Wnt4 overexpression disrupts normal testicular vasculature and inhibits testosterone synthesis by repressing steroidogenic factor 1/β-catenin synergy

Abstract

Genetic studies in mice suggest that Wnt4 signaling antagonizes expression of male hormones and effectively blocks male development in the female embryo. We recently identified an XY intersex patient carrying a chromosomal duplication of the WNT4 locus and proposed that this patient's feminization arises from an increased dosage of WNT4. To test this hypothesis, a transgenic mouse was generated with a large genomic P1 containing the human WNT4. Although a complete male to female intersex phenotype was not observed in WNT4 transgenic male mice, a dramatic reduction in steroidogenic acute regulatory protein was detected consistent with the marked reduction in serum and testicular androgen levels. Furthermore, a mild reduction of germ cells and a disorganized vascular system were observed in testes of WNT4 transgenic males. Consistent with these in vivo data, Wnt4 repressed steroidogenesis in adrenocortical and Leydig cell lines, as evidenced by reduced progesterone secretion and 3β-hydroxysteroid dehydrogenase activity. In vitro studies showed that Wnt4 antagonizes the functional synergy observed between the major effector of the Wnt signaling pathway, β-catenin and steroidogenic factor 1, and chromatin immunoprecipitation showed that Wnt4 attenuates recruitment of β-catenin to the steroidogenic acute regulatory protein promoter. Our findings suggest a model in which Wnt4 acts as an anti-male factor by disrupting recruitment of β-catenin at or near steroidogenic factor 1 binding sites present in multiple steroidogenic genes.

In the mammalian embryo, male and female urogenital ridges are morphologically indistinct until sex-linked and autosomal genes direct testicular or ovarian development. Although the precise genetic interactions responsible for testicular differentiation remain unclear, both human and mouse mutants have shown that several genes, including Sry, Sox9, SF-1, Dax1, and Wnt4, positively or negatively regulate male sexual development. In male mice, expression of Sry at embryonic day 10.5 triggers several distinct morphological events, including differentiation of Sertoli cells, migration of mesonephric cells into the gonad, and reorganization of the gonadal vasculature (1). After expression of Sry, Sox9 is up-regulated in male mouse gonads at embryonic day 11.5. In turn, Sox9 controls expression of target genes, including SF-1 (2), an orphan nuclear hormone receptor that plays a crucial role in regulating expression of three male-specific hormones: testosterone, Müllerian inhibiting substance, and Insl-3. Coordinated action of these three hormones is required for the normal male gonadal and reproductive tract development (3). The overlapping and sexually dimorphic expression patterns of Sry, Sox9, and steroidogenic factor 1 (SF-1) in the developing testis suggest that these genes are needed to direct male sexual differentiation. Moreover, loss-of-function mutations in each of these genes result in a male to female human intersex phenotype.

Conversely, two genes, Dax1 and Wnt4, are thought to antagonize male development in a dosage-dependent manner. Dax1, an orphan nuclear receptor, antagonizes an attenuated Sry allele when overexpressed (4, 5). Furthermore, several in vitro studies have shown that Dax1 represses SF-1-mediated gene activation (6, 7). However, loss-of-function Dax1 mutants exhibit relatively normal sexual development in both male and female mice (8). In contrast, whereas both male and female Wnt4-knockout mice exhibit similar defects in kidney development and in adrenal function (9, 10), gonadal development and steroidogenic function are affected exclusively in the Wnt4^–/– females and not in Wnt4^–/– males (11). Wnt4-null females are masculinized as demonstrated by the absence of Müllerian ducts (presumptive female reproductive tract) and the presence of Wolffian ducts (presumptive male reproductive tract). In addition, mutant Wnt4^–/– females express steroidogenic enzymes required for production of testosterone, which are normally repressed in the female ovary; these include 3β-hydroxysteroid dehydrogenase (3β-HSD) and 17α-hydroxylase (Cyp17) (11). Collectively, these data suggest that Wnt4 normally functions to repress gonadal androgen biosynthesis in females.

