Sex differentiation is a unique developmental process. Starting from a bipotential gonad, it gives rise to the ovary and the testis, two highly specialized organs that differ morphologically and physiologically despite sharing common reproductive and endocrine functions. This highlights the specific plasticity of the gonadal precursors and the existence of complex antagonistic genetic regulation. Mammalian sex determination is controlled by paternal transmission of the Y-linked gene, sex-determining region Y (SRY). Using mouse models, it has been shown that the main role of Sry is to activate the expression of the transcription factor Sox9; either one of these two genes is necessary and sufficient to allow testicular development through Sertoli cell differentiation. Thus, defects in SRY/Sry and/or SOX9/Sox9 expression result in male-to-female sex reversal of XY individuals. Molecular mechanisms governing ovarian differentiation remained unknown for a long time, until the discovery of the roles of R-spondin1 (RSPO1) and WNT4. In XX individuals, activation of the β-catenin signaling pathway by the secreted proteins RSPO1 and WNT4 is required to allow granulosa cell differentiation and, in turn, ovarian differentiation. Thus, mutations in RSPO1 result in female-to-male sex reversal of XX patients, and mouse models have allowed the identification of genetic cascades activated by RSPO1 and WNT4 to regulate ovarian development. In this review, we will discuss the respective roles of RSPO1, WNT4, and the β-catenin signaling pathway during ovarian differentiation in mice.
WNT4 and R-spondin1: state of the art
WNT/β-catenin signaling and the ovary: a woman's story?
It has long been assumed that the female outcome was the default state during sexual development. This came from the work of the endocrinologist Alfred Jost, who carried out castration experiments of rabbit embryos and observed that the external genitalia of both male and female fetuses developed as female (Jost 1947). As a result, the majority of laboratories focused their research on finding the molecular clues triggering male development. In the 1990s, the male-determining factor was discovered and it has been shown that male fate of the gonad depends on expression of the sex-determining region Y (SRY) gene, localized to the Y chromosome transmitted by the father (Sinclair et al. 1990). While male differentiation began to be elucidated, the study of ovarian development received very little attention and remained quite mysterious. More recently, however, the view of a ‘default’ female pathway has been challenged by the description of genetic mutations leading to masculinization of XX gonads despite the absence of any ‘male’ genes in human XX patients (Parma et al. 2006, Mandel et al. 2008), in mice (Vainio et al. 1999), and in goats (Pailhoux et al. 2001). Most of these mutations affect the WNT/β-catenin signaling pathway that controls various steps in mammalian organogenesis and organ homeostasis during adulthood (Niehrs 2012, de Lau et al. 2014). In 2006, Prof. Giovanna Camerino et al. identified R-spondin1 (RSPO1), a WNT signaling pathway activator, as a novel key factor involved in sexual and ovarian differentiation (Parma et al. 2006). After 2 years, Prof. Blanche Capel et al. showed that genetic induction of WNT/β-catenin signaling can modify the fate of the gonad by promoting the ovarian development from a gonad programed to become a testis (Maatouk et al. 2008). This complemented their previous work on Wnt4, another player in the WNT signaling pathway, which showed that sex determination is governed by a balance between two antagonistic pathways (Kim et al. 2006) and firmly established WNT/β-catenin signaling as a key pathway involved in female differentiation. In this review, we will discuss the role of WNT4 and RSPO1, the best studied WNT signaling activators in sexual differentiation. We will also report the data obtained in other species and discuss some of the key questions that remain to be answered.
WNT4, a Wingless family gene involved in many biological processes
The first Wingless gene was discovered in the 1970s in genetic screens aiming to identify genes essential for segment patterning of the Drosophila embryo. Since then, numerous studies have shown the implication of Wingless in a wide spectrum of biological processes. WNT family members (for Wingless-type MMTV integration site family) are highly conserved secreted glycoproteins. Although it was assumed that Wingless was a long-range acting protein, recent data have indicated that, at least in Drosophila, it acts predominantly on neighboring cells (Alexandre et al. 2014). In mammals, 19 secreted WNT ligands have been identified whose function entails binding to Frizzled receptors and LRP co-receptors ((Niehrs 2012) and references herein). Depending on the WNT ligand and its association with Frizzled and LRP, different pathways may be activated.
WNT3A is known to activate the so-called canonical WNT/β-catenin signaling pathway that eventually leads to stabilization of CTNNB1 (β-catenin) and transcription of its targets. By contrast, other WNT proteins can induce CTNNB1-independent pathways including the planar cell polarity (PCP) and WNT/ROR pathways ((Niehrs 2012) and references herein). The cellular context is essential for the choice of the pathway to be activated as exemplified by WNT4. While WNT4 is able to activate canonical WNT signaling in gonadal development (Maatouk et al. 2008, Liu et al. 2010), it appears to induce CTNNB1-independent pathways in the kidney (Burn et al. 2011). In addition to its role in canonical WNT signaling, CTNNB1 is also an important component of adherent junctions, which regulates patterning and morphogenesis. This process is required for gonadogenesis (Fleming et al. 2012) and it has been suggested that WNT4 is involved in CTNNB1 relocalization to the cell membrane during ovarian development (Naillat et al. 2010).
In female patients, heterozygous missense mutations in the WNT4 gene lead to a syndrome characterized by the absence of uterine and fallopian tubes and clinical signs of excess androgen, indicating that WNT4 is required for female reproductive tract development (Biason-Lauber et al. 2004). In mice, Wnt4 has been shown to be important for multiple morphogenetic processes, including formation of the kidney, adrenal and mammary glands and the reproductive tract where it regulates endothelial and steroidogenic cell migration (Jeays-Ward et al. 2003), sex determination (Kim et al. 2006), and female development (Vainio et al. 1999).
RSPO family: four members with a similar structure, activating different receptors and regulating various signaling pathways
The activators of the WNT/β-catenin pathway include RSPO proteins. A few years ago, Rspo1 (roof plate-specific spondin) was the first member of the RSPO family to be identified from screening of genes specifically expressed in the developing spinal cord. It was named for its expression in the roof plate of the neural tube of the mouse embryo and the presence of a protein domain that shares homology with thrombospondin (Kamata et al. 2004). That same year, Xenopus R-spondin2 was isolated in a functional screen for its property to activate WNT signaling (Kazanskaya et al. 2004). Since then, numerous studies have focused on the RSPO family. This family comprises four members (RSPO1, RSPO2, RSPO3, and RSPO4) with pleiotropic functions in embryogenesis, development, and tumorigenesis (Yoon & Lee 2012).
