GLI3-dependent transcriptional repression of Gli1, Gli2 and kidney patterning genes disrupts renal morphogenesis

Regulation of GLI3 repressor formation via cleavage of GLI3 is crucial during mammalian morphogenesis. Gli3 mutations that generate a putative truncated protein similar in size to GLI3 repressor are found in humans with Pallister-Hall syndrome (PHS) and malformations including polydactyly, imperforate anus, hypothalamic harmartoma and renal dysplasia/aplasia (Kang et al.,1997). Moreover, mice homozygous for a targeted mutation that generates a 699 amino acid N-terminal GLI3 protein exhibit numerous malformations, including renal aplasia/dysplasia as observed in PHS(Bose et al., 2002). The molecular mechanisms that control these deleterious GLI3-dependent effects are unknown.

GLI3 is an intracellular transcriptional effector in the sonic hedgehog(SHH) signaling pathway. Among the members of the hedgehog (HH) family of secreted proteins, SHH controls cell fate, determination, proliferation and tissue patterning during embryogenesis (reviewed by Ingham and McMahon, 2001). In Drosophila, Hh signaling is mediated, at the transcriptional level,by a single zinc-finger protein, Cubitus interruptus (Ci)(Methot and Basler, 2001). In the absence of a Hh signal, Ci is processed by proteolysis into an N-terminal fragment that includes the zinc finger region and acts to repress gene transcription (Aza-Blanc et al.,1997). In vertebrates, three Ci homologs, GLI1, GLI2 and GLI3, mediate Hh signals by controlling gene expression. GLI1 and GLI2 exist as full length proteins in cultured mammalian cells(Dai et al., 1999) and act primarily as transcriptional activators during murine embryogenesis(Bai et al., 2002; Park et al., 2000). Absence of a SHH signal provokes processing of GLI3 into a shortened form that can act as a transcriptional repressor in cultured cells and limb explant cultures(Litingtung et al., 2002; Wang et al., 2000). Genetic analyses in mice demonstrate that the predominant activity of GLI3 is to repress expression of Shh dependent genes(Bai et al., 2002; Park et al., 2000). Integration of opposite GLI activities appears to be critical. Genetic inactivation of Gli3 can rescue Shh mutant phenotypes in embryonic tissues including the neural tube(Litingtung and Chiang, 2000),limb (Litingtung et al.,2002), face and forebrain(Rallu et al., 2002), and skin(Mill et al., 2005) suggesting a dynamic interplay between GLI signals during tissue formation. The molecular mechanisms that control such interactions are largely undefined.

The mammalian kidney is a model with which to study embryonic epithelial-mesenchymal interactions and growth factor signaling. In the kidney, interactions between the ureteric bud, an epithelial structure, and the metanephric mesenchyme, a mesenchymal tissue, instigate growth and branching of the ureteric bud, a process known as renal branching morphogenesis, and conversion of the metanephric mesenchyme to those epithelial elements that exist proximal to the ureteric bud branches and their daughter collecting ducts. Growth factors secreted by the ureteric bud and metanephric mesenchymal cells act in a autocrine and paracrine manner to control cellular events including cell proliferation and to control the expression of genes such as Pax2, Sall1 and Mycn, each of which performs crucial functions (reviewed by Bouchard, 2004). The presence of renal hypoplasia/dysplasia in mice deficient in Shh in the ureteric bud lineage (Yu et al.,2002) and in mice expressing a truncated form of GLI3(Bose et al., 2002)demonstrates a crucial role for SHH-GLI signaling during mammalian renal development. However, the molecular mechanisms by which GLI3 controls renal embryogenesis are unknown.

