Vaginal host immune-microbiome interactions in a cohort of primarily African-American women who ultimately underwent spontaneous preterm birth or delivered at term

Abstract

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) In the overall study population, vaginal concentrations of CXCL10, CCL2, CCL3, SLP1 and VEGF negatively correlated with non-Lactobacillus, Community State Type IV (CST IV) members of the vaginal microbiome; 2) CXCL10, in particular, negatively correlated with 15 bacterial phylotypes, most of which are typical members of CST IV, such as Gardnerella vaginalis, Megasphaera spp., and Atopobium vaginae; 3) Gemella spp., also members of CST IV, negatively correlated with vaginal concentrations of VEGF, CCL2, CCL3, SLPI, and CXCL10; 4) when comparing PTB cases to term controls, five soluble immune mediators (CCL26, CCL22, CCL2, CXCL10, and IL-16), especially CCL26, were negatively correlated with five typical members of CST IV: Sneathia sanguinegens, Parvimonas micra, Veillonellaceae, BVAB2, and Gemella spp.; and 5) Sneathia sanguinegens had stronger negative associations with all five soluble immune mediators (CCL26, CCL22, 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 CST IV of the vaginal microbiome. Sneathia sanguinegens, in particular, had stronger negative associations with different immune mediators, including CXCL10 and CCL26, in women who ultimately underwent PTB compared to those who delivered at term. These findings provide insight into the vaginal host immune-microbiome interactions in normal and complicated pregnancies.

1. INTRODUCTION

The vaginal microbiome is a unique microbial community that is typically characterized by Lactobacillus dominance [1–5]. Yet, molecular microbiology studies have now established that the vaginal microbiome comprises five or more broad community state types (CSTs) [4, 6–8]. Four of these CSTs are dominated by one of four Lactobacillus species: L. crispatus (CST-I), L. gasseri (CST-II), L. iners (CST-III), or L. jensenii (CST-V) [4]. The remaining CST (CST-IV) is not dominated by a Lactobacillus spp. but rather is composed of a diverse array of anaerobic bacteria, including Atopobium, Gardnerella, Megasphaera, Prevotella, and Sneathia spp. [4, 8, 9].

The vaginal CSTs vary among asymptomatic women; however, the frequency of their occurrence depends upon ethnicity [4, 6, 7, 10–17]. European [12], European-American [4, 6], Asian [7, 18], and Asian-American [4, 6] women are likely to have a vaginal microbiome dominated by Lactobacillus (i.e. CSTs I, II, III and V). In contrast, 30–60% of African [11–13, 16], African-American [4, 6, 14, 15, 19], and Hispanic [4] women have a diverse vaginal microbiome that is not dominated by Lactobacillus (i.e. CST IV). Although women with a diverse non-Lactobacillus-dominant vaginal microbiome are typically asymptomatic, they are at a higher risk for acquiring human immunodeficiency virus (HIV) [13, 20–23] and other sexually transmitted diseases [20, 23, 24], and are more likely to experience pregnancy complications [14, 15, 25–30].

During normal pregnancy, the vaginal microbiome is characterized by an increase in Lactobacillus dominance and overall stability regardless of ethnicity [15, 17, 19, 25, 31–35]. In contrast, there is preliminary evidence suggesting that a diverse non-Lactobacillus-dominant vaginal microbiome increases the likelihood of experiencing preterm birth [25–29, 36–38], the leading cause of perinatal morbidity and mortality worldwide [39, 40]. However, this association is still under debate [5, 41–44]. One potential explanation for the discrepancies among studies is that most prior investigations did not consider local host immune responses in the vagina. These responses are the first line of defense against microorganisms, including potential pathogens [11, 45–47], and therefore likely play a central role in the pathophysiology of spontaneous preterm birth [48–50].

