Cysteine dependence of Lactobacillus iners is a potential therapeutic target for vaginal microbiota modulation

Abstract

Vaginal microbiota composition affects many facets of reproductive health. Lactobacillus iners-dominant microbial communities are associated with poorer outcomes, including higher risk of bacterial vaginosis (BV), compared with vaginal microbiota rich in Lactobacillus crispatus. Unfortunately, standard-of-care metronidazole therapy for BV typically results in dominance of L. iners, likely contributing to post-treatment relapse. Here we generate an L. iners isolate collection comprising 34 previously unreported isolates from 14 South African with and without BV and 4 previously unreported isolates from 3 US women and we report an associated genome catalog comprising 1,218 vaginal Lactobacillus isolate genomes and metagenome-assembled genomes (MAGs) from >300 women across four continents. We show that, unlike L. crispatus, L. iners growth is dependent on L-cysteine in vitro and we trace this phenotype to the absence of canonical cysteine biosynthesis pathways and a restricted repertoire of cysteine-related transport mechanisms. We further show cysteine concentrations in cervicovaginal lavage samples correlate with Lactobacillus abundance in vivo and that cystine uptake inhibitors selectively inhibit L. iners growth in vitro. Combining an inhibitor with metronidazole promotes L. crispatus dominance of defined BV-like communities in vitro by suppressing L. iners growth. Our findings enable a better understanding of L. iners biology and suggest candidate treatments to modulate the vaginal microbiota to improve reproductive health for women globally.

Female genital tract (FGT) microbiota composition is linked to important women’s health and reproductive outcomes including HIV risk^1,2, preterm birth³, mucosal inflammation^4–6, human papilloma virus (HPV) infection⁷, and cervical dysplasia⁷. Diverse, anaerobe-dominant bacterial communities are associated with negative sequelae and are a hallmark of bacterial vaginosis (BV)^8,9, a syndrome involving vaginal discharge and mucosal inflammation that affects up to 58% of women worldwide, with disproportionately high impact in sub-Saharan Africa¹⁰. By contrast, FGT bacterial communities dominated by Lactobacillus crispatus and most other Lactobacillus species are generally considered optimal for health⁹. One notable exception involves communities dominated by Lactobacillus iners, the most prevalent and abundant FGT species worldwide¹¹, which are associated with many of the same unfavorable outcomes as BV^1,7,12 and have higher risk of transitioning to BV and BV-like communities^3,13–17. However factors influencing the balance between L. iners and more health-associated lactobacilli are incompletely understood^16,18.

The close associations between vaginal microbiota and disease make microbiota-targeted therapies a key strategy to improve women’s health and reproductive outcomes⁸. Microbiota modulation has primarily been studied for BV, where current treatments exhibit low efficacy and high rates of recurrence. Over 50% of women with symptomatic BV who receive standard antibiotic therapy with metronidazole (MTZ) experience recurrence by 12 months, while up to 80% of women with prior history of recurrent BV relapse within 16 weeks^19,20. MTZ typically shifts BV-associated microbiota towards L. iners-dominant communities^17,21–24, which frequently transition back to BV-like communities^13,15,16. Novel strategies to promote L. crispatus-dominance over L. iners-dominance during BV treatment are thus a key priority^16,25.

Despite its ubiquity in human populations, L. iners’ biology remains incompletely characterized¹⁸. Several features distinguish it from other lactobacilli including a putative virulence factor (inerolysin)²⁶, lack of D-lactic acid and hydrogen peroxide production^27–29, and a smaller genome size with reduced metabolic potential, suggesting niche adaptation and dependence on exogenous nutrients²⁸. However, L. iners research has been significantly hampered by its species-defining failure to grow in standard Lactobacillus MRS (de Man, Rogosa, Sharpe) media³⁰, complicating attempts at isolation and in vitro characterization^11,18,31. Consequently, few experimental strains and genomes exist, with many studies reporting little or no success isolating L. iners even when culturing from hundreds of clinical samples^32–34.

Here we show that L. iners’ unique in vitro growth limitation is due to a restricted capacity to exploit exogenous cysteine (Cys) sources. Through experiments and analysis of a collection of >1200 diverse Lactobacillus genomes, we demonstrate that major vaginal Lactobacillus species lack canonical Cys biosynthesis pathways. We find that vaginal Cys concentrations are higher in women without BV and that in vivo abundances of both L. iners and L. crispatus correlate with Cys availability. However, L. iners lacks transport mechanisms for Cys and Cys-containing molecules possessed by other Lactobacillus species, rendering its growth uniquely susceptible to inhibition by cystine uptake inhibitors. Combining an inhibitor with MTZ promotes L. crispatus dominance of defined BV-like bacterial communities by suppressing L. iners, identifying a previously undescribed target for therapeutically modulating vaginal microbiota to promote women’s health.

RESULTS

L-cysteine supports growth of diverse L. iners strains

We investigated the nutritional dependencies underlying L. iners’ failure to grow in Lactobacillus MRS (De Man, Rogosa, and Sharpe) media³¹ by testing various nutrient additives. MRS broth supplemented with IsoVitaleX^™ (a defined micronutrient mixture) supported robust L. iners growth while retaining selectivity against G. vaginalis (Fig. 1a). Testing IsoVitaleX^™ components individually revealed that L-cysteine (L-Cys) was necessary and sufficient for this growth phenotype (Extended Data Fig. 1a-f; reagent details in Supplementary Table 1 and 2). L-glutamine (L-Gln) was neither necessary nor sufficient, but augmented growth in presence of L-Cys, so subsequent experiments employed a base of L-Gln-supplemented MRS broth (“MRSQ”).

Fig. 1. Lactobacillus iners requires L-cysteine supplementation to grow in conditions that support other vaginal lactobacilli.

a, Growth of Lactobacillus crispatus, L. iners (strain F1), and Gardnerella vaginalis in Lactobacillus MRS broth ± supplementation with 2% IsoVitaleX^™. b, Unrooted phylogenetic tree of 198 L. iners genomes from previously unreported and reference bacterial isolates and previously undescribed culture-independent metagenome-assembled genomes (MAGs) derived from geographically diverse populations. Red dots indicate isolates experimentally studied in this work. Genome type, geographic origin, and BV status of the source samples are shown. The tree depicts only genome assemblies exceeding certain quality thresholds to ensure robustness of the phylogenetic reconstruction (see Methods); additional strains and genomes were included in other analyses (Supplementary Tables 3-11 and Extended Data Fig. 2). c, Growth of 8 representative US (prefix “A”) or South African (prefix “F”) L. iners strains cultured in MRS broth supplemented as indicated with IsoVitaleX^™ (2%), L-cysteine (L-Cys, 4 mM), and/or L-glutamine (L-Gln, 1.1 mM). In 1a,c growth was assessed by optical density at 600 nm (OD600) and plotted as median (± range) for 3 replicates from 1 of ≥2 independent experiments per strain and media condition.

We next assessed species-level generalizability of Cys-dependent growth by establishing a large, geographically diverse L. iners isolate collection and genome catalog of strains from women both with and without BV. The isolate collection included 4 previously reported US strains from culture repositories, plus 34 previously unreported isolates from 14 South African women and 4 previously unreported isolates from 3 US women, isolated using L-Cys-supplemented MRS or blood agar. The associated genome catalog comprises 327 genomes derived from studies of disparate human populations in the US, South Africa, Sweden, China, and Italy, including 21 “reference” genomes of previously reported isolates retrieved from RefSeq³⁵, 49 previously unreported isolate genomes, and 257 previously unreported culture-independent metagenome-assembled genomes (MAGs) generated from a combination of both previously published and previously unreported FGT shotgun metagenomic sequencing studies (Hayward et al, manuscript in preparation; Supplementary Table 3 lists isolates tested experimentally in this study; see also additional isolate and genome source information in the Methods, Extended Data Figure 2, and Supplementary Tables 4-11). Species-level phylogenetic reconstruction revealed that L. iners genomic diversity was largely independent of geographic source, genome type, or clinical context (Fig. 1b). Strains broadly representative of species diversity isolated from South African and US women both with and without BV were selected for further experimental characterization (Fig. 1b and Supplementary Table 3). All selected strains grew in MRSQ broth with L-Cys (representative examples in Fig. 1c), confirming Cys-dependent growth as a species-level phenotype.

FGT Lactobacilli lack canonical Cys biosynthesis pathways

We investigated whether L. iners’ in vitro Cys-dependent growth phenotype reflected differences in Cys biosynthetic capacity relative to other common FGT lactobacilli. Bacteria canonically synthesize Cys de novo from serine (Ser) via either the Cys synthase pathway (cysE, followed by cysK, cysM, or cysO) or reverse transsulfuration pathway (cbs or mccA, followed by mccB; Fig. 2a)³⁶. We assessed presence of these pathways in a previously unreported catalog of 1,218 FGT Lactobacillus genomes and MAGs, including 327 L. iners genomes, along with similarly constructed, geographically diverse catalogs for L. crispatus (n = 310 genomes and 198 MAGs), L. gasseri (n = 174 genomes and 42 MAGs), L. jensenii (n = 80 genomes and 57 MAGs), and L. vaginalis (n = 28 genomes and 2 MAGs; see Extended Data Fig. 2 and Supplementary Tables 4-11; Hayward et al, manuscript in preparation). These catalogs represent substantial expansion of known genomic diversity for all species, most notably for L. iners with a >10-fold increase in the number of genomes and a 73% increase in pan-genome size compared to RefSeq-derived genomes alone (Supplementary Tables 4). No genomes from any species were predicted to encode intact Cys synthase or reverse transsulfuration pathways, although interestingly all non-iners species had predicted mccB orthologs (Fig. 2b and Supplementary Table 12). Thus, all common FGT Lactobacillus species, including L. iners, are predicted to lack canonical mechanisms for de novo Cys biosynthesis.

Fig. 2. Vaginal lactobacilli lack canonical Cys biosynthesis pathways.

a, Canonical bacterial Cys biosynthesis pathways. Cysteine synthase pathway: serine O-acetyltransferase (encoded by cysE), then cysteine synthase (cysK, cysM, or cysO). Reverse transsulfuration pathway: cystathionine β-synthase (cbs) or O-acetylserine-dependent cystathionine β-synthase (mccA), then cystathionine γ-lyase (mccB). Asterisks indicate position of ¹³C label in isotopic tracing experiments. b, Predicted presence of genes encoding cysteine biosynthesis enzymes within the cysteine synthase or reverse transsulfuration pathways in isolate genomes and MAGs of the indicated FGT Lactobacillus species (n = number of genomes; detailed results in Supplementary Table 12). Assessment of gene presence was based on annotations from eggNOG 5.0 performed using eggNOG-mapper v2^82,83. Results were supported by BLAST⁸⁴ searches of the genomes for sequences of interest⁸¹. c, Serine isotopic enrichment in L. crispatus and L. iners grown for 24 hours in L-Gln-supplemented MRS broth (“MRSQ”) containing no isotopic tracer (supplemented with unlabeled L-Cys 4 mM) or supplemented with 1-¹³C-L-serine (¹³C-L-Ser) + homocysteine (both 4 mM). Plot depicts fractional abundance of unlabeled (m+0, Ser) and labeled (m+1, ¹³C-Ser) isotopologues in cellular hydrolysates. d, Cystine isotopic enrichment in lysates of L. crispatus or L. iners cultured as shown in (c) or in broth supplemented with 1-¹³C-L-cysteine (¹³C-L-Cys; 4 mM). An oxidation step during sample processing converts Cys to cystine for quantification. Plot depicts fractional abundance of cystine isotopologues: unlabeled (m+0; cystine), partially labeled (m+1; ¹³C₁-L-cystine), and fully labeled (m+2; ¹³C₂-L-cystine) isotopologues. Data in 2c,d depict median values of 3 replicates per group and were un-replicated (full data in Supplementary Tables 13 and 14).

To phenotypically test this result, we used isotopic tracing to measure the ability of L. iners and L. crispatus to convert Ser into Cys in vitro. Representative strains were grown in MRSQ broth supplemented with labeled Ser (¹³C-L-Ser) plus homocysteine or with labeled L-Cys (¹³C-L-Cys; labeling scheme in Fig. 2a), then cellular amino acid isotopic labeling was analyzed via liquid chromatography-mass spectrometry (LC-MS). Both L. iners and L. crispatus took up labeled Ser but failed to convert it to Cys (Fig. 2c,d and Supplementary Table 13 and 14), consistent with mechanistic predictions from genomic analysis of the species genome collections (Fig. 2a,b and Supplementary Table 12). By contrast, both strains readily incorporated labeled Cys, supporting the genomic prediction that, despite their different in vitro growth phenotypes, both utilize exogenous Cys.

