Testicular Differentiation Occurs in Absence of R-spondin1 and Sox9 in Mouse Sex Reversals

In mammals, male sex determination is governed by SRY-dependent activation of Sox9, whereas female development involves R-spondin1 (RSPO1), an activator of the WNT/beta-catenin signaling pathway. Genetic analyses in mice have demonstrated Sry and Sox9 to be both required and sufficient to induce testicular development. These genes are therefore considered as master regulators of the male pathway. Indeed, female-to-male sex reversal in XX Rspo1 mutant mice correlates with Sox9 expression, suggesting that this transcription factor induces testicular differentiation in pathological conditions. Unexpectedly, here we show that testicular differentiation can occur in XX mutants lacking both Rspo1 and Sox9 (referred to as XX Rspo1KOSox9cKO ), indicating that Sry and Sox9 are dispensable to induce female-to-male sex reversal. Molecular analyses show expression of both Sox8 and Sox10, suggesting that activation of Sox genes other than Sox9 can induce male differentiation in Rspo1KOSox9cKO mice. Moreover, since testis development occurs in XY Rspo1KOSox9cKO mice, our data show that Rspo1 is the main effector for male-to-female sex reversal in XY Sox9cKO mice. Thus, Rspo1 is an essential activator of ovarian development not only in normal situations, but also in sex reversal situations. Taken together these data demonstrate that both male and female sex differentiation is induced by distinct, active, genetic pathways. The dogma that considers female differentiation as a default pathway therefore needs to be definitively revised.


Introduction
Mammalian sex determination depends on the primary developmental decision of the gonad to differentiate as testis or ovary. The gonad develops as a bipotential organ with the capacity to respond to two different genetic stimuli: the activation of the SRY/SOX9 pathway that induces testicular development, or the expression of the R-spondin1 (RSPO1)/beta-catenin pathway that regulates ovarian differentiation [1]. Indeed in humans and mice, male sex determination is initiated by the expression of the Ylinked gene SRY [2,3,4]. Sry expression in turn activates the transcriptional regulator SOX9 [5]. Subsequently, SOX9 initiates Sertoli cell differentiation, the supporting cell of the testicular sex cords [6,7]. Signaling pathways initiated in these cells contribute to the organization of the XY gonads [8], as well as to the differentiation of other testicular cell lineages such as the Leydig steroidogenic cells [9,10] and the pro-spermatogonia [11,12], ultimately leading to testis formation and, in turn, male development. In 46,XY patients, loss-of-function mutations in SRY and SOX9 promote male-to-female sex reversal [13,14], whereas translocations of the SRY locus to another chromosome can yield 46,XX patients with female-to-male sex reversal [3]. Loss-of-function mutations [6,7,15,16] and gain-of-function mutations [4,17,18] of Sry and Sox9 have been generated in mouse models, showing that Sry and Sox9 are necessary and sufficient to induce testis differentiation and the associated male development. As a consequence, these genes have been considered as the master inducers of testis differentiation and male development.
In the absence of SRY (XX individuals), up-regulation of RSPO1, an activator of the WNT/beta-catenin signaling pathway, promotes ovarian differentiation. Mutations in RSPO1 are responsible for skin disorders and female-to-male sex reversal in 46,XX patients [19]. Similarly, ablation of Rspo1 in mice yields female-to-male sex reversal and promotes Sox9 up-regulation correlated with differentiation of Sertoli cells and formation of testis cords at birth [20]. This gonadal dysgenesis yields development of an ovotestis, a gonad displaying both testicular and ovarian regions [20,21]. Rspo1 expression in turn activates expression of Wnt4 [21], another activator of the WNT/betacatenin signaling pathway involved in ovarian differentiation [22,23]. When the canonical beta-catenin signaling pathway is activated in XY gonads, this induces male-to-female sex reversal indicating that this pathway acts on top of ovarian differentiation [23]. Indeed, activation of WNT/beta-catenin is required for expression of Foxl2 [24], a transcription factor involved in folliculogenesis [25,26] and homeostasis of the ovary [27]. Thus Rspo1 appears to be the gene instructing the molecular network leading to ovarian development.
