Introduction

Inactivation of the CDKN2A locus at chromosome 9p21 is one of the most frequent genetic changes in human cancers (Kamb et al, 1994). This unusual locus encodes two distinct proteins (Quelle et al, 1995) involved in two pathways crucial for the regulation of cell proliferation and the control of genomic integrity. The p16INK4A protein, encoded by exons E1α, E2, and E3, is a cell cycle regulator that limits the activity of cyclin-dependent kinases and inhibits entry into S-phase when it is accumulated. The p14ARF1 protein is encoded by E1β, E2, and E3 in a different reading frame, is induced during cell proliferation and supports activation of p53 by blocking the p53 inhibitor, MDM2. Reported mechanisms of CDKN2A inactivation in cancer cells comprise deletion, mutation, and promoter hypermethylation. The frequencies of CDKN2A inactivation and the relative importance of each of these inactivation mechanisms differ considerably between different cancers (Ruas and Peters, 1998).

Loss of chromosome 9 has long been recognized as the most characteristic chromosomal aberration in transitional cell carcinoma (TCC) of the urinary bladder (Knowles, 1999; Orntoft and Wolf, 1998). Many TCC display loss of one entire chromosome 9, others show loss of regions only on the short (9p) or the long arm (9q). Although the genes impaired by changes on 9q have not been identified, overwhelming evidence indicates that CDKN2A is an important locus targeted by alterations on chromosome 9p. Homozygous deletions of this locus have been observed in several studies (Balazs et al, 1997; Baud et al, 1999; Orlow et al, 1995, 1999; Packenham et al, 1995; Williamson et al, 1995) and hypermethylation of the CpG island around exon E1α of CDKN2A has been observed in cell lines and primary tumors (Akao et al, 1997; Gonzalez-Zulueta et al, 1995; Gonzalgo et al, 1998; Merlo et al, 1995). Additionally, point mutations affecting both reading frames have been found, albeit with low frequency (Baud et al, 1999; Kai et al, 1995), and re-expression of p16INK4A has been shown to suppress the growth of bladder carcinoma cells harboring an inactivated gene (Bender et al, 1998; Gonzalgo et al, 1998; Grim et al, 1997).

Despite ample work in this area, several issues have not been fully resolved. First, some studies in TCC and other carcinomas have observed two distinct regions of loss of heterozygosity (LOH) around CDKN2A, suggesting that more than one tumor suppressor gene may be located in this region (Czerniak et al, 1999). Second, although hypermethylation of the CpG-island encompassing exon 1α is associated with transcriptional repression, the frequency of hypermethylation reported in different studies on TCC (Akao et al, 1997; Gonzalez-Zulueta et al, 1995; Gonzalgo et al, 1998; Merlo et al, 1995) has varied considerably. Third, the relationship between CDKN2A alterations and progression of TCC is a matter of debate. On one hand, loss of CDKN2A function seems to be required for immortalization of urothelial carcinoma cells in culture (Yeager et al, 1998). Accordingly, several studies have suggested that such alterations are associated with tumor progression (Orlow et al, 1999). On the other hand, loss of chromosome 9, including the region of the CDKN2A locus, has been reported in well-differentiated papillary TCC (Balazs et al, 1997; Hartmann et al, 1999; Jung et al, 1999), which rarely progress to invasiveness.

