Microbe-binding Antibodies in the Female Genital Tract: Associations with the Vaginal Microbiome and Genital Immunology

Endogenous bacteria in the female genital tract (FGT), known as the vaginal microbiota, play a major role in shaping the local immune environment and are important determinants of HIV susceptibility (1). Bacterial vaginosis (BV) is a common vaginal condition in reproductive-aged women, which is characterized by a vaginal microbiome containing a high overall bacterial density and an increased diversity of facultative anaerobes, including Gardnerella vaginalis, Prevotella spp., Atopobium vaginae, and Megasphaera spp., with few lactobacilli (2). BV is associated with an increased incidence of numerous adverse reproductive health outcomes, including preterm birth, spontaneous abortion, pelvic inflammatory disease, endometritis, sexually transmitted infections, and HIV acquisition (3–6). In the absence of BV, the vaginal microbiome is dominated by Lactobacillus spp., most often either Lactobacillus crispatus or Lactobacillus iners (7).

BV causes elevations in the vaginal levels of classical proinflammatory cytokines and chemokines, including IL-1α, IL-6, and IL-8, which may increase HIV risk by 1) recruiting HIV-susceptible CD4⁺ T cells to the genital mucosa, and 2) damaging the genital epithelial barrier, thus enhancing viral access to mucosal target cells (8). However, there is considerable interindividual heterogeneity in the extent to which a BV-type microbiome induces cervicovaginal inflammation, such that the levels of IL-1α and IL-1β can vary by >1000-fold in women with BV (9, 10). Furthermore, host immune mechanisms that may enhance colonization or clearance of BV-associated bacteria are not well described.

Recent studies suggest that colonization by BV-associated bacteria, and the degree to which these bacteria induce vaginal inflammation, may be modulated by host immune factors, including the presence and interactions of local Abs (11, 12). At other mucosal sites, such as the gastrointestinal tract, mucosal Abs interact with the local microbiota to maintain homeostasis by facilitating the clearance and/or dampening of the inflammation induced by pathogenic bacteria (13). Whereas IgA is the most abundant Ig at most mucosal sites, IgG is most common in the FGT (14, 15). In the gut, IgA and IgG play distinct roles in controlling inflammation and shaping the local microbiome. Whereas IgA acts through receptor blockage, neutralization of toxins, steric hindrance, and immune exclusion to facilitate bacterial clearance and immune tolerance (16–18), IgG promotes the death and clearance of bacteria through complement pathways and Fc-mediated phagocytosis (19). Less is known about the roles of these Ig classes in the FGT. Although prior studies have demonstrated that genital levels of proinflammatory cytokines and chemokines are correlated with higher levels of total mucosal Abs in the FGT (12) and reduced IgA coating of bacteria (11), whether and how microbe-binding Abs modulate host immune responses or FGT bacterial colonization are not clear.

In this study, we describe and quantify microbe-binding IgA and IgG in the FGT using a flow cytometry–based assay and investigate the associations of these Abs with total genital levels of IgA and IgG, the vaginal microbiota, and the local immune milieu. We specifically examine IgA and IgG responses against G. vaginalis, Prevotella bivia, L. iners, and L. crispatus. G. vaginalis and P. bivia are common BV-associated bacteria associated with cervicovaginal inflammation (10), L. iners is the predominant Lactobacillus spp. in black women and demonstrates frequent transition to BV (7, 20), and L. crispatus is associated with urogenital protection against BV, sexually transmitted infections, and HIV (21, 22). We hypothesized that microbe-binding IgA and IgG would be associated with dampened inflammatory responses and reduced bacterial abundance.

Female sex workers (FSWs) in Nairobi, Kenya were recruited through the Sex Worker Outreach Program clinics and provided informed, written consent for immune and microbial studies. Consenting participants who were not pregnant, breastfeeding, or had a chronic illness (diabetes, rheumatoid arthritis, asthma, tuberculosis, or chirotherapy in the past 6 mo) were enrolled in the study and completed HIV testing (rapid test and HIV GNA GeneXpert, Cepheid), sexually transmitted infection testing, and a behavioral/biological survey. Participants provided blood samples, cervicovaginal secretions (CVSs; SoftCup; Evofem Biosciences, San Diego, CA), and vaginal swabs for immune and microbial studies. Blood plasma was tested for Abs against syphilis and HSV-2, and urine samples were used to screen for pregnancy, Chlamydia trachomatis (GeneXpert assay, Cepheid), and Neisseria gonorrhoeae (GeneXpert assay, Cepheid). BV was diagnosed by Nugent scoring, and Trichomonas vaginalis was diagnosed using the OSOM Trichomonas Rapid Test (SEKISUI Diagnostics). Exclusion criteria included active infection by C. trachomatis, N. gonorrhoeae, or T. vaginalis; sexual violence in the last 7 d; >10 clients in the past 7 d; and a history of female genital mutilation. Sexual violence, >10 clients in the past 7 d, and female genital mutilation were excluded as they might substantially impact immunology and the microbiome. Syphilis was only assessed by RPR (rapid plasma reagin) and was not followed up with a secondary confirmatory test. HSV serology was recorded; however, active ulceration was not recorded. Based on the Nugent score, we enriched 2:1 for participants with either intermediate flora or BV (n = 134, defined as a Nugent score of 4–10) versus participants with normal flora (n = 66, defined as a Nugent score of 1–3).

