The human microbiome represents the collection of microbes living inside and on the surface of human body. Research efforts by the scientific community worldwide are increasingly focused on understanding the role of microbiota in health and disease. Since the completion of the Human Microbiome Project (HMP), and publication of the characteristics and functions of human microbiota located in different body habitats, this field of science has gained momentum (1). Current microbiome research tries to fill in the missing details in the pathophysiology and explain the seemingly random variation in disease severity and phenotype of confounding factors such as ethnicity, geographical location and societal habits. Due to advances in microbiome research, scientists have obtained valuable insight in many complex disorders such as obesity, cancer and inflammatory bowel disease, and a similar trend is observed in female reproductive tract in both physiological and pathological states (2).

Because of the invasive nature of sampling methods, microorganisms populating the female reproductive tract remain less explored compared to microbiota populating the intestines; nonetheless, they represent an appreciable proportion (around 9%) of the female microbial network (3). Furthermore, disrupted female reproductive tract microbial communities have been implicated in reproductive and pregnancy complications, as reviewed in (4, 5). Reproductive and pregnancy complications are of global health interest, and comprise diverse health problems that occur prior to conception and during gestation. Notably, these involve the mother’s health, the baby’s health or both. Such problems may entail difficulties to conceive, or arise throughout gestation and span from the inability to maintain pregnancy in the first weeks of gestation, to its early termination in the third trimester. An increasing body of evidence associates microorganisms (including mutualistic) to the onset of reproductive health and maternal-fetal conditions, and to some of the major obstetrical syndromes, including premature birth, premature rupture of the membranes, premature labor, intrauterine growth restriction, and stillbirth (6).

The inability to conceive is an often neglected health concern affecting individuals around the globe. In 2010, an estimated 48.5 million couples worldwide were infertile, with little changes over the previous two decades (7). Early pregnancy loss (frequently before 13 weeks of gestation) is estimated to occur in 15–20% of recognized pregnancies, without major geographical differences (8–10). Of note, morbidity and pregnancy complications at more advanced stages of gestation (e.g., preeclampsia, preterm labor and stillbirth) often have a higher burden in low-resources settings, especially in south Asia and sub-Saharan Africa (11). Among such late complications, preterm birth remains the leading cause of worldwide neonatal morbidity and mortality: approximately 10.6% of all live births in 2014 were preterm, 80% of which occurred in Asian and sub-Saharan African countries (12).

Over recent years technological advancements, in particular sequencing-based methods for bacterial detection (metagenomics and 16S rRNA gene amplicon sequencing), have greatly expanded the literature on diverse microbiota colonizing the female reproductive tract both in healthy and disease states (5, 13). However, the wide spread application of sequencing-based methods incorporates a potential for false-positive results (i.e., low specificity) due to contamination as in case of the placenta (14). Our current knowledge is limited on how these microbiota interact with host cells, including local immune mediators, and whether this interaction is causally involved in the pathogenesis of pregnancy complications. It is well known that the mechanisms pivotal in regulating the establishment and maintenance of pregnancy, including epigenetic regulation and immune adaptation, are directly affected by local microorganisms (5, 13). Furthermore, the impact of maternal factors such as BMI, pre-existing disorders or life style habits (e.g., diet and nutrition) on this host-microbiota interaction need to be integrated and thoroughly studied in future studies (15). Understanding the interplay at the fetomaternal interface is essential for developing predictive human biomarkers for implantation and placentation and is a key step toward designing novel therapeutic approaches.

This review aims to present an account of currently available literature on reproductive tract microbiota, describing the composition and function of microbiota and their links to pregnancy complications. The potential mechanisms of interaction between microbes and the immune system are discussed with respect to specific locations, focusing on the unique changes that characterize physiological pregnancy. Furthermore, the clinical impact of abnormal microbiota composition, i.e., “dysbiosis,” on female reproductive biology, from the pre-conceptional stage throughout early and late pregnancy. Common methodological challenges confronting research in the field related to study design and technical issues of sample collection and assay standardization are highlighted. Finally, we present options with respect to translation of various aspects and insights to future therapeutic approaches in reproductive medicine.

The community of microbes in the female lower genital tract plays a fundamental role in the promotion of homeostasis and in the prevention of colonization by pathogenic microorganisms. Compared to other sites, the vagina appears to harbor particularly simple microbial communities of low diversity (1). Although relatively simple at the genus-level, the diversity of Lactobacilli in the vaginal space is nonetheless higher than at other body sites (1). Many studies among non-pregnant women of reproductive age report that Lactobacillus spp. is the predominant species in the vagina, although the possibility of normal vaginal microbiota dominated by bacteria other than Lactobacilli seems to be plausible [as reviewed in (16)]. Current evidence suggests that the vaginal bacterial community is in a state of dynamic equilibrium (17). The composition of the vaginal microbiota correlates with the most dominant bacterial community composition (i.e., community state type; CST) across time (17). This notion appears to be in line with the concept of community resilience. Community resilience suggests that the ability of a microbial ecosystem to mitigate composition changes depends on the presence of beneficial species with stabilizing roles. For instance, Lactobacillus crispatus-dominated communities are less likely to transition to non-beneficial CST, than communities dominated by other Lactobacillus spp., like L. iners (17, 18).

