Boundary stacking interactions enable cross-TAD enhancer-promoter communication during limb development

Abstract

While long-range enhancers and their target promoters are frequently contained within a TAD, many developmentally important genes have their promoter and enhancers within different TADs. Hypotheses about molecular mechanisms enabling such cross-TAD interactions remain to be assessed. To test these hypotheses, we use Optical Reconstruction of Chromatin Architecture (ORCA) to characterize the conformations of the Pitx1 locus on thousands of single chromosomes in developing mouse limbs. Our data supports a model in which neighboring boundaries are stacked with each other as a result of loop-extrusion, bringing boundary-proximal cis-elements into contact. This stacking interaction also explains the appearance of architectural stripes in the population average maps (e.g. Hi-C data). Through molecular dynamics simulations, we further propose that increasing boundary strengths facilitates the formation of the stacked boundary conformation, counter-intuitively facilitating border bypass. This work provides a revised view of the TAD borders' function, both facilitating as well as preventing cis-regulatory interactions, and introduces a framework to distinguish border-crossing from border-respecting enhancer-promoter pairs.

Synergistic insulation of regulatory domains by developmental genes and clusters of CTCF sites

“…3C). This suggests that, in contrast to previous reports 26,27,51 , neither the positioning of Gbx2 close to the 3’ boundary nor the CTCF sites are important for the maintenance of Gbx2 expression. In contrast, both the inversion and the deletion significantly increased Asb18 expression, with the inversion having a larger effect (Fig.…”

“…Furthermore, at these boundaries, developmental genes and CTCF sites are often organized in a sequential and evolutionary conserved manner, with the genes preceding the CTCF clusters. In contrast, genes previously reported as capable of bypassing boundaries 25–27 are typically flanked by CTCF sites and, thus, located within CTFC clusters. Most importantly, through the exhaustive genetic dissection of a couple of representative developmental loci (i.e.…”

“…S5-6), we noticed that they are typically flanked by strong CTCF sites and, thus, located within, rather than next to, CTCF clusters. This organization might favour boundary bypass, TAD intermingling and communication with regulatory elements located on either side of the genes 26,27,31,78 . In contrast, for the two developmental genes that we investigated, we found no evidences indicating that CTCF clusters facilitate their expression or the communication with their cognate enhancers.…”

Increased Enhancer-Promoter Interactions during Developmental Enhancer Activation in Mammals

“…The general mechanism that determines E–P specificity in mammalian genomes is still poorly understood 71 , and further studies are needed to dissect how divergent expression is achieved within the same TAD. Notably, we also observe that 21% of developmental enhancers act across TAD boundaries confirming previous observations 72,73 . These cross-TAD enhancers behave similarly to intra-TAD enhancers (Extended Data Fig.…”

“…6c) but tend to locate closer to TAD borders (Extended Data Fig. 6d) consistent with the boundary staking model that was proposed to facilitate TAD border bypass 73 .…”