Roles of the Microbiota of the Female Reproductive Tract in Gynecological and Reproductive Health

SUMMARY

The microbiome of the female reproductive tract defies the convention that high biodiversity is a hallmark of an optimal ecosystem. Although not universally true, a homogeneous vaginal microbiome composed of species of Lactobacillus is generally associated with health, whereas vaginal microbiomes consisting of other taxa are generally associated with dysbiosis and a higher risk of disease. The past decade has seen a rapid advancement in our understanding of these unique biosystems. Of particular interest, substantial effort has been devoted to deciphering how members of the microbiome of the female reproductive tract impact pregnancy, with a focus on adverse outcomes, including but not limited to preterm birth. Herein, we review recent research efforts that are revealing the mechanisms by which these microorganisms of the female reproductive tract influence gynecologic and reproductive health of the female reproductive tract.

INTRODUCTION

The female reproductive tract is composed of the vagina, the cervix, the uterus, the fallopian tubes, and the ovaries, and the cervix connects the upper reproductive tract to the vagina (Fig. 1) (1). The existence and invasion of microbes in the female reproductive tract have long been known to impact genital and reproductive health. Species of Lactobacillus, generally the most abundant taxa in the vaginal microbiome (VMB), produce lactic acid (2) and probably bacteriocins (3) that inhibit dysbiosis-associated microbes and work to maintain homeostasis and reduce risks of disease. A Lactobacillus-dominated VMB has been the hallmark of female reproductive health. The VMB is less complex than the microbiomes of other body sites, e.g., the oral cavity and the gastrointestinal tract, and the compositions of the microbiomes of the posterior fornix and introitus generally reflect that of the vaginal wall (4). Compared to that in early pregnancy, the VMB in late pregnancy trends toward an even more stable and Lactobacillus-dominated state (5–8), possibly an evolutionarily selected mechanism to ensure a successful pregnancy. VMBs dominated by other taxa, e.g., those associated with bacterial vaginosis (BV), are generally considered suboptimal and have been associated with a higher risk of adverse health (9, 10), including an increased risk of bacterial, viral, and parasitic sexually transmitted infections (STIs) (11), and adverse pregnancy outcomes, including but not restricted to preterm birth (PTB) (5, 12–15). Inflammation caused by microorganisms invading the upper genital tract can lead to adverse pregnancy outcomes, but the source of such microbes remains unclear, and the existence of a natural microbiome in the upper genital tracts of healthy females remains controversial (16, 17).

FIG 1.

Microbiome of the female reproductive tract. In the pie chart showing the composition of the VMB, the taxa are color coded and assigned to the species level. Species belonging to the genus Lactobacillus are highlighted by the gray stripe at the periphery of the pie chart. Data in the pie chart are from a previous study of 2,582 asymptomatic reproductive-age women, 27.2% of whom were pregnant and who had a racial distribution of 52% Black, 20% white, and 28% non-Black Hispanic (8). The VMB can be classified to five community state types (CSTs), and the predominant species in these CSTs are listed. The composition of the cervical microbiome is similar to that of the VMB. The existence of a microbiome in the upper female reproductive tract is still controversial.

Recent studies have provided a deeper understanding of the mechanisms by which the microbiota of the female reproductive tract affects gynecologic and reproductive health (5, 15, 18–23). Herein, we review the microbes in multiple niches of the female reproductive tract and their apparent impacts on human health. Mechanisms by which microbes contribute to maintain overall vaginal health or increase risk for adverse reproductive health are discussed. A particularly impactful target of recent research has been the contributions of microorganisms in the female reproductive tract to adverse pregnancy outcomes. Our review invokes research from the past decade, during which advanced multi-omics technologies, including high-throughput genomic and transcriptomic analyses, have led to rapid advancement of the field. Strategies for the prediction, prevention, and possible intervention to prevent adverse reproductive health outcomes are discussed.

THE VAGINAL MICROBIOME

Composition of the Human Vaginal Microbiome

The initial objective of the National Institutes of Health Human Microbiome Project (HMP; www.hmpdacc.org/hmp), launched in 2008, was to define the human microbiome in health. These early studies confirmed that, compared to the microbiomes of other body habitats, such as the oral cavity and the gastrointestinal tract, the VMB of asymptomatic white reproductive-age females exhibits the lowest community richness and diversity, that it is usually dominated by members of the genus Lactobacillus (4, 24) (Fig. 1), and that the microbiotas of the vaginal introitus, midvaginal wall, and posterior fornix show little distinction (4). Studies of more racioethnically diverse cohorts showed that Lactobacillus spp., i.e., Lactobacillus crispatus, L. gasseri, L. jensenii, and L. iners, and several other taxa, e.g., BV-associated bacterium 1 (BVAB1; “Candidatus Lachnocurva vaginae”), Gardnerella vaginalis, Sneathia amnii, and others (Fig. 1), are the most abundant species in the VMB (5, 8, 25). A landmark study employing high-throughput 16S rRNA taxonomic profiling classified VMBs of nonpregnant reproductive-age women with diverse racioethnicity into four community state types (CSTs) dominated by Lactobacillus spp., i.e., CST I (L. crispatus), CST II (L. gasseri), CST III (L. iners), and CST V (L. jensenii), and a fifth, CST IV, that is more complex and dominated by several anaerobic species (25) (Fig. 1). A more recent study subdivided CST IV into 7 subtypes dominated by different non-Lactobacillus species (26). An alternate but similar approach to classification of VMBs places them into “vagitypes” based on the dominant taxon in the sample (5, 8, 27). Recently, the VIRGO database, which permits classification of the vaginal bacteria at subspecies levels using metagenomic and metatranscriptomic data, was established (28). These classification approaches are based on bioinformatic analyses of taxonomic or gene profiles of the VMB as revealed by high-throughput nucleic acid sequencing. In diagnostic settings, clinical and semiquantitative microscopic observations are generally used to classify the vaginal microbiome (29, 30); i.e., Amsel’s criteria or the semiquantitative Nugent score is used to diagnose BV, and Donder’s score is used for diagnosis of aerobic vaginitis (see below).

Host Factors Affecting the Composition of the Vaginal Microbiome

Estrogen is a key host factor in maintaining the vaginal microbiome. It promotes thickening of the vaginal epithelium and generation of intracellular glycogen (9). Host glycogen can be hydrolyzed to glucose and maltose by host alpha amylase, and lactobacilli ferment glucose and maltose into lactic acid, lowering the local pH (2) (Fig. 2). In addition, both in vitro and in silico studies illustrate that multiple vaginal bacteria encode amylase-like enzymes that can metabolize glycogen into glucose and maltose (31–33). Glycogen is probably released from lysed epithelial cells that are induced by high concentrations of lactic acid and cytolysins excreted by Lactobacillus species, and the release of glycogen could be associated with hyaluronidase-1 and matrix metalloproteinase-8 (2). As estrogen levels increase at puberty, the pH of the female genital tract becomes acidic, and meta-analyses on multiple cohorts indicate that the VMB becomes more stable and more likely to be dominated by acidophilic Lactobacillus spp. (6, 34, 35). Likewise, the stability of the VMB and the concentration of estrogen fluctuate during a menstrual cycle and pregnancy, but a positive correlation between these two subjects is consistently observed; i.e., a higher level of estrogen is associated with an increased stability of the VMB, apparently promoting vaginal health (Fig. 3). The opposite occurs after pregnancy and during menopause as estrogen levels decrease, and estrogen therapy of postmenopausal women maintains a more Lactobacillus-dominated state (36, 37). Not surprisingly, use of estrogen-containing contraceptives tends to increase Lactobacillus prevalence (38) while reducing the incidence of BV (39) in reproductive-age women. These data further support the causal relationship between estrogen and the modulation of the VMB. Unlike estrogen, the impact of progesterone on the VMB is not consistent in different studies, as shown in a recent review (40), while an increased level of testosterone seems to be associated with a more complex VMB of women with polycystic ovary syndrome (41).

