Inactivation of O6-methylguanine-DNA methyltransferase by promoter CpG island hypermethylation in gastric cancers

Promoter hypermethylation of CpG islands in tumour suppressor genes can lead to transcriptional inactivation. To investigate the association between methylation and expression at O6-methylguanine-DNA methyltransferase, we performed methylation-specific PCR and immunohistochemistry in 149 gastric carcinomas. Promoter methylation was found in 14.1% of tumours and loss of expression was detected in 11.4% of tumours. To examine correlation between the O6-methylguanine-DNA methyltransferase expression and the clinical data, we investigated O6-methylguanine-DNA methyltransferase expression in 315 consecutive gastric carcinomas. A similar frequency of loss of O6-methylguanine-DNA methyltransferase expression was confirmed in these cases. The loss of O6-methylguanine-DNA methyltransferase expression was significantly associated with pTNM stage (P=0.037), tumour invasion (P=0.02), microsatellite instability (P=0.041) and overall survival (P=0.01). Among 11 gastric cancer cell lines, SNU-620 showed the loss of O6-methylguanine-DNA methyltransferase expression as well as promoter methylation. After treatment with 5-aza-2-deoxycytidine, a demethylating agent, SNU-620 re-expressed O6-methylguanine-DNA methyltransferase mRNA. In summary, we suggest that during gastric carcinogenesis, the loss of O6-methylguanine-DNA methyltransferase expression frequently occurs via the hypermethylation of the CpG islands of the promoter region, and that this is significantly associated with the clinicopathological characteristics. British Journal of Cancer (2002) 86, 1888–1892. doi:10.1038/sj.bjc.6600372 www.bjcancer.com © 2002 Cancer Research UK

Epigenetics is defined as an inheritable effect that influences gene activity but does not involved a change in the DNA sequence. DNA methylation is one of these epigenetic changes in human cancer and is involved in gene silencing. About 60% of human genes have CpG islands at the 5' region of their DNA sequence. This region contains the promoter and the exons and is normally unmethylated (Toyota et al, 2000). Aberrant methylation of CpG islands in tumour suppressor genes can lead to transcriptional inactivation. Promoter hypermethylation of CpG islands of tumour suppressor genes (p16, p14, and APC), DNA repair genes (hMLH1 and MGMT) and genes related to metastasis and invasion (Ecadherin, TIMP-3, and DAPK) has been shown in various cancers and is associated with the loss of expression (Yoshiura et al, 1995;Herfarth et al, 1999;Eads et al, 2001;Esteller et al, 2001a;Rosas et al, 2001Zöchbauer-Müller et al, 2001. O 6 -methylguanine-DNA methyltransferase (MGMT) is a DNA repair gene, which removes methyl groups as well as larger adducts at O 6 position of guanine. MGMT transfers the alkyl group from O 6 -guanine in DNA to an active cysteine within its own sequence in a reaction that inactivates one MGMT molecule for each lesion repaired (Pegg, 1990;Esteller et al, 1999). Methylating agents such as N-methyl-N-nitrosourea (MNU) and N-methyl-N'-nitro-Nnitrosoguanidine (MNNG) react with guanine in DNA to form O 6 -methylaguanine. The alkylation of DNA at the O 6 position of guanine is associated with formation of DNA mutation in cancers, such as the transition of guanine-cytosine to adenine-thymine pairs in the K-ras mutation or in p53, which occurs because alkylated guanine is recognised as adenine during DNA replication (Esteller et al, 2000a(Esteller et al, , 2001bWhitehall et al, 2001).
Methylation of the MGMT gene has been reported in various carcinomas. In gliomas and colorectal cancers, methylation was shown in 38% of the tumour, whereas in non-small cell lung carcinomas, lymphomas, and head and neck carcinomas, methylation was demonstrated in 23 -28% of tumours (Esteller et al, 1999;Rosas et al, 2001). Promoter hypermethylation of MGMT in colorectal carcinomas results in transcriptional inactivation of MGMT gene (Esteller et al, 2000a).
We examined the loss of expression and the promoter methylation of MGMT in 149 gastric carcinomas and 11 gastric cancer cell lines and investigated an association with loss of expression and clinicopathological characteristics in consecutive gastric carcinomas.

