Sex Determination in the Squalius alburnoides Complex: An Initial Characterization of Sex Cascade Elements in the Context of a Hybrid Polyploid Genome

Background Sex determination processes vary widely among different vertebrate taxa, but no group offers as much diversity for the study of the evolution of sex determination as teleost fish. However, the knowledge about sex determination gene cascades is scarce in this species-rich group and further difficulties arise when considering hybrid fish taxa, in which mechanisms exhibited by parental species are often disrupted. Even though hybridisation is frequent among teleosts, gene based approaches on sex determination have seldom been conducted in hybrid fish. The hybrid polyploid complex of Squalius alburnoides was used as a model to address this question. Methodology/Principal Findings We have initiated the isolation and characterization of regulatory elements (dmrt1, wt1, dax1 and figla) potentially involved in sex determination in S. alburnoides and in the parental species S. pyrenaicus and analysed their expression patterns by in situ hybridisation. In adults, an overall conservation in the cellular localization of the gene transcripts was observed between the hybrids and parental species. Some novel features emerged, such as dmrt1 expression in adult ovaries, and the non-dimorphic expression of figla, an ovarian marker in other species, in gonads of both sexes in S. alburnoides and S. pyrenaicus. The potential contribution of each gene to the sex determination process was assessed based on the timing and location of expression. Dmrt1 and wt1 transcripts were found at early stages of male development in S. alburnoides and are most likely implicated in the process of gonad development. Conclusions/Significance For the first time in the study of this hybrid complex, it was possible to directly compare the gene expression patterns between the bisexual parental species and the various hybrid forms, for an extended set of genes. The contribution of these genes to gonad integrity maintenance and functionality is apparently unaltered in the hybrids, suggesting that no abrupt shifts in gene expression occurred as a result of hybridisation.


Introduction
The genetic basis of sex determination and gonad development processes has been intensively studied in several animal groups and strict regulatory networks have been revealed in birds and mammals [1]. In fish a wide spectrum of reproductive systems is present, ranging from hermaphroditism to gonochorism and a frequent switch between mechanisms as diverse as environmental or strictly genetic based sex determination can be observed [2][3][4]. Among non-model fish, groups that have arisen through hybridisation and polyploidy offer additional challenges, as they often present reproductive and sex distribution alterations that imply an even broader diversity of regulatory processes [5]. Polyploidisation processes, as postulated by Ohno [6], have been important drives for vertebrate evolution [7,8] and a particular relevant basis underlying speciation and diversity of teleost fish [9]. However, little is known about how sex determination cascades are regulated in hybrid fish species. In fish, evolutionary proximity is no guarantee of similarity in sex determination mechanisms. Hybridisation, even between closely related species [10], frequently brings about an abrupt shift in the regulatory networks, as a consequence of the interaction between distinct parental gene hierarchies [11]. As an emblematic example of such a deviation, the Squalius alburnoides complex of hybrid cyprinid fish presents unique features that make it an interesting model to address the question of sex determination in a hybrid context. This complex is endemic from the Iberian Peninsula and resulted from interspecific hybridisation between females of Squalius pyrenaicus (P genome) and males of an unknown species related to Anaecypris hispanica (A genome). The complex includes individuals of various ploidy levels that intercross through highly diverse reproductive modes, ranging from clonal inheritance to normal meiosis, hybridogenesis or meiotic hybridogenesis (in which one genome is excluded from gamete formation) [12]. Sex ratio distribution is clearly altered in the S. alburnoides complex, with a strong bias towards triploid females of PAA genotype. The complex also includes a lineage of individuals of AA genotype, apparently composed exclusively of males [12]. The strong correlation between genomic constitution and sex is an interesting feature to explore in terms of the sex determination process in these fish. Until now, no sex chromosomes that would indicate a chromosomal based sex determination have been identified, although chromosome heteromorphism that would suggest a ZWfemale/ZZmale system has been reported for the maternal ancestor S. pyrenaicus [13]. Data from experimental crosses revealed that such a system could not fully explain sex determination in S. alburnoides and that in addition to female determinants on the W chromosome, a minimum of one non-Wlinked gene would have to be expressed differently in hybrid and nonhybrid genome combinations to account for the results obtained. Moreover, the ''strength'' of these sex-determining factors might be variable depending on populations and parental species [14].
