Identification of Novel Wilms' Tumor Suppressor Gene Target Genes Implicated in Kidney Development*

The Wilms' tumor suppressor gene (WT1) encodes a zinc finger transcription factor that is vital during development of several organs including metanephric kidneys. Despite the critical regulatory role of WT1, the pathways and mechanisms by which WT1 orchestrates development remain elusive. To identify WT1 target genes, we performed a genome-wide expression profiling analysis in cells expressing inducible WT1. We identified a number of direct WT1 target genes, including the epidermal growth factor (EGF)-family ligands epiregulin and HB-EGF, the chemokine CX3CL1, and the transcription factors SLUG and JUNB. The target genes were validated using quantitative reverse transcriptase-polymerase chain reaction, small interfering RNA knockdowns, chromatin immunoprecipitation, and luciferase reporter analyses. Immunohistochemistry of fetal kidneys confirmed that a number of the WT1 target genes had overlapping expression patterns with the highly restricted spatiotemporal expression of WT1. Finally, using an in vitro embryonic kidney culture assay, we found that the addition of recombinant epiregulin, amphiregulin, CX3CL1, and interleukin-11 significantly enhanced ureteric bud branching morphogenesis. Our genome-wide screen implicates WT1 in the transcriptional regulation of the EGF-family of growth factors as well as the CX3CL1 chemokine during nephrogenesis.

Somatic mutations in the Wilms' tumor suppressor gene WT1 are found in about 10 -15% of Wilms' tumors, a pediatric kidney cancer (for review, see Refs. 1 and 2). Germline mutations in WT1 can lead to severe developmental defects such as in WAGR 4 (Wilms' tumor, aniridia, genitourinary defect, and mental retardation) (3,4), Denys-Drash syndrome (5), and Frasier syndrome patients (6), all of whom suffer from an endstage renal failure as a result of glomerulonephropathy together with improper gonadal development (2). In addition, WAGR and Denys-Drash syndrome patients are highly susceptible to Wilms' tumor. The importance of WT1 during development is further demonstrated in mouse studies where Wt1-null embryos die in utero with agenesis of kidneys, gonads, adrenal glands, and spleen (7)(8)(9). Additional defects can be found in the developing eyes of the Wt1-deficient animals due to increased apoptosis and decreased proliferation of retinal ganglion cells (10). Similar to WAGR patients, Wt1 haploinsufficiency in mouse also leads to a kidney failure due to glomerular dysfunction (11,12), demonstrating the requirement for sustained WT1 expression in renal homeostasis.
WT1 encodes a Kruppel-like zinc finger protein that functions as a transcription factor but may also have a non-transcriptional role (2,(13)(14)(15). Multiple WT1 isoforms can be generated from two alternative splicing events (16) and the use of multiple translation initiation sites (17,18). The first alternative splice event inserts or omits 17 amino acids encoded by exon 5, and the second alternative splice determines the inclusion of three amino acids termed KTS (Lys, Thr, and Ser) between zinc fingers 3 and 4. Although the physiological significance of the first splicing event is perhaps compromised by the fact that mice with deletion of exon 5 develop normally (19), the importance of KTS splicing is clearly demonstrated by the ablation of specific KTS isoforms in the mouse (20). Homozygous animals lacking either the ϩKTS or the ϪKTS isoforms fail to develop functional kidneys and gonads and die soon after birth, demonstrating the requirement of both the ϪKTS and the ϩKTS isoforms of Wt1 to fully develop functional kidneys and gonads. This also clearly demonstrates that there are distinct functions mediated by each isoform that cannot be compensated for by the presence of a single isoform. However, unlike Wt1 Ϫ/Ϫ embryos (7), normal progression of the early steps of metanephric kidney induction and development of the heart, adre-nal gland, and spleen occur in both isoform-specific knock-out mice (20), suggesting that the ϩKTS and the ϪKTS isoforms of Wt1 also share some overlapping functions. The diversity of WT1 proteins is further expanded by an epigenetically regulated variant, designated AWT1, which is transcribed from an alternative promoter located within WT1 intron 1 and encodes amino-terminal truncated proteins (21).

