Partial male-to-female reprogramming of mouse fetal testis by Sertoli cell ablation

In mouse early gonadogenesis, the coelomic epithelia covering the mesonephric region are crucial for gonadal formation as a source of gonadal somatic cell precursors that are recruited in a stage-dependent manner (DeFalco et al., 2011; Harikae et al., 2013a; Karl and Capel, 1998). By embryonic day (E) 11.5, the gonadal surface epithelial cells proliferate and form the primary sex cords, including supporting cell progenitors and germ cells in both male and female gonads. In the male gonads, the transient expression of sex-determining region Y (Sry) from E10.5 to E11.5 triggers the differentiation of these bipotential supporting cells into SOX9-positive Sertoli cells (see review by Kanai et al., 2005; Kashimada and Koopman, 2010). These differentiating SOX9-positive Sertoli cells produce several key autocrine/paracrine signals for the maintenance and organization of Sertoli cells into well-defined testis cords (Hiramatsu et al., 2010; Kim et al., 2006), for the mesonephric cell migration of vascular and perivascular progenitor cells (Coveney et al., 2008; Kumar and DeFalco, 2018; Martineau et al., 1997), and for Leydig cell differentiation in interstitial and peri-mesonephric stromal regions in the fetal testes at E12.5 (Brennan et al., 2003; Yao et al., 2002). Indeed, at E11.5, pre-Sertoli cells produce fibroblast growth factor 9 (FGF9), together with prostaglandin D2, to maintain their high SOX9 expression (Hiramatsu et al., 2009; Kim et al., 2006; Moniot et al., 2009; Wilhelm et al., 2007), which also induces male-specific epithelial proliferation and peri-mesonephric/vascular cell migration underneath the surface epithelia and inter-cordal spaces (Colvin et al., 2001; Combes et al., 2009; Schmahl et al., 2004). At E12.0-12.5, the differentiated Sertoli cells express platelet-derived growth factor (PDGF; Brennan et al., 2003) and desert hedgehog (DHH; Yao et al., 2002) to trigger differentiation of 3βHSD-positive Leydig cells from interstitial stroma cells (see review by Svingen and Koopman, 2013). Anti-Müllerian hormone (AMH) is involved in mesonephric/vascular cell migration (Ross et al., 2003), Müllerian duct regression (Behringer et al., 1990, 1994) and the maintenance of Sertoli cells in the testis cords in cooperation with activin (Rodriguez et al., 2022). Testis-specific mesonephric/vascular cell migration may disrupt the direct connection between surface epithelia and primary sex cords (see review by Harikae et al., 2013a; Suzuki et al., 2015), leading to tunica albuginea and interstitial regions being outside the testis cords at E12.5-13.5. Therefore, the paracrine actions of Sertoli cells are crucial for establishing the principal testis architecture at the fetal stage.

In ovarian primary sex cords, supporting cells do not express Sry/Sox9 (Albrecht and Eicher, 2001), but do express Foxl2, an ovary-determining factor in goat, at E12.0-12.5 in mice (Boulanger et al., 2014; Schmidt et al., 2004). These FOXL2-positive cells are mitotically arrested pre-granulosa cells (Gustin et al., 2016; Mork et al., 2012) that contribute to the first wave of folliculogenesis soon after birth (see review by Imaimatsu et al., 2022; Suzuki et al., 2015). In fetal ovaries at E12.5 and thereafter, the surface epithelia continuously undergo proliferation, ingression and expansion into the subepithelial region (Mork et al., 2012; Rastetter et al., 2014), leading to the formation of FOXL2-positive ovarian cords (also known as ‘ovigerous’ or ‘secondary sex’ cords) throughout the late fetal stages (Ng et al., 2014; Niu and Spradling, 2020; Rastetter et al., 2014; Suzuki et al., 2015). The ovarian cords, including some surface epithelial cells, are marked by female-specific expression of the leucine-rich repeat-containing G-protein-coupled receptor 5 (Lgr5) and guanine nucleotide-binding protein (G protein), gamma 13 (Gng13), and contribute to the resting primordial follicles to sustain fertility throughout life (Fujino et al., 2007; Niu and Spradling, 2020; Rastetter et al., 2014; Zheng et al., 2014). Taken together, the above suggest distinct sex-dimorphic cellular events in the surface epithelia between the fetal testis and ovary at E12.5 and later stages. However, the mechanisms are unclear.

Toxin receptor-mediated cell knockout (TRECK) mice that carry a human diphtheria toxin (DT) receptor (HBEGF-mut) transgene driven by a cell linage-specific promoter can achieve conditional loss of lineage-specific cells by the administration of DT (Saito et al., 2001). AMH-TRECK transgenic (Tg) mice carrying the AMH promoter-driven DT receptor can be used for conditional ablation of AMH-positive gonadal supporting cells by DT treatment in vivo and in vitro (Shinomura et al., 2014; Rebourcet et al., 2014). Previous studies revealed the in vivo roles of the fetal and postnatal Sertoli cells at and after E14.5 by using the Sertoli cell ablation in combination with AMH-Cre and Cre recombinase-inducible DT fragment A (DTA) (Rebourcet et al., 2014; Wang et al., 2020). In this study, we focused on the paracrine effects of differentiated Sertoli cells on other testicular somatic cells at E12.5-14.5. Ex vivo direct treatment of fetal AMH-TRECK Tg testes with DT led to Sertoli cell depletion and ectopic FOXL2 expression in testicular somatic cells near surface epithelia and mesonephros, suggesting an unexpected male-to-female sex reversal model of E12.5 testes. Such ectopic FOXL2 induction in fetal testes may be limited to initiation at E12.5-13.5, and its potency was reduced by exogenous FGF9, implicating FGF9 loss in sex reversal in the surface epithelia/subepithelia and peri-mesonephric stroma in early fetal testes with Sertoli cell ablation.

