During mammalian sex determination, expression of the Y-linked gene Sry shifts the bipotential gonad toward a testicular fate by upregulating a feed-forward loop between FGF9 and SOX9 to establish SOX9 expression in somatic cells. We previously proposed that these signals are mutually antagonistic with counteracting signals in XX gonads and that a shift in the balance of these factors leads to either male or female development. Evidence in mice and humans suggests that the male pathway is opposed by the expression of two signals, WNT4 and R-SPONDIN-1 (RSPO1), that promote the ovarian fate and block testis development. Both of these ligands can activate the canonical Wnt signaling pathway. Duplication of the distal portion of chromosome 1p, which includes both WNT4 and RSPO1 , overrides the male program and causes male-to-female sex reversal in XY patients. To determine whether activation of β-catenin is sufficient to block the testis pathway, we have ectopically expressed a stabilized form of β-catenin in the somatic cells of XY gonads. Our results show that activation of β-catenin in otherwise normal XY mice effectively disrupts the male program and results in male-to-female sex-reversal. The identification of β-catenin as a key pro-ovarian and anti-testis signaling molecule will further our understanding of the mechanisms controlling sex determination and the molecular mechanisms that lead to sex-reversal.
The murine gonad forms at 10.5 days post-coitum (dpc) and is initially indistinguishable between XY and XX embryos. At this stage, the cells of the gonad are bipotential and can follow either the male or female pathway. Many lines of evidence suggest that sex determination in mammals centers on a cell fate decision in the supporting cells of the gonad. These cells can either become Sertoli cells in the testis, or their ovarian counterpart, follicle cells. Sex determination occurs in XY embryos when Sry is expressed in supporting cell precursors, which leads to the rapid upregulation of SOX9. Establishment of SOX9 expression in this precursor population triggers a fate decision in these cells, which leads to Sertoli cell differentiation (reviewed in 1 ).
SOX9 performs all known downstream functions of SRY, and its expression is sufficient to cause complete female-to-male sex reversal of XX embryos ( 2 , 3 ). In XY gonads, expression of SRY and SOX9 leads to increased proliferation of the coelomic epithelium, reorganization of somatic cells into testis cord structures and migration of endothelial cells into the gonad from the mesonephros, forming a male-specific blood vessel along the surface of the gonad (reviewed in 4 ).
At a time when male embryos undergo rapid differentiation, reorganization of the female gonad is subtle, with few obvious changes. To date, a female sex-determining gene has not been identified. Several years ago mice lacking Wnt4 were generated; these mice exhibited a partial female-to-male sex-reversal, but lacked embryonic testis cords ( 5 ). Consistent with this result, overexpression of Wnt4 alone in XY gonads does not result in sex-reversal ( 6 , 7 ), suggesting that Wnt4 is not the sole female sex-determining gene and that additional factors are required to effectively disrupt the male pathway and promote female development.
Recent findings implicated an additional gene in female sex-determination. Human patients carrying a mutation in the R-SPONDIN1 ( RSPO1 ) gene exhibited male-to-female sex reversal ( 8 ). Similar to WNT ligands, RSPO proteins also can affect the canonical WNT signaling pathway. Mice lacking Rspo1 exhibit partial female-to-male sex-reversal, similar to mice lacking Wnt4 , suggesting that these genes act within the same pathway to activate β-catenin ( 9 , 10 ). The phenotype of Rspo1 mutant ovaries could be rescued by stabilized β-catenin expression in ovarian somatic cells, suggesting that β-catenin acts downstream of Rspo1 to block testis development in XX gonads ( 9 ). Interestingly, an XY human patient with a duplication of chromosome 1p, which includes both the WNT4 and RSPO1 loci, showed male-to-female sex-reversal ( 11 ), suggesting that the elevation of signals downstream of WNT4 and RSPO1 may be sufficient to override the testis pathway.
To test the hypothesis that β-catenin signaling antagonizes the establishment of the testis pathway, we ectopically expressed a dominant-stable allele of β-catenin in XX and XY gonads. Here, we show that stabilization of β-catenin in XY gonads is sufficient to disrupt the male pathway and promote ovarian development. These results support a role for β-catenin in sex determination and provide a molecular mechanism to account for XY sex-reversal cases that result from gain-of-function mutations.
