Machine learning-driven discovery of aniline phenylacetate derivatives as LasR-targeted biofilm inhibitors against Pseudomonas aeruginosa infections

Pseudomonas aeruginosa biofilm infections pose a severe clinical challenge due to antibiotic resistance driven by quorum sensing (QS). As the master regulator of the P. aeruginosa QS hierarchy, LasR controls biofilm initiation and key virulence factors, establishing it as a high-value therapeutic target. Here, we developed a machine learning pipeline integrating graph neural networks, pharmacophore modeling, molecular docking and molecular dynamics, and determined that the aniline-phenylacetic acid derivative (compound 2g ) is an effective LasR inhibitor. 2g demonstrated exceptional biofilm inhibition (IC 50 = 0.14 ± 0.08 μmol/L) without bactericidal activity (MIC > 256 μmol/L). and bound LasR with high affinity ( K D = 38.97 μmol/L). Mechanistically, transcriptomics revealed selective suppression of LasR-regulated virulence pathways, reducing pyocyanin production and bacterial motility. Notably, 2g synergized with antibiotics in vivo , reducing effective doses of ciprofloxacin and tobramycin by 1000-fold and 500-fold, respectively, while potently suppressing resistance evolution. Pharmacokinetic profiling further demonstrated favorable oral bioavailability (26.8%) and a wide safety margin. This work establishes 2g as a machine learning-optimized LasR inhibitor with potent antibiofilm activity and synergistic antibiotic enhancement, offering a translatable strategy against multidrug-resistant P. aeruginosa infections. Machine learning identified compound 2g , which inhibits Pseudomonas aeruginosa LasR. It promotes LasR degradation, suppresses virulence and biofilm, synergizes with antibiotics, and avoids resistance development.