Wnt4 is a member of the Wnt family of secreted molecules that were originally identified as mammalian homologues of the Drosophila wingless gene. This family of signaling factors functions in a paracrine manner to effect a number of developmental changes, including kidney development and angiogenesis (9, 12–14). Wnts bind to members of the Frizzled family of cell-surface receptors and are known to use at least three separate signaling pathways to effect gene expression (15). It is generally presumed that the molecular details elucidated for Wnt1 signaling will apply directly to other Wnts, including Wnt4. However, the specifics of Wnt4 signaling in gonadal or kidney development remain unclear.

Recently, we identified an XY female with ambiguous genitalia carrying a large duplication of chromosome 1p35 including the WNT4 locus (16). We hypothesized that this patient's intersex phenotype results from a gain of WNT4 function. However, because other duplicated genes are present within this chromosomal region, their contributions could not be ruled out. To investigate the hypothesis that Wnt4 normally represses androgen synthesis in females, we generated a gain-of-function mouse model by using a P1 clone carrying the human WNT4 gene controlled by its endogenous promoter. On the basis of our in vivo and in vitro data, we propose that the high levels of Wnt4 expression directly antagonize the normal functions of SF-1 in the female embryo.

Materials and Methods

Generation of WNT4 Transgenics. A clone (P145) containing the complete human WNT4 gene with flanking regulatory sequences was isolated from a P1 library (Incyte Genomics, Palo Alto, CA). Transgenic founders were generated in CB6F1 and C57BL/6 backgrounds (University of California Transgenic Facility, Irvine). Genotyping for sex and transgene was performed by PCR and Southern blotting as described (16, 17). A 1.6-kb genomic fragment specific for WNT4 exons 3–5 was used as probe. RNA extracted from adult mice gonads and kidneys was analyzed qualitatively by RT-PCR for transgene expression as described (16).

Cell Culture and Transfections. The wild-type and hemagglutinin epitope (HA)-tagged SF-1 plasmids were generated as described (18). The mouse β-catenin expression vector was provided by R. Grosschedl (19), and the mouse axin plasmid was provided by F. Costantini (Columbia University, New York). The mouse Wnt4HA, human β-CATENIN-S37A-HA expression vectors, and RatB1a cells were kindly provided by M. Julius (20). The StAR-LUC was provided by D. M. Stocco (21); the human 3β-HSD LUC was provided by S. Mellon (University of California, San Francisco).

Human embryonic kidney cells (HEK293S), mouse adrenocortical cells (Y1), RatB1a fibroblasts (RB1), and human colorectal adenocarcinoma cells (SW480) were maintained in DMEM supplemented with antibiotics and serum (HEK293S, 5% FBS/5% calf serum; Y1, 2.5% FBS/15% horse serum; RB1, 2.5% FBS/7.5% calf serum; SW480, 10% FBS). The mouse Leydig tumor cells (MA10) were cultured in Waymouth medium supplemented with 15% horse serum, 20 mM Hepes, and 50 mg/ml gentamycin. Transient transfections were performed with calcium phosphate (Specialty Media, Lavellette, NJ) for HEK293S cells or with FuGENE 6 (Roche Molecular Biochemicals) for Y1 cells. For coculture experiments, RB1 cells were treated with or without 1 mM sodium butyrate for 24 h before culturing with MA10 cells. Luciferase activity was determined as described (2). All transfections were performed in triplicate and normalized with β-galactosidase activity (carried by pCMV-βGal, 0.1 μg) for transfection efficiency. TRIzol reagent (GIBCO/BRL) was used to extract total RNA from coculture cells; primers and RT-PCR conditions were as described (2, 22, 23).

Western Blotting Analyses and Immunohistochemistry. Total cell lysate (10–40 μg) was separated by electrophoresis on SDS/polyacrylamide gels, transferred to a nitrocellulose membrane, blocked in 3% milk, blotted overnight at 4°C with primary antibodies, washed, and incubated with either sheep anti-rabbit IgG coupled to horseradish peroxidase (1:10,000) or goat anti-mouse IgG coupled to horseradish peroxidase (1:10,000) for 1–2 h at room temperature. Rabbit anti-SRB1 (1:20,000, Novus Biologicals, Littleton, CO), monoclonal antibodies against β-catenin (1:10,000, BD Biosciences) or against the HA tag on Wnt4HA (1:2,000, Covance) were used. Detection of SF-1 and steroidogenic acute regulatory protein (StAR) proteins was as described (24). Blots were developed by using a chemiluminescence kit (ECL, Amersham Biosciences). Sections of paraffin-embedded testes were probed with c-kit antibody (1:2,000, Santa Cruz Biotechnology) to mark germ and Leydig cells.