Stabilization of β-catenin causes male-to-female sex reversal
To determine whether β-catenin can disrupt male development and drive female development in XY gonads, we have taken advantage of two previously generated mouse lines. The Sf1-cre transgenic mice express Cre recombinase in the gonadal somatic cells of both sexes ∼11.5 dpc ( 12 ). A cross between this transgenic and Catnb lox(ex3) (β -cat fl.ex3 ) mice ( 13 ) results in double heterozygotes (β -cat fl.ex3 /+; Sf1-cre ), in which β-catenin is stabilized in gonadal somatic cells. Litters from this cross yielded approximately equal numbers of XX and XY mice (52 and 48%, respectively; n = 6 litters). However, on the basis of the phenotype of the external genitalia, 70% of the progeny appeared female and only 30% appeared male. We found that 100% of XY β -cat fl.ex3 /+; Sf1-cre mice had completely feminized external genitalia and were indistinguishable from XX wild type and XX β -cat fl.ex3 /+; Sf1-cre littermates (Fig. 1 , row 1). Internal reproductive tracts of XY β -cat fl.ex3 /+; Sf1-cre mice lacked male organs and instead resembled those of females (Fig. 1 , row 2). The sex-reversed gonads were located near the kidneys similar to wild-type ovaries, and reproductive tracts were found to have a uterus and vagina at P0 and 3 weeks after birth.
Histological analysis at P0, prior to the formation of ovarian follicles, showed that both wild-type and XX β -cat fl.ex3 /+; Sf1-cre ovaries contained many meiotic germ cells ( Supplementary Material, Fig. S1 ). Wild-type male testes had numerous cords containing germ cells. In contrast, XY β -cat fl.ex3 /+; Sf1-cre gonads morphologically resembled ovaries and completely lacked testis cords. No germ cells were detected at this stage. At 3 weeks of age, histological sections of wild-type female ovaries showed numerous ovarian follicles (Fig. 1 , row 3). XX β -cat fl.ex3 /+; Sf1-cre mice also contained ovarian follicles; however, large hemorrhagic areas were present, resembling the previously described phenotype of β -cat fl.ex3 /+; Amhr2-Cre mice ( 14 ). Although wild-type male testes had cords containing many germ cells, XY β -cat fl.ex3 /+; Sf1-cre gonads structurally resembled ovaries at P1, but lacked follicles, likely due to the absence of germ cells.
Stabilization of β-catenin in XY gonads disrupts formation of testis cords
We next examined the effect of stabilized β-catenin expression on the initiation of the testis pathway during the fetal stages when sex determination occurs. Despite the lack of testis cords, a coelomic vessel is formed normally along the surface of the XY β -cat fl.ex3 /+; Sf1-cre gonad (Fig. 2 , white arrow). Additionally, the size of the gonad resembled a wild-type XY gonad, suggesting that the early proliferation of the coelomic epithelium, which occurs specifically in XY gonads, occurs in XY β -cat fl.ex3 /+; Sf1-cre gonads. Both of these aspects of the testis pathway normally occur downstream of SOX9.
Stabilization of β-catenin had a dramatic effect on SOX9 expression (Fig. 2 A). At 11.5 dpc, minor differences in SOX9 expression were detectable between XY control and β -cat fl.ex3 /+; Sf1-cre gonads. However, SOX9 expression rapidly declined and was limited to a small number of cells between 12.5 and 13.5 dpc. At 12.5 and 13.5 dpc, testis cords had organized and were clearly apparent in the XY wild-type control gonads; however, XY β -cat fl.ex3 /+; Sf1-cre gonads lacked testis cord structure. (We previously showed that expression of the Sf1-cre transgene at 11.5 dpc is restricted to a small number of cells ( 15 )). The delayed stabilization of β-catenin could account for the transient accumulation of SOX9 at 11.5 dpc. Chaboissier et al . ( 16 ) have shown that delayed deletion of Sox9 with an independent Sf1-cre transgene sometimes allows for the formation of the coelomic vessel, suggesting that transient expression of SOX9 is sufficient to initiate the signals that drive vessel formation.
