Computational prediction of CTCF/cohesin-based intra-TAD loops that insulate chromatin contacts and gene expression in mouse liver

Abstract

CTCF and cohesin are key drivers of 3D-nuclear organization, anchoring the megabase-scale Topologically Associating Domains (TADs) that segment the genome. Here, we present and validate a computational method to predict cohesin-and-CTCF binding sites that form intra-TAD DNA loops. The intra-TAD loop anchors identified are structurally indistinguishable from TAD anchors regarding binding partners, sequence conservation, and resistance to cohesin knockdown; further, the intra-TAD loops retain key functional features of TADs, including chromatin contact insulation, blockage of repressive histone mark spread, and ubiquity across tissues. We propose that intra-TAD loops form by the same loop extrusion mechanism as the larger TAD loops, and that their shorter length enables finer regulatory control in restricting enhancer-promoter interactions, which enables selective, high-level expression of gene targets of super-enhancers and genes located within repressive nuclear compartments. These findings elucidate the role of intra-TAD cohesin-and-CTCF binding in nuclear organization associated with widespread insulation of distal enhancer activity.The human genome contains the complete set of DNA instructions – including all genes – needed to build and maintain an organism. To fit all of this genetic information in the cell’s nucleus, the DNA is neatly wrapped around so-called histone proteins, which help to package the genetic material into chromatin, which forms thread-like structures, the chromosomes.Chromatin is further folded into large DNA loops held together by an anchor protein, CTCF, and by a second protein, cohesin, whose ring-shaped structure ties each loop at its base. DNA segments that are within the same loop may interact frequently, whereas those outside the loop rarely do. Many of these large DNA loops are further pinched off into sub-loops. These sub-loops may help a cell fine-tune whether a gene needs to be turned on or off by limiting the contact between genes and the DNA regions that regulate the activity of genes.Knowing where these DNA sub-loop are located is very important for understanding how each gene is controlled. However, this can be very costly to determine, and therefore, is only known for a few cell types. Now, Matthews and Waxman tackle this issue by creating a computer model that can correctly predict many of these sub-loops. The method used experimental data obtained from mouse liver cells to identify the locations of CTCF and cohesin.The results showed that DNA sub-loops in the liver cells can shield genes from regulatory DNA segments outside the looped area. For example, a small sub-loop that contains a single gene related to obesity is highly active, even though the large DNA loop containing the sub-loop is an otherwise inactive gene region. Similarly, certain genes critical for liver function are positioned within sub-loops containing DNA regions that greatly enhance the gene activity in liver cells. This allows the selected genes to be highly active – unlike other genes that are close by but outside the sub-loop.This new approach will make it easier and cheaper to discover DNA loops and sub-loops across the genome. A better knowledge of where these loops form may also allow us to better understand how genes are turned on and off in different types of cells, and in response to biological stimuli or environmental stresses. This may also help understand and treat conditions that arise from mutations that disrupt the boundaries of DNA loops or sub-loops, which can allow certain DNA segments to activate the wrong genes and can lead to developmental defects and diseases such as cancer.