Computational Characterization of Structural & Thermodynamic Properties of Beta-barrel Membrane Proteins

Beta-barrel membrane proteins are found in the outer membrane of gram-negative bacteria, mitochondria, and chloroplasts. They carry out diverse biological functions, including pore formation, membrane anchoring, enzyme activity, and are often responsible for bacterial virulence. By using a novel statistical mechanical approach that allows the computation of full partition function and thermodynamic properties such as melting temperature, I have developed a method to identify weakly stable regions in the transmembrane (TM) domain and discovered that out-clamps, in-plugs, protein-lipid interactions and oligomerization are four general mechanisms stabilizing the beta-barrel membrane proteins. This method can predict the oligomerization state and can identify the interfaces of protein-protein interaction in the TM region. It is based on fundamental physical principles and only sequence information alone is required. In a blind test, oligomerization state of beta-barrel membrane proteins can be predicted with 100% accuracy, and the protein-protein interfaces can be identified with 86% accuracy. Experiments involving site-directed mutagenesis, SDS-PAGE, CD spectroscopy, tryptophan flourescence, thermal and chemical denaturation have confirmed predictions of important residues in the protein-protein interaction interface of OmpF, VDAC, and Tom40 proteins. Difficulties in experimental determination of structures of beta-barrel membrane proteins have increased the importance of computational modeling for these proteins. I have further developed a computational method called 3D-SPoT for predicting three dimensional structures of the TM domains of beta-barrel membrane proteins. Using a combinatorial model, this method can construct 3D structures of the TM region of beta-barrel membrane proteins using sequence information only. The average RMSD between predicted and true TM region is about 4 Angstrom. This method successfully predicted the structure of the TM domain of VDAC, a mitochondrial membrane protein with no known homologous structure. I also describe a framework to study the assembly of multi-chain proteins into a single beta-barrel structure.