In non-pregnant women of reproductive age, transient variations in the vaginal microbiota’s dynamic equilibrium are the results of physiological changes in response to hormones during the menstrual cycle, or human activities (e.g., sexual intercourse and hygienic practices) (18). In addition, other constitutional and environmental factors, including age, ethnicity, geographical variation and sexual habits influence the bacterial communities detected in the lower urogenital tract (19–21). For example, Lactobacilli-dominated vaginal microbiota has been shown to be less prevalent among non-Caucasian women in several studies (22–24), though, when present, their beneficial role seems maintained. A well-known study performed in North America characterized the vaginal microbiota of women of reproductive age from four ethnic groups (Caucasian, African-American, Hispanic and Asian): five groups of microbial communities, called Community State Types (CSTs) were identified. Whereas CST-IV was the most diverse CST and was also associated with a higher local pH, the remaining four (CST-I, CST-II, CST-III, and CST-V) were dominated by Lactobacilli (20). Lactobacilli thrive in anaerobic environments and produce lactic acid, therefore contributing to the acidic vaginal environment. Several studies have reported how depletion of these microorganisms often leads to vaginal dysbiosis, which is occasionally symptomatic, and at times associated with several important reproductive complications (25, 26).

The microbiota residing in the cervix have been sparsely studied as an independent entity. Current evidence suggests a strong similarity between bacterial communities in the cervix and in the vaginal area, suggesting ascending bacterial colonization from the vagina to the cervix (27). Such microbial communities are sometimes jointly referred to as cervicovaginal microbiota (28).

While in non-pregnant reproductive age women the vaginal microbiota is relatively dynamic, in healthy pregnancies it is characterized by an increase in stability (29). This is one of the physiological changes taking place in response to gestation. Other changes in this microbial community located in the vaginal area are an overall decrease in richness (number of different species present) and diversity (of the microbial ecosystem, i.e., the relative abundance of species). Overall, abundance of Lactobacillus spp. appears to be higher in healthy pregnant women, than in women with complicated pregnancies; whereas Mycoplasma and Ureaplasma appear to be lower (30). Most of the studies in the field seem to agree that Lactobacilli, in particular L. crispatus, L. iners, L. jenseii, and L. gasseri are the dominant bacteria detected in the vaginal microbiota of pregnant women (29, 31, 32).

Some of the factors influencing vaginal microbiota’s diversity during pregnancy are: gestational age (32), previous pregnancy history (33) and ethnicity (31, 34). In particular, L. crispatus was the most dominant species in a Caucasian cohort (35), L. iners in an African American one (29) and, surprisingly, L. acidophilus in a mostly Hispanic population where L. crispatus was not detected, even though tested for (36). Although ethnic and geographical differences in vaginal microbiota are not yet understood, nor consistently observed (32), a combination of colonization by gut microbes, hygienic and sexual practices, and host genetics may contribute to underlying mechanisms (37).

As pregnancy progresses, the vaginal microbiota changes with an increase in the relative abundance of Lactobacilli and decrease in anaerobe or strict-anaerobe species, until around 36 weeks of gestation, when the number of species increases significantly (31). Such composition has been reported to resemble that of the vaginal microbiota in the non-pregnant state (34). Throughout pregnancy, some species of Lactobacilli associate with “normal” (i.e., healthy) vaginal microbiota, whereas the presence of other species reflect an abnormal, less beneficial microbial community. The former has been reported for L. crispatus, whereas the latter for L. gasseri and L. iners (35). Finally, reported changes in the vaginal microbiota after delivery include a decrease in Lactobacillus species and an increase in anaerobe ones (38), irrespective of the vaginal communities during pregnancy and independent of ethnicity (39).

It is becoming increasingly apparent from studies in the field that, besides investigation of the vaginal microbiota as an ecological community (defined by its most abundant species), individual (even low abundant) species with less beneficial roles should be considered, as their mere presence might be sufficiently detrimental to a healthy microbiotal equilibrium.

Unlike the vagina, the endometrium has not been extensively studied as a site of commensal bacterial colonization, in part likely due to the relatively limited access to uncontaminated samples. Historically, the uterus has been considered sterile (40), and presence of bacteria in endometrial, placental or amniotic fluid samples was viewed as pathological. However, with the advent of culture-independent techniques, recently compiled data support both the prevalence and variation of bacterial communities in the endometrium and their possible role in reproductive health (41) but is still under debate (14). Together, much of our knowledge comes from the interpretation and comparison of the role of microbiome in anatomically related sites.

In healthy non-pregnant women, the endometrium appears to harbor a unique, low-biomass microbiota, dominated by a few bacterial species including Bacteroidetes (Flavobacterium spp.) and Firmicutes (Lactobacillus spp.) (5, 41, 42). Compared to the vaginal microbiome, the endometrium harbors a significantly lower quantity of microbes, suggesting that the cervix and/or uterine environment serve as a barrier for ascending microorganisms (42). Since Lactobacilli and Streptococci represent the most dominant bacteria in the vagina and in the cervix, respectively, the occurrence of these species in the uterine cavity may indicate contamination during sample collection. Therefore, the use of paired samples from vagina, cervix and endometrial fluid has been proposed in order to facilitate interpretation (43).

A recent review reported that, in studies using culture-dependent techniques, no bacterial family was reported more than once in uterine or endometrial samples, indicating one of the challenges in determining the complexity of “normal” reference microbiota of the uterine cavity in this manner. However, in culture-independent analyses, Lactobacilli, Bifidobacteriaceae, Comamonadaceae, and Streptococcaceae were reported more than once in the uterine cavity, suggesting that the culture-independent technology approximate microbiotal complexity more truthfully (41).

