Promoter Hypermethylation Profile of Kidney Cancer

Alterations in DNA methylation, an epigenetic process present in mammalian cells, are a hallmark of human cancer. The CpG dinucleotide is not underrepresented in the promoter region of many genes, particularly “housekeeping” genes, as it is underrepresented in the remainder of the genome. These regions are termed “CpG islands” and with the exception of genes on the inactive X chromosome in females and imprinted genes, CpG islands are generally protected from methylation in normal cells. This protection is critical because methylation of CpG islands is associated with loss of expression of that particular gene. In addition to loss of heterozygosity and point mutation, alleles of tumor suppressor and cancer genes can be silenced by promoter hypermethylation. Silencing of tumor suppressor genes, such as p16^INK4a, the mismatch repair gene hMLH1, and BRCA1 have established hypermethylation as a common mechanism for tumor suppressor inactivation in human cancer (1, 2). In the types of cancer analyzed to date, promoter hypermethylation has been found to be frequent, to occur in tumor suppressor and cancer genes involved in many different signaling pathways and to be present in a different pattern in different tumor types (3, 4, 5, 6, 7, 8, 9).

Cancer of the kidney, specifically renal cell carcinomas (RCCs) and transitional cell carcinomas of the renal pelvis, accounts for ∼3% of all solid neoplasms with an incidence (estimated at 31,900 cases in the United States in 2003) roughly equal to that of all forms of leukemia combined (10). Hereditary clear cell RCC arises from an inherited mutation in the VHL tumor suppressor gene located on chromosome 3p. In the sporadic clear cell type of RCC, chromosome 3p deletion and inactivation of the VHL suppressor gene is reported to be the most common genetic alteration (11). That the inactivation of VHL is common in sporadic disease and that it is also the predisposing factor to familial clear cell RCC both suggest that alteration of VHL may be the initiating event in sporadic clear cell cancer. The MET oncogene and the fumarate hydratase (FH) tumor suppressor gene have been identified as predisposition genes for hereditary papillary renal cancer (12, 13), and the Birt-Hogg-Dube (BHD) suppressor gene has been identified as a predisposition gene for oncocytomas and chromophobe renal cancer (14). Adult sporadic cancers arise through the clonal accumulation of multiple genetic alterations, often in a general temporal order (15). Relatively little is known about the secondary and later genetic alterations that drive progression, or about the importance of different pathways in the development and progression of renal cancer. The p53, Rb, p16, and PTEN tumor suppressor genes are infrequently inactivated in RCC, and the target suppressor gene(s) on several of the most frequently deleted chromosomal arms, implicated by cytogenetic, allelotype, and comparative genomic hybridization studies in RCC, has not yet been identified (11).

DNA methylation is known to occur in kidney tumorigenesis. For example, the VHL gene is inactivated by hypermethylation in a subset of clear cell renal cancers (16), and p16^INK4a (17) and RASSF1A (18) can be hypermethylated in clear cell and other histological subtypes. However, for the most part, hypermethylation of a single gene has been examined only in limited types and numbers of kidney tumors (4, 18). To gain insight into the biological background and significance of hypermethylation in kidney tumorigenesis, we selected 10 tumor suppressor or cancer genes known to be hypermethylated with associated loss of expression in human cancer. These included the von Hippel-Lindau (VHL), p16^INK4a, p14^ARF, and adenomatous polyposis coli (APC) tumor suppressor genes; the DNA repair gene O⁶-methyl-guanine-DNA-methyltransferase (MGMT); the detoxifying gene glutathione S-transferase π1 (GSTP1); the putative suppressor genes retinoic acid receptor β2 (RARβ2) and RAS association domain family protein 1A (RASSF1A); and the invasion and metastasis genes E-cadherin and tissue inhibitor of metalloproteinase-3 (Timp-3); as well as the recently described fumarate hydratase (FH) tumor suppressor gene, inherited mutation of which can predispose to familial papillary renal carcinoma (13). Our study profiled promoter hypermethylation status by methylation-specific PCR (MSP) in a series of 100 tumors representative of all major types of kidney cancer. The gene hypermethylation status was correlated to clinicopathological features of the kidney cancer patients.

