In silico design and evaluation of hybrid antimicrobial peptides for combating environmental multidrug-resistant bacteria.

To address the escalating challenge of multidrug resistance (MDR) bacteria and overcome the inherent limitations of conventional antimicrobial peptides (AMPs)-such as structural instability and constrained discovery efficiency-we devised a rational design strategy integrating in silico methods with modular hybridization. Natural short peptides (4-6 amino acids) derived from the biocontrol strain Bacillus velezensis 524B were classified into cationic or hydrophobic modules and systematically assembled using a novel de novo CHCH (cationic-hydrophobic-cationic-hydrophobic) modular hybridization scaffold. Candidate peptides were computationally screened via AMP_scanner, leading to the identification of the optimized hybrid peptide CHCH_KRVL_3. This peptide exhibited potent antibacterial activity against MDR Acinetobacter baumannii, with a MIC of 8 μM, and demonstrated a high selectivity index (SI > 10). Mechanistic investigations defined the bactericidal cascade: CHCH_KRVL_3 specifically binds to phosphatidylethanolamine (PE) and phosphatidylglycerol (PG) as the initial trigger, inducing instantaneous membrane permeabilization followed by a sustained accumulation of reactive oxygen species (ROS). CHCH_KRVL_3 displayed superior antibiofilm performance, effectively eradicating preformed biofilms and inhibiting biofilm formation on clinically relevant surfaces such as PVC endotracheal tubes, surpassing the efficacy of polymyxin B. Beyond identifying a lead compound, this study establishes a generalizable "natural module + chch scaffold" platform that systematically leverages AI-assisted screening to accelerate future AMP discovery.