To identify mechanisms by which SHH-GLI signaling controls renal organogenesis, we generated models of deficient SHH or SMO signaling in mutant mice and cultured embryonic kidneys. Homozygous inactivation of Shhor treatment of pregnant wild-type mice with cyclopamine, a steroidal alkaloid that blocks SMO activity (Chen et al.,2002), caused renal aplasia or dysplasia. Although these states lowered the expression of both GLI1 and GLI2 protein, the relative expression of a shortened form of GLI3 was increased relative to full-length GLI3 protein. Homozygous inactivation of Gli3 in a Shh-deficient background normalized the renal phenotype and rescued kidney expression of Pax2, Sall1, cyclin D1 and MYCN, as well as GLI1 and GLI2. We defined mechanisms by which GLI3 controls expression or these SHH targets using chromatin immunoprecipitation. In normal kidney tissue, GLI1 and/or GLI2 bound a GLI-binding consensus region within each target gene 5′ flanking region. By contrast, in cyclopamine-treated malformed cultured kidneys, we observed decreased associations with GLI1/GLI2 and de novo associations with GLI3. Genetic inactivation of Gli3 alone blocked inhibition of GLI1 and GLI2 expression, and renal branching morphogenesis by cyclopamine. We propose a model in which SHH-SMO signaling controls generation of a GLI3 repressor, which, in turn, controls Gli1 and Gli2 as well as non-Gli target genes crucial to embryonic kidney development.

Shh^+/–(Chiang et al., 1996), Gli2^+/– (Mo et al., 1997) and Gli3^+/–(Schimmang et al., 1992)CD1/129 mice were housed in the Animal Facility of The Hospital for Sick Children (Toronto, Canada) and genotyped by PCR as described previously(Chiang et al., 1996; Mo et al., 1997). Animal experiments were approved by the ethics committee at The Hospital for Sick Children.

Antibodies and reagents were as follows: GLI1 (Abcam, Cambridge, MA);cyclin D1 (BD PharMingen, San Diego, CA); GLI3, MYCN, MYC, cyclin D2 (Santa Cruz Biotechnology, Santa Cruz CA); cyclopamine (Toronto Research Chemicals,Toronto, ON); and SHH-N (B & D Systems, Minneapolis, MN). A partial mouse Gli2 cDNA corresponding to amino acids 327-442 was cloned into the pGEX1 vector (Amersham). Fusion protein was expressed in E. coli(BL21) and purified on glutathione sepharose (Amersham) according to the manufacturer's instructions. Antibodies were affinity purified using a column of Affi-gel 10 beads (BioRad) conjugated with His-tagged GLI2 antigen. Cyclopamine was dissolved in 100% ethanol for culture of kidney explants and dissolved in 45% 2-hydroxypropyl-beta-cyclodextrin (HBC, Sigma-Aldrich, St Louis, MO) in PBS for intraperitoneal injection. Shh-N was dissolved in 1% FBS in PBS.

Immunohistochemistry in paraffin wax-embedded sections (4 μm) of kidney tissue was performed (Hu et al.,2003) with the following antibodies: mouse anti-GLI1 (1:10),rabbit anti-GLI2 (1:700) and rabbit anti-GLI3 (1:50). Biotinylated secondary antibodies were used in a biotin-avidin complex assay (Vector laboratories,Burlingame, CA). Immunoblotting of proteins transferred to nylon membrane was performed using the following antibodies, anti-GLI1 (1:3300 dilution),anti-GLI2 (1: 250 dilution), anti-GLI3 (1:250 dilution, Santa Cruz Biotech),anti-MYC (1:250 dilution), anti-MYCN (1:250 dilution), anti-cyclin D1 (1:250 dilution), anti-cyclin D2 (1:250 dilution) and anti-β-actin (1:3000 dilution, Sigma, St Louis, MO). Secondary anti-mouse IgG and anti-rabbit IgG conjugated HRP (Amersham Pharmacia Biotech, Piscataway, NJ) were used at a 1:3000 and 1:5000 dilution, respectively. Chemiluminescence was performed using commercially available reagents (ECL kit; Amersham Pharmacia Biotech,Piscataway, NJ).

Kidneys were surgically dissected from mouse embryos and cultured on 0.45 mm polyethylene terephthalate membranes (Corning) in multiwell plates in the presence of Richter's modified Dulbecco's modified Eagle's medium-Ham's F-12 nutrient mixture (DMEM-F12) containing 50 μg/ml transferrin (Sigma)(Pichel et al., 1996). Ureteric bud-derived structures were identified in whole-mount kidney specimens with fluorescein isothiocyanate (FITC)-conjugated DBA (20 μg/ml;Vector Labs, Burlington, ON) (Piscione et al., 1997). A ureteric bud branch point was defined as the intersection between two connected branches.