FIG 2.

Role of the vaginal microbiome in gynecologic and obstetric health. Maternal estrogen promotes the production of glycogen in vaginal epithelial cells. Glycogen released into the female reproductive tract by detachment from or lysing epithelial cells is metabolized to glucose and maltose by human and bacterial α-amylases. Glucose and maltose can be further fermented to lactic acid by Lactobacillus, consequently supporting proliferation of members of this genus. Lactic acid decreases the vaginal pH and, as a result, inhibits the colonization and proliferation of dysbiosis-associated species. Bacteriocins may play roles in controlling colonization and proliferation of microbes generally linked to BV and other adverse conditions. However, lactic acid seems to inhibit host inflammation potentially induced by infections with opportunistic microbes by increasing the anti-inflammatory cytokine IL-1RA and inhibiting proinflammatory cytokines and chemokines, including IL-6, IL-8, TNF-α, RANTES, and MIP-3α. Furthermore, Lactobacillus taxa also seem to promote the integrity of the mucosal and epithelial barriers, thus helping to prevent establishment of these deleterious microbes. In contrast, less favorable taxa induce adverse outcomes by inducing inflammation (circle 1), undergoing biofilm formation (circle 2), and producing other virulence factors, including but not limited to toxins (e.g., vaginolysin and inerolysin), proteases, mucinases, or sialidases (circle 3). See Table 2 for additional BV-associated virulence factors.

FIG 3.

Shift of the vaginal microbiome in pregnancy. Estrogen is one of the key factors that modulate the stability of the VMB. The concentration of estrogen and the stability of the VMB increase during pregnancy. The abundance of lactobacilli increases at the expense of species often related to vaginal dysbiosis, e.g., G. vaginalis, A. vaginae, and P. bivia. This shift in the VMB occurs early in pregnancy and is more pronounced in Black women. After delivery, estrogen levels, along with the stability of the VMB, decrease greatly, and the VMB converts from an optimal state to a state of dysbiosis, which often takes over 40 weeks to recover.

Racioethnicity is another important factor that discriminates the composition of the VMB of reproductive-age women (8, 42, 43). The most dominant taxon in white women is generally L. crispatus, whereas L. iners is prevalent in Asian, Hispanic, and Black women (8, 25). The average vaginal pH of white and Asian women is lower than that of Hispanic and Black women, consistent with a higher abundance of Lactobacillus (25). The VMBs of Black women are less stable, consistent with their more complex microbial communities (8, 43), and exhibit a much higher prevalence of BVAB1, G. vaginalis, S. amnii, and several other anaerobes associated with BV. Black women also suffer a higher risk for STIs and adverse pregnancy outcomes (5, 8, 43, 44). Both genetic differences (45, 46), e.g., known sequence variants among individuals of different races, and environmental differences (47–49), e.g., socioeconomic status and stress, among races likely contribute to these racioethnic differences in VMB composition, and several studies have identified a genetic link between a woman’s genetic background and specific bacterial taxa (45, 46, 50). Moreover, these studies have suggested a genetic association between racial background and certain bacterial taxa in the VMB (45, 50). Hence, the impact of racioethnicity on the VMB is likely mediated by multiple factors, including both the individual’s genetic background and environmental conditions. Since structural racism is key in influencing environmental conditions (51), the difference of the VMB associated with racioethnicity could be potentially caused by structural racism (52). However, current studies have yet to quantify the contribution of these factors to the composition of the VMB, and the mechanisms by which these factors impact the VMB remain unclear.

Lactobacilli and Vaginal Health

Lactobacillus spp. are usually the dominant taxa in the VMB, and these taxa play important roles in antagonizing dysbiosis-associated microorganisms. As described above, lactobacilli lower the vaginal pH (2). Low pH inhibits proliferation of anaerobes commonly associated with BV (53, 54) (see below) and is considered partially responsible for reduced susceptibility to human immunodeficiency virus (HIV) infection (22, 55, 56), other viral infections (57) and other STIs (11) (Fig. 2). Lactic acid also permeabilizes the outer membranes of Gram-negative taxa, possibly potentiating the impact of other factors on these bacteria in vitro (58). The ability of many species of Lactobacillus to produce H₂O₂ could provide a selective advantage over other potential vaginal colonizers under aerobic conditions, but the largely anaerobic environment in the female reproductive tract would likely preclude in vivo production of H₂O₂ (59). Lactobacillus spp. also produce bacteriocins, which target other bacterial taxa, permitting the former to proliferate. Multiple putative bacteriocin genes have been identified in the genomes of L. crispatus isolates recovered from cultures of vaginal specimens (60). Gassericin E, identified in a L. gasseri strain, was shown to inhibit other species, including G. vaginalis, in vitro (3). Although the antimicrobial spectrum of these bacteriocins may influence the composition of the VMB, there remains a lack of in vivo evidence that Lactobacillus bacteriocins function to attenuate growth or colonization of other taxa. Finally, in vitro studies showed that various adhesins produced by Lactobacillus spp. promote their colonization of epithelial surfaces (61–64) and inhibit colonization of dysbiosis-associated microorganisms, e.g., G. vaginalis (61) and Escherichia coli (62). L. crispatus also reduces Candida adhesion by producing a biosurfactant in vitro (65) and inhibits Candida albicans infection by promoting epithelial cell defenses through modulation of the production of Toll-like receptors (2 and 4), interleukin 8 (IL-8), and β-defensins 2 and 3 in a HeLa cell model (66).

Lactobacillus spp. interact with the host to impact overall vaginal health. Multiple vaginal epithelial cell models show that d- and l-lactic acid and a mixture of VMB-associated metabolites induce the anti-inflammatory cytokine IL-1RA while inhibiting proinflammatory cytokines, including IL-6, IL-8, TNF-α, RANTES, and macrophage inflammatory proteins 3α (MIP-3α) (67, 68) (Fig. 2). IL-1RA induction and IL-8 inhibition by l-lactic acid was confirmed in an organotypic tissue model of female reproductive tract epithelium (67). Although other in vitro studies suggested that several Lactobacillus strains stimulated proinflammatory responses (69, 70), none of these strains are abundant in the VMB. Both in vivo and cervical epithelial cell model studies show that lactic acid produced by lactobacilli and the resulting acidic vaginal pH also promote integrity of the vaginal epithelial barrier inhibiting colonization by other anaerobes and pathogens (Fig. 2) (22, 71). Thus, it is clear that the predominance of protective lactobacilli is associated with lower risks of suboptimal health states. However, overgrowth of lactobacilli can be associated with cytolytic vaginosis, defined by epithelial cell damage, lysis, and scaling due to overproduction of lactic acid (20, 72–79). Cytolytic vaginosis is less common than BV but emphasizes the importance of quantitative assessment of the VMB in clinical settings.