The current study was reviewed and approved by the Kenyatta National Hospital Ethics and Research Committee (approval no. P778/11/2018), the London School of Hygiene and Tropical Medicine Ethics Committee (approval no. 16629), and the University of Toronto (approval no. 37046).

CVSs were collected by SoftCup, weighed, diluted 1:10 in sterile PBS, and centrifuged at 1730 × g for 10 min. CVS weight ranged from 0.092 to 2.652 g, with the median weight being 0.367 g (interquartile range [IQR]: 0.230–0.5441 g). Supernatants were extracted, and the remaining pellets were resuspended in 500 μl of sterile PBS. Both the supernatants and pellet were frozen at −80°C and transported to the University of Toronto for analysis.

Prediluted CVS supernatants were further diluted 1:10 in PBS with 1% Tween 20 and 10% BSA and centrifuged at 10,000 × g for 10 min. Total IgA and IgG levels were quantified using human IgA and IgG (total) uncoated ELISA kits (Invitrogen) according to the manufacturer’s instructions. The lower limit of quantification (LLOQ) was determined using the lowest value from the standard curve across all plates analyzed and applied to all values below the LLOQ. Only samples above the upper limit of quantification (ULOQ) were diluted and reanalyzed. Internal laboratory controls were used as interassay controls.

The microbe-binding Ab protocol was adapted from Moor et al. (23) and is depicted in Fig. 2. CVS supernatants were diluted 1:2 with FACS buffer (2% BSA, 5 μM EDTA-PBS) and heat-inactivated for 30 min at 56°C and centrifuged at 16,000 × g to remove pelleted debris. Then, 50 μl of serially diluted (1:2 dilutions in FACS buffer) CVS supernatant was combined with an equal volume of bacteria (∼5 × 10⁵ CFU/ml) and incubated for 1 h at 4°C before unbound Abs were washed off with FACS buffer (4000 × g, 10 min, 4°C). Ab-bound bacteria were then stained with secondary Abs conjugated with the following fluorochromes: goat anti-human IgG F(ab′)₂ Ab (AF594, Jackson ImmunoResearch, 109-586-098) and goat anti-human IgA F(ab′)₂ Ab (AF647, SouthernBiotech, 2052-31) for 30 min before unbound Abs were washed off with FACS buffer (4000 × g, 10 min, 4°C). Ab-bound bacteria were fixed with 4% paraformaldehyde for 20 min before being analyzed on an LSRFortessa (Becton Dickinson) with settings optimized for bacterial cell detection. Internal laboratory controls were included for each assay to ensure interassay reproducibility. Flow cytometry analysis of samples was performed using FlowJo (Tree Star), with representative gating strategies presented in Fig. 2C. Geometric mean fluorescence intensity (MFI) values were extracted for each sample at each dilution. Bacteria were incubated without clinical samples and stained with secondary Abs to create a negative control that was subsequently subtracted from all clinical sample measurements. Sample dilution factors and their corresponding MFI values were log₁₀ transformed and plotted with a linear regression using GraphPad Prism to produce regression lines for each sample. Negative log₁₀-transformed values were assigned a zero value. Responses were reported as the area under the curve (AUC) value for each regression line. IgG⁺IgA⁺, IgA⁺IgG⁻, IgA⁻IgG⁺, and IgA⁻IgG⁻ events were reported from the lowest dilution/highest concentration sample for each participant.

After clinical samples were heat inactivated, they were separated into two aliquots. One aliquot was combined with an equal volume of bacteria (∼5 × 10⁵ CFU/ml), and the second aliquot was combined with equal volumes of the FACS buffer; both were incubated for 1 h at 4°C. Afterwards, the bacteria/sample mixture was centrifuged at 16,000 × g for 10 min at 4°C and the supernatant was extracted. Both aliquots were then analyzed using the microbe-binding Ab assay (see above) to detect microbe-binding Abs against a second bacterium.