Since ablation of Rspo1 promotes SOX9 expression concomitantly with Sertoli cell differentiation [20], it was assumed that Sox9 is the sex reversal inducer in XX Rspo1 KO mutants. We now show that i) testicular differentiation occurs in XX Rspo1 KO Sox9 cKO mutants indicating that neither Sry nor Sox9 are required for female-to-male sex reversals; ii) testicular differentiation also occurs in XY Rspo1 KO Sox9 cKO mutants indicating that Rspo1 is required for male-to-female sex reversal in XY Sox9 cKO mutants.

Results/Discussion
Rspo1 is required for ovarian development in XY Sox9 cKO mice Sox9 is required for Sertoli cell differentiation, testis formation and male development. Indeed, deletion of Sox9 in XY Sox9 fl/fl ; Sf1:cre Tg/+ , (referred to as XY Sox9 cKO ) triggers male-to-female sex reversal [16]. However the factor(s) inducing sex reversal in XY Sox9 cKO remained to be identified. Given (i) the prominent role of RSPO1, an activator of beta-catenin signaling, in female sex determination [19], and (ii) the fact that ectopic activation of betacatenin in XY gonads can induce male-to-female sex reversal [23], we hypothesized that Rspo1 expression induced male-to-female sex-reversal in XY Sox9 cKO gonads. According to this scenario, neither testicular (which is Sox9-dependent) nor ovarian (which is Rspo1/beta-catenin-dependent) differentiation should occur in XY Sox9 cKO gonads additionally lacking Rspo1. To test this hypothesis, we have generated and analyzed double loss-of-function mice (i.e. XY Rspo1 2/2 ; Sox9 fl/fl ; Sf1:cre Tg/+ , referred to as XY Rspo1 KO -Sox9 cKO ).
Since previous results have shown that sex reversal can appear quite late during fetal development [20,22], we first analyzed adult stages when the sexual development is likely to be completed. At P60 (postnatal day 60), the anogenital distance in XY Rspo1 KO -Sox9 cKO mice was equivalent to that of XX controls but the internal genitalia contained both male and female organs including oviducts, uterine horns and vaginal tissues, as well as epididymides, vasa deferensia, seminal vesicles and prostate ( Figure S1). The XY Sox9 cKO developed as ovaries (Figure 1b, 1g, 1l, 1q and Figure S1), as expected from a previous report [16]. Interestingly, XY Rspo1 KO Sox9 cKO gonads developed as hypoplastic testes containing well-defined seminiferous tubules as evidenced by histological analysis (Figure 1c, 1h and Figure S1). We next examined whether the supporting cells forming the seminiferous tubules differentiated as granulosa cells, the ovarian supporting cells expressing FOXL2 [25,28] or as Sertoli cells expressing DMRT1 [29]. In P21 gonads, immunostaining experiments showed that the supporting cells forming the seminiferous tubules in XY Rspo1 KO Sox9 cKO gonads were DMRT1-expressing Sertoli cells (Figure 1r), even though SOX9 was clearly missing (Figure 1m). However, a few FOXL2positive granulosa cells were found within the alignment of the Sertoli cells forming the seminiferous tubules (Figure 1m, r) and in a few XY Rspo1 KO Sox9 cKO mice (3 out of 18), rare and abnormal follicles were observed ( Figure S2A). The mixed genetic background of Rspo1 KO Sox9 cKO mice is a likely factor causing the variation of this phenotype.
Altogether this shows that a genetic pathway activated by RSPO1 is required for the male-to-female sex-reversal of XY Sox9 cKO and indicates that Sertoli cell differentiation and seminiferous tubules formation can occur in the absence of SOX9.
