A Novel Wilms Tumor 1 (WT1) Target Gene Negatively Regulates the WNT Signaling Pathway

Abstract

Mammalian kidney development requires the functions of the Wilms tumor gene WT1 and the WNT/β-catenin signaling pathway. Recent studies have shown that WT1 negatively regulates WNT/β-catenin signaling, but the molecular mechanisms by which WT1 inhibits WNT/β-catenin signaling are not completely understood. In this study, we identified a gene, CXXC5, which we have renamed WID (WT1-induced Inhibitor of Dishevelled), as a novel WT1 transcriptional target that negatively regulates WNT/β-catenin signaling. WT1 activates WID transcription through the upstream enhancer region. In the developing kidney, Wid and Wt1 are coexpressed in podocytes of maturing nephrons. Structure-function analysis demonstrated that WID interacts with Dishevelled via its C-terminal CXXC zinc finger and Dishevelled binding domains and potently inhibits WNT/β-catenin signaling in vitro and in vivo. WID is evolutionarily conserved, and ablation of wid in zebrafish embryos with antisense morpholino oligonucleotides perturbs embryonic kidney development. Taken together, our results demonstrate that the WT1 negatively regulates WNT/β-catenin pathway via its target gene WID and further suggest a role for WID in nephrogenesis.

Introduction

Wilms tumor is a childhood kidney cancer arising from cells that failed to differentiate during kidney development (1). About 10–15% of Wilms tumors have loss of function mutations in the Wilms tumor gene (WT1) encoding a zinc finger transcription factor indispensable during multiple stages of kidney development (2–4). More recently, mutations resulting in the activation of the WNT/β-catenin signaling pathway have been identified in Wilms tumors. These include the activating mutations in CTNNB1 (5–7), which encodes β-catenin, and loss of function mutations in WTX (8–10), encoding an X-linked protein that promotes degradation of β-catenin (11). Interestingly, activating mutations in CTNNB1 have been found more frequently in the tumors harboring WT1 mutations (6, 7). Recent studies have shown that WT1 inhibits WNT/β-catenin signaling (12–14), but direct evidence for WT1-mediated inhibition of β-catenin signaling pathway is still lacking.

Multiple isoforms of WT1 are produced as a result of different promoter usage, alternative splicing, internal translation, and alternative translational start (4). These isoforms can be largely classified into two major isoforms, WT1(+KTS) and WT1(−KTS), distinguished by the presence or absence of three amino acids (lysine, threonine, and serine (KTS)) between zinc fingers 3 and 4 (15). WT1(−KTS) functions as a transcription factor, whereas WT1(+KTS) has been implicated in post-transcriptional regulation (1, 3). Mice nullizygous for Wt1 fail to initiate metanephric development, demonstrating a pivotal role for WT1 during the early steps of nephrogenesis (16). Additional studies have also demonstrated clear roles for WT1 in later stages of kidney development, such as the formation of nephrons and differentiation of podocytes (17, 18), as well as in homeostasis of adult kidneys (19, 20). Previously, we conducted genome-wide analyses of WT1 target genes to delineate the pathways that WT1 might regulate (21). Here, we report the identification of a novel WT1 target gene that encodes a negative regulator of the WNT/β-catenin signaling pathway.

EXPERIMENTAL PROCEDURES

Cell Lines and Plasmids

UB27 and UD29, U2OS-derived cell lines with tetracycline (tet)³-repressible WT1(−KTS) or WT1(+KTS) expression, respectively, were maintained as described previously (21). HEK293, mouse NIH3T3, and mouse L and L/Wnt3a fibroblast cell lines (ATCC, Manassas, VA) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Wnt3a and control L CM were prepared according to the protocol provided by ATCC. Zebrafish Wnt8a (pCS2P⁺ wnt8 ORF1) and Super8XTOPFlash plasmids were kindly provided by R. Moon (University of Washington, Seattle). The cytomegalovirus-driven full-length human WID (pCMVSport6-WID; GenBank^TM accession number BC050046) and the human AXIN1 cDNA (GenBank^TM accession number BC044648) were purchased from ATCC. AXIN1 was subsequently subcloned into pcDNA3 (Invitrogen). WID deletion mutants were generated by PCR (see supplemental material for primer sequence). Human IDAX cDNA was PCR-amplified and subcloned into pCMVSport6. All PCR-generated cDNAs were verified by sequencing. Human DVL3 cDNA was purchased from OriGene (Rockville, MD) and subcloned into pCMV-3Tag-2A (Stratagene, La Jolla, CA). FLAG-tagged mouse Dvl2 expression plasmid was kindly provided by X. He (Harvard Medical School, Boston).

