Changes in WT1 Splicing Are Associated with a Specific Gene

Changes in WT1 splicing are associated with a speciﬁc gene expression proﬁle in Wilms’ tumour

Dominique Baudry1, Marine Faussillon1, Marie-Odile Cabanis1, Muriel Rigolet1,6, Jean-Michel Zucker2, Catherine Patte3, Sabine Sarnacki4, Liliane Boccon-Gibod5, Claudine Junien1 and Ce´ cile Jeanpierre*,1

1INSERM U383, Hoˆpital Necker-Enfants Malades, Universite´ Rene´ Descartes, 75743 Paris Cedex 15, France; 2Institut Curie, 75248 Paris, France; 3Institut Gustave Roussy, 94805 Villejuif, France; 4Clinique Chirurgicale infantile, Hoˆpital Necker-Enfants Malades, 75743 Paris Cedex 15, France; 5Service d’Anatomie Pathologique, Hoˆpital Armand Trousseau, 75012 Paris, France

Wilms’ tumour (WT) or nephroblastoma is the most accounting for 7% of cases. A family history of WT is frequent kidney cancer in children. In a previous study, observed in 2% of cases. This tumour may also arise in we reported alterations to WT1 transcription in 90% of association with rare malformation syndromes: WAGR WT tested, with decreased exon 5 +/7 isoform ratio syndrome (WT, aniridia, genitourinary malformations being the most frequent alteration (56% of WT). We and mental retardation), Denys – Drash syndrome now report an approach based on cDNA profiling of (DDS) and Beckwith – Wiedemann syndrome (BWS). tumour pools to identify genes likely to be dysregulated The WT1 gene, located at 11p13, encodes a zinc in association with a decreased WT1 exon 5 +/7 ratio. finger transcription factor involved in renal and gonad We compared the expression profiles of pools of tumours development (Hastie, 2001; Mrowka and Schedl, 2000). classified according to whether this isoform imbalance Two alternative splicing sites give rise to four major was present (five tumours) or not (four tumours), using isoforms. The alternative splicing of exon 5 results in Atlas Cancer cDNA expression arrays. Fourteen of 588 the inclusion or omission of 17 amino acids in the genes tested displayed specific up-regulation (CCND2, transcriptional regulation domain of the protein. The PCNA, N-MYC, E2F3, TOP2A, PAK1, DCC and second alternatively spliced region corresponds to three PCDH2) or down-regulation (VEGF, IGFBP5, TIMP3, amino acids (KTS) located between the third and ARHB, C-FOS and CD9) in the pool of tumours with fourth zinc fingers of the DNA-binding domain. The decreased exon 5 +/7 ratio. These results were four isoforms are produced in a constant ratio that is validated by RT – PCR analysis of four genes (CCND2, conserved between species, suggesting that they have PCNA, VEGF and IGFBP5). We extended the analysis non-overlapping functions. WT1 (-KTS) proteins of VEGF expression to 51 tumours by real-time RT – regulate in vitro the transcription of genes essential PCR and ascertained differential expression of this gene for kidney development, such as IGF2, PDGFA, associated with WT1 expression pattern. Moreover, our EGFR, PAX-2, and WT1 itself (reviewed in Menke results suggest that the VEGF expression level may be of et al., 1998). Other potential targets, such as amphire- prognosis relevance for relapsed patients. gulin and RbAP46, have been identified by expression Oncogene (2002) 21, 5566 – 5573. doi:10.1038/sj.onc. profiling (Guan et al., 1998; Lee et al., 1999). It is now 1205752 clearly established in vivo that the function of WT1 as a transcription regulator is modulated by post-transcrip- Keywords: Wilms’ tumour; expression profiling; WT1 tional modifications and by interaction with other splicing; prognosis proteins including SF1, p53, PAR4 and CBP (Nachti- gal et al., 1998; Maheswaran et al., 1993; Richard et al., 2001; Wang et al., 2001). Exon 5 encodes an Introduction additional regulator domain, with potential repressor and activator functions (Reddy et al., 1995; Wang et Wilms’ tumour (WT) or nephroblastoma is the most al., 1995). WT1 (+KTS) proteins have a low affinity frequent kidney cancer in children. It affects 1 in 10 000 for DNA and may be involved in RNA splicing children, usually around the age of 5 years. Most cases (Larsson et al., 1995; Davies et al., 1998). The presence are sporadic and unilateral, with bilateral tumours of both the WT1 (+KTS) and WT1 (-KTS) isoforms is essential for normal urogenital development (Barbaux et al., 1997; Hammes et al., 2001). Furthermore, the balance between WT1 (+exon 5) and WT1 (-exon 5) *Correspondence: C Jeanpierre; E-mail: [email protected] appears to be essential for the regulation of major 6Current address: UMR 7592 CNRS, Institut Jacques Monod, 75251 Paris Cedex 05, France cellular functions such as proliferation, differentiation Received 2 January 2002; revised 28 May 2002; accepted 7 June and apoptosis (Englert et al., 1997; Hewitt and 2002 Saunders, 1996; Kudoh et al., 1995; Mayo et al., 1999). Expression profiling of Wilms’ tumours D Baudry et al 5567 In Wilms’ tumours in patients with WAGR and in WT(UNB1), WT9 and WT53; (2) similar in Denys – Drash syndromes, a first germline event at the WT(NK-like) and NK; (3) different in WT(UNB1) WT1 locus (deletion or mutation) accounts for and NK; (4) different in WT(UNB1) and FK. Fourteen predisposition and is usually followed by a somatic genes