Primary gastric cancer tissue samples
Initially, 149 stomach carcinomas and matched normal tissues were obtained from surgical resection specimens at Seoul National University Hospital from 1998 to 1999. All samples were fixed using absolute methanol, processed in chloroform and DNA was extracted by the phenol-chloroform methods. Formalin-fixed, paraffin embedded samples were arranged into three tissue array blocks. In addition to the 149 stomach carcinoma specimens, 315 consecutive cases of formalin-fixed, paraffin embedded stomach specimens were arranged into six tissue array blocks (Lee et al, 2001). None of the patients received preoperative chemo-or radiotherapy. These array blocks were cut into 4 um sections, and deparaffinized and dehydrated. Immunohistochemistry was carried out using a Vector ABC kit (Vector Laboratories, Inc., Burlingame, CA, USA) and mouse anti-MGMT monoclonal antibody (Chemicon, Temecula, CA, USA) at a dilution of 1 : 50.
For Western blot analysis, cell lines were harvested with PBSTDS containing 1% Trition X-100, 0.5% sodium deoxycholate, and 0.1% SDS. Equal amounts of protein, as determined by a bicinchoninic acid (BCA) assay (Pierce, Rockford, IL, USA), were dissolved in sample buffer, separated by electrophoresis on 12% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA, USA). The membrane was blocked with 3% non-fat dry milk in tris buffersaline with 0.1% Tween 20 for 1 h at room temperature and incubated with the mouse anti-MGMT monoclonal antibody (1:500, Chemicon) at 48C overnight. The membrane was then washed in the same buffer, incubated in horseradish peroxidase conjugated anti-mouse antibody (Amersham Pharmacia, Buckinghamshire, UK) for 30 min at room temperature, and then immersed in an enhanced chemiluminescence (ECL) Western blotting detection system (Amersham Pharmacia).
For reverse transription-PCR, total RNA of the gastric cancer cell lines was extracted using Trizol (LifeTechnologies) and 0.5 ug of the total RNA was used to generate cDNA. This cDNA was amplified with primers, 5'-GTG GGA GGA GCA ATG AGA GG-3' and 5'-TCC CGC TCC CTT GAG CCA GG-3' (Watts et al, 1997). The primers for b-actin, the positive control, were 5'-ACA CTG TGC CCA TCT ACG AGG-3' and 5'-AGG GGC CGG ACT CGT CAT ACT-3'. The PCR products obtained were analysed by 1.5% agarose gel electrophoresis with ethidium bromide and examined under UV light.
In order to demethylate the methyl-CpG sites of DNA, cell lines were cultured in a 100 mm culture dish (Nunc, Roskilde, Denmark) and were treated with 10 uM of 5-aza-2'-deoxycytidine (Sigma, St. Louis, MO, USA) for 10 days.

Methylation-specific PCR
DNA modification was carried out according to the procedure described by Frommer et al (1992). One ug of DNA was denatured for 5 min at 948C, 10 ul of 1 N HCl was then added, and the mixture was incubated for 10 min at 378C. The denatured DNA obtained was modified using 3.5 M sodium bisulphite per 1 mM hydroquinone (pH 5.0) for 16 h at 508C, and the modified DNA were then purified using a Wizard DNA clean-up system (Promega, Madison, WI, USA). Fifteen ul of 1 N HCl was added to the purified DNA, which was then precipitated with ethanol, and resuspended in 20 ul of water.
After the sodium bisulphite modification, the DNA was amplified in a volume of 10 ul with methylation specific primers (Esteller et al, 1999). The primers for methylated DNA were sense 5'-TTT CGA CGT TCG TAG GTT TTC GC-3' and antisense 5'-GCA CTC TTC CGA AAA CGA AAC G-3'. The primers for unmethylated DNA were sense 5'-TTT GTG TTT TGA TGT TTG TAG GTT TTT GT-3' and antisense 5'-AAC TCC ACA CTC TTC CAA AAA CAA AAC A-3'. The amplification conditions were as followed: 958C for 5 min; followed by 33 cycles of 948C for 30 s, 598C for 30 s, and 728C for 40 s; followed by final extension of 728C for 10 min. PCR products were electrophoresed on 6% polyacrylamide gel, stained with ethidium bromide, and visualised under UV illumination (Herman et al, 1996).