Generation of Rabbit Polyclonal Antibody against WID

A portion of mouse Wid cDNA containing the C-terminal region (residues 109–317) was amplified by PCR (5′-CCC GGA TCC TAT TGG CCA ATG GTC ATG ACC-3′ and 5′-CCC GAA TTC ACT GAA ACC ACC GGA AGG C-3′) and subcloned into pGEX3X plasmid (Amersham Biosciences) to generate a GST-Wid fusion protein. Purified GST-Wid was used to obtain affinity-purified rabbit polyclonal antibodies (Strategic Biosolutions, Newark, DE).

Quantitative Reverse Transcription-PCR

Total RNAs isolated from UB27 and UD29 cells (+ or − tet) were reverse-transcribed, and the expression levels of WID and GAPDH were analyzed by real time PCR using TaqMan probes (Applied Biosystems, Foster City, CA). Data were analyzed by comparative Ct method using GAPDH as an endogenous control and expressed as fold differences to a reference value of WID in the presence of tet. For the quantification of the zebrafish β-catenin target gene (dkk-1), 1 μl of zebrafish cDNA from MO-injected embryos was used employing the QuantiTect SYBR Green real time PCR kit (Qiagen). All samples were measured as triplicates and normalized to the corresponding amounts of Danio rerio elongation factor-1α cDNA measured within the same plate. Relative expression levels were calculated using the Ct method.

Chromatin Immunoprecipitation Assay

UB27 cells were cultured without tet for 16 h to induce WT1 expression, and chromatin immunoprecipitation (ChIP) was performed as described previously (21). Chromatin immunoprecipitated with either rabbit polyclonal α-WT1 antibody (C19, Santa Cruz Biotechnology, Santa Cruz, CA) or normal rabbit IgG was amplified by PCR using primers corresponding to the indicated enhancer regions (E1–E3) or the N1–N2 regions (see supplemental material for primer sequence). For quantitative ChIP, precipitated chromatin was amplified with the same primers and SYBR Green PCR Master Mix (Applied Biosystems). Data are expressed as fold enrichment to a reference value of IgG control.

Luciferase Reporter Assays for Enhancer Regions of WID

The enhancer regions (E1–E3) were subcloned into a pGL3 plasmid containing a minimal promoter (21) in both sense and antisense orientations. NIH3T3 cells were transfected with 0.4 μg of pcDNA3-WT1(−KTS) or the empty vector along with 0.1 μg of the enhancer-reporter and 0.01 μg of Renilla luciferase (Promega, Madison, WI) plasmids using FuGENE 6 (Roche Applied Science). After 48 h, firefly and Renilla luciferase activities were measured using Dual-Luciferase kit (Promega). Renilla luciferase activity was used to normalize transfection efficiency.

Luciferase Reporter Assays for WNT/β-Catenin Signaling

To measure WNT-dependent T cell factor/β-catenin activity, 0.4 μg of pCMVSport6-WID, IDAX, or the empty vector was transfected into HEK293 cells together with 0.1 μg of Super8XTOPFlash and 0.01 μg of Renilla luciferase plasmids. After 46 h, cells were serum-starved for 6 h, stimulated with either control L or Wnt-3a CM for 6 h, and luciferase activities were measured. To examine the effects of WT1 on the WNT/β-catenin pathway, UB27 cells were transfected with 0.1 μg of Super8XTOPFlash and 0.01 μg of Renilla luciferase plasmids, and the following day, tet was removed to induce WT1 expression for 24 h. Cells were serum-starved for 6 h and stimulated with either control or Wnt3a CM for 6 h, and luciferase activity was measured as described. For the siRNA knockdown of WID in UB27 cells, cells were transfected twice with human WID siRNAs (see supplemental material) or control scrambled siRNA using Lipofectamine 2000 (Invitrogen) along with Super8XTOPFlash and Renilla luciferase plasmids. After 24 h post-transfection, tet was removed to induce WT1 expression for 24 h. Cells were stimulated with either control or Wnt3a CM for 6 h, and luciferase activity was measured.

siRNA Knockdown of Wid

Mouse L fibroblasts were transfected with two different Wid siRNAs (see supplemental material) or control scrambled siRNA using Lipofectamine 2000 (Invitrogen) along with Super8XTOPFlash and Renilla luciferase plasmids. After 46 h, cells were serum-starved for 6 h and stimulated with either control or Wnt-3a CM for 6 h, and luciferase activity was measured. For the analysis of β-catenin, L cells were transfected with siRNA against Wid or control scrambled siRNA using Lipofectamine 2000 (Invitrogen). At 48 h, cells were serum-starved for 6 h and treated with Wnt3a or control CM for 1 h, and total cell lysates were analyzed by immunoblotting with antibodies against β-catenin (Santa Cruz Biotechnology), WID, and actin (Sigma). To visualize nuclear localization of β-catenin, L cells cultured on chambered coverglass (Nunc, Rochester, NY) were transfected with siRNA against Wid or control siRNA and treated with Wnt3a CM as described. Cells were fixed, immunostained with antibodies to WID and β-catenin (BD Biosciences), and visualized with Alexa 594 anti-rabbit IgG and Alexa 488 anti-mouse IgG (Invitrogen). Confocal images were captured with the Axio Observer Z1 inverted laser-scanning microscope (Zeiss, Thornwood, NY) equipped with LSM 5 Live confocal scanner (Zeiss).

