Introduction

Helicobacter pylori (H. pylori) has been designated as a human class I carcinogen of gastric malignancy by the International Agency for Research on Cancer [16]. Chronic infection with H. pylori generally increases the risk for gastric carcinoma by five- to six-fold [28]. The risk has been reported to be even greater when exposure to H. pylori infection is estimated based on the presence of serum antibodies to cytotoxin-associated gene product A [11]. Animal models have been developed to demonstrate a close association between H. pylori infection and gastric carcinoma [31, 35]. Although the exact mechanism of H. pylori-associated gastric carcinogenesis is unknown, long-standing bacterial infection, perpetuated chronic inflammation, and sustained mucosal epithelial cell proliferation are thought to produce a carcinogenic environment. Recently, H. pylori infection was reported to be associated with enhanced hypermethylation of multiple CpG islands (CGIs) in chronic gastritis, and its eradication brings about reversal of the methylation status of multiple CGI loci [5, 6, 24, 25, 37]. This fact raises the possibility that aberrant methylation changes induced by H. pylori infection may contribute to H. pylori infection-associated gastric carcinogenesis. To date, however, only a limited number of CGI loci have been studied for their hypermethylation in association with H. pylori infection.

Aberrant DNA methylation in cancer cells is characterized by focal CGI hypermethylation and generalized genomic hypomethylation. Global hypomethylation contributes to chromosomal instability, as evidenced by recent cell line studies that induced genomic hypomethylation by knocking down DNA methyltransferases and demonstrated aberrant chromosomal structure or number in the cells [20]. It is known that the methylated DNA content of stomach-lining cells in chronic gastritis decreases with progression of the lesion along a multi-step progression to gastric carcinoma [8]. However, little is known about whether H. pylori infection leads to genomic hypomethylation.

Polycomb repressive complexes (PRCs) are involved in the initiation of gene repression through epigenetic modification of chromatin structure and three components, namely, enhancer of zeste homolog 2 (Ezh2), suppressor of zeste 12 homolog (Suz12) and embryonic ectodermal development (Eed), forms the core of PRC2 [34]. Ezh2 of PRC2 catalyzes histone H3 lysine 27 (H3K27) trimethylation, which is thought to provide a binding surface for PRC1 [3], which in turn facilitates a closed chromatin structure to repress gene expression. Recent studies have demonstrated that de novo hypermethylation of some promoter CGIs in human cancers may occur in a so-called “instructive manner”, led by the chromatin “marks” of H3K27 trimethylation or binding of Polycomb group proteins Suz12 and Eed [29, 32, 43]. CGI loci carrying these chromatin marks in embryonic stem (ES) cells or normal tissue cells where these same loci are unmethylated may be preferentially targeted by aberrant DNA hypermethylation that occurs in association with cancer development [29, 32, 43]. In light of these facts, it can be speculated that in gastric epithelial stem or progenitor cells, CGI loci with these repressive chromatin marks may be vulnerable to hypermethylation induced by H. pylori infection. To date, nothing is known about whether CGI loci of H. pylori-related hypermethylation have some characteristic chromatin marks for preferred hypermethylation.

In this study, we analyzed the relationship of H. pylori infection to aberrant methylation in chronic gastritis and gastric carcinoma with respect to focal CGI hypermethylation and generalized genomic hypomethylation using quantitative methylation analysis. To determine the change of methylation content caused by H. pylori infection, we measured the methylation levels of LINE-1 and Alu repetitive elements as surrogates for global genomic methylation because these repetitive elements comprise about 30% of genomic DNA [22] and a strong correlation between 5-methylcytosine content in the human genome and measures of Alu and LINE-1 methylation using polymerase chain reaction has been demonstrated in a previous study [41]. Twenty-five CGI loci were analyzed for their methylation levels by quantitative methylation-specific real-time PCR (MethyLight technology). Using previously published information from stem cells [23], we were able to determine the occupancy by Suz12 and Eed and the trimethylation status of H3K27 for the 25 genes we examined.

Materials and methods

Patients

Four hundred sixty-seven patients who visited Seoul National University Bundang Hospital for gastroscopy from September 2003 to March 2005 were enrolled in the study. Patients underwent endoscopic mucosal biopsy and blood sampling for H. pylori detection. Patients with a history of gastric surgery and systemic disease requiring chronic medication were excluded from the enrollment. Neither chemical gastritis nor autoimmune gastritis was included. Written informed consent was obtained from all participating patients, and the study was approved by the Seoul National University Hospital’s Institutional Review Board.

