Inactivation of the DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation and its relationship to aflatoxin B1-DNA adducts and p53 mutation in hepatocellular carcinoma
Article first published online: 21 NOV 2002
Copyright © 2002 Wiley-Liss, Inc.
International Journal of Cancer
Volume 103, Issue 4, pages 440–444, 10 February 2003
How to Cite
Zhang, Y.-J., Chen, Y., Ahsan, H., Lunn, R. M., Lee, P.-H., Chen, C.-J. and Santella, R. M. (2003), Inactivation of the DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation and its relationship to aflatoxin B1-DNA adducts and p53 mutation in hepatocellular carcinoma. Int. J. Cancer, 103: 440–444. doi: 10.1002/ijc.10852
- Issue published online: 11 DEC 2002
- Article first published online: 21 NOV 2002
- Manuscript Accepted: 3 OCT 2002
- Manuscript Revised: 12 AUG 2002
- Manuscript Received: 24 APR 2002
- NIEHS. Grant Numbers: ES05116, P30ES09089
- Department of Health, Executive Yuan, Republic of China
- epigenetic changes;
- hepatocellular carcinoma;
- CpG island;
- O6-methylguanine-DNA methyltransferase
O6-methylguanine-DNA methyltransferase (MGMT) is a repair protein that specifically removes promutagenic alkyl groups from the O6 position of guanine in DNA. MGMT is transcriptionally silenced by promoter hypermethylation in several human cancers. Methylation-specific PCR (MSP) was used to analyze the MGMT promoter methylation status of 83 hepatocellular carcinomas (HCC) and 2 HCC cell lines (HepG2 and Hep3B). Hypermethylation was detected in 32 of 83 (39%) HCC tissues, but it was not found in either HCC cell line. We also analyzed MGMT expression by immunohistochemical analysis of HCC tissue samples. The presence of aberrant hypermethylation was associated with loss of MGMT protein. The relationship between methylation status and risk factors and tumor markers including environmental exposure to aflatoxin B1 (AFB1), measured as DNA adducts, and status of tumor suppressor gene p53 was also investigated. A statistically significant association was found between MGMT promoter hypermethylation and high level of AFB1-DNA adducts in tumor tissues (OR = 5.05, 95% CI = 1.29–19.73). A significant association was also found between methylation and p53 mutation status (OR = 2.97, 95% CI = 1.09–8.11). These results suggest that epigenetic inactivation of MGMT plays an important role in the development of HCC and exposure to environmental carcinogens may be related to altered methylation of genes involved in cancer development. The role of chemical carcinogens in hypermethylation needs further investigation. © 2002 Wiley-Liss, Inc.
Hepatocellular carcinoma (HCC) is one of the most common human malignant neoplasms with almost 500,000 estimated new cases enumerated worldwide during the most recent year surveyed.1 Its etiology is not completely clear, but strong associations exist with hepatitis B and C virus infection and several dietary or environmental factors, including aflatoxin B1 (AFB1). The molecular pathogenesis of HCC appears to involve multiple genetic aberrations in the molecular control of hepatocyte proliferation, differentiation and death and the maintenance of genomic integrity. This process is influenced by the cumulative activation and inactivation of oncogenes, tumor suppressor genes and other genes.
