Variation in the microbiome during pregnancy

During pregnancy, the microbiome across maternal body sites is altered tremendously. Significant microbial changes have been observed in the oral cavity, gut, and vagina during pregnancy (figure 2A).46–48 The richness and diversity of the microbiome in the oral cavity increase during early pregnancy, among which Porphyromonas gingivalis and Aggregatibacter actinomycetemcomitans are the most significant species that exhibit higher abundance during pregnancy.49 In addition, some species of Candida thrive during middle and late pregnancy.49 50

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Microbial variation and transmission during pregnancy. (A) Variation in the abundance of representative bacteria in the maternal oral cavity, gut and vagina during pregnancy. (B). Host–microbe interaction in the maternal gut. Under healthy conditions, the gut microbiome and immune response of pregnant women is similar to that in individuals with metabolic syndromes, characterised by decreasing butyrate-producing bacteria and increasing proinflammatory cytokines, inducing moderate inflammation (I). When mothers have conditions such as T2D, obesity or IBD, deviation of the gut microbiome associated with alteration of the immune response is observed during pregnancy, increasing the risk of inflammatory diseases and gut leakage. Increased intestinal permeability allows the entry of bacterial toxins into the systemic circulation and induces multiple diseases (II and the left panel). Some interventions, such as probiotic, anti-inflammatory diet or FMT interventions, may restore dysbiosis of the gut microbiome and reduce inflammatory responses (III and the left panel). (C) Maternal–fetal interface communication during pregnancy. Maternal antibodies such as IgG are transferred to the fetus via FcRn (case 1). Some bacterial molecules are bound to maternal IgG and are transferred to the offspring (case 2). Such vertical transmission provides protection to the fetus. Fetal immune responses can also be activated by bacterial components or bacterial metabolites from mothers (case 3). Other substances, such as viruses, pathogenic bacteria, virulence factors and parasites that are harmful to the fetus are usually unable to cross the placenta (case 4, black lines with flat ends), except in situations of maternal infections (case 5, red dotted arrows). Translocation of the microbiome between maternal body sites (oral cavity, gut and vagina) and the fetus was observed. Whether the detection of microbes in the placenta is derived from contamination is highly debated, and thus, dotted lines are used for microbial translocation (the left panel). FcRn, neonatal Fc receptor; IBD, inflammatory bowel disease; PAMPs, pathogen-associated molecular patterns; PRRs, pattern reorganisation receptors; SCFAs, short-chain fatty acids; Tregs, regulatory T cells; T2D, type 2 diabetes.

Distinct from the variation in the oral microbiome, the microbial diversity in the gut and vagina significantly decrease during pregnancy (figure 2A).46–48 In women at the beginning of pregnancy, the gut microbiome pattern is similar to that of healthy nonpregnant women, characterised by a high abundance of Bacteroidetes and Firmicutes, such as Clostridiales.51 From the first to the third trimester, profound alterations are observed in the guts of pregnant women, where butyrate-producing bacteria, such as Faecalibacterium, exhibit a significant decrease, while Proteobacteria and some lactic acid-producing bacteria, such as Bifidobacteria, are highly increased.51 52 Some studies revealed the expansion of members of Enterobacteriaceae and Streptococcus in the third trimester,47 51 which were also early colonisers in the infant gut, indicating potential transmission from the maternal gut to the neonatal gut. Changes in the microbial composition, together with immune and metabolic variation, often induce weight gain in normal pregnant women. In one study, the first-trimester and third-trimester faecal microbiome was transplanted to germ-free mice, and greater weight gain and insulin resistance were observed in the mice that received the third-trimester microbiome compared with those that received the first-trimester microbiome.51 These findings indicate that the gut microbiome makes a great contribution to healthy pregnancy and that changes in the gut microbiome potentially lead to changes in host immunology and metabolism.

