Hypermethylation of the p14^ARF Gene in Ulcerative Colitis-associated Colorectal Carcinogenesis¹ Free

UC³ is a chronic disease characterized by inflammation of the mucosa and submucosa of the large intestine. The duration and extent to which a patient suffers from UC correlate directly with an increased propensity to develop colorectal carcinoma (1, 2). For patients who have had UC for >20 years, the incidence of colorectal cancer is 10–20-fold greater than that of the general population, and the average age of onset is 20 years earlier (3). UC-associated colorectal carcinoma is different from sporadic carcinoma; unlike sporadic colorectal carcinoma, which arises from adenomatous polyps, UC-associated colorectal carcinoma progresses from areas of dysplastic mucosa. Although the molecular events that facilitate the progression of adenoma to carcinoma in sporadic colorectal cancer have been well investigated (4), much remains to be learned regarding molecular events underlying the progression of UC mucosa to dysplasia and carcinoma.

The p53 tumor suppressor gene is frequently inactivated in both sporadic and UC-associated colorectal carcinoma (4, 5, 6). Although point mutation and loss of heterozygosity are the most commonly reported mechanisms resulting in p53 inactivation, other genetic and epigenetic factors have been shown to modify p53 activity as well. Amplification of the MDM-2 gene (the protein product of which tags p53 for degradation) and expression of viral oncoproteins (which sequester p53) are two such examples (7).

Recently, p14^ARF has been ascribed a role in modulating the cellular amounts of p53 through direct interaction with MDM-2. MDM-2, which tags p53 for degradation through the ubiquitin/proteosome pathway, is inhibited by p14^ARF (8). Homozygous deletion of the p14^ARF locus has been reported in a variety of cancers (9, 10, 11, 12, 13), and gene knockout of p14^ARF correlates with tumorigenesis (14). In addition to mutation, the p14^ARF gene can also be epigenetically inactivated through hypermethylation of its normally unmethylated CpG island. Esteller et al. (15) demonstrated that p14^ARF hypermethylation occurs frequently in sporadic colorectal cancer. To determine whether p14^ARF hypermethylation occurs during the progression of UC mucosa to carcinoma, the frequency and timing of p14^ARF hypermethylation were investigated in clinical samples ranging from nonneoplastic UC mucosa to colorectal carcinoma.

Matching normal and tumor tissues were obtained at the time of surgical resection from 40 patients with one or more UC-associated colorectal neoplasms. The UC-associated neoplasms consisted of 38 adenocarcinomas and 12 dysplasias. Normal control samples consisted of 30 ileal mucosae, 2 smooth muscle tissues, and 3 nonmetastatic lymph nodes. All tissues were grossly dissected free of normal surrounding tissue, and parallel sections were used for histological characterization. Microdissection was not performed; however, the tissues were selected to include only those tumors containing 70% or more tumor cells by H&E staining.

Genomic normal and tumor DNAs were extracted using published protocols (16, 17).

DNA methylation of p14^ARF was determined by MSP (18). MSP distinguishes ummethylated alleles of a given gene based on DNA sequence alterations after bisulfite treatment of DNA, which converts unmethylated but not methylated cytosines to uracils. Subsequent PCR using primers specific to sequences corresponding to either methylated or unmethylated DNA sequences is then performed. We slightly shifted the MSP primer location used by Esteller et al. (Ref. 15; Fig. 1). Primers for MSP were designed using the GenBank L41934 sequence for p14^ARF. Primer sequences of p14^ARF for the unmethylated reaction were forward (5′-GTTTTTGGTGATTTTTTGGATTTGGT-3′) and reverse (5′-TACCCACTCCCCCATAAACCACAA-3′), which amplify a 94-bp product. Primer sequences for the methylated reaction were forward (5′-GGTGATTTTTCGGATTCGGC-3′) and reverse (5′-CACTCCCCCGTAAACCGCGA-3′), which amplify an 84-bp product. Briefly, 2 μg of genomic DNA were denatured by treatment with NaOH and modified by sodium bisulfite. DNA samples were purified using Wizard DNA purification resin (Promega Corp., Madison, WI), treated with NaOH, precipitated with ethanol, and resuspended in 20 μl of water. Two μl of modified DNAs were PCR amplified in a total volume of 50 μl. The annealing temperature for both unmethylated and methylated reactions was 58°C. PCR was performed in a thermal cycler (Biometra T-Gradient Thermoblock, Goettingen, Germany) for 35 cycles consisting of denaturation at 95°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 30 s, followed by a final 4-min extension at 72°C for all primer sets. PCR products were then loaded directly onto a nondenaturing 6% polyacrylamide gel, stained with ethidium bromide, and visualized under UV illumination. CpGenome Universal Methylated DNA (Intergen, NY) was used as a positive control for methylated alleles. DNA from normal lymphocytes was used as a negative control for methylated genes. The sensitivity of this MSP was established by using totally methylated, positive control DNA serially diluted by normal lymphocyte DNA (18). MSPs with 1:10, 1:100, and 1:1000 diluted positive control DNA produced detectable methylated bands (data not shown).