L. iners and L. crispatus are generally the most abundant Lactobacillus spp. in the VMB, followed by L. gasseri and L. jensenii (Fig. 1). L. crispatus, L. jensenii, and L. gasseri produce H₂O₂, and L. crispatus is associated with lower vaginal pH (25, 80) and is heritable in white women (45). Furthermore, the L. crispatus-dominated VMB yields higher levels of d- but not l-lactic acid than the L. iners-dominated VMB, and decreased d-lactic acid has been associated with PTB (81). As described above, a lower pH both inhibits dysbiosis and generally promotes anti-inflammatory and antibacterial effects (67). The L. crispatus-dominated VMB is more stable during pregnancy (8, 14, 24) and is often reduced in prevalence in pregnant women who go on to experience PTB (5, 12, 14, 82). Supernatants of L. crispatus attenuate disruption of the cervical epithelial barrier mediated by LPS or G. vaginalis and reverse the G. vaginalis-induced inflammation-associated microRNA (miRNA) expression in cervical epithelial cell models (71). L. iners often predominates in Black women (8). It does not produce H₂O₂ (80), is less efficient than other lactobacilli in production of d-lactic acid (81, 83), produces a cholesterol-dependent cytolysin (inerolysin) (84), and does not inhibit disruption of the cervical epithelium by lipopolysaccharide (LPS) or G. vaginalis (71). Moreover, in contrast to L. crispatus, which is largely exclusionary to other bacterial taxa, L. iners often coexists with G. vaginalis in vivo (12) and has been associated with PTB, miscarriage, and instances of an insufficient cervix during pregnancy (14, 82). Although lactobacilli, especially L. crispatus, seem beneficial for overall vaginal health, the role of L. iners in maintaining vaginal health is less clear (85). Recent strain-level studies illustrate that L. iners genomes are quite conserved and similar with regard to single nucleotide polymorphisms (SNPs) and the presence of characterized genes (24, 86) but differ in the presence of phages, plasmids, and some uncharacterized genes (24), and the impact of these genes on the VMB and women’s health requires further exploration. It remains unclear if these apparent strain differences are associated with the differential impact of L. iners on women’s reproductive health. Interestingly, expression of several L. iners genes, including those encoding inerolysin, mucin, glycerol transport and related metabolic enzymes, and proteins belonging to a CRISPR system, are upregulated in the VMB of women with BV (87), suggesting that the modulation of virulence gene expression in L. iners is associated with dysbiosis of the VMB, but the mechanism remains unclear.

A recent large-scale metabolomics study illustrates that vaginal cytokine profiles and the prevalence of Lactobacillus in the VMB can be predicted by metabolite profiles of the VMB using machine learning models (88). These models can further distinguish between L. crispatus-dominated and L. iners-dominated VMBs using metabolite profiles. Thus, this study demonstrates the interaction among the composition of the VMB, microbial metabolites, and host immune responses.

The Vagina Microbiome in Adverse Health Conditions

Bacterial vaginosis.

Vaginitis is a term to describe various conditions of infection or inflammation of the vagina. The most common kinds of vaginitis are BV, vulvovaginal candidiasis or “yeast” vaginitis, and trichomoniasis vaginitis. BV, with a global prevalence ranging from 23% to 29% (89), is characterized by displacement of lactobacilli in the VMB by anaerobic Gram-negative bacteria (44). Although many women with BV-like vaginal microbiomes lack clinical complaints, a recent meta-analysis estimated that ~34.9% are actually symptomatic (89). These women suffer from increased vaginal discharge, odor, and itching, generally without significant local inflammation (90). Diagnosis by Amsel’s criteria requires the presence of three of the following: vaginal discharge, a pH of ≥4.5, presence of sloughed “clue” cells coated with bacteria, and an amine odor with application of potassium hydroxide to the discharge (91). Nugent scoring (30) involves enumerating large Gram-positive rods (lactobacilli), small Gram-variable rods (G. vaginalis), small Gram-negative rods (Bacteroides spp.), and curved Gram-variable rods (other taxa). Several taxa, including Gardnerella, Atopobium, Prevotella, Porphyromonas, Sneathia, Mobiluncus, Mycoplasma, BVAB1, BVAB2, Mageeibacillus indolicus (BVAB3), and Peptostreptococcus, are often enriched in VMBs of BV patients (44) (Table 1). Although Mobiluncus spp. have been considered a primary contributor to the curved rods observed in BV, recent molecular studies suggest that these curved rods are mostly BVAB1 (44). Treatment of BV with metronidazole is often initially successful, but recurrence is common at a rate approaching 50% within as little as 12 weeks (92) to 12 months (93, 94).

TABLE 1.

Composition of the vaginal microbiome in vaginitis^a

Condition	Vaginal microbiome features
Cytolytic vaginosis	Overgrowth of Lactobacillus spp. (20)
Bacterial vaginosis	Enrichment of Atopobium vaginae,b BVAB1, BVAB2, BVAB3, Gardnerella vaginalis, Mobiluncus spp., Mycoplasma spp., Porphyromonas spp., Prevotella spp., Sneathia spp., and Ureaplasma spp. (44); depletion of Lactobacillus spp. (44)
Vulvovaginal candidiasis	Normal to BV-like vaginal microbiome (19, 120, 122); colonization by Candida albicans, Candida glabrata, Candida parapsilosis, Candida tropicalis, or Candida krusei (19, 120, 122)
Trichomoniasis	BV-like vaginal microbiome (133, 134); invasion by Trichomonas vaginalis (133, 134)
Aerobic vaginitis or desquamative inflammatory vaginitis	Depletion of Lactobacillus spp. (29, 139); frequent detection of aerobic, enteric bacteria, e.g., E. coli, Enterococcus spp., Staphylococcus spp., and Streptococcus spp. (29, 139)

^aBVAB1, “Candidatus Lachnocurva vaginae”; BVAB2, Lachnospiraceae BV-associated bacterium 2; BVAB3, Mageeibacillus indolicus.

^bA. vaginae was reclassified to a new genus and renamed Fannyhessea vaginae in 2018 (266).

G. vaginalis is present in at least 95% of clinically diagnosed BV (95), but its role remains unproven. G. vaginalis isolates have recently been reclassified into four clades and 13 putative genomic species, including “true” G. vaginalis, G. leopoldii, G. piotii, and G. swidsinskii (96, 97). A recent study reported that clades 1 and 2, which include G. vaginalis and genomic species 2 and 3, are associated with BV (97). Another study reported that clades 1 (G. vaginalis and genomic species 2) and 3 (genomic species 8, 9, and 10) are associated with BV, while clade 2 (genomic species 3) is associated with intermediate microbiotas (98). Many, but not all, isolates of G. vaginalis produce vaginolysin, a cholesterol-dependent pore-forming cytolysin that kills mammalian cells by punching holes in their membranes. Species-specific variations found in the vaginolysin amino acid sequence could contribute to differential pathogenicity associated with these Gardnerella strains (99). Members of all four clades are found, albeit in low abundance, in women with no symptoms, and most women with Gardnerella in their VMBs harbor multiple Gardnerella species (98). Some strains of Gardnerella are more proficient than others at biofilm formation (100) or in competition with Lactobacillus species (101, 102) in vitro. Much remains to be learned about the various strains of Gardnerella, its diverse subgroups, and their association with gynecologic and obstetric health.