EXPERIMENTAL PROCEDURES
Cell Culture-UB27 cells derived from human osteosarcoma U2OS with tetracycline-repressible expression of WT1(ϪKTS) were maintained as described (35). UD29 cells, also derived from U2OS, expressing WT1(ϩKTS) under the tetracyclinerepressible promoter, were generated as described (35). WT1 expression was induced in UB27 and UD29 cells by three washes with phosphate-buffered saline (Invitrogen) followed by incubation in Dulbecco's modified Eagle's medium without tetracycline. T5A1 cells are human embryonic kidney 293 cells stably transfected with an inducible WT1(ϪKTS) expression vector and were maintained as described (36). WT1 expression was induced with the addition of 1 M cadmium chloride to the media. WiT49 cells (kindly supplied by Dr. Herman Yeger, University of Toronto) were grown in Dulbecco's modified Eagle's medium supplemented with 15% fetal calf serum, 100 l/ml L-glutamine (Invitrogen), 1.1 g/ml insulin (Sigma), and 4 l/liter ␤-mercaptoethanol.
RNA Purification and Microarray Analysis-At indicated times after WT1 induction, total RNA was isolated using RNA STAT-60 (Tel-Test, INC. Friendswood, TX), and further purified with RNeasy kit (Qiagen, Valencia, CA). Double-stranded cDNA synthesized from total RNA was in vitro transcribed, and biotin-labeled cRNA was then fragmented into strands of about 200 bases in length and hybridized to HG133A and HG133B Chips from Affymetrix (Santa Clara, CA). Chips were scanned on an Agilent Gene Array Scanner (Agilent Technologies, Palo Alto, CA). The data were analyzed using the standard MAS 5.0 algorithm by Affymetrix and selected by having consistent change calls in all three time-points; one sample was assayed per time point. The details of the Affymetrix algorithm can be found in the Statistical Algorithms Reference Guide, found online. Routine array quality check was performed, and all arrays were found in the normal range (raw Q values ranged between 2.12 and 2.56, which is below the cutoff of 3.0, and the scaling factor ranged between 2.17 and 3.14). Microarray data have been submitted to GEO data base (accession number GSE5117).
Quantitative Reverse Transcriptase-PCR-To validate the microarray data, qRT-PCR was performed using assay-on-demand TaqMan probes (Applied Biosystems, Foster City, CA). Equal amounts of first-strand cDNA synthesized from total RNA were mixed with each TaqMan probe Mix and TaqMan Universal Master Mix (Applied Biosystems) and amplified using ABI prism 7700 sequence detection system (Applied Biosystems). Data were analyzed by comparative Ct method using glyceraldehyde-3-phosphate dehydrogenase as an endogenous control (37) and expressed as -fold differences to a reference value of each gene's expression in the presence of tetracycline.
Small Interfering RNA (siRNA) Knockdown Assay-500,000 WiT49 cells per well were seeded in 6-well plates 48 h before transfection. 50 -100 nM WT1 SMARTpool or non-targeting control siRNAs (Dharmacon, Lafayette, CO) were combined in serum-free medium with the transfection reagent Dharma-FECT 1 (Dharmacon) following the manufacturer's protocol. The mixture was then added dropwise to the cells in complete WiT49 medium and mixed by gentle swirling. RNA and protein were harvested at 48 h. Experiments were performed in duplicate with similar results. RNA extraction was performed by Tri reagent (Sigma) and DNase I treatment (Ambion, Austin, TX).
Chromatin Immunoprecipitation Assay-ChIP assay was performed following the method of Wells and Farnham (38) with slight modifications. After WT1 induction for 12-14 h by tetracycline withdrawal, formaldehyde (1% of final concentration, Sigma) was added to the culture media for 10 min, and cells were lysed in SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.0). Genomic DNA was fragmented to lengths of 200 -1000 bp by sonication (Branson 450, Danbury, CT), diluted 10-fold with ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl), and subsequently incubated with rabbit polyclonal anti-WT1 antibody (C19, Santa Cruz, Santa Cruz, CA) or anti-rabbit IgG (Santa Cruz) as a negative control overnight at 4°C. Immune complexes pulled down with protein G-agarose were eluted with elution buffer (1% SDS, 0.1 M NaHCO 3 ) and incubated for 4 h at 65°C to reverse the cross-linking. The eluates were treated with proteinase K, and released DNA was purified using Qiaquick PCR purification kit (Qiagen) and amplified by PCR with each primer set corresponding to promoter subregions.