L. crispatus (American Type Culture Collection, 33820) was grown in de Man, Rogosa, and Sharpe (MRS) agar, P. bivia was grown on tryptic soy agar with 5% sheep’s blood (Hardy Diagnostics), L. iners (American Type Culture Collection, 55195) was grown in New York City III agar, and G. vaginalis (American Type Culture Collection, 14019) was grown in New York City III agar. All bacteria were grown anaerobically (80% N₂; 10% CO₂; 10 H₂) at 37°C for 46–48 h. Representative biomass was scrapped from freshly grown plates and resuspended in PBS at an ∼OD₅₉₀ of 1.0. Bacteria were frozen in 100-μl aliquots of frozen glycerol stocks and used in the microbe-binding Ab assay. For use in the microbe-binding assay, bacteria were thawed and washed with PBS at 8000 × g for 10 min. Pelleted bacteria were diluted to the concentration of 10⁸ CFU/ml and stained with CFSE (Thermo Fisher Scientific) at a concentration of 1:1000 for 30 min. CFSE-stained bacteria were then washed with FACS buffer twice (4000 × g, 10 min, 4°C) before resuspending at the concentration of 10⁷ CFU/ml in PBS.

DNA was isolated from 250 μl of CVS pellet using a DNeasy PowerSoil pro kit (Qiagen). DNA was eluted in 60 μl of the Qiagen elution buffer and analyzed through real-time quantitative PCR (RT-qPCR) and 16S rRNA amplicon sequencing.

Extracted CVS DNA was analyzed through 16S rRNA amplicon sequencing on an Illumina MiSeq. The V4 hypervariable region of the 16S rRNA gene was amplified using uniquely barcoded 515F (forward) and 806R (reverse) sequencing primers to allow for multiplexing (24). Amplification reactions were performed using 12.5 μl of KAPA2G Robust HotStart ReadyMix (KAPA Biosystems), 1.5 μl of 10 μM forward and reverse primers, 7.5 μl of sterile water, and 2 μl of DNA. The V4 region was amplified by cycling the reaction at 95°C for 3 min, 18 cycles of 95°C for 15 s, 50°C for 15 s, and 72°C for 15 s, followed by a 5-min 72°C extension. All amplification reactions were analyzed in duplicate to reduce amplification bias, pooled, and checked on a 1% agarose TBE gel. Pooled duplicates were quantified using PicoGreen and combined by even concentrations. The library was then purified using AMPure XP beads and loaded onto the Illumina MiSeq for sequencing, according to the manufacturer’s instructions (Illumina, San Diego, CA). Sequencing was performed using the V2 (150 bp × 2) chemistry. A single species (pseudomonas aeruginosa DNA), a mock community (Zymo microbial community DNA standard D6305), and a template-free negative control were included in the sequencing run. The Qiime2 analysis package was used for sequence analysis, and the following functions were accessed from within the Qiime2 package (25): the quality of the sequencing run was first examined using FastQC and MultiQC (26); Cutadapt was used, following the default settings, to remove sequences with high errors rates (27); paired-end sequences were assembled, and quality trimmed using vsearch –fastq_mergepairs (28, 29) following default settings, with a –fastq_truncqual set at 2, a maxee set at 1, and minimum and maximum assemble lengths set at 250 and 255 (+2 and −3 bp from the expected sequence length of 253 bp); assembled sequences were subjected to an additional filtering step, utilizing the quality-filter function in Qiime2; the resulting high-quality data were then processed following the deblur pipeline. Sequences were clustered into amplicon sequence variant (ASV) groups and singleton sequences were removed. Taxonomy assignment was executed using the Qiime2 classify-hybrid-vsearch-sklearn function and the Average ReadyToWear trained Silva database version 138.1 (30, 31). ASVs with an abundance <0.01% were removed to reduce the potential for observing bleed-through ASVs, and ASVs identified as contaminating chloroplast or mitochondria were removed. A phylogenetic tree was created using the SEPP function available through Qiime2 (32). The speciateIT tool was used to further annotate several operational taxonomic units (OTUs) to the species level. OTUs were only classified to the level compatible with Valencia (classification method for vaginal microbial communities based on composition) (33); as such, several OTUs were only classified to the genus levels, whereas major vaginal microbiome components were classified to the species levels when possible. Valencia was then used to classify participants to their respective vaginal community state types as previously described (33). Women classified as community state type IV were designated as molecular BV (molBV)–positive, and all other classifications were designated as molBV-negative.

