The vaginal microbiota, host defence and reproductive physiology

Abstract

The interaction between the human host and the vaginal microbiota is highly dynamic. Major changes in the vaginal physiology and microbiota over a woman's lifetime are largely shaped by transitional periods such as puberty, menopause and pregnancy, while daily fluctuations in microbial composition observed through culture‐independent studies are more likely to be the results of daily life activities and behaviours. The vaginal microbiota of reproductive‐aged women is largely made up of at least five different community state types. Four of these community state types are dominated by lactic‐acid producing Lactobacillus spp. while the fifth is commonly composed of anaerobes and strict anaerobes and is sometimes associated with vaginal symptoms. The production of lactic acid has been associated with contributing to the overall health of the vagina due to its direct and indirect effects on pathogens and host defence. Some species associated with non‐Lactobacillus vaginal microbiota may trigger immune responses as well as degrade the host mucosa, processes that ultimately increase susceptibility to infections and contribute to negative reproductive outcomes such as infertility and preterm birth. Further studies are needed to better understand the functional underpinnings of how the vaginal microbiota affect host physiology but also how host physiology affects the vaginal microbiota. Understanding this fine‐tuned interaction is key to maintaining women's reproductive health.

Abbreviations

AMP

antimicrobial peptide

AV

aerobic vaginitis

BV

bacterial vaginosis

CCL

chemokine (C‐C motif) ligand

CDC

cholesterol‐dependent cytolysin

CST

community state type

HBD

human β‐defensin

HIV

human immunodeficiency virus

HNP

human neutrophil peptide

HSV

herpes simplex virus

HPV

human papillomavirus

IgA

immunoglobulin A

IFN

interferon

IgG

immunoglobulin G

IL

interleukin

PRR

pattern recognition receptor

TLR

toll‐like receptor

SCFA

short chain fatty acid

SLPI

secretory leukocyte peptidase inhibitor

SP

surfactant protein

NF‐κB

nuclear factor‐κB

TNF

tumor necrosis factor

Introduction

The assemblages of microbes (microbiota) associated with the human body have been shown to affect human physiology, immunity and nutrition (Kau et al. 2011; Chen et al. 2015; Conlon & Bird, 2015). In the vagina, microbes exist in a finely tuned mutualistic relationship with the host and provide the first line of defence against the colonization by opportunistic pathogens. Throughout a woman's lifespan, the vaginal microbiota undergoes major changes associated with transitional reproductive periods such as puberty and menopause, as summarized in Farage & Maibach (2005). During these periods, the vaginal microbiota can affect host reproductive physiology but can also be affected by host physiology.

Recent high‐throughput 16S rRNA gene sequencing studies examining vaginal bacterial species composition and abundance in reproductive‐aged women have shown that there are at least five major types of vaginal microbiota called community state types (CSTs) (Zhou et al. 2007; Ravel et al. 2011; Gajer et al. 2012). Four of these CSTs are dominated by Lactobacillus crispatus (CST‐I), L. iners (CST‐III), L. gasseri (CST‐II) or L. jensenii (CST‐V) and one, CST‐IV, does not contain a significant number of Lactobacillus but is composed of a polymicrobial mixture of strict and facultative anaerobes including species of the genera Gardnerella, Atopobium, Mobiluncus, Prevotella and other taxa in the order Clostridiales (Fig. 1; Fredricks et al. 2005; Campos et al. 2008; Ravel et al. 2011; Gajer et al. 2012). The frequency of these CSTs has been shown to differ in different ethnic backgrounds, with CST‐IV more common (∼40%) in black and Hispanic women (Ravel et al. 2011). The polymicrobial condition known as bacterial vaginosis (BV) is compositionally similar to CST‐IV since it is defined by a loss of Lactobacillus spp., the presence of anaerobes and strict anaerobes, and sometimes associated clinical symptoms including discharge, odour and irritation. In research settings, a Gram‐staining scoring procedure that relies on the identification of bacterial morphotypes known as the Nugent test (appropriately renamed Nugent‐BV by Martin (2012)) is used to establish a BV diagnosis (Nugent et al. 1991). Clinically, the diagnosis of BV is accompanied by an evaluation of the following signs and symptoms: discharge, malodour, the presence of clue cells and vaginal pH > 4.5 as defined by the Amsel criteria (Amsel et al. 1983).

