Vaginal Lactobacillus fatty acid response mechanisms reveal a novel strategy for bacterial vaginosis treatment

Abstract

Bacterial vaginosis (BV), a common syndrome characterized by Lactobacillus-deficient vaginal microbiota, is associated with adverse health outcomes. BV often recurs after standard antibiotic therapy in part because antibiotics promote microbiota dominance by Lactobacillus iners instead of Lactobacillus crispatus, which has more beneficial health associations. Strategies to promote L. crispatus and inhibit L. iners are thus needed. We show that oleic acid (OA) and similar long-chain fatty acids simultaneously inhibit L. iners and enhance L. crispatus growth. These phenotypes require OA-inducible genes conserved in L. crispatus and related species, including an oleate hydratase (ohyA) and putative fatty acid efflux pump (farE). FarE mediates OA resistance, while OhyA is robustly active in the human vaginal microbiota and sequesters OA in a derivative form that only ohyA-harboring organisms can exploit. Finally, OA promotes L. crispatus dominance more effectively than antibiotics in an in vitro model of BV, suggesting a novel approach for treatment.

Figure 1.. cis-9-uLCFAs selectively inhibit L. iners and promote growth of L. crispatus and other FGT lactobacilli.

(A) Relative growth of representative L. crispatus, L. gasseri, L. iners, L. jensenii, and L. mulieris strains in modified Lactobacillus MRS broth (MRS+CQ broth) with varying concentrations of oleic acid (OA, left), linoleic acid (LOA, middle), or palmitoleic acid (POA, right). (B) Relative growth of diverse non-iners FGT Lactobacillus (n=30) and L. iners (n=14) strains in MRS+CQ broth supplemented with varying concentrations of OA. (C) Minimum bactericidal concentration (MBC) assay results for representative L. crispatus, L. gasseri, L. iners, L. jensenii, and L. mulieris strains in MRS+CQ broth. Colony forming units (CFU) were measured after 24 hours by standard serial dilution and colony counting and expressed relative to the number of CFU recovered from the no fatty acid supplementation control. (D) Transmission electron microscopy (TEM) images of L. crispatus (top) and L. iners (bottom) treated with 3.2 mM OA (right) or untreated (left) for 1 hour. (E) Relative growth rescue of diverse non-iners FGT Lactobacillus (n=32) and L. iners (n=5) strains in S-broth supplemented with varying concentrations of OA. Relative growth rescue was calculated as growth relative to the median OD600 measurement in the S-broth supplemented with 0.1% Tween-80. (A, B, and E) Growth was measured by optical density at 600 nm (OD600) after 72 hours of culture. (A and B) Relative growth was calculated relative to the median OD600 measurement in the no fatty acid supplementation control. (A and C) Plotted points represent 3 technical replicates per condition and are representative of ≥2 independent experiments per condition. (B and E) Plotted points represent the median relative growth or growth rescue for 3 technical replicates per condition and are representative of ≥2 independent experiments per condition.

Figure 2.. Non-iners FGT lactobacilli possess a conserved set of OA response genes that L. iners lacks.

(A) Transcriptomic responses of cultured L. crispatus (left), L. gasseri (middle), and L. jensenii (right) grown to exponential phase in MRS+CQ broth, then exposed to OA (3.2 mM) for 1 hour. Bulk RNA-sequencing was performed and DESeq2 analysis was used to identify differentially expressed (DE) genes relative to the matched, no OA supplementation control. The Benjamini-Hochberg procedure was used to control the false discovery rate (FDR) with ɑ=0.05. Plots depict the log2(fold change, FC) of mRNA expression with OA relative to control (x-axis) and −log10(adjusted p-value) for each gene (y-axis). Dotted lines represent cutoffs used to define significant differential expression (−1≥FC≥1 and adjusted p-value≤0.05). DE genes observed in all species included a predicted oleate hydratase (ohyA, purple; COG4716), putative fatty acid efflux pump (farE, teal; COG2409), and its putative regulator (tetR, pink; COG1309). (B) Venn diagram showing the number of DE genes per species with OA treatment and overlap of DE gene functions across species. Three gene functions (ohyA, farE, and tetR; all upregulated in response to OA) were shared among all three species. (C) Presence of gene functions predicted to encode functional oleate hydratase (ohyA) and putative fatty acid efflux pump (farE) activity in isolate genomes and metagenome-assembled genomes (MAGs) of the indicated FGT Lactobacillus species (n=1,167 total genomes and MAGs). Presence of gene functions involved in exogenous fatty acid acquisition and utilization (fakAB, plsC, plsX, and plsY) are shown for comparison. (B and C) Gene function was predicted using eggNOG 5.0 employing eggNOG-mapper v2.1.9.