In developing mouse testes, differentiated Sertoli cells in the testis cords start to express Amh at E12.5 (Münsterberg and Lovell-Badge, 1991). In this study, to examine the paracrine function of fetal Sertoli cells at E12.5 and thereafter, we isolated the testes/ovaries of wild-type and Tg embryos and treated them with DT (200 ng/ml) in 24 h organ culture for Sertoli cell ablation (Fig. 1A-E). After 24 h recovery culture in FCS-DMEM, the phenotypes of gonadal explants were examined by morphometric and immunohistochemical analyses (Fig. 1B,E). Histological analysis revealed that, in wild-type testes, DT treatment induced no appreciable defect in the testis cords (Fig. 1B). By contrast, in the Tg testes, DT treatment caused severe deformation of the testis cords, in which nuclear debris and necrotic death of Sertoli and germ cells were found throughout the presumptive testis cord region. This is in contrast to the normal cell morphology in the surface epithelial and interstitial regions around the deformed testis cords (Fig. 1B). Moreover, morphometric analysis confirmed a significant reduction in the length of the anterior-posterior or dorsal-ventral axis of DT-treated Tg testes compared with DT-treated wild-type and non-treated Tg (control) testes (Fig. 1E).

Anti-SOX9 and AMH immunostaining showed that SOX9- and AMH-positive Sertoli cells formed the well-defined testis cords in the parenchyma of DT-treated wild-type and non-treated Tg control testes (Fig. 1C). In Tg testis explants, DT treatment caused complete loss of SOX9-positive signals in the testicular parenchyma, despite intact SOX9-positive signals in the mesonephric tubules (Fig. 1C). Moreover, most of the anti-AMH-positive signals were reduced in DT-treated Tg explants. In contrast to Sertoli cell depletion in Tg explants, the 3βHSD-positive Leydig cells and NR2F2-positive stromal cells were intact in the presumptive tunica albuginea, and in the interstitial and peri-mesonephric regions of DT-treated Tg testes (Fig. 1D). Therefore, the DT-treated Tg testis explants showed almost complete ablation of Sertoli cells without appreciable damage to the other gonadal soma; e.g. tunica albuginea and interstitium, at E12.5. This ex vivo system enables evaluation of the paracrine actions of fetal Sertoli cells at and after E12.5 ex vivo.

In addition, TUNEL staining revealed that some apoptotic cells were still detectable 1 day culture after DT treatment (Fig. S1A,B), but TUNEL-positive signals, as well as AMH-positive and SOX9-positive signals, were rarely observed at 4-day culture of DT-treated Tg explants (Fig. S1C,D), indicating the complete depletion and no reappearance of Sertoli cells in Tg testes after DT treatment.

Next, we monitored the transcription profile in 4-day culture of fetal testes with Sertoli cell ablation. DT-treated Tg testes at E12.5 were cultured for 0.5, 1, 2, 3 or 4 days, and three sets per culture period (each set had at least four testis explants) were subjected to bulk RNA-seq analysis (Fig. 2A).

Hierarchical clustering of 15 RNA-seq datasets revealed that the transcription profiles could be divided into 0.5/1-day and 2/3/4-day culture periods in DT-treated Tg testes (Fig. 2B), suggesting dynamic transcriptional alterations in Tg testes at 1 and 2 days after 24 h DT treatment. Moreover, based on the transcriptome similarity, all sample sets were classified into the following four stages: pre-stage (three sets of 0.5-day culture samples), stage I (three sets of 1-day samples), stage II (three sets of 2-day samples, one set each of 3-day and 4-day samples) and stage III (two sets of 3-day and 4-day samples). k-Mean gene clustering analysis showed the presence of two major clusters: 750 stage-dependently transcript-reduced genes (cluster I) and 546 stage-dependently transcript-increased genes (cluster II) (Fig. 2C, Table S1A,B). In cluster I, reduced expression of Sertoli cell-specific genes, such as Amh, Sox9, Fgf9, Dmrt1, Pdgfa and Cyp26b1 (Bowles et al., 2006; Brennan et al., 2003; Colvin et al., 2001; da Silva et al., 1996; Kent et al., 1996; Münsterberg and Lovell-Badge, 1991; Raymond et al., 1999), was detectable with significant transcript reduction (Fig. 2D), suggesting Sertoli cell ablation in Tg explants. By contrast, several key genes crucial for the paracrine actions of Sertoli cells, such as Dhh (encoding hedgehog signals to induce Leydig cell differentiation; Bitgood et al., 1996; Yao et al., 2002) and Ptgds (encoding the PGD₂ synthase; Adams and McLaren, 2002), as well as Hsd3b1 (a Leydig cell marker; Baker et al., 1999), showed no appreciable alteration in expression in DT-treated testis explants (Fig. 2D); this is likely due to the rapid reduction of their transcripts in the damaged Sertoli cells, with their levels reaching a minimum by 1 day after 24 h DT treatment (Fig. S2).

