Nephric duct insertion is a crucial step in urinary tract maturation that is regulated by a Gata3-Raldh2-Ret molecular network in mice

The mature urinary tract consists of kidneys that filter and regulate blood homeostasis, the bladder and urethra that store and excrete urine, and the ureters that conduct urine from the kidneys to the bladder. The upper (kidney and ureter) and lower (bladder and urethra) urinary tract compartments form independently, but must be connected during development to generate a functional outflow tract. Kidneys and ureters are derived predominantly from intermediate mesoderm (Grobstein, 1956; Saxen and Sariola, 1987), whereas the bladder and urethra are derived from cloacal endoderm and peri-cloacal mesenchyme (Haraguchi et al., 2007; Seifert et al., 2008).

Formation of the nephric duct (ND) which occurs on embryonic day (E)9.5 in the mouse, is one of the earliest events in urinary tract formation, establishing a primary connection between the upper and lower urinary tract (Obara-Ishihara et al., 1999; Schultheiss, 2003; Vize et al., 1997). During this process the ND extends along the rostrocaudal axis of the embryo, turns towards the midline at the level of the limb buds and inserts into the cloacal epithelium. Kidney development starts at E10.5, when the ureteric bud sprouts from the caudal ND. At this stage the distal ureter is joined to the ND and does not have a direct connection with the bladder. The ureter-bladder connection is established during ureter maturation, which occurs between E11.5 and E14.5, when the newly formed ureter moves to and inserts into the bladder primordium. During sexual differentiation, which begins at later stages, NDs regress in females, or differentiate into the seminal vesicles, vas deferens and epididymis in males.

Mackie and Stephens first hypothesized that the position at which the ureteric bud forms on the ND is an important determinant regulating the final position at which the ureter will join the bladder (Mackie and Stephens, 1975; Tanagho, 1976). According to this model, ureteric buds emerging posterior to the normal sprouting site would insert in the primitive bladder too early, which can lead to ectopically positioned distal ureters and vesico-ureteral reflux. Conversely, ureters emerging at an abnormally anterior site will fail to join the bladder or will join at a later stage than normal, a defect that can result in defective ureter maturation and obstruction (Hains et al., 2010; Kume et al., 2000; Miyazaki et al., 2000; Pope et al., 1999). Recent evidence supports this model and further shows that ureter maturation depends on apoptotic elimination of the common nephric duct (CND), the most caudal segment of the ND (Batourina et al., 2005; Uetani et al., 2009). This process brings the ureter into contact with the bladder epithelium, where the caudal ureter undergoes apoptosis forming a new ureteral orifice in the bladder primordium (Batourina et al., 2002; Batourina et al., 2005; Mendelsohn, 2009; Murawski and Gupta, 2008; Uetani et al., 2009; Uetani and Bouchard, 2009). Subsequent growth of the bladder moves the ureteral orifice to its final position in the bladder. Once the ureter is inserted in the bladder, the smooth muscle surrounding its caudal end will associate with bladder muscle to form the trigone, the site of the valve mechanism that prevents back-flow of urine from the bladder to the kidney, which can cause severe renal damage (Viana et al., 2007).

Ret is a receptor tyrosine kinase that, in the urinary tract, is activated mainly through the Gdnf ligand and the GPI-linked co-receptor Gfrα1 (Cacalano et al., 1998; Enomoto et al., 1998; Moore et al., 1996; Pichel et al., 1996; Sainio et al., 1997; Sanchez et al., 1996). In humans, mutations in the RET gene lead to Hirschsprung's disease, cancer, and renal abnormalities (Jain, 2009; Takahashi, 2001). Recent studies of families with Hirschsprung's disease indicate that RET mutations can also give rise to urinary tract defects known collectively as CAKUT (congenital abnormalities of the kidney and urinary tract), including renal malformations, vesico-ureteral reflux, duplicated kidneys and ureteral obstruction (Pini Prato et al., 2009). In mice, Ret gene mutations lead to similar defects as in humans, including enteric nervous system defects, renal agenesis, hypoplasia and ectopic ureter termination (de Graaff et al., 2001; Ishizaka et al., 1989; Jain et al., 2006; Schuchardt et al., 1996; Skinner et al., 2008; Takahashi et al., 1985).

