Gata3 Acts Downstream of β-Catenin Signaling to Prevent Ectopic Metanephric Kidney Induction

Metanephric kidney induction critically depends on mesenchymal–epithelial interactions in the caudal region of the nephric (or Wolffian) duct. Central to this process, GDNF secreted from the metanephric mesenchyme induces ureter budding by activating the Ret receptor expressed in the nephric duct epithelium. A failure to regulate this pathway is believed to be responsible for a large proportion of the developmental anomalies affecting the urogenital system. Here, we show that the nephric duct-specific inactivation of the transcription factor gene Gata3 leads to massive ectopic ureter budding. This results in a spectrum of urogenital malformations including kidney adysplasia, duplex systems, and hydroureter, as well as vas deferens hyperplasia and uterine agenesis. The variability of developmental defects is reminiscent of the congenital anomalies of the kidney and urinary tract (CAKUT) observed in human. We show that Gata3 inactivation causes premature nephric duct cell differentiation and loss of Ret receptor gene expression. These changes ultimately affect nephric duct epithelium homeostasis, leading to ectopic budding of interspersed cells still expressing the Ret receptor. Importantly, the formation of these ectopic buds requires both GDNF/Ret and Fgf signaling activities. We further identify Gata3 as a central mediator of β-catenin function in the nephric duct and demonstrate that the β-catenin/Gata3 pathway prevents premature cell differentiation independently of its role in regulating Ret expression. Together, these results establish a genetic cascade in which Gata3 acts downstream of β-catenin, but upstream of Ret, to prevent ectopic ureter budding and premature cell differentiation in the nephric duct.


Introduction
In human, urinary tract anomalies rank among the most common birth defects, with an estimated occurrence of 1 in 250 live births [1]. Most of these ontogenic malformations are classified as Congenital Anomalies of the Kidney and Urinary Tract (CAKUT) [2], which is a highly heterogenous condition frequently diagnosed in combination with genital tract anomalies [3]. The most relevant clinical manifestations include absent, dysplastic or obstructed renal systems in infants as well as infertility, pregnancy complications, hypertension and chronic renal failure in adults [4].
The development of the urogenital system (UGS) begins with the formation of the nephric duct (or Wolffian duct) [5,6]. This epithelial duct is a central UGS component among all vertebrates and serves as the primordium for the ureter, kidney collecting duct system and male genital tract [7]. Upon its induction in the intermediate mesoderm at embryonic day (E) 8.5 in the mouse, the nephric duct rapidly elongates caudally until it reaches the cloaca, a pouch from which the bladder and urethra later develop. At E10.5, the formation of the definitive (metanephric) kidney is initiated by sprouting of the ureteric bud from the nephric duct into the adjacent metanephric mesenchyme. The ureteric bud subsequently undergoes several branching cycles to form the collecting duct system, whereas the ureter tips induce nephron formation in the surrounding mesenchyme [8].
Ureteric bud outgrowth and positioning are among the most crucial steps of UGS development, since anomalies at the budding stage account for the majority of kidney and urinary tract developmental defects [9,10]. Extensive evidence has identified GDNF/Ret signaling as a central regulator of ureteric bud induction [11][12][13][14][15][16]. In this system, GDNF secretion in the metanephric mesenchyme activates the Ret receptor tyrosine kinase via its ligand binding GFRa1 co-receptor. In turn, Ret activation results in the initiation of intracellular signaling cascades, which mediate bud outgrowth, proliferation and subsequent ureter branching. Understandably, the activity of GDNF/Ret signaling is tightly regulated by various mechanisms to allow for the generation of a single ureter at the appropriate position [9]. In the mesenchyme surrounding the nephric duct, the forkhead transcription factor FoxCI and the Slit2/Robo2 ligandreceptor pair repress the rostral expression of GDNF [17,18], while Bmp4 antagonizes its activity [19]. At the budding site, Gremlin releases GDNF inhibition by Bmp4 [20], thereby allowing ureteric bud outgrowth. In the nephric duct epithelium, Sprouty1 protein function is crucial to negatively modulate Ret signaling levels [21]. Of notice, gene mutations in any of these regulators of GDNF/Ret signaling result in CAKUT-like phenotypes in mice. In support of the clinical importance of these genes, a growing number of them are found mutated in human developmental diseases affecting the UGS [22][23][24][25][26].
Gata3 is a transcription factor of the Gata Zn-finger family, which perform important functions during organogenesis [27]. In humans, GATA3 haploinsufficiency causes hypoparathyroidism, sensorineural deafness and renal anomalies (HDR) syndrome [28]. The urogenital defects of HDR patients closely resemble CAKUT in combination with genital tract anomalies and include renal aplasia, dysplasia, hypoplasia and vesicoureteral reflux [29,30]. Gene ablation studies in mice further revealed a critical role for Gata3 in the development of several tissues [31,32]. In the urogenital system, Gata3 is necessary for proliferation control and guidance of the nephric duct [33]. Accordingly, it is the only Gata factor expressed in this tissue prior to E12.5 [34].
Here we report the conditional inactivation of Gata3 specifically in the nephric duct, at a stage past the developmental defects observed in germline knockout embryos. These mice display multiple UGS malformations affecting the kidney, ureter and genital tracts. The detailed analysis of this phenotype reveals a genetic cascade whereby b-catenin promotes Gata3 expression in the nephric duct, which in turn activates Ret expression, maintains an undiffererentiated epithelial cell state and prevents the inappropriate response to signaling pathways promoting ureter budding.

