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.

Introduction

Female genital tract (FGT) microbiota composition is linked to numerous adverse health outcomes, such as preterm birth¹, infertility^2–5, cervical dysplasia^6–8, and sexually transmitted infections⁹ including human immunodeficiency virus (HIV) risk^10,11. Bacterial vaginosis (BV) – a syndrome of the FGT microbiota associated with these adverse outcomes – affects up to 58% of women worldwide¹² and has clinical features including watery discharge, odor, and mucosal inflammation^13–15. Microbiologically, BV is characterized by a paucity of lactobacilli and high abundance of diverse obligate anaerobic species^16,17. By contrast, health-associated FGT microbial communities are typically dominated by Lactobacillus species, most notably Lactobacillus crispatus, but also including Lactobacillus gasseri, Lactobacillus jensenii, and Lactobacillus mulieris. However, FGT microbial communities dominated by a different common FGT Lactobacillus species – Lactobacillus iners – have several sub-optimal health associations^6,10,18, including higher risk of transitioning to BV^19–21.

Standard first-line BV therapy with the antibiotic metronidazole (MTZ) has partial efficacy, but ≥50% of treated patients experience recurrence within one year^22–24. One explanation for high recurrence rates is the high probability of MTZ treatment to shift FGT microbiota composition towards dominance by L. iners^21,25–28 instead of more health-associated lactobacilli like L. crispatus. Resistance to MTZ among BV-associated bacteria may also impair treatment efficacy and promote recurrence, although the clinical significance of MTZ resistance in BV remains unclear^29–31. Experimental interventions to improve MTZ efficacy using adjunctive non-antibiotic strategies such as vaginal microbiota transplants and L. crispatus-containing intravaginal live biotherapeutic products have shown some promise compared to MTZ alone, but reported recurrence rates from these studies remain high^24,32,33. To our knowledge, no therapies currently exist to promote L. crispatus dominance in the FGT microbiota by selectively inhibiting L. iners or enhancing L. crispatus growth.

Mammalian mucosal surfaces are rich with LCFAs^34–37, which serve as critical nutrients and building blocks for bacterial membrane components and other biological processes, but can also display antimicrobial properties³⁸. Certain unsaturated long-chain fatty acids (uLCFAs) exert antimicrobial activity against host-adapted Gram positive organisms such as Staphylococcus aureus^39,40. However, uLCFA effects on growth or inhibition of FGT Lactobacillus species have not been systematically assessed. L. iners has a reduced genome size and metabolic capacity relative to other FGT lactobacilli^41,42, suggesting feasibility of exploiting metabolic differences between Lactobacillus species to selectively target L. iners⁴³. We hypothesized that long-chain fatty acid (LCFA) metabolism might constitute a metabolic target to differentially modulate FGT lactobacilli and other bacteria.

In this study, we investigated effects of cis-9-uLCFAs on FGT Lactobacillus species and assessed their potential to selectively modulate FGT microbiota composition. We found that cis-9-uLCFAs selectively inhibited L. iners while simultaneously robustly promoting growth of non-iners FGT lactobacilli. Guided by transcriptomic and genomic analyses, we identified a putative fatty acid efflux pump (farE) and an oleate hydratase enzyme (ohyA) that were upregulated by cis-9-uLCFAs in non-iners FGT Lactobacillus species. These genes were genomically conserved in non-iners lactobacilli but completely absent in L. iners, mirroring the observed growth phenotypes. Using a genetically tractable strain of L. gasseri, we showed that farE is required for both the cis-9-uLCFA resistance and growth enhancement phenotypes. Biochemical characterization of the Lactobacillus OhyA orthologs revealed them to be robustly active in vivo in the human FGT microbiota and demonstrated that non-iners FGT lactobacilli can utilize OhyA to bioconvert and sequester exogenous OA for use in phospholipid synthesis. Taken together, our data support that ohyA and farE confer non-iners lactobacilli a growth advantage in cis-9-uLCFA-rich environments. We leverage this discovery to selectively modulate in vitro BV-like bacterial communities towards L. crispatus dominance using cis-9-uLCFA treatment alone or in combination with MTZ. Collectively, this work identifies important species-level differences in FGT Lactobacillus metabolism, elucidates the mechanisms underlying those differences, and provides evidence for a novel metabolism-targeted intervention to improve female genital and reproductive health.

Results

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

To investigate the effects of cis-9-uLCFAs on FGT Lactobacillus species, we cultured representative strains with varying concentrations of several cis-9-uLCFAs in modified Lactobacillus MRS broth (MRS+CQ broth)⁴³. Strikingly, oleic acid (OA; 18:1 cis-9), linoleic acid (LOA; 18:2 cis-9,12), and palmitoleic acid (POA; 16:1 cis-9) each selectively and potently inhibited L. iners while exerting little or no inhibitory effect on L. crispatus, L. gasseri, L. jensenii, and L. mulieris (Figure 1A; corresponding cis-9-uLCFA chemical structures are shown in Figure S1A). In particular, OA had no inhibitory effect towards any of the non-iners species at concentrations up to 3.2 mM, despite robustly inhibiting L. iners at a half-maximal inhibitory concentration (IC50) of less than 400 μM. To assess whether these phenotypes showed species-level conservation, we tested a diverse collection of strains from each species, including isolates from geographically diverse donors with varying BV status⁴³ (Table S1). All L. iners strains (n=14) were sensitive to OA with a median IC50 of 100 μM in MRS+CQ broth, while none of the non-iners FGT Lactobacillus strains (n=30) were inhibited (Figure 1B). Similar patterns were observed in NYCIII broth, a rich, non-selective media formulation commonly used to culture FGT bacteria⁴⁴ (Figure S1B). These results demonstrate that OA is toxic to L. iners but exerts no inhibitory effect on non-iners FGT Lactobacillus species.

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.

To evaluate whether OA inhibited L. iners via a bactericidal mechanism, we performed a minimum bactericidal concentration (MBC) assay⁴⁵ using representative Lactobacillus strains in MRS+CQ broth. The minimum OA concentration required to kill ≥99.9% of the L. iners inoculum within 24 hours (the MBC) was 400 μM (Figure 1C), which was equal to the corresponding MIC (minimum concentration achieving ≥99.9% growth inhibition, Figure 1A), indicating a bactericidal effect. In contrast, OA had no bactericidal activity against non-iners Lactobacillus species. Transmission electron microscopy (TEM) revealed that exposure to 3.2 mM of OA induced catastrophic cell wall and membrane disruption in L. iners within 1 hour of treatment, whereas L. crispatus cell integrity and morphology remained unaffected by OA (Figure 1D; full images, Supp. Data 1). These results indicate that cis-9-uLCFAs selectively kills L. iners via disruption of their cell wall and membrane.

We next investigated OA growth effects in conditions less optimized for Lactobacillus growth. S-broth is a rich media formulation in which FGT Lactobacillus species grow poorly if not supplemented with Tween-80⁴³. Consistent with observations from other media types, supplementing S-broth with OA inhibited L. iners (n=5 strains; Figure 1E). However, OA supplementation in S-broth robustly promoted growth of non-iners FGT Lactobacillus species (n=32 strains). Thus, OA not only exerts potent, selective toxicity against L. iners, but simultaneously promotes growth of non-iners FGT lactobacilli.

Non-iners FGT lactobacilli possess a conserved set of cis-9-uLCFA-response genes that L. iners lacks

We hypothesized that the similar growth responses of non-iners FGT Lactobacillus species to uLCFAs reflected a shared set of uLCFA-response genes. We therefore assessed transcriptomic responses of representative L. crispatus, L. gasseri, and L. jensenii strains to OA. Bacteria were grown to exponential phase, then treated with OA for 1 hour. We used bulk RNA-sequencing and DESeq2⁴⁶ analysis to identify genes that were differentially expressed (DE) by each species compared to matched, untreated control cultures (Figure 2A). This analysis revealed just three unique gene functions were consistently DE (all upregulated) in all species in response to OA (Figures 2A–B). These included a predicted oleate hydratase enzyme (ohyA), a putative fatty acid efflux pump (farE), and its putative regulator (tetR). To further validate these findings, we investigated transcriptomic responses of the same strains to the cis-9-uLCFAs POA and LOA (Figure S2A). All three OA-induced gene functions (ohyA, farE, and tetR) were also upregulated by POA and by LOA treatment in all three species (Figures S2A–C; full dataset, Supp. Data 2), confirming the hypothesis that non-iners FGT lactobacilli shared a core set of cis-9-uLCFA-inducible response genes.

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 DESeq2⁴⁶ 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)⁴³. 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 5.0⁸⁵ employing eggNOG-mapper v2.1.9⁸⁶.

Given its strikingly different phenotypic response to OA exposure, we hypothesized that L. iners would lack the OA-induced gene functions found in non-iners lactobacilli. We therefore investigated gene function presence in a previously reported FGT Lactobacillus genome catalog comprising 1,167 isolate genomes and metagenome-assembled genomes (MAGs) from geographically and clinically diverse sources, supplemented with six additional completed L. iners isolate genomes (Table S2)^43,47. We excluded tetR, the putative farE regulator, from this analysis due to its nonspecific functional annotation. As hypothesized, ohyA and farE gene functions were completely absent from L. iners genomes but highly conserved among non-iners FGT Lactobacillus genomes (Figure 2C). In contrast, gene functions involved in acquiring (fakA and fakB) and utilizing (plsC, plsX, and plsY) exogenous fatty acids for phospholipid synthesis were conserved among all FGT Lactobacillus species (Figure 2C)⁴⁸, showing that L. iners retains other key fatty acid-related pathways. Collectively, these transcriptomic and genomic analyses suggest roles for ohyA and farE in mediating species-specific uLCFA growth phenotypes in FGT lactobacilli.

Comparative genomics reveals conserved phylogeny of FGT Lactobacillus ohyA and farE

To our knowledge, ohyA and farE have not previously been studied in FGT Lactobacillus species. We therefore investigated their intraspecies diversity and their homology to better-characterized orthologs from other human-adapted bacteria. Most L. crispatus and L. gasseri genomes contained two distinct predicted ohyA orthologs (only one of which was OA-induced), whereas L. jensenii and L. mulieris genomes each contained a single ortholog (Figure S3A). In contrast, each species contained a single farE ortholog (Figure S3B). To compare FGT Lactobacillus ohyA orthologs to other species, we performed a phylogenetic reconstruction comprising 21 distinct OhyA protein sequences from 16 different species, including all 7 distinct FGT Lactobacillus OhyA orthologs (Figure 3A; percent identity matrix shown in Figure S3C). Interestingly, patterns of Lactobacillus OhyA phylogeny differed from underlying species phylogeny. The OA-induced L. crispatus and L. gasseri orthologs (LCRIS_00661 and LGAS_1351, respectively) clustered with previously characterized orthologs from S. aureus^49,50, Bifidobacterium breve⁵¹, S. pyogenes⁵², and Lactobacillus acidophilus^53,54, while the non-OA-induced L. crispatus and L. gasseri orthologs (LCRIS_00558 and LGAS_0484, respectively) clustered separately. The L. jensenii and L. mulieris OhyA orthologs constituted a more distantly related phylogenetic clade, clustering with orthologs from Streptococcus salivarius and Enterococcus faecium. In contrast, phylogenetic reconstruction of FarE orthologs from FGT and non-FGT Lactobacillus species closely matched underlying species phylogeny (Figures S4A and S4B). FGT Lactobacillus FarE orthologs shared more distant homology (18–20% protein sequence identity) to the previously characterized S. aureus FarE, which also neighbors a TetR family regulator (Figures S4C and S4D). Together, these phylogenetic analyses reveal that non-iners FGT Lactobacillus species harbor phylogenetically diverse ohyA orthologs, suggesting prior horizontal gene transfer and/or species-specific gene loss, whereas each species harbors a single farE ortholog that was likely acquired vertically.

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 v5.1⁸⁷-aligned protein sequences using RAxML-NG⁸⁸ (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. aureus⁴⁹ 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.

We next investigated the presence and co-occurrence of farE and ohyA within the Lactobacillaceae family, including species with diverse lifestyles (e.g., vertebrate-associated, insect-associated, free living, or nomadic)⁵⁵. Many species encoded both farE and ohyA, although the genes did not co-occur in all cases (Figure S5). Notably, L. iners was unique among species in the genus Lactobacillus to lack farE and was the only vertebrate-associated Lactobacillus species to lack ohyA. The widespread distribution of farE and ohyA within the Lactobacillaceae family suggests an important conserved role in cis-9-uLCFA responses.