Figure 1. Heat map of vaginal microbiota community state types.

Hierarchical clustering shows that the vaginal microbiota of reproductive‐aged women clusters into five distinct community state types, four of which are dominated by Lactobacillus spp. (Lactobacillus crispatus (CST‐I), L. iners (CST‐III), L. gasseri (CST‐II) or L. jensenii (CST‐V)) and the fifth (CST‐IV) is composed of a polymicrobial mixture of strict and facultative anaerobes including species of the genera Atopobium, Megasphera, Mobiluncus, Prevotella and other taxa in the order Clostridiales. Figure reproduced from Ravel et al. (2011).

Daily fluctuations in the composition of the vaginal microbiota have been previously documented by microscopy and cultivation studies (Hay et al. 1997; Keane et al. 1997; Schwebke et al. 1999). These findings were confirmed and extended in longitudinal culture‐independent analyses such as those of women who self‐collected vaginal swabs twice weekly for 16 weeks (Brotman et al. 2008, 2010; Gajer et al. 2012), or daily for 10 weeks (Ravel et al. 2013) or 4 weeks (Srinivasan et al. 2010). It was observed that some vaginal microbial communities transitioned in and out of CST‐IV. The amount of time spent in a particular CST could vary individually as some women experienced consistent and stable CST longitudinal patterns (defined as community class), while others frequently transitioned between CSTs and most frequently to CST‐IV (Gajer et al. 2012; Ravel et al. 2013). In some cases, CST transitions were triggered by menstruation or sexual behaviours, but in other cases they seem to be driven by uncharacterized factors (Gajer et al. 2012). These longitudinal studies highlight the highly dynamic nature of vaginal microbial communities and emphasize the need to better understand the underlying biological factors modulating fluctuations in composition and functions that affect host physiology.

Historically, Lactobacillus‐dominated vaginal microbial communities have been associated with healthy reproductive‐aged women and are characterized by the production of copious amounts of lactic acid and thus a pH < 4.5 (reviewed in Linhares et al. 2011; Ma et al. 2012; Petrova et al. 2015). This acidic environment is thought to be highly protective against infections or colonization of the vagina by pathogens and non‐indigenous microbes. An additional benefit of Lactobacillus spp. (i.e. L. crispatus and L. gasseri) is the supply of bacteriocins or bacteriocin genes (i.e. gassericin T, acidocin lF221A, type‐A lantibiotic, and Bacteriocins IIa, IIc and J46) to inhibit growth of undesirable species (i.e. Klebsiella spp., Staphylococcus aureus, Escherichia coli or Enterococus faecalis) (Stoyancheva et al. 2014; Zheng et al. 2015). However, the notion that a Lactobacillus‐dominated vaginal microbiota is necessarily the norm has been called into question since mounting evidence suggests that about 25% of asymptomatic women do not possess a Lactobacillus‐dominated microbiota at any given time, a staggering proportion that does not support a diseased state (Zhou et al. 2004; Ravel et al. 2011; Anahtar et al. 2015). These differences between women appear to be driven by a combination of cultural, behavioural, genetic and other unknown underlying factors (Ravel et al. 2011; Gajer et al. 2012; Anahtar et al. 2015). However, a strong association between CST‐IV (as established by Nugent‐BV) and increased risk to sexually transmitted infections (Martin et al. 1999; Peters et al. 2000; Cherpes et al. 2003), including human immunodeficiency virus (HIV) (Cohen et al. 1995; Taha et al. 1998; Cu‐Uvin et al. 2001; Coleman et al. 2007), and reproductive tract and obstetric sequelae has been established through thorough epidemiological studies (Gravett et al. 1986; McDonald et al. 1992; Hillier et al. 1995; Meis et al. 1995; Goldenberg et al. 1997). Hence, while CST‐IV might be normal (asymptomatic) in some women, it is still associated with significantly increased risk of adverse outcomes. An illustration of how CST‐IV can help further foster infections is the case of chlamydial infection, where interferon (IFN)‐γ production is thought to be critical for chlamydia clearance. IFN‐γ activates the human enzyme indoleamine‐2,3‐dioxygenase, which catabolizes tryptophan, eventually leading to tryptophan starvation and chlamydia clearance since genital chlamydia cannot synthesize tryptophan. However, production of indole compounds by anaerobes and strict anaerobes contained in CST‐IV enables chlamydia to shunt its deficiency and produce tryptophan, thus bypassing this host defence mechanism and establishing a long‐term infection (discussed in Aiyar et al. 2014). Similarity, relative to other Lactobacillus‐dominated community states, CST‐IV‐like communities increase the risk of HIV infection (Pyles et al. 2014; Anahtar et al. 2015). However, not all Lactobacillus spp. are necessarily beneficial and protective since, for example, some strains of L. iners might carry pathogenicity factors, such as inerolysin, a cholesterol‐dependent cytolysin (CDC) and a host epithelial cell pore‐forming enzyme, which was found to be up‐regulated at least sixfold in women with BV (Macklaim et al. 2011; 2013). Therefore, when considering the impact of the microbiota on host defence and reproductive physiology, it is important to place it in the context of these dynamic and individualized relationships.