Ret-Gdnf signaling activates downstream pathways, including PLCγ, MAPK and PI3K/AKT, which regulate cell survival, proliferation, chemotaxis, migration and adhesion (Fisher et al., 2001; Tang et al., 2002; Tang et al., 1998; Wong et al., 2005). In the urinary tract, Ret is crucial for outgrowth of the primary ureteric bud from the ND, for branching morphogenesis of the ureteric bud within the kidney (de Graaff et al., 2001; Jain et al., 2006; Schuchardt et al., 1994; Schuchardt et al., 1996; Shakya et al., 2005) and for ureter maturation (Batourina et al., 2005; Shakya et al., 2005; Uetani et al., 2009). Ret signaling mediates ureter formation and ureter maturation, at least in part, by inducing migration and cellular rearrangements (Natarajan et al., 2002; Young et al., 2001).

Ret expression in ND cells depends on Gata3 and retinoic acid (RA) signaling, both of which are potent transcriptional regulators. Gata3 is required for the development of several organs, including the kidney, mammary gland and hematopoietic system (Chou et al., 2010; Patient and McGhee, 2002; Uetani and Bouchard, 2009). In mice, Gata3 inactivation leads to defects in nephric duct proliferation and guidance, and later prevents premature differentiation of nephric duct cells (Grote et al., 2008; Grote et al., 2006). In humans, GATA3 loss-of-function mutations are associated with the HDR (or Barakat) syndrome, which includes hypoparathyroidism, deafness, renal dysplasia, renal agenesis and vesico-ureteral reflux (Grote et al., 2008; Van Esch and Devriendt, 2001; Van Esch et al., 2000). RA, the active form of vitamin A, activates transcription by binding to and activating RA receptors, which are members of the nuclear receptor superfamily (for a review, see Rochette-Egly and Germain, 2009). Studies in humans with mutations in ALDH1A2 (hereafter referred to as RALDH2), the major RA-synthesizing enzyme, suggest that levels of maternal RA during late pregnancy may be important regulators of kidney size (El Kares et al., 2010; Rochette-Egly and Germain, 2009). In rodents, vitamin A deficiency or defective RA signaling can lead to urogenital malformations, including hydronephrosis, megaureter and ectopically terminating distal ureters (Batourina et al., 2002; Batourina et al., 2005; Duester, 2008; Kastner et al., 1997; Luo et al., 1996; Mendelsohn et al., 1994; Niederreither et al., 1999; Quadro et al., 2005; Wilson and Warkany, 1948).

We find here that Ret signaling is required for ND insertion prior to ureteric bud formation. This event establishes the primary connection between the upper and lower urinary tract and appears to depend on formation of cellular extensions commonly associated with migrating cells. A surprising observation from our studies is that delay or failure in ND insertion due to loss of Ret can cause ectopic ureters that join the bladder abnormally or remain fused to the NDs, malformations that have not been previously associated with defective ND morphogenesis. Importantly, mutations in the genes encoding the Ret regulators Gata3 or Raldh2 (Aldh1a2 – Mouse Genome Informatics), result in similar renal anomalies. Accordingly, these factors independently regulate Ret expression in the ND. Together our studies suggest that ND insertion is crucial for proper ureter positioning and delays in this process can result in hydronephrosis and other urinary tract abnormalities.

Gata3^ND−/− (Hoxb7^cre; Gata3^flox/flox), Pax2^GFP, Raldh2⁻, HoxB7GFP and Ret⁻ mice have been described previously (Grote et al., 2008; Grote et al., 2006; Niederreither et al., 1999; Pfeffer et al., 2002; Schuchardt et al., 1994; Srinivas et al., 1999). The HoxB7Cre (Yu et al., 2002) mice were purchased from Jackson laboratories.

To genotype Raldh2 mutant animals, primers RALDH2KO-S (5′-GCCTGACCTATTGCATCTCCCG-3′) and RALDH2KO-AS (5′-GCCATGTAGTGTATTGACCGATTCC-3′) were used for PCR amplification of the mutant allele (annealing temperature 61°C; extension time 45 seconds), and RALDH2WT-S (5′-GAAGCAGACAAGGTGTGTATTGC-3′) and RALDH2WT-AS (5′-CCTTGTCTATATCCACCTGTTA-3′) detected the presence of the wild-type allele (annealing temperature 56°C; extension time 45 seconds). For genotyping of Ret mutant embryos, oligos P3 (5′-AGAGGCTATTCGGCTATGACTG-3′) and P4 (5′-CCTGATCGACAAGACCGGCTTC-3′) were used to amplify the mutant allele (annealing temperature 59°C; extension time 45 seconds), and primers P6 ext (5′-CTCTGGGAGAAGGCGAGTTTGGAAAAGTTGTC-3′) and P7ext (5′-CTCTTCAGGAACACTGGCTACCATGTCCATG-3′) detected the wild-type allele (annealing temperature 57°C; extension time 45 seconds).

