FGFR3 and Ras gene mutations are mutually exclusive genetic events in urothelial cell carcinoma

Transcription

1 (2005) 24, & 2005 Nature Publishing Group All rights reserved /05 $ FGFR3 and Ras gene mutations are mutually exclusive genetic events in urothelial cell carcinoma Adel H Jebar 1,3, Carolyn D Hurst 1,3, Darren C Tomlinson 1, Colin Johnston 1, Claire F Taylor 2 and Margaret A Knowles*,1 1 Cancer Research UK Clinical Centre, St James s University Hospital, Beckett Street, Leeds LS9 7TF, UK; 2 Cancer Research UK Mutation Detection Facility, St James s University Hospital, Beckett Street, Leeds LS9 7TF, UK Fibroblast growth factor receptor 3 (FGFR3) mutations are frequent in superficial urothelial cell carcinoma (UCC).Ras gene mutations are also found in UCC.As oncogenic activation of both FGFR3 and Ras is predicted to result in stimulation of the mitogen-activated protein kinase (MAPK) pathway, we hypothesized that these might be mutually exclusive events. HRAS mutation has been widely studied in UCC, but all three Ras gene family members have not been screened for mutation in the same sample series.we screened 98 bladder tumours and 31 bladder cell lines for mutations in FGFR3, HRAS, NRAS and KRAS2. FGFR3 mutations were present in 54 tumours (55%) and three cell lines (10%), and Ras gene mutations in 13 tumours (13%) and four cell lines (13%). These included mutations in all three Ras genes; ten in HRAS, four in KRAS2 and four in NRAS and these were not associated with either tumour grade or stage.in no cases were Ras and FGFR3 mutation found together.this mutual exclusion suggests that FGFR3 and Ras gene mutation may represent alternative means to confer the same phenotype on UCC cells.if these events have biological equivalence, Ras mutant invasive UCC may represent a novel subgroup. (2005) 24, doi: /sj.onc ; published online 9 May 2005 Keywords: bladder cancer; FGFR3; HRAS; NRAS; KRAS2; mutation Introduction Urothelial cell carcinoma (UCC) of the bladder is the fifth most common cancer in the UK ( cancerresearchuk.org/aboutcancer/statistics/incidence?) and USA (Jemal et al., 2003). These tumours comprise two major groups. At presentation, the majority (70 80%) are low-grade, non-invasive (stage pta or *Correspondence: MA Knowles; margaret.knowles@cancer.org.uk 3 These authors contributed equally to this work. Received 22 December 2004; revised 14 March 2005; accepted 15 March 2005; published online 9 May 2005 pt1), papillary tumours. These superficial tumours recur frequently (B70%), but usually with no increase in grade or progression to muscle invasion. Patients may have multiple recurrences over many years but only 10 20% progress to invade muscle. In contrast, the 20% of tumours that are invasive at diagnosis have a much worse prognosis, with a 50% 5-year survival for pt2 tumours. While non-invasive bladder cancers are not life-threatening, this disease places a heavy burden on both patients and health care providers. Following resection of a primary tumour, long-term follow-up is essential for this large group of patients who suffer the morbidity and anxiety associated with repeated invasive procedures. For these patients, the development of noninvasive tests for recurrence and of more effective therapies to remove all residual tumour cells from the urothelium at the time of initial resection are highly desirable. For the invasive tumours, current therapies are not adequate and the development of tailored systemic therapies is an important goal. There have been major efforts to understand the molecular pathogenesis of both of these groups of bladder cancers to establish the basis for their divergent clinical behaviour and to provide potential markers for disease monitoring and targets for therapy. Many genetic alterations have been identified in muscle invasive UCC including mutations of TP53 and RB1, deletions of chromosome 9 including homozygous deletions affecting CDKN2A/ARF, amplification and overexpression of ERBB2 and many other deletions and amplifications identified by loss of heterozygosity (LOH) and comparative genomic hybridization (CGH) analysis (Knowles, 1999). In contrast, few genetic events have been identified in superficial non-invasive UCC. The only common alterations are deletions of chromosome 9 and mutations of fibroblast growth factor receptor 3 (FGFR3) (Cappellen et al., 1999; Billerey et al., 2001). Interestingly, mutations of FGFR3 are found more frequently in superficial than in invasive UCC in (Billerey et al., 2001), and it has been reported that such mutations are more frequent in UCCs that do not recur (van Rhijn et al., 2001). Mutations of FGFR3 found in UCC are identical to those found in germline skeletal dysplasia syndromes where mutation in hotspot codons in the receptor leads to constitutive kinase activity (Webster and Donoghue, 1997). When present