After approval from the Fox Chase Institutional Review Board, tumor tissue was obtained from the Fox Chase Cancer Center Tumor Bank Facility from 100 patients, ages 30–80 years, who underwent nephrectomy or nephro-ureterectomy for enhancing renal masses. Tumors were graded according to the American Joint Committee on Cancer (19) and staged according to the 1997 tumor-node-metastasis system (Ref. 20; Fig. 1). Specimens of normal kidney distant to the tumor were examined from 10 of the RCC patients as controls. In addition, specimens of histologically confirmed normal ureteral urothelium were collected from five patients with RCC to provide DNA from normal transitional cells. Tumor tissue was obtained immediately after surgical resection and was imbedded in optimum cold temperature medium (OCT). A H&E-stained section was examined by a pathologist (T. A-S.), and the area of highest neoplastic cell content was selected and manually microdissected. DNA was extracted from tumor or normal tissue using a standard technique of digestion with proteinase K in the presence of SDS at 37°C overnight, followed by phenol/chloroform extraction and precipitation with 100% ethanol (21).

Specimen DNA (0.5–1 μg) was modified with sodium bisulfite, converting all unmethylated, but not methylated, cytosine to uracil followed by amplification with primers specific for methylated versus unmethylated DNA. The genes used in the profile were VHL (22), p16^INK4a (22), p14^ARF (23), APC (24), GSTP1 (25), MGMT (26), RASSF1A (27), RARβ2 (28), E-Cadherin (29), and Timp-3 (30). MSP analysis of 6 of the genes in 50 tumors was previously reported (31). The primer sequences used have all been reported previously and can be found in the report referenced after each gene. The primers for RASSF1A include CpG site positions 7–9 on the forward primer and positions 13–15 on the reverse primer, as described previously (27). The primer sequences for RARβ2 are as described previously (28) but with the addition of 2 bp to the 5′ end of both primers for the unmethylated product to aid in discrimination between the unmethylated and methylated RARβ2 PCR products. PCR amplification of template DNA was performed for 31–35 cycles at 95°C denaturing, 58–66°C annealing, and 72°C extension with a final extension step of 5 min. Cycle number and annealing temperature depended on the primer set to be used, each of which had been previously optimized for the PCR technology in our laboratory. For each set of DNA modification and PCR, a cell line or tumor with known hypermethylation as a positive control, normal lymphocyte or normal kidney tissue DNA as a negative control, and water with no DNA template as a control for contamination, were included. If no tumor cell line with known hypermethylation of a particular gene (VHL, APC, E-cadherin) was available, normal human lymphocyte DNA in vitro methylated with SssI methylase according to the manufacturers’ instructions (New England Biolabs, Beverly, MA) was used as a positive control. After PCR, samples were run and analyzed using a 6% nondenaturing acrylamide gel with appropriate size markers.

The FH gene promoter sequence data were obtained from GenBank Database accession no. NM_000143. We selected the CpG island situated between −181 upstream and +166 bases downstream from the ATG codon. This region fulfilled the original CpG island definition criteria of Gardiner-Garden and Frommer (32) and the modifications suggested by Takai and Jones (33). No Alu repetitive elements were detected by REPEAT-MASKER mail server (University of Washington Genome Center, Seattle, WA).⁶ Papillary renal tumor DNAs were modified with sodium bisulfite as described above and were used as template for PCR amplification of the FH promoter CpG island region containing 35 CpG sites with the following primers; Forward: 5′-GAAGGTTTTATATTTTATATTATTAT-3′ positioned at −207 and −181, Reverse: 5′-AACAAAAAAACTAAAAATC-3′ positioned at +166 and +185 from the ATG codon. The resulting 392-bp PCR product was cycle-sequenced.

Binary logistic regression was conducted to assess whether patient age was associated with the hypermethylation of each individual gene. Fisher’s exact test was used to explore the pairwise association between genes with respect to hypermethylation and whether the hypermethylation of a given gene was related to tumor stage, grade, or cell type or with patient gender and smoking status (smoker versus nonsmoker). Results were declared statistically significant at the two-sided 5% comparison-wise significance level (i.e., without correction for multiple comparisons). Kaplan-Meier analysis was used to examine any association between gene hypermethylation and time to recurrence as well as overall survival.

