The WT1 gene encodes a zinc finger transcription factor important for normal kidney development. WT1 is a suppressor for Wilms tumour development and an oncogene for diverse malignant tumours. We recently established cell lines from primary Wilms tumours with different WT1 mutations. To investigate the function of mutant WT1 proteins, we performed WT1 knockdown experiments in cell lines with a frameshift/extension (p.V432fsX87 = Wilms3) and a stop mutation (p.P362X = Wilms2) of WT1, followed by genome-wide gene expression analysis. We also expressed wild-type and mutant WT1 proteins in human mesenchymal stem cells and established gene expression profiles. A detailed analysis of gene expression data enabled us to classify the WT1 mutations as gain-of-function mutations. The mutant WT1Wilms2 and WT1Wilms3 proteins acquired an ability to modulate the expression of a highly significant number of genes from the G2/M phase of the cell cycle, and WT1 knockdown experiments showed that they are required for Wilms tumour cell proliferation. p53 negatively regulates the activity of a large number of these genes that are also part of a core proliferation cluster in diverse human cancers. Our data strongly suggest that mutant WT1 proteins facilitate expression of these cell cycle genes by antagonizing transcriptional repression mediated by p53. We show that mutant WT1 can physically interact with p53. Together the findings show for the first time that mutant WT1 proteins have a gain-of-function and act as oncogenes for Wilms tumour development by regulating Wilms tumour cell proliferation.

## INTRODUCTION

Wilms tumour is a paediatric kidney cancer affecting 1/10 000 children a year. The first protein to be associated with WT development is encoded by the gene WT1 located on chromosome 11p13 (1,2). WT1 is mutated in ∼15–20% of all WT and is an important factor for normal kidney development (3). The gene encodes a protein of 52–54 kDa with exons 7 to 10 encoding four C2-H2 zinc fingers (ZFs) of the Krüppel type that bind DNA and RNA. The first exons encode a proline–glutamine (Pro/Gln)-rich domain that contains a putative RNA recognition motif and is involved in transcriptional repression and activation, dimerization and nuclear localization (47).

Alternative splicing results in four major isoforms, the first leading to inclusion/exclusion of exon 5 and the second to inclusion/exclusion of three amino acids, lysine, threonine and serine (KTS) after exon 9. It was first shown that WT1 lacking KTS binds to a GC-rich EGR1 consensus sequence, as well as to an unrelated TCC repeat motif (8,9). The inclusion of KTS between ZF3 and 4 significantly reduces the DNA-binding affinity of WT1 and the +KTS isoform binds to other DNA targets (10). There is also evidence that both WT1 isoforms + and −KTS are involved in post-transcriptional processes (11). The +KTS isoform co-localizes and co-immunoprecipitates with splice factors, and WT1 can modify splicing by interacting with the splice factor U2AF65 (12,13). Using the RNA selection method SELEX and WT1 ZF constructs, three RNA aptamers that are recognized by WT1 were identified (14). Three of four ZFs were necessary, and deletion of ZF1 resulted in reduced and insertion of KTS abolished binding for the RNA targets (14,15). Using these RNA aptamers, Weiss and Romaniuk showed that ZF2 and 3 are necessary for RNA binding (16).

WT1 was found in poly(A)+ nuclear RNP from foetal kidneys (17) and in mRNP particles in K562 cells, further pointing to a role in post-transcriptional regulation. There is also strong evidence that WT1 binds to mRNA in vivo with an important role of ZF1 in RNA binding (17). α-Actinin1 (ACTN1) was identified as the first cellular RNA target for WT1 (18). In addition, WT1 is also found in the cytoplasm, specifically in cancer cells, suggesting that WT1 could be involved in mRNA export, translation or turnover.

WT1 can form homo- and hetero-dimers with various other proteins and is found in large complexes in Wilms tumours (19,20). WT1 can activate or repress genes depending on the architecture of the promoter; the transcriptional function of WT1 can be modulated through the physical interaction with other proteins (20). The N-terminal transcriptional regulation domain of WT1 was found to bind BASP1, resulting in transcriptional co-suppressor activity (21). WT1 and p53 co-immunoprecipitated in adenovirus E1A-immortalized baby rat kidney cells, and this association resulted in stabilization and increased levels of p53 with enhanced binding to and transactivation of a promoter containing the p53 RGC target sequence. In contrast, repression of a TATA-containing promoter by p53 was inhibited by WT1 (22). The WT1 domain, necessary for binding to p53, was mapped to ZF1 and 2.

Most mutations identified in Wilms tumours lead to a truncated WT1 protein that has lost all or parts of the ZF domain, suggesting that these proteins are impaired in binding to the EGR consensus or the alternative TCC WT1-binding motif. In most cases, however, the N-terminal Pro/Gln-rich repression and activation domain is intact and it is possible that the mutant proteins can still interact with other proteins, thereby modulating their activity. Such a scenario where a truncated mutant WT1 protein inhibits the transcriptional activity of wild-type WT1 has been described (23). In addition, truncated mutant WT1 proteins alter the subcellular localization of the wild-type protein (24). In some cases, the mutations lead to truncation in or after ZF2, 3 or 4, and these proteins may also have retained their RNA-binding capacity or they may bind altered DNA motifs. Recently, a nuclear localization signal in the third ZF was identified that is required for nuclear import (25). We have shown previously that mutant WT1 proteins are predominantly localized to the cytoplasm in Wilms tumours (26) and in Wilms tumour cells as well as in NIH3T3 cells after transfection with mutant WT1 (27).

We have previously described a method for the successful establishment of Wilms tumour cell lines from Wilms tumours with WT1 mutations (27). All cell lines carry a homozygous WT1 mutation owing to loss of heterozygosity of 11p markers. Only one cell line from a WAGR patient has a hemizygous WT1 mutation on the remaining allele (Wilms4). These cell lines can be grown for ∼20 passages but do not have an unlimited life span. With this unique Wilms tumour cell culture model system, where both alleles of WT1 are mutant and no wild-type WT1 allele is present, we can now begin to study for the first time the function of the mutant WT1 proteins in a homologous system (27). We have previously shown that the Wilms2 cell line has a WT1 stop mutation in exon 8 leading to a truncation in ZF2 (p.R362X = WT1Wilms2) and a p.S45Y mutation in CTNNB1. The WT1 frameshift mutation in exon 10 of the Wilms3 tumour cell line leads to an elongation of the WT1 protein by 68 amino acids (p.V432fsX87 = WT1Wilms3); this cell line is wild type for CTNNB1. In Wilms3 cells, the mutant RNA is expressed, the WT1 protein is present in the cytoplasm of the cells, and all isoforms of the WT1Wilms3 mutant proteins are localized in the cytoplasm of transfected NIH3T3 cells (27). Both WT1 mutant proteins retain the Pro/Gln-rich N-terminus and an intact ZF1. In the WT1Wilms3 mutant protein, only ZF4 is disrupted, but both + and −KTS isoforms can be synthesized, whereas in Wilms2, ZF2 is truncated and ZF3 and 4 are missing and the stop codon occurs before the KTS alternative splice site. The aim of this study was to functionally characterize these two naturally occurring mutant WT1 proteins and to establish whether they act via gain or loss of function. Here, we report that two cytoplasmically expressed mutant WT1 proteins regulate a number of genes/proteins as revealed by knockdown of mutant WT1 in the Wilms tumour cell lines. In this work, we show that the mutant proteins retain their ability to interact with p53 and to bind to RNA with a reduced association constant. Loss of WT1 by knockdown in these cells results in reduced proliferation and reduced expression of genes from the G2/M phase of the cell cycle. Expression of the mutant WT1Wilms3 protein in mesenchymal stem cells (MSCs) results in up-regulation of the same cell cycle genes, and these genes are not regulated by wild-type WT1, confirming a gain-of-function for the mutant protein.

## RESULTS

### Intracellular distribution of mutant WT1Wilms2 protein

Wilms tumour protein WT1 is localized to the nucleus in embryonic and adult tissues (28). We have reported recently that contrary to wild-type WT1, mutant WT1-GFP-fusion proteins from Wilms3 tumour cells are located predominantly in the cytoplasm of transfected NIH3T3 cells (27). The mutation in WT1Wilms3 leads to an extended WT1 protein with additional 68 amino acids at the C-terminus. The positions of mutations in WT1Wilms3 and WT1Wilms2 are shown in Figure 1. WT1Wilms2 is a truncated protein, and the stop codon occurs before the KTS alternative splice site. Therefore, only the isoforms + and – exon 5 exist. In both Wilms tumour cell lines, the WT1 mutation is homozygous, i.e. both alleles carry the same mutation and no normal WT1 protein is present in the cells. Here we investigated the intracellular distribution of these proteins in NIH3T3 cells. WT1Wilms2+exon5- and WT1Wilms2−exon5-GFP-fusion proteins were expressed in NIH3T3 cells, and a predominant cytoplasmic localization was observed by fluorescence microscopy (Supplementary, Material Fig. S1). In contrast, the wild-type WT1−/− and −/+ isoforms are exclusively located in the nucleus (Supplementary Material, Fig. S1). Taken together, these results show an important difference between wild-type and mutant WT1 proteins.

Figure 1.

Schematic drawings of wild-type and mutant WT1 proteins. Wild-type WT1 protein (top). Red asterisks indicate the positions of WT1 mutations that we identified previously in Wilms2 and Wilms3 tumour cells. Truncated WT1Wilms2 protein (middle). Extended WT1Wilms3 (bottom). The dotted red flag at the C-terminus of WT1Wims3 indicates the position of the changed 87 amino acids.

Figure 1.

Schematic drawings of wild-type and mutant WT1 proteins. Wild-type WT1 protein (top). Red asterisks indicate the positions of WT1 mutations that we identified previously in Wilms2 and Wilms3 tumour cells. Truncated WT1Wilms2 protein (middle). Extended WT1Wilms3 (bottom). The dotted red flag at the C-terminus of WT1Wims3 indicates the position of the changed 87 amino acids.

### Interaction of mutant WT1 proteins with p53

It has been reported that wild-type WT1 protein has the ability to interact with the tumour suppressor p53 (22). Here we wished to examine whether mutant WT1Wilms2 and WT1Wilms3 proteins have retained this ability. To investigate whether mutant WT1 proteins interact with p53, HeLa cells that express p53 but no WT1 protein were transfected with plasmids driving expression of GFP, or GFP linked to wild-type WT1, WT1Wilms2 and WT1Wilms3. Preparation of whole cell extracts, followed by immunoprecipitation with antibodies specific for p53 or a control antibody and immunoblotting with a WT1 antibody specific for the N-terminus, demonstrated co-immunoprecipitation of these WT1 proteins with p53. As reported before, wild-type WT1 co-immunoprecipitated with p53 (Fig. 2). Moreover, both WT1Wilms2 and WT1Wilms3 retained the ability to interact with p53 (Fig. 2). It is thus possible that mutant WT1 proteins modulate p53 activities in Wilms tumour cells.

Figure 2.