The notion that the uterus represents a sterile environment during pregnancy was first challenged with the advent of the Human Microbiome Project (HMP) (1) and other studies (44). The detection of unique bacterial DNA profiles in human placenta supported the idea that a rich microbial community normally exists in utero. Surprisingly, these early data suggested that the placental microbiome resembles the oral cavity microbiome rather than that of adjacent sites, e.g., the vagina. Since these early reports, several reservations on this notion and its implications on pregnancy complications have been voiced, noting that the low-biomass that bacterial DNA in the placenta represents, is sensitive to capturing background contamination (from DNA extraction kits, polymerase chain reaction reagents, and laboratory environments) (45, 46). Recent meticulous analyses of metagenomic DNA, consistently found no significant differences in the abundance and/or presence of a microbiota between placental tissue (term women without labor) and background technical controls (14, 47). The proposed link between oral dysbiosis and pregnancy complications puts the debate about placental microbiota in the focus: clinical studies on the association between gingivitis and PTB have reported bacteria in the very old structures of the placenta (48). In the context of placental microbiome analysis, the pitfalls associated with technology of choice, in addition to the methodological problems, needs careful consideration (see section “Discussion,” diagnostic challenges).

Like in many other mucosal tissues, the female reproductive tract’s first line of defense against pathogens is a physical barrier that, among other things, consists of a mucous layer, IgA antibodies and commensal bacteria limiting colonization by pathogenic bacteria. In addition, epithelial cells lining the female reproductive tract physically block pathogen invasion and produce protective molecules like antimicrobial peptides (AMPs). AMPs are multifunctional molecules with important roles in direct microbial killing, protection against proteolytic enzymes from various pathogens including bacteria, fungi, and some viruses, and modulation of both innate and adaptive immune responses (52, 53). AMPs are produced either constitutively or after induction by inflammatory stimuli, and are present on the mucosal surface, in decidual stroma, in endometrial fluid and even in amniotic fluid (54). In humans, AMPs have been linked to key regulatory processes in implantation and are implicated in the pathogenesis of various pregnancy complications (52, 55). An example of an AMP found in the uterus is the secretory leukocyte protease inhibitor (SLPI), which has antiviral and antifungal properties, and acts as a bactericidal against gram negative as well as gram-positive bacteria such as Escherichia coli and Staphylococcus aureus (56).

Natural Killer (NK) cells are a critical component of the innate immune system and comprise up to 70% of all endometrial leukocytes during the secretory phase of the menstrual cycle and in early pregnancy, but decline in numbers by mid-gestation (49). The main function NK cells in the vagina is to provide protection against a broad variety of viruses. Like their blood counterparts, vaginal NK cells recognize virus- or stress-associated molecules and they destroy infected cells through the release of toxic granules containing granzymes and perforins or though interaction with death receptors (57). The exact function of uterine NK cells (uNK) is not completely clear as they clearly differ from NK cells in peripheral blood in surface markers and cytokine repertoire. uNK cells are poorly cytotoxic, but are a potent source of cytokines such as Interferon (IFN)γ, TNF-α, Granulocyte-macrophage colony-stimulating factor (GM-CSF) and Interleukin (IL)-10 as well as proangiogenic factors like vascular endothelial growth factor (VEGF) (58–60). Studies using genetic mouse models lacking uNK cells provided histological evidence of failed trophoblast invasion and defective spiral artery remodeling and highlighted the critical role for uNK in for normal placentation (61, 62).

Myeloid cells (macrophages and dendritic cells) represent the second most abundant immune cell subset in the endometrium and account for 10–20% of the decidual leukocyte population (49). They are responsible for the surveillance and scavenging of bacteria present on mucosal surfaces and act as antigen presenting cells (APCs) (58, 63, 64). APCs are equipped with pattern recognition receptors (PRR), such as Toll-like receptors (TLRs), allowing them to recognize the so-called pathogen-associated molecular patterns (PAMPs). PAMPs are species-specific and include molecules from the microbial cell wall (e.g., peptidoglycan) and cell membranes (e.g., LPS) or virus-derived single stranded DNA or double stranded RNA. PAMPs promote the production of cytokines and cell adhesion molecules that lead to the recruitment other immune cells such as neutrophils and NK cells. In addition, APCs have a major function in instructing the adaptive immune response by virtue of expressing major histocompatibility complex (MHC) class II receptors. In contrast to uNK cells, the numbers of decidual macrophages remain relatively constant throughout gestation, however, the diverse repertoire of macrophages in cytokines production, in regulation of T cell responses and in tissue repair suggest an important role in decidualization (49, 59). Together with the uNK cells, uterine macrophages are also postulated to facilitate angiogenesis during placentation and more importantly, spiral artery remodeling by production of growth factors and clearance of cell debris (60, 65).

Uterine dendritic cells (uDC) have a tolerogenic phenotype and both uDC and uterine macrophages produce IL-10, TGF-β, and indolomine2,3 (IDO) contributing to a tolerogenic and receptive micro-environment (66). In addition, uDC have been shown to interact in a bidirectional manner with uNK cells in mouse models, as well as in vitro human studies (67, 68). As a subclass of APC, uDC promote uNK differentiation and activation in a contact-dependent manner and via the production of IL-15 (69).

Adaptive immune cells (B- and T lymphocytes) provide highly specific and long-lasting cellular and humoral immunity against pathogens. While B cells are relatively infrequent in the female reproductive tract, T cells can be consistently found in both the vaginal compartment as well as in the uterus, albeit, in the uterus, their number and phenotype highly vary depending on stage of the menstrual cycle and pregnancy. CD8+ (cytotoxic) T cells represent the most abundant adaptive lymphocyte subset in the pregnant uterus (70). CD4+ (helper) T cells are less abundant, however, important in production of cytokines and interaction with other immune cells upon activation by APC (71). Traditionally, pregnancy was considered an immune privileged state by the different cytokines produced from a dominant T-cell phenotype (Th2) involved in immune tolerance, over another phenotype (Th1) involved in immune rejection. The dominance of Th1 polarized T cells was considered detrimental to embryo implantation and was associated with obstetric complications mainly preeclampsia (72). More recent studies show that the Th1/Th2 paradigm is not inclusive enough and that fetal tolerance is a complex process involving more specialized T cell subtypes, such as Th17 and regulatory T (Treg) cells (51). In humans, Treg cells have been shown to migrate from peripheral blood to the decidua (73), and their levels peak during the second trimester of pregnancy (74). They produce IL-10, leukemia inhibitory factor (LIF), transforming growth factor (TGF)-β, and heme oxygenase 1 (HO-1) contributing to fetal-maternal immune tolerance (75). It is clear that tight regulation of T cell activation and polarization is essential to balance the protection from pathogens, mediated by Th1/Th17 CD4+ T cells and CD8+ (cytotoxic) T cells versus tolerance to paternal antigens expressed by fetal cells, mediated by Th2 and Treg cells (76).

