Genomic Analysis of the Pacific Oyster (Crassostrea gigas) Reveals Possible Conservation of Vertebrate Sex Determination in a Mollusc

Despite the prevalence of sex in animal kingdom, we have only limited understanding of how sex is determined and evolved in many taxa. The mollusc Pacific oyster Crassostrea gigas exhibits complex modes of sexual reproduction that consists of protandric dioecy, sex change, and occasional hermaphroditism. This complex system is controlled by both environmental and genetic factors through unknown molecular mechanisms. In this study, we investigated genes related to sex-determining pathways in C. gigas through transcriptome sequencing and analysis of female and male gonads. Our analysis identified or confirmed novel homologs in the oyster of key sex-determining genes (SoxH or Sry-like and FoxL2) that were thought to be vertebrate-specific. Their expression profile in C. gigas is consistent with conserved roles in sex determination, under a proposed model where a novel testis-determining CgSoxH may serve as a primary regulator, directly or indirectly interacting with a testis-promoting CgDsx and an ovary-promoting CgFoxL2. Our findings plus previous results suggest that key vertebrate sex-determining genes such as Sry and FoxL2 may not be inventions of vertebrates. The presence of such genes in a mollusc with expression profiles consistent with expected roles in sex determination suggest that sex determination may be deeply conserved in animals, despite rapid evolution of the regulatory pathways that in C. gigas may involve both genetic and environmental factors.

organisms from invertebrates to mammals (Matson and Zarkower 2012). Although DM domain genes are deeply conserved in metazoans, upstream sex regulators such as Sry, Sox, and FoxL2 are thought to be recent inventions in vertebrates or placental mammals (Gamble and Zarkower 2012;Matson and Zarkower 2012). In Caenorhabditis elegans, an X chromosome to autosome (X:A) ratio of 0.5 triggers expression of a cascade genes that act on three DM domain genes to initiate male development (Gamble and Zarkower 2012). In Drosophila, sex is also determined by the X:A ratio through alternative splicing of doublesex (dsx), a DM domain gene, into dsx F and dsx M to initiate female and male differentiation, respectively (Hempel and Oliver 2007;Rideout et al. 2010;Robinett et al. 2010).
Mollusca is a major branch of lophotrochozoan where diverse modes of sexual reproduction are observed. In bivalve molluscs, there are dioecious and hermaphroditic species, as well as species capable of sex change, and sex determination can be genetic, environmental, or both (Coe 1943;Guo and Allen 1994;Chavez-Villalba et al. 2011). Oysters in particular have a complex and dynamic sex determination system. Certain species, including many within the genus Ostrea, are functional hermaphrodites, whereas others such as Crassostrea species may exhibit protandric dioecy, hermaphroditism and sex change within the same species (Coe 1943). In the Pacific oyster Crassostrea gigas, most individuals mature as males first and may subsequently change to females, and females may change to males as well (Amemiya 1929). Warm and nutritious conditions may favor the development of females (Quayle 1988). Genetic determinants of sex in oysters also have been demonstrated through analysis of family sex ratios, and genetic models have been proposed without any knowledge of molecular mechanisms of sex determination (Haley 1979;Guo et al. 1998;Hedrick and Hedgecock 2010).
Studies on sex determination genes and pathways in bivalves and other molluscs are few and limited. Several genes homologous to sexdetermining pathway genes in model species have been identified in bivalve molluscs, although their expression profile is often inconsistent with expected roles in sex determination. DM domain genes have been identified in C. gigas, scallop Chlamys farreri, and Chlamys nobilis, but they are not specifically expressed in testis (Naimi et al. 2009a;Feng et al. 2010;Shi et al. 2014). Homologs of Dmrt did show high or specific expression in the pearl oyster Pinctada martensii, the blacklip pearl oyster Pinctada margaritifera, and the abalone Haliotis asinina (Klinbunga et al. 2009;Yu et al. 2011;Teaniniuraitemoana et al. 2014). Sox genes have been reported in C. farreri and C. gigas, but their expression is not restricted to testis (He et al. 2013;Santerre et al. 2014). Cg-FoxL2 and Cg-b-catenin showed high but not specific expression in female gonad (Naimi et al. 2009b;Santerre et al. 2014), although FoxL2 was later found specifically expressed in female gonad of C. gigas and P. margaritifera (Dheilly et al. 2012;Teaniniuraitemoana et al. 2014). Despites the identification of some candidate genes, sexdetermining pathways in molluscs remain elusive. Genome-wide studies have been limited due to a lack of reference genomes (Ghiselli et al. 2012;Dheilly et al. 2012;Teaniniuraitemoana et al. 2014).
