Discs-large homolog 1 regulates smooth muscle orientation in the mouse ureter

Abstract

Discs-large homolog 1 (DLGH1) is a mouse ortholog of the Drosophila discs-large (DLG) tumor suppressor protein, a founding member of the PDZ and MAGUK protein families. DLG proteins play important roles in regulating cell proliferation, epithelial cell polarity, and synapse formation and function. Here, we generated a null allele of Dlgh1 and studied its role in urogenital development. Dlgh1^−/− mice developed severe urinary tract abnormalities, including congenital hydronephrosis, which is the leading cause of renal failure in infants and children. DLGH1 is expressed in the developing ureter; in its absence, the stromal cells that normally lie between the urothelial and smooth muscle layers were missing. Moreover, in ureteric smooth muscle, the circular smooth muscle cells were misaligned in a longitudinal orientation. These abnormalities in the ureter led to severely impaired ureteric peristalsis. Similar smooth muscle defects are observed frequently in patients with ureteropelvic junction obstruction, a common form of hydronephrosis. Our results suggest that (i) besides its well documented role in regulating epithelial polarity, Dlgh1 also regulates smooth muscle orientation, and (ii) human DLG1 mutations may contribute to hereditary forms of hydronephrosis.

Mouse kidney development begins on the 10th day of gestation when the metanephric mesenchyme induces an epithelial outgrowth from the Wolffian duct called the ureteric bud (1). The ureteric bud invades the metanephric mesenchyme and is induced by the mesenchyme to branch; in turn, the ureteric bud tips induce a subpopulation of the mesenchyme to condense and form an epithelial sphere, which becomes a vascularized and functional nephron over the course of a few days. These reciprocal inductive interactions, ureteric bud branching morphogenesis and mesenchyme to epithelium transitions, continue through the first week of postnatal development to form the definitive kidney. Each mouse kidney contains ≈10,000 nephrons derived from the metanephric mesenchyme that are connected to the collecting system derived from the ureteric bud epithelium. The primary stalk of the ureteric bud, which connects the developing kidney first to the Wolffian duct and later to the bladder, matures to become the ureter.

The function of the ureter is to actively propel and guide urine from the renal pelvis to the bladder through peristaltic movements. Any physical obstruction or functional impairment of the peristalsis machinery can lead to hydronephrosis (2, 3). Hydronephrosis is characterized by a distended kidney collecting system (the renal pelvis and ureter) resulting from improper outflow of urine to the bladder. Prenatal ultrasound detects hydronephrosis in 1 of 100 pregnancies, of which 20–30% fail to resolve and are clinically significant (4). Urine accumulation creates pressure that impairs kidney development and damages the renal parenchyma, which, in turn, can lead to renal failure in infants and children (5, 6).

Discs-large homolog 1 (DLGH1; also known as synapse-associated protein 97/SAP97) is a mouse ortholog of Drosophila discs-large (dlg), a founding member of the membrane-associated guanylate kinase (MAGUK) family of scaffolding proteins. DLG proteins contain multiple protein–protein-binding domains, including three eponymous postsynaptic density-95/discs-large/zonula occludens-1 (PDZ) domains, one Src homology domain-3, a protein 4.1 binding motif, and one guanylate kinase-like domain. Drosophila dlg is a tumor suppressor gene; mutations in dlg lead to loss of polarity and overproliferation in both imaginal disc epithelia and the nervous system (7–9).

There are seven dlg homologs in mammals; among them, Dlgh2 (PSD-93), Dlgh3 (NE-dlg), and Dlgh4 (PSD-95) are expressed almost exclusively in the nervous system, whereas Dlgh1 is the most widely expressed outside neuronal tissue (10). In the nervous system, DLGH4 (PSD-95) binds to and organizes ion channels and neurotransmitter receptors at synaptic junctions (10). In epithelial cells, DLGH1 is located at the membrane-cytoskeleton interface and is associated with E-cadherin, F-actin, and CASK (11, 12). Besides structural roles, DLGH1 also binds to APC and p85 to regulate signal transduction (13, 14).

Previously reported Dlgh1 gene-trap mutant mice (Dlgh1^Gt/Gt) express a truncated protein retaining the three PDZ domains linked to a β geo reporter. These mice exhibit growth retardation, craniofacial abnormalities, neonatal lethality, increased proliferation in the lens, and small kidneys associated with impaired ureteric bud branching and reduced nephron formation (15–17). Here, we report the generation and characterization of Dlgh1 null mice. In addition to the phenotypes described for the gene trap mutant, we found that Dlgh1^−/− mice exhibit highly penetrant hydronephrosis associated with a defect in ureteric smooth muscle orientation that dramatically impairs the efficiency of peristalsis. These results suggest a possible link between the orthologous DLG1 gene and congenital hydronephrosis in humans.

