This study was approved by the Kenyatta National Hospital/University of Nairobi Ethics and Review Committee (protocol number P203/03/2022). All the study volunteers signed an informed consent form prior to their enrolment into the study. The study recruited 139 adult women (70 participants with HIV-1 and 69 participants without HIV-1) aged between 18 and 50 years within Nairobi County in Kenya. Of the women with HIV-1, 68 had been on antiretroviral therapy for an average of 89 months and 2 did not provide information on their treatment status. The study excluded pregnant women, women who were currently on or within 3 days of the end of their menstrual period, women who have had either protected or unprotected vaginal intercourse within the past 24 h, women on antibiotics within the past 3 days, or women with any condition that could have precluded provision of informed consent, compromised the volunteer’s health during the conduct of this study, or interfered with achieving the study objectives.

HIV-1 viral load analysis was performed for the participants with HIV-1 using a GeneXpert machine (GeneXpert machine Dx system, Cepheid) according to the manufacturer’s instructions. Briefly, venous whole blood collected in a 4-mL K3 BD EDTA vacutainer tube (BD vacutainer^®) was spun at 1,800 × g for 10 min at room temperature. One milliliter of plasma was then pipetted into a GeneXpert HIV-1 viral load cartridge (GeneXpert^® technology) and loaded into the GeneXpert machine for a run of approximately 2 h. The HIV-1 viral load was displayed in copies/mL of plasma with the limit of detection on the GeneXpert machine being 20 copies/mL and the limit of quantification being 40 copies/mL.

CD4 counts and CD8 counts were analyzed for the participants with HIV-1 using the BD FACS Count instrument according to the manufacturer’s instructions. Briefly, venous whole blood collected in a 4-mL K3 BD EDTA vacutainer tube (BD vacutainer^®) was mixed well by inverting the tube repeatedly. The mixed blood (50 µL) was pipetted into each tube of the reagent tube pair tab (one tube for CD4 PE/CD3 PE-Cy™5 analysis and the second tube for CD8 PE/CD3 PE-Cy5 analysis) previously vortexed for approximately 5 s downwards and 5 s upwards. This was incubated for 1 h at room temperature and in the dark upon vortexing for 5 s. Fixative solution [50 µL; 5.0% formaldehyde and 1.76% methanol in 1× phosphate-buffered saline (PBS)] was then added into each of the tubes and the samples were vortexed for 5 s prior to running the tubes on the BD FACS Count instrument within 48 h of preparation.

CVM samples were collected by a trained study nurse/clinician using a softcup (Softcup^®) following insertion into the vaginal vault for 1 h. The softcup was removed and put in a 50-mL falcon tube and placed on ice for transportation to the laboratory within 2 h. The CVM sample was then centrifuged at 1,160 × g for 10 min at 4°C immediately upon receipt in the laboratory. The supernatant CVM was transferred into a 1.5-mL microcentrifuge tube via a positive displacement pipetting system. Samples were then centrifuged again at 17,000 × g for 15 min at 4°C to separate the clear mucus from other cervicovaginal secretions. Approximately 90 µL of the supernatant (clear CVM) was then transferred into a clean 2-mL vial and stored at 4°C for later use in the viral mobility assay. The pH was measured using a 5-µL aliquot on the Baker-pHix Universal pH Indicator Stick (Avantor - J.T.Baker^®, Germany). The remaining samples (both supernatant and pellet) were mixed with an equal volume of 1× protease inhibitor and mixed well by pipetting up and down several times. The samples were centrifuged again and 2 aliquots of the supernatant (50 µL each) were put in 2-mL vials for storage. Both the supernatant aliquots and the remaining pellet sample were frozen at −80°C.

Concentrations of CVM IgA, IgG1, IgG2, IgG3, IgG4, and IgM were assessed using a Bio-Rad Bio-Plex™ Pro Human Isotyping Assays kit according to the manufacturer’s instructions. Data were acquired on a Luminex 200 reader with the immunoglobulin concentration reported in nanograms per milliliter upon interpolation from the standard curve concentrations (ng/mL) and the median fluorescence intensities. Intra-assay precision was calculated from the median fluorescence intensity of standard dilution points from duplicate wells with an acceptable coefficient of variation of less than 20%. Inter-assay precision was calculated from the median fluorescence intensity of standard dilution points from three different plates with an acceptable coefficient of variation of less than 20% ( Supplementary Figure 1 ).

