8-10-week-old female conventional C57BL/6 mice were obtained from Charles River Laboratories (Saint-Constant, QC, Canada). All mice were maintained under specific pathogen-free and standard temperature-controlled conditions that followed a 12-hour light/dark cycle at the Central Animal Facility at McMaster University. Mice were allowed one week after arrival to acclimate prior to experimental use. All mouse studies performed were approved by and complied with the Animal Research Ethics Board (AREB) at McMaster University (AUP 22-03-07).

Bacteria were prepared as per previously published protocols (19). Briefly, Lactobacillus crispatus SJ-3C-US was grown in American Type Culture Collection (ATCC) medium 416 (Lactobacillus MRS broth) at 37°C in anaerobic conditions for 24 hours. Gardnerella vaginalis ATCC 14019 was grown in ATCC medium 1685 (NYC III medium) at 37°C in anaerobic conditions for 24 hours. Bacteria were then washed once with Phosphate-buffered saline (PBS) and resuspended in 25 µL PBS at 10⁷ colony-forming units (CFU) per mouse. To colonize the mice, mice were inoculated intravaginally with 25 µL of the bacteria and then held facedown for one minute to allow for the bacteria to persist in the vaginal canal. To evaluate colonization, vaginal washes were collected from the mice and quantitative plating assays were performed as described previously using the Miles and Misra technique (19, 24). To maintain consistent colonization in the mice, L. crispatus and G. vaginalis were inoculated every 48 hours for ten days.

10 µL of vaginal washes were pipetted onto a glass slide and viewed under a microscope. The cells were observed and compared to images of vaginal washes from mice under different stages of the estrus cycle (25). Briefly, vaginal washes from mice in the estrus stage of their cycle were primarily composed of keratinized epithelial cells and vaginal washes from mice in the diestrus stage were composed primarily of leukocytes.

Mice were colonized with bacteria for ten days before infection. After bacterial administration, if the mice were in diestrus, they were anesthetized with ketamine (150 mg ketamine/kg) and xylazine (10 mg of xylazine/kg). Mice were intravaginally inoculated with 10⁵ PFU of wildtype HSV-2 333 and left on their backs on a heating pad for approximately 45 minutes to one hour to allow the virus to infect. Vaginal washes were collected daily for seven days post-infection, and genital pathology was monitored. Mice were euthanized when they reached a pathology of 4 or 5 (see next section).

As per endpoint monitoring protocols approved by AREB, mice were monitored every day for seven days post-infection and the pathology in the genital area was recorded following a 5-point scale as previously reported (26). The scale is no infection (0), slight genital redness and inflammation (1), genital swelling and redness (2), genital and surrounding area swelling, redness and hair loss (3), genital ulceration with hair loss (4), and ulceration and hair loss in surrounding areas followed by hind limb paralysis (5). Mice were euthanized when they reached high pathology scores of 4 or 5.

30 μL of PBS was pipetted in and out of the vagina seven to eight times. A total of 50-60 μL/mouse was collected by repeating this twice. Vaginal washes were collected for up to seven days post HSV-2 infection and were stored at −80°C until required.

Vero cells were seeded in a 12-well plate and left for 24 h to grow to confluency in α-Minimum Essential Medium (α-MEM), supplemented with 5% fetal bovine serum (FBS), 1% penicillin-streptomycin, 1% L-glutamate and 1% HEPES. Vaginal washes were diluted from 10^-1 to 10^-6 in serum-free α-MEM and added to the Vero cells. Plates were incubated for 2 h and were rocked every 15 minutes to allow the virus to distribute evenly when infecting cells and to prevent the cells from drying out. After the 2 h incubation, α-MEM with FBS was added on top of the monolayers to stop new viral adsorption. The plates were placed in a 37°C incubator for 48 h. After 48 h, the media was aspirated out of each well and the cells were fixed and stained with crystal violet for approximately 15-20 minutes. The plates were then rinsed in water and left to dry overnight. Plaques were counted in the well that had 30-300 plaques under a light microscope and the plaque forming units (PFU) per mL was calculated (27).

