WT1 Splicing Alterations in Wilms’ Tumors¹

WT³ or nephroblastoma is a childhood renal malignancy that accounts for approximately 6% of all childhood tumors and occurs with a frequency of 1 in 10,000 live births. Most cases are sporadic and unilateral. Bilateral cases account for 7% of cases, and a family history of WT is observed in 2% of cases. This tumor is also observed in association with several rare malformation syndromes: (a) WAGR syndrome linked to a constitutional 11p13 deletion; (b) Denys-Drash syndrome (diffuse mesangial sclerosis, male pseudohermaphroditism, and WT); and (c) Beckwith-Wiedemann syndrome (gigantism,macroglossia, omphalocele, hyperinsulinism, and predisposition to several tumors such as WT, adrenocortical carcinoma, and rhabdomyosarcoma).

The WT1 gene, which is located at 11p13, encodes a zinc finger transcription factor involved in renal and gonad development (1, 2, 3, 4). Exons 1–6 encode domains involved in transcriptional regulation, dimerization, and possibly RNA recognition,whereas exons 7–10 encode the four zinc fingers of the DNA-binding domain (5, 6, 7, 8, 9). Twenty-four potential protein isoforms may be synthesized due to: (a) two alternative splicing regions corresponding to the whole of exon 5 (17 amino acids) and to the three last codons of exon 9 (KTS), respectively (10);(b) a site of RNA editing at codon 281 in exon 6 (a C to T transition producing a leucine to proline substitution; Ref. 11); (c) a non-AUG initiation codon resulting in WT1 proteins with a higher molecular weight (12); and(d) an internal AUG initiation codon resulting in WT1 proteins with a lower molecular weight (13). Biochemical and genetic evidence is accumulating that the WT1(−KTS) and WT1(+KTS)isoforms have different functions. WT1(−KTS) behaves like a transcription factor and can regulate in vitro genes expressed during kidney development, including IGF2, PDGFA, EGFR,PAX-2, and WT1 (14, 15, 16, 17, 18). However, the physiological and functional significance of these regulations is still unknown. Analysis of differential expression profiles in cell lines transfected with WT1 plasmids identified other genes up-regulated by WT1(−KTS), such as RbAp46, a retinoblastoma-associated protein, and amphiregulin, a member of the epidermal growth factor family (19, 20). WT1(+KTS) proteins are supposed to be involved in RNA splicing, based on their subnuclear speckled distribution and physical association with splicing factors (7, 21, 22, 23). The presence of the both WT1(+KTS) and WT1(−KTS)isoforms is essential for normal urogenital development because intronic mutations that prevent the insertion of the KTS amino acids confer Frasier syndrome characterized by focal and segmental glomerular sclerosis, pseudohermaphroditism, and gonadoblastoma (24, 25). Only preliminary results have been obtained regarding the functional effect of the presence or absence of the 17 amino acids encoded by exon 5 (5, 20, 26, 27). At the cellular level, the balance between isoforms with and without the 17-amino acid insertion may be involved in the regulation of proliferation, differentiation, and apoptosis and the prevention of tumor formation (28, 29, 30, 31, 32, 33). In opposition to the KTS insertion that is present in all vertebrates, the 17-amino acid insertion is present only in placental mammals, suggesting a role in biological functions that appeared late in evolution (34).

The role of WT1 as a tumor suppressor gene has been clearly demonstrated in patients with WAGR syndrome (deletion of one allele) and Denys-Drash syndrome (missense changes in exons 8 or 9) and in some patients with cryptorchidism/hypospadias and/or WT (frameshift and nonsense mutations; Ref. 35). Mutation or loss of the normal allele is usually observed in the nephroblastomas developed by these patients.⁴Somatic mutations of WT1 were expected in sporadic cases of WT, but such mutations appeared to be rare, with a frequency of about 10% (36, 37, 38). Disruption of the alternative splicing of exon 5 and aberrant splicing of exon 2 have been reported in a small series of tumors, suggesting that splicing abnormalities could be involved in WT (39, 40, 41, 42). We evaluated in more detail the frequency of the various types of change to WT1 in WT by analyzing mRNA synthesis quantitatively and qualitatively in a large series of 50 tumors, 30 of which were unilateral unifocal WT with no associated malformation. We found changes in the WT1 expression in 90% of unilateral unifocal WT cases, with 63% showing splicing alterations. Disruption of exon 5 splicing was the most frequent alteration, but alteration of exon 9-KTS splicing with an increase in the amount of isoforms with the KTS domain was also observed in some tumors. The causal role of these alterations of WT1 isoform proportions in the tumorigenic process remains to be elucidated.

