Topologically Associating Domain

A topologically associating domain (TAD)—also referred to at fine resolution as a contact domain or loop domain—is a contiguous, self-interacting linear segment of the eukaryotic chromosome (typically spanning 100 kilobases to 2 megabases, with a median size of ~800–900 kb in mammalian genomes) within which DNA sequences physically contact one another in three-dimensional nuclear space 2- to 5-fold more frequently than they contact sequences at equivalent linear genomic distances outside the domain. Discovered simultaneously in 2012 by Jesse Dixon, Bing Ren, and colleagues, Elphège Nora, Edith Heard, Job de Laat, and colleagues, and Tom Sexton, Giacomo Cavalli, and colleagues using genome-wide chromosome conformation capture (Hi-C and 5C), TADs appear on two-dimensional pairwise contact heatmaps as crisp on-diagonal triangles (squares) separated by sharp insulator boundaries.

Functionally, TADs partition the mammalian genome into ~2,000–3,000 discrete cis-regulatory neighborhoods: by confining distal enhancers and target promoters inside the same extruded loop domain while insulating them from neighboring domains, TAD boundaries prevent ectopic enhancer hijacking during embryonic patterning (Hox gene activation, limb-bud morphogenesis) and suppress oncogenic proto-oncogene activation in human cancer.

Across five orders of magnitude in genomic scale, the interphase eukaryotic nucleus organizes chromatin through two mechanistically independent, competing physical processes—polymer phase separation (A/B compartmentalization) and ATP-dependent cohesin loop extrusion (TADs):

How can two DNA boundary elements separated by 500,000 base pairs along a floppy chromatin polymer know whether they are oriented toward or away from one another in 3D space? In 2014–2015, Suhas Rao, Erez Lieberman Aiden, and colleagues discovered from kilo-base-resolution in situ Hi-C maps that >90% of anchored chromatin loops (the bright "corner peaks" at the apex of TAD triangles) occur between pairs of CCCTC-binding factor (CTCF) motifs oriented in a convergent (→ ←, forward–reverse) head-to-head configuration.

This 1D orientation puzzle was solved by the Loop Extrusion Model (proposed theoretically by Arthur Riggs, John Marko, Kim Nasmyth, and formalized via polymer simulations by Geoffrey Fudenberg, Leonid Mirny, and Erez Lieberman Aiden, 2015–2016) and subsequently visualized directly at the single-molecule level in vitro (Ganji et al., 2018; Davidson et al., 2019):

1. Cohesin Loading and Symmetrical ATP-Driven Extrusion: The ring-shaped cohesin SMC complex—comprising the ~50 nm coiled-coil ATPases SMC1A and SMC3, the kleisin subunit RAD21 (Scc1), and theHEAT-repeat subunit STAG1/STAG2 (SA1/SA2)—is loaded onto chromatin by the kollerin complex NIPBL–MAU2. Driven by SMC headgroup ATP hydrolysis, cohesin reels in DNA at rates of ~0.5–1.5 kb s⁻¹, extruding a progressively growing chromatin loop until it either dissociates (released by the unloading factor WAPL, which sets average loop processivity to ~200–300 kb; WAPL knockout extends loops to >1–2 Mb, compacting interphase chromatin into prophase-like "vermicelli") or encounters a directional barrier.
2. Directional Stalling by the N-Terminus of CTCF: CTCF binds its non-palindromic 14–20 bp core consensus motif (5'-CCACNAGGTGGCAG-3') via 11 zinc fingers in a strictly directional orientation: its C-terminal zinc fingers (ZF1–ZF3) face upstream (5'), while its N-terminal domain faces downstream (3', the direction of the motif arrow →). Only when an extruding cohesin complex approaches CTCF from its N-terminal side does a conserved YDF peptide motif (residues 226–228) in the N-terminus of CTCF bind a specific pocket at the STAG1/2–RAD21 interface of cohesin, stabilizing cohesin against WAPL-mediated release and halting extrusion in that direction.
3. Emergence of TAD Triangles and Corner Peaks: Because cohesin loads at random positions inside a domain bounded by a forward-oriented (→) left CTCF site and a reverse-oriented (←) right CTCF site, any cohesin molecule extruding outward toward either boundary encounters the N-terminus of CTCF (→ ←). Ensemble averaging across millions of cells in different stages of extrusion produces the filled TAD contact triangle, while cohesin complexes stalled simultaneously at both boundaries produce the bright corner dot (insulated loop). Inverting a single 20 bp CTCF motif with CRISPR-Cas9 (→ ← to ← ←) abolishes loop anchoring at that boundary and redirects cohesin to form a new loop with the next convergent CTCF site downstream.

Computational 3D genomics pipelines quantify TAD architecture from normalized Hi-C or Micro-C contact matrices W_ij (binned at resolution Δ = 5–25 kb) using three canonical metrics:

• Directionality Index (DI, Dixon et al., 2012): Quantifies the upstream (A, contacts in [i − w, i)) versus downstream (B, contacts in (i, i + w], where w = 2 Mb) contact bias of each genomic bin i, with E = (A + B) / 2:

TAD boundaries correspond to sharp zero-crossings where DI flips from strongly negative (end of the upstream TAD) to strongly positive (start of the downstream TAD).

