BGPT: Paper Review: DNA loop extrusion by human cohesin

Fuel Your Discoveries

Quick Explanation Copied

Human cohesin can extrude DNA loops in vitro—but only when the loader is present.

The paper reconstitutes human cohesin (STAG1/STAG2) on stretched λ-DNA and shows that NIPBL–MAU2 + ATP hydrolysis is required for active loop extrusion with ~kb/s rates, loop-base localization of cohesin and NIPBL–MAU2, ATP-dependence, and a strong test suggesting loop extrusion can occur without opening the tripartite ring (i.e. pseudo-/non-topological mechanisms remain possible).

Key molecular control: ATP vs CTP, ATPase-deficient mutants, and NIPBL–MAU2 presence during extrusion/maintenance determine whether loops form/persist.

Long Explanation

Paper review (mechanistic + skeptical): “DNA loop extrusion by human cohesin”

Target paper DOI: 10.1126/science.aaz3418

Single-molecule reconstitution ATP- and NIPBL–MAU2-dependent Pseudo-/non-topological ring tests

Figure-derived quantitative highlights (from reported counts)

All bars below are computed directly from the paper’s reported “DNAs analyzed” and “DNAs that formed loops” counts.

1) What the paper claims (only what is directly supported)

Core reconstitution outcome

Human cohesin alone (under the reported conditions) did not produce detectable DNA loops on tethered λ-DNA (0/372 and 0/503 reported in the paper’s initial reconstitution logic)
Adding NIPBL–MAU2 together with ATP enables active loop extrusion, with measured loop fractions in the ~40–60% range for simultaneous addition experiments

Mechanistic dependencies

Loop extrusion requires ATP hydrolysis activity and shows strict nucleotide specificity: loops occur with ATP but not with CTP (0/366 with CTP) and are diminished with ATP-binding/hydrolysis-deficient cohesin mutants
NIPBL–MAU2 is required not only to initiate but also to maintain loops over the imaging window; replacing unbound loader conditions dramatically reduces persistence, consistent with loops being sustained by a DNA-bound cohesin–loader holoenzyme

Localization and dynamics consistency with extrusion

Cohesin and NIPBL–MAU2 localize to the base of loops during formation, and they move against buffer flow during extrusion
They also report unidirectional translocation against flow at a mean rate ~0.4 kb/s and suggest this may reflect a partial extrusion reaction

2) Skeptical appraisal: what’s strong vs uncertain

Strong aspects (evidence quality)

Mechanism-linked controls: the paper’s causality chain is supported by multiple orthogonal perturbations—nucleotide substitution (ATP vs CTP), ATPase-deficient mutants, loader dependence via sequential component addition and maintenance tests
Spatial/temporal consistency: loop-base localization of both cohesin and NIPBL–MAU2 plus movement against flow supports the interpretation that these components are mechanically engaged during extrusion rather than merely correlating with loop presence

Key uncertainties / potential blind spots

In vitro context limitation: tethered λ-DNA in a flow cell is a clean substrate, but chromatin in vivo includes nucleosomes, histone modifications, transcriptional activity, and competing DNA-binding proteins. The paper does not directly demonstrate loop extrusion under full chromatin conditions; thus mechanistic conclusions about in vivo topological anchoring must be treated as a best-supported working model, not a proven equivalence
Non-topological vs pseudo-topological inference: covalent crosslinking/single-chain trimer experiments support that ring opening is not strictly required for extrusion. However, incomplete crosslinking efficiency and the possibility of residual un-crosslinked species are always an interpretive concern. The authors attempt to address this by dilution controls suggesting residual un-crosslinked cohesin is insufficient, but without independently verifying “zero” active species, absolute exclusion of all topological routes remains conservative
Force/geometry effects: the assay uses flow to orient/tension DNA; although the authors show looping can occur without buffer flow, other mechanical constraints (DNA stiffness in different configurations; tether spacing) can bias loop-formation probability or rates. Correlations between end-to-end distance and looping are reported, but mechanistic decomposition (e.g., bending energy threshold vs loading kinetics) remains incomplete

3) Mechanism sketch (explicitly separating known vs inferred)

Known from the experiments (supported)

NIPBL–MAU2 + ATP is sufficient to convert “no detectable loops” into robust loop extrusion on tethered λ-DNA with reported loop fractions and kb/s rates
Cohesin and NIPBL–MAU2 occupy the loop base during extrusion and move against buffer flow
ATP hydrolysis and loader presence are required for both initiation and maintenance of loops over multi-minute imaging windows

Most plausible model (explicitly inferred, not fully proven)

The authors’ mechanistic conclusion is that cohesin can extrude loops without requiring ring opening/topological DNA entrapment—consistent with pseudo-topological or non-topological routes. This is strongly suggested by the covalent ring-interface constraints, but it remains an inference because “residual uncrosslinked fraction” and “distinct extrusion geometries” can be hard to rule out completely
The paper additionally connects in vitro activity to cellular loop-extrusion explanations for CTCF convergence and cohesin behavior (“vermicelli”)—but those are outside the experimental scope of this specific in vitro assay

4) What would most directly disprove the main mechanistic claims?

Loader dependence falsification: show loops form with ATP but without NIPBL–MAU2 in chromatin-mimetic substrates (nucleosomes/physiological protein mixtures), contradicting the idea that the functional holoenzyme is needed for extrusion
Ring-opening irrelevance falsification: design assays where ring interfaces are constrained similarly but independently quantify (e.g., with orthogonal labeling) that no ring-opening intermediate exists at all; if loops still require transient interfaces, then the “no ring opening needed” inference weakens
Activity vs geometry artifact falsification: reproduce the same extrusion readout using alternative tension/orientation protocols and quantify how frequently “active extrusion” appears; if loop formation is dominated by mechanical artifacts, the mechanistic model is over-interpreting the in vitro observation

5) Suggested follow-up reading / cross-checks

For how loop extrusion is used as a mechanistic explanation of TAD/loop features, see chromatin-loop extrusion modeling frameworks
For condensin direct real-time loop extrusion imaging (useful comparison baseline), see Ganji et al.
For background on SMC ring/kleisin mechanistic principles, see Uhlmann’s review

Most important takeaway

Cohesin’s ability to form DNA loops is not apparent in a minimal reconstitution unless the loader complex NIPBL–MAU2 and ATP hydrolysis are both functionally engaged; the experimental readouts then show extrusion-like loop bases and translocation. The covalent-ring constraint experiments strongly suggest a loop-extrusion mechanism that does not strictly require topological DNA entrapment, but the in vitro chromatin-free substrate limits how confidently this transfers to in vivo chromatin constraints.

Evidence is taken from the paper’s reported counts and mechanistic assays.

Author review buttons (BGPT)

Feedback:

Updated: March 25, 2026