06-methylguanine-DNA methyltransferase (MGMT) is a 22,000 kd protein that is critical for the rapid reversal of methylation of guanine.2, 3 Alkylation of DNA at the O6 position of guanine is a potentially mutagenic lesion mainly due to the tendency of O6-methylguanine to pair with thymine during replication, resulting in the conversion of a guanine-cytosine to an adenine-thymine base pair.4 The level of MGMT activity varies greatly between species and cell types.5, 6 In humans, liver contains the highest level of MGMT activity.5 The amounts of MGMT protein are decreased in some tumors and tumor cell lines.7, 8 Loss of expression is rarely due to deletion, mutation or rearrangement of the MGMT gene, but methylation of a discrete region of the CpG island of MGMT has been associated with the silencing of the gene in cell lines.9–11 Hypermethylation of normally unmethylated CpG islands in the promoter regions of many genes correlates with loss of transcription.12 Early MGMT inactivation has been linked with gene mutation in some critical growth regulatory genes; K-ras mutations, especially G to A, are more common in colon cancer after MGMT inactivation.13 Inactivation of the MGMT gene by promoter hypermethylation has been found in brain tumors, colon cancer, lung cancer and other cancers.2 Recently, it was reported that MGMT promoter hypermethylation was present significantly more often in tumors with a G to A mutation in p53 than in tumors with other types of p53 mutation or in tumors with wild-type p53.14, 15
In our study, we explored the possible role of inactivation of MGMT in the development of HCC. We had previously analyzed these samples for p53 gene and protein alterations as well as measured DNA damage levels caused by the dietary carcinogen AFB1.16 More recently, we reported a high frequency of promoter hypermethylation of the human ras association domain family 1A (RASSF1A) gene in HCC.17 Thus, we also investigated the relationship between these biomarkers and methylation status.
MATERIAL AND METHODS
Patients population and data on clinical parameters and biomarkers
The study population consisted of 83 consecutive primary HCC patients identified at National Taiwan University Hospital between 1984–1995. Informed consent was obtained from patients and the study was approved by the appropriate institutional review committees. Demographic data are presented in Table I. Clinicopathologic characteristics were obtained from hospital charts. Information on cigarette smoking and alcohol drinking was also extracted from hospital charts using data obtained from interviews carried out by nurses upon hospitalization and only indicate crude categories.
|MGMT methylation (n = 83)||p-value1|
|No (n = 51) n (%)||Yes (n = 32) n (%)|
|MGMT protein expression|
|No||3 (9.1)||30 (90.9)||<0.0001*|
|Yes||48 (96.0)||2 (4.0)|
|<55||23 (59.0)||16 (41.0)||0.66|
|≥55||28 (63.6)||16 (36.4)|
|Male||46 (62.2)||28 (37.8)||0.71|
|Female||4 (50.0)||4 (50.0)|
|Negative||14 (73.7)||5 (26.3)||0.21|
|Positive||37 (57.8)||27 (42.2)|
|Nonsmoker||19 (52.8)||17 (47.2)||0.20|
|Smoker||25 (67.6)||12 (32.4)|
|Never drinker||32 (55.2)||26 (44.8)||0.06|
|Ever drinker||11 (84.6)||2 (15.4)|
|1/2||17 (73.9)||6 (26.1)||0.15|
|3/4||34 (56.7)||26 (43.3)|
Data on AFB1-DNA adducts and p53 were previously reported.16 AFB1-DNA adducts were determined using an immunoperoxidase method with a monoclonal antibody recognizing the stable imidazole ring opened adduct. Staining was independently scored as negative, weakly positive (low) or highly positive (high) by 2 individuals. Discrepancies were reconciled by joint review. p53 protein levels were also determined by an immunoperoxidase procedure with tissues considered positive if >5% of cells were stained. SSCP and sequencing were used to determine p53 mutations. The grading scheme of Edmendson and Steiner was used: G1, well differentiated; G2, moderately differentiated; G3, poorly differentiated; G4, undifferentiated.18
Cell lines and cell culture
Two hepatocellular carcinoma cell lines (HepG2 and Hep3B) were purchased from ATCC (American Type Culture Collection, Rockville, MD) and grown in Eagle's MEM with nonessential amino acid, sodium pyruvate and Eagle's BSS (Life Technologies, Carlsbad, CA) containing 10% fetal bovine serum.
DNA was isolated from frozen tissue specimens and cells as previously described.19 Briefly, tissue was placed in liquid nitrogen and pulverized with a blender. The tissue powder and cells were then lysed with DNA lysing buffer (10 mM Tris, 10 mM NaCl, 0.1% sodium dodecyl sulfate pH 7.9, 200 μg/ml proteinase K); DNA was isolated by RNase treatment, phenol/chloroform extraction and ethanol precipitation.