met these criteria, among which six were down- mutation or loss of the normal allele (Jeanpierre et al., regulated and eight were up-regulated in WT(UNB1) 1998). In sporadic tumours, somatic mutations in WT1 (Figure 1). are rare. However, we recently described changes in WT1 transcription in 90% of cases (Baudry et al., Validation of cDNA pooling for identification of 2000). A disruption of exon 5 splicing, with a relative differentially expressed genes decrease in the amount of the WT1 (+exon 5) isoforms, was the most frequent alteration, observed We analysed two overexpressed genes (CCND2 and in 56% of the tumours. It was thought probable that PCNA) and two underexpressed genes (IGFBP5 and these changes would affect the regulation of genes VEGF), by semi-quantitative RT – PCR using pooled downstream in the WT1 pathway. Moreover, little is cDNAs and individual cDNAs (WT9, WT77, WT84, currently known about the genes and mechanisms that WT97, WT190, WT52, WT72, WT75, WT133, NK183, control WT1 expression and alternative splicing. NK186, NK188 and NK189) at the same time (Figure We recently reported preliminary data on expression 2a). Although different normalization parameters were profiling in Wilms’ tumours using cDNA macroarrays, used for the quantification of array hybridizations resulting in the identification of 153 genes differentially (global normalisation) and RT – PCR assays (HPRT as expressed in tumours and normal kidneys (Rigolet et a control gene), differences in expression pattern al., 2001). We now describe a more specific cDNA between pool WT(UNB1) and pool WT(NK-like) were macroarray approach, intended to identify genes likely consistent for all four genes tested. Moreover, the to be dysregulated in tumours with a decreased WT1 levels of expression of the four genes in the individual exon 5 +/7 isoform ratio. This approach was based tumours were consistent with levels of expression in the on cDNA profiling of pools of tumours classified corresponding pools (Figure 2b). according to whether isoform imbalance was present or not. First, we identified 14 genes that displayed specific VEGF expression in a series of 51 tumours dysregulation in the pool of tumours with imbalance. Second, we validated the pooling strategy for tumour We extended the analysis of VEGF expression to a profiling, by RT – PCR analysis of four genes. Third, series of 51 tumours, 13 normal kidneys and 2 foetal we extended the analysis of one of the down-regulated kidneys, using real-time RT – PCR. Normal and foetal genes, VEGF, to 51 tumours by real-time RT – PCR, kidney VEGF expression levels were evaluated as and showed that the level of expression of this gene 0.99+0.49 and 0.82+0.05 respectively, according to correlated with the pattern of WT1 expression. Fourth, R ratios (see Materials and methods). Thirty-seven we analysed the prognosis significance of VEGF tumours underexpressed VEGF (R50.5) and six expression level. tumours overexpressed VEGF (R41.5) (Table 1). Decreased expression in tumours was confirmed by comparing the level of expression in seven tumours and Results paired normal kidneys (Figure 3). The WT1 expression of the tumours was assessed in a previous study RNA profiling of pools of tumours with decreased and (Baudry et al., 2000). We compared the expression of normal WT1 exon 5 +/7 isoform ratios VEGF and WT1 and found that 29 out of 34 tumours We probed Atlas Human Cancer cDNA expression (85%) with a decreased WT1 exon 5 +/7 ratio or no arrays (Clontech), containing 588 genes involved in WT1 expression (no transcript or truncated protein) various aspects of malignancy and belonging to underexpressed VEGF. There was no difference in different functional classes, with cDNAs from: VEGF expression between tumours with an abnormal (1) tumours with a decreased exon 5 +/7 ratio and KTS +/7 ratio and tumours with a normal KTS a normal KTS +/7 ratio, either as a pool +/7 ratio. Furthermore, all three tumours with an (WT(UNB1)) or individually (WT9, WT53); (2) a pool increased WT1 exon 5 +/7 ratio (UNB2) expressed of tumours with exon 5 +/7 and KTS +/7 ratios VEGF at a higher level. There was no evidence of similar to those in normal kidneys (WT(NK-like)); (3) a correlation between the level of VEGF expression a pool of normal kidneys (NK) and (4) a pool of foetal and treatment or not by preoperative chemotherapy kidneys (FK). Comparison of the expression profiles of (Table 1). WT(UNB1) and WT(NK-like) pools of tumours Finally, we investigated the correlation between revealed 81 differentially expressed genes. We then VEGF expression in tumours and prognosis. We considered the expression pattern of normal kidneys compared two categories of patients with relapse, those (NK), foetal kidneys (FK) and independently analysed who are alive (six patients) and those who died (five tumours WT9 and WT53. Tumours WT9 and WT53 patients). Patients with rhabdoid tumours were had the same WT1 isoform profile as tumours included excluded from this analysis. Although none of the 11 in the WT(UNB1) pool. We selected those of the 81 tumours overexpressed VEGF, VEGF levels were genes which presented levels of expression: (1) similar significantly higher in the tumours from patients who

