Computational Modeling of DNA Sequence Effects on the Nucleosome Core Particle

Abstract

The nucleosome particle is an essential biological macromolecule serving both a structural and gene regulatory roles in eukaryotic genomes. The nucleosome particle has a cylindrical shape, composed of 8 highly conserved histone proteins wrapped by a sequence of 147 base pairs of DNA. Considerable, experimental evidence has shown that different sequences of 147 base pairs have varying preferences for forming stable particles, yet atomic-level descriptions for the preferences are vague at best. Microscopic descriptions contribute to fundamental understanding of genomes process, facilitate rationale approaches to drug design for certain genetic diseases, and can contribute to genetic engineering. We have established a novel basis for computational modeling DNA interactions with the nucleosome core particle. Computational modeling can approach the complexity and vast number of DNA sequences of potential interest. Our method is the first to substitute DNA on the nucleosome core and explore the rotational degree of freedom, crucial to assessing a DNA helix in preferred, low-energy states. This work was carried out along with experimental work used as a reference for the computational studies. Specifically, we experimentally determined the relative binding affinities of 3 DNA sequences for forming nucleosomes. These experimental data provide a ranking of stability as a function of sequence on the free energy of binding. Accurate free-energies of binding for large biological systems are extremely difficult to compute. The computational modeling hinges on a high-resolution crystal structure (1.9 Å) of the nucleosome core particle. In our method, we substitute DNA molecules of interest on the crystal structure in order to study the structural dynamics using sophisticated computational models centered on molecular dynamics simulations. In this work, we performed three separate molecular dynamic simulations, one for each of the sequences, in order to explore the atomic-level basis for the differentials in binding, which were determined experimentally. Crucially, the rotational degrees of freedom are explored applying novel methods to a complex geometric and chemical problem. This method uses systematical sampling and dynamical modeling of the DNA rotational conformers. Through this work we have demonstrated the feasibility and methodology for atomic-level modeling of DNA with potential for high throughput. This thesis describes methods and results of this work.