Results

Generation of a Dlgh1 Null Allele.

We generated a null allele of mouse Dlgh1 (Dlgh1⁻) by replacing exon 4 (the third coding exon) with a neo cassette [see supporting information (SI) Fig. 7A]. Proper targeting and transmission to offspring was confirmed by Southern blotting (SI Fig. 7B). In contrast to the previously reported gene-trap allele (Dlgh1^Gt), which encodes a protein containing all three PDZ domains fused to β geo, our null allele could encode only the first 50 aa of the full-length, 927-aa protein (Fig. 1A). This is because deletion of exon 4, which is 167 bp, alters the reading frame. Western blot analysis by using an antibody to an epitope in the first 163 aa of DLGH1 confirmed the elimination of DLGH1 in Dlgh1^−/− mice, and it revealed a significant reduction of DLGH1 in Dlgh1^+/− mice (Fig. 1B). We note that although the first 50 aa of DLGH1 might be able to interact with CASK (12), the presence of a premature termination codon in the mRNA should result in its degradation because of nonsense-mediated mRNA decay (18).

Fig. 1.

Structure and analysis of DLGH1. (A) Schematic diagrams of wild-type and mutant DLGH1 proteins. DLGH1 contains three PDZ domains, an SH3 domain, a protein 4.1-binding motif, and a guanylate kinase (GUK) domain. The previously reported gene trap allele encodes a fusion between the three PDZ domains and β geo; the null allele reported here encodes only the first 50 aa because of deletion of exon 4 and a reading frameshift. (B) Western blot analysis of lymphocyte extracts shows that DLGH1 is absent from Dlgh1^−/− mice and reduced in Dlgh1^+/− mice. This demonstrates the specificity of the DLGH1 antibody. Antibody to ERK2 was used as a control.

Neonatal Dlgh1 Null Mice Exhibit Urinary Tract Abnormalities, Including Hydronephrosis.

Dlgh1^−/− mice were present at the expected Mendelian ratio during embryogenesis, but they exhibited respiratory distress and died shortly after birth. Renal hypoplasia, also reported in Dlgh1^Gt/Gt mice, was present in all Dlgh1^−/− mice (Fig. 2A and B). However, we found that the Dlgh1^−/− embryos also developed dramatic urinary tract defects (Fig. 2 C–E): All had severely shortened ureters, 20% (12/63) exhibited unilateral renal agenesis, and 35% (22/63) developed severe unilateral or bilateral hydronephrosis perinatally. Hydroureter could be detected as early as embryonic day (E) 16.5 in Dlgh1^−/− embryos (data not shown). Histological analysis confirmed the hydronephrosis phenotype; in severe cases, only a thin rim of parenchyma containing a few developing nephrons remained at E18.5 (Fig. 2 H and I). In kidneys that did not exhibit hydronephrosis, the medullary collecting ducts were dilated frequently (Fig. 2 F and G), an early sign of hydronephrosis. Given the more severe renal phenotypes in our Dlgh1^−/− mice, we suggest that Dlgh1^Gt is actually a hypomorphic allele and that the truncated DLGH1/β geo fusion protein containing all three PDZ domains (15) maintains some residual functions that are missing in our Dlgh1^−/− mice. Alternatively, we cannot rule out the possibility that genetic background differences contribute to the differences in phenotype; our mice have been maintained on a mixed 129/C57BL/6J background, whereas the Dlgh1^Gt allele was maintained on an outbred CD1 background (16).

Fig. 2.

Dlgh1^−/− embryos exhibit renal hypoplasia, renal agenesis, and hydronephrosis. (A–D) Compared with controls (A), Dlgh1^−/− kidneys are small (B), and exhibit severely shortened ureters and hydronephrosis (C and D). The arrow in D points to a dilated ureter and hydronephrotic kidney. (E) Unilateral renal agenesis in a Dlgh1^−/− embryo from a different litter; arrow in E points to the uterus. (F and G) H&E staining of coronal sections of E18.5 kidneys; arrow in G points to dilated collecting ducts in the medulla. (H and I) H&E staining of parasagittal sections of E18.5 kidneys shows eroded parenchyma in the mutant caused by hydronephrosis (I). (Scale bars: A–E, 2 mm; F–I, 0.5 mm.)