The effect of the microbiota on host defence

The vagina contains a number of immune‐related cells and receptors to help sense the microbial environment (Wira et al. 2005). Surveillance for microbes within the female genital tract of both commensal and pathogenic microbes is generally achieved by microbial motif pattern recognition by pattern recognition receptors (PRRs) such as toll‐like receptors (TLRs) or the dectin‐1 receptor (which helps recognize the fungal pathogen Candida albicans; Carvalho et al. 2012; Usluogullari et al. 2014), and nucleotide‐binding oligomerization domain (NOD) receptors present in and on both squamous epithelial cells lining the vagina and the columnar cells lining the upper female genital tract (as reviewed in Wira et al. 2005; Witkin et al. 2007a; Horne et al. 2008; Mitchell & Marrazzo, 2014). Microbial stimulation of PRRs initiates cytokine/chemokine signalling cascades, for example secretion of interleukin (IL)‐1β, IL‐6, IL‐8 and tumor necrosis factor‐α (TNF‐α), in order to recruit or activate specialized cells, such as NK cells, macrophages, CD4⁺ helper T‐cells, and CD8⁺ cytotoxic T‐cell lymphocytes and B lymphocytes (as reviewed in Wira et al. 2005; Brotman et al. 2013a; Nguyen et al. 2014). Genetic variants of PRRs such as the IL‐1R antagonist gene, TLR4, TLR9, IL‐1R2 and TNF‐α may play a role in how a woman responds to a particular microbial challenge or pregnancy outcome, as evidenced by several genetic‐disease association studies (Genç et al. 2004a,b, 2007; Giraldo et al. 2007; Ryckman et al. 2011; Royse et al. 2012; Mackelprang et al. 2015). Women with CST‐IV‐like states show significant increases in IL‐1α, IL‐1β, TNF‐α, IFN‐γ, IL‐4, IL‐8, IL‐10, IL‐12p70, and fms‐like tyrosine kinase 3 ligand relative to CST‐I as well as significantly higher IFN‐γ in CST‐III relative to CST‐I. Specifically, in one study, Prevotella amnii, Mobiluncus mulieris, Sneathia amnii and Sneathia sanguinegens (all commonly found in CST‐IV) were found to induce higher levels of IL‐1α, IL‐1β, and IL‐8 relative to L. crispatus dominated communities (CST‐I), whereas L. iners dominated communities (CST‐III) induced moderate IL‐8 levels relative to CST‐I. The authors also showed how there were significant increases in IL‐1α, IL‐1β and TNF‐α longitudinally in subjects that transition from a CST‐I, to CST‐III and to a CST‐IV (Anahtar et al. 2015). Conversely, mock communities dominated by L. crispatus (CST‐I) and L. jensenii (CST‐V) on reconstructed three‐dimensional vaginal epithelial models do not strongly elicit cytokine IL‐1β or IL‐8 secretion relative to medium control, and also inhibit some pro‐inflammatory responses after TLR 2/6 and 3 agonist induction (Rose et al. 2012). These studies continue to support the notion that the innate immune response is largely driven by vaginal bacterial community states, with CST‐IV potentially having a larger pro‐inflammatory response than CST‐I or CST‐II, and with CST‐III triggering an intermediate response.