Sodium bisulphite genomic sequencing
After the sodium bisulphite modification, the DNA was amplified using hemi-nested primers. The primers were sense, 5'-TTA AGG TAT AGA GTT TTA GGC GGA AGT TGG-3', nested sense, 5'-TTT AGC GAG GAT GTG TAG ATT GTT TTA GGT-3', and antisense, 5'-AAA ACG AAA CGA CCC AAA CAC TCA CCA AAT-3' (Qian and Brent, 1997). The amplification conditions were as followed: 958C for 5 min; followed by 33 cycles of 948C for 30 s, 628C for 30 s, and 728C for 40 s; followed by final extension of 728C for 10 min. The first round PCR products were purified with High Pure PCR product purification kit (Boehringer Manheim, Mannheim, Germany). The final hemi-nested PCR products were amplified at an annealing temperature of 648C. The amplified PCR products were purified using a PCR product pre-sequencing kit (Amersham Pharmacia) and sequencing was carried out with a BigDye Terminator Cycle Sequencing Ready Reaction kit (Perkin Elmer, Foster City, CA, USA) using an ABI Prism 377 DNA sequencer (Perkin Elmer).

Statistics
Statistical analysis was performed using SPSS 9.0 software (SPSS, Chicago, IL, USA). Associations between the discrete variables were assessed using the two-sided Fisher's exact test or Pearson's chi square tests. Overall survival was calculated by Kaplan -Meier log-rank test. Multivariate analysis to assess the independent prognostic value of MGMT staining was performed by Cox regression. P value less than 0.05 was regarded as statistically significant.

Promoter methylation and loss of MGMT expression in 149 gastric carcinomas
To examine promoter methylation, we carried out methylationspecific PCR in the 149 methanol-fixed gastric carcinomas. Methylation was detected in 14.1% (21/149) of tumours. None of the matched normal tissues showed methylated bands ( Figure 1A). To investigate expression, we applied tissue array method and carried out immunohistochemistry in formalin-fixed gastric carcinomas. MGMT protein was normally expressed in the nucleus of most parenchymal and stromal cells (Figure 2A). Seventeen cases (17/149, 11.4%) of tumours showed complete loss of MGMT expression ( Figure 2B) and 13 cases of these (76.5%) were methylated in promoter region. Out of the 132 tumours with MGMT expression, eight tumours (6.1%) showed methylation, and among these eight cases, three showed loss of MGMT expression in the focal area of the tumour ( Figure 2C). In chi square test, promoter hypermethylation of MGMT was significantly associated with a loss of expression in gastric carcinomas (Table 1, P50.001).

Loss of expression and clinicopathological data in consecutive gastric carcinomas
To investigate the association between the loss of MGMT expression and the clinicopathologic characteristics, we carried out immunohistochemistry using six tissue array blocks containing 315 consecutive gastric carcinomas with the follow-up data. Loss of MGMT expression was found in 13.3% of tumours and was significantly associated with pTNM stage (P=0.037), and tumour invasion (P=0.02). We previously investigated microsatellite status using BAT-25 and BAT-26 markers in 315 gastric cancers (accepted in Modern Pathol). Loss of MGMT expression was significantly associated with microsatellite instability (P=0.041) ( Table 2). Furthermore, patients with the loss of MGMT expression had a poorer prognosis than those with a normal expression pattern (P=0.01) (Figure 3). In the multivariate analysis, MGMT did not demonstrate a correlation with survival, when simultaneously assessed with age, stage, histology, and lymph node metastasis (P=0.16).