Oncogene Expression profiling of Wilms’ tumours D Baudry et al 5568 died (0.394+0.125) than in tumours from patients who are still alive (0.148+0.063) (Student’s t-test, P=0.01).

Discussion

The heterogeneity of Wilms’ tumours is reflected in the large number of genes with differential expression in tumours and normal kidneys (Rigolet et al., 2001). The approach described here, based on tumour pooling for profiling, made it possible to reduce this heterogeneity artificially and to select 14 genes more specifically dysregulated in a pool of tumours with decreased exon 5+/7 ratio. Seven of these genes had not been identified by our previous study. We validated the pooling strategy for cDNA profiling, by comparing the levels of expression of four of the 14 genes (PCNA, CCND2, IGFBP5 and VEGF) in individual and pooled samples by RT – PCR. Dysregulation of the 14 genes was associated with decreased WT1 exon 5 +/7 ratio. However, we cannot rule out an effect due to changes in the overall level of expression of WT1, because most tumours with exon 5 +/7 imbalance also had higher overall levels of expression than did tumours with a normal exon 5 +/7 ratio. Moreover, somatic mutations were investigated only in tumours associated with a predisposing syndrome. In sporadic tumours, WT1 mutations are rare. However, we cannot exclude that unidentified somatic mutations could alter the WT1 function in a few cases. At present, we have no evidence that alterations of WT1 expression are causally related to changes in gene expression of the 14 differentially expressed genes, and several hypotheses can be put forward. The first hypothesis is a direct transcriptional regulation by WT1. We identified consensus sequences for WT1 binding in the promoters of N-MYC, CCND2, VEGF, C-FOS and IGFBP5. Direct regulation of the expression of N-MYC by WT1 have already been reported (Zhang et al., 1999). Reporter gene assays are now required to investigate whether the other genes are genuine transcriptional targets of WT1 or not. As a second hypothesis, indirect transcriptional regulation by WT1, or WT1 exon 5 +/7 isoform imbalance as a downstream event, must be considered. As a third hypothesis, WT1 isoform imbalance and dysregulation of expression of the 14 genes may be correlated in tumours, but not causally linked. Analysis of the Figure 1 cDNA macroarray analysis of changes in gene expres- expression of these genes in a larger series of tumours sion in tumour and normal tissue samples. (a) Autoradiographic is required to evaluate by accurate statistical analysis data showing the intensity of the spots corresponding to four differentially expressed genes (CCND2, PCNA, IGFBP5, VEGF) and two control genes (ACTB and UBC), following hybridization with pooled cDNAs from five tumours with decreased exon 5 +/ CCND2 (cyclin D2); CD9 (MRP1/CD9, motility-related pro- 7 isoform ratios (WT(UNB1)), four tumours with normal exon 5 tein-1); C-FOS (c-fos protooncogene); DCC (deleted in colorectal +/7 isoform ratios (WT(NK-like)), four normal kidneys (NK) carcinoma); E2F3 (E2F transcription factor 3); IGFBP5 (insulin- and five foetal kidneys (FK). (b) Quantification of the spot inten- like growth factor-binding protein 5); N-MYC (N-myc proto- sities with AtlasImage 1.5 software (Clontech). Global normaliza- oncogene); PAK1 (P21-activated kinase); PCDH2 (protocadher- tion with the Sum method was used. A=FK; B=NK; in-43); PCNA (proliferating cell nuclear antigen); TIMP3 (tissue C=WT(NK-like); D=WT(UNB1); E=WT53; F=WT9. A, B, inhibitor of metalloprotease-3); TOP2A (topoisomerase II alpha C and D are pooled samples. E and F are individual tumours. gene), UBC (ubiquitin) and VEGF (vascular endothelial growth The gene abbreviations are: ACTB (b-actin); ARHB (rhoB); factor)

Oncogene Table 1 Quantiﬁcation of VEGF expression in a series of 51 WT PreOp Follow-up (years VEGF Tumour Clinical features WT1 mutation WT1 expression Stage Chemo Histological data from 1st diagnosis) expression

