The cellular function of proteins, and their targeting by drug applications, are both governed by biomolecular thermodynamics and kinetics. In order to make meaningful and efficient predictions of these mechanisms, molecular simulations must be able to estimate the binding affinity and rates of association and dissociation of a protein-ligand complex, or the populations and rates of exchange between distinct conformational states (i.e. folding and unfolding, binding and unbinding). The above studies are typically done using different, but complementary approaches. Alchemical methods, including free energy perturbation (FEP) and thermodynamic integration (TI), have become the dominant method for computing high-quality estimates of protein-ligand binding free energies. In particular, the widely-used approach of relative binding free energy calculation can deliver accuracies within 1 kcal mol−1. However, detailed physical pathways and kinetics are missing from these calculations. In principle, all-atom molecular dynamics (MD) simulation, with the help of Markov State Models (MSMs), can be used to obtain this information, yet finite sampling error still limits MSM approaches from making accurate predictions for very slow unfolding or unbinding processes. To overcome these issues, a new approach called multiensemble Markov models (MEMMs) have been developed, in which sampling from biased thermodynamic ensembles can be used to infer states populations and transition rates in unbiased ensembles. In this dissertation, two distinct biophysical problems are investigated. In the first part, we apply expanded ensemble (EE) methods to accurately predict relative binding free energies for a series of protein-ligand systems. Moreover, we propose a simple optimization scheme for choosing alchemical intermediates in free energy simulations. In the second part, we employ MEMMs to estimate the free energies and kinetics of protein folding and ligand binding, to achieve greatly improved predictions. Finally, we combine the above EE method and a maximum-caliber algorithm to study how sequence mutations perturb protein stability and folding kinetics. In summary, this work comprises a wide range of current methodology in biophysical simulation, complementing and improving upon existing approaches. / Chemistry

Protein-small molecule interactions play a central role in various aspects of the structural and functional organization of the cell and are therefore integral for drug discovery. The most comprehensive structural characterization of small molecule binding sites is provided by X-ray crystallography. However, it is often time-consuming and challenging to perform direct experimental analysis. Therefore, it is necessary to have computational methods that can predict binding site locations on unbound structures with accuracy close to that provided by X-ray crystallography. This thesis details four projects which involve the development of a fragment benchmark set, evaluation of allosteric sites in G Protein-Coupled Receptors (GPCRs), computational modeling of binding pocket dynamics, and the development of an Application Program Interface (API) framework for High-Performance Computing (HPC) centers. The first project provides a benchmark set for testing hot spot identification methods, emphasizing application to fragment-based drug discovery. Using the solvent mapping server, FTMap, which finds small molecule binding hot spots on proteins, we compared our benchmark set to an existing benchmark set that with a different method of construction. The second project details the effort to identify allosteric binding sites on GPCRs. We demonstrate that FTMap successfully identifies structurally determined allosteric sites in bound crystal structures and unbound structures. The project was further expanded to evaluate the conservation of allosteric sites across different classes, families, and types of GPCRs. The third project provides a structure-based analysis of cryptic site openings. Cryptic sites are pockets formed in ligand-bound proteins but not observed in unbound protein structures. Through analysis of crystal structures supplemented by molecular dynamics (MD) with enhanced sampling techniques, it was shown that cryptic sites can be grouped into three types: 1) “genuine” cryptic sites, which do not form without ligand binding, 2) spontaneously forming cryptic sites, and 3) cryptic sites impacted by mutations or off-site ligand binding. The fourth project presents an API framework for increasing the accessibility of HPC resources.

<p> The Glucagon-like peptide 1 receptor (GLP-1R) belongs to the pharmaceutically important Class B family of G-protein coupled receptors (GPCRs) and its incretin peptide ligand GLP-1 analogs are adopted drugs for the treatment of type 2 diabetes (T2D). Despite remarkable anti-diabetic effects, Glucagon Like Peptide-1 (GLP-1) peptide-based drugs are limited by the need of injection or high cost oral formulation. On the other hand, developing non-peptide small molecule drugs targeting GLP-1R remains elusive likely due to the large nature of the orthosteric binding site on GLP-1R. A promising approach is to develop small molecule agonistic positive allosteric modulators (ago-PAMs) or positive allosteric modulators (PAMs) of GLP-1R by targeting the potential allosteric sites in the transmembrane (TM) domain of human GLP-1R. </p><p> As the first step of taking this approach, we constructed a three-dimensional structure model of the TM domain of human GLP-1R using homology modeling and conformational sampling techniques. Next, a potential allosteric binding site on the TM domain was predicted computationally. <i>In silico</i> screening of drug-like compounds against this predicted allosteric site has identified nine compounds as potential GLP-1R agonists. The independent agonistic activity of two compounds was subsequently confirmed using cyclic adenosine monophosphate (cAMP) response element (CRE)-based luciferase reporting system. One compound was also shown to stimulate insulin secretion through <i> in vitro</i> assay. In addition, this compound synergized with GLP-1 to activate human GLP-1R. </p><p> In 2017, the crystal structures of GLP-1R in its active state (PDB ID: 5VAI) became available. Hence, we have performed another round of <i> in silico</i> screening employing this structure. First, the potential ligand binding sites in 5VAI were identified using computational tools and <i> in silico</i> screening procedure as described above was carried out again. A new small 8 molecule with low molecular weight and logP was identified. <i> In vitro</i> studies of this compound confirmed that it acts as the ago-Positive Allosteric Modulator (PAM) of GLP-1R that improves GLP-1's affinity and efficacy towards GLP-1R. When used in combination with GLP-1, this compound improves insulin secretion than using GLP-1 alone. Site specific mutagenesis studies confirmed its binding site as predicted in the TM domain of GLP-1R. </p><p> Finally, this ago-PAM molecule was further optimized to improve its potency and specificity towards GLP-1R using structure-based optimization strategy and medicinal synthesis. The newly designed compound, whose molecular weight was less than the parental compound, was found to act as the PAM of GLP-1R and showed improvement in the specificity than the parental compound. Thus, this new compound could be further exploited in the drug development for T2D treatment. </p><p> These results demonstrated that allosteric regulation exists in GLP-1R and can be exploited for developing small molecule agonists. The success of this work will help pave the way for small molecule drug discovery targeting other Class B GPCRs through allosteric regulations.</p><p>

