Introduction

The sex-determining switch mechanism that triggers the indifferent gonad to follow either a male or female fate is highly variable in vertebrates, even among closely related taxa (Ferguson-Smith 2007). There are two main categories of switch mechanisms; genetic sex determination (GSD) where gonadal sex is controlled by one or more sex-determining loci, and environmental sex determination (ESD) where gonadal sex is dependent on external environmental stimuli (Solari 1994). GSD is the most prevalent mechanism and is often associated, but not always, with heteromorphic sex chromosomes. Despite its prevalence, very few sex-determining genes have been identified. In mice and humans, sex is determined by the SRY gene that triggers testis differentiation (Koopman et al. 1991; Sinclair et al. 1990), DMRT1 triggers female development in birds (Smith et al. 2009) and Xenopus leavis (Yoshimoto et al. 2008), while a male-specific orthologue DMY triggers testis development in some medaka fish (Matsuda et al. 2002). The precise molecular mechanism by which ESD triggers gonadal differentiation is even less understood, but there is good evidence that such mechanisms primarily regulate the production of aromatase, the enzyme that synthesizes estrogen from testosterone, to trigger ovarian development (Ramsey and Crews 2009). Despite the highly variable switch mechanisms, the downstream molecular pathways regulating testis and ovarian differentiation are highly conserved across vertebrates, as is the organization of the gonads.

Gonads consist of two complementary cell types: the somatic cells including the supporting cell lineages (the Sertoli cells, Leydig cells, and peritubular myoid cells in the testes and the granulosa cells and theca cells in the ovary) and the germ cells that will become the gametes (spermatozoa in the testes and oocytes in the ovary). The fate of the somatic cells coordinates gonadal differentiation, which in turn regulates the appropriate development of the germ cells (Koubova et al. 2006; Bowles et al. 2006).

In the male gonad, it is the activation of SOX9 in the somatic cells that directs Sertoli cell differentiation (Sinclair et al. 1990; Bishop et al. 2000). SOX9 and its gonadal enhancer, TESCO, are highly conserved, suggesting that it maintains a critical function in vertebrate testis development (Bagheri-Fam et al. 2010). SOX9 initially shows a low level of transcription and a cytoplasmic distribution in the indifferent gonad, but translocates to the nucleus in males and is dramatically upregulated (de Santa Barbara et al. 2000; El Jamil et al. 2008; Morais da Silva et al. 1996). In mammals, expression of the Y-linked sex-determining gene SRY triggers SOX9 upregulation in the gonad (Sekido and Lovell-Badge 2008), where it acts as a transcriptional activator of the testis differentiation pathway. The nuclear translocation of SOX9 is both necessary and sufficient for Sertoli cell development in mice (Qin et al. 2004; Bishop et al. 2000), and is the most critical step in the initiation of the testis pathway. However, even after testicular fate is established, the Sertoli cell phenotype remains plastic. In mice, expression of Dmrt1 is essential to maintain Sertoli cells (Matson et al. 2011). Loss of Dmrt1 even in adult Sertoli cell leads to upregulation of FoxL2 and transdifferentiation into a granulosa cell phenotype. Thus, viable development of the gonad requires a tightly regulated set of key factors to specify and then maintain gonadal cell identity.

Activation of ovarian development

In XX mammals, in the absence of SRY, SOX9 expression is low and its protein product is cytoplasmic in the somatic cells of the indifferent gonad. As a result, the testis pathway is not initiated and granulosa cell development is activated. The precise mechanism preventing SOX9 from entering the nucleus and activating the Sertoli cell program in female gonads is unknown, but recent data suggest that estrogen may play an important role in this process (Pask et al. 2010). This is of special interest since a recent increase in male reproductive abnormalities in humans and wildlife, resulting in lower sperm counts and quality and increased rates of hypospadias and testicular dysgenesis, has been attributed to an increased exposure to estrogenic compounds found in the environment (Giwercman et al. 1993; Carlsen et al. 1992).

