Date of Award

2021

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Chemistry

First Advisor

Jianing Li

Second Advisor

Juan Vanegas

Abstract

Simulating protein complexes on large time and length scales is often intractable at the atomistic resolution. To address this challenge, we have developed new approaches to integrate coarse-grained (CG), mixed-resolution (referred to as AACG throughout this dissertation), and all-atom (AA) modeling for different stages in a single molecular simulation. First, we developed a top-down multiscale modeling approach — a new approach, which combines CG, AACG, and AA modeling — to simulate peptide self-assembly from monomers. We simulated the initial encounter stage with the CG model, while the further assembly and reorganization stages are simulated with the AACG and AA models. Further, a theory was developed to estimate the optimal simulation length for each stage. Finally, our approach and theory have been successfully validated with three amyloid peptides. which highlight the synergy from models at multiple resolutions. This approach improves the efficiency of simulating of peptide assembly process. Furthermore, it serves as proof of concept that applying flexible resolution during the simulation, to adapt to efficiency or accuracy. Second, we gained proof of principle from simulating five heterodimeric models of two G protein-coupled receptors (GPCRs) in the lipid-bilayer membrane on the ns-to-μs timescales. In these simulations of different resolution levels, we observed consistent structural stability, while the AACG and CG models show two- and four-times faster protein diffusion than the AA models, in addition to 4- and 400-fold speedup in the simulation performance. Our findings enable synergy from the combination of AA, AACG, and CG models, which lay the foundation to combine these models in one single simulation. It is also feasible to alternate among different models to represent an efficient solution to investigate complex biophysical systems. To investigation of environmental sensing of histone-like nucleoid-structuring (H-NS) protein, we also apply AA models to simulate H-NS protein at multiple spatial scales. The environmental sensing ability is reflected by residues at binding sites or filaments mechanical properties. With AA simulation of dimers, we investigated potential of the mean force (PMF), to quantitively determine the sensitivity of the environmental change of binding site. The simulation of H-NS tetramers reveals that the site2 rather than site1 takes responsibility for environmental sensing. Through the simulation of H-NS filaments, we were able to reveal the movement of the DNA binding domain, which is sensitive to environmental sensing, also influence the H-NS stability. Then we extended our investigation to H-NS orthologs from different organism. Our findings revealed the adaptive evolution of H-NS in different organism. Our multiscale modeling approaches can be useful tools to simulate biological complexes. We applied different combination of AA, AACG, and CG models of the same system. Our new computational methodology advanced the ability to simulate large systems or long process more efficiently. Our methodology is readily adaptable to other systems, based on the need of sampling, properties of interest, and simulation efficiency. In any circumstances where balance will be reached between efficiency and high-resolution, multiscale modeling would be significantly valuable in molecular modeling.

Language

en

Number of Pages

113 p.

Available for download on Wednesday, November 23, 2022

Included in

Chemistry Commons

Share

COinS