Mouse vaginal tissue was collected, placed in cassettes, and fixed in methacarn (60% methanol, 30% chloroform and 10% glacial acetic acid) for 72 h. Cassettes were transferred to 70% ethanol and samples were taken to McMaster Immunology Research Center (MIRC) Histology Core Facility for processing. Briefly, the tissue was embedded in paraffin, and slides were cut and mounted on microscope slides. Slides were then de-paraffinized and stained with a rabbit polyclonal Desmoglein-1 (DSG-1) antibody (Life Technologies, Cat #BS-6725R) as the primary antibody and a goat anti-rabbit IgG Alexa Fluor™ 488 secondary antibody (Invitrogen, Cat #A-11008) with a DAPI counterstain. An IgG isotype control (Santa Cruz Biotechnology, Cat #sc-2027) and a diluent negative control were also included. All samples were imaged on an inverted confocal laser-scanning microscope (Nikon Eclipse Ti2) at 20X magnification. For each experiment, confocal microscope settings for image acquisition and processing were identical between controls and bacteria-inoculated tissue samples. Three randomized images per sample were obtained for analysis using ImageJ software.

After inoculation with VMB species for 10 days, mice were euthanized by cervical dislocation and the vaginal tissue, spleen and iliac lymph nodes (iLN) were collected in cold Roswell Park Memorial Institute (RPMI) media (Invitrogen, Burlington, Canada). Vaginal tissue was cut open with scissors to expose the interior and then cut into smaller pieces. Collagenase A (Cedarlane, Cat #11088793001) was added to 15 mL of RPMI at a mass of 0.00157 g/mL to digest the tissue. The vaginal tissue suspension was put in 50 mL Falcon tubes and placed in a 225 RPM shaker at 37°C for one hour. The supernatant was filtered through a 40 μm filter and collected in a fresh 50 mL Falcon tube on ice. The tissue suspension was again mixed with collagenase and fresh media and digested for another hour. The suspension was then filtered through a 40 μm filter, where the remaining tissue was also pushed through the filter using the back of a syringe. All samples were centrifuged at 1500 RPM at 4°C for 10 minutes. The supernatant was discarded, and the pellet was resuspended in 200 μL of RPMI media. For spleen and iLN analysis, 40 μm filters were placed in a 6-well plate and 6 mL of RPMI media was added. The spleen or iLN was added to the filter and the back of a syringe was used to break apart the tissue. The spleen or iLN suspension was collected in a 15 mL Falcon tube and the well was washed with another 6 mL media and added to the Falcon tube. The samples were centrifuged for 5 minutes at 1500 RPMs at 4°C. The iLN pellets were resuspended in 500 μL of RPMI. Only the spleen samples were further treated with 2mL Ammonium-Chloride-Potassium lysis buffer for 4 minutes at room temperature to lyse red blood cells. Samples were washed with 10 mL PBS and placed into the centrifuge for 5 minutes at 1500 RPMs at 4°C to obtain a pellet of cells. The spleen pellets were resuspended in 6 mL of RPMI media. Cells from all samples were then counted using a hemocytometer.