Historically, the ovary was viewed as the default gonadal state, such that in the absence of SRY, an ovary would passively form. This view has now changed, and a few key ovary-promoting genes have been identified in mice and humans, the most critical of which include WNT4, RSPO1, and FOXL2. Mutations in all three genes result in a failure of normal ovarian formation in mice and humans (Chassot et al. 2008a, b; Heikkila et al. 2001; Uda et al. 2004; Vainio et al. 1999). FoxL2 appears to be critically important in regulating female somatic cell fate since its ablation in adult mouse ovaries leads to transdifferentiation of the granulosa cells to a Sertoli cell phenotype and an upregulation of SOX9 (Uhlenhaut et al. 2009). FoxL2, in conjunction with activated estrogen receptors, is proposed to maintain granulosa cell fate by suppressing SOX9 transcription through binding to its enhancer element, TESCO (Uhlenhaut et al. 2009). Binding sites for FOXL2 and the estrogen receptor are highly conserved in TESCO across mammals, including in the marsupial, tammar wallaby (Fig. 1). Furthermore, FOXL2 has been shown to play a direct role in the upregulation of CYP19 (aromatase; required for the synthesis of estrogen from testosterone) in both the fish brain (Sridevi et al. 2011) and indifferent XX goat gonad, where it initiates the synthesis of estrogen, promoting ovarian development (Pannetier et al. 2004).

Fig. 1
figure 1

Tammar wallaby SOX9 TESCO enhancer. Green = SOX binding sites; yellow = FKHD (FOXL2) binding sites; red = estrogen receptor binding site; purple = GATA binding site; yellow/green = FOX/SOX binding site

Full size image

A conserved role for estrogen

Estrogen is known to play an essential role in female sex determination in nonmammalian vertebrates regardless of the sex-determining mechanism (Solari 1994; Nakamura 2010). The production of estrogen in the indifferent gonad is controlled by the expression of CYP19, which encodes the aromatase enzyme and causes estrogen production. In the presence of estrogen, the indifferent gonad will follow an ovarian development pathway, while in its absence the gonad will become a testis (Solari 1994). As a result, exogenous estrogen exposure to developing fish, reptile, amphibian, and bird fetuses will trigger ovarian development, while exposure to estrogen inhibitors causes testis development (Solari 1994; Ramsey and Crews 2009). This is in contrast to mammals where SRY triggers testis development and the ovary is the default state. However, in nonmammalian vertebrates, estrogen appears to be the master regulator of ovarian development and in its absence the gonad will default to a testicular fate.

Despite the highly conserved role of estrogen in nonmammalian vertebrates, its function in the development of the mammalian ovary remains less clear. Interestingly, expression of the estrogen receptors, which mediate estrogen actions within the cell, is maintained in the somatic cells of the indifferent gonads of mice, humans, goats, sheep, and marsupials indicative of a highly conserved role for estrogen in the early mammalian gonad (Calatayud et al. 2010). It was a surprising finding then, that estrogen was not required for initial ovarian development in mice (Couse and Korach 1999). Mice deficient for both the alpha and beta estrogen receptors or CYP19 have normal early ovarian differentiation (Britt and Findlay 2003; Britt et al. 2001). However, shortly after birth, germ cells are lost and the somatic cells take on a Sertoli cell phenotype (Fisher et al. 1998; Toda et al. 2001). These Sertoli-like cells express SOX9 and show a characteristic Sertoli cell morphology, with tight junctions, and arrangement (Britt and Findlay 2003). Upon administration of estrogen to aromatase-deficient mice, ovarian histology is restored and SOX9 levels are significantly decreased, along with several other testis markers, to normal female levels (Britt et al. 2004). Together these data show that the somatic cells of the ovary also retain a plasticity that, along with the genes regulating somatic cell differentiation, is directly responsive to estrogen.