The induction of regulatory T cells (suppressor T cells/Treg) is favored over pro-inflammatory Th17 cells through interaction of uDC and uNK cells (77), corroborating the importance of intricate immunological instruction in acquisition of tolerance during implantation. Both macrophages and uDCs can be activated by encountering pathogens in the endometrium and start the process of phagocytosis, internalization, and degradation of the components of the antigen. They subsequently present these bacterial peptides to T-cells via MHC receptors, which activate the T-cells to initiate a cell-mediated and/or humoral immune response via MHC class II molecules (78).

Even though the intricate role of the immune system in pregnancy is not completely understood, collaborative action of innate- and adaptive immune cells appear to be critical for orchestration of the immunological changes required for successful fertilization, implantation and pregnancy.

Successful implantation and subsequent formation of the placenta (placentation) encompass several steps ensuring tissue adhesion between fetal trophoblasts and maternal tissues and the adaptation of their blood vessels and to facilitate nutrient supply (101). These steps involve mechanisms such as angiogenesis (101), decidualization (102) and immune response adaptation (60). This immune adaptation is essential in pregnancy and is required in order to avoid a graft vs. host disease between the semi-allogenic trophoblast and maternal tissues, including immune cells, decidual microbiota and other decidual components such as epithelial and stromal cells and blood vessels. Not surprisingly, this complex interaction is postulated to affect subsequent stages of pregnancy, and is implicated in many pregnancy complications (103).

Published literature points to the existence of a diverse and metabolically active endometrial microbiota and predicts an important physiologically modulatory role of the main function of its host tissue: harboring and nurturing the developing embryo. In healthy women, the presence of commensal bacterial communities in the cycling endometrium mediate physiological responses from various cells at the feto-maternal interface, including epithelial and stromal cells in the endometrium, immune cells and trophoblasts from the implanting embryo. Although the exact molecular nature and extent of these interactions are not fully understood, evidence from in vitro experiments and animal models have provided important insights (13, 55, 104, 105). Based on currently available knowledge, we postulate the following mechanisms:

How the host-microbiota interactions affect the maternal immune response during implantation is not clearly understood. One of the factors, which govern immune modulation and maternal tolerance is the Pre-Implantation Factor (PIF) (108). PIF is a 9–15 amino acid peptide secreted by viable placenta with high concentrations in the maternal circulation in the first trimester (108, 109). PIF shows local effects on the endometrium and trophoblast promoting implantation and invasion of the trophoblast and has a direct impact on immune cell function and targets neutrophils and macrophages, as well as CD4+ and CD8+ T-cells (109). PIF reduces the activation of the NALP3 inflammasome complex (mainly TLR-4 mediated) resulting in reduction of pro-inflammatory cytokines such as IL-1β, IL-18, and IL-33 (110, 111). In addition, PIF creates an anti-inflammatory milieu by reducing IFNγ and stimulating IL-10 secretion, as well as enhancing Th2 cytokines. In context of the macrophages responding to bacterial stimuli, PIF blocks the release of nitric oxide induced by lipopolysaccharides (LPS) (108). This effect is present in case of excessive stimulus only. Thus, PIF operates as an immune modulator, rather than an immune suppressor with minimal impact on the innate, but firm effect on the adaptive immune response (108). Additional protective effects on the embryo protection and development include targeting the protein-disulfide isomerase (PDI) and heat shock proteins (HSP), which impact oxidative stress and protein misfolding (112, 113). Overall PIF is an example of embryonal factor shaping maternal immune response and therefore promoting successful implantation by generating an anti-inflammatory milieu and facilitates immune tolerance. The interaction of microbiome and PIF on pregnancy complications are currently under investigation.

As outlined in preceding sections, the immune system and microbiota play an interactive and collaborative role in maintaining a physiological healthy state in the reproductive tract. Disturbance of this physiological interaction has been implicated in the onset of diverse complications related to female reproductive health.

One of the largest longitudinal cohort studies that assessed vaginal microbiota vs. risk of STI is the Longitudinal Study of Vaginal Flora (LSVF) based in the United States (116). Women diagnosed with BV (by Nugent score assessment) had a nearly 2-fold increased risk of STI, like trichomonal, gonococcal, and/or chlamydial infection (of note: microbiota-testing preceded detection of STI by 3 months). This evidence is in agreement with previous reports that define vaginal dysbiosis as a predictor for gonorrhea and chlamydial infection (117). More recently, individuals with chlamydial infection were found more likely to have a cervicovaginal microbiota dominated by L. iners, or non-L. crispatus anaerobic bacteria (118), although not all associations were statistically significant and may depend on ethnicity (119, 120).

