Vaginal dysbiosis need not rely on antibiotics alone: two Lactiplantibacillus plantarum strains cut Gardnerella in mice to near-healthy levels
Synopsis
From 88 lactobacilli isolates from human milk, infant stool and fermented foods, two Lactiplantibacillus plantarum strains, CYLB2146 and CYLB2189, inhibited Gardnerella vaginalis, Streptococcus agalactiae, Neisseria gonorrhoeae and Candida spp. and lowered inflammatory cytokines in vaginal epithelial cells. In a murine bacterial vaginosis model, live bacteria and their cell-free supernatants reduced G. vaginalis loads from 51.2- and 41.6-fold in controls to 3.2-, 11.1-, 3.2- and 0.7-fold after 5 days, approaching healthy levels by day 10.
Interpretation
Treatment of vaginal dysbiosis gains an option beyond antibiotics: lactobacillus strains that produce inhibitory substances can suppress pathogens and inflammation together. Current therapy relies on metronidazole, clindamycin and azoles, which often fail to restore a stable lactobacilli-dominated microbiota and carry high recurrence and resistance. Of 88 lactobacilli isolates, 53 showed bacteriocin-like inhibitory substance activity, and 7 inhibited G. vaginalis, N. gonorrhoeae, S. agalactiae and Candida spp. broadly.
Two L. plantarum strains brought G. vaginalis loads in mice close to those of healthy animals. Untreated and medium controls still carried 50.6-fold and 34.0-fold expansions at day 10 after treatment. After 5 days, live CYLB2146 and CYLB2189 groups fell to 3.2- and 11.1-fold and supernatant groups to 3.2- and 0.7-fold; by day 10 live groups reached 0.6- and 1.7-fold and supernatant groups 2.8- and 1.7-fold, against 0.1-fold in healthy mice.
The antimicrobial activity comes from several factors acting together rather than one compound. Neutralizing supernatants lowered activity, while uninoculated acidic medium alone strongly inhibited bacteria, marking acidification as a major contributor. Proteinase K, trypsin, heat and catalase treatments reduced activity to different degrees across strain-pathogen pairs, implicating protease-sensitive peptides, heat-sensitive components and hydrogen peroxide.
Machine learning prioritized low-molecular-weight peptides detected by mass spectrometry for experimental testing. Such functional peptides are structurally diverse and low in abundance, so mass spectra alone cannot say which is worth synthesizing and testing. Cross-validated ROC-AUC exceeded 0.92 across the five pathogens; the prioritized cyclo(Tyr-Pro) had a minimum inhibitory concentration of 31.25 micromolar against S. agalactiae and reduced C. tropicalis biofilm at sub-inhibitory concentrations.
Perspective
The results cover in vitro pathogen inhibition, HeLa cell adhesion and cytokine assays, whole-genome and metabolite mass spectrometry analysis, and a bacterial vaginosis model in 28 female BALB/c mice. The intended readers are researchers in probiotics, antimicrobial peptides and vaginal microbial ecology. The paper proposes activity-guided purification, targeted mass spectrometry confirmation, combination testing of candidate compounds and microbiome profiling, and recommends a non-infected probiotic-only control group in future animal work.
The principal antimicrobial compounds were not directly isolated and structurally confirmed, and the contributions of the identified plantaricin gene clusters and low-molecular-weight peptides remain unresolved. Culture-medium components, hydrogen peroxide and strain-derived metabolites could not be fully separated. qPCR measures L. plantarum at species level and cannot distinguish administered strains from resident populations, so overall vaginal microbiota restoration is not shown. Tissue recovery may be secondary to reduced pathogen burden, since no healthy probiotic-only control group was included.