Resolving the 3D landscape of transcription-linked mammalian chromatin folding

Abstract

Chromatin folding below the scale of topologically associating domains (TADs) remains largely unexplored in mammals. Here, we used a high-resolution 3C-based method, Micro-C, to probe links between 3D-genome organization and transcriptional regulation in mouse stem cells. 20Combinatorial binding of transcription factors, cofactors, and chromatin modifiers spatially segregate TAD regions into "microTADs" with distinct regulatory features. Enhancer-promoter and promoter-promoter interactions extending from the edge of these domains predominantly link co-regulated loci, often independently of CTCF/Cohesin. Acute inhibition of transcription disrupts the gene-related folding features without altering higher-order chromatin structures. 25Intriguingly, we detect "two-start" zig-zag 30-nanometer chromatin fibers. Our work uncovers the finer-scale genome organization that establishes novel functional links between chromatin folding and gene regulation. 30ONE SENTENCE SUMMARY Transcriptional regulatory elements shape 3D genome architecture of microTADs. MAIN TEXT 35Chromatin packages the eukaryotic genome via a hierarchical series of folding steps ranging from nucleosomes to chromosome territories (1). Structural analysis of chromosome folding has been revolutionized by the Chromosome Conformation Capture (3C) family of techniques, which uses proximity ligation of cross-linked genomic loci in vivo to estimate contact frequencies (2). Interphase chromosome structures such as compartments (3), topologically-40 associating domains (TADs) (4, 5), and CTCF/cohesin chromatin loops (6) have been characterized using 3C-based methods. Chromosome compartments correspond to large-scale active and inactive chromatin segments and appear as a plaid-like pattern in Hi-C contact maps at the megabase scale (3). At the intermediate scale of tens to hundreds of kilobases, topologically associating domains (TADs) spatially organize the mammalian genome into 45 continuous self-interacting domains. TADs are defined as local domains in which genomic loci contact each other more frequently within the domain than with loci outside. TADs appear as square boxes along the diagonal of 3D contact maps (4, 5). Mounting evidence suggests that CTCF and cohesin likely mediate TAD formation via a loop extrusion mechanism (7, 8) wherein the cohesin ring complex entraps chromatin loci and extrudes chromatin until blocked by CTCF 50 or other proteins. Stabilization of cohesin at CTCF sites result in sharp corner peaks in contact maps and are also referred to as CTCF/cohesin loops or loop domains. Various studies have

Cis-regulatory chromatin loops arise before TADs and gene activation, and are independent of cell fate during development

“…Recent studies reported the existence of enhancer hubs: spatially-localized clusters containing multiple enhancers 27,28,31 that may facilitate transcriptional activation by creating a local microenvironment whereby transcriptional resources are shared, akin to early models of 'transcription hubs' 79 . Formation of enhancer hubs may require interactions between components of the transcriptional machinery which could contribute, or result from, the assembly of phase-separated condensates 30,35,36,[80][81][82] . In this model, enhancers need not directly touch their target promoters but merely come into close proximity 24,83 .…”

“…Early evidence showed that promoters can contact several distant enhancers [15][16][17] , raising the possibility that more than one enhancer may contact a promoter at any given time. More recently, use of multi-way 3C and 4C methods showed that, indeed, enhancers can cluster together to form enhancer hubs that can recruit one or several promoters [27][28][29][30] . This is supported by evidence of nuclear microenvironments containing multiple enhancers and clusters of transcription factors [31][32][33][34][35][36][37] .…”

Loop competition and extrusion model predicts CTCF interaction specificity

“…To validate our model on an additional external dataset, we predicted CTCF loops identified from a recently published high resolution Micro-C dataset. 26 In total, 15,945 significant loops at 1kb resolution were detected in this dataset with HICCUPS. 6 For purposes of predicting CTCFmediated loops, we sampled positive loops with CTCF binding at both ends, and generated a five times larger negative set by sampling from non-interacting CTCF pairs.…”

“…A total of 15,945 loops were called from 2.6B reads of mESC Micro dataset 26 . Chromatin loops were identified by using HiCCUPS 6 .…”

Are the processes of DNA replication and DNA repair reading a common structural chromatin unit?

“…In 2015, Hsieh and colleagues introduced Micro-C, a novel Hi-C method with nucleosome resolution, in which micrococcal nuclease is used instead of restriction enzymes to fragment chromatin [52]. In 2019, Hsieh and colleagues showed by Micro-C that TADs are segregated further into microTADs by the action of transcription factors, cofactors, and chromatin modifiers [53]. Krietenstein and colleagues utilize the same technique to resolve more than 20000 additional looping interactions with single-nucleosome accuracy in comparison to Hi-C [54].…”