WT32 nd nd I yes anaplasic alive (8) 0.01 WT1 nd UNB1*, x11 II yes mixed relapsed/alive (8) 0.05 WT188 nd UNB1, nd II N7 yes blastemal predominant alive (3) 0.07 WT114 WAGR C (del11p13)/S (ins exon 1) truncated prot II yes mixed relapsed/alive (10) 0.11 WT96 BWS nd UNB1, x1.2 I yes mixed alive (5) 0.11 WT84 # mental retardation, del 2q37 nd UNB1, x8 I no blastemal predominant alive (6) 0.11 WT100 S (large rearrangement) no transcript II N7 yes mixed alive (4) 0.12 WT77# nd UNB1, x7 II N7 yes mixed relapsed/alive (8) 0.13 WT180 nd no transcript IV yes post-chemotherapy changes (65 – 90%) alive (3) 0.14 WT190# hemihypertrophy nd UNB1, x9 II N7 no blastemal predominant alive (3) 0.14 WT94 nd UNB1, x0.7 I yes post-chemotherapy changes (65 – 90%) alive (5) 0.15 WT191 nd UNB1, x10 yes nephroblastomatosis alive (3) 0.15 WT165 nd FK-like, x0.6 I yes epithelial alive (4) 0.16 WT65 osteogenesis imperfecta nd NK-like, nd II N7 no mixed relapsed/alive (10) 0.16 WT95 nd UNB1*, x0.8 II N7 yes post-chemotherapy changes (65 – 90%) alive (12) 0.17 WT53 nd UNB1, x6 I yes mixed relapsed/alive (7) 0.19 WT34 WAGR C (del11p13)/S (ins exon 3) no transcript I yes mixed alive (5) 0.20 WT92 nd no transcript I yes rhabdoid alive (6) 0.20 WT137 nd no transcript IV (local III) yes mixed alive (4) 0.21 WT189 C (splice, exon7)/S (ins, exon 7) truncated prot II N7 yes mixed alive (3) 0.21 WT10 S (homo deletion) no transcript I yes stromal predominant died (infection) 0.24 WT71 nd UNB1, nd II yes mixed relapsed/died 0.24 WT56 BWS nd UNB1, x8 I no blastemal and epithelial relapsed/alive (13) 0.25 WT79a/b WAGR, bilateral tumour C (del11p13)/S (del, exon 7) truncated prot V (local I) yes blastemal predominant relapsed/died 0.25/0.19 WT13 WG C (stop, exon 4)/S (LOH) truncated prot II no mixed alive (13) 0.27 WT97# nd UNB1, x3 I yes stromal predominant alive (3) 0.27 WT40 nd FK-like, x23 I yes mixed alive (4) 0.29 Baudry tumours D Wilms’ of profiling Expression WT11 nd UNB1*, x38 I yes mixed alive (6) 0.30 WT98 nd UNB1, x6 II N7 yes mixed alive (4) 0.32

WT193 nd no transcript IV yes rhabdoid relapsed/died 0.34 al et WT21 nd UNB1*, x26 I yes anaplasic alive (4) 0.37 WT185 nd UNB1, x2 II N7 yes mixed alive (3) 0.37 WT163 C (S223N, exon 3)/S (LOH) NK-like, x13 II yes mixed alive (4) 0.43 WT192 nd FK-like, x7 IV (local III) yes post-chemotherapy changes (65 – 90%) relapsed/died 0.44 WT15 nd nd II N7 yes epithelial predominant alive (7) 0.47 WT187 nd nd IV (local IIN7) yes diffuse anaplasia alive (3) 0.47 WT2 bilateral tumour no mutation UNB1, x0.8 V (local II) yes blastemal predoiminant relapsed/died 0.51 WT9# nd UNB1, x3 II no mixed relapsed/died 0.53 WT52# BWS nd NK-like, x4 II yes epithelial alive (10) 0.57 WT4 nd no transcript II N+ yes rhabdoid relapsed/died 0.58 WT70 nd no transcript I yes multicystic alive (7) 0.63 WT133# nd NK-like, nd III no rhabdoid relapsed/died 0.81 WT37 bilateral tumour no mutation UNB1*, x10 V yes mixed alive (9) 0.89 WT73 S (ins, exon 4) UNB2, nd I yes diffuse anaplasia alive (7) 0.96 WT72# nd NK-like, x2 II yes mixed alive (6) 1.65 WT75# nd NK-like, x1.2 IV yes mixed alive (7) 1.86 WT89 DDS C(R394W, exon 9)/S (LOH) nd III yes stromal predominant alive (6) 1.95 WT76 nd nd IV yes mixed alive (8) 2.02 WT58 nd UNB2, nd I yes mixed alive (7) 2.17 WT63 nd UNB2, x27 I yes mixed alive (9) 2.69