The experimental design used in these studies was approved by the institutional animal care and use committees at Columbia and McGill Universities.

For the generation of Raldh2^−/− embryos, retinoic acid (RA) supplementation was performed as described previously (Batourina et al., 2005).

Hematoxylin and Eosin staining was performed on 6 μm thick paraffin-embedded sections, according to standard procedures. In situ hybridization on tissue sections was performed as described previously (Pearse et al., 2007). The Gata3 (George et al., 1994), Ret (Pachnis et al., 1993) and Raldh2 (Batourina et al., 2001) in situ probes have been reported previously. Whole-mount in situ hybridization was performed as described previously (Henrique et al., 1995). Immunohistochemistry was carried out on 8 μm thick frozen sections using standard procedures (Bouchard et al., 2000). The following antibodies and detection reagents were used: rat-anti mouse E-cadherin antibody (Zymed Laboratories, 1:400), In Situ Cell Death Detection Kit (Roche, according to manufacturer's instructions), DAPI (50 μg/ml). Alexa Fluor 488-conjugated anti-rat antibody (Invitrogen, 1:200) was used for secondary detection. Slides were mounted with Slowfade Gold reagent (Invitrogen) and images were acquired on a Zeiss Axioplan2 microscope using Axiovision.

Ret is expressed in a number of sites in the developing urinary tract, including ND epithelia, the CND, the nascent ureteric bud and the ureteric bud tips within the kidney. The majority of Ret mutant embryos display renal agenesis due to impaired ureteric bud formation (Schuchardt et al., 1994; Schuchardt et al., 1996); however, as previously reported, the remaining embryos in which the ureter forms display hydronephrosis (Batourina et al., 2002). Further analysis of Ret-deficient embryos that succeeded in forming a ureter revealed two types of abnormalities at birth: the first, which occurred in ~30% of embryos examined (4/13), was an abnormal junction between the sex ducts (vas deferens) and the ureter, without any connection to the urethra or bladder (compare Fig. 1A,B). In humans, such a malformation is called ectopic vas deferens (Delakas et al., 1996; Sukumar et al., 2010) and is likely to cause obstruction at birth. In the second class of Ret mutant embryos, which was more common, the sex duct was inserted in the urethra, but the ureters joined the bladder abnormally. Analysis of control and Ret^−/− mice bred onto a Hoxb7-Gfp transgenic background (Srinivas et al., 1999) enabled us to clearly visualize the path of the intravesicular ureter. In controls, the ureter crossed through the detrusor muscle and submucosa at a 45° angle following a smooth path (Fig. 1C, white arrow). By contrast, Ret^−/− ureters made a sharp turn and entered the bladder at an ectopic position compared with that in controls (Fig. 1D, white arrow). Histological analysis revealed that, in addition to an abnormal angle of entry, mutant ureters followed a jagged trajectory through the bladder wall (compare Fig. 1E,F, black arrows).

To elucidate the cause of ectopic ureters and vas deferens in Ret mutants at birth, we performed a developmental analysis comparing lower urinary tract formation at different stages. In E11.5 control embryos, the caudal ND segment (the common nephric duct, CND) was inserted in the primitive bladder, and the ureteric bud sprout had formed (Fig. 1G). In Ret^−/− embryos, however, the CND was not joined to the primitive bladder (Fig. 1H). Analysis of embryos at later stages revealed that this phenotype persisted through E14.5 but had partially resolved in most embryos by E16.5 (Fig. 1F; data not shown). At this point in the majority of embryos, ureters and NDs were inserted; however, the position of the ureter was abnormal (Figs 1 and 3). Careful analysis of the cloaca and urogenital sinus in Ret mutant embryos failed to reveal any overt malformations (data not shown). Hence, the observation that the CND fails to join the primitive bladder in Ret mutant embryos points to an early defect in ND morphogenesis. To address this, we analyzed control and Ret knockout embryos at E9.5, when NDs normally fuse with the cloaca. Histological analysis of wild-type embryos revealed that, as expected, NDs joined the cloacal epithelium (Fig. 1I). However, in Ret mutant littermates, NDs extended close to the cloaca but failed to join (Fig. 1J). This phenotype was present bilaterally in all mutants examined (10/10) suggesting that ND defects are likely to underlie both ectopic vas deferens (Fig. 1B) and ectopic ureters (Fig. 1F). This is a surprising finding because ectopic ureters have not been associated previously with ND defects.