Methylation-specific PCR (MSP)
MSP was carried out essentially as described previously20 and was based on the principle that treating DNA with sodium bisulfite results in the conversion of unmethylated cytosine residues to uracil. Thus, the sequence of the treated DNA will differ if the DNA is originally methylated vs. unmethylated and is then distinguishable by sequence-specific PCR primers. Bisulfite modification of DNA was conducted using the CpGnome DNA Modification Kit (Intergen, Purchase, NY). Bisulfite treated DNA was amplified using primers specific for either modified or unmodified DNA. Primer sequences for the methylated reaction were 5′-TTT CGA CGT TCG TAG GTT TTC GC-3′ (sense) and 5′-GCA CTC TTC CGA AAA CGA AAC G-3′ (antisense) and for unmethylated modified reaction were 5′-TTT GTG TTT TGA TGT TTG TAG GTT TTT GT-3′ (sense) and 5′-AAC TCC ACA CTC TTC CAA AAA CAA AAC A-3′ (antisense), respectively.2 PCR was conducted using CpG WIZ™Amplification Kit (Intergen) and AmpliTaq Gold polymerase (Perkin-Elmer, Norwalk, CT). A total 35 cycles was used for amplifying tumor DNA, the annealing temperature was 61°C. PCR products were analyzed by 3.5% agarose gel electrophoresis and ethidium bromide staining. The cell line SW480, known to be methylated at the MGMT gene locus, was used as a positive control and water was used as negative control.
MGMT immunohistochemistry detection
Sections of formalin-fixed, paraffin-embedded tissue were deparaffinized. Immunohistochemistry staining was performed using an immunoperoxidase method (ABC method) with a mouse anti-MGMT monoclonal antibody (clone mT3.1; NeoMarkers, Fremont, CA) at a 1:100 dilution. This antibody has been shown to correlate with O6-alkylguanine-DNA transferase activity.2, 21 A colon carcinoma (LS174T) section was supplied by NeoMarkers as a positive control. Only nuclear staining was regarded as positive staining.
Distribution of demographic characteristics, risk factors for HCC and MGMT protein expression were initially compared between subjects with and without methylation of MGMT using χ2 tests. To evaluate the hypothesis that risk factors of HCC are associated with the methylation status of the MGMT gene, odds ratios (OR) were estimated using unconditional logistic regression models. ORs were estimated for AFB1-DNA adduct level in tumor and adjacent nontumor tissue, p53 mutation and p53 expression, adjusting for potential confounding variables (age at diagnosis and gender). Analyses were also performed by including the other relevant risk factors in the model to assess whether effect estimates changed appreciably.
MGMT promoter hypermethylation
DNA samples obtained from 83 HCC patients and 2 HCC cell lines were analyzed for MGMT promoter methylation by MSP. To determine the specificity of this approach, the human colon carcinoma cell line SW480, previously reported to be methylated at the MGMT locus, was first analyzed and then used as a positive control with every assay.
MGMT promoter methylation was detected in 32 of 83 (39%) HCC DNA samples. Representative examples of the gel analysis of bisulfite-treated DNA samples amplified with methylated- and, as a control for the bisulfite modification process, nonmethylated-specific primers are shown in Figure 1. The 2 HCC cell lines (HepG2 and Hep3B) were negative for promoter methylation. All tissue samples were found to have amplifiable sequences, demonstrating the success of the bisulfite modification process. A prior study of other tumor cell lines2 as well as results on the 2 HCC cell lines investigated here (Fig. 1) demonstrated that the MGMT CpG island DNA is either completely methylated or completely unmethylated. In contrast, for the HCC samples, amplification with the hypermethylation-specific primers was accompanied by amplification with the unmethylated-specific primers as well (Fig. 1). The presence of this unmethylated MGMT DNA could indicate the presence of normal tissue with unmethylated MGMT alleles in these nonmicrodissected samples. However, heterogeneity in the patterns of methylation in the tumor itself might also be present. Immunohistochemistry staining was used address this point.
