Integrative Peptidomics and Deep Learning Identifies Endogenous Antimicrobial Peptides in Periprosthetic Joint Infection

ABSTRACT: Background: Periprosthetic joint infection (PJI) remains a major clinical challenge due to biofilm formation and antimicrobial resistance. This study used an integrated clinical peptidomics and deep learning approach to explore the biochemical features of the infected joint microenvironment, aiming modelsto identify endogenous antimicrobial peptides (AMPs) with translational potential. Results: Untargeted metabolomics revealed pronounced metabolic reprogramming in PJI patients, with enrichment of tryptophan, bile acid, and purine pathways, reflecting an intensified host immune-inflammatory response and heightened oxidative stress. Peptidomic analysis identified 435 DEPs, from which seven candidates were prioritized and synthesized via deep learning prediction. In vitro assays confirmed that c_AMPS15, c_AMPS16, and c_AMPS17 exhibited potent bactericidal activity against Staphylococcus aureus, while c_AMPS15 also showed efficacy against Pseudomonas aeruginosa. These peptides demonstrated superior biosafety profiles with minimal hemolysis and effectively inhibited bacterial biofilm maturation on prosthetic surfaces. In vivo, these peptides significantly reduced S. aureus burdens, while c_AMPS15 retained antimicrobial activity against P. aeruginosa and attenuated local inflammatory response. Conclusion: This integrative peptidomics and deep learning framework identified host-derived AMPs as promising therapeutic candidates for PJI, highlighting c_AMPS15 as a lead molecule with broad-spectrum and antibiofilm activity.

INSTRUMENT(S): Liquid Chromatography MS - negative - reverse-phase, Liquid Chromatography MS - positive - reverse-phase

PROVIDER: MTBLS14275 | MetaboLights | 2026-10-27

REPOSITORIES: MetaboLights