Interaction of p53 and WT1 proteins. Detection of WT1-p53 protein interactions by immunoprecipitation and western blot analysis. HeLa cells were transfected with expression plasmids encoding the WT1-GFP-fusion proteins, wild-type WT1, mutant WT1Wilms3 and mutant WT1Wilms2. As control, cells were transfected with a GFP-expression plasmid alone. Lysates from transfected cells were incubated with control antibodies (lanes, αCtrl) or p53-specific antibodies (lanes, αP53). Immunoprecipitates were resolved by SDS–PAGE and immunoblotted using antibodies specific for the N-terminus of the WT1 protein (αF6WT1). As input controls, 5% of the respective lysates from transfected HeLa cells were loaded on the gel (lanes, In). Asterisk indicates a non-specific band. Wild-type and mutant WT1-GFP-fusion proteins in the input and immunoprecipitated fractions are marked by black boxes.

Figure 2.

Interaction of p53 and WT1 proteins. Detection of WT1-p53 protein interactions by immunoprecipitation and western blot analysis. HeLa cells were transfected with expression plasmids encoding the WT1-GFP-fusion proteins, wild-type WT1, mutant WT1Wilms3 and mutant WT1Wilms2. As control, cells were transfected with a GFP-expression plasmid alone. Lysates from transfected cells were incubated with control antibodies (lanes, αCtrl) or p53-specific antibodies (lanes, αP53). Immunoprecipitates were resolved by SDS–PAGE and immunoblotted using antibodies specific for the N-terminus of the WT1 protein (αF6WT1). As input controls, 5% of the respective lysates from transfected HeLa cells were loaded on the gel (lanes, In). Asterisk indicates a non-specific band. Wild-type and mutant WT1-GFP-fusion proteins in the input and immunoprecipitated fractions are marked by black boxes.

### Equilibrium binding of RNA aptamers to the ZF domains of WT1 and Wilms3

As the mutant WT1Wilms3 protein harbours an intact ZF1, 2 and 3, we wished to study whether the ZF domain from this protein retains its capacity to bind to RNA. The RNA-binding activity of the WT1 ZF domain has been characterized previously using the RNA aptamers RNA22 and RNA38 (29). Here we have determined the affinities of the Wilms3 ZF domains for these two different aptamers using an equilibrium binding assay (Fig. 3). Although the mutation in Wilms3 disrupts proper zinc coordination to finger 4 of the domain, RNA-binding activity is retained but reduced by 4- to 5-fold compared with the wild-type WT1 ZF domain (Fig. 3).

Figure 3.

Equilibrium binding curves for the binding of the ZF domains of WT1 and Wilms3 to RNA. The affinities of RNA22 and RNA38 were measured using a nitrocellulose filter binding assay. Each line represents the best fit for the formation of a simple bimolecular complex (R values of 0.98–0.99). Each data point is the mean of two independent determinations, with the standard deviation indicated with error bars. The table shows the apparent association constant for wild-type WT1 and mutant WT1Wilms3.

Figure 3.

Equilibrium binding curves for the binding of the ZF domains of WT1 and Wilms3 to RNA. The affinities of RNA22 and RNA38 were measured using a nitrocellulose filter binding assay. Each line represents the best fit for the formation of a simple bimolecular complex (R values of 0.98–0.99). Each data point is the mean of two independent determinations, with the standard deviation indicated with error bars. The table shows the apparent association constant for wild-type WT1 and mutant WT1Wilms3.

### Mutant WT1 proteins are linked to gene regulation in Wilms tumour cells

To gain more insight into the function of mutant WT1Wilms3 and WT1Wilms2 proteins, we performed siRNA (siWT1) knockdown analyses. First, we tested different WT1-specific siRNA oligonucleotides and found that a combination of two siRNAs directed against sequences in exon 6 and the 3′ UTR of WT1 achieved a very efficient knockdown 48 h after transfection of Wilms3 cells (Fig. 4A). As control a non-silencing siRNA was used. Gene expression profiles were generated using whole-genome microarrays. To identify WT1 responsive genes, the mean of two biological replicates was determined and the corresponding genes were selected (P < 0.05). A 1.5-fold change (fc) relative to the control and a minimal expression of 200 fluorescence intensity was used as cut–off, and we detected 388 genes with increased and 381 genes with decreased expression. The efficiency of WT1Wilms3 knockdown was examined by Q-RT-PCR, and a reduction of WT1Wilms3 expression by 82% was measured (Fig. 4A). Microarray data were validated for selected genes by Q-RT-PCR (Fig. 4B). It is evident that siWT1 knockdown caused decreased expression of GDF5 and increased expression of MESCD2, CASP7, PODXL1 in Wilms3 tumour cells. We next examined gene expression profiles after siWT1 knockdown in Wilms2 cells, and the efficiency of WT1Wilms2 knockdown was quantified by Q-RT-PCR (Fig. 4C). We identified 361 genes with increased and 422 genes with decreased expression levels. Microarray data showed a 2.7-fold up-regulation of PODXL1, which was validated by Q-RT-PCR (Fig. 4C). Next we compared the siWT1 gene expression data from Wilms2 and Wilms3 tumour cells and identified a common set of 280 down- and 243 up-regulated genes (Supplementary Material, Fig. S2). These results demonstrate that the mutant WT1 proteins are engaged in regulating gene expression in Wilms tumour cells. Moreover, the data show that truncated WT1Wilms2 and extended WT1Wilms3 proteins (Fig. 1) regulate similar gene expression programmes.

Figure 4.

Validation of microarray data from siWT1 and shWT1 knockdown experiments by Q-RT-PCR. All Q-RT-PCR data are displayed relative to a non-silencing si- or shRNA control that is set at 1. Data represent the mean of triplicates, and the standard deviations are indicated for each experiment. PCR reactions were done after transfection of siRNA, and the time points are indicated for each Q-RT-PCR. (A) WT1Wilms3 expression after treatment with siWT1-RNA. (B) Validation of GDF5, CASP7, MESCD2 and PODXL expression after siRNA-mediated WT1 knockdown in Wilms3 cells. (C) WT1Wilms2 and PODXL expression after treatment with siWT1-RNA. (D) WT1Wilms3 expression 48 h after infection with shWT1 lentiviruses (left panel). Expression data are displayed relative to a non-silencing shRNA-lentivirus control that is set at 1. Validation of JARID2 and PODXL1 expression in Wilms3 cells infected with shWT1 lentiviruses (right panel). Q-RT-PCR was performed 48 h after infection with shWT1 lentiviruses.

Figure 4.

Validation of microarray data from siWT1 and shWT1 knockdown experiments by Q-RT-PCR. All Q-RT-PCR data are displayed relative to a non-silencing si- or shRNA control that is set at 1. Data represent the mean of triplicates, and the standard deviations are indicated for each experiment. PCR reactions were done after transfection of siRNA, and the time points are indicated for each Q-RT-PCR. (A) WT1Wilms3 expression after treatment with siWT1-RNA. (B) Validation of GDF5, CASP7, MESCD2 and PODXL expression after siRNA-mediated WT1 knockdown in Wilms3 cells. (C) WT1Wilms2 and PODXL expression after treatment with siWT1-RNA. (D) WT1Wilms3 expression 48 h after infection with shWT1 lentiviruses (left panel). Expression data are displayed relative to a non-silencing shRNA-lentivirus control that is set at 1. Validation of JARID2 and PODXL1 expression in Wilms3 cells infected with shWT1 lentiviruses (right panel). Q-RT-PCR was performed 48 h after infection with shWT1 lentiviruses.

As this result is of great importance for the field of WT1 research, we wished to confirm our findings by another method. We thus performed lentiviral-mediated shWT1 knockdown experiments using Wilms3 tumour cells. The shWT1 targets exon 7 of the WT1 gene, and the most efficient knockdown of WT1Wilms3 expression (73%) occurred 48 h after infection (Fig. 4D). Gene expression analyses were performed under these conditions, and a scrambled non-silencing shRNA was used as control. A 1.5 fc relative to the control and a minimal expression of 200 fluorescence intensity was used as cut-off. The microarrays identified 1012 genes with increased and 1076 genes with decreased expression levels. Increased expression of PODXL and JARID2 was validated by Q-RT-PCR (Fig. 4E). Taken together our WT1 knockdown experiments show for the first time that mutant WT1 proteins do have a biological function and this is linked to gene regulation in Wilms tumour cells. The top 100 genes with decreased and increased expression are listed in Supplementary Material, Tables S1 and S2, showing the fold changes of expression relative to the scrambled control.

### Mutant WT1 proteins promote expression of genes from the G2/M phase of the cell cycle

To gain more insight into the biological function of the mutant WT1Wilms3 protein, we analysed the gene expression data from siWT1 knockdown experiments using MetaCore, an integrated software suite for functional analysis of biological experimental data. To identify biological processes associated with gene expression data, we used an enrichment analysis of MetaCore pathway maps that represent a set of signalling and metabolic maps covering cellular processes in a comprehensive way. These maps also show functions and interactions of individual genes in a given pathway. We first examined the subset of down-regulated genes obtained from siWT1 experiments in Wilms3 cells using MetaCore for enrichment of specific pathways and identified highly significant enrichment of several maps from the cell cycle. A list of the ten most significant pathway maps and their P-values is summarized in Figure 5A (blue bars). The genes from the cell cycle pathways are listed in Supplementary Material, Table S3, left column. It is evident that these pathways are related to the G2/M phase of cell cycle. The most significant pathway maps ‘Cell cycle_Role of APC in cell cycle regulation’, ‘Cell cycle_Spindle assembly and chromosome separation’ and ‘Cell cycle_The metaphase checkpoint’ operate at the end of G2/entry into mitosis.

Figure 5.

Mutant WT1 proteins control expression of cell cycle genes from the G2/M phase in Wilms2 and Wilms3 tumour cells. (A) Enrichment of MetaCore pathway maps in gene expression profiles from Wilms tumour cells after WT1 knockdown. For enrichment analysis, the subsets of down-regulated genes were used. The height of the blue and orange bars indicates the P-values of enrichment for individual pathway maps. Orange bars: WT1 knockdown in Wilms3 cells after infection with shWT1 lentiviruses. Blue bars: WT1 knockdown in Wilms3 cells after transfection of siRNA. (B) MetaCore pathway map ‘Cell cycle_The metaphase checkpoint’. Blue thermometers indicate down-regulated genes. The numbers on top of each thermometer refers to individual WT1 knockdown experiments. 1, shWT1 Wilms3 cells; 2, siWT1 Wilms3 cells. The properties of genes from this pathway are indicated by coloured symbols. These symbols are explained by a MetaCore quick reference guide (Supplementary Material, Table S15). Green arrows indicate positive interactions (activation) and red arrows, negative interactions (inhibition). (C) Enrichment of MetaCore pathway maps in gene expression profiles from Wilms2 cells after siWT1 knockdown. For enrichment analysis, the subsets of down-regulated genes were used. The height of the orange bars indicates the P-values for enrichment of individual pathway maps.

Figure 5.

Mutant WT1 proteins control expression of cell cycle genes from the G2/M phase in Wilms2 and Wilms3 tumour cells. (A) Enrichment of MetaCore pathway maps in gene expression profiles from Wilms tumour cells after WT1 knockdown. For enrichment analysis, the subsets of down-regulated genes were used. The height of the blue and orange bars indicates the P-values of enrichment for individual pathway maps. Orange bars: WT1 knockdown in Wilms3 cells after infection with shWT1 lentiviruses. Blue bars: WT1 knockdown in Wilms3 cells after transfection of siRNA. (B) MetaCore pathway map ‘Cell cycle_The metaphase checkpoint’. Blue thermometers indicate down-regulated genes. The numbers on top of each thermometer refers to individual WT1 knockdown experiments. 1, shWT1 Wilms3 cells; 2, siWT1 Wilms3 cells. The properties of genes from this pathway are indicated by coloured symbols. These symbols are explained by a MetaCore quick reference guide (Supplementary Material, Table S15). Green arrows indicate positive interactions (activation) and red arrows, negative interactions (inhibition). (C) Enrichment of MetaCore pathway maps in gene expression profiles from Wilms2 cells after siWT1 knockdown. For enrichment analysis, the subsets of down-regulated genes were used. The height of the orange bars indicates the P-values for enrichment of individual pathway maps.