Expression of MGMT in HCC tissue samples
Representative examples of MGMT protein expression, determined by immunohistochemical staining, are shown in Figure 2. Forty-eight of 51 unmethylated tumors demonstrated positive nuclear staining with anti-MGMT antibody; 30 of 32 methylated tumors showed loss of expression of MGMT (Table I). Immunostaining results were strongly correlated (p < 0.0001) with MGMT methylation status. As in a prior study,2 variation in the levels of expression in tumor cells was observed in most HCC cases.
Relationship between MGMT methylation and clinical parameters and biomarkers
The distributions in methylation status of MGMT by age, gender, HBsAg, smoking status, alcohol drinking status and tumor grade are given in Table I. Only among never alcohol drinkers was there a slight but not significant association with positive methylation status and never drinking (p = 0.06).
The association between MGMT methylation status and levels of AFB1-DNA adducts and p53 status by both mutant protein expression and gene mutation was also investigated. ORs, adjusted for gender and age, are shown in Table II. Methylation in MGMT was more frequent in subjects with low and high levels of AFB1-DNA adducts in tumor tissues compared to those without detectable levels of adducts (ORs = 1.48, 95% CI = 0.51–4.27 and 5.05, 95% CI = 1.29–19.73, respectively). There was no association with adducts in adjacent nontumor tissues.p53 gene mutations were also more frequent in subjects with methylated MGMT (OR = 2.97, 95% CI = 1.09–8.11). There were no statistically significant correlations between MGMT methylation and HBsAg antigen status, smoking status, alcohol drinking, tumor grade or RASSF1A methylation (data not shown). We also estimated the ORs with further adjustment for HBsAg antigen status, as well as cigarette smoking status; however, the estimates did not change appreciably (results not shown).
|MGMT methylation (n = 83)|
|No (n = 51) n (%)||Yes (n = 32) n (%)||Adjusted ORs1 (95% CI)|
|AFB1-DNA (tumor tissue)2|
|Nondetectable||22 (73.3)||8 (26.7)||1.00|
|Low||24 (63.2)||14 (36.8)||1.48 (0.51–4.27)|
|High||5 (33.3)||10 (66.7)||5.05 (1.29–19.73)*|
|AFB1-DNA (adjacent nontumor tissue)2|
|Nondetectable||25 (58.1)||18 (41.9)||1.00|
|Low||20 (60.6)||13 (39.4)||0.87 (0.34–2.21)|
|High||2 (100.0)||0 (0.0)||–|
|p53 protein expression2|
|No||33 (63.5)||19 (36.5)||1.00|
|Yes||18 (58.1)||13 (41.9)||1.22 (0.49–3.04)|
|p53 gene mutation3|
|No||40 (69.0)||18 (31.0)||1.00|
|Yes||10 (41.7)||14 (58.3)||2.97 (1.09–8.11)*|
Epigenetic inactivation through promoter hypermethylation has been linked both directly and indirectly to several events in neoplasia.12 Numerous examples of aberrant CpG islands promoter hypermethylation have been observed in tumor suppressor, cell-cell adhesion and DNA repair genes.22MGMT promoter hypermethylation and absence of MGMT activity were reported in several human tumors including colon and lung cancers, glioma, lymphoma and gastric carcinoma.2, 6, 8, 23, 24, 25, 26 In our study, MGMT hypermethylation was present in 32 of 83 (39%) HCC cases, but it was not found in 2 HCC cell lines. This frequency is lower than that observed in secondary glioblastoma (75%),27 but similar to that in colon cancer (38%)2 and gastric carcinoma (31%).26
Aberrant MGMT methylation has been associated with loss of mRNA expression,13 lack of MGMT protein2 and loss of enzymatic activity.28 Our data demonstrate that there was loss of MGMT expression in 33 HCC cases (40%). Consistent with these results, a previous study of enzyme activity in HCC found that 6 of 21 (30%) tumors had >3-fold less activity compared to adjacent normal tissues.29 While MGMT methylation correlated well with MGMT expression, 5 samples (6%) gave discordant data (Table I). Because tissue sections were not microdissected, detection of MGMT hypermethylation may be affected by contamination of tumor with adjacent nontumor tissue cells. Immunostaining of small pieces of tissues also limits the detection of protein expression and may also help to explain discordant data. A discrepancy between MSP and immunohistochemistry detection was reported previously.2, 15 In general, the findings are consistent with 2 previous studies in other types of tumors that also used both MSP and immunohistochemistry2, 30 and implicate MGMT hypermethylation and loss of expression as a contributor to malignant progression in HCC. To the best of our knowledge, this is the first demonstration of detection of MGMT promoter hypermethylation in human hepatocellular carcinoma.
