Sex determination must be a very ancient process, with male and female sexes recognised in diverse species, from corals to worms, insects, fishes, birds, and mammals (Loya and Sakai, 2008; Morrish and Sinclair, 2002; Schmieder et al., 2012; Gamble and Zarkower, 2012; Kobayashi et al., 2013). In vertebrates, sex determination has long been equated with gonadal differentiation into ovaries or testes. The logical assumption is that sex-determining genes, inherited at fertilization, become active in the gonads during embryonic or larval life. However, various reports of somatic sexual dimorphisms preceding gonadal development call for a more considered definition of sex determination. A second long-held belief is that the molecular pathways responsible for testis versus ovary formation are conserved among animals. This has largely proven to be true, but the relative positions of these genes in the male and female pathways differ among the major groups. This review focuses on the chicken embryo, drawing comparisons with two other divergent vertebrate lineages: fishes and mammals. We focus on two key pathways, AMH/SOX9 in males and FOXL2-Aromatase in females, asking “just how conserved is vertebrate sex determination?”
WHEN IS SEX DETERMINED AND BY WHAT MECHANISM?
Broadly defined, sex determination is a developmental event during which sex is established. In species with heteromorphic sex chromosomes, such as birds and mammals, sex is set at fertilization by the differential inheritance of sex chromosomes. For example, in mammals, XX embryos are destined to become female, while XY embryos will be male. There is, therefore, little ambiguity as to when sex is determined. Mechanistically, most studies on vertebrate sex determination have focussed upon the embryonic gonads. In eutherian mammals, the Y-chromosome-linked SRY gene initiates a cascade of male-specific gene expression in the gonads of XY embryos, resulting in testicular differentiation (Wilhelm et al., 2007). However, any factors that are differentially inherited could be potential sex determinants (Arnold, 2011). Thus, the inheritance of two versus one X chromosome could play a role in generating sexual dimorphisms. Recent studies have shown, for example, that sex differences in adiposity depend at least partly on X chromosome dosage in mice (Chen et al., 2012). In the mouse, XX and XY embryos show sexual dimorphism in growth, metabolism, and gene expression (reviewed in 6). Meanwhile, in a marsupial, the pouch and scrotum develop prior to the gonads (O et al., 1988). These sex differences can be attributed to either a Y-borne gene/s, or X chromosome dosage (involving genes that would escape dosage compensation). Similarly, in the chicken embryo, sexually dimorphic gene expression has been documented in blastoderms and brain tissues pre-dating gonadal formation (Lee et al., 2009; Scholz et al., 2006; Zhang et al., 2010). Thus, in both birds and mammals, sex-determining genes appear to be operating in various tissues, including both gonadal and non-gonadal sites. In birds, the female is heterogametic (ZW), while the male is homogametic (ZZ) and there is no global Z inactivation system (Arnold et al., 2008). Hence, sex may depend upon W gene expression (females) or Z-dosage, or both mechanisms, in different tissues (Smith, 2007). Birds lack SRY, but the Z-linked DMRT1 gene is required for testis formation (Smith et al., 2009). However, this does not preclude the involvement of other Z-linked genes, in either the gonads or extra-gonadal sites.
Recently, data have been presented in support of at least partial cell-autonomous sex differentiation in birds. Naturally occurring gynandromorphic chickens (male on one side, female on the other) indicate that each cell in the body has an inherent sexual identity (Zhao et al., 2010). This may be triggered by the W sex chromosome, which is poorly understood in birds, or the critical dosage of a single or group of Z-linked genes (Clinton et al., 2012). Sex may be set in each cell by the same combination of sex-linked genes, or perhaps by different genes in different cells. Hence, in birds, as in mammals, different combinations of sex-determining genes may operate in different tissues of the body. The result is a coordinated expression of sexual phenotype, involving multiple parallel pathways (Arnold, 2011). It is unclear at this stage whether cell-autonomous sex differentiation is restricted to avians, which seems unlikely, or whether it is a general phenomenon among vertebrates. As noted above, there is also evidence for cell-autonomous sexual differentiation in mammals.