Surprisingly, in cluster II, pre-granulosa cell-specific genes, such as Foxl2, Hmgcs2 and Fst (Bagheri-Fam et al., 2020; Loffler et al., 2003; Menke and Page, 2002; Yao et al., 2004), and a key ovarian cord marker, Gng13 (Fujino et al., 2007; Niu and Spradling, 2020), were significantly increased, although there was no significant difference in the expression of key ovarian genes (Lgr5 and Wnt4) in the RNA-seq data [Fig. 2E; Table S1B; see also quantitative RT-PCR (RT-qPCR) data in Fig. S3D]. Moreover, expression of Wt1, a marker of both gonadal supporting cells and surface epithelia in both sexes (Armstrong et al., 1993; Liu et al., 2016), was reduced slightly, albeit non-significantly (Fig. 2F). Expression of the gonadal somatic cell markers Nr5a1 and Gata4 was maintained (Fig. 2F), suggesting retention of Wt1-positive gonadal somatic cells in 4-day culture of Tg testes with Sertoli cell depletion. Taken together, these findings suggest male-to-female reprogramming of remaining testicular somatic cells in Tg testes with Sertoli cell depletion.

Differentially expressed gene (DEG) analysis of explants of stage II and III (relative to stage I) showed increased expression of 111 and 464 genes and decreased expression of 289 and 709 decreased genes, respectively [|log₂(fold change)|>1 and adjusted P-value <0.05; Fig. S3A,B]. We next compared these DEGs with the testis- or ovary-specific genes at E12.5-13.5 (i.e. 2984 male and 2950 female genes expressed in a sex-dimorphic manner in vivo; Zhao et al., 2018). As a result, 154 of the 709 decreased DEGs in Tg testes at stage III were found among testis-specific genes in vivo (21.7%; 154/709 genes; Table S2A). Compared with increased DEGs at stage II/III, 35.1% (39/111 genes) at stage II, and 22.0% (102/464 genes) at stage III were found among the 2950 ovary-specific genes, including Foxl2, Hmgcs2, Fst, Irx3 and Gng13 (Fig. S3A,B; Table S2B,C).

RT-qPCR analyses of DT-treated Tg testis explants before and after a 4-day culture confirmed reduced expression of Amh and Sox9 (Fig. S3C). Moreover, RT-qPCR analyses revealed a significant increase of Wnt4 and Lgr5 expression, early key ovarian-specific genes in vivo (Fig. S3D), in contrast to high variations of these two transcripts among the samples in bulk RNA seq data (see Fig. 2E). Taken together, these findings imply that DT-treated Tg testes undergo partial feminization ex vivo, which may be initiated by loss of Sertoli cells from the testes at E12.5. As a side note, the RT-qPCR data showed superior performance compared with bulk RNA-seq data for Wnt4 and Lgr5. This was attributed to the advantage of RT-qPCR, which used a small RNA amount of each Tg or control explant, while bulk RNA-seq necessitated a substantial amount of RNA sourced from multiple gonads obtained from different pregnant mothers at various experimental dates.

To evaluate the partial feminization of Tg testes after Sertoli cell ablation, we examined the spatial pattern of FOXl2 expression in DT-treated Tg testes before and after a 4-day culture (Fig. 3A). At 1 day after DT treatment, no FOXL2-positive signal was observed in Tg or wild-type testes (Fig. 3A; corresponding to stage I, as shown in the RNA-seq data; Fig. 2B). In a 4-day culture of Tg testes, corresponding to stage III, FOXL2-positive signals were found in somatic cells in the testicular parenchymal cells near the surface epithelia and mesonephros (Fig. 3A). Indeed, their FOXL2 signal intensity in Tg gonads was similar to that of the ovarian controls (insets in Fig. 3A). These findings suggest that Sertoli cell ablation of E12.5 testes resulted in the ectopic appearance of two distinct FOXL2-positive cell populations located near surface testis epithelia and in testicular stroma near the mesonephros.

In the surface subepithelial region of XY Tg explants in a 4-day culture, anti-FOXL2 immunostaining, together with anti-NR2F2 staining (a stromal cell marker that marks fetal gonadal somatic cells except Sertoli/granulosa and Leydig cells; McClelland et al., 2015; Rastetter et al., 2014), revealed FOXL2-positive subepithelial cells in the NR2F2-negative cell cluster that is directly associated with the surface epithelial layer (Fig. 3B). Such an NR2F2-negative/FOXL2-positive cord-like structure was also seen in the ovarian cords, including germ cells of female control explants, but not in the subepithelial region (which was replaced by the tunica albuginea), in the wild-type testis explants of a 4-day culture. Moreover, anti-FOXL2 immunostaining together with anti-WT1 antibody (a later marker for surface epithelium and supporting cells; Liu et al., 2016) revealed that some FOXL2-positive cells were located within the WT1-positive subepithelial cell clusters in the 4-day culture of DT-treated Tg testes (Fig. S4). These findings suggest the ectopic appearance of FOXL2-positive cells in the WT1-positive/NR2F2-negative ovarian cord-like structure in DT-treated Tg testes.