Our previous studies reveal that Ret expression in the urinary tract depends on Gata3 and retinoid signaling, raising the possibility that these pathways might also be important for the regulation of Ret at the stage of ND insertion. To test this, we generated Hoxb7Cre;Gata3^f/f (Gata3^ND−/−) mice, in which Gata3 is selectively inactivated in the ND and its derivatives. To investigate whether RA signaling also regulates ND insertion via Ret, we analyzed urinary tract formation in embryos lacking Raldh2, which encodes the major RA-synthesizing enzyme (Duester, 2008; Niederreither et al., 1999). Analysis of Gata3^ND−/− and Raldh2^−/− embryos at birth revealed hydroureter and hydronephrosis (Fig. 2A,B,E,F). To better visualize the cause of hydronephrosis and hydroureter, we performed histological analysis of Gata3^ND−/− and Raldh2^−/− urinary tracts at E18.5. In wild-type controls, ureters were patent and joined the bladder (Fig. 2C,I). However, in Gata3^ND−/− and Raldh2^−/− embryos, ureters ended blindly or joined the urethra-bladder region at an ectopic location (Fig. 2D,H,J; data not shown). We have not observed ND defects in our previous studies of knockouts lacking both Rara and Rarb2 (Rarab2 mutants); however, ND-related defects were present in other classes of RA mutants that display more severe developmental abnormalities, consistent with the findings reported here (Batourina et al., 2002; Kastner et al., 1997; Mendelsohn et al., 1994).

To determine whether genitourinary tract anomalies in Gata3^ND−/−, Raldh2^−/− and Ret^−/− embryos were due to a common cause, we compared their respective phenotypes at different developmental stages. Histological staining at E14.5 revealed that in controls the ND was fully connected to the urogenital sinus at the level of the prospective urethra (Fig. 3A), whereas the ureter was inserted in the bladder region (the black arrowhead in Fig. 3A denotes the position of the ureter, which is out of plane). At this stage, the NDs of all three mutants reached the level of the urogenital sinus but failed to join the epithelium and were separated by a layer of mesenchymal cells (Fig. 3B-D, white arrowheads). E-cadherin staining of vibratome sections of E12.5 urogenital tracts confirmed this result and further showed that the ureter remained connected to the ND in mutant embryos (Fig. 3F-H).

To determine whether, as in Ret mutant embryos, the primary defect leading to abnormal ureter insertion was delayed ND insertion, we performed whole-mount in situ hybridization analysis with Pax2, which is localized in both the ND and cloaca. These experiments revealed that the NDs of wild-type embryos had elongated to the posterior of the embryo and turned to fuse with the cloaca (Fig. 3I). By contrast, in all Gata3^ND−/−, Raldh2^−/− and Ret^−/− embryos examined, NDs extended successfully along the anteroposterior axis to a level below the limb bud but, in most cases, elongation was arrested and NDs failed to turn toward the midline and fuse with the cloaca (Fig. 3J-L). These results suggest a defect in ND elongation and guidance towards the cloaca as the primary cause of genitourinary tract defects in Gata3^ND−/−, Raldh2^−/− and Ret^−/− embryos.

Our previous studies suggest that expression of Ret in the CND and ureteric bud depends on Gata3 and RA signaling. To determine whether Ret expression in the caudal ND also depends on these signaling pathways, we analyzed Ret expression in Gata3 and Raldh2 mutant embryos at the stage of ND insertion. In situ hybridization analysis of wild-type embryos at E9.5 revealed robust Ret expression in the ND extending to the cloaca (Fig. 4A), which was greatly reduced in NDs of Raldh2^−/− and Gata3^ND−/− embryos (Fig. 4B-D). Raldh2 and Gata3 might regulate Ret in a linear fashion or in parallel. To determine whether RA regulation of Ret was via Gata3, we analyzed Raldh2 mutant and control embryos to determine whether Gata3 expression was altered. We observed no significant change in Gata3 expression in the ND of Raldh2 mutants compared with wild-type embryos (Fig. 4E,F), suggesting that RA signaling is unlikely to be an upstream regulator of Gata3 in ND cells. To determine whether Gata3 regulates Ret expression via the RA pathway, we compared expression of Raldh2 and RA receptors in wild-type and Gata3 mutant embryos. These experiments revealed little if any change in expression levels of any of these genes (Fig. 4G,H; data not shown), suggesting that Gata3 is unlikely to be an upstream regulator of RA signaling. Taken together, these experiments suggest that RA and Gata3 are important regulators of Ret expression in the ND, most likely acting in parallel pathways.