So far, only one sex determination gene (the anti-Müllerian hormone, amh) has been isolated in S. alburnoides, and its expression characterized both during development and in the adult [15]. To fully explore the possibility of a genetically based sex determination mechanism, an initial set of possible candidates has to be established. It has been shown that despite the variety of mechanisms, the components of sex determination cascades are apparently conserved throughout the whole vertebrate lineage [16][17][18], although their position in gene hierarchies and their interaction patterns can vary according to group. The study of an initial set of conserved genes in the S. alburnoides complex could then constitute the starting point towards a broader understanding of the impact of abrupt genomic remodeling processes such as hybridisation on gene hierarchies and more broadly elucidate their putative roles within the fish lineage.
The Dmrt1 (doublesex and mab-3 related transcription factor 1) gene is part of a gene family sharing a common zinc-finger DNA binding motif (the DM domain). The DM domain was initially identified in genes that occupy key positions in sex determination pathways of the fruit fly Drosophila melanogaster (the doublesex gene) [19] and in the nematode Caenorhabditis elegans (mab-3 gene) [20], but Dmrt1 homologues have been shown to be expressed mainly in the adult and developing gonads, and correlated to sex determination cascades in a number of species [21]. In humans, although acting downstream in the sex determination cascade, DMRT1 has been implicated in some types of XY sex reversal [22]. In birds, its location on the Z chromosome [23] and the higher dosage in males [16,24,25] make it a very good candidate for the male sex determining gene. In fish, it has been implicated in male sex determination pathways in the Nile tilapia Oreochromis niloticus [26] and rainbow trout Oncorhynchus mykiss [27], and shown to contribute to both male and female gonad development in zebrafish [28]. Most prominently, in the medaka Oryzias latipes, a dmrt1 duplicate on the Y chromosome (dmrt1bY/DMY) [29,30] has been identified as the master male sex determination gene in this species.
The Wilm's tumor suppressor gene (Wt1) is a key regulator of urogenital development. The gene encodes a nuclear protein containing four zinc fingers, acting both as a transcription factor and in RNA processing. In non-mammalian vertebrates two isoforms occur, resulting from alternative splicing [31]. The most important functional difference within all isoforms results from the insertion of three additional amino acids (KTS) by splicing events: the WT1 (-KTS) isoform acts as a transcription factor, while WT1 (+KTS) is mainly involved in RNA processing [32,33]. In the medaka, an additional splice variant of Wt1, wt1a_dE4, with differences in exon 4 is present in both +KTS and -KTS isoforms [34]. The involvement of Wt1 in the differentiation of urogenital structures and the maintenance/proliferation of the cells of the bipotential gonad has been shown in mammals [35], in chicken [36], in reptiles [37] and amphibians [38]. Wt1 orthologs have been isolated in telosts species, such as the medaka O. latipes [39,34], zebrafish D. rerio [40], the Japanese eel Anguilla japonica [41] and the rainbow trout O. mykiss [42]. Considering its role in gonad differentiation a few possible transcriptional targets of Wt1 have been identified, such as Dax1, Sf1 and Gata 4 [43,44].
The third gene, dax1, is a member of the nuclear receptors superfamily, with a ligand-binding domain at the C-terminal and LxxLL motifs, which enable the binding and interaction with potential targets (reviewed in [45]). Dax1 acts as a transcriptional repressor of several genes involved in the development and steroidogenic activity of adrenal and gonadal structures, such as aromatase cyp19 [46], anti-müllerian hormone amh [47], estrogen and androgen receptors [48,49], and most notably was shown to inhibit Sf1-mediated transcriptional transactivation [50]. Dax1 expression is normally restricted to specific tissues, namely the genital ridge and it is largely coexpressed with Sf1, antagonizing its cooperation with WT1 and GATA-4 in the regulation of AMH [51,47]. Dax1 orthologs have been isolated and studied in different groups such as birds [52], reptiles [37,53] and amphibians [54]. In fish, dax1 expression patterns have been investigated both regarding its function in adrenal development (in the zebrafish, D. rerio) [55] and its contribution to gonad development and differentiation in several species, such as the nile tilapia O. niloticus [56], the European sea bass Dicentrarchus labrax [57] and the medaka O. latipes [58].
