Abstract

Mycobacterium abscessus is among the most frequently isolated species in nontuberculous mycobacterial infections and is notoriously challenging to eliminate. To date, there is no universally established treatment regimen proven to be consistently effective against M. abscessus. To accelerate anti-tuberculosis drug discovery, the computational screening has emerged as an attractive and efficient strategy for identifying new therapeutic options against tuberculosis. Our screening approach leverages machine learning in combination with molecular docking using AutoDock and molecular dynamics simulations, and has been previously applied to Mycobacterium tuberculosis. We identified three novel compounds—Izumiphenazines, Hodgkinsine, and Phenazine, 1, 1'-ethylenebis [2-amino]—with strong predicted interaction with M. abscessus NADH pyrophosphatases (NudC). Structural clustering analysis and ADMET-profiling suggest favourable drug-like properties. Docking and molecular dynamics simulations revealed plausible binding modes and mechanisms of action. These findings underscore the potential of computational hit screening to accelerate therapeutic discovery against M. abscessus. They also provide valuable insights for developing and optimizing NudC inhibitors targeting drug-resistant mutations.