The extracted CVS DNA was analyzed using TaqMan-based RT-qPCR through the QuantStudio 6 Flex real-time PCR system (Thermo Fisher Scientific) in either a single-plex or multiplex assay. The protocol for quantification of L. crispatus and L. iners absolute abundance with multiplex qPCR and G. vaginalis absolute abundance single plex were adopted from previously reported assays (34, 35), whereas P. bivia absolute abundance was quantified using primers designed using the pipeline outlined in Schneeberger et al. (36). The total reaction volume for assays was 10 μl. Assays for P. bivia, L. crispatus, L. iners, and total bacterial abundance (16S) were performed at 95°C for 10 min, 45 cycles at 95°C for 15 s, and then at 60°C for 1 min. Assays for G. vaginalis were performed at 95°C for 10 min, 45 cycles for 15 s, and then at 55°C for 1 min. Data analysis was performed with QuantStudio real-time PCR software version 1.3 (Applied Biosystems) and the Thermo Fisher Connect platform. Laboratory-grown pure cultures of bacteria and their quantified DNA (ng/ml; Qubit, Thermo Fisher Scientific) were used to create standard curves and to determine the LLOQ, defined as the lowest technical duplicates with quantifiable Ct values. All values were transformed to account for sample dilutions and concentrations at or below the LLOQ were set to the LLOQ value. All values above the LLOQ were normalized to CVS sample volume and reported as nanograms of DNA per milliliter of CVS.

CVS supernatants were thawed and recentrifuged at 2000 rpm for 5 min. The following soluble immune factors were assayed: IL-1α, IL-1β, IL-6, IL-8, MCP-1, TNF, MIP-1α, MIP-1β, MIP-3α, IP-10, MIG, soluble E-cadherin (sE-cad), and MMP-9 in duplicate by Multiplex MSD according to the manufacturer’s instructions (Meso Scale Discovery, Rockville, MD). sE-cad was a marker of epithelial disruption in the genital tract, and MMP-9 induced epithelial disruption (37, 38). CVS supernatants were plated at 25 μl per well and a standard curve was used to determine the lower and upper limit of detection, as well as the concentration of each analyte (pg/ml). The ULOQ and LLOQ were determined by the lowest value and highest value, respectively, across all plates analyzed. Samples that were below the LLOQ were designated the LLOQ value, and samples above the ULOQ were assigned the ULOQ value.

Data analysis was performed using either R (version 4.2.2) or GraphPad Prism (version 9.0.2). Soluble immune factor concentrations, Ab concentrations, and absolute abundance microbial components were log₁₀ transformed for analysis. When analyzing microbe-binding AUC values between groups, we used an unpaired Welch t test and Mann–Whitney U test for normalized AUC values. Proportional (%) values such as IgA⁺, IgG⁺, and IgA⁺IgG⁺ were treated as ordinal and analyzed using the Mann–Whitney U test. A Pearson correlation was used to analyze correlations of soluble immune factors, Abs (total and microbe-binding), and microbiome composition. Additionally, a Pearson correlation using the R package ppcor was used to control for levels of total bacteria. Analyses were not adjusted for multiple comparisons.

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to the need to protect participant confidentiality and safety. Our commitment to participant confidentiality and safety is detailed in our participant information sheet and consent forms, as approved by three ethics committees: Kenyatta National Hospital (P778/11/2018), The London School of Hygiene and Tropical Medicine (16229), and the University of Toronto (37046).

This study included 200 FSWs from Nairobi, Kenya (see Materials and Methods for exclusion criteria) with deliberate enrichment of participants by Nugent score. Full demographic data are shown in Table I. Participants had a median age of 30 y. The participants had a median of 5 y in sex work, with a median of three clients in the past week. All participants had abstained from vaginal sex for at least 24 h. More than half (52.5%) of the participants reported vaginal washing within the last 30 d. Three participants (1.5%) were seropositive for syphilis based on RPR (rapid plasma reagin) testing.

First, we measured the total IgA and IgG concentration in CVSs from all participants. One participant (0.5%) had levels of IgA below the limits of quantification, and 15 (7.5%) participants had levels of IgG below the limits of quantification, with these participants being assigned the value of the LLOQ. The mean concentration of IgA was log₁₀ 4.944 ± 0.598 pg/ml and the mean concentration of IgG was log₁₀ 4.736 ± 0.994 pg/ml, and within participants, the total cervicovaginal concentration of IgA was higher than IgG (log₁₀ difference of +0.209, paired t test p = 0.004; Fig. 1).

IgA and IgG concentrations were not significantly associated with age, age of first vaginal sex, age when the participant started selling sex, the number of years selling sex, days since last vaginal sex, clients in the last 7 d, condom use, contraceptive method, HSV status, or vaginal douching. Total concentrations of IgA and IgG were substantially lower among women with Nugent BV than among women with either intermediate flora or normal flora (IgA: mean difference of BV versus intermediate flora, log₁₀ −0.581 ± 0.102 ng/ml [p < 0.0001]; mean difference of BV versus normal flora, log₁₀ −0.354 ± 0.089 ng/ml [p = 0.0003]; IgG: mean difference of BV versus intermediate flora, log₁₀ −1.083 ± 0.155 ng/ml [p < 0.0001], mean difference of BV versus normal flora, log₁₀ −1.054 ± 0.138 ng/ml [p < 0.0001]).

