Protein-Protein Interactions (PPIs) are central to all regulatory processes in nature and are therefore considered crucial nodes in disease pathways and particularly attractive drug targets. However, the size and complexity of protein interfaces pose significant challenges to drug discovery, as effective ligand design requires knowledge of the interactions at the atomistic and bond-specific level.
We propose the application of bond-to-bond propensity, a graph-theoretical methodology based on the exploration of the edge space of a protein network constructed from the atomic coordinates and chemical interactions between atoms, as an effective method for the structural characterisation of PPIs.
We showcase how this approach can be used to describe different levels of allosteric complexity — from small ligand-binding pockets to large protein-protein interfaces. The scalability of our procedure also allows for the analysis of multiple structures (such as an entire protein family) without unreasonable computational expenditure. The successful application of this methodology to known systems serves as a stepping stone towards de novo prediction of interaction sites from any given crystal structure.
We identify interaction hotspots across Adenosine Monophosphate-activated Protein Kinase (AMPK) and 3-Phosphoinositide-Dependent Protein Kinase-1 (PDK1), validating the use of our approach across both obligate and non-obligate PPIs. Addition- ally, we analyse Ribulose-1,5-Bisphosphate Carboxylase-Oxygenase (RuBisCO) structures, and show how changes in communication across separate protein subunits and in the cooperativity between active sites depend on the structural composition of the protein complex. Thus, we confirm the functional role played by the small subunits in RuBisCO, and suggest an evolutionary rationale for their existence. Finally, we investigate non-obligate PPIs in G Protein-Coupled Receptors (GPCRs), uncovering extracellular allosteric sites and intracellular protein interaction sites across the Glucagon-Like Peptide 1 (GLP1) receptor and several Class A GPCRs.