Sertoli cell differentiation occurs without Sry and Sox9 in XX Rspo1 KO gonads Our study also allowed us to evaluate the effect of Sox9 removal in a female-to-male sex reversal context (i.e. in XX Rspo1 KO -Sox9 cKO ). Given that homozygous mutations of Rspo1 promote Sertoli cell differentiation around birth, a process that is associated with Sox9 up-regulation in these cells [20], we hypothesized that Sox9 is the inducing factor of testicular differentiation in XX Rspo1 KO mice. If Sox9 is indeed the main switch for female-to-male sex reversal in XX individuals, one expects an impaired differentiation of Sertoli cell and seminiferous tubules in the absence of both Rspo1 and Sox9 in XX Rspo1 KO Sox9 cKO gonads. Unexpectedly, at P60, these XX double mutants displayed hermaphroditism of the reproductive tracts ( Figure S1). Histological analysis revealed that XX Rspo1 KO -Sox9 cKO mice exhibited ovotestes with an extensive presence of sex cords (Figure 1d, 1i and in Figure S2B, f) as do XX Rspo1 KO gonads (shown in Figure S2B, S2e and in previous analyses [20,21]). Thus, the development of XX Rspo1 KO Sox9 cKO mouse genitalia is indistinguishable from that of XX Rspo1 KO mice indicating that the additional deletion of Sox9 in XX Rspo1 KO Sox9 cKO gonads does not change the fate of XX Rspo1 KO gonads. We next examined whether the supporting cells forming the sex cords differentiated as granulosa cells, the ovarian supporting cells expressing FOXL2 [25,28], or as Sertoli cells expressing DMRT1 [29]. In three weeks old mice (P21), Sox9depleted cells forming the seminiferous tubules generally lacked the follicular cell marker FOXL2 and instead expressed DMRT1 (Figure 1n, 1s). These data clearly indicate that Sertoli cell, seminiferous tubule and testis differentiation can occur in the absence of Sry and Sox9 in XX Rspo1 KO gonads.

Steroidogenic cells are present in Rspo1 KO Sox9 cKO embryonic gonads
Previous studies clearly show that the development of male genitalia depends on androgens secreted by the embryonic testis

Author Summary
Mammalian sex determination is controlled by the paternal transmission of the Y-linked gene, SRY. Using mouse models, it has been shown that the main, if not the only, role of Sry is to activate the transcription factor Sox9, and these two genes are necessary and sufficient to allow male development. Indeed, defects in Sry and/or Sox9 expression result in male-to-female sex reversal of XY individuals. In XX individuals, Rspo1 is important for ovarian development as evidenced by female-to-male sex reversal of XX Rspo1 mutants. Since testicular differentiation appears concomitantly with Sox9 expression, it was assumed that Sox9 is the inducer of testicular differentiation in XX Rspo1 mutants. Our genetic study shows that i) neither Sry nor Sox9 are required for femaleto-male sex reversals; ii) other masculinizing factors like Sox8 and Sox10 are activated in sex reversal conditions; iii) Rspo1 is the main effector of male-to-female sex reversal in the XY Sox9 mutants. Together these data suggest that male and female genetic pathways are both main effectors involved in sex determination and that the long-standing dogma of a default female pathway should definitively be revised. [30]. In XX Rspo1 KO gonads, steroidogenic cells appear before Sertoli cell differentiation [20,31] and this was also observed in Wnt4 KO gonads [22,32], Wnt4 being up-regulated upon Rspo1 expression in XX gonads [20,21]. In addition, lack of Wnt4 expression was shown to allow ectopic migration of steroidogenic cells from the neighboring adrenals into gonads [32,33] and subsequent androgen synthesis [34], which explains the development of male genitalia in these mutants. When investigating whether steroidogenic cells were present in XX and XY Rspo1 KO Sox9 cKO gonads, we found that P450Scc, a gene encoding for a precursor involved in androgen synthesis was expressed at 14.5 dpc in XY controls, XY and XX Rspo1 KO Sox9 cKO gonads and XX Rspo1 KO gonads, but not in XX controls ( Figure S3A). However, Cyp21, a marker for adrenal cells [35], was not strongly expressed in Rspo1 KO Sox9 cKO gonads at 13.5 dpc ( Figure S3A), suggesting that the steroidogenic cells in Rspo1 KO Sox9 cKO gonads, did not come from the arenals or, alternatively, have undergone reprogramming as Leydig cells. Whatever the situation, it is likely that male hormones synthesized in the developing mutant gonads can contribute to stimulate epididymides, vasa deferentia and seminal vesicles development.