While mouse studies have been fundamental in developing a basic understanding of gonadal differentiation, there are differences in gene expression, responses to haploinsufficiency of critical genes and, most importantly, in the role of estrogen in the fetal gonad between mice and other mammals (Wilhelm et al. 2007). Comparative analyses across multiple species can be particularly helpful in isolating critical regulatory networks required for developmental events from those that show species-specific variations (Sanchez et al. 2011; Pounds et al. 2011; Crozat et al. 2010; Lu et al. 2009). Outside of the rodent lineage, upregulation of CYP19 has been reported in the fetal ovary of many eutherian species including goats (Pannetier et al. 2004), sheep (Quirke et al. 2001), and cows (Garverick et al. 2010) suggesting that estrogen may play a central role in its early differentiation. Similarly, in humans, exposure of the developing fetus to potent synthetic estrogenic compounds can dramatically affect male gonadal differentiation (Toppari 2008; Arai et al. 1983). Thus, it appears that gonadal development in rodents may be unusually resistant to a loss of estrogen.

The ability for estrogen to direct ovarian development in mammals has been demonstrated in marsupials (Pask et al. 2010; Coveney et al. 2001; Burns 1955). Marsupials have been evolving independently of humans and mice for around 160 million years (Fig. 2; Luo et al. 2011). Sexual differentiation occurs around the time of birth in marsupial, unlike in eutherian mammals where this process occurs in utero. Marsupials develop gonads that are identical in structure to human and mouse gonads and determine sex based on the presence/absence of the SRY-bearing Y chromosome (Pask and Graves 2001). Development of the somatic cell lineages and germ cell entry into either meiotic or mitotic arrest is separated by several days in the tammar wallaby, similar to the developmental timing seen in human, goat, sheep, and cow gonads, whereas in the mouse these events occur concurrently (Harry et al. 1995; Renfree et al. 1996). Thus in marsupials it is possible to examine the effects of exogenous estrogen on the differentiation of the somatic cell lineages ex utero, uncomplicated by changes in germ cell development or the in utero environment.

Fig. 2
figure 2

Phylogenetic tree of the vertebrates showing the divergence of marsupials (red) from mice and humans (eutherian mammals; blue)

Full size image

Estrogen blocks male development by modulating SOX9

Administration of estrogen to genetically male marsupial neonates causes ovarian development of the gonad (Pask et al. 2010; Coveney et al. 2001; Burns 1955). In the presence of estrogen, key male differentiation genes fail to be upregulated in the XY gonad and instead, key ovary-promoting genes are upregulated leading to ovarian development (Pask et al. 2010). Estrogen appears to trigger sex reversal through the exclusion of SOX9 from entering the nucleus in the somatic cells of the developing gonad (Pask et al. 2010). In the absence of nuclear SOX9, Sertoli cell development cannot be initiated and the somatic cells follow a granulosa cell fate (Fig. 3).

Fig. 3
figure 3

Left panel: In the absence of estrogen (E), the estrogen receptors (ER) are not activated and remain cytoplasmic. SOX9 is able to translocate to the nucleus where it acts with SRY to reinforce its own transcription and activate Sertoli cell differentiation. Right panel: In the presence of estrogen, estrogen receptors become activated. Activated estrogen receptors (ER + E) block any SOX9 protein that is made from entering the nucleus, preventing SOX9 upregulation, and resulting in granulosa cell development. Activated estrogen receptor complexes can also move into the nucleus and along with FOXL2, suppress SOX9 transcription by directly binding to the SOX9 enhancer, TESCO

