BMP antagonist required for maintenance of Shh and Fgf signals during limb patterning. Nat Genet 2003, 34: 303-307. 126. Smith A, Avaron F, Guay D, Padhi BK, Akimenko MA: Inhibition of BMP signaling during zebrafish fin regeneration disrupts fin growth and scleroblast differentiation and function. Dev Biol 2006, 299: 438-454. 127. Wang CK, Omi M, Ferrari D, Cheng HC, Lizarraga G, Chin HJ, Upholt WB, Dealy CN, Kosher RA: Function of BMPs in the apical ectoderm of the developing mouse limb. Dev Biol 2004, 269: 109-122. 128. Binato R, Alvarez Martinez CE, Pizzatti L, Robert B, Abdelhay E: SMAD 8 binding to mice Msx1 basal promoter is required for transcriptional activation. Biochem J 2006, 393: 141-150. 129. Beck CW, Christen B, Barker D, Slack JM: Temporal requirement for bone morphogenetic proteins in regeneration of the tail and limb of Xenopus tadpoles. Mech Dev 2006, 123: 674-688. 130. Kawakami Y, Rodriguez Esteban C, Raya M, Kawakami H, Marti M, Dubova I, Izpisua Belmonte JC: Wnt/beta-catenin signaling regulates vertebrate limb regeneration. Genes Dev 2006, 20: 3232-3237. 131. Lin G, Slack JM: Requirement for Wnt and FGF signaling in Xenopus tadpole tail regeneration. Dev Biol 2008, 316: 323-335.

114

132. Stoick-Cooper CL, Weidinger G, Riehle KJ, Hubbert C, Major MB, Fausto N, Moon RT: Distinct Wnt signaling pathways have opposing roles in appendage regeneration. Development 2007, 134: 479-489. 133. Meyer A: Homology and Homoplasy: The Retention of Genetic Programmes. In Homology. Edited by Brock GK, Cardew G. Chichester, UK: Wiley; 1999: 141-157 134. Yokoyama H, Ide H, Tamura K: FGF-10 stimulates limb regeneration ability in Xenopus laevis . Dev Biol 2001, 233: 72-79. 135. Taylor GP, Anderson R, Reginelli AD, Muneoka K: FGF-2 induces regeneration of the chick limb bud. Dev Biol 1994, 163: 282-284. 136. Kostakopoulou K, Vogel A, Brickell P, Tickle C: 'Regeneration' of wing bud stumps of chick embryos and reactivation of Msx-1 and Shh expression in response to FGF-4 and ridge signals. Mech Dev 1996, 55: 119-131. 137. Mercader N: Early steps of paired fin development in zebrafish compared with tetrapod limb development. Dev Growth Differ 2007, 49: 421-437. 138. Kuraku S, Usuda R, Kuratani S: Comprehensive survey of carapacial ridge- specific genes in turtle implies co-option of some regulatory genes in carapace evolution. Evol Dev 2005, 7: 3-17. 139. Abramoff MD, Magelhaes PJ, Ram SJ: Image Processing with ImageJ. Biophoton Int 2004, 11: 36-42. 140. Larget B, Simon DL: Markov chain Monte Carlo algorithms for the Bayesian analysis of phylogenetic trees. Mol Biol Evol 1999, 16: 750-759. 141. Guindon S, Gascuel O: A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 2003, 52: 696-704. 142. Posada D, Crandall KA: MODELTEST: testing the model of DNA substitution. Bioinformatics 1998, 14: 817-818. 143. Felsenstein J: Confidence limits on phylogenies: an approach using the bootstrap. Evolution 1985, 39: 783-791. 144. Huelsenbeck JP, Ronquist F: MRBAYES: Bayesian inference of phylogeny. Bioinformatics 2001, 17: 754-755. 145. Rodriguez F, Oliver JL, Marin A, Medina JR: The general stochastic model of nucleotide substitution. J Theor Biol 1990, 142: 485-501. 146. Tamura K, Nei M: Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Mol Biol Evol 1993, 10: 512-526. 147. Gattiker A, Gasteiger E, Bairoch A: ScanProsite: a reference implementation of a PROSITE scanning tool. Appl Bioinformatics 2002, 1: 107-108. 148. Offen N, Blum N, Meyer A, Begemann G: Fgfr1 signalling in the development of a sexually selected trait in vertebrates, the sword of swordtail fish. BMC Dev Biol 2008, 8. 149. Petersen S, Petersen I: Expression profiling of lung cancer based on suppression subtraction hybridization (SSH). Methods Mol Med 2003, 75: 189- 207. 150. Padhi BK, Joly L, Tellis P, Smith A, Nanjappa P, Chevrette M, Ekker M, Akimenko MA: Screen for genes differentially expressed during regeneration of the zebrafish caudal fin. Dev Dyn 2004, 231: 527-541.