Ureteric Bud Branching Is Reduced in Dlgh1^−/− Mice.

To investigate the role of Dlgh1 in urinary tract development, we examined its expression pattern in the developing kidney and ureter. Our findings confirm and extend earlier studies (16). Besides embryonic kidney, we found that Dlgh1 was expressed in embryonic ureter. By immunohistochemistry, we detected robust expression in the urothelium and low-level expression in ureteric smooth muscle cells (SMCs) that was clearly above the background fluorescence observed in Dlgh1^−/− ureters (Fig. 3A and B; data not shown).

Fig. 3.

Dlgh1 expression in mouse ureter and molecular characterization of Dlgh1^−/− ureters. (A and B) Immunostaining for DLGH1 in E18.5 ureter. DLGH1 levels are high in the epithelium and low but detectable in smooth muscle (A). DLGH1 is not detected in Dlgh1^−/− tissues (B), again demonstrating antibody specificity. (C and D) Immunostaining for uroplakin III (brown) in E18.5 ureter. Nuclei are counterstained with hematoxylin (blue). Uroplakin III is present in both wild-type and Dlgh1^−/− urothelium. (E and F) Immunostaining for cytokeratin 8 (red) and SMA (green) in E18.5 ureter epithelium and smooth muscle, respectively. Nuclei are stained with Hoechst. The cell layer between the ureter epithelium and smooth muscle in E is absent in F. (G and H) In situ hybridization for Raldh2 RNA in E18.5 ureter. Raldh2 staining (dark purple) between the epithelium and smooth muscle in control (G) is absent in the mutant (H). (I and J) In situ hybridization for Patched1 RNA in E14.5 ureter. Patched1 staining (dark purple) is similar in wild-type and Dlgh1^−/− mesenchyme around the urothelium. epi, epithelium; sm, smooth muscle; K8, cytokeratin 8; Ptch1, Patched1. (Scale bars: 50 μm.)

To begin to investigate the cause of the renal hypoplasia, we assayed branching morphogenesis of the ureteric bud (UB) in vitro and found reduced branching in Dlgh1^−/− kidneys (SI Fig. 8). In contrast, the condensation of the metanephric mesenchyme and its further development into glomeruli and tubules were unaffected by the loss of DLGH1 (SI Fig. 9 A–D). In addition, glomerular ultrastructure was comparable in the control and mutant kidneys (SI Fig. 9 E–H). Several genes that are important for kidney development, such as Ret, glial cell line-derived neurotrophic factor (Gdnf), Wnt4, retinaldehyde dehydrogenase 2 (Raldh2), Wilms tumor 1 (Wt1), and Pax2 were all expressed normally in the Dlgh1^−/− kidney (Fig. 4). Thus, reduced UB branching is the primary defect in the hypoplastic Dlgh1^−/− kidneys.

Fig. 4.

Several genes crucial for kidney development are expressed normally in Dlgh1^−/− kidneys. (A–H) In situ hybridization for Ret, Gdnf, and Wnt4 RNA at E18.5 and Raldh2 RNA at E14.5. No significant differences between wild-type and Dlgh1^−/− were observed. (I and J) Immunostaining for Wilms tumor 1 (WT1; red) and Cytokeratin 8 (K8; green) in E18.5 kidneys reveals normal condensation of metanephric mesenchyme (red) around ureteric bud tips (green). (K and L) Immunostaining for Pax2 is normal at E18.5. (Scale bars: A–H, 200 μm; I–L, 100 μm.)

Dlgh1^−/− Mice Develop a Normal Urothelium but Lack Ureteric Stromal Cells.

Based on the hydronephrosis phenotype and the expression pattern of Dlgh1 in the embryonic ureter, we wondered whether Dlgh1^−/− mice developed an abnormal urothelium. Surprisingly, we found normal urothelial ultrastructure (data not shown) and normal expression and localization of uroplakin III, a marker for differentiated urothelium, in Dlgh1^−/− ureters (Fig. 3 C and D).

Between the urothelium and the smooth muscle of normal newborn ureters is a mesenchymal cell population of undefined origin (19). We detected these cells at E17.5 and onwards and further demonstrated that they express Raldh2, a stromal cell marker, suggesting that they are progenitors of ureteral connective tissue that normally separates urothelium from smooth muscle (Fig. 3 E and G). Strikingly, this cell population was absent from Dlgh1^−/− ureters (Fig. 3 F and H). This suggests that Dlgh1 regulates stromal cell differentiation in the developing ureter, but not in the kidney, where Raldh2 was expressed normally (Fig. 4 G and H). These data demonstrate a critical function for Dlgh1 in ureteric architecture. We speculate that the ureteric stromal cells might provide flexibility during the contraction and relaxation phases of peristalsis, such that their absence from the Dlgh1^−/− ureter might contribute to the hydronephrosis phenotype.

