Wilms Tumor Suppressor, WT1, Cooperates with MicroRNA-26a and MicroRNA-101 to Suppress Translation of the Polycomb Protein, EZH2, in Mesenchymal Stem Cells*

Hereditary forms of Wilms arise from developmentally arrested clones of renal progenitor cells with biallelic mutations of WT1; recently, it has been found that Wilms tumors may also be associated with biallelic mutations in DICER1 or DROSHA, crucial for miRNA biogenesis. We have previously shown that a critical role for WT1 during normal nephrogenesis is to suppress transcription of the Polycomb group protein, EZH2, thereby de-repressing genes in the differentiation cascade. Here we show that WT1 also suppresses translation of EZH2. All major WT1 isoforms induce an array of miRNAs, which target the 3′ UTR of EZH2 and other Polycomb-associated transcripts. We show that the WT1(+KTS) isoform binds to the 5′ UTR of EZH2 and interacts directly with the miRNA-containing RISC to enhance post-transcriptional inhibition. These observations suggest a novel mechanism through which WT1 regulates the transition from resting stem cell to activated progenitor cell during nephrogenesis. Our findings also offer a plausible explanation for the fact that Wilms tumors can arise either from loss of WT1 or loss of miRNA processing enzymes.

In 1879, William Osler reported two pediatric patients from Montreal with massive kidney tumors containing bands of muscle-like tissue mixed with epithelial elements (35). Twenty years later, Max Wilms published his celebrated monograph recognizing the malignancy as a unique "mischgeschwulste der Niere" (mixed tumor of the kidney), composed of mixed stromal, epithelial, and undifferentiated mesenchymal cells. This "triphasic" histology suggests that Wilms tumors arise from developmentally arrested stem cells of the metanephric mesenchyme that occasionally exhibit abortive differentiation toward a stromal or an epithelial cell fate. It follows that a disturbance of molecular events governing the transition from renal progenitor cells into these differentiated lineages must be central to the pathogenesis of Wilms tumor.
Wilms tumor (WT) 2 is the most common form of pediatric kidney cancer and affects about 1:10,000 children in North America (1). In a subset of these patients, a germline deletion of the transcription factor, WT1, is accompanied by a somatic mutation of the second WT1 allele, giving rise to clones of incompetent stem cells adjacent to the malignant tumor (2). These "nephrogenic rests" are thought to represent embryonic renal progenitors that have failed to respond to inductive WNT signals during embryogenesis. The pre-malignant cell clusters may persist within the normal kidney (3) until a constitutively activating mutation of the ␤-catenin gene (CTNNB1) (17,18) bypasses the need for a normal WNT signal and drives un-regulated rapid cell growth (4). Wilms tumors retain a chromatin pattern resembling embryonic stem cells in which genes of the differentiation cascade are broadly silenced by tri-methylated lysine 27 residues in the H3 histone associated with their promoter regions (5). We recently showed that the WT1 isoforms lacking the three amino acid insertion lysine-threonine-serine between zinc fingers 3 and 4 (ϪKTS), inhibits transcription of the histone methyltransferase (Enhancer of Zeste Homolog 2, EZH2) that epigenetically silences ␤-catenin gene (CTNNB1) expression in mesenchymal stem cells and facilitates the canonical WNT/␤-catenin pathway response to WNT9b (6).
Only about 20% of Wilms tumors are linked to WT1 mutations (7,8). Recently, it was discovered that a subset of Wilms tumors with normal WT1 alleles are associated with mutations of either DICER1 (9) or DROSHA (10). These genes encode proteins that process microRNAs (miRNAs), implying that loss of either WT1 or miRNA biogenesis can lead to the same Wilms tumor phenotype. We reasoned that the crucial effects of WT1 during development might require both transcriptional and miRNA-mediated post-transcriptional inhibition of EZH2. Here we show that WT1 is evident both in the nucleus and cytoplasm of progenitor cells in embryonic mouse kidney nephrogenic mesenchyme. In mesenchymal stem cells isolated from human amniotic fluid, we show that WT1 induces expression of multiple miRNAs, including miRNA-26a (miR-26a) and miRNA-101 (miR-101). In fibroblasts, we show that the latter two miRNAs suppress EZH2 translation. Furthermore, we show that WT1 binds to the 5Ј untranslated region (UTR) of EZH2 mRNA and cooperates with miRNA-containing RNAinduced silencing complexes (RISCs) to suppress EZH2 translation. These observations suggest a novel mechanism through which the WT1 transcription factor exerts post-transcriptional regulation in cooperation with miRNAs.