Luciferase Reporter Assays-ChIP-positive promoter fragments were cloned into a promoterless pGL3 plasmid (Promega, Madison, WI). NIH3T3 cells were transfected with 0.5 g of cytomegalovirus-driven WT1(Ϫ or ϩKTS) or the empty vector along with both 0.05 g of cloned promoter-reporter construct and 0.005 g of Renilla luciferase (Promega) using FuGENE 6 (Roche Applied Science). Firefly and Renilla luciferase activities were measured using the Dual Luciferase kit (Promega) and expressed as relative luciferase activity. Cotransfected Renilla luciferase was used to normalize for transfection efficiency.
Embryonic Kidney Organ Culture-Pregnant female Sprague-Dawley rats were purchased from Harlan Inc. (Indianapolis, IN). Embryonic kidney rudiments were isolated from 13.5dpc embryos and cultured on Isopore polycarbonate filters (5 m pore-size, Millipore, Billerica, MA) as described (39). Indicated ( Fig. 5B) concentrations of recombinant AREG, EREG, HB-EGF, CX3CL1, and IL-11 (R&D Systems) were added to the culture medium. After 72 h cultured embryonic kidneys were fixed, incubated with anti-laminin antibody (Sigma), and then incubated with Alexa 594-labeled anti-rabbit IgG (Invitrogen) to visualize ureteric bud branching. The tips of ureteric bud branches were counted by fluorescence microscopy (Leica DMLB, Leica Microsystems, Bannockburn, IL). Animals were handled as described in the approved animal study proposal following the guidelines provided by the National Institutes of Health Animal Research Advisory Committee.

Identification and Validation of WT1 Target Genes-We
previously identified amphiregulin (AREG) as a direct WT1(ϪKTS) target gene using an inducible-WT1 expression cell line UB27, which is derived from a human osteosarcoma cell line U2OS, coupled with microarray analysis (22). Although successful, only 6800 genes were examined with the prototype expression array, and the analysis of a single induction time point resulted in significant noise in the expression data. To this end we decided to perform a comprehensive screen for WT1(ϪKTS) target genes using multiple induction time points. Examining expression profiles at multiple time points permits us to eliminate stochastic changes and significantly reduce the amount of noise associated with microarray experiments. As shown in supplemental Fig. 1, A and B, the presence of tetracycline led to a tight repression of WT1 expression, and removal of tetracycline led to a gradual temporal accumulation of both WT1 transcript and protein.
We decided to restrict our analysis to early time points after WT1 induction to identify genes that are immediately affected by WT1 expression. Using total RNA isolated at 0, 4, 8, and 12 h

Validation of WT1(؊KTS)-induced target genes by qRT-PCR
Total RNA was collected at 0, 4, 8, and 12 h after WT1(ϪKTS) induction in UB27 cells, and expression of the indicated target genes was analyzed by qRT-PCR. Data are the mean Ϯ S.D. of three independent experiments performed in duplicate. Bold text indicates genes that were induced greater than 2-fold by qRT-PCR. NT, not tested. hnRNP, heterogeneous nuclear ribonucleoprotein.  after WT1 induction, we interrogated the Affymetrix Human Genome U133 Genechip, representing more than 39,000 transcripts. In our data analysis, we limited our search to those genes whose expression was altered at all three time points compared with the uninduced sample. Among 39,000 transcripts evaluated, only 23 known genes and 10 ESTs exhibited increased expression at all times upon WT1 induction (Table  1), whereas 10 genes were consistently repressed (supplemental Table S2). WT1 was the highest induced transcript (14-to 21-fold) and previously described WT1(ϪKTS) targets, such as AREG, acidic FGF (FGF-1), and IL-11, were also identified (22). Expression of other previously reported WT1 targets such as HSP70 (40), podocalyxin (30), CSF-1 (25), and BAK (33) were not increased at all three time points. Expression of HSP70, CSF-1, and podocalyxin was only induced after 8 and 12 h of induction, whereas BAK was induced only at the 12-h time point (supplemental Table S1). This suggests that some of the authentic WT1 targets whose expression was not continuously up-regulated or down-regulated may have been overlooked in our analysis but further indicates the high signal to noise ratio in our microarray analysis. Consistent with this view, qRT-PCR analysis confirmed the expression changes in many of the genes tested (Table 1 and supplemental Table S2). About two-thirds of the up-regulated genes (16 of 23, including WT1) and all of the 8 down-regulated genes tested, showed significant expression changes by qRT-PCR that were consistent with the microarray data, although the magnitude of expression changes was variable. In subsequent analyses we focused on the 10 known genes whose expression was increased greater than 2-fold by qRT-PCR (Table 1). The up-regulated EST transcripts and the down-regulated genes were not examined further. Target Gene Regulation by WT1 in Human Kidney Cells-To examine whether the target genes identified in U2OS cells can be regulated by WT1(ϪKTS) in renal cells, we used the T5A1 cell line, a human embryonic kidney cell line with cadmiuminducible WT1(ϪKTS) expression (36). WT1 expression was induced by the addition of cadmium (supplemental Fig. 1C), and the target gene expression was examined by qRT-PCR analysis. Consistent with our previous results, induction of AREG, EREG, HB-EGF, and IL-11 was observed 9 h after WT1(ϪKTS) expression (Fig. 1A). We note that IL-11 expression was highly induced by WT1(ϪKTS) expression, whereas the expression of SLUG was only marginally increased. Expression of IL1RAP, CX3CL1, SCL20A1, JUNB, and TIMP3 remained unchanged. This apparent discrepancy might be due to a modest WT1 induction (5-fold) in T5A1 cells compared with UB27 cells (200-to 300-fold, Table 1) or promoter saturation resulting from leaky WT1 expression in the T5A1 cells (supplemental Fig. 1C).

Inhibition of WT1 Expression in Wilms' Tumor Cell Line WiT49 Results in a Concomitant Decrease in Target Gene
Expression-To examine target gene expression after modulation of endogenous WT1, we used WT1 knockdown with siRNA to further validate our target genes. WiT49 is a Wilms' tumor-derived cell line with a high level of wild type WT1 expression and, thus, provides an ideal environment to study acute loss of WT1 function. Introduction of WT1 siRNAs in WiT49 cells led to a sharp decline in WT1 expression (supple-mental Fig. 1D). Importantly, knockdown of WT1 expression in WiT49 cells led to a significant reduction in most of the identified target genes, with the exception of CX3CL1 and, to a lesser degree, IL-11 and TIMP3, as examined by qRT-PCR analysis (Fig. 1B), indicating that the expression of these genes is regulated by WT1. Of note, AREG expression was most severely affected (90% reduced) by the loss of WT1 expression. These results demonstrate that the WT1-induced genes examined represent physiologically relevant targets with possible roles in nephrogenesis and Wilms' tumorigenesis.
To determine whether the identified target genes are also regulated by the WT1(ϩKTS) isoform, we performed qRT-PCR with the tetracycline-regulated cell line UD29 expressing the WT1(ϩKTS) isoform. As shown in Table 2, most of the genes were not transcriptionally activated throughout all time points by the ϩKTS isoform. The only target gene that showed a sustained induction over time was EREG. Some of the genes were induced briefly at 4 h after WT1(ϩKTS) expression, but the expression level returned rapidly to a basal level. This result suggests that the majority of these target genes are preferentially transactivated by the ϪKTS isoform, but some of these genes may also be transcribed at low levels by WT1(ϩKTS), consistent with the idea that the ϩ and the ϪKTS isoforms might have partially overlapping functions (20). Expression of WT1(ϩKTS) in UD29 cells was at least 4-fold higher than WT1(ϪKTS) in UB27 cells (compare the levels of WT1 expression in Tables 1 and 2 and supplemental Fig. 1B), demonstrating that the ϪKTS target genes were not preferentially activated due to a higher ϪKTS expression.