BV-associated bacteria produce biofilms, disrupt the vaginal mucosal and epithelial barriers by sialidase, cytolysins, and other enzymes, increase the vaginal pH, and produce enzymes that enhance their ability to colonize (Table 2). Gene expression in the VMB has been shown to be impacted by the composition of the VMB (103). Since most BV-associated bacteria have higher relative abundances in CST IV (44), genes responsible for sialidase production have the highest expression in the VMB with CST IV, and higher expression of genes encoding cholesterol-dependent cytolysins in L. iners and G. vaginalis is associated with the depletion of Lactobacillus (103). A cervicovaginal epithelial cell model shows that a mixture of BV-associated vaginal metabolites can promote vaginal inflammation by increasing proinflammatory cytokines, i.e., tumor necrosis factor alpha (TNF-α) and IL-8, but inhibit the production of chemokines, i.e., RANTES and IP-10, which could be a mechanism of immune evasion (68). Other epithelial cell models show that pore-forming toxins, such as vaginolysin produced by G. vaginalis (104) and perhaps inerolysin from L. iners (84), bind to tethered lipid rafts embedded in the plasma membranes of vaginal epithelial cells and mediate cytolysis, a plausible pathogenic mechanism in BV. Interestingly, women taking statins to reduce systemic cholesterol generally exhibit reduced prevalence of Gardnerella, and statins protect cultured vaginal epithelial cells from lysis by vaginolysin (105). Other BV-associated taxa, e.g., Sneathia amnii (106), also express toxins that have yet to be clearly implicated in pathogenesis.

TABLE 2.

Bacterial vaginosis-associated virulence factors^a

Mechanism	Bacterial product(s)	Species
Disruption of mucosal barrier	Sialidase	Gardnerella vaginalis, Prevotella bivia, Mycoplasma hominis, Bacteroides fragilis (267, 268) and Prevotella timonensis (269)
	Glycosulfatase	Prevotella spp. (270)
Disruption of epithelial barrier	Vaginolysin	G. vaginalis (268, 271)
	Inerolysin	Lactobacillus iners (84)
	Phospholipase C	G. vaginalis (268), Ureaplasma urealyticum (272)
	Hemolysin	G. vaginalis (273), Sneathia amnii (106)
	Urease	U. urealyticum (274)
Increase of the vaginal pH	Amine production	P. bivia (275), BVAB1, Dialister micraerophilus (276)
Antibiotic resistance	Resistance gene: 5-nitroimidazole, macrolides, tetracycline, β-lactam and aminoglycoside antibiotics	Widely distributed in the vaginal microbiome (277)
Immune evasion	IgA protease	U. urealyticum (278)
	Organic acids (succinic acid, acetic acid, etc.)	Prevotella spp., Mobiluncus spp. (68, 279)
Growth of diseases-associated bacteria	Amino acids	P. bivia (280)
	Ammonia	P. bivia (275)
Proinflammatory responses	Organic acids (53, 68)	G. vaginalis (71), Atopobium vaginae (264), P. timonensis (269, 281), Megasphaera elsdenii (281)
Biofilm formation		G. vaginalis, Atopobium vaginae, Mobiluncus spp., Fusobacterium nucleatum (107)

^aBVAB1, “Candidatus Lachnocurva vaginae.”

Vaginal biofilms protect BV-associated taxa from clearance by lower pH and lactic acid, antibacterials produced by lactobacilli, the host immune system, and antibiotics (44, 107) (Fig. 2). Although still poorly understood, these biofilms are thought to impair the integrity of the epithelial barrier (44, 107) and have been cited as important contributors to establishment of BV and its predilection for recurrence (107–109). Transcription of genes related to growth and vaginolysin production in G. vaginalis are downregulated in biofilms relative to planktonic cultures in vitro, which could be beneficial for long-term survival of G. vaginalis in the vagina (110). However, coculture of G. vaginalis with Enterococcus faecalis and Actinomyces neuii seems to promote biofilm formation and virulence of G. vaginalis at the transcriptional level (111). Known biofilm matrices contain polysaccharide, extracellular proteins, and extracellular DNA. DNase treatment reduces biofilm formation of G. vaginalis in vitro, suggesting that extracellular DNA is an important component in the biofilm matrix of G. vaginalis (112). However, studies on genes related to the production of polysaccharide and extracellular proteins in G. vaginalis are limited by the lack of genetic tools for this taxon. Interspecies interaction of G. vaginalis with BV-associated bacteria apparently promotes biofilm formation. In vitro studies illustrate that initial attachment of G. vaginalis may be mediated by L. iners or Peptoniphilus spp. and biofilm formation is enhanced by additional taxa, e.g., Atopobium vaginae, Prevotella bivia, Fusobacterium nucleatum, and Mobiluncus spp. (107, 111). A better understanding of vaginal biofilms is required to better support female reproductive health.

Several biogenic amines and short- and long-chain fatty acids, e.g., tyramine, N-acetylputrescine, cadaverine, deoxycarnitine, etc., associated with increased vaginal pH, abnormal vaginal odor or discharge, or the presence of clue cells have higher concentrations in BV patients or in VMBs with CST IV (113–116) and are associated with activation of proinflammatory responses (88, 114, 117). Consistent with these results, genes responsible for biogenic amine production have been discovered in the genomes of many BV-associated bacteria (114). In contrast, the concentrations of metabolites associated with health, e.g., lactate, phenylalanine, tyrosine, glutathione, and others, are higher in asymptomatic participants and in VMBs predominated by protective Lactobacillus (113–116).

Vulvovaginal candidiasis.

Vulvovaginal candidiasis often occurs when the vaginal environment is altered by antibiotic treatment, hormonal changes, metabolic disease, immunological incompetence, sexual activity, or other conditions that permit yeast, mainly Candida spp., to colonize the female reproductive tract in hyphal form (118, 119). Candida albicans and other species of Candida, which bind to mannose-binding lectin on the epithelial cell membranes, are the primary etiological agents (120). In response to altered host environmental conditions, Candida undergoes global transcriptional changes while transitioning into hyphal forms. Hyphae from some C. albicans strains form vaginal biofilms and secrete candidalysin, a cytolytic peptide toxin that damages epithelial cell membranes, permitting penetration of epithelial cell layers. Epidermal growth factor receptor is activated, inducing mitogen-activated protein kinase (MAPK) signaling, MKP1 activation and promotion of proinflammatory mediators, neutrophil recruitment, and type 17 immunity (121). The VMB of reproductive-age women with vulvovaginal candidiasis is generally more complex than that of asymptomatic women but is statistically less complex than that of women with BV (122, 123) (Table 1). In vivo studies suggest that colonization by Lactobacillus may not reduce the risk of vulvovaginal candidiasis (124), and colonization by L. crispatus is even associated with increased C. albicans colonization (125, 126). However, an L. crispatus-dominated VMB is associated with lower risks of C. albicans colonization and vulvovaginal candidiasis than an L. iners-dominated VMB in vivo (19, 123), and L. crispatus inhibits hypha formation of C. albicans (127) as well as the innate immune responses induced by yeast in vitro (66). Thus, it is clear that vulvovaginal candidiasis is associated with the VMB, but it is unclear if it is associated with a specific Lactobacillus species, e.g., L. crispatus, or the predominance rather than the presence of lactobacilli.

Sexually transmitted infections.

Vaginal dysbioses are associated with increased risk for acquisition of STIs, including HIV (55), herpes simplex virus (HSV) (128), human papillomavirus (HPV) (129), gonorrhea, chlamydia, and trichomoniasis, as well as an increased persistence of HPV infection (11). HSV infection has also been shown to promote vaginal dysbiosis (11). Dysbiosis-associated taxa in the VMB were reported to be associated with an increase in CD4-positive T cells in cervicovaginal lavage fluid from young South African women and in a murine model (55). This increased level of CD4-positive T cells, the target of HIV, likely favors HIV transmission in vaginal dysbioses. Evidence that BV-associated bacterial taxa enhance HIV RNA expression has also been documented both in vitro and in vivo (130, 131). BV-associated virulence factors, e.g., vaginolysin, inerolysin, mucinase, sialidase, and others, likely impact the integrity of the vaginal epithelium and mucosal barrier, also possibly facilitating these infections (44, 107). In contrast, protective Lactobacillus spp. inhibit the growth of BV-associated species, as outlined above (10), as well as the proinflammatory responses of epithelial cells and protects the integrity of epithelial barrier (22, 67), which may explain the attenuation of HIV infections by these taxa (22, 55).

