CpG islands have been the best-known tourist destination in the epigenome for a long time, but if you’re looking for a change there’s a destination that all the rage: DNA methylation grand canyons. It is well known that DNA methylation at CpG islands is associated with transcriptional repression. However, the precise molecular mechanism linking DNA methylation to chromatin structure and transcription are not fully understood. The Hi-C family of DNA-DNA interaction assays have found that CTCF motifs and the compartmentalization of regions with similar histone modifications are two major mechanisms of chromatin folding. The role that DNA methylation plays in chromatin architecture has not been easy to assess since CpG islands are relatively short (300bp-3kb) and below the resolution of Hi-C technologies.
So, to get a breath of fresh air, the labs of Erez Lieberman Aiden and Margaret Goodell (Baylor College of Medicine, Houston Texas) put on their hiking shoes and explored the role of DNA methylation in genome architecture. During their previous treks of the epigenetic landscape, they identified very long (3.5-25kb) intervals of low cytosine methylation they called “DNA methylation canyons”. These canyons often contain multiple CpG islands and are strongly preserved across cell types and species. They can be marked with H3K27me3 and repressed or marked with H3K4me3 and active. Now, the authors wanted to use DNA methylation canyons as a model to examine the relationship between DNA methylation and 3D chromatin structure. They used primary human Hematopoietic Stem and Progenitor Cells (HSPCs) from umbilical cord blood to generate in situ Hi-C libraries.
Here’s what happened on their voyage:
- They identified ~400 very long-distance interactions, which are much greater in distance than the ~2Mb associated with cohesin loops
- The median distance between interacting loci was 7.5 Mb with the longest being 117 Mb
- They confirmed these features in single-cells via 3D fluorescence in situ hybridization (FISH)
- The sequences at the anchors of the long loops (>3Mb) are enriched for DNA methylation “grand canyons” (>7.3 kb) and depleted for CTCF sites, while the reverse is true in short (<1Mb) loops
- Grand canyons tend to form loops regardless of their linear distance and form links even when located on different chromosomes
- Most (85%) of these long-loop-associated canyons are marked with high H3K27me3 and low H3K27ac
- Inhibition of H3K27me3 deposition using an EZH2 inhibitor reduces grand canyon interactions to 22% their level in untreated HSPCs, suggesting H3K27me3 is necessary for long loop formation
- The long loops were specific to primary stem and progenitor cells: they were not found in other cells types, including immortalized cell lines, but conserved in mouse embryonic stem and neural progenitor cells
The authors hypothesize that DNA methylation grand canyons enable distal interactions in part by guiding deposition of histone marks that then mediate the loop formation. Future work will focus more on the role of these loops in multipotency and other processes in stem cells.
Go trek the CpGenome over at Molecular Cell, May 2020.