Conditional Inactivation of Gata3 in the Urogenital System
The strong mesonephric phenotype observed in Gata3 2/2 embryos [33] precludes the study of Gata3 function later during urogenital system development. To investigate the role of Gata3 in ureteric bud formation, we first generated a Gata3 conditional loss of function allele ( Figure 1A,B). For this, the parental Gata3 allele (Gata3 ex4GFP ) [33] was crossed with a transgenic strain expressing FLPe in the germline [35] to excise the GFP-neo reporter cassette, thereby generating a conditional Gata3 allele with loxP sites flanking exon 4 (Gata3 flox ). These mice were subsequently bred with the More-Cre germline deleter strain [36] to generate the Gata3 D allele in which exon 4 is removed. The splicing from exon 3 to exon 5, expected from this modification, would generate a frameshift leading to protein truncation just upstream of the first zinc-finger DNA-binding domain. This gene mutation is therefore predicted to be null, like other previously reported Gata3 mutant alleles [32,33,37]. To delete Gata3 specifically in the maturing nephric duct and derived collecting duct system, we crossed Gata3 flox mice with the HoxB7-Cre mouse strain [38] to generate HoxB7-Cre; Gata3 flox/D and HoxB7-Cre; Gata3 flox/flox embryos. Both genotypes had the same phenotype and are subsequently referred to as Gata3 ND2/2 embryos. Gata3 flox/flox , Gata3 flox/+ and HoxB7-Cre; Gata3 flox/+ embryos failed to show any overt phenotype and were used as controls.

Gata3 ND2/2 Embryos Show Multiple Urogenital System Abnormalities
We initially characterized whole dissected urogenital systems (UGS) of Gata3 ND2/2 embryos at embryonic day (E)18.5. This gross analysis revealed a broad variety of malformations affecting the kidneys and genital tracts. Kidney defects included agenesis (15%), aplasia (20%) and severe dysplasia (65%) (Figure 2A-G). Moreover, one third of the Gata3 ND2/2 embryos displayed duplex kidneys (arrows in 2F). Histological analysis further revealed that the dysplastic kidneys were associated with hydronephrosis and hydroureter ( Figure 2H,I). As expected from embryos with such poor kidney endowment, no Gata3 ND2/2 pups could be recovered after birth.
In addition to these renal defects, over 80% of male mutant genital tracts displayed a massive enlargement of the vas deferens in comparison to control embryos (Figure 2A,B,J,K). In female embryos, Gata3 inactivation in the nephric duct resulted in a complete loss of uterus in over 85% of UGSs examined. The oviduct, however, was still present in these embryos ( Figure 2C,D,L,M). Hence the conditional inactivation of Gata3 leads to a broad spectrum of urogenital defects, including a high incidence of hydronephrotic kidneys and hydroureters.