Trichomoniasis, the most common nonviral STI (132), causes severe damage to vaginal tissue and disruption of the vaginal ecology by eliciting host inflammatory responses (132). Trichomonas vaginalis is closely associated with some bacterial taxa, including Mycoplasma hominis, “Candidatus Mycoplasma girerdii,” Veillonella montpellierensis, Prevotella amnii, Sneathia sanguinegens, Anaerococcus, and Parvimonas spp. in the VMB (133–135) (Table 1). M. hominis is considered obligately intracellular, and “Ca. Mycoplasma girerdii” may be a facultative endosymbiont. Moreover, there is some evidence to suggest that these taxa coexist with T. vaginalis (136, 137) and that at least M. hominis potentiates the pathogenicity of T. vaginalis (138).

Desquamative inflammatory vaginitis and aerobic vaginitis.

Desquamative inflammatory vaginitis (139, 140), first described over 6 decades ago, is an inflammatory disorder of questionable etiology associated with purulent discharge, vaginal itching, dyspareunia, and an inflamed vaginal wall (29). A similar condition associated with vaginal colonization by aerobic, enteric bacteria, including, among others, Escherichia, Streptococcus, Staphylococcus, and Enterococcus species, has been termed aerobic vaginitis (29, 141) (Table 1). Both conditions are associated with a paucity of vaginal lactobacilli and have an epidemiologic prevalence similar to that of BV (29). Their bacterial etiology is supported by positive responses to antibiotic treatment, but steroids that reduce inflammation similarly reduce the symptoms (29, 139). Thus, the role of the microbiome in aerobic and desquamative inflammatory vaginitis remains controversial.

The bacteria most frequently encountered in aerobic vaginitis, e.g., group B Streptococcus (GBS, or Streptococcus agalactiae), E. coli, and Staphylococcus aureus, usually have low abundances in the VMB (5, 7, 25, 26). However, in utero infections of these bacteria can cause serious reproductive outcomes, including but not limited to stillbirth and neonatal sepsis (142–144). GBS colonization seems not to affect the composition of the VMB of reproductive-age nonpregnant women (145), but G. vaginalis in the VMB has been reported to enhance GBS colonization and infection in a mouse model (146).

The Vaginal Microbiome and Pregnancy

As discussed above, increased estrogen levels during pregnancy promote generation of glycogen and lead to production of lactic acid by Lactobacillus (Fig. 2). Consistent with these observations, the VMB in late pregnancy seems to exhibit greater stability and Lactobacillus dominance than that in early pregnancy (5–8), and the transitions of the VMB during pregnancy occur earlier in gestation (8) (Fig. 3). The higher abundance of lactobacilli in late pregnancy is coupled with a commensurate reduced abundance of taxa associated with vaginal dysbiosis, including but not limited to G. vaginalis, A. vaginae, and P. bivia (5).

Vaginal communities dominated by Lactobacillus species are quite stable or tend to convert to vaginal communities dominated by other Lactobacillus species during pregnancy (5, 8, 147). Vaginal communities associated with dysbiosis tend to shift toward communities dominated by lactobacilli by the second trimester. The G. vaginalis-dominated VMB is less stable, often converting to an L. iners-dominated VMB (5, 8). These shifts are consistent with an evolution toward a more favorable vaginal environment in pregnancy.

Likely because Lactobacillus spp. are already more dominant in white than Black women before pregnancy (8), the transition to Lactobacillus-dominated profiles and the increase of the stability in the VMB is more evident in Black than white women during pregnancy (5, 8) (Fig. 3). Similarly, metagenomics data illustrate a simplification of metabolic activity only in Black women, not in non-Black women (8, 24). Consequently, pregnancy seems to have a greater apparent impact on the VMB of Black women.

Estrogen levels drop rapidly after delivery. Likely as a consequence, the stability of the VMB also drops (147–149), and recovery of a more stable VMB may take several months in asymptomatic women (147, 148) (Fig. 3). This postpartum disturbance of the VMB occurs irrespective of both the community structure during pregnancy and racioethnic background and manifests in reduced prevalence of Lactobacillus and increased prevalence of anaerobes, including Peptoniphilus, Prevotella, and Anaerococcus, that are generally associated with less optimal health outcomes (147, 148) (Fig. 3). Consistent with this observation, an interpregnancy interval shorter than 1 year is associated with increased risks for pregnancy complications, including PTB (150–152).

THE CERVICAL MICROBIOME

The cervix connects the uterus to the vagina (Fig. 1). Cervical brushes or swabs are used to collect cervical samples (18, 153), and vaginal contamination is minimized during this collection. The composition of the cervical microbiome is generally similar to that of the vaginal microbiome in reproductive-age women (18, 153). Most recent studies of the cervical microbiome focus on its association with cervical cancer. Cervical cancer is one of the most prevalent infectious cancers and is tightly associated with infection by high-risk HPV (154). Persistent high-risk HPV infection increases the risk of cervical intraepithelial neoplasia (CIN). In CIN, HPV DNA integrates into chromosomes of cervical epithelial cells, inducing changes that lead to cervical malignancy (155). The cervical microbiome is more complex in patients with CIN (23) and cervical cancer (18, 156, 157). The loop electrosurgical excision procedure for removing intraepithelial lesions was shown to decrease the complexity of the cervical microbiome in a cohort of Asian women (158). The composition of the cervical microbiome in HPV-negative participants is significantly different from that in participants with intraepithelial lesions or cervical cancer (157, 158). Several taxa, e.g., Gardnerella (18, 159), L. iners (160), A. vaginae (160), Mycoplasma (23), Sneathia (157), and Fusobacterium (157), most of which are associated with dysbiosis of the VMB (Table 1) and produce virulence factors in BV (Table 2), have been reported as risk factors for CIN and cervical cancer. Damage to the integrity of the epithelial (71) and mucosal (161) barriers due to overgrowth of dysbiosis-associated microorganisms has been hypothesized to permit HPV attack on cervical epithelial cells, but the mechanism is unproven. Since chronic inflammation seems to favor malignancy (162), local inflammation associated with dysbiosis of the cervical microbiome may also be involved in the progression of cervical and other gynecologic cancers. In contrast, the genus Lactobacillus is associated with a higher clearance of incident high-risk HPV infection (18), and L. crispatus is associated with lower risk of CIN in the cervical microbiome (160), possibly due to Lactobacillus spp. inhibiting growth of dysbiosis-associated microorganisms and host inflammatory responses.

Similar to the cervical microbiome, the VMB is more complex (163, 164) and contains a higher abundance of dysbiosis-associated taxa, e.g., Sneathia (165) and G. vaginalis (164), in individuals with HPV infection and CIN. Since the vagina and the cervix are in close proximity, it is not surprising that their respective microbiomes are similar in composition and exhibit similar changes in the progression of cervical cancer.

THE MICROBIOTA OF THE UPPER GENITAL TRACT

Is There a Microbiome in the Upper Genital Tract?