Staining was done on 1x10⁶ iLN and spleen cells and all cells from the vaginal tissue due to lower yields in 200 μL volumes. Cells were stimulated with a cell stimulation cocktail of ionomycin, PMA, brefeldin A, and monensin (eBioscience™ Cell Stimulation Cocktail, Thermo Fisher Scientific, Cat #00-4975-03) overnight for 16 h as recommended by the manufacturer’s protocol. Samples were incubated with Fc block (rat anti-mouse CD16/32, BD Biosciences, Cat #553142) for 20 minutes to decrease nonspecific Fc receptor binding. Cells were then stained for cell surface markers at 4°C using the following antibodies at 1:100 dilutions for vaginal samples and 1:200 dilutions for spleen and lymph node samples: CD3 PE-CF594 (Invitrogen), CD4 PerCP Cy5.5 (Invitrogen), CD8 BV786 (Invitrogen), CD44 AF700 (BD Bioscience), CD69 PE-Cy7 (Invitrogen), and CD103 PE (Invitrogen). Cells were incubated with these antibodies for 30 minutes and then washed with PBS. Cells were then stained with the Fixable Viability Dye eFluor™ 780 (Invitrogen) at 1:1000 for 20 minutes at 4°C. To prepare for intracellular staining, cells were permeabilized and fixed using the BD Cytofix/Cytoperm™ fixation/permeabilization kit (BD Biosciences, Cat#555028) according to the manufacturer’s protocol. Cells were then stained for intracellular markers at 4°C using the following antibodies at 1:100 dilutions for vaginal samples and 1:200 dilutions for spleen and lymph node samples: IFN-γ BV421 (Invitrogen), TNF-α FITC (Invitrogen), and IL-10 BV510 (BD Biosciences). The validity of staining was verified by fluorescence minus one (FMO) control. Data were collected by flow cytometric analysis using a Cytoflex flow cytometer system (Beckman Coulter Life Sciences, Indianapolis, USA). 50 μL of cell suspensions for iLN and spleen samples and the entire 200 μL was run for vaginal tissue samples. Results were analyzed using FlowJo software (Tree Star, Ashland, USA). For the vaginal tissue, the absolute count was reported as # cells/VAG (vaginal tract) and was calculated as the direct count reported by the flow cytometer, since the entire vaginal tissue was stained and run on the flow cytometer. For the iLN and the spleen, the absolute count was reported as # cells/mL and was calculated by dividing the count from the flow cytometer by 0.05 mL to get the per mL cell count (the volume of cell suspension that was run on the machine).

Vaginal washes were collected from mice ten days after inoculation with bacteria. Washes were analyzed for cytokines and chemokines using the 31-Plex Mouse Cytokine/Chemokine Discovery Luminex Assay from Eve Technologies (Calgary, Canada), as per the manufacturer’s protocol.

All statistical analysis was done using GraphPad Prism version 9.4.1 (GraphPad Software, San Diego, USA). Two-way ANOVA with Tukey’s multiple comparisons was used to determine statistical significance for quantitative plating assays and plaque assays. HSV-2 survival data was analyzed using a Log-rank (Mantel-Cox) test. A one-way ANOVA with Tukey’s multiple comparisons was used to analyze mean fluorescence intensity from histology, all flow cytometry experiments, and multiplex experiments.

To elucidate the cause-effect changes eubiotic and dysbiotic bacteria have on the vaginal microenvironment and susceptibility to HSV-2, a small animal model that can consistently be colonized with bacteria for a short period was required. In a previous study, we reported that L. crispatus and G. vaginalis were able to persist in the mouse vaginal tract for on average 2.6 and 1.75 days, respectively (19). Since these values are close to 48 h and 72 h, these inoculation intervals were used to determine if consistent colonization could be maintained in the model. Mice were inoculated twice 48 h or 72 h apart with the eubiotic bacteria L. crispatus, the dysbiotic bacteria G. vaginalis, or PBS as a negative control. L. crispatus and G. vaginalis loads were evaluated with quantitative plating assays, a method which we have previously validated (19).