2 in the germline, this has a negative effect on the growth of the long bones. In the bladder, FGFR3 is a presumed oncogene, but its precise effects on urothelial cell phenotype and the signalling pathway(s) stimulated have not yet been elucidated. However, it is known that in some cell types in vitro, FGFR3 signalling stimulates the mitogen-activated protein kinase (MAPK) pathway (Hart et al., 2000). Since the first human oncogene identified, an activated HRAS gene, was found in the bladder cancer cell line T24 (Der et al., 1982), many studies have examined HRAS for mutations in bladder cancer (Fujita et al., 1984; Visvanathan et al., 1988; Czerniak et al., 1990; Knowles and Williamson, 1993; Levesque et al., 1993; Burchill et al., 1994; Ooi et al., 1994; Fitzgerald et al., 1995; Uchida et al., 1995). Most studies have examined only HRAS and have reported a wide range of mutation frequencies (0 70%) that may reflect true differences in the tumours examined or technical differences between assays. Currently, there is agreement from several studies that the frequency for HRAS is in the range 10 20%. Few studies have screened NRAS and KRAS2, but mutations have been described previously in KRAS2 (Uchida et al., 1995; Olderoy et al., 1998; Ayan et al., 2001). A single study that examined NRAS using restriction fragment polymorphism analysis reported a high frequency of mutation, but these were not confirmed by sequencing (Przybojewska et al., 2000). To date a comprehensive analysis of all three Ras genes has not been carried out in a large series of bladder tumours. As for FGFR3 mutations, Ras gene mutations in UCC are presumed to activate the MAPK pathway. This raises the question of whether FGFR3 activation and Ras activation could fulfil the same function in the tumour cell. In this study we have examined a series of 98 bladder tumour samples and 31 bladder cell lines for mutations in FGFR3, HRAS, KRAS2 and NRAS to provide a comprehensive assessment of all three Ras genes and to examine the possibility that Ras mutation and FGFR3 mutation are mutually exclusive. Results We screened 29 tumour-derived urothelial cell lines, two normal urothelium-derived cell lines and a panel of 98 tumour samples for mutations in HRAS, NRAS, KRAS2 and FGFR3. Some of the tumours in this series had been screened previously for FGFR3 mutation using SSCP analysis (Sibley et al., 2001). Additional samples were screened by direct sequencing of exons 7, 10 and 15 that contain all of the mutations described in UCC to date. Mutations were present in 54 tumours (55%) (Table 1). One tumour contained two mutations (R248C and S249C). Six different mutations were found, five of which are identical to mutations that in the germline cause thanatophoric dysplasia type I and are believed to result in constitutive activation of the kinase activity of this receptor (Webster and Donoghue, 1997). Mutations were found more frequently in UCCs of lower grade and Table 1 FGFR3 mutations identified in UCC tumours Exon Codon Nucleotide change Amino-acid change Frequency of mutations (%) c.742c>t a R248C c.746c>g S249C c.1114g>t G372C c.1117a>t S373C c.1124a>g Y375C c.1954a>c K652Q 1.82 a Nucleotide positions according to FGFR3 IIIb isoform numbering from the initiating ATG stage (Figure 1a and b). Three cell lines were found to contain mutations. These were J82 (K652E), 97 7 (S249C) and (S249C). In each of these cell lines, the mutation was heterozygous. We used fluorescent single-strand conformation polymorphism (F-SSCP) analysis to screen for mutations in exons 1 and 2 of HRAS, NRAS and KRAS2. Mutations were detected in 13 bladder tumour samples (13%) and four cell lines (Table 2). These included mutations in all three ras genes and involved all exons except KRAS2 exon 2. One tumour (134) contained two mutations (HRAS Q61R and KRAS2 G12R). In total there were four mutations in HRAS codon 12, two in HRAS codon 13, four in HRAS codon 61, four in KRAS2 codon 12, one in NRAS codon 12 and three in NRAS codon 61. Examples of NRAS and KRAS2 mutations are shown in Figure 2. Of the 18 mutations detected, six were G:C to T:A transversions, four were A:T to G:C transitions, three were A:T to T:A transversions, four were G:C to C:G transversions and one was a G:C to A:T