Immunoprecipitation and Immunoblot Analyses

HEK293 cells transfected with pcDNA3-AXIN1, FLAG-Dvl2, and pCMVSport6-WID were lysed in buffer (20 mm Tris-HCl (pH 7.5), 150 mm NaCl, 1% Nonidet P-40, 10% glycerol) containing protease inhibitors (Roche Applied Science). The supernatant was immunoprecipitated with either α-WID or α-FLAG antibodies (M2, Sigma) and immunoblotted with α-AXIN (Santa Cruz Biotechnology), α-FLAG, or α-WID antibodies. For detecting the endogenous WID-DVL2 interaction, HEK293 cell lysates were immunoprecipitated with α-WID antibody followed by immunoblotting with α-Dvl2 antibody (kindly provided by Mikhail V. Semenov, Children's Hospital Boston, Boston).

Identification of the DVL-binding Domain of WID

Myc-DVL3, wild type WID, and the WID deletion mutants (WID-ΔDBD, WID-ΔCXXC, and WID-ΔCXXC-DBD) were synthesized in vitro using [³⁵S]methionine (PerkinElmer Life Sciences) and TnT-coupled reticulocyte lysate systems (Promega). Similar amounts of ³⁵S-labeled wild type WID or the deletion mutants and Myc-DVL3 were mixed in binding buffer (20 mm Tris (pH 7.5), 75 mm NaCl, 0.1% Nonidet P-40, 10% glycerol) containing a mixture of protease inhibitors (Roche Applied Science), and the complex was immunoprecipitated with α-Myc antibody (Sigma) and analyzed on 12% SDS-polyacrylamide gel followed by autoradiography.

Immunohistochemistry of Mouse Embryonic Kidney

Mouse embryonic kidneys at 14.5 days post-coitum were isolated, fixed, and embedded in paraffin. After tissue sectioning, slides were heated in 10 mm citrate buffer, pH 6.0, using a microwave oven after deparaffinization and rehydration. Serial sections were incubated anti-WT1 (BD Biosciences) or anti-WID antibodies and developed with Vectastain ABC kit (Vector Laboratories, Burlingame, CA) according to the manufacturer's recommendations. Tissue sections were counterstained with hematoxylin.

Zebrafish

For the analysis of headless phenotypes, ∼50–100 pg of each DNA, zebrafish Wnt8a expression plasmid (pCS2P⁺ wnt8 ORF1) and pCMVSport6-WID or pCMVSport6-IDAX, was injected through the chorion into the blastomere of one-cell wild type zebrafish embryos. For coinjection of two DNA, plasmids were mixed (1:1) prior to injection and injected into the 1-cell stage at 50 ng/μl (50–100 pg/embryo). After 5 days postfertilization, injected embryos were observed and scored for the headless, small eye, or rescued (wild type) phenotypes. The experiment was repeated three to four times. For the analysis of zebrafish kidney, we used the wt1b::GFP transgenic zebrafish line with pronephros-specific green fluorescent protein expression (22). Embryos were raised at 28.5 °C and staged according to Kimmel et al. (23). Morpholino antisense oligonucleotides (see supplemental material for sequence) against either the first splice acceptor site (intron 1 to exon 2) of zebrafish wid pre-mRNA (called splice MO) or the wid start codon (called ATG MO) were dissolved in water (1 mm) and injected into the yolk of embryos in the 1–4-cell stage. A gradient of different injection volumes was tested for each morpholino to find the lowest morpholino amount, which leads to a consistent and reproducible phenotype (∼0.6 nl (0.6 pmol) for the wid splice and 0.3 nl (0.3 pmol) for the wid ATG morpholino, respectively). Embryos treated with 1-phenyl-2-thiourea and anesthetized with Tricaine (0.016%) were embedded in 1% low melting point agarose (Biozym, Hessisch Oldendorf, Germany) with their back facing the bottom of a μ-dish (ibidi GmbH, München, Germany) and photographed using an inverted fluorescence microscope (Axiovert 200, Zeiss).

RESULTS

Identification of a Novel WT1 Target Gene

We previously performed kinetic expression profiling analysis to identify WT1-regulated genes that might play important roles in kidney development (21). This analysis led us to identify a number of genes and novel ESTs induced by WT1(−KTS). Among them, three ESTs (GenBank^TM accession numbers AK001782, BC006428, and BC002490) were derived from a single gene encoding a novel transcript termed CXXC5 in the data base (herein designated WID, see below for detail), which belongs to a recently identified gene family characterized by a CXXC motif (X is any amino acid) (24). We first confirmed that WID is a direct transcriptional target of WT1 by examining the expression of endogenous WID upon induction of WT1(−KTS) or WT1(+KTS) using tet-repressible WT1-expressing cell lines (21). Following WT1(−KTS) expression, levels of the endogenous WID transcript increased rapidly and gradually decreased over time (Fig. 1A). Upon WT1(+KTS) expression, however, only a marginal increase in WID transcript levels was observed at 4 h. We also observed a concomitant increase in the levels of the endogenous WID protein shortly after the induction of WT1(−KTS) using a rabbit polyclonal antibody raised against WID (Fig. 1B).

