Promoter hypermethylation of E-cadherin and its abnormal expression in Epstein-Barr virus-associated gastric carcinoma



Promoter hypermethylation of various tumor-related genes is extremely frequent in Epstein-Barr virus (EBV)-associated gastric carcinoma (EBVaGC). To investigate the significance of the promoter methylation in EBVaGC, we focused on one of the important proteins in the carcinogenesis of the stomach, E-cadherin. Methylation-specific PCR analysis (MSP) was applied to surgically resected gastric carcinomas, together with immunohistochemistry, PCR-based analysis of mutations and allelic loss, and site-specific MSP of E-cadherin gene. By MSP, nearly all of the carcinomas showed aberrant methylation of E-cadherin promoter in EBVaGC (21/22), and the frequency of this aberration was significantly higher than that in EBV-negative gastric carcinoma (GC; 45/81; p = 0.0003). According to immunohistochemistry of E-cadherin, the frequency of abnormal staining pattern in EBVaGC (87%) was comparable to that in the diffuse type (80%), but higher than that in the intestinal type of EBV-negative GC (47%). Promoter methylation was well correlated with abnormal staining pattern in EBVaGC, but not in EBV-negative GC. Neither mutation nor allelic loss of E-cadherin was observed in EBVaGC. Methylation status of E-cadherin within each carcinoma was heterogeneous as far as examined. Thus, in addition to the known association involving p16, we determined that promoter methylation-mediated silencing of E-cadherin gene was also closely associated with the development of EBVaGC, although it becomes heterogeneous within a given tumor along its progression. © 2003 Wiley-Liss, Inc.

Epstein-Barr virus (EBV)-associated gastric carcinoma (EBVaGC) is a unique type of gastric carcinoma (GC) that accounts for 5–18% of GCs reported around the world. EBV-encoded small RNA (EBER) is present in nearly all of the carcinoma cells in the intramucosal stage. EBV in EBVaGC is monoclonal by Southern blot hybridization analysis with probes adjacent to the unique terminal repeat of EBV DNA. EBVaGC also has some characteristic clinicopathologic features, such as male preference, predominant involvement of the proximal stomach, frequent accompaniment of atrophic gastritis, a moderately differentiated tubular or poorly differentiated solid type of histology1, 2, 3 and specific expression of the splice variants of CD44 and IL-1β.4, 5 Given these many distinctions, the carcinogenic process of EBVaGC is thought to be quite different from that of EBV-negative GC.

EBV nuclear antigen 2 (EBNA2) and latent membrane protein 1 (LMP1) are latency gene products of EBV. The former is capable of immortalizing human lymphocytes, the latter capable of transforming rodent fibroblasts. Since neither is expressed in EBVaGC,6, 7 genetic or epigenetic alterations of the infected cells might be directly responsible for the development of EBVaGC. In an investigation of the genetic changes, the deletion of 5q and/or 17p and the microsatellite instability were found to be extremely rare in EBVaGC, but very frequent in EBV-negative GC.8 On the other hand, Kang et al.9 and our own group10 recently demonstrated that promoter hypermethylation of various tumor-related genes occurs much more frequently in EBVaGC than in EBV-negative GC. The subsequent reduction of gene expression has been observed in at least one tumor suppressor gene, p16.11

In the present study, we focused on E-cadherin, another protein important in the carcinogenesis of the stomach. E-cadherin is a Ca2+-dependent cell-cell adhesion molecule that plays an essential role in the formation and maintenance of the normal architecture and function of epithelial tissues.12, 13, 14 Abnormalities of the gene and gene expression of E-cadherin have been frequently observed in gastric carcinoma,15, 16, 17, 18, 19, 20 and the germline mutation was identified in the hereditary diffuse gastric carcinoma kindred.21 However, few studies have correlated abnormalities of E-cadherin with the EBV infection.22 In the present study, to clarify the significance of the promoter hypermethylation of E-cadherin in EBVaGC, we demonstrated its clinicopathologic features and its characteristic gene expression in GC with and without EBV infection, together with the specific analyses of E-cadherin gene in EBVaGC, such as genetic abnormalities, and regional heterogeneity of methylation status.


EBER, Epstein-Barr virus-encoded small RNA; EBV, Epstein-Barr virus; EBVaGC, Epstein-Barr virus-associated gastric carcinoma; GC, gastric carcinoma; LOH, loss of heterozygosity; MSP, methylation-specific polymerase chain reaction; PCR-SSCP, polymerase chain reaction-single-strand conformational polymorphism.