Recruitment of WT1 to the Promoters of Target Genes in Vivo-WT1(ϪKTS) has been shown to bind two types of DNA sequences: GC-rich or TCC repeats (13). However, in silico prediction of WT1 binding sites in promoters can be difficult due to the high frequency of GC-rich sequences in promoter regions. Thus, we performed ChIP assays to determine whether WT1(ϪKTS) is recruited to the promoters of the identified target genes in vivo. As a control for specificity, we used primers flanking the previously identified WT1 binding sequence (WRE) in the AREG promoter (22) in our ChIP-PCR analysis. As shown in Fig. 2A, we were able to specifically amplify the WRE region in the AREG promoter from chromatin immunoprecipitated with an anti-WT1 antibody, but not with control IgG, thus demonstrating in vivo WT1 recruitment to the promoter. Primers designed to amplify the 5Ј region upstream of the WRE in the AREG promoter did not yield a product, further demonstrating the specificity of our ChIP assay. To test the promoter of each candidate target genes, we designed 3 or 4 sets of primers to amplify ϳ500 bp, covering about 2 kilobases of promoter regions upstream from the transcription start sites. As shown, subregions in the proximal promoters of IL1RAP, EREG, HB-EGF, and SLUG were readily amplified from the WT1-immunoprecipitated chromatin but not from the control antibody (Fig. 2, B and C), demonstrating the recruitment of WT1 to these regions in vivo. In fact, 8 of 10 target genes tested by ChIP analysis demonstrated WT1(ϪKTS) recruitment to the proximal promoter regions

TABLE 2 qRT-PCR analysis of the target genes by WT1(؉KTS)
Total RNA was collected at 0, 4, 8, and 12 h after WT1(ϩKTS) induction in UD29 cells, and expression of indicated target genes was analyzed by qRT-PCR. Data are the mean Ϯ S.D. of three independent experiments performed in duplicate. ( Table 3). A close inspection of the ChIP-positive sequences revealed potential WT1 binding sites such as GC-rich motifs in nearly all promoters (including a WRE-like motif within IL1RAP and SLUG promoters) and TCC repeats in the EREG promoter (Table 3). Multiple regions in the promoters of IL1RAP, JunB, CX3CL1, and SLC20A1 were amplified using WT1 antibody-immunoprecipitated chromatin. However, a more detailed analysis will be required to demonstrate the existence of multiple WT1 binding sites. These results demonstrate that WT1 is recruited to these promoters in vivo and likely results in a direct transcriptional activation. WT1(ϪKTS) was not found in the proximal promoter regions of IL-11 and TIMP3 in the ChIP assay with the primer sets covering about 2 kilobases around the transcription initiation site (data not shown), suggesting the possibility of WT1 binding outside of the amplified region, alternative promoters, or indirect regulation by WT1. Transactivation of Target Gene Promoters by WT1-We next tested the transcriptional activities of the promoters identified by the ChIP analysis in a luciferase promoter-reporter assay. To this end we inserted genomic DNA fragments identified by the ChIP assay into a promoterless luciferase reporter plasmid pGL3. As demonstrated previously (22), the promoter region of AREG identified by ChIP assay was activated about 20-fold by WT1(ϪKTS) and to a lesser extent by WT1(ϩKTS) (Fig. 3). A small 370-bp promoter region of EREG identified by ChIP was activated 6 -7-fold by WT1(ϪKTS), suggesting the presence of a WT1-responsive element in this region, whereas WT1(ϩKTS) resulted in a 3-fold activation (Fig. 3). The genomic fragment of HB-EGF identified by ChIP analysis (region c in Fig. 2C), which lacks the basic transcription initiation elements, was inactive in the promoterless pGL3 plasmid (data not shown). Therefore, we inserted the HB-EGF genomic fragment upstream of the pGL3-C plasmid, which contains the minimal promoter of AREG but lacks the WT1 responsive element (22). The resulting HB-EGF promoter-reporter construct was activated 12-fold by WT1(ϪKTS) (Fig. 3), demonstrating the presence of a strong WT1-responsive element within the small 410-bp region. The proximal promoter regions (up to about 2 kilobases) of SLUG, IL1RAP, CX3CL1, IL-11, and SLC20A1 did not show transcriptional activation by WT1 (Fig.   FIGURE 3. Luciferase promoter-reporter analysis of WT1 target genes. Schematics show ChIP-positive (identified in Fig. 2) proximal promoter regions of AREG, EREG, and SLUG subcloned into a promoterless luciferase pGL3-Basic plasmid and tested in a transient transfection luciferase promoter-reporter assay. The HB-EGF promoter (Ϫ1580 to Ϫ1170), which lacks a transcriptional initiation element, was subcloned into the pGL3-C plasmid containing a minimal promoter (22). Numbering refers to the 5Ј promoter sequence relative to the transcription start (ϩ1). pcDNA3-WT1 (Ϫ or ϩKTS, 0.5 g) or the empty vector (pcDNA3) along with promoter-reporter construct (0.05 g) and Renilla luciferase (0.005 g, Promega) were co-transfected into NIH3T3 cells, and firefly and Renilla luciferase activities were measured at 24 -28 h post-transfection and expressed as relative luciferase activity. Renilla luciferase was used to normalize for transfection efficiency. Data represent the mean Ϯ S.D. from three independent experiments.

