Fifteen Lactobacillus strains (L. crispatus BC1, BC3–BC8; L. gasseri BC9–BC14; and L. vaginalis BC16, BC17), isolated from vaginal swabs of healthy premenopausal women (Parolin et al., 2015), were cultured overnight at 37°C in anaerobic jars containing Gaspak EZ (Becton Dickinson) in modified medium. This modified medium contained 75% Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco BRL, Carlsbad, CA, United States), supplemented with 15% fetal bovine serum (FBS), sodium pyruvate at 1 mM, non-essential amino acids at 1 mM, and 25% de Man, Rogosa, and Sharpe (MRS) broth (Difco, Detroit, MI, United States) supplemented with 0.05% L-cysteine. The turbidity levels of overnight cultures were adjusted to an optical density conversion factor (OD_{600 nm}) of 0.5, corresponding to a cell concentration of 10⁸ colony forming units (CFU)/mL. Culture media were centrifuged at 4,000 × g for 10 min at 4°C and then filtered through a 0.22 μm membrane filter. The resultant Lactobacillus-conditioned medium (CM) was then used to treat tissue explants infected by HIV-1. Lactobacillus-cell pellets (CP) were washed in sterile saline solution (0.9% NaCl supplemented with 0.05% L-cysteine) and resuspended in antibiotic-free modified medium.

Human cervico-vaginal tissue explants obtained from routine hysterectomy (National Disease Research Interchange, Philadelphia, PA, United States) and tonsillar tissue (Children’s National Medical Center, Washington, DC, United States) were dissected and cultured as described in Grivel and Margolis (2009) with slight modifications. Briefly, the tonsillar and mucosa layers from ecto- and endo-cervix tissues were cut in blocks of 2 mm³. Eighteen cervico-vaginal tissue blocks were infected with 0.4 mL of viral stock HIV-1_BaL (120 ng/mL p24_gag obtained from the Virology Quality Assurance Laboratory at Rush University, Chicago, IL, United States) for 2.5 h at 37°C in agitation. After infection, tissue blocks were washed three times with phosphate-buffered saline (PBS) and transferred at the liquid–air interface onto Gelfoam (nine blocks per well) in a 12-well plate containing RPMI 1640 medium at 1 mL/well supplemented with 15% FBS, sodium pyruvate at 1 mM, non-essential amino acids at 1 mM, gentamicin sulfate at 50 μg/mL, amphotericin B at 2.5 μg/mL. Twenty-seven tonsillar tissue blocks (nine blocks per well in 3 mL of RPMI 1640 medium supplemented as above) were placed on collagen sponge gels, and tissue blocks were infected with 7.5 μL of viral stock, on top of each block. Cervico-vaginal and tonsillar tissue were incubated at 37°C for 12 days, with replacement of culture medium every 3 days. 3TC (lamivudine at 10 μM) was used as a positive control for HIV-1 inhibition.

Tonsillar tissues were colonized with 15 vaginal Lactobacillus strains (27 blocks per condition) at a starting inoculum of 10⁴ CFU/mL. At day 3 after inoculation with bacteria, all tissue blocks were collected and digested with collagenase IV (5 mg/mL; Gibco BRL) for 30 min with agitation in a Thermomixer at 900 rpm at 37°C. Following digestion, tissue cells were filtered with 100 μm cell strainers (Corning) and washed with 50 mL of PBS. Cells were then suspended in 1 mL of PBS and stained with 1 μl of live/dead Fixable Viability Dye eFluor 450 (EF 450, Invitrogen) for 15 min. After incubation, cells were washed and diluted in staining buffer [PBS, 1% normal mouse serum, 1% normal goat serum, 1 mM ethylenediaminetetraacetic acid (EDTA)] and stained with anti-CD3-allophycocyanin (CD3-APC) for 20 min. After surface staining, cells were permeabilized with the Fix&Perm Cell Fixation and Cell Permeabilization Kit (Invitrogen) then stained for 20 min with anti-Bcl2-PE, a mitochondrial anti-apoptotic antigen. Data were acquired with a Novocyte flow cytometer (ACEA Biosciences, CA, United States) equipped with 405, 488, and 640 nm laser lines using NovoExpress version 1.2.4 software (ACEA Biosciences) and analyzed using the same software.

