Vaginal lactobacilli produce anti-inflammatory β-carboline compounds

Abstract

The optimal vaginal microbiome is a Lactobacillus-dominant community. Apart from Lactobacillus iners, the presence of Lactobacillus species is associated with reduced vaginal inflammation and reduced levels of pro-inflammatory cytokines. Loss of Lactobacillus-dominance is associated with inflammatory conditions, such as bacterial vaginosis (BV). We have identified that Lactobacillus crispatus, a key vaginal bacterial species, produces a family of β-carboline compounds with anti-inflammatory activity. These compounds suppress nuclear factor κB (NF-κB) and interferon (IFN) signaling downstream of multiple pattern recognition receptors in primary human cells and significantly dampen type I IFN receptor (IFNAR) activation in monocytes. Topical application of an anti-inflammatory β-carboline compound, perlolyrine, was sufficient to significantly reduce vaginal inflammation in a mouse model of genital herpes infection. These compounds are enriched in cervicovaginal lavage (CVL) of healthy people compared with people with BV. This study identifies a family of compounds by which vaginal lactobacilli mediate host immune homeostasis and highlights a potential therapeutic avenue for vaginal inflammation.

Keywords: Lactobacillus; bacterial vaginosis; beta-carbolines; genital herpes; inflammation; perlolyrine; vaginal microbiome; vaginal mucosa.

Figure 1.. Vaginal-lactobacilli-produced effectors suppress inflammatory signaling in human monocyte macrophage cell line

Indicated bacterial strains were grown from single colonies for 24 h in NYC III (L. iners strains only) or MRS media (all other strains) and filtered-cell-free supernatant added to THP1 reporter cells at 5% v/v. Control conditions represent cells that received no treatments. NF-κB (A) and interferon signaling response element (ISRE) (B) activation were read 8 h after addition. In (C) and (D) THP1 cells were first treated with TLR3 (1 mg/mL kasugamycin) (C) or TLR4 (500ng/mL LPS) (D) activators and then received supernatant from indicated lactobacilli strains (C) or L. crispatus strain (MV-1A-US) and a candidate L. iners strain (LEAF 2052A-d) (D). Error bars are SD. Samples were compared using a one-way ANOVA with Dunnett’s correction for multiple comparisons with * indicating p value < 0.05, ***p < 0.0005, ****p < 0.00005. Data in (A)–(C) are representative of 4 experiments with 3 biological replicates per treatment condition. Data in (D) are representative of 2 experiments with 4 biological replicates per treatment condition. See also Figures S1 and S2.

Figure 2.. L. crispatus produce family of anti-inflammatory β-carboline compounds

Supernatant from L. crispatus strain (MV-1A-US, HM-637) was fractionated and added to LPS (500 ng/mL)-treated THP1 cells (A) or aminoglycoside (1 mg/mL)-treated THP1 cells (B). Suppressive active fractions common to both treatments are outlined in blue. Active fractions (#12–16) and a control fraction (#1) were added to aminoglycoside-activated cells at indicated doses and interferon reporter activity shown as fold over untreated cells (C). Structures of β-carboline compounds identified from active fractions are shown in (D). THP1 cells were treated with 50 μM each of all 9 β-carboline compounds and then stimulated with LPS for 24 h (E). Error bars are SEM. Samples were compared using a one-way ANOVA with Dunnett’s correction for multiple comparisons with **** indicating p values < 0.00005. See also Figure S3 and Tables S1 and S2 and Data S1.

Figure 3.. β-carbolines suppress interferon signaling downstream of TLR4 and IFNAR

THP1 cells were treated with 500 ng/mL LPS (A and B) or 20 ng/mL IFNβ (C and D) and received indicated concentrations of β-carbolines 1 h before (A and C) or 0.5–1 h after stimulation (B and D). ISRE activation was quantified 24 h post stimulation. Data from (A and B) are representative of 3 independent experiments with 4 biological replicates per experiment. Data from (C and D) are combined from 2 independent experiments with 3–4 biological replicates per condition. Error bars are SEM. Samples were compared using a one-way ANOVA with Dunnett’s correction for multiple comparisons with * representing p < 0.05, **p < 0.005, ***p < 0.0005 and ****p < 0.00005.

