Identification of cell states using super-enhancer RNA

Abstract

Background: A new class of regulatory elements called super-enhancers, comprised of multiple neighboring enhancers, have recently been reported to be the key transcriptional drivers of cellular, developmental, and disease states. Results: Here, we defined super-enhancer RNAs as highly expressed enhancer RNAs that are transcribed from a cluster of localized genomic regions. Using the cap analysis of gene expression sequencing data from FANTOM5, we systematically explored the enhancer and messenger RNA landscapes in hundreds of different cell types in response to various environments. Applying non-negative matrix factorization (NMF) to super-enhancer RNA profiles, we found that different cell types were well classified. In addition, through the NMF of individual time-course profiles from a single cell-type, super-enhancer RNAs were clustered into several states with progressive patterns. We further investigated the enriched biological functions of the proximal genes involved in each pattern, and found that they were associated with the corresponding developmental process. Conclusions: The proposed super-enhancer RNAs can act as a good alternative, without the complicated measurement of histone modifications, for identifying important regulatory elements of cell type specification and identifying dynamic cell states.