Having demonstrated that the vaginal microbiota has an important effect on total levels of IgA and IgG, we next quantified the binding of cervicovaginal IgA and IgG to four key vaginal bacteria, that is, L. crispatus, L. iners, G. vaginalis, and P. bivia, using a flow cytometry–based method (Fig. 2; see Materials and Methods). We incubated four serial dilutions (1:40, 1:80, 1:160, 1:320) of CVSs with each cultured bacterium and then stained them with fluorophore-labeled secondary Abs specific for either human IgA or IgG and analyzed each by flow cytometry. MFI values were recorded for each serial dilution (Fig. 3A for representative plots; see Supplemental Fig. 3 for all flow cytometry plots).

Microbe-binding Abs were common among the cohort, with most participants having both IgA and IgG that recognize all four bacteria (Fig. 3B). However, the distribution of the MFI responses did not indicate a clear-cut threshold to define a response. Notably, among participants with similar ranges of either total IgA or IgG, we observed at least a 1-log variance in the microbe-binding IgA and IgG MFI responses to any of the four bacterial species. Participants with total vaginal IgG concentrations below the limits of quantification generally exhibited low microbe-binding responses against all four taxa (most within the lower 10% percentile of all MFI values), except one participant who demonstrated moderate responses against L. iners and L. crispatus (MFI of 1774.7 and 647.6, respectively). The one participant with IgA below the limits of quantification, suspected to have IgA deficiency, had very low microbe-binding responses against all taxa (all MFI values <3). Additionally, we observed differences in the Ab-bound proportion of each species. G. vaginalis, P. bivia, and L. crispatus were bound by IgA at a higher proportion than IgG (median difference of IgA⁺ and IgG⁺ bacteria of 47.48, 41.15, and 20.75%, respectively), whereas L. iners was bound by a slightly greater proportion of IgG than IgA (median difference of IgA⁺ and IgG⁺ bacteria of −0.233%; Fig. 3C). BV-associated bacteria P. bivia and G. vaginalis were more commonly bound by only IgA Abs, whereas L. iners and L. crispatus were more commonly bound by both IgA and IgG (Fig. 3D).

We assessed the degree of binding using the AUC value generated by the series of MFI values measured for each sample. The AUC values are strongly correlated with MFI values (all Spearman ρ > 0.895 and p < 0.0001) and highly reproducible across experiments (coefficient of variation < 10% for MFI values >50). The mean IgA AUC response to G. vaginalis, P. bivia, L. iners, and L. crispatus were 2.183 ± 0.514, 2.212 ± 0.521, 2.867 ± 0.571, and 2.682 ± 0.472, respectively, whereas the mean IgG AUC response was 1.600 ± 0.621, 1.586 ± 0.793, 2.854 ± 0.496, and 2.281 ± 0.829 (Table II).

Overall, microbe-binding IgA and IgG AUCs were positively correlated with each other and with total IgA or IgG concentrations (all IgA r > 0.631, p > 0.0001; all IgG r > 0.713, p < 0.0001, Fig. 4A, 4B). Because microbe-binding IgA and IgG responses to different bacteria can only be compared relatively, we used z score and hierarchical clustering to cluster the microbe-binding IgA and IgG AUC responses within the cohort (Fig. 4C, 4D). We observed a spectrum of responses ranging from strong responses against all four bacteria to low responses against all four bacteria, with some participants demonstrating heterogeneity in bacterial responses, suggesting that these responses are not purely dependent on total Ab concentrations.

We conducted a preliminary analysis of whether mucosal IgA and IgG in the FGT can bind to more than one bacterial target (i.e., polyreactive). Using a subset of 20 participants with relatively high (upper quartile) IgA and IgG responses against both P. bivia and L. crispatus, we preadsorbed samples with L. crispatus and analyzed the residual P. bivia–binding IgA and IgG responses, and vice versa (Supplemental Fig. 1A). While preadsorption decreased both L. crispatus and P. bivia–binding IgA and IgG responses (Supplemental Fig. 1B, 1C), a substantial proportion of mucosal IgA and IgG response remained unadsorbed. Furthermore, there was significant heterogeneity between participants in the extent to which IgA and IgG responses were adsorbed away.

In summary, cervicovaginal IgA and IgG frequently bind the key vaginal bacteria species G. vaginalis, P. bivia, L. iners, and L. crispatus, with interindividual heterogeneity and demonstration of microbe specificity.