Steroid Measurements. Progesterone concentrations in culture medium were determined by using an ¹²⁵I RIA kit (ICN). All samples were run in duplicate. The cytochemical assay for 3β-HSD activity in MA10 cells was previously described (25). The blue enzymatic 3β-HSD signal in each transfected well was analyzed with the nih image program. Serum testosterone concentrations were analyzed by Esoterix Endocrinology (Calabasas Hills, CA). Mice were killed by cervical dislocation and immediately dissected to access the heart. After an incision was made in the right atrium, blood was collected as it pooled in the thoracic cavity. For testicular testosterone measurement, testes were homogenized in PBS and extracted with acetone (1:10 ratio). Excess acetone was allowed to evaporate in 37°C water bath for 5–8 h; the remaining supernatant was subjected to testosterone measurement with an ¹²⁵I RIA kit (ICN). The Student's t test was used for all statistical analyses in this study.

Chromatin Immunoprecipitation (ChIP) Assay. HEK293S cells were cotransfected with SF-1 (2 μg), β-CATENIN S37A (2 μg), and the StAR promoter–reporter plasmid (4 μg) with or without Wnt4 (2 μg) in 150-mm dishes. ChIP assays were carried out as described (26). Briefly, cells were fixed with formaldehyde, lysed, and resuspended in buffer. The resulting lysate was sonicated for 3 min on ice and cleared by centrifugation. Samples were immunoprecipitated with the anti-β-catenin antibody (BD Biosciences) and washed. Cross-linking was reversed on the bound immunocomplex, and the reaction was subjected to 30 cycles of PCR at 57°C for the StAR promoter and at 55°C for the control HSP70 promoter gene expression. The following primers were used: StAR-FOR, 5′-GCGAGGACAGGGCTTGAAGT-3′; StAR-REV, 5′-CGCAGATCAAGTGCGCTGCC-3′; HSP70-FOR, 5′-GGATCCAGTGTTCCGTTTCC-3′; and HSP70-REV, 5′-AGTGCTTCATGTCCGACTGC-3′.

Lectin Perfusion and Confocal Microscopy. Adult male mice were anesthetized with i.p. injections of 350 μl of 2.5% avertin solution, followed by injection of 125 μl of FITC-labeled lectin (from Lycopersicon esculentum, Vector Laboratories) into the femoral vein. Mice were perfused with 1% paraformaldehyde, and testes were embedded in 7% SeaPlaque agarose (Biowhit-taker). Sections (100 μm) were mounted on slides and counterstained to label cell nuclei with ToPro3 in VectaShield mounting solution (Vector Laboratories). confocal assistant was used to reconstruct the confocal images.

Results

Overexpression of WNT4 in Mice Disrupts Testicular Vasculature and Testosterone Synthesis. To explore the effect of gain of WNT4 function on gonadal development and function, we generated transgenic mice carrying the human WNT4 gene with its endogenous regulatory sequences (Fig. 1A). The protein encoded by human WNT4 shares 99% identity with mouse Wnt4. Among six identified transgenic founders, only one line, 90-2, expressed the transgene appropriately in the kidney and gonad (data not shown). Southern blot hybridization with a transgene-specific probe indicated that line 90-2 carries two copies of the WNT4 transgene (Fig. 1B). Although analysis of the human WNT4 protein expression is not possible because of the lack of a reliable anti-Wnt4 antibody, semiquantitative RT-PCR analyses showed similar gonadal expression of both the human WNT4 and endogenous mouse Wnt4 transcripts (Fig. 1C).

Fig. 1.