transition. No significant association between tumour grade or stage and mutation was apparent (w 2 ; P>0.25 for each). The 13 tumour samples with mutation included five grade 1, five grade 2 and three grade 3 tumours, five pta, five pt1 and two pt2 tumours. The overall frequencies of Ras and FGFR3 mutations in tumours and cell lines are shown in Figure 1c and d. In total, 68% of the tumours studied had mutation of either a Ras gene or FGFR3. This accounted for 37 of 45 (82%) of grade 1 tumours, and 32 of 39 (82%) of pta tumours, but only 44% of grade 3 and 39% of pt2 tumours. Examination of the distribution of Ras and FGFR3 mutations revealed that these events were totally mutually exclusive in both tumours and cell lines. The conventional odds ratio for RAS mutation in the presence/absence of FGFR3 mutation in a tumour was estimated as 0 and the w 2 test for independence was highly significant (Po0.0001). The median unbiased estimate of the odds ratio was with an exact 95% confidence interval of Discussion Current knowledge of the genetics of UCC is extensive and there are some clear associations of genetic events with the two major groups of tumours defined at 5219

3 5220 Table 2 Ras gene mutations identified in UCC tumours and cell lines Grade/ stage Mutated gene Nucleotide change Amino-acid change Tumour sample 37 G2 pt1 HRAS c.34g>a a G12S 226 G3 pt1 HRAS c.34g>t G12C 267 G2 pta HRAS c.34g>t G12C 222 G1 pt1 HRAS c.37g>c G13R 280 G3 pt2 HRAS c.38g>t G13V 134 b G2 HRAS c.182a>g Q61R 255 G1 pta HRAS c.182a>g Q61R 176 G1 pta HRAS c.182a>g Q61R 279 G3 pt2 HRAS c.182a>t Q61L 99 G1 pta KRAS2 c.35g>t G12V 134 b G2 KRAS2 c.34g>c G12R 153 G1 pta KRAS2 c.35g>c G12A 239 G2 pt1 NRAS c.34g>c G12R 125 G2 pt1 NRAS c.119a>t Q61L Cell line T24 G3 HRAS c.35g>t G12V UMUC3 NI KRAS2 c.34g>t G12C KU19-19 G3 pt3 NRAS c.119a>g Q61R BFTC 905 G3 NRAS c.119a>t Q61L a Reference sequences are NM_ (HRAS), NM_ (KRAS2) and NM_ (NRAS), numbering from the initiating ATG in each case. b Two mutations were found in tumour 134. NI ¼ no information Figure 1 Distribution of FGFR3 and Ras gene mutations in UCC tumours and cell lines. FGFR3 mutations according to tumour stage (a) and grade (b). Percentages of tumours (c) and cell lines (d) with Ras and FGFR3 mutations diagnosis, low-grade superficial UCC and muscle invasive UCC. The majority of known genetic events have been described in muscle invasive UCC and many of these, such as TP53 and RB1 mutation, are associated with poor prognosis. What has not become clear from genetic studies to date is whether subgroups of tumours can be defined, particularly within the very large group of low-grade superficial tumours that represent such a major part of current urological practise. In this group, the only frequent genetic alterations described are LOH affecting chromosome 9 and mutations of FGFR3. As Ras gene mutations are sometimes found in such low grade/stage tumours (Knowles, 1999), we were interested to assess whether such mutations are found in tumours with FGFR3 mutation, an event that might be predicted to result in activation of similar downstream signalling pathways. A previous study of nine multiple myeloma cell lines with a t(4;14) translocation, found four FGFR3 mutations and four Ras gene mutations and these events were mutually exclusive, leading to the suggestion that these events may play an analogous role in the pathogenesis of multiple myeloma (Chesi et al., 2001). Our study demonstrates a striking mutual exclusion of these mutations in UCC. In total, 68% of the tumours studied had a mutation of either a Ras gene or FGFR3 with a maximum frequency of 82% in G1 or pta tumours. Only seven G1/Ta tumours did not contain a mutation. We have some other genetic information for these, one of which is known to have high level amplification and overexpression of ERBB2, which could represent an alternative mechanism by which the same signalling pathways are activated in this tumour. In contrast, the frequency of mutations was much lower in high grade and stage tumours with only 39% of pt2 tumours showing mutation in any gene studied. The frequency of FGFR3 mutation found here was similar to that described in several previous studies and all of the mutations have been described previously in UCC samples (Cappellen et al., 1999; Billerey et al., 2001; van Rhijn et al., 2001). Five of the mutations (S248C, S249C, G372C, S373C, Y375C) are identical to mutations found in the germline in the skeletal dysplasia syndrome thanatophoric dysplasia type I (TDI) (Webster and Donoghue, 1997), all of which induce strong constitutive activation of the receptor. We found one case with K652Q, which in the germline is associated with the milder chondrodysplasia syndrome hypochondroplasia (Bellus et al., 2000) and in vitro shows lower levels of receptor autophosphorylation. It is not yet clear whether there are phenotypic differences in urothelial cells containing mutations with different activation status. Although FGFR3 mutations were found predominantly in tumours of lower grade and stage (Figure 1),

4 5221 Figure 2 Examples of mutations in TCC tumour and cell lines at (a) NRAS exon 1 codon 12, (b) NRAS exon 2 codon 61 and (c) KRAS2 exon 1 codons 12 we did find several mutations in tumours of high grade and stage, which contrasts with some other studies (Billerey et al., 2001; van Rhijn et al., 2001). For this historical tumour series, we had access only to the overall clinical grading recorded in the patients records. One possible explanation is that the recorded grading and staging reflects the most aggressive lesion resected. Thus, if multifocal tumours were present, it is possible that a less aggressive lesion may have contained an FGFR3 mutation, which would be identified by SSCP, even if only present in a minor population of cells in the sample. Studies to confirm and extend our observations will require careful dissection and direct histopathological assessment of the fragments used for mutation analysis. It has been assumed previously, possibly in light of the high degree of specificity for mutation of specific Ras genes in some tissues and also because of the initial identification of mutated HRAS in a TCC cell line (T24), that UCC contains predominantly HRAS mutations. An aim of this study was to carry out comprehensive screening of all three Ras genes in the same series of UCC samples. F-SSCP is widely accepted to be a highly sensitive mutation scanning technique (Taylor and Taylor, 2004). As we had screened some of these samples previously using a lower sensitivity assay (Knowles and Williamson, 1993), we were interested to know whether F-SSCP would detect additional mutations. Since this was not the case, we conclude that all coding sequence mutations have been found. In light of the finding that mutations in intron D of HRAS can affect expression levels and that alternative splicing of HRAS gives rise to a p19 form that is suggested as a negative regulator of p21 (Huang and Cohen, 1997), it is

5 5222 possible that such alterative mechanisms to increase Ras signalling could operate in some UCCs and this must be examined in future. A striking finding was that all three Ras genes showed mutation and that the combined frequency of KRAS2 and NRAS mutations was similar to that of HRAS mutations. No sequenced mutations of NRAS have been described previously in bladder cancer. The mutations we found were G12R, Q61L and Q61R. In other tumour types where NRAS mutation is found (e.g. haematopoietic, laryngeal and skin cancers), codon 61 is most commonly mutated, and Q61R is the most common mutation ( Interestingly, only nine of the 1132 mutations of NRAS recorded are G12R. Therefore, it will be important to analyse more bladder tumours to determine whether this is a chance finding or whether there may be some tissue specificity in the mutation spectrum. Several studies have examined KRAS2 in bladder tumours and a few have been reported (Grimmond et al., 1992; Uchida et al., 1995; Olderoy et al., 1998). We found one G12A and one G12 V, and two mutations not previously described (G12R and G12C). Taken together these results suggest that KRAS2 mutation may be more frequent in bladder tumours than previously thought. In other tumour types there is specificity in the Ras family member that is most commonly activated. For example, most colorectal cancers show KRAS mutation while haematological malignancies show predominantly NRAS mutation. Overall, the vast majority of mutations in human cancers are found in KRAS2 and mutations in HRAS are relatively rare. The exact reasons for this specificity are not yet clear but some recent studies point to possible differences in the ability of different Ras proteins to activate different Raf isoforms (discussed in (Ellis and Clark, 2000)). Whatever the explanation for the differences, it appears that either some shared function of all three Ras genes is important in UCC cells or that any of the potentially different biological effects can contribute to UCC development. Our hypothesis was that the Ras genes and FGFR3 might fulfil equivalent functions in UCC. The data certainly point to