Genome-wide analysis of aberrant CpG island methylation in human tumors

Aberrant DNA methylation of CpG islands (CGIs) is a common alteration during malignant transformation that leads to the abnormal silencing of tumor suppressor genes and plays a role in disease initiation and progression. The major aim of the present thesis was the implementation of methodologies to identify epigenetic marker genes that can be used for the diagnosis as well as for the targeted treatment of tumors. Furthermore, the molecular mechanisms controlling the methylation status of CpG islands in normal and malignant cells should be analyzed. To address these issues, a novel and robust technique, called methyl CpG immunoprecipitation (MCIp) was developed that allows for the unbiased genome-wide profiling of CpG methylation in DNA samples where quantity is limited. This approach is based on a recombinant, antibody-like protein that efficiently binds native CpG methylated DNA and enables the fractionation of DNA fragments depending on the particular methyl-CpG content. This application facilitates the monitoring of CpG island methylation either on single gene or on genome-wide levels. Initial genome-wide methylation profiling of myeloid leukemia cell lines using 12K CpG island microarrays identified over one hundred genes with aberrantly methylated CpG islands. Interestingly, the comparison with gene expression data revealed that more than half of the identified genes were not expressed in various healthy cell types, indicating that hypermethylation in cancer may be largely independent of the transcriptional status of the affected gene. The majority of individually tested genes were also hypermethylated in primary blast cells from AML patients. The MCIp approach was further optimized and adapted for a more suitable microarray platform (Agilent 244K CGI microarrays). The in-depth comparison of MCIp and MassARRAY for two established cell lines showed an excellent correlation over a set of 140 genes (1,150 amplicons covering approximately 13,500 CpG dinucleotides). In order to identify potential marker genes, global comparative CpG island methylation profiles for more than 25 AML samples (of mostly normal karyotype) and ten patients with colorectal carcinoma using MCIp in combination with microarray were generated. Our comprehensive analysis identified a large array of CGIs that are previously unrecognized targets of hypermethylation in AML. For the identification of potential marker genes, approximately 400 regions were selected based on the array results for screening a large set of 200 AML patients. The data are now ready to be subjected to computational analyses. In order to get insights into the process regulating the methylation status of CpG islands, factors should be identified that are responsible for maintaining or establishing methylated states of CGIs in health and disease as well as for de novo methylation in cancer. De novo motif discovery analysis revealed two repetitive sequence motifs (GAGA, CACA) that were commonly enriched in CpG islands that were methylated in cancer. More strikingly, the global analysis demonstrated a highly significant association of unmethylated CpG islands with consensus sequences for GA binding protein (GABP), specific protein (Sp) 1 and 3, nuclear respiratory factor (NRF) 1, nuclear factor (NF) Y, yin-yang (YY) 1 and an unknown factor in all analyzed samples. Using ChIP-on-chip assays we also showed that most of the identified motifs for Sp1, NRF1 and YY1 were actually bound by the respective factors in normal cells and that these regions did not acquire de novo methylation in leukemia cells. In addition, the data provide global evidence that the stable binding of any of these transcription factors to their consensus motif depends on their co-occurrence with neighboring consensus motifs. Thus, the results of the present thesis suggest a major role for cooperative transcription factor binding in maintaining the unmethylated status of CpG islands in health and disease. The data also implies that the majority of de novo methylated CpG islands are characterized by the lack of sequence motif combinations and the absence of activating transcription factor binding.