<p> Block copolymers are composed of chemically distinct polymer chains that can be covalently linked in a variety of sequences and architectures. They are ubiquitous as ingredients of consumer products and also have applications in advanced plastics, drug delivery, advanced membranes, and next generation nano-lithographic patterning. The wide spectrum of possible block copolymer applications is a consequence of block copolymer self-assembly into periodic, meso-scale morphologies as a function of varying block composition and architecture in both melt and solution states, and the broad spectrum of physical properties that such mesophases afford. </p><p> Materials exploration and discovery has traditionally been pursued through an iterative process between experimental and theoretical/computational collaborations. This process is often implemented in a trial-and-error fashion, and from the computational perspective of generating phase diagrams, usually requires some existing knowledge about the competitive phases for a given system. Self-Consistent Field Theory (SCFT) simulations have proven to be both qualitatively and quantitatively accurate in the determination, or forward mapping, of block copolymer phases of a given system. However, it is possible to miss candidates. This is because SCFT simulations are highly dependent on their initial configurations, and the ability to map phase diagrams requires a priori knowledge of what the competing candidate morphologies are. The unguided search for the stable phase of a block copolymer of a given composition and architecture is a problem of global optimization. SCFT by itself is a local optimization method, so we can combine it with population-based heuristic algorithms geared at global optimization to facilitate forward mapping. In this dissertation, we discuss the development of two such methods: Genetic Algorithm + SCFT (GA-SCFT) and Particle Swarm Optimization + SCFT (PSO-SCFT). Both methods allow a population of configurations to explore the space associated with the numerous states accessible to a block copolymer of a given composition and architecture. </p><p> GA-SCFT is a real-space method in which a population of SCFT field configurations &ldquo;evolves&rdquo; over time. This is achieved by initializing the population randomly, allowing the configurations to relax to local basins of attraction using SCFT simulations, then selecting fit members (lower free energy structures) to recombine their fields and undergo mutations to generate a new &ldquo;generation&rdquo; of structures that iterate through this process. We present results from benchmark testing of this GA-SCFT technique on the canonical AB diblock copolymer melt, for which the theoretical phase diagram has long been established. The GA-SCFT algorithm successfully predicts many of the conventional mesophases from random initial conditions in large, 3-dimensional simulation cells, including hexagonally-packed cylinders, BCC-packed spheres, and lamellae, over a broad composition range and weak to moderate segregation strength. However, the GA-SCFT method is currently not effective at discovery of network phases, such as the Double-Gyroid (GYR) structure. </p><p> PSO-SCFT is a reciprocal space approach in which Fourier components of SCFT fields near the principal shell are manipulated. Effectively, PSO-SCFT facilitates the search through a space of reciprocal-space SCFT seeds which yield a variety of morphologies. Using intensive free energy as a fitness metric by which to compare these morphologies, the PSO-SCFT methodology allows us to agnostically identify low-lying competitive and stable morphologies. We present results for applying PSO-SCFT to conformationally symmetric diblock copolymers and a miktoarm star polymer, AB₄, which offers a rich variety of competing sphere structures. Unlike the GA-SCFT method we previously presented, PSO-SCFT successfully predicts the double gyroid morphology in the AB-diblock. Furthermore, PSO-SCFT successfully recovers the A₁₅ morphology at a composition where it is expected to be stable in the miktoarm system, as well as several competitive metastable candidates, and a new sphere morphology belonging to the hexagonal space group 191, which has not been seen before in polymer systems. Thus, we believe the PSO-SCFT method provides a promising platform for screening for competitive structures in a given block copolymer system.</p><p>