TABLE 3 Summary of ChIP and luciferease promoter-reporter assays
ChIP analysis was performed as in Fig. 2, and promoter-reporter assays were done as in Fig. 3. The numbers refer to the promoter sequences relative to the transcriptional start (ϩ1) of each gene. NE, not examined; NT, not tested.

ChIP assay
Luciferase assay

IL1RAP
Ϫ396 to ϩ182 None NE NT NT 3 and Table 3). This could be due to the presence of a weak WT1-responsive element or an absence of an essential co-activator binding site in the tested genomic fragments. Alternatively, other physiological WT1-responsive elements more distal to the proximal promoter regions examined may exist.

Co-expression of WT1 and Novel Target Genes in Rat Embryonic Kidney-WT1
is an important regulator of kidney organogenesis, and its expression during kidney development is highly restricted (41). To determine the potential biological activities of the new WT1 targets, we examined the expression patterns of candidate targets by immunohistochemistry in rat embryonic kidneys. As expected, WT1 expression was detected in the condensing metanephric mesenchyme, the developing glomeruli in the comma-and S-shaped bodies, and the podocytes of the mature glomeruli (Fig. 4B). Interestingly, robust expression of SLUG and IL1RAP can be observed in the condensed mesenchyme and the podocytes of mature glomeruli, overlapping with WT1 expression (Fig. 4, C and D); however, expression of these targets was less restricted than WT1. Expression of CXC3L1 was also detected in the condensed mesenchyme and the podocytes of mature glomerulus, although expression was weaker and also evident in other cell types (Fig. 4E). Jun-B expression was low but detectable in the mesenchyme and the podocytes of glomeruli (Fig. 4F). Expression of IL-11 and TIMP-3 by immunohistochemistry was not detected in the glomerulus, whereas the antibodies to EREG did not provide sat-isfactory results in immunohistochemistry (data not shown). These observations demonstrate that some of the newly identified target genes are co-expressed with WT1 during kidney development, with the expression of SLUG and IL1RAP most closely mirroring the expression of WT1.
Enhancement of Ureteric Bud Branching Morphogenesis in Kidney Organ Culture-In vitro kidney organ culture has been successfully used to examine the involvement of key signaling molecules during kidney development. Because many of the validated WT1 targets encode growth factor/secreted cytokines, we employed the kidney organ culture assay to evaluate their effects on kidney differentiation. We have previously used this assay to demonstrate that AREG, a direct WT1 target gene, can enhance branching morphogenesis of ureteric bud (22). Undifferentiated metanephric kidney rudiments isolated from 13.5-dpc rat embryos were plated onto polycarbonate filters, and kidney differentiation was monitored in the absence or presence of recombinant AREG, EREG, HB-EGF, CX3CL1, or IL-11 (Fig. 5A). As expected, the addition of AREG led to a significant increase in ureteric bud branching compared with the control kidneys grown in the absence of any recombinant growth factors. Similarly, the addition of EREG or CX3CL1 also significantly enhanced ureteric bud branching. The effects of IL-11 were more modest but significant, whereas the addition of HB-EGF did not result in a significant increase in branching morphogenesis. With the exception of HB-EGF, a dosedependent effect on branching morphogenesis was observed with AREG, EREG, CX3CL1, and IL-11 (Fig. 5B). This result demonstrates the specificity of AREG, EREG, and CX3CL1 in promoting ureteric bud branching morphogenesis and further supports their role in renal development.