Cys levels correlate with Lactobacillus abundance in vivo

Since vaginal lactobacilli lack canonical Cys biosynthesis pathways and utilize exogenous Cys, we investigated in vivo relationships between vaginal Cys concentrations, BV, and Lactobacillus colonization levels. We measured Cys concentrations in cervicovaginal lavage (CVL) fluid from 142 participants in a cohort of South African women aged 18–23 years^1,4,37, 53 of whom were also evaluated for BV by Nugent scoring³⁸. Cys concentrations differed strikingly by BV status (P = 6.3 × 10⁻⁹), with significantly higher concentrations among women without BV compared to those with BV (Fig. 3a). We explored FGT microbiota composition in this cohort by bacterial 16S rRNA gene sequencing, classifying, individual bacterial communities into four “cervicotypes” (CTs) based on previously established criteria¹: CT1 (L. crispatus-dominant), CT2 (L. iners-dominant), CT3 (Gardnerella-dominant), and CT4 (non-Lactobacillus-dominant, diverse communities typically featuring high Prevotella abundance) (Fig. 3b; taxonomy assignments in Supplementary Table 15). As expected, CT assignment was strongly related to BV status (P = 1.902 × 10⁻¹¹, two-sided Fisher’s Exact Test; Extended Data Fig. 3a). Cys concentration differed markedly between CTs (P = 2.7 × 10⁻¹⁴ overall), with significantly higher concentrations in Lactobacillus-dominant CTs compared to CT3 and CT4 (Fig. 3c).

Fig. 3. Vaginal Cys concentrations are higher in women without BV and correlate with Lactobacillus-dominant microbiota.

a, Relative Cys concentration by BV status in cervicovaginal lavage (CVL) fluid from 53 South African women (21 without BV, 24 with BV, and 8 intermediate by Nugent method³⁸). Differences in log-transformed concentrations determined by one-way analysis of variance (ANOVA) with post-hoc Tukey’s test; all significant pairwise differences at confidence level 0.95 are shown (No BV-BV: P < 1 × 10⁻⁷; Intermediate-BV: P = 0.0060). b, FGT bacterial microbiota composition among 142 HIV-uninfected South African women (including the 53 from a), determined by bacterial 16S rRNA gene sequencing (taxonomy assignments in Supplementary Table 15). Bacterial communities were classified into “cervicotypes” (CTs) using previously defined criteria¹. c, Relative Cys concentrations per CT in CVL fluid from women in (b). Significance determined by one-way ANOVA with post-hoc Tukey’s test; all significant pairwise differences are shown. CT-CT3: P = 0.000120; CT1-CT4: P < 1 × 10⁻⁷; CT2-CT3: P = 0,00898; CT2-CT4: P < 1 × 10⁻⁷; CT-CT4: P = 0.0279. In a,c, box center lines, edges, and whiskers signify median, interquartile range (IQR), minima and maxima, respectively. d, Two-tailed Spearman rank correlation between Cys concentrations and bacterial relative abundances of the genera Lactobacillus, Gardnerella, and Prevotella, showing correlation coefficients (ρ) with unadjusted p-values. Linear regression lines (solid blue) with 95% confidence intervals calculated from log-transformed abundances and concentrations are shown to assist visualization (Lactobacillus: y = 0.37 + 0.32x; Gardnerella: y = −0.25 – 0.15x; Prevotella: y = −0.47 – 0.27x). Red dotted line represents limit of detection (L.D.). e,f Two-tailed Spearman correlation coefficients (ρ) with adjusted confidence intervals between Cys concentrations and relative abundances of each genus (e) or species (f) detected at >50% prevalence in cohort (n = 142). P-values and confidence intervals in e,f were adjusted for multiple comparisons using the Bonferroni method at significance level 0.05/n (full statistical results in Supplementary Tables 16 & 17). Significance of adjusted P-values depicted as * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001.

We next examined correlations between Cys concentration and abundance of individual taxa within the microbiota. Analysis of CT-associated genera revealed that Lactobacillus relative abundance strongly positively correlated with Cys concentration (ρ = 0.6, P = 5.2 × 10⁻¹⁵ by Spearman correlation), while Prevotella abundance had equally strong negative correlation (ρ = −0.57, P = 1.9 × 10⁻¹³) and Gardnerella abundance was weakly negatively correlated (Fig. 3d). L. crispatus and L. iners each positively correlated with Cys when analyzed individually (Extended Data Fig. 3b). To more comprehensively identify taxa associated with Cys, we calculated correlations for each taxon detected in ≥50% of samples at the genus or species level (Extended Data Fig. 3c,d), adjusting for multiple comparisons. Lactobacillus (genus-level) and L. crispatus and L. iners (species-level) were the only taxa positively correlated with Cys (Fig. 3e,f; full statistical results in Supplementary Tables 16 & 17). Other taxa, including various BV-associated bacteria, showed no correlation or significant negative correlation with Cys. Results were similar for the Cys-containing peptides reduced glutathione (GSH) and cysteinylglycine (Extended Data Fig. 4 and Supplementary Tables 16 & 17). Thus, vaginal Cys concentration is higher in women without BV and correlates with abundance of both L. iners and L. crispatus, consistent with the hypothesis that Cys availability is important for Lactobacillus colonization success in vivo.

L. iners has limited capacity to use complex L-Cys sources

The species-specific nature of L. iners’ Cys-dependence in vitro appeared discordant with the finding that all FGT Lactobacillus species lacked canonical Cys biosynthesis pathways and that in vivo abundance of both L. iners and L. crispatus correlated with vaginal Cys concentrations. We therefore hypothesized that L. iners’ unique in vitro growth phenotype was due to a restricted capacity to utilize exogenous Cys sources in media. Alternatively, Cys might support L. iners growth by acting as a chemical reducing agent rather than a direct nutritional supplement. We found that concentrations of free molecular Cys and cystine (its oxidized counterpart) in MRSQ broth were ≤4.6 μM and 0.7 μM respectively, below levels required by Cys-auxotrophic E. coli strains³⁹. L-cystine supplementation produced similar L. iners growth as L-Cys supplementation (Fig. 4a), including in presence of the oxidizing agent hydrogen peroxide (H₂O₂; Extended Data Fig. 5a), demonstrating that L-cystine supports growth by acting as a direct nutritional supplement rather than by chemically reducing the media. Thus L. iners does not require a reduced environment to grow when it has access to a bioavailable source of L-Cys (e.g., L-cystine), but L-Cys bioavailability in un-supplemented MRSQ broth is inadequate for L. iners.

Fig. 4. L. iners lacks Cys-related transport mechanisms present in other lactobacilli and is selectively inhibited by cystine uptake inhibitors.

a, Growth of L. crispatus, G. vaginalis, and representative L. iners strains in MRSQ broth ± L-Cys (4 mM) or L-cystine (2 mM). b, Predicted presence of putative Cys transport-related gene cjaA, cystine transport loci tcyABC, tcyJKLMN, and tcyP, and the Cys/GSH transport/redox homeostasis locus cydABCD locus in FGT Lactobacillus species genomes and MAGs (n = number of genomes; detailed results in Supplementary Table 12). TcyBC takes up glutathione in some species. c, Cystine isotopic enrichment in L. crispatus or L. iners grown in MRSQ broth containing labeled L-cystine (¹³C₂-L-cystine; 2 mM) or GSH (¹³C-GSH; 4 mM; synthesis described in Supplementary Fig. 1). Plot depicts isotopologue median fractional abundance and was un-replicated. (Full data in Supplementary Tables 13 & 14). d, Growth of representative Lactobacillus strains in L-Cys-supplemented MRSQ broth containing the cystine uptake inhibitor S-methyl-L-cysteine (SMC). Percentages calculated relative to median OD600 measurement in L-Cys-containing no-inhibitor control. e, Median growth inhibition of non-iners Lactobacillus strains (L. crispatus, n = 7; L. jensenii, n = 2; L. gasseri, n = 1) or L. iners (n = 16) by SMC in L-Cys-supplemented MRSQ broth. Significance of differences between L. iners and non-iners strains was determined by two-sided t-test. Box center lines, edges, and whiskers signify the median, IQR, minima and maxima respectively. f, Growth inhibition of L. crispatus and L. iners by SMC (128 mM) or SDLC (2 mM) in NYCIII broth. Plots in a,d,f depict median (± range) for 3 replicates per condition and are representative of ≥2 independent experiments per condition.

The ability of non-iners lactobacilli to grow in un-supplemented MRS broth implied presence of Cys sources that L. iners failed to utilize (MRS contains yeast and beef extracts, which contribute peptides in addition to monomeric amino acids^30,40). Adding chemical reducing agents to MRSQ broth substantially raised Cys and GSH concentrations (Extended Data Fig. 5b), demonstrating that most Cys and Cys-containing molecules in un-supplemented media exist as mixed disulfide compounds. Addition of chemical reducing agents also supported L. iners growth (Extended Data Fig. 5a,c), but in contrast to L-cystine, the oxidized counterparts of these reducing agents did not. Thus, MRS broth contains complex sources of L-Cys, including mixed disulfides, that L. iners has restricted capacity to utilize.

L. iners lacks transporters found in other FGT lactobacilli

We hypothesized that L. iners’s restricted ability to use exogenous Cys sources reflected a limited repertoire of transport mechanisms for Cys and Cys-containing molecules in comparison to other lactobacilli. Uptake of Cys and Cys-containing molecules in bacteria occurs through multiple known or putative mechanisms including the well-characterized cystine transport systems TcyABC, TcyJKLMN, and TcyP⁴¹, a putative Cys ABC transporter associated with Cys-binding protein CjaA^42,43, and the heterodimeric ABC transporter CydDC encoded by the redox-regulating locus cydABCD⁴⁴, a system that exports both Cys and GSH to the periplasm in E. coli⁴⁴ and is proposed to perform glutathione uptake in lactobacilli⁴⁵. Analysis of our Lactobacillus genome catalogs revealed that all Lactobacillus species except for L. iners encoded cydABCD, while L. crispatus, L. vaginalis, and some L. gasseri each encoded one or more predicted cjaA orthologs and L. jensenii, L. gasseri, and L. vaginalis each encoded one or more cystine transport systems (Fig. 4b and Supplementary Table 12). Low-affinity Cys uptake also occurs via the branched-chain amino acid importer LivJ/LivKHMGF⁴³, but no genomes possessed livKHMGF orthologs except 2 L. iners MAGs, nor did any species possess an intact GsiABCD glutathione transport system⁴⁶ (Extended Data Fig. 6a and Supplementary Table 12). Thus, each common FGT Lactobacillus species except L. iners encodes multiple, mechanistically distinct known or putative systems to transport Cys or Cys-containing molecules, while L. iners’ Cys-related uptake occurs via currently unrecognized mechanisms.

L. iners growth is inhibited by cystine uptake inhibitors

Experimental tools to perform genetic screens in L. iners do not currently exist⁴⁷, so we used phenotypic approaches to characterize its Cys-related transport mechanisms. Isotopic tracing (Fig. 4c) revealed that L. iners incorporated labeled L-cystine (¹³C₂-L-cystine) at levels similar to L. crispatus but failed to take up labeled GSH (¹³C-GSH; synthesis detailed in Supplementary Fig. 1), consistent with predicted absence of GSH transport activity (Fig. 4b and Extended Data Fig. 6a). We therefore assessed whether L. iners’ limited repertoire of transport mechanisms would cause its growth in L-Cys-supplemented MRSQ broth to be uniquely impacted by the known cystine uptake inhibitors S-methyl-L-cysteine (SMC) and seleno-DL-cystine (SDLC)⁴¹. SMC caused species-specific, dose-dependent growth inhibition of a diverse collection of L. iners strains at concentrations of 32–64 mM, while SDLC caused selective inhibition at 0.25 mM (Fig. 4d,e and Extended Data Fig. 6b,c). Inhibition occurred for strains from both US and South African women, and from women both with and without BV (Supplementary Table 3). Inhibitor potency varies for different transporters, but these concentrations appear higher for SMC and slightly higher for SDLC than reported inhibitory concentrations for the TcyJKLMN and TcyP transporters in Bacillus subtilis⁴¹. To confirm that growth inhibition was not an artefact specific to MRS media, we assessed inhibition in NYCIII broth, a serum-containing, nutrient-rich medium that supports growth of diverse fastidious FGT bacteria⁴⁸. Both SMC and SDLC inhibited L. iners growth in NYCIII broth, but not growth of L. crispatus (Fig. 4f). They also selectively inhibited L. iners growth in MRSQ broth treated with chemical reducing agents, supporting the conclusion that reducing agents promote L. iners growth by enhancing Cys bioavailability (Extended Data Fig. 6d). Collectively these results from isotopic tracing and growth inhibition assays confirm our genomic prediction that L. iners possesses a uniquely restricted repertoire of Cys-related transport mechanisms.

