Tissue-specific requirements of β-catenin in external genitalia development

External genitalia development starts at around 4 weeks of gestation in humans (Spaulding, 1921; Yamada et al., 2003) and on embryonic day 10.5 (E10.5) in mice(Perriton et al., 2002; Suzuki et al., 2002), when paired genital swellings form on either side of the cloaca. The swellings subsequently merge medially to form the genital tubercle (GT) and continue to grow distally. Within the GT, cloacal endoderm extends distally to form the future urethra (Felix, 1912; Kurzrock et al., 1999a; Perriton et al., 2002). Early GT development involves coordinated growth and patterning of cells from the ectodermally derived surface epithelium, the mesodermally derived mesenchyme and the endodermally derived urethral epithelium (UE)(Kurzrock et al., 1999a; Perriton et al., 2002). Up to E15.5, male and female GTs are morphologically indistinguishable(Suzuki et al., 2002), and their development is presumably controlled by the same genetic program. On and after E16.5, the urethrae in males canalize in the presence of androgen signaling, whereas they remain as an epithelial cord in females(Baskin et al., 2001; Suzuki et al., 2002; Yamada et al., 2003). We focused on studying the genetic program regulating early androgen-independent GT patterning.

Both the GT and the limb bud share similar morphogenetic and signaling pathways, perhaps reflecting a similar evolutionary origin (teleost fins). Shh (Haraguchi et al.,2001; Perriton et al.,2002), Wnt5a (Suzuki et al., 2003; Yamaguchi et al., 1999), Bmp4, Bmpr1a and noggin(Dunn et al., 1997; Suzuki et al., 2003), and HOX genes (Hoxd13 and Hoxa13)(Dolle et al., 1993; Fromental-Ramain et al., 1996; Morgan et al., 2003; Warot et al., 1997; Zakany et al., 1997), are all essential for the development of both appendages. In addition, both processes require an epithelial signaling center marked by Fgf8 expression,namely the distal urethral epithelium (dUE) in the GT(Cohn, 2004; Haraguchi et al., 2000; Suzuki et al., 2003; Yamada et al., 2006) and the apical ectodermal ridge (AER) in the limb(Cohn et al., 1995; Crossley et al., 1996; Lewandoski et al., 2000; Mariani and Martin, 2003; Niswander et al., 1993; Saunders, 1948; Summerbell, 1974). Several lines of evidence support the notion that FGF8 is the GT outgrowth-promoting factor. First, surgical removal of Fgf8-expressing dUE resulted in defective GT outgrowth. Second, neutralizing FGF8 with antibody caused similar outgrowth defects. Finally, outgrowth in dUE-deficient GTs can be restored by the application of FGF8 protein beads(Haraguchi et al., 2000). However, in contrast to the AER, little is known about how the dUE is established and maintained within the endodermal cloaca, and how it functions to promote GT outgrowth. The development of the GT differs from that of the limb bud in that GT growth and patterning has to be coordinated with endodermal urethral development, which requires Fgf10/Fgfr2signaling (Petiot et al.,2005) and Hoxa13(Morgan et al., 2003; Scott et al., 2005). Last, but not least, the GT forms with left-right symmetry, whereas the limb develops asymmetrically. Although both Wnt5a^-/- and Tcf1/Tcf4 double-knockout embryos show GT agenesis(Gregorieff et al., 2004; Suzuki et al., 2003; Yamaguchi et al., 1999), both mutants exhibit severe caudal truncations, raising the concern that the genital phenotype in these mutants might be secondary. Therefore, the involvement of WNT signaling in GT development is not clear. In this study, by using tissue-specific inactivation of β-catenin, a key signal transducer of the canonical WNT pathway, we show that β-catenin-mediated WNT signaling is required at multiple stages for directing GT outgrowth and urethra formation. β-Catenin function is also required in the ventral ectoderm to maintain epithelial integrity.

Msx2-Cre, Shh^Cre/GFP, Shh^tm(Cre/Esr1), TOPGAL and R26R strains were purchased from the Jackson Laboratory (Bar Harbor, MN). Shh^Cre/GFPand Shh^tm(Cre/Esr1) transgenic lines express an EGFP-Cre fusion protein and a tamoxifen-inducible Cre from the endogenous Shhlocus, respectively (Harfe et al.,2004). β-Cat^flox/flox (LOF),β -Cat^{loxEx3/loxEx3} (GOF) and Dermo1-Crestrains were described previously (Brault et al., 2001; Harada et al.,1999; Yu et al.,2003). Briefly, β-Cat^flox/flox mice bear two LoxP sites flanking exons 2-6, which will lead to a loss-of-function deletion upon Cre-mediated excision.β -Cat^{loxEx3/loxEx3} mice have only exon 3 floxed,deletion of which leads to a stabilized form of β-catenin. Tamoxifen(Sigma-Aldrich, St Louis, MO) was dissolved in corn oil at a concentration of 20 mg/ml and was delivered by gavaging pregnant females at a dose of 0.2 g/kg of body weight. For studies involving embryos from E14.5 onward, only males were presented (except for the study shown in Fig. S5D in the supplementary material).