Pregnant wild-type or Gli3^+/– mice were injected via the peritoneum at E 9.5 with cyclopamine (6 mg/kg body weight) once per day for 4 consecutive days. Embryos were then isolated surgically and fixed in 4% paraformaldehyde-PBS. Alternatively, for treatment of cultured embryonic kidney explants, cyclopamine was added to culture medium at a final concentration of 10 μM. Shh-N was dissolved in 1% FBS and added to culture medium at final concentration of 1.0 μg/ml for 4 days.

In situ hybridization was performed in paraffin wax-embedded tissue sections as described (Mo et al.,1997) using probes encoding Pax2 [kindly provided by Dr Peter Gruss (Stoykova and Gruss,1994)] and Sall1 [kindly provided by Dr Ryuichi Nishinakamura (Nishinakamura et al.,2001)]. For whole-mount in situ hybridization, embryos were fixed in 4% paraformaldehyde at 4°C and processed according to published methods(Conlon and Herrmann,1993).

Cell proliferation in kidneys was assayed by incorporation of 5-bromo-2′-deoxyuridine (BrdU, Roche Molecular Biochemicals, Mannheim,Germany). Pregnant female mice received an intraperitoneal injection of BrdU(100 μg/g body weight) 4 hours prior to scarifying these mice. Identification of BrdU-positive cells was performed by immunostaining as described previously (Cano-Gauci et al.,1999).

Chromatin immunoprecipitation was carried out using published methods(Hu and Rosenblum, 2005). GLI consensus binding sites were identified as TGGGTGGTC or GACCACCCA(Lai et al., 2004). Oligodeoxynucleotide primers were designed to encompass these binding sites as follows. Mouse cyclin D1 promoter (GenBank Accession Number AF212040):forward, 5′-AATTCTAAAGGTGGGGGAACA-3′, reverse 5′-GAGACACGATAGGCTCCTTCC-3′ (148 bp PCR product). Mouse Mycn promoter (GenBank Accession Number X06993): forward 5′-TAATATCCCCCGAGCTTCAA-3′, reverse 5′-AGCTTCGCAAGTACCGCTTC-3′ (254 bp PCR product at annealing temperature 56°C for 30 seconds). Mouse Sall1 promoter (GenBank Accession Number AC147558): forward 5′-AGGCGGTGCCTAGGGTCT, reverse 5′-CTGAAGTTTTCGGGAGAAGC-3′ (299 bp PCR product at annealing temperature 54°C for 45 seconds). Mouse Pax2 promoter (GenBank Accession Number MMU13975): forward 5′-GGGCTTTGCAGCTTTTAGAG-3′,reverse 5′-TTGGCAGAGAAGTAGCAATCC-3′ (256 bp PCR product at annealing temperature 54°C for 45 seconds), Mouse Gli1 promoter(ENSEMBL gene ID ENSMUSG00000025407): forward 5′-AGCTTCAGTGCCACCCCAC-3′, reverse 5′-TCGGTCCGAAGGAAGGATATA-3′ (197 bp PCR product at annealing temperature 55°C for 40 seconds). Mouse Gli2 promoter (ENSEMBL gene ID ENSMUSG00000051835): forward 5′-GTGCTGAGGCTCCTGACAAT-3′and reverse 5′-CAGCAGAGCAAGGTGAACAG-3′ (124 bp PCR product at annealing temperature 55°C for 40 seconds).