In situ hybridization revealed positive signals for Lgr5 and Gng13, which are markers of ovarian cords (Niu and Spradling, 2020; Rastetter et al., 2014), in part of the epithelial and subepithelial regions of DT-treated Tg testes after a 4-day culture (Fig. 3C). These findings suggest that the FOXL2-positive ovarian cord-like structure may be ectopically induced in the surface epithelial region of testis explants after the removal of Sertoli cells.

In the interstitial regions at the mesonephric side of Sertoli cell-ablated testes, FOXL2-positive signals were mainly found in NR2F2-positive cells and rarely in 3βHSD-positive Leydig cells (Fig. 4A,B). Moreover, anti-FOXL2 immunostaining, together with anti-PAX8 antibody (a marker for supporting-like cells and rete testis epithelia; Mayère et al., 2022; Uchida et al., 2022), confirmed no overlap in the expression of FOXL2 and PAX8 signals in the DT-treated testes (Fig. 4C). These data suggest that FOXL2-positive signals are present in certain interstitial stromal cells within the testicular parenchyma, rather than in the rete testis tissues located at the border of the mesonephros.

In XY gonads after E11.5, male-specific vascular/perivascular migration takes place from the adjacent mesonephros (Coveney et al., 2008; Martineau et al., 1997). These mesonephric cells then contribute to certain steroidogenic and smooth muscle cells, along with the formation of a vascular network, within the testicular interstitium (Kumar and DeFalco, 2018). Next, we used E11.5 XY Tg gonads combined with GFP-positive XY mesonephros at E11.5 (Hiramatsu et al., 2009). These XY gonads with GFP-positive mesonephros were cultured for 1 day to reach E12.5 testis development, including mesonephric migration. They were then treated with DT and cultured to recovery for 4 days to induce FOXL2-positive cells (upper scheme in Fig. 4D). In the DT-treated testes, GFP-positive cells originating from the mesonephros were observed in the testicular parenchyma. However, no detectable FOXL2-positive signals were found in the GFP-positive cells derived from the mesonephros in the DT-treated combined explants (Fig. 4D). These data suggest that the FOXL2-positive stromal cells in DT-treated Tg testes may originate from the gonadal parenchyma, rather than the mesonephros, before E11.5.

Finally, most germ cells degenerated inside the deformed testis cords (Fig. S5A), as in previous reports (Rebourcet et al., 2017; Shinomura et al., 2014). Interestingly, a subset of surviving germ cells initiated meiosis and was detected by staining for meiosis markers, including SCP3, REC8 and H2AFX (Fig. S5A,B). These findings align with the tendency towards increased transcript levels of these genes in the bulk RNA-seq analysis (Fig. S5C), further supporting the notion of meiotic initiation in these surviving germ cells.

In fetal ovaries, the surface epithelia continuously proliferate throughout the fetal stages. The result is the ovarian cord structure, which subsequently forms the ovarian cortex, including primordial follicles (Gustin et al., 2016; Mork et al., 2012; Niu and Spradling, 2020). Next, to examine epithelial cell dynamics in DT-treated Tg testes, explant surfaces were labeled by Qdot probes (fluorescent nanocrystals) after DT treatment (Fig. 5A,B, asterisk), and the dynamics of labeled surface cells were traced in a 4-day culture. Cells with Qdot probes contribute to the thickening subepithelial region of DT-treated Tg testes, as in ovarian control explants (Fig. 5B). Anti-FOXL2 staining confirmed the presence of several FOXL2-positive cells with cytoplasmic Qdot probes in the subepithelial region, suggesting FOXL2-positive cells derived from the surface epithelia of the E12.5 testes. Moreover, a pulse labeling experiment using 5′-ethynyl-2′-deoxyuridine (EdU) for 3 h immediately after DT treatment revealed a cord-like cluster with EdU-positive cells in the epithelial/subepithelial region of DT-treated Tg testes, as in the ovarian control explants, albeit at higher density (Fig. 5C,D). In the surface epithelia/subepithelia of DT-treated Tg testes, these EdU-positive cell clusters were FOXL2 positive/NR2F2 negative, suggesting that the testicular epithelia proliferate and contribute to the surface FOXL2-positive cell population in Tg testes after Sertoli cell ablation.

Sertoli cell ablation of E12.5 testes caused ectopic Foxl2 expression in the other testicular somatic cells, suggesting Foxl2 inducibility is maintained in testicular somatic cells, except Sertoli cells, for some time after testis differentiation. To evaluate the time window, Tg testes were isolated at E12.5, E13.5 and E14.5, treated with DT for 24 h, and cultured for 4 days to undergo ex vivo gonadal development in the absence of Sertoli cells (Fig. 6A). After 4-day culture following DT treatment of Tg testes isolated at E14.5, SOX9-positive Sertoli cells were completely depleted in the testicular parenchyma, which also lacked FOXL2-positive signals (Fig. 6B). Moreover, RT-qPCR confirmed that the Foxl2 inducibility in DT-treated Tg testes was maintained in those initiated at E13.5, but lost in those initiated at E14.5 (Fig. 6C). DT-dependent induction of the ovarian cord marker genes Lgr5 and Gng13 (Niu and Spradling, 2020) was reduced in these explants at E13.5 and E14.5, compared with E12.5. However, there was no significant difference in Gng13 expression due to the marked variation among samples even at E12.5. Therefore, the Foxl2 inducibility in the testicular soma is likely maintained at E12.5-13.5 but lost by E14.5.