Next, we analysed the subset of genes that were down-regulated after infection of Wilms3 cells with shWT1 lentiviruses. The Venn diagram in Supplementary Material, Fig. S3 Panel A, shows that 70 genes are regulated in common in shWT1 and siWT1 knockdown experiments. Enrichment analysis of MetaCore pathway maps revealed ‘Cell cycle_The metaphase checkpoint’, ‘Cell cycle_Chromosome condensation in prometaphase’ and ‘Cell cycle_initiation of mitosis as highly significant’. A list of the ten top-ranking pathway maps is shown in Figure 5A, orange bars. The most significant pathway map ‘Cell cycle_The metaphase checkpoint’ is visualized in Figure 5B, and the down-regulated genes are indicated by blue thermometers. The genes corresponding to the cell cycle pathways are listed in Supplementary Material Table S3, right column. Although only 70 identical genes are regulated by si and shRNA, Supplementary Material, Table S3 shows that additional but different genes from the same pathways are regulated in either approach. These data show for the first time that mutant WT1 from Wilms3 tumour cells promote the expression of a significant number of genes from the G2/M phase of the cell cycle. Pathway maps, however, show only selected genes and do not represent all genes of a given biological process. To get a comprehensive overview, we performed a GO process analysis using MetaCore and the subset of down-regulated genes. We identified the most significant biological processes associated with the set of down-regulated genes (Supplementary Material, Fig. S4). These included ‘cell cycle’ (P-value 1.113E−15), ‘mitosis’ (P-value 1.511E–14) and ‘cell division’ (P-value 5.459E−14).

We next performed a MetaCore pathway enrichment analysis of the subset of genes whose expression was down-regulated after siRNA-mediated knockdown of mutant WT1Wilms2. The enrichment analysis of MetaCore pathway maps revealed that ‘Cell cycle_Role of APC in cell cycle regulation’, ‘DNA damage_ATM/ATR regulation of G1/S checkpoint’ and ‘Cell cycle_the metaphase checkpoint’, are the three most significant pathway maps. A list of the ten top-ranking pathway maps and their P-values for enrichment is shown in Figure 5C, and the corresponding genes are shown in Supplementary Material, Table S4. We can conclude here that the functions of truncated WT1Wilms2 and extended WT1Wilms3 proteins are very similar in the context of gene regulation.

### WT1Wilms3 inhibits the expression of genes from multiple signalling pathways

Next, we wished to examine the biological function of genes that are repressed by mutant WT1Wilms3. We therefore used the subsets of genes that were up-regulated in siWT1 and shWT1 knockdown experiments and performed enrichment analyses of pathway maps using MetaCore. We noticed that the P-values for pathway enrichment were lower in siWT1 than in shWT1 experiments and selected the shWT1 data from Wilms3 cells for our analysis. Enrichment analysis by pathway maps identified ‘Transcription_Androgen Receptor nuclear signalling’ and Cytoskeleton remodelling_TGF, WNT and cytoskeleton remodelling’ as most significant. A list of the top-ranking pathway maps is presented in Supplementary Material, Fig. S5 and Table S5. This analysis shows an overrepresentation of enriched pathway maps related to signalling, and we decided to investigate this issue in a more comprehensive fashion. We therefore inspected all pathway maps from the MetaCore database related to signalling and identified a large number of genes that are repressed by mutant WT1. These include TGFBRII, IGF1R, AXL, MET, WNT5 and FZD6 and the downstream signalling components DSH, CTNNB1, AKT, DUSP3, DUSP16, ERK2, FAK1, MAPK, MLCK, MYLK1, STAT2/3, FOS, SOS and others. The proteins encoded by DUSP3 and DUSP16 are involved in the negative regulation of Erk signalling and MAPK pathways, respectively (Supplementary Material, Fig. S6). Moreover, WT1Wilms3 seems to repress expression of CDKN1B (p27) and CDKN2B (p15) genes, negative regulators of progression from the G1 to S phase transition of the cell cycle. A comparison of the siWT1 with shWT1 knockdown experiments is shown in Supplementary Material, Fig. S3, Panel B. The top-ranking pathway ‘signal transduction Erk interactions: inhibition of Erk’ is identified by the separate analysis of the 70 identical genes that are regulated in both approaches.

Most Wilms tumours with WT1 mutations also have mutations in CTNNB1 and in consequence an activated Wnt signalling pathway. However, Wilms3 cells do not carry a CTNNB1 mutation, and the Wnt/β-catenin pathway is not active as measured by TOP/FOP reporter plasmids (27). To check whether WT1Wilms3 is involved in regulating expression of genes from Wnt signalling pathways, we used the subset of genes with increased expression in siWT1 and shWT1 experiments and examined the Wnt signalling pathway maps of the MetaCore database. Our analysis revealed that WT1Wilms3 negatively regulates the expression of several genes from the Wnt signalling pathways including WNT5, DKK1, FZD6, FZD8, MESDC2, DSH (DVL3), CTNNB1, as well as the downstream target VIM (Supplementary Material, Fig. S7). It is known that the regulation of this pathway is highly complex, with several negative regulators being up-regulated in a cell with active Wnt signalling. The up-regulation of ligands, receptors and downstream signalling components upon WT1 removal suggests that the mutant WT1 protein may be a negative regulator of the Wnt signalling pathway. A detailed analysis of this issue will be published elsewhere.

### Mutant WT1 proteins promote proliferation of Wilms tumour cells

Our data show that mutant WT1 proteins are involved in regulating the activities of several genes from the G2/M phase of the cell cycle. We therefore investigated whether mutant WT1 proteins also affect the proliferation of Wilms tumour cells. To this end, we determined the proliferation rates of Wilms2 and Wilms3 cells after WT1 knockdown using shWT1 lentiviruses. Figure 6 shows that the shWT1 knockdown significantly reduces the proliferation of both Wilms2 (P = 0.037) and Wilms3 cells (P = 0.014) in comparison with control cells transduced with a non-silencing shRNA. A highly significant inhibition of cell proliferation is also visible in tissue culture dishes 72 and 96 h after infection of Wilms3 cells with shWT1 lentiviruses in comparison with the non-silencing control (Supplementary Material, Fig. S8, A). Next, we wished to verify these results by a WST proliferation assay (Supplementary Material, Fig. S8, B). In this assay, cell survival is analysed and we found significantly reduced cells numbers in shWT1 expressing cells. This suggests that some cells do not survive without the presence of WT1Wilms3. We conclude from these results that the proliferation of Wilms2 and Wilms3 tumour cells is dependent on the activity of mutant WT1Wilms2 and WT1Wilms3 proteins. We report here that mutant WT1 proteins facilitate the expression of G2/M phase genes, and we asked whether the distribution of cell cycle phases would be altered in Wilms3 tumour cells after shWT1 knockdown. To this end, we used the Cycletest™ Plus kit in initial experiments and analysed Wilms3 cells 48 h after infection with shWT1 lentiviruses using a fluorescence-activated cell sorter (Supplementary Material, Fig. S9). At this time point, we did observe a slight increase of Wilms3 cells in G2/M phase in comparison with the controls; however, apoptosis was not observed in these studies. Future experiments will determine the kinetics of siWT1 and shWT1-mediated knockdowns in Wilms3 cells and correlate these data with an analysis of cell cycle phases in Wilms3 tumour cells.

Figure 6.

Mutant WT1 proteins promote proliferation of Wilms tumour cells. Wilms2 and Wilms3 cell number after shWT1 and scrambled RNA (shC) knockdown at different time points, performed in two independent analyses (a and b), showing a significantly lower median cell count after shWT1 knockdown compared with scrambled RNA, Wilms2: P = 0.037 (Wilcoxon test) and Wilms3 P = 0.014 (Wilcoxon test).

Figure 6.

Mutant WT1 proteins promote proliferation of Wilms tumour cells. Wilms2 and Wilms3 cell number after shWT1 and scrambled RNA (shC) knockdown at different time points, performed in two independent analyses (a and b), showing a significantly lower median cell count after shWT1 knockdown compared with scrambled RNA, Wilms2: P = 0.037 (Wilcoxon test) and Wilms3 P = 0.014 (Wilcoxon test).

### WT1Wilms3 protein modulates the activities of selected transcription factors including p53

Our data show that mutant WT1 proteins are linked to gene regulation. As both Wilms2 and Wilms3 mutant proteins lack a functional DNA binding domain, it was of interest to gain more insight into the underlying molecular mechanisms. To investigate this issue, we used the MetaCore program ‘What are the key transcription factors and target genes in my data?’ This program uses a database containing all published information regarding transcription factors and their target genes, and the results show which transcription factors play a significant role in individual microarray experiments. First we uploaded the down-regulated genes from shWT1 Wilms3 knockdown experiments as the input list. Using these data, the transcription regulation workflow generates sub-networks centred on transcription factors. Sub-networks are ranked by a P-value and interpreted in terms of Gene Ontology. The resulted sub-networks of the whole network are displayed in Supplementary Material, Table S6, a network list table, in the order reflecting their statistical significance in terms of P-values. The network list shows the top 21 most significant sub-networks centred around CREB1, c-MYC, SP1, YY1, OCT 3–4, GCR-alpha, ESR1, p53, E2F1, Androgen Receptor, HIF1A and others. The column ‘P-value for enriched transcription factor’ indicates the level of significance. The column ‘enriched GO processes’ lists specific GO processes that were identified in each transcription factor sub-network. For example, the sub-networks centred around CREB1, SP1, YY1, ESR1, p53, E2F1, Androgen Receptor, REL-A p65, ZNF143, NANOG, c-JUN, EGR1, ETS1 and NFKB p50 showed an enrichment of GO processes related to cell cycle and cell proliferation. We also uploaded the up-regulated genes from shWT1 Wilms3 knockdown experiments as input list for the MetaCore program ‘What are the key transcription factors and target genes in my data?’ and identified enriched transcription factors and related GO processes (Supplementary Material, Table S7). It is recognized that different GO processes were identified related to cell metabolism, organismal development, differentiation and many others.

The bioinformatics data provide good evidence that mutant WT1 proteins modulate transcription factor activity, and we wished to further explore this issue and selected p53 as a model system. Wild-type WT1 interacts with p53, and we have shown that mutant WT1Wilms2 and WT1Wilms3 proteins also bind to p53 (Fig. 2). Wilms2 and Wilms3 cells express very low levels of p53 protein, which are below the detection limit of western blot analysis. However, we could detect serine 15 (S15), serine 46 (S46) and serine 392 (S392) phosphorylation of p53 using an antibody array (Supplementary Material, Fig. S10, Panel A). To investigate whether p53 is transcriptionally active, we used p53-responsive and p53 non-responsive reporter plasmids (pG13-Luc and MG15-Luc) that contain p53 wild-type and p53 mutant binding sites. The relative luciferase activity of the p53-responsive reporter compared with the p53 non-responsive reporter was highly significant, demonstrating that p53 in Wilms2 and Wilms3 tumour cells is transcriptionally active (Supplementary Material, Fig. S10, Panel B).

