Longevity & AgingArticle de rechercheAccès libre

New single-cell tool maps weak transcription factor binding across the genome

D&D-seq uses a base-editing enzyme to mark where proteins sit on DNA in single cells, including weak, transient binders and closed chromatin. Cell lines and human blood tested.

jeudi 19 novembre 2026 1 vue
Publié dans Cell
Glowing nanobody-enzyme complex tethered to a DNA double helix, marking cytosines on a transcription factor binding site, dark blue lab-style render

Résumé

Researchers built D&D-seq, a method that records where DNA-binding proteins sit on the genome in individual cells. An antibody-guided nanobody carries a split cytosine deaminase (DddA) that converts nearby cytosines to uracil only when activated, leaving a sequenceable footprint. It works with standard single-cell ATAC-seq and whole-genome workflows. In K562 and CA46 cell lines and in human blood cells, it recovered CTCF, GATA1 and other factor binding, including in closed chromatin. It showed higher specificity than the CUT&Tag/CUT&RUN datasets it was compared against. This is a tool-development study in cell lines and a few primary samples. It does not test aging interventions or clinical outcomes, but it could help study how chromatin regulation changes in aging and clonal hematopoiesis.

Résumé détaillé

Gene regulation depends on transcription factors (TFs) and chromatin remodelers binding specific DNA sites. Aging and cancer reshape these binding landscapes. Mapping weak or transient binding in single cells has been hard, because existing methods such as ChIP-seq, CUT&Tag and DamID are bulk-oriented or capture only strong binders. Better maps could help explain how aging chromatin drives dysfunction.

The authors developed D&D-seq (docking and deamination followed by sequencing). A secondary nanobody, or protein A/G, fused to an engineered split DddA11 deaminase binds a primary antibody against the target protein. The enzyme stays inactive until a C-terminal peptide and Zn2+ are added. It then converts cytosines to uracil near the bound site, and the edits are read out by ATAC-seq, scATAC-seq or whole-genome sequencing (via PTA). This avoids the high-salt conditions that disrupt weak interactions.

In bulk K562 cells, edits were enriched at ENCODE ChIP-seq peaks for CTCF, GATA1 and GATA2, with bimodal footprints around motifs. The assay also detected p300, JUN, SP1 and TAL1 signals. Combined with whole-genome sequencing, it found 2,045 CTCF peaks outside open chromatin, 196 of them overlapping heterochromatin marks. In a single-cell mixing experiment (1,732 K562 and 4,108 CA46 cells), signals stayed specific to the antibody-stained population. Fraction of reads in peaks was about 57% for CTCF regardless of input size, versus about 14% for uliCUT&RUN. Against public scCUT&Tag data, D&D-seq showed less open-chromatin bleed-through. Aggregate footprints needed about 250 or more cells.

In 5,358 primary human PBMCs, CTCF binding was mapped across cell types. As a preliminary demonstration, the signal was used with C.Origami to predict 3D genome contact maps comparable to those from ChIP-seq input. A variant, D&D-GoT-ChA, linked genotype to CTCF binding in PBMCs from one clonal hematopoiesis (CHIP) patient with IDH2 R140Q. Of 15,807 cells, 33% were genotyped, and mutant cells were enriched in CD8 T cells, which was confirmed by nanopore sequencing of sorted populations.

The main caveats are sparse edits per cell, which require pooling cells, and moderate performance for GATA1 at single-cell level. The 3D genome result is preliminary, and the CHIP work is one patient. No effects on aging, function or lifespan were tested, and the paper is a methods advance. The text provided ends mid-results, so later findings are not covered.

Principales conclusions

  • D&D-seq footprints recapitulated ENCODE ChIP-seq peaks for CTCF, GATA1 and p300 in K562 cells, with motif-centered bimodal edit patterns.
  • Paired with whole-genome sequencing, it found 2,045 CTCF peaks outside open chromatin, 196 of them overlapping H3K9me3 or H3K27me3 heterochromatin.
  • Single-cell CTCF D&D-seq reached ~57% FRiP versus ~14% for uliCUT&RUN, with less open-chromatin bleed-through than scCUT&Tag.
  • In 5,358 human PBMCs, CTCF binding was mapped; in one CHIP patient (IDH2 R140Q), mutant cells were enriched in CD8 T cells.
  • Edits per cell are sparse; robust footprints needed at least ~250 cells, and single-cell GATA1 sensitivity was only moderate.

Méthodologie

Tool-development study: bulk and single-cell D&D-seq in K562 and CA46 cell lines, plus primary human PBMCs from healthy donors and one CHIP patient. Performance was benchmarked against ENCODE ChIP-seq, uliCUT&RUN and public scCUT&Tag using FRiP, motif enrichment and a logistic regression peak classifier (AUC 0.73-0.89). Findings are mechanistic and technical; no function or lifespan outcomes were measured.

Limites de l'étude

Few samples (healthy PBMC donors and a single CHIP patient), sparse edits per cell requiring pooling, and only moderate single-cell GATA1 sensitivity. Direct single-cell benchmarks do not exist, and the 3D genome prediction is preliminary. The supplied full text is truncated, so later results are not summarized.

Ce résumé vous a plu ?

Recevez les dernières recherches sur la longévité dans votre boîte de réception chaque semaine.

Saisissez votre e-mail pour vous abonner :