The availability of the C. gigas genome (Zhang et al. 2012) provides an opportunity for a comprehensive analysis of sex-determining pathways in this species that has an interesting sex determination system. In this study, we investigated genes related to sex determination by sequencing and analyzing transcriptomes of female and male gonads of C. gigas. Our analysis reveals that several key genes of the vertebrate sex-determining pathway are present in C. gigas with expression profiles consistent with roles in sex determination, and the C. gigas pathway appears more similar to that of vertebrates than to that of worms and flies.

METHODS
To discover genes related to sex-determining pathways, we sequenced transcriptomes of mature gonads from one female (F3) and two male (M2 and M3) Pacific oysters. In addition to the three transcriptomes obtained in this study, we also included gonadal transcriptomes of two females (F1 and F2) and one male (M1) that were generated in our previous study (Zhang et al. 2012). Thus, a total of three female and three male gonadal transcriptomes were included in this analysis. To establish expression profile and infer function of selected genes, we consulted seven somatic transcriptomes from six organs (gill, outer and pallial mantle, adductor muscle, digestive gland, labial palp, and hemocytes) and transcriptomes from 38 developmental stages that were obtained in our previous study (Zhang et al. 2012).
All animals used for gonadal transcriptome sequencing in this and the previous study were mature 2-yr-old oysters obtained from Laodong Aquaculture Breeding Company, Qingdao, China. Sex of oysters was determined by observing the presence of eggs or sperm under microscope. Gonadal tissues were dissected, frozen in liquid nitrogen and stored at 280°before RNA extraction. RNA extraction, cDNA synthesis, library construction, and transcriptome sequencing were performed in the same way as in our previous study (Zhang et al. 2012). In brief, total RNA was extracted using TRIzol reagents and protocol (Invitrogen). Poly-A RNA was isolated with oligo-dT-coupled beads and sheared for first strand cDNA synthesis with random hexamers and Superscript II reverse transcriptase (Invitrogen). The second strand was synthesized with Escherichia coli DNA PolI (Invitrogen). Double-stranded cDNA was purified with the Qiaquick PCR purification kit (QIAGEN, Gaithersburg, MD). After end repair and addition of a 39 dA overhang, the cDNA was ligated to Illumina adapters and size selected to about 200 bp by gel purification. The selected fragments were amplified for 15 polymerase chain reaction cycles and sequenced for 49 bp single-end reads using Illumina sequencing platform. After cleaning, RNA-seq reads were mapped to the oyster genome with Tophat (Trapnell et al. 2009). Expression level for each gene was measured with reads per kilobase per million (RPKM). Only genes with a RPKM value larger than 3 in at least 1 of the 13 transcriptomes (6 gonad and 7 somatic) were used for further analysis.
Ovary-specific genes were identified by comparing expression in three female gonadal transcriptomes with that in seven somatic and three testis transcriptomes and vice versa for testis-specific genes. To qualify as ovary or testis-specific genes, genes should satisfy the following criteria: 1) All three ovary or testis samples had greater RPKM values than any of the other 10 samples; 2) The RPKM mean of the three ovary or testis samples should be at least fivefold of that in other 10 samples. For a general characterization of ovary-and testisspecific genes, Gene Ontology (GO) term enrichment analysis was conducted with the Fisher's exact test in classic algorithm of topGO (Alexa and Rahnenfuhrer 2010). KEGG enrichment was conducted with the algorithm implemented in GOstats (Falcon and Gentleman 2007). The Benjamini-Hochberg false discovery rate control was used to adjust the P value (Benjamini and Hochberg 1995).