Ureteric Smooth Muscle Is Disorganized in Dlgh1^−/− Mice.

Ureteric mesenchyme differentiates into SMCs in a proximal to distal wave starting at E15.5 (19). We found that the expression of smooth muscle actin (SMA) was greatly reduced in the E16.5 Dlgh1^−/− ureter compared with controls, but its expression in the E18.5 Dlgh1^−/− ureter appeared normal (Fig. 5A and B and data not shown). Immunostaining for smooth muscle myosin heavy chain, a marker for late-stage differentiated smooth muscle (20), in E18.5 ureters provided additional evidence that mutant SMCs were well differentiated by this stage (data not shown).

Fig. 5.

Defects in smooth muscle in the Dlgh1^−/− ureter. (A and B) Immunostaining for SMA in E16.5 ureter reveals greatly reduced expression in the mutant. (C and D) Immunostaining for connexin 45 in E18.5 ureter. Connexin 45 expression is similar in wild-type and Dlgh1^−/− smooth muscle. (E and F) Masson's Trichrome staining of E18.5 ureter cross-sections. SMCs are organized into circular muscle in the wild-type, but appear disorganized in the mutant. (G and H) Transmission electron micrographs of E18.5 ureter cross-sections. SMCs are spindle-shaped in the wild-type (G) but rounded in the mutant (H). (I and J) Toluidine blue staining of E18.5 ureter longitudinal sections. SMCs are rounded in the wild-type (I) but spindle-shaped in the mutant (J). (K and L) Transmission electron micrographs of E18.5 ureter longitudinal sections. SMCs are rounded in the wild-type (K) but spindle-shaped in the mutant (L). These results indicate that in the absence of DLGH1, SMCs that are normally organized into circular muscle are instead oriented longitudinally. sm, smooth muscle; Cx45, connexin-45. (Scale bars: A–F, 50 μm; G–L, 10 μm.)

Masson's Trichrome staining on cross-sections of E18.5 ureter revealed that SMCs were organized into typical circular muscle in the WT ureter, but SMCs appeared disorganized in the Dlgh1^−/− ureter (Fig. 5 E and F). EM revealed spindle-shaped SMCs in controls but showed rounded SMCs in Dlgh1^−/− ureter cross-sections (Fig. 5 G and H). Examination of longitudinal sections of ureters by Toluidine blue staining (Fig. 5 I and J) and EM (Fig. 5 K and L) revealed the opposite: rounded cells in controls and spindle-shaped cells in Dlgh1 nulls, indicating that SMCs were misaligned by 90° in the absence of DLGH1. Interestingly, this abnormal SMC organization was not observed in intestinal smooth muscle (data not shown), which also is organized into circular and longitudinal muscle, suggesting that regulation of SMC orientation by Dlgh1 is ureter-specific.

Sonic hedgehog (SHH) signaling has been shown to play crucial roles in regulating both the formation/maintenance of the subepithelial ureteric mesenchymal cells (we refer to these cells as stromal cells) and the differentiation of ureteric SMCs (19). We therefore looked for evidence of SHH signaling in Dlgh1^−/− ureters. We found that Shh was expressed in both wild-type and Dlgh1^−/− ureters (data not shown). Ptch1 and Bmp4, two downstream target genes of SHH signaling whose expression serves as readouts for response to SHH, were expressed normally in the mutant ureters (Fig. 3 I and J and data not shown). Thus, the ureteric stroma and SMC defects observed in Dlgh1 mutant ureters do not result from a diminution in SHH signaling.

Ureteral Peristalsis Is Severely Impaired in Dlgh1^−/− Mice.

