Single-fiber nucleosome density shapes the regulatory output of a mammalian chromatin remodeling enzyme

Nour J. Abdulhay, Laura J Hsieh, Colin P McNally, Mythili Ketavarapu, Sivakanthan Kasinathan, Arjun S. Nanda, Megan S Ostrowski, Ke Wu, Camille M. Moore, Hani Goodarzi, Geeta J. Narlikar, Vijay RamaniZenodo 2021 · DOI

TL;DR

Researchers developed SAMOSA-ChAAT, a single-molecule footprinting platform that maps nucleosome positions on individual chromatin fibers with high resolution. Using this tool, they discovered that the chromatin remodeler SNF2h acts as a "length-sensor" that adjusts its behavior based on local nucleosome density, explaining how a single enzyme can both open and close DNA accessibility across the genome.

What Problem Does This Paper Address?

In the nucleus, DNA is wrapped around histone proteins to form nucleosomes, which act as gatekeepers for genetic information. ATP-dependent chromatin remodelers, like the ISWI-family enzyme SNF2h, are the "motors" that slide or eject these nucleosomes to control DNA accessibility.

Until now, our understanding of these enzymes was limited by a "resolution gap." Biochemical studies usually look at single nucleosomes, while genomic studies look at "bulk" averages across millions of cells. Bulk averages hide the fact that individual chromatin fibers are highly heterogeneous—two identical DNA sequences in the same cell might have nucleosomes in completely different spots. The authors sought to resolve a long-standing debate: Does SNF2h act as a "clamping" tool (placing nucleosomes at a fixed distance like a ruler) or a "length-sensing" tool (moving nucleosomes based on the available space between them)?

Methods

The authors introduced SAMOSA-ChAAT (Chromatin Accessibility of Assembled Templates), which combines several cutting-edge techniques:

  • In Vitro Reconstitution: They assembled chromatin fibers on specific genomic sequences using salt gradient dialysis at varying densities.
  • Adenine Methylation Footprinting: They used a non-specific methyltransferase (EcoGII) to "paint" accessible DNA with methyl groups ($m^6A$). Areas covered by nucleosomes remain unmethylated.
  • Single-Molecule Sequencing: They used PacBio SMRT sequencing, which can detect $m^6A$ modifications by measuring the "Inter-Pulse Duration" (IPD)—the time the polymerase pauses at each base.
  • NN-HMM Pipeline: To turn raw sequencing pulses into nucleosome maps, they built a custom computational pipeline using a Neural Network to account for sequence-specific noise and a Hidden Markov Model to call the final "footprints" (nucleosome positions) on single fibers.

Datasets

  • Synthetic Genomic Templates: Two ~3kb sequences from the mouse genome (S1 and S2) containing CTCF binding motifs, assembled at four different histone-to-DNA ratios (5:1 to 20:1).
  • In Vivo mESCs: Murine embryonic stem cells, comparing wild-type cells, SNF2h knockouts ($Smarca5^{-/-}$), and "addback" lines where SNF2h expression was restored.
  • Epigenomic Benchmarks: ENCODE-defined domains (heterochromatin, promoters, etc.) and gold-standard CTCF binding sites.

Key Insights and Ideas

  • Density-Dependent Logic: SNF2h doesn't have a single "target" spacing. Instead, the final distance between nucleosomes is a function of how many nucleosomes are already on the fiber. This confirms the length-sensing model.
  • A Unified Theory for Opening and Closing: The paper explains a biological paradox. In dense, "closed" heterochromatin, SNF2h senses the lack of space and slides nucleosomes into regular, tightly packed arrays to keep the DNA repressed. In less dense, "open" regions, it slides nucleosomes away from specific motifs (like CTCF sites) to increase accessibility.
  • Overriding DNA Sequence: While DNA sequence often dictates where nucleosomes want to sit (positioning sequences), SNF2h is powerful enough to override these preferences, creating a "remodeled steady state" that is highly heterogeneous.
  • The "State Space" of Chromatin: By using UMAP (dimensionality reduction) on single fibers, the authors showed that chromatin exists in a vast landscape of structural states that bulk methods simply cannot see.

Key Results

  • Footprint Accuracy: The pipeline called footprints of ~130–160 bp, matching the known size of DNA wrapped around a histone octamer.
  • Spacing Dynamics: In vitro, as nucleosome density increased, the average spacing (Nucleosome Repeat Length or NRL) decreased, directly contradicting the "clamping" model.
  • CTCF Accessibility: SNF2h remodeling increased the accessibility of CTCF binding motifs by up to 60-fold in some contexts, specifically by sliding occluding nucleosomes into "offset" positions.
  • In Vivo Impact: In mESCs, losing SNF2h caused a global shift toward longer, more irregular nucleosome spacing. Restoring the enzyme specifically condensed satellite DNA (heterochromatin) while opening up promoters and CTCF sites.

Limitations and Future Work

  • Complex Composition: This study focused primarily on the SNF2h ATPase alone. In cells, SNF2h works as part of larger complexes (like ACF or WICH) with non-catalytic subunits that likely fine-tune its "sensing" ability.
  • Static Snapshots: While SAMOSA-ChAAT provides high resolution, it is a "terminal" assay. It captures the state of the fiber at the moment of methylation, not the real-time movement of the enzyme.
  • Cell Cycle Averaging: The in vivo data represents an average of cells across different stages of the cell cycle (e.g., mitosis vs. interphase), which the authors note could be disentangled in future work using synchronized cell populations.
  • Future Directions: The authors suggest applying this platform to study how other factors, like transcription, DNA replication, or histone modifications, compete with remodelers to shape the chromatin landscape.

Summary updated

Abstract

Processed data from the manuscript "Single-fiber nucleosome density shapes the regulatory output of a mammalian chromatin remodeling enzyme", including predicted DNA accessibility states along each molecule, predicted nucleosome counts for each molecule, and alignments of circular consensus sequences to mm10 for mESC samples. Data are separated by the sample type. Each zipped folders also contain a readme explaining the file contents, a table with clear information on the conditions of each sample, a script to uncompress the accessibility path representations, and for the S1 and S2 molecules a reference sequence.
BibTeX
@inproceedings{abdulhay2021single,
  author = {Nour J. Abdulhay and Laura J Hsieh and Colin P McNally and Mythili Ketavarapu and Sivakanthan Kasinathan and Arjun S. Nanda and Megan S Ostrowski and Ke Wu and Camille M. Moore and Hani Goodarzi and Geeta J. Narlikar and Vijay Ramani},
  title = {{Single-fiber nucleosome density shapes the regulatory output of a mammalian chromatin remodeling enzyme}},
  booktitle = {Zenodo},
  year = {2021},
  doi = {10.5281/zenodo.5770727},
}

Loading related content...

Figure