Figure 3.. FGT Lactobacillus OhyA enzymes are functional and physiologically active.

(A) OhyA protein phylogenetic tree for representative orthologs from the indicated species. Tree was constructed from MUSCLE v5.1-aligned protein sequences using RAxML-NG (see methods). Starred leaf tips correspond to FGT Lactobacillus orthologs shown in bold. OhyA orthologs upregulated in response to OA (Figure 2A) are marked with asterisks. (B) OhyA9 enzymatic activity reaction diagram with OA substrate. (C) Extracted ion chromatograms from supernatants of ohyA-gene deleted strain of S. aureus complemented with an empty vector (ΔSaohyA/empty vector), SaohyA-expressing plasmid (ΔSaohyA/pSaohyA), LCRIS_00558-expressing plasmid (ΔSaohyA/pLCRIS_00558), and LCRIS_00661-expressing plasmid (ΔSaohyA/pLCRIS_00661). Strains were cultured with OA for 1 hour. Annotated peaks include OA (18:1) and 10-HSA (h18:0). (D) MS2 spectra with major fragmentation labels for the h18:0 peak shown in the lower right panel of Figure 3C for the supernatant of ΔSaohyA/pLCRIS_00558 cultured with OA. (E) Detection of ¹³C-labeled 10-hydroxystearic acid (¹³C₁₈-10-HSA; h18:0) in supernatants of L. crispatus, L. gasseri, and L. jensenii cultured for 72 hours in NYCIII broth with and without universally ¹³C-labeled OA (¹³C18-OA; 3.2 mM, which is a lethal concentration for L. iners). (F) Detection of ¹³C₁₈-10-HSA in supernatants of L. crispatus and L. iners cultured for 72 hours in NYCIII broth with and without ¹³C₁₈-OA (100 μM, which is a sublethal concentration for L. iners). (E and F) The no-OA control data shown for media and L. crispatus supernatant in E and F are derived from the same samples. Plotted points represent 3 technical replicates per condition.

Figure 4.. Women with non-iners lactobacilli have uniquely elevated vaginal concentrations of OhyA products.

(A) FGT bacterial microbiota composition of 180 distinct FGT swab samples from 106 total South African women, determined by bacterial 16S rRNA gene sequencing (top, stacked barplot). Samples were classified into cervicotype (CT) based on microbiota composition as previously described^,. Middle and bottom bar plots show relative concentrations of OhyA products 10-HSA (h18:0) and hydroxy 18:1 (h18:1) respectively, measured in paired cervicovaginal lavage (CVL) samples by targeted lipidomics. The top colorbar shows Nugent score-based BV status. (B) Relative h18:0 (top) and h18:1 (bottom) concentrations within each CT for the 180 CVL samples shown in (A). Significance was determined by one-way ANOVA with post-hoc Tukey’s test; selected pairwise differences are shown (****p < 0.0001; full statistical results in Table S4). (C) The change in relative h18:0 (top) and h18:1 (bottom) concentrations within 74 paired consecutive samples from participants whose microbiota transitioned to CT1 (n=5), away from CT1 (n=6), remained CT1 (n=11), or remained non-CT1 (n=52). Significance was determined by paired t-test on log-transformed values (**p < 0.01; ***p < 0.001; ****p < 0.0001; ns: p ≥ 0.05).

Figure 5.. farE is required for OA resistance and growth enhancement.