Although not all studies agree, the overall published data on the mechanisms of CDKN2A inactivation in TCC suggest that deletions are the most common alteration in this tumor type, whereas point mutations or DNA hypermethylation predominate in other tumors (Ruas and Peters, 1998). Because the various modes of inactivation are probably functionally equivalent, the causes for this difference must be sought in the mechanisms of mutation characteristic for each tumor type. In this regard, it may be relevant that early invasive TCC or their presumptive precursors such as high grade papillary tumors (pTaG3) and carcinoma in situ (CIS) often show a high number of chromosomal alterations (Kallioniemi et al, 1995; Presti et al, 1991; Zhao et al, 1999). This chromosomal instability may be promoted by the loss of p53 function in many TCC that is associated with a high risk of progression (Esrig et al, 1994). However, TCC are also characterized by a high prevalence of genome-wide DNA hypomethylation, which is extensive in some cases (Florl et al, 1999; Jürgens et al, 1996). Although focal DNA hypermethylation contributes to the inactivation of individual genes such as CDKN2A, there is evidence that genome-wide DNA hypomethylation in tumors is associated with an increased risk of illegitimate recombination, gene deletion and chromosome loss (Ehrlich, 2000; Schulz, 1998). Moreover, increased frequencies of chromosomal loss and gene deletions have been observed in patients with the inherited immunodeficiency, centromeric region instability, facial anomalies (ICF) syndrome in which hypomethylation of certain repetitive sequences occurs (Jeanpierre et al, 1993; Xu et al, 1999), and in mice with decreased levels of methylation caused by targeted deletions of DNA methyltransferase 1 (Chen et al, 1998). Therefore, the high prevalence of DNA hypomethylation might explain the predominance of deletions being the mechanism of CDKN2A gene inactivation in TCC.

These issues were addressed in the present study, based on the analysis of 86 TCC samples employing LOH analysis by quantitative multiplex PCR, detection of homozygous deletions by quantitative PCR, DNA hypermethylation analysis by restriction enzyme PCR (RE-PCR) and methylation-specific PCR (MS-PCR), and quantification of genome-wide hypomethylation by Southern blot analysis of LINE L1 sequence methylation.

Results

LOH Analysis

LOH analysis was first performed using five microsatellite markers surrounding the CDNK2 locus and distributed across 9p21 (Fig. 1). DNA from each tumor sample was compared with normal DNA isolated from leukocytes or with morphologically normal corresponding tissue. LOH analysis was performed in multiplex reactions (Fig. 2) with quantitative evaluation of signal intensities as detailed in the “Materials and Methods” section. Thirty-four tumors reproducibly showed LOH for at least one marker. In eight specimens, LOH encompassed all informative markers. However, D9S942, D9S157, or both D9S942 and D9S157 showed apparent retention of heterozygosity in 23 tumors, although all flanking markers displayed definitive LOH. Few specimens (eg, specimens 60 and 79) exhibited localized loss (Fig. 3).

Figure 1
figure 1

Partial map of chromosome 9p21. A, Locations of the CDKN2A and CDKN2B loci and the microsatellites analyzed. B, DNA fragments amplified by PCR are indicated.

Full size image
Figure 2
figure 2

Loss of heterozygosity (LOH) at 9p21 in bladder cancers. Typical results of LOH analysis. PCR products from paired normal (N) and tumor (T) specimens were loaded next to each other. Left, Microsatellite markers D9S171 and D9S942: Cases 197 and 94 show LOH for both markers, Case 193 shows LOH for D9S942 and D9S171 is noninformative. Sample 86 shows LOH for D9S171 and retention of both alleles of marker D9S942. Right, Microsatellite markers D9S168, D9S162, and D9S157: Cases 124 and 85 show LOH for all three markers; Sample 91 shows LOH only at D9S168 and D9S162. Cases 152 and 74 are noninformative or retain heterozygosity at the three markers.

Full size image
Figure 3
figure 3

Summary of LOH analysis data. □, retention of heterozygosity; ⍁, LOH; ░ not informative; ▦, microsatellite instability; ░, homozygous deletion.