Recently, two reports have explored the relationship between the vaginal microbiome and local soluble immune responses in the context of preterm labor and birth [14, 15]. Elovitz et al. found that associations between non-Lactobacillus vaginal bacteria (e.g. Atopobium, Mobiluncus, and Sneathia spp.) and preterm birth were modulated by high vaginal concentrations of β-defensin-2 in African-American women [14]. Fettweis et al. reported that eleven bacterial taxa that are generally members of CST IV were more relatively abundant in the vaginal microbiomes of women ultimately delivering preterm than of those delivering at term [15]. Among women ultimately delivering preterm, there were positive correlations between many of these bacteria and the vaginal fluid concentrations of proinflammatory cytokines [15]. Nevertheless, the contribution of specific vaginal immune-microbiome interactions to the likelihood of spontaneous preterm birth remains poorly understood and requires further investigation.

The aim of the current study was to characterize the local soluble immune mediator profile and its association with the vaginal microbiome in pregnant women who ultimately underwent spontaneous preterm birth and those who delivered at term.

2. MATERIALS AND METHODS

2.1. Clinical specimens

Vaginal fluid samples were obtained at the Perinatology Research Branch, an intramural program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, U.S. Department of Health and Human Services, Wayne State University (Detroit, MI), and the Detroit Medical Center (Detroit, MI). The collection and use of human materials for research purposes were approved by the Institutional Review Boards of the National Institute of Child Health and Human Development and Wayne State University (#110605MP2F(RCR)). All participating women provided written informed consent prior to sample collection.

2.2. Study design

This was a longitudinal case–control study evaluating the correlations between bacterial taxa and soluble immune mediator concentrations (cytokines, chemokines, β-defensins 1, 2 and 3, and SLPI) in the vaginal fluid of pregnant women who ultimately underwent spontaneous preterm birth or delivered at term (see 2.3 Clinical definitions). The study included 18 spontaneous preterm birth cases and 34 term controls (Figure 1A). Sample collection started as early as 10 weeks of gestation and occurred until 32 weeks of gestation (Figure 1A). The number of available samples per patient in this gestational age interval ranged from one to six. A sample of vaginal fluid was collected under direct visualization from the posterior vaginal fornix using a Dacron swab and E-Swab for molecular microbiology and immune mediator quantification, respectively. Vaginal swabs were stored at −80°C until analysis. Data from a bacterial survey of these samples using pyrosequencing of the V1-V3 regions of the 16S rRNA gene were obtained from Romero et al. [51] (Figure 1B).

Figure 1. Study design.

(A) Samples of vaginal fluid from 52 pregnant women were collected in a case-control study: 18 delivered preterm and 34 had a term delivery. The concentrations of 33 immune mediators were determined using sensitive and specific immunoassays. For each patient, 16S rRNA gene pyrosequencing data were also available. Linear mixed effect models were utilized to explore interactions between immune mediators and bacterial phylotypes in the vaginal ecosystem. (B) Relative abundances of bacterial phylotypes in the vaginal samples of women who ultimately underwent spontaneous preterm birth or delivered at term.

2.2.1. Inclusion and exclusion criteria

Samples collected within one week of spontaneous preterm birth were excluded from analysis to mitigate any potential effects of the inflammatory process of labor on vaginal fluid bacterial and immune profiles [52, 53]. Patients with medically indicated preterm birth (e.g. preeclampsia, intrauterine growth restriction, or congenital anomalies) or cerclage placement were excluded from the cases. Patients with a history of preterm contractions, preterm labor, or cerclage placement were excluded from the term controls.

2.3. Clinical definitions

Preterm labor was diagnosed by the presence of at least two uterine contractions every 10 minutes associated with cervical changes in patients with a gestational age between 20 and 34 weeks. Preterm birth was considered as delivery before or at 34 weeks of gestation [51]. Most patients underwent spontaneous preterm labor with intact membranes, and a subset (N=7) of the preterm cases was diagnosed with preterm prelabor rupture of membranes (PPROM). No samples in this study were collected after rupture of membranes. Term controls were defined as women who delivered between 37 to 42 weeks of gestation without congenital anomalies or obstetrical, medical, or surgical complications. Histopathological examination of the placenta was performed by perinatal pathologists blinded to clinical diagnoses and obstetrical outcomes according to standardized Perinatology Research Branch protocols [54]. Acute inflammatory lesions of the placenta (maternal inflammatory response and fetal inflammatory response) were diagnosed according to established criteria, including staging and grading [54].