Data concerning WT1 mutations and expression are described in Baudry et al., 2000. C=Constitutional; S=Somatic; LOH=loss of heterozygosity; UNB1=decreased exon 5+/7 ratio; UNB2=increased exon 5+/7 ratio; NK-like and FK-like=exon 5+/7 ratio as in normal and foetal kidneys respectively; *increased KTS+/7 ratio; the overall level of WT1 expression in tumour is shown in comparison with that in normal kidneys. Stage, histological and follow up data are from SIOP studies (SIOP 6, 9 and 93). #tumours used for hybridization on macroarrays Oncogene 5569 Expression profiling of Wilms’ tumours D Baudry et al 5570

Figure 3 Comparison of VEGF expression in paired tumour and normal kidney samples, by real-time RT – PCR. VEGF expression was quantiﬁed with respect to three control genes, ACTB, HPRT and S9 (ribosyl protein S9), then normalized with respect to the mean level of expression in normal kidneys

the validity of a classification based on expression profiles. The WT1 exon 5 +/7 ratio has been reported to be involved in the control of proliferation and differentiation (Hewitt and Saunders, 1996; Kudoh et al., 1995). Four of the eight overexpressed genes, CCND2, E2F3, PCNA, and N-MYC, are involved in cell cycle control and proliferation. CCND2, encoding the regulator subunit of cyclin D2/cyclin dependent kinase complexes, and the transcription factor E2F3 are critical for control of the G1/S transition (Sherr, 1996; Humbert et al., 2000). PCNA plays an essential role in nucleic acid metabolism and is a sensitive indicator of cell proliferation (Kelman, 1997). The transcription factor N-MYC is down-regulated when kidney differentiation occurs, and increases in the expression of N-MYC have already been reported in Wilms’ tumours (Stanton and Parada, 1992; Nisen et al., 1986). In vitro experiments previously demonstrated G1 phase arrest by WT1, with exon 5+ isoforms displaying greater efficiency than exon 57 isoforms (Kudoh et al., 1995). The association between overexpression of these four genes and decreased WT1 exon 5+/7 ratio suggest a specific mechanism in tumours with this imbalance. In accordance with SIOP (International Society of Paediatric Oncology) protocols, most patients under- went preoperative chemotherapy treatment. The effect of such a treatment on expression levels in tumours has been reported for PCNA and TOP2A. A significant decrease in PCNA expression has been reported in Wilms’ tumours after preoperative chemotherapy, the maintenance of a high level of expression being associated with a poor prognosis (Kim et al., 1996). Figure 2 Quantification by semi-quantitative RT – PCR of gene Topoisomerases are known targets of antitumour drugs expression; (a) fragments specific to CCND2, PCNA and VEGF and a decrease in TOP2A expression has also been were co-amplified with a fragment specific to HPRT (hypox- anthine-guanine phosphoribosyl transferase) as a control. Pooled reported in nephroblastomas after preoperative cDNAs were used, as previously described. For each cDNA pool, chemotherapy (Granzen et al., 2001). In this report, four 1 : 2 serial dilutions were amplified for 26 cycles to obtain re- three of the five tumours of pool WT(UNB1) with high liable quantification in exponential conditions. (b) Comparison of levels of PCNA and TOP2A expression (WT9, WT84 the levels of expression of PCNA, CCND2, IGFBP5 and VEGF, with that of HPRT as a control, in pooled and individual cDNA and WT190) were not treated by preoperative samples. The results were normalized with respect to the mean chemotherapy. We therefore cannot exclude the level of expression in normal kidneys possibility that the overexpression of these genes was