Full size image

In most cases with apparent retention of heterozygosity at D9S942 or D9S157, for both alleles the band intensities were lower than the other microsatellite marker in the same lane, and also lower than the same microsatellite marker for the corresponding normal tissue sample (Fig. 2, Lanes 86 and 94). This situation strongly suggests loss of both alleles, ie, a homozygous deletion encompassing the microsatellites. To verify this assumption, all respective cases were analyzed for deletions of exon 1α and exon 2 of the CDKN2A gene by quantitative PCR (see “Materials and Methods”). In most of these cases, a homozygous deletion of exon 2 or both exons was observed (Fig. 3). Likewise, in cases where apparent retention of heterozygosity of DS157 or DS162 was observed alongside that of D9S942 (specimens 50, 91, and 186), homozygous deletions were present in the CDKN2A gene. Overall, homozygous deletions of the CDKN2A gene were detected in 17 of 34 cases (50%) with LOH.

Hypermethylation Analysis

In tumors with LOH but no homozygous deletion at the CDKN2A gene, hypermethylation of the CpG island at E1α was investigated as a possible cause of gene inactivation. Two different methods were used, RE-PCR and MS-PCR. DNA from the bladder carcinoma cell line T24 displaying hypermethylation of this CpG island was used as a positive control in all assays, and DNA from the bladder carcinoma cell line 5637, which lacks Rb and over-expresses p16INK4A, was used as a negative control.

In RE-PCR, DNA from the CpG-island of interest that is not digested by the methylation-sensitive enzyme HpaII is amplified using a primer pair flanking the restriction sites (cf. Fig. 1B). A PCR signal is obtained only if all HpaII recognition sites are methylated. However, a signal can also be obtained from small amounts of remaining undigested DNA after extensive amplification. This amount can be determined by digestion with the methylation-insensitive HpaII isoschizomer MspI. Figure 4 shows that some samples yielded a weak signal after a limited number of PCR cycles even after extensive digestion with the methylation-insensitive restriction enzyme (see “Materials and Methods”). This constitutes a background against which the HpaII signal must be measured to detect and quantify hypermethylation. Dilution of DNA from T24 cells, which contain a strongly hypermethylated E1α, with DNA from 5637 cells, which lack methylation in this region, showed that approximately 5% hypermethylation can be reliably detected over this background (data not shown). Among 16 tumor samples showing LOH but no evidence for homozygous deletions at the CDKN2A locus, 3 samples (specimens 102, 182, and 193) were clearly and reproducibly positive in this assay. In contrast, in 17 samples without LOH, no samples had hypermethylation of the E1α CpG island above background. Furthermore, in several specimens with homozygous deletions, no samples showed evidence of hypermethylation.

Figure 4
figure 4

Methylation analysis by restriction enzyme PCR (RE-PCR). Analysis of CDKN2A exon 1α in two human bladder cell lines and five bladder tumor samples. DNA was digested with HpaII (H), or MspI (M), or left undigested (U) before PCR amplification with primers described in Table 1. Uncut DNA (U) was amplified in all cases. After digestion with MspI, either no signals or weak signals were obtained, indicating completeness of the restriction enzyme digestion. Cases 102, 182, and 193 and the T24 cell line show significant methylation as evidenced by a strong signal in the H lane. * Molecular weight marker pUC-MspI digest.

Full size image

In MS-PCR, DNA treated with bisulfite, which converts cytosine to uracil but does not react with methylcytosine, is amplified using two distinct primer pairs (Table 1), which selectively recognize either the converted or non-converted sequence. By diluting T24 DNA with 5637 DNA, the sensitivity of the method was determined to be less than 1:10. Using this method, the same three tumor samples with LOH in the 9p21 region (specimens 102, 182, and 193) showed hypermethylation, which was not found in any of the other samples from this group (Fig. 5). Among 21 carcinomas without evidence of LOH at 9p21, no samples were positive. Likewise, as expected, hypermethylation was detected in none of the eight specimens with homozygous deletions.

Table 1 Oligonucleotides Used
Full size table
Figure 5
figure 5

Methylation-specific polymerase chain reaction (MS-PCR) analysis of CDKN2A exon 1α. Amplification of bisulfite-treated DNA from bladder cancer cell line T24 and 11 bladder tumor samples. Primer sets for unmethylated (U) and methylated (M) DNA were used as described in Table 1. The same samples as in Figure 4 react positively. * Molecular weight marker pUC-MspI digest.