Figure 4.. L. crispatus-produced β-carbolines suppress inflammatory signaling in primary human cells

Primary human monocytes (A–C) and human fibroblasts (D–E) were treated with indicated β-carbolines and inflammatory signaling assessed. Primary human monocytes were isolated from peripheral circulating mononuclear cells and treated with 500 ng/mL LPS. 1 h post stimulation cells received 10 μM of all three β-carbolines and 24 h post treatment cells were stained for phospho-NF-κB and CD14. Representative FACS plots gated on live cells are shown in (A), pNF-κB+CD14⁺ cells quantified in (B) and pNF-κB fluorescence intensity quantified in (C). Human foreskin fibroblasts were treated with indicated amounts of BC6 and gene expression quantified after 24 h (D). Fibroblasts were treated with 10 μM and 1 μM BC6 for 24 h, then stimulated with 1 μg/mL PAM3CSK4 and secreted IL-6 quantified via ELISA (E). Error bars represent SEM in (B)–(D) and SD in (E). Data from (A–C) are representative of 3 independent experiments with 4 replicates per condition, whereas data from (D and E) are representative of 2 independent experiments with 3 replicates per condition. Data in (B)–(E) were quantified using 1 way ANOVA with Dunnett’s correction for multiple comparisons with ** representing p < 0.005, ***p < 0.0005 and ****p < 0.00005. See also Figures S4 and S5 and Table S3.

Figure 5.. Vaginal lactobacilli supernatant alleviates inflammation during genital herpes infection

Mice (n = 9 per condition) were injected with 2 mg depot medroxyprogesterone (DMPA) and received 10 μL of either cell-free supernatant from L. crispatus (HM-637), L. reuteri (HM-102), or equivalent volume of media at indicated days (A). Mice were infected with 0.5–1 × 10⁴ PFU HSV-2 and disease symptoms were scored as follows with 1 = mild erythema, 2 = fur loss and visible ulceration, 3 = severe ulceration and signs of sickness behavior (lack of grooming), 4 = hindlimb paralysis, and 5 = moribund (B). Day 12 scores were further broken down in (C). Survival was quantified in (D) and infectious virus from vaginal wash quantified in (E). Error bars represent SEM. Disease scores were compared using two-way ANOVA and survival curves were compared using Mantel-Cox test for significance with *p < 0.05. Frequency distributions in (C) were compared using Kolmogorov-Smirnov test and were not significantly different. All data are combined from 2 independent experiments with n = 4–5 mice per condition.

Figure 6.. Therapeutic β-carboline application suppresses vaginal inflammation during genital herpes infection

Mice were injected with 2 mg DMPA and received 10 μL of either β-carboline BC6 or equivalent volume of DMSO (A). Mice were infected with 0.5–1 × 10⁴ PFU HSV-2 and disease scores quantified (B). Day 11 scores were further broken down in (C). Survival was quantified in (D) and infectious virus from vaginal wash quantified in (E). Vaginal IL-1β was quantified on indicated days post infection (F). Error bars represent SEM. Disease scores were compared using two-way ANOVA and survival curves were compared using Mantel-Cox test for significance with *** indicating p < 0.0005. Data are combined from 2 independent experiments with n = 9–10 mice per condition, except for (E), where data from a single experiment with n = 5 per group are shown. Data in (F) were compared using an unpaired t test with * indicating p < 0.05. See also Figure S7.

Figure 7.. β-carbolines are enriched in cervicovaginal lavage from people with low Nugent scores

We collected cervicovaginal lavage (CVL) from 35 patients who had either low (0–3, n = 24) or high (7–9, n = 11) Nugent scores and quantified BCs using a standard curve. Groups were compared using Welch’s t test. See also Table S4.