Additional factors contributing to vaginal defence include mannose binding lectin (MBL), vaginal antimicrobial peptides (AMPs) and immunoglobulin A and G (IgA, IgG). As its name suggests, MBL binds mannose, N‐acetylglucosamine and fucose carbohydrate moieties present on microbial cell surfaces. Eventually, this interaction leads to cell lysis or targeting for the immune system (Neth et al. 2000; Turner, 2003). IgA and IgG may help to prevent vaginal epithelial cell adherence and uptake, as well as contribute to the neutralization and clearance of infectious microbes from the vagina (Tramont, 1977; Wang et al. 2014). Vaginal AMPs exist in various classes and may recruit immune cells via chemotaxis or possess anti‐endotoxin activity. Mechanisms for each AMP have been thoroughly reviewed elsewhere (Ding et al. 2009; Wilson et al. 2013; Yarbrough et al. 2015), and while the specific association between AMPs and vaginal microbiota has not been extensively investigated, key findings are emphasized here. Defensins are a class of cationic and amphipathic AMPs with diverse mechanisms of action against common vaginal bacteria, pathogens and viruses including HIV, herpes simplex virus (HSV) and human papillomavirus (HPV). In organotypic models of the vaginal epithelium, human β‐defensin (HBD)‐2 expression, but not that of HBD‐1, was associated with colonization by L. iners, Atopobium vaginae and Prevotella bivia (Doerflinger et al. 2014), while in another study using similar experimental in vitro conditions, L. jensenii but not Gardnerella vaginalis was shown to induce HBD‐2 transcription (Valore et al. 2006). As expected, many human defensins bind to viral‐specific proteins to prevent viral attachment to a human cell surface, as for example, with retrocyclin‐1, retrocyclin‐2, human neutrophil peptide (HNP)‐1, HNP‐2, HNP‐3 and to a much lesser degree HNP‐4 (Maddon et al. 1986; Münk et al. 2003; Wang et al. 2003, 2004; Wu et al. 2005; Furci et al. 2007); however, the topic is outside the scope of this review. In addition to defensins, other AMPs are found in the human vagina and include the secretory leukocyte protease inhibitor (SLPI), human epididymis protein 4 (HE4), LL‐37, and surfactant protein (SP)‐A and SP‐D. SLPI expression is associated with BV organisms (Nasioudis et al. 2015) but not with L. crispatus, L. iners, A. vaginae or P. bivia (Doerflinger et al. 2014; Orfanelli et al. 2014). HE4 is associated with G. vaginalis (Orfanelli et al. 2014) and LL‐37 inactivates the sexually transmitted pathogen Neisseria gonorrhoeae while having no effect on L. crispatus and L. jensenii and comparatively little effect on L. iners (Moncla et al. 2012). The lack of AMP stimulation in response to some Lactobacillus spp. is associated with their needed maintenance in the vagina (Valore et al. 2002). Similar to defensins, SP‐A and SP‐D contribute to viral inhibition, including HIV where they act via binding to the viral protein gp120 and human CD4, but with SP‐A simultaneously enhancing gp120 binding to dendritic cells and therefore also facilitating HIV uptake. (Gaiha et al. 2008; Pandit et al. 2014). Thus, overall, microbes, environments, immune regulatory actions and genes tightly interact to govern homeostasis of the vaginal environment.