³⁵S in situ hybridizations were performed on paraformaldehyde(PFA)-fixed, paraffin-embedded, 10-μm sections, as described previously(Wawersik and Epstein, 2000). Whole-mount in situ hybridization was performed as previously described(Wilkinson, 1992). Probes for Tcf1, Lef1, Tcf4 and Ptc1(Hu et al., 2005), Shh (Bitgood and McMahon,1995), Msx2 (Yin et al., 2006), Wnt5a(Huang et al., 2005), Hoxa13 and Hoxd13, Bmp4(Jones et al., 1991) and Fgf8 (Crossley and Martin,1995) were described previously. The Wnt11 probe was a gift from Dr Andy McMahon (Harvard University, Cambridge, MA). Wnt9b,Wnt2 and Wnt3 probes were generated using ATCC clones or PCR amplification.

Embryos were fixed in Bouin's fixative, embedded in paraffin after dehydration and sectioned at 5 μm. Hematoxylin and eosin staining and X-gal analyses were performed following standard protocols. Immunofluorescence was performed as described previously (Yin et al., 2006). Primary antibodies used in this study (all in 1:300 dilutions) were as follows: anti-β-catenin, anti-E-cadherin,anti-plakoglobin (BD biosciences, San Jose, CA), and anti-phosphoH3(Millipore, Billeric, MA).

For scanning electron microscopy (SEM) analysis, samples were fixed in 3%glutaraldehyde, post-fixed with 1% aqueous osmium tetroxide for 4 hours, then processed using the Osmium-Thiocarbohydrazide-Osmium (OTO) method, dehydrated in alcohol and critical-point dried in liquid CO₂. Mounted samples were sputter-coated and examined in a Hitachi S-450 SEM. For transmission electron microscopy (TEM) and semi-thin histology, samples were fixed in paraformaldehyde and glutaraldehyde, and post-fixed with aqueous 1%OsO₄ for 2 hours. Samples were dehydrated, subjected to three changes of propylene oxide, and infiltrated with Polybed 812 epoxy embedding resin. Specimen blocks were polymerized at 60°C in a vacuum oven. Thin sections were generated, post-stained with 2.5% uranyl acetate and lead citrate, and examined in a Hitachi H-600 TEM, or post-stained with Epoxy Tissue Stain (catalogue number 14950, Electron Microscopy Sciences, Hatfield,PA) and followed by light microscopy.

TUNEL staining was performed on PFA-fixed, paraffin-embedded, 10-μm sections using the In Situ Cell Death Kit (Roche Diagnostic, Indianapolis,IN), according to the manufacturer's instructions.

Data were analyzed by unpaired Student's t-test, and results are expressed as means±s.e.m. The number of independent experiments is specified in the Results.

To determine whether canonical WNT signaling is activated in the developing mouse GT, we used TOPGAL transgenic mice to report canonical WNT-β-catenin activity (DasGupta and Fuchs, 1999). In the cloaca region, transgene activity was first detected at E10.5 (Fig. 1A). Sections of stained embryos revealed that β-galactosidase(β-gal)-positive cells were localized to the cloacal endoderm(Fig. 1E, arrows). As GT development progressed, this distal urethral β-gal expression persisted(Fig. 1B-F, arrows). Distinct populations of β-gal-positive cells were also detected on the proximolateral sides of the GT at E13.5(Fig. 1C, arrowheads) and in the labioscrotal swellings at E14.5 (Fig. 1D, arrowheads). Next, we examined the expression of transcriptional mediators of WNT signaling, T-cell factors, in the GT. RT-PCR analysis showed that Tcf1 (Tcf7 - Mouse Genome Informatics), Lef1 and Tcf4 (Tcf7l2 - Mouse Genome Informatics)were expressed throughout GT development (see Table S1 in the supplementary material), and ³⁵S-in situ hybridization revealed that Tcf1 and Tcf4 were co-expressed in the dUE where TOPGAL activity was detected (Fig. 1G,H, arrows). In addition, Tcf1 transcripts were also detected in the ventral ectoderm and distal mesenchyme, while Tcf4was also expressed in the proximal UE (Fig. 1G,H, arrowheads). By contrast, Lef1 was only expressed in distal GT mesenchyme (see Fig. S1F in the supplementary material). To identify WNT ligands that may activate canonical WNT signaling in the GT, we performed RT-PCR using RNA extracted from E11.5, E12.5 and E14.5 GTs and ³⁵S-in situ hybridization in E12.5 GTs. Several WNT ligands,including Wnt2, Wnt3, Wnt4, Wnt5a, Wnt9b and Wnt11 were expressed in the developing GT, with Wnt5a and Wnt9bexpression detected in the dUE (see Table S1 and Fig. S1A-E in the supplementary material). Together, these data demonstrate that WNT signaling is activated during GT development.