FIGURE 1.

WID is a WT1(−KTS) target gene. A, quantitative reverse transcription-PCR analysis of WID. WID transcript level was measured by quantitative reverse transcription-PCR at indicated times after removal of tet in UB27 and UD29 cells. Data are the means ± S.D. of three independent experiments. B, Western blot analysis. Total cell lysates were prepared from UB27 cells (− or + tet) and immunoblotted with α-WT1, α-WID, and α-actin antibodies. C, ChIP analysis. UB27 cells were induced to express WT1(−KTS), and the cross-linked chromatin was immunoprecipitated with α-WT1 or rabbit IgG antibodies, followed by PCR amplification with primers corresponding to the WID enhancer regions E1–E3 (black boxes) or to the adjacent regions N1 and N2 (gray boxes) located ∼10 kb upstream of the transcriptional start (+1). Distance between regions is as follows: E1–E2 (420 bp), E2–N1 (1540 bp), N1–E3 (2170 bp), and E3–N2 (1610 bp). Quantitative PCR analysis of ChIP was performed independently using SYBR Green, and the result is presented as the fold increase over IgG (lower panel). D, luciferase (LUC) reporter assay for WID enhancer regions. NIH3T3 cells were cotransfected with plasmids containing the E1, E2, or E3 regions in either sense (SE) or antisense (AS) orientations and with either pcDNA3-WT1(−KTS) or empty vector and Renilla luciferase. Data represent the mean ± S.D. from three independent experiments.

Next, we examined the WID proximal promoter region (∼5 kb upstream from exon 1) to determine whether WT1 might directly activate its transcription, but we found no evidence for WT1-mediated transactivation of the proximal WID promoter in a reporter assay (data not shown). Thus, we searched for conserved regulatory regions by comparing the genomic sequences of WID from human, rat, and mouse and identified three highly conserved regions, designated E1–E3, each containing multiple potential GC-rich WT1-binding sites (25) located ∼10 kb upstream of the transcriptional start site (Fig. 1C and supplemental Fig. S1). To determine whether WT1 was recruited to these sites, ChIP analysis was performed. The results demonstrated that WT1(−KTS) was present in the putative enhancer regions E1 and E3 but not at E2 nor in the adjacent regions designated N1 and N2 (Fig. 1C). An independent quantitative ChIP analysis confirmed this finding (Fig. 1C, lower panel). Each putative enhancer region was then inserted upstream of a minimal promoter-luciferase reporter plasmid and tested in the reporter assay. The E1 region, positioned in either direction, was activated 5–6-fold by WT1(−KTS), whereas the E2 and E3 regions were only modestly activated (Fig. 1E). To further examine WT1-mediated transactivation of the enhancers, we deleted at least three to four putative WT1-binding sites from each of the enhancer regions and tested in the reporter assay. Deletion of multiple potential WT1 sites in the E1 had no effect on the ability of WT1 to activate the reporter, suggesting the existence of other WT1-binding sites (supplemental Fig. S2). This is not surprising given the presence of high GC-rich sequences throughout the enhancer (supplemental Fig. S1). Interestingly, deletion of four putative binding sites in the E3 region resulted in 40% reduction in the WT1-mediated activation, demonstrating that WT1, at least partially, activates E3 enhancer through one or more of the deleted WT1 sites (supplemental Fig. S2). Because WT1 was not recruited to the E2 region by ChIP (Fig. 1, C and D), a modest activation from the E2 region by WT1(−KTS) might be due to an indirect mechanism in a manner similar to the WT1-SF1-mediated activation of the Müllerian-inhibiting substance promoter, which is independent of WT1 binding to the DNA (26). Thus, deletion of three putative WT1-binding sites in E2 had no effect on the reporter activity (supplemental Fig. S2). Collectively, these results suggest that WT1(−KTS) activates the transcription of WID through the upstream enhancer region.