To determine whether the aberrantly oriented ureteric smooth muscle affected peristalsis, we examined peristalsis in dissected E18.5 ureters by video microscopy. In wild-type ureters (13 of 13 examined that fully contracted), a spontaneous proximal squeezing (diameter-reducing) contraction was transduced to the distal end, resembling a typical peristaltic wave (Fig. 6A–D and SI Movie 1). However, this was not observed in Dlgh1^−/− ureters; instead, a primarily longitudinal (length-reducing) contraction occurred (7 of 7 examined that fully contracted; Fig. 6 E–H and SI Movies 2 and 3), although some minor diameter-reducing movements were observed occasionally. This is consistent with the observation that most of the circular muscle is replaced by longitudinal muscle in the Dlgh1^−/− ureter (Fig. 5). Furthermore, the apparently smooth and regular contractions and lack of obvious twitching movements suggest that the mutant SMCs are organized and do communicate with each other to propagate a coordinated contraction. Consistent with this, expression and localization of connexin 45, a component of ureteric smooth muscle gap junctions (21, 22), were comparable in wild-type and Dlgh1^−/− ureter (Fig. 5 C and D). Thus, we have obtained direct evidence that misorientation (rather than disorganization and miscommunication) of SMCs caused by the lack of DLGH1 impairs ureteric peristalsis. This provides a plausible mechanism to explain the hydronephrosis phenotype.

Fig. 6.

Defective ureteral peristalsis in the Dlgh1^−/− ureter. (A–D) Serial images from a digital video recording of a peristaltic cycle in a WT ureter. A squeezing (diameter-reducing) contraction (arrows) was transduced from the proximal end to the distal end. (E–H) Serial images from a digital video recording of a peristaltic cycle in a mutant ureter. A primarily longitudinal (length-reducing) contraction (indicated by bars) occurred (see SI Movies 1–3). Note: The kidney is up; the bladder is down. (Scale bar: 200 μm.)

Discussion

Disorganization of SMCs has been associated with congenital hydronephrosis in humans (23). In addition, several mouse models of congenital hydronephrosis exhibit defects in smooth muscle differentiation or proliferation (19, 24, 25). Here, we present a model of congenital hydronephrosis in which most of the circular smooth muscle is replaced by longitudinal muscle in the absence of DLGH1. Importantly, similar defects have been observed frequently in ureteropelvic junction obstruction in humans (2).

Little is known regarding the regulation of smooth muscle orientation during normal development of the ureter. Dlgh1 is expressed highly in the urothelium but at a lower level in SMCs. To our surprise, Dlgh1^−/− urothelium appeared normal, but the SMCs exhibited abnormalities, in terms of both their timing of differentiation and their orientation in the muscle. Thus, Dlgh1 may function non-cell-autonomously in this context by acting primarily in urothelial cells to control the alignment of the nearby SMCs. Indeed, Drosophila dlg has been shown to be able to function both cell- and non-cell-autonomously. For example, at dlg mutant neuromuscular junctions, a presynaptic cell-specific dlg transgene rescues the postsynaptic structural defect better than a postsynaptic cell-specific dlg transgene, suggesting that neurotransmitter release promoted by dlg influences postsynaptic cell phenotype (26).

Similarly, here Dlgh1 might function non-cell-autonomously by regulating the secretion of soluble factors from urothelium to SMCs or their precursors. One such possible secreted factor with a demonstrated role in SMC development is SHH (19). Mice with a urothelium-specific knockout in Shh exhibit hydroureter at the newborn stage, and then approximately half progress to hydronephrosis in adulthood. Analysis of mutant ureters showed that SMC differentiation is delayed, especially in the distal ureter, and the number of SMCs is reduced; these defects are consistent with the impaired flow of urine to the bladder. With regard to mechanism, those authors found a SHH-responsive population of subepithelial ureteral mesenchymal cells in the proximal ureter that were missing in the absence of SHH. They proposed that a high level of SHH near the urothelium maintains the proliferative state of these cells, whereas lower levels away from the urothelium are associated with differentiation into smooth muscle. These subepithelial ureteral mesenchymal cells appear to be the same cells that we observed to express Raldh2; we refer to them as ureteric stromal cells because Raldh2-expressing cells in the developing kidney are considered stromal cells (27). That these ureteric stromal cells also were absent in the Dlgh1^−/− ureter (Fig. 3) suggested a potential mechanistic link between DLGH1 function and SHH signaling. However, we were unable to obtain any evidence for a diminution in SHH signaling in Dlgh1 mutant ureters. Although it is still possible that altered SHH signaling is somehow involved in the Dlgh1^−/− phenotype, the fact that both the smooth muscle defects and the functional obstruction are more severe in the Dlgh1 mutant than in the Shh mutant suggests that additional pathways must be involved.