The MetaCore program ‘What are the key transcription factors and target genes in my data?’ revealed that the group of genes (n = 1076) that are down-regulated by shWT1 contains a subset of 54 p53-target genes where p53 is directly engaged in transcriptional regulation. This information is displayed in a p53 network diagram (Fig. 7A). The p53 network diagram consists of a central p53 node; red arrows are connected to p53-target genes whose expression is inhibited (n = 20), and green arrows are connected to genes whose expression is activated by p53 (n = 34). If all genes of this p53 network are positively regulated by WT1Wilms3, we can conclude that the mutant WT1 protein can either enhance or antagonize the transcriptional activity of p53. Similar data were obtained when we investigated the group of genes (n = 1012) that are up-regulated by shWT1. Here we identified a subset of 44 p53-target genes where 34 are activated and 10 are repressed by p53 (Fig. 7B). These results show that the WT1Wilms3 protein has the ability to modulate the transcriptional activity of p53 where it most likely acts either as a co-repressor or co-activator in the context of individual genes. It is important to note that the transcriptional activity of p53 in general is controlled by a highly complex set of co-activators and co-repressors (30).

Figure 7.

Identification of p53 networks in cohorts of WT1Wilms3 target genes. The MetaCore program ‘What are the key transcription factors and target genes in my data’ was used to identify p53 networks in subsets of genes whose expressions are either down-regulated (A) or up-regulated (B) by shWT1 knockdown in Wilms3 cells. Blue circles mark down-regulated genes (A), and red circles mark up-regulated genes (B). The properties of individual genes are indicated by coloured symbols. These symbols are explained in Supplementary Material, Table S15. Green arrows point to genes that are activated by p53 and red arrows to genes that are repressed by p53.

Figure 7.

Identification of p53 networks in cohorts of WT1Wilms3 target genes. The MetaCore program ‘What are the key transcription factors and target genes in my data’ was used to identify p53 networks in subsets of genes whose expressions are either down-regulated (A) or up-regulated (B) by shWT1 knockdown in Wilms3 cells. Blue circles mark down-regulated genes (A), and red circles mark up-regulated genes (B). The properties of individual genes are indicated by coloured symbols. These symbols are explained in Supplementary Material, Table S15. Green arrows point to genes that are activated by p53 and red arrows to genes that are repressed by p53.

To further investigate the influence of WT1Wilms3 on p53 transcriptional activity, we used a p53-responsive reporter construct (pG13-Luc) in transient transfection assays of Wilms3 cells after infection with shWT1 lentiviruses (Supplementary Material, Fig. S11). As control, a p53 non-responsive reporter construct (MG 15-Luc) was used in transient transfection assays. This initial experiment revealed diminished transcriptional activity of p53 in shWT1-treated Wilms3 cells (Supplementary Material, Fig. S11, right panel). We can conclude that WT1Wilms3 acts as a co-activator for p53 in the context of this p53-responsive promoter in pG13-Luc.

### Classification of WT1 mutations as gain-of-function mutations

Next, we wished to validate the siWT1 and shWT1 knockdown data from Wilms tumour cells by expressing WT1Wims3 protein in a cell line devoid of WT1. It is a necessary prerequisite to identify the spectrum of genes whose expression is controlled by mutant WT1 in order to investigate in a second step whether the mutant protein gained any novel functions in comparison with the wild-type protein. To address these issues, we selected human MSCs as a model system as these do not express WT1. It is important for these experiments to use a cell line related to Wilms tumour cells in order to avoid major context-specific effects. We previously showed that the transcriptomes and differentiation potentials of MSC and our Wilms tumour cells are similar (27). To express wild-type and mutant WT1 in MSCs, we generated lentivirus constructs that express the wild-type WT1+/−, WT1++, and the mutant WT1Wilms3+/− and WT1Wilms3+/+ isoforms. After infection of MSCs, all four lentivirus constructs showed similar expression levels as determined by Q-RT-PCR (Supplementary Material, Fig. S12). Gene expression profiles were established using the same parameters as in WT1 knockdown experiments. In MSCs infected with WT1Wilms3++ lentiviruses, 425 genes were up-regulated and 556 down-regulated. Infection with WT1Wilms3+− lentiviruses caused an up-regulation of 605 and a down-regulation of 755 genes. The top 100 up- and down-regulated genes in MSCs infected with WT1Wilms3+− lentiviruses are listed in Supplementary Material, Tables S8 and S9, respectively. We selected the subset of up-regulated genes from WT1Wilms3+− lentivirus infected MSCs and performed an enrichment analysis of pathway maps using MetaCore. The pathway maps ‘Cell cycle_The metaphase checkpoint’, ‘Cell cycle_Chromosome condensation in prometaphase’ and ‘Cell cycle_Spindle assembly and chromosome separation’ are most significant (Fig. 8A, orange bars). Similar results were obtained using WT1Wilms3++ lentiviruses (Fig. 8A, blue bars). These results show that the mutant WT1 protein can activate the expression of a significant number of genes from the G2/M phase of the cell cycle. A comparison of pathway enrichment data after WT1 knockdown in Wilms3 cells (Fig. 5A) with those after WT1Wilms3 overexpression in MSCs (Fig. 8A) shows complete concordance. This is also demonstrated by the pathway map ‘Cell cycle_The metaphase checkpoint’ where the shWT1 knockdown data from Wilms3 cells and the WT1Wilms3+− expression in MSCs have been compared (Fig. 8B). To illuminate quantitative aspects, we compared the fold change values of shWT1 down-regulated genes (in Wilms3 cells) with those of WT1Wilms3+−-induced genes in MSCs, and a list of common regulated genes is shown in Supplementary Material, Table S10. This gene list comprises a large group of cell cycle genes from G2/M, and it is evident that overexpression of WT1Wilms3+− in MSCs induced these genes to a much higher level compared with the fold change repression by shWT knockdown. We conclude here that the mutant WT1 protein is connected to the cell cycle where it is involved in regulating expression of genes that promote the G2/M transition and the proliferation of Wilms tumour cells.

Figure 8.

Mutant WT1Wilms3 controls expression of cell cycle genes from the G2/M phase in mesenchymal stem cells. (A) Enrichment of MetaCore pathway maps in the subset of genes whose expression is induced by WT1Wilms3++ and WT1Wilms3+− isoforms in MSC. The two isoforms of WT1Wilms3 were expressed in MSCs using lentiviral constructs. The length of orange bars indicates the P-value for enrichment of individual pathway maps in WT1Wilms3++ infected MSCs and blue bars show those for the WT1Wilms+− isoform. (B) Most significant pathway map ‘Cell cycle_The metaphase checkpoint’: comparison of gene expression levels after shWT1 knockdown in Wilms3 cells and lentiviral expression of the mutant WT1Wilms3+− isoform in MSCs. Red thermometers indicate up-regulated genes by expression of WT1Wilms3+− in MSC (number 1 on top of the thermometer) and blue thermometers indicate down-regulated genes in shWT1 knockdown in Wilms3 cells (number 2 on the top of the thermometer). Properties of individual genes are indicated by coloured symbols. Symbols are explained in a (Supplementary Material, Table S15).

Figure 8.

Mutant WT1Wilms3 controls expression of cell cycle genes from the G2/M phase in mesenchymal stem cells. (A) Enrichment of MetaCore pathway maps in the subset of genes whose expression is induced by WT1Wilms3++ and WT1Wilms3+− isoforms in MSC. The two isoforms of WT1Wilms3 were expressed in MSCs using lentiviral constructs. The length of orange bars indicates the P-value for enrichment of individual pathway maps in WT1Wilms3++ infected MSCs and blue bars show those for the WT1Wilms+− isoform. (B) Most significant pathway map ‘Cell cycle_The metaphase checkpoint’: comparison of gene expression levels after shWT1 knockdown in Wilms3 cells and lentiviral expression of the mutant WT1Wilms3+− isoform in MSCs. Red thermometers indicate up-regulated genes by expression of WT1Wilms3+− in MSC (number 1 on top of the thermometer) and blue thermometers indicate down-regulated genes in shWT1 knockdown in Wilms3 cells (number 2 on the top of the thermometer). Properties of individual genes are indicated by coloured symbols. Symbols are explained in a (Supplementary Material, Table S15).

To address the issue whether mutant WT1 proteins gained any function in comparison with wild-type WT1, we infected MSCs with lentiviruses expressing the wild-type WT1++ and WT1+− isoforms. Gene expression analysis was done as described earlier, and we generated datasets of up- and down-regulated genes. Lentiviruses expressing the WT1++ isoform induced the expression of 882 individual genes, and 690 genes showed reduced expression levels in MSC. Infection with WT1+− lentiviruses was associated with an up-regulation of 425 and a down-regulation of 696 genes. The top 100 up- and down-regulated genes in MSCs infected with WT1Wilms3+− lentiviruses are listed in Supplementary Material, Tables S11 and S12, respectively. We then performed a MetaCore pathway enrichment analysis using the datasets for the WT1++ and WT1+− isoforms (Fig. 9A). This figure shows that identical pathway maps are significantly enriched in gene expression profiles after infection of MSCs with the two WT1 isoforms. The P-values for pathway enrichment are indicated for WT1++ and WT1+− by orange and blue bars, respectively (Fig. 9A). Although identical pathways were enriched, the P-values indicate subtle differences between the WT1++ and WT1+−-dependent gene expression profiles. For the WT1++ isoform ‘Development regulation of epithelial to mesenchymal transition’ and for the WT1+/− isoform ‘Cytoskeleton remodeling_Regulation of actin cytoskeleton by RHO GTPases’ were the most significant pathway maps. Next, we looked for any enrichment of cell cycle pathway maps in gene expression data from WT1++ and WT1+−-infected MSCs and visually inspected the top 50 enriched pathway maps (Supplementary Material, Fig. S13). We did not find any pathway map related to the cell cycle in this analysis. We can conclude from this data that mutant WT1 differs from wild-type WT1 by the capacity to regulate ‘cell cycle genes’. This function is novel, and therefore, these mutations can be classified as gain-of-function mutations.

Figure 9.

Classification of WT1Wilms2 and WT1Wilms3 mutations as gain-of-function mutations. (A) Enrichment of MetaCore pathway maps for genes induced by wild-type WT1 in MSCs. The lengths of orange and blue bars correspond to the significance level of individual maps. Orange bars refer to WT1++-induced gene expression data and blue bars to gene expression data induced by the WT1+− isoform. There was no significant enrichment of any pathway map related to the cell cycle. (B) Gene expression data from MSCs infected with WT1Wilms3+− lentiviruses show proof of gain-of-function. To generate the heatmap, we selected a subset of genes from the cell cycle whose expression was down-regulated by siWT1 and shWT1 knockdown in Wilms tumour cells and compared their expression levels in diverse experiments that are indicated in the Figure. The colour corresponds to the log ratio of WT1 knockdown versus scrambled control transfected cells (blue negative log ratio and yellow positive log ratio). Lanes Wilms3 SH-48h.1 and Wilms3 SH-48h.2 are biological replicates of lentivirus-mediated shWT1 knockdown in Wilms3 cells, 48h after infection, and are log ratios versus scrambled RNA transduced cells. Wilms3 SI3-48h.1 and Wilms3 SI3-48h.2 are biological replicates of siWT1-mediated knockdown in Wilms3 cells 48 h after transfection versus non-silencing RNA. Wilms2 SI3-48h corresponds to siWT1-mediated knockdown in Wilms2 cells. MSC WT1Wilms3++ and MSC WT1Wilms3+− refer to MSCs infected with lentiviruses expressing two isoforms of mutant WT1Wilms3, and MSC wt-WT1++ and MSC wt-WT1+− refer to MSCs infected with lentiviruses expressing two isoforms of wild-type WT1, expressed as log ratios versus control virus-transduced cells.

