Allelic Transcription Factor binding shape transcriptional kinetics observed with scRNA-seq in human cell lines

Abstract

Gene expression from bulk RNA-seq studies is an average measurement between two chromosomes and across cell populations. Both allelic and cell-to-cell heterogeneity in gene expression results from promoter bursting patterns that repeatedly alternate between an activated and inactivated state. Increased cell-to-cell heterogeneity in gene expression has been shown as a hallmark of aging, which is potentially induced by the change of promoter bursting patterns. Despite their importance in humans, bursting kinetics studies have been studied only in a limited number of genes within selected model organisms due to technical restrictions in measuring multiple transcript levels over time. Here, we construct a transcriptomic kinetics map using single-cell RNA-seq (scRNA-seq) data and systematically investigate the regulatory effect of genetic variants and transcription factor (TF) binding on transcriptional kinetics. We obtain allelic level expression from scRNA-seq data with phased genotypes for two clonal cell lines and estimate transcriptional bursting kinetics for a single chromosome. We found that among transcriptional kinetics, the transcription initiation rate and burst frequency correlate most with eQTL effect sizes from bulk RNA-seq studies, suggesting that eQTLs affect average gene expression mainly through altering the transcription initiation rate and burst frequency. Notably, eQTL studies focused on mean expression changes cannot identify scenarios where burst size and frequency are regulated in opposite directions, which is common amongst genes from LCL. We further found that ~90% of the variance of burst frequency can be explained by TF occupancy within the core promoter and allele-specific binding events of individual TFs are typically associated with the change of transcription initiation rate and burst frequency. Finally, we demonstrate how a genetic variant alters TF binding, resulting in changes to promoter burst kinetics of HLA-DQA1 to influence multiple hematological traits.