LINE-1 Hypomethylation in Cancer Is Highly Variable and Inversely Correlated with Microsatellite Instability

Marcos R. Estecio(The University of Texas MD Anderson Cancer Center), Vazganush Gharibyan(The University of Texas MD Anderson Cancer Center), Lanlan Shen(The University of Texas MD Anderson Cancer Center), Ashraf E.K. Ibrahim(University of Cambridge), Ketan D. Doshi(University of Minnesota), Rong He(The University of Texas MD Anderson Cancer Center), Jaroslav Jelı́nek(The University of Texas MD Anderson Cancer Center), Allen S. Yang(Southern California University for Professional Studies), Pearlly S. Yan(Cancer Genetics (United States)), Tim H-M. Huang(Cancer Genetics (United States)), Eloíza H. Tajara(Faculdade de Medicina de São José do Rio Preto), Jean‐Pierre J. Issa(The University of Texas MD Anderson Cancer Center)
PLoS ONE
May 1, 2007
Cited by 265Open Access
Full Text

Abstract

BACKGROUND: Alterations in DNA methylation in cancer include global hypomethylation and gene-specific hypermethylation. It is not clear whether these two epigenetic errors are mechanistically linked or occur independently. This study was performed to determine the relationship between DNA hypomethylation, hypermethylation and microsatellite instability in cancer. METHODOLOGY/PRINCIPAL FINDINGS: We examined 61 cancer cell lines and 60 colorectal carcinomas and their adjacent tissues using LINE-1 bisulfite-PCR as a surrogate for global demethylation. Colorectal carcinomas with sporadic microsatellite instability (MSI), most of which are due to a CpG island methylation phenotype (CIMP) and associated MLH1 promoter methylation, showed in average no difference in LINE-1 methylation between normal adjacent and cancer tissues. Interestingly, some tumor samples in this group showed increase in LINE-1 methylation. In contrast, MSI-showed a significant decrease in LINE-1 methylation between normal adjacent and cancer tissues (P<0.001). Microarray analysis of repetitive element methylation confirmed this observation and showed a high degree of variability in hypomethylation between samples. Additionally, unsupervised hierarchical clustering identified a group of highly hypomethylated tumors, composed mostly of tumors without microsatellite instability. We extended LINE-1 analysis to cancer cell lines from different tissues and found that 50/61 were hypomethylated compared to peripheral blood lymphocytes and normal colon mucosa. Interestingly, these cancer cell lines also exhibited a large variation in demethylation, which was tissue-specific and thus unlikely to be resultant from a stochastic process. CONCLUSION/SIGNIFICANCE: Global hypomethylation is partially reversed in cancers with microsatellite instability and also shows high variability in cancer, which may reflect alternative progression pathways in cancer.


Related Papers