Multiple Ectopic Buds Are Observed in Gata3 ND2/2 Embryos
The combination of kidney hydronephrosis, hydroureter and duplex kidneys seen in Gata3 ND2/2 embryos pointed to a primary defect at the level of ureter budding. In order to easily visualize nephric duct cells, we crossed the Rosa26 STOPLacZ allele [38] into control and Gata3 ND2/2 genetic backgrounds and stained the UGSs for b-Galactosidase activity.
During normal development, a localized swelling of the caudal portion of the nephric duct indicates the site of ureteric bud outgrowth at E10.5 ( Figure 3A). The bud quickly emerges and undergoes the first dichotomous branching event at E11.5, forming the T-stage kidney ( Figure 3C). Subsequently, the ureter lengthens and multiple branching cycles mark the development of the metanephric kidney ( Figure 3E). In Gata3 ND2/2 mutant embryos at E10.5, nephric duct swelling was sometimes observed ( Figure 3B), but normal ureteric bud formation failed in most embryos analyzed. In addition, the nephric duct looked more

Author Summary
In humans, kidney development originates during embryonic development by the sprouting of an epithelial budcalled the ureteric bud-from a simple epithelial structure-the nephric duct. The ureteric bud quickly grows and branches in a treelike fashion to form the kidney collecting duct system, while the emerging ureteric tips induce nephron differentiation. One of the most important steps during kidney development is the positioning of a single ureteric bud along the nephric duct, since mutations of genes implicated in this process lead to severe urogenital malformations. In this study, we identified the Gata3 protein as a crucial regulator of ureteric bud positioning by using genetically modified mice. Deleting the Gata3 gene in the mouse resulted in the development of multiple kidneys emerging at improper positions. We show that this defect was caused by a hypersensitivity of nephric duct cells in their response to local growth signals. Interestingly, this phenomenon was partly triggered by premature differentiation of a subset of nephric duct cells. Furthermore, we report a genetic pathway in which Wnt/bcatenin signaling activates the Gata3 gene, which in turn positively regulates the Ret gene. In summary, we introduce a mouse model system that can be used to study human birth defects affecting the urogenital system. sinuous in appearance ( Figure 3B). Strikingly, at E11.5, ectopic epithelial buds formed along the entire length of the nephric duct, with a preferential accumulation in the middle segments of the duct ( Figure 3D). By E12.5, the majority of the ectopic buds started to regress, while some buds expanded to form ectopic kidneys at a position far more rostral than the normal kidney induction site ( Figure 3F). Following ectopic bud regression, the male nephric duct started to enlarge ( Figure 4A   normal in these embryos ( Figure 4E-H and data not shown), it is possible that this elongation defect is simply caused by a physical obstruction by ectopic ureteric buds. Together, these results indicate that the urogenital defects observed in Gata3 ND2/2 embryos are largely caused by the emergence of ureteric buds at aberrant positions along the nephric duct.

The Formation of the Ectopic Buds Depends on GDNF/ Ret as well as Fgf Signaling
To determine the cause of ectopic ureter budding in Gata3 ND2/2 embryos, we first looked at Ret expression by in situ hybridization. Consistent with the previous analysis of Gata3 germline knockout embryos [33], most Gata3 ND2/2 nephric duct cells lost Ret expression. Curiously, however, the ectopic buds of Gata3 ND2/2 embryos remained positive for Ret ( Figure 5A,B). In situ hybridization with a cRNA probe against Gata3 exon 4 (which is excised in the Gata3 conditional allele) confirmed that the bud cells still expressed Gata3 ( Figure 5C,D), whereas nephric duct cells had already lost Gata3 expression ( Figure 5C-F). The remaining Ret expression therefore resulted from the incomplete action of Cre in the nephric duct at this stage. Interestingly, these ectopic buds consisted of Gata3+/Ret+ and Gata32/Ret2 cells segregated from each other ( Figure 5A-D). The continuous inactivation of Gata3 as well as the downregulation of Ret expression was confirmed in dysplastic Gata3 ND2/2 metanephric kidneys derived from such ectopic buds ( Figure 5G-J). The severity of kidney dysplasia in these embryos correlated well with the amount of remaining Ret expression ( Figure 5J insert, data not shown). As expected, these dysplastic kidneys additionally showed severely impaired nephron differentiation, as evidenced by the strong reduction in Fgf8-positive nephron precursors in comparison to control embryos ( Figure 5K,L).
Since Ret expression was specifically localized in the ectopic ureteric buds of Gata3 ND2/2 embryos ( Figure 5A,B), we hypothesized that Ret signaling might be causally involved in the budding process. To investigate whether the ectopic buds required GDNF/ Ret signaling, we performed organ culture experiments with dissected Gata3 ND2/2 and control UGSs starting at E10.5. After 42 hours in culture, their development progressed, forming either T-stage metanephric kidneys in control UGSs or multiple ectopic buds in Gata3 ND2/2 UGSs ( Figure 6A,B). Interestingly, addition of the Ret tyrosine kinase inhibitor SU5416 [39] efficiently suppressed ectopic budding in Gata3 ND2/2 UGSs in culture  ( Figure 6D) as well as primary ureteric bud formation in control UGSs ( Figure 6C,D). To further assess the role of GDNF/Ret signaling in ectopic bud formation, we treated cultured UGSs with recombinant GDNF [40]. This treatment, sufficient to induce ectopic budding in control UGS cultures ( Figure 6E), significantly increased the size of ectopic buds in Gata3-deficient cultures ( Figure 6F, compare with 6B). In addition, an antibody blocking GDNF activity successfully inhibited both primary and ectopic bud formation in control and Gata3 ND2/2 UGS cultures, respectively ( Figure G,H), thereby demonstrating the direct implication of GDNF/Ret signaling in ectopic ureteric bud formation. Interestingly, a 50% lower concentration of GDNF blocking antibody still inhibited primary bud induction but failed to inhibit the formation of ectopic buds (data not shown), suggesting the presence of additional factors involved in the budding process. One candidate is the Fgf signaling pathway, which has the potential to induce ectopic ureter budding [41,42]. Supplementing the culture medium with a soluble FgfR2-Fc chimeric protein [43] did not perturb primary ureteric bud outgrowth in control cultures ( Figure 6I). In striking contrast, however, it could efficiently suppress ectopic ureter budding in Gata3 ND2/2 cultures ( Figure 6J). In conclusion, these results demonstrate that the generation of the ectopic buds in Gata3 ND2/2 embryos is mediated through the combined action of GDNF and Fgf signaling activities.