2.4. Cytokine profiling

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 10spot 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, IL-5, IL-7, TNFα, TNFβ, vascular endothelial growth factor (VEGF)], and the 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(macrophage inflammatory protein (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. The detection limits of the assays and the inter-assay and intra-assay coefficients of variation, are displayed in Supplemental Table 1.

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. The detection limits of the assays and the inter-assay and intra-assay coefficients of variation are displayed in Supplemental Table 1.

Total protein concentrations were measured by a BCA Protein Assay kit from (ThermoFisher, catalogue number 23225). The vaginal immune mediator dataset and sample metadata for this study are available in Supplemental Data File 1.

2.5. Statistical analysis

2.5.1. Demographics analysis

Clinical characteristics of the patient population were summarized as median and interquartile ranges (IQRs) for continuous variables and as percentages for categorical variables. The comparison of demographic variables between the preterm and term groups was performed using the Fisher’s exact test for binary variables or the Mann-Whitney U test for continuous variables (Table 1).

Table 1.

Clinical and demographic characteristics of the study population

	Patients who ultimately delivered at term (n = 34)	Patients who ultimately underwent spontaneous preterm birth (n = 18)	p-value
Maternal age (y; median [IQR])a	24 (21–28.8)	21(20–26)	.3
Body mass index (kg/m2; median [IQR])a	28.7(24.6–34.3)	25.7 (21.6–33.2)c	.5
Primiparityb	11.8% (4/34)	5.6% (1/18)	.7
Race/ethnicityb			.7
African-American	94.1% (32/34)	94.4% (17/18)
White	2.9% (1/34)	5.6% (1/18)
Others	2.9% (1/34)	0% (0/18)
Gestational age at delivery (wk; median [IQR])a	39.7 (38.7–40.3)	30.5 (28.4–33.1)	<0.001
Mode of deliveryb			.1
Cesarean section	17.7% (6/34)	38.9% (7/18)
Vaginal	82.3% (28/34)	61.1% (11/18)
Birthweight (g; median [IQR])a	3257.5 (3127.5–3522.5)	1402.5 (1037.5–1986.3)	<0.001
Apgar score at 1 min (median [IQR])a	9 (8–9)	6 (3–8)	<0.001
Apgar score at 5 min (median [IQR])a	9 (9–9)	8 (6–8)	<0.001
Acute maternal inflammatory responseb
Stage 1 (Early acute subchorionitis or chorionitis)	21.2% (7/33)c	5.6% (1/18)	.2
Stage 2 (Acute chorioamnionitis)	0% (0/33)c	44.4% (8/18)	<0.001
Stage 3 (Necrotizing chorioamnionitis)	0% (0/33)c	5.6% (1/18)	.4
Acute fetal inflammatory responseb
Stage 1 (Chorionic vasculitis or umbilical phlebitis)	12.1% (4/33)c	33.3% (6/18)	.3
Stage 2 (Umbilical arteritis)	0% (0/33)c	22.2% (4/18)	0.01
Stage 3 (Necrotizing funisitis)	0% (0/33)c	5.6% (1/18)	.4

^aMann-Whitney U test.

^bFisher’s exact test.

^cOne missing data.

2.5.2. Filtering out low-abundance bacterial taxa

The 16S rRNA gene bacterial phylotype data were retained for analysis if (1) bacterial phylotypes had a relative abundance of at least 1% in at least 5% of the samples, or (2) had a relative abundance of at least 0.1% in at least 15% of the samples, as informed by Fettweis et al. [15]. Twenty-five bacterial phylotypes met these criteria. (Figure 1B)

2.5.3. Data transformation

The percent relative abundances of bacterial phylotypes were log₂-transformed prior to analysis. The concentrations of soluble immune mediators (pg*ml⁻¹) were normalized to the total protein concentration (mg*ml⁻¹). Concentrations lower than the measurable threshold were replaced with 99.9% of the lowest value detected in the study for each immune mediator. The normalized concentrations were also log₂-transformed.