The between-individual diversity (beta diversity) of the gut microbiome greatly increases during pregnancy.47 48 In contrast, the vaginal microbial composition of pregnant women exhibits considerable convergence across different populations (figure 2A).53 54 During pregnancy, Lactobacillus becomes the only predominant bacteria in the vagina in most women, leading to a dramatic decrease in alpha diversity (figure 2A).55 56 Numerous studies have demonstrated that a core microbial pattern exists in the vaginas of healthy reproductive-age women, characterised by the dominance of Lactobacillus species, termed community state types. The dominance of Lactobacillus spp is very important to the maintenance of a healthy vaginal microenvironment, where the production of lactic acid lowers the vaginal pH and inhibits the growth of other harmful bacteria.57–60 In contrast, a lower proportion of Lactobacillus is frequently associated with adverse pregnancy outcomes, such as miscarriage and preterm birth.35 61 62

Deviation of the maternal gut microbiome and interventions

It was demonstrated that the alteration of the gut microbiome and immune responses in late pregnancy resembled that in metabolic syndromes, such as obesity or diabetes, characterised by low microbial diversity, reduced levels of butyrate-producing bacteria, decreased insulin sensitivity and moderate intestinal inflammation (figure 2B).63 64 Short-chain fatty acids (SCFAs, eg, acetate, propionate and butyrate) are often considered to alleviate low-grade inflammation,65 and the loss of butyrate-producing bacteria is associated with increased production of proinflammatory cytokines and decreased generation of regulatory T cells (Tregs) (figure 2B).66 Nevertheless, the causal relationship between the gut microbiome and immune responses in pregnant women and the underlying mechanisms need further exploration.67

Dysbiosis of the gut microbiome refers to the loss of beneficial microbes and the enrichment of pathobionts, which are often associated with unhealthy lifestyles and may disturb the ecological balance in the intestines, leading to undesirable consequences (figure 2B).68 For example, many studies have emphasised associations between microbial dysbiosis and poor habits, such as smoking, drinking alcohol, consuming a high-fat diet (HFD) and drug abuse.14 69–76 Among these habits, consuming an HFD, which is a diet consisting of at least 35% of the total calories consumed from fats, including but not limited to animal fat, chocolate and butter,77 is one of the most common factors that disrupts the normal gut microbiome.71 Pregnant women who are accustomed to an HFD often suffer from obesity or gestational diabetes mellitus (GDM), exhibiting lower microbial diversity and persistent perturbation of Staphylococcus, Bacteroides, Bifidobacterium and Lactobacillus, as well as considerable variations in insulin secretion and insulin sensitivity.78 Nonetheless, several studies suggest that various outcomes may occur depending on fat types and compositions. Specifically, fish and unsaturated fatty acids such as vegetable oils and olive oils are suggested to have protective effects, whereas it is recommended that trans fatty acids and saturated fatty acids be consumed as little as possible.79 80

Deviation of the gut microbiome may disrupt the ecological balance in the intestines, leading to an increase in pathogen-associated molecular patterns, which can be recognised by pattern recognition receptors and activate the immune response through downstream signalling pathways (figure 2B).81–84 In addition, an abnormal gut microbiome is frequently associated with a higher risk of intestinal permeability,85 which impairs the gut barrier and allows the leakage of pathogens from the gut lumen into the lamina propria, increasing inflammatory responses (figure 2B).1 86–88 Persistent inflammation leads to the disruption of immune homeostasis and is associated with many autoimmune disorders and chronic metabolic diseases, such as GDM, T2D, non-alcoholic fatty liver disease, chronic kidney disease and atherosclerosis.86 Deviation of the gut microbiome not only has a detrimental effect on mothers but also influences offspring.72 78 89 90 In an intergenerational microbiome study of multiple body sites, Wang et al found an obvious concordance of microbial variation between neonates and mothers suffering from GDM. The deviation of the neonatal gut microbiome was associated with metabolic depletion and virus prevalence in the meconium, indicating the importance of microbial inheritance during pregnancy.91 Other studies also demonstrated that for pregnant women who use antibiotics (eg, ampicillin, penicillin and cefazolin), great microbial alteration is observed in their offspring, with markedly reduced microbial diversity and abnormal microbial composition.75 92 93