To analyze the function of β-catenin in all three tissue layers during GT development, we used transgenic Cre lines to either conditionally remove or activate β-catenin. We first analyzed tissue specificity of the Cre lines by crossing them with Rosa26-lacZ reporter mice (R26R)(Soriano, 1999). At E10.5, Shh^cre/Gfp;R26R embryos showed Cre-mediated recombination exclusively in the cloacal endoderm(Fig. 1J), whereas Msx2-Cre and Dermo1-Cre lines conferred Cre activity in the ectodermal epithelium and mesodermal mesenchyme,respectively (Fig. 1K,L). The tissue-specific Cre expression in all three lines persisted throughout early GT development (data not shown).

As strong TOPGAL activity was detected in the dUE, we first examined the function of β-catenin in GT endoderm. Shh^cre/Gfp mice were crossed to either β-Cat^c/c(Brault et al., 2001) orβ -Cat^loxEx3(Harada et al., 1999) mice to generate endodermal loss- or gain-of-function (LOF or GOF) embryos. At E10.5,no morphological difference in the cloaca region was observed between wild-type and LOF embryos. At E12.5, scanning electron microscopy (SEM)revealed a cone-shaped GT with a centered urethral seam on the ventral side in wild-type embryos (Fig. 2A). By contrast, Shh^cre/Gfp;β-cat^c/c(LOF) embryos exhibited a severe outgrowth defect, in which GT failed to form and, instead, a crater-like structure was present(Fig. 2B). Both male and female mutants were equally affected. At E18.5, only a small remnant was detected in the presumptive genital region (data not shown). Immunofluorescence confirmed the complete removal of β-catenin protein from the dUE(Fig. 2E). Becauseβ-catenin is required in the limb ectoderm to establish the AER and to regulate Fgf8 expression (Barrow et al., 2003), we reasoned that a similar mechanism might also apply to GT development. This hypothesis was supported by the colocalization of TOPGAL activity with Fgf8 expression in wild-type dUE at both E10.5 and E12.5 (see Fig. S2A-D in the supplementary material). Consistent with this hypothesis, Fgf8 was never detected in the distal cloacal endoderm of Shh^cre/Gfp;β-cat^c/cembryos from E10.5 onwards (Fig. 2H, data not shown). Consequently, expression of Bmp4, a downstream target of Fgf8(Haraguchi et al., 2000), was markedly reduced in mutant GT mesenchyme(Fig. 2K). Previous reports demonstrated that disrupting AER by either physical or genetic means caused increased cell death and decreased cellular proliferation(Barrow et al., 2003; Dudley et al., 2002). Similarly, increased apoptosis in both endoderm and mesenchyme of E10.5 Shh^cre/Gfp;β-Cat^c/c GTs(Fig. 2M,N,O; 2.48±0.6%in controls versus 10.9±2.49% in mutants, n=10, P<0.0001) was revealed by TUNEL assay. We also detected a decrease in cell proliferation, evidenced by a twofold reduction in phospho histone-H3(PHH3)-positive cells in E10.5 mutant endoderm(Fig. 2P,Q,R; 6.84±0.99%in controls versus 3.03±0.6% in mutants, n=10, P<0.0001). By contrast, Hoxa13 and Fgf10 were properly expressed in the LOF mutants, which provides evidence against a global gene expression change as a result of reduced cell numbers (data not shown). Altogether, the loss of Fgf8 induction, the increased cell death and the decreased proliferation are all consistent with a failure of dUE establishment in the LOF mutant embryos, indicating an obligatory role for WNT-β-catenin in establishing the GT signaling center.