To confirm the ectopic stabilization of β-catenin in XY β -cat fl.ex3 /+; Sf1-cre gonads, we examined β-catenin expression at 12.5 dpc. In control XY gonads, β-catenin was expressed on germ cell membranes and some Sertoli cell membranes. In the XY β -cat fl.ex3 /+; Sf1-cre gonads, β-catenin was extensively expressed throughout the gonad in somatic cell cytoplasm and nuclei (Fig. 2 B). Additionally, immunostaining for the Sertoli cell marker anti-Mullerian hormone (AMH) showed a rapid loss of AMH expression in the mutant gonads. Therefore, stabilization of β-catenin in XY β -cat fl.ex3 /+; Sf1-cre gonads disrupted the establishment of the Sertoli cell lineage, marked by the loss of both SOX9 and AMH.
In order to investigate the effect of earlier stabilization of β-catenin in all cells of the gonad and mesonephros, we performed in vitro culture of whole gonads in the presence of a β-catenin activator. Lithium chloride (LiCl), an inhibitor of GSK3β, activates β-catenin by blocking the pathway which leads to its phosphorylation and subsequent degradation. When XY gonads carrying an ECFP transgene under the control of the Sox9 promoter ( Sox9 -ECFP) were cultured in control media, a male-specific blood vessel was formed normally as well as testis cords consisting of Sox9 -ECFP positive cells surrounding germ cells. β-catenin was expressed in germ cell membranes (Fig. 3 A). When treated with 50 m m LiCl, XY gonads expressed β-catenin throughout the gonad, and the Sox9 -ECFP reporter was rapidly downregulated. This downregulation of Sox9 is consistent with the rapid downregulation of SOX9 in XY β -cat fl.ex3 /+; Sf1-cre gonads.
Sex-reversal due to exposure to LiCl in culture was variable in our samples; therefore, in vitro culture of gonads with or without LiCl was initiated between 16 and 20 tail somites (∼11.5 dpc), to examine the effects of LiCl at different stages. XX and XY gonads were cultured in the presence of 50 m m LiCl or in control medium for 24 h and then assayed for three sexually dimorphic characteristics: the presence of a coelomic vessel, expression of SOX9 and gonad size. LiCl-treated XX gonads appeared normal or similar to XX control gonads (data not shown). XY gonads cultured in LiCl lacked an organized coelomic vessel, had decreased numbers of SOX9 expressing cells and resembled female gonads in size; these effects were more severe with cultures of the earliest staged gonads (Fig. 3 B and C). Although we cannot rule out the possibility that the effects of inhibiting GSK3β, independent of β-catenin activation, cause some of the observed results, the similarities with the in vivo results suggest that stabilization of β-catenin is sufficient to disrupt male sex determination. Additionally, this effect is more severe when β-catenin activation occurs early in all cells of the XY gonad and is not limited to the cells that express the Sf1-cre transgene.
Sex-reversed XY gonads upregulate markers of ovarian development
To determine whether β-catenin activation is sufficient to induce an ovarian cell fate in XY cells, we examined the expression of ovarian markers. FOXL2 is expressed specifically in XX somatic cells starting at 12.5 dpc and is a marker of pre-granulosa cells ( 17 ). Examination of FOXL2 in 12.5 and 13.5 dpc XX and XX β -cat fl.ex3 /+; Sf1-cre gonads showed a wild-type expression pattern, with FOXL2 expressed in the nuclei of somatic cells (Fig. 4 ). Interestingly, FOXL2 appeared to be expressed in more cells of XX β -cat fl.ex3 /+; Sf1-cre gonads, compared with XX controls. Additionally, mutant female gonads were larger than XX controls. Although no staining was detected in XY gonads, XY β -cat fl.ex3 /+; Sf1-cre gonads showed a small number of cells with nuclear FOXL2 staining at 12.5 dpc and large numbers of FOXL2-expressing cells by 13.5 dpc.