Bacterial vaginosis and aerobic vaginitis

As mentioned, the vaginal microbiota can be characterized by one of five CSTs, with CST‐IV lacking a relatively high abundance of Lactobacillus spp. Generally, CST‐IV can clinically manifest as aerobic vaginitis (AV) or BV, so the immune response to CST‐IV outlined above overlaps considerably with BV or AV. AV is mainly differentiated from BV by the presence of an inflammatory response predominately associated with aerobes, such as group B Streptococcus, Staphylococcus aureus, Escherichia coli, and Enterococcus (Donders et al. 2002; Donders, 2007). The AV inflammatory response manifests symptomatically as itching or burning, molecularly as increased IL‐6 and IL‐1β, and cellularly as the presence of leukocytes or primary blood cells in a microscopic wet mount (Han et al. 2014). In contrast, the clinical definition of BV does not involve any overt inflammatory responses such as recruitment of neutrophils, redness, itching or burning (reviewed in Cauci, 2004). A number of immune factors including IL‐1β, IL‐2, IL‐4, IL‐6, IL‐8, IL‐10, IL‐12, TNF‐α, IFN‐γ, chemokine C‐C motif ligand 5 (CCL5) and SLPI have been variably and inconsistently associated with BV (summarized in Mitchell & Marrazzo, 2014). These conflicting findings may be due to different study designs (longitudinal versus cross‐sectional, or in vitro versus in vivo), different definitions of BV (symptomatic versus asymptomatic BV or Nugent‐BV versus BV diagnosed according to the Amsel criteria) or that additional features actively suppress the inflammatory response in BV, such as IgA degradation, TLR expression inhibition, or immune‐related genetic variants (Cauci et al. 2003; Witkin et al. 2007b). As an example of BV's effect on host defence, cytokine analysis from a vaginal epithelial cell model co‐colonized with mock communities representing CST‐I to ‐IV as well as Nugent scores corresponding to respective BV diagnosis showed significant increases in IL‐1β, IL‐8, TNF‐α, CCL5 and IL‐1RA in CST‐III or CST‐IV, but not CST‐I or CST‐II (Pyles et al. 2014). The ability of individual BV‐associated bacterial species to elicit an in vitro immune response has also been studied, as in the cases of A. vaginae, which induces expression of chemokine C‐C motif ligand 20 (CCL20), HBD‐2, IL‐1β, IL‐6, IL‐8 and TNF‐α via nuclear factor‐κB (NF‐κB), TLR2 and MyD88 signalling pathways; G. vaginalis, which induces IL‐6 and IL‐8 transcripts; and L. iners, which stimulates PRR signalling but not downstream inflammatory response cytokines IL‐6, IL‐8 or mucins (Libby et al. 2008; Doerflinger et al. 2014). Bacteria‐derived short chain fatty acids (SCFAs), namely acetate, butyrate, propionate and succinate, some of which exist at relatively higher proportions during BV, can induce a pro‐inflammatory response under the hypothesis that SCFAs may act to ultimately inhibit chemotaxis and inflammation in BV (Al‐Mushrif et al. 2000; Chaudry et al. 2004; Mirmonsef et al. 2011; Gajer et al. 2012; O'Hanlon et al. 2013). Relatively high concentrations (2–20 mm) of acetate and butyrate, but not propionate, induce cytokine IL‐6, IL‐8 and IL‐1β secretions and also induce IL‐8 and TNF‐α with TLR2 and TLR7 ligand stimulation in a dose‐ and time‐dependent manner in vitro (Mirmonsef et al. 2011). However, whether or not the host is actively downplaying the sensing of BV‐associated microbes or a specific attribute of BV is evading inflammation remains to be demonstrated since the aetiology of BV is still unknown and the necessary longitudinal studies are lacking (Schwebke, 2009).

