Identification of Novel Antimicrobial Compounds in Colostrum-Associated Lactiplantibacillus plantarum ZFS by Integrating In vitro, Machine Learning and Bioinformatics Approaches

Abstract

Current study was designed to identify safe, novel antimicrobial compounds in the cell free supernatant of two colostrum-associated probiotic strains, “Lactiplantibacillus plantarum, ZFS-1 and ZFS-2” by using in vitro, advanced machine learning, and bioinformatics approaches. The cell free supernatant (CFS) and ethyl acetate fraction of CFS of both isolates, were evaluated for antagonistic properties against four pathogens. and anti-inflammatory properties of ethyl acetate fraction were also assessed in vitro. The secondary metabolites in the ethyl acetate fraction of CFS were identified by Gas chromatography-Mass spectrometry. Further, Machine learning techniques, including models such as Support Vector Machine, k-Nearest Neighbor, Naive Bayes, and Random Forest, were applied to analyze the data using Python. Molecular descriptors were calculated using the RDKit library and further analyzed through Principal Component Analysis. Among the models, the RF achieved 100% accuracy on the test set, with a 10-fold cross-validation accuracy of 93.84%. Additionally, docking analysis was conducted to evaluate the binding affinities between the identified metabolites and microbial proteins. The CFS of both isolates displayed inhibition zones of 12-18 mm, while ethyl acetate fraction exhibited ZI of 16-30 mm against four pathogens. Minimal Inhibitory and bactericidal concentrations ranged from 3.12-6.5 and 6.5-12.5 mg/ml, respectively. Significant differences (p <0.05) were noted in their antimicrobial and antioxidant efficiencies. Acridine, 1H-Phenalen-1-one, and (NE)-N-[(4-amino-1,2,5-oxadiazol-3-yl)-(triazirin-1-yl)methylidene]hydroxylamine in CFS of Lactiplantibacillus plantarum strain ZFS-1 and 5,6,7,8-tetrahydro-2-methoxy-5,5-dimethyl-1,4-Anthracenedione in CFS of strain ZFS-2 were identified as novel antimicrobial compounds. The compounds are validated as potentially safe drug candidates employing comprehensive in vitro and molecular docking analyses.