Delayed testicular formation in Rspo1 KO Sox9 cKO mice
We next investigated the timing of testicular cord formation in XY and XX Rspo1 KO Sox9 cKO gonads. In wild-type embryos, the earliest morphological sign of testis development occurs at 12.0-12.5 dpc when testis cord are formed [36]. Accordingly at 13.5 dpc, the testis cords were highlighted in XY controls by the prominent expression of SOX9 and AMH, two markers of Sertoli cells ( Figure 2a). In contrast, Rspo1 KO Sox9 cKO gonads did not show a clear testicular organization as they lack AMH at 13.5 dpc (Figure 2b, 2c, 2f, 2g). As AMH synthesis and secretion by Sertoli cells promotes the elimination of the female reproductive tract during embryogenesis [37], the absence of AMH in Rspo1 KO -Sox9 cKO gonads provides an explanation for the maintenance of the Mullerian derivatives (oviducts, uterine horns and vaginal tissues) in these mutant mice. In addition, Sertoli cell differentiation is delayed in gonads lacking both Sox9 and Rspo1, as indicated by the maintenance of SRY expression, in the XY Rspo1 KO Sox9 cKO gonads at 13.5 dpc (Figure 2f), a stage at which SRY expression has already ceased for one day in the control situation [38,39,40]. Along these lines, the maintenance of SF1 expression in XX Rspo1 KO Sox9 cKO gonads at 13.5 dpc (Figure 2j, 2k), a factor whose Immunofluorescence of SOX9 (k-o) or DMRT1 (p-t) (a Sertoli cell marker, in red), FOXL2 (k-t) (a follicular cell marker, in green) and DAPI (a nuclear marker in blue) (scale bar, 50 mm). Deletion of Sox9 with Sf1:cre (Sox9 cKO ) eliminates SOX9 expression in Sertoli cells (l, m, n), and promotes male-to-female sex reversal in XY Sox9 cKO gonads as highlighted by robust FOXL2 expression (l, q). However, Sox9 deletion no longer allows ovarian cells differentiation when Rspo1 is deleted in the XY (m, r) and XX (n, s) Rspo1 KO Sox9 cKO mice. This is evidenced by the robust expression of DMRT1 in 3 week-old XY (s) and XX (r) mutant gonads and XY controls (p), and the low or absent expression of FOXL2 in these gonads (k, m, n, p, r, s). XY (a, f, k, p) and XX (e, j, o, t) Rspo1 +/2 ; Sox9 flox/flox controls, XY Sox9 cKO gonads (b, g, l, q), XY (c, h, m, r) and XX (d, i, n, s) Sox9 cKO Rspo1 KO respectively. XX Rspo1 KO and XX Sox9 cKO Rspo1 KO gonads appeared similar (see Figure S2B). doi:10.1371/journal.pgen.1003170.g001 expression is normally down-regulated between 13.5 and 16.5 dpc in the ovary [41] (Figure 2l), also suggests that the XX Rspo1 KO Sox9 cKO gonads are still undifferentiated or have differentiated as testis. However, with respect to the latter, the absence of AMH expression shows that no Sertoli cell differentiation has occurred (Figure 2c, 2g). Altogether these data indicate that the Rspo1 KO Sox9 cKO gonads are still undifferentiated at 13.5 dpc.
The first signs of Sertoli cell differentiation appeared at 16.5 dpc in Rspo1 KO Sox9 cKO gonads, with some rare DMRT1-positive cells in comparison to XY controls ( Figure S3B). Then, few rudimentary testis cords were observed around 17.5 dpc ( Figure S3B). At P0, Sertoli cells aligned to form sex cords as evidenced by the localization of DMRT1 positive-cells ( Figure S4A c, d). Quantitative PCR experiments further confirmed that Dmrt1 expression was strongly expressed in XY Rspo1 KO Sox9 cKO gonads and weakly in XX Rspo1 KO Sox9 cKO gonads at P0, highlighting that more Sertoli cells were present in XY Rspo1 KO Sox9 cKO gonads in comparison to XX Rspo1 KO Sox9 cKO gonads ( Figure S4C). In addition, some FOXL2-positive cells were also detected in Rspo1 KO Sox9 cKO gonads ( Figure S4A c, d). However, quantitative PCR experiments showed that Foxl2 expression was significantly reduced in comparison to XX control or XY Sox9 cKO gonads ( Figure S4B) as expected for a gonad developing as ovotestis or testis.
