Experimental and computation study of protein interactions with lipid nanodomains
MetadatosMostrar el registro completo del ítem
Protein lipid interactions are significantly relevant to understanding of a wide variety of biological phenomena in general. In particular, human beta-amyloid protein is closely related to the pathogenesis of Alzheimer's disease. Due to its high propensity to self-aggregate, beta-amyloid protein is difficult to study with experiments. Molecular dynamics simulations is capable of providing atomistic details of the protein lipid interactions; therefore, is an important theoretical tool to investigate these subtle interactions and offer insights to the pathogenesis of Alzheimer's disease. In this dissertation, I studies the protein lipid interactions with several systems with different lipid composition and protein conformations. I developed computational tools to quantitatively analyze lipid perturbations due to protein interactions, since it is commonly believed that the neurotoxicity of beta-amyloid protein is through perturbation of the lipid membrane. I discovered that for the case of a beta-amyloid dimer on the surface of lipid bilayers, the perturbation effect of protein is correlated to the degree of disorder of the protein in term of its secondary structure. Meanwhile, for a system where a beta-amyloid protein was partially inserted into the bilayer, the protein insertion rate was regulated by both the secondary structure of the protein and the lipid environment. Especially, a scaling relation between the insertion rate and degree of disorder was found. Even though molecular dynamics simulations is a powerful tool in studying atomistic protein lipid interactions, it is not efficient in sampling the free energy landscape of the system; hence results are biased by the initial structure of the system. I developed a multiscale molecular simulation scheme to increase the efficiency in free energy landscape sampling by switching the system between different spatial resolutions, i.e., atomistic and coarse-grain representations of the system. Using this method, I discovered a novel protein lipid orientation, which has implications in understanding the biochemical pathway of the protein as well as developing therapeutic interventions. Finally, I also developed a Monte Carlo method to estimate molecule volumes accurate to atomistic scale. This method is directly applicable to lipid membrane system with heterogeneous components including proteins; it is a useful tool for not only investigating protein lipid interactions but also calibration of force field parameters for classical molecular dynamics simulations.