Genome-wide mapping and characterization of hypomethylated sites in human tissues and breast cancer cell lines

Abstract

We have developed a method for mapping unmethylated sites in the human genome based on the resistance of TspRI-digested ends to ExoIII nuclease degradation. Digestion with TspRI and methylation-sensitive restriction endonuclease HpaII, followed by ExoIII and single-strand DNA nuclease allowed removal of DNA fragments containing unmethylated HpaII sites. We then used array comparative genomic hybridization (CGH) to map the sequences depleted by these procedures in human genomes derived from five human tissues, a primary breast tumor, and two breast tumor cell lines. Analysis of methylation patterns of the normal tissue genomes indicates that the hypomethylated sites are enriched in the 5′ end of widely expressed genes, including promoter, first exon, and first intron. In contrast, genomes of the MCF-7 and MDA-MB-231 cell lines show extensive hypomethylation in the intragenic and intergenic regions whereas the primary tumor exhibits a pattern between those of the normal tissue and the cell lines. A striking characteristic of tumor cell lines is the presence of megabase-sized hypomethylated zones. These hypomethylated zones are associated with large genes, fragile sites, evolutionary breakpoints, chromosomal rearrangement breakpoints, tumor suppressor genes, and with regions containing tissue-specific gene clusters or with gene-poor regions containing novel tissue-specific genes. Correlation with microarray analysis shows that genes with a hypomethylated sequence 2 kb up- or downstream of the transcription start site are highly expressed, whereas genes with extensive intragenic and 3′ untranslated region (UTR) hypomethylation are silenced. The method described herein can be used for large-scale screening of changes in the methylation pattern in the genome of interest.