Science is full of paradoxes, and while we humble epigeneticists have yet to come across a cat that is both alive and dead, we have discovered a mysterious states of DNA methylation! For instance, some CpG sites can be both methylated and unmethylated at the same time, depending on which strand it is located on.
During DNA replication, DNA strands split, and the methylation profile is not written at the same time; rather, a complex containing the maintenance DNA methyltransferase DNMT1 later recognizes the hemimethylated CpG (hemiCpG) site and methylates the nascent strand. But does it always? Exciting new findings from the lab of Victor Corces at Emory University (Atlanta, USA) demonstrate that hemiCpGs actually hang around for multiple cell divisions and have functional roles that involve CTCF.
By using nascent DNA bisulfite sequencing (nasBS-seq) to probe human embryonic stem cells (hESCs) they found:
- The methylation status of almost all CpG sites, and even CpH sites, is faithfully maintained 20 minutes after division
- However, some 2467 hemiCpGs are inherited across multiple cell divisions
- The methylation level of most hemiCpGs is conserved in pluripotent cells but not non-pluripotent cells
- Profiling of the methyltransferases (DNMT1, DNMT3A, DNMT3B) by chromatin immunoprecipitation on nascent chromatin followed by bisulfite sequencing (nasChIP-BS-seq) revealed the details of how the maintenance and de novo methyltransferases transiently bind hemiCpGs that later become fully methylated
- Interestingly, some DNMT3A bound sites remain hemimethylated after multiple cell divisions, suggesting that DNMT3A regulates loci-specific hemiCpGs, while the other de novo methyltransferase (DNMT3B) does not
- nasCHIP-BS-seq of CTCF revealed enrichment of hemiCpGs at regions flanking CTCF binding sites in human and mouse pluripotent cells, which along with the computational analysis of MeCP2 binding, suggests a functional role for hemiCpGs in chromatin architecture
Overall, these findings enable an unprecedented resolution of the dynamics of DNA methylation maintenance while also providing crucial evidence for a functional role of hemiCpGs in chromatin architecture. However, these findings leave us eager to find out all the details of the how hemiCpG sites are chosen in a locus-specific manner as well all their functional consequences during embryonic development and beyond!
Catch the perspective and the rest of this paradoxical paradigm over at Science, March 2018.