Fifty tumor samples were collected from 49 patients. Thirty tumors were sporadic unilateral WT. Patients 34, 79, and 114 had WAGR syndrome and a 11p13 deletion. Germ-line WT1 mutations were identified in patients 13 and 163 (35). Patients 52, 56,and 96 had Beckwith-Wiedemann syndrome, and patient 190 had hemihypertrophy. Patient 84 presented with 2q3 deletion, urogenital abnormalities, and mental retardation. Patients 2, 37, 117, and 164 had bilateral tumors, whereas patient 77 had bifocal WT, and patient 191 had multifocal WT. Patient 65 developed a late tumor, at 17 years of age. Patients 4, 92, 133, and 193 presented with renal rhabdoid tumor. The histological characteristics of the tumors are given in Table 1. Most tumor samples were obtained after chemotherapy, except for tumor samples 9T, 13T, 56T, 84T, 133T, and 190T. Normal control kidney samples were available for 15 patients, and NR sample was available for one patient (117NR). FK samples were obtained from 23-week-old (FK1) and 20-week-old (FK3) fetuses. RNA was extracted from frozen tissue using Trizol reagent according to the manufacturer’s protocol (Life Technologies, Inc.). FK (19–23 weeks)polyadenylated RNA was purchased from Clontech.

For analysis of mutations, DNA was extracted from either tumor samples or blood samples.

Five μg of each RNA were reverse transcribed with 200 units of SuperScript II reverse transcriptase and hexanucleotide random primers according to the manufacturer’s instructions (Life Technologies,Inc.). The cDNA reaction was diluted to a final volume of 100 μl, and 2 μl were used for PCR amplification. Several pairs of WT1 primers were used: (a) p125 (5′-CCCAACCACTCATTCAAG-3′) and p4AS(5′-CATTCAAGCTGGGATGTCAT-3′; (b) p386(5′-CTAACGCGCCCTACCTGCCC-3′) and p611 (5′-GGTGCAGCTGTCGGTGGGGG-3′);(c) pSPL4 (5′-CTTGAATGCATGACCTGGAA-3′) and pSPL10(5′-TTGGCCACCGACAGCTGAAG-3′); and (d) pSPL4 and p6AS(5′-TATTCTGTATTGGGCTCCGC-3′). The regions corresponding to the primers are presented in Fig. 1. Primers for HPRT were used as control: HPRT-exon 3 (5′-TGTGTGCTCAAGGGGGGC-3′)and HPRT-exon 7 (5′-CGTGGGGTCCTTTTCACC-3′). The PCR products were resolved by electrophoresis in 2% agarose or 6% polyacrylamide gels.

For each sample, four 2-fold serial dilutions of the cDNA were amplified for 26 cycles using WT1 and HPRT primers.[³²P]dCTP was added to the reaction mix for radiolabeling of the PCR products. The ratios of WT1:HPRT were determined by phosphorimaging and analysis with ImageQuant software(Molecular Dynamics). For each data point, the results of two to three successive cDNA dilutions in the nonsaturated range of the PCR were averaged.

For each sample, four PCR reactions were set up in parallel with one-fiftieth of the reverse transcribed product and amplified for 26,29, 32, and 35 cycles using primer pSPL10 and fluorescein-labeled primer pSPL4. Aliquots of the PCR products were run on Ready-Mix acrylamide gels for 400 min at 30 W on an ALF sequencer (Pharmacia Biotech). The volume of the aliquots (1–4 μl) was adjusted to allow reliable peak detection and to avoid saturation. Four fragments were resolved (502, 511, 552, and 561 bp in length), corresponding to isoforms A (exon 5−, KTS−), B (exon 5−, KTS+), C (exon 5+, KTS−),and D (exon 5+, KTS+), respectively. The area under the peaks was quantified using AlleleLinks software, and the amount of each isoform was evaluated relative to the amount of isoform A. Data were accepted when they were confirmed by independent experiments and different numbers of cycles. The exon 5+:exon 5− and the KTS+:KTS−ratios were calculated. Student’s t test was used to evaluate the statistical significance of the differences in these ratios.