Sertoli cell ablation caused partial feminization in E12.5 testes, implicating the paracrine function of Sertoli cells in the maintenance of testicular somatic cells, especially the testicular surface epithelia. To identify the factors derived from Sertoli cells that modulate Foxl2 inducibility, DT-treated Tg testes were cultured with PDGF (a promoter of mesonephric cell migration and epithelial cell proliferation; Brennan et al., 2003), BW245C (an agonist for the PGD2 receptor Dp1 with PGD2-like activity; an autocrine factor for Sertoli cell establishment; Malki et al., 2005; Moniot et al., 2009), AMH (a paracrine factor that promotes testis-specific mesonephric migration; Ross et al., 2003), DHH (a paracrine factor that promotes Leydig cell differentiation; Yao et al., 2002) and FGF9 (a key factor for Sertoli cell establishment, mesonephric migration and coelomic epithelial proliferation; Colvin et al., 2001; Kim et al., 2006; Schmahl et al., 2004) (Fig. 7A), all of which are involved in the male-specific pathway for testis formation during the sex differentiation period. Among these additives, only exogenous FGF9 repressed the DT-dependent induction of Foxl2 expression in Tg explants (Fig. 7B). In addition to its dose-dependent repression of Foxl2 inducibility in E12.5 testes (Fig. 7C), FGF9 also repressed expression of Lgr5 and Gng13 in DT-treated Tg explants (Fig. 7D). Anti-FOXL2 immunostaining and morphometric analysis of serial sections of testis explants confirmed a marked reduction of FOXL2-positive signals throughout the testicular parenchyma in DT-treated Tg testes cultured with FGF9 (Fig. 7E,F). Moreover, there was a significant reduction in the number of FOXL2-positive cells in the surface epithelial/subepithelial region or the remaining parenchyma, including the peri-mesonephric region (reduced levels: 23.2% in SE and 47.6% in non SE; Fig. 7G).

In situ hybridization confirmed that Fgfr1 and Fgfr2 (Kim et al., 2007; Schmahl et al., 2004), as well as anti-heparan sulfate proteoglycan (HSPG) (low-affinity receptors/reservoirs for FGF ligands; Harikae et al., 2013b), were highly expressed in the surface epithelial/subepithelial region and the peri-mesonephric region, in contrast to high Fgf9 expression in Sertoli cells (Fig. S6A). Taken together with the RNA-seq data showing no altered Fgfr1 and Fgfr2 expression in 4-day cultures of DT-treated testes (Fig. S6B), the reduction in the FGF9 signal caused by Sertoli cell ablation may trigger the ectopic appearance of FOXL2-positive cells among remaining testicular somatic cells.

Using the AMH-TRECK Tg line, we found that E12.5 testes with Sertoli cell ablation undergo ectopic induction of several ovary-specific genes, including Foxl2, Irx3 and Fst (Jorgensen and Gao, 2005; Loffler et al., 2003; Menke and Page, 2002; Yao et al., 2004). This suggests that partial male-to-female reprogramming is induced in testicular somatic cells by the loss of Sertoli cells ex vivo. Two studies examined the testicular phenotypes of in vivo Sertoli cell ablation using AMH-Cre and Cre recombinase-inducible DT fragment A (DTA) or DT receptor (Rebourcet et al., 2014; Wang et al., 2020). However, Sertoli cell ablation in these testes in vivo could not be induced before E14.5, possibly owing to a considerable time lag in the death of Sertoli cells between the in vivo and ex vivo conditions. No male-to-female sex reversal phenotype, albeit of aberrant testicular vasculature and peritubular stroma, was observed in these fetal testes initiated after E14.5 in vivo. Together with the lack of DT-dependent FOXL2 inducibility in Tg testis initiated at E14.5 (Fig. 6), these data indicate that the cut-off time for FOXL2 inducibility is E13.5. This timing coincides with the establishment of the tunica albuginea, which is a thickened basement membrane layer underlying the surface epithelia and vasculature, after E13.5, as reported by Karl and Capel (1998).

In mouse testiculogenesis, all fetal Sertoli cells are recruited from the gonadal surface epithelia around E11.0-11.5 and tightly packed inside the well-defined testis cords, which are separated from the surface epithelia by E12.5 (Karl and Capel, 1998; see review by Harikae et al., 2013a). In DT-treated Tg testes at E12.5, FOXL2 was upregulated mostly in testicular somatic cell populations, except for the Sertoli cells, suggesting peripheral, not central, localization of ectopic FOXL2-positive cells at the surface epithelium and mesonephric border (Fig. 3A). This is in contrast to the sex reversal phenotypes of bipotential supporting cells in the primary sex cords/testis cords discovered using mutants of several key genes – Rspo1, Wnt4 and Ctnnb1 (Maatouk et al., 2008; see review by Chassot et al., 2014), Dmrt1 (Matson et al., 2011; Zhao et al., 2015), Znrf3 (Harris et al., 2018), Nedd4 (Windley et al., 2022), and Amh and Inhbb (Rodriguez et al., 2022). Therefore, our DT-dependent FOXL2 inducible system using E12.5 Tg testes enables evaluation of feminization in testicular somatic cell components, except for the Sertoli cell lineage in early fetal testes.