DISCUSSION
Although microarray analyses have previously been used to identify WT1 target genes (22,(42)(43)(44)(45)(46)(47), this study represents the first genome-wide analysis of WT1 targets. The use of tetracycline repressible expression system to profile expression changes at multiple time points allowed us to eliminate most of the stochastic changes and identify a small number of induced and repressed transcripts. The effects of tetracycline removal on global gene expression (i.e. activation of tetracycline repressor-VP16 transactivator) should be minimal based on our previous microarray analysis with a control cell line (UV9) expressing an empty tetracycline-responsive vector (22), which demonstrated that none of the identified WT1 target genes is up-regulated in UV9 cells. The 23 known genes induced by WT1 (Table 1) can be classified into 4 groups: 1) growth or secreted factors, 2) membrane and signaling proteins, 3) transcription factors, and 4) splicing factors (supplemental Table  S3). Through a series of experiments to validate the array result, we demonstrated members of the EGF family, AREG, EREG, and HB-EGF, the chemokine CX3CL1, and the cytokine IL-11 to be direct target genes of WT1(ϪKTS) that may contribute to ureteric bud branching morphogenesis during metanephric development. Other target genes including IL1RAP, SLUG, JUNB, SLC20A1, and TIMP3 were also identified as likely direct target genes of WT1(ϪKTS) based on qRT-PCR, ChIP and immunohistochemistry analyses. Furthermore, siRNA-medi- ated knockdown of WT1 expression in Wilms' tumor cell line WiT49 resulted in a dramatic decrease in 7 of 10 target genes examined (Fig. 1B), providing strong evidence for the regulation of these genes by WT1. Remarkably, ChIP assays revealed that WT1 was recruited to the proximal promoter regions in 8 of 10 target genes (Table 3), with the promoters of EREG, HB-EGF, and AREG being significantly activated by WT1(ϪKTS) in reporter assays (Fig. 3). Together, these results demonstrate that most of the identified genes are likely direct targets of WT1(ϪKTS), which could have important roles during the development of kidneys and other WT1 expressing tissues. Our multifaceted approach of expression profiling, siRNA knockdown, ChIP, promoter-reporter assay, and functional validation assays could be generalized for identifying physiological target genes of other transcriptional regulators. Recently Davies et al. (48) demonstrated the importance of WT1 in different stages of renal development using the siRNA to knockdown WT1 expression in embryonic kidney organ culture assay. However, for reasons that are unclear, we could not achieve specific knockdown of WT1 or any of the target genes using the siRNA approach in fetal kidney explants. It is clear from mouse studies that WT1 is required throughout kidney development and plays multiple roles depending on the stage of development (7,20). During the initial reciprocal induction period, WT1 is required for the generation of a soluble inductive signal(s) to promote ureteric bud invasion of the mesenchyme and for the recognition of signals from the bud (7). Thus, it is intriguing to note that more than half of the WT1 target genes identified in this study fall into the growth factor and membrane/signaling pathways (supplemental Table S3). Specifically, the EGF family of growth factors, AREG, EREG, and HB-EGF, cytokine IL-11, and chemokine CX3CL1 were identified as the direct WT1 target genes. By using kidney explants, WT1-regulated growth factors EREG, AREG, and CX3CL1 significantly enhanced ureteric bud branching morphogenesis (Fig. 5). A cytokine IL-11 also displayed a modest but significant enhancement on renal branching morphogenesis. The effect of HB-EGF on branching morphogenesis was subtle, but the membrane-bound form of HB-EGF has previously been shown to induce survival of renal epithelial cells and enhance epithelial branching (49,50). The expression of these growth factors can be detected by qRT-PCR in rat embryonic kidneys and increases steadily during renal development (data not shown). Interestingly, Areg-null mice are defective in mammary duct branching morphogenesis (51) but do not show defects in renal differentiation, probably due to a redundancy in EGF signaling during kidney development. Importantly, EGF receptor (EGFR) is expressed in the ureteric bud, and EGFR mutant animals display defects in the collecting ducts of the kidney (52), which are derivatives of ureteric bud branching. However, EGF receptor is not required during the early induction of ureteric bud formation, suggesting that whereas the EGF signaling may contribute toward ureteric bud branching, an as yet unidentified WT1-regulated FIGURE 5. Induction of ureteric bud branching by WT1 targets in rat metanephric kidney organ culture. A, rat metanephric kidney rudiments dissected at 13.5-dpc were cultured in vitro in the absence (control) or presence of indicated recombinant proteins. After 72 h cultured metanephric rudiments were fixed and stained with anti-laminin antibody (Sigma) and Alexa 594-labeled anti-rabbit IgG (Invitrogen) to visualize ureteric bud branching. A representative branching morphogenesis from four independent experiments is shown. B, kidney rudiments were cultured at different doses of recombinant proteins, and the tips of ureteric bud branches were stained as in A and counted under fluorescence microscopy. Data represent the mean number (ϮS.D.) of branch tips from four independent experiments. Student's t test. *, p value Ͻ0.05; **, p value Ͻ0.01. secreted factor(s) is responsible for the initial induction of ureteric bud outgrowth. As well as identifying WT1-mediated transcriptional regulation of EGF family members in this study, our preliminary data also suggest that the imprinted WT1 variant, AWT1 (21), can modulate AREG and EREG expression (data not shown). Thus, a WT1/AWT1 regulatory axis may orchestrate and fine-tune EGF signaling pathways during nephrogenesis.