Full size image

Genome-Wide Hypomethylation

Genome-wide hypomethylation was investigated in 84 TCC specimens with sufficient available DNA by Southern blot analysis of the methylation status of LINE L1 sequences, which comprise between 10% and 15% of the human genome (Kazazian and Moran, 1998). These sequences are almost completely methylated in normal somatic tissues. Their hypomethylation in TCC can be used as a measure of overall decreased methylation (Florl et al, 1999). In line with this previous report, L1 hypomethylation was detected in almost all of the tumor specimens (Fig. 6). As is evident in Figure 6, many of the samples without alterations at chromosome 9p21 had low degrees of hypomethylation, with an average level of hypomethylation of 11 ± 13% (mean ± sd). By comparison, the average level of hypomethylation in the samples with LOH at 9p21 but no homozygous deletion was 26 ± 18% and the average level of the samples with homozygous deletions was 28 ± 18%. The difference between either of the latter groups and the first group was statistically significant (Wilcox rank sum test: p = 0.0015 and 0.0001, respectively). All three specimens with hypermethylation at CDKN2A also displayed pronounced genome-wide hypomethylation.

Figure 6
figure 6

Genome-wide hypomethylation and deletions at CDKN2A in bladder cancers. The extent of DNA hypomethylation, as reflected by L1 LINE. Hypomethylation is indicated for each transitional cell carcinoma (TCC) specimen. The three cases showing hypermethylation at CDKN2A in RE-PCR and MS-PCR (see Figs. 4 and 5) are indicated by a filled circle. Left column, Homozygous deletion at CDKN2A with LOH at flanking markers; center column, LOH at 9p21; right column, no alterations detected at 9p21. Average values for each group are indicated by horizontal bars. (HD, homozygous deletion)

Full size image

Relation to Clinical Parameters

The relationship between alterations on chromosome 9p21 and clinical parameters is compiled in Table 2. Neither the presence of LOH at 9p21 nor the presence of homozygous deletions showed any significant relationship to tumor stage or grade. Specifically, the percentage of LOH at 9p21 or homozygous deletions at CDKN2A was almost the same between muscle-invasive tumors and earlier stages and between tumors with or without metastases. The relationship to tumor grade was also not statistically significant, although none of three well-differentiated tumors, but all G4 carcinomas, exhibited alterations. However, the frequencies of LOH and homozygous deletion were almost the same between moderately and poorly differentiated tumors.

Table 2 Relationship of Alterations at 9p21 to Stage and Grade in TCC *
Full size table

Discussion

In accord with cytogenetic data identifying loss of chromosome 9 as the most frequent chromosomal alteration in TCC, molecular analyses have consistently shown a high rate of LOH at 9p21, often in the range of 30% to 40%, as was found in the present study. The reported frequencies of homozygous deletions at the CDKN2A locus have been more variable (Balazs et al, 1997; Baud et al, 1999; Orlow et al, 1995, 1999; Packenham et al, 1995; Williamson et al, 1995), the frequency in the present series (17 of 86, 20%) being in the higher range. These differences may reflect patient selection and geographical variation to some extent, but may be compounded by differences in methods. The high frequency of homozygous deletions detected in this study may be because of the use of a combination of multiplex PCR LOH analysis and a quantitative assay for gene copy number, as suggested by others (Cairns et al, 1998). Of note, a remarkably sensitive indicator of homozygous deletion in our study was an apparent retention of heterozygosity at microsatellite markers close to the CDKN2A locus (Fig. 2). This apparent retention, probably caused by contamination with stromal cells, may have inspired reports of distinct regions of LOH at 9p in TCC and other cancers. Indeed, few studies have been performed on microdissected material, presumably because most TCC contain a high proportion of tumor cells anyhow, the amount of DNA obtained is limited, and the concentration and quality of DNA are difficult to ascertain for subsequent analyses. However, if the amounts of DNA used and the intensity of the PCR signal in multiplex reactions are closely monitored, this experimental artifact can obviously be used to identify cases with homozygous deletions.

