Topology-informed regulatory element collections coordinate cell identity gene expression programs[scATAC-Seq]

Cis-regulatory elements (CREs) contact target genes to regulate their transcription, but CRE interactions are complex and predicting their function is difficult. Contemporary approaches to predict CRE targets generally assume one-to-one associations and linear genomic proximity. These methods fail to account for long-range CRE contacts caused by folding of the genome, or for circumstances in which a single CRE regulates multiple genes or, conversely, a single gene is regulated by multiple CREs. To address these challenges and investigate CRE accumulation, we developed BOUQUET, an integrative, graph-theory-based approach that utilizes multiple aspects of genome topology to probe communities of CREs, their bound apparatus, and their target genes. BOUQUET discovers CRE communities that are undetectable by existing unbiased methods, such as those with interactions that span insulating loop boundaries as well as regulatory communities that are comprised of clustered promoter elements. A subset of these communities, which we term “3D super-enhancers” (3D SEs), is consistent with aspects of the super-enhancer model and accrues exceptional amounts of regulatory apparatus at crucial cell identity-defining genes. In single-cell analyses, pairs of genes found within the same community are co-expressed, and pairs of CREs are co-accessible, suggesting that the regulation of both types of elements within a topologically defined community is tightly coordinated. In confocal microscopy experiments, we regularly observe co-localization and co-expression of in-community genes, an occurrence which frequently takes place within transcriptional co-factor condensates. Our collective approaches more accurately reflect the complexity of CRE interaction networks, underscores the need for incorporating spatial organization in the study of enhancer biology, and provides a framework for the reinterpretation of SEs, their condensates, and their target genes through the lens of 3D chromatin structure. Overall design: Signle cell chromatin profiling of mESC v6.5 cell line is conducted by performing scATAC-seq