• Log₂ Insulation Score (IS, Crane et al., 2015): Slides a square off-diagonal diamond window of width s (typically 100–500 kb) along the main diagonal of the contact matrix, computing the log₂ ratio of mean contacts crossing locus i relative to the chromosome-wide average. Local minima (valleys) of the Insulation Score pinpoint TAD boundaries, and the peak-to-trough depth quantifies boundary insulation strength.
• Graph Laplacian Algebraic Connectivity (λ₂, Fiedler Value) and Normalized Cut: Treating a locus contact matrix W as a weighted undirected graph with degree matrix D (D_ii = ∑_j W_ij) and normalized graph Laplacian L_sym = I − D^−1/2 W D^−1/2, the second-smallest eigenvalue λ₂ measures intra-domain 3D compaction, while the Fiedler eigenvector (v₂) partitions adjacent TADs or A/B compartments along minimal cross-boundary contact cuts.

In 2015, Darío Lupiáñez, Stefan Mundlos, and colleagues demonstrated that structural variants (deletions, inversions, or duplications) that disrupt TAD boundaries without altering a single protein-coding exon cause human congenital limb malformations by rewiring enhancer–promoter topology ("TADopathies"):

• The WNT6–IHH–EPHA4–PAX3 Limb Malformation Locus (Chromosome 2q35–36): In normal human limb buds, a cluster of potent distal limb enhancers sits inside the large EPHA4 TAD, separated from the flanking WNT6/IHH TAD (upstream) and PAX3 TAD (downstream) by robust CTCF boundaries.
  1. Brachydactyly (Shortened Digits): Deletions removing the telomeric boundary between the EPHA4 TAD and PAX3 TAD fuse the two domains (neo-TAD formation), allowing the EPHA4 limb enhancers to ectopically activate PAX3 in the developing autopod.
  2. Polydactyly (Extra Digits): Inversions or deletions disrupting the centromeric boundary between EPHA4 and IHH (Indian Hedgehog) place IHH under the control of the EPHA4 enhancers.
  3. F-Syndrome (Syndactyly): Duplications spanning the boundary between WNT6 and the EPHA4 enhancers create a duplicated neo-TAD in which WNT6 is ectopically driven in the limb bud.
• Oncogenic Boundary Disruption in Cancer: In T-cell acute lymphoblastic leukemia (T-ALL), somatic microdeletions removing CTCF boundary sites flanking silent proto-oncogenes such as TAL1 or LMO2 allow constitutively active super-enhancers from adjacent housekeeping TADs to hijack and activate the oncogene; similarly, in IDH1/IDH2-mutant gliomas, accumulation of the oncometabolite 2-hydroxyglutarate (2-HG) inhibits TET DNA demethylases, causing CpG hypermethylation of a CTCF boundary site upstream of PDGFRA—which blocks CTCF binding, erases the TAD boundary, and lets a constitutive enhancer drive oncogenic PDGFRA overexpression.

If acute CTCF or cohesin depletion erases nearly all TADs on Hi-C maps, why do only a modest subset of genes immediately change expression?

When Elphège Nora (2017) and Suhas Rao (2017) used auxin-inducible degron (AID) tags to acutely degrade CTCF or the cohesin kleisin subunit RAD21 within 2–6 hours in mammalian cells, Hi-C contact maps showed complete loss of loop domains (TADs), yet nascent RNA sequencing (PRO-seq / RNA-seq) showed immediate >2-fold transcriptional deregulation in only ~300–1,000 genes (~5%–15% of active genes). Three complementary mechanisms explain this resilience:

1. Promoter-Proximal vs. Long-Range Enhancer Dependence: Housekeeping genes driven primarily by promoter-proximal elements or short-range enhancers (<20 kb) contact their promoters efficiently via random 3D polymer diffusion without requiring active cohesin loop extrusion; in contrast, inducible developmental, neuronal, and inflammatory genes that rely on long-range distal enhancers (>100–500 kb away) strictly require cohesin extrusion to bring the enhancer and promoter together.
2. Single-Cell Chromatin Globules Persist dynamically (Bintu et al., 2018): Multiplexed super-resolution oligo-PAINT imaging of single chromosomes revealed that individual cells always exhibit globular domain structures at variable positions; cohesin and CTCF do not create chromatin compaction out of nothing, but rather phase-lock and statistically align domain boundaries at fixed CTCF genomic coordinates across the cell population.
3. Epigenetic Compatibility and Promoter Competition: Even when a TAD boundary is deleted, an enhancer will only ectopically activate a neighboring gene if that gene's promoter is biochemically compatible and lacks repressive Polycomb (H3K27me3) or heterochromatin silencing.

1. Dixon, J. R., Selvaraj, S., Yue, F., et al. (2012). "Topological domains in mammalian genomes identified by analysis of chromatin interactions." Nature, 485(7398), 376–380. https://doi.org/10.1038/nature11082
2. Rao, S. S. P., Huntley, M. H., Durand, N. C., et al. (2014). "A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping." Cell, 159(7), 1665–1680. https://doi.org/10.1016/j.cell.2014.11.021
3. Lupiáñez, D. G., Kraft, K., Heinrich, V., et al. (2015). "Disruptions of topological chromatin domains cause pathogenic rewiring of gene-enhancer interactions." Cell, 161(5), 1012–1025. https://doi.org/10.1016/j.cell.2015.04.004
4. Fudenberg, G., Imakaev, M., Lu, C., Goloborodko, A., Abdennur, N., & Mirny, L. A. (2016). "Formation of chromosomal domains by loop extrusion." Cell Reports, 15(9), 2038–2049. https://doi.org/10.1016/j.celrep.2016.04.085