We then evaluated whether BV was associated with differences in microbe-binding Ab responses as well as total Ab concentration. Specifically, we compared total Ab and microbe-binding Ab responses associated with molBV using 16S rRNA sequencing of CVS samples and the Valencia classification program: 123 participants were identified as having molBV, whereas 75 participants were identified as having normal flora. Two participants had an insufficient number of reads (<1000 reads) to ascertain molBV status using Valencia and were subsequently excluded from all BV-specific analyses.

First, we compared the concentration of total IgA and IgG between women with and without molBV. Women with molBV had an ∼33% reduction in total IgA (Fig. 5A; molBV positive, log₁₀ 4.861 ± 0.624 pg/ml; molBV negative, log₁₀ 5.066 ± 0.532 pg/ml; p = 0.0148) and an 80% reduction total IgG (Fig. 5A; molBV positive, log₁₀ 4.465 ± 1.138 pg/ml; molBV negative, log₁₀ 5.149 ± 0.452 pg/ml; p < 0.0001) when compared with women without molBV. Of note, all participants with IgG below the limit of quantification had molBV, and those with quantifiable IgG levels were less likely to have molBV (odds ratio: 0 [confidence interval: 0–0.369], p < 0.0006). Even with the exclusion of undetectable IgG values, total IgG levels were lower among participants with molBV (p = 0.0001) than participants without molBV. There was relative enrichment of IgA in women with molBV (Fig. 5B), with an IgA/IgG ratio of 1.71 (IQR: 0.56–8.29) versus 0.77 (IQR: 0.51–1.50) (p < 0.0001) in those who were molBV negative. Women with molBV had higher levels of IgA than IgG (comparison with 1.0 ratio, p < 0.0001), and women without molBV had comparable levels of IgA and IgG (p = 0.5199).

Next, we compared IgA and IgG AUC responses to the four key vaginal bacteria analyzed in our flow cytometry assay: G. vaginalis, P. bivia, L. crispatus, and L. iners. In keeping with the reduced overall Ab concentrations, women with molBV also had lower levels of IgA binding to G. vaginalis and L. crispatus, a trend toward lower levels of IgA binding to P. bivia (p = 0.0754), and decreased IgG binding to all four taxa measured (all p < 0.0001, Fig. 5C). When binding was normalized for the reduced total IgA and IgG concentrations, women with molBV demonstrated higher microbe-binding IgA AUCs for all four bacterial species, as well as higher IgG responses against G. vaginalis, L. crispatus, and L. iners (Fig. 5D).

We then assessed whether the presence and absolute abundance of the four key vaginal bacteria in the vaginal microbiome were associated with their respective microbe-binding IgA and IgG responses. In this study, the four key vaginal bacteria were quantified using qPCR and normalized based on CVS volume. Two participants were excluded from qPCR analyses because their CVS volumes were not recorded. P. bivia was detectable by qPCR above the LLOQ in 38.9% (77/198) of participants, L. crispatus in 22.7% (45/198) of participants, L. iners in 80% (156/198) of participants, and G. vaginalis in 93.4% (185/198) of participants. Instead of demonstrating distinct associations with the detectability (presence/absence defined by assay LLOQ) of their respective species, microbe-binding IgA and IgG exhibited overall class-specific associations with all four microbes analyzed: all analyzed microbe-binding IgG responses were lower in the presence of G. vaginalis and L. iners, whereas they were elevated in the presence of L. crispatus. Only L. iners–binding IgA was reduced in the presence of detectable L. iners in the vaginal microbiome, whereas all other microbe-binding IgA did not exhibit associations with the presence or absence of any species (Table III).

Next, we assessed whether microbe-binding IgA and IgG demonstrated a dose-dependent relationship with the abundance of their respective bacterial species. The absolute abundance of detectable P. bivia, L. iners, and L. crispatus was generally not correlated with microbe-binding IgA and IgG responses, except for a weak correlation between L. crispatus–binding IgG and L. iners abundance, and between G. vaginalis–binding IgG and P. bivia abundance (Table IV). The abundance of G. vaginalis was negatively correlated with all measured microbe-binding IgA and IgG responses, likely due to its collinearity with total bacterial abundance (Supplemental Table I). When we control for total bacterial abundance (Supplemental Table II), all correlations between microbe-binding IgA and IgG and G. vaginalis abundance resolve. This was true for many vaginal bacteria, as they coassociate with each other and with total bacterial abundance. Overall, our findings suggest that microbe-binding IgA and IgG responses were generally not associated with the presence or abundance of their corresponding microbe.

