INTRODUCTION:
Carbapenem‑resistant Pseudomonas aeruginosa is a World Health Organization critical‑priority pathogen associated with high morbidity, mortality, and diminishing therapeutic options. Carbapenemases encoded by blaIMP, blaVIM, blaKPC, and blaGES hydrolyse carbapenems and other beta‑lactams, driving multidrug‑resistant and extensively drug‑resistant phenotypes.

METHODS:
Plant‑origin antimicrobial peptides were retrieved from a curated database and sequentially filtered using AI‑based predictors of haemolysis, allergenicity, toxicity, and physicochemical suitability. A lead Arabidopsis thaliana peptide underwent rational sequence optimisation via targeted amino acid substitutions, followed by three‑dimensional structure prediction with AlphaFold, molecular docking against blaIMP, blaVIM, blaKPC, and blaGES, molecular dynamics simulations to assess complex stability, and comprehensive in silico ADMET profiling. Findings / Expected outcomes From 579 initial peptides, seven non‑toxic, non‑allergenic candidates were retained, and one optimised peptide satisfied stringent criteria (molecular weight 2792.15 Da, instability index 24.68, net charge +5.5, favourable half‑life). Docking and dynamics analyses indicate high‑affinity, structurally stable interactions with all four carbapenemases, with preferential binding to metallo‑beta‑lactamases, and an ADMET profile suggesting high oral bioavailability, low systemic toxicity, and primary limitations related to permeability and efflux.

CONCLUSIONS:
This AI‑integrated workflow, anchored by AlphaFold‑based tertiary structure prediction, prioritised a plant‑derived antimicrobial peptide with promising target engagement, structural stability, and pharmacokinetic‑toxicological attributes. These findings support experimental validation and further optimisation of this peptide as an adjuvant to restore or potentiate carbapenem efficacy against carbapenem‑resistant P. aeruginosa.