The fourth gene that is reported to be involved in early sex differentiation of several groups is the Factor in the germline alpha (Figla), a germ cell specific basic helix-loop-helix transcription factor. This factor plays a key role in folliculogenesis, by coordinating the regulation of the zona pellucida genes [59]. Absence of figla expression has been shown to affect female, but not male fertility in mice, suggesting a critical role in female germline development [60].
In a system such as the S. alburnoides complex, in which no genomic information is available, our strategy was to isolate potential candidates for sex determination and assess, based on their conservation, functional characteristics and expression patterns whether they could have a role in the establishment of the phenotypic sex. In the present work we report the isolation of the dmrt1, wt1, dax1 and figla Squalius orthologs, the characterization of their specific features with respect to potential functionality, and the description of their expression in the gonads of the bisexual species and hybrids by in situ hybridisation. Finally, we evaluated the potential for a more extensive analysis of their contribution for male or female gonad development in the S. alburnoides complex. This constitutes a first attempt of comparing gene expression patterns between fish hybrids and one of their parental species, in the pursuit of a better understanding how gene hierarchies underlying sex determination might be affected by the hybridisation process.

Results
Isolation of sexual development genes -structural and phylogenetic analysis dmrt1. Amplification of a partial fragment of the dmrt1 gene was performed using primers DMRT1-F1 D and DMRT1 zf-R3 and sequences from S. pyrenaicus (P genome) and S. alburnoides (A genome) were obtained (Table 1). Both fragments included the specific DM DNA binding domain that characterizes this class of proteins ( Figure S1). The conservation of functional motifs, in comparison to other DMRT1 proteins ( Figure S1) and the phylogenetic proximity to the Dmrt1 genes of other teleosts, revealed by the analysis of different proteins that share the conserved DM domain (Figure 1) confirmed the isolation of the Dmrt1 ortholog in S. pyrenaicus and S. alburnoides.
A two amino acid (Threonine and Aspargine) deletion followed by an amino acid substitution (Leucine by Isoleucine) was observed on the predicted sequence of S. alburnoides (A genome) compared to S. pyrenaicus (P genome), but no relevant differences in the secondary structure predicitions of the protein were found as a consequence of the genome-specific polymorphism.
wt1. Of all primer combinations tested, only the ones involving WT1-F1a and, WT1-R1a and WT1-R1b, respectively (Table 1) resulted in the production of sequence segments that could be identified as wt1. BLAST comparison revealed an overall similarity of all amplified fragments with WT1a transcripts of other teleosts. A larger PCR fragment was amplified with primers WT1 F5 and WT1-R1a (Table 2), using S. pyrenaicus gonad cDNA as template. The closer homology of our isolate to a-form Wt1 proteins of other telelosts was confirmed both in the phylogenetic tree comparing other Wt1 orthologs ( Figure 2) and the alignments of the predicted amino acid sequence ( Figure S2). Three distinct domains were identified in the S. pyrenaicus Wt1 isolate ( Figure S2): a Wilm's tumour specific domain, and two (of the four) zinc finger DNA-binding motifs that characterize the C terminus of this transcription factor. dax1. Fragments corresponding to a partial coding sequence of the dax1 gene were amplified with primers DAX1-F1 and DAX1-R1 using gonad cDNA samples of S. pyrenaicus and S. alburnoides as templates ( Table 1). Alignment of the predicted amino acid sequences revealed 96% identity with the Dax1 protein of zebrafish and 60-64% identity with Dax1 proteins of  other teleosts ( Figure S3). The phylogenetic reconstruction ( Figure 3) revealed a close evolutionary proximity to the Dax1 protein of zebrafish, with both cyprinids grouping with the Dax1 group of proteins of other teleosts. The domain prediction analysis confirmed that the Squalius isolates included functional regions that characterize Dax1 proteins and that are relevant for their role as transcription factors ( Figure S3). The Squalius dax1 gene encodes for two LxxLL motifs, an important feature for mediating protein-protein interactions and a typical characteristic of non-mammalian Dax1 proteins (as opposed to the four LxxLL motifs in mammals) [55]. Furthermore, a conserved ligand-binding domain, characteristic of hormone receptors was predicted within our isolate. Thus, although it corresponds to a partial coding sequence, the dax1 fragment isolated from S. alburnoides and S. pyrenaicus includes all relevant features that characterize this family of proteins.