All four RSPOs are cysteine-rich (CR) secreted proteins containing a single thrombospondin type 1 repeat domain. The amino acid sequences of the RSPO proteins are highly conserved (40–60%), especially within vertebrate species, and the four members of the mammalian RSPO family have a similar domain organization (Kamata et al. 2004, Kazanskaya et al. 2004).
Although RSPOs contain an amino-terminal signal peptide, these secreted glycoproteins remain associated with the cell surface (Kazanskaya et al. 2004). This suggests that RSPOs act mainly as juxtacrine factors. Nevertheless, the missing link for clarifying how RSPOs activate the WNT signaling pathway remained yet to be found. In 2011, it has been demonstrated that RSPOs stimulate WNT signaling by binding to the leucine-rich repeat-containing G protein-coupled receptors LGR4, LGR5, and LGR6 (de Lau et al. 2011). RSPO can also bind the transmembrane RING-type E3 ubiquitin ligases ZNRF3 or its homolog RNF43, two negative-feedback regulators of WNT signaling leading to their clearance at the membrane level (Koo et al. 2012). Upon binding, degradation by ubiquitination of the WNT receptors can no longer occur and canonical WNT signaling becomes activated (de Lau et al. 2014). As RSPO1 can bind to ZNRF3 independent of LGR4, this suggests that RSPO1 could transduce WNT signaling without interacting with LGR receptors. However, RSPO1 binds ZNRF3 or RNF43 with a low affinity that is highly enhanced when RSPO1 is already bound to LGR. This implies that LGRs act as recruitment receptors for RSPOs at the membrane. The CR domain of RSPO is necessary and sufficient for this interaction ((de Lau et al. 2014) and references herein).
Moreover, RSPOs can also bind some cell membrane heparan sulfate proteoglycans such as syndecans. Thus, RSPO2 and RSPO3 bind syndecan 4 through their TSP domain (Glinka et al. 2011). In combination with WNT5A, this leads to the activation of the MAP kinase JNK and WNT/PCP signaling pathways during Xenopus gastrulation.
LGRs can also activate non-canonical WNT signaling including the WNT/PCP, G-RhoGTPase pathways (Glinka et al. 2011) and can act independent of RSPO (Fafilek et al. 2013). Whether RSPOs have receptors other than LGR remains yet to be elucidated.
RSPO, a family of genes regulating morphogenesis and maintenance of numerous organs
RSPOs are involved in both embryonic development and homeostasis of adult tissue in many species. Disruption of the human RSPO1 gene in a recessive syndrome is characterized by female-to-male sex reversal, palmoplantar hyperkeratosis, a predisposition to squamous cell carcinoma, corneal opacity, onychodystrophy, and seminoma (Parma et al. 2006, Tomaselli et al. 2008). This suggests that RSPO1 functions as a tumor suppressor. However, RSPO1 can act as a potent specific mitogen by promoting the proliferation of intestinal cells (Kim et al. 2005). During development, Rspo1 plays key roles in sex determination and ovarian differentiation (Chassot et al. 2008a, Tomizuka et al. 2008). Rspo1 has also been shown to be involved in mammary gland formation in mice and angiogenesis in zebrafish. In addition to its different roles in embryogenesis, this gene can also promote organ homeostasis, as evidenced by its role in β-cell neogenesis in the pancreas (Gore et al. 2011, Yoon & Lee 2012).
The importance of Rspo2 in development has also been revealed by loss-of-function experiments. Indeed, null mutations of Rspo2 cause a variety of abnormalities in branchial arches, lungs, and limbs in mice (Yoon & Lee 2012). Abnormal expression of RSPO2 or RSPO3 has been identified in 10% of colorectal cancer cases (Wu et al. 2014). Moreover, RSPO2 can also induce proliferation of another epithelium, the epidermis during keloid scarring (Chua et al. 2011). Rspo3 ablation in mice is lethal at 10 days post coitum (dpc). Although this has led to the description of this gene's key roles in vasculogenesis, angiogenesis, and blood cell speciation (Kazanskaya et al. 2008), it has precluded, thus far, the analysis of its potential role in later development. The congenital mutations in RSPO4, the last member of the RSPO family, result in anonychia in humans (Blaydon et al. 2006).
In the future, it is likely that new physiological functions of RSPOs in organogenesis and maintenance of organs will be unraveled.
Sex determination: two sexes, common progenitors
Sex differentiation is a unique developmental process. Starting from an undifferentiated bipotential gonad, it gives rise to two very specialized organs, the ovary and the testis (Fig. 1). Initially, the morphogenesis of the gonadal primordium appears to be similar between male and female embryos, although the transcriptomic signature is different between both sexes in mice (Jameson et al. 2012a), in which the genital ridges are formed at ∼9.5 dpc (reviewed by Svingen & Koopman (2013)). Proliferation of coelomic epithelial cells, combined with the recruitment of precursor cells from the mesonephroi, leads to the outgrowth of future gonads. The early somatic precursors are specified and proliferate between 10.5 and 12.0 dpc, corresponding to ∼6th week of gestation in humans (Schmahl et al. 2004). At this stage, the gonadal primordium is morphologically similar in male and female embryos. The cells localized to the coelomic domain delaminate and enter the underlying mesenchyme to differentiate into Sertoli or pre-granulosa cells, and a few interstitial cells in XY and XX gonads respectively (Karl & Capel 1998). Thus, the supporting cell types, Sertoli cells in males and granulosa cells in females, stem from a common progenitor population present in the bipotential gonad before its commitment to a testicular or ovarian fate. In the same way, interstitial cell types, Leydig cells in males and theca cells in females, are thought to derive from a common progenitor (Albrecht & Eicher 2001). Supporting and interstitial cell types are responsible for germ cell support and hormone production in the gonad.
In the gonadal primordium, the insulin/insulin-like growth factor (IGF) pathway promotes the expression of key genes involved in gonad formation and sex determination, such as the orphan nuclear receptor SF1 (Pitetti et al. 2013), also known as nuclear receptor subfamily group A member 1 (NR5A1). In turn, Sf1/Nr5a1 expression controls the differentiation into gonadal precursor cells. Moreover, MAP kinase signaling cascades, especially p38MAPK and MAP3K4, lead to the expression of Sry specifically in the XY gonad (Warr et al. 2012). SIX1 and SIX4 transcription factors are also involved both in gonadal precursor differentiation through Sf1/Nr5a1 upregulation, and in Sry upregulation through direct regulation of Zfpm2 (Fog2) expression, a zing finger protein interacting with GATA4 to stimulate Sry transcription (Fujimoto et al. 2013).
