Nonlinear control of transcription through enhancer-promoter interactions

Abstract

Chromosome structure in mammals is thought to regulate transcription by modulating the three-dimensional interactions between enhancers and promoters, notably through CTCF-mediated interactions and topologically associating domains (TADs)1–4. However, how chromosome interactions are actually translated into transcriptional outputs remains unclear. To address this question we use a novel assay to position an enhancer at a large number of densely spaced chromosomal locations relative to a fixed promoter, and measure promoter output and interactions within a genomic region with minimal regulatory and structural complexity. Quantitative analysis of hundreds of cell lines reveal that the transcriptional effect of an enhancer depends on its contact probabilities with the promoter through a non-linear relationship. Mathematical modeling and validation against experimental data further provide evidence that nonlinearity arises from transient enhancer-promoter interactions being memorized into longer-lived promoter states in individual cells, thus uncoupling the temporal dynamics of interactions from those of transcription. This uncovers a potential mechanism for how enhancers control transcription across large genomic distances despite rarely meeting their target promoters, and for how TAD boundaries can block distal enhancers. We finally show that enhancer strength additionally determines not only absolute transcription levels, but also the sensitivity of a promoter to CTCF-mediated functional insulation. Our unbiased, systematic and quantitative measurements establish general principles for the context-dependent role of chromosome structure in long-range transcriptional regulation.

How subtle changes in 3D structure can create large changes in transcription

“…Similarly, manipulations of the Sox9 locus, which progressively increased the contact frequency between Sox9 enhancers and the adjacent Kcnj2 gene by deleting first the TAD border between them and then further shortening the distance, also exhibited just such a non-linear response (Despang et al, 2019). Most notably, systematic variation of the distance between the Sox2 promoter and one of its enhancers, both inserted into an ectopic site, reveals a strikingly hypersensitive dependence of transcription on E-P contact frequency (Zuin et al, 2021).…”

Enhancer-promoter interactions and transcription are maintained upon acute loss of CTCF, cohesin, WAPL, and YY1

“…We propose that this timebuffering model reconciles our observations with the unambiguous genetic evidence that CTCF and cohesin regulate some E-P interactions. This evidence includes the following: 1) Insertion of CTCF sites between an enhancer and a promoter can both reduce E-P interactions and strongly reduce gene expression [98][99][100] ; 2) CTCF binding site silencing 101,102 or genetic CTCF binding site loss 21,103,104 can cause aberrant E-P interactions and gene expression and drive disease; 3) Inversion or repositioning of CTCF sites can redirect E-P interactions that cause gene misexpression and diseases 105,106 . Two recent studies have also proposed variants of a time-buffering model based on mathematical modeling of E-P interactions and gene expression 100,107 .…”

“…This evidence includes the following: 1) Insertion of CTCF sites between an enhancer and a promoter can both reduce E-P interactions and strongly reduce gene expression [98][99][100] ; 2) CTCF binding site silencing 101,102 or genetic CTCF binding site loss 21,103,104 can cause aberrant E-P interactions and gene expression and drive disease; 3) Inversion or repositioning of CTCF sites can redirect E-P interactions that cause gene misexpression and diseases 105,106 . Two recent studies have also proposed variants of a time-buffering model based on mathematical modeling of E-P interactions and gene expression 100,107 . In both models, individual E-P interactions are memorized -either as long-lived promoter states 100 or as long-lived "promoter tags" 107 -such that gene expression can be temporally uncoupled from E-P interactions, yet still be causally linked.…”

“…Two recent studies have also proposed variants of a time-buffering model based on mathematical modeling of E-P interactions and gene expression 100,107 . In both models, individual E-P interactions are memorized -either as long-lived promoter states 100 or as long-lived "promoter tags" 107 -such that gene expression can be temporally uncoupled from E-P interactions, yet still be causally linked. An alternative, more conservative interpretation of our data and the evidence cited above is that CTCF and cohesin only regulate a very small, unique sets of genes in specific biological processes, and their effect on a handful of loci simply cannot be generalized as a universal rule.…”

Transcriptional enhancers and their communication with gene promoters

“…In addition, 3C-derived methods can pick up rare events that imaging is underpowered to identify [404]. Finally, recent biophysical modelling suggests that functionally causal enhancer-promoter contacts do not have to be temporally concomitant with transcriptional bursts [405,406]. The exact mechanisms underlying the observed phenomenon, however, remain to be elucidated.…”