Human WNT4 construct in transgenic mice. (A) Schematic representation of the WNT4 transgene. Transgenic mice were generated with an 80-kb P1 clone containing the five exons of human WNT4 and at least 25 kb of 5′ and 3′ flanking sequence. (B) Two copies of WNT4 constructs were integrated into the genomes of transgenic mice as shown by Southern blotting. Lane 1 shows the WNT4 hybridization signal of a transgenic mouse. Lanes 2 and 3 are negative controls containing DNA from nontransgenic littermates. Lanes 4 and 5 are positive controls containing the same amount of human genomic DNA. (C) Evaluation of WNT4 transgene expression level. RT-PCR of the β-actin control (upper bands) and WNT4 transgene (lane 1, lower band) and the endogenous mouse Wnt4 (lane 2, lower bands) from transgenic (Tg, Upper) and wild-type (Wt, Lower) testes. Lanes 3 and 4 are minus RT controls corresponding to lanes 1 and 2, respectively. (D and E) On the surface testicular vasculature of transgenic males (E), a marked reduction in the number of blood vessels and in the degree of vessel branching was observed compared with the testes of wild-type littermates (D). (F–I) Confocal images of the internal testicular vasculature (labeled with lectin, a green fluorescent endothelial cell marker) showing organized vasculature in a wild-type testis (F and H) and disorganized vascular branching in a WNT4 transgenic testis (G and I). (F and G, ×40; H and I, ×100.)

We examined the testicular vasculature, given that this is a major characteristic feature of the testis, and found the vasculature to be disrupted in WNT4 transgenic males. While the testicular artery was present on the surface of both transgenic and wild-type testes, the number of collateral vessels branching off this artery was markedly reduced in transgenic animals (Fig. 1 D and E). To examine the internal vasculature, we perfused adult males with an FITC-labeled lectin that binds specifically to endothelial cells of blood vessels. In contrast to the highly organized vasculature observed in wild-type testes, blood vessels in the transgenic testes seemed disorganized and failed to encircle the seminiferous tubules (Fig. 1 F–I). In addition, the internal vasculature of transgenic testes showed an increase in vascular branching with thread-like projections. This defect was restricted to testes; all other vascular development appeared normal (data not shown).

Previous studies suggested that loss of Wnt4 function in mice leads to abnormal testosterone synthesis in embryonic ovaries (11). Here, in WNT4 transgenic male mice, total serum and testicular testosterone concentrations were significantly lower than in wild-type littermates (Fig. 2 A and C). Lowered testicular androgen levels suggest that a reduction in serum testosterone is not due to the abnormal testicular vasculature, but instead represents a primary defect in androgen synthesis in WNT4 transgenic testes. The physiological effects of low testosterone were evident by both a weight reduction (Fig. 2B) and altered morphology of the androgen-sensitive organ, seminal vesicles. Seminal vesicles in WNT4 transgenic males appeared underdeveloped, lacking the deep invaginations characteristic of the wild-type organ (Fig. 2 D and E). Testicular histology in the transgenic animals also revealed elongation of seminiferous tubules in all planes of sections, and thinning of the epithelium with a moderate reduction in round spermatids; these features are consistent with low testosterone (Fig. 2 F and G). WNT4 transgenic males are fertile despite these low levels of testosterone. We also noted that the fertility and ovarian vasculature are normal in WNT4 transgenic females (data not shown). This observation is consistent with the hypothesis that Wnt4 functions as an anti-male factor and that overexpression of Wnt4 in the female gonad does not interfere with ovarian function.

Fig. 2.

Lower testosterone levels in WNT4 transgenic mice. Testosterone levels in peripheral blood (A) and testes (C) are significantly reduced in WNT4 transgenic adult males compared with age-matched wild-type (Wt) controls (A; **, P < 0.01). (B) Seminal vesicles weigh significantly less in WNT4 transgenic males than in the wild type (**, P < 0.01). (C) Testicular testosterone was analyzed from three different pairs of age-matched (12 or 18 months) wild-type and transgenic testes. Wild-type testosterone level (ranges from 12 to 40 ng per testis) is taken as 100% for each pair. (D and E). The deep invaginations (white arrows) characteristic of wild-type seminal vesicles (D) are absent from the seminal vesicles of transgenic males (E). (F and G) Immunohistostaining of c-kit, a germ and Leydig cell marker, on equivalent testes sections (identical orientation) revealed the abnormal elongation and thinning of the seminiferous tubules epithelium in transgenic animals (G) compared with the wild type (F). (D and E, ×2; F and G, ×100.)