Genes belonging to sex-determining pathways in model species (see the Results) were collected from the literature and searched against the oyster genome and our transcriptome dataset. Homologs were identified and their expression profiles in ovary, testis, somatic tissues, and at various developmental stages were extracted to infer function in C. gigas. Although the published oyster genome provided annotation for most genes, genes of interest were manually checked with ESTs and RNA-seq reads for possible assembly errors and characterized by searching against databases, including InterPro (Hunter et al. 2009), GO (Ashburner et al. 2000), SWISS-PROT (Magrane and Consortium 2011), TrEMBL (Magrane and Consortium 2011), and KEGG (Kanehisa and Goto 2000). Domain structures of selected genes were determined using SMART (http://smart.embl-heidelberg.de/). To confirm homology of selected genes, conserved domain sequences were identified, aligned, and compared. Unrooted maximum likelihood phylogenetic trees of protein sequences were constructed with RAxML (Stamatakis 2014), with the model PROTGAMMAJTT and bootstrap for 1000 replications. Genes in the oyster genome are coded by a prefix of "CGI_100" plus a five-digit unique gene number and for brevity especially in figures, the prefix of "CGI_100" was replaced with "Cg."

RESULTS AND DISCUSSION
Ovary-and testis-specific genes Comparative analyses of RNA-seq data between gonad and somatic organs identified 621 genes that were specifically expressed in female gonad, and 552 specifically expressed in male gonad (Supporting Information, File S1). The finding of more ovary-specific than testisspecific genes is in agreement with the finding of 1570 ovary-enriched and 1370 testis-enriched genes in zebrafish (Santos et al. 2007), but differs from other studies where more genes are found in testis (Rinn et al. 2004;Small et al. 2009). Analyses of these ovary-and testisspecific genes reveal some similarities and interesting differences between two sexes. Genes related to meiosis are overrepresented in both ovary-and testis-specific genes. Among ovary-specific genes, GO terms related to DNA replication, nucleic acid metabolism, DNA repair, chromosome organization, cell cycle, gene expression regulation, DNA recombination, and telomere maintenance are significantly enriched (Table 1). Among testis-specific genes, GO terms related to protein phosphorylation and dephosphorylation, protein metabolism, and sex differentiation are significantly enriched. KEGG enrichment analysis shows similar patterns that pathways related to DNA replication, DNA repair, nucleotide metabolism, recombination, oocyte maturation, cell cycle, and proteolysis are enriched in ovary-specific genes, whereas KEGG pathways related to protein digestion and absorption, protein interaction, flagellar assembly, proteolysis, and focal adhesion are enriched in testis-specific genes (Table 1). These findings suggest that ovary-specific genes are more likely involved in DNA replication, DNA metabolism, DNA repair, DNA organization, and DNA transcription, whereas testis-specific genes are primarily involved in protein phosphorylation, protein interactions and protein metabolism.
Protein phosphorylation is important for sperm capacitation in mammals (Visconti and Kopf 1998). In mammalian testis, abnormal or damaged spermatozoa are ubiquitinated for proteolytic destruction (Sutovsky et al. 2002;Purdy 2008). The fact that ubiquitin mediated proteolysis is enriched among testis-specific genes suggests ubiquitination may play a role in maintaining sperm quality in oysters also. GO terms related to cellular component movement, RNA metabolism, and response to stress are enriched in ovary-specific genes, suggesting these pathways are also important to ovary biology. Overall, it seems that ovary-specific genes are enriched for more diverse pathways than testis-specific genes (Table 1), which may indicate that the production of large and yolk-containing oocytes involves more metabolic pathways than the production of sperm.

Sex-determining pathway genes
Clearly, not all ovary-and testis-specific genes are involved in sexdetermining pathways. To identify genes related to sex-determining pathways, we searched the oyster genome for such genes previously identified in model organisms and examined their expression profiles in gonadal transcriptomes of C. gigas. Of the 24 genes examined, homologs were found for nine genes or gene families (Table 2). Among these nine genes, only three genes (CgDsx, SoxH or Sry-like and FoxL2) showed sex-specific expression as expected for sex-determining genes. The other six genes (Fem, Gata4, Wnt4, beta-catenin, Run, Sox9) did not show sex-specific expression in our samples ( Table 2), suggesting that they may not be involved in sex determination or maintenance in mature gonads. We cannot rule out the possibility that they may have sex-specific expression at an earlier stage.