FGT Lactobacillus OhyA enzymes are functional and physiologically active

Enzymatic activities of the predicted OhyA orthologs from non-iners FGT lactobacilli have not previously been determined, so we assessed their function by heterologous expression in a well-characterized ohyA-knockout strain of S. aureus (ΔSaohyA)⁴⁹. We complemented the ΔSaohyA strain with each of the two most common L. crispatus ohyA orthologs (LCRIS_00661 and LCRIS_00558), along with the S. aureus ohyA^49,50,56 (SaohyA) as a positive control, and empty vector as a negative control. Supernatants from the complemented ΔSaohyA strains cultured with OA or LOA were harvested for identification of OhyA-produced metabolites by targeted lipidomics. The strain expressing the OA-inducible OhyA ortholog from L. crispatus (LCRIS_00661) produced the same hydroxy fatty acid (hFA) metabolites as the SaohyA-expressing strain when cultured with OA or with LOA, confirming they shared the biochemical activity of hydrating the cis-9 double bond to produce 10-hydroxystearic acid (10-HSA or h18:0) from OA and 10-hydroxy-12-octadecenoic acid (h18:1) from LOA (OA reaction, Figures 3B–D; 10-HSA standard MS2 spectra, Figure S6A; LOA reaction, Figures S6B–D). Thus, we named this L. crispatus ortholog OhyA9. In contrast, when complemented with the L. crispatus OhyA ortholog that was not induced by OA (LCRIS_00558), the ΔSaohyA strain did not produce hFAs when cultured in presence of OA, showing it lacked ability to hydrate OA’s cis-9 double bond (Figure 3C). However, it did hydrate the cis-12 double bond of LOA to produce 13-hydroxy-9-octadecenoic acid, so we called this ortholog OhyA12 (Figures S6B and S6E–F). These results show that OhyA orthologs in L. crispatus have distinct enzymatic activities, with only the OA-induced ortholog (OhyA9) exhibiting the ability to hydrate cis-9 double bonds of OA and related cis-9-uLCFAs.

We next sought to assess physiologic OhyA9 enzymatic activity in L. crispatus and other non-iners FGT Lactobacillus species and to confirm the genomic prediction that L. iners lacks OhyA activity. We cultured representative strains of L. crispatus, L. iners, L. gasseri, and L. jensenii in NYCIII broth supplemented with universally ¹³C-labeled OA (¹³C₁₈-OA). Concentrations of the labeled OhyA9 product 10-HSA (¹³C₁₈-10-HSA) were measured in culture supernatants and cell pellets. We observed robust production of ¹³C₁₈-10-HSA by all three non-iners species in both media and cell pellets, confirming that each species has a physiologically active OhyA9 (supernatants, Figure 3E; pellets, Figure S6G). By contrast, L. iners did not produce ¹³C18–10-HSA when cultured in a sub-lethal concentration of ¹³C₁₈-OA (supernatants, Figure 3F; pellets, Figure S6H). Collectively these results confirmed transcriptomic and genomic predictions that non-iners FGT lactobacilli encode and express physiologically active OhyA9 enzymes, while L. iners does not.

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

We hypothesized that the OhyA activity observed in vitro would also be active in the vaginal microbiota of women with bacterial communities dominated by non-iners FGT lactobacilli. We tested this hypothesis by measuring concentrations of OhyA enzymatic products in cervicovaginal lavage (CVL) fluid samples from participants in the FRESH (Females Rising through Education, Support and Health) study, which enrolls South African women aged 18–23 years who provide cervicovaginal samples at three-month intervals⁵⁷. We examined 180 distinct samples from 106 unique FRESH study participants (Table S3). We profiled microbiota composition via bacterial 16S rRNA gene sequencing of matched vaginal swabs and classified bacterial communities into distinct cervicotypes (CTs) using previously validated criteria^10,19,43 (full microbiome dataset, Supp. Data 4). Briefly, samples were classified as cervicotype 1 (CT1, n=34 samples), defined based on dominance by non-iners Lactobacillus species consisting primarily of L. crispatus; CT2 (n=57), defined by dominance of L. iners; CT3 (n=60), defined by Gardnerella predominance; or CT4 (n=29), which consists of non-Lactobacillus-dominant diverse anaerobes, typically with high Prevotella abundance¹⁰ (Figure 4A). We measured concentrations of hydroxy fatty acids (hFAs) including h18:0 and hydroxy 18:1 (h18:1) in paired cervicovaginal lavage (CVL) supernatant samples using a validated targeted lipidomics approach involving picolylamine-based derivatization^56,58 (representative human sample chromatograms shown in Figures S7A–B). h18:0 and h18:1 concentrations were significantly and uniquely elevated in CT1 compared to all other CTs, with h18:0 exhibiting 4.8-fold higher median concentration in CT1 samples compared to other samples and h18:1 exhibiting a 3.6-fold higher median concentration (Figure 4B). Remarkably, h18:0 and h18:1 concentrations segregated CT1 samples from other samples with near-perfect sensitivity and specificity, as shown by respective area under the curve (AUC) values of 0.99 and 1.00 for h18:0 and h18:1 in receiver operating characteristic analysis. These results confirm robust in vivo physiologic activity of non-iners Lactobacillus OhyA enzymes in the human FGT microbiota.

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^10,43. 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 status⁸⁹.

(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).

To further establish the relationship between microbiota composition and OhyA activity, we investigated how individual women’s cervicovaginal h18:0 and h18:1 levels changed over time in serially collected samples. Since some participants provided samples at multiple study visits, the dataset included 74 pairs of consecutively collected samples, including 5 sample pairs in which a participant transitioned from a non-CT1 community to a CT1 community; 6 pairs in which a participant transitioned from a CT1 to a non-CT1 community; 11 pairs in which a participant remained CT1; and 52 pairs in which a participant remained non-CT1. Relative concentrations of h18:0 and h18:1 significantly increased in CVL samples from participants who transitioned from non-CT1 to CT1 (p=0.00006 with median fold-increase 7.11 for h18:0; p=0.00003 with median fold-increase 4.50 for h18:1), and significantly decreased in samples from participants who transitioned away from CT1 (p=0.000256 with median fold-decrease 4.12 for h18:0; p=0.00118 with median fold-decrease 3.88 for h18:1; Figure 4C). By contrast, h18:0 and h18:1 levels did not significantly change in participants who remained in CT1 or remained in non-CT1 states (median fold changes for these transitions ranged between 0.96 and 1.02 with p>0.05 for all types of sample pairs and metabolites). Thus, changes in levels of OhyA enzymatic products within cervicovaginal fluid of individual women closely tracked with bacterial community changes over time, further confirming the close relationship between OhyA activity and microbiota composition.

farE is required for OA resistance and growth enhancement

We used a genetic approach to interrogate the relationship between OA growth phenotypes and the conserved, OA-induced genes, ohyA9 and farE. Tools to genetically modify L. crispatus and L. iners were not available, so we conducted these studies in L. gasseri because of the close phenotypic, genomic, transcriptional, and phylogenetic resemblance of its cis-9-uLCFA responses to those of L. crispatus. Using adaptations of previously reported methods, we made genetic knockouts of ohyA9 (ΔohyA9) and farE (ΔfarE) in the ATCC 33323 strain of L. gasseri via in-frame gene deletions generated by homologous recombination (Figure S8A; WGS verification, Supp. Data 3; see methods). We found that farE, but not ohyA9, was required for the OA resistance and growth enhancement phenotypes (Figures 5A–B). ΔfarE was fully inhibited by OA in MRS+CQ broth (Figures 5A) and failed to exhibit growth with OA supplementation in delipidated MRS+CQ broth (Figure 5B). Complementing ΔfarE with plasmid-overexpressed farE (ΔfarE/pfarE) fully rescued both the inhibition and growth phenotypes (Figures 5A–B). However, OA-mediated growth inhibition in ΔfarE was not as potently bactericidal as in L. iners, indicating that additional factors may also contribute to L. iners susceptibility (Figure 5C). Together, these results show FarE is required for OA resistance and growth enhancement in non-iners FGT lactobacilli.

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.

To genetically confirm the role of ohyA9 in 10-HSA production, we grew L. gasseri wild-type (WT), ΔfarE, ΔfarE/pfarE, ΔohyA9, and ΔohyA9 complemented with plasmid-overexpressed ohyA9 (ΔohyA9/pohyA9) in media supplemented with a sublethal concentration of ¹³C₁₈-OA and measured ¹³C₁₈-10-HSA production in media supernatants and cell pellets. As hypothesized, knocking out ohyA9 fully ablated ¹³C₁₈-10-HSA production, which was rescued by genetic complementation (Figures 5D and S8B). Interestingly, ¹³C₁₈-10-HSA production and secretion into supernatant was increased in ΔfarE but decreased in ΔfarE/pfarE relative to wild-type, indicating that presence of FarE modifies 10-HSA production and export, but is not required. Based on these results, we propose that OA-induced genes in non-iners FGT Lactobacillus species are required for two major mechanisms of response to exogenous cis-9-uLCFAs: FarE-mediated transport to prevent toxic intracellular accumulation and OhyA9-mediated bioconversion to produce their hFA counterparts (model schematic, Figure S8C).

FGT Lactobacillus species are fatty acid auxotrophs

The observation that OA supplementation enhanced non-iners Lactobacillus growth in S-broth, a media not optimized for lactobacilli, prompted us to hypothesize that FGT Lactobacillus species are fatty acid auxotrophs. To confirm this hypothesis, we investigated FGT Lactobacillus genomes for presence of fatty acid synthesis II (FASII) pathway gene functions⁵⁹. FGT Lactobacillus species were largely predicted to lack an intact FASII pathway, including genes predicted to perform the functions of AccABCD, FabH, FabB/F, and FabA/Z (Figures 6A and S9A). Additionally, L. iners lacked genes predicted to perform the functions of FabG and FabD. Among all FGT Lactobacillus genomes examined, only a minority of L. crispatus genomes were predicted to contain an intact FASII pathway. We therefore investigated growth of diverse FGT Lactobacillus species and strains in delipidated MRS+CQ broth supplemented with OA or with acetate, a precursor to the FASII initiating molecule, acetyl-Coenzyme A. All strains, including predicted FASII-containing L. crispatus strains, failed to grow in delipidated media and in acetate-supplemented delipidated media, suggesting all were fatty acid auxotrophs (Figure 6B). In contrast, OA supplementation in delipidated media largely restored growth of non-iners FGT Lactobacillus species (n=45 strains) at levels equivalent to growth in untreated media, while L. iners failed to grow with OA supplementation (n=13 strains; Figure 6B). Acetate supplementation was not inhibitory towards non-iners FGT Lactobacillus species in non-delipidated media, demonstrating that acetate’s failure to rescue growth was due to these species’s inability to use it as a FASII precursor rather than direct toxicity (n=45 strains; Figure S9B). Taken together, these results confirm that all FGT Lactobacillus species are fatty acid auxotrophs but only farE-harboring species can withstand and utilize high concentrations of OA to enhance growth.

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.

Based on these findings, we hypothesized that FGT lactobacilli may utilize exogenous OA as precursors for phospholipid synthesis. To characterize exogenous OA utilization, we used ¹³C₁₈-OA to isotopically trace the ¹³C label in potential downstream metabolites and biosynthetic pathways. As hypothesized, we observed the ¹³C label to be incorporated into structural lipid metabolites, such as phosphatidylglycerol (PG) and diglyceride lipids, in all FGT Lactobacillus species (Figures 6C and S9C). These were primarily observed in PG34:1, PG36:1, and PG36:2. Notably, membranes of L. crispatus, L. gasseri, and L. jensenii grown in high OA concentrations consisted largely of the OA-containing PGs, PG34:1 and PG36:2, indicating immense versatility of these organisms to incorporate exogenous OA into their membranes (Figures 6C and S9D). In contrast to their ability to use exogenous fatty acids for phospholipid synthesis, genomic analysis of FGT Lactobacillus species revealed no predicted beta-oxidation pathway gene functions. We experimentally confirmed this prediction by showing that no ¹³C label was incorporated into TCA cycle metabolites, indicating that FGT lactobacilli do not exploit OA for central carbon metabolism (Figure S9C). Thus, all FGT Lactobacillus species rely on environmental fatty acids to build their membranes while farE-harboring non-iners FGT lactobacilli possess a greater ability to resist and utilize high levels of exogenous OA.

Non-iners FGT lactobacilli use OhyA to sequester OA in a derivative form only they can exploit

We hypothesized that since OhyA activity can be bidirectional, OhyA-mediated bioconversion of uLCFAs to their hFA counterparts might allow non-iners FGT Lactobacillus species to sequester OA in a derivative form that they could exploit but that competing ohyA-deficient species could not. To test this hypothesis, we assessed whether supplementation with 10-HSA (h18:0), the OhyA9-produced metabolite that is highly elevated in bacterial communities dominated by non-iners FGT lactobacilli, could enhance growth of these species in a similar manner to unmodified OA. As hypothesized, non-iners FGT Lactobacillus species (n=17 strains) exhibited growth enhancement when cultured in delipidated MRS+CQ broth supplemented with either OA or 10-HSA, while L. iners (n=5 strains) failed to grow in either condition (Figure 6D). 10-HSA was also inhibitory towards L. iners, albeit with somewhat lower potency than unmodified OA (~4-fold higher MIC; Figure S9F). These results confirmed that only non-iners FGT Lactobacillus species derived a growth benefit from 10-HSA.