The inhibitory effects of Wnt4 on testosterone synthesis were tested in steroidogenic Leydig MA10 cells by coculturing with an inducible Wnt4-expressing fibroblast RatB1a (RB1) cell line. In these cocultures, 3β-HSD was measured by a cytochemical colorimetric assay after induction of Wnt4 in RB1 cells by butyric acid (25). We noted a significant reduction in 3β-HSD activity (40%, Fig. 3A) and transcripts (Fig. 3B) in MA10 cells after Wnt4 induction compared with control cocultures (without butyric acid, Fig. 3A). Moreover, coculturing MA10 cells with increasing ratios of Wnt4-expressing RB1 cells resulted in a dose-dependent inhibition of 3β-HSD (Fig. 3C). Collectively, these cellular studies support our in vivo findings that an increased dosage of Wnt4 lowers testosterone in male mice.

Fig. 3.

Wnt4 inhibits steroid biosynthesis in mouse testicular Leydig cells. (A) 3β-HSD expression was determined in MA10 cells cocultured with either RB1 cells (+RB1) or RB1 cells secreting Wnt4 (+RB1/Wnt4). Representative plates are shown from five independent experiments done in triplicate. The intensity of blue reflects 3β-HSD activity and is reduced after coculture of MA10 cells with RB1 cells expressing Wnt4 (Left). The signal intensity for 3β-HSD was quantified by nih image (Right, bar graph; **, P < 0.01). (Inset) HA-tagged Wnt4 protein expression is shown in RB1 cells treated with (+) or without (–) butyric acid, which induces Wnt4 protein expression. (B) RT-PCR analysis of Wnt4 and 3β-HSD mRNA expression in MA10-RB1 coculture. β-Actin was used as a control to indicate the relative RNA input. (C) Dose-dependent reduction of 3β-HSD expression in MA10-RB1 coculture assay. For every point showing the ratio of RB1 to MA10 cells (x axis), 3β-HSD was measured in MA10 cells cocultured with either RB1 cells (+RB1) or RB1 cells expressing Wnt4 (+RB1/Wnt4). Values obtained for MA10 cells cocultured with control RB1 cells are taken as 100%.

β-Catenin Enhances SF-1-Mediated Transcription. Given that Wnt4 repressed steroidogenesis in MA10 cells and in transgenic mice, we investigated how β-catenin, a downstream effector of Wnt signaling, might affect SF-1 action. We tested the activities of SF-1 and β-catenin on a reporter driven by five tandem SF-1 response elements, 5x-SF-1RE LUC (27). Cotransfection of SF-1 and the constitutively active β-catenin S37A dramatically increased SF-1 reporter construct activity 10-fold over the activity induced by SF-1 alone (Fig. 4A). Enhancement of SF-1 activity by β-catenin increased in a dose-dependent manner and was observed also with wild-type β-catenin, although increased concentrations of wild-type β-catenin plasmid were needed to achieve the same effect. A fully intact SF-1 was required for enhancement by β-catenin; SF-1 mutants in the DNA-binding domain (Fig. 4B) or in the activation function 2 domain failed to show this effect (data not shown). Furthermore, overexpression of axin, a direct inhibitor of β-catenin, attenuated the functional synergism between SF-1 and β-catenin (Fig. 4B).

Fig. 4.