The materials consisted of 103 gastric carcinomas that had been surgically resected at Jichi Medical School or Tokyo Metropolitan Komagome Hospital from 1988 to 1998. Fresh tissues of gastric carcinoma were frozen in liquid nitrogen immediately after surgical resection and stored at −80°C until use. Histologic typing of the carcinoma and the pathology of the resected stomachs were evaluated according to the Japanese Classification of Gastric Carcinoma.23 We also adopted Lauren's classification of gastric carcinoma,24 intestinal and diffuse.

To determine the presence or absence of EBV, EBER in situ hybridization was applied to the formalin-fixed and paraffin-embedded sections as described previously.3 Genomic DNA was isolated from the frozen tissues by a standard phenol/chloroform procedure. Due to limited sample size and relatively long duration of the tissue storage, DNA, but not RNA, could be used for the mutation analysis.

Methylation-specific polymerase chain reaction of E-cadherin gene promoter

Bisulfite modification of genomic DNA was performed25 with the CpGenome DNA modification kit (Intergen, Purchase, NY) before the methylation-specific polymerase chain reaction (MSP) of the E-cadherin promoter. MSP was performed with the CpG WIZ E-cadherin Amplification kits (Intergen) according to the manufacturer's recommendation with slight modifications.26 Reaction buffer at a final volume of 12.5 μl contained 1 unit of AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA) and 1 μl of modified DNA. The temperature profiles for the amplification were as follows: initial heating at 95°C for 10 min, 40 cycles of denaturation at 95°C for 45 sec, annealing at 60°C for 45 sec and extension at 72°C for 1 min, followed by a final extension at 72°C for 10 min. A pair of positive (universal methylated and unmethylated DNA; Intergen) and negative controls (distilled water) accompanied every amplification reaction; 6 μl of each PCR product was electrophoresed in a 2% agarose gel, stained with ethidium bromide and visualized under an ultraviolet illuminator. All experiments were performed in duplicate.

Immunohistochemistry of E-cadherin

Since the high incidence of E-cadherin promoter hypermethylation was observed in EBVaGC, immunohistochemistry of E-cadherin was further performed to evaluate the expression of E-cadherin. For the further analysis of immunohistochemistry, the formalin-fixed and paraffin-embedded sections were available for EBVaGC when the size of the carcinoma was more than 1 cm (15 cases). For EBV-negative GCs, the specimens of Jichi Medical School were subjected to immunohistochemistry (19 intestinal and 15 diffuse type, respectively). Sections 4 μm thick were cut from formalin-fixed and paraffin-embedded specimens, deparaffinized in xylene and rehydrated in alcohol. The sections were autoclaved in 0.01 M citrate-phosphate buffer (pH 6.0) with 0.1% of Tween20 at 121°C for 10 min for antigen retrieval. The monoclonal antibody to human E-cadherin (clone HECD-1, TaKaRa, Shiga, Japan) was applied to the sections and incubated at room temperature for 1 hr. Following blockage of endogenous peroxidase activity by treatment with 0.3% hydrogen peroxide in methanol for 10 min, a standard avidin-biotin immunoperoxidase technique was used for visualization of the reactive product.27, 28 The sections were incubated with avidin-biotin complex (Vectorstain ABC kit, Vector Laboratories, Burlingame, CA), developed in 3′-3′ diaminobenzidine, counterstained with Mayer's hematoxylin for 5 min and dehydrated in alcohol prior to mounting. The epithelium of noncancerous parts within each section (if available) and/or normal colonic epithelium served as positive controls. To obtain negative controls, the primary antibodies were omitted.

For the evaluation of the immunohistochemical results, we classified the staining patterns into 3 categories according to classification recommended by Oka et al.17 Tumors showing positive signal in more than 90% of tumor cells were classified as the normal type, those showing positive signal in 10–90% of tumor cells were classified as the heterogeneous type and those expressing very weak or no signal (less than 10%) were classified as the reduced type. Both reduced and heterogeneous types were considered abnormal.

PCR-single-strand conformational polymorphism analysis and loss of heterozygosity analysis of E-cadherin gene

Due to the amount of DNA available, only 11 out of the 15 EBVaGC cases examined in the immunohistochemical study were further subjected to PCR-single-strand conformational polymorphism (SSCP) and loss of heterozygosity (LOH) analyses. Exons 6–9 of E-cadherin gene were evaluated by PCR-SSCP analysis with subsequent sequencing of the abnormal bands. Exons 8 and 9 of the E-cadherin gene have been identified as the region of the mutation hot spot in diffuse-type gastric carcinoma.29 PCR-SSCP analysis was performed according to the established protocols.20, 30 Electrophoresis was performed on 6% neutral polyacrylamide gels at 15 mA/1,500 V for 4 hr. The gels were dried and exposed to X-ray film at −80°C overnight. All experiments were performed in duplicate. MKN45, a cell line that harbors a mutation on exon 6, was used as a mutation-positive control.30 When mobility shift SSCP bands were present, they were isolated from the gels and subjected to PCR using the same primer sets as those used in the prior PCR. The PCR products were cloned into the vector pCR II-TOPO (Invitrogen, Leek, The Netherlands). Sequencing was performed using a DYEnamic ET terminator cycle sequencing kit (Amersham Pharmacia Biotech, Uppsala, Sweden) with a DNA sequencer (Model 373A, Applied Biosystems).