To determine whether ohyA9 was required for this 10-HSA-driven growth benefit, we performed 10-HSA growth experiments in delipidated media using the L. gasseri genetic mutant strains (WT, ΔfarE, ΔfarE/pfarE, ΔohyA9, and ΔohyA9/pohyA9). The ΔohyA9 strain failed to grow when supplemented with 10-HSA, but this growth defect was rescued in the ΔohyA9/pohyA9 complemented strain (Figure 6E), confirming that ohyA9 was necessary for 10-HSA-dependent growth enhancement. Importantly, 10-HSA did not inhibit ΔohyA9 in non-delipidated MRS+CQ broth, indicating that the lack of 10-HSA growth enhancement was due to inability to bioconvert 10-HSA to OA rather than direct toxicity (Figure S9E). These results show that non-iners FGT lactobacilli can employ OhyA9 to bioconvert and sequester OA in a derivative form that they are able to exploit in an OhyA9-dependent manner for growth.

OA inhibits growth of key BV-associated bacteria, including MTZ-resistant strains

To evaluate the potential of OA to improve BV treatment, we characterized effects of OA alone and in combination with MTZ on growth of various non-Lactobacillus FGT bacteria, including canonically BV-associated species (Figures 7A and S10A–B). Remarkably, mono-cultures of the BV-associated species Gardnerella vaginalis, Gardnerella piotii, Fannyhessea (formerly Atopobium) vaginae, Sneathia amnii, and Prevotella timonensis were all robustly inhibited by OA, while other Prevotella species including P. bivia, P. amnii, and P. disiens were largely unaffected in NYCIII broth (Figure 7A). These assays were repeated in S-broth with similar results except for additional OA-driven inhibition of P. amnii (Figure S10B). Interestingly, even MTZ-resistant G. piotii and F. vaginae strains were susceptible to OA inhibition. To further characterize these OA-sensitive non-Lactobacillus species, we performed dose-response assays in NYCIII broth and observed Gardnerella MICs to be approximately 2-fold greater than the median MIC of diverse L. iners strains (Figure S10C). Taken together, these results suggest that OA may additionally improve existing BV treatment via inhibition of key BV-associated bacteria, including MTZ-resistant strains, without any expected detriment to L. crispatus.

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).

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

The effects of OA on both FGT lactobacilli and BV-associated bacteria suggested it could have therapeutic potential to shift vaginal bacterial communities towards L. crispatus dominance, alone or in combination with MTZ. We tested this hypothesis using an established approach employing defined, BV-like bacterial communities in vitro⁴³. The input communities, which included a predominance of BV-associated bacteria, L. iners at lower abundance, and L. crispatus at very low abundance (<2%), were cultured in rich, non-selective media with or without addition of OA and/or MTZ for 72 hours (Figure 7B). Composition of the input mixtures and the resulting in vitro communities was assessed by 16S rRNA gene sequencing, and total community growth was confirmed by optical density (Figures S11A–C). Untreated communities maintained diverse, BV-like compositions, while communities treated with MTZ alone became dominated by L. iners, recapitulating patterns observed in human studies of BV treatment^21,25–28 (Figure 7B). By contrast, treatment with OA alone significantly promoted L. crispatus dominance and suppressed L. iners and most BV-associated species, although some Prevotella species were not inhibited. Combining OA with MTZ generally preserved or enhanced L. crispatus dominance relative to OA alone, while decreasing Prevotella abundance. Similar results were observed using defined, BV-like communities containing L. gasseri or L. jensenii instead of L. crispatus, and in experiments culturing these same communities in S-broth (Figures S12A–B). Thus, OA showed robust ability to enhance microbiota dominance by L. crispatus and other non-iners FGT Lactobacillus species alone or combined with MTZ in this model, highlighting its potential to improve existing BV therapies.

Discussion

In this study, we described important differences in fatty acid metabolism among FGT Lactobacillus species that reveal fundamental principles of nutrient utilization in the FGT microbiota and point to new strategies for improving BV treatment. Specifically, we found that cis-9-uLCFAs, exemplified by OA, selectively inhibit L. iners and key BV-associated species while robustly promoting growth of L. crispatus and other health-associated non-iners FGT lactobacilli. Transcriptional profiling of non-iners FGT lactobacilli revealed a core set of genes upregulated in response to OA and other cis-9-uLCFAs, including a predicted oleate hydratase enzyme and putative fatty acid efflux pump. These OA-induced genes were genomically conserved in non-iners FGT Lactobacillus species but universally absent in L. iners, mirroring the observed growth phenotypes. We characterized the predicted Lactobacillus OhyA orthologs, confirming they exhibited OhyA activity by hydrating uLCFA cis-9 double bonds in vitro. We further showed that their biochemical products were uniquely elevated in cervicovaginal fluid of women with L. crispatus-dominated microbiota. Notably, we overcame the historical technical challenge of conducting genetic experiments in FGT lactobacilli to assess the roles of farE and ohyA in mediating these observed growth phenotypes. Knocking out farE abolished the cis-9-uLCFA resistance phenotype, which could be restored by genetic complementation. We further showed that FGT lactobacilli are fatty acid auxotrophs, but only farE-harboring species could withstand and utilize high levels of OA for phospholipid synthesis. Additionally, we found that non-iners FGT Lactobacillus species can employ OhyA to sequester exogenous OA in a derivative form that only they can exploit, demonstrating an important nutrient competition strategy for these fatty acid auxotrophic species. Finally, we demonstrated that OA treatment alone or in combination with MTZ could robustly shift defined BV-like bacterial communities towards L. crispatus-dominated states. Together, these results advance our mechanistic understanding of FGT Lactobacillus metabolism and identify novel therapeutic strategies to modify the vaginal microbiota.

Our description of a role for uLCFAs in selectively inhibiting L. iners adds to existing knowledge about antimicrobial properties of uLCFAs and their modes of action in mammalian-adapted bacteria³⁸. Antimicrobial properties of host-produced free fatty acids contribute to innate defense mechanisms on the human skin, in oral mucosa, and likely other mucosal surfaces^40,60,61. The potency and mechanisms of antibacterial activity of these fatty acids vary across bacterial strains and species. For example, POA and sapienic acid potently inhibit S. aureus via physical disruption of its cell membrane, leading to leakage of low molecular weight proteins and solutes and subsequent interference with energy metabolism³⁹. Similarly, sapienic acid and lauric acid physically disrupt the membranes of Porphyromonas gingivalis and lead to cell lysis⁶². In contrast, high concentrations of OA and LOA are reported to induce bacteriostatic inhibition in certain S. aureus strains by inhibiting the FASII enzyme, FabI⁶³. Based on the rapid lysis induced by OA in L. iners, we propose that its bactericidal action results from a membrane-disrupting mechanism. We speculate that L. iners’ inability to prevent toxic intracellular accumulation of these fatty acids and regulate their membrane insertion contributes to its cis-9-uLCFA susceptibility.

Our identification of a conserved cis-9-uLCFA-induced response mechanism in non-iners FGT lactobacilli is consistent with uLCFA resistance mechanisms previously described in other organisms. For example, S. aureus FarE was reported to confer resistance to LOA and arachidonic acid via uLCFA efflux⁶⁴ and Tet38, a separate S. aureus fatty acid efflux pump, promotes resistance to POA⁶⁵. A functionally similar MtrCDE efflux system in Neisseria gonorrhoeae exports hydrophobic molecules, such as LCFAs, and enhances survival in fatty acid enriched environments, including the genital tract of a female mouse model⁶⁶. Thus, our finding that FarE mediates cis-9-uLCFA resistance in non-iners FGT lactobacilli, agrees with this larger body of literature showing that fatty acid efflux pumps serve as an important resistance mechanism against exogenous fatty acids and can confer a competitive advantage in LCFA-rich environments.

Interestingly, knocking out ohyA9 in L. gasseri did not increase cis-9-uLCFA susceptibility in our in vitro experimental conditions, contrasting with findings in S. aureus and S. pyogenes, where OhyA was protective against POA toxicity^49,52. Thus, the role of OhyA in protecting bacteria against uLCFAs may be specific to certain species or environmental contexts. In addition to protecting against uLCFA toxicity, OhyA activity may play a role in tolerogenic host signaling through hFA products as shown for S. aureus infections⁵⁶ or in producing products inhibitory to other bacteria or fungi⁶⁷. Future studies will be required to determine how OhyA enzymes contribute to host-microbe crosstalk and competitive microbial chemical warfare.

Our finding that FGT Lactobacillus species are fatty acid auxotrophs offers insight into a previously underappreciated nutrient requirement and their metabolic niche in the FGT microbiome. These results agree with previous reports of closely related non-FGT Lactobacillus species (Lactobacillus johnsonii, Lactobacillus helveticus, and L. acidophilus) being fatty acid auxotrophs as well^68,69. Such findings imply that many Lactobacillus species rely on fatty acids derived from their environment, host, or other microbial species. A critical next step will be understanding which specific sources within the FGT support Lactobacillus species growth in vivo.

Based on our finding that ohyA9 confers a growth advantage to non-iners Lactobacillus species in the presence of 10-HSA, we propose that these species employ OhyA9 to sequester environmental cis-9-uLCFAs in a derivative form inaccessible to non-ohyA9-harboring organisms in the FGT microbiota. This proposed strategy aligns with previously described competitive microbial mechanisms for molecularly encrypting limited nutrients to prevent access by competing species⁷⁰. For example, bacteria produce species-specific siderophores to bind and sequester environmental metals^71,72, rendering them inaccessible to organisms lacking the cognate siderophore receptor. Siderophores also serve as only a temporary molecular modification of the acquired nutrient, which is shed following entry into the bacterial host^71,72. Corrinoid cofactors are another example of molecular nutrient encryption, with their diverse structures dictating which bacteria can take up and utilize the cofactor with corrinoid-compatible transporters and enzymes^73,74. We propose that OhyA9 activity in FGT lactobacilli is another example of the diverse mechanisms bacteria employ to acquire, sequester, and utilize limited nutrients in competition with other species.

Finally, we demonstrated the therapeutic potential of OA to improve upon standard treatment for BV. Promotion of L. crispatus over L. iners in the FGT microbiota is a core objective of BV treatment for which novel approaches are needed. Methods to target MTZ-resistant BV-associated bacteria may also be important, although the extent to which MTZ-resistant bacteria contribute to BV recurrence remains unclear^29–31. We find that – in addition to inhibiting L. iners growth – OA also inhibits several BV-associated species, including MTZ-resistant G. piotii and F. vaginae strains. Of particular interest is OA-driven inhibition of F. vaginae, a frequently MTZ-resistant species that has been linked to BV treatment failure³¹. We found that treatment of defined, in vitro BV-like communities with OA (alone or in combination with MTZ) strongly promoted L. crispatus dominance while suppressing both L. iners and BV-associated anaerobes, such as Gardnerella species, P. timonensis, F. vaginae, and S. amnii. Current investigational therapies for BV include vaginal microbiome transplants and single- or multi-strain Lactobacillus live biotherapeutic products that aim to shift microbiota composition by introducing health-associated species^24,32,33. However, preliminary clinical efficacy has been modest. We propose that supplementing delivery of a L. crispatus-containing live biotherapeutic and MTZ treatment with OA or other uLCFAs could improve efficacy by promoting L. crispatus colonization and inhibiting competition from L. iners and BV-associated bacteria.

In summary, we identified and characterized important species-level differences in fatty acid resistance, metabolism, and nutrient sequestration mechanisms among FGT Lactobacillus species, demonstrated in vivo relevance of these mechanisms, and provide preclinical evidence to support translating the findings into a novel therapeutic strategy for BV. This study illustrates how functional, metabolic, and genomic approaches can inform development of microbiota-targeted therapies to improve human health.

Limitations of the study

While this study provides mechanistic and functional support for its primary findings, the work has several limitations. Genetic manipulation experiments were performed only in L. gasseri due to lack of accessible genetic tools for other FGT Lactobacillus species. Genetic manipulation of these organisms (including L. gasseri) has been a major challenge for the field^75–79, but we succeeded in generating and complementing the necessary mutants in L. gasseri, which closely resembles L. crispatus in genomic, phylogenetic, transcriptional, enzymatic, and phenotypic features relevant to uLCFA resistance. Similarly, lack of genetic tools for manipulating L. iners prevented us from performing gain-of-function genetic experiments in L. iners to heterologously express uLCFA-induced genes from non-iners lactobacilli⁸⁰. Finally, we employed a previously described in vitro model of BV⁴³ because animal models (including gnotobiotic mice and non-human primates) have been largely intractable to vaginal colonization by human FGT lactobacilli or fully diverse BV-like communities^81–84, precluding their use for preclinical testing of uLCFAs in vivo. Human trials will likely be needed to fully assess the effects of uLCFA treatments on relevant FGT bacterial species and communities.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Douglas S. Kwon (dkwon@mgh.harvard.edu).