We examined the hypermethylation status of a panel of 10 normally unmethylated tumor suppressor or cancer genes: VHL, p16^INK4a, p14^ARF, APC, MGMT, GSTP1, RARβ2, RASSF1A , E-cadherin, and Timp-3 in 100 kidney tumor (50 clear cell, 20 papillary, 6 chromophobe, 5 collecting duct, 5 unclassified RCC, 7 oncocytoma, 6 transitional cell of renal pelvis, and 1 Wilms’ tumor) DNAs using the MSP assay (Fig. 1). The frequency of promoter hypermethylation of the tumor suppressor gene loci included in the panel was VHL 8%, p16^INK4a 10%, p14^ARF 17%, APC 14%, MGMT 7% GSTP1 12%, RARβ2 12%, RASSF1A 45%, E-cadherin 11%, and Timp-3 58% of the 100 tumors (Table 1). Fig. 2 shows representative examples of MSP analysis of each gene. Hypermethylation was observed in all of the histological cell types, grades, and stages of kidney cancer examined and in patients of all ages. Ninety-three tumors showed methylation of at least 1 gene, and 7 tumors showed no methylation of any of the 10 genes. A total of 33% of kidney tumors had one gene, 34% two genes, 15% three genes, 8% four genes, 2% five genes, and 1% seven genes hypermethylated (Fig. 1). The mean number of genes hypermethylated in each tumor was 1.94 (SD, 1.229), and the median number was 2. No methylation was observed in 10 normal renal or 5 normal ureteral tissue DNAs. We did not find any evidence of methylation of the CpG island in the promoter of the FH gene after bisulfite sequencing of 15 papillary renal tumors (Fig. 3). We, therefore, did not perform further analysis of the FH gene.

Using statistical analysis, we examined methylation with regard to the kidney cancer patient clinicopathological parameters of age, gender, cell type, grade, stage, size (T_1a versus T_1b), and smoking history (yes or never) and with regard to associations between methylation of one particular gene and methylation of another gene in a particular tumor. VHL methylation was only found in clear cell tumors (P = 0.006) and only in males (P = 0.048). Hypermethylation of RASSF1A was significantly more frequent in papillary renal tumors (P = 0.011) and in higher grade tumors of all cell types (P = 0.003). MGMT methylation was more frequent in women (P = 0.049) and also in never-smokers of both sexes (P = 0.045). MGMT methylation also showed a trend to associate with high stage but not at a statistically significant level (P = 0.0868). GSTP1 methylation was associated with older patients (P = 0.029). Hypermethylation of several individual genes was negatively associated with hypermethylation of other individual genes (all, P < 0.05), specifically, p14^ARF with RASSF1A, RARβ2 or p16^INK4a; p16^INK4a with APC or RARβ2; and RASSF1A with APC. There was no association between gene hypermethylation and survival.

The growing number of tumor suppressor and other cancer genes reported to be hypermethylated with associated transcriptional silencing provides an opportunity for the examination of the pattern of epigenetic alteration in kidney cancer cells (1, 2, 4). The majority of kidney cancers (80–85%) are RCCs originating from the renal parenchyma. The remaining 15–20% are mainly transitional cell carcinomas of the renal pelvis. The classification of RCC includes several histological subtypes with different genetic backgrounds and natural histories. Conventional (clear cell) carcinoma (70%) and papillary carcinoma (10–15%) account for the majority of RCCs. The remaining types include chromophobe carcinoma (5%), the benign tumor oncocytoma (5–10%), and rarer forms such as collecting duct carcinoma (<1%) and RCC unclassified (≤5%; Ref. 34). The 100 tumors studied here included all major cell types and are representative of grade and stage at presentation of kidney cancer. We examined the methylation status of 10 tumor suppressor or cancer genes selected on the basis of both the biological significance of the gene and the methylation associated with loss of expression being well described for the gene (16, 17, 23, 24, 25, 26, 27, 28, 29, 30).

The frequency of methylation of the genes examined varied from 58% for Timp-3 and 45% for RASSF1A to 7% for MGMT. The gene methylation frequencies observed in our study were broadly similar to previous reports on primary kidney tumors (4, 18) but lower compared with kidney tumor cell lines (17, 35). Previous studies had examined the methylation status of either a single gene only in kidney cancer or several genes in a smaller number of kidney tumors (4, 18, 36). Although the genes examined in our study have been reported to be unmethylated in normal tissue (22, 23, 24, 25, 26, 27, 28, 29, 30), we confirmed that the methylation was specific to cancer cells because our analysis of 15 nonmalignant renal or upper-tract tissue specimens found all 10 genes to be unmethylated (Ref. 31; data not shown). The finding of hypermethylation in kidney tumors of the lowest pathological stage (T_1a) and grade (I), as well as tumors <3 cm in diameter, indicates that suppressor gene methylation can be a relatively early event in kidney tumorigenesis. There is also evidence in cancers from other organ sites that methylation is an early event, e.g., in breast (37), colorectal (24), and lung tumorigenesis (38). Hypermethylation was found in all of the pathways examined including the p16/Rb and p53/p14 tumor suppressor pathways (39) and the Wnt signaling pathway (40) emphasizing the widespread role of epigenetic silencing in kidney tumorigenesis.