The occurrence of many diseases can be associated with dysbiosis of the gut microbiome.94 95 To better improve maternal health and fetal outcomes, it is necessary to restore the maternal gut microbiome via interventions. Recently, probiotic interventions for pregnant women have been introduced in many clinical trials.96–98 Probiotics are specific live microbial cultures that benefit the host by improving its intestinal microbial balance.99 Currently, many metabolic syndromes and immune diseases, such as allergies, obesity, GDM and T2D, have been treated with such a strategy, and promising progress has been made.97 98 100 For example, Isolauri et al have focused on probiotic interventions for both pregnant women and infants for a long time. In a double-blind, placebo-controlled study including 256 pregnant women, they demonstrated that a probiotic-supplemented diet was an effective method for the prevention of obesity during pregnancy and could significantly reduce the occurrence of GDM for mothers.101 102 Another follow-up study from birth to 10 years indicated that a perinatal probiotic intervention helped reprogram the infant gut microbiome and avoid offspring obesity in childhood, especially for children with a parent or sibling who suffered from immune diseases.103 104 Such studies suggest the inheritance of the maternal microbiome during pregnancy and emphasise the importance of maintaining homeostasis of the maternal microbiome before and after delivery. Other groups also confirmed the protective effect of probiotic interventions on restoring the maternal gut microbiome and preventing multiple gestation disorders such as GDM, dyslipidaemia, preeclampsia and excessive gestational weight gain.98 105

The most frequently used probiotics include strains from Bifidobacterium and Lactobacillus. On the one hand, such strains modulate the gut microbiome by competing with and inhibiting pathogen adhesins. On the other hand, they produce antibacterial substances, SCFAs, and anti-inflammatory cytokines, increasing the mucus layer and cell junctions (figure 2B). Despite the low biomass, probiotics help restore homeostasis of the gut microbiome and reduce intestinal inflammation.106–108 The gut microbiota is a balanced ecosystem, and the occurrence of many diseases is associated with the disturbance of some key species.109–111 Most recently, Xiao et al explored common characteristics of the gut microbiome in multiple diseases. Based on a novel biomarker identification algorithm, they constructed microbial networks related to different diseases and revealed a high prevalence of multirelated bacteria, which exhibited wide associations with multiple diseases, in global populations. The similar disorder pattern of the gut microbiome in different disease networks suggested the driving effects of some keystone microbes in the development of disease.112 Consequently, identifying keystone species in microbial networks and developing new probiotic interventions to target key taxa can be a promising strategy for disease treatment.

In addition to probiotic interventions, other diet-based interventions and microbiome-based therapies, such as faecal microbiota transplantation (FMT), have also demonstrated successful results for the restoration of the normal microbiome (figure 2B).113–116 Nonetheless, a recent study implemented FMT in one pregnant woman and reported the detection of donor-derived bacterial strains in the later-born infant, indicating vertical transmission during the perinatal period.117 No further disorders have been observed in the offspring. This case reminds us that the choice of therapy for pregnant women and neonates must be made very carefully, and the safety of such interventions, especially during pregnancy and at an early age, needs further evaluation.

Vertical transfer from mothers to fetuses

Multiple studies have shown that maternal exposure during gestation may alter the microbiome and immunity of offspring, indicating the importance of vertical transmission during pregnancy.69 91 118 The placenta is an organ that is essential during the development of fetuses, providing indispensable nutrients and oxygen to growing babies,119 and frequent substance exchange occurs at the maternal–fetal interface (figure 2C).11 120 Many maternal components, such as antibodies and immune cells, are able to traverse the placental cell layers from mothers to fetuses.121–125 For example, maternal IgG is the dominant antibody that can cross the human placenta via neonatal Fc receptor (FcRn), providing initial protection to the fetus from pathogens (figure 2C).11 120 126

In contrast to the direct transfer of antibodies, the maternal microbiome impacts and promotes fetal immune development in an indirect way. Many studies have indicated that bacterial antigens or metabolites can cross the maternal–fetal interface, inducing immune tolerance (figure 2C).120 126 One elegant experiment, designed by Gomez de Agüero et al, demonstrated the association between the maternal microbiome and the fetal innate immune system.127 A modified Escherichia coli strain was used to colonise germ-free pregnant mice, and before delivery, the dams were returned to sterile conditions. Compared with the offspring of the control groups, the offspring of the strain-colonised group showed altered intestinal transcriptional profiles and increased type 3 innate lymphoid cells and F4/80 mononuclear cells, indicating that the bacteria-induced immune development of infants starts before birth and that the gut microbiota, even transient residents, has the capability to prime fetal immune programming.127 Intriguingly, this study also showed that maternal IgG antibodies, which were bound to microbial molecules and transmitted to the offspring, played an essential role in postnatal innate immune development.127