β-Catenin enhances SF-1-mediated activation of SF-1-responsive promoter, and this activation is repressed by the Wnt4 signaling pathway. (A) The effects of β-catenin on SF-1 transcriptional activity are shown on a reporter containing five copies of SF-1 response element (5x-SF1-RE LUC, 0.2μg). SF-1 expression plasmid (0.2 μg) was cotransfected with increasing amounts of either β-catenin S37A (10–200 ng, gray bars) or wild-type β-catenin (0.2–2 μg, black bars) in HEK293S cells. Plus (+) and minus (–) indicate the presence or absence of transfected plasmids. Luciferase activity is shown as fold activation, where transfection of reporter alone (5x-SF1-RE LUC) is taken as 1-fold. (Inset) Protein expression of β-catenin, detected by using an anti-β-catenin antibody, is shown when empty vector (–) or HA-tagged β-catenin S37A expression plasmid (+) was transfected into HEK293S cells. Endogenousβ-catenin from SW480 cell lysate was used as positive control (SW, far left lane). (B) β-Catenin enhancement of SF-1 activity was not observed with an SF-1 mutant deleted of DNA-binding domain (SF-1 ΔDBD) or after cotransfection of axin (0.2 μg) in HEK293S cells. (C)(Left) Progesterone levels in cultured medium were measured after Y1 cells were transfected with different combinations of SF-1, β-catenin S37A, or Wnt4 expression plasmids (0.2 μg each; *, P < 0.05). (Right) Protein expression of transfected β-catenin S37A (+, Upper), Wnt4 (+, Lower), and empty expression vector (–) is shown for Y1 cells.

This functional synergism between SF-1 and β-catenin was further tested in mouse adrenocortical Y1 cells by measuring the basal levels of progesterone secretion. By contrast to HEK293S cells, which are negative for both SF-1 and β-catenin, Y1 cells express modest amounts of β-catenin and high levels of SF-1 (data not shown and ref. 2). Transfection of the β-catenin S37A mutant alone resulted in a slight increase in progesterone secretion. Overexpression of both SF-1 and β-catenin resulted in a significant increase in hormone secretion, whereas cotransfection of a Wnt4 expression vector inhibited progesterone secretion below basal levels (Fig. 4C). Thus, these results in Y1 cells are similar to those obtained in MA10 cells, showing that steroidogenesis is enhanced by SF-1 and β-catenin and repressed by Wnt4.

Wnt4 Represses β-Catenin Enhancement of SF-1-Mediated Transcription. Repression of the SF-1/β-catenin functional interaction by Wnt4 was tested further on three different SF-1-responsive reporter constructs. These included the tandem SF-1-responsive elements (5x-SF-1RE LUC) used above, the proximal promoter of 3β-HSD (3β-HSD LUC), and the proximal promoter of StAR (StAR-LUC). StAR shuttles cholesterol into the mitochondria and is therefore a rate-limiting step in steroidogenesis. Exogenous SF-1, Wnt4, and β-catenin were transfected in HEK293S cells. For all three reporters, we observed increased activity with cotransfection of both SF-1 and β-catenin (Fig. 5), although this increase was much less robust with the 3β-HSD LUC reporter. By contrast, Wnt4 attenuated the SF-1/β-catenin activity, but not SF-1 activity alone.

Fig. 5.

Wnt4 inhibits the SF-1 and β-catenin synergism on SF-1 target gene promoters. Luciferase activities of three different SF-1-responsive promoters (A, 5x-SF1-RE LUC; B,3β-HSD LUC; C, StAR LUC, 0.2 μg each) were determined in HEK293S cells transfected with different combinations of SF-1, β-catenin S37A, and Wnt4 expression plasmids (all plasmids 0.2 μg, except 0.4 μg of Wnt4 in C, far right bar). Plus (+) and minus (–) indicate the presence or absence of transfected plasmids. (Inset) Protein expression of Wnt4 is shown when empty vector (–) or HA-tagged Wnt4 expression plasmid (+) was transfected in HEK293S cells.

To further investigate Wnt4-induced repression in vivo, we analyzed the expression of SF-1, β-catenin, StAR, and scavenger receptor class B type 1 (SRB1) in the WNT4 transgenic testes. Both StAR and SRB1 are transcriptionally regulated by SF-1, and SRB1 is the predominant cell-surface receptor that supplies serum cholesterol into cytoplasm of steroidogenic tissues (28). A significant decrease in StAR protein expression was observed in WNT4 transgenic adult testes compared with wild-type littermates, whereas no obvious differences were noted for SF-1, β-catenin, and SRB1 (Fig. 6A). The lowered StAR levels are consistent with reduced testosterone observed in the WNT4 transgenic mice. To further elucidate the repression by Wnt4 signaling on StAR expression, ChIP assays were carried out in HEK293S cells transfected with SF-1/β-catenin and the StAR promoter. Using PCR primers that spanned the two SF-1 binding sites in the proximal StAR promoter, we found that recruitment of β-catenin to the StAR promoter reporter is attenuated markedly after activation of Wnt4 signaling (Fig. 6B).