Molecular Characterization of Ectopic Ureteric Bud Formation in Gata3 ND2/2 Embryos
To study ectopic bud formation at the molecular level, we looked at the expression of key components of the GDNF/Ret signaling cascade in E10.5 and E11.5 embryos. Already at E10.5, prior to budding, Ret expression was lost in a subset of nephric duct cells in Gata3 ND2/2 embryos ( Figure 7A,B). Staining of adjacent sections with a GDNF in situ probe revealed that the remaining Ret-expressing cells were concentrated near the mesenchymal source of GDNF ( Figure 7C,D). At E11.5, only the caudal ectopic buds that could maintain GDNF and Pax2 expression in the surrounding mesenchyme were able to develop into ectopic kidneys ( Figure 7E,F and data not shown). In some Gata3 ND2/2 embryos, the forming buds induced more Wnt11 expression than the presumptive ureteric bud induction site in control embryos, pointing to an elevated GDNF/Ret signaling response in those cells ( Figure 7G,H). By E12.5, the expression of Wnt11 and Ret was lost in the regressing ectopic buds of Gata3 ND2/2 embryos (data not shown). Surprisingly, assessing the expression of the GDNF/ Ret modulators Spry1, Spry2, Spry4, Slit2, Robo2, FoxcI, FoxcII, Bmp4 and Grem1, only revealed a modest downregulation of Spry1 and Slit2 expression in Gata3-mutant nephric duct cells at E10.5, while the other markers remained unaffected (Figure 7I-L and data not shown). In order to further support the involvement if Fgfs in ectopic bud formation, we additionally probed control and Gata3 ND2/2 embryos with the soluble Fgfr2-Fc fusion protein, which recognizes several Fgfs [44]. This experiment revealed strong and similar Fgf expression levels surrounding the nephric duct in both Gata3-mutant and control embryos at E11.5 ( Figure 7M,N). This protein localization was consistent with the expression of one of the known FgfR2 ligands, Fgf10 ( Figure 7O,P). Hence, the molecular marker analysis reveals no obvious mesenchymal defects, pointing to a primary defect in the response of the nephric duct to GDNF and Fgf signals.   The developmental defects of Gata3 ND2/2 embryos described above show striking similarities with the phenotypes resulting from the conditional inactivation of b-catenin (ctnnb1) in the nephric duct [45]. The fact that both animal models have the same spectrum of genital tract and kidney defects, prompted us to verify whether Gata3 and b-catenin act in the same genetic pathway. For this, we first performed in situ hybridization against Gata3 in E11.5 HoxB7-Cre; Ctnnb1 flox/2 embryos (Ctnnb1 ND2/2 ) and found a strong downregulation of Gata3 expression in the nephric duct in comparison to control embryos ( Figure 8A,B). Accordingly, in situ hybridizations with a Ret cRNA probe revealed a loss of Ret expression in Ctnnb1 ND2/2 embryos ( Figure 8C,D), which mimics the loss of Ret expression in Gata3 ND2/2 embryos ( Figure 5). Staining of adjacent tissue sections with in situ probes for the canonical Wnt target genes Axin2, Sp5 and Daple [46,47] confirmed the loss of b-catenin transcriptional response in Ctnnb1 ND2/2 embryos ( Figure 8E,F and data not shown). In order to clarify the genetic hierarchy between Gata3 and b-catenin, we next stained Gata3 ND2/2 embryos with antibodies against b-catenin and phospho-b-catenin. These experiments failed to show any modification of b-catenin expression levels or activity following Gata3 inactivation ( Figure 8G,H and data not shown). In support of this, the expression of the canonical Wnt-signaling target genes Axin2, Daple and Sp5 was unchanged in Gata3 ND2/2 embryos (Figure 8I-L and data not shown). From these data, we conclude that Gata3 acts genetically downstream of b-catenin but upstream of Ret in the nephric duct. To further characterize the molecular basis of the b-catenin-Gata3-Ret pathway, we first performed a bioinformatics analysis of the 50 kb genomic region upstream of the mouse and human Gata3 genes. Highly conserved sequences shared by the two species were further analyzed for putative binding sites for TCF/Lef. In total, 10 putative TCF/Lef binding sites, located in 5 of the 8 conserved regions, were identified ( Figure 9A). In this analysis, we additionally included a Gata3 urogenital enhancer located at 2110 kb [48], but failed to detect any conserved TCF/ Lef binding sites in this element. A similar bioinformatics analysis of the Ret regulatory region revealed putative Gata3 binding sites in 4 of the 11 conserved regions 50 kb upstream of the Ret-ATG.
( Figure 9B). The potential of b-catenin to regulate endogenous Gata3 expression was further evaluated in mouse IMCD3 collecting duct-derived cells. Using the GSK3b inhibitor BIO to stabilize the b-catenin protein [49], we observed a significant increase of Gata3 expression levels in those cells ( Figure 9C). In order to assess the activity of Gata3 on the Ret regulatory region, we took advantage of the fact that one of the Gata3 binding sites mapped to a region previously reported to drive reporter gene expression in the zebrafish pronephros ( Figure 9B, large asterisk) [50]. To test whether Gata3 acts on this site, we isolated the 1.2 kb conserved fragment and introduced a specific point mutation in the Gata3 binding site ( Figure 9B). The wild-type and mutated fragments cloned upstream of a b-Gal reporter construct were transfected in IMCD3 cells stably expressing Gata3. The inactivation of the Gata3 binding site led to a significant downregulation of b-Gal expression relative to the wild-type control ( Figure 9D), thereby suggesting that Gata3 may regulate Ret expression directly from this binding site. Together these data are consistent with a b-catenin-Gata3-Ret genetic cascade in the nephric duct.