In addition to the beneficial substances described above, other detrimental substances or microbes, such as viruses, pathogenic bacteria, virulence factors and parasites, may cross the placenta and induce fetal infection (figure 2C).128 Severe fetal infection may result in adverse pregnancy outcomes, such as preterm labour, miscarriage and maternal and fetal death.128 Even when no infection occurs, an unfavourable intrauterine environment can contribute to an abnormal gut microbiome and metabolome in offspring. In a large twin study focused on fetal growth restriction (FGR), Yang et al demonstrated that adverse intrauterine environmental factors related to selective FGR-dominated genetics in their effects of altering microbial diversity and composition in the offspring. Dysbiosis of the gut microbiome in early life is not only correlated with pronounced metabolic alterations but also has long-term effects on the neurobehavioural development of infants.129 Furthermore, previous studies have shown the vertical transmission of antibiotics through the placenta and demonstrated that the overuse of antibiotics during pregnancy may influence the resistome profiles of offspring.14 Such evidence emphasises the importance of maternal health to unborn babies.

Whether the microbiome already exists in utero before delivery has been highly debated for a long time.130 131 Recently, an increasing number of studies have shown that certain bacteria exist in the placenta or uterus.132 133 While some researchers claimed that this could be due to contamination, as the placenta or uterus contains very little biomass,130 134 others believed that potential transmission of the microbiome from mothers to fetuses may be involved in normal gestation.132 133 For example, as the closest organ, the maternal vagina is considered to be the most likely source to transfer the microbiome into the uterus (figure 2C). Multiple studies have revealed Lactobacillus spp, the dominant bacteria in the vagina, to be one of the most frequent microbes present in placental or endometrial samples, suggesting putative transmission of the microbiome ascending from the vagina to the uterine cavity.135–140

In addition to the vagina, the maternal oral cavity is another potential source involved in the seeding of the microbiome during pregnancy (figure 2C).136 Bacteria residing in the oral mucosa, such as Streptococcus and Fusobacterium, may translocate to the placenta through the maternal circulation, sometimes leading to preterm birth or miscarriage.137 138 141–143 With some gut-derived bacteria, such as Enterobacter and Enterococcus, being observed in placenta or fetal meconium samples, the maternal gut is also thought to be a main contributor to the microbiome of babies (figure 2c).6 138 144–146 Contrary to the potential adverse outcomes associated with microbial translocation from the maternal vagina and oral cavity to the uterus, the gut-derived microbiome is often identified in samples without any inflammation or infection, indicating the supporting role of the maternal gut microbiome in the development of the fetus.119 138

However, most ‘source-tracking’ studies focused on microbiome profiling between different organs and thus provide only indirect evidence of transmission. As no culture-dependent methods were used, it was difficult to determine whether the detected microbes were live bacteria or bacterial fragments or simply contamination derived from sample processing. Do such transmissions truly exist during pregnancy? What is the mechanism and further impact underlying the transmission? Additional research is required to answer these questions, which will be discussed in the next section.

Sterile womb hypothesis

It has long been thought that the uterus is a sterile environment free of microorganisms. One of the strongest pieces of evidence supporting this hypothesis is the successful cultivation of gnotobiotic animals (figure 3A). Such animals, although they are different from normal animals and along with their immune and neurological deficiencies, can survive for a very long time, corroborating the possibility of a sterile womb environment.

Over the past few decades, a number of researchers have used sequencing-based methods to detect potential bacteria in placentas or fetuses (figure 3A). For example, Lauder et al conducted one of the earliest case‒control comparisons using 16S rRNA sequencing and shotgun metagenomic sequencing.147 In their studies, no difference was observed between placental samples and negative control samples. Subsequently, several studies also denied the existence of the prenatal microbiome based on 16S rRNA sequencing and suggested that the acquisition of bacteria was more likely to come from the contamination of kits and reagents in the laboratory.148–152 Similarly, in an analysis among spontaneous preterm-delivered, non-spontaneous preterm-delivered and term-delivered placentas, Leon et al found that contamination during delivery contributed greatly to the artificial signals of low-biomass samples.153 Sterpu et al also found that more bacteria could be cultured from vaginally delivered placentas than from caesarean-delivered placentas, enhancing the possibility of delivery-derived contamination.154

Apart from detecting the microbiome using 16S rRNA sequencing, Lager et al modified 18S sequencing but detected no eukaryotic pathogen signals in placental biopsies either in women with adverse pregnancy outcomes or in healthy controls.155 Six different methods were used by Kuperman et al to identify bacterial signals in placenta samples from C-section deliveries.156 Except for a small number of bacterial cells detected by immunohistochemistry, none of the other methods supported the presence of microbial colonisation. Recently, Kennedy et al collected fetal meconium from 20 term fetuses during caesarean section before birth and compared it with first-pass meconium and infant stool157 but did not detect any microbial signal that was distinct from the fetal meconium of negative controls. Through aerobic and anaerobic culture, they found that the frequently isolated Staphylococcus epidermidis from fetal meconium might result from skin contamination.