The concept that the uninfected upper genital tract has an intrinsic microbiome remains controversial. Multiple studies have supported the existence of the microbiome in the upper genital tract (166–168). However, because of low microbial biomass in these sites, potential contamination in the process of sample collection and processing or the presence of even low levels of bacterial DNA in the so-called “kitome” confounds results based on DNA sequencing. Recent reviews have argued that the uterus lacks a native microbiome (169, 170), whereas others find evidence of a microbiome in the upper genital tract (171). Several carefully controlled studies did not detect a normal microbiome in placental samples (16, 17, 172), and another did not detect a microbiota in fetal meconium before birth (173). Nevertheless, it is clear that pathogens can cause infections in the placenta (16, 174), uterus, fallopian tubes, and ovaries (175–178), and additional study is required to confirm the existence of and characterize a native microbiome in the upper female reproductive tract.

Origin of the Microbiota in the Upper Genital Tract

The correlation between dysbiosis of the VMB and PTB (5, 13, 15, 179, 180) and the similarity of taxa found in the amniotic fluid (174, 181) and membranes (182) to the VMB suggests that pathogens ascend from the lower reproductive tract into the uterine cavity causing uterine infections and problems in pregnancy, including PTB. Consistent with this hypothesis, a study in pregnant mice observed the ascension of bioluminescent E. coli from the vagina into the uterine cavity, resulting in premature delivery (183). However, the mouse and human female reproductive tracts are quite divergent in both anatomy and physiology, rendering these observations difficult to generalize. It has also been proposed that bacteria spread hematogenously from the oral cavity, as taxa similar to those found in the mouth have been found in the uterus (174) and periodontitis has been identified as a risk factor of PTB (184). Taxa common to those in both vaginal and oral microbiomes have also been reported in chorioamnionitis (185), and oral bacteria have been reported in the amniotic fluid in women and mice that experience premature rupture of membranes due to intrauterine infections (186–188). However, human oral taxa are also common in the VMB (25, 148), and their presence in the uterus does not preclude contiguous spread from the lower reproductive tract. Other hypotheses on the origin of microbes in the upper genital tract include migration from the infected fallopian tubes (189) and accidental transmission in invasive procedures (190). Metagenomic studies that provide strain-level resolution will be useful to confirm or refute the association between the VMB, the oral microbiome, intrauterine infections, and adverse pregnancy outcomes.

Ascension of Microbes to the Uterine Cavity

Ascension of microbes to the uterine cavity is generally prevented by the cervical mucus plug (191). However, this plug could be compromised by bacterial products (192–194). GBS produces hyaluronidases that digest hyaluronan, a protective polysaccharide (195), and can promote ascending infection (193, 196). Surprisingly, sloughing (also known as exfoliation or shedding) of infected epithelial cells does not reduce GBS vaginal colonization but increases GBS dissemination and ascension (197). Sialidases and mucinases produced by BV-associated bacteria, e.g., G. vaginalis, P. bivia, M. hominis, and Bacteroides fragilis, compromise mucosal membranes and the cervical plug, are conducive to invasion of the upper reproductive tract in a murine model (198), and increase the risk of very early PTB in vivo (199). However, it remains uncertain whether these sialidases promote ascending infections of other microbes, e.g., GBS or E. coli, in humans, although a short cervical length is thought to be beneficial for microbe ascension and has been shown as a risk factor of PTB (190, 200, 201). Additionally, cervical-fundal uterine peristaltic contractions have been shown to promote the transport of albumin macrospheres of the size of spermatozoa through the cervix (202), which could also promote ascension of pathogens.

THE MICROBIOTA OF THE FEMALE REPRODUCTIVE TRACT AND PRETERM BIRTH

The Microbiota and Preterm Birth

PTB is defined as childbirth after less than 37 weeks of gestation due to multiples causes, e.g., cesarian section or labor induction for medical reasons, preterm premature rupture of the membranes (PPROM), spontaneous labor with intact membranes, etc. (190). PTB is the leading cause (17.7%) of deaths among children under 5 years of age worldwide (203). Intrauterine infections caused by microorganisms have been hypothesized to activate the innate immune system and consequently enhance the risk for spontaneous PTB (190). It has been estimated that up to 25 to 40% of spontaneous PTBs have microbial etiology.

Bacterial species, e.g., Chlamydia trachomatis (204–206), Klebsiella pneumoniae (207, 208), E. coli (208, 209), Fusobacterium nucleatum (210), GBS (16, 208, 211), Mycoplasma hominis (204), Neisseria gonorrhoeae (206), Staphylococcus aureus (194, 209), and Streptococcus mitis (209), have been detected in intrauterine infections and induce PTB (Table 3). Although it is uncertain whether ascension and intrauterine infections are required for vaginal microbes to induce spontaneous PTB, several microorganisms frequently detected in the vagina have increased relative abundances in the VMB of women who later experience PTB, e.g., Ureaplasma, GBS, and several BV-associated taxa. Other microbes, e.g., HIV (212, 213) and HPV (214), have been reported to be associated with increased risk of PTB, but the PTB induced by HIV seems to be independent from the VMB (212).

TABLE 3.

Risk factors for preterm birth associated with the microbiome or microbial infections of the female reproductive tract^a

Classification	Risk factor(s)
Factors in the VMB
Vaginal taxa increased in relative abundance in PTB	Aerococcus spp. (5, 15), Atopobium spp. (15, 226, 282), BVAB1 (5), BVAB2 (5), Chlamydia spp. (208), Clostridium sensu stricto (15), Coriobacteriaceae species (5), Dialister spp. (5, 15), Escherichia coli (208), Fusobacterium nucleatum (282), Gardnerella spp. (12, 147), GBS (208, 211), Klebsiella pneumoniae (208), BVAB3 (282), Megasphaera spp. (15, 226), Mobiluncus curtisii/Mobiluncus mulieris (226), Mycoplasma hominis (282), Olsenella sp. (15), Parvimonas sp. (5), Porphyromonas asaccharolytica (226), Prevotella spp. (5, 15, 148), Sneathia amnii (5), Sneathia sanguinegens (5, 226), TM7-H1 (5), Trichomonas vaginalis (208, 283), Ureaplasma spp. (81, 147, 209, 283)
Vaginal taxa reduced in relative abundance in PTB	Lactobacillus crispatus (5, 12, 14, 228, 229), Lactobacillus spp. (5, 12, 15, 147, 226, 227, 229)
Bacterial virulence factors	BV-associated bacteria: sialidase (199, 284); GBS: β-hemolysin/cytolysin (192) and hyaluronidase (193); unidentified taxa: lipopolysaccharide (285); unidentified taxa: volatile organic compound (286)
Bacterial load	A higher vaginal bacterial load is a risk for PTB recurrence (245)
Host factors associated with the VMB
Host cytokines/chemokines	Amniotic fluid: IL-1β (287), IL-6 (184, 217, 287), IL-8 (184), IL-10 (218), and IL-18 (219); cervicovaginal fluid: IL-1β (5, 88), IL-6 (5, 216, 217), eotaxin (5, 225), and MIP-1β (5); plasma: GM-CSF (220)
Host antimicrobial peptide	Reduced β-defensin-2 (226)
Other factors	Short cervix (14, 234), cervical cerclage with braided suture (compared to monofilament suture) (236)
Microbial invasion
Bacterial intrauterine infectionsb	Chlamydia trachomatis (204, 205), Escherichia coli (208, 209), Fusobacterium nucleatum (210), GBS (16, 208, 211), Klebsiella pneumoniae (207, 208), Mycoplasma hominis (204), Staphylococcus aureus (194, 209), Streptococcus mitis (209)
Viral infections	HIV (213, 288, 289), HPV (214)
Other sexually transmitted infections	Neisseria gonorrhoeae (289–291), Treponema pallidum (290), Chlamydia trachomatis (289, 291)

^aUnless stated otherwise, risk factors show elevated concentrations/abundances in women who later experience PTB. BVAB1, “Candidatus Lachnocurva vaginae”; BVAB2, Lachnospiraceae BV-associated bacterium 2; BVAB3, Mageeibacillus indolicus; TM7-H1, “Candidatus Saccharibacteria” genomospecies TM7-H1; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

^bMost of the vaginal taxa whose abundance increased in PTB were also reported as risk factors for intrauterine infections (174).