Following Sry expression, a genetic cascade is induced, triggering a testis-specific wave of proliferation of pre-Sertoli cells. Together with SF1, SRY promotes the expression of another Sry-related HMG box (SOX) gene, Sox9. SOX9 induces Sertoli cell differentiation and, in turn, testis differentiation (Chaboissier et al. 2004, Sekido & Lovell-Badge 2008). Their expression is then relayed by additional factors ensuring testis development ((Svingen & Koopman 2013) and references herein).
From the bipotential gonad to the ovary: roles of RSPO1 and WNT4
In the bipotential gonad, the insulin family of growth factors and its related receptors are required for gonadal differentiation in both sexes (Pitetti et al. 2013). XX or XY embryos, deficient for both insulin receptor (Insr) and IGF receptor 1 (Igf1r) genes, are characterized by smaller gonads that remain undifferentiated for several days before activation of the ovarian genetic program. In this delayed gonad, expression of Sf1 is decreased and proliferation of the somatic cell progenitors severely affected, indicating that the insulin/IGF signaling is participating in early proliferation of the somatic precursor cells (Pitetti et al. 2013). Similarly, ablation of both Six1 and Six4 genes triggers the development of hypoplastic gonads, suggesting that they are also required for the proliferation of Sertoli or granulosa cell progenitors (Fujimoto et al. 2013).
Rspo1 and Wnt4 also participate together in the early coelomic proliferation and mutant mice, deficient for both of these genes (Rspo1 −/− ; Wnt4 −/− embryos), display a significant decrease in coelomic cell proliferation between 10.5 and 11.5 dpc (Chassot et al. 2012). Nevertheless, Sf1 expression is not significantly downregulated in the Rspo1 −/− ; Wnt4 −/− gonads, suggesting that RSPO1 and WNT4 act downstream of SIX1, SIX4, and SF1. Moreover, Wnt4 expression is significantly reduced in Insr; Igf1r double mutant ovaries (Pitetti et al. 2013), implying either that Wnt4/Rspo1 expression depends directly on the insulin signaling pathway, or that the low level of Wnt4 expression indirectly results from decreased cell proliferation of Insr; Igf1r mutant gonads. However, Igf1r expression is strongly reduced in the Rspo1 −/− ; Wnt4 −/− gonads (Chassot et al. 2012), suggesting that potential feedback interactions may exist between these two signaling pathways.
The proliferation defect of the Rspo1 −/− ; Wnt4 −/− gonads is observed independent of the sex of the embryo. Nevertheless, this process is more severely impaired in mutant XY embryos than in mutant XX embryos. In normal XX embryos, the pre-granulosa cells, which are devoid of Sry expression, display ovarian-specific upregulation of cell cycle inhibitors such as p21kip1, p57kip1, and p27kip1 at 11.5 and 12.5 dpc, which may maintain their proliferation rates at a lower level than the proliferation of XY precursor cells (Cederroth et al. 2007). In addition, in XY embryos, proliferation of coelomic epithelial cells is activated following the expression of Sry and its downstream effectors, leading to an increase in testis size (Schmahl et al. 2004). Defects of proliferation in XY Rspo1 −/− ; Wnt4 −/− gonads lead to hypoplastic testis development.
As described above, RSPO1 is an activator of the β-catenin signaling pathway. Activation of the WNT/β-catenin signaling pathway is detected in the coelomic epithelium of the bipotential gonad in XY and XX embryos at 11.5 dpc, suggesting that WNT4/RSPO1 may activate proliferation of these cells through WNT signaling (Chassot et al. 2012). At this early stage, Lgr5, a recruitment receptor of RSPO1, and Lrp4, a WNT co-receptor involved in epithelial–mesenchymal cell communication during development, are specifically enriched in gonadal XY somatic precursor cells (Jameson et al. 2012a). However, one cannot exclude that RSPO1 and WNT4 act through different receptors. Nevertheless, no functional study of potential receptors has been carried out in the gonad and it remains to be shown that LGRs act as RSPO1 receptors in this organ. How RSPO1 and WNT4 signal synergistically in the bipotential gonad remains another open question.
Rspo1, Wnt4, and female sex determination
RSPO1, WNT4, and WNT/β-catenin signaling: an ovarian-specific activation
At 12.5 dpc, Rspo1 and Wnt4 expression becomes ovary specific (Vainio et al. 1999, Parma et al. 2006). RSPO1 and WNT4 are secreted by the somatic cells of the ovary. In addition, RSPO1 expression is detected at the membrane of both somatic and germ cells at 12.5 and 14.5 dpc (Kocer et al. 2008, Smith et al. 2008), suggesting that it may act on both cell types (see below).
At 12.5 dpc, Wnt4 expression is abolished in the Rspo1 −/− gonad, whereas Rspo1 expression is maintained in the Wnt4 −/− gonad, suggesting that RSPO1/WNT signaling promotes the maintenance of Wnt4 expression (Chassot et al. 2008a, Tomizuka et al. 2008). This loss of Wnt4 expression might highlight abnormal differentiation of pre-granulosa cells.
Following Rspo1 and Wnt4 expression, WNT/β-catenin signaling is activated in a sex-specific manner in somatic and germ cells of the ovary from 12.5 dpc onwards, as evidenced by Axin2 expression, a universal activation marker for this signaling pathway (Chassot et al. 2011, Jameson et al. 2012a). In XY gonads, WNT/β-catenin signaling is downregulated following Sry expression. Consistent with this observation, in vitro studies have shown that SRY can antagonize CTNNB1 by physically interacting with it at the protein level, thus targeting CTNNB1 to nuclear bodies, to trigger its degradation and inhibit CTNNB1-mediated transcriptional activity (Bernard et al. 2008). Nevertheless, this mechanism has not yet been confirmed in vivo during sex determination. Conversely, ectopic activation of CTNNB1 in XY somatic cells results in male-to-female sex reversal by disrupting the testis fate and promoting ovarian development. This study established that β-catenin signaling is a female-determining pathway (Maatouk et al. 2008; Fig. 1). However, the conditional knockout of Ctnnb1 in somatic tissue of the ovary does not impair ovarian differentiation (Manuylov et al. 2008, Liu et al. 2009), suggesting that either the ablation of Ctnnb1 must occur within all ovarian cell types (somatic and germ cells) to promote sex reversal, or Rspo1 and Wnt4 activate pathways other than the WNT/CTNNB1 signaling pathway.