Fig. 6.

Wnt4 reduces β-catenin recruitment to the proximal StAR promoter. (A) Protein expression of endogenous mouse SF-1 (52 kDa), β-catenin (92 kDa), SRB1 (82 kDa), StAR (30 and 37 kDa), and actin (48 kDa) was analyzed from adult testes of WNT4 transgenic mice (WNT4) and wild-type (Wt) age-matched controls. Decreased expression of both the functional (37-kDa) and the cleaved (30-kDa) StAR proteins was observed in WNT4 transgenic mice. Protein molecular weight markers are as indicated on the left. (B) A schematic diagram of the StAR proximal promoter is shown with the two SF-1 response elements (ovals), the two C/EBP-binding sites (black boxes), and the transcriptional initiation site (+1) (21). Arrows indicate the primers used for PCR after ChIP. Two independent ChIP experiments are shown (I and II) in HEK293S cells after cotransfection of StAR LUC, SF-1, and β-catenin (+SF1/βCAT) with (+Wnt4) or without Wnt4 (–Wnt4). Cells transfected with StAR LUC reporter and empty expression plasmids were used as controls (C, left lanes). Purified HEK293S DNA was immunoprecipitated without (input, Top) or with anti-β-catenin antibodies (Middle). HSP70 primers were used as controls to show nonspecific DNA binding to the anti-β-catenin antibody (HSP70, Bottom).

Discussion

In the last decade, new insights into mammalian sexual differentiation have suggested that the morphological fate of the indifferent gonad depends not on a single testis-determining gene but on a delicate balance between genes that promote or inhibit testis development (29). On the basis of human and mouse genetic studies, Dax1 and Wnt4 are both proposed to counteract male development. Our recent identification of an XY female carrying a duplication including the WNT4 locus (16) strengthens the hypothesis that WNT4 functions as an anti-male factor and is consistent with the masculinized phenotype of female embryos observed in Wnt4-null mice. In the present study, our transgenic mouse model showed that WNT4 overexpression interferes with the normal development of male gonadal vasculature and with testosterone biosynthesis. In vitro studies strongly suggest that Wnt4 suppresses steroid biosynthesis by antagonizing SF-1-mediated gene transcription and disrupting the functional synergism between SF-1 and β-catenin. Our findings lead to the proposal that Wnt4 signaling disrupts recruitment of β-catenin at or near SF-1 binding sites within multiple steroidogenic promoters.

Our results demonstrate that overexpression of WNT4 in male mice disrupts normal testicular vasculature and function, but does not lead to an XY sex-reversed phenotype observed in a human patient carrying a duplication of the WNT4 locus (16). However, it should be noted that among the four known XY human patients with duplications of chromosome 1p, and who presumably overexpress WNT4, symptoms range from isolated cryptorchidism to severe genital ambiguity. Thus, the phenotype in our one WNT4 transgenic mouse line may simply recapitulate the milder symptoms exhibited by human patients, or may reflect the fact that other duplicated genes present in this region are required to observe a fully XY sex-reversed phenotype. Discrepant phenotypes between engineered mouse mutants and human patients have been observed for other genes involved in sex determination. In an analogous situation, duplication of the DSS region containing DAX1 leads to XY female sex reversal in humans, whereas Dax1 transgenic mice exhibit only a minor delay in testicular development (4). Moreover, the XY intersex phenotype observed in heterozygous SOX9 and SF-1 human patients is not exhibited by either SF-1 or Sox9 heterozygous targeted mice (24, 30). These results could imply that humans and mice differ in their mechanisms controlling sexual development or that they may differ in their dosage sensitivity to gene products because of genetic backgrounds. Indeed, in Dax1 transgenic mice, an overt intersex phenotype was observed only in a weakened Sry allelic background (4). At present, it is still unclear whether altering the genetic background would exacerbate the gonadal phenotype observed in WNT4 transgenic male mice. However, it is of interest to note that all other WNT4 transgenic founders generated on a pure C57BL/6 background were infertile (data not shown); whether these infertile WNT4 transgenic founders mimic the more severe phenotypes found in some WNT4-duplication XY individuals remains unknown.

