Abstract
BACKGROUND: Epigenetic reprogramming during early mammalian embryonic and germ cell development is a genome-wide process. CpG islands (CGIs), central to the regulation of mammalian gene expression, are exceptional in terms of whether, when and how they are affected by epigenetic reprogramming. RESULTS: We investigated the DNA sequences of CGIs in the context of genome-wide data on DNA methylation and transcription during oogenesis and early embryogenesis to identify signals associated with methylation establishment and protection from de novo methylation in oocytes and associated with post-fertilisation methylation maintenance. We find no evidence for a characteristic DNA sequence motif in oocyte-methylated CGIs. Neither do we find evidence for a general role of regular CpG spacing in methylation establishment at CGIs in oocytes. In contrast, the resistance of most CGIs to de novo methylation during oogenesis is associated with the motif CGCGC, the recognition site of E2f1 and E2f2, transcription factors highly expressed specifically in oocytes. This association is independent of prominent known hypomethylation-associated factors: CGI promoter activity, H3K4me3, Cfp1 binding or R-loop formation potential. CONCLUSIONS: Our results support a DNA sequence-independent and transcription-driven model of de novo CGI methylation during oogenesis. In contrast, our results for CGIs that remain unmethylated are consistent with a model of protection from methylation involving sequence recognition by DNA-binding proteins, E2f1 and E2f2 being probable candidates.