2.5.4. The pairwise analysis of immune mediator-bacterial phylotype associations in the overall study population

Linear mixed effects models were used to assess associations between the relative abundances of bacterial phylotypes (control variables) and soluble immune mediator concentrations (response variables) in the overall patient population. In order to account for repeated sample collection during pregnancy, a random effect was included for each patient. The derived association represented the significant change in soluble immune mediator concentration (response variable) with every doubling of the relative abundance of a specific bacterium (explanatory variable). The p-values were corrected using the false discovery rate method to obtain q-values. Significance was inferred based on a false discovery rate less than 20%.

2.5.5. Differential association of immune mediators and bacterial phylotypes between preterm cases and term controls

To determine whether immune mediator-bacterial phylotype associations differed between term controls and spontaneous preterm birth cases, the models described above also included an interaction term between bacterial phylotype relative abundance and adverse delivery outcome (spontaneous preterm birth). The correlations displayed in figures correspond to significant differences in the regression slopes between patients who subsequently had a spontaneous preterm birth and those who delivered at term. The effect size in this analysis represents a difference in the rate of change in the immune mediator concentration with an increase in microbial relative abundance between cases and controls, and it was further converted into a linear fold change. The p-values corresponding to the models were corrected using the false discovery rate method to obtain q-values. Significance was inferred based on a false discovery rate less than 20% and a fold change ≥1.3.

2.5.6. Software packages utilized in this study

Statistical analyses were conducted using the lme4 package [55] in R (version 3.6.1) [56] and GraphPad Prism version 8.0.1 for Windows (GraphPad Software, San Diego, CA, USA, www.graphpad.com).

3. RESULTS

3.1. Characteristics of the study population

The clinical and demographic characteristics of women who had a spontaneous preterm birth or a term delivery are displayed in Table 1. The study population was largely African-American (94%). There were no significant differences in maternal age, ethnicity, pre-pregnancy body mass index, rate of primiparity, or the mode of delivery between the two groups. Gestational age at delivery, neonatal birthweight, and Apgar scores were significantly lower in preterm birth cases than in term controls (p<0.001). The frequency of stage 2 acute maternal and fetal inflammatory responses was higher in women who ultimately underwent spontaneous preterm birth (p<0.001).

3.2. Associations between vaginal immune mediators and bacterial phylotypes in the overall study population

Associations between the concentrations of vaginal soluble immune mediators and the composition of the vaginal microbiome across all samples from all women were investigated using linear mixed effect models in a pair-wise fashion between the 33 immune mediators (Supplemental Table 1) and the 25 bacterial phylotypes (Figure 1B). Of the 33 immune mediators examined, eight (β-defensins 2 and 3, IL-1β, CXCL10, CCL2, CCL3, SLPI, and VEGF) were found to correlate with 18 different bacterial phylotypes in the overall patient population, regardless of whether they ultimately delivered preterm or at term (Figure 2; Supplemental Table 2). Notably, the vaginal fluid concentrations of CXCL10 negatively correlated with 15 vaginal bacterial phylotypes, most of which are typical members of vaginal CST IV [4], such as Gardnerella vaginalis (Figure 3A), Megasphaera spp. type 1 (Figure 3B), and Atopobium vaginae (Figure 3C). The vaginal fluid concentrations of VEGF were lower (p=0.002) in samples with a higher relative abundance of Gemella spp. (Figure 3D), bacteria also found in CST IV [4]. The same trend was observed between Gemella spp. and CCL2, CCL3, SLPI, and CXCL10 (Figure 2).

Figure 2. Network of associations between vaginal immune mediators and bacterial phylotypes in the overall study population.

