An In silico Algorithm for Identifying Amino Acids that Stabilize Oligomeric Membrane-Toxin Pores through Electrostatic Interactions

Abstract

Pore forming toxins (PFTs) are a class of proteins which have specifically evolved to form unregulated pores in target plasma membranes, and represent the single largest class of bacterial virulence factors. With increasingly prevalent antibiotic-resistant bacterial strains, next generation therapies are being developed to target bacterial PFTs rather than the pathogens themselves. However, structure-based design of inhibitors that could block pore formation are hampered by a paucity of structural information about pore intermediates. On similar lines, observations of the inter-subunit interfaces in fully-formed pore complexes to identify druggable residues, whose interactions could potentially be blocked to hamper pore formation or destabilize pore assemblies, are often limited because of the presence of a large number of protein-protein interaction sites across pore inter-subunit interfaces. Narrowing down the list of plausible target residues requires a quantitative assessment of their contributions towards pore stability, which cannot be gleaned from a single, static, crystal or cryo-EM pore structure. We overcome this limitation by developing an in silico screening algorithm that employs fully atomistic molecular dynamics simulations coupled with knowledge-based screening to identify residues engaged in persistent and stabilizing electrostatic interactions across inter-subunit interfaces in membrane-inserted PFT pores. Application of this algorithm to prototypical α-PFT (cytolysin A) and β-PFT (α-hemolysin) pores yielded a small predicted subset of highly interacting residues, blocking of which could destabilize pore complexes as shown in previous mutagenesis experiments for some of these predicted residues. The algorithm also yielded a novel set of residues in both cytolysin A and α-hemolysin pores for which no mutagenesis and stability data exists to the best of our knowledge, and therefore could serve as hitherto un-recognised potential targets for PFT inhibitors. The algorithm worked equally well for both α and β-PFT pores, and could thus be potentially applicable to all pores with known structures to generate a database of pore-destabilizing mutations, which could then serve as a starting point for experimental validation and structure-based PFT-inhibitor design.