In the developing testis, the formation of the coelomic vessel and the concurrent reorganization of the existing gonadal vasculature mark one of the earliest sex-specific morphological events (1). The finding that WNT4 transgenic mice exhibit a disruption in the testicular vasculature suggests that WNT4 signaling influences this male-specific process during early gonadogenesis. Our findings are reminiscent of other studies linking Wnt signaling and vascular development. For example, Wnt2 has been shown to play a role in the vascularization of the placenta (12), and members of the Frizzled receptors for Wnts, FZD4 and Fzd5, have been implicated in normal angiogenesis of the retina and in the vascularization of the yolk sac, respectively (13, 14). In WNT4 transgenic testes, the increased internal vasculature contrasts the decreased branching of the testicular artery and might suggest that WNT4 primarily influences branching of small vessels. Consistent with this notion, Wnt4 signaling has been shown to stimulate the side branching of terminal ducts, but not major ducts, in mammary glands (31).

The in vivo and in vitro disruption of steroidogenesis by overexpressing Wnt4 implies that this signaling factor dampens expression of some proteins, such as StAR, which are required for steroid biosynthesis. Here, we focused on the possibility that Wnt4 signaling might repress the activity of SF-1, as this orphan nuclear hormone receptor is known to regulate many steroidogenic enzymes in conjunction with other regulatory factors. The canonical Wnt signaling pathway is known to affect the nuclear association of β-catenin and members of the TCF/LEF family to effect gene expression (15). Our analyses of the functional synergism between SF-1 and β-catenin strongly suggest that once SF-1 is tethered to DNA, β-catenin is recruited. Using a mammalian two-hybrid Lex-A fusion system, we found that the hinge and N-terminal portion of the ligand-binding domain of SF-1 are required for this functional interaction (data not shown). Previous studies have shown a ligand-dependent functional and physical association between the androgen receptor and β-catenin (32–34). However, a robust interaction between SF-1 and β-catenin was not observed in our hands, despite repeated attempts to show direct interaction. Instead, we found a weak, but reproducible, interaction between SF-1 and β-catenin (data not shown), comparable to the modest interaction reported by Mizusaki et al. (35). The weak association between SF-1 and β-catenin may reflect the lack of an SF-1 ligand or, alternatively, may suggest that additional bridging factors are needed for the integrity of the SF-1/β-catenin complex.

Although our data show that Wnt4 attenuates β-catenin recruitment to an SF-1 target promoter, the precise molecular nature for this inhibitory effect remains unclear. Clearly, Wnt4 could activate repressors of SF-1, such as Dax1, as recently proposed (35). However, our studies support a Wnt4 repression that is independent of Dax1 because neither Y1 nor HEK293S cells express endogenous Dax1 (data not shown). Other potential targets affected by Wnt4 signaling might include Sox proteins; we found that Sox3, Sox8, and Sox9 markedly inhibit the SF-1/β-catenin functional synergy (data not shown). These data are consistent with the known interference of the canonical Wnt/β-catenin pathway by Sox proteins via direct interaction with β-catenin (36, 37). The use of a noncanonical signaling pathway by Wnt4, as proposed in the original classification of Wnts (38), suggests that Wnt4 may antagonize the canonical Wnt/β-catenin signaling pathway as shown in other systems (39). Activation of these noncanonical pathways could disrupt the SF-1/β-catenin functional interaction either by destabilizing the β-catenin protein or by disrupting SF-1 DNA binding. However, endogenous β-catenin was not decreased after Wnt4 overexpression (data not shown), and there is little evidence that β-catenin competes directly for SF-1 DNA binding. Clearly, further experiments are needed to delineate the precise molecular mechanisms that account for Wnt4 repression of SF-1-mediated transcription.