Materials availability

All unique/stable reagents directly used in this study are available from the lead contact with a completed Materials Transfer Agreement. ATCC 33323 upp gene deletion mutant and the empty vector, pMZ7, used for double homologous recombination will be deposited at public repositories.

Data and code availability

Supplemental Data

Original data will be deposited in a public repository at XX.

• Supp. Data 1. Full sized TEM images

• Supp. Data 2. RNA-sequencing data and reference genomes

• Supp. Data 3. WGS of genetic knockout strains

• Supp. Data 4. 16s rRNA gene sequencing data for competition assays

• Supp. Data 5. 16s rRNA gene sequencing data for human samples

Supplemental Code

Original code will be deposited in a public repository at XX.

Methods

Bacteria strains and culture conditions

Bacterial strains used in this study for each experiment/figure are summarized in Table S1. All vaginal bacterial isolates were cultured under anaerobic conditions at 37–40°C using a palladium catalyst-based anaerobic chamber (COY) with an atmosphere of 5% carbon dioxide, 5% hydrogen, and 90% nitrogen (Airgas #X03NI90C3001054). All media, culture reagents, and plastic-ware were pre-reduced by placement overnight in the anaerobic chamber before use. Vaginal bacterial isolates were first revived on solid agar plates to obtain single colonies. After 4–5 days of incubation, representative colonies were picked into liquid culture. All Lactobacillus species isolates were revived from frozen 25% glycerol stocks on MRS agar plates (Hardy Diagnostics #89407–144), except for Lactobacillus iners, which was revived on Columbia Blood Agar (CBA) agar plates (Hardy Diagnostics #A16). Non-Lactobacillus species were revived from frozen 25% glycerol stocks on CBA agar plates. For cultivation of Lactobacillus species in liquid media, MRS+CQ broth was used unless otherwise noted. For cultivation of non-Lactobacillus species, NYCIII broth was used unless otherwise noted.

ATCC 33323 Lactobacillus gasseri and derivative mutant strains were cultured under anaerobic conditions for solid media and aerobic conditions for liquid media at 37°C. Strains were revived from frozen 25% glycerol stocks on MRS agar plates for 48 hours under anaerobic conditions using an anaerobic box with oxygen-eliminating sachets padded with a methylene blue anaerobic indicator (BD GasPak^™ EZ Anaerobe Container System Sachets with Indicator #260001). After incubation, a single colony was picked into liquid culture and incubated aerobically at 37°C with shaking. MRS+CQ broth was used for liquid culture unless otherwise noted.

Escherichia coli cloning strains, EC1000 (Addgene #71852) and MC1061 (Molecular Cloning Laboratories #MC1061), Staphylococcus aureus, and S. aureus derivative mutant strains were cultured under aerobic conditions at 37°C with shaking for liquid cultures. EC1000 was revived from frozen 25% glycerol stocks on LB agar plates supplemented with 40 μg/mL kanamycin (KAN). After overnight incubation, a single colony was picked into liquid culture and incubated aerobically at 37°C with shaking. LB media supplemented with 40 μg/mL KAN was used for liquid culture unless otherwise noted. MC1061 strains with a pTRK892-derived plasmid (Addgene #71803) were revived from frozen 25% glycerol stocks on LB agar plates supplemented with 200 μg/mL erythromycin (ERY). LB media supplemented with 200 μg/mL ERY was used for liquid culture.

Media additives were added to autoclave-sterilized broth or agar plates after cooling to room temperature. Broth was re-sterilized using a 0.22 μM PES filter after all additives were added. To prepare MRS+CQ broth, BD Difco-formulated Lactobacillus MRS (De Man, Rogosa and Sharpe) broth (BD Biosciences #288130) was prepared as per the manufacturer’s instructions and then supplemented with L-cysteine (4 mM) and L. glutamine (1.1 mM) as previously described⁴³. To prepare delipidated MRS+CQ broth, four parts MRS+CQ broth was combined with one part Cleanascite^™ Lipid Removal Reagent (Biotech Support Group #X2555–10) and shaken (220 rpm) at room temperature for 10 min, then centrifuged at 2,000 x g for 15 min. The media supernatant was collected and re-sterilized using a 0.22 uM PES filter. NYCIII broth (ATCC medium 1685) pre-media was prepared with 15 g/L Proteose Peptone No. 3 (BD Biosciences #211693), 5 g/L sodium chloride, and 4 g/L HEPES in 875 mL of distilled water, pH-adjusted to 7.3, and autoclaved. After autoclaving and cooling the pre-media, dextrose solution (3 g / 45 mL distilled water) was added at 7.5% v/v, Gibco yeast extract solution (Gibco #18180–059) was added at 2.5% v/v, and heat inactivated horse serum (Gibco #26050070) was added at 10% v/v. The complete NYCIII broth was then sterilized by passage through a 0.22 μm PES filter. S-broth⁴³ pre-media was prepared with 37 g/L Brian Heart Infusion (BHI) broth (BD Biosciences #211059), 10 g/L yeast extract powder, and 1 g/L dextrose in 880 mL distilled water, brought to a boil until dissolved, and then autoclaved. After autoclaving and cooling the pre-media, fetal bovine serum (Millipore Sigma #F4135) was added at 5% v/v, vitamin K1-Hemin solution (BD Biosciences #B12354) was added at 5% v/v, and IsoVitaleX (BD Biosciences #B11875) enrichment was added at 2% v/v. The complete S-broth was then sterilized by passage through a 0.22 μm PES filter. To prepare LB media and agar, BD Difco-formulated Lysogeny broth (LB) media (BD Biosciences #244610) or agar (BD Biosciences #240110) plates were prepared as per the manufacturer’s instructions and then supplemented with pre-sterilized stocks of antibiotics or sodium acetate where indicated. For all fatty acid supplementations, oleic acid (≥99% purity Millipore Sigma #O1008), universally ¹³C₁₈-labeled oleic acid (¹³C₁₈-OA; ≥99% purity, Millipore Sigma #490431), palmitoleic acid (≥98.5% purity, Millipore Sigma #P9417), and linoleic acid (≥99% purity, Millipore Sigma #L1012) were pre-sterilized by passage through a 0.22 μm filter and directly added to broth media.

FRESH cohort and human samples

The FRESH cohort is an ongoing prospective observational study based in Umlazi, South Africa, that enrolls 18–23 year old, HIV-uninfected, sexually active, healthy women. Exclusion criteria included pregnancy, anemia, any chronic medical condition or other conflict likely to prevent study protocol adherence, and/or enrollment in any other study that involves frequent blood sampling or that might otherwise interfere with the FRESH study protocol. Study characteristics and inclusion and exclusion criteria have been described in detail elsewhere^10,13,43,90. The study protocol was approved by the Massachusetts General Hospital (MGH) Institutional Review Board (IRB, 2012P001812/ MGH) and the Biomedical Research Ethics Committee of the University of KwaZulu-Natal (UKZN; Ethics Reference Number BF131/11). All participants provided written informed consent.

Twice per week at the study site, participants received HIV RNA viral load testing by PCR and attended classes focused on personal empowerment, job skills training, and HIV prevention. Once every 12 weeks, participants provided peripheral blood and mucosal specimens described in more detail below. Participants were additionally provided a light meal during study site visits and cumulative monetary compensation over a 36-week period of ZAR3,700 (~US$280), meant to help defray transportation expenses for the twice-weekly study site visits.

For FRESH study mucosal specimen collections, cervicovaginal swab samples (Puritan 6” sterile standard foam swab with polystyrene handle) were collected by swabbing the ectocervix in two full revolutions under direct visualization during speculum exam. The swabs were then used to make a slide preparation for Gram stain analysis, cryopreserved in thioglycolate broth with 20% glycerol for bacterial isolation, or frozen without cryopreservatives for bacterial isolation or nucleic acid extraction and microbiota profiling. Cervicovaginal lavage (CVL) samples were collected using a flexible plastic bulb pipette to dispense 5 ml of sterile normal saline into the vagina and wash the cervix four times. Fluid was then re-aspirated into a 15 ml conical tube. CVL and swab samples were stored on ice for 1–4 hours during transport to the processing laboratory at the HIV Pathogenesis Programme, Doris Duke Medical Research Institute at the Nelson R. Mandela School of Medicine at UKZN, where swabs were stored at −80°C. CVL samples were centrifuged (700xg, 10 min at 4°C). Supernatants from the centrifuged CVL samples were then transferred to cryovials and stored at −80°C.

Bacterial culture conditions for growth inhibition, growth enhancement, and killing assays

To assess mono-culture growth effects, bacterial isolate stocks were first revived on solid media for 2–5 days until fully formed colonies were present. Starter liquid cultures were then inoculated with a few representative colonies for each strain and grown for 48 hours. After 48 hours, the optical density at 600 nm (OD600) of each liquid mono-culture was measured and each cultivated strain was back-diluted to 40X the starting OD600 (e.g. 40X OD600=0.8 for a starting OD600=0.02) for each experiment using the base media specific to that experiment. In cases where growth enhancement was being assessed, starter cultures were pelleted (5,000xg, 5 min at RT), washed with phosphate-buffered saline (PBS) three times, resuspended in PBS, and back-diluted to 40X the starting OD600 prior to assay inoculation to prevent nutrient carryover. Back-diluted starter cultures were used to inoculate the experimental media conditions. At the selected time point for each experiment, growth was assessed by optical density at 600 nm (OD600) with a SpectraMax M5 (Molecular Devices).

To assess bactericidal activity of OA, a minimum bactericidal concentration (MBC) assay was performed in MRS+CQ broth with varying concentrations of OA. Starter cultures and experimental conditions were prepared as described above. After 24 hours of incubation, colony forming units (CFU) were measured by serially diluted each culture condition 10-fold (e.g. 10 μL of culture into 100 μL of media) and 10 μL of each dilution was plated from neat (undiluted) to 10 ⁻⁷ dilution. Colonies were counted manually after 48–72 hours of incubation.

Competition and mock community culture experiments

Starter cultures of representative strains of L. crispatus, L. gasseri, L. iners, L. jensenii, G. piotii, G. vaginalis, P. amnii, P. bivia, P. disiens, P. timonensis, F. vaginae, and S. amnii were prepared in NYCIII broth. Aliquots of mono-cultured strains were mixed in defined ratios (see CFU input for each strain relative to L. crispatus 1 in Table S9) and then divided and used to inoculate replicate cultures across different treatment conditions for incubation (6 replicates per condition) by adding 150 μL of mixture in V-bottom 96 well plates. Treatment conditions included NYCIII broth (untreated), NYCIII broth supplemented with 3.2 mM oleic acid, NYCIII supplemented with 50 μg/mL metronidazole, and NYCIII broth supplemented with both 3.2 mM oleic acid and 50 μg/mL metronidazole. CFU (colony forming unit) titres were determined for each axenic culture as described above and used to calculate starting ratios within the mixed cultures. At 72 hours, cultures were harvested by centrifuging at 5,000xg for 10 min at 4°C, supernatant was removed, and pellets were frozen at −20°C for subsequent DNA extraction and analysis. Relative growth within mixed cultures was assessed by bacterial 16S rRNA gene sequencing as described below. OD600 was determined for each input community mixture (t=0 hours) and the final time point (t=72 hours) in the no-treatment control condition to confirm growth of the defined, in vitro community cultures. Aliquots of the mono-cultured strains were assayed separately in all treatment conditions to confirm expected growth patterns, and assessed by OD600 at the corresponding time points to confirm purity.

Sample preparation and conditions for TEM imaging

For TEM sample preparation, representative strains of L. crispatus and L. iners were revived on solid media, representative colonies were picked into liquid cultures, and incubated for 48 hours under standard culture conditions. Cultures were back diluted to a starting OD600 of 0.02 in untreated media and grown to exponential phase, OD600=0.4–0.6. Exponential cultures were then either left untreated or treated with 3.2 mM of oleic acid for 1 hour. After fatty acid treatment, cultures were removed from the anaerobic chamber, combined in a 1:1 ratio with fixative solution prepared at 2X the working concentration, and promptly pelleted (5,000xg, 5 min at RT). Samples were kept at room temperature for 2 hours to allow for pellet fixation. Fixative solution at 2X the working concentration contained 5% Glutaraldehyde (Electron Microscopy Services (EMS) #16000), 2.5% Paraformaldehyde (EMS #19200), 0.06% picric acid (EMS #19552), 20 mM Lysine (EMS #L5626), and 0.2% Ruthenium Red (EMS #20600) in 0.2 M cacodylate buffer pH 7.4.