figla. Fragments of approximately 300 bp corresponding to a putative figla product were obtained using S. alburnoides (AA) and S.pyrenaicus (PP) as templates. Successful amplification was obtained both with male and female gonad samples and no sequence differences were found according to sex. This is a surprising finding, as figla was expected to be an exclusive ovarian marker and the presence of a specific transcript of this gene was only expected in female gonads. The alignment of the predicted amino acid sequences ( Figure S4) revealed a high identity with Figla proteins of several teleosts. A Helix-loop-helix domain was identified ( Figure S4), which is a diagnostic feature of the Figla transcription factor, enabling the conversion of monomers to trans-activating dimmers. The phylogenetic analysis ( Figure 4) revealed a preferential grouping with zebrafish Figla, and a closer evolutionary relatedness to other teleosts' proteins, thus confirming our isolate as a partial sequence of the Squalius Figla ortholog.
Gene expression patterns dmrt1. Expression of dmrt1 was observed in specific cellular locations in males of the bisexual ancestor S. pyrenaicus (PP) and both in nuclear non-hybrid males (AA) and diploid hybrids (PA) of S. alburnoides (Figure 5a, b and c). The location of dmrt1 signal is observed in positions expected for Sertoli cells, as predicted previously from morphological comparison and gonad structural analysis [15]. Since the signal is more diffuse in AA gonads, we cannot exclude that dmrt1 may also be expressed in some peripheral spermatogonia (Figure 5b). Otherwise, no differences were observed between hybrids and parental species in terms of the location of positive sites of dmrt1 expression, although a comparatively stronger signal was usually obtained in PP and AA gonads (Figure 5a and b) compared to PA male hybrids (Figure 5c). Dmrt1 expression was observed in adult ovaries of diploid (PA) and triploid (PAA) hybrid females of S. alburnoides. Expression was confined to the more developed cortical alveolar and yolk vesicle oocytes and was apparently absent from earlier stage oocytes (Figure 5d and e).
Dmrt1 transcripts were detected early during male gonad development in the hybrids of S. alburnoides, suggesting a potential role already at these early stages. The presence of a dmrt1 specific transcript was also detected by RT-PCR at early stages in S. alburnoides AA progeny and its expression in the presumptive region of gonad development was observed by in situ hybridisation, as early as 6 days after hatching (dah) ( Figure S5).    wt1. Wt1 expression in the adult testes of S. pyrenaicus (PP) and S. alburnoides (AA and PA) was observed in an area where Sertoli cells are located. No clear differences between parental and hybrid samples were observed, but again the gonads of the AA genotype appear to have a more extensive wt1 expression, possibly in peripheral spermatogonia or interstitial cells (Figure 5g, h and i). Very low levels of wt1 expression were observed in adult ovaries of hybrid females of S. alburnoides, in oocytes at different stages of development (Figure 5j and k).
A faint signal was detected for wt1 by RT-PCR at early developmental stages. Two distinct bands were obtained from amplification with wt1 primers using embryos as templates, as opposed to the unique band obtained from adult gonads. Expression in male embryos of AA genotype was observed at 4, 6 and 14dah ( Figure S6).
dax1. Dax1 expression was observed in adult gonads of both sexes in the S. alburnoides complex. Expression in males was seen exclusively outside germ cells in a position compatible with expression in Sertoli cells (Figure 5l). In females of S. alburnoides, expression was observed in the adult gonad, in follicle cells surrounding perinuclear stage oocytes (Figure 5m and n), but was absent from cortical alveolar and vitellogenic oocytes.
Dax1 expression was not observed in early S. alburnoides male development at any of the stages analysed ( Figure S7).
Figla. No sexual dimorphic expression of figla was observed in S. alburnoides. Amplification of a figla specific fragment was conducted successfully both in male and female gonad cDNA templates ( Figure S8). Expression was observed in testes in locations compatible with expression in Sertoli cells (Figure 6a). In ovaries figla was highly expressed in late primary oocytes, both in bisexual parental species S. pyrenaicus and in the S. alburnoides hybrids, while only a faint signal was detected in primary oocytes (Figure 6b and c).

Discussion
In the present study, we report the isolation of four genes known to be involved in the sex determination/gonad differentiation pathways of mammals, predicted to play similar roles in fish, but so far not analysed in the S. alburnoides hybrid complex.