115

151. Gebicke-Haerter P: Expression profiling methods used in drug abuse research. Addict Biol 2005, 10: 37-46. 152. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol 1990, 215: 403-410. 153. Bechara IJ, Joazeiro PP, Mari-Beffa M, Becerra J, Montes GS: Collagen- affecting drugs impair regeneration of teleost tail fins. J Submicrosc Cytol Pathol 2000, 32: 273-280. 154. Yamada S, Wirtz D, Coulombe PA: Pairwise assembly determines the intrinsic potential for self-organization and mechanical properties of keratin filaments. Mol Biol Cell 2002, 13: 382-391. 155. Lu HC, Swindell EC, Sierralta WD, Eichele G, Thaller C: Evidence for a role of protein kinase C in FGF signal transduction in the developing chick limb bud. Development 2001, 128: 2451-2460. 156. McCahill A, Warwicker J, Bolger GB, Houslay MD, Yarwood SJ: The RACK1 scaffold protein: a dynamic cog in cell response mechanisms. Mol Pharmacol 2002, 62: 1261-1273. 157. Zakrzewicz A, Hecker M, Marsh LM, Kwapiszewska G, Nejman B, Long L, Seeger W, Schermuly RT, Morrell NW, Morty RE, Eickelberg O: Receptor for activated C-kinase 1, a novel interaction partner of type II bone morphogenetic protein receptor, regulates smooth muscle cell proliferation in pulmonary arterial hypertension. Circulation 2007, 115: 2957-2968. 158. Atkins GB, Jain MK: Role of Kruppel-like transcription factors in endothelial biology. Circ Res 2007, 100: 1686-1695. 159. Oates AC, Pratt SJ, Vail B, Yan Y, Ho RK, Johnson SL, Postlethwait JH, Zon LI: The zebrafish klf gene family. Blood 2001, 98: 1792-1801. 160. Bhattacharya R, Senbanerjee S, Lin Z, Mir S, Hamik A, Wang P, Mukherjee P, Mukhopadhyay D, Jain MK: Inhibition of vascular permeability factor/vascular endothelial growth factor-mediated angiogenesis by the Kruppel-like factor KLF2 . J Biol Chem 2005, 280: 28848-28851. 161. Lee JS, Yu Q, Shin JT, Sebzda E, Bertozzi C, Chen M, Mericko P, Stadtfeld M, Zhou D, Cheng L, et al: Klf2 is an essential regulator of vascular hemodynamic forces in vivo. Dev Cell 2006, 11: 845-857. 162. Kinney CM, Chandrasekharan UM, Mavrakis L, DiCorleto PE: VEGF and thrombin induce MKP-1 through distinct signaling pathways: role for MKP- 1 in endothelial cell migration. Am J Physiol Cell Physiol 2008, 294: C241-250. 163. Bayliss PE, Bellavance KL, Whitehead GG, Abrams JM, Aegerter S, Robbins HS, Cowan DB, Keating MT, O'Reilly T, Wood JM, et al: Chemical modulation of receptor signaling inhibits regenerative angiogenesis in adult zebrafish. Nat Chem Biol 2006, 2: 265-273. 164. Caunt CJ, Rivers CA, Conway-Campbell BL, Norman MR, McArdle CA: Epidermal growth factor receptor and protein kinase C signaling to ERK2: spatiotemporal regulation of ERK2 by dual specificity phosphatases. J Biol Chem 2008, 283: 6241-6252. 165. Teng CH, Huang WN, Meng TC: Several dual specificity phosphatases coordinate to control the magnitude and duration of JNK activation in signaling response to oxidative stress. J Biol Chem 2007, 282: 28395-28407.

