Rethinking how DNA methylation patterns are maintained

Abstract

DNA methylation patterns are set up early in mammalian development and are then copied during the division of somatic cells. A long-established model for the maintenance of these patterns explains some, but not all, of the data that are now available. We propose a new model that suggests that the maintenance of DNA methylation relies not only on the recognition of hemimethylated DNA by DNA methyltransferase 1 (DNMT1) but also on the localization of the DNMT3A and DNMT3B enzymes to specific chromatin regions that contain methylated DNA.

Figure 1. Current model for the establishment and inheritance of DNA methylation patterns

The current model for the establishment and inheritance of DNA methylation patterns relies heavily on the original hypotheses of Riggs and Holliday and Pugh as exemplified in several recent reviews,. The basis of this model is that DNA methylation patterns are established in germ cells and in developing embryos by the activity of de novo DNA methyltransferases DNMT3A and DNMT3B. Subsequently, methylation patterns (● methylated CpG site; ○ unmethylated CpG site) are inherited after DNA replication primarily due to the activity of DNMT1 which has a preference for hemi-methylated sites generated through DNA synthesis. The concept is that the enzyme is copying a pattern which is present on naked DNA. The model seems adequate in principle, however its failure to incorporate a correction mechanism or the constraints of chromatin structure suggests that additional factors are involved in copying DNA methylation patterns in mammalian cells.

Figure 2. Revised model for the maintenance of DNA methylation patterns

In this revised model the DNMT3A and 3B enzymes remain bound to chromatin in somatic cells, and in particular to nucleosomes that contain methylated CpG sites (A). When DNA replicates (B) the bulk of methylation copying is achieved by DNMT1, which is the predominant DNA methylase in the cell and is localized to the replication fork by PCNA, and possibly by UHFR1. Soon after DNA replication we propose that DNMT3A and 3B complete the methylation process and correct errors left by the DNMT1 enzyme. These enzymes do not “read” the parental strand for DNA methylation patterns but rather methylate newly replicated CpG sites which are unmethylated because the are compartmentalized to the chromatin region containing methylated DNA. In this way, the enzymes function similarly to other chromatin modifying enzymes such as EZH2 which are also localized to their product after methyl transfer. As described in the text, DNMT3A and 3B seem to be compartmentalized to CpG islands and repetitive elements whereas DNMT1 has a preferential ability to maintain the majority of DNA methylation in the cell which is found at non-CpG islands. UHFR1 might help locate sites missed during the replication process and be responsible for the delayed methylation seen after the DNA has left the replication fork.