Viral Particle Transduction-amMSC were transduced with viral particles expressing the two principal subgroups of human WT1 isoforms containing exon 5 (17 amino acid sequence) with (ϩ/ϩ) or without (ϩ/Ϫ) the KTS insertion in exon 9 or control GFP particles. The WT1 isoforms cDNA were cloned into the pWPI bicistronic vector (Dr. Trono Lab/Addgene). Each plasmid was co-transfected with pMD2.G, the envelope plasmid, and psPAX2, the packaging plasmid, into 293T cells. Viral particles were harvested and used for the transduction. Specific WT1 isoforms are denoted by plus or minus symbols; the first indicating the presence or absence of a 17-amino acid sequence in exon 5 and the second KTS between zinc fingers 3 and 4 status.
Indirect Immunofluorescence Staining-Cryosections (5 m) from embryonic mouse kidneys at 17.5 dpc stage were fixed with acetone at Ϫ20°C for 7 min, incubated in 1% Triton X-100, PBS for 30 min, blocked with normal goat serum (10% in 1% BSA, 0.1% Triton X-100/PBS) for 1 h, and incubated with primary antibody WT1 C19 (sc192, 1/400, Santa Cruz Biotechnology, Inc.) and Alexa Fluor 568-conjugated secondary antibodies (1/1000) at 4°C overnight and room temperature for 1 h, respectively. The sections were also incubated with Dolichos biflorus agglutinin antibody (1/300, Vector Laboratories, Inc., Burlingame, CA) and the nuclei were counterstained with DAPI. Images were obtained with a laser scanning confocal microscope (LSM780) and the ZEN2010 software (Carl Zeiss Canada Ltd., Toronto, ON, Canada) at room temperature and processed by Adobe Photoshop and Illustrator software.
RNA Isolation and Real-time PCR Analysis-RNA, including miRNA, was purified using the miRNeasy mini kit according to the manufacturer's instructions (Qiagen, Toronto, ON, Canada). Reverse transcription for mRNA was done with the iScript cDNA synthesis kit (Bio-Rad), and for miRNA using the miScript II RT kit (Qiagen, Toronto, ON, Canada). Quantitative PCR for reverse transcribed mRNA was performed using with the SsoFast EvaGreen Supermix with Low ROX (Bio-Rad) and specific primer sets. Quantitative PCR for reverse transcribed miRNA was performed using the miScript SYBR Green PCR kit and specific primer assays for miRNA (Qiagen). Both PCR were performed in a LightCycler 480 II (Roche Applied Science, Laval, QC, Canada).
miRNA Expression Profile-RNA was extracted and sent for miRNA microarray and analysis (EXIQON, Vedbaek, Denmark). The samples were labeled using the miRCURY LNA TM microRNA Hi-Power Labeling Kit, Hy3 TM /Hy5 TM , and hybridized on the miRCURY LNA TM microRNA array (7th generation), following a dual color experimental design.
Electromobility Shift Assay-Synthetic oligonucleotides to the 5Ј UTR of EZH2 were labeled by back filling with the Klenow fragment of DNA polymerase I using [␣-32 P]dCTP (3000 Ci/mmol) (PerkinElmer Life Science). WT1 protein was synthesized using a T7-coupled transcription/translation reticulocyte lysate system. The in vitro translated products were incubated with 32 P-labeled probes at room temperature for 15 min in a solution containing 4% Ficoll, 1 mM EDTA (pH 8.0), 10 mM HEPES (pH 7.9), 1 mM dithiothreitol (DTT), and 1 g of poly(dI)⅐poly(dC). For specificity tests, the protein preparation was preincubated with 1 g of anti-WT1 (C-19) antibody for 20 min at room temperature and then incubated with the radiolabeled probe for 20 min at room temperature. For competition experiments, reaction mixtures were preincubated for 15 min at room temperature with unlabeled oligonucleotide prior to the addition of radiolabeled probes. Following the binding step, reaction mixtures were electrophoresed on a 6% polyacrylamide gel (acrylamide/bis-acrylamide ratio, 29:1) in 0.25ϫ TBE buffer (22.25 mM Tris⅐HCl, 22.25 mM boric acid, and 1 mM EDTA) at 90 V at room temperature. The gels were dried and exposed to Kodak X-Omat film at room temperature.