116

166. Chakroborty D, Chowdhury UR, Sarkar C, Baral R, Dasgupta PS, Basu S: Dopamine regulates endothelial progenitor cell mobilization from mouse bone marrow in tumor vascularization. J Clin Invest 2008, 118: 1380-1389. 167. Roth LW, Bormann P, Bonnet A, Reinhard E: beta-thymosin is required for axonal tract formation in developing zebrafish brain. Development 1999, 126: 1365-1374. 168. Malinda KM, Goldstein AL, Kleinman HK: Thymosin beta 4 stimulates directional migration of human umbilical vein endothelial cells. Faseb J 1997, 11: 474-481. 169. Malinda KM, Sidhu GS, Mani H, Banaudha K, Maheshwari RK, Goldstein AL, Kleinman HK: Thymosin beta4 accelerates wound healing. J Invest Dermatol 1999, 113: 364-368. 170. Liu JM, Lawrence F, Kovacevic M, Bignon J, Papadimitriou E, Lallemand JY, Katsoris P, Potier P, Fromes Y, Wdzieczak-Bakala J: The tetrapeptide AcSDKP, an inhibitor of primitive hematopoietic cell proliferation, induces angiogenesis in vitro and in vivo. Blood 2003, 101: 3014-3020. 171. Goldstein AL, Hannappel E, Kleinman HK: Thymosin beta4: actin- sequestering protein moonlights to repair injured tissues. Trends Mol Med 2005, 11: 421-429. 172. Yialouris PP, Coles B, Tsitsiloni O, Schmid B, Howell S, Aitken A, Voelter W, Haritos AA: The complete sequences of trout ( Salmo gairdneri ) thymosin beta 11 and its homologue thymosin beta 12. Biochem J 1992, 283 (Pt 2): 385-389. 173. Huang WQ, Wang QR: Bone marrow endothelial cells secrete thymosin beta4 and AcSDKP. Exp Hematol 2001, 29: 12-18. 174. Sambrook J, Fritsch EF, Maniatis T: Molecular Cloning: A Laboratory Manual. 2nd edn. New York: Cold Spring Harbor Laboratory Press; 1989. 175. Volff JN: Genome evolution and biodiversity in teleost fish. Heredity 2005, 94: 280-294. 176. Stolting KN, Wilson AB: Male pregnancy in seahorses and pipefish: beyond the mammalian model. Bioessays 2007, 29: 884-896. 177. Pickford GE, Atz JW: The physiology of the pituitary gland of fishes. In Book The physiology of the pituitary gland of fishes (Editor ed.^eds.). pp. 613 pp. City: New York Zoological Society; 1957:613 pp. 178. Chazaud C, Chambon P, Dolle P: Retinoic acid is required in the mouse embryo for left-right asymmetry determination and heart morphogenesis. Development 1999, 126: 2589-2596. 179. Duester G: Retinoic Acid Regulation of the Somitogenesis Clock. Birth Defects Res C Embryo Today 2007, 81: 84-92. 180. Maden M, Corcoran J: Role of thyroid hormone and retinoid receptors in the homeotic transformation of tails into limbs in frogs. Dev Genet 1996, 19: 85- 93. 181. Thompson JD, Higgins DG, Gibson TJ: CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 1994, 22: 4673-4680.