Figure 9.

Classification of WT1Wilms2 and WT1Wilms3 mutations as gain-of-function mutations. (A) Enrichment of MetaCore pathway maps for genes induced by wild-type WT1 in MSCs. The lengths of orange and blue bars correspond to the significance level of individual maps. Orange bars refer to WT1++-induced gene expression data and blue bars to gene expression data induced by the WT1+− isoform. There was no significant enrichment of any pathway map related to the cell cycle. (B) Gene expression data from MSCs infected with WT1Wilms3+− lentiviruses show proof of gain-of-function. To generate the heatmap, we selected a subset of genes from the cell cycle whose expression was down-regulated by siWT1 and shWT1 knockdown in Wilms tumour cells and compared their expression levels in diverse experiments that are indicated in the Figure. The colour corresponds to the log ratio of WT1 knockdown versus scrambled control transfected cells (blue negative log ratio and yellow positive log ratio). Lanes Wilms3 SH-48h.1 and Wilms3 SH-48h.2 are biological replicates of lentivirus-mediated shWT1 knockdown in Wilms3 cells, 48h after infection, and are log ratios versus scrambled RNA transduced cells. Wilms3 SI3-48h.1 and Wilms3 SI3-48h.2 are biological replicates of siWT1-mediated knockdown in Wilms3 cells 48 h after transfection versus non-silencing RNA. Wilms2 SI3-48h corresponds to siWT1-mediated knockdown in Wilms2 cells. MSC WT1Wilms3++ and MSC WT1Wilms3+− refer to MSCs infected with lentiviruses expressing two isoforms of mutant WT1Wilms3, and MSC wt-WT1++ and MSC wt-WT1+− refer to MSCs infected with lentiviruses expressing two isoforms of wild-type WT1, expressed as log ratios versus control virus-transduced cells.

To confirm this result at the level of individual genes, we generated a heat map using the down-regulated ‘cell cycle genes’ from WT1 knockdown experiments (Fig. 9B). The lanes labelled Wilms3 and Wilms2 show all experimental data of WT1 knockdown experiments using either siWT1 or shWT1 lentiviruses and are represented as log ratios versus scrambled control RNAs. The lanes MSC WT1Wilms3 show the log ratios of gene expression data after infection of MSC with lentiviruses expressing mutant WT1Wilms3, versus control virus-transduced cells, and the lanes MSC wt-WT1 show the gene expression levels after infection of MSC with lentiviruses expressing wild-type WT1 versus control virus. The heat map visualizes our results and shows that all ‘cell cycle genes’ that are repressed by siWT1 and shWT1 knockdown approaches are up-regulated by lentivirus-mediated expression of mutant WT1Wilms3 in MSC. Moreover, lentivirus-mediated expression of wild-type WT1 in MSC failed to activate expression of these genes. We conclude that the WT1 mutations that have been identified in Wilms2 and Wilms3 tumour cells are gain-of-function mutations. This gain-of-function phenotype of mutant WT1Wilms3 is also evident when the top 100 genes induced by WT1Wilms3+− and wtWT1+− lentiviruses are compared (Supplementary Material, Table S13). In this cohort of genes, only APOL6 and MNS1 are induced by both lentiviruses (intensity >200). It is very interesting to note that in the cohort of the top 100 genes that were down-regulated by WT1Wilms3+− and wtWT1+− lentiviruses (Supplementary Material, Table S14), most genes are repressed by both lentivirus types. This is also visualized in the Venn diagrams shown in Fig 10. Here the large overlap of genes down-regulated by mutant and wild-type WT1 is evident (Fig. 10A), whereas only a smaller set of genes is up-regulated by both lentiviruses (Fig. 10B). We can conclude that the gain-of-function mutation in mutant Wilms tumour proteins is confined to transcriptional activation, whereas transcriptional repression is mediated by both, the wild-type and mutant WT1 proteins. Future work is needed to exactly work out how transcriptional activation and repression by mutant WT1 is brought about.

Figure 10.

Comparison of mutant and wild-type WT1-dependent gene expression in MSCs. Lentiviruses expressing the +− isoforms of mutant WT1Wilms3 and wild-type WT1 were used in these experiments. The expression data are displayed in Venn diagrams. (A) Comparison of down-regulated genes and (B) up-regulated genes.

Figure 10.

Comparison of mutant and wild-type WT1-dependent gene expression in MSCs. Lentiviruses expressing the +− isoforms of mutant WT1Wilms3 and wild-type WT1 were used in these experiments. The expression data are displayed in Venn diagrams. (A) Comparison of down-regulated genes and (B) up-regulated genes.

## DISCUSSION

Most WT1 mutant Wilms tumours have lost the wild-type WT1 allele, following Knudson’s two-hit hypothesis for the inactivation of a tumour suppressor gene (31,32). In contrast in many adult human tumours, WT1 is highly expressed and sometimes mutations are found, suggesting that it acts as an oncogene (32,33). Some Wilms tumours with WT1 mutations express the mutant RNA and protein (27). Wilms tumour-derived Wilms2 and Wilms3 cell lines were used here to address the question whether WT1 mutations have an impact for the biology of Wilms tumours. To investigate whether mutant WT1Wilms2 and WT1Wilms3 proteins gained a new function (gain-of-function mutation), we compared the gene expression profiles of MSCs, infected with lentiviruses expressing mutant and wild-type WT1, and found that mutant WT1 proteins activate the expression of a large number of genes from the G2/M phase of the cell cycle, whereas wild-type WT1 was unable to do so. These data show unequivocally that WT1Wilms2 and WT1Wilms3 mutations are gain-of-function mutations. This result is particularly interesting because it contradicts the currently accepted model that mutant WT1 proteins contribute to Wilms tumour development because they no longer function as tumour suppressor (32). Mutant WT1Wilms2 and WT1Wilms3 promote the proliferation of Wilms tumour cells in tissue culture, and it has been shown for a variety of human cancers that their growth and survival can be impaired by the inactivation of a single oncogene, a phenomenon termed oncogene addiction (34).

This gain-of-function mutation in WT1Wilms3 is illustrated in a Venn diagram (Fig. 10A and B). Here we compared the datasets of up-regulated genes from MSCs infected with WT1Wilms3+− and wtWT1. Gain-of-function is demonstrated by the blue area in Figure 10B. However, the picture is more complex as there are genes that are regulated by both, wtWT1 and the mutant protein (green area). In addition, as expected there are quite a number of genes that are induced by wtWT1 only (pink area). We also compared the top 100 induced genes from MSCs infected with wtWT1+− and WT1Wilms3+− lentiviruses. Here it is evident that amongst these, only two genes were induced in common (Supplementary Material, Table S13). This clearly demonstrates that gain-of-function is the dominant consequence of this mutation. Different results are evident in the group of down-regulated genes (Fig. 10A). The green area indicates that a large fraction of genes is repressed by wtWT1+− and WT1Wilms3+− in MSCs. We also compared the top 100 repressed genes from MSCs infected with wtWT1+− and WT1Wilms3+− lentiviruses and found that the majority of genes was repressed by both, the wtWT1 and the mutant protein (Supplementary Material, Table S14). Taken together these data show that a novel transcription activation function is the major phenotype of the WT1 mutation in Wilms3 cells. The repression function of wtWT1 and the mutant protein may be explained by the presence of the Pro/Gln-rich transregulatory and dimerization domain in the N-terminal domain. This domain also contains a putative RNA recognition motif (5,8,17), and mutant Wilms3 protein has retained its ability to bind to RNA aptamers, albeit with reduced affinity. Therefore, the mutant proteins can interact and modulate the function of other proteins and/or bind to RNA as well. However, the mechanisms by which the mutant proteins exert these functions remain to be identified.

It is interesting to note that a leukaemia-associated mutant WT1 protein devoid of the ZF domain, WT1(delZ), has also gained new functions (35). Whereas WT1 inhibited the proliferation of haematopoietic progenitor cells (36), the WT1(delZ) protein promoted expansion of human haematopoietic progenitor cells in tissue culture. WT1 mutations were detected in 10–15% of adult AML cases at diagnosis, and the presence of WT1 mutations is an independent predictor of poor clinical outcome (35). Gene expression profiling revealed that WT1(delZ) expression was associated with increased gene expression levels of certain genes related to proliferation and decreased expression levels of genes related to myeloid differentiation.

Mutant WT1 controls genes from the G2/M phase of the cell cycle. For example, the expression of BIRC5, BUB1, CCNB1, CCNB2, CDC20, CDK1, CENPF, PLK1 and UBE2C was down-regulated by shWT1 in Wilms3 cells and up-regulated in MSCs infected with lentiviruses expressing the WT1Wilms3+− isoform. It is remarkable that CCNB1, CCNB2 and CDK1 are amongst these genes as they are key regulators of cell cycle progression in G2/M phase. The most significant transcriptional signature of cancer cells and malignant tumours is the up-regulation of cell cycle and proliferation-related genes (37). An analysis of gene expression data from a large variety of human cancer types identified a core proliferation cluster (core PC) predominantly consisting of genes from the G2/M phase of the cell cycle (38). A comparison of our gene expression data with those from the core PC showed a high degree of overlap. The core PC consists of 46 genes and 21 of these genes showed reduced expression in lentiviral shWT1 knockdown experiments. Similar results were obtained by siWT1 knockdown in Wilms2 and Wilms3 cells. We can conclude from these data that the WT1 gain-of-function mutations in Wilms2 and Wilms3 cells are oncogenic events affecting gene expression of important cell cycle regulators and in consequence increased proliferation of Wilms tumour cells. This is supported by our WT1 knockdown experiments, which showed that the mutant WT1 proteins promote proliferation of Wilms2 and Wilms3 tumour cells. This finding provides convincing experimental evidence that the WT1 mutations may contribute to the development of Wilms tumours in vivo.

A large number of cell cycle regulatory genes are also principal targets of p53-mediated repression (38,39). Transcriptional repression mediated by p53 is complex and involves other transcription factors like E2F1 and NFY (38). Mutant WT1 proteins can activate the transcription of cell cycle genes that are repressed by p53. These include CCNB1, CDK1, PLK1, CDC20, MAD2a and others (3943). The promoter of the Cyclin B (CCNB1) gene for example contains a p53 head-to-tail response element, and binding of a p53 dimer to this response element is sufficient to abrogate transcription (40). We show here that mutant WT1 can activate CCNB1 and CCNB2 expression, and this suggests that the mutant protein can antagonize the repression function of p53 at the Cyclin B promoter. Future experiments will elucidate how the activation of the Cyclin B1 gene expression is brought about. WT1 interacts with p53 and modulates its function (22). Mutant WT1Wilms2 and WT1Wilms3 proteins can bind to p53 in vitro, and it is possible that this property enables WT1Wilms3 to neutralize p53-dependent repression of the Cyclin B promoter in vivo. The major phenotype of the WT1 mutation is the gain of a transactivation function, and an interaction of mutant WT1 with p53 at the Cyclin B1 promoter might enable this transactivation function to overcome repression by p53. Another possible scenario is the sequestration of p53 by mutant WT1 in the cytoplasm, resulting in loss of p53 function in the nucleus. In the context of this issue, it is important to study in future experiments, whether mutant WT1 proteins can be detected in the nucleus, even in minor amounts. Our immunofluorescence analysis (Supplementary Material, Fig. S1) and recent work (27) showed a predominant cytoplasmic localization of mutant WT1 proteins. Despite of this, it is possible that small amounts of these proteins are transported to the nucleus by a piggyback mechanism after association with other nuclear proteins. This is a speculative argument at this stage; however, a mutant WT1 protein lacking the ZF domain (WT1-delZ) was identified in nuclei of transfected osteosarcoma cells (44).