SMC inhibits L. iners in competition with L. crispatus

To assess the functional significance of L. iners growth inhibition by cystine uptake inhibitors, we assessed whether an inhibitor could shift the balance between L. iners and L. crispatus in culture competition assays. We tested pairwise strain combinations of L. crispatus and L. iners in L-Cys-supplemented MRSQ broth at varying SMC concentrations, then measured ratios of L. iners to L. crispatus in the mixed cultures by 16S rRNA gene sequencing. Experimental controls were included to confirm expected growth patterns and absence of significant contamination during sample processing (Extended Data Fig. 7). SMC significantly suppressed L. iners relative to L. crispatus in a dose-dependent fashion (representative experiments in Fig. 5a, Supplementary Table 18), confirming that it offers selective advantage to L. crispatus in direct competition.

Fig. 5. SMC inhibits L. iners in competition with L. crispatus and combining SMC with metronidazole enhances L. crispatus dominance of mock BV-like communities.

a, Ratios of L. iners to L. crispatus in representative mixed culture competition assays after 28 hours incubation in L-Cys-supplemented MRSQ broth with varying concentrations of SMC. Ratios were determined via sequencing the bacterial 16S rRNA gene in DNA isolated from the mixed cultures; plots depict results for 5 replicates per condition. Between-group differences were determined by one-way ANOVA. Pairwise comparisons to the reference condition (“Ref”: no-inhibitor, L-Cys-supplemented control) were calculated using Dunnett’s test. All significant comparisons are shown (full statistical results in Supplementary Table 18). Input ratios of colony-forming units (C.F.U.) of L. iners to L. crispatus were 8.01:1, 1.08:1, and 5.96:1 for the mixes containing L. iners strains A4, F1, and F3 respectively. b, Growth inhibition of L. crispatus, L. iners, and three BV-associated species at 48 hours incubation in S-broth with or without SMC and/or metronidazole (MTZ). Plots depict median (± range) for 3 replicates per condition and are representative of 2 independent experiments per condition. c, Relative abundance of bacterial species in three representative mock BV-like communities (each with a different L. iners strain) grown in S-broth with or without SMC and/or MTZ at 28 hours incubation. Composition was determined by bacterial 16S rRNA gene sequencing. Input ratios of strains for each community are shown in Supplementary Table 19. d, Ratios of L. crispatus to the sum of all other taxa in the mock communities shown in (c). (c,d) depict 6 replicates per condition. Between-group differences in (d) were determined via one-way ANOVA with post-hoc Tukey test; selected pairwise differences are shown (full statistical results in Supplementary Table 20). In 5a,e significance is depicted as * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001. (b,d) Box center lines, edges, and whiskers signify the median, IQR, minima and maxima respectively.

SMC plus MTZ enriches L. crispatus in mock BV communities

To assess whether cystine uptake inhibitors have potential to augment BV therapy by preferentially promoting L. crispatus expansion relative to L. iners, we investigated SMC’s effects in combination with MTZ on mock BV-like bacterial communities cultured in a rich nutritional milieu. Multiple South African L. crispatus isolates unexpectedly failed to grow in NYCIII broth, so we developed a previously unreported nutrient-rich, serum-containing formulation named “S-broth” that supported South African L. crispatus strains as well as diverse BV-associated bacteria (Extended Data Fig. 8a). Testing effects of SMC and MTZ on pure cultures of L. crispatus, L. iners, G. vaginalis, Prevotella bivia, and Atopobium (Fannyhessea) vaginae grown in S-broth revealed that neither MTZ nor SMC blocked L. crispatus growth, SMC inhibited L. iners, and MTZ suppressed both G. vaginalis and P. bivia (Fig. 5b). The A. vaginae strain was relatively MTZ-resistant, but susceptibility was enhanced by addition of SMC.

We next cultured defined communities of these species in S-broth and assessed whether combining SMC with MTZ enhanced L. crispatus dominance. Each community contained a different pairwise combination of L. crispatus and L. iners strains (input ratios as in Supplementary Table 19) and we quantified mock community composition by 16S rRNA gene sequencing (controls shown in Extended Data Fig. 8b,c). Cultivation without inhibitors produced diverse mixtures of all 5 species (representative examples in Fig. 5c). As hypothesized, SMC alone diminished L. iners relative abundance, MTZ alone suppressed BV-associated species, and MTZ combined with SMC preferentially favored L. crispatus. To quantify L. crispatus’ competitive advantage, we calculated the ratio of L. crispatus relative abundance to the summed abundances of all other taxa (Fig. 5d, Supplementary Table 20). Combining MTZ with SMC increased this ratio significantly more than MTZ alone, suggesting cystine uptake inhibitors could interact with MTZ to promote L. crispatus dominance during BV treatment.

DISCUSSION

L. iners has long been recognized as possessing diminished metabolic potential compared to other lactobacilli – indeed, the name “iners” refers to relative inertness in biochemical assays³¹. But despite being the most abundant FGT species worldwide and having adverse health associations compared to L. crispatus^{1,3,52–54,7,11,12,16,18,49–51}, L. iners’ biology remains poorly characterized because it fails to grow in Lactobacillus MRS media³¹, resulting in a paucity of isolates and genomes^11,18. Here we resolve this growth defect by identifying a species-specific requirement for Cys supplementation. Employing an isolate collection encompassing both reference and previously unreported strains and >1200 Lactobacillus genomes from diverse geographical and clinical sources, we extend and experimentally confirm predictions from smaller-scale genomic studies suggesting all common FGT lactobacilli lack canonical Cys biosynthesis pathways⁵⁵. We further show that vaginal Cys concentrations strongly correlate with Lactobacillus relative abundance in vivo in a South African cohort. However, L. iners lacks transport mechanisms for Cys and Cys-containing molecules present in other lactobacilli, rendering its growth selectively susceptible to cystine uptake inhibitors. Combining a cystine uptake inhibitor with MTZ (a standard antibiotic used for BV treatment) enhances L. crispatus dominance in mock, BV-like bacterial communities by suppressing L. iners growth. These advances establish a foundation for future study of L. iners’ biology, diversity, and effects on human health while identifying a previously undescribed target for therapeutic vaginal microbiota modulation.

The in vivo correlation between Cys concentrations and Lactobacillus abundance suggests Cys availability may influence FGT microbiota composition. A US-based case-control study comparing women with and without BV found that vaginal Cys concentrations positively correlated with non-iners lactobacilli and negatively correlated with various BV-associated species⁵⁶. That study appeared to show a weaker correlation between L. iners and Cys than the correlation we identify here⁵⁶, which may reflect differences in study design and/or species abundances between the two populations. Since FGT lactobacilli appear to lack de novo Cys biosynthetic capacity, we hypothesize that the high Cys concentrations in Lactobacillus-dominant states are likely primarily host-derived. Possible mechanisms explaining the correlation could thus include preferential Lactobacillus colonization of hosts with high levels of mucosal Cys secretion, induction of host Cys secretion by lactobacilli, Cys degradation by host cells in setting of BV-associated inflammation⁵⁷, and/or degradation of Cys by BV-associated bacteria⁵⁶ as a means to outcompete lactobacilli. Further investigation of FGT microbiome-metabolome relationships will be required to fully assess these possibilities.

A striking aspect of our findings is the paucity of transport mechanisms for Cys and Cys-containing molecules in L. iners compared to other FGT lactobacilli. In contrast to cystine transporters⁴¹, bacterial Cys transporters are not well characterized due to technical challenges inherent to redox chemistry⁴³. Our genome analysis predicted that non-iners FGT Lactobacillus species each encode multiple known or putative transporters for Cys, cystine, and/or Cys-containing molecules, whereas we identified no predicted transporters for these molecules in L. iners. However, both L-Cys and L-cystine supported L. iners growth, and isotopic tracing demonstrated uptake when L. iners was grown in the presence of labeled L-Cys or L-cystine, pointing to one or more currently unrecognized uptake mechanisms. The finding that cystine uptake inhibitors selectively inhibit L. iners growth provides further evidence that it lacks the multiple, mechanistically distinct Cys-related transporters present in other species.

In addition to providing mechanistic insights, our results have important implications for BV treatment. Standard BV therapy with MTZ often results in disease recurrence^19,20,25. This may be partially due to MTZ’s tendency to promote L. iners-dominant FGT bacterial communities^17,21–24, which to transition back to BV-like states more frequently than L. crispatus-dominant communities^13–17. We recently showed in a South African population that L. iners-dominant communities are a “gateway” for transitions from Lactobacillus dominance to more diverse, BV-like communities and our modeling of microbiota dynamics suggested that interventions shifting community transition probabilities towards L. crispatus over L. iners would most effectively increase the prevalence of L. crispatus-dominance within the population¹⁶. Therapies to promote L. crispatus dominance during BV treatment are under active investigation. In a trial of a live L. crispatus biotherapeutic administered after MTZ treatment, BV recurred less frequently among biotherapeutic recipients than placebo recipients²⁵. However recurrence in the treatment arm remained high (30% within 3 months)²⁵. A small Israeli cases series reported achieving L. crispatus-dominance and long-term BV remission in four of five women treated for refractory BV with MTZ plus vaginal microbiome transplants from healthy donors⁵⁸, but most patients required multiple rounds of therapy. Interestingly, that study did not report detecting L. iners in any donors or recipients – a notable difference from typical FGT microbiota composition in North and South America, Europe, Asia, Africa, and Australia^{1,2,23,24,48–54,56,3,59,60,4,5,12,13,17,21,22}. Thus, live biotherapeutic and microbiome transplant approaches for BV show promise but are unlikely to be a panacea. Our study shows that combining MTZ with a previously undescribed L. iners growth inhibitor was superior to MTZ alone in promoting L. crispatus community dominance of defined, BV-like bacterial communities. The specific inhibitors identified here have potential limitations for use in humans including low potency (SMC) and possible host toxicity (SDLC), and their effects (if any) on the FGT mucosa remain to be characterized. However, our results demonstrate proof of concept for a previously unrecognized therapeutic approach and provide a rationale to identify additional inhibitors targeting L. iners through this mechanism. Further development of this approach as an adjunct to antibiotics has potential to improve therapies for microbiota-linked reproductive health conditions worldwide, including for women in low- and middle-income countries where the burden of adverse reproductive health outcomes is greatest.

Methods

Overview of cohorts and sources of metagenomic data

This study includes previously unreported FGT microbiota sequencing data, FGT metabolite concentration data, bacterial isolates, isolate genomes, and/or MAGs assembled from whole-genome shotgun metagenomic data (WGS) from the Females Rising through Education, Support, and Health (FRESH) study, a South African cohort (details below). It also includes previously unreported bacterial isolates, isolate genomes, and/or MAGs assembled from WGS data from participants in 6 US-based cohorts, as well as previously unreported MAGs assembled de novo from published WGS data from 5 additional cohorts from the US, Italy, and China (details below).

FRESH cohort characteristics

The FRESH cohort is an ongoing prospective observational study based in Umlazi, South Africa, that enrolls 18–23-year-old, HIV-uninfected, non-pregnant, otherwise healthy women willing to have frequent HIV testing and 3-monthly collection of blood and mucosal specimens during study enrollment. Exclusion criteria included hemoglobin level <10 g/dL, any chronic medical condition or other conflict likely to prevent adherence to the study protocol, 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 also been described in detail elsewhere^1,4,37. The study protocol was approved by the Massachusetts General Hospital Institutional Review Board (2012P001812/MGH) and the Biomedical Research Ethics Committee of the University of KwaZulu-Natal (UKZN; Ethics Reference Number BF131/11). Participants were recruited at local sites frequented by young people including shopping malls, cafes, and nightclubs. All participants provided written informed consent. The study protocol includes non-monetary benefits encompassing job-skills, health, and empowerment training, a light meal during study site visits, and cumulative monetary compensation over a 36-week period of 3,700 rand (~$280 US dollars), which helps defray transportation expenses to twice-weekly study site visits. Participant self-selection bias may potentially occur based on participants’ perception of their own HIV risk, but it is unknown whether or how such a bias would affect FGT microbiota composition or metabolite concentrations.

FGT sample collection within the FRESH study included multiple cervical swab samples (Puritan 6” Sterile Standard Foam Swab w/Polystyrene Handle), which 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, were cryopreserved in thioglycolate broth with 20% glycerol for bacterial isolation or were 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 vaginal vault, wash the cervix four times, then re-aspirate the fluid 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 Nelson R. Mandela School of Medicine at UKZN, after which swabs were stored at −80°C and CVL samples were centrifuged at 700xg for 10 minutes at 4°C, then supernatants were transferred to cryovials and stored at −80°C.