We next evaluated the relationship between microbe-binding IgA and IgG and the absolute abundance of key vaginal bacteria. We quantified the total bacterial abundance based on 16S copies measured by qPCR. In keeping with our finding that molBV was associated with reduced total IgA and IgG levels, total levels of IgA and IgG, as well as all measured microbe-binding IgA and IgG responses, were strongly negatively correlated with total bacterial abundance (Fig. 6; IgA, r = −0.438, p < 0.0001; IgG, r = −0.539, p < 0.0001; Table IV). The relationship between total Ig concentration and bacterial abundance was primarily driven by participants with molBV (n = 123; IgA, r = −0.596, IgG, r = −0.531, both p < 0.0001). In those without molBV, there was no significant correlation between total IgA/IgG and overall bacterial abundance (n = 75; IgA, r = 0.069; IgG, r = −0.159), regardless of whether participants were dominated by either L. crispatus or L. iners.

The molBV was strongly associated with elevated levels of the proinflammatory cytokines IL-1α, IL-1β, and TNF, as well as elevated levels of epithelial disruption markers sE-cad and MMP-9, and with decreased levels of the chemokines IP-10, MCP-1, MIG, and MIP-3α (Supplemental Fig. 2A). High bacterial abundance is a hallmark of molBV and correlated with similar soluble immune factors. Specifically, total bacterial abundance correlated with elevated vaginal levels of sE-cad, IL-1α, and IL-1β, and with reduced IL-6, IP-10, MCP-1, MIG, MIP-1α, MIP-1β, and MIP-3α (Supplemental Fig. 2B). Therefore, subsequent analyses of microbiome–immune associations controlled for total bacterial abundance.

We hypothesized that microbe-binding IgA and IgG would have anti-inflammatory properties; therefore, we next assessed their correlation with both proinflammatory cytokines and chemokines, using partial correlations to control for total bacterial density. Higher concentrations of total IgA and IgG were correlated with elevated levels of the proinflammatory cytokines (IL-6, TNF) and chemokines (IL-8, IP-10, MCP-1, MIG, MIP-1α, MIP-1β, MIP-3α; Fig. 7A). Total IgA correlated with elevated IL-1β and markers of epithelial disruption (sE-cadherin and MMP-9), whereas total IgG was inversely correlated with vaginal levels of IL-1α. Next, we assessed whether there were taxonomically specific immune effects of microbe-binding IgA and IgG. Once again, controlling for total bacterial abundance, we generally observed a positive correlation with inflammatory cytokines (IL-1β, IL-6, TNF) and chemokines (IL-8, IP-10, MCP-1, MIG, MIP-1α, MIP-1β, MIP-3α), and inversely correlated with IL-1α (Fig. 7B). Of note, microbe-binding IgA and IgG responses did not associate with sE-cadherin, and L. crispatus–binding IgA and IgG and L. iners–binding IgG had the strongest inverse associations with IL-1α (all r > 0.215, p > 0.01). In summary, total and microbe-binding IgA and IgG correlate with elevated levels of proinflammatory cytokines and chemokines.

In the gut, mucosal Abs interact with luminal bacteria to shape the local microbiome composition and modulate the host immune response (13). This can occur through mechanisms including, but not limited to, immune exclusion, where bacteria are coated by Abs to prevent colonization and inflammation-inducing interactions with the host epithelium (39). Limited evidence from the genital tract suggests that microbe-coating IgA may limit vaginal bacterial diversity and modulate inflammation in the context of BV (11). In this cross-sectional study, we use a flow cytometry–based assay to quantify total and microbe-binding IgA and IgG in CVSs. We demonstrate that vaginal IgA and IgG are abundant and bind the key vaginal species G. vaginalis, P. bivia, L. iners, and L. crispatus with microbe specificity. Importantly, the total density of vaginal bacteria was increased during molBV and was strongly correlated with reduced concentrations of both total and microbe-binding IgA and IgG. When normalized for absolute bacterial density, total and microbe-binding IgA and IgG were correlated with elevated levels of several proinflammatory cytokines and chemokines previously linked to HIV acquisition.

A key finding was the decrease in both total IgA and IgG concentrations and microbe-binding IgA and IgG, in the context of high bacterial burden, specifically during molBV. Recent studies using IgA/IgG sequencing (IgA/IgG-seq), such as those by Murphy et al. (11), showed increased IgA coating of vaginal bacteria following BV treatment. Similarly, Schuster et al. (40) observed that IgA and IgG coating was lowest among study participants with BV. IgA/IgG-seq analyzes the relative proportion of endogenous bacteria coated by host IgA and IgG but does not account for the total bacterial abundance (41). Therefore, a lower ratio of coated bacteria assessed by IgA/IgG-seq could either represent a reduced concentration of microbe-binding Abs, an increased absolute bacterial density, or a combination of both. Likewise, our microbe-binding flow cytometry assay measures the microbe-binding capacity of the unbound fraction of mucosal IgA/IgG, and so reductions during molBV could represent adsorption by the increased bacterial load. However, both our flow-based assay and previous IgA/IgG-seq FGT studies are consistent with the hypothesis that, as in the gut, microbe-binding IgA/IgG play a role in the clearance of inflammatory bacteria, and thus the reduction observed in our studies and those of others reflects a decreased capacity to eliminate these bacteria, resulting in higher bacterial loads. In observational studies, the direction of this relationship cannot be defined.

