Computational predictions of 3D chromatin many-body interactions reveal roles of chromatin hubs in gene expression

Chromatin 3D organization plays important roles in the regulation of gene expression. Hi-C measurements provide a projection of 3D chromatin organization onto a 2D heatmap, which depicts the population-averaged pairwise contact frequencies between all genomic elements. These 2D heatmaps highlight structural units of chromatin organization such as TADs. As distal cis-regulatory elements such as enhancers interact with promoters to increase the transcriptional output of target genes in a tissue-specific manner, quantifying cell-type specific chromatin spatial organization is therefore critical. However, TADs have been found to be largely invariant across cell types, suggesting that features of 2D Hi-C heatmaps are insufficient to gain understanding of tissue-specific gene expression. Another challenge is when multiple genomic elements are involved: It is not possible to obtain higher-order many-body interactions from Hi-C data. In order to address these problems, we apply CHROMATIX, a computational method which reconstructs ensembles of single-cell chromatin conformations by deconvolving Hi-C data. We reconstructed large ensembles (2×104) of independent 3D single-cell chromatin conformations which reproduce imaging, Dip-C, and 3D FISH single cell measurements. The reconstructed ensembles of single-cell chromatin structures further enable us to uncover statistically significant 3D hubs of higher-order many-body interactions. We uncovered chromatin hubs of many-bodies across hundreds of loci in B-lymphocyte cells. We illustrate the functional roles of these 3D hubs of many-body interactions in regulating gene expression. Using the example of eQTL associated genes, we show that higher-order 3D hubs provide the physical basis of differential gene expression, with detailed examples of hubs of 1-gene-to-many-eQTLs and many-genes-to-1-eQTL. Overall, our results show 3D chromatin many-body hubs are important for regulating gene expression, and they can be discovered through CHROMATIX computational modeling single-cell 3D chromatin conformations using population Hi-C data.