To further examine the role of WNT signaling in dUE establishment, we analyzed GT development in endodermal GOF embryos. GTs of Shh^cre/Gfp;β-cat^loxEx3 embryos were much larger than those of wild types at E12.5, and immunostaining showed accumulation of β-catenin in the UE(Fig. 2C,F). In situ hybridization revealed that Fgf8 was overexpressed in the GOF cloacal endoderm at E10.5 (Fig. 2I). Combined with LOF studies, these results clearly demonstrate that during GT initiation, WNT-β-catenin signaling is both necessary and sufficient to activate Fgf8 expression in the endodermal cloaca. To test whether this genetic regulation between WNT-β-catenin and Fgf8 is conserved in the limb bud, we generated limb ectodermal β-catenin GOF embryos by crossing Msx2-Cre withβ -Cat^loxEx3 mice. At E10.5, R26R reporter assays showed sporadic Cre activity in the inter-limb ectoderm in addition to in the dorsal and ventral ectoderm and the AER of the limb bud, as previously reported (Barrow et al., 2003)(Fig. 2V). Intriguingly,ectopic Fgf8 expression was detected in similar sporadic areas in the inter-limb ectoderm of Msx2-Cre;β-Cat^loxEx3 embryos(Fig. 2U), likely corresponding to Cre-mediated β-catenin activation in those cells. Furthermore, ectopic outgrowth was observed in the inter-limb ectoderm of Msx2-Cre;β-Cat^loxEx3 embryos at E12.5(Fig. 2X, arrows). These results indicate that activation of Fgf8 by WNT-β-catenin signaling is conserved between the limb ectoderm and dUE, and that this genetic regulation is essential for directing appendage outgrowth.

The early phenotype in Shh^cre/Gfp;β-Cat^c/c mice prevented us from studying the function of β-catenin during GT outgrowth and urethra formation. To circumvent this limitation, we employed a tamoxifen(Tm)-inducible Shh^Cre/esr line(Harfe et al., 2004). In this experiment, Cre-mediated recombination can be detected as early as 12 hours after Tm treatment and robust recombination was evident in the UE 24 hours after Tm treatment (data not shown). We generated Shh^Cre/esr;β-Cat^c/c and Shh^Cre/esr;β-Cat^loxEx3 embryos for LOF and GOF studies, respectively. Cre activity was induced at E9.5, E10.5 or E11.5 for LOF studies and at E10.5 for GOF studies. Embryos were collected at E12.5 for molecular analysis and at E14.5 for histological analysis. To confirm that WNT-β-catenin signaling was perturbed in the mutants, we examined the expression of Tcf1, which is a transcriptional target of WNT signaling (Tu et al.,2007). As expected, Tcf1 expression was markedly reduced in the LOF dUE (Fig. 3F′)and was activated in the entire UE of GOF GT at E12.5(Fig. 3F″). At E14.5, SEM revealed that Shh^Cre/esr;β-Cat^c/cGTs showed phenotypes with graded severity correlated with the timing of Cre induction (Fig. 3A-D). Specifically, earlier Tm treatments led to reduced distal growth(Fig. 3B-D, arrows) and larger proximal openings (Fig. 3B-D,arrowheads). By contrast, Shh^Cre/esr;β-Cat^loxEx3 GT displayed excessive distal growth (Fig. 3E, arrow) with no proximal urethral opening(Fig. 3E, arrowhead). These phenotypes suggest a role for β-catenin in both GT outgrowth and urethra formation.

To test whether the reduced distal growth in LOF GTs resulted from decreased FGF8 signaling, we examined Fgf8 expression in E12.5 Shh^Cre/esr;β-Cat^c/c mutants treated with Tm at different times. Fgf8 expression in the dUE was moderately reduced in E11.5-treated LOF embryos, dramatically reduced in E10.5-treated LOF embryos and completely absent in E9.5-treated LOF embryos(Fig. 3B′-D′). These results therefore indicate that the defective outgrowth in these LOF mutants is associated with altered FGF8 signaling. Thus, in addition to its function in Fgf8 induction, WNT-β-catenin signaling is also required to maintain Fgf8 expression during GT outgrowth. To investigate the function of endodermal WNT-β-catenin signaling in regulating mesenchymal gene expression, we examined the expression of mesenchymal genes in E10.5 Tm-treated wild-type and LOF embryos. Strong expression of Msx2, Lef1 and Wnt5a was detected in the distal mesenchyme surrounding the wild-type dUE(Fig. 3H-J), but their expression domains were reduced in LOF GTs(Fig. 3H′-J′). By contrast, in the GOF GT, Fgf8 expression was not only elevated in the distal UE but also ectopically induced in the proximal UE, indicating that at this stage ectopic WNT-β-catenin signaling can still activate Fgf8 expression in UE (Fig. 3E′). Consistently, Msx2, Lef1 and Wnt5aexpression were ectopically expanded to more proximal mesenchyme(Fig. 3H″-J″). By contrast, contrary to previous results from an ex vivo study(Haraguchi et al., 2000), Hoxa13 and Hoxd13 were not significantly altered, despite dramatic changes in Fgf8 expression(Fig. 3K-L″). Thus,endodermal WNT-β-catenin signaling controls the expression of Msx2,Lef1 and Wnt5a in the distal mesenchyme but not expression of the HOX genes.