WID Encodes a Negative Regulator of the WNT/β-Catenin Signaling Pathway

Sequence analysis of WID by a BLASTP search identified another CXXC family protein, CXXC4, also known as IDAX (Inhibitor of Dishevelled and Axin) (27), with a high degree of homology to the C terminus of WID (supplemental Fig. S3). The C-terminal regions of WID and IDAX contain the CXXC domain as well as highly conserved sequences flanking the CXXC motif. The N-terminal half of WID does not share significant homology with any other protein and contains an unusually high number of serine, alanine, and glycine residues (comprising 51.6% of the first 167 residues). Because IDAX has been shown to inhibit WNT/β-catenin signaling (27), we tested whether WID might also function as a negative regulator of the WNT pathway by using a Super8XTOPFlash luciferase reporter assay to measure the activation of the WNT/β-catenin pathway (28). Upon transfection of the reporter plasmid into HEK293 cells and treatment with Wnt3a CM, a 10-fold induction of the luciferase activity was observed as compared with the control CM (Fig. 2A). In contrast, cotransfection of WID with the luciferase reporter led to a severe repression (∼5-fold reduction) of Wnt3a-mediated reporter activity. As expected, coexpression of IDAX also led to the inhibition of WNT/β-catenin signaling. We observed a similar inhibition of Wnt3a signaling by WID in NIH3T3 mouse fibroblast, mouse L-fibroblast, and U2OS human osteosarcoma cell lines (data not shown).

FIGURE 2.

WID inhibits the WNT/β-catenin signaling. A, TOPFlash luciferase reporter assay. Super8XTOPFlash plasmid was transfected into HEK293 cells either with pCMV-WID, IDAX, or empty vector. Luciferase activity was measured following treatment with control L (white bar) or Wnt3a CM (black bar) for 6 h. B, luciferase reporter assay after siRNA knockdown of Wid. Super8XTOPFlash plasmid was transfected into mouse L cells either with scrambled (control) or Wid siRNAs, and luciferase activities were measured. Data represent the mean ± S.D. from three independent experiments. Student's t test. *, p value = 0.017; **, p value = 0.019. C, stabilization of β-catenin after Wid depletion. L cells were transfected with either scrambled (control) or Wid siRNAs, treated with L (−) or Wnt3a CM for 1 h, and total cell lysates were immunoblotted with antibodies against β-catenin, WID, and actin. D, nuclear localization of β-catenin after Wid depletion. L cells grown on chambered coverglass were transfected with control or Wid siRNAs and treated with Wnt3a CM as in C. Fixed cells were double-immunostained with antibodies against WID and β-catenin. Images were captured under an oil immersion objective. Scale bar, 20 μm. Lower panel, percentage of nuclear β-catenin from 3 to 4 randomly chosen fields was calculated from three independent experiments. Data represent the mean ± S.D. Student's t test. *, p value = 0.008; **, p value = 0.016. E, inhibition of WNT signaling by WID in zebrafish. Plasmids containing zebrafish wnt8a alone or together with either WID or IDAX were injected into 1-cell zebrafish embryos and scored for the headless, small eye, or wild type phenotypes. Representative phenotypes of zebrafish embryos are shown.

We next examined the effects of Wid depletion on the activation of the WNT/β-catenin pathway using the Super8XTOPFlash reporter assay. Suppression of endogenous Wid in mouse L fibroblasts using two different small interfering RNAs (siRNA) resulted in a significantly higher level of the Wnt3a-mediated luciferase activity than the control siRNA-transfected cells (Fig. 2B). To further study the inhibitory effect of WID on the WNT/β-catenin signaling pathway, we examined the level of β-catenin after reducing endogenous Wid levels by siRNAs. In the absence of Wnt3a, suppression of Wid led to a slight increase in β-catenin levels compared with the control siRNA-transfected cells (Fig. 2C, compare lane 1 with lanes 3 and 5). In the presence of Wnt3a, depletion of Wid resulted in a greater accumulation of β-catenin compared with the control (Fig. 2C, compare lanes 2, 4, and 6). This was further confirmed by confocal microscopy, which showed about 2-fold increase in the nuclear localization of β-catenin in Wnt3a-treated Wid depleted cells compared with the control (Fig. 2D). Together, these observations clearly demonstrate the role of WID as the negative regulator of the WNT/β-catenin signaling pathway.

WID Is a Negative Regulator of WNT Signaling in Zebrafish

Next, we examined whether WID inhibits WNT signaling in vivo. To this end, we used a previously described zebrafish model in which ectopic expression of zebrafish Wnt8 cDNA driven from a constitutive promoter causes either a headless or a small eye phenotype (29). When zebrafish Wnt8 cDNA was injected into fish embryos, approximately two-thirds of the injected embryos displayed either the headless (∼60%) or the small eye (∼17%) phenotype (Fig. 2E). Coinjection of Wnt8 and WID cDNAs, however, resulted in less than 5% of the embryos with the headless phenotype, and ∼70% of the embryos displayed normal head development, demonstrating that WID functions as a negative regulator of the WNT pathway in zebrafish. As expected, coinjection of IDAX also reversed the Wnt8-induced headless phenotype but less efficiently than WID (Fig. 2E).