In the female-type supporting cell populations, two major FOXL2-positive granulosa cell populations contribute to folliculogenesis in a cortical and medullary axis-dependent manner (Suzuki et al., 2015). One medullary population is bipotential supporting cells in the primary sex cords, which share an origin with fetal Sertoli cells in the testis cords (Albrecht and Eicher, 2001; Mork et al., 2012). The cortical population is the ovarian cords (i.e. secondary sex cords, which are continuously recruited from the ovarian epithelia at the late fetal and perinatal stages; Mork et al., 2012; Rastetter et al., 2014). These two distinct granulosa cell lineages were confirmed by a single-cell RNA sequence and lineage-tracing study (Niu and Spradling, 2020), in which Gng13 and Lgr5 expression marks the ovarian cords at the fetal stage in a female-specific manner (Niu and Spradling, 2020). In our ex vivo model of the fetal testis with Sertoli cell depletion, the FOXL2-positive/WT1-positive/NR2F2-negative cord-like structures were ectopically induced in the surface subepithelial region of DT-treated Tg testes (Fig. 3B, Fig. S4), together with Gng13 and Lgr5 upregulation in the testicular surface region (Figs 2E, 3C and Fig. S3D). Moreover, FOXL2-positive cells were, at least in part, derived from the proliferation of surface epithelial cells (Fig. 5). Therefore, such FOXL2-positive cells in the testicular surface region reflect the ectopic induction of ovarian cord-like FOXL2-positive cell clusters that are typically found in the ovarian cortical region at late fetal and perinatal stages (Mork et al., 2012; Niu and Spradling, 2020; Rastetter et al., 2014). This, in turn, suggests that fetal Sertoli cells repress ovarian cord formation in the surface epithelia, possibly via their paracrine actions, at E12.5-E13.5.

In DT-treated Tg testis explants, Foxl2 inducibility, as well as Lgr5 and Gng13 expression, were repressed by exogenous FGF9 (Fig. 7B,D). FGF9 is a Sertoli cell-derived signal that induces mesonephric/vascular cell migration and coelomic vessel formation during early testis morphogenesis (Colvin et al., 2001), and maintains high SOX9 expression for Sertoli cell establishment in an autocrine manner (Hiramatsu et al., 2010; Kim et al., 2006). Genetic studies of FGF9-FGFR signals revealed male-to-female sex reversal in early fetal testes (Bagheri-Fam et al., 2017; Colvin et al., 2001; Jameson et al., 2012; Kim et al., 2006; Schmahl et al., 2004; Siggers et al., 2014). In fact, some Fgf9^−/− testes, albeit with various feminization phenotypes, showed reduced or depletion of subepithelial mesenchyme (Colvin et al., 2001), similar to the phenotype of ovarian cord-like formation in DT-treated Tg testes. Fgfr1/2 and its mandatory co-factor HSPG were expressed in these two sites (Fig. S6A; Harikae et al., 2013b; Kim et al., 2007; Schmahl et al., 2004), thus Sertoli cell-derived FGF9 signals may contribute to repression of feminization in the surface epithelia and peri-mesonephric stroma in early fetal testes. This is consistent with the concept that FGF9 acts as an antagonist for WNT4-RSPO1 signals to regulate supporting cell establishment (Hiramatsu et al., 2009, 2010; Kim et al., 2006) via FOXL2 upregulation by RSPO1/WNT4/β-catenin signaling (Maatouk et al., 2008). Furthermore, the initiation of meiosis in certain surviving germ cells within DT-treated Tg explants (Fig. S5) is consistent with this finding. The absence of FGF9, which acts as an antagonist to retinoic acid, a meiosis initiation factor (Bowles et al., 2006, 2010; Koubova et al., 2006), further supports the role of FGF9 in regulating these processes.

In addition to FOXL2-positive supporting cell types, theca cells were also derived from at least two distinct FOXL2-positive progenitor populations: WT1-positive and -negative progenitors (Liu et al., 2015; Zhou et al., 2022). Although 3βHSD-positive theca cells were not detected in growing follicles by postnatal day (P) 14 in vivo (Zhou et al., 2022), some theca cells shared an origin with fetal Leydig cells in a Nr5a1 promoter-EGFP line (Miyabayashi et al., 2015). The present study revealed that, in DT-treated Tg testes, most FOXL2-positive stromal cells in the peri-mesonephric region were 3βHSD-negative/NR2F2-positive/PAX8-negative cells that originated from the gonadal parenchyma, rather than the mesonephros, before E11.5 (Fig. 4). Based on the localization of ectopic FOXL2-positive cells with 3βHSD-positive Leydig cells (arrowheads in Fig. 4B), these FOXL2-positive stromal cells are similar to FOXL2-positive theca cell progenitors in fetal ovaries (Liu et al., 2015; Zhou et al., 2022). However, as the perivascular multipotent progenitors in the fetal ovary have been shown to contribute to granulosa, thecal and pericyte cell lineages (Li et al., 2022), the origin and fate of these FOXL2-positive stromal cells in DT-treated Tg testes warrant further research.

In conclusion, Sertoli cell ablation from early fetal testes induced the emergence of FOLX2-positive granulosa/theca-like progenitor cells in the surface epithelial and peri-mesonephric regions (Fig. 7H). Moreover, ectopic FGF9 addition repressed their FOXL2 expression. Our newly designed ex vivo system enables evaluation of the direct actions of signaling factors in testicular somatic cell components after Sertoli cell establishment in the early fetal testis.