Premature and Ret-Independent Differentiation of Gata3 Mutant Cells
Another aspect of b-catenin-loss in the nephric duct is a premature differentiation of the affected cells [45]. To test whether this phenotype was also present in Gata3 ND2/2 embryos, we stained mutant and control embryos with the differentiation markers DBA and Zo1+. In the developing metanephric kidney, DBA and Zo1+ are strongly expressed in the distal collecting duct and downregulated in Ret+ ureter tip cells [45,51]. At E11.5, DBA and Zo1+ expression were almost undetectable in the nephric duct of control embryos ( Figure 10A,B). However, Gata3-deficient cells upregulated both differentiation markers already at this stage, suggesting a premature acquisition of a collecting duct identity in these cells ( Figure 10C,D). To clarify whether this premature differentiation was a result of Gata3 inactivation only or secondary to the loss of Ret expression in Gata3-mutant cells, we next tested the expression of the two differentiation markers in Ret mutant embryos. Interestingly, no difference in DBA and Zo1+ expression levels could be observed between wild-type and Ret deficient embryos ( Figure 10A,B and E,F). To quantifiy the differentiation defect in Gata3 ND2/2 embryos, we counted the total amount of DBA-positive nephric duct cells in the three different genotypes. Strikingly, we observed a ten-fold increase in differentiated nephric duct cells in Gata3-deficient embryos when compared to control or Ret mutant embryos ( Figure 10G). From these data, we conclude that Gata3 acts independently of Ret to maintain a precursor state in the nephric duct epithelium, downstream of b-catenin.

