Theses and Dissertations from UMD
Permanent URI for this communityhttp://hdl.handle.net/1903/2
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a give thesis/dissertation in DRUM
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
7 results
Search Results
Item Molecular dynamics simulation and machine learning study of biological processes(2022) Ghorbani, Mahdi; Klauda, Jeffery B; Brooks, Bernard R; Chemical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)In this dissertation, I use computational techniques especially molecular dynamics (MD) and machine learning to study important biological processes. MD simulations can effectively be used to understand and investigate biologically relevant systems with lengths and timescales that are otherwise inaccessible to experimental techniques. These include but are not limited to thermodynamics and kinetics of protein folding, protein-ligand binding free energies, interaction of proteins with membranes, and designing new therapeutics for diseases with rational design strategies. The first chapter includes a detailed description of the computational methods including MD, Markov state modeling and deep learning. In the second chapter, we studied membrane active peptides using MD simulation and machine learning. Two cell penetrating peptides MPG and Hst5 were simulated in the presence of membrane. We showed that MPG enters the model membrane through its N-terminal hydrophobic residues while Hst5 remains attached to the phosphate layer. Formation of helical conformation for MPG helps its deeper insertion into membrane. Natural language processing (NLP) and deep generative modeling using a variational attention based variational autoencoder (VAE) was used to generate novel antimicrobial peptides. These in silico generated peptides have a high quality with similar physicochemical properties to real antimicrobial peptides. In the third chapter, we studied kinetics of protein folding using Markov state models and machine learning. We studied the kinetics of misfolding in β2-microglobulin using MSM analysis which gave us insights about the metastable states of β2m where the outer strands are unfolded and the hydrophobic core gets exposed to solvent and is highly amyloidogenic. In the next part of this chapter, we propose a machine learning model Gaussian mixture variational autoencoder (GMVAE) for simultaneous dimensionality reduction and clustering of MD simulations. The last part of this chapter is about a novel machine learning model GraphVAMPNet which uses graph neural networks and variational approach to markov processes for kinetic modeling of protein folding. In the last chapter, we study two membrane proteins, spike protein of SARS-COV-2 and EAG potassium channel using MD simulations. Binding free energy calculations using MMPBSA showed a higher binding affinity of receptor binding domain in SARS-COV-2 to its receptor ACE2 than SARS-COV which is one of the major reason for its higher infection rate. Hotspots of interaction were also identified at the interface. Glycans on the spike protein shield the spike from antibodies. Our MD simulation on the full length spike showed that glycan dynamics gives the spike protein an effective shield. However, breaches were found in the RBD at the open state for therapeutics using network analysis. In the last section, we study ligand binding to the PAS domain of EAG potassium channel and show that a residue Tyr71 blocks the binding pocket. Ligand binding inhibits the current through EAG channel.Item Examining the role of water and hydrophobicity in folding, aggregation, and allostery(2018) Custer, Gregory Scott; Matysiak, Silvina; Bioengineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Solvation and hydrophobicity drive many critical processes in nature, playing an important role in the folding of proteins, aggregation of surfactants into micelles, and in the disorder to order transitions that occur in some allosteric proteins upon ligand binding. Understanding how solvation and hydrophobicity affect these processes at a molecular level is important to finding new ways to use these processes, but it can be difficult to characterize these molecular details using experimental methods. Molecular dynamics (MD) simulations have proven useful in exploring details and thermodynamic conditions inaccessible in experiment, as MD captures the time evolution of the system at a molecular level. The phenomena which can be studied with an MD simulation depend on the mathematical model employed. Atomistic models provide the most detail for a simulation, but due to the computational costs required are not typically used to study phenomena which require large systems and time scales greater than several μs. Coarse-grained (CG) models reduce the complexity of the system being studied, enabling the exploration of phenomena that occur at longer time scales. We have developed CG models to study protein folding and surfactant aggregation. Our CG surfactant model uses a three-body potential to account for hydrogen bonding without an explicit electrostatic potential, reducing the computational cost of the model. With our surfactant model we studied the stability of non-ionic micelles at extremes of temperature, capturing a window of thermal stability with destabilization of the micelles at both high and low temperatures. We observed changes in structure and solvation of the micelle at low temperatures, with a shift in enthalpy of solvation water providing the driving force for destabilization. Solvation and hydrophobicity are also critical in the folding and stability of proteins. With a modified version of our surfactant model we characterized the folding landscape of a designed sequence which folds to a helical bundle in water. We found two competing folded states which differ by rotation of a helix and trade between hydrophobic packing and solvation of protein's core. Changes in hydrophobic packing can also be involved in the disorder to order transitions that occur upon liganding binding in an allosteric protein, such as the E. Coli biotin ligase/repressor (BirA), in which ligand binding promotes dimerization. We have used atomistic simulations of BirA mutants in collaboration with an experimental group to identify structural changes, accompanied by changes in solvation, at both the dimer interface and ligand binding regions for