Background: Recent studies suggest that alterations in the vaginal microbiome allow for the assessment of the risk for spontaneous preterm birth (PTB), the leading cause of neonatal morbidity and mortality worldwide. However, the associations between the local immune response and the vaginal microbiome are still poorly understood. Herein, we characterize the vaginal host immune-microbiome interactions in women who ultimately underwent PTB and in those who delivered at term.

Methods: Vaginal fluid samples from 52 pregnant women (of whom 18 underwent PTB and 34 delivered at term) were collected from 10-32 weeks in a case-control study. Concentrations of 33 immune mediators were determined using sensitive and specific immunoassays. The previously published 16S rRNA gene sequence and bacterial phylotype data of these subjects were utilized in this study. Linear mixed effects models were utilized to test associations between vaginal immune mediator concentrations and bacterial phylotype relative abundances.

Results: 1) Specific immune mediators (β-defensins 2 and 3, IL-1β, CXCL10, CCL2, CCL3, SLPI, and VEGF) correlated with 18 different vaginal bacterial phylotypes in the overall study population; 2) vaginal concentrations of CXCL10, CCL2, CCL3, SLP1 and VEGF negatively correlated with non-Lactobacillus members of the vaginal microbiome; 3) vaginal concentrations of CXCL10 were negatively correlated with 15 bacterial phylotypes, most of which are typical members of Community State Type IV of the vaginal microbiome, such as Gardnerella vaginalis, Megasphaera sp. type 1, and Atopobium vaginae; 4) Gemella spp. were negatively correlated with vaginal concentrations of VEGF, CCL2, CCL3, SLPI, and CXCL10; 5) when comparing PTB cases to term controls, five soluble immune mediators (CCL26, CCL22 and CCL2, CXCL10 and IL-16), and especially CCL26, were negatively correlated with five typical members of Community State Type IV: Sneathia sanguinegens, Parvimonas micra, Veillonellaceae, BVAB2, and Gemella spp; and 6) Sneathia sanguinegens had stronger negative associations with all five soluble immune mediators (CCL26, CCL22 and CCL2, CXCL10 and IL-16) in PTB cases than in term controls.

Conclusions: The assessment of vaginal host immune-microbiome interactions revealed that specific soluble immune mediators, mainly CXCL10, negatively correlated with typical members of Community State Type IV of the vaginal microbiome. In addition, this assessment particularly showed that Sneathia sanguinegens had stronger negative associations with different immune mediators, including CXCL10 and CCL26, in women who ultimately had a PTB compared to those who delivered at term. These findings provide insight into the vaginal host immune-microbiome interactions in normal and complicated pregnancies.

A sample of vaginal fluid was collected under direct visualization from the posterior vaginal fornix using an E-Swab for immune mediator quantification. Vaginal swabs were stored at −80°C until analysis. Vaginal fluid samples were assessed using sensitive and specific V-PLEX immunoassays (Meso Scale Discovery, Gaithersburg, MD, USA) to measure vaginal fluid concentrations of several soluble immune mediators: the pro-inflammatory V-PLEX 10-spot assay K15049D-2 [IFN-γ, IL-10, IL-12p70, IL-13, IL-1β, IL-2, IL-4, IL-6], cytokine VPLEX 10-spot assay K15050D-2 [GM-CSF, IL-12/IL-23p40, IL-15, IL-16, IL-17A, IL-1α, 99 IL-5, IL-7, TNFα, TNFβ, VEGF], and chemokine V-PLEX 10-spot assay K15047D-2 [CCL11(Eotaxin), CCL26(Eotaxin-3), CXCL8(IL-8), CXCL10(IP-10), CCL2(MCP-1), CCL13(MCP-4), CCL22(MDC), CCL3(MIP-1α), CCL4(MIP-1β), CCL17(TARC)], according to the manufacturer’s instructions. The plate signals were read by QuickPlex SQ 120 (Meso Scale Discovery). Standard curves were generated and the assay values  of the samples were interpolated from the curves. Secretory leukocyte protease inhibitor (SLPI) (R&D Systems, Minneapolis, MN, USA), β-defensin-1 and 3 (Aviscera Bioscience, Santa Clara, CA, USA), and β-defensin 2 (ALPCO, Salem, NH, USA) concentrations were measured in vaginal fluid samples  using sensitive and specific single-analyte immunoassays, according to the manufacturers' instructions. The intensity of developed color was measured by a SpectraMax M5 Microplate Reader (San Jose, CA, USA). The concentrations of SLPI, and β-defensins 1, 2 and 3 were determined by interpolation from the standard curves. Total protein concentrations were measured by a BCA Protein Assay kit from 117 (ThermoFisher, catalogue number 23225).