Screening antimicrobial peptides and probiotics using multiple deep learning and directed evolution strategies

We designed a high throughput AMP screening approach to filter AMPs in random peptide sequences. A plasmid pool containing 150 bp of random DNA sequences (150N) located at the downstream of start codon ATG and RBS was constructed. To minimalize the impact of gene expression leakage on host growth, we rigorously regulated random peptide gene expression using arabinose operons. After transferring the plasmid pool into Escherichia coli DH5-alpha strain and culturing it to the logarithmic phase with or without arabinose. we extracted the plasmid pool and amplificated the 150N using universal primer pairs at its both ends.In order to produce complex L. plantarum mutant strain pools, Atmospheric and Room Temperature Plasma (ARTP), a safe and powerful technology, was applied, followed that FADS was utilized to screen L. plantarum mutants with high AMPs production using split GFP sensor. 50 colonies were picked, and then the byproducts of protein fermentation were extracted and hydrolyzed to check for antibacterial effect. Many colonies are more effective against E. coli or B. subtilis than wildtype (WT). Only 3 mutations were able to intensify the growth inhibition on both strains.To dissect the gene mutations contributing to high bacteriostatic activity, we carried out whole genome sequencing (WGS) and drew a complete genome map of the three screened L. plantarum mutants. The DNA libraries which were sequenced on the illumina NovaSeq 6000 platform.