some degree of equivalence of these events and this will require further examination of the phenotype of such tumours, including detailed examination of clinical follow-up data. Several other genetic events and expression changes known to occur in UCC could also have the same effect. These include overexpression of certain tyrosine kinase receptors, for example, EGFR and ERBB2, alterations in expression of Ras effectors such as RAF1, which is sometimes amplified and overexpressed in UCC, increased expression of the Ral proteins, one of which (RalA) has been implicated in EGFR-regulated motility in a bladder cancer cell line, and decreased expression of RhoGDI2, an inhibitor of guanine nucleotide dissociation for Rho proteins, which has been implicated recently in UCC invasion and metastasis (reviewed in (Oxford and Theodorescu, 2003)). We predict that the MAPK pathway might be the common effector pathway for mutant FGFR3 and Ras signalling but until the exact signalling effects of FGFR3 in urothelial cells are elucidated, the involvement of the PI3 K pathway, PLCg and STATs, all of which are known to be involved in FGFR3 signalling in chondrocytes and other nonepithelial cells, must be considered. We have been unable to access tissue samples from the current historical tumour series to assess pathway activation status and this will be priority for future studies. Two recent studies have shown that FGFR3 and TP53 mutations are almost mutually exclusive in UCC (Bakkar et al., 2003; van Rhijn et al., 2004). These events appear to characterize the two major groups of UCCs, which may define two alternative pathways in the pathogenesis of these cancers. TP53 is associated with tumours of high grade and stage and with carcinoma in situ which is a high-risk superficial lesion believed to represent a precursor for invasive UCC. The finding of FGFR3 mutation in a population of tumours that is virtually distinct from these is compatible with the previous finding that FGFR3 mutations are strongly associated with low grade and stage UCC and with lower frequency of recurrence (Billerey et al., 2001; van Rhijn et al., 2001). This distinction of FGFR3 and TP53 mutant UCCs fits well with the proposed two pathway model for UCC development and provides a specific marker for the pathway leading to the development of low grade superficial UCC. In the present study we have found total mutual exclusion of Ras and FGFR3 mutations in UCC but in this case, unlike the situation for TP53 and FGFR3, the mutations do not segregate with tumour grade and stage in both cases. Thus, the distribution of Ras mutations, though nonoverlapping with FGFR3 mutations, does not distinguish tumours that fall predominantly into either of these proposed pathways. This lack of association of Ras mutation with tumour grade or stage is in agreement with previous studies of UCC (Levesque et al., 1993). What then is the significance of Ras gene mutation in the urothelium? Several studies have examined the effects of expression of Ras genes on urothelial cells both in culture and in animal models. Overexpression of either wild type or mutant HRAS (Val12) in the UCCderived cell line RT4 induced invasion in an orthotopic model (Theodorescu et al., 1990). The effects of HRAS (Val12) have also been assessed in normal and immortal human urothelial cells in culture (reviewed in Reznikoff et al., 1993). Normal urothelial cells were not immortalized and failed to form tumours in nude mice and tumorigenic conversion of SV40-immortalized cells was not induced. However, there was an upgrading effect in cells that formed low-grade tumours and stimulation of the growth rate of a high-grade tumour-forming cell line. Together these studies suggest that in some human UCC cell lines, mutant Ras can contribute to a more aggressive phenotype, but cannot achieve immortalization of normal cells or induce tumorigenicity in immortal cells. Interestingly, experiments in mice have recently provided evidence for a role of Ras mutation in the

