Systematic DNA methylation analysis of multiple cell lines reveals common and specific patterns within and across tissues of origin.

DNA methylation is a key functional regulatory mechanism in human genome, which plays critical roles in development, differentiation and many diseases. With rapid progress of large-scale projects (e.g. ENCODE), many DNA methylation data across diverse cell lines have been produced. However, common methylation patterns, cell lineage- and cell line-specific DNA methylation patterns across multiple cell lines have not yet been explored completely. Using the DNA methylation data across 54 human cell lines, we identified 35 276 local DNA methylation regions called local clusters of CpG sites (LCCSs). We constructed an LCCS co-methylation network and investigated the common DNA methylation patterns across all cell lines, which reveal two distinct groups in terms of their methylation level and genomic characteristics. We further detected diverse sets of cell lineage-specific high- and low-methylation patterns, which were depleted in promoter, CpG island (CGI) and repeat regions but enriched in gene body and non-CGI regions, especially the CGI shore regions. We discovered that the cell lineage-specific low-methylated LCCSs were enriched with functional transcriptional factor binding motif regions. Moreover, the detected cell line-specific high- and low-methylated patterns show distinct enrichments in cell line-specific chromatin states and functional relevance with the corresponding cell lines.