On the other hand, the fact that we can detect some DLGH1 in SMCs (Fig. 3) suggests that Dlgh1 might function cell-autonomously, perhaps by either clustering signaling receptors in SMCs (10) or regulating their orientation via effects on the cytoskeleton, as shown in astrocytes (28). The generation of ureteric epithelium- and SMC-specific mutations in Dlgh1, using a conditional Dlgh1 allele and appropriate Cre transgenes, will allow us to address these issues.

Interestingly, Dlgh1 null mice do not have widespread defects in epithelial polarity and control of cell proliferation; there were no tumor-like overgrowths in any tissues examined. This is in contrast to predictions from in vitro studies in which (i) RNAi knockdown of human DLG1 in CaCO2 cells causes mislocalization of E-cadherin and disruption of adherens junctions (13), and (ii) overexpression of DLGH1 and DLGH3 in 3T3 cells inhibits cell proliferation (29, 30). The absence of epithelial polarity and proliferation defects in vivo may be explained by the presence of functionally redundant DLG family members. The generation of mice with multiple family members mutated likely will be required to resolve this issue.

Materials and Methods

Generation of Dlgh1^−/− Mice.

A 7.5-kb EcoRI-StuI fragment of genomic DNA (strain 129) upstream and a 5.2-kb EcoRI-EcoRI fragment downstream of exon 4 were cloned into pLNTK-flanking PGKneo. The targeting vector was electroporated into B6/129 embryonic stem cells (31). Targeted clones were injected into C57BL/6J blastocysts to produce chimeras, which were mated to C57BL/6J mice. Genotypes were determined by Southern blotting and PCR analysis of tail DNA. Mice were kept on a mixed 129/C57BL/6J background. Studies were approved by the Washington University Animal Studies Committee.

Antibodies, Immunostaining, and Histology.

The antibodies or reagents that were used are as follows: mouse anti-DLGH1 (Stressgen Bioreagents, Ann Arbor, MI), mouse anti-uroplakin III (RDI Division of Fitzgerald Industries, Concord, MA), FITC-mouse anti-SMA (Sigma, St. Louis, MO), rat anti-cytokeratin 8 (TROMA-1, Developmental Studies Hybridoma Bank, Iowa City, IA), Dolichos biflorus agglutinin and Lotus tetragonolobus lectin (Vector Laboratories, Burlingame, CA), rabbit anti-Tamm-Horsfall glycoprotein (Biomedical Technologies, Stoughton, MA), rabbit anti-Wilms Tumor-1 (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-PAX2 (Covance/Babco, Berkeley, CA), rabbit anti-SM-MHC (Biomedical Technologies), rabbit anti-connexin 45 (Chemicon, Temecula, CA), Hoechst 33342 (Sigma), and FITC- and Cy3-conjugated secondary antibodies (Molecular Probes, Eugene, OR).

Immunostaining was performed on 7-μm frozen or 5-μm paraffin sections as described in ref. 32. Whole-mount staining of embryonic kidney cultures was performed as described in ref. 33. For light microscopy, 5-μm paraffin sections were stained with H&E, 10-μm frozen sections were stained with Masson's Trichrome, and 2-μm plastic sections were stained with Toluidine blue. All sections were prepared as described in refs. 32 and 34.

Embryonic Kidney Culture and Video Microscopy of Peristalsis.

E12.5 kidneys were cultured on transwell filters (BD Biosciences, San Jose, CA; pore size 1 μm) as described in ref. 35. E18.5 kidneys and associated ureters were removed and cultured for 2 h. Ureter movements were captured by using a MicroPublisher Imaging System driven by StremPix software (NorPix, Montreal, QC, Canada).

In Situ Hybridization.

In situ hybridization was performed according to the Nonradioactive In Situ Hybridization Application Manual from Roche Applied Science (Indianapolis, IN). Digoxigenin-labeled cRNA probes were generated by using cDNA fragments encompassing nt 72–529 of Raldh2, nt 1444–2389 of Ret, nt 122–781 of Gdnf, or nt 339–760 of Wnt4 (adenine of the initiator Met was assigned as nt 1) as a template. Patched1 riboprobe was generated from a full-length cDNA. Ret and Gdnf templates were gifts from Robert Heuckeroth (Washington University School of Medicine), Wnt4 template was a gift from Andrew McMahon (Harvard University, Cambridge, MA), and Patched1 template was a gift from Matthew Scott (Stanford University School of Medicine, Stanford, CA).

Abbreviations

Dlgh1

discs-large homolog 1

En

embryonic day n

PDZ

postsynaptic density-95/discs-large/zonula occludens-1

SHH

sonic hedgehog

SMA

smooth muscle actin

SMC

smooth muscle cell.