Another proposed function of WT1 during the initial stage of metanephric development is its anti-apoptotic role in the uninduced metanephric mesenchyme. Wt1-null mesenchyme undergoes apoptosis in the absence of ureteric bud invasion (7), but the precise mechanism by which apoptosis is triggered is not known. One possibility is that the mesenchyme requires signals from the invading ureteric bud to survive and differentiate, which are absent in the Wt1 mutants. However, Wt1-null mesenchyme fails to respond to ureteric bud signals when the two tissues are juxtaposed in vitro, whereas in the same assay the wild type mesenchyme responds readily (7). This strongly implies a cell autonomous role for WT1 in mesenchymal cell survival and differentiation. In this regard, WT1-induced growth factors which we have identified, such as CX3CL1, IL-11, and FGF-1, may confer mesenchymal cell survival and differentiation through autocrine circuits. CX3CL1/Fractalkine inhibits apoptosis in neurons and microglia (53,54), and IL-11 protects skin from UVB-induced apoptosis (55) and from nephrotoxic serum-induced glomerulonephritis (56). The importance of FGF signaling in kidneys has been demonstrated using a transgenic mouse expressing dominant negative FGF receptor (57). The expression of FGF-1 and FGF receptor 1 can be detected in the metanephric mesenchyme during early renal development (58,59), and FGF-1 stimulates ureteric bud branching in organ culture (60). Our study has also identified a number of signaling molecules as WT1 targets, such as IL1RAP, GRB10, SH3-BP5, and other membrane and signaling proteins (supplemental Table S3), that could potentially play important roles to autocrine or paracrine signaling from the ureteric bud for mesenchymal cell survival and differentiation. WT1(ϪKTS) also transcriptionally activates a number of transcription factors. SLUG/SNAI2 is a zinc finger transcriptional repressor that is expressed by cells undergoing epithelial-tomesenchymal transition (61)(62)(63). Our results demonstrate that SLUG is likely to be a direct target of WT1(ϪKTS), as suggested by qRT-PCR, siRNA, and ChIP assays. Moreover, expression of SLUG in embryonic kidney overlaps closely with WT1, with the highest expression present in the condensed mesenchyme and the podocytes of glomeruli (Fig. 4C). SLUG has also been shown to confer resistance to radiation-induced (64) and p53-induced apoptosis (65). This raises an intriguing possibility that the intrinsic anti-apoptotic function of WT1 within the metanephric mesenchyme may also be mediated by the transcriptional induction of SLUG, in conjunction with other signaling mechanisms. Another interesting WT1(ϪKTS) target gene is SOX9, although this gene was not characterized beyond qRT-PCR analysis due to its low level of induction. Recently, using a conditional WT1 knock-out approach, Gao et al. (66) demonstrated that ablation of Wt1 expression in Sertoli cells led to a concomitant loss of Sox9 expression, suggesting that WT1 either directly or indirectly regulates Sox9 expression. Our study supports this in vivo observation and further suggests that WT1 may play a direct role in regulating SOX9 expression.
In sum, our comprehensive expression profiling using an inducible WT1 expression has identified additional bona fide WT1 target genes. Our study represents the first genome-wide analysis of WT1 target genes and alludes to new differentiation and development pathways that are under the control of WT1. In particular, the EGF family of ligands and the chemokine CX3CL1 were shown to be directly regulated by WT1(ϪKTS) and strongly implicated in renal development. Our study implicates new genes and pathways that may be integral to normal nephrogenesis and also aberrant kidney development where WT1 functions are perturbed, as exemplified by Wilms' tumor.