henhaobio
Sign in

bioRxivMolecular BiologyPreprint

Cooperative CTCF-nucleosome oligomerization stabilizes chromatin loop anchors

Manuel Osorio‐ValerianoiD, Alexander C. StoneiD, Masahiro NaganoiD, Bonnie G. SuiD, Laura CaccianiniiD, Anders S. HanseniD, Lucas FarnungiD, Seychelle M. VosiD

PDFDOI
In-depth full-text digest available · Jump to digest ↓

Abstract

Eukaryotic DNA is organized across multiple scales to support genome compaction, appropriate gene expression, and DNA recombination. A central player in these processes is the CCCTC binding factor (CTCF), which defines specific chromatin loop structures and insulates enhancer elements from promoters. Chromatin is organized in a distinct pattern around CTCF-bound sites; however, the role of this patterning remains unclear. Here, we report cryo-electron microscopy structures of reconstituted CTCF-nucleosome complexes, revealing that CTCF dimerization promotes the oligomerization of nucleosomes into defined higher-order assemblies involving specific histone-histone and CTCF-CTCF interactions. Notably, CTCF does not oligomerize efficiently on non-chromatinized DNA substrates. Disruption of identified CTCF interaction interfaces in cells results in a marked decrease in chromatin looping and functionally impairs cellular differentiation. These results indicate that chromatin structure at CTCF sites plays an important role in supporting higher-order interactions between distal regions of the genome and that these interactions are important for supporting cell-type-specific gene expression.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (12 Sept 2026) · CC BY 4.0

CTCF dimerization on stacked nucleosomes is resolved by cryo-EM through a ZnF6-7-8 interface. Disrupting it in mouse stem cells weakens CTCF-cohesin loops more than insulation and impairs germline differentiation.

Nucleosomes are phased around CTCF sites and CTCF is reported to self-associate, but the roles of this phasing and of CTCF-CTCF contacts in genome organization were unclear. The paper links a specific CTCF dimerization interface and nucleosome stacking to loop stability and cell differentiation.

Key results

  1. ShownCTCF drives oligomerization of mononucleosomes into dimers and trimers in vitro, but does not oligomerize on non-chromatinized motif DNA.

    Mass photometry: nucleosomes alone were monomers; with CTCF, dimers (2 nucleosomes + 2 CTCF) and trimers appeared. Cy3/Cy5 FRET rose with 0-1.5 µM CTCF. Motif-containing 42 bp DNA gave only monomeric CTCF complexes; random-sequence nucleosomes bound no CTCF. · CTCF forms higher-order assemblies with…

  2. ShownCryo-EM shows a pseudo-symmetric stacked nucleosome dimer bridged by a parallel CTCF-CTCF contact (ZnF6 against ZnF7-8). Mutating histone interface residues or deleting the H4 tail severely impairs oligomerization, and the CTCF ZnF6-7-8 mutant almost completely loses it.

    Map of the dimer at 3.7 Å (2.8 Å focused on the nucleosome stack). H2B triple mutant or H4 tail deletion: severe defect. CTCF ZnF6-7-8 mutant almost lost oligomerization; ZnF6 and ZnF7-8 mutants each showed a defect. Mutants bound linear DNA like wild type; ZnF1-11 alone sufficed. · Cryo-EM structure of CTCF-nucleosome complex; Disruption of dimerization interfaces reduces…

  3. ShownIn mouse embryonic stem cells, ZnF7-8 and ZnF6-7-8 interface mutants grow slowly, show a striking defect in germ-cell-like induction and have markedly weakened CTCF-cohesin loops, while insulation is largely retained.

    At least two clones per genotype. Micro-C (17.3 billion unique ligations total) showed markedly weaker CTCF-cohesin loops in ZnF7-8 and ZnF6-7-8, moderate in ZnF6. Enhancer-promoter and promoter-promoter loops were only weakly affected. Loss was less severe than acute CTCF degradation. · Disruption of the CTCF dimerization interface…; Disruption of the CTCF dimerization interface…

  4. ShownLower CTCF occupancy in the mutants does not explain most of the loop loss; the dimerization interface itself stabilizes loops more than it supports boundaries.