CONSERVATION OF KEY GONAD-differentiating GENES
In light of the discussion above, genes operating within the embryonic gonads to induce ovary or testis formation might best be described as “gonadal sex-differentiating” rather than simply “sex-determining genes.” Overall, key gonadal sex-differentiating genes are conserved in vertebrates, i.e., they show conservation of structure and are expressed sexually dimorphically. However, the master trigger differs among groups. The vertebrate gonadal sex-determining pathway shows great plasticity and it is becoming clear that genes can readily become co-opted to participate in gonadal sex differentiation, often at the top of the molecular cascade, or genes already in the pathway can become more senior. Hence, in eutherian mammals, Sry is a recently evolved key Y-linked testis determinant (Fig. 1). It is likely that Sry evolved from its nearest relative Sox3, which is X-linked but unrelated to sex per se. It is hypothesised that acquisition of male-specific functions by the ancestral Y chromosome lead to the emergence of Sry (Graves, 2002).
Sry is absent from the genome outside therian mammals (marsupials and so-called “placentals”). In birds and lower vertebrates, there is a pervasive role for DM domain genes in gonadal sex differentiation and it is considered that these genes have an ancient association with sex (Raymond et al., 1998)(Fig. 1). DM domain genes are highly conserved and encode transcription factors with a zinc-finger-like DM domain. One of these genes, DMRT1, is Z-linked in birds, is male up-regulated, and is required for testis formation during embryonic development (Raymond, 1999; Smith, 2009; Shan, 2000). A duplication of DMRT1, DMY/Dmrt1bY is the primary testis determinant in the Japanese medaka fish (Oryzias latipes) (Matsuda et al., 2002, 2003; Nanda et al., 2002), while a W-linked DM-domain gene, DM-W, is involved in ovary determination in the amphibian, Xenopus laevis (Yoshimoto et al., 2008) (Fig. 1). These DM domain genes have acquired new roles in gonadal sex differentiation via gene duplication and translocation (fish), duplication and truncation (DM-W on the female W sex chromosome in X. laevis), or loss of function of one allele (female birds, ZW). While Dmrt1 is expressed in the developing mouse testis, targeted deletion reveals a role in testis maintenance postnatally (Matson et al., 2011; Raymond et al., 2000).
TESTIS differentiation, SOX9 AND AMH
A key player in the testis-differentiating pathway is the Sry-related gene, Sox9. Sox9 is expressed in the developing testes of all vertebrate embryos that have been examined (da Silva et al., 1996; Kent et al., 1996; Kobayashi et al., 2005; Rodriguez-Mari et al., 2005; Shoemaker et al., 2007; Torres Maldonado et al., 2002; Western et al., 1999). In mammals, current evidence favours the idea that Sry acts with the orphan nuclear receptor, Sf1, to activate Sox9 expression in the developing XY gonad (Sekido and Lovell-Badge, 2008). Sry is turned off by Sox9, which then maintains its own expression. In the absence of Sry, other factors must activate Sox9 in non-mammalian vertebrates. This may involve DM domain genes. However, in the chicken embryo, DMRT1 expression precedes that of Sox9 by at least 2 days (embryonic day 4 vs. day 6), implying that other intervening genes are involved. In fishes, whole genome duplication events have yielded Sox9 paralogues. Interestingly, in the medaka fish, Sox9b (the orthologue of tetrapod Sox9) is not involved in testis determination, but has a function in germ cells (Nakamura et al., 2012a) (Fig. 1). This suggests that Sox9 neofunctionalisation has occurred during vertebrate evolution, such that the gene now has a somatic role in groups such as birds and mammals. How and when this occurred remains to be clarified.
Sry is likely to be the trigger for Sox9 activation in eutherian mammals, but the former is absent in monotremes. In lower vertebrates, SRY is also absent and other triggers must exist. Given the different, independent origins of sex chromosomes among birds, reptiles, and the anamniotes, a variety of different testis-determining triggers is likely. In birds, testis development requires the conserved Z-linked DMRT1 gene. Yet DMRT1 is not sex-linked in various reptiles, pointing to another factor being involved. In mammals, Sox9 activates expression of the Fgf9 and Prostaglandin D synthase (PDGS) genes. The former is required for proper testis development, at least partly by positive feedback regulation of Sox9 expression (Colvin et al., 2001). The product of prostaglandin D synthase (PGD2) is involved in recruiting Sertoli cells of the developing testis, and facilitates Sox9 nuclear import (Fig. 1A). (Wilhelm et al., 2005). In the chicken embryo, PDGS is a conserved gene implicated in testis development (Moniot et al., 2008), but Fgf9 does not show a sexually dimorphic expression pattern during gonadal sex differentiation (Smith, unpublished data). It is possible that other Fibroblast Growth Factor genes substitute for FGF9 in the avian system.