Human Papilloma Virus (HPV) and the Human Immunodeficiency Virus (HIV) represent sexually transmitted viral infections increasingly studied in association with reproductive tract microbiota. Women with detected or persistent HPV infections showed a more diverse vaginal microbiota, in studies conducted among African/Caribbean and Italian women and suggested an association with Atopobium spp. and G. vaginalis (121, 122). Among Nigerian women, the prevalent high-risk HPV (hrHPV) infection was associated with a decrease in Lactobacilli and abundance of anaerobes, particularly of the general Prevotella and Leptotrichia (123). In Asia an association between increased vaginal bacterial diversity and presence of HPV were reported, with a Korean study suggesting Fusobacteria, in particular Sneathia spp. to be particularly implicated (124). A Chinese study identified several microbial genera in hrHPV-infected women (Bifidobacterium, Bacillus, Megasphaera, Sneathia, Prevotella, Gardnerella, Fastidiosipila, and Dialister), while another set (Bifidobacterium, Megasphaera, Bacillus, Acidovorax, Oceanobacillus, and Lactococcus) in hrHPV-infected pregnant women. In pregnancy, this study showed an association between a more diverse cervical microbiota and HPV (125). In addition, several genera and species were associated to HPV positivity (Ureaplasma parvum), HPV negativity (Brochothrix, Diplorickettsia, Ezakiella, Faecalibacterium, and Fusobacterium), likelihood of reinfection (Actinomyces) or persistence (Prevotella, Dialister, and Lachnospiraceae) (126). Recent data also suggested altered microbiota in placenta, cervix and mouth in the presence of HPV infection (119).

Two decades ago, the absence of Lactobacilli was already associated with an increased risk of acquiring HIV infection in a cohort of Kenyan women (120). Similarly, vaginal dysbiosis was suggested as a contributor to the acquisition of HIV in Ugandan and Zimbabwean women (127). Conversely, Rwandan women with a Lactobacillus-dominated (particularly L. crispatus) cervicovaginal microbiota were less likely to be infected with HIV, hrHPV, as well as Herpes Simplex Virus 2 (HRV 2) (128). Low pH, due to lactic acid production by Lactobacilli, was suggested as a main strategy to prevent HIV infection, either by means of inactivating the virus, or inactivating T lymphocytes, thus decreasing their susceptibility to HIV infection (129). Other possible defense mechanisms of vaginal microbiota to HIV include the production of peroxide or bacteriocins by Lactobacilli, although their effect on viral biology is not fully understood (129). These findings are seemingly congruous with the protective role of Lactobacilli in urinary tract infections [as reviewed in (130)].

The above mentioned sexually transmitted infections, among others, have been extensively implicated in diverse reproductive tract complications, from infertility, to adverse pregnancy outcomes. A growing body of evidence collectively supports the association between vaginal dysbiosis and different genital tract infections and corroborates an important physiological role for microbiota in protection from, or susceptibility to pathogenic infections. Interaction and cross-regulation between the vaginal microbiota and the immune responses in the lower genital tract are therefore crucial in creating a protective environment against external pathogens, thus ensuring the right condition for the initiation of pregnancy.

In recent years, the availability of culture-independent sequence techniques has led to a rise in the number of studies investigating the association of disturbed vaginal and endometrial microbiome composition and reproductive failure, focusing mainly on implantation failure. In women undergoing in vitro fertilization (IVF), the percentage of vaginal and endometrial Lactobacilli were significantly lower than non-IVF patients and healthy volunteers (131). In addition, studies have shown that presence of various bacterial contaminants, such as Enterobacteriaceae, Streptococcus, Staphylococcus, and Gram-negative bacteria, from catheter tips at the time of embryo transfer had a negative impact on pregnancy outcome, as reviewed in (13). Using 16S ribosomal RNA sequencing of paired endometrial and vaginal samples from 13 fertile women and 35 infertile patients undergoing IVF, Moreno et al. showed that the presence of a non-Lactobacillus-dominated microbiota (defined as <90% Lactobacillus spp.) was associated with significant decreases in implantation (60% vs. 23%), pregnancy (70% vs. 33%), ongoing pregnancy (59% vs. 13%) and live birth (59% vs. 7%) rates (43). Using a similar approach in a cohort of 31 women, Bernabeu et al. showed that women achieving pregnancy after IVF (cryotransfer of a single embryo) showed a greater presence of Lactobacillus spp., while a trend toward higher alpha diversity in vaginal samples was found in patients who did not achieve pregnancy and no difference in beta diversity (132). In a recent large prospective cohort study, Koedooder et al., used a new technique “IS-PRO” (see section “Diagnostic Challenges”) to examine microbial profiles of vaginal microbiota in 192 women undergoing IVF. Women with a low percentage of Lactobacillus in their vaginal sample were less likely to have a successful embryo implantation (133). This failure was correctly predicted in 32 out of 34 women based on the vaginal microbiota composition, resulting in a predictive accuracy of 94% (sensitivity, 26%; specificity, 97%). Additionally, the degree of dominance of Lactobacillus crispatus was an important factor in predicting pregnancy: none of the women who had a negative prediction (low chance of pregnancy) became pregnant. Taken together, these data suggest that a balanced, less diverse vaginal microbiota, dominated by Lactobacillus species increased the chances of a successful outcome.

In women with recurrent reproductive failure, ascribing a causative or correlative connection to aberrant microbiota is controversial. This group is composed of women with recurrent miscarriage (RM) (defined as loss of two or more clinically or biochemically established pregnancies) (134) and women with recurrent implantation failure (RIF), defined as loss of two or more pregnancy losses after transfer of good-quality embryos (135). RM and RIF both have heterogeneous etiology, with diverse risk factors being implicated covering genetic, metabolic, hormonal, immune maternal aspects. Although several groups have studied microbiota disturbances in women with recurrent reproductive failure, the complex pathogenesis has hampered any meaningful conclusion on the role of reproductive tract dysbiosis in this early pregnancy disorder. Analysis of endometrial samples from women investigated for recurrent reproductive failure showed that the uterine microbiota was dominated by Bacteroides species in >90% of the women (35). However, dissimilarities in dominance of Prevotella spp. or L. crispatus due to possible contamination from the vagina limits the interpretation of these data.

