Understanding Chromatin Organization and Dynamics with Coarse-Grained Modeling

Abstract

The genome is the blueprint of human life, and it is crucial to understand its organization. The genome organization is hierarchical with different principles dominating at different scales. At the near-atomistic level, nucleosomes are organized as ordered chromatin fibers or disordered chromatin arrays. Furthermore, chromatin and related proteins can function within condensate environments. Computational modeling provides valuable insights into such complex biological processes. Considering the complexity of chromatin and biomolecular condensates, coarse-grained (CG) modeling is essential to achieve the biologically relevant timescales. We have developed CG models and toolkits to facilitate modeling chromatin and related proteins. We have also applied CG protein and DNA models to study chromatin folding and phase separation.

In Chapter 1, we begin with an overview of the hierarchical scales of genome organization. We also introduce CG modeling as a powerful tool to understand the chromatin structures and dynamics. In Chapters 2 and 3, we demonstrate the development of CG simulation force fields and toolkits. In Chapter 2, we present novel CG force fields trained with contrastive learning. We have achieved a new set of hydropathy parameters trained with a99SB-disp all-atom force field trajectories of intrinsically disordered proteins, which accurately reproduces their average radius of gyration. In addition, we have developed a unified force field that captures the average radius of gyration of both ordered and disordered proteins in the training set. In the future, we will focus on benchmarking our models and existing CG models with condensate simulations, which enables more appropriate selections of CG models based on specific conditions. In Chapter 3, we introduce OpenABC, a versatile toolkit designed to streamline the setup of CG simulations, especially condensate simulations. OpenABC incorporates diverse CG force fields within an extensible framework and is built on a simulation platform that supports GPU acceleration, thus speeding up CG simulations.

In Chapters 4 and 5, we shift our focus to the applications of CG simulations. In Chapter 4, we discuss the force extension and inter-chain contacts of chromatin fibers. Our CG simulations reveal that the chromatin fiber behaves like an elastic spring under forces no more than 3 pN, while it dramatically unstacks and unwraps at approximately 4 pN. Meanwhile, inter-chain contacts can help unfold the native two-start fibril-like structures. The study demonstrates that biologically relevant pN-level forces and crowding environments contribute to the absence of 30-nm fibers in vivo. In Chapter 5, we apply Markov state models and non-Markovian dynamics models to study the folding dynamics of tetra-nucleosomes. The tetra-nucleosome with 10n+5-bp linkers shows more diverse structures without dominant native structures, while 10n-bp linkers lead to funnel-shaped free energy landscape with a strong folding trend. Within the condensate, the transition rates slow down, while the unfolding and folding rates are comparable. These two studies highlight that the intrinsic physical chemistry properties of chromatin are fundamental to the genome organization in cells.