Total RNA from frozen tissue samples was extracted using the RNeasy Mini kit (Qiagen, Mississauga, ON, Canada) according to the manufacture's protocol. First-strand cDNA was generated from 2 μg total RNA using random hexamers(SuperScript First Strand Synthesis System, Invitrogen). The following primers were used. Gli1: forward 5′-ATCACCTGTTGGGGATGCTGGAT-3′,reverse 5′-GGCGTGAATAGGACTTCCGACAG-3′ (317 bp product at annealing temperature 56°C for 45 seconds). Gli2: forward 5′-GTTCCAAGGCCTACTCTCGCCTG-3′, reverse 5′-CTTGAGCAGTGGAGCACGGACAT-3′ (304 bp product at annealing temperature 56°C for 45 seconds). Gli3: forward 5′-AGCAACCAGGAGCCTGAAGTCAT-3′, reverse 5′-GTCTTGAGTAGGCTTTTGTGCAA-3′ (270 bp product at annealing temperature 56°C for 40 seconds). β-actin: forward 5′-TGTTACCAACTGGGACGACA-3′, reverse 5′-CTCTCAGCTGTGGTGGTGAA-3′ (393 bp product). PCR was performed using HotStarTag. The RT-PCR amplified products were separated by agarose gel electrophoresis and stained with ethidium bromide.

Mean differences between groups were analyzed using Student's t-test (two-tailed) or by ANOVA using Stat-View statistical analysis program (version 4.01; Abacus Concepts, Berkeley, CA) for more than two groups. Statistical significance was taken at a value of P<0.05.

We determined the requirement for SHH signaling during murine kidney development by analyzing the renal phenotype in SHH-deficient mice and in mice or kidney explants treated with cyclopamine. Analysis of 20 Shh^–/– E14.5 embryos revealed bilateral renal aplasia in 50% (Fig. 1C). The remaining 50% exhibited a single ectopic kidney located in the pelvis(Fig. 1D). These ectopic kidneys demonstrated histological features of renal dysplasia including disorganization of structures within the renal parenchyma and cystic renal tubules (Piscione and Rosenblum,1999) (Fig. 1E). Using a complementary strategy, we used cyclopamine to block SMO function and investigate the function of the SHH-SMO signaling during renal morphogenesis. An intraperitoneal injection of 6 mg/kg cyclopamine daily for 4 consecutive days starting at E9.5 exerted a major deleterious effect on kidney development(Fig. 1H,I). Compared with embryos treated with drug vehicle alone(Fig. 1F,G), the kidneys in cyclopamine treated embryos demonstrated primordial branching of the ureteric bud and a paucity of metanephric-derived epithelial structures. To determine SMO functions after the onset of mesenchymal-epithelial interactions, kidneys were harvested at E11.5 and cultured in the presence of cyclopamine for 4 consecutive days. Qualitative analysis demonstrated marked inhibition of ureteric bud branching identified by Dolichos Biflorus Agglutinin (compare Fig. 1K with 1J). Quantitation of this effect, performed by counting branch point number, showed that cyclopamine inhibited formation of ureteric bud branch points by 40% (branch point number – drug vehicle versus cyclopamine: 61.6±2.4 versus 37.6±1.6; P<0.0001. n=5 kidneys/group). Analysis of histological sections generated from these kidneys revealed a paucity of metanephric-derived epithelial elements (see Fig. S1 in the supplementary material). Taken together, these results in two complementary models of disrupted SHH-SMO signaling demonstrate a crucial role for SHH-SMO signaling during induction of the metanephric blastema by the ureteric bud and during branching morphogenesis and nephrogenesis.

In mammals, GLI1, GLI2 and GLI3 are the intracellular effectors of SHH-SMO signaling. Yet the regulation of GLI protein expression in response to SHH is not well understood. We used antibodies specific for distinct GLI proteins to investigate the effects of SHH deficiency and SMO inactivity on GLI expression. The specificity of anti-GLI antibodies was demonstrated by lack of detection of GLI2 and GLI3 in kidneys isolated from Gli2^–/–(Fig. 2F) and Gli3^–/–(Fig. 2K) mice, respectively. Expression of GLI1, GLI2 and GLI3 was detected in both the metanephric-derived epithelium and ureteric bud of kidneys isolated from E14.5 wild-type mice(Fig. 2A,E,I). By contrast,homozygous inactivation of Shh but not Gli3 decreased GLI1 and GLI2 protein expression to undetectable levels in dysplastic kidneys(Fig. 2C,D,G,H). GLI3 expression was not affected by SHH deficiency(Fig. 2L).