Chronic endometritis (CE) is an inflammatory condition typified by dysregulated interactions between endometrial pathogens and the endometrium. Chronic endometritis is a persistent inflammation of the endometrium, characterized by the presence of plasma cells syndecan-1 (CD138) on immunohistochemical staining of endometrial biopsies [reviewed in (136)]. Although various pathogens have been implicated in causing CE, the most commonly reported species were common bacteria (Escherichia coli, Enterococcus faecalis, and Streptococcus agalactiae) in 77.5%, followed by Mycoplasma/Ureaplasma (25%) and Chlamydia (13%) (137).

Recently, research has focused on the role of CE in reproductive failure, with various studies reporting a wide range prevalence depending on the clinical characteristics of the studied group and reflecting heterogeneity of diagnostic methodology and definitions. Studies have found an increased prevalence of CE in women with recurrent pregnancy loss (13%) (138) and RIF (30%) (139), while the rate of CE in the general infertility population was suggested to be much lower, with a prevalence of 2.8% among 606 infertility patients (140). A recent meta-analysis of five studies (total of 796 patients) concluded that women receiving antibiotic therapy for CE did not show any reproductive advantage in comparison with untreated controls (141). However, patients with cured CE, confirmed by a repeat biopsy, showed higher clinical pregnancy rate (OR 4), ongoing pregnancy rate/live birth rate (OR 6.8) and implantation rate (OR 3.2) (141). The exact mechanism of how CE affects implantation is yet unknown, but a negative effect on endometrial receptivity by abnormal infiltration of plasma cells (B lymphocytes) and antibody production is suggested (142). Similarly, as the causal connection between CE and dysbiosis is unknown, it is unclear whether a status of dysregulated immune response in the endometrium is responsible for the higher prevalence of dysbiosis, or whether dysbiosis is the cause of CE.

How the disturbance of the endometrial microbial ecosystem can have a negative impact implantation process is not fully understood. Parallel to the proposed mechanism of interaction between the microbiota and endometrial cells, a disturbed balance can act upon the following mechanisms (Figure 3). (1) First, the dominance of non-commensal bacterial communities could weaken the integrity of the endometrial mucosal barrier by affecting the epithelial tight junctions and reducing AMP and mucin secretion; (2) This in turn could further weaken host defense mechanisms and allow pathogens to enter the endometrial stroma and elicit a profound immune reaction from APC and other immune cells harboring pattern recognition receptors (PRR); (3) Aberrant stimulation of T-cells, either directly from invading pathogens breaching the mucosal barrier, or indirectly from absorbed bacterial products, results in disbalance in cytokine production in favor of the pro-inflammatory Th-1 types, predominated by TNF-a, IFN, IL-2, and IL-10. Abnormal uNK cell maturation, either primary or secondary to shallow extravillous trophoblast (EVT) invasion, is a possible link between a disturbed endometrial microbiome and incomplete remodeling of maternal spiral arteries, characteristic of the great obstetric syndromes (143).

Alteration of the ecology of the female reproductive tract has been linked to maternal and fetal health, and to adverse pregnancy outcomes (27, 79, 144, 145). The most widely studied pregnancy complication in relation to the vaginal microbiota is preterm birth (PTB) (146–148). The prevalence of a Lactobacillus-poor microbiota was inversely correlated with gestational age at delivery in some studies (38), not in others (149). Distinctive species of Lactobacilli were reported to be associated with differential pregnancy outcomes. Indeed, in two ethnically distinct cohorts, L. crispatus was associated with a low PTB risk, whereas L. iners was not (42). Also, Gardnerella was also associated with PTB and coexisted with L. iners, but not with L. crispatus (42). Based on these findings, a model was suggested to help describe the interplay between key vaginal bacterial species: the presence of G. vaginalis correlates with adverse outcomes, such as PTB and symptomatic Bacterial Vaginosis (BV); G. vaginalis and L. crispatus are strongly mutually exclusive, while this is not the case for L. iners and Gardnerella (150). Comparison of vaginal samples of 90 women who delivered at term and 45 women with preterm birth suggested that a specific signature: the presence of BV-associated bacterium (BVAB) 1, Prevotella cluster 2, Sneathia amnii and BVAB-TM7 in early pregnancy, may be useful for prediction of PTB risk, particularly in high-risk populations of African ancestry (147).

Abnormal vaginal colonization in the second trimester was also associated with an increased PTB risk (151). Similarly, in a predominantly African-American population, an increased vaginal microbial community richness and diversity between the first and second trimester was associated with PTB (152). In addition to decreased Lactobacillus spp., specific pathogens have been linked to PTB in different populations, such as Klebsiella pneumonia, Gardnerella, Ureaplasma and other genera, including Prevotella, Atopobium, Sneathia, Gemella, Megasphaera, Dorea, Streptococcus, and Escherichia/Shigella (38).

At this point, although one out of four preterm births appears to be associated with intra-amniotic infection, and some associations between microbial states or abundance of individual species with PTB have been reported, it does not appear clear whether changes in the bacterial communities of the lower genital tract allow for a clear identification of women at risk (144). A recent overview summarizing studies on the association between vaginal microbiota and PTB emphasized the methodological heterogeneity, and scarcity of, studies in the field (146). Overall, research on the topic has produced conflicting outcome, often related to ethnical background of the women included and the associated risk degree of PTB. Nonetheless, more recent studies more consistently report an association between vaginal dysbiosis and PTB, possibly resulting from the improved understanding of the contribution of L. iners to this association (146).

