3D Genome Structure Variation Across Cell Types Captured by Integrating Multi-omics

Abstract

Background3D genome structure contributes to the establishment or maintenance of cell identity in part by organizing genes into spatial active or inactive compartments. Less is known about how compartment switching occurs across different cell types. Rather than analyze individual A/B compartment switches between pairs of cell types, here, we seek to identify coordinated changes in groups of compartment-scale interactions across a spectrum of cell types.\n\nResultsTo characterize the impact of genome folding on cell identity, we integrated 35 Hi-C datasets with 125 DNase-seq, 244 RNA-seq, and 893 ChIP-seq datasets. We first find physical associations with the nuclear lamina inform the most dramatic changes in chromosome structure across cell types. By examining variations in chromosome structure, transcription, and chromatin accessibility, we further observe that certain sets of correlated chromosome structure contacts also co-vary in transcription and chromatin accessibility. Analyzing ChIP-seq signals, we find that sets of chromosome contacts that form and break in sync tend to share active or suppressive histone marks. Finally, we observe that similar principles appear to govern chromosome structure fluctuations across single cells as were found across cell types.\n\nConclusionOur results suggest that cells adapt their chromosome structures, guided by variable associations with the lamina and histone marks, to allocate up-regulatory or down-regulatory resources to certain regions and achieve transcription and chromatin accessibility variation. Our study shows E-PCA can identify the major variable interaction sets within populations of single cells, across broad categories of normal cell types, and between cancer and non-cancerous cell types.