We also examined whether promoter methylation of the recently identified FH tumor suppressor gene (13), inherited mutation of which can predispose to papillary renal cancer, occurs in sporadic papillary renal tumors. We examined the FH promoter region in 15 papillary renal cell tumors by bisulfite sequencing and found all to be unmethylated. Point mutation of FH has also been reported to be absent or rare in sporadic papillary renal cell tumors (41). The p53 and PTEN tumor suppressor genes, known to be inactivated in kidney cancer by point mutation and deletion, have been found to be unmethylated in human cancer cells and were, therefore, not examined for promoter methylation (2). The Rb suppressor gene appears to be methylated at an appreciable frequency in retinoblastoma only (2, 42), and, because we have found no case of any urological tumor to be methylated (43), this gene was also not examined. The recently identified BHD gene has been reported to be methylated (44), but another study found no evidence of methylation in renal cancer (45).

The comparison of the clinicopathological data and methylation data revealed that hypermethylation of the VHL gene was specific for clear cell renal cancer, as expected (16). We also noted that hypermethylation of RASSF1A was significantly more frequent in papillary RCC compared with other cell types. RASSF1A methylation was also significantly associated with high-grade tumors. Although hypermethylation of p14^ARF or APC was more common in non-clear cell cancers, the difference in frequency was not statistically significant in the present sample size. Analysis of larger numbers of specimens will determine whether this tendency is significant. MGMT hypermethylation was more frequent in tumors from nonsmokers. A similar finding has been reported in lung tumors from nonsmokers (46). It is important to note that, because of the statistical analyses performed, some significant associations could potentially be type I errors. For example, there is no obvious biological rationale for differences in gene methylation frequencies between genders. The statistical tests to identify patient characteristics associated with hypermethylation were conducted without explicit multiple comparison correction (e.g., Bonferroni) to the significance level of the individual tests. That is, in order for the tests to have reasonable statistical power, no formal effort was made to control the family-wise type I error rate. Therefore, until further validation, the statistical data should be considered as preliminary.

We found no significant correlation between methylation and pathological stage, which is the most important determinant of survival for kidney cancer patients. Several profiles of tumors from other organ sites have also failed to display a clear association between gene methylation and tumor grade or stage (5, 6, 8, 9). Survival data were available for 86 patients with a median follow-up period of 29 months. Kaplan-Meier analysis indicated that there was no association between methylation of any gene and survival or time to recurrence. A more lengthy follow-up will likely be necessary to determine whether gene hypermethylation can predict survival. Another issue to be considered is that present knowledge provides, at best, a partial picture of gene methylation in cancer. It has been estimated that several hundred, as yet unidentified, genes are hypermethylated in human cancer (33, 47); therefore, many more genes remain to be discovered in kidney cancer. The majority of the known hypermethylated genes have been identified through a candidate gene approach (1, 2, 4). In the future, global analysis (48, 49, 50) will be of increasing importance in the identification of novel hypermethylated genes in kidney cancer.

We recently reported on the detection of tumor suppressor gene hypermethylation in urine DNA from kidney cancer patients (31), which provided further impetus to profile gene hypermethylation in kidney tumorigenesis. For a potential diagnostic test to have maximal utility, it should be able to detect all types of kidney cancer and provide as much information as possible. Simultaneous differential diagnosis and molecular prognosis might be possible with an optimal panel of hypermethylated genes.

Our profile has demonstrated that aberrant promoter hypermethylation of tumor suppressor and cancer genes is frequent, widespread in terms of cell type, and can occur relatively early and in many cancer pathways in kidney tumorigenesis. The associations of VHL, RASSF1A, and MGMT hypermethylation with cell type, grade, and stage, respectively, if confirmed in larger studies, outline the potential as markers for the prediction of patient clinical outcome and for simultaneous differential diagnosis and molecular prognosis in urine-based screening.