Metagenomics reveals potential antimicrobial peptides in Chinese baijiu fermentation.

Antimicrobial peptides (AMPs) from food fermentation microbiota hold promise for food preservation and as potential antimicrobial agents. However, the biosynthetic potential of AMPs in food fermentations remains largely unclear. Here, using Chinese baijiu fermentation as a model, we provided a workflow for AMP mining by combining metagenomics and machine learning. We recovered 389 metagenome-assembled genomes (MAGs) spanning both bacteria and archaea from 18 fermented samples. In total, 414 AMPs, including 290 novel AMPs, were predicted in 59.38 % of these MAGs using a machine learning model. Correlation network analysis showed that AMP-producing microorganisms potentially mediated negative microbial interactions. We selected ten AMPs for experimental validation, and eight AMPs exhibited antimicrobial activity against five human pathogens and two food spoilage microorganisms. One peptide, AMP_22, showed a broad-spectrum activity (all seven test strains) with high potency (MIC = 3.06-200 μg/mL) and cytotoxicity was not observed below 25 μg/mL using HepG2 and A549 cell lines. We further investigated the antimicrobial mechanism of AMP_22 using Escherichia coli as a model. Treatment with AMP_22 caused severe damage to the bacterial cell membrane, inhibited intracellular protein synthesis, and led to a significant accumulation of reactive oxygen species (ROS). Furthermore, molecular docking analysis indicated that AMP_22 can bind to DNA gyrase and dihydrofolate reductase via hydrogen bonding. This study highlights the potential of food-derived AMPs for application as preservatives and antimicrobial agents.