FRESH cohort microbiota and metabolite profiling data

Paired bacterial microbiota composition and cervicovaginal metabolite concentrations (Figure 3 and Extended Data Figures 3 and 4) were determined for 143 HIV-uninfected FRESH study participants consecutively sampled in May through October of 2017 (methodological details below). The sample size was initially selected for microbiota profiling based on a separate, unpublished investigation of microbiome-immune parameters; paired CVL samples from these participants were later used for FGT metabolite concentration measurements due to availability of paired microbiota sequencing data. No a priori sample size calculation was performed in relation to analysis of FGT microbiota-cysteine correlations because it was (to our knowledge) the first investigation of this association in a sub-Saharan African cohort and most studies in non-African populations have used case-control designs that are not adapted to evaluate general population-level correlations – we therefore lacked information upon which to make informed estimates about metabolite distributions for power calculations. However the cohort was of similar or larger size as other reported studies of the FGT metabolome^56,61,62. One of the 143 participants was excluded from analysis of vaginal metabolite concentrations based on the pre-specified criterion of low bacterial 16S rRNA gene sequence read count (<10,000 reads). Assessment of BV status among this group of FRESH study participants using the Nugent scoring method³⁸ was performed sporadically prior to September 2017, when a protocol change instituted universal BV testing for all participants. We restricted analysis of BV-metabolite associations (Figure 3a and Extended Data Figures 4a,b) to the subset of 53 out of 143 participants assessed under the universal testing protocol to minimize risk of bias. Nugent scoring was performed at Global Laboratories (now Neuberg Global Laboratories, Durban, South Africa), an accredited commercial laboratory diagnostics company, by trained laboratory technologists blinded to all participant information other than study ID.

Sources of bacterial isolates and MAGs

This study reports previously undescribed bacterial isolates used in experiments (experimentally studied strains are listed in Supplementary Table 3, isolation details are described below) and analyzes a genome catalog including previously unreported Lactobacillus bacterial isolate genomes and MAGs (Figures 1b, 2b, 4b, and Extended Data Figures 2 and 6a; see also Supplementary Tables 4-11. The genome catalog was abstracted from a more extensive, unpublished, global characterization of FGT microbiome phylogenetic structure and function analyzing a combination of published bacterial isolate genome and WGS data, as well as previously unreported US and South African genomic and metagenomic data (manuscript in preparation by M.R. Hayward, S.M. Bloom, F.X. Ceasar, U. Das Adhikari, S.M.R. Demby, K.L. Dong, M. Dong, J.A. Elsherbini, M.T. France, M.S. Ghebremichael, T. Gumbi, F.A. Hussain, C. Huttenhower, N. Ismail, N.A. Mafunda, J.M. Marrazzo, C.M. Mitchell, T. Ndung’u, J. Ravel, D.A. Relman, J.K. Rice, J.M. Urbach, J. Xu, N. Xulu, and D.S. Kwon; additional methodological details are provided below). The previously undescribed datasets contributing to the analysis by Hayward et al include isolates and MAGs from 138 total FRESH study participants (6 of whom contributed both isolates and MAGs, 11 of whom contributed isolates only, and 121 of whom contributed MAGs only). This number included 35 participants from among the 143 participants listed above, as well as 103 additional FRESH study participants for whom microbiota composition based on bacterial 16S rRNA gene-based profiling has previously been described^1,4.

Additional previously unreported US-based primary isolates tested experimentally (Supplementary Table 3), as well as bacterial isolate genomes and WGS sequence data that served as the source of MAGs (Supplementary Tables 5, 6, 8, 9, and 11) were obtained via the Vaginal Microbiome Research Consortium (VMRC; https://vmrc4health.org/). The VMRC data and isolates were derived from two different IRB-approved protocols. The first of these IRB-approved protocols (Stanford University IRB protocol #21956; principal investigator D.A.R.) encompasses two previously described⁶³ cohorts that enrolled 135 pregnant women aged at least 18 years presenting for obstetrical care at either Lucille Packard Children’s Hospital at Stanford University or at the University of Alabama, Birmingham, who provided written informed consent. Exclusion criteria included antibiotic or probiotic use within twelve weeks prior to enrollment, significant immunosuppression, or (retrospectively) delivery of a baby who had congenital defects. Vaginal swab specimens were collected during pregnancy and (in the Stanford cohort) in the post-partum period as well. Cumulative monetary compensation totaled as much as $350 per participant. Isolates and MAGs obtained from participants enrolled via this protocol were derived from 18 total study participants, 11 of which contributed both isolates and MAGs, 5 of which contributed isolates only, and 2 of which contributed MAGs only. The second VMRC-associated cohort (University of Maryland Baltimore IRB protocol #HP-00041351; principal investigator J. Ravel), which has also been previously described²³, enrolled 135 non-pregnant, HIV-uninfected women at University of Alabama, Birmingham, who were recruited via newspaper advertisements, fliers, general gynecology clinics, and the Jefferson County STD clinic. Participants were aged 18–45 with regular menses who provided written informed consent. Exclusion criteria included antibiotic or antimycotic use within 30 days prior to enrollment, pregnancy or current lactation, active infection with Trichomonas vaginalis, Neisseria gonorrhoeae, Chlamydia trachomatis, or syphilis, symptomatic BV or vulvovaginal candidiasis for which treatment was requested at time of enrollment, immunocompromised status (including HIV), significant affective or psychotic disorders or intellectual limitations precluding informed consent, history of prior or pending hysterectomy or pelvic radiotherapy, non-fluency in English, current use of NuvaRing^® contraception, recent receipt of a bacterial vaccine, or concurrent participation in other trials involving blinded administration of antibiotics or topical microbicides. Participants self-collected daily vaginal swabs longitudinally over 10 weeks. Cumulative monetary compensation totaled $440 per participant. Isolates and MAGs obtained from participants enrolled via this protocol were derived from 43 total study participants, 11 of which contributed both isolates and MAGs, 12 of which contributed isolates only, and 20 of which contributed MAGs only.

Additional previously unreported US-derived genomes of 111 non-iners Lactobacillus isolates (Supplementary Tables 5, 6, 8, 9, and 11) were generated from samples collected as part of the Vaginal Health Project (VHP) study and the isolates were provided as a kind gift from VHP principal investigator Dr. Jeanne M. Marrazzo⁶⁴. The VHP was an IRB-approved longitudinal study of BV dynamics based in Seattle, Washington, USA, that enrolled women aged 16–30 years who reported having sex with ≥ 1 female partner in the prior year. Recruitment occurred through media, advertisements, and community referrals. Participants provided written informed consent and attended at least quarterly study visits over a year at which mid-vaginal swabs were collected, with more frequent study visits if BV was diagnosed via a combination of Amsel criteria⁶⁵ and Nugent scoring^38,64. Cumulative monetary compensation totaled $100 per participant. Bacteria were subsequently isolated from cryopreserved swabs via cultivation on Lactobacillus MRS agar (thus L. iners was not isolated). For isolate genome sequencing, a subset of 6 participants were retrospectively selected from the overall VHP cohort based on presence of both BV and cultivable lactobacilli at the baseline study visits and genome sequencing was performed on 15 to 32 isolates per participant, isolated from samples collected at between 4 and 6 study visits per participant.

Other previously unreported US-derived genomes of 4 non-iners Lactobacillus isolates (Supplementary Tables 5, 6, 8, 9, and 11) were derived from 4 participants in the V2 cohort⁶⁶, an IRB-approved cohort at Massachusetts General Hospital (MGH) in Boston, MA, (MGH IRB Protocol #2014P001066) that enrolled women presenting for gynecological care at MGH who provided written informed consent, excluding those with HIV or who were otherwise immunocompromised. Participants received no monetary compensation and ages ranged from 24–38 years. Samples were collected via swabs.

MAGs were assembled from previously unreported US-based WGS data from a single participant (Supplementary Tables 5, 6, 8, 9, and 11) in an IRB-approved Female Reproductive Tract biopsy protocol based at the Ragon Institute of MGH, MIT and Harvard in Cambridge, MA, (MGH IRB Protocol #2012P001812). The protocol enrolls non-pregnant, premenopausal women aged 18 to 50 years with normal Pap smears, normal results on routine screening laboratory tests, and willingness to provide informed consent. Exclusion criteria included history of severe gynecologic disease, symptomatic vaginal infection, treatment for vaginal infection within the prior 30 days, being <8 weeks postpartum, sexual intercourse, tampon use or application of vaginal medicines within the prior 2 days, use of immunomodulatory therapy, prior receipt of an HIV vaccine, history of significant medical illness, or history of requiring antibiotics prophylaxis for invasive procedures. Samples collected included cervical and mid-vaginal swabs, with optional consent for cervical and vaginal biopsy. Participants were compensated $150 if no biopsy was performed and $250 if biopsy was performed. The participant for whom MAGs were assembled from WGS data was 26 years old and HIV uninfected.

Additional previously unreported MAGs (Supplementary Tables 5, 6, 8, and 9-11) were assembled from previously reported FGT WGS data from three additional US cohorts (95 total participants; NCBI BioProject accession numbers PRJNA48479, and PRJNA288562)^67–70, one Italian cohort (10 participants; NCBI BioProject accession number PRJNA352475)⁷¹, and one Chinese cohort (4 participants; NCBI BioProject accession number PRJEB24147)⁷².

Bacterial culture conditions and additives for bacterial growth media

Bacterial cultivation was performed at 35–37°C under anaerobic conditions using an AS-580 anaerobic chamber (Anaerobe Systems) in an atmosphere of 5% carbon dioxide, 5% hydrogen, and 90% nitrogen (Airgas^®). All media and other culture reagents were pre-reduced (deoxygenated) by placement overnight in the anaerobic chamber prior to use. Nutritional additives or inhibitors (reagents listed in Supplementary Table 1; additives prepared as in Supplementary Table 2) were added to autoclave-sterilized broth media after cooling to room temperature (broth media), then broth was re-sterilized by passage through a 0.22 μm vacuum filter before being transferred to the anaerobic chamber to pre-reduce (deoxygenate) for use in experiments. For testing of nutrient pools (Extended Data Figure 1), a single concentrated stock comprising all pool constituents was added to media as per Supplementary Table 2 rather than stocks of each additive being added separately. In contrast to other additives, stocks of hydrogen peroxide (H₂O₂) and metronidazole (MTZ) were added to pre-reduced broth media as freshly prepared, filter-sterilized stock solutions immediately prior to inoculation of bacterial cultures due to potential volatility or chemical instability. For preparation of solid media, filter-sterilized additives were added to autoclave-sterilized agar media after cooling to 50°C, mixed using a magnetic stir/hot plate, then used to pour agar plates.

Bacterial broth culture media

Growth characteristics and nutritional requirements distinguishing L. iners from other Lactobacillus species were investigated using Lactobacillus MRS broth. We tested two different commercial MRS formulations (BD Difco^™ and Hardy Criterion). The BD Difco^™-formulated media is referred to as “MRS” while the Hardy Criterion-formulated media is referred to as “HMRS” throughout this paper. The broth was prepared by autoclaving according to manufacturer instructions and allowed to cool to room temperature. We confirmed that L. crispatus, L. jensenii, and L. gasseri grew rapidly and robustly in both formulations whereas L. iners failed to grow even after >10 days in HMRS and grew only after a prolonged delay of ~48–72 hours in BD Difco^™-formulated MRS. Supplementation of MRS broth (BD Difco^™) with IsoVitaleX^™ (2% v/v), L-Cys or L-Cys-containing mixtures (augmented by L-Gln) enabled robust growth of L. iners by 20–36 hours, depending on the experimental strain (Fig. 1a,c, and Extended Data Fig. 1a-e), while supplementation of HMRS broth had similar pro-growth effects after a more prolonged lag phase (Extended Data Fig. 1f). Unless otherwise indicated, figures depict growth in MRS (BD Difco^™) broth; key growth results were reproduced where shown using HMRS broth. MRS broth supplemented with L-Gln (1.1 mM) is abbreviated “MRSQ” (or “HMRSQ” for L-Gln-supplemented HMRS broth) in figures and text.

NYCIII broth (ATCC medium 1685) was prepared using a slightly modified version of the standard ATCC protocol. Pre-media consisted of 4 g/L HEPES, 15 g/L Proteose Peptone No.3 and 5 g/L Sodium Chloride in 875 mL distilled water, was pH-adjusted to 7.3 and autoclaved. Prior to use, complete NYCIII broth was prepared from autoclaved, cooled pre-media by adding dextrose (3g/45mL) at 7.5% v/v, yeast extract solution at 2.5% v/v, and heat inactivated horse serum at 10% v/v, then sterilized by passage through a 0.22 μm vacuum filter. Where indicated, NYCIII broth was supplemented with IsoVitaleX^™ (2% v/v), Vitamin K1-Hemin Solution (5% v/v), and/or Tween-80 (1 g/L); volumes of distilled water in the pre-media were decreased accordingly to ensure all other components were present at standard concentrations in the final solution.