It should also be noted that, in contrast to Axin2 + /LacZ reporter, no positive staining was observed with the TOPGAL mouse model, another reporter line for β-catenin signaling (reporter lines are described by Barolo (2006)). TOPGAL mice carry three multimerized LEF/TCF consensus binding sites in front of the LacZ reporter, which may suggest that CTNNB1 in the developing gonad may act in a LEF/TCF-independent way (A A Chassot, unpublished data).
Rspo1 and Wnt4 control WNT/β-catenin activation in the XX gonads as evidenced by rescue experiments (Chassot et al. 2008a, Maatouk et al. 2008, Liu et al. 2010). However, in the absence of Wnt4 or Rspo1, Ctnnb1 transcriptional activity is only partially decreased as Axin2 remains expressed in the coelomic region of the XX gonads (Chassot et al. 2012), suggesting that there remain other positive regulators of WNT/β-catenin signaling to be identified.
Sex determination is governed by two antagonistic pathways, taken together, involving, on the one hand, WNT signaling, which promotes female differentiation, and, on the other hand, Sry/Sox9/Fgf9, which activate male development (Sekido & Lovell-Badge 2008, Jameson et al. 2012b, Lavery et al. 2012; Fig. 1).
Lack of Rspo1 and Wnt4 both leading to partial sex reversal
Human 46, XX SRY-negative patients carrying mutations in the RSPO1 gene display male-to-female sex reversal with hypospadia and hypogenitalism (Parma et al. 2006). Histological analysis of one gonad from another XX patient carrying an RSPO1 homozygous mutation revealed the presence of both testicular and ovarian tissues (Tomaselli et al. 2008), indicating that sex reversal was partial. Moreover, seminiferous tubules yield germ cell neoplasia and tumor cells were found in the stroma. The ovarian part contained corpora albicans, indicating the presence of a former follicle. Consistent with this finding, adult XX Rspo1 −/− mice display sex reversal and are characterized by ovotestes in which the testicular part, evidenced by the presence of male-like seminiferous tubules with Sertoli cells and interstitial tissue, can form the main part of the mutant gonad, giving it the appearance of a hypoplastic testis (Chassot et al. 2008b). Within the ovarian part, the presence of large abnormal cysts was noticed in 11-week-old mice (Chassot et al. 2008a). No tumors have been described in these mutant mice; this could be due, however, to genetic background susceptibility as shown for other mutant models (Cook et al. 2011).
Homozygous mutations of WNT4 in XX patients are associated with developmental defects such as renal and adrenal agenesis and female-to-male sex reversal (Mandel et al. 2008). Histological analysis of a 19-week-old 46, XX fetus revealed testicular development similar to a control testis, suggesting that complete sex reversal might have occurred. The gonad of another 24-week-old 46, XX affected fetus from the same family displayed partial sex reversal with both ovarian and testicular tissues. Both fetuses carried the same mutation affecting the stability of WNT4 mRNA, a fact that could explain the phenotypic variability (Mandel et al. 2008). In the Wnt4 mouse model, sex reversal with the appearance of cord-like structures is observed around birth. These testis cords are similar to those observed in Rspo1-deficient females and express male markers such as DHH and SOX9 (Vainio et al. 1999, Maatouk et al. 2013). However, in mice, no adrenal agenesis has been described. The differences between humans and mice may be due to functional redundancy between Wnt genes in mice. Wnt2b and Wnt9a are expressed in early gonadal development (Cederroth et al. 2007, Jameson et al. 2012a) and could activate WNT signaling in the absence of WNT4. Moreover, Rspo1 and Wnt4 are also involved in maintaining the quiescent state in pre-granulosa cells after germ cells have entered meiosis ((Maatouk et al. 2013) and below). Their role at this later stage may account for the late appearance of sex reversal approximately at the time of birth, rather than at the time of sex determination.
Absence of Rspo1 or Wnt4 stimulates a male-like vascularization of the gonad
The first morphological abnormality observed in the XX Rspo1 −/− embryo is the presence of a blood vessel at 12.5 dpc that is characteristic of male development. This coelomic vessel is usually formed at the surface of the embryonic testis and is required to pattern the gonad and guide sex cord formation (Cool et al. 2011). The male-like vasculature of mutant XX gonads suggests that Rspo1 is required to inhibit the crosstalk between mesenchymal and endothelial cells. Interestingly, screening performed on isolated endothelial cells indicates that, at 12.5 and 13.5 dpc, Axin2 and known RSPO1 receptors, such as Lgr4, Lgr5, or Lgr6, are not expressed in this cell population (Jameson et al. 2012a), suggesting that RSPO1 acts independent of canonical WNT/β-catenin signaling or indirectly in this context. Thus, the role of Rspo1 in vasculature development remains to be analyzed.
Wnt4 deficiency also induces male-like vascularization of the embryonic XX gonad. Jeays-Ward et al. (2003) demonstrated that WNT4 is required to antagonize both endothelial and steroidogenic cell migration from the mesonephros within the XX gonad, thus preventing the formation of the testis-specific coelomic vessel and the production of steroids. WNT4 is required to stimulate follistatin (Fst) expression. FST is an activin-binding protein expressed in the XX gonad from 11.5 dpc onwards, whereas it is absent in XY gonad. FST is known to antagonize Activin B action to ensure that no testis-specific vasculature is formed (Yao et al. 2004). To date, it is not clear whether the effect of Wnt4 on Fst and endothelial cells is direct or indirect. In Rspo1-deficient XX gonads, Wnt4 and Fst expressions are abolished. The maintenance of Rspo1 expression in Wnt4-deficient gonads indicates that the appearance of the coelomic vessel is not a direct effect of Rspo1 (Chassot et al. 2008a). Therefore, it appears that Rspo1, Wnt4, and Fst have an epistatic relationship during ovarian development, with Rspo1 acting at the top of this genetic cascade. In accordance with this model, XX Rspo1 – Wnt4-deficient mice (double mutant Rspo1 −/− ; Wnt4 −/− mice) are not more severely affected in their phenotypes than the single XX Rspo1 −/− or Wnt4 −/− mutants (Chassot et al. 2012), indicating that Rspo1 and Wnt4 act along the same pathway, at least in somatic cells, at the time of sex determination.