Doublesex and MAB-3 related transcription factor 1 (Dmrt1): Dmrt1 is a transcription factor that contains a characteristic zinc finger DM domain and plays deeply conserved roles in sex determination and differentiation (Kopp 2012). Members of the family include the doublesex (dsx) gene in fruit fly, MAB-3 in C. elegans and the Dmrt1 in vertebrates, all of which promote male-specific development. In Drosophila melanogaster, dsx is alternatively spliced to produce maleand female-specific isoforms in male and female gonads, respectively (Burtis and Baker 1989). MAB-3 in C. elegans and Dmrt1 in vertebrates are exclusively expressed in testis and promote male-specific development (Yi and Zarkower 1999;Smith et al. 2009;Kopp 2012). The oyster genome encodes three DM domain containing genes: Cg19568, Cg01830, and Cg15952, compared with 11 in nematodes, 4 in flies, and 7 in human (Matson and Zarkower 2012). We named one of the three genes, Cg19568, as CgDsx (GenBank accession No. KJ489413) after manual correction to remove three misassembled exons. It contains a DM domain showing closest homology (45%, E-value = 9e-13) to Dsx isoform A found in D. melanogaster ( Figure 1A). CgDsx has three exons, but it shows no sex-specific alternative splicing as described in D. melanogaster. Similar to MAB-3 from C. elegans, CgDsx does not contain any other recognizable c-terminal domains that are found in Dsx of Drosophila or in vertebrate Dmrt1 ( Figure 1B).
CgDsx is exclusively expressed in gonads with virtually no expression in somatic organs (0.2 RPKM). Its expression in testis (15.8 RPKM) is 5.3-fold greater than that in ovary (3 RPKM) ( Figure 1C). The high expression in testis supports a possible role for CgDsx in determining or promoting male-specific development. The finding of low levels of CgDsx transcripts in female gonad may indicate that dormant male germline cells exist in female gonad, permitting sex change in coming seasons. Oyster gonads are known to contain both male and female germline cells possibly as a prerequisite for sex change (Cole 1941;Guo et al. 1998).
The other two DM domain genes (Cg15952 and Cg01830) both have a DmrtA domain c-terminal to the DM domain and show the highest homology to DmrtA2 (aka Dmrt5) from many species (identity 38-83%, E-value , e-40). They are expressed in all organs and appeared to be unrelated to sex determination or differentiation. Cg15952 has been previously identified as Cg-DM1 and shown to be related to gonad development by Naimi et al. (2009a). In our data, however, Cg-DM1 is primarily expressed in gill, labial palp, and mantle (61.5 RPKM) rather than in female (5.3 RPKM) or male (19.6 RPKM) gonads. Our results suggest that Cg-DM1 may have broad functions in organ development and may not be involved in sex determination in C. gigas. Cg01830 has low expression in gonads (2.9 RPKM), and its greatest expression is at the umbo larval stages (35.6 RPKM) and in adult labial palp (15.1 RPKM), suggesting it is also unrelated to sex determination. Our analysis indicates that this novel CgDsx is a DM domain gene and a close relative of Dsx that may be involved in promoting male development in C. gigas. This finding along with reports of testis-specific Dmrt-like genes in other molluscs (Klinbunga et al. 2009;Yu et al. 2011;Teaniniuraitemoana et al. 2014) supports the idea that the sex-determining role of DM domain genes is deeply conserved in evolution (Matson and Zarkower 2012).
Sox genes: Sox (Sry-related HMG box) proteins are a family of transcription factors that possess a DNA binding HMG-box (high mobility group) domain. Sox genes are highly conserved in animal kingdom and play key roles in determining cell fate in development and differentiation (Lefebvre et al. 2007). Some members of the Sox family function in sex determination and differentiation, including the founding member Sry (sex-determining region on the Y-chromosome) and Sox9 from mammals (Kent et al. 1996;Koopman 2001;Kashimada and Koopman 2010). Expression of Sry activates Sox9 and the male determining pathway leading to Dmrt1 (Koopman 2005). Sox9 inhibits ovary development through the induction of anti-Müllerian hormone in Sertoli cells and promotes male sex-development through the activation of glia-activating factor 9 (Schepers et al. 2003). It is specifically expressed in testis germ cells in humans (Su et al. 2004). The C. gigas genome encodes 32 proteins containing the HMG domain, and 10 of them can be classified as Sox genes (Table 3). One of the oyster Sox genes (Cg22931) recently has been identified as a member of the SoxE family (Cg-SoxE) that includes the sex-determining Sox9 from mammals (Santerre et al. 2014). Its HMG domain shares high similarity (86.1%) to that of Sox8 and Sox9 in vertebrates (Figure 2, A  and B). However, Cg-SoxE is probably not involved in sex determina-tion. In Santerre et al.'s (2014) study, Cg-SoxE was expressed in both male and female gonads, higher at undifferentiated than mature stages. In our transcriptome data, it was mostly expressed in somatic organs (averaging 197 RPKM) and at gastrula stage (194 RPKM); its expression was lower in male (90 RPKM) and female (61 RPKM) gonads ( Figure  2C). These results suggest that the primary function of Cg-SoxE may not be related to sex determination or differentiation in C. gigas.
