Insights From Multidimensional Analyses Of The Pan-cancer Promoter DNA Methylome

Promoter DNA methylation marks play critical roles in transcriptional regulation of gene expression,thereby involved in many biological processes.Although multiple“omics” profiles have been available in large cancer cohorts such as TCGA,integrative analysis of the DNA methylome architectures in a pan-cancer manner remains limited.In this study,based on the tumor and normal DNA methylation profiles of 21 cancers and other supportive multi-omics data,we systematically interrogated the DNA methylation dysregulations,the tumor methylome heterogeneities,and the associations between promoter DNA methylation and gene expression.First,TF genes tend to bear more dynamic promoter DNA methylation profiles than other genes,which serves as a potential source of transcriptome heterogeneity in cancers.The two major factors determining the gene expression levels,DNA methylation and CNV,are complementary and,at some level,mutually exclusive.Next,a clustering analysis of the highly heterogeneous DNA methylomes revealed convergence of cancer types such as adenocarcinomas or squamous cell carcinomas,and meanwhile,provided new classifications of the cancer subtypes.Examples showed that such reclassifications of the tumors are often associated to prognostic differences.Thirdly,the comparisons between paired tumor and adjacent normal tissues suggest that the cancer DNA methylomes are associated to,but not directly determined by the tissues of origin.The cancer context-specificity of the DNA methylomes are rather products of largely cancer type dependent dysregulations.Finally,we found unanticipated significant numbers of the non-canonical promoter Cp G sites that are positively correlated with the gene expression.Distribution patterns of these Cp G sites in Cp G islands,TF binding sites,open chromatin regions,Partially Methlated Domains and histone modification landscapes suggested against a pervasive mechanism of transcriptional activation due to m Cp G-dependent binding of TFs.In summary,our deep mining of the highly heterogeneous DNA methylome data in a pan-cancer context generated novel insights into the architecture of cancer epigenetics and provided a series of resources for further investigations.