High-density methylation of p14ARF and p16INK4A in Epstein-Barr virus–associated gastric carcinoma
Article first published online: 16 JUN 2004
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 112, Issue 2, pages 273–278, 1 November 2004
How to Cite
Sakuma, K., Chong, J.-M., Sudo, M., Ushiku, T., Inoue, Y., Shibahara, J., Uozaki, H., Nagai, H. and Fukayama, M. (2004), High-density methylation of p14ARF and p16INK4A in Epstein-Barr virus–associated gastric carcinoma. Int. J. Cancer, 112: 273–278. doi: 10.1002/ijc.20420
- Issue published online: 17 AUG 2004
- Article first published online: 16 JUN 2004
- Manuscript Accepted: 15 APR 2004
- Manuscript Received: 5 JAN 2004
- Ministry of Education, Science, Sports, and Culture of Japan
- gastric carcinoma;
- Epstein-Barr virus;
Promoter hypermethylation of various tumor-related genes is extremely frequent in gastric carcinoma (GC) associated with Epstein-Barr virus (EBV). To investigate the significance of the promoter methylation in this type of GC, we examined the methylation densities of the promoter regions of p14ARF and p16INK4A in EBV-associated (n = 7) and EBV-negative (n = 14) GC. Bisulfite sequencing demonstrated a high frequency of concurrent methylation of p14ARF and p16INK4A promoter regions in EBVaGC. Methylation was observed in all 29 CpG sites of p14ARF and all 16 sites of p16INK4A with equally high densities. In EBV-negative GC, the methylation profiles differed between the 2 genes. Promoter methylation was sporadic and variable in p14ARF, and only the last position of CpG in p14ARF was methylated at high frequency. High-density methylation in p16INK4A was observed in a subset of GC, but the first position of CpG was never methylated in EBV-negative GC. These findings suggest the presence of mechanisms of de novo and maintenance methylation specific to EBVaGC that might be associated with EBV infection. © 2004 Wiley-Liss, Inc.
EBV is a human oncogenic virus that has been identified in a wide variety of malignancies.1, 2, 3 EBVaGC is the most common malignant neoplasm associated with EBV, comprising 10% of all GCs. This malignancy has several distinct clinicopathologic features, including male predominance, frequent involvement of the proximal stomach, histopathology of moderately differentiated tubular or poorly differentiated solid type and accompaniment of lymphocytic infiltration to various degrees.1, 2 Our previous investigation of the genetic changes in GC demonstrated that chromosomal alterations and microsatellite instabilities are rare in EBVaGC.4 As in Burkitt's lymphoma, viral gene expression in EBVaGC is restricted to a small number of latency genes (EBNA1, EBER1, BARF0 and LMP2A), and not EBNA2 and LMP1, the products of which are important for the immortalization of human lymphocytes and the transformation of rodent fibroblasts, respectively.5
These findings led us to the hypothesis that an epigenetic alteration such as aberrant DNA methylation6 may play a primary role in the genesis of EBVaGC. In studies using MSP,7 EBVaGC in vivo exhibited global and nonrandom DNA methylation of the promoter regions of various cancer-associated genes, including p16INK4A and E-cadherin.8, 9 This promoter methylation correlated well with the repression of p16INK4A and E-cadherin expression in EBVaGC, but no such correlation was observed in EBV-negative GC.10, 11 These findings suggest that the mechanisms of de novo or maintenance methylation differ quite markedly between GCs with and without the EBV association.
Following along the same lines, we further investigated the methylation densities of the promoter regions of p14ARF and p16INK4A. These 2 genes are quite unique. Both are encoded by the INK4A/ARF loci on chromosome 9p21 and share common exons, but each consists of a different exon 1s and is driven by a different set of promoters. The gene products have different homologies and play different inhibitory roles in the cell cycle. Specifically, the p16INK4A product inhibits G1–S progression by preventing phosphorylation of the RB1 protein through binding to CDK4,12 and the p14ARF product binds to MDM2 and stabilizes both p53 and MDM2.13, 14 Promoter hypermethylation represses both genes in various carcinomas, either independently or concurrently. Thus, the methylation profiles of the promoter regions of p16INK4A and p14ARF provide important insights into the mechanism of promoter methylation in EBVaGC.