In utero colonisation hypothesis

Few doubted the sanctity of the dogma that the womb was sterile until the experiment performed by Stout et al, where they found that in nearly one-third of the 195 placenta samples collected using a sterile technique in their study, intracellular bacteria could be histologically visualised in the basal plate.158 Aagaard et al performed a larger microbial analysis of placentas in a 320-subject population. They processed samples and isolated DNA in a strictly controlled, decontaminated and sterile environment and found that the placenta harboured a unique microbial niche characterised by the phyla Firmicutes, Tenericutes, Proteobacteria, Bacteroidetes and Fusobacteria.141 Further 16S rRNA sequencing analysis revealed that this bacterial community resembled the human oral microbiome, indicating a trend of microbial transmission from the maternal oral cavity to the fetus during pregnancy. Subsequently, with standard techniques to ensure sterile and clean conditions during sample collection and DNA isolation, Antony et al demonstrated that in women with excess gestational weight gain, both the microbial composition and metabolic profiles of the placenta were altered, which was associated with preterm birth.159 Although no causality was examined in this study, it has opened the door for research on the presence of intrauterine bacteria during pregnancy in recent years.

Since then, a number of studies have identified bacterial signals in placental villi,160–163 the endometrium,164 the uterine cervix,162 decidual tissue,165 fetal membranes,160 161 the basal plate,160 amniotic fluid161 165 166 and meconium161 166 167 using 16S rRNA sequencing,137 160–167 the traditional histological Warthin-Starry and Gram stain methods,163 fluorescence in situ hybridisation (FISH)167 and clinical culture methodologies (figure 3B).163 167 These studies, including both term and preterm populations, all claimed sample processing under clean and sterile conditions and demonstrated that although bacteria existed as low-abundance, low-biomass and sparse populations, in utero bacterial colonisation did occur during healthy pregnancy.

Among these studies, Gomez-Arango et al confirmed the oral-derived hypothesis in the development of the fetal microbiome in a study that included 37 women with OWOB, in which they further suggested that the gut microbiome was also one of the main contributors to the microbial seeding of the placenta.137 Intriguingly, Liu et al and Parnell et al both found that microbial patterns from the placenta were not altered by delivery mode, suggesting that contamination from delivery made little difference in the identification of bacterial signals.160 161 In addition, Parnell et al employed 16S rRNA sequencing of multiple variable regions and demonstrated that the microbial community in utero exhibited distinct spatial profiles depending on placental location, implying that the sampling sites also greatly affected the detection of bacteria in utero.160

More convincing evidence came from two recent studies conducted by Rackaityte et al and Mishra et al. In the first study, Rackaityte et al examined mid-trimester fetal tissue using both culture-dependent and culture-independent methods.132 According to the results of 16S rRNA sequencing, scanning electron microscopy (SEM), and FISH, they confirmed the existence of a limited microbiome in human fetal meconium, dominated by Micrococcaceae and Lactobacillus. Furthermore, strains of Micrococcus luteus were isolated from fetal meconium samples and cultured in a fetal intestinal-like environment with placental steroid hormones or THP1 human monocyte cells. These isolates exhibited heightened expression of immune cell recruitment and promotion to the tolerogenic environment. In the second study, Mishra et al further revealed the contribution of microbes to the activation of T cells in the fetus.133 They collected fetal tissues in the second trimester and detected consistent microbial signals across different fetal organs (fetal gut, lung, skin and placenta) based on sequencing and clinical culture methodology. Specifically, Lactobacillus and Staphylococcus were identified as the most prevalent bacteria in multiple tissues. In addition to these microbes, the activation of T cells was observed in fetal tissues, implying the presence of antigenic stimuli before birth. In vitro experiments confirmed that bacterial antigens in utero, rather than bacterial contamination from the external environment, induced fetal T-cell expansion and memory activation. These findings strongly suggest the existence of viable microbes in utero and that even such a low-biomass community can educate the fetal immune system, promoting fetal and infant development before and after birth. The two studies provided strong evidence on the presence of the prenatal microbiome; however, contamination in low biomass samples such as placental, uterine, and fetus tissues is still a difficult problem that needs to be addressed.134 Next, we will discuss the problem and propose available solutions.