(A) Relative growth of L. gasseri ATCC 33323 wild-type (WT) strain, and derivative genetic mutant strains created by double crossover homologous recombination, including knockout strains of ohyA9 (ΔohyA9) and farE (ΔfarE), and ΔfarE complemented with plasmid-overexpressed autologous farE (ΔfarE/pfarE) in MRS+CQ broth supplemented with varying concentrations of OA. (B) Relative growth rescue of L. gasseri ATCC 33323 WT and derivative genetic mutant strains in delipidated MRS+CQ broth supplemented with varying concentrations of OA. (C) MBC assay results for L. gasseri ATCC 33323 WT and derivative genetic mutant strains in MRS+CQ broth. MBC assay performed following the same methods described in Figure 1C. (D) Detection of ¹³C₁₈-10-HSA in blank media and culture supernatants from L. gasseri WT, ΔohyA9, ΔohyA9 complemented with plasmid-overexpressed autologous ohyA9 (ΔohyA9/pohyA9), ΔfarE, and ΔfarE/pfarE cultured for 24 hours in NYCIII broth with and without ¹³C₁₈-OA (100 μM). Plotted points represent 2 or 3 technical replicates per condition. (A and B) Growth was measured by OD600 after 24 hours of culture. Relative growth or growth rescue was calculated relative to the median OD600 measurement in the non-delipidated media, no OA supplementation control. (A, B, and C) Plotted points represent 3 technical replicates per condition and are representative of ≥2 independent experiments per condition.

Figure 6.. FGT lactobacilli are fatty acid auxotrophs and OhyA9 activity permits non-iners lactobacilli to utilize 10-HSA.

(A) Fatty acid synthesis II (FASII) and phospholipid synthesis pathways annotated with the predicted presence of gene functions in L. iners and in non-iners FGT Lactobacillus genomes. Genes marked as missing for a species if the gene function that was predicted to be absent in >50% of genomes for each species. (B) Relative growth rescue of diverse L. crispatus (n=19), L. gasseri (n=3), L. iners (n=13), L. jensenii (n=8), and L. mulieris (n=5) strains in delipidated MRS+CQ broth supplemented with 3.2 mM acetate or 3.2 mM OA after 72 hours of culture. (C) Detection of phosphatidylglycerol lipids in cell pellets from L. crispatus (top) and L. iners (bottom) cultured for 72 hours in NYCIII broth with no OA supplementation (left), 100 μM OA (middle), or 3.2 mM OA (right). Unlabeled OA (¹²C₁₈-OA) or universally ¹³C-labeled OA (¹³C₁₈-OA) was used for supplementation. Plots represent MS1 spectra of one representative sample per condition with major unlabeled (black) and ¹³C-labeled (red) lipid species annotated. (D) Relative growth rescue of diverse non-iners FGT Lactobacillus (n=17) and L. iners (n=5) strains in delipidated MRS+CQ broth supplemented with varying concentrations of OA (left) or 10-HSA (right) after 72 hours of culture. (E) Relative growth rescue of L. gasseri ATCC 33323 WT and derivative genetic mutant strains in delipidated MRS+CQ broth supplemented with varying concentrations of 10-HSA after 24 hours of culture. (B, D, and E) Relative growth rescue was calculated as growth relative to the median OD600 measurement in non-delipidated MRS+CQ broth with no OA supplementation. (B and D) Plotted points represent the median relative growth for 3 technical replicates per condition. (E) Plotted points represent 3 technical replicates per condition and are representative of ≥2 independent experiments per condition.

Figure 7.. OA treatment shifts in vitro BV-like communities towards L. crispatus-dominance alone or in combination with MTZ.

(A) Relative growth of the indicated bacterial species in NYCIII broth supplemented with or without metronidazole (MTZ; 50 μg/mL) and/or OA (3.2 mM) after 72 hours of culture. Growth was measured by OD600. Relative growth was calculated relative to the median OD600 measurement in the no OA supplementation control. Plotted points represent 3 technical replicates per condition. (B) Relative bacterial abundance in 4 representative, defined BV-like communities grown for 72 hours in NYCIII broth with or without MTZ (50 μg/mL) and/or OA (3.2 mM). Composition of the cultured communities and of the input mixture (T0) was determined by bacterial 16S rRNA gene sequencing. Plots depict 6 technical replicates per condition. (C) Ratios of L. crispatus to the sum of all other taxa in the mock communities shown in (B). The gray dotted line represents the input ratio measured in the input sample (T0). Between-group differences were determined by one-way ANOVA with post-hoc Tukey’s test; selected significant pairwise differences are shown (****p < 0.0001; full statistical results in Table S7).