Gastroscopic biopsy and serologic evaluation

Three biopsy samples were taken from the greater curvature of the antrum and the body of the stomach. Two samples from the antrum and two samples from the stomach body were fixed in neutral buffered formalin solution and processed for hematoxylin–eosin (H&E) staining and modified Giemsa staining for histologic evaluation and assessment of the presence of H. pylori, respectively. The remaining two specimens were subjected to a rapid urease test (CLO test, Delta West, Bentley, Australia), which was monitored for up to 24 h. Blood samples were obtained from the patients and evaluated for the level of anti-H. pylori immunoglobulin (Ig) G using an enzyme-linked immunosorbent assay (Genedia H. pylori ELISA; Green Cross Sang A Co., Seoul, South Korea), and duplicate determinations were performed according to the manufacturer’s protocol. Cases negative for three tests (CLO test, H. pylori IgG test, and histological H. pylori evaluation) were regarded as H. pylori-negative.

DNA extraction and bisulfite modification

Because previous studies [18, 26] revealed that intestinal metaplasia is accompanied by enhanced promoter CGI hypermethylation compared with chronic gastritis without intestinal metaplasia and our preliminary study revealed enhanced CGI hypermethylation in intestinal metaplasia regardless of H. pylori infection, cases that showed intestinal metaplasia or epithelial neoplasia in any of the four tissue samples were excluded from the study in order to focus on the effect of H. pylori infection on DNA methylation changes in chronic gastritis. Gastric carcinoma cases or chronic gastritis cases without intestinal metaplasia were included in the study. Twenty archival tissue sections of 4-μm thickness were used for DNA extraction. In gastric carcinoma cases, manual microdissection was performed in order not to include non-neoplastic mucosal epithelial cells. The dissected tissues were pooled together in a microtube containing lysis buffer and proteinase K. Sodium bisulfite conversion of genomic DNA was performed as described [41].

MethyLight assay

Twenty-five CGI loci were analyzed by the MethyLight technology using primers and probes targeting promoter sequences or the first exon [9, 41]. The oligonucleotide sequences of the primers and probes used have been described [42]. Briefly, two sets of primers and probes designed specifically to bind to bisulfite-converted DNA were used in each reaction; one set of primers and probe were used for every methylated target to be analyzed (methylated reaction), and another pair of primers and probe were for the reference locus, ALU (normalization control reaction). Normalization control reactions were methylation-independent measurements to control for DNA amplification and normalize for input DNA. M.SssI-treated genomic DNA was used as a reference sample for complete methylation to determine the percentage of methylated reference (PMR) at a particular locus. PMR was defined as 100 × (methylated reaction / normalization control reaction)sample / (methylated reaction/normalization control reaction)M.SssI-Reference.

Combined bisulfite restriction analysis (COBRA) assay

Bisulfite-modified genomic DNA was analyzed by COBRA to determine the methylation level of LINE-1 and Alu, as described [7]. In brief, LINE-1 and Alu were amplified with their respective primers that anneal to short sequences without CpG sites to amplify both methylated and unmethylated alleles equally. For the Alu reaction, TaqI was used to digest the methylated alleles, and the digested products were electrophoresed on a 12% polyacrylamide gel. The intensities of the digested and undigested bands were determined using Image J software (http://rsb.info.nih.gov/ij), and the percentage of methylated alleles was calculated by dividing the intensity of the digested band by the sum of the intensities of the digested band and the undigested band and multiplying by 100. For the LINE-1 reaction, TaqI and TasI were used to digest the methylated alleles and unmethylated alleles, respectively. The percentage of methylated alleles was determined by dividing the intensity of TaqI-digested bands by the sum of TaqI-digested bands and TasI-digested bands and multiplying by 100.

Statistical analysis

All statistical analyses were performed using SPSS software (SPSS, Ver.11.0, Chicago, IL). The relationship of CGI hypermethylation to H. pylori infection or cancer development was assessed by comparing the PMR values of examined CGIs using the Student’s t test. The methylation levels of Alu or LINE-1 were compared between groups using the Student’s t test. Comparison of the number of methylated CGI loci between groups was also performed using the Student’s t test. A P value < 0.05 was considered to be statistically significant.