“S-broth” pre-media consisting of 37 g/L BHI Broth, yeast extract (powder) 10 g/L, and dextrose 1 g/L was brought to a boil in 880 mL distilled water, then autoclaved (121°C for 15 min) and cooled, followed by addition of fetal bovine serum 5% v/v, Vitamin K1-Hemin Solution 5% v/v, and IsoVitaleX^™ Enrichment 2% v/v to a final volume of 1L. The complete broth was then sterilized by passage through a 0.22-micron vacuum filter. Unless otherwise indicated, S-broth was supplemented with 1g/L Tween-80.

Except where otherwise indicated, lactobacilli were propagated in MRSQ broth with L-Cys (4 mM), Prevotella species were propagated in Wilkins-Chalgren Anaerobe Broth (Thermo Scientific^™), and other species including Gardnerella vaginalis were propagated in NYCIII broth.

Bacterial solid culture media

Non-iners Lactobacillus isolates were maintained and quantitatively titered on Lactobacillus MRS Agar plates (purchased as prepared media from Hardy Diagnostics). Most other species, including L. iners, were maintained and titered on Columbia Blood Agar (“CBA”; purchased as prepared media from Hardy Diagnostics). Prevotella species were maintained on CBA agar or CDC Anaerobe Laked Sheep Blood Agar with Kanamycin and Vancomycin (“LKV”; purchased as prepared media from BD BBL^™). For bacterial isolations from primary clinical samples, we used commercial CBA and LKV agar plates and prepared in-house MRS agar plates by autoclaving Lactobacillus MRS Broth (BD DIFCO^™) in Noble Agar (BD DIFCO^™, 2% w/v), cooling to 50°C with agitation, then supplementing with IsoVitaleX^™ Enrichment (BD BBL^™; 2% v/v) with or without Vitamin K1-Hemin Solution (BD BBL^™; 5% v/v), or with freshly filter-sterilized stocks of L-Cys and L-Gln (at final concentrations of 4 mM and 1.1 mM respectively). All solid media was stored at 4°C in the dark after purchase or preparation until being introduced to the anaerobic chamber for deoxygenation and use.

Bacterial isolate sources and isolation protocols

Experiments were performed using a combination of previously reported bacterial strains (n = 9 strains), previously unreported South African primary isolates from cervicovaginal samples from the FRESH cohort (n = 20 strains), and previously unreported US strains from cervicovaginal samples (n = 4 strains; See Supplementary Table 3 and additional details below). Isolation of South African strains from FRESH cohort samples was performed at the Ragon Institute. Cervicovaginal swabs were thawed in the anaerobic chamber. Swabs that had been frozen dry (without cryopreservatives) were immersed in 500uL of room temperature pre-reduced PBS. Swabs that had been frozen in cryopreservative solution were rapidly thawed directly without additional dilution. Thawed swabs were agitated vigorously within the solution for 30 seconds using sterile forceps to dislodge bacteria, then removed from the cryovial using a sterile micropipette tip to strain excess fluid from the swab. The solution was then serially diluted in 10-fold dilutions in sterile, pre-reduced PBS and dilutions were plated on Lactobacillus MRS agar plates (with or without nutritional additives), CBA, and LKV agar. After 2–5 days incubation (depending on growth rate), colonies were selected based on distinct morphology, and individual representative colonies were sub-cultured onto reduced agar plates, then further clonally sub-cultured into broth media. Aliquots from the purified broth cultures were frozen at −80°C in 20% glycerol, with additional aliquots reserved for nucleic acid extraction and sequence-based taxonomic identification as described below. L. iners colonies were preferentially identified for isolation from primary swab cultures on supplemented MRS or CBA plates based on characteristic morphology after 3–4 days incubation (circular, white-translucent colonies between 0.5 and 2 mm in diameter with entire vs slightly irregular edges, smooth surface, and convex/umbonate contour that were non-hemolytic on CBA after ~4 days). In experienced hands (S.M.B. and N.A.M.), morphology-based identification of L. iners colonies for isolation from CBA plates resulted in equally high or higher rates of L. iners recovery from clinical samples compared to isolation using plates with supplemented MRS media (which had lower CFU yield than CBA), enabling us to perform targeted isolations of L. iners from >80% of attempted clinical samples, including BV-associated, anaerobe-dominated samples known to have <5% L. iners relative abundance based on 16S rRNA gene sequencing. Isolated bacteria were identified based on Illumina-based sequencing of genomic DNA and/or Sanger sequencing of the full-length bacterial 16S rRNA gene as described below. BV status of the samples from which isolates were derived was determined based on Nugent scoring when available (Supplementary Table 3).

The following 4 experimental strains were obtained through BEI Resources, NIAID, NIH as part of the Human Microbiome Project: L. iners, Strain LEAF 2053A-b, HM-705; L. iners, Strain LEAF 3008A-a, HM-708; L. iners, Strain UPII 143-D, HM-126; L. crispatus, Strain JV-V01, HM-103. Other strains (4 in total) were obtained from the American Type Culture Collection or the Culture Collection University of Gothenburg culture repositories as indicated (Supplementary Table 3). Genomes corresponding to these strains were obtained from RefSeq. Information about the BV status of the samples from which these isolates and other isolate genomes were derived was obtained (when available) from metadata accompanying the isolate genome entries in RefSeq or the Genomes OnLine Database (GOLD; https://gold.jgi.doe.gov/) or from associated information from the reference strain repositories from which they were obtained. One previously reported US-based experimental strain of L. crispatus¹ was genome-sequenced for this project using the Methods described below. Additional previously unreported US-based strains used experimentally in this study were provided by Jacques Ravel and the VMRC (4 strains; strain details are reported in Supplementary Table 3). Unless otherwise indicated, all growth, growth inhibition, isotopic tracing, and mock community experiments employed L. crispatus strain 233, L. iners strain F1, and G. vaginalis strain ATCC 14018. Reference bacterial strains used experimentally in this study are available from the indicated culture repositories (see Supplementary table 3) and previously unreported strains used in these experiments are available upon reasonable request to the corresponding author.

Nucleic acid extraction

Total nucleic acids from cervicovaginal swab samples were extracted with a phenol-chloroform method, which includes a bead beating process to disrupt bacteria as previously described^4,60. 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 (Fisher) in Tris-EDTA buffer with sterile 0.1mm glass beads (BioSpec), vigorously rubbed against the walls of the tube to dislodge microbial material, then incubated on ice for 5–10 minutes. Swabs were then removed, squeezing out excess fluid by pressing the swab against the side of the tube using a sterile pipette tip as it was withdrawn. A bead beater was used to homogenize samples for 2 minutes at 4°C, then centrifuged at 6,800xg for 3 minutes at 4°C. The aqueous phase was transferred to a clean tube with equal volume of PCI solution, mixed by vortexing, centrifuged again at 16,000xg for 5 minutes at 4°C, and the aqueous phase was transferred to a clean tube, precipitated using 0.8 volume of −20°C isopropanol (molecular biology grade, Sigma) with 0.08 volume (relative to initial sample) 3M sodium acetate at pH 5.5 (Life Technologies), inverted to mix, and incubated overnight at −20°C. Samples were then centrifuged for 30 minutes at 21,100xg 4°C, then washed in 0.5 mL 100% ethanol (Decon), and centrifuged for 15 minutes at 21,100xg at 4°C. The ethanol supernatant was discarded, the pellet allowed to dry, and the sample was resuspended in 20 μL molecular-grade Tris-EDTA buffer. Genomic DNA (gDNA) from bacterial isolates or mock communities cultured in vitro was extracted using a plate-based adaptation of the above protocol that included a bead beating process and combined phenol-chloroform isolation with Qiagen QIAamp 96 DNA QIAcube HT kit (Qiagen) procedures.

Bacterial full-length 16S rRNA gene PCR and Sanger sequencing

Bacterial isolate identity was confirmed using near-full-length bacterial 16S rRNA gene was PCR-amplified from gDNA of individual bacterial isolates using the broad-range primers Bact-8F (5’-AGAGTTTGATCCTGGCTCAG-3’) and Bact-1510R (5’-CGGTTACCTTGTTACGACTT-3’; Integrated DNA Technologies, Inc.). PCR amplicons were confirmed via agarose gel electrophoresis, then purified and directly Sanger-sequenced on an ABI3730XL DNA Analyzer at the MGH Center for Computational and Integrative Biology DNA Core using the Bact-8F and Bact-1510R PCR primers as sequencing primers in separate forward and reverse sequencing reactions. Isolates identities were determined by BLAST searches against the NCBI nucleotide collection (nt) database and Bacterial 16S Ribosomal RNA RefSeq Targeted Loci Project database (NCBI accession PRJNA33175). Isolate identities were further confirmed in most cases by bacterial whole genome sequencing as detailed below.

Shotgun library preparation

Shotgun sequencing libraries for cultured bacterial isolate genomes or culture-independent swab samples were prepared following a modified protocol of Baym et. al⁷³ using the Nextera DNA Library Preparation Kit (Illumina) and KAPA HiFi Library Amplification Kit (Kapa Biosystems). Briefly, DNA concentration of each sample was standardized to 0.6 ng/μL after quantification with SYBR Green I. Simultaneous fragmentation and sequencing adaptor incorporation was performed by mixing 0.6 ng DNA with 1.25 μL TD buffer and 0.25 μL TDE1 (Tagment DNA Enzyme, Nextera) and incubating for 10 min at 55°C. Tagmented DNA fragments were PCR-amplified using KAPA high fidelity library amplification reagents and primers incorporating Illumina adaptor sequences and sample barcodes. Products were pooled, purified with magnetic beads, and paired-end sequenced on an Illumina NextSeq with a 300-cycle kit.

Culture-independent bacterial 16S rRNA gene amplification and Illumina MiSeq sequencing

Bacterial microbiota composition in study participants and results of bacterial growth competition experiments were determined via Illumina-based amplicon sequencing of the V4 region of the bacterial 16S rRNA gene. The V4 region of the bacterial 16S rRNA gene was PCR-amplified following standard protocols^4,60,74. Samples were amplified using 0.5 units of Q5 High-Fidelity DNA Polymerase (NEB) in 25 μl reaction with 1X Q5 Reaction Buffer, 0.2 mM dNTPs (Sigma), 200 pM 515F primer (5’- AATGATACGGCGACCACCGAGACGTACGTACGGTGTGCCAGCMGCCGCGGTAA-3’, where the underlined sequence represents the region of complementarity to the bacterial 16S rRNA gene; IDT) and 200 pM barcoded 806R primer (5’-CAAGCAGAAGACGGCATACGAGATXXXXXXXXXXXXAGTCAGTCAGCCGGACTACHVGGGTWTCTAAT-3’, where the underlined sequence represents the region of complementarity to the bacterial 16S rRNA gene and the X characters represent the barcode position; IDT) in PCR-clean water (Invitrogen Ultra Pure DNase/RNase-Free Distilled water). A water-template negative control reaction was performed in parallel for each barcode master mix. Additional blank extraction and amplification controls performed using unique barcoded primers were also included in sequencing libraries. DNA from clinical swab samples for FGT microbiota profiling was amplified in triplicate reactions that were then combined prior to library pooling to minimize stochastic amplification biases. DNA from in vitro mock community experiments (for which each experimental condition had been cultured in multiple replicates) was amplified in a single reaction per replicate culture. Amplification was performed at 98°C for 30 seconds, 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 were checked via agarose gel electrophoresis in parallel with the matching water-template control reactions 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 based on estimated relative amplicon concentration) were combined into 100 μl sub-pools and purified using an UltraClean 96 PCR Cleanup Kit (Qiagen). Blank extractions, water-template, and (for in vitro experiments) blank media controls were included in the sequencing libraries although they did not produce visible PCR bands. Concentration of the sub-pools were quantified using a Nanodrop (Thermo Scientific), then pooled at equal molar concentrations to assemble the final library. The pooled library was diluted and supplemented with 10% PhiX according to standard Illumina protocols, 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)⁷⁴.

Genome and meta-genome sequence processing and assembly

We constructed Lactobacillus genome catalogs from 127 reference isolate genomes, as well as from genome sequences of previously unreported bacterial strains and MAGs assembled from culture-independent WGS sequences of genital tract samples derived from the cohorts detailed above (Hayward et al, manuscript in preparation; see cohort descriptions above and details in Supplementary Tables 4-11). Briefly, we retrieved all reference genomes annotated as Lactobacillus crispatus, Lactobacillus iners, Lactobacillus jensenii, Lactobacillus gasseri, or Lactobacillus vaginalis that were deposited in the NCBI RefSeq database³⁵ as of February 2020 (Supplementary Table 7). Genomes of previously unreported FGT strains from the South African FRESH cohort (isolated as described above) and of non-iners Lactobacillus strains isolated from the US-based VHP and V2 cohorts were sequenced on an Illumina NextSeq at the Ragon Institute as described above. Additional US FGT Lactobacillus isolate genomes were provided by the VMRC from the cohorts detailed above. FGT WGS metagenomic samples from cohorts in South Africa, the US, Italy, and China (see above) were used to generate MAGS. Sequence reads from isolate genomes and WGS samples were trimmed and filtered to high quality reads. Human reads from WGS samples were removed by mapping to GRCh38 (Genbank accession GCA_000001405.15). We assembled genomes and MAGs from high quality reads, then binned contigs and removed contamination. Bin completeness, contamination, and strain heterogeneity metrics were determined, then genomes and MAGs were assigned quality scores based on criteria from the Genome Standards Consortium⁷⁵. Species genome bins (SGBs) were determined based on 95% pairwise absolute nucleotide identity (ANI).