distal mutations which impact the functionality of BirA.Item DEVELOPING A NEW MODEL OF THE GROEL FUNCTIONAL CYCLE AND ITS IMPLICATIONS FOR THE GROEL-OPTIMIZED SUBSTRATE PROTEIN REFOLDING(2014) Ye, Xiang; Lorimer, George H; Biochemistry; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Despite years of research work, many aspects of the fundamentally important GroEL functional cycle are still in dispute. The work of this dissertation mainly focuses on three major disputes in the field: the identity of the rate determining step (RDS), the physiological order of arrival of ligands (ATP, SP and GroES) to the GroEL trans ring, and the role of the symmetric GroEL-GroES2 "football" complex in the overall chaperonin cycle. With multiple carefully designed spectroscopic probes, a pre-steady state survey has been conducted on the kinetics of the GroEL functional cycle. From the survey, a two cycle model emerges: in the absence of SP, ADP release is the RDS of the asymmetric cycle and consequently, the asymmetric GroEL-GroES1 ,"bullet" which precedes this step, is the pre-dominant species. In this mode, the machine turns over very slowly, minimizing futile ATP consumption. Due to the slow release of ADP, the system turns over in a well defined manner with the two rings operating 180o out of phase of each other, analogous to a two-stroke motor. In the symmetric cycle, which operates in the presence of SP, the release of ADP is greatly accelerated while the intrinsic ATPase activity of GroEL remains unaffected. Consequently ATP hydrolysis becomes the RDS and the symmetric GroEL-GroES2 "football" becomes the predominant species. Contrary to previous chaperonin dogma, the symmetric complex is a highly dynamic species exchanging its two bound ligands, GroES and encapsulated SP from both rings with a half time ~1sec. Switching to a parallel processing machine, the chaperonins turns over rapidly, ultimately driven by stochastic hydrolysis of ATP which causes the symmetric complex to undergo breakage of symmetry (BoS). With such a dynamic system, folding in the `folding cage' seems less important in GroEL-mediated SP refolding as suggested by the passive refolding model. Instead, GroEL may play a more active role in achieving its central biological function as indicated by this two cycle model. This may be the very reason why employing even as low as one GroEL ring per ten SP can achieve SP refolding to a similar extent as using a stoichiometric amount.Item ROLE OF SALT BRIDGES IN GROEL ALLOSTERY(2014) Yang, Dong; Lorimer, George H; Biochemistry; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Chaperonin GroEL facilitates protein folding with two stacked back-to-back, identical rings and the "lid", co-chaperonin GroES. The mis-folded/unfolded substrate protein (SP) adjusts the chaperonin cycling from an asymmetric to a symmetric cycle by catalyzing the release of ADP from the trans ring of GroEL, thus promoting the R to T allosteric transition. ATP binding to the SP bound ring promotes the association of a second GroES and subsequently a GroEL-GroES 2 "football" complex is formed as the folding functional form. However, ADP does release spontaneously, albeit at very slow rate, in the absence of SPs. The intrinsic mechanism by which GroEL relaxes to the lower potential energy T state remains poorly understood. A network of salt bridges forms and breaks during the allosteric transitions of GroEL. Residue D83 in the equatorial domain forms an intra-subunit salt bridge with K327 in the apical domain, and R197 in the apical domain forms an inter-subunit salt bridge with E386 in the intermediate domain. These two salt bridges stabilize the T state and break during the T to R state transition. Removal of these salt bridges by mutation destabilizes the T state and favors the R state of GroEL. These mutations do not alter the intrinsic ATPase activity of GroEL. However, the affinity for nucleotides becomes enhanced and ADP release is hindered such that SP cannot displace the equilibrium to the T state, as normally it does in the wild type. The exchange of ADP to ATP and association of a second GroES is compromised with the following GroEL-GroES 2 "football" formation is hindered. These mutations do not completely eliminate the T state, in the absence of nucleotide, as shown biochemically and by crystal structures. The biased allosteric equilibrium hampers the formation of folding active "football" complex as the mutant GroEL's incompetency to revisit T state in the presence of nucleotide, but not due to the elimination of its T state. This study revealed the critical role of salt bridges in regulating the allosteric transitions of GroEL and conjugated formation of the "football" complex.Item Protein folding and amyloid formation in various environments(2008-11-21) O'Brien, Edward Patrick; Thirumalai, Devarajan; Brooks, Bernard; Chemical Physics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Understanding and predicting the effect of various environments that differ in terms of pH and the presence of cosolutes and macromolecules on protein properties is a formidable challenge. Yet this knowledge is crucial in understanding the effect of cellular environments on a protein. By combining thermodynamic theories of solution condition effects with statistical mechanics and computer simulations we develop a molecular perspective of protein folding and amyloid formation that was previously unobtainable. The resulting Molecular Transfer Model offers, in some instances, quantitatively accurate predictions of cosolute and pH effects on various protein properties. We show that protein denatured state properties can change significantly with osmolyte concentration, and that residual structure can persist at high denaturant concentrations. We study the single molecule mechanical unfolding of proteins at various pH values and varying osmolyte and denaturant concentrations. We find that the the effect of varying solution conditions on a protein under tension can be understood and qualitatively predicted based on knowledge of that protein's behavior in the absence of force. We test the accuracy of FRET inferred denatured state properties and find that currently, only qualitative estimates of denatured state properties can be obtained with these experimental methods. We also explore the factors governing helix formation in peptides confined to