Previous studies have shown that MGMT methylation is often associated with p53 mutation. Promoter methylation was strongly associated with G to A mutations in p53 in brain27 and nonsmall cell lung cancers.15 Our results demonstrate that MGMT methylation is significantly correlated with p53 mutation in HCC. But MGMT methylation was not associated with p53 mutation status in gastric carcinoma.26 The relationship between methylation status and p53 mutations is controversial and the detailed mechanisms behind this phenomenon are not clear.
Cigarette smoking is known to increase MGMT expression in both normal and neoplastic lung tissue, suggesting that MGMT may protect the lung from carcinogen-induced guanine alkylation.31, 32 However, MGMT inactivation may be an early event in tumorigenesis conferring an increased susceptibility to nitrosamines and other alkylating agents.15 It was previously demonstrated that cytosine methylation enhances guanine alkylation by a variety of bulky carcinogens (benzo(a)pyrene diol epoxide, AFB1 8,9-epoxide, benzo(g)chrysene diol epoxide and N-acetoxy-2-acetylaminofluorene).33 In addition, many types of DNA damage (including oxidative stress, alkylation of bases, photodimer, abasic sites, etc.) interfere with the ability of mammalian DNA to be methylated at CpG dinucleotides by DNA-methyltransferase. Thus, the role of methylation in chemical carcinogenesis is complex. The process of carcinogenesis has frequently been associated with increased expression of DNA-MTase activity accompanied by either hypermethylation or hypomethylation of target cell DNA.34 A variety of chemicals can alter the extent of DNA methylation including AFB1.35
In our previous study, a statistically significant association was found between RASSF1A methylation status and the level of AFB1-DNA adducts in HCC tumor samples. In our study, there was also a significant correlation between AFB1-DNA adduct levels and MGMT hypermethylation in the same HCC tumor tissues. These findings suggest that exposure to chemical carcinogens, as measured by detectable levels of AFB1-DNA, may result in alteration of methylation status in several genes. However, a limitation of our study was that this association was stronger in tumor tissue than in adjacent nontumor tissue. The mechanism by which chemical carcinogens can result in altered methylation is not known.
In summary, our present study demonstrated a high frequency of MGMT hypermethylation in HCC samples. Epigenetic changes by methylation of CpG islands in the MGMT promoter region may play an important role in hepatocarcinogenesis. In HCC, MGMT hypermethylation was correlated with p53 tumor suppressor gene mutation. We also found a significant association between MGMT hypermethylation and AFB1-DNA adducts level in tumor tissue. The biologic basis and mechanisms of the relationship between gene inactivation by methylation and exposure to chemical carcinogens is still not clear at the present time. Further investigation of the relationship between carcinogen exposure and changes in methylation status and gene-gene interaction is needed.
We thank Dr. D. Kennedy and Dr. M. Agrawal for their very helpful assistance on our manuscript.
- 18AJCC. American joint committee on cancer AJCC cancer staging manual. Philadelphia: Lippincott Williams & Wilkins, 1997.