Problems & solutions

Despite the accumulating evidence on in utero colonisation, many researchers remain opposed to the presence of the prenatal microbiome. One of the most concerning reasons is potential contamination during experimental operations. Potential sources of contamination include the genital tract or perineum during labour and delivery, laboratory-derived bacterial DNA during biopsy collection, sample processing, library preparation and sequencing.148 168 169 As the microbes in utero (if present) are usually under the limit of detection, their low biomass impedes the distinction of authentic biological features from false signals. Although nearly all studies claimed that rigorous aseptic procedures were conducted during sample collection, DNA extraction and downstream processes, the contamination of microbes was rarely ruled out.

To address this challenge, positive and negative controls must be considered in the analysis. The source and type of contamination can be greatly determined by examining the bacterial load in case and control samples. de Goffau et al highlighted the impact of batch effects between case and control samples in the identification of bacterial colonisation.131 For this reason, it must be ensured that the experimental procedure is identical for all samples to avoid batch effects.

In addition, multiple methods, including culture-independent and culture-dependent methods, are highly recommended for use in parallel in the detection of microbial signals between case and control groups. Culture-independent techniques include molecular-based methods, such as 16S rRNA or shotgun metagenomic sequencing, and FISH, histological methods (eg, H&E, Warthin-Starry and Gram stains), and image-based methods, such as SEM. Efficiency is one of the advantages of these methods, while a disadvantage is that they cannot determine the biological activity of bacteria. The culture-dependent technique mainly refers to the clinical culture of bacteria directly from biological samples. This method provides solid evidence for the presence of a viable microbiome in the tissue, although with some possibility of contamination. If both culture-independent and culture-dependent methods yield the same result, there is a good chance that the detected signal is biologically significant. Most importantly, all these methods must be applied very carefully during experiments to avoid the additional introduction of bacterial contamination.

The second problem corresponds to the real source of the intrauterine microbiome detected using various methods. Several studies have indicated that diseases such as endometritis and bacterial vaginosis may result in bacteria accessing the uterus, seriously affecting the development of the fetus.135 148 On the other hand, bacteria in maternal blood and the oral cavity may be transferred into the uterus and fetus through the maternal circulation, which may be caused by serious blood diseases, periodontitis or mild inflammation or infection during pregnancy.136 152 Various abnormal conditions could result in the detection of bacteria in uterine or fetal tissues. Pathogens from other locations of the maternal body are obviously a poor example of bacterial colonisation, which cannot strongly support a stable community in utero. Further evidence is needed to demonstrate that the colonisation of the microbiome truly occurs in utero rather than via transient resident transfer from other organs or tissues under a diseased or inflammatory condition.

Due to the lack of culture-dependent methods in most studies, there is doubt that the bacteria detected in these experiments are likely to be bacterial fragments rather than viable microbes. Some have questioned whether no differences will be observed in prenatal and postnatal development if such fragments or low biomass microbes neither form a stable community nor perform biological functions. While numerous previous studies have failed to prove the functionality of the detected microbiome, recent studies have demonstrated that the presence of live bacteria in utero is not required for immunomodulatory effects, and some bacterial molecules can be transferred to offspring with maternal antibodies.120 126 127 133

Whether bacterial colonisation occurs in utero remains an open question, and the present evidence is not sufficient to completely reject the sterile womb hypothesis. If the uterus is free of a microbiome, how can infants survive and immediately adapt when they leave a completely sterile environment and are exposed to a complex environment filled with bacteria at the time of delivery? If the microbiome truly exists, divides and colonises in utero, what function does it actually have? Where do these bacteria come from and when do they colonise? These are all problems that must be handled with caution.