A cotton swab was used to swab the vaginal walls of a participant and a smear was made on a glass slide. Gram staining procedure was performed as per Smith and Hussey’s protocol (10). Air-dried, heat-fixed smear of cells on a glass slide was flooded for 1 min with crystal violet staining reagent before washing the slide in a gentle and indirect stream of tap water for 2 s. The smear was then flooded with a mordant, Gram’s iodine, for 1 min followed by washing in a gentle and indirect stream of tap water for 2 s. Next 95% (vol/vol) ethanol was added drop by drop to the slide to decolorize it before flooding the slide with a counterstain, safranin, for 30 s. The slide was then washed in a gentle and indirect stream of tap water until no color appeared in the effluent and then blot dried with an absorbent paper. The slide was examined under oil immersion using a brightfield microscope. BV diagnosis was done using the Nugent scoring system as proposed by Robert P. Nugent in 1991 where a score of 0 to 3 is considered BV negative, a score of 4 to 6 is considered intermediate, and a score of 7 or higher is considered BV positive (11).

Deoxyribonucleic acid (DNA) was extracted from the frozen CVM pellet samples using the DNeasy^® Blood & Tissue kit (Qiagen). The samples were first equilibrated to RT and then mixed well. Two hundred microliters of this suspension and 10 µL of an internal control (IC) DNA provided in the Sacace Multiplex Real Time PCR kit (Sacace technologies) were added into a 1.5-mL centrifuge tube containing 20 µL of QIAGEN proteinase. Buffer AL (200 µL) was then added. This mixture was then pulse-vortexed for 15 s and incubated at 56°C for 10 min. Two hundred microliters of 96% ethanol was added to the sample and mixed by pulse-vortexing for 15 s. The mixture was carefully applied to the kit-provided DNeasy Mini spin columns placed in a 2-mL collection tube and centrifuged at 8,000 rpm for 1 min. The DNA was washed twice with consecutive addition of 500 µL of buffer AW1 and AW2 as per the manufacturer’s instructions. The DNA was eluted with 200 µL of elution buffer AE into a clean 1.5-mL tube.

To test for infection with Neisseria gonorrhoeae, Chlamydia trachomatis, Mycoplasma genitalium, and Trichomonas vaginalis, a real-time multiplex PCR reaction was done using a multiplex real-time PCR kit for the detection of N. gonorrhoeae, C. trachomatis, M. genitalium, and T. vaginalis (Sacace technologies) following the manufacturer’s instructions: A 25-µL PCR reaction mix for each sample was prepared by the addition of 10 µL of PCR-mix-1-FL N. gonorrhoeae/C. trachomatis/M. genitalium/T. vaginalis, 5 µL of PCR-mix-2-FRT, 0.5 µL of polymerase (TaqF), and 10 µL of the DNA sample/positive control/negative controls. The data were acquired on a 6-plex rotor gene real-time PCR thermocycler. The amplification protocol had a 15-min hold at 95°C and two cycling sets. The first cycling set had five cycles of 95°C for 5 s, 60°C for 20 s, and 72°C for15 s. Fluorescence was recorded at the second step of cycle 2 having 40 cycles with the same temperature conditions as the first cycling set.

The hormones were measured using competitive chemiluminescent emission assay on an automated Cobas e 411^®, a Roche 411 series as per the manufacturer’s instructions. Briefly, whole blood was collected in 5-mL BD serum separation tubes and approximately 1 mL of serum was obtained upon centrifugation at 1,800 × g for 10 min. Serum (500 µL) was pipetted into a sample cup and loaded into the Cobas analyzer, after which the concentrations of the hormones in the sample were obtained in picograms per milliliter.

Comparisons among sample groups were made using the non-parametric Kruskal–Wallis test with Dunn’s multiple comparisons to compare the mean rank of each group with the mean rank of every other group. Non-parametric Mann–Whitney U test was used for comparison between sample groups while a multiple linear regression analysis was used to describe multiple associations. The multiple associations were adjusted for the participant’s age, serum concentrations of estrogen and progesterone, the CVM pH, BV, and STIs including HIV-1. Normality distribution was tested using the Shapiro–Wilk test while the level of significance was tested at 95% confidence interval (p < 0.05). All statistical analyses were performed on GraphPad Prism software version 10.4.2.

