Reiterative de novo methylation maintainsmethylation levels in somatic cells

Abstract

DNA methylation is a pervasive epigenetic mark in normal cells. DNA methylation abnormalities are a fundamental hallmark of cancer that can promote carcinogenesis. DNA methylation is lost specifically in heterochromatic regions in tumours. These hypomethylated regions are termed partially methylated domains (PMDs) and replicate during late S-phase. The late replication of PMDs has been proposed to play a key role in their hypomethylation. Specifically, it has been suggested that PMDs passively lose methylation due to incomplete maintenance of methylation after consecutive cell divisions. This model directly implicates DNMT1 as the maintenance methyltransferase and suggests that it does not have enough time to fully methylate late replicating regions. I aimed to elucidate how PMDs become hypomethylated during tumorigenesis and address this ‘passive loss’ model. I investigated the levels and patterns of DNA methylation in HCT116 colorectal cancer cells and their DNMT1 Knock-Out (DNMT1KO) derivatives. I identified that PMDs show distinct hypomethylation in HCT116 cells, depending on their heterochromatic state. Constitutive heterochromatic PMDs, marked by H3K9me3, showed more pronounced hypomethylation than facultative ones, marked by H3K27me3. In DNMT1KO cells, I observed global loss of methylation levels. However, hypomethylation was particularly prominent within PMDs, suggesting that hindering DNMT1 activity led to poorer maintenance of methylation, in agreement with the model. I also observed a subgroup of PMDs that were predominantly marked by H3K9me3 and bordered by H3K27me3 in HCT116 cells, which unexpectedly showed increased methylation levels in DNMT1KO cells. These hypermethylated PMDs remain late replicating in DNMT1KO cells despite their high methylation. However, these regions were no longer marked by H3K9me3 and H3K27me3 in DNMT1KO cells, indicating the loss of their heterochromatic state. Finally, using ChIP, I identified that DNMT3A and DNMT3B were not recruited in constitutive and facultative heterochromatic regions. DNMT3A, but not DNMT3B, recruitment was detected in these hypermethylated PMDs in DNMT1KO but not HCT116 cells, aligning with the loss of the heterochromatic marks in the hypermethylated PMDs. Taken together, my results suggested that hypermethylated PMDs in DNMT1KO cells could maintain high methylation levels, despite their late replication timing, due to the recruitment of DNMT3A. More generally, this suggested that de novo DNMTs play an important role in maintenance of methylation levels via reiterative de novo methylation, while highlighting that chromatin environment and its role in DNMT recruitment might play a more important role than replication timing in the hypomethylation observed in cancers.