Lactic acid and host defence

Lactic acid is produced mainly by vaginal microbes (Boskey et al. 2001) and helps maintain healthy host physiological functions since it has been shown to directly inhibit Chlamydia trachomatis infection (Gong et al. 2014), and potentially both HSV‐2 and HIV in vitro and in vivo if there is sufficient lactic acid to acidify the vagina to pH < 4 (Conti et al. 2009; Aldunate et al. 2013; Isaacs & Xu, 2013, 2014). Lactic acid also inactivates a broad range of BV‐associated microbes at pH < 4.5 (O'Hanlon et al. 2011). When Lactobacillus spp. dominate the vaginal microbiota, they acidify the vagina to a strongly acidic mean pH of 3.5 ± 0.2 that is likely to help protect against a broad range of infections (O'Hanlon et al. 2013). Recent studies aimed to uncover the mechanism by which lactic acid can directly affect host immune functions, as for example by directly inhibiting pro‐inflammatory responses IL‐6, IL‐8 and IL‐1RA, (Hearps et al. 2014), inducing the Th17 lymphocyte pathway via IL‐23 in a dose‐dependent manner upon lipopolysaccharide co‐stimulation (Witkin et al. 2011), and helping to release mediators from vaginal epithelial cells and stimulating antiviral response by release of transforming growth factor‐β (Mossop et al. 2011). In the gut, lactate and acetate from L. casei and Bifidobacterium breve inhibit cell proliferation, but whether these molecules play a similar role in the vagina has not been studied (Matsuki et al. 2013). Interestingly, lactic acid isomers may also play a role in determining host response and the subsequent host–microbiota relationship. Lactic acid exists in the vagina in both d‐(–)‐ and l‐(+)‐isomers, with the host contributing only about 4–30% of the total lactate (Boskey et al. 2001), suggesting a large reliance on microbes to supply the majority of lactic acid for protection. In one study, only d‐(–)‐lactic acid was correlated with α‐amylase, SLPI, hyaluronidase‐1, neutrophil gelatinase‐associated lipocalin and matrix metalloproteinase 8 (MMP‐8) expression in vitro (Nasioudis et al. 2015). The authors suggest that epithelial cell exfoliation and subsequent breakdown of glycogen helps favour Lactobacillus spp. growth, and thus helps sustain d‐(–)‐lactic acid production (see discussion on α‐amylase and glycogen below). Moreover, women with BV were found to be deficient in both isomers, while those with vulvovaginal candidiasis have elevated l‐(+)‐lactic acid as well as CD147 and MMP‐8 genes (Beghini et al. 2014). Lactobacillus iners does not produce d‐(–)‐lactic acid and fails to produce the l‐(+)‐lactic acid in abundance as high as L. crispatus or L. gasseri while L. jensenii produces only d‐(–)‐lactic acid (Witkin et al. 2013), suggesting potential Lactobacillus species‐specific effects on the host. Consequently, the composition of the vaginal microbiota, and specifically the ability of vaginal microbes to produce d‐(–)‐lactic acid, may help to inhibit pathogens and inflammatory responses while also favouring Lactobacillus spp. survival by using host cells resources for carbon sources.

The effect of the microbiota on the vaginal mucosa

The vaginal mucosa plays an important role as a physical barrier to separate host epithelial environment from harmful pathogens, including HIV (Nunn et al. 2014, 2015), whereas vaginal microbes can affect the integrity of the mucosa (Arnold et al. 2014). The BV‐associated species G. vaginalis secretes sialidases, which have been shown to deglycosylate secretory immunoglobulin A (sIgA) and other sialoglycan substrates via the cleavage of sialic acid from gylocoprotein at the α2–3 and α2–6 linkage Neu5Ac present on both N‐ and O‐glycans, thereby hydrolysing protective mucosal sialoglycoproteins (Lewis et al. 2012). Gardnerella vaginalis can consume and neutralize liberated sialic acid residues to further evade the host response (Lewis et al. 2013). In addition, G. vaginalis produces vaginolysin, a cholesterol‐dependent cytolysin that could contribute to BV symptomology by forming pores in the vaginal epithelium with or without CD59 (Gelber et al. 2008; Zilnyte et al. 2015). BV‐associated bacteria quantified by Nugent score also significantly associate with mucins‐1, ‐4, ‐5AC and ‐7 (Moncla et al. 2014). Conversely, certain types of vaginal communities could enhance the integrity of the mucosal barrier. A recent study showed that L. crispatus‐dominated vaginal microbiota were able to reinforce the diffusional barrier properties of cervicovaginal mucus against HIV, hence hindering HIV penetration, while communities dominated by L. iners facilitated the penetration of HIV through the cervicovaginal mucus barrier (Nunn et al. 2015). Thus, a change in vaginal community composition and function is strongly associated with the integrity of the protective mucus layer. Therefore, vaginal bacteria, including species of Lactobacillus, can reduce or increase susceptibility to infectious agents such as HIV.

