Skip navigation
Please use this identifier to cite or link to this item:
Title: Investigation of Biological Systems at Low Temperatures using Molecular Simulation
Authors: Kozuch, Daniel Jeffrey
Advisors: Debenedetti, Pablo G
Contributors: Chemical and Biological Engineering Department
Keywords: Machine Learning
Molecular dynamics
Protein engineering
Protein folding
Subjects: Biophysics
Computational chemistry
Chemical engineering
Issue Date: 2021
Publisher: Princeton, NJ : Princeton University
Abstract: A major outstanding problem in the field of biological physics is to properly understand the behavior of biological systems outside physiological conditions, especially at the molecular level. In this thesis, we focus mainly on the behavior of biological systems at low temperatures. These conditions are fundamentally interesting from a theoretical perspective, but they are also particularly relevant for engineering applications, such as anti-icing, crop hardiness, cryogenic electron microscopy, and cryopreservation. To probe the behavior of these systems at the molecular level, we ran advanced molecular dynamics simulations which were processed with a variety of computational and machine learning techniques. In the second chapter of this thesis, we investigated the metastable folding behavior of a model peptide, Trp-cage, at deeply supercooled temperatures using a combination of parallel tempering and metadynamics. From these simulations, we observed novel re-folding of the peptide at temperatures more than 50 K below the freezing point of the system. This re-folding was found to correlate with the continuous transition of the water from a high density liquid state to a lower density liquid state. Protein-water hydrogen bonds were found to be less favorable in the low density water, suggesting that this cold re-folding is driven by the expulsion of water bound to the protein core, which allows the protein to re-fold into the same structure observed at ambient temperatures. In the third chapter, we used standard molecular dynamics simulations to study the dynamic properties of antifreeze proteins (AFPs), which are a diverse class of proteins that non-colligatively depress the freezing point of water by binding to nascent ice crystals. From these simulations, we developed a neural network model that could predict the experimentally observed antifreeze activity of the AFPs from the structure of the AFPs and the dynamics of hydrogen bonding measured near the surface of the AFPs. Then, in the fourth chapter, we implemented this neural network model as a fitness function in a genetic algorithm. This algorithm was applied to three naturally occurring AFPs, and we found that for two of the three AFPs, the algorithm produced mutants that have predicted activities that are significantly higher than the wild type. In the fifth chapter, we considered how the cryopreservative trehalose impacts the behavior of a model membrane composed of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) lipids under rapid cooling. To do this, we ran molecular dynamics simulations in which the temperature of the system was progressively dropped in a step-wise manner from 350 K to 250 K using both all-atom (AA) and coarse-grained (CG) resolution. Analysis of the AA simulations showed that at sufficiently high concentrations, trehalose inhibits the formation of the major domain of the ripple phase at low temperatures, favoring the formation of the thinner, interdigitated minor domain. In the CG simulations, the ripple phase does not form at low temperatures; instead, the system transitions directly from the liquid crystalline phase to the gel phase. Even so, we observed that increasing the concentration of trehalose in the CG system inhibits the formation of the gel phase at low temperatures, leading to a thinner membrane as was observed with the AA system. Finally, in the sixth chapter, we departed from our focus on low temperature systems. In collaboration with an experimental group, we studied the molecular level details of hydrophobic ion pairing (HIP) in which the hydrophobicity of a target drug, polymyxin B (PB), was effectively increased by electrostatically pairing it with an ionic surfactant, oleic acid (OA). Using molecular dynamics, we measured how variables such as solvent composition, OA:PB ratio, and system size affected the hydrophobicity of the resulting complex. From these simulations, we also monitored the kinetics of complexation, allowing us to extrapolate the expected time scales for particle assembly on experimentally observed length scales. Lastly, we probed the morphology of the resulting complexes, but calculation of charge distributions within the complex were unable to explain the surface charge stabilization observed in experiment.
Alternate format: The Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog:
Type of Material: Academic dissertations (Ph.D.)
Language: en
Appears in Collections:Chemical and Biological Engineering

Files in This Item:
File Description SizeFormat 
Kozuch_princeton_0181D_13733.pdf24.3 MBAdobe PDFView/Download

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.