The LOH study was performed using the tumor DNA and the matched DNA from the nonneoplastic mucosa. Radioactive PCR amplification of microsatellite markers D16S265 and D16S301 was performed according to the procedures described by Machado et al.19 PCR products were run in a 6% denaturing polyacrylamide gel with 5% crosslinking and exposed to X-ray film at room temperature overnight.

Regional heterogeneity of promoter hypermethylation of E-cadherin

Regional heterogeneity of promoter hypermethylation of E-cadherin was assessed in EBVaGC by the MSP analysis of the microdissected tumor tissue from the formalin-fixed, paraffin-embedded section. Two serial sections, 8 μm in thickness, were deparaffinized in xylene and rehydrated in alcohol. After the sections were briefly stained with hematoxyline, a 5 × 5 mm sized region was dissected from each with an 18 G needle tip. The tissue samples were treated with proteinase K (1 mg/ml) at 37°C overnight. After phenol-chloroform-isoamyl alcohol extraction and ethanol precipitation, each 1 μg of DNA was subjected to bisulfite modification as described above. Since the DNA was obtained from the formalin-fixed and paraffin-embedded sections, different primer sets were used to obtain shorter PCR products according the method of Herman et al.25 The temperature profiles for the amplification were as follows: initial heating at 95°C for 5 min, 45 cycles of denaturation at 95°C for 45 sec, annealing at 57.0°C for the methylated primer set and at 53.0°C for the unmethylated primer set and extension at 72°C for 45 sec, followed by a final extension at 72°C for 10 min.

Statistical analysis

Statistical analysis of the results was performed using the chi-square test or Fisher's exact test. Differences were considered to be significant at p < 0.05.


Promoter methylation of E-cadherin

Twenty-two cases of EBVaGC and 81 cases of EBV-negative GC were included in this study. The only significant difference between the 2 groups was the tumor location (cardia, gastric body in total cases: 22/22 in EBVaGC and 53/81 in EBV-negative GC).

The results of MSP analysis (Fig. 1) are presented in Table I. Nearly all of the carcinomas showed aberrant methylation of E-cadherin promoter in EBVaGC, and the frequency of this aberrant methylation was significantly higher in EBVaGC than in EBV-negative GC (p = 0.0003). The diffuse type of histology appeared to be correlated with the E-cadherin hypermethylation, albeit not to a statistically significant extent (p = 0.0502). Since 2/3 (15 of 22) of EBVaGC cases showed the diffuse type of histology, EBVaGC contributed to the relatively high frequency of aberrant methylation of E-cadherin in the diffuse type. Otherwise, the frequency of promoter hypermethylation did not differ significantly between any of the other groups classified by clinicopathologic factors such as age, gender, or location, depth of invasion and lymph node metastasis of the carcinoma.

Figure 1.

MSP analysis of promoter hypermethylation of E-cadherin gene in EBV-associated and EBV-negative gastric carcinoma. Representative result of MSP of 4 EBVaGC cases (J1, J49, J84 and J183) and 4 EBV-negative GC (EBVnGC; J3, J7, J55 and J148). All 4 EBVaGC cases and 1 EBV-negative GC case (case J7) were considered methylation-positive. U, unmethylated primer set; M, methylated primer set; V, DNA marker V; NC, negative control; PC, positive control.

Table I. Correlation Between Promoter Hypermethylation of E-Cadherin and Clinicopathologic Factors in Gastric Carcinoma
Clinicopathologic factorsPromoter methylation statusp
  • a

    EBVnGC; EBV-negative GC.

  • b

    Location of tumor; upper/middle/lower third portion of the stomach.

  • c

    Tumor invasion within submucosa.

  • d

    Tumor invasion beyond muscularis propria.

EBV infection   
Mean age61.6 ± 14.064.9 ± 11.5NS
Gender (male:female)48:1832:5NS
 Upper and middle5124NS
Depth of invasion   
Lymph node metastasis   
Lymphatic invasion   
Venous invasion   

Since E-cadherin is reported to relate to the histologic subtype of gastric carcinoma, EBV-negative GC was further divided into diffuse and intestinal types (Table II). The frequencies of aberrant methylation in these 2 subtypes of EBV-negative GC did not differ, and both were lower than the frequency in EBVaGC (p = 0.0113 and < 0.0001, respectively). However, the significance of the E-cadherin methylation appeared to be different in each subtype. When early and advanced stages were compared, the frequency of hypermethylation in the advanced stage (59%) was relatively lower than that in the early stage (100%) in the diffuse type of EBV-negative GC (p = 0.0743). On the other hand, the frequency of promoter hypermethylation increased as the tumor progressed from early (29%) to advanced carcinoma (50%) in the intestinal type of EBV-negative GC, albeit not to a statistically significant extent. The presence of lymph node metastasis was also found to be significantly correlated with promoter hypermethylation in this subtype of EBV-negative GC (p = 0.0038).

Table II. Pathologic Significance of Promoter Hypermethylation of E-Cadherin in EBV-Associated Gastric Carcinoma, and Diffuse and Intestinal Subtypes of EBV-Negative Gastric Carcinoma
 Frequencies of promoter hypermethylation
TotalDepth of invasionaLymph node metastasis
  • a

    Early/advanced, tumor invasion within submucosa/tumor invasion beyond muscularis propria.

  • b

    EBV-negative GC showing diffuse type of histology.

  • c

    EBV-negative GC showing intestinal type of histology.

EBVaGC95% (21/22)100% (7/7)94% (14/15)NS93% (14/15)100% (7/7)NS
EBVnGC diffuseb65% (26/40)100% (6/6)59% (20/34)0.074370% (7/10)63% (19/30)NS
EBVnGC intestinalc46% (19/41)29% (2/7)50% (17/34)NS14% (2/14)63% (17/27)0.0038

Immunohistochemistry of E-cadherin and its correlation with promoter methylation

Next, we investigated the expression of E-cadherin by immunohistochemistry in order to evaluate the correlation with promoter methylation. Both heterogeneous and reduced staining patterns were considered abnormal pattern (Fig. 2). Abnormal staining pattern was observed in 13 of 15 EBVaGC cases, 12 of 15 diffuse-type EBV-negative GC cases and 9 of 19 intestinal-type EBV-negative GC cases (Fig. 3). When the relationship between methylation status and immunohistochemical staining was analyzed in each subgroup, a significant correlation was noted only in the EBVaGC cases. On the other hand, significant number of the unmethylated cases showed abnormal staining pattern in the diffuse and intestinal types of EBV-negative GC cases.

Figure 2.

Immunohistochemical staining pattern of E-cadherin in EBV-associated and EBV-negative gastric carcinoma. (a) Normal expression of E-cadherin in normal colonic mucosa. (b) Example of normal staining pattern in intestinal-type EBV-negative GC. (c) Heterogeneous staining pattern in EBVaGC. (d) Reduced staining pattern in diffuse-type EBV-negative GC. Original magnification × 400 for all panels.

Figure 3.

Promoter methylation status and immunohistochemical staining patterns of E-cadherin in EBV-associated gastric carcinoma and diffuse and intestinal subtypes of EBV-negative gastric carcinoma. Twelve of 15 EBVaGC cases showed a heterogeneous staining pattern with promoter methylation, whereas this correlation was not observed in EBV-negative GC (EBVnGC). Methyl, methylated promoter; Unmethyl, unmethylated promoter; N, normal staining pattern; H, heterogeneous staining pattern; R, reduced staining pattern

Mutation and LOH of E-cadherin gene in EBVaGC

To exclude a possible contribution of genetic changes of E-cadherin in EBVaGC, PCR-SSCP was performed, and no mutation was observed in EBVaGC (Table III). MKN45 has 18 bp deletion at the exon 6-intron 6 boundary. At the experimental condition in the present study, the mutation of MKN45 was detected even when the percentage of MKN45 DNA was 10% in total DNA when mixed with DNA of nonneoplastic tissue. Similarly, LOH of E-cadherin was not observed in any of 11 cases examined (Table III).

Table III. Evaluation of Mutation, LOH and Regional Heterogeneity of Promoter Hypermethylation of E-Cadherin in EBV-Associated Gastric Carcinomas
 CaseAgeGenderHistologyDepthaMutationbLOHcMethylation in frozen sectiondMethyle/samplef
  • a

    Early/advanced, tumor invasion within submucosa/tumor invasion beyond muscularis propria.

  • b

    Result of mutation analysis by PCR-SSCP; −; no mutation was detected; ND, not done.

  • c

    −, no LOH was detected.

  • d

    +, methylated; −, unmethylated.

  • e

    Number of samples that showed promoter hypermethylation.

  • f

    Number of samples obtained from one tumor by microdissection.

  • g

    MSP was not successful.


Regional heterogeneity of promoter hypermethylation of E-cadherin in EBVaGC

To evaluate the regional heterogeneity of promoter hypermethylation of E-cadherin in EBVaGC, several samples were dissected from a section of each EBVaGC tumor (Table III). In 13 of 15 cases, the methylation status was successfully evaluated using the tissue samples of formalin-fixed and paraffin-embedded specimens (Table III). One of the cases evaluated did not show promoter methylation in the first evaluation using DNA derived from frozen tissue, but all of the cases showed methylation at least in one of the formalin-fixed samples. Heterogeneity of methylation status within the carcinoma was observed in all cases except 1 of 3 early carcinomas.


Epigenetic silencing of the gene expression by CpG island hypermethylation has been shown to play an important role in the carcinogenesis of various organs, including the stomach.31, 32, 33 In recent investigations of EBVaGC cases, Kang et al.9 and our own group10 observed high frequencies of promoter methylation in various cancer-related genes such as p14, p15, p16, DAPK, GSTP, TIMP-3 and E-cadherin, but not in hMLH1 and MGMT. This finding suggests that a global but selective DNA hypermethylation of the cellular genes may occur in the EBV-infected epithelial cells. In the present study focusing on the E-cadherin protein of the stomach, we confirmed on a larger scale that nearly all of the carcinoma tissues of EBVaGC showed CpG island methylation of E-cadherin. Furthermore, we demonstrated that abnormal expression occurred in nearly all of EBVaGC cases, while such a correlation was not strictly observed in EBV-negative GC. Since neither mutation nor LOH was observed in EBVaGC, epigenetic silencing is a major mechanism of E-cadherin dysfunction in EBVaGC.

Most of the cases with hypermethylation in the present study showed a heterogeneous staining pattern rather than the reduced type in EBVaGC. This was consistent with our findings from earlier immunohistochemistry studies on p16 showing a marked decrease of p16 expression rather than a total loss of expression. Current studies on gene regulation by promoter methylation fit well with these findings. In cell lines from the human bladder, colon cancers and melanoma, Bender et al.34 found that the expression of p16 is highly affected by the frequency of methylation of the p16 promoter, but not by the presence of the methylation itself. Our study of the microdissection confirmed heterogeneity in the methylation status within each tumor. There are 2 explanations for this heterogeneity: hypermethylation occurs in a subpopulation of the carcinoma. Alternatively, the maintenance of hypermethylation is considerably disturbed in the progression of the carcinoma,35, 36 even though hypermethylation occurs in a uniform manner at early stage of cancer development. Methylation was observed in nearly all of the cases of EBVaGC, and the case showing methylation in all of the samples was the early gastric carcinoma in the present study. Thus, the latter possibility would constitute the most likely mechanism in EBVaGC.

As for the discrepancy between promoter methylation and gene expression in EBV-negative GC, regional heterogeneity of promoter methylation might exert in the cases of positive methylation but normal expression. On the other hand, in the cases of negative methylation but abnormal expression, at least 3 mechanisms37 are possible: gene mutation,18, 30 posttranslational truncation or modification38 and transcriptional repressor.38, 39, 40, 41 Thus, other than EBVaGC, further studies are necessary to disclose the regulation of E-cadherin gene expression in GC.

As an additional result, the present study provides some detailed statistical insights into the correlation between promoter methylation, gene expression/localization, tumor stages and tumor type in EBV-negative GC. In the intestinal-type EBV-negative GC, promoter hypermethylation of E-cadherin was correlated with both lymph node metastasis and invasion beyond the muscular layer. Since promoter methylation was not faithfully reflected in overall gene expression, promoter-methylated subpopulation within the tumor may play a primary role in the progression of this type of gastric carcinoma. On the other hand, the frequency of hypermethylation decreased from 100% to 59% as the carcinomas progressed from early to advanced in diffuse type. The maintenance of hypermethylation may be considerably disturbed in the progression of the carcinoma of the diffuse type of EBV-negative GC, as in the case of EBVaGC.

In conclusion, in addition to p16, the abnormality of E-cadherin expression caused by the aberrant methylation of E-cadherin gene promoter is closely associated with the development of EBVaGC. As the significance of aberrant methylation of E-cadherin differs between EBVaGC and EBV-negative GC, we can conclude that the recognition of this type of gastric carcinoma is critical for the evaluation of carcinogenesis of the stomach.