Analysis of GLI protein expression in kidney tissue lysates confirmed the decrease in GLI1 and GLI2 proteins observed by immunohistochemistry(Fig. 2M). Western analysis provided additional insight into the expression of GLI3 demonstrating a marked decrease in the full-length activator form of GLI3 (190 kDa) with preserved expression of the short repressor form of GLI3 (89 kDa). This resulted in a marked decrease in the ratio of GLI3 activator to GLI3 repressor in tissue isolated from Shh^–/– mice (GLI3 activator:GLI3 repressor: wild type versus Shh^–/–, 3.25 versus 0.23). We observed similar changes in GLI protein expression in cyclopamine-treated wild-type embryonic kidney explants(Fig. 2N). First, cyclopamine decreased GLI1 and GLI2 protein expression. Second, cyclopamine increased formation of GLI3 repressor and decreased levels of GLI3 activator, albeit to a more modest degree than that observed in Shh^–/– mice. Third, cyclopamine decreased the ratio of GLI3 activator to GLI3 repressor (GLI3 activator:GLI3 repressor– culture medium alone versus drug vehicle versus SHH-N versus cyclopamine, 3.68 versus 3.8 versus 11.9 versus 1.43, respectively). Together,these data demonstrate that SHH deficiency or SMO inhibition decreases GLI1 and GLI2 and increases the relative expression of GLI3 repressor.

To investigate the impact of increased GLI3 repressor on kidney development, we analyzed Shh^–/–;Gli3^–/– mice at different developmental stages. During murine kidney development, Pax2, a member of the paired box family of transcription factors, is expressed in the Wolffian Duct, ureteric bud and induced metanephric blastema and is required for metanephric development(Dressler et al., 1990; Torres et al., 1995)(Fig. 3A). Thus, we used Pax2 mRNA expression as a marker of early inductive events(Fig. 3A). Consistent with the finding of renal aplasia in 50% Shh^–/– mice, Pax2 mRNA expression was barely detectable in ∼50% of Shh^–/– mice(Fig. 3B). In the remaining Shh^–/– embryos, Pax2 mRNA was detected in the metanephros (Fig. 3C). Although loss of Gli3 function alone had no effect on Pax2 mRNA expression (Fig. 3D), it restored Pax2 mRNA expression in all Shh^–/–;Gli3^–/– mice(Fig. 3E). Next, we investigated the role of Gli3 at a later developmental stage when branching morphogenesis and metanephric epithelialization are established. By E13.5, Pax2 is normally expressed in both the induced metanephric mesenchyme and ureteric bud branches (Fig. 3K). In addition, Sall1 expression is localized to the metanephric mesenchyme (Fig. 3P). Both Pax2 and Sall1 are required for metanephrogenesis at this stage(Nishinakamura et al., 2001; Rothenpieler and Dressler,1993). Using Pax2 and Sall1 as markers of ureteric bud and metanephric development, we observed rescue of kidney number,histology and Pax2 and Sall1 expression in Shh^–/–;Gli3^–/– mice(Fig. 3J,O,T) compared with Shh^–/– mice(Fig. 3G,L,Q). Interestingly,deficiency of neither Gli2 nor Gli3 interfered with expression of Pax2 (Fig. 3M,R) and Sall1 (Fig. 3N,S). Taken together, these data demonstrate that the deleterious effect of SHH deficiency on kidney development is dependent on GLI3.

Growth factor-dependent control of cell proliferation is tightly regulated during kidney development. Dysregulation of cell proliferation results in renal dysplasia (Michael and Davies,2004). We determined the impact of altered SHH signaling on cell proliferation by measuring incorporation of 5-bromodeoxyuridine (BrdU), a surrogate marker of cell proliferation, in embryonic kidneys(Fig. 4). Qualitative analysis of kidney sections stained with anti-BrdU antibody(Fig. 4A-D) revealed incorporation of BrdU in epithelial and mesenchymal derivatives located on the outer regions of kidneys isolated from wild-type, Gli3^–/– and Shh^–/–;Gli3^–/– kidneys. By contrast, BrdU incorporation was markedly diminished in Shh^–/– kidneys. Quantitation of the number of BrdU-positive cells in ureteric bud branches, identified by Dolichos Biflorus Agglutinin (Grisaru et al.,2001), and in mesenchymal cells demonstrated an approximate twofold and fourfold diminution of cell proliferation in the ureteric bud and metanephric mesenchyme, respectively, in kidney tissue from Shh^–/– mice compared with wild type(Fig. 4E). By contrast, in kidneys isolated from Shh^–/–;Gli3^–/– mice,cell proliferation in both these tissue compartments was rescued to levels observed in wild-type tissue. Interestingly, a higher rate of ureteric bud cell proliferation was observed in Gli3 deficient mice compared with wild type.