    Spike-in ChIP-seq: CTCF about 80% of wild type in ZnF6, about 65% in ZnF7-8 and ZnF6-7-8. At matched CTCF/RAD21 occupancy, modeled loop intensity was about 40% (ZnF6) and about 10% (ZnF7-8, ZnF6-7-8); boundary weakening about 20% and about 40%. A dTAG titration gave consistent results. · Quantitative dissection of CTCF occupancy and…

  5. SuggestedMicro-C read-level patterns at convergent CTCF loops are consistent with the in vitro symmetric dimer-stack geometry and are reduced in the interface mutants.

    Ligation classes expected from a symmetric dimer stack were enriched in wild type and reduced in ZnF7-8 and ZnF6-7-8. Phasing amplitude fell in mutants, but with matched CTCF occupancy, median phasing scores mostly stayed above 90% of wild type. · Micro-C read-level analyses favor a ‘symmetric…

Methods, data and reproducibility

Reconstituted human CTCF with nucleosomes and characterized complexes by mass photometry, FRET, gradient fixation and cryo-EM, then tested structure-guided point mutants in vitro. The same mutations were knocked into mouse embryonic stem cells and assessed by growth, germ-cell-like differentiation, Micro-C and spike-in ChIP-seq.

  • Cryo-EM of CTCF-mononucleosome dimer (Map B final particles 44,141 from 20818 movies; Map A 461,738 particles) · Human CTCF; Xenopus laevis histones · n = 44,141 particles (Map B); 461,738 particles (Map A)
  • Cryo-EM of CTCF-dinucleosome dimer and trimer samples (2D classes only) · Human CTCF; Xenopus laevis histones · n = 9577 and 16008 movies
  • Micro-C in wild-type and ZnF6, ZnF7-8, ZnF6-7-8 mutant mESCs (BVSC reporter line) · mouse · n = 17.3 billion unique ligations; 1.6 to 2.4 billion per replicate
  • Micro-C and spike-in CTCF/RAD21 ChIP-seq after 4 h dTAG-13 titration in CTCF-dTAG mESCs · mouse · n = 6 dTAG doses (0 to 500 nM); 2 replicates; about 16.9 billion contacts
  • Quantitative CTCF and RAD21 ChIP-seq in mutant and wild-type mESCs with HEK293T spike-in · mouse (human spike-in)
Methods & tools
cryoSPARC, UCSF ChimeraX/ISOLDE, Phenix, AlphaFold3, Refeyn mass photometry (DiscoverMP), pairtools, cooler, cooltools, Mustache, coolpup.py, Bowtie2, deepTools
Reagents & environment
Tool versions stated for the genomics workflow, e.g. Snakemake v9.1.1, pairtools v1.1.0, cooler v0.10.3, cooltools v0.7.1, Mustache v1.3.3, Bowtie2 v2.5.5, deepTools v3.5.2, Phenix v1.21.2, ISOLDE v1.10.1.
Reproducibility gaps
No code or deposition accessions (maps, models, sequencing data) are printed; materials are available on request with a materials transfer agreement with MIT and Harvard.
Limitations (5)
  • Dinucleosome dimer and trimer complexes were too flexible for 3D reconstruction, so only 2D classes are available for them. · stated by the authors
  • CTCF density was at lower resolution than the nucleosomes and side chains could not be modeled unambiguously, so interface residues were chosen from Cα positions. The Zn2+ assignment between nucleosomes is tentative. · stated by the authors
  • The dTAG titration used a different cell line from the mutants, so the dose-matched comparison is not strictly one-to-one. Micro-C read-level signatures are modest and cannot resolve interfaces at Ångström scale. · stated by the authors
  • The cellular conclusions rest on at least two isogenic clones per genotype in one mouse embryonic stem cell background. · noted in this summary
  • The mutants reduce chromatin-bound CTCF, and the cause (lost dimerization versus altered recruitment or retention) is not separated experimentally. · noted in this summary

Read this if you study CTCF, cohesin loop anchors or nucleosome phasing and want structural and Micro-C evidence for CTCF dimerization.

Citations
3

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISDo '24Lucero '25Clarkson '19Hafner '22Owens '19Nora '17
70% semantic matchGenomics
Catherine Do, Guimei Jiang, Giulia Cova+3 more·bioRxiv·Nov 2, 2024· 5 citations