Discussion
We previously reported the critical role played by Gata3 in proliferation control and nephric duct guidance in the pro/ mesonephros [33]. To circumvent this early renal phenotype and the embryonic lethality of Gata3 2/2 embryos at midgestation [31][32][33], we generated a conditional knockout allele of Gata3. In order to specifically address the later role of Gata3 in urogenital system morphogenesis, we inactivated Gata3 in the nephric duct using the HoxB7-Cre transgenic line (Gata3 ND2/2 ). This resulted in severe malformations of the urogenital system including kidney agenesis, aplasia, dysplasia, duplex systems, uterine agenesis and vas deferens hyperplasia. This spectrum of malformations overlaps with the urogenital phenotypes observed in HDR syndrome patients (heterozygous for GATA3) and is generally reminiscent of human congenital anomalies of the kidney and urinary tract (CAKUT). Interestingly, most of these phenotypes can be either directly or indirectly attributed to ectopic ureteric budding observed in Gata3 ND2/2 embryos. Furthermore, our results identify Gata3 as a critical mediator of b-catenin signaling, which regulates both cell differentiation and Ret expression in the nephric duct.

Regulation of Gata3 by b-Catenin
The UGS malformations observed in Gata3 ND2/2 embryos were strikingly similar to the ones reported recently for Ctnnb1 ND2/2 embryos [45]. Gene expression analyses in both genotypes revealed a genetic cascade whereby Gata3 acts downstream of b-catenin in the nephric duct to maintain Ret expression and prevent premature epithelial differentiation. These genetic interactions were further supported by promoter analyses and cell culture assays. Hence, our results identify Gata3 as a crucial mediator of b-catenin activity in the nephric duct. Using Ret 2/2 embryos, we could further determine that the premature differentiation phenotype observed in Ctnnb1 and Gata3 mutant embryos is not mediated by Ret, thereby establishing at least two distinct cellular functions regulated by the b-catenin/Gata3 pathway ( Figure 11A). The possibility remains that b-catenin also has a Gata3-independent effect on Ret expression. However, our data indicate that, if present, this effect is not sufficient for Ret expression in the absence of Gata3. The identification of b-catenin acting upstream of Gata3 also raises the question of the transcriptional control of Gata3 expression in the nephric duct. The only other regulators of Gata3 identified in this tissue are the transcription factors Pax2 and Pax8 [33]. It is thus possible that Pax2/8-mediated activation of Gata3 acts through the b-catenin pathway. Alternatively, Pax2/8 and b-catenin may act independently to regulate Gata3 expression in the nephric duct either together or as activation and maintenance factors, respectively. We favor the latter model based on the observation that the canonical Wnt signaling readouts Axin2 and Daple are not yet expressed in 18-somite stage embryos (E9.0; D.G., M.B. unpublished results), whereas Gata3 is already under the control of Pax genes at this stage [33]. The recent identification of a Gata3 kidney enhancer active in the nephric duct may help clarify some aspects of Gata3 regulation in this tissue [48].