117

182. Hasegawa M, Kishino H, Yano T: Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. J Mol Evol 1985, 22: 160-174. 183. Yang H, Hazlewood L, Heater SJ, Guerrero PA, Walter RB, Tiersch TR: Production of F1 interspecies hybrid offspring with cryopreserved sperm from a live-bearing fish, the swordtail Xiphophorus helleri . Biol Reprod 2007, 76: 401-406. 184. Douard V, Robinson-Rechavi M, Laudet V, Guiguen Y: Molecular evolution of angrogen receptors in fish. Fish Physiol Biochem 2003, 28: 207-208. 185. Thoms SD, Stocum DL: Retinoic acid-induced pattern duplication in regenerating urodele limbs. Dev Biol 1984, 103: 319-328. 186. Chithalen JV, Luu L, Petkovich M, Jones G: HPLC-MS/MS analysis of the products generated from all-trans-retinoic acid using recombinant human CYP26A. J Lipid Res 2002, 43: 1133-1142. 187. Nakanishi M, Tomaru Y, Miura H, Hayashizaki Y, Suzuki M: Identification of transcriptional regulatory cascades in retinoic acid-induced growth arrest of HepG2 cells. Nucleic Acids Res 2008, 36: 3443-3454. 188. Lai L, Bohnsack BL, Niederreither K, Hirschi KK: Retinoic acid regulates endothelial cell proliferation during vasculogenesis. Development 2003, 130: 6465-6474. 189. Dubrulle J, Pourquie O: Coupling segmentation to axis formation. Development 2004, 131: 5783-5793. 190. Moreno TA, Kintner C: Regulation of segmental patterning by retinoic acid signaling during Xenopus somitogenesis. Dev Cell 2004, 6: 205-218. 191. Niederreither K, Subbarayan V, Dolle P, Chambon P: Embryonic retinoic acid synthesis is essential for early mouse post-implantation development. Nat Genet 1999, 21: 444-448. 192. Yashiro K, Zhao X, Uehara M, Yamashita K, Nishijima M, Nishino J, Saijoh Y, Sakai Y, Hamada H: Regulation of retinoic acid distribution is required for proximodistal patterning and outgrowth of the developing mouse limb. Dev Cell 2004, 6: 411-422. 193. Abe G, Ide H, Tamura K: Function of FGF signaling in the developmental process of the median fin fold in zebrafish. Dev Biol 2007, 304: 355-366. 194. Galdones E, Lohnes D, Hales BF: Role of retinoic acid receptors alpha1 and gamma in the response of murine limbs to retinol in vitro. Birth Defects Res A Clin Mol Teratol 2006, 76: 39-45. 195. Pecorino LT, Entwistle A, Brockes JP: Activation of a single retinoic acid receptor isoform mediates proximodistal respecification. Curr Biol 1996, 6: 563-569. 196. Lohnes D, Kastner P, Dierich A, Mark M, LeMeur M, Chambon P: Function of retinoic acid receptor gamma in the mouse. Cell 1993, 73: 643-658. 197. Lohnes D, Mark M, Mendelsohn C, Dolle P, Dierich A, Gorry P, Gansmuller A, Chambon P: Function of the retinoic acid receptors (RARs) during development (I). Craniofacial and skeletal abnormalities in RAR double mutants. Development 1994, 120: 2723-2748. 198. White JA, Boffa MB, Jones B, Petkovich M: A zebrafish retinoic acid receptor expressed in the regenerating caudal fin. Development 1994, 120: 1861-1872.

118

199. Geraudie J, Monnot MJ, Ridet A, Thorogood P, Ferretti P: Is exogenous retinoic acid necessary to alter positional information during regeneration of the fin in zebrafish? Prog Clin Biol Res 1993, 383: 803-814. 200. Sachidanandan C, Yeh JR, Peterson QP, Peterson RT: Identification of a novel retinoid by small molecule screening with zebrafish embryos. PLoS ONE 2008, 3: e1947. 201. Trasino SE, Harrison EH, Wang TT: Androgen regulation of aldehyde dehydrogenase 1A3 (ALDH1A3) in the androgen-responsive human prostate cancer cell line LNCaP. Exp Biol Med (Maywood) 2007, 232: 762-771. 202. Huang HF, Li MT, Von Hagen S, Zhang YF, Irwin RJ: Androgen modulation of the messenger ribonucleic acid of retinoic acid receptors in the prostate, seminal vesicles, and kidney in the rat. Endocrinology 1997, 138: 553-559.

119

Appendix

Figure S1.1. Sequence and domain structure of X. helleri fgfr1. The 1248 bp fgfr1 sequence from X. helleri (A) codes for parts of the IG domain 2 (blue), IG domain 3 (blue) and parts of the tyrosine receptor kinase (red). B shows a schematic drawing of the isolated cDNA fragment and the domain-coding portions.

120

Figure S1.2. Sequence and domain structure of X. helleri fgfr24 and fgf20a. The 633 bp fgf24 sequence from X. helleri represents the full ORF of the gene, while the 663 bp sequence of X. helleri fgf20a misses a part of the 5’ region of the ORF (A). The Heparin-binding growth factors/fibroblast growth factor (HBGF/FGF) family signature is marked in blue. B shows a schematic drawing of the two isolated cDNA fragment and their HBGF/FGF-coding portions.

121

Table S1.1: Accession Numbers of sequences used for the phylogenetic trees . cDNA Danio rerio Tetraodon nigroviridis Takifugu rubripes Homo sapiens fgfr1 AF389400 GSTENT00034863001 SINFRUT00000128473 NM_000604 fgfr2 NM_178303 GSTENT00025098001 NM_000141 fgfr3 NM_131606 GSTENT00023332001 SINFRUT00000174828 M58051 fgfr4 NM_131430 GSTENT00014622001 SINFRUT00000143771 NM_213647 fgf8 NM_131281 scaf13770 (47766 to 43187) scaffold_52 (147851 to 151899) AF520763 fgf17b BC083269 SCAF7880 (181 to 1923) scaffold_4659 (4421 to 2507) fgf17 AY358869 fgf18 NM_001013264 SCAF7783 (655 to 6588) scaffold_362 (111038 to 104939) AY358811 fgf24 NM_182871 SCAF14744 (1671104 to scaffold_70 (668977 to 661812) 1664185) fgf10 BC094986 fgf20a NM_001037103 GSTENT00035824001 SINFRUT00000135793 fgf20b NM_001039172 GSTENT00005561001 SINFRUT00000137831 fgf20 NM_019851 fgf16 NM_001040407 GSTENT00017034001 SINFRUT00000152407 NM_003868 fgf9 BC103979