Results

A total of 151 mucosal samples were analyzed for their methylation status in 25 CGI loci by MethyLight and in two repetitive elements (LINE-1 and Alu) by COBRA. The selected CGI loci except for CDH1 were found to show higher methylation frequency or level in gastric carcinoma than in non-neoplastic gastric mucosa in our previous studies [17, 18] in which 170 CGI loci were analyzed for their methylation status in 16 gastric carcinoma and paired non-neoplastic gastric mucosa samples. CDH1 was selected because of its known association of promoter CGI hypermethylation with H. pylori infection in the stomach [5, 6, 24]. The information regarding PRC2 occupancy was available for 153 of the 170 genes, and 60 genes contain occupancy of at least one component of PRC2. All of the PRC2-occupied 60 CGI loci showed promoter DNA hypermethylation in gastric carcinoma or non-neoplastic gastric mucosa samples, whereas 55 (59%) of 93 PRC2-nonoccupied genes exhibited CGI hypermethylation in gastric carcinoma or non-neoplastic gastric mucosa. Of the selected 24 CGI loci (except for p14 for which PRC2 occupancy information is not available), 50% harbor occupancy of at least one component of PRC2, whereas 37.2% of the remaining 129 CGI loci contain occupancy of at least one component of PRC2. Forty three H. pylori-positive chronic gastritis samples (mean age, 55.1 years; range, 25–78 years), 39 H. pylori-negative chronic gastritis (mean age, 55.6 years; range, 24–85 years), 41 H. pylori-positive gastric carcinoma (mean age, 63.6 years; range, 40–82 years), and 28 H. pylori-negative gastric carcinoma (mean age, 64.7 years; range, 41–83 years) were included in the study. H. pylori-positive chronic gastritis cases were selected to match an age distribution and gender distribution comparable to those of the H. pylori-negative chronic gastritis cases. The H. pylori-positive chronic gastritis was composed of two subgroups, a current H. pylori infection group (n = 33) and a prior H. pylori infection group (n = 10). The current infection group was defined by the identification of H. pylori on H&E-stained biopsy slides or by a positive CLO test, and the prior infection group was defined by a positive H. pylori IgG test but negative CLO test and absence of H. pylori infection on histological examination.

CGI hypermethylation of chronic gastritis versus gastric carcinoma

To determine CGI loci showing cancer-related hypermethylation, the methylation levels of 25 CGI loci were compared between H. pylori-negative chronic gastritis (n = 39) and H. pylori-negative gastric carcinoma (n = 28). Twenty-three CGI loci (all except for CDH1 and DAPK1) showed significantly higher methylation levels in H. pylori-negative gastric carcinoma than in H. pylori-negative chronic gastritis (P < 0.05), indicating a cancer-related methylation (Fig. 1a). The remaining two loci showed apparent higher methylation levels in gastric carcinoma than in chronic gastritis, but the difference was statistically insignificant. When the methylation levels of examined CGI loci were compared between H. pylori-positive chronic gastritis (the current infection group, n = 33) and H. pylori-positive gastric carcinoma, the cancer-related methylation loci except for seven loci (p14, HOXA1, NEUROG1, SOCS1, CRABP1, MLH1, and SMAD9) also displayed significantly higher methylation levels in gastric carcinoma than in chronic gastritis (P < 0.05) (Fig. 1b). The level of global genomic methylation, represented by methylation levels of LINE-1 and Alu, was significantly lower in gastric carcinoma than in chronic gastritis (P < 0.05), regardless of the status of H. pylori infection (Fig. 2a).

Fig. 1
figure 1

a Comparison of the methylation levels of 25 CGI loci between H. pylori-negative (HP−) chronic gastritis specimens from non-cancer patients and HP− gastric carcinoma specimens. b Comparison between chronic gastritis cases with current H. pylori infection (from non-cancer patients) and H. pylori-positive gastric carcinoma cases. The examined CGI loci were sorted ordered from highest to lowest P values from left to right along the x axis. Error bars indicate S.E.M.