RNA Immunoprecipitation Assay-qPCR-Cells were grown andharvestedusing0.25%trypsin/EDTA(Gibco)followingstandard techniques and resuspended in RIP buffer (150 mM KCl, 25 mM Tris, pH 7.4, 5 mM EDTA, 0.5 mM DTT, 0.5% Nonidet P-40, 100 units/ml of RNase inhibitor and protease inhibitor). Embryonic mouse kidney at 17 dpc stage were flash frozen and pulverized using standard protocol and resuspended in RIP buffer. Sonication was performed on a Sonolab 7.1 (Covaris, Toronto, ON, Canada). The immunoprecipitation was performed using Protein A/G-agarose beads. The following antibodies and titers were used: normal rabbit IgG (2729, 2 g, Cell Signaling), WT1 C19 (sc192, 2 g, Santa Cruz Biotechnology, Inc.), and anti-Ago2/eIF2C2 (ab32381, 2 g, Abcam). Purification of the RNA bound to the RNA-binding proteins was performed using the miRNeasy mini kit according to the manufacturer's instructions (Qiagen). cDNA synthesis was performed as mentioned before. Primers for the qPCR are miScript Primer FIGURE 1. miRNA expression profile. The heat map shows the differential expression pattern of miRNAs in mesenchymal stem cells stably transfected with WT1 (ϩKTS) or WT1(ϪKTS) isoforms relative to the overall mean for both cell lines and an empty vector control. Each row represents a microRNA and its expression level relative to control. Red represents expression below the mean; green represents expression above the mean.
Statistical Analysis-Data are presented as mean Ϯ S.E. of three or more independent results. Statistical significance was assessed using t test, one-way or two-way analysis of variance followed by comparison tests.

WT1 Induces an Array of miRNAs in Mesenchymal Stem
Cells-Because mutations of WT1 or DICER1 lead to a similar Wilms tumor phenotype, we reasoned that WT1 might nor-mally induce key miRNAs in metanephric mesenchyme that are essential for differentiation. We isolated RNA from human mesenchymal stem cells (6) stably transduced with either a WT1(ϪKTS) or a WT1(ϩKTS) isoform. Fig. 1 shows the miRNA heat map, indicating up-regulation (Ͼ2-fold) of 12 miRNAs in the WT1(ϩKTS)-expressing cell line and 28 miRNAs in the WT1(ϪKTS) cell line, compared with the mean for parental and empty vector-transfected controls. Eight of the WT1(ϩKTS) miRNAs overlapped with the WT1(ϪKTS) group, giving a total of 32 miRNAs induced by one isoform or the other. Because our previous studies suggest that down-regulation of EZH2 is a crucial event in de-repressing genes of the differentiation cascade (6), we screened the 3Ј UTR of EZH2 for seed sequences for these 32 miRNAs. Interestingly, 16/32 miRNAs are predicted to target members of the Polycomb complex or associated methyltransferases, of these, 12/16 are predicted to target the EZH2 3Ј UTR (Table 1). These 12 included miR26a and miR101, previously shown to suppress translation of EZH2 in myocytes, lung tissue, and breast epithelia, respectively (12)(13)(14)(15). We decided to focus on these latter   two miRNAs as representative of the putative concerted mechanism by which WT1 suppresses epigenetic silencing in stem cells (6). To validate these two miRNAs, we performed quantitative RT-PCR on mesenchymal stem cells stably transfected with WT1(ϪKTS) or WT1(ϩKTS) isoforms. As seen in Fig. 2, miR-26a and miR-101 levels were significantly increased in both cell lines expressing WT1.
