Cis-regulatory control of transcriptional timing and noise in response to estrogen

Abstract

Cis-Regulatory Elements (CREs) control transcription levels, temporal dynamics, and cell-cell variation — often referred to as transcriptional noise. However, the combination of regulatory proteins and epigenetic features necessary to control different transcription attributes is not fully understood. Here, single-cell RNA-seq (scRNA-seq) is conducted during a time course of estrogen treatment to identify genomic predictors of expression timing and noise. We find that genes associated with multiple active enhancers exhibit faster temporal responses. Synthetic modulation of enhancer activity verifies that activating enhancers accelerates expression responses, while inhibiting enhancers results in a more gradual response. Noise is controlled by a balance of promoter and enhancer activity. Active promoters are found at genes with low noise levels, whereas active enhancers are associated with high noise. Finally, we observe that co-expression across single cells is an emergent property associated with chromatin looping, timing, and noise levels. Overall, our results indicate a fundamental tradeoff between a gene's ability to quickly respond to incoming signals and maintain low variation across cells.

Estrogen-induced chromatin looping changes identify a subset of functional regulatory elements

“…The Boruta analysis was performed as described by Ginley-Hidinger et al 40 including the feature data sets analyzed. For the analysis looking at differences between differential and non-differential loops, each anchor was included separately.…”

“…The majority of these ERBS (71.9 % in Ishikawa and 79.2 % in HCI-EC-23) were associated with gained 3D interactions. To understand the differences between ERBS associated with differential or non-differential interactions, we used Boruta, a random forest feature selection approach to look for genomic features in Ishikawa cells that distinguish ERBS anchoring differential 3D interactions 40 . Several transcription factors as well as chromatin accessibility were enriched at ERBS with differential 3D interactions, while ERBS associated with non-differential interactions exhibited features of actively transcribed regions, including RNA polymerase II binding, PRO-seq signal, and H3K36me3 (Figure 3C).…”

“…We found that differential and non-differential ERBS containing loops were similarly enriched for E2 up-regulated genes (Figure 3D). In addition, ERBS with differential or non-differential interactions were equally likely to regulate genes that respond early or late during an 8-hour time course 40 (p = 0.55, Fisher's Exact Test). We next used linear regression to determine if ERBS associated with differential 3D interactions were predictive of larger transcriptional responses to E2.…”