Next, we examined urethra formation in LOF mutants. In E14.5 wild-type GTs,endodermal cells formed a solid urethral plate in the distal region(Fig. 4A,A′) and a urethral tube at the proximal end (Fig. 4D,D′). In LOF embryos, the distal urethral plate failed to form with endodermal cells displaying a tube-like structure(Fig. 4B,B′), while the proximal urethra was open (Fig. 4E,E′). This phenotype may result from either defective epithelial differentiation or a reduction in cell proliferation. We first assessed epithelial differentiation by examining the expression of K14 and p63, markers for progenitor cells in stratified epithelium. The expression of both markers was maintained in the urethra, suggesting the progenitor cells are still present in the mutant (Fig. 4B‴,E‴, data not shown). By contrast, PHH3 immunostaining at E12.5 showed reduced proliferation in Shh^Cre/esr;β-Cat^c/c urethra(Fig. 5A′;6.14±1.03% in controls versus 2.08±0.5% in mutants, n=10, P<0.001). We did not observe any change in apoptosis in the UE (see Fig. S3D′ in the supplementary material). The lack of cellular proliferation in the LOF urethra could result from the loss of Shh and/or Fgf8 signaling, because these genes have been directly associated with proliferation in the GT(Haraguchi et al., 2001; Haraguchi et al., 2000; Suzuki et al., 2003). Consistently, both Fgf8 (Fig. 3G′) and Shh were markedly downregulated in the mutant urethra (Fig. 5B′,see also Fig. S3A′ in the supplementary material). As expected, Ptch1, a transcriptional target of HH signaling was also reduced in the mesenchyme (Fig. S3B′ in the supplementary material). These results indicate that WNT-β-catenin signaling is required to maintain FGF8 and SHH signaling during urethra formation and to promote cell proliferation, but is not required for maintaining the progenitor cell population.

Shh^Cre/esr;β-Cat^loxEx3 (GOF)GT, however, exhibited adisorganized urethral plate in the distal region(Fig. 4C,C′) and severe excessive endodermal growth confined to the proximal end(Fig. 4F,F′). This region-specific phenotype correlated with an increase in PHH3 staining specifically at the proximal end (Fig. 5A″). The molecular basis for the differential response in proliferation between the distal and proximal endoderm is not clear. However,we noted that Shh was downregulated in the distal but not the proximal UE (Fig. 5B″,see also Fig. S3A″, arrows and arrowheads, respectively). Correspondingly, Ptch1 was reduced in the distal but not the proximal mesenchyme (see Fig. S3B″, arrows and arrowheads). In addition, we found that Bmp4 was ectopically expressed in the distal UE of Shh^Cre/esr;β-Cat^loxEx3 embryos where Shh expression was reduced (compare Fig. 5B″,C″, arrows and arrowheads). Intriguingly, this ectopic Bmp4 activation in the distal UE appeared to require WNT-β-catenin signaling, because it was not observed in LOF urethra where Shh was also repressed(Fig. 5C′). Consistent with Bmp4 expression, phosphorylated-Smad1/5/8 was strongly upregulated in the GOF UE (see Fig. S3C″ in the supplementary material),but was reduced in LOF GT (Fig. S3C′). The different response to ectopic WNT signaling probably reflects intrinsic differences in gene regulation between the distal and proximal regions of the urethral epithelium.

Notably, both male and female endodermal mutants (both LOF and GOF) were equally affected. They either do not reach the stage of sexual differentiation(shh^Cre/esr lines) or have an early developmental arrest in GT morphogenesis that does not allow normal sexual differentiation to occur(Shh^Cre/gfp lines). Thus, these phenotypes reflect a requirement ofβ-catenin during the androgen-independent phase of GT development.