A total of 139 adult women were enrolled in the study, of whom 70 participants had HIV-1 and 69 participants had no HIV-1. Of the participants with HIV-1, 42 were negative for BV, 6 were intermediate, and 22 were positive for BV. In the group of participants who were negative for HIV-1, 41 were negative for BV, 6 were intermediate, 20 were positive, and 2 samples were not of sufficient quality to obtain a valid result. A total of 37 participants with HIV-1 and 34 participants without HIV-1 reported that they were using hormonal contraception while 33 participants with HIV-1 and 32 participants without HIV-1 were not on any contraceptive. One of the participants with HIV-1 and two of the participants without HIV-1 reported to be using condoms as a way of contraception. Four of the participants with HIV-1 and five of the participants without HIV-1 were positive for N. gonorrhoeae. A total of 11 participants with HIV-1 and 18 participants without HIV-1 were positive for C. trachomatis and 5 participants with HIV-1 and 4 participants without HIV-1 were positive for M. genitalium. A total of 20 participants with HIV-1 and 17 participants without HIV-1 tested positive for T. vaginalis ( Table 1 ). Of the participants with HIV-1, 43 had undetectable viral load (<20 copies per mL), 19 had detectable but unquantifiable viral load (between 20 and 40 copies per mL), while 8 had detectable and quantifiable viral load with a median of 236 viral copies per mL of plasma. Of the participants with HIV-1, 68 were on antiretroviral treatment for a median time of 87.5 months. One participant with HIV-1 reported to have defaulted from treatment while one did not give any information on her treatment status. The median CD4 and CD8 counts for the participants with HIV-1 were 709 and 727, respectively ( Table 1 ).

Immunoglobulin subtypes and subclasses have distinct biological effector functions; therefore, variations in the antibody composition in CVM may have an impact on the CVM barrier function. We therefore sought to assess immunoglobulin composition and concentration in the CVM using a Luminex assay utilizing a 6-plex kit for detection of six immunoglobulin isotypes/subclasses (IgA, IgG1, IgG2, IgG3, IgG4, and IgM). In general, there were significant differences (p < 0.0001) in the concentrations of the different immunoglobulin isotypes and subclasses in the CVM samples by Kruskal–Wallis test ( Figure 1A ). Upon comparison between each immunoglobulin concentration using a two-tailed Mann–Whitney U test, we observed that IgG2 had the highest concentrations in comparison to IgA (p = 0.0373) and in comparison to all the other immunoglobulins (p < 0.0001). The concentrations of IgA and IgG1 were comparable. However, both IgA and IgG1 concentrations were higher in comparison to the concentrations of IgG3, IgG4, and IgM (p < 0.0001). IgG3 concentrations were significantly higher compared to those of IgG4 (p = 0.0155). There was no difference between the concentrations of IgG3 and those of IgM. However, the concentrations of IgM were significantly higher relative to those of IgG4 (p < 0.0001) ( Figure 1B ). Differential CVM immunoglobulin isotype and subclass concentrations were subsequently shown to associate with clinically relevant variations in the local CVM microenvironment.

The FRT mucosa is the primary portal of entry for STIs including HIV-1. The microbiome at this pivotal site plays an important role in shaping protective barrier properties as microbial dysbiosis has been associated with increased risks of HIV-1 acquisition (12). As immunoglobulins form part of the CVM protective component, we sought to assess the effect of microbial dysbiosis on the abundance of the measured immunoglobulin isotypes and subclasses among the BV status across the whole cohort using the Kruskal–Wallis test with Dunn’s multiple comparisons test. We found that CVM IgG1 concentrations were significantly lower in BV-positive samples relative to BV-negative (p < 0.0001) and BV-intermediate (p = 0.0291) samples ( Figure 2A ). This was independent of HIV status but negatively associated to the CVM sample pH ( Figure 3 ). This finding suggests an increased susceptibility of IgG1 degradation under BV conditions, which may be associated with increased concentrations of degradative enzymes or from elevated pH environments associated with BV. Additionally, there were trends towards increased IgA concentrations in BV-intermediate (p = 0.0511) and BV-positive (p = 0.0605) CVM samples compared to BV-negative samples, suggesting a protective role in response to the microbial dysbiosis ( Figure 2A ). We also noted a trend towards higher concentrations of IgG3 in CVM samples from participants with HIV-1 (p = 0.0677) relative to the samples from participants without HIV-1. Although this is not statistically significant in our results, it may suggest a possible elevation of IgG3 expression in HIV-1 infection, which may be due to HIV-1-associated hypergammaglobulinemia (13) ( Figure 2B ). We did not see any difference in the concentrations of the tested immunoglobulins across the CVM samples from women with any of the tested STIs relative to those without any of the tested STIs ( Supplementary Figure 2 ).

