By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Metachronous occurrence of gastric cancers is becoming an important issue as localized resection of early gastric cancers by endoscopic submucosal dissection (ESD) has become common.1 The incidence of secondary primary gastric cancers after ESD reaches as high as 2.0% per year2 whereas the incidence of gastric cancer in the general Japanese population is 0.14% per year.3 This indicates that noncancerous gastric mucosae are already predisposed to developing gastric cancers, forming a field defect (field for cancerization). High incidences of metachronous cancers have been known not only for gastric cancers but also for bladder, liver, and esophageal cancers4–6 and are becoming recognized for lung, breast and colorectal cancers.7–9
A molecular basis for the field defect has been considered as an accumulation of genetic and epigenetic alterations in normal-appearing tissues. Traditionally, cells with a genetic alteration were considered to form a physically continuous patch, producing a genetically altered field.10 Recently, we found that aberrant DNA methylation of specific genes can be induced in as high as several percentage of cells in noncancerous gastric mucosae (thus in multiple independent gastric glands), and the degree of methylation is associated with gastric cancer risks.11, 12 Importantly, Helicobacter pylori infection, a major carcinogenic factor in the stomach, was shown to potently induce aberrant DNA methylation in gastric epithelial cells.11 In addition to gastric cancers, the presence of aberrant DNA methylation in noncancerous tissues and possible association with cancer risks have been reported for liver,13 colon,14 esophageal,15 breast16 and renal cancers.17
Genes so far analyzed in noncancerous gastric mucosae are those methylated in gastric cancers, including tumor-suppressor genes, such as CDKN2A, MLH1, CDH1, LOX and APC,2, 11, 18, 19 and genes with little or no expression in normal gastric mucosae, such as FLNc, HAND1 and THBD. The latter group of genes is methylated in parallel with tumor-suppressor genes but with higher frequencies, and is considered as a good marker to detect the presence of an epigenetic field defect.11 In contrast with these protein-coding genes, involvement of microRNA (miRNA) silencing in field defect formation has not been clarified yet. Since the role of aberrant expression or reduction of various miRNAs in human multistep carcinogenesis is now clear,20, 21 there is a possibility that miRNAs silencing by aberrant DNA methylation is involved in field defect formation. Indeed, several tumor-suppressor miRNAs, including miR-124a,22miR-137, miR-193a23 and miR-127,24 are reported to be silenced by aberrant DNA methylation of their promoter CpG islands (CGI) in cancers.
In this study, we aimed to clarify whether or not miRNA silencing by DNA methylation can be involved in the formation of a field defect for gastric cancers. First, we searched for miRNAs that are reported to have tumor-suppressive functions and be controlled by DNA methylation, and confirmed methylation-silencing of these candidate genes. Then, we quantified their methylation levels in gastric mucosae of healthy volunteers, noncancerous gastric mucosae of gastric cancer patients, and primary gastric cancer tissues.
Material and methods
Cell lines and tissue samples
Six gastric cancer cell lines, AGS, KATOIII, MKN28, MKN45, MKN74 and NUGC3 were obtained from the Japanese Collection of Research Bioresources (Tokyo, Japan) and the American Type Culture Collection (Manassas, VA). Three gastric cancer cell lines, HSC39, HSC44 and HSC57 were gifted by Dr. K. Yanagihara, National Cancer Center Research Institute, Tokyo, Japan. GC2 was developed by M. T. TMK1 was gifted by Dr. W. Yasui, Hiroshima University, Hiroshima, Japan. 5-Aza-2′-deoxycitidine (5-aza-dC) treatment was performed with AGS, HSC57 and MKN28. Cells were seeded on day 0, media was added with freshly prepared 5-aza-dC on days 1 and 3, and cells were harvested on day 5. The concentrations of 5-aza-dC were determined as minimum concentrations that deplete DNMT1.25
Gastric mucosae were obtained by endoscopic biopsy of antral regions from 56 healthy volunteers (25 male and 31 female; average age 53, ranging from 27 to 91) and 45 gastric cancer patients (35 male and 10 female; average age 66, ranging from 38 to 89). Gastric cancer tissues were obtained from 28 gastric cancer patients (21 male and 7 female; average age 66, ranging from 49 to 81; 13 intestinal and 15 diffuse types) who underwent gastrectomy due to gastric cancers. Gastric epithelial cells were obtained by the gland isolation technique from eight noncancerous gastric tissues. Informed consents were obtained from all the patients and healthy volunteers. Gastric mucosae, noncancerous mucosae and cancer tissues were frozen in liquid nitrogen immediately after biopsy or resection, and stored at −80°C until extraction of genomic DNA. High molecular weight DNA was extracted by the phenol/chloroform method. RNA was isolated with ISOGEN (Nippon Gene, Tokyo, Japan).