We analyzed two additional aspects of ovarian development in the XY β -cat fl.ex3 /+; Sf1-cre gonads: expression of XX somatic cell markers and germ cell entry into meiosis. By in situ hybridization, Bmp2 ( 18 ) was detected in the coelomic domain of wild-type and XY β -cat fl.ex3 /+; Sf1-cre gonads at 14.5 dpc (Fig. 5 A). The expression of Rspo1 , Wnt4 and Follistatin ( Fst ) ( 18 ) was examined by comparing mRNA expression levels between 13.5 dpc control and XY β -cat fl.ex3 /+; Sf1-cre gonads (Fig. 5 B). By quantitative PCR (qPCR), Rspo1 expression was found to be lower in mutant gonads, compared with controls, whereas Wnt4 and Fst expression was upregulated. We next examined germ cell entry into meiosis. At 15.5 dpc, we detected nuclear accumulation of γH2AX (a characteristic of meiotic germ cells) in XY β -cat fl.ex3 /+; Sf1-cre gonads, similar to XX controls (Fig. 5 A). As germ cell differentiation depends on the somatic environment and is independent of the germ cell's own sex chromosomes ( 19–21 ), these results suggest that stabilization of β-catenin in somatic cells of XY gonads results in the activation of the ovarian pathway.
Ectopic stabilization of β-catenin rescues the Wnt4 mutant phenotype
Wnt4 is necessary for ovarian development and for antagonizing aspects of male development. In XX Wnt4 null gonads, a coelomic vessel forms and steroidogenic cells are present, both normal characteristics of male, and not female gonads ( 6 ). To determine whether WNT4 signals through the canonical WNT signaling pathway, we attempted to rescue the partial sex-reversal of XX Wnt4 null gonads by stabilization of β-catenin. XX control and Wnt4 null gonads were explanted at 11.5 dpc and cultured in the presence of LiCl-containing medium, or in control medium (Fig. 6 ). After 24 h of culture, male-specific vasculature was apparent in XX gonads lacking Wnt4 . However, XX Wnt4 null gonads treated with LiCl lacked a coelomic vessel. These results suggest that stabilization of β-catenin at 11.5 dpc throughout the Wnt4-/- gonad and mesonephros can directly or indirectly rescue the Wnt4 mutant phenotype and block the formation of the male coelomic vessel.
In the XX gonad, two secreted ligands, WNT4 and RSPO1, are capable of activating the β-catenin canonical signaling pathway, and loss of either Wnt4 or Rspo1 in mice results in a partial sex-reversal ( 5 , 9 , 10 ). The existence of human sex-reversed XY males carrying chromosomal duplications suggests that overexpression of female sex-determining genes could lead to sex-reversal. Attempts to sex reverse XY mice by overexpression of Wnt4 have been unsuccessful, suggesting that Wnt4 alone is insufficient to override the male pathway ( 6 , 7 ). However, an XY patient carrying a duplication that includes WNT4 and RSPO1 exhibited sex-reversal ( 11 ), leading to the hypothesis that upregulation of both of these ligands may be required to antagonize male development. Here, we show that stabilization of β-catenin, the downstream effector of WNT4 and RSPO1 signaling, was sufficient to disrupt the male pathway in XY gonads. This led to a loss of SOX9 and AMH expression, a failure of testis cord formation and the increased expression of several ovarian somatic cell markers including FOXL2, Bmp2 , Wnt4 and Fst , suggesting that the somatic lineages have switched from a male to female fate.
Although we observed male-to-female sex-reversal of XY β -cat fl.ex3 /+; Sf1-cre gonads, a male-specific vasculature was formed. However, when β-catenin was stabilized in XY gonads cultured in the presence of LiCl, formation of the coelomic vasculature was disrupted. This discrepancy is likely related to the timing of β-catenin stabilization relative to the time of establishment of the male program. SOX9 expression is sufficient to initiate all subsequent aspects of testis differentiation, including the migration of mesonephric endothelial cells ( 2 , 3 ). Transient SOX9 expression in XY β -cat fl.ex3 /+; Sf1-cre gonads may initiate the male vascular program, in contrast to the case in early LiCl-treated XY samples, where SOX9 expression is absent. However, we cannot rule out the possibility that β-catenin signaling has a more direct role in vascular development. Wnt4 has been shown to block the migration of endothelial cells into XX gonads ( 6 ). Thus, the early stabilization of β-catenin in LiCl-treated gonads may have a more direct effect on blocking endothelial migration.