Promoter methylation and loss of expression in gastric cancer cell lines
Promoter methylation was detected in the SNU-620 and SNU-719 cell lines. The SNU-620 cell line contained only methylated allele but SNU-719 contained both unmethylated and methylated alleles ( Figure 1B). To compare MGMT expression vs methylation, Western blot analysis and RT -PCR were performed. The protein and the mRNA of MGMT were absent in the SNU-620 cell line ( Figure 4A,B). To confirm the methylation status of the CpG site, we performed bisulphite genomic sequencing of the nt 7128 to 744 promoter region. In the SNU-620 cell line, which contained only the methylated allele, 9 of 14 CpG sites (7125, 7122, 7106, 790, 774, 770, 763, 752, and 746) were found to be completely methylated (data not shown). The significance of the methylation status upon MGMT expression in vitro was confirmed by adding a demethylating agent to SNU-620, which does not express MGMT mRNA. After 10 days of treatment with 10 uM 5-aza-2'-deoxycytidine, MGMT mRNA was detected by RT -PCR ( Figure 3C).

DISCUSSION
Normal tissues exhibit methylation in CpG sites in exons randomly and global hypomethylation occurs in these methylated CpG sites during carcinogenesis (Toyota et al, 2000). In contrast, CpG sites in the promoter regions of many tumour suppressor genes are normally unmethylated but are heavily methylated in cancer tissue (Esteller, 2000b). In the present study, promoter methylation was detected in 14.1% of gastric carcinomas and a loss of MGMT expression was found in 11.4 -13.3% of tumours. Our result is lower than those of Oue et al (2001) or Park et al (2001) but is similar to those of Esteller et al (2001a). The study by Park et al (2001) showed that there was no significant difference in the protein expression between cases with MGMT methylated and unmethylated tumours. But in our results, as well as the study by Oue et al (2001), there were significant differences associated between the loss of MGMT protein expression and promoter methylation (P50.001). Among the cases with a loss of MGMT expression, 76.5% were methylated while 6.1% of tumours were methylated in cases showing MGMT expression. Furthermore in tumours showing both MGMT expression and promoter methylation, three cases showed loss of expression in the focal area of the tumour, i.e., 1 or 2 glands in intestinal type or a small area in diffuse type ( Figure 2C). Because MGMT protein was expressed in more than 10% of the above cases, they were regarded positive. Due to high sensitivity of methylation-specific PCR, the methylated band may be observed  even if the methylated DNA represents only a minor portion. In the 315 consecutive cases examined, 12 cases showed focal loss of MGMT expression and eight of these cases were at stage I or II.

Molecular and Cellular Pathology
Therefore, it appears that during carcinogenesis, methylated cancer cells in a small region may expand clonally, and subsequently dominate the whole tumour tissue.
To investigate an association between MGMT expression and clinicopathological characteristics, we additionally investigated MGMT expression in consecutive cases and compared the results with clinicopathologic data. Cases with loss of MGMT expression were associated with advanced gastric cancer (P=0.02) and microsatellite instability (P=0.041) and cases with MGMT expression were associated with longer survival (P=0.01).
Recently, several studies concerning the relations between MGMT and MSI was reported (Whitehall et al, 2001;Laiho et al, 2002) and Yamamoto et al (2002) revealed that the sporadic colorectal cancers with MSI-H showed frequently methylation pattern in various genes and showed MGMT promoter methylation in 23% of the cases. In our gastric cancer cases, the loss of MGMT expression is significantly associated with microsatellite instability (P=0.041) as well as with hMLH1 expression (P=0.002, data not shown). Our results may suggest that cancers with simultaneous methylation of CpG islands, so-called CpC islands methylator phenotype or CIMP+, may demonstrate the promoter methylation of both MGMT and hMLH1 genes. In addition, cells with loss of MGMT expression are left unprotected from mutagens and may predispose to additional mutations in the genes involved in the progression of tumours which may lead to advanced stage and poor survival.
In SNU-620 cell line, which contains only methylated allele, loss of MGMT protein and mRNA expression were reversed with treatment of 5-aza-2'-deoxycytidine, a demethylating agent. It may support that the promoter methylation of MGMT could silence the protein expression in gastric cancer cells.
In summary, we suggest that during gastric carcinoma progression, loss of expression at the MGMT gene is frequently caused by promoter hypermethylation of CpG islands and that this is significantly associated with tumour progression and prognosis.