H. pylori infection status was analyzed by a serum anti-H. pylori IgG antibody test (SRL, Tokyo, Japan), rapid urease test (Otsuka, Tokushima, Japan), or culture test (Eiken, Tokyo, Japan). All cancers were histologically diagnosed according to the Japanese classification of gastric carcinoma,26 and classified according to the Lauren classification system.27
Fully methylated DNA and fully unmethylated DNA were prepared by methylating genomic DNA with SssI methylase (New England Biolabs, Beverly, MA) and by amplifying genomic DNA with the GenomiPhi amplification system (GE Healthcare, Buckinghamshire, UK), respectively. Bisulfite modification was performed using 1 μg of BamHI-digested genomic DNA as previously described,28 and the modified DNA was suspended in 40 μl of Tris-EDTA buffer. An aliquot of 1 μl was used for methylation-specific PCR (MSP) and Quantitative real-time MSP (qMSP) with a primer set specific to methylated (M) or unmethylated (U) sequences.
For MSP, the fully methylated and unmethylated DNA was used to determine an annealing temperature that specifically amplifies only methylated or unmethylated DNA. A minimum number of PCR cycles to obtain visible bands was determined using the fully (un)methylated DNA, and four cycles were added for analysis of gastric cancer cell lines. The primers were designed just upstream of reported transcription start sites within the CGI (Table I; Fig. 1a), whose methylation statuses are now known to be critical for induction of gene silencing.29, 30
Table I. Primers and Conditions for MSP and Real-Time MSP
Number of cycles for MSP
Reverse (5′ →3′)
M, Primers specific to methylated DNA; U, Primers specific to unmethylated DNA.
CAAAAA AAA AAAATAAAAAACAACAC
Primers for bisulfite sequencing
Number of cycles
Reverse (5′ →3′)
qMSP was performed by real-time PCR using SYBR® Green I (BioWhittaker Molecular Applications, Rockland, ME) and an iCycler Thermal Cycler (Bio-Rad Laboratories, Hercules, CA). Although the same primer set was used for qMSP, a specific annealing temperature in the presence of SYBR® Green I was re-determined using the fully methylated and unmethylated DNA. The number of molecules in a sample was determined by comparing its amplification with those of standard DNA that contained exact numbers of molecules (101–106 molecules). Based on the numbers of M molecules and U molecules for a genomic region, a methylation level of the region was calculated as the fraction of M molecules in the total number of DNA molecules (# of M molecules + # of U molecules). The standard DNA samples were prepared by cloning PCR products of methylated and unmethylated sequences into the pGEM-T Easy vector (Promega, Madison, WI), respectively, or by purifying the PCR products using the Wizard SV Gel and PCR clean-up system (Promega).
For bisulfite sequencing, an aliquot of 1 μl of the sodium bisulfite-treated DNA was amplified by PCR with the primers common to methylated and unmethylated DNA sequences (Table I). The PCR product was cloned into pGEM-T Easy vector (Promega), and 15 clones or more were cycle-sequenced for each sample.
For quantitative RT-PCR, cDNA was synthesized from 10 ng of total RNA using TaqMan® MicroRNA-specific primers and a TaqMan® MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Real-time PCR was performed using the ABI Prism 7300 Fast Real-Time PCR System (Applied Biosystems). Expression levels of target miRNAs were normalized to that of a small nuclear RNA RNU6B transcript.
A difference in mean methylation levels was analyzed by the t-test Welch method, and differences in methylation incidence in gastric cancer tissues were analyzed by the chi-square test. Correlation between the age and methylation levels of miRNA genes, and correlation between methylation levels of each gene were analyzed using Spearman's rank correlation coefficient. All the analyses were performed using SPSS (SPSS, Inc., Chicago, IL), and the results were considered significant when a p value less than 0.05 was obtained by two-sided tests.
Identification of miRNAs silenced in gastric cancer cell lines
Six genes of four miRNAs (miR-124a, miR-137, miR-193a and miR-127) were reported to have a tumor-suppressive function and be controlled by DNA methylation in colon, bladder and oral cancers.22–24 We first analyzed methylation statuses of their putative promoter regions in 11 gastric cancer cell lines and two normal gastric mucosae of healthy individuals without H. pylori. It was found that miR-124a-1, miR-124a-2, miR-124a-3 and miR-137 were unmethylated in the normal gastric mucosae, but were completely methylated (no unmethylated DNA molecules detected) frequently in the cell lines (Figs. 1b–1d). miR-193a was partially methylated in one of the two normal gastric mucosae, and completely methylated frequently in gastric cancer cell lines. In contrast, miR-127 was completely methylated in the normal gastric mucosae, but unmethylated in the gastric cancer cell lines.