In the mammalian gonad, a key role for Sox9 during testicular development is the activation of Amh (Anti-Müllerian hormone). In the mouse embryo, Amh is expressed only in male embryos, within the Sertoli cells of developing testis cords, shortly after Sox9 expression. While Amh is not required for mouse testis development, it mediates the degradation of the paired Müllerian ducts (which otherwise form the Fallopian tubes and upper vagina of female embryos). Sox9 is required with the Sf1 and Wt1 transcription factors for Amh expression in the embryonic mouse testis (Arango et al., 1999). In the chicken embryo, AMH precedes SOX9 expression, and the gene is expressed in both males and females (Oreal et al., 1998). Hence, SOX9 does not activate AMH in the avian system, although it might up-regulate it (Fig. 1). The activator of AMH expression in the chicken and other vertebrates is unclear, but is likely to involve the orphan nuclear receptor, SF1, which is expressed in early gonads, as in mouse (Smith et al., 1999b). However, AMH is always expressed more highly in male gonads (ZZ) compared to females (ZW) suggesting that a dosed Z-linked gene such as DMRT1 may activate its expression in avians (Fig. 1).
In birds and other non-mammalian vertebrates, AMH signalling may play a more central role in testis determination than in mammals. In the chicken embryo, gonadal AMH is expressed in a sexually dimorphic manner from very early stages (day 4.5; stage 25) (Smith et al., 1999a; Smith et al., 2007; Ayers et al., 2013). Furthermore, experiments conducted over 20 years ago showed that late-stage embryonic chicken testes grafted into the extraembryonic coelom of day-3 chicken embryos can induce testicular differentiation in genetically female embryos (Maraud et al., 1990; Rashedi et al., 1983). Although the secreted factor responsible was not definitively identified, it was considered to be AMH, since the Müllerian ducts also regressed (Rashedi et al., 1990). (The other major testicular hormone, testosterone, does not have the same masculinising effect when administered to early chicken embryos) (Faucounau et al., 1995). It is, therefore, possible that endogenous AMH has an important role in organising the early developing testis in birds, possibly by activating SOX9, but this remains to be shown. In the chicken embryo, AMH can antagonize the well-characterised feminizing influence of estrogen (Stoll et al., 1993). Since estrogen is a key player in ovarian differentiation in birds (Scheib, 1983), AMH may participate in testis determination by blocking estrogen synthesis, namely by repressing expression of the Aromatase (CYP19A1) gene (di Clemente et al., 1992) (Fig. 1).
In the medaka fish (O. latipes), Amh and its specific receptor (Amhr2) are equally expressed in the developing gonads of both sexes, with no clear sexual dimorphism (Kluver et al., 2007). However, dysfunctional AMH signaling is responsible for feminisation of the gonads in the medaka hotei mutant model (Nakamura et al., 2012b). Intriguingly, recent data indicate that a paralogue of AMH on the Y chromosome is the master testis determinant in a species of fish, the Patagonian pejerrey (Odontesthes hatcheri) (Hattori et al., 2012). It can be appreciated from these examples that AMH signaling occupies different positions within the testis-determining pathway in different groups (fishes vs. birds vs. mammals). In fishes, AMH is expressed in gonads despite the fact that fishes generally lack Müllerian ducts (Kluver et al., 2007). This indicates that the evolution of AMH probably pre-dates the evolution of Müllerian ducts in higher vertebrates, supporting a more ancient role for the hormone in the gonads themselves.