The WT1 exons were sequenced directly from PCR fragments, as described previously (35), or after cloning with the TA cloning kit (Invitrogen). Primers were purchased from Life Technologies, Inc. or Amersham-Pharmacia Biotech for fluorescent primers. To characterize the abnormal splice variants in tumors 79T1,79T2, and 189T, we amplified a cDNA fragment using primers pSPL4 and pSPL10 and sequenced it using primer p1093 specific for exon 8 (Fig. 1). Sequences were resolved on an ALF sequencer (Pharmacia Biotech). For Southern blot analysis, genomic DNA was digested with PstI and hybridized with the cDNA probe WT33 (10). Quantification of the alleles was performed using a PhosphorImager (Molecular Dynamics) after hybridization with a control probe specific for the dystrophy myotonic protein kinase locus on chromosome 19.

A WT1 fragment spanning exons 2–4 was coamplified with a HPRT fragment as a control (Fig. 1). Forty tumors, a NR sample (117NR), normal mature kidneys, and FKs clearly expressed WT1. The expression of WT1 was 4-fold higher in FKs than in mature kidneys and was increased more than 6-fold in 19 tumors and in the NR sample (Table 1). We were unable to amplify the WT1-specific fragment from 10 tumors with a 35-cycle procedure, whereas the HPRT fragment was clearly present. Only a very faint WT1 band was obtained for tumors 82T and 180T. Deletion of exon 2 has been reported in WT xenograft cell lines and in a primary tumor (39, 41). We therefore thought it likely that deletion of this exon had occurred in some of the tumors, preventing hybridization with the exon 2 primer p125. Using primers p386 and p611 specific for exons 1 and 3,respectively, very little or no WT1 expression was detected in the 10 tumors (Table 1), confirming our initial findings.

The cDNAs from the 40 tumors expressing WT1, the NR sample (117NR), 15 NKS and 3 FKs were amplified using primers pSPL4 and pSPL10, which span the two alternative splice sites (Figs. 1 and 2). The relative proportions of the splice variants, exon 5+ versus exon 5− and KTS+versus KTS−, were evaluated in two to four independent experiments using the AlleleLinks program.

The values obtained for the normal tissues are presented in Table 2. The increase in both ratios with maturation of the kidney was statistically significant with a P of 0.01. It was mainly due to a relative increase in the amount of isoform D, which contains both alternatively spliced regions.

We defined five tumor groups based on the exon 5+:exon 5− ratio(Tables 1 and 3). Groups T1 and T2 included tumors with a ratio significantly lower than the mean ratios for NKS and FKs (T1 and FK, P = 0.001; T2 and FK, P = 0.01). Group T3 consisted of tumors with a ratio similar to that seen in FKs. Group T4 consisted of tumors with a ratio similar to that seen in NKS. Group T5 consisted of tumors with a ratio higher than that in NKS and FKs (T5 and NK, P = 0.001). Whether the ratio increased or decreased depended on the relative increase or decrease in the amount of the two isoforms with exon 5(isoforms C and D). To visualize a possible correlation between the overall level of WT1 expression and the alterations in WT1 isoform abundance, we quantified the amount of isoforms with and without exon 5 relative to HPRT. Interestingly, in tumors with an overall increased expression of WT1, overexpression concerned all of the isoforms,but with an even higher level for isoforms lacking exon 5 (Fig. 3).

The KTS+:KTS− ratio was in the normal range in 31 tumors but was significantly higher in 9 tumors belonging to groups T1, T2, and T4.(Table 3). This increase was due to the relative increase in the amount of the two isoforms with KTS (isoforms B and D).

Sequencing of the whole gene in tumors 2T, 37T, and 117T revealed no mutation in the coding sequence, demonstrating that isoform imbalance in these tumors is not related to exonic or consensus splice sequence mutations (data not shown).