6 development of superficial papillary UCC. When a mutant HRAS was expressed in the bladder of transgenic mice using a urothelium-specific promoter of the uroplakin II gene, urothelial hyperplasia and superficial papillary tumours were induced (Zhang et al., 2001). These results are in accord with the previous finding that when mouse urothelial cells infected in vitro with a virus carrying a mutant Ras gene were implanted under the renal capsule, cell proliferation was increased, leading to hyperplasia (Wagner et al., 1990). These results indicate a role for Ras in generating hyperplasia in the normal mouse bladder. This is in sharp contrast with the ability of uroplakin II promoter-driven expression of SV40 T to induce CIS and invasive UCC in transgenic mice (Zhang et al., 1999). Thus, the expression of SV40 T, which inactivates both p53 and Rb, leads to aggressive tumours similar to human bladder tumours in which mutations of TP53 and RB1 are found and expression of mutant Ras results in tumours resembling the common low-grade papillary tumours. Assuming that mouse and human urothelial cells respond similarly, these in vivo and in vitro studies indicate that Ras can participate both early in UCC development and during tumour progression. This contrasts with what is observed for FGFR3 where the strong association with low grade/stage, nonrecurrent tumours indicates a role early in tumour development and defines a group of UCCs with favourable prognosis. Our present findings are compatible with this but raise important questions about the biological similarity of Ras mutant and FGFR3 mutant tumours. It will now be important to assess the clinical course of patients with Ras mutant tumours in both the noninvasive and invasive groups and to examine the relationship between Ras and FGFR3 mutation, TP53 mutation and other known molecular characteristics of UCC. As both FGFR3 and Ras gene mutation are rare in muscle invasive UCC, it will be particularly interesting to assess whether these features characterize a less aggressive or lower risk group of invasive UCCs. Materials and methods Tumour samples and cell lines The tumour series used has been described previously (Proctor et al., 1991). The total number analysed was 98. Tumours were classified according to the WHO and TNM guidelines. Information on grade was available for 94 samples and on stage for 86 samples. There were 45 grade 1, 31 grade 2 and 18 grade 3 tumours. In Total, 39 were pta, 23 were pt1, 23 were pt2 and one was carcinoma in situ. A total of 29 bladder tumour cell lines were used (Table 3). One cell line, SVHUC, was established following SV40 T immortalization of normal urothelial cells and one, HCV29, was derived from apparently normal urothelium. DNA was extracted from cultured cells using standard proteinase K digestion followed by phenol extraction and ethanol precipitation Table 3 Cell lines used in this study, and their origins Cell line Origin Other information Reference T24 TCC G3 Primary tumour, untreated Williams (1980) J82 TCC G3 T3 Primary tumour, treated O Toole et al. (1978) 5637 TCC NR a Primary tumour Cannon, unpublished 647V TCC G2 Primary tumour Williams (1980) VM-CUB-1 TCC NR Primary tumour Williams (1980) VM-CUB-2 TCC NR Metastatic tumour, lymph node Williams (1980) VM-CUB-3 TCC NR Primary tumour Williams (1980) 253J TCC G4 T4 Metastatic tumour, lymph node Elliot et al. (1974) RT4 TCC G1 T2 Recurrent tumour, treated Rigby and Franks (1970) RT112 G2 papillary Primary tumour, untreated Masters et al. (1986) SD TCC NR Primary tumour Paulie et al. (1983) UMUC3 TCC NR Grossman (1986) SW1710 TCC G3 Primary tumour Kyriazis et al. (1984) BFTC905 TCC G3 Primary tumour from blackfoot disease endemic area in Taiwan Tzeng et al. (1996) BFTC909 TCC G3 Primary tumour from blackfoot disease endemic area in Taiwan Tzeng et al. (1996) DSH1 G2 T1a Recurrent tumour, untreated Williams et al. (2002) SCaBER SCC moderate differentiation Untreated tumour O Toole et al. (1976) BC-3C TCC G4 Pratsinis et al. (1998) HT1376 TCC G3 invasive Untreated tumour Rasheed et al. (1977) TCCSUP TCC G4 Primary tumour, untreated Nayak et al. (1977) KU19-19 TCC G3 T3 Metastatic tumour Tachibana et al. (1995) 92-1 TCC G3 T4 Patient had metastases Yeager et al. (1998) TCC G1-2 T1 Primary Sarkar et al. (2000) 97-7 TCC G2-3 T1 Primary Sarkar et al. (2000) TCC G3 T2 Primary Sarkar et al. (2000) TCC G3 T4 Metastasis from same patient as Sarkar et al. (2000) TCC G3 T3 Primary Sarkar et al. (2000) 96-1 TCC G2-3 T3 Recurrent tumour Yeager et al. (1998) 97-1 TCC G1-2 T1/2 Primary Yeager et al. (1998) SVHUC Normal urothelial cells Immortalized in vitro by SV40 Reznikoff et al. (1988) HCV29 Apparently normal urothelium Fogh (1978) a NR ¼ not recorded