Cervico-vaginal and tonsillar tissue blocks were cultured in Lactobacillus-CM from six Lactobacillus strains (L. crispatus BC3, BC5; L. gasseri BC12, BC13; and L. vaginalis BC16, BC17), obtained as described above. Tissue blocks were pre-incubated with Lactobacillus-CM undiluted and diluted 1:5 with normal medium for 2 h before HIV-1 infection. After HIV-1 infection, tissue cultures were kept in the same medium (undiluted or diluted 1:5) for the next 3 days of culture, then the medium was replaced with complete RPMI 1640 medium every 3 days, and the culture was kept until day 12.

We carried out virucidal experiments by pre-treating HIV-1 with Lactobacillus-CP or Lactobacillus-CM. HIV-1 viral suspensions at 400 μL were mixed with 100 μL of Lactobacillus-CP (stock 5 × 10⁸ CFU/mL), corresponding to a final concentration of 10⁸ CFU/mL, or 100 μl of undiluted Lactobacillus-CM (corresponding to a final 1:5 dilution) or with 100 μL of normal medium (experimental control condition). Cultures under these three experimental conditions were then incubated for 60 min at 37°C and centrifuged at 4,000 × g for 10 min at 4°C. Supernatants were used to infect cervico-vaginal tissue, as described above.

We quantified titers of lactate isomers D and L from overnight-cultured Lactobacillus-CM using a lactate quantification assay kit according to the manufacturer’s instructions (BioAssay Systems, EFDLC-100 and EFLLC-100). Isomers D (3 mM), L (23 mM), and D + L (3 mM + 23 mM), corresponding to the average titers found in all undiluted Lactobacillus-CM, were tested for HIV-1 inhibition in human cervico-vaginal and lymphoid tissues. Isomers D and L at concentrations corresponding to those found in dilution 1:5 were also tested in lymphoid tissues. pH values in all Lactobacillus-CM, undiluted or diluted 1:5, were measured as well. Furthermore, in order to evaluate the effect of low pH on HIV-1 replication in tissues ex vivo, as measured in Lactobacillus-CM (undiluted average around pH 4 and diluted 1:5 up to pH 6.9), we evaluated HIV-1 infectivity in ex vivo tissue at pH 4 and pH 6.9, buffering the medium with hydrochloric acid (HCl).

We evaluated HIV-1 replication on tissue by measuring the levels of p24_gag in tissue culture medium using a dynamic immunofluorescent cytometric bead assay as described by Biancotto et al. (2009).

We performed all statistical analyses using ANOVA test GraphPad Prism version 7 (GraphPad Prism Software Inc., San Diego, CA, United States). Results were deemed significant for p-values < 0.05.

Fifteen Lactobacillus strains belonging to L. vaginalis, L. gasseri, and L. crispatus were evaluated for their capacity to colonize ex vivo tissue blocks. All lactobacilli colonized tissue explants with similar kinetics, reaching maximum colonization (approximately 10^8.5 CFU/mL) after 3 days of culture, and then plateaued for the entire 12 days of culture duration (Figure 1A). Cell depletion in tissue explants by lactobacilli was evaluated 3 days after bacterial inoculation (Figure 1B). Colonization of explants with 6 out of 15 Lactobacillus strains, L. crispatus (BC3, BC5), L. gasseri (BC12, BC13), and L. vaginalis (BC16, BC17), did not result in cell depletion, as compared with control (Figure 1B, lower panel, BC5 representative of this group). In contrast, the colonization of tissue blocks by the remaining nine Lactobacillus strains, L. crispatus (BC1, BC4, BC6, BC7, BC8) and L. gasseri (BC9, BC10, BC11, BC14), resulted in a loss of T (CD3⁺) cells as well as an increase in the expression of the apoptotic marker Bcl2 (data not shown). The losses of CD3⁺ cells were not characteristics of particular species of Lactobacillus, as some strains of L. crispatus, L. gasseri, and L. vaginalis induced cell depletion while others did not. Lactobacillus strains that induced CD3⁺ cell depletion were not used in further experiments.