Dmrt1 transcripts were found both in male and female adult gonads. The absence of a sexual dimorphic expression pattern is apparently in contrast to the reported expression patterns for the majority of fish species. In the Nile tilapia O. niloticus, dmrt1 expression occurs exclusively in male gonads [26] and the gene is specifically expressed in male fry very early during development, even before any signs of morphological sex differentiation [61]. In the rainbow trout O. mykiss, dmrt1 is also expressed early in testis development, but not in the ovary [27]. In the more closely related cyprinid zebrafish, though, expression patterns seem to be in agreement to what we report for S. alburnoides and S. pyrenaicus: dmrt1 is expressed in the developing gonads of both sexes [28], suggesting that it might play a role in ovary differentiation (as opposed to the classical view of sex dimorphic dmrt1 expression and an exclusive association with testis development). In fact, evidence from various species suggest that timing and levels of dmrt1 expression could be associated to sex-specific differentiation fates, with high expression levels correlating with testis development and low dmrt1 levels being compatible with ovary differentiation [53,[62][63][64]. However, differences emerge when we directly compare the cellular location of dmrt1 expression in the adult gonads of zebrafish and of S. alburnoides. In males of zebrafish, dmrt1 transcripts were identified in the developing germ cells, while in the Squalius samples expression is apparently restricted to Sertoli cell positions. The Sertoli cell lineage specificity and the absence of expression in the germ cell lineage are in accordance with what has been reported for the Nile tilapia [26], Fugu [65] and dmrt1a and dmrt1bY in the Medaka [66]. Conversely, dmrt1 expression in both the Sertoli and the germ cell lineage has been reported in the platyfish Xiphophorus maculatus [67]. Thus, the observed differences could be a consequence of different roles in sexual development or differential gene evolution in fish sublineages. The same could apply for gene expression patterns in females. In zebrafish, dmrt1 gene expression is high in developing perinuclear oocytes and fainter in more mature stages [28], while in S. alburnoides females it is mainly observed in the more developed yolk vesicle oocytes.
We have isolated a single wt1 ortholog in S. pyrenaicus in the present work, despite having tested primer combinations based on the two zebrafish genes (wt1a and wt1b), but we cannot exclude the possibility that a second wt1 gene is present in these Iberian species. In adult males of S. alburnoides and S. pyrenaicus, wt1 is expressed in the Sertoli cell lineage and is apparently absent from germ cells, which is in agreement with what has been reported in other vertebrates [68]. In the medaka, expression of wt1 genes has been observed in somatic cells during development and in the adult gonads. In adult testis, the location of wt1a expression suggested a probable association with germ cell supporting Sertoli cells [34]. Thus, it is possible that wt1 is playing a similar function in the male gonads of the Squalius cyprinids. The apparently lower expression in female gonads could also be related to the role of wt1 in follicle development. In different mammals and birds [69], higher expression in early stage oocytes and a progressive reduction in wt1 expression during follicle maturation implied an additional role of this gene: the repression of genes involved in follicle differentiation, and the maintenance of a number of oocytes that can later on be recruited, in early developmental stages. It is not clear whether or not this is the role that wt1 is playing in S. alburnoides ovary but the fact that most oocytes present in the analysed samples are of later stages of maturation and have low wt1 expression is consistent with that possibility.
The non-dimorphic expression pattern exhibited by dax1 in gonads of adult Squalius is in accordance with what has been described in a number of other teleost species. Non sex specific expression between female and male adult gonads was found in the nile tilapia O. niloticus [56] and throughout the gonadal sex differentiation period in the European sea bass D. labrax [57]. Conversely, in medaka, dax1 mRNA was absent from the adult testis, but expression was detected in postvitellogenic follicles of the adult ovary [58]. The expression pattern of this gene in the Squalius gonads, restricted to Sertoli cells in males and granulosa cells in females, could well correlate with the functions of modulation of the activity of other factors and participation in the maintenance of testis and ovary structure and organization [50]. In fact, it has been proposed that the effects observed in mouse Dax1 mutants, which affect normal granulosa cell organization around oocytes in females, could be functionally related to the defects in Sertoli cell support of germ cells observed in males [70]. Taking into account the expression patterns of dax1 in the same cell types in Squalius, it is possible that in the adult gonads of both sexes, dax1 could also be contributing to the regulation of these somatic cell lineages and thus indirectly affecting gametogenesis.