7 5224 Ras gene mutation analysis HRAS, NRAS and KRAS2 were screened for mutations in exons 1 and 2 by F-SSCP analysis. Primers with FAM or HEX labels were used to amplify exons 1 and 2 of each gene (Table 4). Sequences were amplified in 25 ml reactions using GeneAmp buffer II or Gold buffer (Applied Biosystems, Foster City, CA, USA), 2.5 mm MgCl 2, 200 mm dntps, 5 pmol of each primer, 2.5 U AmpliTaq Gold and 10 ng template DNA. Cycle parameters were 951C for 10 min, then 35 cycles of 951C for 30 s, 551C (HRAS exon 2, KRAS2 exon 1, KRAS2 exon 2, NRAS exon 2) or 601C (HRAS exon 1, NRAS exon 1) for 30 s and 721C for 30 s followed by a final incubation at 721C for 10 min. Controls were samples containing known HRAS mutations and no template controls. Aliquots of all reactions were checked by agarose gel electrophoresis. PCR products were diluted with water by a factor of between 1 in 10 and 1 in 40 depending on the yield of the reaction. In total, 1 2 ml of the diluted product was mixed with 0.5 ml of ROX-500 size standards (Applied Biosystems) and 10.5 ml of HiDi Formamide (Applied Biosystems). Samples were denatured at 951C for 2 min and snap cooled on ice. F-SCCP analysis was on a 3100 Genetic analyser (Applied Biosystems) at 181C and 301C using 5% Genescan polymer with 10% glycerol and 1 Tris-TAPS-EDTA (Applied Biosystems). Data was analysed using Genescan and Genotyper 3.7 software (Applied Biosystems). Mutation detection was by visual inspection of electropherograms by two observers. Products that showed a mobility shift under at least one of the F-SSCP conditions were sequenced following reamplification with unlabelled primers. Unincorporated primers and deoxynucleotides were removed using shrimp alkaline phosphatase and exonuclease I (Amersham Biosciences, Chalfont St Giles, UK). Sequencing reactions were carried out using the PCR primers and a BigDye Terminator V1.1 Cycle Sequencing Kit (Applied Biosystems). Data analysis was carried out using Sequencing Analysis 3.0 software (Applied Biosystems) and by visual inspection of the electropherograms. FGFR3 mutation analysis Some tumours were previously screened for mutations in all coding exons of FGFR3 by SSCP analysis (Sibley et al., 2001). Additional samples were screened by direct sequencing of exons 7, 10 and 15. Primers used for PCR amplification and sequencing are shown in Table 4. Sequences were amplified in 25 ml reactions using Gold buffer (Applied Biosystems), 1.5 mm MgCl 2, 200 mm dntps, 5 pmol of each primer, 2.5 U AmpliTaq Gold and 10 ng template DNA. Cycle parameters were 951C for 10 min, then 35 cycles of 951C for 30 s, 601C (exons 10 and 15) or 651C (exon 7) for 30 s and 721C for 30 s followed by a final incubation at 721C for 10 min. Products were sequenced as described above. Statistical analysis An hypothesis of mutual exclusivity can not be proved statistically, but is disproved by a single counter-example. Rather than hypothesis testing, it was therefore decided to produce an exact confidence limit on the odds ratio of Ras mutation in the presence or absence of FGFR3 mutations in a tumour, to provide a ceiling on the odds of Ras mutation with FGFR3. An exact logistic model was fitted to the data, with the presence/absence of Ras mutations as the outcome and the presence/absence of FGFR3 mutations as the explanatory variable. The median unbiased estimate of the odds ratio was produced, together with a 95% confidence interval. We also quote the exact w 2 test for independence of Ras and FGFR3 mutations. All calculations were carried out in SAS v8.1. Table 4 Primers used for F-SSCP of HRAS, NRAS and KRAS2 and direct sequencing of FGFR3 Primer Sequence Label a Fragment length (bp) HRAS exon1f CAGGAGACCCTGTAGGAGGA FAM 139 HRAS exon1r TCGTCCACAAAATGGTTCTG FAM HRAS exon 2F TCCTGCAGGATTCCTACCGG FAM 194 HRAS exon 2R GGTTCACCTGTACTGGTGGA FAM KRAS2 exon 1F GGCCTGCTGAAAATGACTG HEX 162 KRAS2 exon 1R GTCCTGCACCAGTAA HEX KRAS2 exon 2F TTCCTACAGGAAGCAAGT HEX 128 KRAS2 exon 2R CACAAAGAAAGCCCTCCCCA HEX NRAS exon 1F GACTGAGTACAAACTGGTGGTGG HEX 118 NRAS exon 1R GGGCCTCACCTCTATGGTG HEX NRAS exon 2F GGTGAAACCTGTTTGTTGGA HEX 103 NRAS exon 2R ATACACAGAGGAAGCCTTCG HEX FGFR3 exon 7F CGGCAGTGGCGGTGGTGGTG 120 FGFR3 exon 7R AGCACCGCCGTCTGGTTGGC FGFR3 exon 10F CCTCAACGCCCATGTCTTTTC 173 FGFR3 exon 10R GGGAGCCCAGGCCTTTCTTG FGFR3 exon 15F CCGCAATGTGCTGGTGAC 150 FGFR3 exon 15R GGCGTCCTACTGGCATGA a FAM ¼ carboxyfluorescein; HEX ¼ 6-carboxy-1,4-dichloro-2 0,4 0,5 0,7 0 -tetrachlorofluorescein

