CpG island-mediated global gene regulatory modes in mouse embryonic stem cells

Both transcriptional and epigenetic regulations are fundamental for the control of eukaryotic gene expression. Here we perform a compendium analysis of 4200 large sequencing data sets to elucidate the regulatory logic of global gene expression programs in mouse embryonic stem (ES) cells. We define four major classes of DNA-binding proteins (Core, PRC, MYC and CTCF) based on their target co-occupancy, and discover reciprocal regulation between the MYC and PRC classes for the activity of nearly all genes under the control of the CpG island (CGI)-containing promoters. This CGI-dependent regulatory mode explains the functional segregation between CGI-containing and CGI-less genes during early development. By defining active enhancers based on the co-occupancy of the Core class, we further demonstrate their additive roles in CGI-containing gene expression and cell type-specific roles in CGI-less gene expression. Altogether, our analyses provide novel insights into previously unknown CGI-dependent global gene regulatory modes.