carbon nanotubes. We find that the interplay of the peptide's sequence and dimensions, the nanotube's diameter, hydrophobicity and chemical heterogeneity, lead to a rich diversity of behavior in helix formation. We determine the structural and thermodynamic basis for the dock-lock mechanism of peptide deposition to a mature amyloid fibril. We find multiple basins of attraction on the free energy surface associated with structural transitions of the adding monomer. The models we introduce offer a better understanding of protein folding and amyloid formation in various environments and take us closer to understanding and predicting how the complex environment of the cell can effect protein properties.Item SINGLE MOLECULE FLUORESCENCE INVESTIGATION OF PROTEIN FOLDING(2008-09-30) liu, jianwei; Munoz, Victor; Chemical Physics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)In addition to the well known two-state folding scenario, the energy landscape theory of protein folding predicts the possibility of downhill folding under native conditions. This intriguing prediction was extended by Victor Muñoz and coworkers to include global downhill folding. i.e. a barrierless free energy surface and unimodal conformational distributions at all degrees of unfolding stress. A small protein, BBL, has been shown to follow this behavior as evidenced from experiments and simulations. However, the identification of BBL as a global downhill folder has raised a significant amount of controversy with some groups claiming that it still folds in a two-state fashion. The objective of this thesis is to characterize the conformational distribution of BBL using single molecule Förster resonance energy transfer (SM-FRET) to obtain direct evidence for the downhill folding in BBL. We carried out SM-FRET measurements at 279 K to slow down the protein dynamics to 150 μs thus enabling the use of a 50 μs binning time (the short binning time being a first in SM measurements). By optimizing the microscope system setup and employing a novel Trolox-cysteamine fluorophore protection system, we obtained sufficient signal to construct reliable 50 μs SM-FRET histograms. The data show clear unimodal conformational distributions at varying denaturant concentrations thus demonstrating the downhill folding nature of BBL. Further SM-FRET measurements on a two-state folder, α-spectrin SH3 produced bimodal histograms indicating that our experimental setup works well and that the unimodal distributions of BBL are not due to instrumental errors. The comparison of ensemble FRET measurements on labeled proteins (both BBL and α-spectrin SH3) with CD measurements on the corresponding unlabeled proteins shows that the fluorophores do not affect the protein stability. We also simulated the expected histograms if BBL were a two-state folder using Szabo's photon statistics theory of SM-FRET. The two-state simulation results are inconsistent with the experimental histograms even under very conservative assumptions about BBL's relaxation time. Therefore, all the control experiments and simulations exclude any possible artifacts, which shows our results are quite robust. Additionally, we estimated the relaxation time of BBL from the histogram width analysis to be consistent with independent kinetic measurements.Item One-Dimensional Free Energy Surface Models of Protein Folding: Connecting Theory and Experiments(2007-04-27) Doshi, Urmi Rajnikant; Munoz, Victor; Biochemistry; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Experimental techniques have now reached the sub-microsecond timescale necessary to study fast events in protein folding. However, analysis of fast folding experiments still commonly rely on conventional procedures that provide an oversimplified picture i.e. an all-or-none transition between the unfolded and native states, which is not valid for all cases. Moreover, due to the presence of discrepancies between theoretical predictions and experimental observations, discerning the correct mechanisms of protein folding becomes difficult. This is true even for the most elementary processes such as a-helix formation. Recent laser-induced temperature jump experiments on a-helical peptides have revealed unprecedented complexity in relaxation kinetics. These observations are suggested to be incompatible with the nucleation-elongation theory for a-helix formation. However, the detailed kinetic model based on nucleation-elongation theory developed in this work quantitatively reproduces all the observed complex kinetics. The results are rationalized using a simple one-dimensional projection of free energy surface. It is concluded that the observed probe-dependent and thermal perturbation size-dependent multiphasic relaxation kinetics are consequences of helix fraying and heterogeneity of peptide sequence. Remarkably, all the kinetic behaviors predicted by the detailed model are successfully reproduced by diffusion on one-dimensional free energy surface. The one-dimensional free energy approach thus validated empirically is then extended for the analysis of protein folding experiments. For this purpose a simple mean field model is formulated that is consistent with the size-scaling properties of thermodynamic parameters as well as with the observation of entropy convergence at high temperatures. The model describes the effects of chemical and thermal denaturation, making it amenable for direct comparison with experimental observables i.e. folding rates and heat capacity changes on a quantitative level. The main advantage of the model is the treatment in which free energy barrier on one-dimensional profile is allowed to modulate by just one parameter, that can be directly related to protein size, structure- and sequence- dependent energetics. Recently the one-dimensional free energy surface model has been applied for analyzing the dependence of rates on temperature and chemical denaturant in fast folding proteins. This analysis has allowed simultaneous investigation of energetic and dynamic factors governing folding kinetics. Unlike traditional methods the model serves as an analytical tool without making any a priori assumptions about the presence of a barrier. With its simplicity and versatility the model provides the foundation for exploring general trends in protein folding as well as prediction of folding properties at the level of individual proteins.