We have developed a new technology that provides whole-genome methylation profiles. The new technology has large advantages in coverage, simplicity, accuracy, and economy over other methods. The approach combines a new method of enzymatic fractionation of DNA into methylated and unmethylated compartments and the application of emerging technologies in ultrahigh throughput DNA sequencing to profile genomic methylation patterns. We will determine sites of demethylation and de novo methylation in the genomes of a panel of cell lines derived from mammary carcinomas and of a large panel of primary mammary carcinomas that are well-characterized in terms of histopathology and expression of molecular markers. Whole-genome methylation profiling of DNA of 2 ductal carcinomas and adjacent normal tissues has been completed, as has a complete methylation profile of the mammary carcinoma cell line MCF7. We will then use fully-implemented computational methods to identify those methylation abnormalities that are most highly correlated with type, grade, and stage of mammary carcinomas. The goal is a set of diagnostic and prognostic criteria that can be used in the development of biomarkers for mammary carcinoma. A diagnostic procedure will be developed that will test for de novo methylation and demethylation of the sequences found to be most often hypermethylated in mammary carcinoma, and a simple and novel method for the evaluation of methylation abnormalities in biopsy DNA is described. Our preliminary results, together with some long-standing enigmas in the literature, lead us to propose that the methylation abnormalities so common in mammary carcinoma may reflect the operation of a system (the methylation suicide system) that kills cells that have lost growth control.
We will perform the first genome-wide methylation profiling of the human genome to catalog the methylation abnormalities that occur in mammary carcinoma. Our methylation profiling method is superior to competing methods by an order of magnitude and is the only method that can address the methylation status of repeated sequences. We have already profiled two mammary carcinoma genomes (one of a luminal A type and one of a basal-like type) and have found that tumor suppressor genes are no more likely than other genes to acquire methylation over their promoters. We provide preliminary evidence in the form of many millions of sequences to both provide a robust, economical, and state of the art solution to whole-genome methylation profiling. The preliminary results cause us to challenge the prevailing epimutation model, which ascribes a major role to de novo methylation of tumor suppressor promoters in human cancer. We also develop the first genome-wide and unbiased approach to the discovery of biomarkers that will be useful in the early detection of mammary carcinoma.