In nonrenal tissues, members of the cyclin and MYC families regulate cell proliferation in a SHH-dependent manner(Li et al., 2004; Mill et al., 2005). Thus, we determined the effect of SHH deficiency on the expression of cyclin D1, cyclin D2, MYCN and MYC in the kidney of SHH-deficient mice using specific antibodies and western analysis (Fig. 4F). Although each of these gene products is expressed in the embryonic kidney in wild-type mice, the expression of both cyclin D1 and MYCN was greatly diminished in kidneys isolated from Shh^–/–mice but was rescued in the kidneys of Shh^–/–;Gli3^–/– mice. Together, these data demonstrate a crucial role for GLI3 in controlling the response to SHH deficiency at the level of cell proliferation and target gene expression.

We determined how differential control of GLI1, GLI2 and GLI3 by SHH-SMO signaling controls the expression of SHH target genes in the kidney. First, we identified putative GLI-binding consensus sequences(Ikram et al., 2004; Kinzler and Vogelstein, 1990; Sasaki et al., 1997) in the 5′ flanking regions of Pax2, Sall1, cyclin D1 and Mycn(Fig. 5A). Next, we used chromatin immunoprecipitation to analyze the association of GLI1, GLI2 or GLI3 with regions encompassing these sequences. Our results indicate that GLI1 and/or GLI2, but not GLI3, bind these promoter elements in wild-type kidneys either cultured in vitro (Fig. 5B) or processed immediately after isolation (see Fig. S2 in the supplementary material). Treatment of embryonic kidney explants with cyclopamine induced changes in GLI protein binding with these target promoters(Fig. 5B). Most remarkably, the binding of GLI1 and GLI2 with Pax2 and Sall1 was barely detectable. Strikingly, we observed de novo binding of GLI3 with each of these promoters. These results demonstrate that decreased SMO signaling changes induces the de novo association of GLI3 and decreases the association of GLI1 and GLI2 with target promoters.

Our genetic analyses in Shh^–/– mice demonstrated that Shh deficiency decreases GLI1 and GLI2 as well as expression of Shh target genes. By contrast, removal of both Gli3 and Shh restored expression of Shh target genes. To determine whether Gli3 plays a primary role in orchestrating these events, we investigated whether Gli3 controls Gli1 and Gli2 in Shh^–/–;Gli3^–/– mice. Analysis of GLI protein expression in kidney tissue lysates demonstrated rescue of GLI1 and GLI2 levels in Shh^–/–;Gli3^–/– mice to the levels observed in wild-type mice (Fig. 6A). To determine whether GLI3 controls GLI1 and GLI2 expression at a transcriptional level, we examined mRNA in kidney tissue isolated from wild-type and Shh^–/–;Gli3^–/– mice using reverse transcriptase PCR (Fig. 6B). Although Gli1 and Gli2 mRNA levels were decreased in Shh^–/– mice, the levels of these mRNAs were rescued to wild-type levels in Shh^–/–;Gli3^–/– mice. These results provided a basis for determining how GLI3 controls Gli1and Gli2 transcription. First, we identified GLI binding sites in the 5′ flanking regions of Gli1 and Gli2(Fig. 6C). Next, we examined the association of GLI proteins with these sites by chromatin immunoprecipitation. Analysis of the regions encoding GLI binding sequences in wild-type kidneys either cultured in vitro(Fig. 6D) or processed immediately after isolation (see Fig. S3 in the supplementary material)demonstrated that GLI2 bound both Gli1 and Gli2. A weaker association was observed between GLI1 and Gli1 or Gli2 and between GLI3 and Gli2. cyclopamine enhanced binding between GLI3 and the Gli1 promoter and between GLI3 and the Gli2 promoter. By contrast, the associations with GLI2 were decreased to almost undetectable levels. Our concurrent finding that cyclopamine decreased mRNAs encoding Gli1 and Gli2 and Pax2 showed that these effects in vitro paralleled our findings in vivo (see Fig. S4 in the supplementary material). Together, these results demonstrate that GLI3 binds to the Gli1 and Gli2 promoters when SHH-SMO signaling is inhibited.