OVARIAN DIFFERENTIATION, β-CATENIN, AND THE FOX2-AROMATASE PATHWAYS
The key ovary determinant in mammals has not yet been defined, but the canonical β-catenin signalling pathway is required for ovarian morphogenesis (Chassot et al., 2008; Maatouk et al., 2008). This involves the Wnt4 and R-Spondin1 signaling molecules, which exert their effects through the stabilisation of β-catenin and signal transduction to the nucleus (Liu et al., 2009; Tomizuka et al., 2008). β-catenin is dispensable for mammalian testis development, while Sry and/or Sox9 expression in males can block the Rspo1/Wnt4/β-catenin pathway (Bernard et al., 2008; Lau and Li, 2009; Tamashiro et al., 2008). The β-catenin signalling pathway appears to be conserved in the other vertebrates, including fishes, reptiles with temperature-dependent sex determination, and chickens. In these cases, R-Spo1, Wnt4, and/or β-catenin show female up-regulation (Ayers et al., 2013; Smith et al., 2008; Zhang et al., 2011)(Fig. 2). This “female” pathway appears to be deeply conserved among vertebrates. Based upon studies of mouse embryos, it has been proposed that the fate of the developing mammalian gonad involves an antagonism between Sry/Sox9-induced Fgf9 signaling (male promoting) versus Wnt signalling (female promoting) (Jameson et al., 2012; Kim et al., 2006). It remains to be seen whether this antagonistic system also applies in other vertebrates. In the chicken, Fgf9 is not expressed dimorphically at the time of sexual differentiation.
A major difference between mammal and non-mammalian vertebrates is the requirement of estrogen for ovarian differentiation in the latter. Estrogen synthesis from androgen substrates is catalysed by the aromatase enzyme, encoded by CYP19A1. Egg-laying species such as most fishes, reptiles, and birds express CYP19A1 at the onset of ovarian differentiation (Andrews et al., 1997; Ramsey et al., 2007; Vaillant et al., 2001) and inhibition of aromatase can cause testicular differentiation in genetic females (Elbrecht and Smith, 1992; Guiguen et al., 1999, 2010; Wibbels and Crews, 1994). Conversely, androgens can masculinise the developing gonads in various fishes, but not decisively in reptiles and birds. In eutherian mammals, the embryonic gonads are resistant to sex steroid effects, although estrogen is required to maintain the postnatal ovary (Britt et al., 2002). There appears to have been a gradual evolutionary loss in the requirement of sex steroid hormones for gonadal sex differentiation, from fishes to mammals. It is logical to suppose that eutherian mammals abandoned the use of estrogen as a gonad-determining factor with the evolution of placentation and intrauterine development (Wolf, 1999). Such an environment would expose all embryos to maternally derived estrogens. (Interestingly, marsupials are susceptible to estrogens, but gonadal sex differentiation typically occurs after birth in pouch young, and hence presumably in an estrogen-free environment.) The regulation of CYP19A1 gene therefore becomes a key question in vertebrate ovarian differentiation. Steroidogenic factor 1 (SF1) has been shown to positively regulate CYP19A1 in mammals and other species, but this factor alone cannot account for female-specific aromatase expression. (In the chicken, SF1 is initially expressed in both female and male gonads, but CYP19A1 expression is female-specific, Smith et al., 1999b.) The conserved forkhead/winged helix transcription factor, FOXL2, is a likely key regulator of CYP19A1 in females. Gonadal FOXL2 is expressed female-specifically and co-localises with aromatase, in mouse, chicken, and fishes, and is therefore likely to be a deeply conserved element of the vertebrate ovary-determining pathway (Baron et al., 2005; Govoroun et al., 2004; Nakamoto et al., 2006; Uhlenhaut and Treier, 2006) (Fig. 2).
However, the position of FOXL2 in the ovarian pathway appears to vary among the major groups. In the chicken embryo, FOXL2 expression from embryonic day 5.5 is one of the earliest known markers of ovarian development. FOXL2 activates the CYP19A1 5′ regulatory region in the Tilapia fish (Wang et al., 2007) and in mammals (Fleming et al., 2010; Pannetier et al., 2006), making it likely to also do so in chicken. In the mouse, FOXL2 is required for granulosa cell differentiation, folliculogenesis, and to repress the testicular developmental program, but postnatally (Ottolenghi et al., 2005; Uda et al., 2004; Uhlenhaut et al., 2009). Given the requirement of aromatase for embryonic ovarian development in non-mammals, the early female-specific expression of FOXL2 in these groups points to a more critical role for FOXL2 in the female pathway. However, analysis of FOXL2 function in goat embryos cautions against drawing conclusions about mammalian gonadal sex differentiation based solely on mouse studies. In goats with the Polled Intersex mutation (PIS), embryos show FOXL2 loss-of-function and masculinisation of the gonads during embryonic stages, which is not seen in mice.