Because all previous studies agree that point mutations in the coding region of the CDKN2A gene are rare in bladder cancer (Kai et al, 1995; Packenham et al, 1995), we have not performed a systematic search for mutations in the present series of tumors. A preliminary screen by PCR, single-strand conformation polymorphism (SSCP), and sequencing has identified four cases with potential mutations yielding a figure in the expected range. However, even when the number of homozygous deletions and of point mutations affecting CDKN2A was somewhat underestimated, a substantial number (up to 13 of 34 cases) of carcinoma specimens with LOH without either of these alterations would remain. In three of these cases, DNA hypermethylation was detected by two independent methods (Figs. 4 and 5). Because hypermethylation was not detected in any specimen without LOH, the overall frequency of hypermethylation was lower than that in other studies (Akao et al, 1997; Gonzalez-Zulueta et al, 1995; Gonzalgo et al, 1998; Merlo et al, 1995). Technical reasons may account for this discrepancy. Specifically, RE-PCR tends to yield false positives because of incomplete digestion of DNA and over-amplification, particularly when used on archival tissue. False positives in MS-PCR can result from incomplete conversion of cytosines in the bisulfite reaction (Rein et al, 1998). However, because the highest frequency of hypermethylation has been reported in Japanese patients (Akao et al, 1997), geographical factors cannot be excluded as a cause of the differences. Nevertheless, the present data suggest that in initial reports the frequency of CDKN2A inactivation by hypermethylation in TCC may have been overestimated. Moreover, our data raise the possibilities that either a further unidentified mechanism exists for inactivation of the CDKN2A locus or that a sizable number of cases with LOH are associated with hemizygosity rather than complete inactivation of this gene.

Because monosomy of chromosome 9 is a frequent change in superficial TCC, it is not surprising that, in most investigations, LOH at 9p21 was not related to tumor stage or grade, as was found in the present study. However, because homozygous deletion of CDKN2A provides an alternative pathway to simultaneous loss of p53 and Rb function in vitro (Markl and Jones, 1998; Yeager et al, 1998), the finding in the present study that the frequency of homozygous deletions did not increase with tumor stage and grade (Table 2) is somewhat surprising. Notably, the specimens in this study did not include well-differentiated papillary tumors (pTaG1), but had all progressed to require major surgical intervention. The cumulative findings are best explained by the hypothesis that inactivation of CDKN2A is permissive for the progression of TCC but is not the actual stimulus.

Taken together, the data corroborate the hypothesis that chromosomal loss and deletions are the most important causes underlying CDKN2A inactivation in TCC. Because DNA hypomethylation has been experimentally linked to chromosome instability, we previously speculated that the high rate of chromosomal alterations found in TCC are related to the high prevalence of hypomethylation (Schulz, 1998). According to this hypothesis, the extent of DNA hypomethylation should be related to the frequency of deletions overall and at any particular locus. The present study demonstrates this relationship for the CDKN2A locus (Fig. 6). All carcinomas with LOH or homozygous deletions displayed significant DNA hypomethylation and their average extent of hypomethylation was significantly higher. Significant hypomethylation was also observed in the three tumors exhibiting hypermethylation of the CDKN2A exon 1α underlining our previous demonstration that hypermethylation of the calcitonin gene and genome-wide hypomethylation coexist in individual bladder carcinoma cells (Jürgens et al, 1996). Because different types of tumors differ considerably in the frequency of genome-wide DNA hypomethylation (Ehrlich, 2000; Schulz, 1998), the present data provide an explanation why deletions are the most prevalent mechanism of CDKN2A inactivation in TCC, whereas point mutations predominate in certain other tumors.