Impaired Rspo1 or Wnt4 expression results in ectopic appearance of adrenal-like steroidogenic cells
The concomitant feature of abnormal vascularization in the XX Rspo1 −/− or Wnt4 −/− gonad is the appearance of steroidogenic cells expressing markers such as Cyp11a1 (Cyp11 α 1) or P450scc, and Hsd17b3 (Hsd17 β 3) from 12.5 dpc onwards (Jeays-Ward et al. 2003, Chassot et al. 2008a). In addition, Cyp21a1 (Cyp21), a marker of adrenal steroidogenic cells, is also expressed in the Wnt4 −/− mutant ovaries, suggesting that these ectopic steroidogenic cells migrate from the neighboring adrenals into the gonad in the absence of Wnt4 (Heikkila et al. 2002, Jeays-Ward et al. 2003). The origin of these cells is not as clear in the XX Rspo1 −/− mutants, where Cyp21a1 does not seem to be expressed (Lavery et al. 2012). In the XX Wnt4 −/− mutant ovaries, testosterone is produced from 12.5 dpc onwards and several genes encoding enzymes regulating steroidogenesis (Hsd17b1, Hsd17b3, and Cyp11a1) are upregulated (Heikkila et al. 2002). Similarly, Hsd17b3 and Cyp11a1 are strongly upregulated in the Rspo1 −/− mutant (Chassot et al. 2008a). It is likely that these steroidogenic cells produce enough hormones to stimulate the development of epididymis, vas deferens, and hypoplastic seminal vesicles in both mutant situations.
RSPO1, WNT4, and the sex determination of female germ cells
Although gonadal somatic cells are specified in situ within the developing gonad, the primordial germ cell precursors, future spermatocytes and oocytes, are committed to the germ cell lineage much earlier during embryonic development. Commitment occurs at ∼6.5 dpc in mice, in the proximal epiblast. Subsequently, germ cells migrate from the allantois through the primitive forming gut to reach and colonize the gonads between 10.5 and 11.0 dpc. Once they have reached their destination, they proliferate and initiate divergent sexual differentiation programs depending on their somatic environment, as well as on intrinsic signals (reviewed by Kocer et al. (2009)). Thus, at ∼13.5 dpc for female embryos, germ cells progressively enter meiosis, whereas they become quiescent in male embryos at ∼14.5 dpc. Reinitiation of proliferation, differentiation, and meiosis occurs in males only after birth, at the onset of puberty.
Time- and sex-specific activation of several genes controls these different events (for a review, see Kocer et al. (2009)). Both Rspo1 and Wnt4 mutants display defects in germ cell development (see below).
RSPO1 regulates proliferation while WNT4 acts as a survival factor in female germ cells
The Rspo1-deficient XX gonad is characterized by a reduced number of germ cells already evidenced at 12.5 dpc and caused by a decrease in germ cell proliferation (Chassot et al. 2011). This is not surprising as Rspo1 has been shown previously to stimulate the proliferation of crypt cells (intestinal progenitor cells) in gain-of-function experiments (Kim et al. 2005). However, proliferation affects only about half of the germ cells and remains partial in Rspo1 mutant mice, indicating that RSPO1 is not the only factor controlling the cell cycle of female germ cells. Interestingly, at 16.5 dpc, germ cells in the Rspo1-mutant ovary resemble G0–G1-arrested gonocytes, and thus are similar to male germ cells at the same stage (Chassot et al. 2011). This suggests that Rspo1 is also required for germ cell sex determination. Again, this is an incomplete phenotype, suggesting that Rspo1 is not the only factor implicated in ovarian germ cell fate. This demonstrates that RSPO1 is likely to affect germ cells in two ways: i) it stimulates their proliferation and ii) it inhibits their entry into a quiescent phase.
In contrast to the Rspo1-deficient ovary, the number of germ cells in the Wnt4 −/− ovary is very similar to that of WT XX gonads between 11.5 and 15.5 dpc (Yao et al. 2004), indicating that they proliferate normally.
Germ cell apoptosis is considered to be a normal process in ovarian development. It is restricted to the medullar region of the ovary and occurs at ∼16.5 dpc in the developing gonad in mice (De Felici et al. 2005). This apoptosis contributes to eliminating oocytes having undergone meiotic defects. Germ cell death also occurs in case of somatic defects, as these supporting cells are crucial for germ cell survival. At 16.5 dpc, in the XX Wnt4-deficient gonad, germ cells undergo massive apoptosis throughout the entire gonad, with ∼90% being lost at this stage, compared with 30% in a WT gonad (Yao et al. 2004). By contrast, no increase in apoptosis was detected in the germ cells of Rspo1 mutant gonads before birth. Thus, 17.5 dpc Rspo1 −/− ovaries display much higher germ cell numbers compared with Wnt4 −/− ovaries, and these germ cells are distributed throughout the entire gonad (Chassot et al. 2011, Maatouk et al. 2013). This indicates that Wnt4 is required for female germ cell survival, unlike Rspo1. In addition, this suggests that downregulation of Wnt4 in Rspo1 KO gonads is not complete, and that persistently low levels of Wnt4 may be sufficient to support germ cell survival (Chassot et al. 2008a, Tomizuka et al. 2008).
RSPO1 and WNT4 regulate meiosis entry of female germ cells, potentially through distinct signaling pathways
Loss of Rspo1 expression also impairs germ cell entry into meiosis (Chassot et al. 2011), and at 14.5 dpc, different germ cell populations can be observed in the mutant XX gonad: about half of the germ cells still express premeiotic and meiotic markers such as Stra8 and Ctdspl (Sycp3). The other half expresses markers of pluripotency, such as Pou5f1 (Oct4), or of quiescence in male germ cells, including Nanos2 and Dnmt3L. In the Wnt4 mutant gonads, the expression of Stra8 is also weak while the expression of Pou5f1 is maintained at 14.5 dpc, indicating a delay in meiosis entry (Naillat et al. 2010). However, the number of meiotic germ cells in the Wnt4 −/− ovaries is normal when compared with a WT ovary at 15.5 dpc (Yao et al. 2004) and surviving germ cells usually express meiotic markers (Maatouk et al. 2013), before undergoing apoptosis through increasing Activin βb expression (Liu et al. 2010). The precise mechanisms of how Rspo1 and Wnt4 affect meiosis entry remain yet to be elucidated.
