A computational approach to map nucleosome positions and alternative chromatin states with base pair resolution

Xu Zhou, Alexander W. Blocker, Edoardo M. Airoldi, Erin K. O’SheaeLife 2016 · DOI

TL;DR

This paper introduces a new computational method, Template-Based Bayesian (TBB), that precisely maps nucleosome positions at base-pair resolution using standard MNase-seq data. This advancement reveals widespread "alternative nucleosome positioning" in yeast and human cells, suggesting dynamic chromatin states that influence gene expression.

What Problem Does This Paper Address?

Understanding how DNA is packaged into chromatin is crucial for deciphering gene regulation. Nucleosomes, the basic units of chromatin, are formed by DNA wrapped around histone proteins. Their precise location is thought to significantly impact DNA accessibility for transcription factors and the cellular machinery. While MNase-seq is a common technique for mapping nucleosomes genome-wide, it typically provides an "average" nucleosome position across millions of cells, masking important biological variability. This averaging can obscure the true, fine-grained positioning of individual nucleosomes and the existence of alternative configurations within a cell population, making it challenging to understand dynamic chromatin states and their role in gene regulation.

Methods

The core innovation is the Template-Based Bayesian (TBB) approach. It tackles the limitations of MNase-seq by:

  1. Estimating Digestion Variability: The method first models the inherent variability in how MNase cuts DNA around nucleosomes. This is done by analyzing the distribution of DNA fragment lengths from sequencing data, creating a "digestion variability template." This template accounts for technical variations (like nuclease trimming) and biological factors (like nucleosome breathing).
  2. Bayesian Deconvolution: Using this template, TBB employs a Bayesian framework to "deconvolute" the sequencing reads. This process disentangles the signal from individual nucleosomes, even when they overlap, and corrects for the digestion variability.
  3. Base-Pair Resolution: The output is a precise map of nucleosome positions at single-base-pair resolution, distinguishing between "consensus centers" (the average position) and individual TBB nucleosome positions.

This approach is designed to be applicable to both paired-end and single-end sequencing data.

Datasets

  • Yeast (S. cerevisiae): Paired-end MNase-seq data from two independent experiments (T1 and T2) were used.
  • Human: Paired-end MNase-seq data from a previously published study (Gaffney et al., 2012) was analyzed for a specific region of chromosome 12.
  • In Silico Simulations: Simulated MNase-seq datasets were generated to rigorously test the TBB method's accuracy and precision under various controlled conditions (coverage, spacing, effective magnitude).

Key Insights and Ideas

  • Widespread Alternative Nucleosome Positioning: The most striking finding is that alternative nucleosome positions are not rare but are prevalent across the yeast and human genomes. This means that within a population of cells, a single genomic locus can be occupied by nucleosomes in slightly different, yet stable, configurations.
  • Dinucleotide Periodicity is a Universal Signal: TBB's high resolution allowed the direct observation of a conserved 10 bp periodicity of specific dinucleotide sequences (like AA/AT/TA/TT and CC/CG/GC/GG) around nucleosome centers genome-wide. This suggests a fundamental sequence-based code for nucleosome positioning.
  • Alternative Nucleosomes at TSSs Dictate Transcription Initiation: Nucleosomes near transcription start sites (TSSs) exhibit distinct positioning states ("proximal" vs. "distal") that correlate with the accessibility of the promoter and the assembly of the transcription pre-initiation complex (PIC). This suggests a dynamic model for how chromatin states at promoters control gene activation.
  • Uniquely Positioned Nucleosomes Link to High Expression: Genes with uniquely positioned nucleosomes at their TSSs, particularly ribosomal protein genes, tend to be highly expressed, suggesting a connection between stable chromatin structure and robust gene output.

Key Results

  • High Accuracy: The TBB approach achieved base-pair resolution and demonstrated high accuracy, with simulated nucleosome positions being detected within 2-3 bp for primary positions and 3-7 bp for alternative positions under various conditions.
  • Reproducibility: The method showed excellent reproducibility between biological replicates (T1 and T2), with a median difference of only 2 bp.
  • Dinucleotide Periodicity Confirmed: The 10 bp dinucleotide periodicity was clearly observed around TBB-mapped nucleosomes, a pattern that was less apparent or absent with previous methods.
  • Alternative Nucleosome States Identified: Analysis of TSSs revealed distinct proximal and distal nucleosome states, with proximal states being less compatible with PIC binding and distal states appearing to define boundaries for PIC assembly.

Limitations and Future Work

  • Statistical Confidence: The TBB approach identifies nucleosome positions with strong statistical confidence, meaning it captures configurations likely relevant to biological processes but may not capture every transient or low-occupancy state.
  • Dynamic Processes: While TBB provides a snapshot of chromatin states, further work is needed to fully reconstruct dynamic processes and understand how these states transition.
  • Model Extension: The authors suggest that similar analytical approaches could be applied to other systems to reconstruct dynamic processes. Future work could focus on applying TBB to study how nucleosome positions change in response to other regulators of gene activity and during gene activation/repression. The authors also propose a 3-step model for transcription initiation influenced by nucleosomes, which warrants further experimental validation.

Summary updated

Abstract

Understanding chromatin function requires knowing the precise location of nucleosomes. MNase-seq methods have been widely applied to characterize nucleosome organization in vivo, but generally lack the accuracy to determine the precise nucleosome positions. Here we develop a computational approach leveraging digestion variability to determine nucleosome positions at a base-pair resolution from MNase-seq data. We generate a variability template as a simple error model for how MNase digestion affects the mapping of individual nucleosomes. Applied to both yeast and human cells, this analysis reveals that alternatively positioned nucleosomes are prevalent and create significant heterogeneity in a cell population. We show that the periodic occurrences of dinucleotide sequences relative to nucleosome dyads can be directly determined from genome-wide nucleosome positions from MNase-seq. Alternatively positioned nucleosomes near transcription start sites likely represent different states of promoter nucleosomes during transcription initiation. Our method can be applied to map nucleosome positions in diverse organisms at base-pair resolution.
BibTeX
@inproceedings{zhou2016computational,
  author = {Xu Zhou and Alexander W. Blocker and Edoardo M. Airoldi and Erin K. O’Shea},
  title = {{A computational approach to map nucleosome positions and alternative chromatin states with base pair resolution}},
  booktitle = {eLife},
  year = {2016},
  doi = {10.7554/elife.16970},
}

Loading related content...

Figure