Mechanisms of transcriptional repression by DNA methylation

Modification at the fifth carbon of cytosines (5mC) in the context of CpG (cytosine-phosphate-guanin) dinucleotides is a wide-spread DNA modification that is essential for mammalian development. DNA methylation is associated with transcriptional repression and required for silencing of evolutionary young repetitive elements and some CpG rich promoters in somatic cells. However, how DNA methylation translates into transcriptional repression remains enigmatic. Accumulating in vitro evidence suggests that DNA methylation can directly repel TF binding by motif methylation, although in vivo evidence remains scarce. An indirect repression model suggests that methyl-CpG binding domain (MBD) proteins recognize densely methylated DNA and cause transcriptional repression by the interaction with co-repressors. Importantly, a direct and indirect repression model are not mutually exclusive and it is likely that both mechanisms are at play. However, individual or combinatorial deletions of some MBD proteins during mouse embryogenesis, do not result in a phenotype that is associated with the loss of DNA methylation. While it has been argued that this could be explained by functional redundancy between MBD proteins, genetic evidence is missing, as a simultaneous deletion of all MBD proteins has not been reported to date. Here, we test the indirect repression model by deleting all 5mC binding MBD proteins in different mammalian cell lines. We show that upon deletion, we do not detect upregulation of methylated CpG-rich promoters found for instance in germline-specific genes or repetitive elements in mouse embryonic stem cells or derived neurons. In contrast, neurons that lack DNA methylation die after several days in culture, which is associated with strong de-repression of repetitive elements. Mouse ES cells tolerate the loss of DNA methylation and do not show strong upregulation of TEs, which is attributed to an alternative repression mechanism involving H3K9me3. We suggest that absence of this mark in neurons explains the essentiality of DNA methylation for repeat repression in somatic cells.
In order to further dissect a DNA methylation mediated repression model, we explore the TF repertoire of young transposable elements that are highly de-repressed in methylation deficient neurons. By systematically investigating CREB1, we provide evidence that this regulation entails the direct repulsion of methylation-sensitive TFs.
In the second part of this thesis, we investigate MBD proteins outside the murine lineage. Therefore, we comprehensively delete all MBD proteins in a human cancer-derived cell line and contrast this to cells with a hypomethylated genome. This reveals, in line with the observations in mouse cells, a minor role of MBD proteins in translating DNA methylation into transcription repression. Taken together, this work provides evidence that the dominant mode of gene and repeat repression is the direct repulsion of transcription factors by DNA methylation.