When mice were inoculated with PBS every 48 h ( Figure 1A ) or every 72 h ( Figure 1B ), no exogenous bacteria were detected, as expected. All mice inoculated with exogenous bacteria L. crispatus or G. vaginalis had detectable bacterial counts in the quantitative plating assays, indicating successful colonization ( Figures 1C–E ). Mice inoculated with L. crispatus every 48h had a high number of bacteria between the first and second inoculation, with none of the mice showing a period with low exogenous bacterial load, indicating consistent colonization ( Figure 1C ). In mice inoculated with G. vaginalis every 48 h, all except one showed a high exogenous bacterial count 24 h after the first inoculation ( Figure 1E ). 3/5 successfully colonized mice in this group showed declining G. vaginalis load at 48 h post-inoculation ( Figure 1E ). One mouse did not show any exogenous bacterial counts after the first inoculation, indicating unsuccessful colonization, which aligns with the lower rates of successful G. vaginalis colonization we have reported before (19). When L. crispatus was inoculated every 72 h, there was notably less consistency in colonization, and three mice had a period greater than 24 h with no colonization before the second inoculation ( Figure 1D ). Similarly, when mice were inoculated every 72 h with G. vaginalis, of the 5/6 mice colonized after the first inoculation, 3/5 mice had a period of more than 24 h with no bacteria detected ( Figure 1F ). Given the increased consistency in colonization when mice were inoculated every 48 h compared to 72 h, we proceeded with an inoculation regimen of every 48 h for this study. In this study, mice needed to be colonized with bacteria for a long enough duration of time to elicit a change in the vaginal microenvironment. Therefore, we maintained colonization for ten days (five inoculations every 48 hours). When mice were inoculated for ten days following this regimen, they consistently showed high counts of exogenous bacteria indicating short-term colonization ( Supplementary Figure S1 ).

After determining the best inoculation regimen in the human microbiota-associated mouse models, mice were infected with HSV-2 following colonization with L. crispatus or G. vaginalis to determine if differential outcomes were seen in these mice. Clinically, a dysbiotic VMB colonized with BV-associated bacteria such as G. vaginalis has been shown to increase susceptibility to HSV-2 in women (13). To elucidate if this could be recapitulated in our model, mice were inoculated with L. crispatus, G. vaginalis, or PBS every 48 h for ten days (five inoculations) and on day ten, infected with a sublethal dose of wildtype HSV-2 333 at 10⁵ PFU. Administration of depot-medroxyprogesterone acetate (DMPA), a progestin-based contraceptive, before HSV-2 infection is a common practice to thin the vaginal epithelium to make mice more susceptible to HSV-2 (28), however, mice were not colonized with exogenous bacteria following DMPA administration (data not shown). Due to this, mice were staged to be in diestrus on the day of infection, the period when the epithelial barrier is thin during the normal mouse reproductive cycle (29). While not as successful as DMPA pre-treatment, this model of HSV-2 infection has previously been validated (30). The presence of bacteria following HSV-2 infection was quantitated with plating assays 24 h and five days post-infection. No bacteria were detected in the vaginal tract 24 h post-infection, indicating the virus was detrimental to vaginal bacterial growth ( Supplementary Figure S2 ). At five days post-infection, no endogenous bacteria were detected, indicating the infection may be disrupting the ability for recolonization as well ( Supplementary Figure S2 ).

Survival and pathology were recorded, and vaginal washes were collected up to seven days post-infection to determine viral titers. Cumulative data from three independent experiments showed that G. vaginalis inoculated mice had significantly decreased survival rates compared to L. crispatus and PBS groups ( Figure 2A ). All mice survived in the PBS group, 1/17 mice did not survive in the L. crispatus group, and 9/19 mice succumbed to infection in the G. vaginalis group ( Figure 2A ). From the cumulative data, G. vaginalis inoculated mice also had significantly higher viral titers at two days post-infection compared to other groups ( Figure 2B ). When looking at the cumulative pathology, G. vaginalis inoculated mice also had increased scores compared to other groups ( Table 1 ). Individual mouse data for pathology ( Figure 2C ) and viral titers ( Figure 2D ) from one independent experiment representative of the three experiments are also included. In this experiment, 4/7 mice in the G. vaginalis group developed severe lesions and pathology, whereas no mice in the L. crispatus and PBS-negative control groups developed pathology over 7 days of monitoring ( Figure 2C ). When viral titers were examined in the vaginal washes, all mice in the G. vaginalis group showed high viral shedding, whereas only 3/7 mice in the PBS group and 2/6 mice in the L. crispatus group had high viral titers ( Figure 2D ). All measures indicated G. vaginalis was associated with increased HSV-2 infection compared to PBS and L. crispatus inoculated groups. L. crispatus did not appear to increase or decrease susceptibility to HSV-2 in vivo compared to PBS controls.