The effect of the microbiota on reproductive functions

The vaginal microbiota in combination with other factors is associated with adverse reproductive and obstetric outcomes. For example, a meta‐analysis revealed that BV‐like vaginal microbiota are significantly more prevalent in women with tubal infertility when compared with women with other causes of infertility, but is not associated with decreased conception rates (van Oostrum et al. 2013). Preterm labour and delivery has been thought to be in part associated with changes in the vaginal microbiota composition, namely bacteria found in CST‐IV (i.e. G. vaginalis and Ureaplasma), AV or BV (preterm labour and delivery defined as occurring before 37 weeks in Caucasian women (Donders et al. 2009), mostly African American women (Nelson et al. 2014), and mostly Caucasian women (DiGiulio et al. 2015)). Jakovljević et al. assessed differences of gestational time to delivery in BV versus non‐BV women, with a statistically significant difference of 37.72 ± 3.9 versus 39.59 ± 1.1 weeks (Jakovljević et al. 2014). However, another study did not find any association between CST‐IV and preterm births (defined as 28–33.1 weeks versus term births of 38.8–40.7 weeks in mostly African American women) even though the study was well‐powered and preterm births were phenotypically well controlled (Romero et al. 2014). It should be noted that differences in ethnicity, definition of preterm birth and analytical methods of microbiota data could explain these different observations. The possibility remains that functional differences exist and that hypotheses should be further explored using metagenomics‐ or metatranscriptomics‐based approaches.

The interplay between host polymorphism and the vaginal microbiota could play an important role in the mechanisms by which microbes affect reproductive health. Polymorphisms in genes that control inflammatory response (protein kinase Cα, fms‐like tyrosine kinase 1, IL‐6 and TNF‐α) are associated with preterm delivery in combination with CST‐IV vaginal microbiota, although the direct functional impact of these polymorphisms is unknown (Gómez et al. 2010; Jones et al. 2010). These studies and others have suggested that the inflammatory and antimicrobial peptide response, associated with certain vaginal microbiota, exhibit a role in rupturing and invading cervical plug or amniotic membranes, eventually triggering pro‐inflammatory cascades that could lead to premature labour and delivery (reviewed in Goldenberg et al. 2000; Witkin, 2014; Yarbrough et al. 2015). In a rhesus monkey model infected with group B Streptococcus, there was an observed increase in amniotic fluid cytokines TNF‐α, IL‐1β and IL‐6 occurring before uterine contractility or any clinical signs of infection, suggesting a direct role of infection in preterm labour (Gravett et al. 1994). Specific AMPs are expressed upon exposure in vitro and in vivo to the BV‐associated bacteria A. vaginae (CCL20, HBD‐2) or pathogens, such as C. trachomatis (elafin) and N. gonorrhoeae (SLPI), but not L. crispatus or L. jensenii (Libby et al. 2008; King et al. 2009; Cooper et al. 2012; Eade et al. 2012; Doerflinger et al. 2014). The immune response to pathogens can trigger signalling cascades, which could ultimately lead to miscarriage, intrauterine infection, preterm labour, and tubal and ectopic pregnancy. For example, if C. trachomatis ascends past the cervix, such an infection in the upper genital track could lead to tubal scarring and potentially tubal infertility, ectopic pregnancy and chronic pelvic pain (Hafner, 2015, and discussed in Barlow et al. 2001). C. albicans triggers IL‐6 secretion in the placenta ultimately leading to an increase in NF‐κB and an inflammatory response (Tyutyunnik et al. 2014). A study measuring the effects of immunomodulation in pregnant women with BV found that C. albicans and Trichomonas vaginalis, but not C. trachomatis or N. gonorrhoeae, had an effect on vaginal cytokines and none of these pathogens had any effect on anti‐G. vaginalis haemolysin IgA, sialidase or prolidase activity (Cauci & Culhane, 2007). Clearly there is still much to learn about the dynamics, function and mechanisms driving the role of the vaginal microbiome in reproduction health.