Multiple studies provide evidence of placental and amniotic fluid microorganisms affecting miscarriage, chorioamnionitis, premature rupture of membranes (PROM), stillbirth, preeclampsia (PE), and intra uterine growth restriction (IUGR) rates (143, 153). In some cases of spontaneous preterm delivery, microorganisms have been found to invade the amniotic cavity, leading to increased maternal/neonatal morbidity and mortality (154). Infectious bacteria gain access, typically via ascending route and/or perturbations of the vaginal microbiota and can have an adverse impact on pregnancy outcomes (146). Chorioamnionitis is an inflammatory disease of the extraembryonic membranes, placenta and amniotic fluid due to microbial invasion mostly commonly Ureaplasma and Mycoplasma spp. infections (155). Culture-dependent studies identified members of the genera Prevotella, Bacteroides, Peptostreptococcus, Gardnerella, Mobiluncus and genital mycoplasmas in the placentae of women delivering preterm with or without PE, suggestion the involvement of multiple bacterial strains (15). In line with this, antibiotic treatments have not reduced the rates of preterm birth, suggesting that a single inflammatory/infectious pathway may not fully explain the problem (144). In contrast, DNA-based investigations of the placental microbiota in PTB showed increased enrichment of Burkholderia spp. and an increased relative abundance of Alphaprotoebacteria and Actinomycetales and mixed non-cultivable anaerobes (15). However, in case of chorioamnionitis a higher abundance of Streptococcus agalactiae, Fusobacterium nucleatum and Ureaplasma parvum was reported (15). These results fuel the debate whether a prenatal bacterial microbiota really exist (45). In physiological pregnancies or in the presence and absence of active labor, many “causal” microorganisms or their DNA are detectable in the placenta and amniotic fluid (156, 157). Since microbiotal ratios of specific microbial species change throughout gestation, the placental response to such environmental cues possibly does as well (158). This is supported by the observation that PIF is differently expressed in the placenta throughout gestation or in response to an inflammatory insult (109). Since, as of yet, the amniotic fluid and the membranes are not available for non-invasive sampling, the time of onset of chorioamnionitis is not available for clinical stratification and decision making. Besides placental inflammation, systemic inflammation in concert with oxidative stress and endothelial dysfunction plays a role in pregnancy complications such as IUGR or PE (159).

The interaction of vaginal and endometrial microbiota with local maternal components modulates the maternal immune system. Although the contribution of the fetal immune system to this interplay is currently unknown, it is conceivable that when the fetus initiates an inflammatory response, premature labor may impose a risk to fetal wellbeing. Conversely, there is an inherent risk that the fetus may be injured by a longer stay in utero either directly through microbial toxins or through proinflammatory cytokines. This delicate balance between maternal and fetal needs possibly dictates the course and outcome of pregnancy and may contribute to long-term health of the offspring. This may be another example how future health is primed by the intrauterine environment according to the principles of the DOHAD (Developmental Origins of Health and Disease) hypothesis (160).

As discussed throughout this review, one of the major limitations in microbiome research is the diversity of the techniques used and the databases linked to those techniques to identify individual species and determine the composition of microbiota. Such differences potentially introduce significant variations in analysis and constitute a major source of difference in interpretation of data. Although currently no compelling evidence points to the existence of a universal mammalian placental or fetal microbiota, consensus on the importance of uniform technological and analytical approaches warrants further investigation into standardization of microbiota research and incorporating geographical, ethnic and societal data (45) (Table 1). These recommendations are general for all microbial community profiles, including the female reproductive tract microbiota, and are expected to reduce variation and inconsistency between studies on microbiomes.

Moreover, most of the techniques described have not undergone rigorous validation steps and certification by regulatory organs, such as the European commission in vitro diagnostic (CE-IVD) label or the American Food and Drug Administration (FDA) approval, limiting their commercial availability. The availability of CE-IVD-certified microbiome analysis tools would meet part of the recommendations (see: Table 1) and allow more reliable comparisons between studies in different countries and settings. Besides the widespread Next-Generation Sequencing (NGS) approaches, our group developed another detection technique called IS-pro, first described in 2009 (161). The IS-pro test is already CE-IVD certified and is currently being used in many different disease profiling activities and applications including analyses of the vaginal microbiome as a predictor for outcome of in vitro fertilization (133). A wider application of this technique to study the role of reproductive tract microbiome in various pregnancy complications is expected to be available in the near future.

The general concept of interactions between commensal microbiota and human cells has recently been embraced as a principal of human physiology. The realization that a dysbalanced and/or harmful microbiotal composition frequently correlates with specific clinical conditions, has led to a sharp rise in studies on host-microbiota interaction over the last decade (25, 55, 114, 162, 163). Although it is becoming increasingly clear that the interplay between host and microbiota also affects human reproductive biology, the exact molecular mechanisms underpinning these interactions are far from understood.

All long-lasting physiological adaptations of cells and tissues in response to altered environmental conditions have their basis in altered epigenomic programming. Environmental changes are detected by a myriad of cellular sensing mechanisms and, via signal transduction routes, ultimately reach the nucleus where environmental cues are translated to epigenetic regulation and chromatin remodeling. This epigenetic regulation of genetic input and potential environmental cues underlies maternal physiological plasticity and embryonal development during gestation. Programming of immune cells during immunological responses is mediated by cellular stress responses and is typically accompanied by metabolic changes in the cell. Prolonged disturbance of these interactions is associated with the development of immunological disorders, such as chronic inflammatory conditions (164). The close link between cellular metabolism and epigenetic responses has its origin in early evolution, has enabled multicellularity and is closely connected to organismal survival (165). Hence, fine-tuning of epigenetic control occurs in conjunction with cellular metabolic status, as available energy ultimately directs and limits cell responses. Recent advances in this field have identified oxygen, numerous metabolic intermediates, cellular reduction (NADH, FADH₂) and energy equivalents (ATP, GTP), as direct molecular effectors of epigenetic regulatory activity (165). Conversely, cellular metabolic adaptation is controlled by epigenetic regulation, substantiating the reciprocal nature of the physiological interaction between metabolism and epigenetics (166). Sex hormones, metabolic profiles, nutrition, maternal stress, drugs, smoking and air pollution represent obvious examples of environmental cues that, via active modulation of epigenetic regulatory mechanism.

