Kilobase Pair Chromatin Fiber Contacts Promoted by Living-System-Like DNA Linker Length Distributions and Nucleosome Depletion

Abstract

Nucleosome placement, or DNA linker length patterns, are believed to yield specific spatial features in chromatin fibers, but details are unknown. Here we examine by mesoscale modeling how kilobase (kb) range contacts and fiber looping depend on linker lengths ranging from 18 to 45 bp, with values modeled after living systems, including nucleosome free regions (NFRs) and gene encoding segments. We also compare artificial constructs with alternating versus randomly distributed linker lengths in the range of 18-72 bp. We show that nonuniform distributions with NFRs enhance flexibility and encourage kb-range contacts. NFRs between neighboring gene segments diminish short-range contacts between flanking nucleosomes, while enhancing kb-range contacts via hierarchical looping. We also demonstrate that variances in linker lengths enhance such contacts. In particular, moderate sized variations in fiber linker lengths (∼27 bp) encourage long-range contacts in randomly distributed linker length fibers. Our work underscores the importance of linker length patterns, alongside bound proteins, in biological regulation. Contacts formed by kb-range chromatin folding are crucial to gene activity. Because we find that special linker length distributions in living systems promote kb contacts, our work suggests ways to manipulate these patterns for regulation of gene activity.

Figure 1:

Persistence length for systems with uniform linker length = 18 bp, uniform linker length = 27 bp, life-like, life-like with NFRs, and gene encoding-like fibers. Persistence length is calculated according to Eq. 2. A short persistence length is associated with a more flexible fiber. NFRs decrease the persistence length by ∼20 nm. Gene-encoding like fibers show slightly increased persistence length as compared to randomly distributed life-like fibers with NFRs, indicating that localizing short linkers near the 5^′ NFR stabilizes the fiber slightly. Error bars depict standard deviations, which are computed across 1,000 conformations taken from the last 10 millions steps of several independent trajectories (See Methods for details).

Figure 2:

Contact maps and contact probability profiles for uniform linker length = 18 bp, uniform linker length = 27 bp, life-like, life-like with NFRs, and gene encodinglike fibers. Contact maps are determined by counting distances between any two fiber constituents (tail, core, or linker DNA bead) that are less than 2 nm. The 1D contact profiles are shown in the top right, indicating that long-range contacts (i± ≥ 7) are absent for short uniform linker length fibers but present for other systems. Gene-encoding like fibers show the least amount of long-range contacts. Contact matrices indicate that NFRs decrease local contacts along the diagonal, and gene-encoding like linker length distributions slightly decrease long-range (i± ≥ 7) contacts.

Figure 3:

Contact probability profiles, contact maps, and fiber structures for systems of gene encoding-like fibers with different linker histone densities: a) no linker histone (−LH), b) 1 linker histone per 2 nucleosomes $\left(\frac{1}{2}\text{LH}\right)$, and c) 1 linker histone per nucleosome (+LH). Linker histone beads are drawn in light blue.

Figure 4:

Matrix of structures (lower triangle) and contact maps (upper triangle) for various combinations of linker lengths randomly distributed throughout the 100 core fibers. Systems near the top left (shortest linkers) are the most compact fibers, and fibers become less condensed locally as linker lengths increase. Kb range contacts are strongest for systems with 36 and 45 bp, the median average linker length for living systems.

Figure 5:

Matrix of structures (lower triangle) and contact maps (upper triangle) for various combinations of NRLs alternating in sequence along the 100 core fiber. While general characteristics are shared with randomly distributed NRL sequences, subtle differences in contact types are observed, such as in the 27/45 bp systems, where long-range contacts are significantly higher than in the randomly distributed case. Additionally, hairpins form spontaneously in the 36/45 bp and 45/54 bp case which are not seen in the randomly distributed case.

Figure 6:

Select ΔLL = 9 bp fibers with random distribution of x,y linker lengths of a) 18/27 bp, b) 36/45 bp, and c) 63/72 bp. Shorter linkers encourage long-range contacts, and large linker lengths diminish local contacts as well as long-range contacts. See supporting information for similar plots of all other fibers.

Figure 7:

Select ΔLL = 9 bp fibers with alternating sequence (x,y,x,y...) linker lengths of a) 18/27 bp, b) 36/45 bp, and c) 63/72 bp. Shorter linkers encourage long-range contacts, and large linker lengths diminish local contacts as well as long-range contacts, similar to the randomly distributed case. Fibers with 36/45 bp linker lengths, however, show more compaction and a hairpin fold, absent in the random case. See supporting information for similar plots of all other fibers tested.

Figure 8:

Self-association metric (bottom right) computed for solenoid (i ± 1), zigzag (i ± 2), local (i ± 3)+(i ± 4), medium (i ± 5)+(i ± 6), and long-range (i± ≥ 7) contacts for randomly distributed linker length constructs (solid black line) versus alternating linker length sequences (dashed grey line) plotted against the ΔLL of each system. ΔLL is defined as the magnitude of the difference between DNA linker length values in the fiber, and all points are averaged across all systems presented in Figs. 4 and 5. Alternating linker lengths show marked increase in solenoid contacts and a decrease in zigzag systems when ΔLL is large. Near contacts are slightly higher in the ΔLL = 27 bp range, where medium range contacts are slightly higher in all ΔLL values considered. Finally, far contacts are ∼20% higher in the ΔLL = 27 bp range.