To verify MSP data and investigate the distribution and density of methylated CpG sites, bisulfite sequencing was performed in 20 samples from eight patients with a range of MSP results. Primer sequences of p14^ARF for bisulfite sequencing were forward (5′-GTTGTTTATTTTTGGTGTTA-3′) and reverse (5′-ACCTTTCCTACCTAATCTTC-3′), which amplify a 272-bp product (Fig. 1). This area contains 28 CpG sites, 6 of which were detected by the MSP primers (forward, 9th–11th sites; reverse, 19th–21th sites). Amplified PCR products were gel purified (Gel Extraction kit; Qiagen) and ligated into the pCR4-TOPO plasmid vector using the TA-cloning system (InVitrogen). Plasmid-transformed Escherichia coli were cultured, and plasmid DNA was isolated (QIAprep 96; Qiagen). Purified plasmid DNA containing the p14^ARF sequence was sequenced using an ABI 377 automated sequencer with BigDye Terminator chemistry and the M13 reverse primer. Samples that had clones with >50% methylation of CpGs were designated as partially methylated. Samples containing clones with methylation at all 6 MSP-recognition CpGs sites were deemed methylation positive. All other samples were designated as being methylation negative.

A proportional analysis between the sample groups was performed using Fisher’s exact probability test in 2 × 2 contingency tables and with the χ² or Kruskall-Wallis tests in n × n contingency tables. All tests were two sided, and the results were considered to be significant when P < 0.05.

DNAs obtained from 40 patients were analyzed by MSP. p14^ARF hypermethylation was present in 19 of 38 (50%) adenocarcinomas, 4 of 12 (33%) dysplasias, and 3 of 5 (60%) nonneoplastic UC mucosae. Conversely, 3 of 40 (7.5%) normal tissues showed p14^ARF hypermethylation (χ² test: P = 0.0003). No other correlations were found between hypermethylation status of p14^ARF and clinicopathological features such as age, gender, histology, anatomical location of sample (right versus left colon), or Dukes’ stage. All three methylated bands obtained from normal tissue were faint. Examples are shown in Fig. 2.

Two hundred thirty-two clones from 20 samples (>10 clones from each sample) were investigated using bisulfite sequencing. Bisulfite sequencing showed 3 negative, 2 partial, and 3 positive results among 8 MSP-positive samples, whereas all MSP-negative samples were sequence negative (Table 1). As shown in Table 2, no densely methylated clones were found in normal samples. An example is shown in Fig. 3. Partially methylated clones were observed in dysplastic samples, and densely methylated clones were detected only in carcinoma samples (Kruskall-Wallis test: P = 0.0005) In 4 samples, we detected 12 clones with methylation at all 6 MSP primer-recognized CpGs. The number of methylated CpGs in these 12 clones ranged from 20 to 28 (mean, 25.3) of 28 CpGs. No particular pattern could be observed in the distribution of methylated CpGs in different samples. We confirmed that the methylated and unmethylated control DNAs used in MSP were totally methylated and unmethylated, respectively, by bisulfite sequencing.

The purpose of this study was to determine the frequency and timing of p14^ARF hypermethylation in UC-associated colorectal carcinogenesis. In this study, hypermethylation of p14^ARF was present in 19 of 38 (50%) carcinomas and in 4 of 12 (33%) dysplasias. Esteller et al. (15) and Burri et al. (19) reported that the frequency of p14^ARF hypermethylation is 28–33% in sporadic colorectal cancers and 32% in colon adenomas. In comparison with these data, our hypermethylation frequencies are somewhat higher.

Although several groups have reported genetic abnormalities in nonneoplastic UC mucosa, such as microsatellite instability (20, 21) and p53 mutation (22, 23, 24, 25), hypermethylation of cancer-related genes has not been reported in this setting. Hypermethylation of tumor suppressor genes has been demonstrated in other precancerous lesions associated with chronic inflammation. Hypermethylation of E-cadherin has been reported in chronic gastritis (26), and hypermethylation of APC, p16^INK2a, and E-cadherinhas been demonstrated in the esophagitis-associated lesion, Barrett’s esophagus (27). Previously, Esteller et al. (15) found no evidence of p14^ARF hypermethylation in normal uninflamed colonic mucosa; however, we have noted p14^ARF hypermethylation in 3 (60%) of 5 (60%) nonneoplastic inflamed UC mucosae. Therefore, it is possible that longstanding chronic inflammation induces aberrant methylation of the p14^ARF gene in colonic mucosa. It is also possible that p14^ARF methylation varies according to location within the intestinal tract (e.g., absent in the ileum but present in the colon). However, our study suggests that p14^ARF methylation is a potential marker of early carcinogenesis in the setting of UC. In this regard, p14^ARF may resemble the p53 tumor suppressor gene, in that alterations of p53 are regarded as relatively early events in UC-associated carcinogenesis (5, 22, 23, 24, 25).

In this study, 3 of 20 samples studied by bisulfite sequencing showed results different from MSP. Because our samples had varying degrees of contamination by normal and stromal cells, our observed ratios of methylated CpGs could underrepresent actual methylation ratios in tumor cells. However, according to our bisulfite sequencing data, the density of methylated CpGs increased during the progression from dysplasia to carcinoma. These findings suggest that aberrant methylation of the p14^ARF gene may be progressive during the dysplasia-carcinoma sequence of UC.

Regarding the distribution pattern of methylated CpGs, no significant pattern was found in this study. Zheng et al. (28) described that the 3′ region of exon 1β was more densely methylated than the promoter and 5′ region of exon 1β. However, their primers for bisulfite sequencing were situated on several CpGs, and the forward primer sequence appeared to be based on the unmethylated sequence, whereas the reverse primer was based on the methylated sequence. Primers for bisulfite sequencing are generally placed in regions lacking the sequence CpG.

In conclusion, our data suggest that hypermethylation of p14^ARF is a relatively common and early event in UC-associated carcinogenesis. Thus, p14^ARF is worthy of study to explore its potential as a biomarker for the early detection of cancer or dysplasia in UC. Furthermore, future analyses of p14^ARF methylation in other organs should focus not only on frank cancers but on other premalignant lesions as well.