Quantitative test of the barrier nucleosome model for statistical positioning of nucleosomes up- and downstream of transcription start sites

The positions of nucleosomes in eukaryotic genomes determine which parts of the DNA sequence are readily accessible for regulatory proteins and which are not. Genome-wide maps of nucleosome positions have revealed a salient pattern around transcription start sites, involving a nucleosome-free region (NFR) flanked by a pronounced periodic pattern in the average nucleosome density. While the periodic pattern clearly reflects well-positioned nucleosomes, the positioning mechanism is less clear. A recent experimental study by Mavrich et al. argued that the pattern observed in S. cerevisiae is qualitatively consistent with a `barrier nucleosome model’, in which the oscillatory pattern is created by the statistical positioning mechanism of Kornberg and Stryer. On the other hand, there is clear evidence for intrinsic sequence preferences of nucleosomes, and it is unclear to what extent these sequence preferences affect the observed pattern. To test the barrier nucleosome model, we quantitatively analyze yeast nucleosome positioning data both up- and downstream from NFRs. Our analysis is based on the Tonks model of statistical physics which quantifies the interplay between the excluded-volume interaction of nucleosomes and their positional entropy. We find that although the typical patterns on the two sides of the NFR are different, they are both quantitatively described by the same physical model, with the same parameters, but different boundary conditions. The inferred boundary conditions suggest that the first nucleosome downstream from the NFR (the +1 nucleosome) is typically directly positioned while the first nucleosome upstream is statistically positioned via a nucleosome-repelling DNA region. These boundary conditions, which can be locally encoded into the genome sequence, significantly shape the statistical distribution of nucleosomes over a range of up to ~1000 bp to each side.

💡 Research Summary

The paper addresses a long‑standing question in chromatin biology: to what extent the characteristic nucleosome arrangement observed around transcription start sites (TSSs) in budding yeast can be explained by a purely physical “barrier nucleosome” model versus intrinsic DNA sequence preferences. Genome‑wide nucleosome maps reveal a nucleosome‑free region (NFR) flanked on both sides by a series of well‑positioned nucleosomes that generate a damped oscillatory pattern in average nucleosome density. The barrier nucleosome model, originally proposed by Kornberg and Stryer, posits that a fixed “barrier” (for example a tightly bound transcription factor or a nucleosome that is directly positioned) creates a statistical positioning effect: nucleosomes, treated as hard rods of 147 bp, cannot overlap and therefore arrange themselves with a characteristic spacing that decays with distance from the barrier due to entropic considerations.

Mavrich et al. previously argued that the yeast pattern is qualitatively consistent with this model, but the presence of strong sequence‑dependent nucleosome preferences left the quantitative contribution of the physical mechanism unclear. To resolve this, the authors adopt the Tonks gas model from one‑dimensional statistical physics, which explicitly incorporates the excluded‑volume interaction of hard rods and the configurational entropy of their positions. The model requires only three biologically meaningful parameters: (i) the nucleosome length (147 bp), (ii) the average nucleosome density (or equivalently the mean inter‑nucleosome spacing), and (iii) the boundary condition that defines the nature of the barrier at the edge of the NFR.

Using high‑resolution MNase‑Seq data from Saccharomyces cerevisiae, the authors compute average nucleosome occupancy profiles upstream (‑1 side) and downstream (+1 side) of thousands of NFRs aligned at the TSS. They then fit the Tonks model separately to each side, allowing the boundary condition to vary while keeping the physical parameters identical. Downstream of the NFR, the best‑fit boundary corresponds to a “fixed” nucleosome: the +1 nucleosome sits at a well‑defined location, effectively acting as an impenetrable wall for downstream nucleosomes. Upstream of the NFR, the optimal boundary is a “repulsive” region: a stretch of DNA that disfavors nucleosome formation, thereby creating a soft barrier that statistically positions the –1 nucleosome. Remarkably, the same excluded‑volume length and average density parameters reproduce both profiles with high quantitative accuracy, demonstrating that the oscillatory patterns on both sides arise from the same underlying physics, differing only in how the barrier is encoded.

The inferred upstream repulsive barrier is likely encoded in the DNA sequence (e.g., high A/T content or specific transcription‑factor binding motifs) and can extend the influence of the NFR up to ~1000 bp, shaping nucleosome organization far beyond the immediate vicinity of the TSS. Conversely, the downstream fixed +1 nucleosome may be directly positioned by transcription‑initiation complexes or chromatin remodelers. By quantitatively linking these boundary conditions to measurable sequence features, the study bridges the gap between purely statistical positioning and sequence‑driven nucleosome preferences.

In summary, the authors provide a rigorous quantitative test of the barrier nucleosome model. They show that a simple Tonks‑gas description, with identical physical parameters but distinct boundary conditions, accurately captures the nucleosome density oscillations on both sides of yeast NFRs. The work underscores that while DNA sequence can encode the nature of the barrier, the bulk of the nucleosome arrangement over hundreds of base pairs is governed by excluded‑volume interactions and entropic positioning. This insight refines our understanding of chromatin organization, suggesting that statistical positioning and sequence specificity act together, with the former providing a robust, long‑range scaffold upon which sequence‑encoded regulatory cues are superimposed.