To investigate whether Gli3 is primarily involved in controlling Gli1 and Gli2 expression, we analyzed their expression in cyclopamine-treated embryonic kidney explants isolated from Gli3^–/– mice. Our prior results(Fig. 6) predicted that cyclopamine would fail to decrease GLI1 and GLI2 in GLI3 deficient tissue. In kidneys isolated from wild-type mice, cyclopamine markedly decreased GLI1 and GLI2 (Fig. 7A) and branching morphogenesis (Fig. 7B). By contrast, we observed no effect of cyclopamine on GLI1 and GLI2 or ureteric bud branching in tissue isolated from Gli3^–/–mice (Fig. 7A,B). Thus, GLI3 controls GLI1 and GLI2 expression and renal morphogenesis when SMO is inhibited.

The experiments reported here demonstrate that decreased SHH-SMO signaling disrupts kidney development and regulates GLI3 expression in a manner distinct from that of GLI1 and GLI2. Expression of the repressor form of GLI3 is increased, while its full-length form and that of GLI2 and GLI3 is decreased. The significance of GLI3 repressor formation was demonstrated by our finding that homozygous GLI3 deficiency in Shh deficient mice rescued the dysplastic renal phenotype and the expression of Pax2, Sall1, cyclin D1 and Mycn. Analysis of GLI protein association with GLI consensus binding regions in the 5′ flanking regions of these target genes provides a novel insight into potential mechanisms by which GLI3 controls gene expression. Binding of GLI1 and GLI2 to the Pax2, Sall1, cyclin D1 and Mycn promoters established these genes as direct targets of SHH. De novo GLI3 binding to these same regions in a state of SMO inactivity suggests that GLI3 directly represses these genes. Decreased expression of GLI1 and GLI2 and decreased binding of these proteins to their target genes raised the possibility that GLI3 regulates GLI1 and GLI2 expression. Binding of GLI3 to the Gli1 and Gli2 promoters in tissue treated to inhibit SMO supported this hypothesis. Our demonstration that Gli3 deficiency abrogates the deleterious effects of SMO inhibition on GLI1 and GLI2 expression and normalizes branching morphogenesis provided functional support for our hypothesis. Together, our results indicate that SHH-SMO signaling regulates the formation of GLI3 repressor. In a state of decreased SHH-SMO signaling,GLI3 represses GLI1, GLI2 and genes crucial to renal morphogenesis.

Our results demonstrate a requirement for SHH-SMO signaling during kidney development. What are the mechanisms that underlie this requirement? One possibility is loss of one or more GLI activator functions. Gli1,Gli2 and Gli3 have all been shown to function as SHH-SMO-dependent activators in nonrenal tissues in a context-dependent manner(Bai et al., 2004; Buttitta et al., 2003; McDermott et al., 2005). Yet,homozygous deficiency of Gli1, Gli2 and Gli3 does not negatively impact murine kidney development(Bai et al., 2002; Kim et al., 2001; Schimmang et al., 1992). By contrast, mutations that generate expression of a shortened repressor-like form of GLI3 in humans with Pallister-Hall syndrome(Kang et al., 1997) and in genetically engineered mice (Bose et al.,2002) also disrupt kidney development. Our studies provide novel insights into the potential actions of GLI3 repressor showing that when SMO is inhibited, GLI3 controls Gli1 and Gli2 expression and associates with the Gli1 and Gli2 promoters. These findings suggest that in Pallister-Hall syndrome, GLI3 represses Gli1 and Gli2 as well as Pax2, Sall1, cyclin D1 and Mycn via interactions with their respective promoter elements. More generally, our findings provide insight into the observation that genetic inactivation of Gli3 can rescue Shh mutant phenotypes in nonrenal tissues(Litingtung and Chiang, 2000; Litingtung et al., 2002; Mill et al., 2005; Rallu et al., 2002). Our results suggest that in these different developmental contexts GLI3 repressor directly controls Gli transcription and that of Gli-dependent target genes.