XX gonads lacking Wnt4 exhibit a partial sex-reversal and lose expression of ovarian somatic markers including Fst and Bmp2 ( 5 , 18 ). As these genes are thought to be downstream of Wnt4 signaling, we hypothesized that they would be upregulated by β-catenin signaling. Additionally, gonads lacking Rspo1 show decreased Wnt4 expression, suggesting that β-catenin signaling might also positively regulate Wnt4 ( 9 , 10 ). We examined expression of these genes in XY β -cat fl.ex3 /+; Sf1-cre gonads and found that increased β-catenin signaling leads to increased expression of Wnt4 , Fst and Bmp2 . However, Rspo1 was downregulated, suggesting that its expression is negatively correlated with high levels of β-catenin signaling. Additionally, we observed that both XX and XY β -cat fl.ex3 /+; Sf1-cre gonads were larger than controls. This might suggest that β-catenin promotes the survival or proliferation of an ovarian somatic cell population. Consistent with this, we observed a larger number of FOXL2-expressing cells in XX β -cat fl.ex3 /+; Sf1-cre gonads when compared with XX controls (Fig. 4 ). In wild-type gonads, a negative-feedback loop may regulate levels of β-catenin to maintain the size of the ovary; this regulation would be lost when β-catenin is artificially stabilized, leading to the increased size. However, more work is required to test the role of β-catenin on patterning of ovarian somatic cells.
Previously, we proposed a model in which an antagonistic relationship between SOX9 and WNT4 exists within the bipotential gonad ( 22 ). This relationship is likely to be indirect as WNT4 is an extracellular ligand and SOX9 is a nuclear transcription factor. Our data suggest that β-catenin mediates this antagonistic relationship. Similarly, Chang et al . ( 23 ) recently performed a similar experiment in which β-catenin was stabilized specifically in Sertoli cells. Although β-catenin stabilization occurred later than in our mouse model (∼13.5 dpc), they also observed a decrease in the SOX9 expression by E14.5. Stabilized β-catenin could antagonize SOX9 in several ways. As β-catenin from the recombined β -cat fl.ex3 allele cannot be degraded, β-catenin can translocate to the nucleus and could interact directly with SOX9, compete for a binding partner or target site or bind to the Sox9 promoter to negatively regulate its expression. A negative interaction between SOX9 and β-catenin has previously been documented during chondrocyte differentiation, in which heterodimerization of SOX9 and β-catenin leads to the mutual degradation of both proteins ( 24 ).
Our work strongly suggests that antagonism between SOX9 and β-catenin is the molecular mechanism through which the fate of the supporting cell lineage in the gonad is established to drive male or female sex determination. This competition is pushed toward the ovarian pathway by RSPO1 and WNT4, which act to establish β-catenin as the dominant signal in XX gonads. This provides a potential mechanistic explanation for how male-to-female sex-reversal can occur in XY patients who lack mutations in genes required for the activation of the male program; stabilization of β-catenin counteracts the effects of SRY by destabilizing its only known downstream target, SOX9.
MATERIALS AND METHODS
Mice and matings
The Sf1-cre transgenic ( 12 ), the Wnt4 deletion strain ( 5 ) and the Catnb lox(ex3) ( β-cat fl.ex3 ) ( 13 ) mouse lines were maintained on the C57BL/6 background. To generate mice expressing the stabilized form of β-catenin specifically in the gonad, Sf1-cre males were mated to β-cat fl.ex3 females. Mice carrying a single copy of the Sf1-cre transgene, or a single copy of the β-cat fl.ex3 allele, were indistinguishable from wild-types and used interchangeably as controls. Mutant mice were designated as β-cat fl.ex3 /+; Sf1-cre .
Immunohistochemistry, histology and in situ hybridization
Following timed matings, gonads were dissected from embryos and fixed overnight at 4°C in 4% paraformaldehyde. Samples were immunostained as whole mounts using the following antibodies: SOX9 (1:200; a gift of F. Poulat, Institut deGenetique Humaine, Montpelier, France), FOXL2 (1:250; a gift of Reiner Veitia, Paris, France), PECAM1 (1:250; BD BioScience, San Jose, CA, USA), γH2AX (1:100; Calbiochem, San Diego, CA, USA), β-catenin (1:2000; Sigma, St Louis, MO, USA) and AMH (1:500; Santa Cruz, Santa Cruz, CA, USA). Double immunohistochemistry was detected by Cy3- and Cy5-conjugated secondary antibodies (1:500; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) and imaged using confocal scanning microscopy. For γH2AX, β-catenin and AMH staining, samples were embedded in OCT and sectioned at 12–14 µm before staining.
