Dissertation/Thesis Abstract

Computer Simulation of Molecular Systems: Glycoproteins and Carbon Nanotubes with Ionic Liquid
by Bernardi, Austen Verbrugge, D.Engr., University of California, Davis, 2020, 148; 27667557
Abstract (Summary)

This dissertation contains two major components: simulations that are concerned with glycoprotein characterization (chapters 2,3 and 4) and simulations that are concerned with filling carbon nanotubes with ionic liquid (chapter 5). While the subjects of these components are very different (biology vs. materials science), the simulations by which they are studied here are extremely similar. Both simulations employ atomistic molecular dynamics (MD) to generate trajectories and energies to characterize their equilibrium ensembles.

Prior to exhibiting the glycoprotein simulations, we first present a paper that focuses on enabling the accurate simulation of glycosylated systems in GROMACS, the simulation suite employed for our glycoprotein simulations. Ensuring that proper simulation strategies are employed is critical for obtaining useful simulation results. In our case, we had to develop modifications for and validate a force field transfer program. The results of this work is presented in chapter 2. Chapters 3 and 4 detail glycoprotein simulations that make use of the tool described in chapter 2. In the glycoprotein simulations, we employed classical MD with 100 ns trajectories to gain insight on the dynamics and function of the systems.

The first glycoprotein simulations, given in chapter 3, focus on the structural characterization of butyrylcholinesterase (BChE). BChE is an antidote for Sarin gas, and is heavily glycosylated. Chapter 3 provides the methodological framework for studying glycoproteins with MD, from virtual construction to simulation and analysis. This framework forms the basis for the simulation presented in chapter 4. Chapter 4 features experimental and computational analysis of a chimeric anthrax-decoy protein CMG2-Fc. CMG2-Fc contains two glycosylation sites that directly contribute to receptor binding. The paper presented in chapter 4 is the first to comprehensively study a glycoprotein using real world experiments and MD. While the full paper is provided in chapter 4, my personal contribution to the experimentation and analysis was limited to MD simulation and analysis.

The glycoproteins exhibited in chapters 3 and 4 are expensive to produce, while being of critical importance to the security of the nation. The glycoprotein simulations presented in this work develop a methodology for studying these proteins' utility as a function of glycosylation; a factor that is heavily dependent on the expression system. UC Davis is investigating the production of these proteins in planta, where the glycosylation profile is different from that of humans. From the research in this work, we learn that this change in glycosylation does have effects on protein utility. We have also developed the methodology for characterizing these systems in high detail using classical MD.

The final chapter in this work, chapter 5, presents work concerned with simulating the filling of carbon nanotubes (CNTs) with ionic liquid (IL). While this appears widely different from simulating glycoproteins, these simulation also employ MD, demonstrating the versatility of the method. The focus of this work was to develop a filling method that results in an energetically favorable, filled state for periodic CNTs with IL. Since simulating this filling process is impossible using classical MD, we developed a novel method using accelerated MD, employing soft-core potentials (SCPs) and simulated annealing (SA). This new method is termed SCP/SA. We validated SCP/SA on both charged and uncharged carbon nanotubes, illustrating the method's utility. Additionally, SCP/SA is broadly applicable to other systems and readily modifiable for generating new accelerated MD methods. More specifically, one could directly apply SCP/SA to the problem of generating initial configurations for the glycans in glycoproteins. This would be quite useful for glycoprotein simulations, as the protein coordinates can be experimentally generated from X-ray crystallography, but most of the glycan residues are too flexible to be resolved experimentally, and one must use approximate methods to generate initial coordinates.

Indexing (document details)
Advisor: Faller, Roland
Commitee: McDonald, Karen, Grønbech-Jensen, Niels
School: University of California, Davis
Department: Chemical Engineering
School Location: United States -- California
Source: DAI 81/11(E), Dissertation Abstracts International
Subjects: Chemical engineering, Computational physics, Condensed matter physics
Keywords: Carbon nanotubes, Computer simulation, Glycoproteins, Ionic liquid, Molecular dynamics, Soft-core potentials
Publication Number: 27667557
ISBN: 9798645452681
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy