Protein-protein interactions: a structural bioinformatics approach

Molecular function in cellular processes is governed by protein-protein interactions (PPIs). With the exponential growth of PPI in the drug discovery field, understanding the key principles governing PPI is of immense current interest. Investigation of protein interfaces of known complexes is an important step towards understanding the molecular basis of PPIs. The overall objective of this thesis is to study known PPI complexes from the Protein Data Bank (PDB) using computational tools to capture their driving force and relate structural interfaces to their biological functions. PPI features, analysis data and conclusions drawn are documented to facilitate prediction of interaction sites and partners and also to facilitate prediction of potential protein function of novel complexes. The interface features were analysed for all non-redundant protein heterodimers (278) in the PDB. The relative interface-surface polarities of each complex in the dataset were estimated to understand predominant forces driving binding. Structural analysis revealed two classes of interfaces - class A with less polar residues and class B with more polar residues, at the interface than the rest of the surface. Five distinguishing features (interface area, interface property abundance, interface charged residues, solvation free energy gain, binding energy) among these classes were identified. These results verify the need for classification of complexes based on residue-level properties in determining the features driving binding. Also, all functional categories are represented in the interface classes. This led to the study on relating structural features to their biological functions. PPIs are essential for catalysis, regulation, assembly, immunity and inhibition in a cell. However, it is unclear whether structural features can define protein functionality. Therefore, analysis of non-redundant protein complexes has been carried out to determine the structural basis for functional preferences. Structural interface of each complex has been characterized using a range of physico-chemical properties. The dataset is grouped using known function for molecular preferences. Five interface features (interface area, interface property abundance, hydrogen bonds, salt bridges, solvation free energy gain, and binding energy) are observed to be significantly different among functional groups. Preliminary application of using PPIs for the characterisation of protein interfaces in integrin αvβ6 heterodimer and its interactions with other proteins especially urokinase plasminogen activator receptor (uPAR) is carried out. The integrin αvβ6•uPAR interaction promotes cancer progression. Therefore, a comprehensive analysis of αvβ6 using modelling data and docking simulations helped gain insights into binding of αvβ6 with uPAR suggesting an interaction site. These results provide preliminary evidence for potential targets in cancer therapies. In conclusion, the work presented in this thesis investigates interface features of known protein complexes to gain insights into the binding principles of PPIs. Structural analysis of heterodimer dataset and grouping complexes based on interface classes and functional groups lead to the identification of discriminatory features amongst these groups. Incorporation of these combinatorial features is necessary to develop models for PPI prediction and analysis, and also in utilizing PPI information for the prediction of potential functions in future studies. Novel observations using modelling and docking data to obtain significant information on key PPIs (involved in cancer) are discussed.