On the binding affinity of macromolecular complexes : daring to ask why proteins interact

The last twenty years we have reached the conclusion that most of the cellular functions are orchestrated by interacting protein molecules. It has also become clear that modifying or preventing these protein-protein interactions may have great therapeutic potential, especially for curing diseases such as cancer and Alzheimer’s where protein-protein interactions (PPIs) are of central role. Despite all efforts in understanding cellular complexity, either by identifying hundreds of thousands of biologically-related protein-protein interactions or by determining the three-dimensional structures of hundreds of these, we haven’t yet understood how and why all these molecules can function together. Each chapter of this thesis tackles exactly the aforementioned puzzle, following a holistic approach to understand protein-protein interactions based on three specific steps: (a) Protein-Protein Interaction Classification. Categorize protein-protein interactions according to available functional, structural and biochemical data. Structural data must be available for both reactants and products of any interaction whereas binding constants must also be available for the specific interactions. (b) Structure-Affinity Connection. Relate structural or biophysical properties of protein-protein interactions to their binding affinity. In this way, all current knowledge concerning factors governing the interactions of proteins will be put to the test. Furthermore, factors previously neglected that may influence complexation or dissociation can be identified. (c) Data Rationalization and Causality Interpretation. Findings from (b), concerning the categorization performed in (a) are validated through independent datasets and interpreted considering physical forces and/or evolutionary conservation. Results from this Ph.D. Thesis show that contributors to binding affinity of protein-protein interactions must be re-evaluated, with as key points: (1) Buried Surface Area, interface properties and accompanying interfacial models partially relate to binding affinity, conformational changes being an additional limiting factor for binding affinity prediction. (2) Properties of the Non-interacting surface are causally affecting binding affinity through long-range electrostatics and surface-solvent interactions. (3) Interface waters are critical for accurate structure prediction as reflected by the improved prediction performance when explicitly accounted for. Water can assist the formation of the native protein-protein complex by water-mediated hydrogen bonds and salt bridges. (4) The Buried Surface Area is a critical descriptor for binding, especially for small-molecule inhibitors of protein-protein interactions. However, its contribution to affinity is decreasing with increasing degrees of freedom (and conformational changes) of the biomolecular reactants under study. In conclusion, the benchmarks assembled and analyzed in this thesis and the novel contributions to the theory about structure-affinity relations of protein-protein interactions provide critical information to understand and model macromolecular recognition. The impact on current and future protein research should be obvious: This thesis will have a major contribution to our understanding of how and why these proteins interact. It opens the route to the design of novel interactions (or modulators of these interactions) where alternate metabolic pathways can be exploited at will in order to fight diseases such as cancer and Alzheimer’s.