Files
Abstract
Protein-ligand binding is among the most important interactions in physiological events. There has been enormous endeavor from the science community to study and characterize various systems involving protein-ligand interactions. Several experimental techniques have been developed and employed in studying protein-ligand binding, including but not limited to, immunoprecipitation, X-ray crystallography, cryogenic electron microscopy (cryo-EM), nuclear magnetic Resonance (NMR), surface plasmon resonance (SPR), biolayer interferometry (BLI), etc. However, experimental approaches can frequently be extremely laborious, resulting in difficulty in being employed on very large scales and relatively long production cycles. On the other hand, computational modeling, a newly emerging field with increasing potential, can be employed and has been employed in studying protein-ligand binding. Given its relatively high-throughput nature, computational modeling can aid experimentalists by prioritizing effort towards the most promising targets. In this thesis, research work is dedicated to computational simulation of protein-ligand binding, with a focus on carbohydrate-based ligands. A series of mono- or biantennary novel glycopeptides has been synthesized by our collaborator, with their binding affinity to wheat germ agglutinin characterized experimentally. Molecular dynamic (MD) simulations of these binding events performed in this thesis provided visualization and theoretical explanation that bivalent binding is indeed possible and is likely responsible for stronger binding affinities. Given the significance of the non-classical π interactions in protein-ligand binding, structural-informatic and quantum-mechanical (QM) study was performed for the cation-π and CH-π interaction. Molecular-mechanical functional forms were also developed and validated to model the two types of interactions. Finally, an automated high-throughput virtual screening protocol has been programmed for the generation of a large number of glycomimetic analogs based on a system involving an endogenous carbohydrate ligand and its target receptor protein. MD simulations were performed for each such analog, with average binding affinity computed from the ensemble trajectory.