We then examined the effect of methylation of the putative promoter regions on miRNA expression (miR-124a for miR-124a-1, miR-124a-2 and miR-124a-3 genes; miR-137; miR-193a; miR-127) in the 11 gastric cancer cell lines and gastric epithelial cells obtained by the gland isolation technique (Fig. 2). miR-124a was consistently unexpressed in six cell lines with simultaneous methylation of its three isoforms (miR-124a-1, miR-124a-2 and miR-124a-3), but was expressed in the gastric epithelial cells. In contrast, miR-137, miR-193a, and miR-127 were expressed even in cell lines with complete methylation. This showed that these three miRNA genes were not silenced by their “promoter” methylation, and indicated that, in contrast, miR-124a was silenced by promoter methylation of its three isoforms.
Methylation-silencing of miR-124a was further confirmed by analyzing its re-expression in association with its promoter demethylation after treatment with a demethylating agent, 5-aza-dC, in three cell lines (AGS, HSC57 and MKN28). Re-expression and appearance of unmethylated DNA molecules were observed in all the three cell lines, HSC57 being prominent. This further orted that miR-124a was methylation-silenced.
The presence of miR-124a methylation in primary gastric cancers
Since methylation-silencing was identified only for miR-124a, we analyzed methylation levels of its three genes (miR-124a-1, miR-124a-2 and miR-124a-3), along with a representative protein-coding gene (LOX), in 28 primary gastric cancer tissues (13 intestinal and 15 diffuse types) by qMSP. The fact that densely methylated DNA molecules were being measured was confirmed by bisulfite sequencing (Supp. Info. Fig. 1). miR-124a-1 showed a distribution of methylation levels similar to LOX, some having no methylation and the others having various levels of methylation (Fig. 3a). This was consistent with our previous finding that cancer samples could be essentially classified into two groups (cancers with and without methylation), and that the various degrees of methylation levels in methylation-positive cancer samples were mainly due to contamination of normal cells.18 On the other hand, miR-124a-2 and miR-124a-3 showed a unimodal distribution of methylation levels, suggesting that they are susceptible to methylation induction in cancer tissues. Using a cut-off value of 6%, as in previous reports,31, 32miR-124a-1, miR-124a-2 and miR-124a-3 were methylated in 11, 23 and 26 of the 28 samples, respectively. Between the two histological types, the incidences of methylation were the same for miR-124a-1, miR-124a-2 and miR-124a-3 (p = 0.95, 0.84 and 0.67) (Supp. Info. Fig. 2a).
We further analyzed an association between methylation and expression of miR-124a in an additional 19 gastric cancer samples. Using a cut-off value of 6%, eight samples had methylation of all the three miR-124a genes, and the other 11 samples had methylation of only one or two genes and retained at least one unmethylated gene. miR-124a was barely expressed in all the eight samples with methylation of the three genes whereas it was expressed in 5 of 11 cancer samples with at least one unmethylated gene (Fig. 3b).
Accumulation of methylation in H. pylori positive gastric mucosae, and its association with gastric cancer risk
Methylation levels of miR-124a-1, miR-124a-2 and miR-124a-3, again along with LOX, were analyzed by qMSP in gastric mucosae of 56 healthy volunteers (28 volunteers with H. pylori and 28 without) and noncancerous gastric mucosae of 45 gastric cancer patients (29 patients with H. pylori and 16 without) (Fig. 3b). Among the healthy volunteers, the mean methylation levels of miR-124a-1, miR-124a-2, miR-124a-3 and LOX in the H. pylori-positive individuals were 13.1-, 7.8-, 8.9- and 46.7-fold, respectively, as high as those in H. pylori-negative individuals. This showed that H. pylori infection was associated with aberrant methylation of not only protein-coding genes but also miRNA genes.
Next, methylation levels in gastric mucosae of healthy volunteers were compared with those of noncancerous gastric mucosae of gastric cancer patients. Since potent methylation induction by H. pylori can mask a difference in H. pylori-positive individuals, the comparison was made among H. pylori-negative individuals only (28 healthy volunteers and 16 gastric cancer patients) (Table II; Fig. 3c). The mean methylation levels of the three miRNA genes and LOX were much higher in noncancerous gastric mucosae of gastric cancer patients than those of gastric mucosae of healthy volunteers (15.5-, 7.2-, 13.3- and 24.7-fold, respectively). Between the two histological types, the mean methylation levels were not different (Supp. Info. Fig. 2b).