We then studied SDMG1 expression, a cytoplasmic marker of

SOX9-negative Sertoli cells can support germ cell differentiation until initiation of meiosis
We next investigated whether the Sertoli cells that differentiate in the Rspo1 KO Sox9 cKO gonads can support germ cell differentiation. Since XX germ cells cannot survive in a testicular environment [42,43], the analysis was only carried out in XY Rspo1 KO Sox9 cKO gonads. In the normal fetal testis, following Sertoli cell differentiation, prospermatogonia become quiescent from 14.5 dpc and express the multipotency marker Oct4 until 17.5 dpc [44]. At that time, Cyp26b1, a protein involved in retinoic acid degradation, contributes to prevent germ cells from entering meiosis [45,46]. As expected, the majority of prospermatogonia in XY Rspo1 KO Sox9 cKO gonads expressed Oct4 at 14.5 dpc ( Figure S6o). Nonetheless, some cells had already committed to meiosis ( Figure S6k) and expressed the meiotic marker Stra8 [47], possibly because of the low level of Cyp26b1 expression in XY Rspo1 KO Sox9 cKO gonads ( Figure S6g). The reduced level of Cyp26b1 expression is however not sufficient to allow all germ cells to enter meiosis in XY Rspo1 KO Sox9 cKO gonads.
At P10, GATA1, Androgen Receptors (AR) and Clusterin (Clu) were normally expressed in Sertoli cells of XY Rspo1 KO Sox9 cKO gonads (Figure 4c, 4i, 4l), suggesting that these cells have acquired their identity and may be capable to support spermatogenesis. Accordingly, XY germ cells had committed to meiosis at P10, as assessed by immunodetection of the pre-meiotic and meiotic markers STRA8 and cH2AX, respectively (Figure 4c, 4f, 4i). However, later stages of spermatogenesis cannot however be analyzed, as hypoplasia of germ cells occurred within the seminiferous tubules of adult XY Rspo1 KO Sox9 cKO gonads (Figure 1h and Figure S1m), most likely because of cryptorchidism.
Sox8 and Sox10 are expressed in the absence of Sox9 Interestingly, we found that AMH was expressed in Sertoli cells of both XX and XY Rspo1 KO Sox9 cKO gonads at P12 ( Figure 5A). Given that (i) Amh is a target-gene of SOX9 [48,49], (ii) Amh expression can be induced by SOX8 [50] and SOX10 [51], and (iii) Sox10 ectopic up-regulation in XX gonads can induce testis differentiation [51], we hypothesized that a Sox factor distinct from Sox9 could have induced late AMH expression in Rspo1 KO Sox9 cKO gonads and delayed testicular differentiation. In agreement with this possibility, expression of both Sox8 and Sox10 was activated in Rspo1 KO Sox9 cKO mutants at P12 and P0 respectively ( Figure 5B, 5C). Previous data have shown that Sox8 becomes crucial from 14.5 dpc onwards, for the maintenance of testis development [52], suggesting that Sertoli cell differentiation can be induced by Sox genes other than Sox9 during late embryogenesis. However, the function of these Sox genes during late development is likely not sufficient to replace the role of Sox9 in early Sertoli cells development, thus leading to the formation of an hypoplastic testis, as is the case in the XY Rspo1 KO Sox9 cKO mice. To date, the only factors that have been shown to be able to induce Sertoli cell differentiation are Sox genes [51,53], while other factors such as Dmrt1 are required after birth (P7) for the maintenance of Sertoli cell identity [54]. Further studies are required to address whether DMRT1 is able to allow Sertoli cell differentiation from undifferentiated supporting cells. Given that Sox9 expression is controlled by Wt1 when Sry expression has ceased [55], we can speculate that Wt1 might also be involved in Sox8 and Sox10 expression in these mutants. Furthermore, FGF9 or PGD2, two secreted factors synthesized in the undifferentiated gonads, [56,57] can also contribute to Sertoli cell differentiation [58,59,60]. Whether Wt1, PGD2 or FGF9 signaling also regulate Sox8 and Sox10 remains to be investigated. In addition, when XX and XY Rspo1 KO Sox9 cKO gonads are compared at the same stage, XY gonads always appear more masculinized than XX gonads (Figure 1, Figure 3, Figure S1, Figure S3, Figure S5), because they contain more sex cords/seminiferous tubules. At a molecular level, the main difference between XX and XY Rspo1 KO Sox9 cKO gonads is the expression of SRY in XY gonads ( Figure 2). Indeed, SRY expression is . Sertoli cells support germ cell differentiation in XY Rspo1 KO Sox9 cKO gonads. Immunofluorescence (a-i) of GATA1 (Sertoli cell marker, in green), AR (Androgen Receptor) (Sertoli, peritubular myoid and Leydig cell marker, in red), STRA8 (a premeiotic marker, in red), and cH2AX (a meiotic marker, in green) at P10. Counterstain is DAPI (in blue). In situ hybridization (j-l) using a probe for Clu transcripts, another marker for mature Sertoli cells, illustrated as computer-generated bright field superimpositions of the blue counterstain (DAPI) with the hybridization signal (red false color). GATA1, AR and Clu expression show that the Sertoli cells mature in XY controls (a, g, j) and XY Rspo1 KO Sox9 cKO testes (c, i, l), and are able to support germ cell differentiation until meiosis initiation as revealed by STRA8 (a, c, d, f) and cH2AX (d, f, g, i) expression. Note that both Sertoli, peritubular myoid and Leydig cells of XY Sox9 cKO mutant gonads normally expressed AR (h). (scale bars: 50 mm). XY (a, d, g, j) Rspo1 +/2 ; Sox9 flox/flox controls, XY Sox9 cKO gonads (b, e, h, k) and XY Rspo1 KO Sox9 cKO (c, f, i, l) gonads. doi:10.1371/journal.pgen.1003170.g004  in XY Rspo1 KO Sox9 cKO gonads at 13.5 dpc, while at this time point its expression has ceased in XY control gonads. This suggests that SRY participates in the masculinization of the XY Rspo1 KO Sox9 cKO gonads by inducing the expression of genes other than Sox9 to promote sex cord formation.
In summary, here we have shown that (i) both SRY and SOX9 are dispensable for female-to-male sex reversal in Rspo1 KO , (ii) RSPO1 signaling is required for male-to-female sex reversal in Sox9 cKO , (iii) Sertoli cell differentiation and seminiferous tubule formation can occur in the absence of SOX9, possibly because of a functional redundancy with other SOX proteins such as SOX8 and SOX10. Indeed, ectopic expression of Sox10 in XX gonads has been shown to promote testicular differentiation [51]. Altogether these data show that SRY and SOX9 are not the only masculinizing factors, since other SOX proteins can induce female-to-male sex reversal in pathophysiological conditions ( Figure 6). Following Sertoli cell differentiation, DMRT1 expression becomes required for the maintenance of the Sertoli cells and the testicular tissue [54]. Furthermore, our data suggest that the feminizing factors remaining in Rspo1 KO mice can be overtaken by SOX proteins, when they are activated in XX gonads. Testicular differentiation in the absence of Rspo1 expression in XY Sox9 cKO gonads was unexpected since female development is thought to be a default pathway [61,62]. Our results imply that instead the female pathway needs to be activated. Therefore our genetic study suggests that mammalian sex determination is regulated by a finely tuned balance between two main factors [56,63], which are the SOX genes on the one hand and the RSPO1/WNT/ beta-catenin signaling pathway on the other hand.

Histological analysis
Urogenital organs were dissected, fixed in Bouin's solution overnight, and then embedded in paraffin. Microtome sections of 5 mm thickness were stained with periodic acid Schiff (PAS) or hematoxylin and eosin (H&E) according to standard procedures. Pictures were taken with an Axiocam mrm camera (Zeiss) and processed with Adobe Photoshop.

Immunological analyses
Gonad samples were fixed with 4% (w/v) paraformaldehyde overnight and then processed for paraffin embedding. Gonad samples for cryosections were successively fixed for 2 hours in 4% (w/v) paraformaldehyde, washed in cold phosphate-buffered saline (PBS), equilibrated in 10% (w/v) sucrose solution during 3 hours, then in 30% (w/v) sucrose solution overnight at 4uC, embedded in Cryomount (Histolab) and stored at 280uC.