n The first + sign indicates presence and the second + sign indicates sex-specific expression or confirmed role in sex-determining pathways. Among all Sox genes that we identified in C. gigas, only one showed testis-specific expression in our transcriptome data. This novel Sox gene Cg06950, classified as CgSoxH, is closely related to Sox30, a member of SoxH. Within family identity in vertebrate Sox gene families usually ranges between 70 and 95% (Lefebvre et al. 2007), and the classification of CgSoxH is tentative and needs to be confirmed in future studies. Nevertheless, phylogenetic analysis shows CgSoxH is a close relative of Sox30 and Sry of vertebrates ( Figure 2B). CgSoxH encodes a protein of 1283 aa divided into eight exons, whereas Sox30 in mammals has five exons alternatively spliced producing two polypeptides of 753 and 501 amino acid residues (Osaki et al. 1999). Its HMG domain shares a 54% identity (e-value = 1.4e215) with Sox30 from human as well as a 49% identity (e-value = 2e221) to Sry from Mus musculus domesticus (Figure 2A). Phylogenetic analysis also shows that CgSoxH is clustered with Sox30 and then with Sry ( Figure  2B). At four positions, CgSoxH shared identical amino acids with Sry instead of Sox30 so it may be considered as an Sry-like gene. It is possible that CgSoxH may be closely related to the ancestral gene before the divergence of Sry and Sox30.
In our data, CgSoxH is exclusively expressed in testis ( Figure 2C). Its expression in testis is 55.7 RPKM compared with ,1 RPKM in all other transcriptomes, including ovary, all somatic organs, and at all developmental stages. In mammals, Sry is the primary male-determining gene, and Sox30 is exclusively expressed in normal testis, but not in maturing germ cell-deficient testis, suggesting a role in differentiation of male germ cells (Osaki et al. 1999). The homology between CgSoxH n   and male-determining Sox30 and Sry and the fact that CgSoxH is exclusively expressed in testes suggest that CgSoxH may play a key role in determining or promoting male sex development in oysters. As far as we can determine, CgSoxH is the first Sry-like gene with possible roles in male-determination identified in a mollusc. A member of SoxE family (Sox100B) has been found in D. melanogaster showing male-specific expression, although it is clearly not a homolog of Sry (Figure 2, A and B). SoxE genes have been identified in molluscs but without male-specific expression or roles in sex determination (He et al. 2013;Santerre et al. 2014, this study). It has been suggested that Sry and Sox9 assumed their roles in sex determination in vertebrates (Matson and Zarkower 2012). The finding of an Sry-like gene in C. gigas with testis-specific expression suggests that Sry and its role in sex determination may not be inventions specific to vertebrates but deeply rooted in evolution. It is possible that an Sry-like gene existed in the common ancestor of bilaterians and had a role in sex determination already. The gene may have been lost in some lineages (such as worms and flies) but conserved in others (molluscs and vertebrates). All other oyster Sox genes are not specifically expressed in gonads. Many of them are highly expressed at certain embryonic stages, suggesting that they may play roles in determining cell fate during early development.
Fox genes: Fox (forkhead-box) proteins are a family of transcription factors with a characteristic DNA-binding forkhead domain. They regulate gene expression and play roles in diverse biological processes including development, differentiation, metabolism and immunity. One member of the Fox gene family, FoxL2, is a key gene involved in ovarian determination in vertebrates (Uhlenhaut and Treier 2006). In mammals, FoxL2 is expressed in the ovary and promotes ovarian development while suppressing the key male promoting Sox9 gene (Crisponi et al. 2001;Ottolenghi et al. 2007;Uhlenhaut et al. 2009).