MATERIAL AND METHODS
Patients and samples
Seven cases of EBVaGC and 14 cases of EBV-negative GC were previously identified and surgically resected at Jichi Medical School. Mean patient age (±SD) and male/female ratio were 50 ± 10 years and 5:2 in EBVaGC and 50 ± 10 years and 14:0 in EBV-negative GC. When histologic type was determined according to Lauren's classification,15 EBV-negative GC consisted of 8 intestinal and 6 diffuse subtypes. Fresh tissue from each GC was taken immediately after surgical resection, frozen in liquid nitrogen and stored at −80°C. Genomic DNA was extracted by a standard phenol/chloroform procedure. EBER-1 in situ hybridization was applied to formalin-fixed and paraffin-embedded specimens to determine the presence or absence of EBV.2, 4
All DNA samples underwent bisulfite sequencing to investigate the distribution and density of methylated CpG sites. Samples were subjected to bisulfite modification with the CpGenome DNA Modification Kit (Intergen, New York, NY). Forward and reverse primer sequences of p14ARF for bisulfite sequencing were 5′-GTGGGTTTTAGTTTGTAGTT-3′ and 5′-AAACCTTTCCTACCTAATCT-3′, respectively, and together they amplified a 309 bp product containing 29 CpG sites in the promoter/exon 1β (Fig. 1a). Bisulfite sequencing of p16INK4A was performed using the method of Sato et al.,16 with amplification of a 384 bp product containing 16 CpG sites (Fig. 1b). The target regions of both genes covered those of our previous studies.9, 10 PCR products were gel-purified (Gel Extraction Kit; Qiagen, Hilden, German) and ligated into the PCR 2.1-TOPO plasmid vector using the TA Cloning System (Invitrogen, Carlsbad, CA). Plasmid-transformed Escherichia coli (10–20 colonies) was cloned in each sample by overnight culture, and then each plasmid DNA was purified using the Miniprep Kit (Sigma-Aldrich, St. Louis, MO) and sequenced with M13 reverse primers using an ABI PRISM (R) 310 Genetic Analyzer (Applied Biosystems, Foster City, CA) by Big-Dye Terminator sequencing. Samples were only included in the analysis if more than 5 colonies could be successfully sequenced.
Statistical analysis was performed using the χ2 test or Wilcoxon test in n × n contingency tables. All tests were 2-sided, and the results were considered significant at p < 0.05.
|Methylated/total clones (%)||Number of methylated sites in methylated clones (mean ± SE)||Methylated/total clones (%)||Number of methylated sites in methylated clones (mean ± SE)||Distribution of methylated sites (<3, 4–13, ⩾14)|
|1||3/6 (50%)||27 ± 0||13/13 (100%)||8 ± 2||5, 3, 5|
|2||5/7 (71%)||27 ± 4||5/9 (56%)||10 ± 4||2, 0, 3|
|3||nd||13/14 (93%)||10 ± 2||5, 1, 7|
|4||10/11 (91%)||26 ± 2||10/14 (71%)||11 ± 2||3, 1, 6|
|5||5/7 (71%)||25 ± 2||6/6 (100%)||16 ± 0||0, 0, 6|
|6||8/11 (73%)||28 ± 1||11/17 (65%)||16 ± 0||0, 0, 11|
|7||5/6 (84%)||28 ± 1||8/10 (80%)||16 ± 0||0, 0, 8|
|1||0/5 (0%)||3/5 (60%)||14 ± 1||0, 1, 3|
|2||2/13 (15%)||2 ± 1||0/13 (0%)|
|3||0/6 (0%)||0/18 (0%)|
|4||5/17 (29%)||4 ± 2||12/19 (63%)||15 ± 0||0, 0, 12|
|8||4/9 (44%)||2 ± 1||15/15 (100%)||15 ± 0||0, 0, 15|
|9||1/7 (14%)||2||2/13 (15%)||2 ± 0||2, 0, 0|
|10||5/16 (31%)||16 ± 6||15/17 (88%)||15 ± 0||0, 0, 15|
|11||13/19 (68%)||5 ± 1||0/16 (0%)|
|12||6/9 (67%)||3 ± 1||nd|
|13||3/19 (16%)||2 ± 1||11/12 (92%)||15 ± 0||0, 0, 11|
|14||4/11 (36%)||3 ± 1||0/5 (0%)|
|Case||Age (years)||Gender||Location||Histology||Lymphocytic infiltration||Stage||Lymph node metastasis||Methylation frequency1 (density)2|
|1||45||M||Cardia||Intestinal||+(Grade 2)||Early||0||High (++)||High (+/++)|
|2||62||M||Body||Intestinal||+(Grade 2)||Advanced||0||High (++)||High (+/++)|
|3||75||M||Body||Intestinal||+(Grade 3)||Advanced||+||nd||High (+/++)|
|4||22||F||Body||Intestinal||+(Grade 3)||Advanced||+||High (++)||High (+/++)|
|5||60||M||Body||Diffuse||+(Grade 2)||Advanced||0||High (++)||High (++)|