Fixed pellets were washed three times in 0.1% Ruthenium Red in 0.1 M cacodylate buffer (EMS #11650) and postfixed with 1% Osmium tetroxide (OsO4, EMS #19100) + 0.1% Ruthenium Red for 2 hours, washed two times in water + 0.1% Ruthenium Red, and subsequently dehydrated in grades of ethanol (10 min each; 50%, 70%, and 90% ethanol once, and 100% ethanol twice). The samples were then put in propylene oxide (EMS #20401) for 1 hour and infiltrated overnight in a 1:1 mixture of propylene oxide and Spurr’s Low Viscosity Embedding media (EMS Catalog #14300). The following day the samples were embedded in Spurr’s Low Viscosity Embedding media (EMS #14300) and polymerized at 60°C for 48 hours. Ultrathin sections (about 80 nm) were cut on a Reichert Ultracut-S microtome, picked up on to copper grids (EMS #G-GVPB-Cu) stained with lead citrate (EMS 22410) and examined in a JEOL 1200EX Transmission electron microscope and images were recorded with an AMT 2k CCD camera.

Sample preparation for bulk RNA-sequencing of bacterial isolates

For transcriptomic profiling of cis-9-uLCFA-treated non-iners Lactobacillus cultures, representative strains of L. crispatus, L. iners, L. gasseri, and L. jensenii were revived on solid media, then representative colonies were picked into liquid cultures and incubated for 48 hours under standard culture conditions. Cultures were back-diluted to a starting OD600 of 0.02 in untreated media and grown to exponential phase, OD600=0.4–0.6. Exponential cultures were then either left untreated or treated with OA, LOA, or POA (3.2 mM each). To collect samples for bulk RNA sequencing after 1 hour of treatment, cultures were pelleted (5,000xg, 5 min at 4°C), the supernatant removed, and the pellets were resuspended in 500 μL of Trizol (Life Technologies Corporation #15596026), immediately put on dry ice, and stored at −80°C until RNA extraction and library preparation. Three replicates, each from an independent starter culture, were included for each strain and condition.

RNA extraction from bacterial isolates

Trizol-preserved bacterial samples were thawed, transferred to 2 mL FastPrep tubes (MP Biomedicals #115065002) containing 500 μL of 0.1 mm Zirconia/Silica beads (BioSpec Products #11079101z), and bead beaten for 90 seconds at 10 m/sec speed using the FastPrep-24 5G (MP Biomedicals #116005500) with a Metal QuickPrep adapter. After bead beating, samples were incubated on ice for 3 min, and 200 μL chloroform was added to each sample and mixed by tube inversion. After 3 min incubation at room temperature, samples were centrifuged (12,000xg, 15 min at 4°C) to form separated phase layers between the organic & aqueous sections. For each sample, 200 μL of the clear aqueous phase was transferred to separate clean tubes, mixed with an equal volume (200 μL) of 100% ethanol by tube inversion, and incubated for 5 min at RT. Using a Direct-zol RNA Purification kit (Zymo Research #R2070), each sample was then transferred to a Direct-zol spin column and centrifuged (16,500xg, 1 min at RT). Columns were washed twice with 400 μL Direct-zol RNA Pre-Wash Buffer, and once with 700 μL Direct-zol RNA Wash Buffer and then centrifuged at max speed for 2 min. To dry the pellet and remove any ethanol carryover, columns were centrifuged with lids open (12,000xg, 1 min at RT). Finally, columns were moved to RNase-free 1.5mL tubes, incubated with 100 μL of nuclease-free water for 5 min, and then eluted by centrifugation (12,000xg, 1 min at RT). Extracted RNA samples were kept on ice for use and QC or stored at −80°C.

Library preparation for bulk RNA-sequencing of bacterial isolates

Illumina cDNA libraries were generated using a modified version of the RNAtag-seq protocol⁹¹. Briefly, 250 ng of total RNA was fragmented, depleted of genomic DNA, dephosphorylated, and ligated to DNA adapters carrying 5’-AN₈-3’ barcodes of known sequence with a 5’ phosphate and a 3’ blocking group (IDT). Barcoded RNA molecules were pooled and depleted of rRNA using the Pan-Bacteria riboPOOL depletion kit (siTOOLs Biotech, Galen Laboratories #dp-K096). Pools of barcoded RNAs were converted to Illumina cDNA libraries in 2 main steps: (i) reverse transcription of the RNA using a primer designed to the constant region of the barcoded adaptor with addition of an adapter to the 3’ end of the cDNA by template switching using SMARTScribe Reverse Transcriptase (Takara ClonTech #639538) as described⁹²; (ii) PCR amplification using primers whose 5’ ends target the constant regions of the 3’ or 5’ adaptors and whose 3’ ends contain the full Illumina P5 or P7 sequences. cDNA libraries were sequenced on the Illumina NovaSeq SP 100 platform to generate paired end reads.

Analysis of RNA-sequencing data

Since samples were barcoded in library preparation and then pooled for sequencing, reads from each sample were demultiplexed based on their associated barcode sequence using custom scripts. Up to 1 mismatch in the barcode was allowed provided it did not assign the read a separate barcode included in the sequencing pool. Barcode sequences were removed from the first read as were terminal G’s from the second read that may have been added by SMARTScribe during template switching.

For each sample, reads were aligned to the sample species reference genome (GCF_022455535.1 / ASM2245553v1 for L. crispatus; GCF_022456925.1 / ASM2245692v1 for L. gasseri; GCF_022456915.1 / ASM2245691v1 for L. jensenii) using BWA⁹³ and read counts were assigned to genes and other genomic features using custom scripts. Differential expression analysis was conducted using DESeq2⁴⁶.

Preparation of Lactobacillus spent media and pellets for untargeted lipidomics and isotopic tracing

Starter cultures were prepared as described above, pelleted, and washed 3 times with PBS. Washed pellets were resuspended in PBS using the same volume as the initial start culture. Washed starter cultures were then back-diluted into 0, 0.1, or 3.2 mM OA or ¹³C₁₈-OA the tested media conditions and allowed to incubate under standard culture conditions for 72 hours. After incubation, cultures were pelleted (5,000xg, 10 min at 4°C). Supernatants were removed, collected into separate tubes, and placed onto dry ice until storage at −80°C. Cell pellets were washed once with ice cold PBS. Washed pellets were also placed onto dry ice and then stored at −80°C once frozen. Three replicates were included for each strain and condition.

Plasmid construction methods for strain generation in L. gasseri ATCC 33323

All vectors and primers with their respective annealing temperatures are noted in Tables S5 and S6, respectively. Primers were purchased/synthesized by IDT. All cloning host strains and mutants generated are noted in Table S7. All plasmids were verified with nanopore-based whole plasmid sequencing using Primordium Labs services. All mutant strains were verified with Illumina-based WGS using SeqCenter, LLC.

Genomic DNA extractions were performed using the DNeasy Blood & Tissue Kit (Qiagen Beverly LLC #69504) following manufacturer protocols for Gram positive bacterial pellet samples. Minipreps were performed using the QIAprep Spin Miniprep Kit (Qiagen Beverly LLC #27104) following the manufacturer’s protocol. Gel extractions were performed using the QIAquick Gel Extraction Kit (Qiagen Beverly LLC #28704) following the manufacturer’s protocol. Extracted and amplified DNA products were quantified using a NanoDrop 2000. DNA bands were visualized using E-Gel^™ EX Agarose Gels (2%, Invitrogen #G401002) under UV light.

Q5^® High-Fidelity DNA Polymerase (New England Bio Labs (NEB) #M0491S) was used for PCR amplification of products to be used for plasmid construction following the manufacturer’s protocol. Briefly, Q5^® High-Fidelity DNA Polymerase PCR reactions were performed in 25 μL reactions such that the final concentration of reagents were as follows, 1X Q5 Reaction Buffer, 200 μM dNTPs, 0.5 μM Forward Primer, 0.5 μM Reverse Primer, 0.02 units/μL Q5 High-Fidelity DNA Polymerase, and 1–10 ng of template DNA; and, thermocycling was performed at 98°C for 30 s, followed by 30 cycles of 98°C for 10 s, annealing temperature (noted in Table S6 for each primer pair) for 30 s, and 72°C for 20 s/kb of desired amplicon, with a final 2 min extension at 72°C. KAPA HiFi HotStart ReadyMix (Roche Holding AG #07958935001) was used for all colony PCR-screening reactions following the manufacturer’s protocol. KAPA HiFi HotStart ReadyMix PCR reactions were performed in 25 μL reactions such that the final concentration of reagents were as follows, 1X Ready Mix, 0.3 μM Forward Primer, 0.3 μM Reverse Primer, and a pipette-tip sample of a single colony; and, thermocycling was performed at 95°C for 3 min, followed by 35 cycles of 98°C for 20 s, annealing temperature (noted in Table S6 for each primer pair) for 15 s, and 72°C for 20 s/kb of desired amplicon, with a final 1 min extension at 72°C.

For restriction and ligation-based cloning, restriction enzyme-based digestion was performed at 37°C for 15 min following manufacturer’s protocols. Digestion reactions were performed in 50 μL reactions such that the final concentration of reagents were as follows, 1X rCutSmart Buffer and 0.4 units/μL per enzyme. Ligation was performed at room temperature for 10 min and then heat inactivated at 65°C for 10 min. Reactions were performed in 20 μL reactions such that the final concentration of reagents were as follows, 1X T4 DNA Ligase Buffer, 37.5 ng of insert fragment, 50 ng of linearized backbone, and 1 unit/μL of T4 DNA Ligase (NEB #M0202). A no-insert with backbone only control was prepared in parallel with each ligation reaction to serve as a backbone self-ligation control. For Gibson cloning, Gibson Assembly^® Master Mix (NEB #E2611) was used for all Gibson assembly reactions following the manufacturer’s protocol. Gibson assemblies were performed in 20 μL reactions such that the final concentration of reagents were as follows, 1X Gibson Assembly Master Mix with 0.02–0.5 pmols of DNA fragments comprising 3–5 parts per insert fragment and 1 part linearized backbone. For plasmid construction, E. coli EC1000 strain⁹⁴ (Addgene #71852), a kanamycin resistant strains carrying a single copy of the repA gene in the glgB gene, was used as the cloning host for all pORI28-based plasmids, and E. coli MC1061 strain (Molecular Cloning Laboratories #MC1061) was used as the cloning host for all pTRK892-based plasmids.

Plasmid transformations methods for strain generation in L. gasseri ATCC 33323

Competent cells of E. coli EC1000 strain were prepared using the Mix & Go! E. coli Transformation Kit (Zymo Research #T3001) following the manufacturer’s protocol. 0.5 mL of an overnight culture of EC1000 grown in LB media at 37°C was transferred to 50 mL of ZymoBroth (Zymo Research #M3015) and cultured for 15–16 hours at 22°C with shaking to obtain a culture of OD600=0.4. Wash and Competent Buffers were diluted to 1X working concentration using the provided Dilution Buffer and kept on ice. The grown ZymoBroth culture was then pelleted (5,000xg, 10 min at 4°C), supernatant was removed, and pellets were washed with 5 mL of 1X Wash Buffer. The washed pellets were gently resuspended in 5 mL of ice cold 1X Competent Buffer and aliquoted into 100 μL volumes on ice. Aliquots were immediately used or stored at −80°C for transformation at a later time. For transformation of competent EC1000, 5 μL of plasmid DNA was added to 100 μL of cells, gently mixed, and incubated for 10 min on ice. After incubation, 400 μL of S.O.C. medium (Invitrogen #15544034) was added to the cells, which were then incubated for 1 hour at 37°C with shaking plated onto pre-warmed LB agar with antibiotic selection, and incubated overnight at 37°C. The no-insert cloning control and no DNA plasmid control samples were transformed and plated in parallel as negative controls for each transformation.

For transformation of E. coli MC1061, competent MC1061 were thawed on ice, aliquoted into 100 μL volumes into chilled, clean 1.5 mL microcentrifuge tubes, gently mixed with 5 μL of plasmid DNA, and incubated for 15 min on ice. Cells were heat-shocked for 45 seconds in a 42°C water bath and then immediately placed on ice for 2 min. 0.9 mL of S.O.C. medium was added, cells were incubated for 1 hour at 37°C with shaking, plated onto pre-warmed LB agar with antibiotic selection, and incubated overnight at 37°C. The no-insert cloning control and no DNA plasmid control samples were transformed and plated in parallel as negative controls for each transformation.

Competent cells of L. gasseri ATCC 33323 and derivative mutants were prepared fresh for each transformation. 3.5X sucrose:MgCl2 electroporation buffer (3.5X SMEB) buffer was used as the transformation buffer and prepared such that the final concentration was 952 mM sucrose and 3.5 mM MgCl2 at pH 7.2 in DI water and then sterilized using a 0.22 uM PES filter. A 500 mL stock of 3.5X SMEB was prepared fresh each month and stored at 4°C. A single colony of L. gasseri wild-type or mutant was picked and cultured aerobically in 100 mL of MRS broth for 15–16 hours at 37°C with shaking. After 15–16 hours, cells were pelleted (5,000xg, 10 min at 4°C) and resuspended on ice with 100 mL of ice cold 3.5X SMEB buffer. The resuspension was carefully mixed until the pellet was completely resuspended into a homogenous solution, re-pelleted (5,000xg, 10 min at 4°C), and then resuspended on ice again in 5 mL of ice cold 3.5X SMEB to obtain the electrocompetent cell mixture. 100 μL of the electrocompetent cells were aliquoted into pre-chilled 0.2 cm gap Gene Pulser Electroporation Cuvettes (Bio-rad Laboratories #1652086) and kept on ice. 1 ug of DNA plasmid (volume kept < 10 μL to prevent sparking) was added to the cuvette, gently mixed without creating any air bubbles, and incubated on ice for 5 min. After incubation, electroporation was performed with a Gene Pulser Xcell Electroporation System using the following conditions: 1.25 kV, 25 μF, and 200 Ω. Immediately after electroporation, the cuvette was held on ice for 5 min, then the sample was added to 1 mL of pre-warmed MRS broth and incubated for 3 hours at 37°C with shaking. After incubation, cells were gently pelleted (300xg, 5 min), plated onto MRS agar plates with selection antibiotic, and incubated for 48 hours at 37°C in an anaerobic box.