To examine the role of ectodermal β-catenin in GT development, we removed β-catenin from the ventral ectoderm using Msx2-Cre. Msx2-Cre;β-Cat^c/c embryos exhibited a severe GT phenotype. SEM showed that, at E12.5, the urethral seam was evident in wild-type GTs (Fig. 6A,arrows), but was not present in Msx2-Cre;β-Cat^c/c GTs(Fig. 6B). Mutant GTs developed a large proximal opening at E13.5, and distal GT bifurcated at E14.5(arrowheads in Fig. 6D,F). At E16.5, mutant GTs were severely dysmorphic and preputial swellings failed to join on the ventral side (Fig. 6H, arrowhead). On E18, urethrae in wild-type males were canalized but they remained as an epithelial cord in females. By contrast, both mutant males and females showed complete open urethrae, evidenced by positiveβ-catenin staining in the outer-most epithelial lining (Fig. S6G,H,arrows, in the supplementary material). Thus, the phenotype reflected a role for ectodermal β-catenin during early GT patterning, but not a disrupted androgen response. To track the fate of ectodermal cells in the mutant GT, we performed lineage-tracing experiments by using the R26R reporter allele in combination with Msx2-Cre. In wild-type embryos, the ectodermally derived surface epithelium of the Msx2-Cre lineage covered the entire GT, except for a small proximal opening from E12.5 to E16.5(Fig. 6I,K,M,O). By contrast,mutant ectoderm was disrupted at the ventral midline as early as E12.5,evidenced by an unstained region (Fig. 6J, arrowheads) that continued to expand as the GT grew(Fig. 6L,N,P). At E16.5, the entire ventral side of the mutant GT was devoid of ectodermal cover(Fig. 6P). Immunostaining of E12.5 mutant GT sagittal sections revealed that β-catenin-positive endodermal cells were not covered by β-catenin-negative ectodermal cells(see Fig. S4D, arrowheads, in the supplementary material). Consistently, Shh-expressing UE was exposed(Fig. 6Q,R). At E13.5, the mutant developed an open urethra, evidenced by histology, lineage-tracing and Shh expression (Fig. 6S-X). The defects were unlikely to be caused by a disruption in WNT signaling because TOPGAL-positive cells were still present in the urethral epithelium of Msx2-Cre;β-Cat^c/c GT (Fig. S4F in the supplementary material). Moreover, the expression of Tcf1 and Fgf8 remained largely unchanged, except for a distal shift in expression domain; the shift was probably secondary to an overall structural change in these mutants (see Fig. S4H in the supplementary material, data not shown).

Conversely, cell-cell adhesion among ectodermal cells, and between ectodermal and endodermal cells, appeared to be compromised in the Msx2-Cre;β-Cat^c/c mutants. Toluidine Blue-stained plastic sections and transmission electron microscopy (TEM)analyses revealed that the mutant ectoderm was much thinner than that of the control at E10.5 (Fig. 7A-D). At E12.5, nuclei of the ectodermal cells in the mutants were elongated,indicative of a stretched epithelium (Fig. 7G,H). Both plastic section and TEM showed that contact between the ectoderm and the endoderm was disrupted in the mutants(Fig. 7B,D, asterisks). Interestingly, the apparent deficit in epithelial integrity in the mutant occurred despite the seemingly normal distribution of α-catenin and E-cadherin at the cell membrane (data not shown, Fig. 7M,N). We also noted that plakoglobin expression was elevated in the ectoderm(Fig. 7L, arrows), which might partially compensate for the loss of β-catenin in adherens junctions. This upregulation was also confirmed by real-time PCR analysis (data not shown). To assess mutant epithelial differentiation, we examined the expression of p63 and K14. In E10.5 and E12.5 embryos, normal p63 expression was detected in the ventral ectoderm and the endodermal urethra of both wild-type and mutant GTs (data not shown, Fig. 7O,P). By contrast, K14 was not expressed in the genital ectoderm until E12.5 in wild-type GTs, and its expression was absent in the ventral ectoderm of mutant GTs(Fig. 7Q,R). This loss of K14 expression was not specific to the genital epithelium but was also observed in other regions of ectoderm where β-catenin was deleted (data not shown). Together, these results indicate that ectodermal β-catenin is required to maintain the integrity of the genital epithelium.

Unlike the ecto- and endodermal Cre lines, mesenchymal Dermo1-Cre can only achieve patchy β-catenin deletion despite global recombination at the R26 locus in the GT mesenchyme (see Fig. S5H in the supplementary material). Nevertheless, SEM analysis showed that Dermo1-Cre; β-Cat^c/c GTs were much smaller and were severely dysmorphic (Fig. S5B,D,F in the supplementary material). PHH3 staining demonstrated a reduction in cellular proliferation in the mutant mesenchyme (Fig. S5J,L in the supplementary material). Mitotic index was calculated in eight different mutants and seven wild types by counting PHH3-positive cells in a 0.03-mm² region. A more than twofold reduction was detected in the mutant (5.29±0.95% in controls versus 2.24±0.53% in mutants, P<0.0001). Consistently, mesenchymal expression of cyclin D1, a cell cycle regulator and a direct WNT target, was also downregulated (see Fig. S5N in the supplementary material). These results indicate a role for β-catenin in GT mesenchyme in promoting cell proliferation. However, activation of β-catenin by this Cre line resulted in early lethality, which prevented us from further analyzing its function in GT mesenchyme.