The Microbiota and Immune Responses

GBS generates a β-hemolysin/cytolysin that induces inflammation and the disruption of maternal-fetal barriers, leading to higher risk for PTB in a mouse model (192). Although commonly present in the VMB with relatively high abundance, several BV-related bacteria stimulate proinflammatory responses by producing short-chain fatty acid in ectocervical, endocervical, and dendritic cell models (53) (Table 2) and the increased proinflammatory responses caused by non-Lactobacillus taxa in the VMB are associated with PTB (88) (Table 3). Lipopolysaccharide is widely used to promote inflammation to induce PTB in mouse models (183, 215), but the function of lipopolysaccharide in the female reproductive tract is not clear. Several proinflammatory cytokines and chemokines, i.e., IL-1β (5), IL-6 (5, 184, 216–218), IL-8 (184), IL-10 (218), IL-18 (219), granulocyte-macrophage colony-stimulating factor (GM-CSF) (218, 220), MIP-1β (5), and eotaxin (5), have been reported to have elevated concentrations in the amniotic fluid, cervicovaginal fluid, and plasma of pregnant individuals who later experience spontaneous PTB, and cervical levels of IL-1β, IL-6, IL-8, IL-10, eotaxin, MIP-1β, and GM-CSF have been associated with bacteria in the VMB (221–225) (Table 3). β-Defensin-2, a peptide with broad antimicrobial activity, has been reported to be associated with the lower risk of spontaneous PTB previously linked to cervicovaginal microorganisms, but this phenomenon has been observed only in Black women (226).

The Vaginal Microbiome in Preterm Birth

16S rRNA taxonomic profiles of the VMBs in pregnancies that deliver prematurely are generally more complex than the VMBs in pregnancies that go to term (5, 13, 15). Because of the protective effects of Lactobacillus spp. on overall vaginal health (Fig. 2), it is not surprising that a decrease in prevalence of lactobacilli, primarily L. crispatus, in pregnancies that end in PTB has been observed (5, 12, 15, 147, 226–229) (Table 3). Moreover, several taxa have been identified as putative risk factors for PTB (Table 3). These taxa are generally more readily observed earlier in pregnancy (5, 13, 15), likely due to the general homogenization effect of pregnancy toward a more Lactobacillus-dominated VMB (8). Similarly, the vaginal pH and Gram staining in early pregnancy can better predict PTB in asymptomatic women (230). Many of the taxa associated with PTB, e.g., Gardnerella, Atopobium, Dialister, Megasphaera, Prevotella, and Sneathia, and others, are components of CST IV (25) (Table 3). Thus, not surprisingly, CST IV has a positive association with PTB (82, 147, 229, 231). As described above, virulence factors that compromise mucosal membranes and the cervical plug and promote proinflammatory responses, e.g., sialidase, β-hemolysin/cytolysin, and hyaluronidase, increase ascension of microbes to the uterine cavity. It is reasonable that bacteria that produce these virulence factors (192, 193, 199) and increase proinflammatory responses (92, 221, 222, 232), e.g., Gardnerella spp., Mycoplasma hominis, Prevotella spp., and GBS, increase the risk for PTB (Table 3). However, PTB-associated virulence factors have yet to be identified in other PTB-associated taxa, e.g., BVAB1, TM7-H1, Dialister spp., etc. Continued study is required to define which taxa are keys in the enhanced risk of PTB and explore more PTB-associated virulence factors and the mechanisms by which they induce PTB. Other studies have failed to identify components of the microbiome associated with risk of PTB (7, 13, 180). These studies may have missed signals due to the racioethnic composition of their cohorts, gestational age of sampling, low sample size, or treatments (e.g., antibiotics and estrogen) received by the participants. There were also significant technical differences in these studies, including the definition of spontaneous PTB, the varied sequencing technologies employed, and the differential analysis pipelines and databases applied.

Metagenomic and metatranscriptomic data of the VMB, not surprisingly, have identified essentially the same panel of bacteria as risk factors for PTB in 16S rRNA-based studies (5). Recent metabolomics studies on the VMB and PTB have been inconsistent in study design, sample sources, metabolomic techniques, and statistical methods, leading to inconsistent conclusions (233). A recent comprehensive review of metabolomic publications listed 163 metabolites that were potentially associated with PTB, only four of which, i.e., myoinositol, creatinine, histidine, and 5-oxoproline, were associated with PTB across multiple studies (233). Thus, more in-depth metabolomic studies are warranted.

Clinical Issues Associated with the Vaginal Microbiome and Preterm Birth

As introduced above, a short cervix is a risk factor of spontaneous PTB, likely by promoting microbe ascension (190, 200, 201) (Table 3). Furthermore, a short cervical length in reproductive-age women is associated with suboptimal VMBs dominated by L. iners or anaerobes rather than L. crispatus (14, 234), which could further increase the risk of PTB. Cervical cerclage is widely used to prevent PTB in pregnancies at risk for PTB (235). Cerclage with a braided suture is associated with greater dysbiosis in the VMB and higher risk for PTB than cerclage with a monofilament suture, likely due to its irregular structure of the former providing more readily colonized niches (236).

Cervical remodeling is a term to describe the changes of the cervix in extracellular matrix structure and mechanical properties after pregnancy (237). Premature cervical remodeling leads to 12.5% of PTB (238) and can be induced by lipopolysaccharide in a mouse infection model (239). A study with a cohort dominated by white women shows that extreme cervical shortening is associated with a nonoptimal VMB (240). According to these observations, proinflammatory responses caused by microbes are thought to be a risk factor stimulating premature cervical remodeling and subsequent PTB (238, 241).

Similar to the association between the VMB and PTB, a more complex VMB with reduced Lactobacillus spp. is associated with PPROM (227, 242–244). The association between PTB and PPROM was recently reviewed (242).

It is worth noting that not only the composition of the VMB but also a higher vaginal bacterial load detected in the second trimester is associated with recurrent PTB (245), which again emphasizes the importance of quantitative assessment of the VMB in clinical settings.

Bacterial Vaginosis and Preterm Birth

Interestingly, G. vaginalis, which has long been the hallmark of a VMB of women with BV and which produces known toxins, e.g., vaginolysin and sialidase, has not been universally associated with PTB (5, 227). However, as discussed above, G. vaginalis has recently been redefined into four clades and up to 13 putative genomic species (8, 96, 97), and although at least one subgroup was tentatively associated with PTB (12), additional work is required to verify the possible contributions of each of these taxa to reproductive health and risk of PTB. Although BV and some BV-associated taxa are identified as risk factors for PTB (179, 246) (Tables 1 and 3) and an effective BV treatment seems to attenuate the PTB-associated proinflammatory responses (221, 222), current treatment strategies, usually with metronidazole or clindamycin, do not seem to reduce the risk for PTB (208, 247). Although this observation might suggest that the BV-associated VMB is not contributory to PTB risk, other interpretations are plausible. Thus, it is also possible that resistance to metronidazole by the relevant bacteria may be responsible for the persistence or rapid regrowth of these bacteria or that treatment fails to reverse the negative impacts; e.g., inflammatory responses, of these taxa (92, 221, 248, 249). Alternatively, sequestration of metronidazole by L. iners could reduce efficacy of the antibiotic, permitting BV recurrence (250) and a consequent failure to reduce PTB risk. It seems possible that a more effective elimination of the BV-associated VMB, e.g., with multiple antibiotics, followed by reestablishment of a Lactobacillus-dominant microbiome, perhaps with prebiotics or probiotics, would both improve treatment of BV and reduce risk of PTB.