We have shown that Wilms2 and Wilms3 proteins are able to interact with RNA. It is possible that cytoplasmic localization of mutant WT1 proteins affects gene expression by altering the metabolism of RNA in the cells. This is, however, a complex issue and investigations addressing the functions of cytoplasmic mutant WT1 proteins are beyond the scope of the current work. In addition, we provided evidence based on an analysis using the MetaCore program ‘What are the key transcription factors and target genes in my data?’ that mutant WT1 modulates the activities of selected transcription factors also in the context of cell cycle-related gene expression. Again these results revealed an additional level of complexity, and this will be addressed in future studies.

Our WT1 knockdown data show that mutant WT1 inhibits the expression of a large number of genes from signalling pathways including several members of the canonical Wnt and planar cell polarity signalling pathways. The regulation of these pathway activities is complex, and in some cases, gene expression levels of inhibitors (e.g. AXIN2) can be up-regulated in the active state of the pathway. To ultimately determine activity of a given pathway of interest, functional studies are required, which are beyond the scope of this investigation and will be part of future work. With regard to Wnt signalling, it is important to note that wild-type WT1 inhibits Wnt signalling because it competes with TCF4 for binding to CBP (45). Moreover, depletion of wild-type WT1 results in up-regulation of CTNNB1 expression and WT1 negatively regulates Wnt signalling in Sertoli cells (46). Mutant WT1Wilms3 down-regulates CTNNB1 expression that might also inhibit the Wnt signalling pathway. In this context, it is of special interest to note that Wilms3 cells express wild-type CTNNB1. Therefore, Wilms3 tumour cells can survive without CTNNB1 mutations that are a common phenotype of Wilms tumours where WT1 is mutated. Mutant WT1Wilms3 was classified as a gain-of-function mutation with oncogenic potential, and the WT1 knockdown data showed that the growth and survival of Wilms3 tumour cells is dependent on the expression of WT1Wilms3.

Furthermore, several genes that are repressed by mutant WT1 are involved in chromatin modification and some are positive and negative regulators of MAPK signalling. It was reported that wild-type WT1 negatively regulates the RAS/MAPK signalling pathway by inducing the expression of MKP3 (DUSP6), a phosphatase and negative regulator of this pathway (47). Here we found that mutant WT1 inhibits the expression of the dual specificity phosphatase genes DUSP3 and DUSP16. Another group of genes that are down-regulated by mutant WT1 are differentiation regulating genes, e.g. GREM2, HMOX1, MYO5a, MYO9B, GDNF, NES, NKX3–1, PODXL, HGFL and SNPO. By using mouse models, it was shown that PODXL is a WT1 target gene with high expression levels in podocytes and highly differentiated kidney epithelial cells (48). PODXL gene expression is induced by WT1 and repressed by p53 (49). Wild-type WT1 activates PODXL by direct binding to its promoter, and WT1 and PODXL are co-expressed in the developing kidney (50). We have reported recently that WT1 and BASP1 interact with the promoter of PODXL, BASP1 inhibiting WT1. During differentiation, BASP1 promoter binding is down-regulated and in consequence WT1 can activate PODXL1 gene expression (51). Since the expression of PODXL is inhibited by WT1Wilms3 and PODXL expression is increased during kidney differentiation, one could argue that mutant WT1 keeps the tumour cells in an undifferentiated or less differentiated state.

In summary, we have identified novel gain-of-function mutations of WT1 that occurred in vivo during the development of Wilms tumours in patients. These mutations may drive tumour development as they promote cell proliferation in Wilms tumour cells. This result is of special interest for the biology of Wilms tumours as an ablation of Wt1 alone does not induce Wilms tumours in transgenic mice. However, in a transgenic mouse model with carrying a WT1 truncation mutation (396 stop), one animal developed a WT (52). It is interesting that in this Wilms tumour, one allele carried the targeted mutation and the other allele showed an exon 9 skipping event. Therefore, this tumour developed when both WT1 alleles carried mutations, corresponding to our Wilms tumour cell lines, where both alleles harbour the same the mutation (52). These data are reminiscent of the situation with another tumour suppressor, p53. A mouse model for pancreatic cancer revealed that a mutant form of p53 had an increased tumorigenicity through a gain-of-function mechanism. In this case, mutant p53 had extra functions over and above the loss of p53 in a tumour model (53).

## MATERIALS AND METHODS

### Cell culture

Human mesenchymal stem cells were purchased from Lonza and maintained in MSCGM (Lonza). The primary Wilms tumour cell lines Wilms2 and Wilms3 were described previously (27) and were cultivated in MSCGM with a medium change every 3 days.

### Plasmids and lentiviral expression vectors

Full-length human wild-type and mutant (p.V432fsX87) WT1 genes with an optimal Kozak sequence in front of the start ATG were cloned through PCR from mutant WT1Wilms3++ and WT1Wilm3+− isoforms in pCRII-TOPO vectors and from wild-type WT1++ and WT1+− isoforms in pDNA3.3-Topo-TA vectors. Primers for the PCR were Kozak-WT1-for (5′ AACGAATTCACCACCATGGGCTCCGACGTGCGG 3′) and wt-WT1-rev (5′ GTTGAATTCTCAAAGCGCCAGCTGGAGTT 3′) or mut-WT1-rev (5′ CAGTCGACTCAACTAAAAGTAGGCAGGGC 3′). The PCR products were cloned into the pCRII-TOPO vector and excised by digestion with EcoRI (wild-type WT1) or KpnI/XbaI (mutant WT1) and ligated into the pENTR4 vector [Addgene, 17 424 (54)], digested with the same restriction enzymes. The wild-type and mutant WT1 isoforms were finally cloned into the pLenti CMV Puro Dest vector [Addgene, 17 452 (54)] through the Gateway Technology as described in the ‘Gateway LR Clonase II Enzyme Mix’ protocol (Invitrogen). As a control for the transduction of expression vectors, a FOP luciferase pLenti X1 dest puro vector was used, containing the FOP expression cassette, cloned from the M51 super 8 FOP Flash [Addgene (55)]. All vector constructs were verified by sequencing.

The construction of a plasmid vector used to express the ZF domain of wild-type WT1 has been described previously (56). The cDNA used in this study to express the ZF domain (−KTS) of Wilms3 was constructed by the polymerase chain reaction (PCR) using an upstream primer at the start of ZF 1, containing a NcoI recognition sequence and a downstream primer including the termination codon for the Wilms3 reading frame, containing a EcoRI recognition sequence. PCR products were digested with NcoI and EcoRI and ligated into plasmid pET30a, digested with the same restriction enzymes. The NcoI site puts the Wilms3 cDNA in frame with the vector-included his-tag. Plasmids were isolated from appropriate colonies, and the correct plasmid constructs were confirmed by DNA sequencing.

The ‘Mission shRNA Plasmid DNA’ with a pLKO.1-puro backbone used in the WT1 knockdown experiments and the non-mammalian shRNA control used in all transduction experiments were purchased from Sigma–Aldrich; shWT1 (exon7, TRCN0000040067) and sh scrambled (SHC002).

### Production of lentiviral viral particles, transduction and siRNA transfection

For virus production, HEK293 T cells were transfected with 6 µg of each plasmid DNA: packaging vector pczVSV-G (57), pCD NL-BH (57) and the desired expression vector in the presence of 45 µg PEI (Polyethyleneimine, branched, Aldrich). As internal transfection control, a lentiviral GFP-expression Vector (pCL7EGwo, obtained from Dr H. Hanenberg) was transfected in each experiment. The medium was changed after 1 day to IMDM with 10% FCS and 1% penicillin/streptomycin, and 48 h after transfection, viral supernatants were harvested, filtered (0.45 µm), concentrated by centrifugation and cryo-conserved.

WT1 knockdown experiments were performed with (1) transfection with siRNA oligonucleotides (siWT1) and (2) infection with lentiviruses expressing shRNA (shWT1). For siWT1 knockdown, Wilms2 and Wilms3 cells were seeded in 24-well plates (4 × 104 cells/well), transfected after 24 h with the ‘HiPerfect transfection reagent’ (Qiagen) with a combination of two different siRNAs (Qiagen: Hs_WT1_1_HP in 3′ UTR and Hs_WT1_8_HP in exon 6) and a non-silencing control siRNA, Alexa Fluor 488 labelled (Qiagen). Total RNA was isolated at the indicated time points.

For shWT1 experiments, Wilms3 cells were seeded in a 6-well plate (5 × 104 cells/well). After 24 h, the cells were infected with shWT1 lentiviruses or as control with lentiviruses expressing a non-silencing shRNA in the presence of polybrene (5 µg/ml, H9268, Sigma), and the medium was changed after 4 h. RNA was isolated 48 h after transduction. All experiments were performed in duplicates.

Overexpression of wild-type and mutant proteins in MSCs was achieved by seeding 1 × 105 MSC in 6-well plates and infection after 24 h with 2 ml of the respective lentiviruses in the presence of polybrene (1 mg/ml). For a biological replicate, the experiments were done in duplicates. The medium was changed after 24 h to MSCGM, and RNA was isolated at 48 h. The negative control was a scrambled lentiviral vector.

### RNA isolation, gene expression analysis and Q-RT-PCR

Total RNA was isolated after WT1 knockdown or overexpression using the RNeasy Mini Kit (Qiagen). Labelling of total RNA in the Two-Colour format was performed as described by the manufacturer (Agilent Technologies). The cRNA was hybridized to 4 × 44 K ‘Whole Human Genome Oligo Microarrays’ (Agilent, V1), and ‘Spike-In positive controls’ (Agilent) were added as internal controls. Expression arrays were scanned (Agilent Technologies Scanner G2505B) and quantified using the Agilent Feature Extraction V10.1.1.1 Software. The software package LIMMA (R-library, bioconductor, http://www.R-project.org) was used for statistical analysis and the search for differentially expressed genes. The gene expression data were submitted and are found at GEO, GSE54635.

For Q-RT-PCR analyses, cDNAs were synthesized using TaqMan Reverse Transcription Reagents (Applied Biosystems). All Q-RT-PCR experiments were performed in triplicates using the TaqMan gene expression assay probes from Applied Biosystems for WT1 (Hs00240913_m1), MESDC2 (Hs00398059_m1), GDF5 (Hs00167060_m1) CASP7 (Hs01029847-_m1), PODXL (hCG18687) and JARID2 (Hs1004460_m1) with the TaqMan Universal PCR master Mix (no AmpEraseUNG) or Brilliant II QPCR Master Mix with Low Rox (Agilent Technologies). In the siWT1 experiments, the expression levels were normalized with the 18SRNA (Hs99999901_s1), in the shWT1 experiments with GAPDH (4333764 T). The WT1 knockdown was normalized versus ACTB (4333762F) and the overexpression versus RER1 (Hs00199824_m1). The reason for using these different calibrator probes was that we noted a regulation of the calibrator gene in some of these experiments, and therefore, we have identified a new gene that was basically never regulated in any of our gene expression studies: RER1. The siWT1 Q-RT-PCRs were run on an ABI PRISM 7900HT, and the Q-RT-PCRs from shWT1 and expression of WT1 in MSC were run on a Mx3000P Sequence Detection System (Stratagene).