Immune mediators and bacterial phylotypes are represented on the left and right sides of the plot, respectively. The lines indicate significant associations between vaginal immune mediators and bacterial phylotypes in the study population (N=52). Red lines indicate negative associations between the concentrations of vaginal immune mediators and the relative abundances of bacterial phylotypes. Green lines indicate positive associations between immune mediators and bacterial phylotypes. The analyses reflect linear mixed effects models, accounting for repeated measures and with an FDR<20%.

Figure 3. Correlation analyses of vaginal immune mediators and bacterial phylotypes in the overall study population.

The X-axes represent the relative abundances of the indicated bacterial phylotypes, and the Y-axes represent the concentrations of either CXCL10, VEGF, or β-defensin-3 in the vaginal fluid. Correlation between vaginal concentrations of CXCL10 and the relative abundance of Gardnerella vaginalis (A), Megasphaera spp. type 1 (B), or Atopobium vaginae (C). Correlation between the vaginal concentrations of VEGF and the relative abundance of Gemella spp. (D). Correlation between the vaginal concentrations of β-defensin-3 and the relative abundance of (E) Clostridiales and (F) BVAB1. The lines represent the linear mixed-effects model fit to the data. The fold changes represent the change in concentration of the immune mediator given a doubling of bacterial phylotype relative abundance (1 unit on the x-axis).

Conversely, the vaginal fluid concentrations of IL-1β and β-defensins 2 and 3 positively correlated with the relative abundances of specific vaginal bacterial phylotypes (Figure 2; Supplemental Table 2). Specifically, the vaginal concentrations of the antimicrobial peptide β-defensin 3 were higher in samples with higher abundances of the order Clostridiales (Figure 3E) or Bacterial Vaginosis Associated Bacteria 1 (BVAB1) (Figure 3F). Yet, the vaginal fluid concentrations of the antimicrobial peptide β-defensin 2 positively correlated solely with Lactobacillus coleohominis (Figure 2). The pro-inflammatory cytokine IL-1β had positive correlations with Gardnerella vaginalis and Aerococcus spp. (Figure 2).

These results indicate that, in our study population, there are negative associations between specific soluble immune mediators, mainly CXCL10, and increasing relative abundances of CST IV bacteria in the vaginal fluid.

3.3. Differential association between vaginal soluble immune mediators and bacterial phylotypes in women who delivered preterm compared to those who delivered at term

Next, it was determined whether the associations between vaginal immune mediators and bacterial phylotypes differed across all samples between women who ultimately delivered preterm and those who delivered at term. Regression analysis using linear mixed effects models was performed in a pair-wise fashion between the 33 immune mediators (Supplemental Table 1) and the 25 bacterial phylotypes (Figure 1B). Unlike in the overall association analyses above, in this analysis, each group (women who ultimately delivered preterm or at term) was allowed to have a different slope. To add stringency, in addition to a cut-off on adjusted p-values stating the significance of the difference in slope, a minimum difference of ≥1.3-fold between slopes was required (Figure 4A; Supplemental Table 3). When comparing preterm birth cases to term controls, five soluble immune mediators (CCL26, CCL22, CCL2, CXCL10, and IL-16) were more negatively correlated with five typical members of vaginal CST IV: Sneathia sanguinegens, Parvimonas micra, Veillonellaceae, Bacterial Vaginosis Associated Bacteria 2 (BVAB2), and Gemella spp. Sneathia sanguinegens had the highest number of differentially negative associations (n=5) with vaginal pro-inflammatory immune mediators between the preterm birth cases and term controls: CXCL10 (Figure 4B), IL-16 (Figure 4C), CCL22 (Figure 4D), and CCL26 (Figure 4E). Notably, CCL26 had the highest number of differentially negative correlations (n=5) with bacterial phylotypes, including Sneathia sanguinegens (Figure 4E), Gemella spp. (Figure 4F) and BVAB2 (Figure 4G), when comparing preterm birth cases to term controls. Collectively, these results show that specific immune mediators, namely CCL26, CCL22, CCL2, CXCL10, and IL-16, were more negatively correlated with stereotypical CST IV bacterial phylotypes in women who ultimately underwent spontaneous preterm birth than in those who delivered at term.