Gene knockout and complementation in L. gasseri ATCC 33323

L. gasseri ATCC 33323 was selected as a genetically tractable, representative non-iners Lactobacillus strain exhibiting cis-9-uLCFA resistance and OA growth enhancement phenotypes similar to other strains of L. gasseri, L. crispatus, L. jensenii, and L. mulieris. To generate gene knockout mutants in L. gasseri ATCC 33323, we adapted a previously reported uracil phosphoribosyltransferase (upp)-based two-plasmid homologous recombination system^95–97. This system exploits the 5-fluorouracil (5-FU) resistance of a upp-deficient parent strain for knockout construction. To briefly summarize the underlying principles, uracil phosphoribosyltransferase, central to the pyrimidine salvage pathway, catalyzes the conversion of uracil to uridine monophosphate. When provided 5-FU, uracil phosphoribosyltransferase will produce 5-fluorouridine-5’-monophosphate, which is a suicide inhibitor to thymidylate synthase, a required enzyme in DNA synthesis. Vectors used to generate the upp-deficient strain and other gene deletion mutants included pTRK669⁹⁵, pORI28⁹⁴, and pORI28-derived plasmids. pORI28 is an empty backbone that requires RepA for stable plasmid replication and propagation and that encodes an erythromycin (ERY) resistance gene as a selectable marker; it is used for chromosomal integration in Gram positive bacteria. pMZ7 is a pORI28-derived backbone inserted with lacZ and the L. acidophilus upp gene (Laupp), which serves as the counterselection gene when integrated chromosomally. pTRK669 is a chloramphenicol (CHLOR) resistant, temperature sensitive helper plasmid that encodes RepA. To summarize the overall approach, we first generated an upp gene-deleted mutant in L. gasseri ATCC 33323 (∆1245-WT; generated as described below), which was resistant to 5-FU. This ∆1245-WT mutant served as the parent strain for all additional mutants and is therefore referred to as the WT strain in the figures and the Results and Discussion text. The pTRK669 helper plasmid was electro-transformed into ∆1245-WT to create ∆1245-WT/pTRK669. Then pMZ7-derived vectors containing ~600-bp homology arms within each gene-of-interest were constructed and respectively electro-transformed into ∆1245-WT/pTRK669, where RepA expression from pTRK669 enabled them to be replicated and propagated. ∆1245-WT/pTRK669 strains containing each gene-of-interest-specific pMZ7-derived vector were cultured under ERY and CHLOR selection (7.5 μg/mL CHLOR and 5 μg/mL ERY) at 37°C, then subcultured 3–5 times under ERY selection only (2.5 μg/mL) at 42°C to cure the pTRK669 helper plasmid and select for single-crossover chromosomal integrants of the pMZ7-derived vector. After these subculturing steps, cells were plated onto MRS agar plates with ERY (5 μg/mL) and colony PCR-screened for pMZ7-derived vector chromosomal integration. Confirmed clones of ∆1245-WT containing single-crossover integrants of the pMZ7-derived vector targeting the gene-of-interest were next cultured without antibiotic selection to allow for vector resolution from the chromosome, then subcultured in 5-FU to counter-select for the desired gene deletion mutant strain via a second crossover event to generate an in-frame gene deletion knockout by double homologous recombination. All subculturing steps were performed by transferring 5% of the grown culture volume into the newly prepared broth of equal volume to make a 5% inoculum.

We generated the ∆1245-WT parent strain used for the above knockout approach as follows. To construct the pMZ4 vector for generating ∆1245-WT, we generated a pORI28 derivative with homology arms flanking the endogenous upp gene (LGAS_1245), each being ~600 bp for the upstream and downstream arms, cloned into pORI28 using restriction digest and ligation cloning methods. In brief, a 594-bp upstream region and a 616-bp downstream region flanking LGAS_1245 were each amplified from genomic DNA extracted from L. gasseri ATCC 33323 to produce PCR amplicons, LGAS_1245-up and LGAS_1245-dwn, (primers for LGAS_1245-up: 1245-SOE-1 and 1245-SOE-2; primers for LGAS_1245-dwn: 1245-SOE-3 and 1245-SOE-4). Splicing by overlap-extension PCR was used to fuse LGAS_1245-up and LGAS_1245-dwn, producing a single 1187-bp amplicon (primers: 1245-SOE-1 and 1245-SOE-4), which was gel-purified to yield only the fused amplicon product. The fused LGAS_1245 homology arm amplicon and pORI28 were each digested separately using BamHI-HF and SacI-HF. The digested pORI28 backbone was additionally treated with shrimp Alkaline Phosphatase (rSAP) during the digest reaction and gel-purified before use in cloning. The digested pORI28 backbone and fused LGAS_1245 homology arm amplicon were ligated using T4 ligase and subsequently deactivated by heating. 10 μL of the ligation reaction was used to transform competent EC1000 cells, which were then plated onto LB agar plates with 200 μg/mL ERY. The constructed plasmid, pMZ4, was miniprepped from a single colony and confirmed by sequencing.

To use the pMZ4 plasmid to generate the ∆1245-WT strain, the helper plasmid pTRK669 was first electro-transformed into freshly prepared competent cells of L. gasseri ATCC 33323, then plated onto MRS agar plates with 7.5 μg/mL CHLOR and incubated anaerobically for 48 hours at 37°C. Clones containing pTRK669 (L. gasseri ATCC 33323/pTRK669) were verified by colony PCR screening (primers: pTRK699_F1 and pTRK699_R1). Next, pMZ4 was electro-transformed into freshly prepared competent cells of verified L. gasseri ATCC 33323/pTRK669, which were then plated onto MRS agar plates with 7.5 μg/mL CHLOR and 5 μg/mL ERY and incubated anaerobically for 48 hours at 37°C. ATCC 33323/pTRK669+pMZ4 clones were verified by colony PCR screening (primers: pTRK699_F1 and pTRK699_R1; pORI28F1 and pORI28R1). ATCC 33323/pTRK669+pMZ4 was then broth cultured under 7.5 μg/mL CHLOR and 5 μg/mL ERY selection at 37°C, then subcultured 3–5 times under 2.5 μg/mL ERY selection only at 42°C to cure the pTRK669 helper plasmid and select for pMZ4 single-crossover chromosomal integrants, then plated onto MRS agar plates with 5 μg/mL ERY to identify pMZ4 single-crossover chromosomal integrant clones. pMZ4-integration clones verified by colony PCR screening (using primers 1245-up and 1245-dwn) were cultured without antibiotic selection, then subcultured into MRS broth with 100 μg/mL 5-FU to counter-select for a second crossover event to produce the upp gene deletion mutant strain, then plated onto MRS media agar with 100 μg/mL 5-FU and incubated anaerobically for 48 hours at 37°C. Multiple subcultures were grown in parallel to increase chances of obtaining an upp gene deletion strain. Obtained colonies were PCR-screened (primers: 1245-up and 1245-dwn) and single-colony purified 1–2 times. Purified clones were verified by WGS to confirm deletion of upp by double homologous recombination. This upp-deleted (∆1245-WT) strain (referred to in figures and main text as WT) was thus 5-FU resistant.

To construct the pMZ7 backbone vector for generating gene-specific knockouts, lacZ and the L. acidophilus upp gene (Laupp) were cloned into pORI28 using two-piece Gibson cloning. In brief, lacZ and Laupp were each amplified from pUC19 and genomic DNA extracted from L. acidophilus ATCC 4356, respectively, to produce PCR amplicons, lacZ and Laupp (primers for lacZ: pUC19lacZ-F and pUC19lacZ-R; primers for LAupp: upp-F and upp-R). Splicing by overlap-extension PCR was used to fuse lacZ (780 bp) and LAupp (759 bp), producing a single 1519-bp amplicon (primers: pUC19lacZ-F and upp-R), which was gel-purified. The fused lacZ-LAupp amplicon and pORI28 were each PCR-amplified separately (primers for pORI28: lacZupp_intF1 and lacZupp_intR1; primers for lacZ-LAupp: lacZupp_intF2 and lacZupp_intR2) and assembled using NEB Gibson Master Mix (NEB #E2611). 10 μL of the Gibson reaction was used to transform competent EC1000 cells, then plated onto LB agar plates with 200 μg/mL ERY. The constructed plasmid, pMZ7, was miniprepped from a single colony and confirmed by sequencing.

To generate pMZ7-derived vectors for generating gene-specific deletion mutants in L. gasseri ∆1245-WT, pMZ9 (containing homology arms for ohyA9, LGAS_1351) and pMZ10 (containing homology arms for farE, LGAS_1630) were constructed using three-piece Gibson cloning in EC1000 cells. For pMZ9 construction, pMZ7 was linearized by PCR amplification (primers: LGAS_1351F9 and LGAS_1351R9). A 600-bp in the upstream region of LGAS_1351 (primers: LGAS_1351F10 and LGAS_1351R10) and a 600-bp in the downstream region of LGAS1351 (primers: LGAS_1351F11 and LGAS_1351R11) were PCR-amplified. All amplicons products were combined via Gibson reaction to make pMZ9. For pMZ10 construction, pMZ7 was linearized by PCR amplification (primers: LGAS_1630F9 and LGAS_1630R9). A 628-bp in the upstream region of LGAS_1630 (primers: LGAS_1630F10 and LGAS_1630R10) and a 627-bp in the downstream region of LGAS1630 (primers: LGAS_1630F11 and LGAS_1630R11) were PCR-amplified. All amplicons products were combined via Gibson reaction to make pMZ10.

To generate gene deletion mutants in ∆1245-WT, pTRK669 was electro-transformed into freshly prepared competent cells of ∆1245-WT, then plated onto MRS agar plates with 7.5 μg/mL CHLOR and incubated anaerobically for 48 hours at 37°C. Clones containing pTRK669 (∆1245-WT/pTRK669) were verified by colony PCR screening (primers: pTRK699_F1 and pTRK699_R1). Next, the pMZ7-derived vector containing homology arms for the gene of interest (pMZ9 or pMZ10) was electro-transformed into freshly prepared competent cells of ∆1245-WT/pTRK669, then plated onto MRS agar plates with 7.5 μg/mL CHLOR and 5 μg/mL ERY and incubated anaerobically for 48 hours at 37°C. ∆1245-WT/pTRK669+pMZ7-derived_vector clones (pMZ9 or pMZ10) were verified by colony PCR screening (primers for PTRK669: pTRK699_F1 and pTRK699_R1; primers for pMZ7-derived vector: pORI28F1 and pORI28R1). To select for pMZ7-derived_vector chromosomal integration clones, ∆1245-WT/pTRK669+pMZ7-derived_vector was subjected to the same protocol described above for generating pMZ4 single-crossover chromosomal integrant clones from ATCC 33323/pTRK669+pMZ4. Colonies were PCR-screened for pMZ7-derived_vector chromosomal integration (primers for LGAS_1351: S1351_F1 and S1351_R2; primers for LGAS_1630: S1630_F1 and S1630_R2). To isolate the desired gene deletion mutant, verified pMZ7-derived_vector-integration clones were subjected to the same protocol described above for generating the ∆1245-WT strain from the pMZ4-integration clones. Obtained colonies were PCR-screened with primers flanking the gene of interest (primers for LGAS_1351: S1351_F1 and S1351_R2; primers for LGAS_1630: S1630_F1 and S1630_R2) and single-colony purified 1–2 times. Purified clones were verified by WGS to confirm deletion of the gene.