Table II. Mean Methylation Levels and Standard Deviations of the Four Genes in Gastric Mucosae of Healthy Volunteers and Gastric Cancer Patients
(1) Healthy volunteers
0.76 ± 1.70
5.75 ± 7.73
4.45 ± 9.29
0.43 ± 1.22
(2) Gastric cancer patients
11.82 ± 12.94
41.66 ± 19.33
59.42 ± 30.71
10.64 ± 11.33
(3) Healthy volunteers
9.96 ± 12.28
44.79 ± 23.96
39.66 ± 30.08
20.10 ± 15.72
(4) Gastric cancer patients
12.28 ± 14.31
46.33 ± 28.36
37.48 ± 25.13
15.93 ± 12.58
(1) vs. (3)
(1) vs. (2)
(3) vs. (4)
Correlations among methylation levels of miRNA genes and LOX were examined by calculating correlation coefficients. Correlations among the three miRNA genes were very strong, but correlations between a miRNA gene and LOX were weak or absent (Table III; Supp. Info. Fig. 3).
Table III. Correlation among Methylation Level of miR-124a-1, miR-124a-2, miR-124a-3 and LOX
r, correlation coefficient.
No effect of age and sex on methylation levels on miRNA genes
Methylation of various CGIs is reported to be correlated with age.33, 34 Also, males have twice as high an incidence of gastric cancers as females.1 In H. pylori-negative healthy volunteers, methylation levels of miR-124a-1, miR-124a-2 and miR-124a-3 were not correlated with age (Spearman correlation test: r = 0.19, 0.01 and 0.29; p = 0.35, 0.94 and 0.15), and not associated with sex (p = 0.05, 0.68 and 0.19). Also, in H. pylori-positive healthy volunteers, methylation levels were not correlated with age (r = 0.13, 0.18 and −0.1; p = 0.51, 0.35 and 0.51), and not associated with sex (p = 0.70, 0.20 and 0.67).
The present study showed that significantly higher methylation levels of three miRNA genes (miR-124a-1, miR-124a-2 and miR-124a-3) were present in gastric mucosae of H. pylori-positive healthy volunteers, indicating that H. pylori infection can induce DNA methylation of miRNA genes, in addition to protein-coding genes. Moreover, it was also shown that methylation levels of the miRNA genes in noncancerous gastric mucosae of gastric cancer patients were higher than those in gastric mucosae of healthy volunteers among H. pylori negative individuals, indicating that miRNA silencing is involved in the formation of a field defect for gastric cancers. To our knowledge, the presence of miRNA silencing in a field for cancerization was shown here for the first time.
Recent studies demonstrated that expression of some miRNAs is regulated by epigenetic mechanisms.24, 35 From six miRNA genes that were reported to be silenced by promoter methylation and to have tumor-suppressor functions, we were able to confirm that three genes of miR-124a were methylation-silenced in gastric cancer cell lines. The other three genes, miR-137, miR-193a and miR-127, were expressed even in cell lines with complete methylation, and were unlikely to be silenced by promoter methylation in gastric cancers. Since methylation of putative promoter regions consistently represses transcription of their downstream genes,29, 30 the presence of the expression of the three genes in gastric cancer cell lines with complete methylation of their “promoter” CGI indicated that the three genes had additional or alternative promoters.
Lujambio et al.22 discovered that miR-124a was silenced by promoter methylation after screening 320 miRNA genes. They also found that miR-124a down-regulates CDK6, a demonstrated oncogene involved in cell cycle progression and differentiation, and induces hypophosphorylation of RB.22 Therefore, it is possible that miR-124a silencing is also involved in gastric carcinogenesis, and the presence of its silencing in noncancerous tissues could be directly associated with predisposition to developing gastric cancers.
As repeatedly shown by epidemiological studies, the majority of H. pylori-negative individuals with a gastric cancer are considered to have past exposure to H. pylori.36 Methylation levels of protein-coding genes, including LOX, in the gastric mucosae of individuals with past infection (gastric cancer patients without H. pylori) were lower than those of individuals with current infection (both healthy volunteers and gastric cancer patients) in our previous study.11 Actually, incidences of aberrant methylation and methylation levels of CDH1 are reported to decrease after the eradication of H. pylori,19, 37 showing that DNA methylation in gastric mucosae decreases when H. pylori infection discontinues. Interestingly, methylation levels of the three miRNA genes in the gastric mucosae of individuals with past infection by H. pylori (gastric cancer patients without H. pylori) did not decrease compared with those of individuals with current infection (healthy volunteers and gastric cancer patients). Since aberrant methylation induced in stem cells is expected to persist even after H. pylori infection discontinues, DNA methylation of these miRNA genes might be relatively more easily induced in gastric stem cells than those of protein-coding genes.
In conclusion, our data indicated that DNA methylation of certain miRNA genes was associated with H. pylori infection, in addition to protein-coding genes, and involved in the formation of field defect for gastric cancers.
The authors thank Drs. Eriko Okochi-Takada and Satoshi Yamashita for their advice.