Oncogene Expression profiling of Wilms’ tumours D Baudry et al 5571 due to the lack of preoperative chemotherapy rather with WT1 pathways. They also provide preliminary than to changes in WT1 expression. data towards a classification based on expression TIMP3 and IGFBP5 belong to gene families whose profiles. Such a classification will be helpful not only members have already been implicated in Wilms’ to determine how many Wilms’ tumour subtypes exist, tumours. Tissue inhibitors of metalloproteases (TIMPs) but also to determine whether there is any particular play a key role in extracellular matrix metabolism and relationship between the changes in gene expression are involved in tumour progression (DeClerck et al., and pathological features. 1992). A lack of expression of TIMP4 has been reported in Wilms’ tumours (Celiker et al., 2001). In our experiments, the level of TIMP4 expression in Materials and methods tumours and normal tissues was too faint to allow Samples quantification, but we report a decrease in TIMP3 expression. TIMP3 has also been incriminated in Tumours were collected from 50 patients. All these patients various tumours (Bachman et al., 1999). IGFBPs are were included in International Society of Paediatric Oncology important regulators of IGF1 and IGF2 function (SIOP) treatment protocols. Clinical and histological data are (Rechler and Clemmons, 1998). IGFBP3 and IGFBP4 summarised in Table 1. WT1 analysis for these tumours has been reported elsewhere (Baudry et al., 2000). The WT(UNB1) have been reported to be underexpressed in cells pool included five tumours with lower than normal exon 5 +/ displaying truncation of WT1 (Wagner et al., 2001). 7 isoform ratios and a normal KTS +/7 isoform ratio: WT9, Here we report a down-regulation of IGFBP5, a gene WT77, WT84, WT97 and WT190. The WT(NK-like) pool associated with mesenchyme – epithelial transition included four tumours with exon 5 +/7 and KTS +/7 during kidney development (Plisov et al., 2000). As isoform ratios similar to those in normal kidney: WT52, WT72, IGF2 is frequently overexpressed in Wilms’ tumours, WT75 and WT133. The NK pool included five normal kidneys: the down-regulation of IGFBPs may be a mechanism NK182, NK183, NK186, NK188 and NK189. Foetal kidney by which an excess of functional IGF2 could be (FK) samples were obtained from 23-week-old (FK1) and 20- generated. week-old (FK3) foetuses. FK (five samples, 19 – 23 weeks) Analysis of the expression of VEGF in 51 tumours polyadenylated RNA was purchased from Clontech and used as the FK pool. showed underexpression, not only in tumours with a decreased exon 5 +/7 ratio, but also in tumours devoid of normal WT1 protein. Conversely, the highest Total RNA extraction and cDNA preparation levels of expression were observed in the few tumours Total RNA was prepared from frozen tissues using Trizol with a high exon 5 +/7 ratio. This suggests that two reagent, according to the manufacturer’s protocol (Life subtypes of WT may be defined according to WT1 and Technologies). We reverse transcribed 5 mg of each RNA VEGF expression. The observation of VEGF under- with 200 U of SuperScriptII reverse transcriptase and expression in most tumours contrasts with the frequent hexanucleotide random primers according to the manufac- observation of VEGF overexpression in various turer’s recommendations (Life Technologies). The resulting tumours, which is correlated with a poor prognosis cDNA solution was diluted to a final volume of 100 ml, 2 ml of which was used for PCR amplification. (Linderholm et al., 2001; Yuan et al., 2001). Over- expression of VEGF was observed in only six tumours and all these patients are alive without relapse, with a cDNA macroarray hybridisation and data acquisition follow-up of at least 6 years. However, among patients Gene expression was analysed using Atlas Human Cancer with relapse, we observed that the level of VEGF cDNA expression arrays (Clontech). For expression profiling expression, studied in initial tumours, was significantly of tumour pools, equal amounts of total RNA from the higher if fatal outcome occurred than if patients tumour samples to be pooled were mixed and cDNA was recovered. It suggests that the level of VEGF synthesized from 5 mg of each mix, as previously described expression, albeit low, may have a prognosis value (Rigolet et al., 2001). Following hybridisation, membranes for Wilms’ tumours. This is in agreement with data were placed against a Phosphor Screen for 2 days to 1 week and scanned with a PhosphorImager (Molecular Dynamics). showing that VEGF production is associated with Data were saved as ‘gel format’ files and analysed using the tumour growth and metastases in a murine model of AtlasImage 1.5 software (Clontech). The hybridization of Wilms’ tumour (Kayton et al., 1999; Rowe et al., individual tumour samples and normal controls has already 2000). Analysis of a larger number of tumours is now been reported (Rigolet et al., 2001) and ‘gel format’ files were necessary in order to validate the use of VEGF as a used for analysis with the AtlasImage 1.5 software. We prognosis marker. considered a spot signal to be meaningful if its adjusted cDNA arrays have proven to be powerful tools not intensity (spot signal intensity minus background signal only for tumour classification but also for identification intensity) was at least equal to background signal intensity. of regulation pathways. Our analysis of differential As each cDNA sample was hybridized in triplicate, we expression profiles in tumour pools allowed us to performed a ‘composite’ analysis: the software calculated the mean of the three results for each spot position. Global identify 14 genes dysregulated in association with a normalization with the Sum method was used to compare decreased WT1 exon 5 +/7 isoform ratio. These data composites two by two. A gene was regarded as differentially provide a starting point for investigations of the way in expressed if the adjusted intensity of the signal for one of the which these genes may be involved in the nephro- composites was higher or lower, by a factor of at least two, blastoma tumorigenic process, and their relationship than that of the other.