To construct expression vectors for genetic complementation, genes were cloned into pTRK892⁹⁸, an erythromycin resistant vector backbone containing a strong Ppgm promoter, using Gibson cloning. In brief, pTRK892 was linearized by PCR amplification such that the strong Ppgm promoter was retained and original gene insert, a mutated form of GusA, was excluded (primers: pTRK892_F1G and pTRK892_R1G). To construct pMZ12 (the ohyA9 expression vector, also referred to a pohyA9), ohyA9 (LGAS_1351) was amplified from purified genomic DNA from L. gasseri ATCC 33323 (primers: 1351.FOR and 1351.REV) and combined with the linearized pTRK892 via a Gibson reaction. To construct pMZ13 (the farE expression vector, also referred to as pfarE), farE (LGAS_1630) was amplified from purified genomic DNA from L. gasseri ATCC 33323 (primers: 1630G.FOR and 1630G.REV) and combined with the linearized pTRK892 via a Gibson reaction. Constructs were transformed into competent MC1061 and plated onto LB agar with 200 μg/mL ERY. Plasmids were miniprepped from a single colony and confirmed by sequencing. pMZ12 was electro-transformed into freshly prepared competent cells of ∆ohyA9 to make ∆ohyA9/pohyA9, and pMZ13 was electro-transformed into freshly prepared competent cells of ∆farE to make ∆farE/pfarE. Transformed cells were plated onto MRS agar plates with 5 μg/mL ERY and incubated anaerobically for 48 hours at 37°C. Complementation was verified by colony PCR-screening (primers: Erm_F1 and Erm_R1).

Plasmid construction and complementation in Staphylococcus aureus USA300 ΔSaohyA

All vectors and genetic mutants generated in S. aureus are noted in Tables 5 and 7, respectively. To construct vectors for genetic complementation in the ohyA-knockout S. aureus USA300 strain (ΔSaohyA), His-tagged LCRIS_00661 and LCRIS_00558 genes with the appropriate restriction sites were ordered from Invitrogen and subcloned into a previously constructed S. aureus expression vector⁴⁹ (pPJ480) using restriction enzymes NcoI and XhoI in TOP10 chemically competent E. coli (Invitrogen #C4040) to make pLCRIS_00661 and pLCRIS_00558. Constructed plasmids were verified by nanopore-based whole plasmid sequencing using Primordium Labs services. Plasmids were next laundered through S. aureus strain RN4220, which can accept plasmids propagated through E. coli due its inactivated restriction system^99,100, and then S. aureus RN4220-derived plasmids were purified and used to transform ΔSaohyA. Electroporation was used to transform all S. aureus strains. Briefly, 1.5–2 μg of DNA was incubated with the S. aureus strain on ice for 5 min, the mixture was then electroporated in pre-chilled 0.1 cm gap Gene Pulser Electroporation Cuvettes (Bio-rad Laboratories #1652089) with a Gene Pulser Xcell Electroporation System using the following conditions: 1.6 kV, 25 μF, and 200 Ω. Immediately after electroporation, cells were incubated for 2 hours in Brain Heart Infusion broth (BHI; BD Biosciences #DF0037) at 37°C with shaking. After incubation, cells were gently pelleted (300xg, 5 min at RT), plated onto BHI agar plates with 10 μg/mL CHLOR, and incubated overnight at 32°C.

Enzyme product characterization

In order to measure production of OhyA metabolites, strains ΔSaohyA/empty vector, ΔSaohyA/pSaohyA, ΔSaohyA/pLCRIS_00558, and ΔSaohyA/pLCRIS_00661 were grown to an OD600 of 0.5 in Tryptone broth containing 1% DMSO. OA or LOA was added to a final concentration of 20 μM, and the cultures were grown for 1 hour at 37°C with shaking. The cells were separated from media by centrifugation, and the medium was extracted by adding methanol to a final concentration of 80%. Extracts were centrifuged to pellet debris, and the supernatant was analyzed by LC-MS as described below to detect the presence of the hFAs.

Culture supernatants containing the fatty acid substrate and hFA product were analyzed with a Shimadzu Prominence UFLC attached to a QTrap 4500 equipped with a Turbo V ion source (Sciex). Samples were injected onto an XSelect^® HSS C18, 2.5 μm, 3.0 × 150-mm column (Waters) at 45°C with a flow rate of 0.4 ml/min. Solvent A was water, and solvent B was acetonitrile. The HPLC program was as follows: starting solvent mixture of 60% B, 0–1 min isocratic with 60% B; 1–16 min linear gradient to 100% B; 16–21 min isocratic with 100% B; 21–23 min linear gradient to 0% B; and 23–28 min isocratic with 0% B. The Sciex QTrap 4500 was operated in the negative mode, and the ion source parameters were: ion spray voltage, −4500 V; curtain gas, 30 psi; temperature, 320°C; collision gas, medium; ion source gas 1, 20 psi.; ion source gas 2, 35 psi; and declustering potential, −35 V. The system was controlled by Analyst^® software (Sciex).

The Sciex QTrap 4500 mass spectrometer was operated in the negative mode using the product scan to determine the position of the hydroxyl group by direct injection. The source parameters were: ion spray voltage, −4500 V; curtain gas, 15 psi.; temperature, 250°C; collision gas, high; ion source gas 1, 15 psi; ion source gas 2, 20 psi; declustering potential, −25 V; and collision energy, −35 V. The system was controlled by Analyst^® software (Sciex).

Targeted lipidomics for the detection of hFAs in cervicovaginal lavage samples

Picolylamide derivatization was used to sensitively and accurately detect hFA abundance in CVL supernatant samples using the unique ions generated from breakage at the hydroxyl group position^56,58. 250 μL of each human CVL sample was added to 750 μL of methanol containing 200 nM ¹³C₁₈-OA (Cambridge Isotope Laboratories, Inc. #CLM-460-PK), incubated on ice for 15 min, and centrifuged at 4,000 rpm for 10 min. Supernatant was removed to a new tube and dried in a speed-vac overnight. 500 μL of oxalyl-chloride was added to the dried samples and incubated at 65°C for 15 min. After drying the samples under N₂ gas, 500 μL of 1% 3-picolylamine in acetonitrile was added and incubated at room temperature for 15 min. After drying the samples under N₂ gas, the samples were resuspended in 100 μL of ethanol for analysis.

Picolylamide-hFA were analyzed using a Shimadzu Prominence UFLC attached to a QTrap 4500 equipped with a Turbo V ion source (Sciex). Samples (5 μL) were injected onto an XSelect HSS C18, 2.5 μm, 3.0 × 150 mm column (Waters) at 45°C with a flow rate of 0.4 mL/min. Solvent A was 0.1% formic acid in water, and solvent B was acetonitrile with 0.1% formic acid. The HPLC program was the following: starting solvent mixture of 70% A/30% B; 0 to 5 min, isocratic with 30% B; 5 to 15 min, linear gradient to 100% B; 15 to 23 min, isocratic with 100% B; 23 to 25 min, linear gradient to 30% B; 25 to 30 min, isocratic with 30% B. The QTrap 4500 was operated in the positive mode, and the ion source parameters for the picolylamide-hFA multiple reaction monitoring (MRM) parameters were: ion spray voltage, 5,500 V; curtain gas, 15 psi; temperature, 300°C; ion source gas 1, 15 psi; ion source gas 2, 20 psi; declustering potential, 25 V, and a collision energy, 40 V. MRM masses (Q1/Q3) were: picolylamide-h18:0, 391.1/109.0; picolylamide-h18:1, 389.1/109.0; and picolylamide-¹³C₁₈-OA, 391.1/109.0. The system was controlled by the Analyst software (Sciex) and analyzed with MultiQuant^™ 3.0.2 software (Sciex). The relative concentration of each hFA was calculated based on the known amount of ¹³C₁₈-OA spiked into the sample at the beginning of the sample preparation.

Targeted lipidomics for the detection of PG metabolites in cell pellets

Lipids were extracted from bacterial cell pellets using the Bligh and Dyer method¹⁰¹. In brief, cell pellets were homogenized with a 1:2 mixture of chloroform:methanol with bead beating. The mixture was then filtered through Whatman No. 1 filter paper (Millipore Sigma #WHA1001090), and the filtrate was allowed to separate into two layers, an alcohol and chloroform layer. The alcohol layer was removed and the remaining chloroform layer contained the lipid extract. Lipid extracts were resuspended in chloroform/methanol (1:1). PG was analyzed using a Shimadzu Prominence UFLC attached to a QTrap 4500 equipped with a Turbo V ion source (Sciex). Samples were injected onto an Acquity UPLC BEH HILIC, 1.7 um, 2.1 × 150 mm column (Waters) at 45°C with a flow rate of 0.2 ml/min. Solvent A was acetonitrile, and solvent B was 15 mM ammonium formate, pH 3. The HPLC program was the following: starting solvent mixture of 96% A/4% B; 0 to 2 min, isocratic with 4% B; 2 to 20 min, linear gradient to 80% B; 20 to 23 min, isocratic with 80% B; 23 to 25 min, linear gradient to 4% B; 25 to 30 min, isocratic with 4% B. The QTrap 4500 was operated in the Q1 negative mode. The ion source parameters for Q1 were as follows: ion spray voltage, −4,500 V; curtain gas, 25 psi; temperature, 350°C; ion source gas 1, 40 psi; ion source gas 2, 60 psi; and declustering potential, −40 V. The system was controlled by the Analyst software (Sciex). The sum of the areas under each peak in the mass spectra were calculated, using LipidView software (Sciex).

Untargeted lipidomics and isotopic tracing methods and analysis

C18-neg:

Reversed-phase C18 chromatography/negative ion mode MS detection was used to measure metabolites of intermediate polarity. Analyses of metabolites of intermediate polarity, including free fatty acids and bile acids, were conducted using an LC-MS system comprised of a Shimadzu Nexera X2 U-HPLC (Shimadzu Corp.) coupled to a Q-Exactive orbitrap mass spectrometer (Thermo Fisher Scientific). Media samples (30 μL) were extracted for analyses using 90 μL of methanol containing 50ng/mL 15-methyl PGE1, 15-methyl PGA2, 15-methyl PGE2 (Cayman Chemical Co.) as internal standards. Cell pellet samples (30 μL) were extracted for analyses using 90 μL of methanol containing 50ng/mL 15-methyl PGE1, 15-methyl PGA2, 15-methyl PGE2 (Cayman Chemical Co.) as internal standards. The cell pellets were homogenized using the QIAGEN TissueLyser II with 3mm Tungsten beads for 4 min at a frequency of 20 Hz. Samples were centrifuged (10 min, 15000xg at 4°C). After centrifugation, supernatants (2 μL) were injected directly onto a 150 × 2.1 mm, 1.8 μm ACQUITY HSS T3 C18 column (Waters). The column was eluted isocratically with 80% mobile phase A (0.01% formic acid in water) for 3 min followed by a linear gradient to 100% mobile phase B (0.01% acetic acid in acetonitrile) over 12 min. MS analyses were carried out using electrospray ionization in the positive ion mode using full scan analysis over 70–850 m/z at 70,000 resolution and 3 Hz data acquisition rate. Other MS settings were: sheath gas 45, in source CID 5 eV, sweep gas 10, spray voltage −3.5 kV, capillary temperature 320°C, S-lens RF 60, probe heater temperature 300°C, microscans 1, automatic gain control target 1e6, and maximum ion time 250 ms. Raw data were processed using TraceFinder software (Thermo Fisher Scientific) for targeted peak integration and manual review of a subset of identified metabolites and using Progenesis QI (Nonlinear Dynamics) for peak detection and integration of both metabolites of known identity and unknowns.

C8-pos:

Reversed-phase C8 chromatography/positive ion mode MS detection was used to measure lipids. Analyses of polar and non-polar plasma lipids were conducted using an LC-MS system comprising a Shimadzu Nexera X2 U-HPLC (Shimadzu Corp.) coupled to an Exactive Plus orbitrap mass spectrometer (Thermo Fisher Scientific). Media samples (10 μL) were extracted for lipid analyses using 190 μL of isopropanol containing 1,2-didodecanoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids) as an internal standard. Cell pellets (~30ul) were extracted for lipid analysis using 570ul isopropanol containing 1,2-didodecanoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids) as an internal standard. The cell pellets were homogenized using the QIAGEN TissueLyser II with 3mm Tungsten beads for 4 min at a frequency of 20Hz. The media and cell pellets were centrifuged (10 min, 9,000xg at 4°C), and were injected directly onto a 100 × 2.1 mm, 1.7 μm ACQUITY BEH C8 column (Waters). The column was eluted isocratically with 80% mobile phase A (95:5:0.1 vol/vol/vol 10mM ammonium acetate/methanol/formic acid) for 1 minute followed by a linear gradient to 80% mobile-phase B (99.9:0.1 vol/vol methanol/formic acid) over 2 min, a linear gradient to 100% mobile phase B over 7 min, then 3 min at 100% mobile-phase B. MS analyses were carried out using electrospray ionization in the positive ion mode using full scan analysis over 200–1100 m/z at 70,000 resolution and 3 Hz data acquisition rate. Other MS settings were: sheath gas 50, in source CID 5 eV, sweep gas 5, spray voltage 3 kV, capillary temperature 300°C, S-lens RF 60, heater temperature 300°C, microscans 1, automatic gain control target 1e6, and maximum ion time 100 ms. Raw data were processed using TraceFinder software (Thermo Fisher Scientific) for targeted peak integration and manual review of a subset of identified lipids and using Progenesis QI (Nonlinear Dynamics) for peak detection and integration of both lipids of known identity and unknowns. Lipid identities were determined based on comparison to reference plasma extracts and are denoted by total number of carbons in the lipid acyl chain(s) and total number of double bonds in the lipid acyl chain(s).