Mus musculus Gallus gallus Rattus norvegicus Ciona intestinalis fgfr1 BC033447 NM_205510 fgfr2 NM_010207 NM_205319 fgfr3 NM_008010 NM_205509 fgfr4 BC033313 fgfr5 AF321300 fgf8 BC048734 NM_001012767 fgf17 NM_008004 fgf18 NM_008005 NM_204714 fgf20 BC125509 ENSGALT00000022223 fgf16 AB049219 NM_001044650 fgf9 AB108842 NM_012952 fgf9/16/20 NM_001032477

Table S2.1: Primer used for RT PCR and probe sequence amplification Primer locus Primer sequence

14.3.3 -fw 14.3.3 5'-GTG TGT CAA GCT CTG CGA TG-3' 14.3.3 -rev 14.3.3 5'-CTG CAA TGA CGT TCT GGG TC-3' dsP -1-fw dual specific phosphatase 1 5'-GGA CCA CCC TGA TCA TAT AG-3' dsP -1-rev dual specific phosphatase 1 5'-GGA CGA ACG GAG CAC AGA C-3' calpain -fw m-calpain 5'-GCT GCT GGT AGA TGG TGT TG-3' calpain -rev m-calpain 5'-CCT GTT CAC TAA ACT CGC GG-3' c-fos -fw c-fos 5'-CAG CAT GCA CCA CCT ACA CG-3' c-fos -rev c-fos 5'-CCA TAG CCC TGT AAT CTG CAC-3' gapdh -fw gapdh 5'-CAA CGC TGG CGC CAA ATA CG-3' gapdh -rev gapdh 5'-GAA TGA CTT TGC CCA CAG CC-3' KLF2 -fw krueppel like factor 2 5'-CCT GCT TGT TTG CCG GCT GC-3' KLF2 -rev krueppel like factor 2 5'-GAA TTT GCC CAC ATG GAG AGC-3' rack -fw receptor for activated PKC 5'-GTG TCT TGT GTT CGC TCC TC-3' rack -rev receptor for activated PKC 5'-CCT TGC TGT TTG TGC TGA TC-3' Thy -fw thymosin beta a-like 5'-GGC GAC AAT CAA CCC AGT C-3' Thy -rev thymosin beta a-like 5'-CGT GGC AGA CAG CAG TAA AG-3' lib -universal* λ-phage vector primer 5'-CAC TAT AGG GCG AAT TGG CTA CCG-3'

*: Primer binds to the λ-phage vector DNA downstream of the 3’ UTR of the cloned insert and was used to amplify a 3’ fragment of the cDNA insert for probe synthesis

122

Figure S2.1. Caudal, anal and regenerating fins hybridised with sense probe. To confirm the observed expression pattern, treated caudal and anal fins and rgenerating caudal fins were hybridised with sense probes for rack1 , dusp1 , klf2 and tmsb a-like . Neither treated caudal (A, B, C, D), nor treated anal fins (E, F, G, H), nor regenerating caudal fins (I, J, K, L) showed a specific signal, when hybridised with sense probes. (n= 3 for every fin and probe; scale bars: 200 µm).

123

Table S3.1: Correlation between segment size before and after treatment Ray n Pearson correlation (r) Probability one tail two tail 3 28 -0,0721 0,35 77 0,7154 4a 29 -0,0206 0,4578 0,9155 4p 29 -0,2026 0,1459 0,2919 5a 29 0,1228 0,2628 0,5257

Table S3.2: Correlation between Segment number before treatment and new segment size Ray n Spearman’s rank Probability correlation (r) one tail two tail 3 28 0,3168 0,0505 0,1011 4a 29 0,1122 0,2800 0,5601 4p 29 0,0757 0,3498 0,6996 5a 29 0,2003 0,1493 0,2985

124