After determining that G. vaginalis increased susceptibility to HSV-2 in the model, potential mechanisms as to why this is the case were explored. As the epithelium plays a vital role in susceptibility to HSV-2 (29), changes in the expression of an important epithelial barrier protein, Desmoglein-1 (DSG-1), were investigated. DSG-1 is expressed in vaginal tissue and decreased expression of this protein has previously been implicated with increased HSV-2 infection in mice (31). To determine if the VMB plays a role in DSG-1 expression, mice were inoculated every 48 h with L. crispatus, G. vaginalis, or PBS as a negative control for ten days (five inoculations). The vaginal tissue was collected on day ten and fixed and stained for DSG-1. When comparing the groups, G. vaginalis inoculated mice had significantly decreased expression of DSG-1 compared to L. crispatus and PBS groups ( Figure 3 ). This was demonstrated by the decreased green fluorescent staining for DSG-1 in the G. vaginalis group compared to the L. crispatus and PBS groups ( Figure 3A ). When the mean fluorescence intensity was analyzed, there was significantly decreased expression in the G. vaginalis group compared to L. crispatus and PBS groups ( Figure 3B ). L. crispatus did not alter the expression of DSG-1 compared to the PBS control ( Figures 3A, B ).

To further elucidate why G. vaginalis may be increasing susceptibility to HSV-2 in vivo, changes in the immune microenvironment were evaluated through flow cytometry analysis. Mice were inoculated with L. crispatus, G. vaginalis, or PBS every 48 h for ten days (five inoculations). The vaginal tissue, spleen and iliac lymph nodes, which drain the reproductive tract, were collected from mice on day ten after inoculations and processed and stained for flow cytometry. The gating strategy is depicted in Supplementary Figure S3 . No significant differences were seen in the cells from the spleen or the lymph nodes, indicating that the effect of vaginal bacterial colonization was site-specific and limited to local tissues ( Supplementary Figure S4 ).

Since T cells have been shown to play a critical role in protection against HSV-2, our analysis was focused on these cells (32, 33). Vaginal cells were first gated on CD4+CD3+ T cells. Inoculation with L. crispatus or G. vaginalis did not significantly alter the proportions or absolute number of CD4+ T cells ( Figure 4A ), indicating these bacteria did not affect the overall numbers of CD4+ T cells in the vaginal mucosa. Next, the quality of T cell response in the vaginal tract following inoculation with different bacteria was evaluated by examining CD4+ T cells expressing the tissue-resident marker CD103 (34), the tissue-resident and activation marker CD69 (35), and the tissue-resident and adhesion marker CD44 (36). G. vaginalis inoculated mice had significantly decreased absolute numbers of CD44+CD4+ T cells ( Figure 4B ), CD69+CD4+ T cells ( Figure 4C ), and CD103+CD4+ T cells ( Figure 4D ) compared to L. crispatus inoculated mice. L. crispatus inoculated mice had significantly increased proportions of CD69+CD4+ T cells compared to the PBS group. All of these markers are mucosal tissue residency markers, signifying there was a decrease in tissue-resident helper T cells in the vaginal immune environment in the G. vaginalis inoculated mice compared to L. crispatus. Decreased numbers of resident CD4+ T cells in the vaginal tissue could be linked to the increased susceptibility to HSV-2 in the G. vaginalis inoculated mice.

Next, the CD8+CD3+ T cell compartment was examined ( Figure 5A ). Similar to the CD4+ T cells, there was not a significant change in the CD8+ T cells in the G. vaginalis or L. crispatus inoculated mice compared to the PBS group, indicating the bacterial inoculations did not significantly alter the total amount of CD8+ T cells in the vaginal mucosa. Once again, cells were further gated for mucosal markers including CD44+CD8+ T cells, CD69+CD8+ T cells, and CD103+CD8+ T cells. When compared to the PBS group, G. vaginalis inoculated mice had significantly decreased percentages and trends towards decreased absolute numbers of CD44+CD8+ T cells ( Figure 5B ). There was also significantly decreased absolute counts of CD69+CD8+ T cells ( Figure 5C ). There were trends toward decreased absolute counts of CD103+CD8+ T cells in the G. vaginalis group ( Figure 5D ). Although not as strong as the data for CD4+ T cells, there was a decrease in mucosal CD8+ T cells in G. vaginalis inoculated mice. This could also contribute to the increased susceptibility to HSV-2 in the G. vaginalis group.