For histology, samples were fixed in Bouin's fixative overnight at 4°C, dehydrated through an ethanol series and embedded in paraffin. Samples were sectioned at 5–7 µm and then stained with hematoxylin and eosin (H&E).
For in situ hybridization, samples were fixed overnight at 4°C in 4% paraformaldehyde and processed as described ( 25 ). A digoxigenin-labeled RNA probe was detected by using an alkaline phosphatase-conjugated anti-digoxigenin antibody (1:1000; Roche, Indianapolis, IL, USA).
Total RNA was extracted from E13.0–13.5 dpc gonads (separated from mesonephroi) using Trizol (Invitrogen, Carlsbad, CA, USA). Gonads of the same genotype were pooled for each RNA preparation (between 4 and 10). The following was performed on three independent pools of RNA (biological replicates): RNA was DNase-treated (Sigma) and converted to cDNA using an iScript™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA), according to the manufacturer's protocol. Quantitative PCR was performed in triplicate using SensiMix Plus SYBR+Fluorescein (Quantace, Norwood, MA, USA) and run on the iCycler™ Thermal Cycler (Bio-Rad). PCR conditions were as follows: 95°C for 3 min and 30 s (one cycle); 95°C for 30 s, 55°C for 45 s, 72°C for 45 s (40 cycles) and 72°C for 5 min (one cycle). Primer sequences are listed 5′–3′: Fst F-AAACCTACCGCAACGAATGTG, R-GGTCACACAGTAGGCATTATTGGTC; β-actin F-GGC TGT ATTCCCC TCCATCG, R-CCAGTTGGTAACAATGCC ATGT; Wnt4 F-AGCCGGGCACTCATGAATCT, R-GC A CGCCAGCACGTCTTTAC and Rspo1 F-GTCT ATCTTGG GGGTGGTTC, R-AGGGGTGGTCTCTTGCTAA.
Organ culture using the droplet method
Organ cultures were performed using a modified version of a previously described protocol ( 26 ). Briefly, gonad/mesonephros complexes were cultured at 37°C with 5% CO 2 /95% air in a 30 µl droplet of Dulbecco's modified Eagle's medium (Gibco), supplemented with 10% fetal calf serum (Cambrex), and 50 µg/ml ampicillin with or without 50 m m LiCl. Gonads were collected from embryos at stages between 16 and 23 tail somites ( 27 ) and cultured for 24 h. Contralateral gonads were used as controls, and genetic sex was determined by PCR genotyping of embryonic tail samples ( 28 ). Two gonad/mesonephros complexes were dissected and immediately placed in round droplets of media on opposite sides of an inverted lid of a 5 cm Petri dish. Using a pipette tip, the droplets were spread out until the gonad/mesonephros complex lay on its side and the surface tension of the medium just held the gonads in place against the dish. The droplet was approximately the size of a dime once spread out. This dish was then floated on water in a larger, covered 10 cm Petri dish.
Treated gonads were scored relative to untreated XX and XY controls based on the presence or absence of the coelomic vessel (based on PECAM staining), levels of SOX9 protein (using an antibody to detect SOX9) and size. When a disrupted vasculature or intermediate levels of SOX9 postive cells were observed, those characteristics were considered ‘partial male’. Although size was a continuous variable, the LiCl treated gonads were considered a ‘male’ size when the width of the gonad (from the mesonephros to the coelomic surface) closely resembled control XY gonads; gonads were considered ‘female’ when they were observed to be smaller and closer in size to control XX gonads (see Fig. 3 A for comparison).
Funding was provided by the National Institutes of Health (HL63054 and HD39963 to B.C. and DK54480 and HD046743 to K.L.P.).
We would like to acknowledge Yuna Kim for her helpful suggestions and Reiner Veitia and Francis Poulat for contributing valuable antibody reagents. The project described was supported by grant number F32HD055791 (to D.M.M.) from the National Institute of Child Health and Human Development. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Child Health and Human Development or the National Institutes of Health.
Conflict of Interest statement . None declared.