Full size image
Fig. 2
figure 2

a COBRA assay of LINE-1 and Alu repetitive elements reveal significantly lower methylation levels in H. pylori-negative gastric carcinoma than in H. pylori-negative chronic gastritis. Error bars indicate S.E.M. b COBRA assay of LINE-1 or Alu methylation levels demonstrated no difference between H. pylori-positive and H. pylori-negative chronic gastritis samples

Full size image

CGI hypermethylation of H. pylori-positive chronic gastritis versus H. pylori-negative chronic gastritis

Of 25 CGI loci hypermethylated in gastric carcinoma, four loci (RUNX3, SOCS1, TERT, and TP73) were never methylated in chronic gastritis samples, regardless of H. pylori infection. Figure 3 depicts the summary of the PMR values generated by the MethyLight analysis. When the methylation levels of 21 CGI loci were compared between H. pylori-positive chronic gastritis (the current H. pylori infection group, n = 33) and H. pylori-negative chronic gastritis, close association of hypermethylation with H. pylori infection was noted in eight CGI loci (CACNA1G, CALCA, CDH1, CRABP1, CYP1B1, DAPK1, GRIN2B, and TWIST1; Fig. 4). Of these, six loci (CALCA, CACNA1G, CRABP1, CYP1B1, GRIN2B, and TWIST1) showed both cancer-related and H. pylori-related hypermethylation. Of the 23 cancer-related methylation loci, 13 loci showed more than a two-fold increase in methylation levels in H. pylori-positive chronic gastritis samples than in H. pylori-negative chronic gastritis samples. No difference was noted in LINE-1 and Alu methylation levels between H. pylori-positive and H. pylori-negative chronic gastritis specimens (Fig. 2b).

Fig. 3
figure 3

Methylation map of the percent of methylated reference (PMR) values obtained for each CpG island (CGI) locus in H. pylori-positive (HP-positive) or H. pylori-negative chronic gastritis (HP-negative CG) cases. In total, 82 CG samples (horizontal axis) were examined for their methylation status at 25 CGI loci (vertical axis). The data shows that HP-positive CG samples tend to be more hypermethylated than HP-negative CG samples with regard to methylation levels or methylation frequencies. Colored boxes represent four classes of methylation levels [0 < PMR < 4 (light blue), 4 ≤ PMR < 20 (light yellow), 20 ≤ PMR < 50 (light orange), PMR ≥ 50 (brown)], as indicated in the figure

Full size image
Fig. 4
figure 4

Comparison between chronic gastritis cases with current H. pylori infection (from non-cancer patients) and chronic gastritis cases with no current H. pylori infection and no past history of H. pylori infection (from non-cancer patients). The examined CGI loci were sorted ordered from highest to lowest P values from left to right along the x axis. The left eight CGI loci showed a statistically significant difference in methylation levels (P < 0.05)

Full size image

When a CGI locus was arbitrarily regarded methylated when its PMR value was ≥4, the number of CGI loci methylated varied from 0 to 11 in the H. pylori-positive chronic gastritis samples (the current H. pylori infection group, n = 33) and from 0 to 5 in the H. pylori-negative chronic gastritis samples (mean, 4.3 vs. 1.4; P < 0.001) (Fig. 5a).

Fig. 5
figure 5

a Distribution of the number of methylated CGI loci in H. pylori-positive chronic gastritis (current H. pylori infection) and H. pylori-negative chronic gastritis (no current H. pylori infection and no history of H. pylori infection). A CGI locus was considered methylated if the PMR was more than 4. b Comparison of the number of methylated CGI loci between chronic gastritis with current H. pylori infection and chronic gastritis with past H. pylori infection. Note the marked difference in the number of methylated CGI loci between the two groups

Full size image

CGI hypermethylation in current H. pylori infection versus past H. pylori infection

The H. pylori-positive chronic gastritis cases were divided into two groups, namely current infection (n = 33) and prior infection (n = 10). Comparison of methylation levels of the examined CGI loci revealed that the methylation levels of the eight CGI loci showing H. pylori-related hypermethylation were apparently lower in the prior infection group than in the current infection group, but the differences were not significant (data not shown). When the number of methylated CGI loci (PMR ≥ 4) was compared between the two groups, the current infection group showed a significantly higher number of methylated loci than the prior infection group (mean number, 4.3 vs. 2.6; P = 0.014; Fig. 5b). Moreover, the number of CGI loci methylated was significantly higher in the prior infection group than in the H. pylori-negative chronic gastritis group (mean, 2.6 vs. 1.4; P = 0.022).