Physiology affects vaginal microbiota composition

The composition of the vaginal microbiota changes throughout a woman's lifetime from birth, through puberty, reproductive age and menopause. The early childhood vaginal microbiota comprise a variety of anaerobes, diphtheroids, coagulase‐negative staphylococci, and E. coli, whereas postmenopausal women often experience a loss of Lactobacillus spp. associated with the decrease in oestrogen controlling vaginal epithelial proliferation, maturation, and accumulation of glycogen, which is directly or indirectly nutritionally necessary for the maintenance of Lactobacillus spp. (Hammerschlag et al. 1978a,b; Hillier & Lau, 1997; Galhardo et al. 2006; Brotman et al. 2013b). Indeed, oestrogen levels peak during reproductive age and contribute to shaping the composition of the vaginal microbiota. In menopause, vaginal application of oestrogen cream is associated with vaginal epithelial maturation, the accumulation of glycogen and acidic pH (<4.0), the latter indicative of the presence of high number of Lactobacillus spp. (Nilsson et al. 1995). Interestingly, Lactobacillus spp. was originally thought to directly ferment glycogen in the vagina. However, this idea was gradually refuted and recent evidence suggests that human α‐amylase catabolizes glycogen into smaller polymers, namely maltose and maltotriose, which can then be used by Lactobacillus spp. for metabolism, even in newborns who have residual circulating maternal oestrogen (Cruickshank, 1934; Cruickshank & Sharman, 1934; Stewart‐Tull, 1964; Bernbaum et al. 2007; Spear et al. 2014). This model puts forward that the influence of oestrogen, glycogen and especially α‐amylase provides a positive selection pressure for a Lactobacillus‐dominated microbiota (Spear et al. 2014). These findings highlight the tight interplay between host physiology and the vaginal microbiota.

Conclusion

The human vaginal ecosystem is a dynamic environment in which microbes can affect host physiology but also where host physiology can affect the composition and function of the vaginal microbiota. Species of Lactobacillus have been historically associated with vaginal health in reproductive‐age women due to the direct and indirect protective nature afforded by Lactobacillus products, such as lactic acid and bacteriocin among others, against mucus degradation and inhibition of pathogens. The reported inconsistent innate immune response observed with non‐Lactobacillus‐ or L. iners‐dominated microbiota (CST‐IV, BV, AV and CST‐III, respectively), coupled with recent findings that question the definitions of normality, highlight the need for more in‐depth functional understandings of the interaction between the vaginal microbiota and host physiology, reproduction and defence.

Additional information

Competing interests

None declared.

Author contributions

Both authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported in part by the National Institute of Allergy and Infectious Diseases and the National Institute for General Medical Sciences of the National Institutes of Health under awards number R01GM103604, R01AI089878, R21AI107224, R01AI116799 and U19AI084044.

Biographies

Steven Smith is a PhD candidate at the University of Maryland with a focus on bioinformatics and computational biology. His past and current experiences include integration of experimental and high‐throughput sequencing datasets for use in systems biology approaches. He is currently studying how changes in vaginal bacterial community composition and abundance and associated BV status affect human vaginal transcriptomic responses.

Jacques Ravel is a Professor of Microbiology and Immunology and the Associate Director for Genomics at the Institute for Genome Sciences (IGS) at the University of Maryland School of Medicine in Baltimore, Maryland. His research focuses on applying combinations of modern genomics technologies, systems biology approaches and ecological principles to understand the relationship between the host and the genital microbiota as it relates to women's health.