To investigate the effects of metabolites secreted by lactobacilli on HIV-1 replication in ex vivo tissues, tissue blocks were pre-incubated with Lactobacillus media conditioned by L. crispatus (BC3, BC5), L. gasseri (BC12, BC13), and L. vaginalis (BC16, BC17), infected with HIV-1, and cultured as described in Section “Materials and Methods.” In both cervico-vaginal and lymphoid tissues undiluted Lactobacillus-CM suppressed replication of HIV-1 compared with the control by 91.9 ± 4.3 and 98.3 ± 1.4% (L. crispatus BC3, p < 0.0001, n = 5), 95.9 ± 4.8 and 97.7 ± 1.8% (L. crispatus BC5, p < 0.0001, n = 5), 93.5 ± 3.6 and 98.2 ± 1.9% (L. gasseri BC12, p < 0.0001, n = 5), 91.9 ± 1.5 and 98.3 ± 2.3% (L. gasseri BC13, p < 0.0001, n = 5), 95.9 ± 5.0 and 98.1 ± 2.9% (L. vaginalis BC16, p < 0.0001, n = 5), 85.8 ± 11.7 and 95.0 ± 6.5% (L. vaginalis BC17, p < 0.0001, n = 5), respectively (Figure 2). Lactobacillus-CM had an inhibitory effect on HIV-1 replication even when diluted fivefold. Depending on the Lactobacillus strain, inhibition of HIV-1 replication by such diluted medium was ranging from 44.3 ± 31.7% (L. crispatus BC3, p = 0.0038, n = 5) to 77.3 ± 7.3% (L. vaginalis BC16, p < 0.0001, n = 5) in cervico-vaginal tissue (Figure 2B) and from 55.5 ± 13.2% (L. vaginalis BC17, p < 0.0001, n = 5) to 93.1 ± 5.2% (L. crispatus BC5, p = 0.0001, n = 5) in tonsillar tissue (Figure 2D).

In order to understand whether Lactobacillus-CM can suppress HIV-1 infectivity before interaction with tissues, we pre-incubated HIV-1 for 1 h with Lactobacillus-CM (diluted 1:5) and tested HIV-1 infectivity in cervico-vaginal tissues ex vivo. We studied the virucidal capacities of Lactobacillus-CM of six strains: L. crispatus (BC3, BC5), L. gasseri (BC12, BC13), and L. vaginalis (BC16, BC17). As shown in Figure 3A, HIV-1 replication was reduced when cervico-vaginal tissues were infected with HIV-1 pre-treated with Lactobacillus-CM from L. crispatus BC3 (47.7 ± 7.0%, p = 0.005, n = 5), L. crispatus BC5 (60.9 ± 13.8%, p = 0.0005, n = 5), L. gasseri BC12 (64.0 ± 11.4%, p < 0.0001, n = 5), and L. vaginalis BC16 (57.4 ± 8.1%, p = 0.003, n = 5) (Figure 3B). No statistically significant inhibition was observed due to Lactobacillus-CM from L. gasseri BC13 (28.6 ± 6.6%, p = 0.13, n = 5) and L. vaginalis BC17 (31.1 ± 24.8%, p = 0.06, n = 5).