For all genes, an initial assessment was performed of whether, besides their contribution to regulation in the adult gonads, they might also have a direct role in the earlier process of sex determination. The evaluation of a putative role and thus interest for further analysis would be based on whether they were expressed early during development in the developing gonadal structures. As no sex marker is available for the Squalius species, we selected the putative all-male progeny of AA genotype to perform the test, avoiding the ambiguities that could result from the use of a diversity of individuals of unknown sex. We observed that, in agreement with other teleosts [27,40,61], both dmrt1 and wt1 are expressed in S. alburnoides male progeny, at early stages and in the developing gonad, suggesting a contribution to its differentiation. However, the real participation of wt1 and dmrt1 as sex determination genes can only be assessed when male and female embryos can be analysed in parallel. The same is true for dax1 expression. No dax1 transcripts were observed in the early developmental stages of S. alburnoides males. It is possible that levels of dax1 expression might vary during development, as shown in mouse [71] and that the stages analysed in S. alburnoides could correspond to ''windows'' of dax1 downregulation. However, the absence of dax1 expression at early developmental stages argues against its involvement in male sex determination, without excluding the possibility of a later role in gonad differentiation.
Regarding figla, and its potential use as an early ovary marker in Squalius, the expression results obtained in adult gonads indicate a completely opposite scenario as initially predicted. Female restricted figla expression has been described for a number of species, namely in mammals and fish [60,73], so it was unexpected to isolate a figla transcript from S. pyrenaicus and S. alburnoides adult testis. No differences were found between female and male figla isolates in terms of sequence, domain prediction and putative functionality.
In fish like the gilthead seabream Sparus aurata, sex change occurs at a certain stage of the life cycle, from a heterosexual gonad, with gene expression profiles changing accordingly [73]. Thus, an initial hypothesis would be that gonad maturation in S. alburnoides could also involve a similar process and that figla transcripts detected by RT-PCR could originate from a vestigial female structure. Data from in situ hybridisation did not support this hypothesis. Figla transcripts were detected in Sertoli cell locations in males with normal testicular organization and with no morphological indication of the presence of any female specific gonadal tissues. In females, on the other hand, figla expression apparently fits the expectations. Expression was restricted to cells surrounding early stage oocytes, which is in close agreement with what has been reported in the medaka [74] and consistent with the role of figla in follicle development. A more reasonable possibility to account for non-dimorphic expression would be that figla transcripts would occur both in Squalius males and females and differential expression levels of figla during development would promote male or female differentiation. In fact, a peak in figla expression has been correlated with ovary differentiation in zebrafish [72]. An alternative hypothesis would be that figla female specific expression would only occur during development and that in the adult it would be expressed in cells supporting the germ cell lineage. These possibilities can only be tested more systematically when a straightforward identification of male and female Squalius embryos becomes possible and the expression patterns can be followed in parallel in both sexes. Additionally, it would be of interest to follow expression of figla and of the other sex determination candidates quantitatively, at different stages of gonad development, as timing and expression levels of genes in sex determination cascades have been shown to be critical to promote the switch between male and female determination pathways [62][63][64][65]73].
A more comprehensive analysis of the putative roles and interactions of the isolated candidate genes is currently hampered by the impossibility of unambiguous sexing of embryos. While trying to overcome this difficulty, our aim was to create the framework that could subsequently be extended in terms of further analysis. No significant differences were found regarding the cellular location and the expression patterns of individual genes in gonads of the parental species S. pyrenaicus and in the hybrid S. alburnoides, even in an heterogeneous context of germ cell development, implying that, at least for the set of genes analysed, the normal patterns of gene expression are being resumed even upon hybridisation and that the normal process of cyclical differentiation and maintenance of gonad identity is taking place in hybrids as in parental species. Globally, the gene expression patterns observed in S. alburnoides are apparently in accordance with what has been described for other teleost species, implying that there is a good conservation of players and functions that might ''resist'' genome remodeling events such as hybridisation. Therefore, rather than a definitive answer, the present work has provided the first clues and the initial working basis that can be expanded in the attempt to elucidate a gene-mediated sex determination in S. alburnoides. This then may contribute to a broader understanding of the complex and diverse pathways of sex determination in fish.

Samples
The specimens used in this study were collected from the Tejo River Basin (River Raia), as previously described [15]. A total of 9 gonad samples, 3 of S. pyrenaicus and 6 of S. alburnoides (2 nuclear non-hybrid, 2 diploid, and 2 triploid hybrids) were selected from the global sample already analysed and genotyped using microsatellites [15]. An additional S. pyrenaicus sample was included and processed as described in [15] (a summary of the selected samples is presented in Table 2-a).

