PRC1 organizes 3D chromatin architecture in mouse ES cells

Abstract

Polycomb is a major epigenetic pathway involved in developmental gene regulation. While currently it is not understood how Polycomb Repressive Complexes (PRCs) silence genes, regions of the genome bound by polycomb are highly compacted, and cluster together in nuclear space. An appealing model is that this structural property of polycomb complexes contributes to gene silencing. To investigate this, here I apply Hi-C to analyse chromatin organization in 3D in mouse embryonic stem cells (mESCs) lacking RING1B, a core component of PRC1, and in cells with impaired catalytic function of RING1B. Hi-C is a molecular method coupled to high-throughput sequencing that can interrogate the 3D organisation of genomes. The main approach used to quantify enrichment interactions in Hi-C datasets is creation of so-called pile-ups: averaging of multiple 2D regions of the Hi-C map to visualize the average profile of interactions between features of interest. Pileups are a crucial analytical approach, however there was no convenient tool available for the task. Therefore, first I developed my own versatile tool - coolpup.py. I show that Coolpup.py works for extremely sparse single-cell Hi-C data. Moreover, I have used coolpup.py to discover a novel dynamic pattern of polycomb-associated loops during cell cycle progression that is a completely different pattern from CTCF-mediated loops, with most prominent polycomb-mediated interaction enrichment occurring just before and just after mitosis. Second, using computational analysis including application of coolpup.py, I describe properties of polycomb-mediated chromatin structures and the role of PRC1 in creating them, in mouse embryonic stem (ES) cells. Using a RING1B knock-out line, I showed that the PRC1 complex creates both local compaction of its targets, and distal interaction between them. Using cells with a RING1B I53A point mutation which has greatly impaired E3 ubiquitin ligase activity, my data suggest that there is no direct role of the H2AK119 ubiquitination in formation of 3D chromatin structures. I investigate what determines formation of distal interactions between RING1B binding sites. Surprisingly, distance turned out not to be an important factor, since I detect enriched interactions even at distances as large as 50-100 Mbp. I detect a clear role of canonical PRC1 binding, unlike PRC2 or variant PRC1 complexes in forming these interactions. I show that transcriptional activation of polycomb targets in cells lacking RING1B is not required for loss of interactions between them. Finally, I have investigated 3D genome re-organization in ES cells grown in conditions of ground state pluripotency in “2i” medium. I have shown that 2i conditions lead to a depletion of chromatin compaction and looping mediated by polycomb. Moreover, I have shown that the PRC1-mediated interactions, while weakened, are still present in 2i grown cells, unlike the inner cell mass where no interactions can be observed. Overall, I have demonstrated the abundance of PRC1-mediated structures present in the ES cell nucleus, the role of canonical PRC1 complexes in their formation, the plasticity of the 3D genome organization which follow epigenome changes, during both cell cycle progression and upon entering ground state pluripotency.