Full size image

A conserved role for estrogen-mediating SOX9 action is consistent with several observations in mammals. In mice, Sox9 is able to autoregulate by binding to its own promoter (Sekido and Lovell-Badge 2008). However, despite Sox9 expression in the somatic cells of the adult mouse ovary, it does not become upregulated (Notarnicola et al. 2006). High levels of estrogen could prevent the nuclear translocation of SOX9 in adult female gonads, preventing its own upregulation. Conversely, Sox9 is upregulated in the absence of estrogen in aromatase-deficient mice ovaries, suggesting it can translocate to the nucleus to propagate its own transcription. However, when these mice are exposed to exogenous estrogen the effect is reversible, and Sox9 is repressed (Britt et al. 2002; de Santa Barbara et al. 2000). Data from marsupials would suggest that this repression occurs by trapping SOX9 in the cytoplasm preventing its autoregulation. Furthermore, activated estrogen receptors, in conjunction to FoxL2, could directly suppress SOX9 transcription through binding to TESCO (Uhlenhaut et al. 2009). Either way, in mammals, estrogen still remains a critical factor in maintaining ovarian somatic cell fate. This fundamental role of estrogen is also consistent with observations in nonmammalian vertebrates. In the presence of estrogen, SOX9 remains cytoplasmic in the early developing gonads of many nonmammalian vertebrates, permitting granulosa cell development (de Santa Barbara et al. 2000). However, in the absence of estrogen, SOX9 becomes nuclear and Sertoli cell development is initiated (Sim et al. 2008). Therefore, the ability of estrogen to mediate the subcellular localization of SOX9 and directly mediate its transcription, could explain the primary mechanism by which estrogen establishes sex in nonmammalian vertebrates.

Further investigations are needed to determine how estrogen mediates the subcellular localization of SOX9 within the somatic cells. SOX9 contains two defined nuclear localization signals (NLS) found in the C- and N-termini that are 100% conserved between mouse, human, and the wallaby (Pask et al. 2002). Active transport through nucleopore complex is facilitated in part by importin-β, binding directly to the C-terminal NLS (Sim et al. 2008). This binding is enhanced by phosphorylation of SOX9 by phosphokinase A, facilitating increased nuclear import (Malki et al. 2005). The N-terminal NLS binds calmodulin, another factor that facilitates nuclear transport of SOX9 (Argentaro et al. 2003). SOX9 is also subject to SUMOylation and ubiquitination (Sim et al. 2008). SUMOylation has been shown to regulate nucleocytoplasmic trafficking of several proteins while ubiquitination marks proteins for degradation. SUMOylation of SOX9 in COS7 cells has been shown to alter its subnuclear localisation and transcriptional activity (Hattori et al. 2006). Estrogen may affect one or many of these different pathways to regulate the subcellular localization and activity of SOX9.

A conserved model for determining vertebrate somatic cell fate

While the switch mechanisms that trigger the development of the ovary or testis pathways vary widely among vertebrates, the fundamental control mechanisms regulating somatic cell fate share many commonalities. This suggests a highly conserved and antagonistic relationship between SOX9 and estrogen-driving Sertoli cell and granulosa cell differentiation, respectively. In mammals, the somatic cell decision is initially determined by the presence or absence of SRY. When SRY is present, SOX9 is upregulated and can translocate to the nucleus to activate Sertoli cell differentiation. In females, in the absence of SRY, SOX9 is not upregulated and the granulosa cell program is initiated. While it is yet to be shown if estrogen plays a critical role in early mammalian ovary formation outside of the rodent lineage, estrogen is essential for maintaining granulosa cell fate in the mature gonads, possibly by ensuring that any SOX9 protein produced remains trapped in the cytoplasm. In nonmammalian vertebrates, the primary sex-determining mechanism, be it GSD or ESD, leads to either the presence or absence of CYP19 expression. In the presence of aromatase and estrogen, basal SOX9 cannot enter the nucleus and the Sertoli cell program is blocked. In the absence of aromatase and estrogen, SOX9 can translocate to the nucleus, trigger its own upregulation, and initiate Sertoli cell development. More work is needed to confirm this model and determine the precise mechanism by which activated estrogen receptors mediate the subcellular localization of SOX9. However, these findings provide a simple explanation for the dramatic switch in vertebrate sex determination mechanisms from primarily hormonal control to primarily genetic control, converging through the modulation of SOX9.