A major question relating to ovarian development is how the FOXL2 and R-SPO1/Wnt4 pathways interact to coordinate ovarian development. The two pathways appear to be independent. In goat and in chicken embryos, FOXL2 and R-SPO1 localise to different ovarian compartments (medulla and cortex, respectively). In the PIS goat mutants, R-SPO1 expression appears unaffected (Kocer et al., 2008). Meanwhile, targeted deletion of Foxl2 or R-spo1 or Wnt4 alone does not induce testis development in mice. However, FOXL2−/−;WNT4−/− and FOXL2−/−;R-SPO1−/− compound mutants do show more robust masculinisation (Ottolenghi et al., 2007; Auguste et al., 2011).
A general feature of vertebrate gonadal sex-differentiation pathways is that master switches appear to have been added at the top of the hierarchy, with more conserved core genes appearing downstream (Wilkins, 1995). Hence, AMH and SOX9 in males are conserved across groups, but the upstream regulator differs (SRY in mammals, DM domain genes in non-mammals). Among fishes, the sex-determining switch is remarkably labile. Hence, a duplicated copy of DMRT1 controls testis development in some species of Medaka fish, but not in all medaka, nor in all fishes. In other cases, a novel gene has been co-opted into the pathway. For example, an immune-related gene, sdY, has recently been identified as the master Y-linked regulator of testis development in the rainbow trout, Oncorhynchus mykiss (Yano et al., 2012). This gene encodes a truncated form of an interferon regulatory factor 9, previously unassociated with sex. With respect to the ovarian development, the R-SPO1/ Wnt4 and FOXL2 pathways are conserved from fishes to birds and mammals. However, the genetic networks are in fact more complex than this simple scenario. It is clear that genes can become shuffled within the pathway; SOX9 is upstream of AMH in mammals, but the reverse applies in birds, for example. It can been seen from Figures 1 and 2 that less is known about the ovarian pathway versus the testicular pathway; with the possible exception of DMW in X. laevis, master ovary-determining genes remain undiscovered. In contrast, several different master testis-determining factors have been identified. Superficially, it would appear that the testis pathway is more changeable than the ovarian pathway, although this cannot be confirmed in the absence of master ovary factors.
These differences among species reflect differences in the cell biology and morphogenesis of gonads that are specific to the different vertebrate clades. For example, in male embryos, the Sertoli cell progenitors in the chicken apparently have a different developmental origin than those of the mouse (Sekido and Lovell-Badge, 2007). In female chicken and goat embryos, the embryonic gonad elaborates a thickened outer cortex, which is not seen in XX mouse embryos at early stages. Aromatase/ CYP19A1 is a key player in gonadal differentiation in the egg-laying groups, with FOXL2 likely to mediate its activation, a pathway that does not apply in the mouse embryo. The estrogen synthesised by aromatase is important for regulating the proliferation of the outer gonadal cortex in the chicken embryo, whereas the mouse gonad does not elaborate a cortex during embryonic stages. Most fishes exhibit an ovarian cavity, which is not seen in the tetrapod gonad. In some fishes, the germ cells are required for proper gonadal sex differentiation (Kurokawa et al., 2007), whereas this is not the case in other fish species or in chicken or mouse (Maatouk et al., 2012; McCarrey and Abbott, 1978). Such differences between species must logically involve differences in gene expression and this is reflected in the plasticity of the genetic networks underlying gonadal morphogenesis. In the past, the embryonic gonads of vertebrates have been considered highly conserved in structure. While this is generally true, it is clear that the molecular pathways underlying this commonality of structure are actually quite plastic.
Advances in whole genome sequencing will facilitate the identification of additional components of vertebrate sex-determining pathways, both within and beyond the gonads. The challenge is to build interactive gene networks in the different species, to elucidate conserved and divergent elements. RNA-seq and ChiP-seq approaches are now being used to address these issues. Functional analyses are now possible in most groups, such as fishes and chicken, in addition to well-established strategies in mouse. Another area worthy of investigation is exactly how novel genes are incorporated to the top of the pathway, and determining the evolutionary pressures that cause the relative shuffling of genes in the male and female pathways. These efforts will broaden our understanding of vertebrate sex determination and how it has evolved.