Ret signaling mediates a number of cellular behaviors including migration and chemoattraction, processes that have been shown to be important for ND formation in other species (Drawbridge et al., 2000). Analysis of whole-mount and sectioned Hoxb7-Gfp control embryos at E9.5 revealed the presence of thick extensions emanating from cells at the ND tip, which were more abundant in the caudal ND segment that has turned toward the midline (Fig. 5A,B). To investigate further whether these cellular processes might be important for ND morphogenesis, we analyzed the morphology of NDs in E9.5 Ret^−/−, Gata3^ND−/− and Raldh2^−/− embryos using GFP expression to visualize ND cells. These studies revealed few, if any, extensions in the posterior ND in the mutants compared with controls (Fig. 5C-F). The observation that ND tip cells display prominent cellular processes in wild-type embryos that are absent in all three mutant lines that also display delayed ND extension, suggests that these processes may normally be crucial for ND morphogenesis, and that their formation is regulated by the Gata3-Raldh2-Ret signaling network in ND cells.

The majority of the lower urinary tract anomalies identified to date originate from defects in the induction, growth or differentiation of the kidneys and ureters (Mackie and Stephens, 1975; Murawski and Gupta, 2008; Tanagho, 1976). ND insertion occurs prior to the onset of kidney development; hence, little attention has been paid to developmental defects at this early stage. Here, our studies suggest that a nephric duct extension or fusion defect may be a novel cause of urogenital system malformations. We show that in Gata3^ND−/−, Raldh2^−/− and Ret^−/− embryos, NDs fail to complete their caudal extension and, as a result, either fail to fuse with the cloacal epithelium or fuse later than normal. Consequently, ureters remain joined to the ND (ectopic vas-deferens), a likely cause of hydronephrosis at birth (Fig. 6).

Importantly, our results suggest that delayed ND insertion can also cause ectopic ureters and hydronephrosis by disrupting later stages of distal ureter maturation, the process by which apoptotic elimination of the CND and distal-most ureter positions the ureteral orifice in the bladder (Fig. 6). Examination of the migrating ND revealed the presence of cellular extensions at the ND tip that are regulated by the Gata3-Raldh2-Ret network and are likely to reflect the active process of ND guidance toward the cloaca. That abnormal ND insertion can lead to ectopically terminating ureters is in retrospect not surprising, as ND morphogenesis and ureter maturation occur at a stage when the bladder and urethra are undergoing dramatic growth and differentiation. Hence, developmental defects that occur prior to ureteric bud formation and that alter the timing of alignment or fusion of the ureters with the bladder, may be an important cause of ectopically terminating ureters and hydronephrosis (Fig. 6).

Our studies suggest that there are at least four phases in ND morphogenesis: epithelialization of the ND from intermediate mesoderm, extension along the rostrocaudal axis, migration to the midline and fusion with the cloaca. Pax2 and Pax8 mediate one of the earliest events in ND formation. Deletion of these two transcription factors results in failure at the stage of specification of intermediate mesoderm, and failure of the ND to form (Bouchard et al., 2002). Studies in the chick and amphibians suggest that epithelializaton or elongation of the NDs may be regulated by Bmp4 signals from the overlying ectoderm and laminin in extracellular matrix (Morris et al., 2003; Obara-Ishihara et al., 1999). In the present study, we find that NDs of Ret mutant embryos can extend successfully to the posterior of the embryo, but fail to turn toward the midline, suggesting that there is a guidance mechanism that normally regulates late stages of ND extension. We find a similar phenotype in Gata3^ND−/− and Raldh2^−/− and we show that Ret expression is downregulated in the NDs of these mutants. These observations suggest that Ret is normally important in ND cells for the final phase of insertion, whereas Gata3 and RA are important upstream regulators of Ret expression in the ND. As Ret is also expressed in the cloaca, it is possible that ND defects could be non-cell autonomous, caused by a defect in cloacal morphogenesis. However, a careful examination of Ret mutant embryos has not revealed detectable defects in the cloaca, bladder or urethra, arguing against this possibility. In addition, chimeric embryos accumulate wild-type but not Ret^−/− cells at the ND tip, suggesting a cell-autonomous function in ND tip cell guidance (Chi et al., 2009). Interestingly, we previously identified a ND guidance mechanism regulated by Gata3 at an earlier stage of extension. Germline inactivation of Gata3 indeed results in a misguidance of the ND that ultimately fuses with the surface ectoderm (Grote et al., 2006), raising the idea of an early Ret-independent and a late Ret-dependent guidance mechanism, both of which would require Gata3 function.