The C. gigas genome encodes 26 Fox genes (Table 4), compared with more than 40 found in human (Katoh and Katoh 2004a). All 26 Fox genes showed evidence of expression in our transcriptome data with two (Cg06159 and Cg24546) specifically expressed in ovary. We initially suspected that these two ovary-specific Fox genes might function like FoxL2 and determine ovarian development in C. gigas, however their expression profiles suggested otherwise. These two Fox genes had their greatest expression in oocytes, which progressively declined during embryonic development, to undetectable levels by late D-stage (Cg24546) or juvenile stage (Cg06159). They likely play important roles in embryogenesis, but seem unrelated to sex-determining pathways. Cg06159 is a homolog of FoxQ2 ( Figure 3A). In sea urchin, FoxQ2 is progressively restricted to the animal plate during cleavage stage and provides the linkage of the primary animal-vegetal and secondary oral-aboral axes (Yaguchi et al. 2008). Hydrozoan has two FoxQ genes: FoxQ2a is expressed in early embryos and maintained through larval stages, while FoxQ2b is not expressed in embryos or larvae or polyp, but specially expressed in the gonad of medusa (Chevalier et al. 2006). Cg24546 is closely related to FoxN2 ( Figure  3A), which in murine is involved in differentiation of multiple tissues during embryogenesis (Tribioli et al. 2002). In sea urchin, FoxN2/3 is a key gene involved in the formation of the larval skeleton (Rho and Mcclay 2011).
A homolog of FoxL2 has previously been identified in C. gigas (Naimi et al. 2009b), which is also identified in the oyster genome assembly ( Figure 3A) although the assembled copy (Cg11004) incorrectly included an additional 39 exon. In Naimi et al.'s study, CgFoxL2 is expressed in both female and male gonads, with a significant increase in females earlier during sexual development. In our transcriptome data, CgFoxL2 is also expressed in both sexes, highly although not specifically expressed in the ovary (due to abnormally high expression in one male) ( Figure 3B). Its expression is high in gonads of all three n females (24-35 RPKM) and one male (29 RPKM), but low in the other two males (0.322.2 RPKM). Without the exceptional male (M2 in Figure 3C), CgFoxL2 would qualify as an ovary-specific gene (Dheilly et al. 2012). Although the number of oysters sampled is small, the large variation in males is consistent with the sex determination model that recognizes two types of males: fake males that change sex and true male that do not (Guo et al. 1998). M2 with high CgFoxL2 expression may be a fake male that has a higher tendency to change sex. Interestingly, the exceptional male (M2) with high CgFoxL2 expression also had exceptionally low expression of male-promoting CgDsx and CgSoxH genes ( Figure 3C). It is possible that all three genes collectively and through their interactions make M2 prone to sex change. This is largely speculative at this time but can be tested in future studies.
CgFoxL2 is not expressed in oocytes, early embryos, or somatic organs except at moderate levels in digestive gland (possibly due to contamination by gonad) and labial palp ( Figure 3B). A small peak of CgFoxL2 expression is observed at trochophores stage, which together with ovary-specific expression during sexual development, point to a likely role for CgFoxL2 in germline or ovarian determination.
It is interesting to note that one Fox gene (Cg23645) is specifically expressed in testis ( Figure 3D). Cg23645 or CgFoxN5 is closely related to FoxN5, which in mouse is expressed in embryonic germ cells and zygote (Katoh and Katoh 2004b). CgFoxN5 is not expressed at embryonic and larval stages, nor in any somatic organs in C. gigas. Its highly specific expression in testis is novel and may point to a possible role for this novel Fox gene in testis development in C. gigas.
A working model for sex determination in C. gigas The Pacific oyster has a complex sex-determining system that is characterized by protandry, sex change, and rare but consistent hermaphroditism, and how such a dynamic system is controlled and maintained has been the subject of considerable interest (Coe 1943, Haley 1979, Guo et al. 1998, Naimi et al. 2009a,b, Hedrick and Hedgecock 2010, Santerre et al. 2014. We identified possible  sex-determining pathway genes in C. gigas based on sequence homology and functions inferred from transcriptome data. The assumption is that if the gene is involved in sex determination in other organisms and it shows sex-specific expression that is consistent with its known function, it may be related to the sex-determining pathway in C. gigas. We recognize that sequence homology and expression data can only identify possible candidates that require further experimental verification. Given the state of knowledge about sex determination in molluscs, identification of candidate genes and working models are necessary steps for further analysis.
