Identifying Novel Therapeutics for the Resistant Mutant "F533L" in PBP3 of Pseudomonas aeruginosa Using ML Techniques.

Abstract

Pseudomonas aeruginosa (P. aeruginosa) is a highly infectious and antibiotic-resistant bacterium, which causes acute and chronic nosocomial infections. P. aeruginosa exhibits multidrug resistance due to the emergence of resistant mutants. The bacterium takes advantage of intrinsic and acquired resistance mechanisms to resist almost every antibiotic. To overcome the drug-resistance problem, there is a need to develop effective drugs against antibiotic-resistant mutants. Therefore, in this study, we selected the F533L mutation in PBP3 (penicillin-binding protein 3) because of its important role in β-lactam recognition. To target this mutation, we screened 147 antibacterial compounds from PubChem through a machine-learning model developed based on the decision stump algorithm with 75.75% accuracy and filtered out 55 compounds. Subsequently, out of 55 compounds, 47 compounds were filtered based on their drug-like activity. These 47 compounds were subjected to virtual screening to obtain binding affinity compounds. The binding affinity range of all 47 compounds was -11.3 to -4.6 kcal mol-1. The top 10 compounds were examined according to their binding with the mutation point. A molecular dynamic simulation of the top 8 compounds was conducted to understand the stability of the compounds containing the mutated PBP3. Out of 8 compounds, 3 compounds, namely, macozinone, antibacterial agent 71, and antibacterial agent 123, showed good stability and were validated by RMSD, RMSF, and binding-free analysis. The findings of this study revealed promising antibacterial compounds against the F533L mutant PBP3. Furthermore, developments in these compounds may pave the way for novel therapeutic interventions.