CGI hypermethylation of H. pylori-positive gastric carcinoma versus H. pylori-negative gastric carcinoma

Although no significant difference was noted in the comparison of the methylation levels of the 24 CGI loci between H. pylori-positive and H. pylori-negative gastric carcinomas, THBS1 showed a statistically insignificant 2-fold increase in methylation levels in H. pylori-infected gastric carcinoma samples. Alu and LINE-1 methylation levels did not differ between the H. pylori-positive and H. pylori-negative gastric carcinoma samples.

Discussion

In this study, H. pylori-positive and -negative chronic gastritis cases were analyzed for their methylation status of 25 promoter CGI loci using the MethyLight assay. The methylation levels of 15 CGI loci in H. pylori-positive chronic gastritis cases were elevated more than two-fold compared with those in H. pylori-negative chronic gastritis, and of these CGI loci the differences in the methylation levels in eight loci reached statistical significance. Furthermore, the number of CGI loci hypermethylated (PMR ≥ 4) was significantly higher in the current infection group than in the prior H. pylori infection group. These results indicate that H. pylori infection is closely associated with aberrant CGI hypermethylation in chronic gastritis, consistent with the findings of other researchers [5, 24, 25]. However, it should be noted that variability existed in the susceptibility to hypermethylation induced by H. pylori infection depending on the type of CGI loci: 15 CGI loci showed more than two-fold increases in methylation levels in H. pylori-positive chronic gastritis compared to H. pylori-negative chronic gastritis cases, whereas the remaining ten CGI loci did not. Even within the current H. pylori-infection chronic gastritis subjects, the number of CGI loci hypermethylated per case displayed a wide distribution, indicating the existence of other factors, including host factors, in the vulnerability of CGIs to hypermethylation provoked by H. pylori infection. Host factors may include the duration of H. pylori infection and inflammatory response, and genetic variables. For example, genetic polymorphisms of the inflammatory cytokines involved in H. pylori infection may affect susceptibility to H. pylori-induced hypermethylation [4].

Histologically, H. pylori-positive chronic gastritis tends to show more mononuclear and polymorphonuclear cell infiltration in the lamina propria than H. pylori-negative chronic gastritis, which was also the case in our study. Because we did not perform laser capture microdissection, inflammatory cells may give rise to a bias in the comparison of the methylation levels between H. pylori-positive and H. pylori-negative chronic gastritis. To identify whether the increased methylation levels of CGI loci in H. pylori-positive chronic gastritis were related to increased infiltration of inflammatory cells in the gastric mucosa, we examined peripheral blood leukocytes (from 50 healthy volunteers) for the methylation status of 25 CGI loci using the MethyLight technology. None of the CGI loci gave a PMR ≥ 4 in peripheral blood leukocyte DNA (Supplementary data).

The present study provides supportive evidence that H. pylori infection is closely associated with aberrant CGI hypermethylation in the stomach. However, the mechanism of methylation induction in H. pylori-infected gastric mucosa remains unclear. A plausible explanation for the mechanism is as follows: an intense inflammatory response and an increased production of pro-inflammatory cytokines in the gastric mucosa provoked by H. pylori infection might participate in the induction of aberrant hypermethylation. (1) Interleukin-6, one of the cytokines produced in the H. pylori-infected gastric mucosa [1], has been shown to regulate the DNMT1 promoter and the resulting enzyme activity [14, 15] and to increase methylcytosine content and modulate CGI hypermethylation [40]. (2) DNA methyltransferase expression has been reported to increase in the presence of H. pylori infection [36]. Overexpression of DNA methyltransferase is itself capable of bringing about de novo methylation of selective CGI loci [12]. (3) 5-Halogenated cytosine damage products, including 5-chlorocytosine and 5-bromocytosine, are formed in vivo by inflammation-mediated reactive molecules [2, 13] and it has been suggested that these modified pyrimidines mimic 5-methylcytosine and induce de novo methylation of previously unmethylated CpG sites [38].