miR-26a and miR-101 Cooperate with WT1 to Suppress EZH2 Translation-To confirm that miR-26a and miR-101 suppress EZH2 translation, we transfected human fibroblasts with an expression plasmid containing the full-length EZH2 mRNA driven by a CMV promoter; at baseline, fibroblasts had no detectable endogenous EZH2 or WT1. In the transfected cells, we confirmed high levels of exogenous EZH2 transcript (Fig. 3A). When fibroblasts were transfected with varying amounts of miR-26a or miR-101 mimics, we noted maximal suppression of EZH2 with 25 nM miR-26a vector and 20 nM miR-101 (Fig. 3B). In previous studies, we noted that the WT1(ϩKTS) isoforms suppress EZH2 protein levels in mesenchymal stem cells (6). If this effect is due to transcriptional inhibition, there should be no effect of WT1(ϩKTS) on the CMV-driven EZH2 expression in fibroblasts. However, when we transiently transfected fibroblasts with WT1(ϩKTS), we noted a 50% reduction (arrows) in exogenous EZH2 mRNA and protein levels (Fig. 4, A and B). When WT1(ϩKTS) was combined with miR-26a or miR-101 mimics (25 nM), EZH2 expression was further suppressed to 10 -15% of baseline (Fig. 4, A  and B).
To confirm that the effect of WT1(ϩKTS) on the exogenous EZH2 transcript is mediated by endogenous miRNAs, we preformed rescue experiments with antago-miRs specific for miR-101 and miR-26a. As seen in Fig. 4C, addition of either antago-miR blocks the effect of WT1 on the EZH2 transcript level.

WT1(ϩKTS) Binds to a Recognition Motif in the EZH2 5Ј UTR-
Although it is known that the WT1(ϪKTS) isoforms regulate transcription of target genes in the nucleus, other investigators have noted that the WT1(ϩKTS) isoform can interact with RNA in the cytoplasm (16 -22). This suggests that WT1(ϩKTS) may participate in the post-transcriptional regulation of targets such as EZH2. We screened the full-length EZH2 transcript for the consensus WT1 recognition motif (23) using MatInspector software. This identified a classical motif, located 5 base pairs upstream of the ATG translational start codon in the EZH2 5Ј UTR (Fig. 5A). To confirm binding to this site, we incubated in vitro-translated WT1(ϩKTS) protein with a radiolabeled oligomer corresponding to the sequence of the 5Ј UTR containing the putative recognition motif. In Fig. 5B, a  FEBRUARY 19, 2016 • VOLUME 291 • NUMBER 8 high molecular weight complex is evident; interaction between the WT1(ϩKTS) protein and oligomer can be competed out with excess cold probe or blocked with WT1-specific antibody (Fig. 5B). To confirm that WT1 binds to EZH2 mRNA in whole cells, we extracted RNA from fibroblasts transfected with a CMV-EZH2 plasmid and incubated this with in vitro-translated WT1(ϩKTS). Following immunoprecipitation with anti-WT1 antibody, we quantified (qRT-PCR) EZH2 mRNA associated with the WT1 protein. As seen in Fig. 5C, the EZH2 transcript was detected in the anti-WT1 precipitate. To confirm that WT1 binds to the endogenous EZH2 transcript, we performed RNA-IP experiments followed by qRT/PCR. As seen in Fig. 5D, anti-WT1 antibody immunoprecipitates EZH2 mRNA in embryonic day E17 mouse kidneys and in a mouse mesonephric kidney cell line (M15) (11).