The cellular protrusions extending from posterior ND cells are reminiscent of processes that form in migrating neural crest cells (Druckenbrod and Epstein, 2007), in protrusions at the tips of neuronal growth cones (Lowery and Van Vactor, 2009) and in sprouting vasculature (Gerhardt, 2008). In these systems, attractive and repulsive signals are regulating the directional migration of cells toward their targets. In this case, ND cells would normally respond to attractive signals from the cloaca (such as Gdnf) or repulsive signals from the lateral part of the embryo originating from limb buds or other tissues. Ret-Gdnf signaling regulates migration and cellular process formation in a number of different contexts. For example, Gdnf from gut mesenchyme guides Ret-expressing enteric nervous system precursors to their destination (Druckenbrod and Epstein, 2007; Natarajan et al., 2002), a process that might also be important in ND formation.

Studies in amphibians and fish suggest that Ret signaling acts in a cell-autonomous manner in ND cells, controlling ND migration by a chemotactic mechanism that can be regulated by exogenous Gdnf (Drawbridge et al., 2000). Consistent with this, in mouse embryos, Gdnf is localized in mesenchyme at the caudal-most portion of the ND adjacent to Ret-expressing ND cells (Hellmich et al., 1996; Pachnis et al., 1993), and its inactivation leads to arrested ND extension (Pichel et al., 1996), a defect similar to the one we observe in Ret mutant embryos. Hence, Gdnf might act as a chemoattractant during ND insertion in mammals, guiding the trajectory of ND tip cells toward the cloaca via a Ret-dependent mechanism. Interestingly, experiments in Axolotls suggest that NDs follow a gradient of Gdnf, enabling signaling over several cell diameters that guides the connection of NDs to the cloaca (Drawbridge et al., 2000). Ret-Gdnf signaling also plays an important role in migrating MDCK cells and other cell types, where Ret controls cytoskeletal arrangements that underlie process formation and polarity (Tang et al., 1998; Wang et al., 2010). It is unclear at present how protrusions of ND tip cells regulate migration to the cloaca. Further experiments will be needed to address this question. Together, our experiments suggest that ND insertion depends on Ret, and that Ret expression in ND cells depends on RA and Gata3. This novel developmental pathway performs multiple functions during kidney development and our findings suggest that the same is true in the lower urinary tract. Here, we find that Gata3 and RA regulation of Ret is required for ND insertion that establishes a primary connection between the upper and lower urinary tract. These signaling pathways are likely to be crucial again during ureter maturation, when the ureter makes final connections with the bladder. To our knowledge, this is the first report directly linking ND insertion to ureter positioning, opening up a new dimension in our thinking about the causes of obstruction and vesicoureteral reflux (VUR).

We thank Dr Meinrad Busslinger (Vienna, Austria) for providing the Gata3^flox mice; Frank Costantini for the Ret mutant mice, Hoxb7-Gfp line and Ret probe; Pierre Chambon, Pascal Dollé, Karen Neiderreither and Greg Duester for the Raldh2 mutant mice; and Thomas Jessell for the Raldh2 in situ hybridization probes. We are also grateful to Odyssé Michos for comments on the manuscript, Fanghua Li for technical assistance and to the McGill GCRC Developmental Histology Facility for histology services. This project was supported by operating grants from the Canadian Institutes for Health Research (CIHR; MOP-84343 and MOP-84470) to M.B. and NIH (R01 DK061459-07 and R01 DK061459-07S1) to C.M. M.B. holds a Canada Research Chair in Developmental Genetics of the Urogenital System. Deposited in PMC for release after 12 months.