Lactobacillus genome taxonomy assignment

We assigned taxonomy to SGBs with FGT Lactobacillus genomes based on presence of reference genomes with defined NCBI taxonomy, confirming that each contained reference genomes only from a single species. Of note, for both L. jensenii and L. gasseri, publicly available reference genomes classified in RefSeq as belonging to these species actually segregate into two separate SGBs per species, indicating that L. jensenii and L. gasseri (as traditionally defined) each comprise two distinct genomic species (Hayward et al, manuscript in preparation)⁷⁶. However we analyzed L. jensenii and L. gasseri as single species units in this study to correspond with prevailing paradigms in literature on 16S rRNA gene-based FGT microbiota profiling⁷⁷.

Cervicovaginal Lactobacillus pan-genome construction and analysis

We constructed a cervicovaginal Lactobacillus pan-genome using genes from catalogs of L. iners, L. crispatus, L. jensenii, L. gasseri, and L. vaginalis genomes and MAGs. To maximize comprehensiveness of the pan-genome, we included genomes and MAGs classified as high-quality (>90% completeness and <5% contamination) or medium-quality (≥50% completeness and <10% contamination) assemblies based on Genome Standards Consortium criteria⁷⁵. Some of the medium-quality assemblies are classified as “partial” by NCBI standards based on size criteria. Genes were identified within individual genomes by Prokka v1.14.5⁷⁸, which predicts genes using Prodigal v2.6.3⁷⁹. The resulting genes were clustered into a comprehensive pan-genome at 95% nucleotide identity using Roary v3.13.0⁸⁰ to generate a multi-fasta file containing gene sequences and a per-genome gene presence-absence table⁸¹. The pan-genome was annotated with eggNOG 5.0 using eggNOG-mapper v2^82,83. We used custom R scripts to parse the eggNOG output for genes predicted to encode enzymatic or transporter activities of interest based on gene names, KEGG numbers, KEGG reaction numbers, Clusters of Orthologous Groups (COGs), EC numbers, Transporter Classification (TC) numbers, and GO (Gene Ontology) terms, followed by manual curation of the initial search results and BLAST-based sequence review for genes with unclear annotations⁸¹. We then determined the number of genomes from each species predicted to encode each gene function of interest. Since the Genome Standards Consortium criteria for high- and medium-quality genomes and MAGs allow for low-level sequence contamination within assemblies⁷⁵, we restricted gene presence-absence analysis to gene sequences of interest that were detected in at least two genomes or MAGs from a species in order to exclude singleton contaminant sequences from our analysis. Importantly, including partial genomes and MAGs with completeness as low as 50% maximizes genome catalog diversity, thus increasing pan-genome size and sensitivity for detecting genes of interest within each species, but also results in a fraction of genomes and MAGs appearing to lack universally present genes due to incompleteness of assemblies. The absence of intact Cys synthase and reverse transsulfuration pathways in these species determined by EggNOG-based analysis (Figure 2b) was further supported by results of BLAST⁸⁴ searches against the genome collections using gene sequences of interest from related species (see custom code posted on Zenodo⁸¹).

Maximum likelihood phylogenetic distances and phylogenetic reconstruction

Phylogenetic reconstruction of L. iners genomes and MAGs (Fig. 1b) was performed using assemblies with >60% completeness and <5% contamination to maximize robustness of the analysis. fetchMG v1.0⁸⁵ was used to extract DNA sequences for each of forty single-copy universal bacterial marker genes from each genome (called using Prodigal⁷⁹). Genome assemblies containing fewer than 10 of the 40 universal marker genes were omitted from the subsequent alignment. A phylogenetic reconstruction was then produced using ETE3 v3.1.1 (parameters: “ete3 build -w clustalo_default-trimal-gappyout-none-none -m cog_85-alg_concat_default-fasttree_default”)⁸⁶, and the tree was visualized using iTOL v4⁸⁷. An additional phylogenetic reconstruction encompassing other major FGT Lactobacillus species (Extended Data Fig. 2a) was performed using L. iners genomes and MAGs as well as isolate genomes from L. crispatus, L. jensenii, L. gasseri, and L. vaginalis, filtered based on the same quality and completeness criteria.

Preparation of bacterial inocula for growth experiments

For initial experiments identifying cysteine and cystine as key nutrient requirements for L. iners growth (Fig. 1a,c, Fig 4a, and Extended Data Fig. 1), bacteria from frozen stocks were plated on solid media, incubated for 3 days, then suspended in sterile, pre-reduced Dulbecco’s phosphate-buffered saline (PBS), adjusted to an OD600 of 0.3+/−0.05, and then inoculated into the indicated broth media formulations for measurement of growth kinetics. In subsequent growth experiments for Lactobacillus species, experimental bacterial inocula were prepared from liquid starter cultures in MRSQ broth with L-Cys (4 mM) that were incubated for 18–20 hours. The starter cultures were then pelleted by centrifugation for 10 min at 3716 x g, spent media was decanted, and bacteria were washed 2 times by re-resuspending in sterile, pre-reduced PBS followed by centrifugation to avoid carryover of nutrients from the original starter culture media. Washed bacteria were resuspended in PBS, adjusted to OD600 0.3 (± 0.05), then inoculated into experimental media at 3.5% (v/v), equating to bacterial titers ranging from ~1×10⁵ to 1×10⁶ colony-forming units (C.F.U.) per mL, depending on the experimental species and strain.

Growth kinetics quantification

We found that Lactobacillus growth kinetics in broth media were adversely affected by either continuous or intermittent agitation, therefore we grew broth cultures without agitation. Mono-culture growth kinetics were assessed using separate cultures prepared in parallel for each experimental timepoint to avoid repeatedly agitating a single culture by performing serial measurements. For timepoint series lasting ≤48 hours, broth cultures were grown in a volume of 250 μL in technical triplicate in clear 96-well flat bottom plates (Falcon), with a blank for each media condition per plate. Measurements were taken at ~16, 20, 24, 28, ~42, and 48 hours by removing plates from the anaerobic chamber, pipetting to re-suspend bacteria, and measuring OD600 on a Tecan Infinite^® M1000 PRO plate reader. Due to slower growth kinetics in the HMRS broth formulation, L. iners growth measurements in HMRS were taken daily for up to 10 days, precluding use of 96-well plates due to evaporation at later timepoints. Bacteria were therefore incubated in low-evaporation 1.2 mL 96-well cluster tubes (Corning^™) in triplicate, with separate parallel culture plates used for each experimental timepoint, then 250 μL from each culture was transferred to a clear 96-well flat-bottom Falcon plate for optical density measurement. Data from mono-culture growth experiments are depicted as median ± range for the three replicates. For each bacterial strain and media condition, figures depict representative results from 1 of ≥2 independent experiments with distinct batches of freshly prepared media and freshly prepared bacterial input inocula unless otherwise indicated. Growth data collection and analysis were not performed blind to the conditions of the experiments.

Inhibitor experiments and analysis

To test mono-culture growth inhibition, bacteria were cultured in media containing inhibitors at varying concentrations as indicated, including a reference (no-inhibitor) control. Since growth kinetics differed between species and strains, inhibition was determined for each strain at the first timepoint fulfilling the European Commission on Antimicrobial Susceptibility Testing (EUCAST) criterion of “definite turbidity” in the reference control (which we defined experimentally as OD600 >0.2), unless otherwise specified. At the selected timepoint for each strain, the median OD600 value for the reference control was set to a reference value of 100%, which was used to calculate percentage growth in the other conditions. Growth experiments employing metronidazole (MTZ) used a concentration of 50 μg/mL, approximating the concentration in 0.75% intravaginal MTZ gel (a first-line treatment for BV)²⁵. Data from inhibition experiments are depicted as median ± range for three technical replicates. For each bacterial strain and media condition, figures depict representative results from 1 of ≥2 independent experiments with distinct batches of freshly prepared media and freshly prepared bacterial input inocula, unless otherwise indicated. Inhibition data collection and analysis were not performed blind to the conditions of the experiments.

Competition and mock community culture experiments

For pairwise competition between L. iners and L. crispatus strains in MRS broth ± SMC and for mock BV-like community experiments with L. iners (multiple strains), L. crispatus (multiple strains), G. vaginalis (ATCC 14018), P. bivia (ATCC 29303), and A. vaginae (0795_578_1_1_BHK4; see Supplementary Table 3) in S-broth ± SMC and/or MTZ, bacteria were initially prepared in individual suspensions in broth as described above for bacterial monoculture growth and inhibition experiments. Aliquots of the mono-bacterial suspensions were mixed in defined ratios, then mixtures were divided into replicate cultures for incubation (5 replicates per conditions for MRSQ experiments; 6 replicates per condition for S-broth experiments) by adding 150 μL of mixture per well into V-bottom 96-well plates (Falcon). C.F.U. titers were determined for each input mono-bacterial suspension as described above and used to calculate starting ratios within the mixed cultures. At 28 hours, cultures were harvested by centrifuging at 4700xg for 25 minutes at 4C, spent media was decanted, and pellets were frozen for later DNA extraction and analysis. Relative growth within mixed cultures was assessed by bacterial 16S rRNA gene sequencing as described below. Aliquots of mono-bacterial suspensions in the corresponding media type were cultured separately to confirm expected growth patterns (e.g., Fig. 5b and Extended Data Fig. 7a).

Processing of Illumina MiSeq 16S rRNA gene sequences

Demultiplexing of Illumina MiSeq bacterial 16S rRNA gene sequence data was performed using QIIME 1 version 1.9.1⁸⁸. Mapping files in standard QIIME 1 format were created and validated using validate_mapping_file.py, sequences were demultiplexed using split_libraries_fastq.py with parameter store_demultiplexed_fastq and no quality filtering or trimming, and demultiplexed sequences were organized into individual fastq files using split_sequence_file_on_sample_ids.py⁶⁰. We then used dada2 version 1.6.0⁸⁹ in R 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. Sequences were then inferred and initial taxonomy assigned using the dada2 assignTaxonomy function employing the RDP training database rdp_train_set_16.fa.gz (https://www.mothur.org/wiki/RDP_reference_files). Taxonomic assignments were refined and extended via extensive manual review; see Supplementary Table 15 for amplicon sequence variant (ASV) taxonomy. The denoised dada2 results with final taxonomic assignment were analyzed in R using phyloseq version 1.30.0⁹⁰ and custom R scripts. Initial sequencing-based analysis of FGT microbiota composition of FRESH cohort participants, taxonomy assignment, and cervicotype assignments were performed blinded to participants’ corresponding Nugent scores and metabolite concentrations in CVL fluid.

16S rRNA gene-based microbiome analysis

For 16S rRNA gene-based microbiome profiling of clinical specimens, microbial communities were classified into four cervicotypes (CTs) as previously defined 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 population⁴); 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 species¹. Although L. jensenii and L. gasseri were detected in microbial communities of some FRESH study participants by both sequencing and cultivation, they were not dominant in any individuals, a common finding among studies of sub-Saharan African populations^1,4,6,50 that contrasts with observations in most North American and European cohorts, where a consistent minority of women tend to have L. gasseri-dominant or L. jensenii-dominant communities^12,49,59,77. For further sequence processing and analysis, amplicon sequence variants (ASVs) that could not be defined at least to the level of taxonomic class were pruned from the dataset and one sample was excluded due to sequence read count <10,000. The remaining 142 samples were rarefied without replacement to a uniform depth of 16,603 reads (the minimum read count among remaining samples), yielding a relative abundance limit of detection (L.D.) of 6.02 × 10^-5. ASVs were collapsed at the species or genus level as indicated for further visualization and statistical analysis. In plots displaying taxon relative abundance using a logarithmic scale, a pseudocount equal to 0.5 x L.D. was added to assist visualization of any taxon with read count 0. Correlation analysis between bacterial relative abundances and cervicovaginal metabolite concentrations was restricted to taxa detected (≥1 sequence read after rarefaction) in ≥50% of samples (Extended Data Fig. 3c,d).