Figure 4. Network of immune mediators differentially associated with changes in bacterial phylotype relative abundances in preterm cases and term controls.

(A) Immune mediators and bacterial phylotypes are presented on the left and right sides of the plot, respectively. Red lines indicate a lower slope of the association between the concentration of an immune mediator and the relative abundance of a bacterial phylotype in preterm birth cases than in term controls. The fold changes represent a difference in the rate of change in the immune mediator concentration given a unit increase in bacterial phylotype relative abundance between cases and controls. Correlation analysis between vaginal concentrations of CXCL10 (B), IL-16 (C), CCL22 (D) and the relative abundance of Sneathia sanguinegens. Association between the vaginal concentrations of CCL26 and the relative abundance of Sneathia sanguinegens (E), Gemella spp. (F), and BVAB2 (G).

4. DISCUSSION

The assessment of vaginal host immune-microbiome interactions revealed that specific soluble immune mediators, mainly CXCL10, negatively correlated with typical members of CST 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 underwent spontaneous preterm birth than in those who delivered at term. These findings provide insight into the vaginal host immune-microbiome interactions in normal and complicated pregnancies.

In our overall study population, which was largely African-American, there were negative associations between specific soluble immune mediators and increasing relative abundances of CST IV bacteria in the vaginal fluid. Consistently, previous reports showed that the vaginal ecosystem of African-descendent pregnant [14, 17, 57] and non-pregnant [4, 6, 8, 10, 58–60] women is enriched in CST IV bacteria. Such a vaginal phenotype has been attributed to the presence of asymptomatic bacterial vaginosis in non-pregnant [61] and pregnant [62] women. In the current study, we did not evaluate the incidence of bacterial vaginosis; yet, historically, our study population is at high risk of presenting with this clinical condition [63, 64].

Herein, we report that a high relative abundance of CST IV bacteria was associated with low concentrations of CXCL10 in the vaginal fluid. This result is in agreement with the following evidence: 1) in pregnant and non-pregnant African women, the presence of G. vaginalis and A. vaginae (i.e. bacteria associated with bacterial vaginosis) is strongly associated with low vaginal fluid concentrations of CXCL10 [66, 67]; 2) non-pregnant South African women with bacterial vaginosis-associated microbial communities display reduced vaginal fluid concentrations of CXCL10 [68]; 3) the vaginal fluid concentrations of CXCL10 negatively correlate with BVAB1, a CST IV and bacterial vaginosis associated bacterium, in women (mostly African-American) who ultimately delivered preterm [15]; and 4) the low vaginal fluid concentrations of CXCL10 observed in non-pregnant African women with bacterial vaginosis can be boosted upon successful treatment with metronidazole [69], a drug commonly used to treat this clinical condition [70, 71].

CXCL10 is primarily a T-cell chemokine that induces the migration of immune cells to sites of inflammation [72–74]. This chemokine plays a central role in host defense mechanisms against pathogens [75–78], including those which take place in the amniotic cavity [79, 80], and in the mechanisms involved in the pathophysiology of preterm labor and birth [54, 79, 81, 82]. CXCL10 also mediates the induction of apoptosis and regulates cell growth and proliferation as well as angiogenesis [83]. These diverse functions of CXCL10 depend on the splice variants of its receptor CXCR3 (CXCR3-A, CXCR3-B, and CXCR3-alt). Specifically, chemotaxis and proliferation are mediated by CXCR3-A in several cell types [84, 85], whereas CXCL10 inhibits migration and proliferation and induces apoptosis via CXCR3-B [85]. These receptors are expressed on various immune (e.g. T cells, natural killer (NK) cells, and NKT cells) as well as non-immune (e.g. bronchial epithelial cells and astrocytes) cells [83, 86, 87]. Pertinent to the current investigation, mechanistic studies have shown that CXCL10-deficient mice are more susceptible to viral infections [88, 89] and, even more importantly, this chemokine can directly kill Gram-positive bacteria such as Bacillus anthracis [90]. Furthermore, polymorphisms in CXCL10 and its receptor, CXCR3, are associated with susceptibility to infection [91, 92] and recurrent preterm birth [81], respectively. Therefore, it is tempting to suggest that, in the vaginal ecosystem of African-American women (our study population), low concentrations of CXCL10 modulate local bacterial growth, favoring the proliferation of CST IV bacteria. Yet, additional mechanistic studies are needed to test this hypothesis.

