A topologically associating domain (TAD) is a contiguous region of a genome in which DNA sequences show a higher frequency of physical contact with one another than with sequences in neighboring regions. TADs are features of three-dimensional chromatin organization, usually identified through chromosome-conformation experiments rather than directly from the DNA sequence. They help describe how chromosomes are folded and how regulatory interactions are spatially constrained. A TAD is not necessarily a stable, enclosed structure present in every cell: its experimental signature often represents contact preferences averaged across many cells.
Definition and structural context
Within the cell nucleus, DNA is packaged with proteins into chromatin. Its folding brings regions that are far apart along the linear DNA sequence into spatial proximity. TADs describe an intermediate level of this organization, between individual local contacts and larger-scale patterns of chromosome compartmentalization.
In mammalian genomes, reported TADs commonly span hundreds of thousands of base pairs to around a million base pairs, although their measured sizes vary substantially with the organism, experimental resolution, and computational definition. Smaller domains can occur within larger ones, producing a nested or hierarchical organization.
A TAD boundary is a region across which contact frequencies are relatively reduced. Boundaries are therefore zones of partial insulation, not impermeable walls. Contacts can cross them, and the strength of insulation varies among boundaries and biological conditions.
The word topologically is used in a broad structural sense. A TAD is not ordinarily defined by a mathematical topological invariant, nor does its identification establish that it is a separate DNA supercoiling domain.
Domains, loops, and compartments
Several features of chromosome organization are related but should not be treated as interchangeable:
- TADs are regions with enriched internal contacts relative to contacts across their boundaries.
- Chromatin loops are preferential contacts between particular genomic sites. A loop joining the ends of a domain can help produce its contact pattern, but many domains lack a single prominent boundary-to-boundary loop.
- Chromatin compartments reflect preferential association among regions with similar biochemical or transcriptional properties. Such regions can be widely separated along a chromosome.
A domain may contain multiple loops, and a compartment may contain multiple domains. Compartment segregation and local domain formation can also respond differently to experimental perturbations, indicating that they are not simply different names for one folding mechanism.
Discovery and experimental detection
Domain-like organization became a major framework in genome biology following chromosome-conformation studies published in 2012, including work in mammalian cells and fruit flies. These studies identified widespread regions of enhanced internal contact and relatively sharp transitions in interaction patterns.
The principal methods belong to the chromosome conformation capture family. In a typical experiment, chromatin is chemically crosslinked, fragmented, and ligated so that DNA fragments near one another in the nucleus can be joined. The resulting junctions are identified by DNA sequencing.
Hi-C extends this approach genome-wide. Its results are commonly displayed as a contact matrix, in which each entry represents the frequency of contacts between two genomic intervals. TADs often appear as squares along the matrix diagonal, or as triangles when only one half of the symmetric matrix is displayed.
Higher-resolution methods, including Micro-C, can resolve finer contact features. Imaging methods, including fluorescence in situ hybridization, provide complementary information about spatial distances and structural variation between individual cells. Contact frequency and spatial distance are related measurements, but they are not equivalent.
Computational identification
TAD boundaries are inferred using an algorithm, rather than read directly from a single universal experimental marker. Common approaches include:
- Insulation scores, which measure contacts crossing a candidate boundary.
- Directionality measures, which detect shifts between preferential contacts with upstream and downstream regions.
- Segmentation and clustering methods, which partition a contact map into internally enriched regions.
- Multiscale methods, which identify nested structures at different genomic scales.
Results depend on sequencing depth, bin size, normalization, and the parameters of the domain-calling method. Different analyses can therefore produce different domain counts or boundaries from the same experiment. Comparisons are most meaningful when the experimental resolution and operational definition are specified.
Mechanisms of domain formation
Loop extrusion
Loop extrusion is a major mechanistic explanation for many mammalian TAD patterns. In this model, a protein complex associates with chromatin and progressively enlarges a DNA loop. The cohesin complex is central to this process.
The DNA-binding protein CTCF can act as a directional barrier to extrusion. Many prominent mammalian loop anchors contain CTCF-binding sites arranged in convergent orientations. Repeated extrusion events constrained by such sites can generate enriched contacts within a region and reduced contacts across its boundaries.