The retinoic acid (RA) pathway is a key regulator of meiosis entry and ectopic addition of exogenous RA on cultured testis stimulates premature meiosis entry and induces Stra8 expression (Bowles et al. 2006, Koubova et al. 2006). Stimulated by retinoic acid 8 (STRA8) gene is necessary for DNA replication preceding meiosis. Despite these convincing in vitro data, a study aiming to monitor the endogenous role of RA in fetal mouse ovary in vivo reported that RA is required to induce neither Stra8 nor meiosis (Kumar et al. 2011). Cyp26B1 is a RA-metabolizing cytochrome P450 enzyme expressed in the male embryonic gonad from 12.5 dpc onwards and absent from the ovary. In XY mice deficient for Cyp26b1 (Cyp26b1 −/− mice), germ cells prematurely enter into meiosis only to undergo massive apoptosis from 13.5 dpc onwards, despite generally normal Sertoli and Leydig cell specification (MacLean et al. 2007). Thus, CYP26B1 appears to maintain low RA levels in the male embryonic gonad, in order to inhibit entry of germ cells into meiosis and prevent their apoptosis.
Stra8 is downregulated in the Rspo1 −/− XX gonad, while Cyp26b1 is not upregulated (Chassot et al. 2011). The absence of Cyp26b1 expression suggests that RA is usually present in mutant Rspo1 −/− gonad and that the reduced expression of Stra8 is independent of RA. Nevertheless, further experiments using RA reporter assays are required to confirm definitively RA presence in mutant gonads. It has been recently shown that germ cell differentiation and entry into meiosis are two dissociable events that contribute to the development of a functional oocyte (Dokshin et al. 2013). We currently hypothesize that Rspo1 participates in establishing female germ cell identity, allowing germ cells to become competent to respond to meiotic signals. Another hypothesis would be that RA regulates RSPO1 signaling in the developing ovary to promote meiotic gene expression. Further experiments will be required to analyze the relative importance of the two pathways and determine their relationship.
RSPO1, WNT4, and β-catenin signaling in the female germ cell: transcriptional activity and cell adhesion
Canonical β-catenin signaling is activated in XX germ cells, as evidenced by Axin2 expression (Chassot et al. 2011, Jameson et al. 2012a). The active form of CTNNB1 and its transcriptional activity detected in the nuclei of control XX germ cells are lost in germ cells of the Rspo1 −/− mutant ovary. Thus, RSPO1 appears to activate canonical β-catenin signaling in both somatic cells and germ cells of the ovary (Chassot et al. 2011). Consistent with this model, RSPO1 protein is detected on the membrane of both cell types (Kocer et al. 2008, Smith et al. 2008), suggesting that CTNNB1 is directly activated in female germ cells to regulate their fate. Consistent with this hypothesis, germ cells of the XX embryos normally enter meiosis when Ctnnb1 is specifically ablated in somatic cells of the developing ovary (Sf1:cre; Ctnnb1 flox/flox mice; Manuylov et al. 2008, Liu et al. 2009). This suggests that WNT/β-catenin signaling in somatic ovarian cells does not affect germ cell fate. Instead, the intrinsic activation of CTNNB1 within the germ cells contributes to female germ cell fate.
Early gonadal patterning is strictly correlated with the cell-specific membrane expression of CTNNB1 and CDH (cadherins) in adherent junctions during sex determination (Fleming et al. 2012). In the absence of Rspo1, β-catenin signaling is downregulated in the nucleus and CTNNB1 seems to be relocalized to the cell membrane, where it may function as an adhesion molecule to organize germ cells in clusters and establish cell–cell contacts, a feature characteristic of testes at the same stage (Chassot et al. 2008a). RSPO1 could also directly act on cell surface adhesion molecules through its thrombospondin 1 domain to regulate germ cell adhesion differentially between XX and XY gonads. Interestingly, Wnt4-deficient gonads also display enhanced cell-to-cell contacts (Naillat et al. 2010), suggesting that Wnt4 is also involved in cell adhesion. Whether RSPO1 and WNT4 directly regulate the expression and/or localization of cell adhesion molecules or whether this increased adhesion is a consequence of an abnormal germ cell identity remains yet to be investigated.
Rspo1/Wnt4 pathways prevent pro-spermatogonia differentiation and sex reversal, in turn
It has been demonstrated that the decision of germ cells to enter meiosis or to become quiescent depends on their somatic environment (McLaren 1991). Conversely, the involvement of germ cells in the differentiation of supporting cells is less documented.
In Rspo1 −/− and Wnt4 −/− mutant ovaries, Pou5f1 expression is maintained as it is in quiescent male germ cells, and the expression of Nanos2, a marker of male germ cells, is upregulated (Chassot et al. 2011). This indicates that in both mutants, germ cells have undergone abnormal differentiation. In addition, germ cells of the Rspo1 −/− and Wnt4 −/− ovary have the morphology of G0–G1 arrested gonocytes at 16.5 and 18.5 dpc respectively ((Chassot et al. 2011) and unpublished data). This suggests that, while proliferation is not affected by the absence of Wnt4 expression, a proportion of germ cells does not usually progress in meiosis and become quiescent. It is noteworthy that these masculinized germ cells are always found within or in close vicinity of developing testicular cords. However, these masculinized germ cells differentiate before becoming mature Sertoli cells. This is in contrast to normal gonadal development during which somatic cells determine the sexual fate of germ cells (Adams & McLaren 2002).
Meiotic germ cells antagonize male development in embryonic mouse ovaries (Yao et al. 2003) and female germ cells exert strict control on their surrounding environment under the influence of RSPO1 and WNT4 (Maatouk et al. 2013). Thus, Wnt4-deficient as well as Rspo1-deficient ovaries fail to undergo sex reversal, when germ cells are depleted from the gonad at mid-gestation by bisulfan treatment (Maatouk et al. 2013). This implies that abnormally differentiated germ cells of these mutants participate in Sertoli cell differentiation and sex reversal. Moreover, the improvement of cell–cell contacts through sex-specific localization of adhesion molecules such as CTNNB1 and CDH1 (E-cadherin) may play a role in this process. Thus far, the signals generated by germ cells upon RSPO1 and WNT4 action and the pathways induced by these signals are unknown.