HILIC-neg:

Targeted negative ion mode MS detection was used to measure central metabolites. HILIC (hydrophilic interaction chromatography) analyses of water soluble metabolites in the negative ionization mode (HILIC-neg) were conducted using an LC-MS system comprised of an AQUITY UPLC system (Waters) and a 5500 QTRAP mass spectrometer (SCIEX). Media samples (30 μL) were extracted with the addition of four volumes of 80% methanol containing inosine - 15N4, thymine-d4 and glycocholate-d4 internal standards (Cambridge Isotope Laboratories). Cell pellets (~30ul) were extracted with the addition of four volumes of 80% methanol containing inosine-15N4, thymine-d4 and glycocholate-d4 internal standards (Cambridge Isotope Laboratories. The cell pellets were homogenized using the QIAGEN TissueLyser II with 3mm Tungsten beads for 4 min at a frequency of 20 Hz. The samples were centrifuged (10 min, 9,000xg at 4°C), and the supernatants were injected directly onto a 150 × 2.0 mm Luna NH2 column (Phenomenex). The column was eluted at a flow rate of 400 μL/min with initial conditions of 10% mobile phase A (20 mM ammonium acetate and 20 mM ammonium hydroxide in water) and 90% mobile phase B (10 mM ammonium hydroxide in 75:25 v/v acetonitrile/methanol) followed by a 10 min linear gradient to 100% mobile phase A. MS analyses were carried out using electrospray ionization and selective multiple reaction monitoring scans in the negative ion mode. To create the method, declustering potentials and collision energies were optimized for each metabolite by infusion of reference standards. The ion spray voltage was −4.5 kV and the source temperature was 500°C. Raw data from the 5500 QTRAP MS system were processed using MultiQuant 2.1 software (SCIEX).

Nucleic acid extraction for 16s rRNA gene sequencing

For cervicovaginal swab samples, total nucleic acids were extracted from the swab samples using the phenol-chloroform method, which includes a bead beating process to disrupt bacteria, as previously described^13,102. Briefly, swabs were thawed on ice, transferred into a solution consisting of phenol:chloroform:isoamyl alcohol (PCI, 25:24:1, pH 7.9, Ambion) and 20% sodium dodecyl sulfate in Tris-EDTA buffer with sterile 0.1 mm glass beads (BioSpec Products #11079101), vigorously rubbed against the walls of the tube to dislodge microbial material, and then incubated on ice for 5–10 min. Swabs were then removed by pressing the swab against the side of the tube using a sterile pipette tip as being lifted out to squeeze out excess fluid. Samples were homogenized using a bead beater for 2 min at 4°C, and then centrifuged at 6,800×g for 3 min at 4°C. The aqueous phase was transferred to a clean tube with equal volume of PCI solution, vortexed, and centrifuged again at 16,000×g for 5 min at 4°C. The aqueous phase was transferred to a second clean tube, precipitated using 0.8 volume of −20°C isopropanol with 0.08 volume (relative to initial sample) 3 M sodium acetate at pH 5.5, inverted to mix, and incubated overnight at −20°C. Samples were then centrifuged for 30 min at 21,100×g at 4°C, washed in 0.5 ml 100% ethanol and centrifuged for 15 min at 21,100×g at 4°C. The ethanol supernatant was discarded while keeping the pellet, which was allowed to air-dry and then resuspended in 20μl molecular-grade Tris-EDTA buffer. Genomic DNA from mock communities cultured in vitro was extracted using a plate-based adaptation of the above protocol including a bead beating process combined with phenol–chloroform isolation with QIAamp 96 DNA QIAcube HT kit (Qiagen Beverly LLC #51331) protocols⁴³.

16s rRNA gene sequencing for vaginal swab and mock community samples

Bacterial microbiota composition from cervicovaginal swabs collected from FRESH study participants and compositions of defined bacterial mock community experiments were determined using Illumina-based amplicon sequencing of the V4 region of the bacterial 16S rRNA gene. Standard PCR-amplification protocols were used to amplify the V4 region of the bacterial 16S rRNA gene^13,102,103. Briefly, samples were amplified using 0.5 units of Q5 high-fidelity DNA polymerase (NEB #M0491S) in 25 μl reaction with 1X Q5 reaction buffer, 0.2 mM deoxyribonucleotide triphosphate mix, 200 pM 515F primer (5’-AATGATACGGCGACCACCGAGACGTACGTACGGTGTGCCAGCMGCCGCGGTAA-3’, the underlined sequence representing the complementary region to the bacterial 16S rRNA gene; IDT) and 200 pM barcoded 806R primer (5’-CAAGCAGAAGACGGCATACGAGATXXXXXXXXXXXXAGTCAGTCAGCCGGACTACHVGGGTWTCTAAT-3’, the underlined sequence representing the complementary region to the bacterial 16S rRNA gene and the X characters representing the barcode position; IDT) in PCR-clean water. For each prepared barcode master mix, a water-template negative control reaction was performed in parallel. Blank extraction and amplification controls were additionally performed using unique barcoded primers in sequencing libraries. For FGT microbiota profiling from cervicovaginal swab samples, DNA was amplified in triplicate reactions and then triplicates were combined before library pooling to minimize stochastic amplification biases. For defined bacterial mock community experiments, DNA was amplified in a single reaction per replicate culture. Amplification was performed at 98°C for 30 s, followed by 30 cycles of 98°C for 10 s, 60°C for 30 s, and 72°C for 20 s, with a final 2 min extension at 72°C.

PCR products from all samples and the matched water-template control were visualized via agarose gel electrophoresis to confirm successful target amplification and absence of background amplification. Gel band strength was used to semi-quantitatively estimate relative amplicon concentrations for library pooling. To prepare the sequencing libraries, 3–20 μl of individual PCR products (adjusted on the basis of estimated relative amplicon concentration) were combined into 100 μl subpools and purified using an UltraClean 96 PCR cleanup kit (Qiagen Beverly LLC #12596-4). Despite not producing visible PCR bands, blank extractions, water-template and (for in vitro experiments) blank media controls were included in the sequencing libraries for additional quality control verification. Concentrations of the subpools were quantified using a Nanodrop 2000 and then pooled at equal molar concentrations to assemble the final library. Following standard Illumina protocols, the pooled library was diluted and supplemented with 10% PhiX, and then single-end sequenced on an Illumina MiSeq using a v2 300-cycle sequencing kit with addition of custom Earth Microbiome Project sequencing primers (read 1 sequencing primer: 5’-ACGTACGTACGGTGTGCCAGCMGCCGCGGTAA-3’; read 2 sequencing primer: 5’-ACGTACGTACCCGGACTACHVGGGTWTCTAAT-3’; index sequencing primer: 5’-ATTAGAWACCCBDGTAGTCCGGCTGACTGACT-3’; IDT)¹⁰³.

Analysis of 16S rRNA gene sequencing results

QIIME I v1.9.188¹⁰⁴ was used to demultiplex Illumina MiSeq bacterial 16S rRNA gene sequence data. QIIME 1-formatted mapping files were used and validated using validate_mapping_file.py, sequences were demultiplexed using split_libraries_fastq.py with parameter store_demultiplexed_fastq with no quality filtering or trimming, and demultiplexed sequences were organized into individual fastq files using split_sequence_file_on_sample_ids.py¹⁰². Dada2 v1.6.089¹⁰⁵ in R was used to filter and trim reads at positions 10 (left) and 230 (right) using the filterAndTrim function with parameters truncQ=11, MaxEE=2 and MaxN=0. Then, sequences were inferred and initial taxonomy assigned using the dada2 assignTaxonomy function, employing the Ribosomal Database Project training database rdp_train_set_16.fa.gz (https://www.mothur.org/wiki/RDP_reference_files). Taxonomic assignments were refined and extended via manual review (see Table S8 for amplicon sequence variant (ASV) taxonomy). Phlyoseq v1.30.090¹⁰⁶ in R was used to analyze and process the denoised dada2 results with final taxonomic assignment and custom R scripts. Final analysis and visualization of results were performed in python using jupyter notebooks. Initial sequencing-based analysis of FGT microbiota composition from FRESH cohort swab samples, taxonomy assignment and cervicotype assignments was performed blinded to participants’ corresponding Nugent scores.

For 16S rRNA gene-based microbiome profiling of clinical samples, microbial communities were classified into 4 cervicotypes (CTs) as previously defined¹³ in a non-overlapping subset of participants from the FRESH cohort: CT1 includes communities with >50% relative abundance of non-iners Lactobacillus species (which consists almost entirely of L. crispatus in this population); CT2 consists of communities in which L. iners is the most dominant taxon; CT3 consists of communities in which the genus Gardnerella is the most dominant taxon; and CT4 consists of communities dominated by other species, typically featuring high abundance of one or more Prevotella species. ASVs that could not be defined to the level of taxonomic class were pruned from the dataset. Taxonomically defined ASVs were collapsed at the species or genus level as indicated for further visualization and statistical analyses.

For 16S rRNA gene sequence analysis for bacterial mock community experiments, sequences were generated, processed, and annotated as described above. For each community replicate, relative abundances of each experimental strain were determined and the ratios of non-iners FGT Lactobacillus species read counts to the sum of the read counts of all of the other experimental strains were determined to quantify non-iners FGT Lactobacillus species enrichment for each condition. Significance of between-group differences for each mixture was determined by one-way ANOVA, and significance of pairwise comparisons was calculated using Tukey’s test, with statistical values for all pairwise conditions reported in Table S4.

Genomic analyses of FGT Lactobacillus and other species

All genomes of experimentally tested strains were obtained from RefSeq or Genbank (see Table S5 for accession numbers). We utilized the extensive genome collection of FGT Lactobacillus species reported by Bloom et al. (including isolate genomes and metagenome-assembled genomes)⁴³ and the type strain genome sequences of species in the family Lactobacillaceae reported in Zheng et al.⁵⁵ to identify presence/absence profiles of genes of interest across FGT Lactobacillus and Lactobacillaceae species. All genomes were downloaded from NCBI RefSeq or GenBank. Gene prediction for all genomes was performed using Prodigal, and gene functions were predicted using eggNOG 5.0⁸⁵ employing eggNOG-mapper v2.1.9⁸⁶. Custom Python scripts were used to parse the eggNOG outputs to identify the presence of genes of interest in each genome. We constructed a gene presence and absence map for all genes of interest across all genomes. For each gene of interest, MUSCLE v5.1⁸⁷ was used for multiple sequence alignment of representative orthologs, ModelTest-NG¹⁰⁷ was used to select the optimal substitution model, and RAxML-NG⁸⁸ used for tree construction employed via raxmlGUI 2.0¹⁰⁸ to map their phylogenetic relationships. For the construction of species phylogeny, core ribosomal genes present in all Lactobacillaceae genomes were aligned using MUSCLE v5.1⁸⁷ and FastTree v2.1¹⁰⁹ was used to construct the Lactobacillaceae species tree. Trees were constructed using ortholog sequences from each species that were representative of the majority of sequences found in their respective orthologous groups to ensure robustness of the phylogenetic reconstruction. Tree and corresponding metadata visualization was done using Interactive Tree Of Life (iTOL) v5¹¹⁰.

Quantification and statistical analyses

For bacterial growth assays, figures depict representative results from 1 of ≥2 independent experiments prepared with distinct batches of media and bacterial input inocula. Growth data collection and analysis were not performed blind to the conditions of the experiments. For bulk RNA-sequencing samples, all conditions were performed in triplicate repeats with each inocula coming from a unique bacterial colony. For mass spectrometry experiments, all conditions were performed in duplicate or triplicate technical repeats. For mock community assays, all conditions were performed with six technical repeats.

Data analysis, statistics and visualization were performed using custom scripts written in python v3.9 using Jupyter Notebook v6.5.2 or R v.3.6.3. R packages used for analyses and plotting include seqinr v.4.2.5, tidyverse, v.1.3.1, knitr v.1.33, ggpubr v.0.4.0, DescTools v.0.99.41, gtools v.3.8.2, gridExtra v.2.3, cowplot v.1.1.1, scales v.1.1.1, grid v.3.6.3, broom v.0.7.6, e1071 v.1.7.6, and table1 v.1.4. Python packages used for analyses and plotting include biopython v1.79, matplotlib v3.7.1, numpy v1.22.3, pandas v1.5.1, scikit-bio v0.5.8, scipy v1.9.3, seaborn v0.11.2, statannot v0.2.3, and statsmodels v0.13.2. All P values are two-sided with statistical significance defined at α=0.05, unless otherwise indicated.

Supplementary Material

Supplement 1

Reagent Table

Table S1. FGT bacterial isolates.

Table S2. Isolate genomes and MAGs.

Table S3. Human specimens and associated targeted lipidomic and microbiome data.

Table S4. Statistical values for all figures.

Table S5. Vectors.

Table S6. Primers.

Table S7. Genetic mutant strains.

Table S8. ASV taxonomy.

Table S9. Relative CFU input for competition assays.