Cytoplasmic WT1 Protein Is Detected in Progenitor Cells of Mouse Embryonic Kidney Cap Mesenchyme-If WT1(ϩKTS)
interacts with cytoplasmic EZH2 mRNA during normal kidney development, it should be detectable in the cytoplasm of progenitor cells in the cap mesenchyme of embryonic kidney. To confirm this, we performed immunofluorescent studies of embryonic day E17.5 mouse kidneys. WT1 protein was detected in both the nuclear and cytoplasmic compartments of cells in the cap mesenchyme (Fig. 6). Fig. 4, we noted that WT1(ϩKTS) and either miR26a or miR101 had a stronger inhibitory effect on fibroblast CMV-driven EZH2 expression than either molecule alone. We considered the possibility that this might involve direct interactions between WT1 at the EZH2 5Ј UTR and the  2 and 4). This complex disappears in the presence of excess cold probe (lane 3) or specific antibody against WT1 (lane 5). The high molecular weight complex formed remains intact with IgG antibody (lane 6). C, mRNA associated with exogenous WT1 protein was immunoprecipitated from fibroblasts transfected with EZH2 plasmid (RIP) and EZH2 RNA levels were quantified by qRT-PCR; WT1-associated EZH2 mRNA levels are expressed as percent of total mRNA input. D, endogenous EZH2 mRNA was immunoprecipitated with anti-WT1 antibody from a mouse mesonephric kidney cell line (M15 cells, black bars) and from mouse embryonic E17 kidneys (gray bars) and quantified by qRT-PCR. Unpaired t test, p Ͻ 0.0001 (****). microRNA-containing RISC at the 3Ј UTR. We lysed WT1(ϩKTS)-expressing amMSCs (or controls) and performed immunoprecipitation assays using antibodies specific for WT1, AGO2, DICER1, or IgG control (Fig. 7A). WT1 was precipitated with each member of the RISC. Conversely, when AGO2 or DICER1 were immunoprecipitated from the lysate, we were able to detect WT1 protein (Fig. 7A). To confirm that WT1 associates with the miRNA-containing RISC, we immunoprecipitated WT1 or AGO2 from cell lysates and screened for the presence of miR-26a, miR-101, and the EZH2 transcript by qRT-PCR. As seen in Fig. 7B, both miR-26a and miR-101 (but not miR-93, an unrelated miRNA lacking seed sequences in the transcript, not shown) were detectable in the WT1 and AGO2 precipitates. Similarly, we detected the EZH2 transcript by qRT-PCR in both anti-WT1 and anti-AGO2 precipitates from mesenchymal stem cells transfected with WT1(ϩKTS). To confirm that the association of these molecules was not due to their common interaction with EZH2 mRNA, we treated the cell lysates with RNase A after immunoprecipitating with anti-DICER1, WT1, or AGO2 antibodies. Although full-length EZH2 mRNA was no longer detectable by RT-PCR, we were still able to detect WT1 in association with each of its putative RISC partners (Fig. 7C).

Discussion
Mammalian kidneys arise from stem cells of the intermediate mesoderm. A subset of these cells begin to express WT1 and then acquire properties of nephron progenitors capable of responding to a canonical WNT signal; another subset of WT1(Ϫ) cells form the epithelial nephric duct and its caudal offshoot, the ureteric bud. As the ureteric bud begins to arborize, nearby nephron progenitors form a cap of tightly clustered cells around the tip of each ureteric bud branch these cells express WT1. As we show here, WT1 can be visualized both in the nucleus and cytoplasm of nephron progenitors in cap mesenchyme. In response to WNT9b released by the ureteric bud, nephron progenitors undergo a dramatic mesenchyme-to-epithelium transition and proceed to form all segments of the nephron.
In the nucleus of renal progenitors, we previously found that the WT1(ϪKTS) isoform suppresses transcription of a Polycomb histone methyltransferase, EZH2 (6). This event is likely to be essential if the differentiation cascade is to be released by inductive WNT signals from ureteric bud. Here we show that transcriptional inhibition of EZH2 is paralleled by translational inhibition in the cytoplasm by the WT1(ϩKTS) isoform. WT1 was co-immunoprecipitated with endogenous EZH2 mRNA from embryonic kidney and a mesonephric cell line.
These findings support the observations of Hastie and coworkers (24), identifying a specific role for the WT1(ϩKTS) isoform in regulating RNA. We identified four classical WT1 recognition motifs in the 5Ј untranslated end of the EZH2 transcript and studied one of these in detail, showing that it binds WT1(ϩKTS) protein with high affinity. All four putative WT1 binding sites are positioned amid sequences recognizing proteins involved in mRNA translation. Thus WT1(ϩKTS) binding to the EZH2 5Ј UTR is likely to disturb the translation initiation complex assembly. Exogenous EZH2 expression from a CMV-driven plasmid was suppressed in vitro by WT1(ϩKTS).