In the current study, we also show that Sneathia sanguinegens is negatively associated with multiple vaginal immune mediators in women who subsequently underwent spontaneous preterm birth compared to those who delivered at term. Sneathia species are rod-shaped Gram-negative anaerobic bacteria whose common habitat is the human vagina [93–95]. The traditional view is that Sneathia species are of a fastidious nature since their culture is not always possible [96–103]. Nevertheless, Sneathia species have been detected using molecular microbiology techniques in the amniotic fluid of women with intra-amniotic infection associated with PPROM [98, 104–108], a sonographic short cervix [109], preterm labor and birth with intact membranes [96, 97, 101, 110], and clinical chorioamnionitis [111]. Importantly, Sneathia species in the vaginal fluid are associated with preterm birth in African-American [14, 15, 112] and Caucasian [29] women. Sneathia species have also been associated with neonatal infection [27, 93, 113]. Therefore, investigation focused on the mechanisms whereby Sneathia species invade the intra-amniotic space and fetal tissues is warranted.

To our knowledge, this is the first study to show that Sneathia species are negatively associated with immune mediators such as CXCL10 and CCL26 in the vaginal fluid. The former chemokine is discussed above. CCL26 is a chemokine that belongs to the eotaxin family (eotaxin-1/CCL11, eotaxin-2/CCL24, and eotaxin-3/CCL26) [114]. Eotaxins are released by epithelial, mesenchymal, and endothelial cells in response to various mediators, including IL-4, IL-13, and TNF-α [115–117]. Eotaxins are potent chemoattractants for eosinophils [115, 117], which play a central role in type-2 immunity such as host defense against parasitic helminth infections, tissue repair and remodeling, and allergic diseases [118, 119]. Specifically, CCL26 is preferentially recognized by CCR3, a common and specific receptor for eotaxins [114, 117, 120]. CCL26 is the most effective eotaxin to induce eosinophil migration in asthma patients [121]. Therefore, this chemokine is implicated in several diseases associated with eosinophilic infiltration such as atopic dermatitis [122] and eosinophilic gastrointestinal disorders [123]. In addition, CCL26 displays host defense activities by exerting bactericidal potency against several airway pathogens [124]. The effect of CCL26 on Sneathia species, or any other members of the CST IV vaginal microbiome, has yet to be described. Therefore, further mechanistic studies are required to investigate the translational relevance of the negative chemokine-microbiome associations described herein, and whether such interactions play a central role in the mechanisms leading to preterm labor and birth.

5. CONCLUSION

Specific soluble immune mediators, mainly CXCL10, negatively correlate with typical members of CST IV of the vaginal microbiome. Sneathia sanguinegens, a CST IV bacterium that is commonly associated with intra-amniotic infection, displays stronger negative associations with different immune mediators, including CXCL10 and CCL26, in women who ultimately experience spontaneous preterm birth compared to those who deliver at term. These findings provide insight into the vaginal host immune-microbiome interactions in normal and complicated pregnancies.

HIGHLIGHTS:

• Vaginal fluid immune mediators negatively correlated with CST IV vaginal bacteria

• CXCL10, in particular, was negatively correlated with 15 CST IV bacteria

• Specific cytokines were negatively correlated with CST IV bacteria in preterm births

• Sneathia sanguinegens was negatively correlated with cytokines in preterm births

Abbreviations:

BVAB

bacterial vaginosis-associated bacterium

VEGF

vascular endothelial growth factor

CST

community state type

MIP

macrophage inflammatory protein

PPROM

preterm prelabor rupture of membranes

PTB

preterm birth

SLPI

secretory leukocyte protease inhibitor