WT1 Inhibits WNT/β-Catenin Signaling Pathway via WID

Because WT1(−KTS) activates endogenous WID expression (Fig. 1), a potent inhibitor of WNT/β-catenin signaling, we tested whether WT1(−KTS) expression alone can inhibit WNT/β-catenin signaling. To this end, we transfected the Super8XTOPFlash plasmid into the WT1(−KTS)-inducible cell line and examined the reporter activity with or without WT1 expression. In the absence of WT1(−KTS) expression (+tet), Wnt3a activated the luciferase reporter to ∼5-fold (Fig. 3A). In contrast, expression of WT1(−KTS) (−tet) resulted in a marked decrease (∼2-fold) in the Wnt3a-mediated reporter activity, demonstrating that WT1(−KTS) can negatively regulate the WNT/β-catenin signaling. Reduction of WID by siRNA following WT1(−KTS) expression resulted in a partial loss of the WT1-mediated inhibition of WNT/β-catenin signaling as compared with the control siRNA-transfected cells (Fig. 3B), suggesting that WT1-induced repression of WNT/β-catenin signaling is at least partially mediated through WID. Western blot analysis confirmed the induction of WT1 and reduction of WID expression (Fig. 3B, bottom panel).

FIGURE 3.

WT1 inhibits the WNT/β-catenin signaling. A, luciferase reporter assay. Super8XTOPFlash and Renilla luciferase plasmids were cotransfected into UB27 cells, and luciferase activity was measured with (−tet) or without (+tet) WT1 expression. B, siRNA knockdown of WID. Super8XTOPFlash and Renilla luciferase plasmids were transfected into UB27 cells with either scrambled (control) or WID siRNA, and WT1 expression was induced (−tet) or uninduced (+tet), and luciferase activity was measured. Data represent the mean ± S.D. from three independent experiments. Cell lysates were immunoblotted with antibodies against WT1, WID, and actin.

Physical Interaction of WID and DVL

Activation of WNT/β-catenin signaling is mediated by the direct binding of Dishevelled (DVL) to a complex containing AXIN-APC-GSK3β (30). A previous study showed that IDAX might inhibit WNT/β-catenin signaling by directly interacting with DVL (27). To determine whether WID can physically interact with DVL, we introduced FLAG-tagged DVL2, and WID into HEK293 cells followed by immunoprecipitation with an anti-WID antibody. As shown in Fig. 4A, WID readily formed a complex with DVL2 but not with AXIN. A reciprocal immunoprecipitation with an anti-FLAG antibody also demonstrated the interaction of DVL2 and WID. We further confirmed the interaction between endogenous WID and DVL2 in HEK293 cells by immunoprecipitating the endogenous WID, followed by immunoblotting with an anti-DVL2 antibody (Fig. 4B).

FIGURE 4.

WID interacts with DVL. A, interaction of WID and Dvl2. HEK293 cell lysate transfected with WID, FLAG-Dvl2, and AXIN1 was immunoprecipitated (IP) with α-WID or α-FLAG antibody and immunoblotted (IB) with α-FLAG, α-AXIN, or α-WID antibodies. B, interaction of endogenous WID and DVL2. HEK293 cell lysate was immunoprecipitated with α-WID antibody and immunoblotted with α-Dvl2 antibody. C, in vitro binding assay. In vitro synthesized ³⁵S-labeled WID (wild type or deletion mutants) was incubated with ³⁵S-labeled Myc-DVL3, immunoprecipitated with α-Myc antibody, and analyzed by SDS-PAGE and autoradiography. D, DBD and the CXXC are essential domains. HEK293 cells were cotransfected with Super8XTOPFlash plasmid and either with wild type WID or WID deletion mutants, and luciferase activity was measured after control or Wnt3a CM treatment. Data represent the mean ± S.D. from three independent experiments. E, Western blot analysis. HEK293 cells were transfected with wild type FLAG-WID or WID deletion mutants, and cell lysates were analyzed by immunoblotting with α-FLAG and α-actin (loading control) antibody. Because of the lower expression of wild type WID, more cell lysates from the empty vector or wild type WID-transfected cells were loaded.

Identification of the DVL-binding Domain of WID

A “KTXXXI” motif (X is any amino acid) within IDAX has been identified as the minimal DVL binding domain (DBD) (31). The KTXXXI motif, as well as the immediate surrounding residues (herein designated as the DBD), is perfectly conserved in WID (supplemental Fig. S3). To examine whether the KTXXXI motif is required for binding to DVL, we deleted the putative DBD domain of WID (WID-ΔDBD, residues 284–296) and tested the ability of the WID-ΔDBD mutant to bind DVL in vitro. As controls, we also generated two additional WID deletion mutants, one lacking the CXXC motif (WID-ΔCXXC, residues 263–281) and the other lacking both the CXXC and the DBD motifs (WID-ΔCXXC-DBD, residues 263–296). Similar amounts of in vitro ³⁵S-labeled wild type WID or the deletion mutants were mixed with ³⁵S-labeled Myc-DVL3 followed by immunoprecipitation with α-Myc antibody. Compared with the wild type WID, the WID-DVL3 interaction was markedly reduced with the deletion of DBD, but the ΔDBD mutant still retained some binding to DVL (Fig. 4C, compare lanes 6 and 7). In contrast, deletion of the CXXC motif had no effect on the WID-DVL3 interaction. Interestingly, deletion of both DBD and CXXC domains (ΔCXXC-DBD) resulted in a nearly complete loss of the WID-DVL3 interaction (Fig. 4C, lane 9), demonstrating that both domains are required for the full interaction with DVL.