Prediction of Preterm Birth

A variety of parameters; e.g., cervical length (200, 201), the composition of the VMB (5, 12, 15, 147, 226, 227), and the local expression of inflammatory cytokines (5, 184, 217, 220), have been reported to be predictive of PTB. Several reports suggest that some of these biomarkers are more relevant in earlier stages of pregnancy, likely due to the tendency of the VMB to become more homogeneous and Lactobacillus dominated as pregnancy progresses (5, 13, 15) (Fig. 3). A model for PTB prediction using the abundance of four taxa, i.e., S. amnii, BVAB1, Prevotella cluster 2, and TM7-H1, in the VMB early in pregnancy showed high sensitivity and specificity (5). Other recent models for prediction of risk for PTB invoked integrated proteomics and transcriptomics data from plasma and metabolomics from urine samples (251) or a combination of cervical length, gestational age, amniotic fluid glucose, and IL-6 (200). Honing of such models to include vaginal taxa abundance, human gene expression, metabolomics (e.g., d-lactic acid) (81) and cytokine profiling (e.g., IL-6) (217) as well as demographic and clinical parameters (e.g., cervical length, preterm-birth history, racioethnicity, and socioeconomic status) has promise for accurate early prediction of risk for preterm birth and other adverse pregnancy outcomes.

THE VAGINAL VIROME AND VAGINAL HEALTH

Still largely missing in studies of the microbiomes of the female reproductive tract is the virome component. Bacteriophages have been shown to help maintain stability in the gut microbiome (252) and are also abundant in the VMB (253). Recent studies suggest that the vaginal virome is associated with the composition of the VMB and BV status, and vaginal viruses targeting BV-associated taxa have been identified (254, 255). Also, CRISPR genes, which function as an antibacteriophage defense system, have been reported to be upregulated in the VMBs of women with BV (87, 255), implying that bacteriophages may play a role in modulating bacterial components of the VMB. Additionally, one early study implied that higher viral diversity and richness of the eukaryotic virome in the first trimester of pregnancy were associated with higher risk for PTB (256), thus suggesting that the virome could play a significant role in obstetric health. Thus, the virome of the reproductive tract should be further investigated.

ADDRESSING DYSBIOSES OF THE VAGINAL MICROBIOME

Approximately 50% of women with BV experience a recurrence within as few as 12 weeks (92) to 12 months (93, 94) after treatment with metronidazole. Treatment with other antibiotics more effective against Gram-positive bacteria, e.g., erythromycin, seems to alter the balance of the microbiome, leading to vaginal dysbiosis (227). As outlined above, estrogen enriches the vaginal epithelium in glycogen and favors proliferation of lactobacilli. Estrogen is used to treat atrophic vaginitis associated with estrogen deficiency, e.g., menopause and postpartum dysbiosis (257). However, certain clinical criteria, such as gynecological, breast, or other malignancy, or risk thereof, could be a contraindication to the use of estrogen. Thus, low-dose transvaginal topical estrogen has been shown to be effective at engendering a more optimal Lactobacillus-rich VMB while not increasing the risk of estrogen-sensitive cancers (257, 258). Additionally, more attention should be focused on nonestrogen therapy, e.g., probiotics, prebiotics, VMB transplantation, or possibly statins. Women taking statins are more likely to have a VMB dominated by lactobacilli, possibly due to inhibition of toxicity of the vaginolysin toxin from G. vaginalis (105). A recent report showed that introduction of L. crispatus into the reproductive tract after metronidazole treatment decreases BV recurrence by 15% at week 12 (259). Similarly, Lactobacillus-rich VMB transplants from healthy donors to five reproductive-age women with recurrent BV resulted in long-term remission in at least four of the recipients (21). Although such transplants introduce risks and other challenges, these results are promising for future development of noninvasive probiotic treatment of dysbioses of the VMB.

PERSPECTIVES

Our understanding of the microbiome of the female reproductive tract has advanced rapidly in the past decade. Although not universally true, an optimal VMB is usually dominated by species of Lactobacillus (25), particularly L. crispatus, L. jensenii or L. gasseri, and VMBs dominated by other taxa, e.g., G. vaginalis, A. vaginae, BVAB1, Mycoplasma spp., Sneathia spp., Ureaplasma spp., E. coli, and GBS, are associated with adverse conditions, including BV, aerobic vaginitis, vulvovaginal candidiasis, STIs, viral infections, cervical cancer, and adverse pregnancy outcomes (e.g., PTB). However, this dogma may be overstated, as VMB profiles that might normally be considered adverse are common in the absence of any clinical symptoms, complaints, or complications. Although much progress has been made, it is clear that there is still much lacking in our understanding of the contribution of the microbiomes of the female reproductive tracts to human health and disease.

Dysbioses of the female reproductive tract are generally treated with antibiotics or antifungals with considerable initial response. Although BV usually responds initially to antibiotic therapy, its high recurrence rate remains an enigma (92–94). Several virulence factors, e.g., biofilm formation and antibiotic resistance, are relevant to BV recurrence (260), and in-depth strain-level taxonomic and gene-centric studies are required to dissect the mechanisms by which the relevant taxa persist. Application of probiotics in combination with antibiotics and estrogen may be an option that aids in improving the homeostasis of the VMB and thereby vaginal health. Vaginal protective lactobacilli, in contrast to other Lactobacillus strains, have attracted significant attention as possible probiotics (259). Other combinations of pharmaceuticals, e.g., statins (105) and vitamin D (261, 262), may be helpful as well. Future study will undoubtedly clarify these possibilities. However, it is abundantly clear that much work remains to be done to find a combination of therapies that leads to a stable health-promoting VMB.

Although mice and epithelial cell models have been developed, study of the VMB is hampered by the lack of better models, as even nonhuman primates have vaginal physiology that differs greatly from that of the human female reproductive tract. In the absence of good animal models, establishment of three-dimensional (3D) organoid models would advance the field significantly (263). Most of the abundant taxa in the VMB are culturable, e.g., G. vaginalis (110), S. amnii (106), A. vaginae (264), and P. bivia (264), but other taxa, e.g., BVAB1 and “Ca. Mycoplasma girerdii,” remain difficult to manipulate in vitro. Molecular methods such as single gene deletion, genome-wide mutagenesis, and spatial genomics and transcriptomics are also important in clarifying the mechanism by which these bacteria cause disease. Although metagenomic and metatranscriptomic results have largely confirmed that the compositions of the VMB (4, 5, 24, 265) and bacterial risk factors for PTB are those identified in 16S rRNA-based studies (5), more in-depth longitudinal metagenomic and metatranscriptomic studies would be helpful to identify clades and genes relevant to BV, particularly in G. vaginalis (98) and L. iners (87).

In sum, recent studies have modified and greatly improved our understanding of the impact of the microbiome of the female reproductive tract. Continued study is required to elucidate the specific taxa that are relevant and the mechanisms by which they exert their effects. The contribution of host genetics and the virome are yet to be explored. Overall, a better understanding of the microbiome of the reproductive tract holds great promise to improve human health and well-being.