The structures, vibrational frequencies, and potential energy functions of several molecules in their ground and excited electronic states were determined using various spectroscopic and theoretical methods. High-level ab initio and density functional theory (DFT) calculations were utilized to investigate the previously reported structures and vibrational spectra of 1,3- disilacyclobutane (13DSCB) and its 1,1,3,3-d4 (13DSCB-d4) isotopomer. These calculations confirmed the finding from earlier microwave work that the CSiC angles of the 13DSCB ring are unexpectedly larger than the SiCSi angles. The calculated vibrational spectra using density functional theory agreed well with the experimental data and showed CH2 modes to have unusually low values. The calculations also confirmed that the individual molecules in the vapor phase are puckered whereas in the solid they become planar. The one-dimensional potential energy surfaces (PESs) for the ring inversion vibration of 2-cyclohexen-1-one and its 2,6,6-d3 isotopomer in its ground and singlet S1(n,π*) electronic states were determined using ultraviolet cavity ringdown spectroscopy (CRDS). The CRDS data allowed several of the quantum states of the ring inversion vibration to be determined for both the ground and excited electronic states, and the data were fit very well with PESs with high barriers to inversion. The infrared and Raman spectra and DFT calculations were utilized to complete a vibrational assignment of 2CHO and 2CHO-d3. A remarkable agreement was seen between the experimental and calculated spectra. The fluorescence excitation spectra (FES) and the single-vibronic level fluorescence (SVLF) spectra of jet-cooled 1,4-dihydronaphthalene (14DHN) were acquired to determine its ring-puckering potential energy function for the ground and singlet S1(π,π*) electronic states. Ultraviolet, infrared, and Raman spectra were also recorded to complement the analysis. The potential energy functions showed that the molecule is planar in both the ground and S1(π,π*) states. A complete vibrational assignment was carried out for 14DHN using the infrared and Raman data and aided by DFT calculations. The ab intio calculations carried out on 2-methyl-2-cyclopenten-1-one (2MCP) showed that the molecule can have 3 different conformers. Infrared and Raman spectra of the liquid-phase molecule were recorded and analyzed to complement the theoretical calculations.

The structures, vibrational frequencies, and potential energy functions of several molecules in their ground and excited electronic states were determined using various spectroscopic and theoretical methods. High-level ab initio and density functional theory (DFT) calculations were utilized to investigate the previously reported structures and vibrational spectra of 1,3- disilacyclobutane (13DSCB) and its 1,1,3,3-d4 (13DSCB-d4) isotopomer. These calculations confirmed the finding from earlier microwave work that the CSiC angles of the 13DSCB ring are unexpectedly larger than the SiCSi angles. The calculated vibrational spectra using density functional theory agreed well with the experimental data and showed CH2 modes to have unusually low values. The calculations also confirmed that the individual molecules in the vapor phase are puckered whereas in the solid they become planar. The one-dimensional potential energy surfaces (PESs) for the ring inversion vibration of 2-cyclohexen-1-one and its 2,6,6-d3 isotopomer in its ground and singlet S1(π,π*) electronic states were determined using ultraviolet cavity ringdown spectroscopy (CRDS). The CRDS data allowed several of the quantum states of the ring inversion vibration to be determined for both the ground and excited electronic states, and the data were fit very well with PESs with high barriers to inversion. The infrared and Raman spectra and DFT calculations were utilized to complete a vibrational assignment of 2CHO and 2CHO-d3. A remarkable agreement was seen between the experimental and calculated spectra. The fluorescence excitation spectra (FES) and the single-vibronic level fluorescence (SVLF) spectra of jet-cooled 1,4-dihydronaphthalene (14DHN) were acquired to determine its ring-puckering potential energy function for the ground and singlet S1(π,π*) electronic states. Ultraviolet, infrared, and Raman spectra were also recorded to complement the analysis. The potential energy functions showed that the molecule is planar in both the ground and S1(π,π*) states. A complete vibrational assignment was carried out for 14DHN using the infrared and Raman data and aided by DFT calculations. The ab intio calculations carried out on 2-methyl-2-cyclopenten-1-one (2MCP) showed that the molecule can have 3 different conformers. Infrared and Raman spectra of the liquid-phase molecule were recorded and analyzed to complement the theoretical calculations.

Dialkylstannylene acetals are tin-containing species employed extensively as intermediates to facilitate high-yielding and regioselective monosubstitution reactions of diols or polyols with various electrophiles, which is an important application of organotin compounds in organic synthesis. Although an abundance of experimental studies of these reactions have been reported, the mechanism of the reaction has not been well defined. High-level theoretical methods are used in this thesis to investigate the chemistry of organotin systems at a molecular level. This involves the exploration of the geometry characteristics of the gas-phase structures along the reaction paths in order to understand the mechanism of the organotin-mediated alkylations of diols. Alkylation reactions which require strict conditions can be dramatically enhanced by the presence of nucleophiles. The effects of added nucleophiles were examined computationally by comparing reaction profiles obtained for alkylations of dimethylstannylene acetals in the presence of different nucleophiles.

Univ., Diss., 2009--Marburg.

Mannheim, Univ., Diss., 2002.

Darmstadt, Techn. Univ., Diss., 2000.