Structure‐Guided De Novo Design of Carbapenem Analogs Targeting Penicillin‐Binding Protein 1B Using Pharmacophore‐Based Strategies

Abstract Carbapenems, once considered “last‐resort” antibiotics, are now facing resistance due to misuse, highlighting the urgent need for new antimicrobial drug development. This study utilized computational methods to discover and optimize compounds targeting antibiotic‐resistant bacteria. A shared featured pharmacophore (SFP) map for three compounds (PubChem CIDs: 441130, 104838, and 150610) was developed, incorporating hydrophobic regions, hydrogen bond acceptors, donors, and aromatic interactions. The model achieved a goodness‐of‐hit score of 0.525, validating its potential to identify promising compounds. Virtual screening with LigandScout software identified eight hits with pharmacophoric fit scores between 95.72 and 115.9. These ligands were fused with carbapenem core rings via genetic algorithms and fragment‐based design, yielding 50 synthetic carbapenem models with over 70% structural similarity and Synthetic Accessibility Scores (SAS) ≤ 4.4. In the docking analysis, Molecules 25, 42, and 48 showed stronger binding to Penicillin‐binding protein 1B (PDB ID: 5FGZ) with scores of −8.7, −8.6, and −8.5 kcal/mol, which were stronger than Meropenem's (control) score of −7.5 kcal/mol. Molecular dynamics simulations (200 ns) confirmed the structural stability of Molecule 42. Retrosynthesis and scaffold analysis confirmed that the newly designed compounds are chemically feasible and could be synthesized in laboratory settings. Overall, this study offers a promising computational strategy for developing next‐generation antibiotics to combat resistant bacterial infections.