In the gut, the mucosal IgA response consists of both highly specific, high-affinity binding IgA and polyreactive, low-affinity binding IgA (17). The low-affinity IgA can bind a fixed repertoire of bacteria taxa, likely due to conserved surface Ags such as insulin, LPS, carbohydrates, or DNA, whereas the high-affinity binding IgA is taxonomically specific (42). Our study provides initial evidence that in the FGT, a substantial proportion of mucosal IgA and IgG does not demonstrate polyreactive binding to both P. bivia and L. crispatus, as preadsorption with one bacterium only partly abrogates microbe-binding IgA and IgG responses to the other. Although we did not extensively assess the binding capacity of these Abs against all common vaginal bacteria, this suggests that vaginal microbe-binding IgA and IgG responses are likely predominantly taxonomically specific. Future studies will be important to further explore the degree to which cervicovaginal Abs are comprised of high-affinity Abs, and whether these Abs preferentially modulate colonization or inflammation.

Microbe-binding IgA in the gut typically targets the most abundant bacterial taxa and adapts to changes in bacterial composition (43). Our results do not support the hypothesis that the presence of bacterial species induces vaginal microbe-binding Ab responses or that there is a dose-dependent response to individual bacterial taxa. However, our cross-sectional study format limits our ability to define the vaginal microbiota at prior time points when microbe-binding Abs may have been induced. The vaginal microbiome varies over time (44), with changes linked to factors such as the menstrual cycle, initial bacterial community composition (BV- or L. iners–dominated communities are less stable) and sexual activity (44); the latter two factors may play particularly important roles in our cohort of FSWs, deliberately enriched for the presence of BV. While not possible with the IgA sequencing method, our microbe-binding Ab assay allows us to investigate Ab responses against bacteria not present at the time of sampling. As such, we observed women with high L. crispatus–binding IgA and IgG responses despite undetectable vaginal L. crispatus. This may be explained (in part) by polyreactive Abs, previous colonization with L. crispatus, or repetitive challenges through sexual partners. Prospective studies during microbiota flux or probiotic treatments should investigate the induction of these Ab responses after microbe exposure, and the relationships between exposure, microbe-binding Abs, and microbial persistence or clearance.

There are several limitations to our study. First, our assay only assesses the ex vivo binding of the unbound fraction of IgAs and IgGs collected from CVSs. However, similarities observed between our method and IgA sequencing provide confidence in our results and that biological phenomena are analyzable through our assay. Second, we only used a single laboratory strain of each key vaginal bacteria species, potentially resulting in less antigenic diversity or Ags that are not native to this cohort. However, the existence of such taxonomic specificities to different strains or clinical isolates remains unclear. Third, the stage of the menstrual cycle was not recorded (although samples were not collected during menses), and the menstrual stage can impact Ig levels, microbiota composition, and genital immunology (44, 45). Fourth, our study was focused on FSWs in Kenya, potentially limiting generalizability to women of different ethnicities or those in monogamous relationships. Among black women, there is a low prevalence of non–iners Lactobacillus–dominated microbiota, specifically L. crispatus, L. gasseri, and L. jensenii, and therefore our results might not comprehensively address the interactions between these bacteria and the local immune environment or contrast the differences that exist between Lactobacillus species. Furthermore, the heightened exposure to diverse bacteria from multiple sexual partners could alter Ig responses in unknown ways, making it important to reproduce our findings in other populations.

This study suggests that cervicovaginal Abs may be important modulators of the vaginal microbiota and genital immunology, with potential implications for HIV prevention. We describe the performance characteristics of using a flow cytometry–based assay to quantify microbe-binding Abs, showing that cervicovaginal IgA and IgG are abundant and bind with microbe specificity to the key vaginal bacteria G. vaginalis, P. bivia, L. iners, and L. crispatus. Cervicovaginal IgA and IgG correlate with reduced total bacterial abundance and increased proinflammatory cytokines and chemokines linked to HIV acquisition. Cervicovaginal microbe-binding IgA and IgG do not appear to be immunoregulatory but may indirectly mitigate bacteria-induced inflammation by reducing total bacterial abundance. However, it is unclear whether inducing these Abs through either vaccines or Ig therapy would be beneficial for HIV prevention; further studies are needed to evaluate casual relationships and the biological significance of our findings in prospective clinical studies, potentially utilizing probiotics and antibiotics to induce microbial changes.

The authors have no financial conflicts of interest.