In situ hybridization
Embryos were fixed with 4% paraformaldehyde (PFA) in 16PBS at 4uC overnight. Further processing of embryos and in situ hybridization were carried out essentially as described [67]. Sox9 riboprobe was synthesized according to [68] and Sox8 to [69], P450scc, Stra8 and Oct4 riboprobes synthesis were carried out as described previously [20]. In situ hybridisation (ISH) with digoxigenin-labelled probes was performed as described [70], using 10 mm-thick cryosections. Each experiment was repeated on at least two gonads. Post-hybridization washes were done in 100 mM maleic acid pH7.5, 150 mM NaCl, 0.1% (v/v) tween-20 (MABT). To increase the sensitivity, 5% (v/v) polyvinyl alcohol (Sigma) was added to the staining solution [71]. Nuclei were counterstained with DAPI diluted in the mounting medium at 10 mg/ml (Vectashield, Vector laboratories). ISH signals corresponding to Clu-positive cells were converted into a red false color on the merged pictures. The plasmids containing Lgals1 (366 bp-long; exons 2-4; MGI:96777) or Clu (942 bplong; exons 5-9; MGI:88423) cDNA fragments were linearized and used as a templates for the synthesis of the sense or antisense riboprobes.

Quantitative PCR analysis
Individual gonads were dissected in PBS from P0 animals (day of birth) and immediately frozen at 280uC. RNA was extracted using the RNeasy Qiagen kit, and reverse transcribed using the RNA RT-PCR kit (Stratagene). Primers and probes were designed at Roche Assay design center (https://www.   ggcgtcgtgaactcctaca-39 and 59-tgcagatgatgtgcgtgag-39 (probe 51) for Foxl2. All real-time, quantitative, PCR assays (QPCR) were carried out using the LC-Faststart DNA Master kit Roche, according to the manufacturer's instructions. QPCR was performed on cDNA from one gonad and compared to a standard curve. QPCR were repeated at least twice. Relative expression levels of each sample were quantified in the same run, and normalized by measuring the amount of Hprt1 cDNA (which represents the total amount of gonadal cells).

Statistical analysis
For each genotype (n = 6 individuals), the fold-change was the mean normalized expression levels divided by the mean normalized expression levels of the XY samples considered as the reference. Graphs illustrate fold-changes +1 s.e.m. The results were analyzed using Graphpad for statistical significance that was assessed using one-way ANOVA followed by Tukey-Kramer post-test for selected pairs of genotypes. Asterisks indicate : * p,0.05, ** p,0.01 and *** p,0.001.   (d, g). Empty and filled arrowheads indicate testis cords and follicles, respectively. (TIF) Figure S3 A-Expression of the steroidogenic marker P450Scc and Cyp21 in XY Rspo1 KO Sox9 cKO gonads. Whole-mount in situ hydridization of gonads at 14.5 dpc and 13.5 dpc. P450Scc is expressed in XY Rspo1 KO Sox9 cKO gonads (b), and in XY controls (a) but not in XX controls (e). P450Scc, was expressed in cells at the anterior part of the XX Rspo1 KO Sox9 cKO and XX Rspo1 KO gonads (c and d respectively) at 14.5 dpc. Cyp21 was strongly expressed in the adrenals (f, g, h, i, j), whereas no signal was detected in the gonads at 13.5 dpc (f, g, h, i, j). Ad: adrenal, G: gonad, K: kidney. B-Delayed testicular cords formation in XY and XX Rspo1 KO Sox9 cKO gonads. Haematoxylin and eosin stained histological sections of XY and XX Rspo1 KO Sox9 cKO gonads at 17.5 dpc show that some sex cords are forming in the XY Rspo1 KO Sox9 cKO (b) gonads in contrast to the XY controls (a) containing already formed sex cords. In the littermates XX Rspo1 KO Sox9 cKO and Rspo1 KO gonads (c and d respectively), no sex cords were observed at this stage (scale bar, 10 mm). Insets