Animal experiments were carried out in accordance with the Guidelines for Animal Use and Experimentation of the University of Tokyo. The procedures were approved by the Institutional Animal Care and Use Committee of the Graduate School of Agricultural and Life Sciences of the University of Tokyo (approval ID: P18-046). The C57BL/6-Tg AMH-TRECK (toxin receptor-mediated cell knockout) male mice carrying human AMH-promoter-driven DT receptor-EGFP transgene (#94 line established by Shinomura et al., 2014) were mated with wild-type females (C57BL/6 or ICR strain; SLC) overnight and the next morning checked for the presence of a vaginal plug. Noon on the day when a vaginal plug was detected was considered embryonic day 0.5 (E0.5). Genotypic wild-type littermates were used as controls. C57BL/6-Tg (CAG-EGFP) mice (EGFP mice; SLC) were used for visualization of migrated mesonephric cells in combined culture with XY Tg gonads.

For RNA-sequencing (RNA-seq) analysis, pregnant females were pretreated with busulfan (intraperitoneal injection, 40 mg/kg, twice at E10.5 and 11.5) to eliminate the influence of germ cells and their transcripts on expression profiles in gonadal samples (Miura et al., 2019). We confirmed that the fetal testes of busulfan-treated embryos at E12.5 could induce ectopic appearance of FOXL2-positive cells in a similar manner to those in non-treated embryos in this Tg line by immunofluorescence (see Fig. S7).

At E12.5, fetal testes and ovaries were isolated from wild-type and Tg embryos in ice-cold phosphate-buffered saline (PBS). The sex and genotype of the isolated gonads were determined by their morphology and GFP fluorescence, in addition to genotyping PCR (Shinomura et al., 2014). Gonads were treated with or without DT (D0564-1MG; Sigma-Aldrich; 100-200 ng/ml) in 10% fetal calf serum (FCS)-Dulbecco's modified Eagle medium (DMEM; D5796; Sigma-Aldrich) for 24 h on ISOPORE membrane filters (TSTP02500; Millipore) at 37°C with 5% CO₂. Next, the culture medium was exchanged for 10% FCS-DMEM, and the incubation was continued for up to 4 days, as described previously (Hiramatsu et al., 2003). To evaluate paracrine signals, Knockout Serum Replacement (KSR; 10828010; Gibco)-DMEM was used instead of FCS. DT-treated gonads were cultured in 10% KSR-DMEM supplemented with FGF9 (recombinant human fibroblast growth factor 9, F1168; Sigma; 50 ng/ml; Hiramatsu et al., 2009), BW245C [a hydantoin compound with prostaglandin D2 (PGD2)-like activity, 12050; Cayman Chemical; 50 nM], MIS/AMH (recombinant human MIS/AMH protein, 1737-MS-010/CF; R&D Systems; 100 ng/ml; Yamamoto et al., 2018), DHH (recombinant human desert hedgehog protein N-terminus, 4777-DH; R&D Systems; 100 ng/ml; Pathi et al., 2001) or PDGF-AA (recombinant human platelet-derived growth factor-AA, 165-25541; Fujifilm Wako; 100 ng/ml; Fujikawa et al., 2017) for 4 days. The medium was exchanged every 24 h.

For RNA analysis, germ cell-depleted testes from busulfan-pretreated embryos were used as described previously (Miura et al., 2019). At E12.5, busulfan-pretreated testes were carefully separated from the adjacent mesonephros under a dissecting microscope. The testes were pretreated with DT in FCS-DMEM for 24 h, and incubated for 0.5, 1, 2, 3 and 4 days. Total RNA was extracted using TRIzol reagent (15596026; Invitrogen). Three sets of total RNA samples for each culture period (each set comprising four testes) were used to prepare RNA-seq libraries. The libraries were 50 bp single-end sequenced on an Illumina Novaseq6000 (Illumina). After removal of ribosomal RNA by mapping to the ribosomal RNA sequences using bowtie2 v. 2.3.4.1 (Langmead and Salzberg, 2012), sequences were mapped to the mouse genome (GRCm38/mm10) using STAR v. 2.6.0a (Dobin et al., 2013) and only uniquely mapped tags were extracted. Transcripts mapped to exons in each gene were quantified using featureCounts v. 1.6.2 (Liao et al., 2014).

Differential gene expression was analyzed using DESeq2 v. 1.18.1 (Love et al., 2014) in R (v. 3.4.3; R Development Core Team). Gene expression changes with |log2 fold change|>1.0 and adjusted P-value (Benjamini-Hochberg method) <0.05 were considered significant. The 2984 testis-specific and 2950 ovary-specific gene lists at E12.5-E13.5 (bulk RNA-seq data; Zhao et al., 2018) were compared with these differentially expressed genes. TPM counts were used as normalized counts for hierarchical clustering and temporal gene expression plots. Hierarchical clustering was conducted using the Python Scipy module. K-means clustering was conducted in the iDEP web application (Ge et al., 2018). The top 2000 genes ranked by standard deviations were divided into four clusters.