### Microarray data analysis

Microarray data analysis was performed using the MetaCore software suite (http://thomsonreuters.com/metacore/) to identify biological processes associated with gene expression profiles. For analysis of gene expression profiles, normalized microarray data were manually curated before uploading individual data files. We selected only named genes from datasets and used a cut-off of 200 fluorescence intensity and a fold change of 1.5. The Agilent Identification numbers of selected genes were then uploaded and analysed using the MetaCore software program Enrichment Analysis (EA) for Datasets:

Enrichment analysis consists of mapping gene IDs of the dataset(s) of interest onto gene IDs in entities (terms) of built-in functional ontologies represented in MetaCore™/MetaDrug™ by pathway maps and networks. Mapping procedure involves calculating statistical relevance of the matches found. We use both ontologies, which are the result of our in-house extensive classification effort, and public ontologies such as Gene Ontology (GO, www.geneontology.org). Ontologies are represented by canonical pathway maps, cellular process networks, disease biomarker networks, drug target networks, toxicity networks and metabolic networks. GO ontologies include GO processes, GO molecular functions and GO localizations.

Statistical relevance of the found ontology matches is calculated as P-value, or a probability of a match to occur by chance, given the size of the database. The lower the P-value, the higher is the ‘non-randomness’ of finding the intersection between the dataset and the particular ontology. That, in turn, translates into a higher rating for the term matched. In particular, the more genes/proteins belong to a process/pathway, the lower the P-value. MetaCore program ‘What are the key transcription factors and target genes in my data?’ uses a database containing all published information regarding transcription factors and their target genes. We uploaded the up- or down-regulated genes from shWT1 knockdown experiments in Wilms3 cells as the input list. The transcription regulation workflow generates sub-networks centred on transcription factors. Sub-networks are ranked by a P-value and interpreted in terms of Gene Ontology. The resulted sub-networks of the whole network are displayed in Supplementary Material, Table S6 and Table S7, a network list table, in the order reflecting their statistical significance in terms of P-values.

### Cell proliferation after WT1 knockdown

For the determination of the proliferation rates of Wilms2 and Wilms3 cells after WT1 knockdown, 5 × 104 cells per well were seeded in MSCG medium in a 6-well plate. After transduction with shWT1 and scrambled control lentiviral particles, the cells were counted in a Neubauer chamber 24 h, 48 h, 72 and 96 h. Each time point was analysed in duplicate. RNA was extracted from these cells for gene expression analysis.

### Expression and purification of recombinant ZF proteins

Preparations of N-terminal his-tagged ZF proteins were carried out as described previously (58). Protein purity was confirmed by SDS-polyacrylamide gel electrophoresis (PAGE), and the concentration of each protein preparation was determined by the method of Bradford.

### Radiolabelling of RNA ligands and equilibrium binding of RNAs to ZF proteins

Preparation of radiolabelled RNA aptamers by in vitro transcription was as described previously (29). The apparent association constants for the binding of radiolabelled RNAs to ZF proteins were determined using a double-filter binding assay. The binding buffer consisted of 20 mm Tris–HCl pH 7.5 (20°C), 5 mm MgCl2, 100 mm KCl, 10 µm ZnCl2, 0.5 mm Tris (2-carboxyethyl)phosphine hydrochloride, 100 µg/ml BSA and 1 µg/ml polydIdC. The affinity of each ZF protein for RNA was determined using three or more independent assays. Apparent dissociation constants (Kd) for the binding of the mutant and wild-type proteins to RNA were calculated by fitting the data to a simple bimolecular equilibrium model using the general curve fitting function of Kaleidagraph software (Synergy Software, Reading, PA, USA) and the following equation:

(1)
$$\displaystyle{{[\hbox{RNA-protein}]} \over {{{[\hbox{RNA}]}_{{\rm total}}}}} = \displaystyle{{{{[\hbox{protein}]}_{{\rm total}}}} \over {{{[\hbox{protein}]}_{{\rm total}}} + {K_{\rm d}}}}$$
where [RNA]total<< Kd and [RNA-protein]/[RNA]total is reported as the fraction of RNA bound. Association constant (Ka) values were derived as the reciprocal of the measured Kd values and are reported as the mean of three independent determinations with the associated standard deviations. Relative affinities were arrived at by dividing the apparent Ka for the mutant protein by the apparent Ka for the wild-type protein determined in parallel, or by dividing the apparent Ka for the mutant RNA by the apparent Ka for the wild-type RNA determined in parallel. The errors for relative affinities are given by the expression σ = {(σ 1/M1)2 + (σ2/M2)2}1/2 × M2/M1, where M1 and M2 are the respective association constants for wild-type and mutant protein and the σ values are the corresponding standard deviations for these determinations.

### Analysis of endogenous p53 activity in Wilms cells

Endogenous p53 activity in Wilms3 cells was measured with the p53-responsive reporter pG13-Luc [13 p53 binding sites, Addgene, 16 442 (59)] and the non-responsive reporter pG15-Luc [15 mutant p53 binding sites, Addgene, 16 443 (59)]. As control, Wilms3 cells were transfected with a promoter-less pGL3-basic luciferase vector. All cells were co-transfected with pGL4.75 (Renilla-Luc) for normalization. Cells were lysed after 24 h in 1× Passive Lysis Buffer (Promega), and the luciferase activity was measured with the Beetle-/Renilla-Juice Big Kit (PJK GmbH, Germany) as described by the manufacturer. The relative p53 luciferase activity was determined as quotient of the firefly-luciferase and renilla-luciferase activity. The analysis was performed in triplicates.

For the determination of the status of the p53 protein, cell lysates from Wilms2 and Wilms3 cell lines were used for probing the Human Phospho-Kinase Array (Catalog # ARY003B, R&D). The analysis was performed as described by the manufacturer (R&D).

### Protein interaction of WT1 and p53

HeLa cells were cultured in Dulbecco's modified Eagle's medium and transfected with Effectene (Qiagen) with pAcGFP vectors expressing N-terminally GFP-tagged WT1Wilms3 or WT1Wilms2 or wild-type WT1 fusion proteins (27). Preparation of whole cell extracts and IP western blot were carried out as described previously (60). Antibodies anti-WT1 (F6) and anti-p53 (DO1) sc-126 were obtained from Santa Cruz Biotechnology.

## SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

Conflict of Interest statement. None declared.

## FUNDING

This work was supported by a grant from the Deutsche Forschungsgemeinschaft, Germany (RO 501/12-2 to B.R.-P.); from the National Institute of General Medical Sciences, USA (NIH, 1R01GM098609, to S.R.) and from the Natural Sciences and Engineering Research Council of Canada (to P.R.). Funding to pay the Open Access publication charges for this article was provided partially by DFG funds and the Institute of Human Genetics.

## REFERENCES

1
Call
K.M.
Glaser
T.
Ito
C.Y.
Buckler
A.J.
Pelletier
J.
Haber
D.A.
Rose
E.A.
Kral
A.
Yeger
H.
Lewis
W.H.
Isolation and characterization of a zinc finger polypeptide gene at the human chromosome 11 Wilms' tumor locus
Cell

1990
60
509
520
2
Gessler
M.
Poustka
A.
Cavenee
W.
Neve
R.L.
Orkin
S.H.
Bruns
G.A.
Homozygous deletion in Wilms tumours of a zinc-finger gene identified by chromosome jumping
Nature

1990
343
774
778
3
Hastie
N.D.
Wilms' tumour gene and function
Curr. Opin. Genet. Dev.

1993
3
408
413
4
S.L.
Cook
D.M.
Morris
J.F.
Gashler
A.
Sukhatme
V.P.
Rauscher
F.J.
III
Transcriptional repression mediated by the WT1 Wilms tumor gene product
Science

1991
253
1550
1553
5
Kennedy
D.
Ramsdale
T.
Mattick
J.
Little
M.
An RNA recognition motif in Wilms' tumour protein (WT1) revealed by structural modelling
Nat. Genet.

1996
12
329
331
6
Englert
C.
WT1—more than a transcription factor?
TIBS

1998
23
389
393
7
Roberts
S.G.E.
Transcriptional regulation by WT1 in development
Curr. Opin. Genet. Dev.

2005
15
542
547
8
Rauscher
F.J.
Morris
J.F.
Tournay
O.E.
Cook
D.M.
Curran
T.
Binding of the Wilms' tumor locus zinc finger protein to the EGR-1 consensus sequence
Science

1990
250
1259
1262
9
Wang
Z.-Y.
Qiu
Q.Q.
Enger
K.T.
Deuel
T.
A second transcriptionally active DNA-binding site for the Wilms tumor gene product, WT1

1993
90
8896
8900
10
Bickmore
W.A.
Oghene
K.
Little
M.H.
Seawright
A.
van Heyningen
V.
Hastie
N.D.
Modulation of DNA binding specificity by alternative splicing of the Wilms tumor wt1 gene transcript
Science

1992
257
235
237
11
Morrison
A.A.
Viney
R.L.
M.R.
The post-transcriptional roles of WT1, a multifunctional zinc-finger protein
Biochim. Biophys. Acta.

2008
1785
55
62
12
S.H.
Charlieu
J.P.
Miyagawa
K.
Engelkamp
D.
M.
Ross
A.
Cuzin
F.
van Heyningen
V.
Hastie
N.D.
Subnuclear localization of WT1 in splicing or transcription factor domains is regulated by alternative splicing
Cell

1995
81
391
401
13
Davies
R.C.
Calvio
C.
Bratt
E.
S.H.
Lamond
A.I.
Hastie
N.D.
WT1 interacts with the splicing factor U2AF65 in an isoform-dependent manner and can be incorporated into spliceosomes
Genes Dev.

1998
12
3217
3225
14
Bardeesy
N.
Pelletier
J.
Overlapping RNA and DNA binding domains of the wt1 tumor suppressor gene product
Nucleic Acids Res.

1998
26
1784
1792
15
Caricasole
A.
Duarte
A.
S.H.
Hastie
N.D.
Little
M.
Holmes
G.
Todorov
I.
Ward
A.
RNA binding by the Wilms tumor suppressor zinc finger proteins

1996
93
7562
7566
16
Weiss
T.C.
Romaniuk
P.J.
Contribution of individual amino acids to the RNA binding activity of the Wilms' tumor suppressor protein WT1
Biochemistry

2009
48
148
155
17
M.R.
Slight
J.
Mc Ghee
S.
Hastie
N.D.
Presence of WT1, the Wilm's tumor suppressor gene product, in nuclear poly(A)(+) ribonucleoprotein
J. Biol. Chem.

1999
274
36520
36526
18
Morrison
A.A.
Venables
J.P.
Dellaire
G.
M.R.
The Wilms tumour suppressor protein WT1 (+KTS isoform) binds alpha-actinin 1 mRNA via its zinc-finger domain
Biochem. Cell Biol.