in a and b show AMH (Red) and DMRT1 (green), two markers of Sertoli cells highlighting sex cords at 16.5 dpc. Sex cords were rare in XY Rspo1 KO Sox9 cKO gonads and absent from XX Rspo1 KO Sox9 cKO gonads at this stage. (TIF) Figure S4 Detection of DMRT1 and FOXL2 in XY and XX Rspo1 KO Sox9 cKO gonads at P0. A-Immunofluorescence analysis of DMRT1 and FOXL2 in XY and XX Rspo1 KO Sox9 cKO gonads at P0. DMRT1 (red), a marker of postnatal Sertoli cells, was detected in the XY control (a), XY Rspo1 KO Sox9 cKO and XX Rspo1 KO Sox9 cKO gonads (c and d respectively) while in XY Sox9 cKO (b) and XX (e) control gonads DMRT1 was not detected. The ovarian protein FOXL2 (green) was detected in XY Sox9 cKO and XX control gonads, and few FOXL2-positive cells (granulosa cells) were found within the XY Rspo1 KO Sox9 cKO (c) and XX Rspo1 KO Sox9 cKO (d) gonads. B-qPCR anaylsis of Foxl2 in XY and XX Rspo1 KO Sox9 cKO gonads at P0. Foxl2 is significantly up-regulated in the XY Sox9 cKO compared to XY and XX Rspo1 KO Sox9 cKO indicating that ablation of Rspo1 leads to reduced Foxl2 expression at a transcriptional level. C-qPCR anaylsis of Dmrt1 in XY and XX Rspo1 KO Sox9 cKO gonads at P0. Dmrt1 is significantly up-regulated in the XY Rspo1 KO Sox9 cKO gonads, indicating that ablation of both Rspo1 and Sox9 promotes Sertoli cell differentiation. Dmrt1 is not up-regulated in XX Rspo1 KO Sox9 cKO gonads. Indeed, at P0 few Sertoli cells have differentiated in XX Rspo1 KO Sox9 cKO gonads in comparison to XY Rspo1 KO Sox9 cKO gonads (see Figure 3). In addition it is noteworthy that Dmrt1 transcription occurs in the XY Sox9cKO but the protein is not detected (see same figure panel A). This suggests that the Rspo1 signaling pathway antagonizes the male pathway, at least partly at the post-transcriptional level, as shown for two other genes Wnt4 and Fgf9 [56,72]. (TIF) Figure S5 Histological analysis of XY and XX Rspo1 KO Sox9 cKO gonads at P0, P12 and P21. Differentiation of seminiferous tubules containing germ cells in XY Rspo1 KO Sox9 cKO hypoplasic testis (c, h, m) and XY controls (a, f, k). In XX Rspo1 KO Sox9 cKO (d, i, n), the gonads develop as ovotestes containing both seminiferous tubules and follicles (white star indicates a section of a follicle and white arrowhead shows a section (n) of a seminiferous tubule). However, as previously shown [73], the XX germ cells did not survive in post-natal seminiferous cords. In contrast, germ cells survived until P12 and some until P21 in the XY Rspo1 KO Sox9 cKO seminiferous tubules (h and m respectively) (scale bar: 50 mm). Controls XY (a, f, k) and XX (e, j, o), XY Sox9 cKO (b, g, I) and XY Rspo1 KO Sox9 cKO (c, h, m), XX Rspo1 KO Sox9 cKO (d, i, n). (TIF) Figure S6 Mixed germ cell differentiation in XY Rspo1 KO Sox9 cKO gonads at 14.5 dpc. In situ hybridization using a riboprobe for Sox9 (a-d), Cyp26b1 (e-h), Stra8 (i-l) and Oct4 transcripts (m-p) shows lack of Sox9 expression in the XY Sox9 cKO (b) XY Rspo1 KO Sox9 cKO (c) and the XX control (d) gonads. Germ cells in XY Sox9 cKO (j) and XX (i) control gonads have entered meiosis in as evidenced by robust Stra8 expression and weak expression of primordial germ cell marker Oct4 (n and p respectively). XY Rspo1 KO Sox9 cKO mutants (k) showed Stra8 expression at the periphery of the E14.5 gonad indicating these few cells have undergone meiosis (k) while the remaining germ cells were quiescent, thus, express Oct4 (o). Few Cyp26b1 expressing cells were detected in XY Rspo1 KO Sox9 cKO gonads (g). G: gonad, K: Kidney. XY (a, e, i, m) and XX (d, h, l, p) Rspo1 +/2 ; Sox9 flox/flox controls, XY Sox9 cKO (b, f, j, n) and XY Rspo1 KO Sox9 cKO (c, g, k, o). (TIF)