The cervicovaginal local microenvironment and some host factors could potentially impact the presence and abundance of the different immunoglobulin isotypes and subclasses in the CVM. To understand these nuances better, we performed a multiple linear regression analysis to assess the potential impact of factors that could affect immunoglobulin concentrations in CVM samples, including participant’s age, use of contraceptives, HIV-1, BV and other STI status, pH, and concentrations of serum estrogen and progesterone ( Figure 3 ). The β coefficient values were standardized by dividing the unstandardized values with the standard deviation of the data points from the predicted values (denoted as Sy.x in the GraphPad Prism regression model). The unstandardized and standardized β coefficient values are shown in Supplementary Tables 1, 2 . Of the data combined, we found significant, independent positive associations between IgA and M. genitalium infection (p = 0.0197, β = 0.946) and pH (p ≤ 0.0001, β = 0.875). pH was also negatively associated with the concentrations of IgG1 (p = 0.0104, β = −0.341). In section 3.3, we showed decreased concentration of IgG1 in BV-positive CVM samples relative to the BV-intermediate and -positive samples. However, no associations between the IgG1 concentrations with BV were detected, but rather only negative associations with CVM pH. This observation suggests that CVM pH may have more impact in the modulation of IgG1 concentrations in CVM as opposed to the increased concentrations of degradative enzymes associated with BV. We also found that being BV intermediate or positive was independently negatively associated with IgG3 (p = 0.0066, β = −1.595 and p = 0.0013, β = −2.710, respectively) and IgG4 (p = 0.0302, β = −1.266 and p = 0.0156, β = −2.0200, respectively). The negative associations of IgG3 and IgG4 with BV suggest possible degradation of these immunoglobulins with BV-associated degradative enzymes. IgM was positively associated with N. gonorrhoeae infection (p = 0.0408, β = 0.7992), suggesting possible induction of IgM expression by N. gonorrhoeae infection. Our results suggest modulations of the humoral immunity in the CVM by the local microbial dysbiosis and STIs.

Immunoglobulins in CVM work by binding pathogens and potentially limiting their motion and accessibility to their target cells (14). In addition, a previous study demonstrated that immunoglobulins may neutralize pathogens by binding to their epitopes and disabling the infection of their target cells (2). To better understand this phenomenon, we performed multiple linear regression analysis adjusted for HIV infection status, age, pH, BV status, concentrations of sex hormones, contraceptive use, and STIs to evaluate the relationships between the concentrations of the immunoglobulin isotypes and subclasses and the mobility of several T/F HIV-1 strains in CVM samples. The β coefficient values were standardized as explained in section 3.4. The unstandardized and standardized β coefficient values are shown in Supplementary Tables 3, 4 . Although we did not notice associations between any of the tested immunoglobulins with the clade B T/F HIV-1 strain CH040 and the clade B laboratory adapted R9Bal, we did observe a significant negative association between concentrations of IgG3 and the mobility of the non-depleted clade C T/F HIV-1 CAP045 (p = 0.0419, β = −0.000001177). Additionally, we found that IgG1 concentrations had a negative correlation with the VRC01 depleted set of the circulating recombinant clade 92TH023 (p = 0.0113, β = −0.000002997) ( Figure 4 ). On the other hand, the mobility of the non-depleted set of the recombinant clade A/E 92TH023 virions had a positive association (p = 0.0086, β = 0.00001745) with the concentrations of IgG4 in the CVM samples ( Figure 4 ). Our results indicate viral mobility hindrance in the CVM, possibly by immunoglobulin binding to different regions of the envelop glycoprotein, which may be strain dependent.