Samples were fixed in 4% paraformaldehyde (PFA)/PBS and embedded in paraffin wax or OCT compound. Deparaffinized sections (4 μm) or cryosections (10 μm) were subjected to conventional histological (Hematoxylin-Eosin) or protein immunodetection. For immunohistochemical staining and immunofluorescence, sections were incubated with the primary antibodies listed in Table S3 in Tris-buffered blocking solution (TNB) for 12 h at 4°C. The reactions were visualized by incubation with biotin-conjugated secondary antibodies in combination with the Elite ABC Kit (PK-6100; Vector Laboratories) or Alexa-488/594/647 conjugated secondary antibodies listed in Table S3 with 4′,6-diamidino-2-phenylindore (DAPI). Stained sections were analyzed using an Olympus fluorescence microscope (BX51N-34-FL-2) and a Leica TCS SP8 confocal laser microscope.

To measure testis size, cultured explants were photographed using a dissection microscope (SZX16; Olympus) and the lengths of the long (anteroposterior) and short (dorsoventral) axes were measured using ImageJ Fiji (Schindelin et al., 2012). To quantify the number of FOXL2-positive cells, serial sections covering the entire testicular parenchyma were processed for immunostaining and photographed (×10 magnification). The number of FOXL2-positive cells per 10,000 µm² was estimated separately in the surface epithelial and subepithelial (SE) regions, and in the other parenchymal (non-SE) region at the mesonephric side using 4 µm sections 40 µm apart (i.e. every 10 sections) in combination with the cell counter plug-in from Fiji (n=3 testes each).

To detect proliferating cells, gonadal explants were cultured in the presence of 20 µM 5′-ethynil-2′-deoxyuridine (EdU; 052-08843; Fujifilm Wako) for 3 h following 24 h DT treatment. The culture was continued in FCS-DMEM without EdU for 4 days. EdU labeling was visualized with the Click-iT Plus EdU Alexa Fluor 488 Imaging Kit (C10637; Invitrogen) according to the manufacturer's instructions. Paraffin wax sections of EdU-stained samples were co-stained using anti-FOXL2 and anti-NR2F2 antibodies, as described above.

Samples were fixed with 10% neutral-buffered formalin and embedded in paraffin wax. Deparaffinized sections were processed for RNA in situ detection using the RNAscope Target Probe Mm-Lgr5 (312171, NM_010195.2), Gng13 (462531, NM_022422.5), Fgf9 (499811, NM_013518.4), Fgfr1 (443491, NM_010206.3) and Fgfr2 (443501, NM_010207.2) with the RNAscope 2.5 HD Reagent Kit-RED System (ACDBio) as described previously (Imura-Kishi et al., 2021; Uchida et al., 2022).

For quantitative RT-PCR (RT-qPCR) analysis, total RNA was purified from gonadal samples using TRIzol reagent (15596026; Invitrogen) and reverse-transcribed with Superscript VILO MasterMix (1175050; Invitrogen). Taqman gene expression probes for Dhh (Mm01310203_m1), Foxl2 (Mm00843544_s1), Gng13 (Mm00458152_m1), Lgr5 (Mm00438890_m1), Ptgds (Mm01330613_m1) and Wnt4 (Mm00437341_m1) were purchased from Applied Biosystems. PCR was performed using the Step One Real-Time PCR System (Applied Biosystems). Relative levels of the transcripts were normalized to that of Actb (4351315; Applied Biosystems) or Gapdh (4351309; Applied Biosystems) as an endogenous reference.

To visualize the dynamics of the surface epithelium, gonadal surface epithelia were labelled using the Qtracker 655 Cell Labeling Kit (Q25029; Invitrogen). Briefly, whole gonads with mesonephros were treated with DT for 24 h and incubated with 10 nM Qdot probes in FCS-DMEM for 1 h. The labeled gonads were washed in culture medium several times and incubated in FCS-DMEM for 4 days. Some gonads were fixed immediately after labeling to ensure the initially labeled surface area. Cultured gonads were finally fixed and embedded in OCT compound. Cryosections were processed for immunohistochemistry as described above.

Gonads were separated from the XY Tg at E11.5 (16-18 tail-somite stage). GFP-positive mesonephroi were dissected from the XY CAG-EGFP embryos at E11.5. After genotyping by PCR, each Tg gonad was assembled with GFP-positive mesonephros and cultured on 1.5% agar blocks in FCS-DMEM for 24 h to develop the combined testes correspond to E12.5 (Hiramatsu et al., 2009). These explants of the combined gonad-mesonephros were treated with DT for 24 h and recovery-cultured in FCS-DMEM for 4 days. Paraffin wax-embedded sections of these combined explants were co-stained using anti-FOXL2 and anti-GFP antibodies, as described above.

Quantitative data are presented as mean±s.e.m. or as box-and-whisker plots. Two-tailed Student's t-test was performed for single comparisons between two groups. For more than two groups, one-way ANOVA followed by Dunnett's test or Tukey's test were performed. Statistical analysis was carried out using R (v. 3.4.3; R Development Core Team) software. n refers to the number of samples. P<0.05 was considered indicative of significance, and levels of significance are shown as *P<0.05 and **P<0.01.

The authors thank Drs Ryuichi Yamada, Yuki Okada and Kataaki Okubo (The University of Tokyo) for their advice and support, and Ikuno Oike (The University of Tokyo) for her secretarial assistance.