Most studies on genitalia development have focused on the role of androgen signaling, while much less is known about the genetic program governing early androgen-independent GT development. With the advent of Cre/LoxP technology,we have now investigated the function of β-catenin in GT development in a tissue-specific manner, and have demonstrated distinct functions forβ-catenin in each tissue layer.

Our data showed that the timing of β-catenin removal from the endodermal urethra correlated with the severity of GT outgrowth defects. Deletion around GT initiation completely abrogated Fgf8 induction in the dUE and subsequent GT outgrowth, whereas later removal resulted in reduced Fgf8 expression and an underdeveloped GT. The graded GT phenotypes are similar to earlier findings in the limb, in which AER removal at successive time points caused limb truncations at increasingly distal positions (Saunders, 1948),supporting a conserved function of dUE and AER in directing appendage outgrowth (Cohn, 2004; Minelli, 2002; Yamada et al., 2006). Although it has long been recognized that the distal signaling centers are essential for appendage outgrowth, less is known about how they are initially established and restricted to a specific region within a seemingly homogeneous epithelium. Our data have demonstrated that the activation of WNT-β-catenin signaling is necessary and sufficient to activate Fgf8 expression both in early cloacal endoderm and in later UE during GT development. Similarly, when we activated WNT-β-catenin signaling in the inter-limb ectoderm and non-AER limb ectoderm, Fgf8 was ectopically expressed and, as a result, ectopic outgrowth was observed. These data demonstrated that, within a developmental window, the ectoderm and cloacal endoderm are competent to respond to activated WNT-β-catenin signaling and induce Fgf8 expression. The fact that ectopic WNT-β-catenin signaling is sufficient to initiate Fgf8expression and outgrowth suggests that restricting WNT-β-catenin activity to a precise location represents a crucial step in determining the position and physical dimensions of the AER and dUE. Unfortunately, what mechanism restricts WNT-β-catenin activity to the signaling center remains largely unknown.

Epithelial-mesenchymal interactions play crucial roles during organogenesis(Hogan, 1999), including external genitalia development (Kurzrock et al., 1999b; Murakami and Mizuno, 1986). Once established, the dUE regulates mesenchymal gene expression and directs outgrowth. In this study, we identified a set of regulatory genes, including Msx2, Lef1, Bmp4 and Wnt5a,whose expression depends on dUE signaling. These data provide in vivo evidence for the function of dUE in orchestrating and maintaining mesenchymal gene expression.

Our data also demonstrated a requirement for WNT-β-catenin signaling in the growth and patterning of endodermal urethra, as LOF mutant GTs showed an open urethra accompanied by reduced cellular proliferation. We specifically focused on analyzing the expression of two known important regulators of cell proliferation and apoptosis, Shh and Bmp4(Haraguchi et al., 2001; Perriton et al., 2002; Suzuki et al., 2003). The genetic hierarchy for Wnt, Shh and Bmp4 has not been established in GT development. Our data suggest that Bmp4 acts genetically downstream of WNT-β-catenin in dUE in a cell-autonomous manner and in GT mesenchyme in a non-cell-autonomous manner, possibly mediated by Fgf8, as the application of FGF8-soaked beads can stimulate Bmp4 expression in GT mesenchyme(Haraguchi et al., 2000). Conversely, Shh can repress Bmp4 expression in dUE,evidenced by the ectopic Bmp4 induction in Shh^-/-GT (Haraguchi et al., 2001). In our mutants, Shh downregulation also correlates with ectopic Bmp4 activation in dUE, but only in the presence of WNT-β-catenin and FGF8 signaling. As Fgf8 induction occurs in Shh^-/- GT and Shh expression is downregulated inβ-catenin LOF dUE, these data suggest that β-catenin is at the top of a genetic hierarchy regulating Fgf8, Shh and Bmp4expression (Fig. 8). However,we obtained an unexpected result in which Shh exhibited a bimodal response to activated WNT-β-catenin signaling in GOF mutants(Fig. 5B″). One possibility is that ectopic Bmp4 expression in the distal urethral cells is responsible for Shh repression in this region. In support of this notion, Bmp4 and Shh have been shown to repress the transcription of one another in other developmental systems(Monsoro-Burq and Le Douarin,2001; Zhang et al.,2000). Alternatively, ectopic Bmp4 expression might be secondary to Shh downregulation, and, in this case, it is not clear what mediates the bimodal response of Shh. Taken together, we propose a GT signaling pathway (summarized in Fig. 8) in which WNT-β-catenin signaling regulates both Fgf8 and Shh expression, and Shh in turn inhibits Bmp4. Thus, it appears that the balance between the positive regulators of cell proliferation, Fgf8 and Shh, and the negative regulator Bmp4 controls cellular proliferation in the urethra to maintain homeostasis. In addition to Fgf8 and Shh,Fgfr2 is also required for urethral cell proliferation(Petiot et al., 2005). Fgfr2 expression was visibly reduced in both β-catenin LOF and GOF urethrae (see Fig. S3E′,E″ in the supplementary material), as evidenced by in situ analysis, although real-time PCR confirmed such a reduction only in LOF UE. The urethra phenotype in the β-catenin LOF mutant was similar to that observed in Fgfr2IIIb^-/-mutants (Petiot et al., 2005). However, unlike in Fgfr2IIIb^-/- embryos, the expression of K14, a progenitor cell marker for squamous epithelium, was maintained. Thus,the defects observed in our LOF GT were unlikely to be caused by reduced FGFR2 signaling. The interaction between the WNT-β-catenin and FGFR2 signaling pathways needs further investigation.

