Structure/function studies of membrane proteins: From molecular modeling and ligand binding to protein -protein interactions of the dopamine transporter

摘要

This thesis describes theoretical and computational efforts to develop a model of the human dopamine transporter (DAT), and its use to study relationships between structure and function, including the regulation of DAT by ligands and protein-protein interactions. In the process of developing a model of DAT, several key problems in the field of membrane protein (MP) structure/function have been addressed. These include (1) the development of computational tools for predicting structural properties from sequence data, (2) the application of such predictive tools in modeling MPs based on low resolution electron microscopy maps, (3) the use of experimental data and bioinformatics tools to refine structure based sequence alignments of MPs, (4) the mode of binding of small molecules to MPs, and (5) the interactions of MPs with regulatory proteins such as PDZ domains. These problems are characterized by different levels of available structural detail, and specific solutions to each problem had to be developed in each case. Thus, when no structures were available for modeling, prediction algorithms were developed that were entirely based on sequence analysis, and which yielded valuable information on the secondary structure of MPs. Such prediction algorithms were then used to complement constraints from low resolution structural data in creating atomic models of MPs. When a low similarity template became available for molecular modeling of DAT, it was necessary to use bioinformatics tools and experimental data to refine a sequence alignment, in order to generate atomic models. These models of DAT were used for ligand docking studies, and revealed binding modes for substrates and inhibitors. In cases where many structures and sequences were available, integration of this information yielded insights into specificity. To study the interaction of DAT with PDZ domains, a combination of protein-protein interaction databases and molecular dynamics simulations was used to describe specificity determinants.