Oncogene Expression profiling of Wilms’ tumours D Baudry et al 5572 Table 2 Primer sequences and PCR conditions Primer name Sequence 5’ –3’ Expected product

For radioactive RT-PCR CCND2 sense TGGAGCTGCTGTGCCACG CCND2 antisense GTGGCCACCATTCTGCGC 181 bp PCNA sense GCGCTAGTATTTGAAGCACCAA PCNA antisense CGATCTTGGGAGCCAAGTAGTA 469 bp IGFBP5 sense CGGGGGAGCCGAGAACAC IGFBP5 antisense GGCGCTGGCTGGAGTCGG 385 bp VEGF sense CCTCGTCATTGCAGCAGC VEGF antisense TGTCTTGGGTGCATTGGA 258 bp HPRT sense TGTGTGCTCAAGGGGGGC HPRT antisense CGTGGGGTCCTTTTCACC 314 bp

For real time PCR VEGF sense CTGCTGTCTTGGGTGCATTGGA VEGF antisense GGATGGGCAGTAGCTGCGCTGAT 152 bp ACTB sense TTGCCGACAGGATGCAGAAGGA ACTB antisense AGGTGGACAGCGAGGCCAGGAT 129 bp S9 sense AGCGGACGCAACATGCCAGTGG S9 antisense TCGCCGATCAGCTTCAGCTCTTGCT 113 bp HPRT sense AATTATGGACAGGACTGAACGTCTTGCT HPRT antisense TCCAGCAGGTCAGCAAAGAATTTATAGC 117 bp

kidney sample, R ratios were then normalized with respect RT – PCR analysis of candidate genes to the appropriate R(NK) ratio: R(norm)HPRT=RHPRT/ For each sample, four 1 : 2 serial dilutions of the cDNA were R(NK)HPRT; R(norm)S9=RS9/R(NK)S9; R(norm)ACT- ampliﬁed for 26 cycles, using candidate gene and HPRT B=RACTB/R(NK)ACTB. Finally, we calculated an R ratio primers (Table 2). 32P-dCTP was added to the reaction as the mean of these three R(norm) ratios. We considered mixture for radiolabelling of the PCR products. We the level of VEGF expression to be lower or higher in a determined the ratios of candidate gene to HPRT by tumour sample than in normal kidneys if R was lower or phosphorimaging and analysis with ImageQuant software higher than the mean normal kidney R ratio minus or plus (Molecular Dynamics). For each data point, the results of one standard deviation. For analysis, we considered the two or three successive cDNA dilutions in the non-saturated mean of the results obtained in two independent experi- range of PCR were averaged. Two independent experiments ments. were performed.

Quantification of VEGF expression by real-time RT – PCR Real-time RT – PCR was performed with an ABI Prism 7700 (PE Biosystems) according to the manufacturer’s Acknowledgements instructions. We independently amplified VEGF and three We thank Drs P Boute, M Fabre, J-C Fournet, B Gilbert, control genes, HPRT, S9 and ACTB. Reactions were F Jaubert, F Pein, B Pautard and H van Tinteren for performed in a 20 ml volume with 10 pmoles of each providing tumour samples and clinical data. This work was primer. Deoxyribonucleotides, Taq DNA polymerase and supported by the Institut National de la Sante´ et de la buffer were included in the SYBR Green PCR Master Mix Recherche Me´ dicale, the Association pour la Recherche sur (PE Biosystems). We quantified the amount of amplified le Cancer and the Ligue Nationale contre le Cancer. D fragments with Sequence Detector v1.6 software. For each Baudry was supported by grants from the Association pour sample, three ratios were calculated: RHPRT=VEGF/HPRT, la Recherche sur le Cancer and from the Fondation pour la RS9=VEGF/S9 and RACTB=VEGF/ACTB. Ratios for the Recherche Me´ dicale and M Faussillon was supported by a 13 normal kidneys were averaged to give R(NK)HPRT, grant from the Ministe` re de la Recherche et de la R(NK)S9 and R(NK)ACTB. For each tumour and normal Technologie.

References

BachmanKE,HermanJG,CornPG,MerloA,CostelloJF, Celiker MY, Wang M, Atsidaftos E, Liu X, Liu YE, Jiang Y, Cavenee WK, Baylin SB and Graff JR. (1999). Cancer Valderrama E, Goldberg ID and Shi YE. (2001). Res., 59, 798 – 802. Oncogene, 20, 4337 – 4343. Barbaux S, Niaudet P, Gubler MC, Grunfeld JP, Jaubert F, Davies RC, Calvio C, Bratt E, Larsson SH, Lamond AI and Kuttenn F, Fekete CN, Souleyreau-Therville N, Thibaud Hastie ND. (1998). Genes Dev., 12, 3217 – 3225. E, Fellous M and McElreavey K. (1997). Nature Genet., DeClerck YA, Perez N, Shimada H, Boone TC, Langley KE 17, 467 – 470. and Taylor SM. (1992). Cancer Res., 52, 701 – 708. Baudry D, Hamelin M, Cabanis MO, Fournet JC, Tournade Englert C, Maheswaran S, Garvin AJ, Kreidberg J and MF, Sarnacki S, Junien C and Jeanpierre C. (2000). Clin. Haber DA. (1997). Cancer Res., 57, 1429 – 1434. Cancer Res., 6, 3957 – 3965.

