1. A model that generates contact-domain patterns
Loop extrusion proposes that factors enlarge DNA loops, with boundary interactions influencing their extent. Fudenberg and colleagues tested a minimal chromatin-polymer model containing assumed loop-extruding factors and boundaries. It reproduced selected features of published GM12878 contact maps. This establishes the model's sufficiency for those patterns under its assumptions; it does not directly observe a molecular motor or prove that no other mechanism could produce similar contacts. [1]
2. Cohesin-dependent architecture in human cells
Rao and colleagues used acute RAD21 degradation in engineered human HCT116 colorectal carcinoma cells. Hi-C loop domains disappeared after cohesin loss while compartment organization remained. The perturbation identifies cohesin-dependent contact architecture in this system and separates loop domains from compartments. It does not establish that every form of domain organization disappears, nor does a before-and-after population contact map show cohesin physically moving along DNA. [2]
This audit checked the official primary abstract and detailed figure legends for that experiment. Full article access remained unavailable, and no full-method appraisal or cross-tissue replication is claimed. [2]
3. Reconstitution tests biochemical capability
Davidson and colleagues visualized cohesin-mediated loop growth on surface-tethered DNA using purified components, NIPBL–MAU2 and ATP. CTCF acted as an asymmetric barrier whose effectiveness depended on DNA tension; the barrier could be bypassed. This directly supports biochemical loop-growth capability and a conditional CTCF barrier. It corrects the idea that CTCF is an absolute gate that always stops extrusion in living chromatin. [3]
A controlled DNA assay differs from nucleosome-rich chromatin in a living nucleus. Its loop-growth behavior therefore cannot supply a universal cellular speed, processivity or loading mechanism. The retained conclusion is the capability measured under the assay's conditions. [3]
4. Cellular dynamics are measured indirectly
Gabriele and colleagues tracked anchors of one 505-kb Fbn2 domain in mouse embryonic stem cells during G1/early S. Distances fluctuated, and Micro-C plus architectural-factor perturbations supported an extrusion-related interpretation. Bayesian analysis inferred infrequent, transient fully boundary-bridged states. It tracked locus labels rather than cohesin translocation; the inferred states and lifetimes depend on the model and cannot be assigned to every TAD. Fbn2 was silent, so this is not a direct enhancer–transcription test. [4]
5. Strong architectural evidence is not universal gene-control evidence
The model, cellular perturbation, reconstitution and live-locus results provide converging support for extrusion-related architecture, with different inferential strengths. Regulatory consequences remain a separate question: Hsieh and colleagues found that most enhancer–promoter contacts and expression in mouse stem cells persisted during three-hour CTCF/cohesin/WAPL depletion, despite altered structural-loop architecture. Weak overall contact changes and sensitive subsets remained. This concerns short-term maintenance, not all establishment or developmental functions. [5]
An expanded live human-cell comparison — 12 October 2026
Sabaté and colleagues extended live anchor tracking to three endogenous TADs in human HCT116 cells, alongside Capture Micro-C, cohesin perturbation and polymer modeling. The sampled domains showed dynamic anchor proximity and were interpreted as being continuously extruded by multiple cohesin complexes. This extends the earlier single-locus evidence, while retaining locus-label, cell-line and model limits; the paper’s inferred uniform cohesin dynamics across its sampled regions is not a measured universal constant for every tissue or locus. Sabaté et al. (2025).
This addition uses primary studies. Access depth and claim limits are recorded with each added reference; it is a dated literature appraisal, not independent experimental replication.
Primary studies and review scope
- Fudenberg et al. (2016), Cell Reports 15:2038–2049. Formation of Chromosomal Domains by Loop Extrusion. DOI: 10.1016/j.celrep.2016.04.085. Primary research.
- Rao et al. (2017), Cell 171:305–320.e24. Cohesin Loss Eliminates All Loop Domains. DOI: 10.1016/j.cell.2017.09.026. Primary research.
- Davidson et al. (2023), Nature 616:822–827. CTCF is a DNA-tension-dependent barrier to cohesin-mediated loop extrusion. DOI: 10.1038/s41586-023-05961-5. Primary research.
- Gabriele et al. (2022), Science 376:496–501. Dynamics of CTCF- and cohesin-mediated chromatin looping revealed by live-cell imaging. DOI: 10.1126/science.abn6583. Primary research.
- Hsieh et al. (2022), Nature Genetics 54:1919–1932. Enhancer–promoter interactions and transcription are largely maintained upon acute loss of CTCF, cohesin, WAPL or YY1. DOI: 10.1038/s41588-022-01223-8. Primary research.
- Sabaté T, Lelandais B, Robert MC, et al. (2025). Uniform dynamics of cohesin-mediated loop extrusion in living human cells. Nature Genetics 57:3152–3164. DOI: 10.1038/s41588-025-02406-9. Primary research.
Independent update access: Consensus fetched primary abstract and Scite selected main-text passages, including study design and initial Results; not every supplementary analysis inspected. Primary source.
The retained scientific text and the identity and fit of these references were checked on 3 October 2026. The discussion covers these experiments rather than an exhaustive literature search. Source-access limitations and outstanding work remain.