Lack of Rspo1 or Wnt4 triggers Sertoli cell differentiation around birth
In the developing ovary, pre-granulosa cells are usually in a quiescent state: they are mitotically arrested and they express cell cycle inhibitors such as p27kip1. In pubescent females and throughout reproductive life, quiescent pre-granulosa cells are organized in ovarian follicles that resume proliferation when the follicle is recruited and matures (Edson et al. 2009). In Rspo1 and Wnt4 mutant gonads, the pre-granulosa cells are not quiescent and differentiate precociously and transiently into proliferative granulosa cells in an anterior-to-posterior wave across the ovary. Then, approximately at the time of birth, germ cells promote these cells to transdifferentiate into Sertoli cells (Maatouk et al. 2013) that, with the onset of Sox9 expression, form tubular structures reminiscent of seminiferous tubules (Vainio et al. 1999, Chassot et al. 2008a; Fig. 2). Surprisingly, this transdifferentiation can still occur in the absence of Sry or Sox9, indicating that other genes can induce female-to-male sex reversal (Lavery et al. 2012). Together, these data indicate that Rspo1 and Wnt4 are involved in maintaining the immature progenitor state of the ovarian supporting cells.
Rspo1, Wnt4, and Foxl2 relationship in ovarian development
Mutations in FOXL2 in humans lead to blepharophimosis/ptosis/epicanthus inversus syndrome (BPES), an autosomal dominant genetic disorder characterized by drooping eyelids and/or premature ovarian failure in women (Crisponi et al. 2001). FOXL2 is required for the transition from primordial to primary follicle approximately at the time of birth and Foxl2 −/− mice show progressive apoptosis of functional granulosa cells, leading to follicular depletion and ovary atresia (Schmidt et al. 2004). Interestingly, conditional ablation of Foxl2 in adult mice induces reprograming of granulosa cells into Sertoli cells, demonstrating that maintenance of the sexual identity of supporting cells is determined by the antagonistic action of SOX9 and FOXL2 (Uhlenhaut et al. 2009).
In the embryonic mouse ovary, Foxl2 expression is partly dependent on RSPO1 and CTNNB1 (Manuylov et al. 2008, Auguste et al. 2011), whereas it is normally expressed in the Wnt4 −/− ovary (Manuylov et al. 2008), suggesting that Foxl2 is expressed in two different somatic cell populations. Moreover, simultaneous deletion of Foxl2 −/− and Wnt4 −/−, as well as Foxl2 −/− and Rspo1 −/−, promotes early female-to-male sex reversal (Ottolenghi et al. 2007, Auguste et al. 2011), indicating that Foxl2 and Rspo1/Wnt4 have complementary roles during ovarian development. FOXL2 mutations are also sufficient to induce XX sex reversal in the goat (Pailhoux et al. 2001). The different phenotypes observed in Foxl2 mutant mice and goats may indicate a difference in the identity/maturation time of pre-granulosa cells between these species.
R-spondin and Wnt4 genes in ovarian differentiation throughout evolution
Despite the diversity of ‘master sex-determining genes’ among species, the key role of WNT signaling in sexual differentiation in humans and mice raises questions about the conservation of this pathway during evolution. Expression and potential roles of WNT genes in other species are largely unclear. WNT4 has been found in various metazoan phyla and the embryonic expression pattern is well conserved throughout vertebrate evolution ((Nicol et al. 2012) for references). RSPOs are expressed in vertebrates and invertebrates, such as hemichordates, chordates, and echinoderms, but screening of Drosophila melanogaster and Caenorhabditis elegans failed to identify any RSPO homolog (Yoon & Lee 2012). However, the WNT/β-catenin pathway is required for male cell fate commitment in the gonad of C. elegans (Kalis et al. 2010).
RSPO1 is upregulated in humans and goats during critical stages of ovarian development. In goats, RSPO1 is expressed in both male and female mesonephroi and the level of expression peaks just before germ cell entry into meiosis (55 dpc; Kocer et al. 2008). In addition to RSPO1, RSPO2 is expressed at the time of folliculogenesis (from 70 dpc until before birth) in goat gonads, while RSPO4 begins to be very faintly expressed in 50 dpc ovaries (Kocer et al. 2008).
In humans, RSPO1 is expressed between 6 and 9 weeks after conception in the fetal ovary (Tomaselli et al. 2011), whereas WNT4 is detected in both the fetal and adult ovary at different follicular stages (Jaaskelainen et al. 2010) as in mice. In marsupials, the WNT4 mRNA level increases after birth, reaching a peak of expression by days 9–13 post partum when the ovarian cortex and medulla become distinguishable (Yu et al. 2006).
The amniotes split into the ancestral lineages of mammals and reptiles ∼320 million years ago. In birds, some reptiles, and insects, sex determination depends on the ZZ/ZW chromosomes with ZZ homogametic males and ZW heterogametic females. In ZZ embryos, Rspo1 and Wnt4 expression levels remain low while Rspo1 and Wnt4 are specifically expressed in chicken ovaries (Smith et al. 2008). Wnt4 mRNA is detected in oocytes of postnatal chick ovaries. Moreover, Rspo1 expression is downregulated when sex reversal is induced by hormone manipulation of ZW embryos (Lambeth et al. 2013). Thus, Rspo1 and Wnt4 also seem to be key determinants of the female differentiation pathway in the chicken.
Besides XY or ZW genetic sex determination, temperature has a dominant role in the establishment and maintenance of sexual fate in some reptiles. The thermosensitive period of the red-eared slider turtle (Trachemys scripta) spans from embryonic stages 14–19/20, during which a 31 °C temperature of incubation triggers Rspo1 upregulation and the production of female hatchlings (Smith et al. 2008). Moreover, in this species, Wnt4 transcription appears to be upregulated following estrogen signaling (Mork & Capel 2013).
In Rana rugosa, Wnt4 is not expressed in a sexually dimorphic fashion in the early stages of gonadal development. Wnt4 is transcribed in the embryos at the late gastrula stage, and its expression is maintained until the undifferentiated gonad differentiates into a testis or an ovary (Oshima et al. 2005).