DBD and the CXXC Domains Are Essential for WID Function

We next examined the ability of the WID deletion mutants to inhibit WNT/β-catenin signaling using the TOPFlash reporter assay. As expected, expression of wild type WID led to a significant inhibition of Wnt3a-stimulated TOPFlash luciferase reporter (Fig. 4D). However, expression of WID-ΔDBD mutant failed to inhibit the Wnt3a-stimulated TOPFlash luciferase activity, indicating that the WID-DVL interaction is critical for the inhibition of the WNT signaling pathway. Surprisingly, the WID-ΔCXXC mutant completely lost the ability to inhibit WNT signaling (Fig. 4D), despite retaining the ability to bind to DVL. In fact, both WID-ΔCXXC and WID-ΔDBD-CXXC mutants resulted in a stronger WNT-reporter response than either the control (empty vector) or the WID-ΔDBD mutant, suggesting a possible dominant-negative effect on the endogenous WID. These results demonstrate that the WID-DVL interaction is necessary but not sufficient to inhibit WNT signaling and further highlight the importance of the CXXC domain in addition to the DBD. Western blotting demonstrated that the WID deletion mutants were expressed at a higher level than wild type WID (Fig. 4E), indicating that the inability of the WID mutants to inhibit WNT signaling was not due to a difference in expression levels.

Wt1 and Wid Are Coexpressed in Podocytes of Developing Kidney

Expression of WT1 in kidney is restricted to metanephric mesenchyme, renal vesicles, the proximal portion of S-shaped bodies, and finally podocytes, which are specialized visceral epithelial cells that form the glomerular filtration barrier in nephrons (1). To determine whether Wt1 and Wid are coexpressed in developing kidney, we examined the expression of Wt1 and Wid in the developing mouse kidney. In the developing kidney, Wt1 expression can be seen in the condensed mesenchyme, proximal region of S-shaped bodies and podocytes of maturing glomeruli (Fig. 5A, arrows, left panel). Immunostaining of the serial section of the kidney revealed that Wid is coexpressed with Wt1 in the proximal region of S-shaped bodies and the developing podocytes (Fig. 5A, arrows, right panel), but Wid expression was also detected in non-Wt1-expressing cells of the tubules (Fig. 5A, white arrowheads).

FIGURE 5.

Depletion of zebrafish wid results in defective kidney development. A, colocalization of Wt1 and Wid in podocytes of developing kidney. Serial sections of mouse embryonic kidney at 14.5 days post-coitum were immunostained with α-WT1 or α-WID antibodies. Arrows indicate pre-podocytes of S-shaped bodies and podocytes of maturing glomeruli, which are positive for both Wt1 and Wid. White arrowheads indicate Wt1-negative tubules that express Wid. Scale bars, 100 μm. B, representative images of the pronephros of wt1b::GFP transgenic zebrafish embryos injected with either the splice (Splice MO) or the start codon (ATG MO) wid antisense morpholino oligonucleotides are shown. Cystic glomeruli are clearly visible in the wid morpholino-injected animals but not in the control. Arrowheads indicate glomeruli, and arrows indicate pronephric tubules, and asterisks indicate exocrine pancreas. C, quantitative reverse transcription-PCR analysis of wid and dkk-1. Total RNAs were collected from 10 pooled control morpholino or wid morpholino-injected embryos at the indicated stage. For comparison, expression in 30-h-old control embryos was set to 1. Data are the mean ± S.D. of three independent experiments.

Inactivation of Wid in Zebrafish Leads to Kidney Defect

We searched for WID orthologs in the NCBI data base and identified a hypothetical protein in zebrafish (GenBank^TM accession number XP_686158) that shares 55% identity and 61% similarity with human WID (supplemental Fig. S4). Given such high evolutionary conservation, we examined whether zebrafish wid might function in kidney development. To this end, we utilized a transgenic zebrafish expressing green fluorescent protein under the zebrafish wt1b promoter, wt1b::GFP (22), which expresses green fluorescent protein in the glomerulus and the pronephric tubule of developing pronephros. A splice antisense MO directed against the first splice acceptor site of wid (supplemental Fig. S5A) or an antisense MO directed against the start codon of wid (ATG MO) was used to determine the effects of ablating wid expression in wt1b::GFP embryos. At 48 h post-fertilization, control MO-injected embryos displayed normal pronephros development, forming two glomeruli (arrowheads) and the tubules that extend laterally (arrows) (Fig. 5B, control panel). Remarkably, injection of either the splice MO or the ATG MO against wid resulted in malformed pronephros with large cysts in the glomerular-tubular region in ∼80% of the injected embryos at 48 h post-fertilization (Fig. 5B, see also Table 1). These results demonstrate that normal pronephric development in zebrafish requires wid expression. The specificity of the splice MO was demonstrated by the dose-dependent decrease in the spliced wid mRNA by reverse transcription-PCR analysis (supplemental Fig. S5B), and the use of two morpholino oligonucleotides with unrelated sequences further demonstrates the specificity of the observed phenotypes.

