Predicting drug targets by homology modelling of Pseudomonas aeruginosa proteins of unknown function

Abstract

Efficacies of antibiotics to treat bacterial infections rapidly decline due to antibiotic resistance. This stimulated the development of novel antibiotics, but most attempts failed. As a response, the idea of mining uncharacterised genes of pathogens to identify potential targets for entirely new classes of antibiotics raised. Without knowing the biochemical function of a protein it is difficult to validate its potential for drug targeting; therefore progress in the functional characterisation of bacterial proteins of an unknown function must be accelerated. Here we present a paradigm for comprehensively predicting biochemical functions of a large set of proteins encoded by hypothetical genes in human pathogens, to identify candidate drug targets. A high-throughput approach based on homology modelling with ten templates per target protein was applied on the set of 2103 P. aeruginosa proteins encoded by hypothetical genes. Obtained >21000 homology modelling results and available biological and biochemical information about several thousand templates was scrutinised to predict the function of reliably modelled proteins of unknown function. This approach resulted in assigning, one or often multiple, putative functions to hundreds of enzymes, ligand-binding proteins and transporters. New biochemical functions were predicted for 41 proteins whose essential or virulence-related roles in P. aeruginosa were already experimentally demonstrated. Eleven of them were shortlisted as promising drug targets which participate in essential pathways (maintaining genome and cell wall integrity), virulence-related processes (adhesion, cell motility, host recognition) or antibiotic resistance, which are general drug targets. These proteins are conserved among other WHO priority pathogens but not in humans, therefore they represent high-potential targets for pre-clinical studies. These and many more biochemical functions assigned to uncharacterised proteins of P. aeruginosa, available as PaPUF database may guide the design of experimental screening of inhibitors which is a crucial step toward validation of the most potential targets for the development of novel drugs against P. aeruginosa and other high-priority pathogens.