Mechanisms of Ectopic Ureteric Bud Formation
Defects in ureteric budding account for most UGS defects in humans and mice. The underlying cause is typically a deregulation of GDNF/Ret signaling [9]. Accordingly, our data show a direct implication of GDNF/Ret signaling in the formation of Gata3 ND2/2 ectopic buds. Recombinant GDNF treatment indeed sustained ectopic bud growth in organ culture, while blocking the signaling pathway with a GDNF blocking antibody or a chemical inhibitor against Ret was sufficient to prevent ectopic ureter budding. We also demonstrated that the emerging buds consist of Ret+ cells, which respond to mesenchymal GDNF by upregulating Wnt11 expression. Importantly, however, we show that Ret expression was lost in Gata3-mutant nephric duct cells and that these Gata32/Ret2 cells failed to contribute to ectopic ureteric bud formation. This suggests that ectopic ureteric buds form as a consequence of Gata3 loss in neighboring nephric duct cells. It further implies that Ret-mediated signaling played a role in the segregation of Gata3+/Ret+ from Gata32/Ret2 cells. Interestingly, a biased contribution of Ret+ and Ret2 cells was also observed in the ureteric bud and metanephric kidney of embryos chimeric for the Ret gene [40]. However, the precise mechanism leading to Gata3+/Ret+ and Gata32/Ret2 cell segregation remains unclear.
Importantly, a number of observations argue against a simple GDNF/Ret-based budding mechanism. Among them are 1) the different sensitivity of ectopic buds towards GDNF-blocking antibody concentration and 2) the fact that ectopic buds emerge in all directions whereas GDNF is expressed in the inner intermediate mesoderm only. Together, these observations point to the existence of a parallel pathway to GDNF/Ret promoting ectopic ureter budding in Gata3 ND2/2 embryos. Using a soluble recombinant Fgf receptor 2, we identified Fgf signaling as a crucial component of this alternative pathway. In support of this, Fgfs are able to induce ectopic budding in organ culture [41,42]. Of notice, the RTK inhibitor (SU5416) we used to inhibit Ret signaling in culture was also reported to have an effect on FgfRs [39], which may have enhanced its activity against ectopic ureteric bud formation.
The genetic regulation of Ret by Gata3 corroborates our previous observations in germline Gata3 mutant embryos [33], but raises the intriguing question of the relationship between Gata3+/ Ret+ budding cells and their surrounding Gata32/Ret2 cells in ectopic ureteric bud formation. Remarkably, the expression of the GDNF/Ret signaling regulators, Foxc1, Foxc2, Robo2, Bmp4, Grem1 and GDNF was not significantly affected in Gata3 ND2/2 embryos. The modest dowregulation of Spry1 and Slit2 expression was additionally restricted to non-budding Gata3-mutant cells and might be secondary to the loss of Ret expression. Hence, none of the major known mechanisms of GDNF/Ret modulation are likely to cause ectopic ureter budding in Gata3 ND2/2 embryos. Instead, this suggests that the juxtaposition of nephric duct cells with different Gata3/Ret status is central to the ectopic budding phenotype. Hence, the simplest model to explain ectopic ureteric bud formation would be that the combined expression of GDNF and Fgf along the defective Gata3 ND2/2 nephric duct at E10.5 is  sufficient to initiate ectopic bud formation by wild-type nephric duct cells. At E11.5, the source of GDNF becomes restricted to the metanephric mesenchyme adjacent to the caudal region of the nephric duct. Thus, only the ectopic buds which are close enough to this mesenchymal source of GDNF and still harbor a sufficient number of Ret+/Gata3+ cells will be able to sustain the GDNF/ Ret/Wnt11 positive feed back loop [52] and continue to grow to form ectopic kidneys. At this stage, most of the rostral buds are devoid of local GDNF expression and will therefore regress.
Importantly, the trigger for the aberrant responses to GDNF and Fgf inductive signals necessarily results from the loss of Gata3 in the surrounding nephric duct cells ( Figure 11B). Ectopic budding may thus be a consequence of the premature differentiation and changes in adhesive properties observed in Gata3-mutant cells, as reported for Ctnnb1 ND2/2 embryos [45]. Alternatively, the loss of Gata3 may result in the activation of yet unidentified signals affecting the behavior of neighboring cells ( Figure 11B). By avoiding these aberrant cellular responses in wild-type embryos, Gata3 maintains epithelial homeostasis and thus prevents the formation of ectopic buds by GDNF and Fgf in nephric duct regions that normally express their respective receptor.

Gata3 and HDR Syndrome
In humans, heterozygous inactivation of the GATA3 gene is responsible for HDR syndrome [28,30]. As many as 35 of the 38 mutations identified to date result in a loss of GATA3 DNA binding activity, mostly as a consequence of complete gene deletion or truncation of GATA3 DNA-binding domains [53]. As in Gata3 ND2/2 embryos, this GATA3 haploinsufficiency in HDR patients leads to a variety of UGS defects such as kidney agenesis, dysplasia, hypoplasia, sometimes associated with female genital tract malformations [29,30]. This suggests that a critical threshold level of GATA3 is necessary to perform its normal function in the UGS and that cells or tissues expressing GATA3 below this level behave aberrantly. The Gata3 threshold is apparently reached at 50% gene dosage in human. Such a threshold, however, is not observed in Gata3 heterozygous mice [32]. Instead, UGS anomalies in mice are only observed at Gata3 expression levels below 50% of the wild-type levels [48], complicating the use of Gata3 mutant mice as disease models. The mosaic inactivation we observe in Gata3 ND2/2 embryos may, in fact, better mimic the threshold mechanism of HDR malformations whereby some cells respond normally to Gata3 while others behave aberrantly. In this respect, it is possible that the urogenital anomalies of HDR patients are caused by premature nephric duct progenitor cell differentiation and involve an aberrant response to GDNF and Fgf signaling. Hence, the Gata3 ND2/2 embryos may provide a suitable model to understand the molecular mechanisms and morphogenetic defects underlying HDR syndrome and other CAKUTrelated UGS diseases.