Translation of EZH2 is also known to be controlled by specific microRNAs directed to seed sequences in its 3Ј UTR. Two such miRNAs (miR-101 and miR-26a) suppress EZH2 translation in other cellular settings (myoblasts and cancer cells) (12)(13)(14). Interestingly, we observed that both major WT1 isoforms induce expression (Ͼ2fold) of an overlapping array of 32 miRNAs in mesenchymal stem cells. These included miR-101 and miR-26a. About half of the 32 miRNAs have putative recognition motifs in the 5Ј end of many other Polycomb protein transcripts. This suggests a concerted inhibitory mechanism by which WT1 isoforms broadly suppress expression of many Polycomb complex proteins that epigenetically silence differentiation cascade genes. In this study, we show that both WT1(ϩKTS) and exogenous miRNAs inhibit EZH2 in human fibroblasts. Remarkably, we found that addition of miRNA antagonists completely block the effects of WT1. Thus, cytoplasmic WT1(ϩKTS) bound to the 5Ј UTR appears to cooperate with endogenous RISC-associated miRNAs bound at the 3Ј UTR to disrupt translation.
miRNAs are a class of single-stranded non-coding RNA molecules containing 22 nucleotides (25,26). Primary miRNAs (pri-miRNA) are transcribed in the nucleus by RNA polymerase II and processed by the DROSHA-DGCR8 complex into a stem-loop structure called the precursor miRNA (pre-miRNA) (27). Exportin-5 (XPO-5) transports the pre-miRNA out of the nucleus (28). In the cytoplasm, DICER1 processes the RNA molecule into the miRNA duplex (29). The strands are separated, the mature miRNA is recognized and bound by an AGO protein at the 3Ј end and is integrated into the RNA-induced silencing complex (30). Through complementarity, the miRNA binds the 3Ј UTR of the target messenger RNA (mRNA) (31). Perfect base pair match leads to degradation of the mRNA, whereas imperfect pairing will lead to sequestration of the mRNA and inhibition of translation (32). miRNAs play a crucial role in the regulation of mammalian development, their expression is tissue-specific, and tightly regulated spatial-temporally (33,34).
Our immunoprecipitation studies indicate direct interaction between the WT1(ϩKTS) protein and DICER1/DROSHA (members of the RISC complex that mediates post-transcriptional mRNA decay and mRNA translation). Furthermore, immunoprecipitates containing WT1 and DICER/DROSHA also contained miR26a and miR101. Association between WT1 FIGURE 8. Schematic model of WT1 post-transcriptional regulation of EZH2. A, WT1(ϩKTS) (red) binds the 5Ј UTR of the EZH2 transcript and the miRNA(green)-containing RISC (gray) binds to the 3Ј UTR seed sequence (green) to block access of the translation machinery. B, in the absence of WT1(ϩKTS) and miRNA-containing RISC, the 40S ribosome and translation complexes bind to the EZH2 transcript 5Ј UTR and promote protein translation. and the RISC was unaffected by prior experimental RNase degradation of the EZH2 transcript. These observations prompt a model, shown as a schematic in Fig. 8, in which WT1(ϩKTS) binds to motifs in the EZH2 transcript 5Ј UTR, which displace translation factors and inhibit translation; WT1 also interacts with RISC proteins targeted to the EZH2 3Ј UTR by specific miRNAs and enhances their effects on transcript stability and protein translation.
In hereditary forms of Wilms tumor, biallelic loss of WT1 is thought to preclude the WNT response to WNT signals, resulting in a genetically unstable, developmentally arrested clone of progenitor cells; a second genetic "hit" then drives malignant transformation. Our observations demonstrate that the role of WT1 in de-repressing stem cell genes are, in part, dependent on specific microRNAs. We propose that this may explain why Wilms tumors may also arise from mutation of genes in miRNA processing.