Molecular recognition: From statistical mechanics to statistics of protein interfaces.

摘要

Understanding protein-protein interactions is vital for a complete picture of cellular processes. This thesis describes both quantitative and qualitative approaches to studying molecular recognition. At the quantitative level, calculations are performed on model protein systems, evaluating the energetics of short beta-turn forming peptides, and chemically modified hydrogen bonds in a helix of T4 lysozyme. This work, using numerical solutions to the Poisson-Boltzmann equation, coupled with standard molecular mechanics force fields, provides a physical basis for understanding the detailed energetics of the interactions of biological molecules in water. To address biochemical questions regarding protein-protein interaction on a broader, more comparative level, a comprehensive database of protein-protein interfaces, using surface properties defined in Grasp. This database, SPIN (Surface Properties of Interfaces) was designed as a World Wide Web tool to access a large variety of structural, chemical, and informatic data on protein surfaces, and represents the first attempt to organize protein structures based on molecular surface properties. Distinctive features of protein-protein interfaces were cataloged, and correlated with functional information, and a taxonomic classification scheme was developed.