Background: Breast cancer is the most prevalent neoplasia diagnosed in women in the Western world. Recent genomic studies are proving invaluable in discovering the underlying basis of this disease; however, further mechanisms for tumorigenesis exist. Changes to DNA methylation during cancer progression are well established. For example, decreases in global methylation levels are correlated with genomic instability, whereas increases in locus-specific methylation levels are associated with pathological silencing of tumour suppressor genes. Though the importance of DNA methylation in normal development and disease is known, very little is understood about its genomic distribution in normal and tumour cells. Methods: DNA was obtained from ten pairs of breast cancer cell lines and their patient-matched normal lymphoblasts. DNA was fragmented by sonication and samples were split into immunoprecipitation (IP) and input (IN) samples. The methylation-dependent immunoprecipitation (MeDIP) technique, in which anti-5-methylcytidine antibodies enrich methylated DNA fragments, was then applied to the IP samples. IN and IP DNA samples were then differentially labelled with fluorescent dyes and were competitively hybridized to the sub-megabase resolution tiling-set (SMRT) array. The SMRT array contains ∼27,000 BACs that contiguously span the human genome at an average resolution of 80 kb. Slides were scanned and analyzed using the ArrayWoRxe Auto Biochip Reader and the resulting images were analyzed using SoftWoRx software. Data was aligned and visualized using custom SeeGH software. Results: Epigenomic profiles were obtained for both breast cancer cell lines and matched normal lymphoblasts. Multiple differentially methylated regions were observed on each chromosome. They range in size from ∼100 kb to whole cytobands. Regions of recurrently altered DNA methylation were identified in at least half of the samples. Specifically, over 100 discrete regions were found to be recurrently hypermethylated in the breast cancer genomes and over 80 regions were hypomethylated. Many hypermethylated regions contain the promoter region of one gene, for example a 200 kb region on 1p33, whereas many hypomethylated regions are gene-poor. Conclusions: This study describes the first high-resolution global methylation profiles of matched normal and malignant breast cancer samples. Using this technique, we have identified multiple novel recurrent regions of epigenetic alterations in breast cancer. These regions may harbour oncogenes or tumour suppressor genes, or may contribute to the general genomic instability that drives tumourigenesis. These findings will be important in the design of novel epigenetic drug therapies.

[Proc Amer Assoc Cancer Res, Volume 47, 2006]