8 References Ayan S, Gokce G, Kilicarslan H, Ozdemir O, Yildiz E and Gultekin EY. (2001). Int. Urol. Nephrol., 33, Bakkar AA, Wallerand H, Radvanyi F, Lahaye JB, Pissard S, Lecerf L, Kouyoumdjian JC, Abbou CC, Pairon JC, Jaurand MC, Thiery JP, Chopin DK and de Medina SG. (2003). Cancer Res., 63, Bellus GA, Spector EB, Speiser PW, Weaver CA, Garber AT, Bryke CR, Israel J, Rosengren SS, Webster MK, Donoghue DJ and Francomano CA. (2000). Am. J. Hum. Genet., 67, Billerey C, Chopin D, Aubriot-Lorton MH, Ricol D, Gil Diez de Medina S, Van Rhijn B, Bralet MP, Lefrere-Belda MA, Lahaye JB, Abbou CC, Bonaventure J, Zafrani ES, van der Kwast T, Thiery JP and Radvanyi F. (2001). Am. J. Pathol., 158, Burchill SA, Neal DE and Lunec J. (1994). Br. J. Urol., 73, Cappellen D, De Oliveira C, Ricol D, de Medina S, Bourdin J, Sastre-Garau X, Chopin D, Thiery JP and Radvanyi F. (1999). Nat. Genet., 23, Chesi M, Brents LA, Ely SA, Bais C, Robbiani DF, Mesri EA, Kuehl WM and Bergsagel PL. (2001). Blood, 97, Czerniak B, Deitch D, Simmons H, Etkind P, Herz F and Koss LG. (1990). Br. J. Cancer, 62, Der CJ, Krontiris TG and Cooper GM. (1982). Proc. Natl. Acad. Sci. USA, 79, Elliot AY, Cleveland P, Cervenka J, Castro AE, Stein N, Hakala TR and Fraley EE. (1974). J. Natl. Cancer Inst., 53, Ellis CA and Clark G. (2000). Cell. Signal., 12, Fitzgerald JM, Ramchurren N, Rieger K, Levesque P, Silverman M, Libertino JA and Summerhayes IC. (1995). J. Natl. Cancer Inst., 87, Fogh J. (1978). Natl. Cancer Inst. Monogr., 49, 5 9. Fujita J, Yoshida O, Yuasa Y, Rhim JS, Hatanaka M and Aaronson SA. (1984). Nature, 309, Grimmond SM, Raghavan D and Russell PJ. (1992). Urol. Res., 20, Grossman HB, Wedemeyer G, Rem L, Wilson GL and Cox B. (1986). J. Urol., 136, Hart KC, Robertson SC, Kanemitsu MY, Meyer AN, Tynan JA and Donoghue DJ. (2000)., 19, Huang MY and Cohen JB. (1997). Oncol. Res., 9, Jemal A, Murray T, Samuels A, Ghafoor A, Ward E and Thun MJ. (2003). CA Cancer J. Clin., 53, Knowles MA. (1999). BJU Int., 84, Knowles MA and Williamson M. (1993). Cancer Res., 53, Kyriazis AA, Kyriazis AP, McCombs III WB and Peterson WBJ. (1984). Cancer Res., 44, Levesque P, Ramchuuren N, Saini K, Joyce A, Libertino J and Summerhayes IC. (1993). Int. J. Cancer, 55, Masters JR, Hepburn PJ, Walker L, Highman WJ, Trejdosiewicz LK, Povey S, Parkar M, Hill BT, Riddle PR and Franks LM. (1986). Cancer Res., 46, Nayak SK, O Toole C and Price ZH. (1977). Br. J. Cancer, 35, O Toole C, Nayak S, Price Z, Gilbert WH and Waisman J. (1976). Int. J. Cancer, 17, O Toole C, Price ZH, Ohnuki Y and Unsgaard B. (1978). Br. J. Cancer, 38, Olderoy G, Daehlin L and Ogreid D. (1998). Anticancer Res., 18, Ooi A, Herz F, Ii S, Cordon-Cardo C, Fradet Y, Mayall BH and Cancer NMNfB. (1994). Int. J. Oncol., 4, Oxford G and Theodorescu D. (2003). J. Urol., 170, Paulie S, Hansson Y, Lundblad ML and Perimann P. (1983). Int. J. Cancer, 31, Pratsinis H, Saetta A, Gagos S and Davaris P. (1998). In vitro Cell. Dev. Biol. Anim., 34, Proctor AJ, Coombs LM, Cairns JP and Knowles MA. (1991)., 6, Przybojewska B, Jagiello A and Jalmuzna P. (2000). Cancer Genet. Cytogenet., 121, Rasheed S, Gardner MB, Rongey RW, Nelson-Rees WA and Arnstein P. (1977). J. Natl. Cancer Inst., 58, Reznikoff CA, Kao C, Messing EM, Newton M and Swaminathan S. (1993). Semin. Cancer Biol., 4, Reznikoff CA, Loretz LJ, Christian BJ, Wu SQ and Meisner LF. (1988). Carcinogenesis, 9, Rigby CC and Franks LM. (1970). Br. J. Cancer, 24, Sarkar S, Julicher KP, Burger MS, Della Valle V, Larsen CJ, Yeager TR, Grossman TB, Nickells RW, Protzel C, Jarrard DF and Reznikoff CA. (2000). Cancer Res., 60, Sibley K, Cuthbert-Heavens D and Knowles MA. (2001)., 20, Tachibana M, Miyakawa A, Tazaki H, Nakamura K, Kubo A, Hata J, Nishi T and Amano Y. (1995). Cancer Res., 55, Taylor CF and Taylor GR. (2004). Methods Mol. Med., 92, Theodorescu D, Cornil I, Fernandez BJ and Kerbel RS. (1990). Proc. Natl. Acad. Sci. USA, 87, Tzeng CC, Liu HS, Li C, Jin YT, Chen RM, Yang WH and Lin JS. (1996). Anticancer Res., 16, Uchida T, Wada C, Ishida H, Egawa S, Ao T, Yokoyama E and Koshiba K. (1995). Urol. Int., 55, van Rhijn BW, Lurkin I, Radvanyi F, Kirkels WJ, van der Kwast TH and Zwarthoff EC. (2001). Cancer Res., 61, van Rhijn BW, van der Kwast TH, Vis AN, Kirkels WJ, Boeve ER, Jobsis AC and Zwarthoff EC. (2004). Cancer Res., 64, Visvanathan KV, Pocock RD and Summerhayes IC. (1988). Res, 3, Wagner HE, Joyce AD, Beatrice K and Summerhayes IC. (1990)., 5, Webster MK and Donoghue DJ. (1997). Trends Genet., 13, Williams RD. (1980). Invest. Urol., 17, Williams SV, Sibley KD, Davies AM, Nishiyama H, Hornigold N, Coulter J, Kennedy WJ, Skilleter A, Habuchi T and Knowles MA. (2002). Genes Chromosomes Cancer, 34, Yeager TR, DeVries S, Jarrard DF, Kao C, Nakada SY, Moon TD, Bruskewitz R, Stadler WM, Meisner LF, Gilchrist KW, Newton MA, Waldman FM and Reznikoff CA. (1998). Genes Dev., 12, Zhang ZT, Pak J, Huang HY, Shapiro E, Sun TT, Pellicer A and Wu XR. (2001)., 20, Zhang ZT, Pak J, Shapiro E, Sun TT and Wu XR. (1999). Cancer Res., 59,