|6||83||F||Body||Diffuse||+(Grade 2)||Advanced||0||High (++)||High (++)|
|7||50||M||Body||Diffuse||−(Grade 0)||Advanced||+||High (++)||High (++)|
|1||47||M||Body||Intestinal||−(Grade 1)||Early||0||−||High (++)|
|2||80||M||Body||Intestinal||−(Grade 0)||Early||0||Low (+)||−|
|4||62||M||Antrum||Intestinal||−(Grade 1)||Early||0||High (+)||High (++)|
|6||48||M||Body||Intestinal||−(Grade 1)||Advanced||0||Low (+)||nd|
|8||54||M||Body||Intestinal||−(Grade 0)||Advanced||+||High (+)||High (++)|
|9||55||M||Antrum||Diffuse||+(Grade 3)||Early||−0||Low (+)||Low (+)|
|10||64||M||Body||Diffuse||−(Grade 0)||Advanced||+||High (+)||High (++)|
|11||79||M||Antrum||Diffuse||−(Grade 0)||Advanced||+||High (+)||−|
|12||69||M||Antrum||Diffuse||−(Grade 1)||Advanced||+||High (+)||nd|
|13||62||M||Antrum||Diffuse||−(Grade 0)||Advanced||+||Low (+)||High (++)|
|14||56||−M||Antrum||Diffuse||−(Grade 0)||Advanced||+||High (+)||−|
Methylation status of CpG Sites in p14ARF
The methylation status of the 29 CpG sites of the p14ARF gene could be evaluated in 6 cases of EBVaGC and 12 cases of EBV-negative GC (Table I). In EBVaGC, 50% or more of the clones (3/6–10/11 clones, 50–91%, mean 73 ± 14%) showed methylation in each case, and the mean number of methylated sites in the methylated clones per case ranged 25.4–28.0. As each CpG site showed methylation in >75% of the methylated clones (36 of 48 examined clones), no specific site was thought to show resistance to methylation (Fig. 2a).
Methylation was observed in 10 of 12 cases of EBV-negative GC, and the frequency of methylated clones in each methylation-positive case was variable (1/7–13/19 clones, 14–68%, mean 34 ± 21%) (Table I). The mean number of methylated sites in the methylated clones ranged 1.0–16.0 per case. Methylation at the CpG sites (Fig. 2a) was variable, occurring at a rate of 0–89% in methylated clones (44 of 138 examined). The highest frequency was observed at the 28th and 29th sites, which showed 47% and 89% rates of methylation when methylation occurred, respectively.
Methylation status of CpG Sites in p16INK4A
The methylation status of p16INK4A was evaluated in 7 cases of EBVaGC and 12 cases of EBV-negative GC (Table I). As seen in the separate analyses for the p14ARF gene, methylation was observed in all cases of EBVaGC in which 50% or more of the clones (5/9–13/13 clones, 56–100%, mean 81 ± 18%) showed methylation. The 16 CpG sites of p16INK4A in the methylated clones (total 66 clones in 83 examined) showed bimodal distribution (Fig. 2b): 46 clones showed more than 14 methylated sites, whereas 15 showed fewer than 3 sites (sparsely methylated clone). Highly methylated clones were observed in all EBVaGCs, while sparsely methylated clones were observed in only 4 of 7 clones. Although the frequencies were somewhat lower than those in EBV-negative GC due to the presence of low-density clones, each CpG site showed methylation in >64% of the methylated clones. No specific site in the highly methylated clones showed resistance to methylation (Fig. 2b), and no hot spots were observed in the sparsely methylated clones.
Methylation was observed in 6 of 12 EBV-negative GC, and the frequency of methylation was >50% in all but one of the methylation-positive cases (2/13 clones, 15%). In the 5 cases with >50% methylation, the mean number of methylated sites was 15 in 56 clones showing methylation, and the first position of CpG was never methylated in EBV-negative GC (Fig. 2b). In the one methylation-positive case with <50% methylation, the 2 clones showed only 2 common methylation sites, at positions 13 and 15 CpG.
Promoter methylation of p14ARF and p16INK4A occurred concurrently in EBVaGC (Table II), and the density of methylation was high in both genes. In p16INK4A methylation, however, sparsely methylated clones were observed in a subset of EBVaGC, i.e., in carcinoma with intestinal type histology accompanied by dense lymphocytic infiltration.
Promoter methylation of p14ARF and p16INK4A was not concordant in EBV-negative GC. When GC cases were divided into 2 groups according to the frequency of p14ARF methylation (0–29% and ≥30%), methylation was somewhat more frequent in advanced carcinoma (5 of 7 cases) and in lymph node-positive cases (5 of 6 cases) than in early carcinoma (1 of 6, p = 0.1026) and lymph node-negative cases (1 of 6, p = 0.0801). When cases were compared based on the frequency of the promoter methylation of p16INK4A, no correlations with histology, lymphocytic infiltration, stage or lymph node metastasis were observed. However, the highly methylated cases were younger (mean 56 ±8, n = 5) than the others (mean 68 ± 10, n = 7) (p = 0.0344).
The EBVaGCs examined in the present study exhibited concurrent methylation of p14ARF and p16INK4A promoters with extremely high density. This finding was in contrast to that for EBV-negative GC, in which the methylation profile was discordant and different in both genes. Specifically, promoter methylation was sporadic and variable in p14ARF, whereas it occurred with high density in p16INK4A in a subset of GC. Findings on the target CpG of methylation also differed between the 2 GCs. While all 29 CpG sites of p14ARF were similarly methylated in EBVaGC, only the last position of CpG was commonly methylated in the methylated clones in EBV-negative GC. The first position of CpG in p16INK4A was never methylated in EBV-negative GC, although the other positions were commonly and ubiquitously methylated in both types of GC. These findings further strengthen our hypothesis that there are mechanisms of de novo and maintenance methylation characteristic of EBVaGC.
According to Lorincz et al.,17 a low level of methylation was not stable when a proviral reporter premethylated at different densities was introduced into a defined chromosomal site in murine erythroleukemic cells, while a high density of methylation was faithfully propagated in vivo. They also found that Dnmt3a/3b-independent de novo methyltransferase activity was stimulated by the presence of preexisting methylation. Since mRNA expression of Dnmt3a/3b is considerably low in EBVaGC,9 the presence of highly methylated DNA suggests that methylation independent of Dnmt3a/3b may provide de novo and maintenance functions in this type of GC. One of the important functions of DNA methylation is genome defense against foreign invaders in mammalian cells.18, 19 EBV infection may induce Dnmt3a/3b-independent methylation in infected cells, which would primarily methylate the viral promoter genes and suppress the viral latent genes.20, 21 Further studies are necessary to clarify the mechanisms of hypermethylation in EBVaGC, including the putative role of EBV gene products.
Widespread hypermethylation of CpG islands over the entire genome is also observed in NPC.22 Although deletion of chromosomes 3p and 9p appears to precede EBV infection and hypermethylation in NPC,23 the tumorigenesis pathway of EBVaGC has not been clarified. zur Hausen et al.24 demonstrated transcription of the BARF1 gene in EBVaGC, which has transforming and immortalizing capacities. There might also be upregulation of the molecules derived from infected cells, which has an autocrine growth function, such as IL-1β25 and IGF-I.26 Thus, it is interesting to clarify the significance of these mechanisms and their relationship with promoter hypermethylation along the development of EBVaGC. In one such attempt, we are now investigating promoter methylation profiles in GC cell lines with and without recombinant EBV infection.
In our examinations of p16INK4A methylation in EBVaGC, we found sparsely methylated clones in a subset of cases. It may be that most neoplastic cells show ubiquitous and high-density methylation, with some cells representing intermediate forms. However, the distribution of methylation density was bimodal, and the sparse methylation was more likely derived from some component other than the carcinomas, e.g., infiltrating lymphocytes. In an evaluation of the methylation phenotypes of gastric lymphomas of the MALT type, Kaneko et al.27 identified CIMP in Helicobacter pylori–dependent, but not in H. pylori–independent, cases. Clones of sporadic methylation in EBVaGC may have been derived from the mucosal lymphocytes induced to infiltrate the neoplastic cells containing the virus.
We also evaluated correlations of the methylation profile with pathologic features in the small number of EBV-negative GCs included in our study. GC patients with highly methylated p16INK4A were younger than those with negative or sparsely methylated p16INK4A, indicating the importance of dense methylation of p16INK4A in a subset of EBV-negative GCs. In our previous study, there was no correlation between age and methylation status of p16INK4A in EBV-negative carcinoma.10 Since the simple MSP method could not discriminate between high- and low-density methylation, the difference in the results may be due to the methods used in both studies. Nevertheless, according to MSP analyses of the promoter regions of various cancer-related genes, including p16INK4A, 25–40% of GCs are thought to show CIMP,28 but the clinicopathologic features of CIMP GC have not been precisely identified. We believe that the major component of CIMP GC is EBVaGC. On this basis, it will be crucial to identify EBV and clarify the significance of CIMP or promoter methylation in all types of GC.