TABLE 1.

Phenotypic alterations in wid morpholino-injected embryos

Morpholino	na	Pronephric cystsb	Curved body
Uninjectedc	40	0	0
wid splice	37	32 (86%)	34 (92%)
wid ATG	32	23 (72%)	30 (94%)

^a n indicates number of embryos.

^b Evaluation was 2 days after injection using a fluorescence stereomicroscope.

^c No phenotypic differences could be observed between uninjected and control morpholino-injected embryos.

We analyzed the expression levels of several known zebrafish β-catenin target genes after the suppression of wid and compared it with the control zebrafish embryos. Among the target genes examined, we found consistent up-regulation (2–3-fold) of dkk-1 mRNA after wid depletion (Fig. 5C). Treatment with wid MO was very effective because wid transcripts were hardly detectable.

DISCUSSION

In Wilms tumors, loss of WT1 often occurs concomitantly with nuclear localization of β-catenin resulting from gain-of-function (activating) mutations within CTNNB1 itself (7) or by other as yet unknown mechanisms (32). In this study, we demonstrate that a novel target of WT1, WID, functions to negatively regulate WNT/β-catenin signaling. Recently, a newly described tumor suppressor gene, WTX, was found to be inactivated in a subset of Wilms tumors (8–10). The WTX protein forms a complex with AXIN-APC-GSK3β and actively promotes the degradation of β-catenin (11). Collectively, these observations suggest that activation of the WNT/β-catenin pathway may be a common mechanism underlying the formation of Wilms tumors (33). However, preliminary examination of expression profiling data base for any alteration of WID in Wilms tumor and colorectal cancer did not reveal any significant changes in WID transcript levels (data not shown).

Consistent with our findings reported in this study, exogenous expression of WT1 in breast cancer cell line resulted in reduced β-catenin levels (14), and deletion of Wt1 in Sertoli cells of the developing testis resulted in increased β-catenin levels as a result of β-catenin stabilization (12). Another recent study also demonstrated that WT1 inhibits WNT/β-catenin signaling by competing with limited amount of coactivator CREB-binding protein, which is also required by the T cell factor-β-catenin transcription complex (13). However, a direct mechanism by which WT1 regulates β-catenin levels or activity has not been demonstrated. In this study, our results demonstrate a direct role for WT1 in the negative regulation of the WNT/β-catenin signaling pathway by activating the transcription of WID, which directly binds to DVL and prevents WNT-mediated stabilization of β-catenin.

During the course of our study, a recent report identified WID/CXXC5 as the BMP4-induced inhibitor of WNT/β-catenin signaling in neural stem cells (34), but how WID/CXXC5 inhibits WNT signaling was not demonstrated. In the same study, it was also reported that exogenously expressed WID/CXXC5 is predominantly nuclear. Our results are consistent with the role of WID in the inhibition of WNT/β-catenin signaling; however, our data clearly demonstrate that endogenous WID is predominantly cytoplasmic (Fig. 2D). Curiously, we observed a small amount of WID translocated into the nucleus upon Wnt3a stimulation (Fig. 2D and data not shown). Our findings further demonstrate that WID-DVL interaction requires the CXXC and DBD domains of WID and that both domains are essential for disabling DVL, which is responsible for the stabilization of β-catenin (30). Interestingly, these results further imply that the inhibition of WNT/β-catenin signaling by WID requires more than just binding to DVL because the WID-ΔCXXC mutant, which retained the ability to bind DVL, was unable to inhibit the WNT/β-catenin signaling (Fig. 4, C and D). The precise mechanism by which the CXXC domain functions to negatively regulate the WNT pathway will require further study. We also note that, by virtue of its ability to interact with DVL, WID might also regulate β-catenin-independent WNT pathways, such as the planar cell polarity pathway that regulates cell polarity and movement or the Ca²⁺-related signaling pathway that involves cell adhesion (35).

WID orthologs are present in all mammals and in lower vertebrates such as zebrafish (supplemental Fig. S4). The conservation is especially high in the last 110 residues (>90% identity in all species). This region spans the CXXC and DBD motifs, which are essential for the WID-DVL interaction and for the WID-mediated inhibition of the WNT/β-catenin signaling pathway. The results from this study further demonstrated a potential role for WID in kidney development as ablation of zebrafish wid disrupted normal embryonic kidney development and resulted in the formation of large cysts in the glomerular-tubular regions. In this regard, it is interesting to note that activated WNT signaling also causes cystic kidney formation in the zebrafish (36) and the mouse (37). These results suggest that the physiological function of WID in nephrogenesis might be conserved in mammals.