Our results provide new insights into the target genes regulated by SHH and SMO, demonstrating that Shh acts upstream of Pax2 and Sall1. Both Pax2 and Sall1 perform crucial functions during murine renal organogenesis(Nishinakamura et al., 2001; Rothenpieler and Dressler,1993). Mutations in PAX2 are associated with renal coloboma syndrome and renal hypoplasia(Sanyanusin et al., 1995). Homozygous inactivation of Pax2 in mice results in renal aplasia or severe dysgenesis (Rothenpieler and Dressler, 1993), indicating a requirement for Pax2 during outgrowth of the ureteric bud and invasion of the metanephric blastema. The observation that Pax2 controls expression of Gdnf(Brophy et al., 2001), which encodes a secreted growth factor essential for outgrowth and invasion by the ureteric bud (Moore et al.,1996), suggests that Shh acts upstream of the GDNF-RET signaling pathway during renal organogenesis. SALL1 mutations are found in individuals with Townes Brock syndrome and renal aplasia/dysplasia(Kohlhase et al., 1998). During murine kidney development, Sall1 expression is restricted to the metanephric mesenchyme where it is essential as homozygous inactivation of Sall1 abrogates outgrowth of the ureteric bud(Nishinakamura et al., 2001). Our results provide novel insight into Sall1 regulation showing that it is controlled by Shh.

Our results using chromatin immunoprecipitation extend our genetic analyses and demonstrate that promoter elements in both Pax2 and Sall1 are bound by GLI1, GLI2 and GLI3 to variable degrees and in different combinations, depending, in part, on the state of SHH-SMO signaling.

Our results demonstrate that SHH controls cell proliferation and expression of cyclin D1 and MYCN during kidney development as in non-renal embryonic tissues (Long et al., 2001; Mill et al., 2005). Interestingly, while cyclin D2 is regulated by SHH in the developing hair follicle (Mill et al., 2005),cyclin D2 expression is independent of Shh expression in the embryonic kidney. The mechanisms by which GLI proteins mediate the actions of SHH to control cell proliferation appear to differ between skin and kidney. In skin, both Shh and Gli2 control cell proliferation(Mill et al., 2005). In kidney, Gli2 deficiency does not affect cell proliferation (data not shown). In both skin and kidney Gli3 deficiency rescues decreased cell proliferation observed in Shh^–/– mice. However, the degree of rescue is much greater in kidney compared with skin,suggesting that other factors modulate the actions of GLI proteins to control cell proliferation. These factors may include GSK3β and β-catenin(Mill et al., 2005). The relevance of this mode of regulation to kidney development requires further investigation.

The experiments reported here provide a basis for a model that predicts that Gli3 acts downstream of SMO and upstream of Gli1 and Gli2 in the developing kidney and that Shh programs kidney morphogenesis by restricting the activity of GLI3. In our model, GLI3 repressor orchestrates the effects of SMO inhibition, thereby disrupting kidney morphogenesis (Fig. 7C). GLI3 repressor acts by inhibiting the transcription of several classes of genes, the actions of which are crucial to kidney development(Fig. 7C). The first is the Gli family, specifically Gli1 and Gli2. The second class of genes includes those crucial to renal patterning. Members of this class include Pax2 and Sall1. The third class of genes is that controlling cell proliferation and includes cyclin D1 and MYCN. Formation of GLI3 repressor is critical to repression of these genes. Our findings provide a basis for investigating molecular mechanisms that control the interplay of GLI1, GLI2 and GLI3 in the regulation of these SHH-SMO target genes.

This work was supported by operating grants awarded by the Canadian Institutes of Health Research (to N.D.R. and C.C.H.). We thank Ya-Jun Li for technical assistance. Work in the Chuang laboratory was supported by an NIH grant (HL67822). N.D.R. is supported by a Canada Research Chair.