Canonical WNT activity was not detected in the ventral ectoderm. Consistently, WNT signaling in the ectodermal β-catenin LOF GT does not seem to be affected, as evidenced by the presence of TOPGAL-positive cells and normal Fgf8 and Tcf1 expression. Nonetheless, the removal of ectodermal β-catenin still resulted in severe GT malformations. Tissue lineage analysis revealed an ectodermal breakdown at the ventral midline and subsequent urethral exposure, which implied that the ability ofβ-catenin^-/- ectoderm to contain the growing GT was compromised. Consistently K14, an intermediate filament protein that gives tensile strength to epithelium, is not expressed in the mutant ectoderm, and the fact that mutant ectoderm shows signs of stretching suggest that a lack of tensile support could directly contribute to the phenotype. Inβ-catenin-deficient ectoderm, we observed abnormal epithelial morphology and disrupted ectodermal-endodermal connection during initial GT outgrowth. Therefore, although we could not formally exclude disrupted WNT signaling as a cause of the mutant phenotype, it is more likely that weakened cell-cell adhesion was responsible for the ectodermal rupture. During early androgen-independent GT development, the endodermal urethra remains attached to the ventral surface epithelium(Perriton et al., 2002). As the GT protrudes from the ventral body wall, the ectoderm contacting the urethra would receive an increased physical force. A well-formed ectodermal-endodermal connection in this region may be required for structural support and to coordinate the growth of the two epithelia. This function forβ-catenin in cell adhesion has also been implicated in other developmental systems (Cattelino et al.,2003; Fu et al.,2006). Why could overexpression of plakoglobin in the mutant GT ectoderm not compensate for the function of β-catenin in cell adhesion?We propose that although plakoglobin can connect E-cadherin to α-catenin and maintain the basic structure of adherens junctions, such a junction might not provide the same adhesive force as the wild-type congeners. In plakoglobin knockout mice, it is known that β-catenin can only partially compensate its function in desmosomes (Bierkamp et al., 1999). These results suggest that both plakoglobin andβ-catenin have unique functions in cell adhesion that cannot be fully compensated.

In humans, hypospadias is defined as an abnormal urethral opening anywhere along the ventral side of the penis and scrotum. Both genetic and endocrine factors are implicated in the etiology of hypospadias. Androgen signaling is particularly important because hypospadias is commonly considered a male disease, although cleft clitoris does occur(Baskin and Ebbers, 2006). Both our endodermal and ectodermal β-catenin conditional-knockout mice exhibited a severe hypospadias phenotype in both sexes. These results raise the possibility that the disruption of WNT-β-catenin signaling and/or proper cell adhesion by either somatic mutation or endocrine disruption during GT development may lead to external genital abnormalities, including hypospadias in humans. Future research on genetic mechanisms and endocrine-genetic interaction could shed light on external genitalia development, as well as on the pathogenesis of hypospadias and other congenital malformations.

We thank Drs Yongjun Yin and David Ornitz for sharing Dermo1-Cre;β-Cat^c/c mice and for their invaluable comments on the manuscript. We also thank Dr David Beebe for Phospho-Smad antibody, Mike Veith for help on electron microscopy, Dr Gen Yamada for exchanging progress before publication and Dr Raphael Kopan for critical reading of the manuscript. This work was supported by National Institutes of Health Grants HD41492 and ES014482.