RNA extraction and cDNA synthesis
Total RNA was extracted from adult gonads of S. pyrenaicus and S. alburnoides using the TRIZOL reagent (Gibco-BRL) according to the supplier's recommendation. First strand cDNA was synthesized with RevertAid TM First Strand cDNA Synthesis Kit (Fermentas), using random hexamer primer.
Isolation of dmrt1, wt1, dax1 and figla A partial fragment of the dmrt1 gene of S. pyrenaicus (PP) and S. alburnoides (AA) was amplified using degenerate and specific primers ( Table 1), according to conditions previously described [75]. To obtain initial sequence information for the isolation of the wt1 gene, degenerate primers (Table 1 -WT1 F1a; WT1 R1a; WT1 F1b; WT1 R1b; WT1 F5) were designed based on coding sequences of this gene from D. rerio (NM_131046; NM_001039634), A. japonica (AB030741) and O. mykiss (AF334670; AF334671). Amplification of a partial sequence of the dax1 gene of S. pyrenaicus and S. alburnoides was performed using primers based on zebrafish dax1 (EN-SDART00000020212) ( Table 1). Primers for the Figla gene were designed based on zebrafish sequences (NM_198919) ( Table 1). All primers were tested on cDNA samples of S. pyrenaicus (PP) and S. alburnoides (AA and PAA) gonads, according to the following PCR conditions: pre-heating at 96uC for 2 min 30 s, 35 cycles at 96uC for 30 s, 54uC or 56uC (dax1) for 45 s and 72uC for 1 min 15 s and a final extension at 72uC for 10 min. Products corresponding to the A and P genomes were sequenced and analysed using Sequencher ver. 4.0 (Gene Codes Corporation, Inc.).

Cloning
Amplification products of the four genes were cloned into pDrive Cloning Vector (Qiagen), positive colonies were identified by blue/white selection and screened with M13F and M13R universal primers. Positive colonies were randomly picked and sequenced. Sequences were analysed using Sequencher ver. 4.0 (Gene Codes Corporation, Inc.) and compared with sequences corresponding to the P and A genomes, previously isolated from S. pyrenaicus (PP) and S. alburnoides nuclear non-hybrid (AA) males.  [77]. Support values for the observed topologies were generated by 1000 bootstrap replicates.

Protein structure comparison
Secondary structure was predicted, based on the amino acid sequences of the dmrt1 gene fragments of S. pyrenaicus and S. alburnoides using the PSIPRED method [78] as implemented in PSIPRED server [79].
Analysis of candidate gene expression in adult gonads of S. pyrenaicus and S. alburnoides Probe synthesis. Colonies selected after the cloning procedure (representing the P or A genomes) were cultured overnight and purified with aMiniprep Kit (Invitrogen). Sense and antisense DIG-labelled riboprobes were obtained by digestion with HindIII or BamHI (in the case of dmrt1, wt1 and figla), HindIII and KpnI (dax1) and transcription with SP6 and T7 polymerases (DIG RNA Labelling Kit -Roche).
Tissue preparation and sectioning and in situ hybridisation. Testis from adult males of S. pyrenaicus and S. alburnoides (AA and PA) and ovaries of diploid (PA) and triploid (PAA) females of S. alburnoides were fixed and processed according to [80]. Embedded gonads were then frozen and stored at 280uC. Cryosections of 10 mm were obtained and collected on Superfrost Plus slides (VWR). In situ hybridisation was performed as previously described [15].
Collection of embryos and in situ hybridisation. Embryos of different developmental stages and of specific genotype (diploid AA) were selected from the progeny of the experimental Cross I (see [15] for the cross and Table 2-b for information about the selected progeny). In the absence of a molecular sex marker to unambiguously identify sexes in the S. alburnoides complex we relied on the extremely strong correlation between AA genotype and male phenotype, with only one report of a single female individual of AA genotype in 20 years of intensive study of the complex [81]. Thus, a group of samples was selected, for which sex could be inferred (in this case male sex) with a high degree of confidence. Progeny had been fixed, processed, frozen and stored at 280uC according to [80]. Cryosections of 14 mm were obtained and collected on Superfrost Plus slides (VWR). In situ hybridisation was performed as previously described [15].