In tumors 73T, 79T1, 79T2, and 189T, PCR amplification of cDNA using primers pSPL4 and pSPL10 produced fragments that were abnormal in size. In tumor 73T, in addition to the four normal fragments, four additional fragments, each of which was 80 bp longer than the corresponding normal fragment, were coamplified. We mapped the insertion to the region between primer pSPL4 in exon 4 and primer p6AS in exon 6 (data not shown). Tumors 79T1 and 79T2, both of which were developed by patient 79 with WAGR syndrome, and sporadic tumor 189T produced four variant transcripts 150 bp shorter than the normal transcripts (Fig. 2). The abnormal transcripts were also detected in the NK sample 189NK. For the three tumors, sequencing of the cDNA fragments showed that exon 7 was totally absent (data not shown). Sequencing of exon 7 and the adjacent intron sequences of DNA from tumor 79T2 revealed a 60-bp deletion spanning the beginning of the exon, from position −23 to position +39. Sequencing of exon 7 in DNA from tumor 189T identified a C→A change at position −3 for one allele and a CAGT duplication at position 951 of the coding sequence in the other. The intronic mutation was present in the paired NK DNA.

Tumors from WAGR patients were systematically screened for the presence of somatic mutations. We identified an insertion of one base (A) in exon 3 in the tumor from patient 34 and an insertion of five bases(GCGGC) in exon 1 in the tumor from patient 114. We also used Southern blotting to look for rearrangements and deletions of the gene in the DNA of tumors that did not express WT1. Tumor 10T displayed a homozygous deletion, and tumor 100T displayed a large rearrangement (Table 1).

The role of WT1 in sporadic WT is still poorly understood. To evaluate in more detail the various types of changes to WT1 at the somatic level, we analyzed this gene in a series of 50 tumors, 46 WTs and 4 renal rhabdoid tumors. Thirty WTs were unilateral unifocal tumors with no associated malformation and were therefore a priori considered to be sporadic. Unilateral multifocal tumors and bilateral tumors were not included in the sporadic group because the association with a predisposing event was unclear. Although sequencing of the whole WT1 coding sequence revealed no germ-line WT1 mutation in patients 2, 37, and 117 with bilateral WT, we cannot exclude the possibility of a mutation in another predisposing gene possibly involved in the regulation or splicing of WT1. We also excluded patient 163 with unilateral WT from the sporadic group because she carried a germ-line WT1 mutation (35).

Our analysis did not allow us to determine the frequency of somatic mutations in the whole series, but several mutations were identified following characterization of changes in expression. Tumor 34T in a patient with WAGR syndrome carried a 1-bp insertion in its unique WT1 allele, which may account for the absence of transcripts by nonsense-mediated mRNA decay (43). The homozygous deletion of the gene in sporadic tumor 10T clearly account for the lack of WT1 expression, but it is unclear how the large rearrangement detected in sporadic tumor 100T could account for the lack of expression. The identification of transcripts of abnormal size made it possible to characterize somatic mutations in four tumors. In tumors 79T1 and 79T2, both from a WAGR patient, the second inactivating mutation was identified as a somatic deletion resulting in abnormal splicing and loss of exon 7. In tumor 189T, two different mutations accounted for the abnormal splicing of exon 7. The first mutation, at position −3 in the “donor site” consensus sequence, was also detected in the paired NK in which abnormal splicing also occurred. It is not known whether this is a germ-line mutation or a first somatic mutation in the renal tissue. The second mutation, a 4-bp insertion, was present only in the tumor DNA and is also likely to affect splicing because far fewer normal-sized transcripts were detected in the tumor than in the paired NK. In the three tumors, abnormal splicing not only resulted in the loss of the first zinc finger encoded by exon 7 but also gave rise to an aberrant reading frame at the 5′ end of exon 8 ending in a stop codon 32 amino acids downstream, thus resulting in proteins with no zinc finger domain. In sporadic tumor 73T, both alleles were affected, one by an insertion of 80 bp into the coding sequence and the other by an imbalance in the isoform ratio. We identified somatic mutations in all of the tumors developed by patients with WAGR syndrome.

Three of the 10 tumors that did not express WT1 were renal rhabdoid tumors. Analysis of a larger series of rhabdoid tumors would help to determine more clearly whether WT1 is involved in this highly malignant tumor, initially described as a“rhabdomyosarcomatoid” variant of WT (44, 45, 46, 47). The histology of tumors 10T and 180T is consistent with previously reported data showing correlations of the inactivation of WT1 with predominant stromal histology and with ectopic myogenesis in WTs (48, 49). However, we cannot exclude the possibility that in some tumors, the absence of expression may be representative not of the whole tumor but of a distinct histological region possibly selected by chemotherapy and from which RNA has been extracted. Analysis of the status of methylation of WT1 and WT1-antisense promoters could allow us to decipher the mechanism responsible for absence of expression (50).