Organ Culture and b-Galactosidase Staining
Mouse urogenital ridges were micro-dissected in cold PBS supplemented with 1% FBS, 1 mM CaCl 2 and 0.5 mM MgCl 2 . The ridges were collected into 10% FBS/DMEM medium (Wisent) containing Penicillin/Streptomycin (Gibco) and L-Glutamin (Gibco), preincubated at 37uC in 5% CO 2 and kept under these conditions until all ridges were dissected. Subsequently, the urogenital ridges were cultured in the presence of the appropriate compound on 0.4 mm Transwell filters (Corning) at 37uC in 5% CO 2 for 42 hrs. The following compounds were used: rat recombinant GDNF (300 ng/ml, R&D systems) anti-GDNF blocking antibody (10 or 20 mg/ml, R&D systems), SU5416 (20 mM, Calbiochem), IgG (20 mg/ml), FgfR2-Fc fusion protein (20 mg/ml, R&D systems) or DMSO (0.01%, Fisher Scientific). b-Galactosidase activity was detected by X-gal staining of dissected or cultured urogenital ridges as described [68]. and HAM's F12 media (Wisent) supplemented with 10% fetal bovine serum. To induce the canonical Wnt-pathway, mIMCD3 cells were stimulated with 3 mM BIO ((29Z,39E)-6-Bromoindirubin-39-oxime, EMD) or 0.3% DMSO for 4 hours. RNA was extracted (Rneasy Mini kit, Qiagen) and reverse transcribed (Superscript III, Invitrogen) according to the manufacturer's instructions. Quantitative PCR was performed using iQ Sybr Green Supermix (BioRad) in a RealPlex 2 cycler (Eppendorf) using the following primers: Gata3 sense 59-CTCTGGAGGAGGAACGCTAA- 39 and antisense 59-TTTGCACTTTTTCGATTTGC-39, S16 sense 59-GTACAAGT-TACTGGAGCCTGTTTTG-39 and antisense 59-GCCTTTGA-GATGGACTGTCGGATGG-39. For the production of mIMCD3 cells, stably overexpressing Gata3, mouse Gata3 cDNA was cloned into pMSCV-HA3-IRES-GFP vector (kindly provided by Dr. Jerry Pelletier). The Gata3-pMSCV vector was then co-transfected with pVPack-GP and pVPack-VSV-G vectors (Stratagene) in HEK-293T cells, for virus production. The virus containing supernatant was harvested after 48 hrs by filtration and added to mIMCD3 cells. 48 hours post-infection, the cells were sorted according to GFP expression on an FACSAria-sorter (BD-Bioscience). Gata3 expression levels were assessed by western blotting and the medium level expressing cells were used for the subsequent b-Galactosidase reporter analysis. The conserved RET regulatory element was isolated from human RCC cells (ACHN) using published primers [50] and subcloned into the pGEM-T-easy vector (Promega). The point mutation in the Gata3 binding site was introduced via sitespecific mutagenesis using the overlap extension method. Subsequently, the wild-type and mutated element were cloned into the b-Galactosidase expression vector pTrap [68] using the unique SalI, SphI restriction sites. The pGAL4-SV40-Luc vector (kindly provided by Dr. Xiang-Jiao Yang), constitutively expressing firefly luciferase, was used as a transfection control. The Gata3 expressing mIMCD3 were transiently co-transfected with either the wild-type or mutated pTrap vector in combination with pGAL4-SV40-Luc using Lipofectamine 2000 (Invitrogen) according to manufacturer's instructions. The analysis was carried out 24 hours post-transfection with the DualLight Combined Reporter Gene Assay System (Applied Biosystems) according to manufacturer's instructions. All experiments were done in triplicates.

Bioinformatics
The sequence alignments were performed, using blast2seq (NCBI). The transcription factor binding sites were identified with Mac Vector.