Sequence analysis for in vitro competition experiments and mock communities

For pairwise competitions between L. crispatus and L. iners, 16S rRNA gene sequences were generated, processed, and annotated as described above. Ratios of L. iners reads to L. crispatus reads in each sample were calculated. Significance of between-group differences for each mixture was determined by 1-way ANOVA. Significance of pairwise differences between the positive control condition (MRSQ + L-Cys without inhibitor) and each experimental condition was calculated using Dunnett’s test and all significant pairwise comparisons were plotted (see Supplementary Table 18 for numeric p-values). Results of representative competition experiments are shown in Fig. 5a.

For mock BV-like community experiments, 16S rRNA gene sequences were generated, processed, and annotated as described above. Relative abundances of each experimental strain were determined for individual replicates and displayed in Fig. 5d. To assess L. crispatus enrichment, read counts of all experimental strains except L. crispatus were summed for each replicate sample, then ratios of L. crispatus to the summed taxa were calculated. Significance of between-group differences for each mixture was determined by 1-way ANOVA, and significance of pairwise comparisons were calculated using Tukey’s test, with significance of selected pairwise comparisons plotted in Fig. 5e and shown in detail in Supplementary Table 20. Competition and mock community assay data collection and analysis were not performed blind to the conditions of the experiments.

Measurement and analysis of metabolites in cervicovaginal lavage fluid

Concentrations of Cys, reduced glutathione (GSH, γ-L-glutamyl-L-cysteinyl-glycine), and cysteinylglycine (Cys-Gly) were measured in CVL supernatants from the 143 FRESH cohort participants whose microbiome composition was concurrently profiled by 16S rRNA gene sequencing in this study. CVL supernatant samples underwent a single freeze-thaw cycle during preparation of aliquots for metabolite analysis. Metabolites of interest were quantified using ultraperformance liquid chromatography/tandem mass spectrometry (UPLC-MS/MS) by Metabolon, Inc., as part of an untargeted metabolomics analysis. To avoid bias in assay performance, measurements were performed after randomized reordering of samples by laboratory staff at Metabolon, Inc., who were blinded to all participant information except sample ID. Metabolon’s method generates relative instead of absolute metabolite concentrations. For analysis, concentrations were volume-normalized and median values were adjusted to 1, then percent missingness (reflecting samples with analyte concentrations below the limit of detection, or L.D.) was calculated for each analyte. For analytes with missingness >0%, analyte limit of detection was inferred as equaling the lowest measured relative concentration in the cohort, then missing values were imputed at half the limit of detection. The Shapiro-Wilk test of normality was then performed on relative concentration values after log-transformation. In subsequent analysis of concentration differences between cervicotypes, differences were analyzed by 1-way ANOVA with post-hoc Tukey test for Cys (which had 0% missingness and was normally distributed, Fig. 3a,c) and by Kruskal-Wallis test with post-hoc Dunn’s test for GSH and Cys-Gly (which had 21.7% and 7.0% missingness respectively and were therefore not normally distributed after imputation, Extended Data Fig. 4a-d). We calculated Spearman rank-order correlations between individual metabolites and bacterial taxa at both the genus and species levels, adjusting p-values for multiple comparisons where indicated using the Bonferroni method via the R stats package function p.adjust(). The R package DescTools was used to calculate confidence intervals for the Spearman correlation coefficients (ρ), including Bonferroni-corrected confidence levels at (1 – 0.05/n), where n represents the number of taxa (Fig. 3e,f, Extended Data Fig. 4e,f, and Supplementary Tables 16 & 17).

Quantification of oxidized and reduced cysteine and glutathione in MRS

Oxidized and reduced glutathione and cysteine were quantified by UPLC-MS/MS (Waters Acquity/TQ-S), according to a modification of the protocol by Sutton et al⁹¹. Briefly, 100 µL anaerobic broth was mixed with 90 µL buffer (100 mM ammonium bicarbonate, pH 7.4), followed by 10 µL fresh N-ethylmaleimide solution (25 mg/mL in ethanol), and allowed to react at room temperature for 10 minutes. Samples were then diluted 100-fold in water before analysis by UPLC-MS/MS, with a 1 µL injection volume. Separation was carried out with an Acquity BEH/C18 UPLC column (Waters 186002350), with the following gradient: 0 minutes, 2% B; 6 minutes, 98% B; 7 minutes, 98% B; 7.1 minutes, 2% B; 9 minutes, 2% B. Mobile phase A was water with 0.1% formic acid, and B was acetonitrile with 0.1% formic acid. Compounds were quantified by comparison of peak area to authentic standards processed via the same method.

Synthesis and analysis of isotopically labeled cystine, GSSG and GSH

¹³C-labeled cystine (1,1’−¹³C₂-L-cystine) was prepared from 1-¹³C-L-cysteine (Cambridge Isotope Labs, CLM-3852–0.5) via a modification of the procedure of Hill, Coy, and Lewandowski⁹². Briefly, 100 mg of labeled cysteine was mixed with 200 µL 10% NaOH, and 800 µL 3% hydrogen peroxide. The resulting precipitate was filtered and washed extensively with ice cold water, then finally redissolved in 1 M aqueous HCl. Complete oxidation was confirmed by UPLC-MS/MS analysis of the resulting solution, as described above.

Labeled glutathione-(1-¹³C-cysteine) was prepared enzymatically using E. coli GshA and GshB overexpressed and purified as N-terminal 6xHis fusions. gshA and gshB were amplified from genomic DNA of E. coli BL21(DE3) using Q5 polymerase (New England Biolabs) and the following primers (purchased from Millipore-Sigma): HIS-gshA-F, gcctggtgccgcgcggcagcATCCCGGACGTATCACAG; HIS-gshA-R, cagcttcctttcgggctttgTCAGGCGTGTTTTTCCAGCC; HIS-gshB-F gcctggtgccgcgcggcagcATCAAGCTCGGCATCGTGAT; and HIS-gshB-R, cagcttcctttcgggctttgTTACTGCTGCTGTAAACGTGC, according to the manufacturer’s instructions. The expression vector backbone pET28 was amplified from purified stock using primers: pET28-F, GCTGCCGCGCGGCACCAG; and pET28-R, CAAAGCCCGAAAGGAAGCTG. After DpnI digest and purification, expression plasmids pETgshA and pETgshB were assembled using HiFi Assembly MasterMix (NEB) and transformed into chemically competent E. coli DH5a. Positive colonies were verified via Sanger sequencing, and correctly assembled plasmids were transformed into E. coli BL21(DE3). Cells for protein expression were inoculated from overnight culture with 100-fold dilution into 4 L flasks containing 1 L LB broth, and grown at 37°C and 200 RPM until reaching OD600 ~ 0.5, after which they were induced with 0.5 mM IPTG, and grown for a further 3 hours at 37 ⁰C. Pellets were harvested by centrifugation at 6,000 g for 10 minutes, then resuspended in ice cold 98% Buffer A (50 mM HEPES, 300 mM KCl, 10% glycerol, pH 7.5) and 2% Buffer B (Buffer A with 500 mM imidazole). Cells were lysed by three passages through an Emulsiflex C5 (Avestin) at 15,000 psi, and lysates were clarified by centrifugation at 20,000 g, 4°C for 30 minutes. Target proteins were bound to 1 mL Ni-NTA His-Bind resin (Millipore Sigma) in gravity columns, then washed with 5 mL wash buffer (90% Buffer A, 10% Buffer B) and eluted with 3 mL elution buffer (100% Buffer B). Fractions containing purified proteins were identified via SDS-PAGE, then combined and dialyzed overnight against Buffer A with 3.5k MWCO Slide-A-Lyzer Cassettes (ThermoFisher), then concentrated by centrifugation through 3k MWCO Microsep spin filters (Pall).

Glutathione (γ-L-glutamyl-L-cysteinyl-glycine) synthesis was carried out in 100 mM sodium phosphate buffer (pH 7.2), with 80 mM 1-¹³C-L-cysteine, 120 mM glycine, 120 mM glutamic acid, 40 mM MgCl₂, and 100 mM ATP, using 50 mg 1-¹³C-L-cysteine, 1 mg each of GshA and GshB, incubating overnight at 37°C. The resulting glutathione-(1-¹³C-cysteine) was treated with 3% hydrogen peroxide to produce the oxidized form of glutathione (GSSG) for purification via preparative HPLC on a Thermo Scientific Dionex UltiMate 3000 HPLC system with a Thermo Scientific Hypersil GOLD aQ C18 preparative column (20 × 20 mm, 5 μm) using a gradient from 100% A (water + 0.1% formic acid) to 70% B (acetonitrile + 0.1% formic acid). Fractions containing labeled GSSG (¹³C₂-GSSG) were pooled and lyophilized. Analysis of the purified product was conducted by ¹H NMR (400 MHz) in the Magnetic Resonance Laboratory in the Harvard University Department of Chemistry and Chemical Biology on a Jeol J-400, as well as by LC-MS/MS as described above, and by high-resolution LC-MS using an Agilent 6530 Q-TOF (confirming expected product characteristics as detailed in HMDB: https://doi.org/10.13018/BMSE000906). This analysis confirmed the authenticity of the desired product and complete conversion of cysteine, and also revealed contamination with ADP, which was not removed by preparative HPLC (Supplementary Fig. 1). This material was used without further purification. Reduced labeled glutathione (¹³C-GSH, 4 mM) for use in experiment measuring GSH uptake was produced from the oxidized form (¹³C₂-GSSG) by treating a stock solution of 81.6 mM labeled GSSG with TCEP at a 0.9:1 molar ratio to reduce the disulfide bonds prior to addition to MRSQ broth.

Quantification and isotopic analysis of amino acids

For isotopic analysis of proteinogenic amino acids, L. crispatus (strain 233) and L. iners (strain F1) grown in MRS/Q broth supplemented with isotopically labeled or unlabeled substrates were harvested via centrifugation. The pellets washed 3 times with water and resuspended in 0.6 mL of 6 N aqueous HCl and heated at 100°C overnight to facilitate cell lysis, protein hydrolysis, and Cys oxidation. The hydrolysate was dried under air, then resuspended in 100 µL of 90% acetonitrile, 10% water. For determination of amino acid concentration in media, samples were prepared by diluting media 10–100-fold in 90:10 acetonitrile:water, followed by centrifugation to remove precipitated material. All amino acid samples were analyzed via UPLC-MS/MS using instrumentation described above. Separation was carried out with an Acquity BEH/Amide UPLC column (Waters 186004800) in HILIC mode, with the following gradient: 0 minutes, 90% B; 0.3 minutes, 73% B; 1 minute, 73% B; 1.5 minutes, 30% B; 2 minutes, 30% B; 2.2 minutes, 90% B; 4 minutes, 90% B. Mobile phase A was water with 0.1% formic acid, and B was acetonitrile with 0.1% formic acid. Quantification of amino acids was confirmed by comparison of peak area to authentic standards (Sigma # AAS18–5ML) processed via the same method. Isotopic distribution data were corrected for the natural abundance of ¹³C using IsoCor v2.1.3⁹³ (Supplementary Tables 13 & 14). Isotopic assay data collection and analysis were not performed blind to the conditions of the experiments.

Statistics, software, and visualization

Data analysis, statistics, and visualization were performed in R v3.6.3, except where otherwise indicated, using packages including seqinr v4.2.5, tidyverse, v1.3.1, knitr v1.33, ggpubr v0.4.0, DescTools v0.99.41, gtools v3.8.2, gridExtra v2.3, cowplot v1.1.1, scales v1.1.1, grid v3.6.3, broom v0.7.6, e1071 v1.7.6, and table1 v1.4. All p-values are two-sided with statistical significance defined at α = 0.05 unless otherwise indicated.

Data availability

Compressed directories containing data files sufficient to reproduce (1) analysis of pan-genome composition and gene content, (2) analysis for human FGT microbiota-metabolite analysis, and (3) analysis for competition cultures of L. iners and L. crispatus and mixed community cultures are posted at zenodo.org under DOI https://zenodo.org/record/5900469 ⁸¹. The dataset containing the raw Illumina MiSeq read data for genital tract bacterial 16S rRNA gene profiling analyzed in this study (Figure 3 and Extended Data Figures 3 & 4) is available in the NCBI Sequence Read Archive (SRA) under BioProject PRJNA729907. The dataset containing the raw Illumina MiSeq read data for the bacterial 16S rRNA gene sequences from competition cultures of L. iners and L. crispatus (Fig. 5a) and from mixed community cultures (Figs 5c&d) is available in the NCBI SRA under BioProject PRJNA777644. The taxonomic assignments used for amplicon sequence variants (ASVs) from bacterial 16S rRNA gene sequencing are supplied in Supplementary Table 15. The Lactobacillus genomic catalogs included a total of 1,091 previously unreported isolate genomes, partial genomes, and MAGs from multiple human cohorts, as detailed above. The assemblies are available in the NCBI SRA under BioProjects PRJNA799384, PRJNA799634, PRJNA799626, PRJNA799445, PRJNA799630, PRJNA799633, PRJNA799642, PRJNA799746, PRJNA799744, PRJNA799737, PRJNA799762, and PRJNA797778; additional details on the individual studies associated with these BioProjects are contained in Supplementary Tables 5, 6, and 10, and individual NCBI BioSample accession numbers for each of the 1,091 assemblies are listed in Supplementary Table 8. In addition, the genome catalogs included 127 previously reported isolate genomes that were retrieved from RefSeq; the individual accession numbers for these genomes are listed in Supplementary Table 7. The raw and corrected cystine and serine isotopologue measurements associated with Fig 2c,d and 4c are available in Supplementary Tables 13 and 14. Some metadata related to previously reported isolate genomes was obtained from corresponding entries in RefSeq (https://www.ncbi.nlm.nih.gov/refseq/) or the Genomes OnLine Database (GOLD; https://gold.jgi.doe.gov/). Source data are supplied for plots and phylogenetic trees, including for Figures 1–5 and Extended Data Figures 1-8.