Material and Methods

Cell Lines

The human bladder carcinoma cell lines T24, 5637, and SD were cultured in DMEM with Glutamax I, sodium pyruvate, 4.5 g/l glucose, supplemented with 15% fetal bovine serum (Biochrom, Berlin, Germany), and 100 μg/ml of penicillin/streptomycin (Life Technologies, Inc., Eggertshausen, Germany) at 37° C in a humidified atmosphere with 5% CO2.

Patients and Tissue Samples

TCC samples were obtained by cystectomy, nephroureterectomy, or transurethral resection from 86 patients between 1994 and 1999. Tumors and corresponding normal tissue were dissected after surgery, frozen in liquid nitrogen, and stored at −80° C. Blood samples from tumor patients were routinely collected for DNA extraction. Grading and staging were performed according to the 1997 Union Internationale Contre le Cancer (UICC) classification for bladder carcinoma. Among the 86 patients, 64 were male and 22 female. Ages ranged from 41 to 97 years with a mean age of 58 years. In all, 6 tumors were staged as pTa, 10 carcinomas as pT1, 22 as pT2, 32 as pT3, 15 as pT4, and 1 as pCIS. Three tumors were graded as G1, 23 as G2, 55 as G3, and 4 as G4, respectively. One case of carcinoma in situ was not graded.

DNA Extraction

High molecular weight genomic DNA from cell lines, tissue, and whole blood was isolated using the blood and cell culture DNA kit (Qiagen, Hilden, Germany). Frozen tissues were crushed to a fine powder with a mortar and pestle before extraction.

LOH Analysis

Matched pairs of normal and tumor DNA samples were screened for LOH at 9p21 at the five microsatellite loci: D9S168, D9S157, D9S942, D9S162, and D9S171. Primer sequences (Table 1) were obtained from the Genome DataBase. Testing of markers was performed in two multiplex PCR analyses. Template DNA (100 ng) was amplified in a total volume of 50 μl containing 150 μm of each dNTP, 1.5 mm MgCl2, 15 pmol of each primer (forward primers modified at the 5′end with IRD800), and 2 U of DyNAzymeII Taq polymerase (Biozym, Göttingen, Germany). After initial denaturation at 95° C for 5 minutes, 26 cycles of 30 seconds at 94° C, 30 seconds at 55° C, and 1 minute at 72° C each were performed for primer sets 3 and 5, and 26 cycles of 94° C for 45 seconds, 55° C for 45 seconds, and 72° C for 45 seconds each for primer sets 1, 2, and 4. All reactions included a final elongation step at 72° C for 10 minutes.

PCR products were analyzed with an automated infrared DNA electrophoresis system (LI-COR DNA 4000/4200, MWG-Biotech, Ebersberg, Germany). After addition of 4 μl of formamide loading buffer to the 1:6-diluted PCR product, the sample was heated to 95° C for 5 minutes and snap-cooled. From each mixture, 1.5 μl per lane was loaded onto a 6% denaturing polyacrylamide gel (SequaGel, National Diagnostics, Atlanta, Georgia). Before loading the samples, the gel was pre-run for 30 minutes at 1000 V. Electrophoresis was carried out at a constant voltage of 1000V at 50° C. The data are presented as an radioautograph-like image that is stored in TIF format using the image manipulation subprogram of the LICOR Base ImagIR software package. Band analysis was performed using OneDScan 1.2 software (MWG-Biotech). An allele was regarded as lost when the intensity of remaining signal was less than 60% of the signal from the same allele in the matching control DNA of the same patient, as previously described (An et al, 1996).