2006
84
789
798
19
Iben
S.
Royer-Pokora
B.
Analysis of native WT1 protein from frozen human kidney and Wilms' tumors
Oncogene

1999
18
2533
2536
20
Roberts
S.G.E.
The modulation of WT1 transcription function by cofactors
Biochem. Soc. Symp.

2006
73
191
201
21
Carpenter
B.
Hill
K.J.
Charalambous
M.
Wagner
K.J.
Lahiri
D.
James
D.I.
Andersen
J.S.
Schumacher
V.
Royer-Pokora
B.
Mann
M.
Ward
A.
Roberts
S.G.E.
BASP1 is a transcriptional cosuppressor for the Wilms' tumor suppressor protein WT1
Mol. Cell. Biol.

2004
24
537
549
22
Maheswaran
S.
Englert
C.
Bennett
P.
Heinrich
G.
Haber
D.A.
The WT1 gene product stabilizes p53 and inhibits p53-mediated apoptosis
Genes Dev.

1995
9
2143
2156
23
Reddy
J.C.
Morris
J.C.
Wang
J.
English
M.A.
Haber
D.A.
Shi
Y.
Licht
J.D.
WT1-mediated transcriptional activation is inhibited by dominant negative mutant proteins
J. Biol. Chem.

1995
270
10878
10884
24
Englert
C.
Vidal
M.
Maheswaran
S.
Ge
Y.
Ezzell
R.M.
Isselbacher
K.J.
Haber
D.A.
Truncated WT1 mutants alter the subnuclear localization of the wild-type protein

1995
92
11960
11964
25
Depping
R.
Schindler
S.G.
Jacobi
C.
Kirschner
K.M.
Scholz
H.
Nuclear transport of Wilms' tumour protein Wt1 involves importins α and β
Cell Physiol. Biochem.

2012
29
223
232
26
Schumacher
V.
Schuhen
S.
Sonner
S.
Weirich
A.
Leuschner
I.
Harms
D.
Licht
J.
Roberts
S.
Royer-Pokora
B.
Two molecular subgroups of Wilms' tumors with or without WT1 mutations
Clin. Cancer Res.

2003
9
2005
2014
27
Royer-Pokora
B.
Busch
M.
Beier
M.
Duhme
C.
de Torres
C.
Mora
J.
Brandt
A.
Royer
H.-D.
Wilms tumor cells with WT1 mutations have characteristic features of mesenchymal stem cells and express molecular markers of paraxial mesoderm
Hum. Mol. Genet.

2010
19
1651
1668
28
Mundlos
S.
Pelletier
J.
Darveau
A.
Bachmann
M.
Winterpacht
A.
Zabel
B.
Nuclear localization of the protein encoded by the Wilms' tumor gene WT1 in embryonic and adult tissues
Development

1993
119
1329
1341
29
Zhai
G.
Iskandar
M.
Barilla
K.
Romaniuk
P.J.
Characterization of RNA aptamer binding by the Wilms' tumor suppressor protein WT1
Biochemistry

2001
40
2032
2040
30
Coutts
A.S.
La Thangue
A.B.
The p53 response: emerging levels of co-factor complexity
Biochem. Biophys. Res. Commun.

2005
331
778
785
31
Schumacher
V.
Schneider
S.
Figge
A.
Wildhardt
G.
Harms
D.
Schmidt
D.
Weirich
A.
Ludwig
R.
Royer-Pokora
B.
Correlation of germ-line mutations and two-hit inactivation of the WT1 gene with Wilms tumors of stromal-predominant histology

1997
94
3972
3977
32
Huff
V.
Wilms' tumours: about tumour suppressor genes, an oncogene and a chameleon gene
Nat. Rev. Cancer

2011
11
111
121
33
Yang
L.
Han
Y.
Suarez Saiz
F.
Minden
M.D.
A tumor suppressor and oncogene: the WT1 story
Leukemia

2007
21
868
876
34
Weinstein
I.B.
Joe
A.
Cancer Res.

2008
68
3077
3080
35
Vidovic
K.
Ullmark
T.
Rosberg
B.
Lennartsson
A.
Olofsson
T.
Nilsson
B.
Gullberg
U.
Leukemia associated mutant Wilms' tumor gene 1 protein promotes expansion of human hematopoietic progenitor cells
Leuk Res.

2013
37
1341
1349
36
Svedberg
H.
Richter
J.
Gullberg
U.
Forced expression of the Wilms tumor 1 (WT1) gene inhibits proliferation of human hematopoietic CD34(+) progenitor cells
Leukemia

2001
15
1914
1922
37
Whitfield
M.L.
George
L.K.
Grant
D.
Perou
C.M.
Common markers of proliferation
Nat. Rev. Cancer

2006
6
99
106
38
Brosh
R.
Rotter
V.
Transcriptional control of the proliferation cluster by the tumor suppressor p53
Mol. Biosyst.

2010
6
17
29
39
Spurgers
K.B.
Gold
D.L.
Coombes
K.R.
Bohnenstiehl
N.L.
Mullins
B.
Meyn
R.E.
Logothetis
C.J.
McDonnell
T.J.
Identification of cell cycle regulatory genes as principal targets of p53-mediated transcriptional repression
J. Biol. Chem.

2006
281
25134
25142
40
Lipski
R.
Lippincott
D.J.
Durden
B.C.
Kaplan
A.R.
Keiser
H.E.
Park
J.H.
Levesque
A.A.
p53 Dimers associate with a head-to-tail response element to repress cyclin B transcription
PLoS One

2012
7
e42615
1–9
41
McKenzie
L.
King
S.
Marcar
L.
Nicol
S.
Dias
S.S.
Schumm
K.
Robertson
P.
Bourdon
J.C.
Perkins
N.
Fuller-Pace
F.
Meek
D.W.
p53-dependent repression of polo-like kinase-1 (PLK1)
Cell Cycle

2010
9
4200
4212
42
Kidokoro
T.
Tanikawa
C.
Furukawa
Y.
Katagiri
T.
Nakamura
Y.
Matsuda
K.
CDC20, a potential cancer therapeutic target, is negatively regulated by p53
Oncogene

2008
27
1562
1571
43
Banerjee
T.
Nath
S.
Roychoudhury
S.
DNA damage induced p53 downregulates Cdc20 by direct binding to its promoter causing chromatin remodeling
Nucleic Acids Res.

2009
37
2688
2698
44
Englert
C.
Vidal
M.
Maheswaran
S.
Ge
Y.
Ezzel
R.M.
Isselbacher
K.J.
Haber
D.A.
Truncated WT1 mutants alter the subnuclear localization of the wild-type protein

1995
92
11960
11964
45
Kim
M.K.-H.
McGarry
T.J.
O Broin
P.
Flatow
J.M.
Golden
A.A.-J.
Licht
J.D.
An integrated genome screen identifies the Wnt signalling pathway as a major target of WT1

2009
106
11154
11159
46
Chang
H.
Gao
F.
Guillou
F.
Taketo
M.M.
Huff
V.
Behringer
R.R.
Wt1 negatively regulates beta-catenin signaling during testis development
Development

2008
135
1875
1885
47
Morrison
D.J.
Kim
M.K.H.
Berkofsky-Fessler
W.
Licht
J.D.
WT1 induction of mitogen-activated protein kinase phosphatase 3 represents a novel mechanism of growth suppression
Mol. Cancer Res.

2008
6
1225
1231
48
Guo
J.K.
Menke
A.L.
Gubler
M.C.
Clarke
A.R.
Harrison
D.
Hammes
A.
Hastie
N.D.
Schedl
A.
WT1 is a key regulator of podocyte function: reduced expression levels cause crescentic glomerulonephritis and mesangial sclerosis
Hum. Mol. Genet.

2002
11
651
659
49
Stanhope-Baker
P.
Kessler
P.M.
Li
W.
Agarwal
M.L.
Williams
B.R.
The Wilms tumor suppressor-1 target gene podocalyxin is transcriptionally repressed by p53
J. Biol. Chem.

2004
279
33575
33585
50
Palmer
R.E.
Kotsianti
A.
B.
Boyd
T.
Gerald
W.
Haber
D.A.
WT1 regulates the expression of the major glomerular podocyte membrane protein Podocalyxin
Curr. Biol.

2001
11
1805
1809
51
Green
L.M.
Wagner
K.J.
Campbell
H.A.
K.
Roberts
S.G.
Dynamic interaction between WT1 and BASP1 in transcriptional regulation during differentiation
Nucleic Acids Res.

2009
37
431
440
52
Patek
C.E.
Little
M.H.
Fleming
S.
Miles
C.
Charlieu
J.P.
Clarke
A.R.
Miyagawa
K.
Christie
S.
Doig
J.
Harrison
D.J.
et al.
A zinc finger truncation of murine WT1 results in the characteristic urogenital abnormalities of Denys-Drash syndrome

1999
96
2931
2936
53
Morton
J.P.
Timpson
P.
Karim
S.A.
Ridgway
R.A.
Athineos
D.
Doyle
B.
Jamieson
N.B.
Oien
K.A.
Lowy
A.M.
Brunton
V.A.
et al.
Mutant p53 drives metastasis and overcomes growth arrest/senescence in pancreatic cancer

2010
107
246
251
54
Campeau
E.
Ruhl
V.E.
Rodier
F.
Smith
C.L.
Rahmberg
B.L.
Fuss
J.O.
Campisi
J.
Yaswen
P.
Cooper
P.K.
Kaufman
P.D.
A versatile viral system for expression and depletion of proteins in mammalian cells
PLoS One.

2009
4
e6529
55
Veeman
M.T.
Slusarski
D.C.
Kaykas
A.
Louie
S.H.
Moon
R.T.
Zebrafish prickle, a modulator of noncanonical Wnt/Fz signaling, regulates gastrulation movements
Curr Biol.

2003
13
680
685
56
Veldhoen
N.
You
Q.
Setzer
D.R.
Romaniuk
P.J.
Contribution of individual base pairs to the interaction of TFIIIA with the Xenopus 5S RNA gene
Biochemistry

1994
33
7568
7575
57
Hartmann
L.
Neveling
K.
Borkens
S.
Schneider
H.
Freund
M.
Grassman
E.
Theiss
S.
Wawer
A.
Burdach
S.
Auerbach
A.D.
et al.
Correct mRNA processing at a mutant TT splice donor in FANCC ameliorates the clinical phenotype in patients and is enhanced by delivery of suppressor U1 snRNAs
Am. J. Hum. Genet.

2010
87
480
493
58
Hamilton
T.B.
Barilla
B.
Romaniuk
P.
High affinity binding sites for the Wilms' tumour protein WT1 High affinity binding site suppressor protein WT1
Nucleic Acids Res.

1995
23
277
284
59
el-Deiry
W.S.
Tokino
T.
Velculescu
V.E.
Levy
D.B.
Parsons
R.
Trent
J.M.
Lin
D.
Mercer
W.E.
Kinzler
K.W.
Vogelstein
B.
WAF1, a potential mediator of p53 tumor suppression
Cell

1993
75
817
825
60
Goodfellow
S.J.
Rebello
M.R.
Toska
E.
Zeef
L.A.H.
Rudd
S.G.
Medler
K.F.
Roberts
S.G.E.
WT1 and its transcriptional cofactor BASP1 redirect the differentiation pathway of an established blood cell line
Biochem. J.

2011
435
113
125