In conjunction with mucosal cell surface markers, cytokine production by T cells was also assessed using flow cytometry to determine functional T cell populations in VMB-administered mice. Mice were inoculated with L. crispatus, G. vaginalis, or PBS as a negative control for ten days (five inoculations every 48 h). Following bacterial inoculation, the vaginal tissue, spleen and iliac lymph nodes were collected from mice on day ten after inoculations and processed and stained for flow cytometry. Cells were gated for CD4+ and CD8+ T cells and then further gated for the proinflammatory cytokines TNF-α and IFN-γ and the regulatory cytokine IL-10. Once again, no significant differences were noted in the cytokine expression by T cells from the spleen or the lymph nodes ( Supplementary Figure S5 ), indicating that the effect of bacterial inoculation was site-specific and limited to the vaginal tract.

G. vaginalis inoculated mice had significantly increased proportions of TNF-α+CD4+ T cells compared to PBS and L. crispatus; however, the absolute numbers were not significantly different, indicating a possible selective enrichment of TNF-α producing CD4+ T cells ( Figure 6A ). There was no significant difference in the TNF-α+CD8+ T cells ( Figure 6A ). G. vaginalis inoculated mice had significantly increased proportions of IFN-γ+CD4+ and IFN-γ+CD8+ T cells compared to the PBS group ( Figure 6B ), but no difference in the absolute counts, once again indicating the possibility of selective enrichment. Lastly, the L. crispatus inoculated group has significantly increased proportions of IL-10+CD4+ T cells compared to PBS and significantly increased absolute counts of IL-10+CD4+ T cells compared to G. vaginalis ( Figure 6C ). L. crispatus inoculated mice also had significantly increased proportions of IL-10+CD8+ T cells compared to PBS and G. vaginalis groups and significantly increased absolute counts compared to G. vaginalis ( Figure 6C ). Overall, these results suggest G. vaginalis inoculation was associated with an inflammatory signature in vaginal T cells while L. crispatus inoculation was associated with regulatory cytokine production by vaginal T cells.

Since the flow cytometric analysis indicated functional differences in T cell populations, we next examined if there were changes in the cytokine levels in the vaginal immune milieu following colonization with different bacteria. A multiplex assay was used to assess cytokine and chemokine levels after bacterial inoculation. Mice were inoculated for ten days (five inoculations 48 h apart) with L. crispatus, G. vaginalis, or PBS as a negative control. On day ten, vaginal washes were collected from mice and a 31-Plex mouse cytokine/chemokine assay was conducted to analyze cytokine levels in the vaginal tract. While the majority of the 31 cytokines and chemokines examined did not show any significant changes following colonization with different bacteria, some key cytokines and chemokines were altered. Overall G. vaginalis was associated with an inflammatory signature in the vaginal canal compared to L. crispatus. In congruence with the flow cytometry experiments, G. vaginalis inoculated mice had significantly increased TNF-α levels in the vaginal secretions compared to L. crispatus ( Figure 7A ). There were also trends towards increased IFN-γ, IL-1α, and IL-1β levels in the G. vaginalis group. L. crispatus had significantly decreased levels of IL-12p70 expression compared to PBS ( Figure 7A ). IL-10 levels were not significantly affected in bacteria-inoculated mice ( Figure 7A ). When the expression of chemokines was examined, G. vaginalis also had increased expression of chemokines that attract inflammatory cells, including a significant increase in LIX, a chemokine known to attract neutrophils (37), and a trend towards increased expression of MIP-1α and MIP-1β ( Figure 7B ). Other cytokines and chemokines from the 31-Plex assay did not show any trends or significance (data not shown).