Oncogene Expression profiling of Wilms’ tumours D Baudry et al 5573 Granzen B, Efferth T, Keller U, Beniers AJ, Mertens R, Menke A, McInnes L, Hastie ND and Schedl A. (1998). Jakse G and Fuzesi L. (2001). Anticancer Res., 21, 771 – Kidney Int., 53, 1512 – 1518. 776. Mrowka C and Schedl A. (2000). J. Am. Soc. Nephrol., 11, Guan LS, Rauchman M and Wang ZY. (1998). J. Biol. S106 – S115. Chem., 273, 27047 – 27050. Nachtigal MW, Hirokawa Y, Enyeart-VanHouten DL, HammesA,GuoJK,LutschG,LehesteJR,LandrockD, Flanagan JN, Hammer GD and Ingraham HA. (1998). Ziegler U, Gubler MC and Schedl A. (2001). Cell, 106, Cell, 93, 445 – 454. 319 – 329. Nisen PD, Zimmerman KA, Cotter SV, Gilbert F and Alt Hastie ND. (2001). Cell, 106, 391 – 394. FW. (1986). Cancer Res., 46, 6217 – 6222. Hewitt SM and Saunders GF. (1996). Anticancer Res., 16, Plisov SY, Ivanov SV, Yoshino K, Dove LF, Plisova TM, 621 – 626. Higinbotham KG, Karavanova I, Lerman M and Humbert PO, Verona R, Trimarchi JM, Rogers C, Perantoni AO. (2000). Genesis, 27, 22 – 31. Dandapani S and Lees JA. (2000). Genes Dev., 14, 690 – Rechler M and Clemmons D. (1998). TEM, 9, 176 – 183. 703. Reddy JC, Morris JC, Wang J, English MA, Haber DA, Shi Jeanpierre C, Be´ roud C, Niaudet P and Junien C. (1998). Y and Licht JD. (1995). J. Biol. Chem., 270, 10878 – 10884. Nucleic. Acids Res., 26, 271 – 274. Richard DJ, Schumacher V, Royer-Pokora B and Roberts Kelman Z. (1997). Oncogene, 14, 629 – 640. SG. (2001). Genes Dev., 15, 328 – 339. Kayton ML, Rowe DH, O’Toole KM, Thompson RB, Rigolet M, Faussillon M, Baudry D, Junien C and Jeanpierre Schwarz MA, Stolar CJ and Kandel JJ. (1999). J. Pediatr. C. (2001). Pediatr. Nephrol., 16, 1113 – 1121. Surg., 34, 743 – 747. Rowe DH, Huang J, Kayton ML, Thompson R, Troxel A, Kim CJ, Oh MH, Shin HY, Ahn HS, Kim CW and Chi JG. O’Toole KM, Yamashiro D, Stolar CJ and Kandel JJ. (1996). Pathol. Int., 46, 137 – 141. (2000). J. Pediatr. Surg., 35, 31 – 32. Kudoh T, Ishidate T, Moriyama M, Toyoshima K and Sherr CJ. (1996). Science, 274, 1672 – 1677. Akiyama T. (1995). Proc.Natl.Acad.Sci.USA,92, 4517 – Stanton BR and Parada LF. (1992). Brain Pathol., 2, 71 – 83. 4521. Wagner KJ, Patek CE, Miles C, Christie S, Brookes AJ and Larsson SH, Charlieu JP, Miyagawa K, Engelkamp D, Hooper ML. (2001). Biochem. Biophys. Res. Commun., Rassoulzadegan M, Ross A, Cuzin F, van Heyningen V 287, 977 – 982. and Hastie ND. (1995). Cell, 81, 391 – 401. Wang W, Lee SB, Palmer R, Ellisen LW and Haber DA. LeeSB,HuangK,PalmerR,TruongVB,HerzlingerD, (2001). J. Biol. Chem., 276, 16810 – 16816. Kolquist KA, Wong J, Paulding C, Yoon SK, Gerald W, Wang ZY, Qiu QQ, Gurrieri M, Huang J and Deuel TF. Oliner JD and Haber DA. (1999). Cell, 98, 663 – 673. (1995). Oncogene, 10, 1243 – 1247. Linderholm BK, Lindahl T, Holmberg L, Klaar S, Yuan A, Yu CJ, Kuo SH, Chen WJ, Lin FY, Luh KT, Yang Lennerstrand J, Henriksson R and Bergh J. (2001). Cancer PC and Lee YC. (2001). J. Clin. Oncol., 19, 432 – 441. Res., 61, 2256 – 2260. Zhang X, Xing G and Saunders GF. (1999). Anticancer Res., Maheswaran S, Park S, Bernard A, Morris JF, Rauscher 19, 1641 – 1648. 3rdFJ, Hill DE and Haber DA. (1993). Proc. Natl. Acad. Sci. USA, 90, 5100 – 5104. Mayo MW, Wang CY, Drouin SS, Madrid LV, Marshall AF, Reed JC, Weissman BE and Baldwin AS. (1999). EMBO J., 18, 3990 – 4003.

Oncogene