To investigate whether lactic acid produced by lactobacilli is responsible for HIV-1 inhibition, we measured the concentrations of lactic acid isomers D and L in Lactobacillus-CM (Table 1). Depending on the strain, the concentrations of lactic acid isomers D and L ranged from 1.8 to 3.6 mM and from 9.0 to 24.7 mM, respectively. L. gasseri BC12 was the strain that produced the highest concentrations of both isomers, while L. crispatus BC17 was the strain that produced the lowest concentrations. Next, we tested the effects of lactic acid isomers at the concentration found in undiluted Lactobacillus-CM or CM diluted 1:5 on HIV-1 replication in tissues ex vivo (Figure 4). As shown in Figures 4A,C, lactic acid isomers D (3 mM), L (23 mM), and D + L (3; 23 mM) significantly reduced HIV-1 replication in both cervico-vaginal and tonsillar tissues. We found that D lactate (3 mM) inhibited HIV-1 replication by 48.2 ± 6.2% in cervico-vaginal (p = 0.0004, n = 3) and by 57.6 ± 33.2% in tonsillar (p = 0.0125, n = 3) tissue cultures, while L lactate (23 mM) suppressed HIV-1 replication by 94.3 ± 5.5% (p < 0.0001, n = 3) and by 99.3 ± 21.9% (p < 0.0001, n = 3) in cervico-vaginal and tonsillar tissues, respectively. The mixture of D + L lactate suppressed HIV-1 replication by 92.1 ± 7.7% in cervico-vaginal and by 94.4 ± 30.4% in tonsillar tissue (p < 0.0001, n = 3) (Figures 4B,D). Afterward, we evaluated the effect of lactic acid isomers at the concentrations found in fivefold-diluted Lactobacillus-CM on HIV-1 replication in lymphoid tissue (Figures 4C,D). We found that isomer D did not inhibit HIV-1 replication while isomer L and the mixture of isomers D + L significantly reduced HIV-1 replication in tonsillar tissues by 67.8 ± 0.8% (isomer L, p = 0.0033, n = 3) and by 56.5 ± 9.5% (isomers D + L, p = 0.0142, n = 3) (Figure 4D).

Furthermore, we evaluated whether the effect of lactobacilli on HIV-1 replication is due to the acidic pH of the Lactobacillus-CM. As shown in Table 1, pH values of undiluted Lactobacillus-CM ranged from 3.8 to 4.6 and of fivefold-diluted Lactobacillus-CM from 6.3 to 6.9. To mimic the effect of pH on HIV-1 replication, we acidified the culture medium of human cervico-vaginal and tonsillar tissues with HCl. In the culture medium buffered to pH 4, HIV-1 replication was reduced in both cervico-vaginal (90.1 ± 0.1%, p < 0.0001, n = 3) and tonsillar tissue (88.0 ± 17.5%, p = 0.0003, n = 3) compared with control tissue blocks cultured in regular medium (Figures 4B,D). No statistically significant inhibition of HIV-1 replication in cervico-vaginal or tonsillar tissues was observed when the culture medium was buffered to pH 6.9 (21.6 ± 18.8%, p = 0.1492, n = 2 and 14.28 ± 10.19%, p = 0.9, n = 3, respectively).

In order to understand if vaginal Lactobacillus themselves are able to suppress HIV-1 infectivity, the virucidal capacities of six strains, L. crispatus (BC3, BC5), L. gasseri (BC12, BC13), and L. vaginalis (BC16, BC17), were studied. HIV-1 was first incubated with Lactobacillus-CP, and after bacteria have been washed off, the infectivity of HIV-1 was tested in cervico-vaginal tissues ex vivo, as described in Section “Materials and Methods.” As shown in Figure 5A, tissue infection with HIV-1 pre-incubated with Lactobacillus-CP for 1 h resulted in inhibition of HIV-1 replication by 64.7 ± 14.9% for L. crispatus BC5 (p < 0.0001, n = 5), by 39.3 ± 18.4% for L. gasseri BC12 (p = 0.0124, n = 5), and by 59.8 ± 13.2% for L. vaginalis BC17 (p = 0.0002, n = 5). No statistically significant inhibition was observed with L. crispatus BC3 (19.4 ± 18.5%, p = 0.46, n = 5), L. gasseri BC13 (16.1 ± 27.9%, p = 0.64, n = 5), and L. vaginalis BC16 (9.0 ± 15.6%, p = 0.96, n = 5) (Figure 5B).

Thereafter, to investigate whether this suppression of HIV-1 replication of cervico-vaginal tissue by L. crispatus BC5, L. gasseri BC12, and L. vaginalis BC17 in the above-described experiments was due to viral binding to Lactobacillus cells, we measured the concentration of p24_gag on Lactobacillus-CP after bacteria were separated by centrifugation. We found that these three strains adsorbed 36.2 ± 21.5, 29.6 ± 25.6, and 39.2 ± 9.6% of HIV-1, respectively, as evaluated from measurements of p24_gag (Figure 5C). In the CP of the remaining strains (L. crispatus BC3, L. gasseri BC13, and L. vaginalis BC16) the p24_gag was less than 10% of the original HIV-1 preparation (data not shown).