Code availability

Compressed directories containing R analysis code sufficient to reproduce (1) analysis of pan-genome composition and gene content, (2) analysis for human FGT microbiota-metabolite analysis, and (3) analysis for competition cultures of L. iners and L. crispatus and mixed community cultures is available at zenodo.org under DOI https://zenodo.org/record/5900469 ⁸¹. Each compressed directory contains a README file describing dependencies and other details, an R Project file, an R Markdown file containing the analysis code with additional information, and associated sub-directories used in the analysis.

Extended Data

Extended Data Fig. 1. L-cysteine supplementation supports L. iners growth in Lactobacillus MRS broth, augmented by L-glutamine.

a,b Growth of L. crispatus, L. iners, and G. vaginalis at 24 hours incubation in MRS broth (BD DIFCO^™) ± supplementation with 2% IsoVitaleX^™ (“Iso”) or with the indicated sub-pools of IsoVitaleX^™ components. c, Growth in MRS broth supplemented with 2% IsoVitaleX^™ or various combinations of the nutrients in “Pool 2”. d, Growth of L. iners at 24 hours in MRS broth + L-Gln (1.1 mM) supplemented with varying concentrations of L-Cys or e, in MRS broth + L-Cys (4 mM) supplemented with varying concentrations of L-Gln. Experiments in (a-e) all used BD DIFCO^™-formulated MRS broth base. (f) Growth of L. iners in Hardy Criterion-formulated MRS (“HMRS”) broth supplemented with IsoVitaleX^™ 2% v/v, L-Cys (4 mM), and/or L-Gln (1.1 mM) produced similar results, although with a substantially longer lag phase. Each plot depicts median (± range) for 3 replicates per condition and each plot is representative 1 of ≥2 independent experiments per strain and media condition except (a), which was un-replicated.

Extended Data Fig. 2. Phylogeny and per-genome gene content of genomes and MAGs within FGT Lactobacillus genome catalogs.

a, Phylogenetic tree of L. iners isolate genomes and MAGs as in Fig 1b, plus isolate genomes of L. crispatus (n = 182), L. jensenii (n = 39), L. gasseri (n = 28), and L. vaginalis (n = 8). Isolates further experimentally studied in this work are indicated. The tree depicts only genome assemblies exceeding certain quality thresholds to ensure robustness of the phylogenetic reconstruction (see Methods); additional strains and genomes were included in other analyses. b, Number of genes per genome for each genome retrieved from RefSeq (“Reference Genome”), previously unreported isolate genome (“Novel Genome”), or MAG within the Lactobacillus genome catalogs analyzed in Fig. 2b and 4b and Extended Data Fig. 6a. All MAGs present in the analysis represent previously unreported assemblies (Hayward et al, manuscript in preparation, see additional information in Methods and Supplementary Tables 4-11). To maximize comprehensiveness of the pan-genomes, the catalogs included high- and medium-quality genomes and MAGs, defined as assemblies with minimum estimated completeness >50% (some of which are classified as partial assemblies by NCBI size criteria) and maximum estimated contamination <10%. Gene count analysis excludes gene sequences observed in only 1 genome or MAG per species to eliminate singleton contaminating sequences within individual genome assemblies. Box center lines, edges, and whiskers signify the median, interquartile range (IQR), minima and maxima respectively.

Extended Data Fig. 3. In vivo association of BV status and vaginal Cys concentrations with microbiota composition.

a, Relationship between Nugent score-based BV status and cervicotype among the 53 women depicted in Fig. 3a. BV status and cervicotype were significantly associated (P = 1.902 × 10⁻¹¹; two-sided Fisher’s Exact Test). b, Two-tailed Spearman correlation between relative Cys concentrations in cervicovaginal lavage (CVL) fluid and relative abundances of the species L. iners, and L. crispatus among the 142 women depicted in Fig. 3b-f, showing correlation coefficients (ρ) with unadjusted p-values. Linear regression lines (solid blue) with 95% confidence intervals calculated based on log-transformed abundances and concentrations are shown to assist visualization (L. crispatus: y = 0.77 + 0.25x; L. iners: y = 0.30 + 0.22x). The red dotted line represents the bacterial limit of detection (L.D.). c,d Per-sample relative abundances and cohort-level prevalence (fraction of samples from the cohort in which each taxon was detected) for each genus (c) or species (d) with ≥50% prevalence (panels correspond to main Fig. 3e and 3f, respectively). Purple and blue lines respectively represent median and interquartile range of relative abundances for each taxon.

Extended Data Fig. 4. Vaginal concentrations of the Cys-containing peptides reduced glutathione (GSH) and cysteinylglycine (Cys-Gly) in cervicovaginal fluid are higher in women without BV and correlate with Lactobacillus dominance of the microbiota.

a,b Relative concentrations of (a) GSH and (b) Cys-Gly by BV status in CVL fluid from the 53 women in (Fig. 3a). c,d Relative concentrations of (c) GSH and (d) Cys-Gly by cervicotype in CVL fluid from the 142 women depicted in (Fig. 3b,c). In (a-d) the red dotted line represents the metabolite limit of detection (L.D.). For samples in which an analyte was below the L.D., concentrations were imputed at 0.5 x L.D. Log-transformed concentrations were not normally distributed due to the imputed values, so between-group differences were determined via Kruskal-Wallis test with post-hoc Dunn’s test, adjusting for multiple comparisons using the Bonferroni method. All significant pairwise differences are displayed. a No BV-BV: P = 5.7 × 10⁻⁷; No BV-Intermediate: P = 0.0037. b No BV-BV: P = 0,0032. c CT1-CT4: P = 7.7 × 10⁻⁵; CT2-CT3: P = 0.0292; CT2-CT4: P = 2.9 × 10⁻¹¹. d CT1-CT4: P = 0.0191; CT2-CT4: P = 0.00014. Box center lines, edges, and whiskers signify the median, IQR, minima and maxima respectively. e,f Forest plots depicting Spearman correlation coefficients (ρ) between concentrations of GSH and Cys-Gly and relative abundances of each bacterial genus (e) or species (f) detected at >50% prevalence in the cohort (n = 142). P-values and confidence intervals in e,f were adjusted for multiple comparisons using the Bonferroni method at significance level 0.05/n (full statistical results in Supplementary Tables 14 & 15). Significance is depicted for adjusted p-values as * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001.

Extended Data Fig. 5. Cys and Cys-containing molecules in MRS exist primarily as mixed disulfides and addition of chemical reducing agents permits L. iners growth.

a, Growth of L. iners and L. crispatus at 28 hours in MRSQ broth supplemented as indicated with the reduced thiols L-Cys, D-Cys (the non-physiological enantiomer of L-Cys), GSH, or homocysteine (each 4 mM), with their oxidized counterparts L-cystine, D-cystine, oxidized glutathione (GSSG), or homocystine (each 2 mM), or with TCEP (4 mM), H₂O₂ (0.4 mM), or L-cystine + H₂O₂. b, Concentrations of reduced Cys (baseline median concentration 1.11 μM) and glutathione (GSH; baseline median concentration 1.70 μM) in MRSQ broth supplemented with the oxidizing agent H₂O₂ (0.4 mM), the non-sulfur-containing reducing agents Tris(2-carboxyethyl)phosphine (TCEP) or homocysteine (each 4 mM), or homocysteine’s oxidized counterpart homocystine (2 mM). c, Growth at 7 days of L. crispatus and L. iners in HMRS broth with 1.1 mM L-Gln (“HMRSQ”) supplemented as indicated with L-Cys, L-cystine, TCEP, homocysteine, or homocystine at the above concentrations. All plots depict median ± range for 3 replicates per condition and each plot is representative of 1 of ≥2 independent experiments per strain and media condition. Bar coloring highlights the pairing of media conditions with each thiol-containing reducing agent and its oxidized counterpart.

Extended Data Fig. 6. FGT Lactobacillus genomes lack predicted alternate Cys and GSH transporters and L. iners is selectively inhibited by cystine uptake inhibitors.

a, Predicted presence of the branched-chain amino acid transport locus livFGHKM (which has low-affinity Cys transport activity in E. coli) and the glutathione transport locus gsiABCD in isolate genomes and MAGs of common FGT Lactobacillus species (n = number of genomes. Detailed statistics are in Supplementary Table 4). b, Selective growth inhibition of L. iners in MRSQ broth with or without L-Cys (4 mM) and varying concentrations of the cystine uptake inhibitor seleno-DL-cystine (SDLC). c, Growth of L. crispatus and L. iners in HMRSQ broth with or without L-Cys (4 mM) ± SMC. d, Growth inhibition of L. crispatus and L. iners by SMC (128 mM) or SDLC (2 mM) in MRSQ supplemented with L-cystine (2 mM) or D-Cys, TCEP, or GSH (4 mM each). For each growth additive, percentage growth in in presence of inhibitor was calculated relative to median growth in broth containing that additive without inhibitor. Plots in b-d depict median ± range for 3 replicates per condition and each plot is representative 1 of ≥2 independent experiments per strain and media condition.

Extended Data Fig. 7. Sample preparation and controls for L. iners / L. crispatus growth competition assays.

a, Mono-culture growth of L. crispatus (strain FRESH1) and 3 representative L. iners strains at 28 hrs incubation in MRSQ broth with or without L-Cys (4 mM) and varying concentrations of the cystine uptake inhibitor S-methyl-L-cysteine (SMC), exhibiting expected growth patterns for the respective species. Plots depict median ± range for 3 replicates per condition and each plot is representative 1 of ≥2 independent experiments per strain and media condition. The competition assays between L. iners and L. crispatus depicted in Fig. 5a were prepared by mixing the input inocula from each of these L. iners monocultures pairwise with the input inoculum for the L. crispatus monoculture at the colony-forming unit (C.F.U.) ratios listed in the legend of Fig. 5a. b, Bacterial 16S rRNA gene read counts in cultured samples (n = 105 individual cultures), with negligible background read counts in blank growth media controls (n = 7 controls, one each per media type) and extraction controls (n = 7 controls) associated with the competition experiments in main Fig. 5a. (Control sample reactions were included in the sequencing library despite absence of visible PCR bands.) Box center lines, edges, and whiskers signify the median, IQR, minima and maxima respectively.

Extended Data Fig. 8. Development of “S-broth” and controls for mock BV-like community growth experiments.

a, Monoculture growth at 48 hours of experimental US (strain “233”) and South African (strains FRESH1 and FRESH2) L. crispatus strains, as well as L. iners, G. vaginalis, Prevotella bivia, Prevotella disiens, and Sneathia sanguinegens strains in various broth media including MRSQ broth + L-Cys (4 mM), NYCIII broth with or without 2% IsoVitaleX^™ plus 5% Vitamin K1-Hemin solution (“IHK”) and/or Tween-80 (1 g/L), and in “S-broth” (see Methods) with or without Tween-80 (1 g/L). S-broth + Tween was used in subsequent experiments. (Detailed strain information is in Supplementary Table 3). Plots depict median ± range for 3 replicates per condition and each plot is representative 1 of ≥2 independent experiments per strain and media condition. b, Bacterial 16S rRNA gene read counts in mock mixed communities (n = 72 individual cultures) and pure strain monocultures (n = 7), with negligible background read counts in blank growth media controls (n = 4 controls, one each per media type) and extraction controls (n = 2 controls) associated with the experiments in Fig. 5c,d. (Control sample reactions were included in the sequencing library despite absence of visible PCR bands.) Box center lines, edges, and whiskers signify the median, IQR, minima and maxima respectively. c, Confirmation of identity of input strains in mock BV-like communities by 16S rRNA gene sequencing of bacterial monoculture controls, prepared from the same input inocula used for the bacterial mock communities shown in Fig. 5c,d.