Interestingly, microbial metabolites, among which Short-Chain Fatty Acids (SCFA), like acetate, butyrate and propionate, are known to harbor the ability to alter epigenetic status of numerous cell types; studied examples thereof include the effect of such compounds on immune cells (78). In the context of human reproduction, it is conceivable that microbial metabolites, directly (e.g., via SCFA) or indirectly (e.g., acidification, alkalization, inflammatory responses), can either support (healthy microbiota) or upset (unbalanced/harmful microbiota) local cell-cell communication and tissue physiology and adaptation. As such, all relevant reproductive processes, including fertilization, implantation, placentation, immune tolerance, embryonal development, infant and adult health and may be harmfully altered by dysbiosis (167). Such gene-environment interactions are yet to be examined in detail in the context of reproductive health and disease.

Given the association between pregnancy complications and microbiota, the question of modulation strategies is valid. Postnatal dietary strategies, including human colostrum/milk or prebiotics/probiotics, reduce morbidities in preterm infants (163). Evidence of prenatal strategies to support reproductive success and fetal health is slowly emerging: modulation of early microbiota in pregnancy shows promising effects (168). Maternal bacteria were shown to enter the gastrointestinal tract of the fetus (169) and microbiota alteration in the neonate and placenta is detectable in pregnant women receiving probiotics (170) but still under debate (14).

Recently our group published a meta-analysis on the relation between vaginal microbiota and early pregnancy development after IVF and the effect of probiotics thereon (71). It provides an overview of published studies describing long term modulation of the vaginal microbiome using Lactobacillus-based probiotics. Patients were often treated by Metronidazole, and followed up with a Lactobacillus-based probiotic. Besides lactobacillus-based probiotics, mixtures of Lactobacillus, Bifidobacterium and Streptococcus strains were also used in different studies (71). Further studies assessing the potential to modulate pregnancy outcomes are needed.

The notion that the neonatal immune system can be shaped by early fetal microbial colonization by inference implies that reciprocal interactions between the host and microbiota exist (171). This hypothesis is in line with the evidence that factors like PIF mediate maternal immune tolerance during pregnancy (108). Fine tuning of immune modulation is not only relevant for embryo implantation but also for defense against potentially harmful microbes. An exaggerated maternal immune response is putatively linked to preterm birth and fetal loss (172). Therefore, a novel strategy could be the maternal immune system modulation by synthetic PIF (110). Inflammatory challenge during pregnancy results in endogenous PIF expression and additional administration of synthetic PIF could prevent fetal loss.

The neonatal microbiota, mainly that of the gut, skin and oral cavity, has long been postulated to be acquired postnatally known as “postnatal microbiota seeding”. Alternatively, the hypothesis of perinatal microbiota transfer and its potential relevance to infant and adult general health is gaining attention. Given the importance of transfer of maternal microbiota to the child via colonization, the significance of birth mode is of high interest (173). Vaginal birth exposes the baby to maternal vaginal and intestinal microbiota, whereas cesarean section limits exposure of the newborn to parental dermal microbiota and any microbes present in the surgical theater. The “prophylactic” antibiotic treatment of the mother during labor or cesarean section may aggravate any effect of non-natural birth on colonization and may also negatively affect intestinal microbiota of the offspring (174). Hence, birth mode may have long-lasting effects on the composition of the newborn gut microbiome, and predispose for adverse health outcomes (174). The notion that cesarean delivery deprives the infant of exposure to vaginal microbiota and consequently leads to neonatal dysbiosis has led to the popularized, yet unsubstantiated and potentially hazardous practice of “vaginal seeding”. In a pilot study of 18 mother-infant pairs, there was partial restoration of microbiome (mainly of the skin and oral cavity and less so of the gut) in infants exposed to vaginal fluids from a vaginally placed gauze after cesarean delivery (n = 4), compared to non-exposed infants (n = 7), and resembling the microbiome of vaginally delivered infants (n = 7) (175). This widely cited trial was criticized for the small sample size and the potential bias from confounders such as intrapartum antibiotic prophylaxis and maternal BMI (173, 174). Although postnatal vaginal seeding altered newborn intestinal microbiota composition over several months, this intervention is believed to introduce an inherent risk of transferring pathogenic microbes (e.g., HPV, GBS) onto the newborn (176). For this reason, perinatal seeding is currently not encouraged as standard practice by the American College of Gynecology and Obstetrics (177). Whether and how vaginal seeding has any long-term beneficial health effects for the offspring requires large properly conducted studies with robust microbiome analysis.

As microbiota represents an environmental factor which affects host-microbe interactions at the epigenetic level, detailed understanding of the molecular workings of the close functional interplay between host cell systems and microbiota holds the promise that therapeutic intervention strategies can be designed for the benefit of general and reproductive health. These include the use of probiotics, beneficial microbial metabolites, rational diets and/or (ant)agonists of specific biological response pathways (162). Combined the research cited in his review define opportunities for modulation of the female reproductive tract microbiota, while harnessing its protective and immuno-regulating role during pregnancy. Such opportunities should take into consideration individual differences in microbial communities, and tailor therapeutics to different anatomical and gestational factors in an attempt to provide precision tools for reproductive health.