CGI loci that are methylated in human cancers may be marked for preferred methylation. In recent studies, genes hypermethylated in human cancers were found to be often those genes which are marked by H3K27 trimethylation or Polycomb component occupancy in ES cells, where they are unmethylated [29, 32, 43]. In other words, PRC2 target genes of ES cells are predisposed to DNA hypermethylation occurring in association with cancer development. Because PRCs are not only essential regulators of embryonic development but also key players in the maintenance of adult stem cells [27, 39], gastric epithelial stem cells are expected to carry many genes under the control of PRCs and these Polycomb target genes may preferentially succumb to DNA hypermethylation induced by H. pylori infection. Although we could not explore the PRC occupancy of the examined 25 genes in gastric epithelial stem cells, published information about the occupancy by Polycomb group components (Suz12 and Eed) or H3K27 trimethylation at the genes in ES cells enabled us to analyze the association [23]. Based on the results of our methylation analysis, the examined 25 genes could be classified into three groups: (1) H. pylori-associated methylation only (CDH1, and DAPK), (2) cancer-associated methylation only (BCL2, BDNF, CHFR, DLEC1, HOXA1, MLH1, NEUROG1, p14, RUNX3, SEZ6L, SFRP4, SMAD9, SOCS1, TERT, THBS1, TIMP3, and TP73), and (3) both H. pylori- and cancer-associated methylation (CACNA1G, CALCA, CRABP1, CYP1B1, GRIN2B, and TWIST1). Seventy five percent of the genes showing H. pylori-associated methylation (the first and third groups) contain at least one of the three repressive marks, namely Suz12 occupancy, Eed occupancy, or H3K27 methylation, whereas 69% of the genes with cancer-associated methylation (the second group) do not carry any of the three repressive marks (Fig. 6). Furthermore, all of the genes with methylation associated with both H. pylori and cancer (the third group) contain one or more of the repressive marks. This fact suggests that repressive chromatin marks may leave genes vulnerable to later aberrant DNA hypermethylation induced by H. pylori infection, on the assumption that repressive chromatin marks established on genes early in development are maintained in gastric epithelial stem cells and progenitor cells. However, it should be noted that CDH1 and DAPK, showing association of their hypermethylation with H. pylori infection, are not occupied by PRC2 in ES cells, and also that a large portion of the genes showing cancer-related methylation do not contain the repressive marks. Thus, other potential factors capable of directing targeted methylation should exist and may include the local DNA sequence motifs. Genome-wide DNA methylation studies have demonstrated the presence of such motifs which are significantly enriched in methylated promoter CGIs relative to the bulk of unmethylated CGIs [21, 33].

Fig. 6
figure 6

Venn diagram showing CGI loci of H. pylori-associated methylation and those of gastric cancer-associated methylation. CGI loci having significantly higher methylation levels in H. pylori-negative gastric cancer specimens compared to H. pylori-negative chronic gastritis specimens are referred to as cancer-associated loci. The loci for which DNA hypermethylation was significantly associated with H. pylori infection are referred to as H. pylori-associated loci. The occupancy of the three components of PRC2 (Suz12 (S), Eed (E), and Ezh2 (H)) at the examined 25 genes in ES cells could be established from the published information and is given in parentheses and the numbers in parenthesis indicates the total number of occupying components. The status of PRC2 occupancy or H3K27 methylation for p14 in ES cells is not known

Full size image

Generalized genomic hypomethylation as well as focal CGI hypermethylation constitute aberrant methylation changes characteristic of human cancer cells and aged cells [10, 30]. Both methylation changes are frequently found in gastric carcinoma cells and even in multi-step lesions prior to gastric carcinoma [8, 19]. Given these facts, H. pylori infection was expected to affect the global genomic methylation status of the gastric epithelial cells. However, Alu and LINE-1 methylation levels, surrogates for genomic methylation levels, did not decrease in H. pylori-positive chronic gastritis samples compared with those of H. pylori-negative chronic gastritis samples, but they were significantly lower in gastric carcinoma compared with chronic gastritis cases, regardless of H. pylori infection. Although the number of analyzed cases was low, the findings suggest that H. pylori infection affects methylation changes of CGI loci but does not change the methylation content of the repetitive elements.

We have analyzed the relationship of aberrant DNA methylation changes in the stomach relative to H. pylori infection using the MethyLight assay on 25 genes and the COBRA assay on Alu and LINE-1 repetitive elements. The results suggest that H. pylori infection is strongly associated with focal CGI hypermethylation but does not affect the genomic methylation contents. Identification of Polycomb group protein occupancy in stem cells of the 25 genes examined revealed that genes occupied by Polycomb group proteins during embryonic development seem to be vulnerable to H. pylori-related DNA hypermethylation.