In mammals, testes and ovaries retain the ability to transdifferentiate in adult life, as documented using mouse mutants (Uhlenhaut et al. 2009, Matson et al. 2011). The plasticity of the adult gonad is most evident in fish species that spontaneously change sex in adult life, shifting from male to female (protandrous) or female to male (protogynous). In the protandrous Black Porgy (Acanthopagrus schlegeli), wnt4 expression is increased during the late ovarian growth and during the transition from male to female (Wu & Chang 2009), suggesting a conserved role of wnt4 in ovarian development. While in mammals, the transdifferentiation of adult gonads not only affects the somatic cells, but also is associated with sterility; sex-changing fishes retain their fertility. Zebrafish, salmonids, and medakas are teleost fishes, but they have been separated for 115–200 million years. In zebrafish, Rspo1 expression starts to be detected at 30 day post fertilization (dpf) in the somatic cells of the undifferentiated gonads, until 150 dpf in the ovary (granulosa cells and theca cells) and in the testis (Leydig cells), while its expression is present only from 30 to 60 dpf in female germ cells and until 150 dpf in male germ cells (Zhang et al. 2011). Unlike other teleost fishes, the salmonid rainbow trout (Oncorhynchus mykiss) has a third wnt4 gene (Nicol et al. 2012). Protein comparisons with WNT4 from other vertebrates show a high level of amino acid identity (>80%). However, wnt4 is not strongly expressed in the ovary during early gonadal differentiation in the rainbow trout. Thus, gonadal differentiation may involve other members of the Wnt gene family. Indeed, other wnt genes have been shown to display a sex-specific expression at different stages of development in the rainbow trout (Nicol & Guiguen 2011). Moreover, transgenic inhibition of wnt signaling results in male-biased sex ratios, suggesting that Wnt signaling is a conserved key pathway during gonadal differentiation in zebrafish (Sreenivasan et al. 2014). In medaka (Oryzias latipes), Rspo1 and Rspo2 are prominently expressed in both germ cells and somatic cells (Zhou et al. 2012). Their functions in this species remain yet to be clarified. Recently, three genes belonging to the WNT pathways, wnt4, rspo1, and ctnnb1, have been identified in Latimeria menadoensis transcripts and in the Latimeria chalumnae genome, two coelacanths considered to be true ‘living fossils’ (Forconi et al. 2013).
Despite the diversity of sex determinants in these different phyla, WNT/RSPO1/β-catenin signaling appears as an ancient, conserved pathway of ovarian determination.
Remaining questions and prospects
It is clear today that the RSPO1/WNT/β-catenin pathway acts at the top of the ovarian differentiation cascade (Fig. 1). The development of male or female traits appears to be antagonistic, and two key events orient the developing gonad toward a male or a female fate: female (or male) genes need to be expressed and male (or female) genes actively repressed (Jameson et al. 2012b; Fig. 1). The molecular mechanisms regulating the expression of Rspo1 and Wnt4 in the bipotential gonad, as well as those involved in their ovary-specific upregulation remain yet to be discovered.
Thus far, it is speculated that Rspo1 and Wnt4 act through the WNT signaling pathway during gonadal development. However, when Ctnnb1 is deleted in somatic cells (Sf1cre; Ctnnb1 flox/flox mice), the gonads do not undergo sex reversal at 18.5 dpc while sex reversal of somatic cells highlighted by Sox9 expression is already noticeable in the Rspo1 −/− and Wnt4 −/− XX gonads at this stage (Chassot et al. 2011, Maatouk et al. 2013). It is likely that sex reversal is promoted by a signal derived from abnormally differentiated germ cells in Rspo1 and Wnt4 mutants, whereas this signal is not produced in Sf1cre; Ctnnb1 flox/flox embryos. Thus, the differentiation of germ cells is unaffected before they undergo apoptosis in Sf1cre; Ctnnb1 flox/flox gonads unlike in Rspo1 −/− and Wnt4 −/− gonads (Manuylov et al. 2008, Liu et al. 2009, Naillat et al. 2010, Chassot et al. 2011). However, it is not clear as to how germ cells promote sex reversal to occur in Rspo1 and Wnt4 knockout XX gonads (Maatouk et al. 2013). Germ cells might be involved in the early differentiation of granulosa cells via a paracrine signal after which these granulosa cells could transdifferentiate into Sertoli cells. Whether pre-granulosa cells differentiate early in Sf1cre; Ctnnb1 flox/flox XX gonads has not been investigated thus far.
The most obvious difference between Rspo1 −/− and Wnt4 −/− XX mutant embryos is the differentiation of germ cells. Thus, RSPO1 appears to contribute to female germ cell proliferation, while WNT4 rather acts as a survival factor, protecting them from apoptosis, either directly or indirectly. Although both secreted factors are able to activate WNT/β-catenin signaling, it becomes obvious that they may act through different receptor complexes (LGR/RNF43 for RSPO1 and LRP/Fzl for WNT4) that may be differentially expressed depending on the cell type. Nevertheless, we cannot exclude that these molecules act through distinct signaling pathways in the gonad, which remains yet to be identified. Moreover, their specific receptors and the genetic network they are regulating during sex determination still need to be clarified.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the review.
This work was supported by the French Government (National Research Agency, ANR) through the ‘Investments for the Future’ LABEX SIGNALIFE: program reference #ANR-11-LABX-0028-01 and by L'Agence Nationale pour la Recherche (ANR-09-GENM-009-03 GENIDOV, ANR-13-BSV2-0017-02 ARGONADS) and Association pour la Recherche sur le Cancer (PJA 20131200236).
The authors would like to apologize to all researchers whose work is not listed in the References; unfortunately, we did not have enough space for this. The authors are grateful to Andreas Schedl for his thoughtful contribution to this review and to Catherine Ungar for proofreading.
FafilekBKrausovaMVojtechovaMPospichalovaVTumovaLSloncovaEHuranovaMStancikovaJHlavataASvecJ2013Troy, a tumor necrosis factor receptor family member, interacts with lgr5 to inhibit wnt signaling in intestinal stem cells. Gastroenterology144381–391. (doi:10.1053/j.gastro.2012.10.048)
JaaskelainenMPrunskaite-HyyrylainenRNaillatFParviainenHAnttonenMHeikinheimoMLiakkaAOlaRVainioSVaskivuoTE2010WNT4 is expressed in human fetal and adult ovaries and its signaling contributes to ovarian cell survival. Molecular and Cellular Endocrinology317106–111. (doi:10.1016/j.mce.2009.11.013)