This model explains several experimental observations:
- Removing cohesin strongly weakens many local domains and anchored loops.
- Removing CTCF reduces insulation at many CTCF-associated boundaries.
- Changing the orientation or position of particular CTCF-binding sites can alter local folding.
- Changing the residence time of cohesin on chromatin can change the range and organization of contacts.
These effects do not mean that each TAD is a permanently closed loop. Loop formation, enlargement, and release are dynamic, and the structures present at one moment can differ among cells.
Additional mechanisms
Not all domain-like patterns arise from CTCF-constrained loop extrusion. Preferential interactions among chromatin regions with similar properties can also generate self-interacting regions. Transcriptional activity, chromatin-associated proteins, and contacts with nuclear structures can contribute to domain organization.
The balance of these mechanisms differs among organisms. Mammalian CTCF-associated domains are particularly well characterized, but domain-like structures in fruit flies, plants, and other organisms need not share the same molecular basis. Similar-looking contact patterns therefore do not, by themselves, establish a common mechanism.
Relationship to gene regulation
TADs are important in the study of gene expression because regulatory sequences can act over long genomic distances. An enhancer, for example, can influence the activity of a promoter without being immediately adjacent to it along the chromosome.
Domain boundaries can reduce opportunities for regulatory contacts between neighboring regions. In some genomic contexts, this helps keep an enhancer associated with an appropriate set of target genes. Nevertheless, TAD membership is not a complete rule for enhancer targeting:
- Not every enhancer regulates every gene within its domain.
- Some regulatory interactions cross called boundaries.
- Promoter compatibility, chromatin accessibility, and bound transcription factors also influence regulatory outcomes.
- A detected contact does not by itself demonstrate a functional regulatory interaction.
Experiments that delete or rearrange boundaries have produced outcomes ranging from substantial changes in expression to limited effects. Such differences depend on which regulatory elements are present, whether alternative barriers remain, and whether the affected genes can respond to newly accessible enhancers.
Developmental organization
Many domain boundaries are shared across cell types, whereas specific contacts and transcriptional states within domains can change during development. TADs can consequently provide a relatively persistent structural context within which regulatory programs vary.
This stability is not universal. Domain strength, boundaries, and internal contacts can change with differentiation, chromatin state, or experimental treatment. In mammalian mitosis, the interphase TAD pattern is largely lost as chromosomes undergo a different mode of compaction, and it is re-established after mitotic exit.
Genome rearrangements and disease research
A mutation that changes a boundary or rearranges a genomic region can alter the regulatory environment of nearby genes. Deletions, inversions, duplications, and translocations may remove insulation, create new domains, or place a gene near regulatory elements that normally act elsewhere.
One experimentally established mechanism is enhancer adoption, also called enhancer hijacking: a rearrangement allows a gene to come under the influence of an enhancer from a different regulatory neighborhood. Such mechanisms have been documented in particular congenital developmental disorders and in cancers.
However, a predicted boundary disruption alone does not establish that a variant causes disease. Its consequences depend on the precise rearrangement, the relevant cell type, and the regulatory elements involved. TAD maps therefore provide contextual evidence for variant interpretation rather than a stand-alone diagnostic criterion.
Single-cell organization and interpretive limits
Population contact maps average across large numbers of chromosomes. A strong domain in such a map does not require every cell to contain the same compact structure or sharply defined boundary. Single-cell contact measurements and imaging reveal substantial variability in chromatin conformations.
An important distinction is between:
- Population-level domains, which describe reproducible statistical contact preferences.
- Single-cell conformations, which are particular spatial arrangements sampled from a variable and dynamic ensemble.
Individual cells can contain domain-like structures whose boundaries differ from those inferred in population averages. Conversely, a reproducible average boundary can reflect a bias in a dynamic folding process rather than a fixed physical partition.
The term TAD is also used somewhat differently across studies. Some definitions emphasize contact enrichment, others insulation, and others loop-bounded structures. This variation complicates comparisons of domain number, size, conservation, and function. Clear interpretation requires distinguishing the observed contact-map feature from the molecular mechanism proposed to produce it.