Our analyses indicate that CgDsx, CgSoxH, and CgFoxL2 are probably involved in sex determination in C. gigas. All three genes or their close relatives are key elements of sex-determining pathways in vertebrates and exhibited sex-specific expressions. Other than the DM domain containing Dsx that has been shown to have a deeply conserved role in sex determination in both invertebrates and vertebrates, Sry and FoxL2 are thought to be new recruits to sex-determining pathways in vertebrates or placental mammals (Gamble and Zarkower 2012;Matson and Zarkower 2012). The finding of these key vertebrate sex-determining genes with expected expression profile for sex determination in C. gigas is novel and suggests that these vertebrate genes may not be inventions of vertebrates. Their role in sex determination may be deeply rooted in evolution and at minimum conserved in a mollusc, despite rapid evolution of the regulatory pathways that in C. gigas may involve both genetic and environmental factors. Except for the deeply conserved DM domain gene Dsx, sex-determining or regulating genes in C. elegans and D. melanogaster are either not found in C. gigas or without expression profiles expected for sexdetermining pathway genes (Table 2 and Figure 4). Our analysis suggests that sex determination in C. gigas may share more similarities with that in vertebrates than with that of worms and flies (ecdyspzoans). Although the three groups of bilaterians, Lophotrochozoa, Ecdysozoa, and Deuterostomia, are well-recognized, their relationship to each other is not clear. Phylogenetic analysis based on whole-genome sequences indicates that although molluscs and annelids (lophotrochozoans) are related to worms and flies within protostomes, their genomes in many aspects are more similar to those of invertebrate deuterostomes (Simakov et al. 2013). Also, molluscs share the same telomeric sequence with vertebrates, but not with worms and flies (Zakian 1995, Guo and Allen 1997, Sakai et al. 2005. The conservation of genes related to sex determination between the oyster and vertebrates provides additional argument that molluscs may be closer to the common bilaterian ancestor than ecdysozoans.
As a working model, we speculate that CgSoxH may play a leading role in the sex-determining pathway of C. gigas as it is closely related to the up-stream regulator Sry in vertebrates and strictly expressed in testis. It may directly or indirectly activate CgDsx (Figure 4), which as a DM domain gene is a master switch for testis development in all metazoans studied so far. DM domain genes have been suspected for roles in male-determination in bivalve molluscs (Yu et al. 2011). Both CgSoxH and CgDsx may interact or inhibit CgFoxL2, which is specifically expressed in ovaries with the exception of one male (M2 in Figure 3C and partly supported by Naimi et al. 2009b). The abnormally high expression of CgFoxL2 and low expression of CgSoxH and CgDsx in M2 provided preliminary evidence for possible interaction among these male-and female-promoting genes ( Figure 3C). This finding is preliminary but consistent with the reported interaction among Sry, Sox9, Dmrt1, and FoxL2 in mammals (Veitia 2010, Matson et al. 2011. The finding of large variation in expression of sex-determining genes in males supports one of the genetic models of sex determination Figure 4 Hypothesized sex-determining pathway in C. gigas compared with that in model organisms as summarized by Gamble and Zarkower (2012). For clarity, only selected key sex-specific regulators are shown. Dashed black lines indicate temporal relationships, and dashed red lines indicate hypothetical relationships based on expression data only. [Modified based on Gamble and Zarkower (2012)]. FF genotype permits sex change (Guo et al. 1998). that recognizes two types of males: fake males (FF) that change sexes and true males (FM) that do not (Guo et al. 1998). Although the variation could be caused by various factors such as different stages of sexual development, it is possible that M2 is a FF male where low expression of male-promoting CgSoxH results in low expression of male-promoting CgDsx and high expression of female-promoting CgFoxL2, which in turn may promote sex change to female. The number of oysters studied here is limited, and further studies are needed. If the proposed model is correct, it would be interesting to ask how the expression of CgSoxH is controlled by cis/trans genetic elements and by environmental factors. The working model and insights provided in this study should stimulate further investigation on sex-determining pathways in molluscs and other invertebrates.
This study identified two novel genes, CgDsx and CgSoxH (or Sry-like), that are likely involved in sex determination in C. gigas and provided supporting data for the involvement of CgFoxL2. The sex-determining functions of Sry and FoxL2 are thought to have emerged late during the evolution of vertebrates. The finding of such genes in C. gigas with sex-specific expression indicates that these vertebrate sex-determining genes may not be inventions of vertebrates as suggested by previous studies. Their role in sex determination may be deeply conserved in evolution, despite rapid evolution of the regulatory pathways that in C. gigas may involve both genetic and environmental factors.