The isoform ratios obtained for the tumors were compared with those obtained for normal mature kidneys and FKs. Several estimates of the isoform ratios in normal fetal and mature kidneys have already been reported (10, 40, 42, 51). These estimates are similar for mature kidneys but are highly variable for FKs, ranging from 1.0–2.25 for the exon 5+:exon 5− ratio and from 1.23–3.4 for the KTS+:KTS−ratio. This may be due to the developmental age of the FKs used in the various studies and/or to differences in the techniques used for quantification. This report is the second to compare the isoform ratios in fetal and mature kidneys with the same technique. Reverse transcription-PCR was used because only a small amount of tumor and kidney RNA was available for most of the patients. Consistent with the first report (51), we found that the exon 5+:exon 5−ratio increased significantly with maturation of the kidney. We also observed an increase in the KTS+:KTS− ratio, which has not been reported previously. We considered isoform proportions in tumors to be abnormal if they were significantly different not only from those in mature kidneys but also from those in 19–23-week-old FKs. This showed that tumors do not simply correspond to an embryonic state. Consistent with previously reported data, the most frequent alteration was a decrease in the exon 5+:exon 5− transcript ratio (40, 42). This change affected 56% of the WTs. However, other types of imbalance may occur, as we report for the first time an increase in the exon 5+:exon 5− ratio in 10% of the WTs and an increase in the KTS+:KTS− ratio in 20% of the WTs. Although most of the tumors were postchemotherapy samples, six tumors were removed before treatment. The observation that five of these six tumors presented with an isoform imbalance demonstrates that chemotherapy is not responsible for WT1 splicing alterations.

In total, this brought to 91% the proportion of WTs (rhabdoid tumors excluded) with quantitative and/or qualitative alterations of WT1 RNA expression. If only unilateral unifocal WTs were considered, an identical proportion, 90%, displayed abnormal expression. Alteration of the alternative splicing affected 69% of the whole series and 63% of the sporadic tumors. This clearly demonstrates that changes in WT1 expression are not restricted to tumors with predisposing events at WT1. These alterations can be specifically involved in the tumorigenic process. Although previous data obtained with cell lines suggest that the different isoforms have different effects on the tumorigenic process,in tumors, the situation is obviously more complex because the four splice isoforms are coexpressed (32, 39). These alterations may also reflect the types of cells in the tumor sample,their proliferative status, and/or their stage of differentiation. Demonstration of such a relationship is hindered because nothing is known concerning the expression of the different WT1isoforms in specific cell types. However, the different isoform ratios could reflect the different categories of tumors. These analyses did not allow us to establish how many different isoform proteins,possibly due to alternative translation initiation sites, are synthesized from these RNA isoforms. However, there was no evidence of RNA editing in either normal fetal and mature kidneys or tumors (data not shown).

WT1 isoform imbalance may be involved in various types of cancer because it has also been reported in breast tumors (52). However, the mechanism that controls the splicing of WT1 in normal tissues and is disrupted in tumors is unknown. Whether splicing alterations are WT1 specific or are a more general process remains to be elucidated. The balance between isoforms with and without exon 5 was reported to be involved in the control of proliferation and differentiation (29, 30);hence, its disruption is likely to be involved in tumorigenesis. In addition, the increase in the KTS+:KTS− isoform ratio that we described for nine tumors contrasts with the decrease in the same ratio associated with constitutional intron 9 mutations in patients with Frasier syndrome (24, 25). These intronic mutations predispose patients to gonadoblastoma rather than to WTs, suggesting that a decrease in KTS+:KTS− ratio may protect against nephroblastoma development.

In conclusion, this analysis of a large series of tumors was used to accurately evaluate the frequency of WT1 alterations in WT tumorigenesis. We show here that WT1 alterations occurred in the vast majority of sporadic WTs. Changes to WT1 are mostly related to disruption of the alternative splicing of exon 5, but alternative splicing of the KTS region is also affected in some tumors. To look for a relationship between these alterations and the tumorigenic process, we are investigating the consequences of these alterations on gene expression using the microarray approach.

We thank Drs. A. Abel, C. Henry, A. Leblanc, L. de Lumlay, B. Pautard, D. Sommelet, P. Valayer, M. Zucker, and the International Society of Pediatric Oncology for providing patients, tumor samples,and clinical information. We are also grateful to Muriel Rigolet for helpful discussions and comments on the manuscript.