MS-PCR

MS-PCR was performed as described by Herman et al (1996). Briefly, 0.5 to 1 μg of DNA from each sample was treated with bisulfite using the CpGenome DNA modification kit (Oncor, Heidelberg, Germany). Aliquots from the reaction mixture were used in two separate PCR reactions with primer pairs for CDKN2A exon 1α specific for converted (non-methylated) or unconverted (methylated) DNA (Herman et al, 1996). PCR products were separated on 6% polyacrylamide gels (AccuGel 29:1, National Diagnostics, Atlanta, Georgia) and visualized by silver staining. DNA from the bladder carcinoma cell line T24 was used as a positive control with each batch of samples.

RE-PCR Methylation Analysis

One microgram of DNA was digested with 10 U of methylation-sensitive HpaII or methylation-insensitive MspI restriction enzymes overnight. Digestion was then continued for 2 hours with 5 U of freshly added enzyme to ensure complete digestion. After 10 minutes of incubation at 65° C to inactivate the enzymes, 100 ng of each DNA sample was amplified in a total volume of 50 μl containing 150 μm each of dNTP, 5% DMSO, 5 mm MgCl2, 40 pmol of primer set 9, and 1.2 U of Taq polymerase. Cycling consisted of an initial denaturation at 95° C for 5 minutes, 30 cycles each of 95° C for 30 seconds, 67° C for 30 seconds, and 72° C for 60 seconds. The final elongation step was at 72° C for 10 minutes. PCR products were separated on 6% polyacrylamide gels and visualized by silver staining.

Homozygous Deletion Testing

Testing for homozygous deletion was performed using primer set 11 for CDKN2A exon 2, primer set 10 for exon 1α, and primer set 14 for GAPDH as an internal standard. One hundred nanograms of template DNA was amplified in a total volume of 50 μl containing 150μm dNTP, 1.5 mm MgCl2, 5% DMSO, 20 pmol of each primer, and 2 U of Taq polymerase. Forward primers were modified at the 5′-end with IRD 800. For exon 1α, the PCR conditions were initial denaturation at 95° C for 5 minutes, and 30 cycles each of 95° C for 1 minute, 60° C for 1 minute, and 72° C for 1 minute. The final elongation step was at 72° C for 10 minutes. For p16 exon 2 and GAPDH, cycling consisted of an initial denaturation at 95° C for 5 minutes, and 30 cycles each of 95° C for 45 seconds, 56° C for 45 seconds, and 72° C for 1 minute. The final elongation step was at 72° C for 10 minutes. For separation of the PCR products, an 8% denaturing polyacrylamide gel was used on the automated sequencer. Control PCR was performed using different ratios of DNA from normal lymphocytes and from the bladder carcinoma cell line SD with a homozygous deletion of CDKN2A.

Mutation analysis

For mutation analysis by PCR-SSCP, exon 2 of CDKN2A was amplified as two overlapping fragments of 285 bp and 260 bp, using primer sets 12 and 13, respectively. Thirty-five PCR cycles were carried out consisting of 45 seconds each at 95° C, 55° C or 66° C, and 72° C, preceded by initial denaturation and final elongation as above. The 50 μl reaction mix contained 150 μm NTPS, 6 mm or 1.5 mm MgCl2, 5% DMSO, 20 pmol of primer, and 2 U of Taq polymerase. Sample aliquots were diluted 1:5 in 95% formamide, 20 mm EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol, heated at 90° C for 3 minutes, electrophoresed on an 8% partially denaturing polyacrylamide gel at 300V and 4°C for 26 hours. DNA was visualized by silver staining.

LINE-1 Methylation Analysis

Genome-wide hypomethylation was analyzed on Southern blots using a LINE L1 probe as described (Florl et al, 1999; Jürgens et al, 1996). In normal somatic tissues, L1 sequences are almost completely methylated. Hypomethylation results in the appearance of lower size bands (0.5 to 4 kb) after digestion with HpaII. The data are expressed as percent hypomethylation, which is the intensity of these bands after HpaII digestion compared with MspI digestion. The value in normal bladder tissue is 0.5% to 1% (Florl et al, 1999).