There is still a long way to go before the debate is completely resolved. On the one hand, new ideas and observations are greatly encouraged to open our minds and reshape our views, pushing the field forwards; at the same time, rigorous and well-designed experiments are highly required to look for more evidence to support or oppose the hypothesis. On the other hand, novel technologies are urgently needed to improve the sensitivity and accuracy in analysing low biomass samples, as well as to improve bioinformatic strategies based on sequencing data.

Delivery mode

At the beginning of this process, the mode of birth plays a critical role in the establishment and development of the microbiome at an early age. Numerous studies have concentrated on the intergenerational transmission of the microbiome during delivery, and most of them have reached a similar conclusion that the microbial community of vaginally delivered neonates exhibits high diversity, dominated by Bacteroides, Bifidobacterium, Parabacteroides and Escherichia. Conversely, infants delivered via caesarean section harboured more species of Klebsiella, Clostridia, Enterobacter, Staphylococcus and some opportunistic pathogens (figure 4A).5 12 170–174

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Microbial transmission during and after delivery. (A) Different transmission patterns between vaginal and C-section deliveries. The microbiome from the maternal gut persists much longer in the infant gut than those from other sources. Microbial divergence between different birth modes decreases with the growth of infants. (B) Transmission during breast feeding. Microbes in maternal breast milk benefit the establishment of the infant gut community. Other bioactive components, such as HMOs, antibodies, immune cells and cytokines, are largely involved in the regulation of the neonatal immune system. (C). Transmission during physical contact. Microbes from different sources contribute to the colonisation of the neonatal microbiome in early life. A wider range of microbial exposure (eg, living on farms) is associated with decreased inflammation and a low risk of autoimmune diseases. HMOs, human milk oligosaccharides.

During delivery, infants are exposed to complex communities with abundant bacteria, such as the maternal vagina, faeces, skin and the hospital environment. To further determine the transmission route at this critical time point, Dominguez-Bello et al performed the first study to explore microbial transmission patterns.170 They collected initial microbiome samples from multiple body sites of mother–infant pairs and observed that the microbiome greatly varied between different birth modes. Vaginally delivered infants harboured more maternal vagina-derived microbes, such as Lactobacillus, Prevotella and Sneathia spp. In contrast, the microbiome of caesarean-delivered infants was characterised by Staphylococcus, Corynebacterium and Propionibacterium spp, which were more frequently found on the maternal skin surface.170

Following this study, several studies further revealed transmission during this process: with partial bacterial transfer from the maternal oral cavity and vagina, the main source of the microbial community in vaginally delivered infants was the mother’s gut, where strains of Bacteroides and Bifidobacterium, as well as Escherichia, were the bacteria most frequently transmitted from mothers to newborns. In contrast, caesarean-delivered neonates acquired their microbiome mainly from the maternal skin and hospital environment.5 172 174–176

Except for differences in initial acquisition in the first few days, persistence patterns of these bacteria also vary between infants delivered vaginally and by caesarean section (figure 4A). Bacteria such as Bifidobacterium longum, Bifidobacterium bifidum and Parabacteroides diastonis, which are enriched in vaginally delivered infants, may come from the maternal gut and have a better fitness for colonisation, while others such as Streptococcus salivarius, Staphylococcus hominis, Staphylococcus massiliensis and Veillonella parvula come from external environments (maternal skin and the hospital environment) and tend to be transient passengers.174 175 One possible explanation is that the bacteria with high abundance that are transmitted from mothers have a better adaptation to the intestinal environment and thus have a stronger competitiveness in the development of the infant microbiome. Despite high variability at the early stage, microbial convergence between vaginal and caesarean deliveries is observed with infant growth (figure 4A).

Feeding

As one of the most important sources providing the initial bacterial community, human breast milk is the optimal food for infants, while the feeding guidelines vary in different regions.177–179 Previous studies have confirmed the benefits of breast feeding for both mothers and infants.11 178 180 181 Through breast feeding, a variety of nutrients that are important in early life are transferred to infants (figure 4B). These bioactive components, including human milk oligosaccharides (HMOs), immune cells, lactoferrin, cytokines, antibodies, antimicrobial proteins and peptides, provide strong protection to infants to effectively avoid the occurrence of multiple diseases, such as asthma, obesity, T2D and allergies, in childhood.11 177 182 Breast milk is the primary source of antibodies and immune cells for newborns who are characterised by an immature immune system.11 These immune cells are crucial for reducing inflammation and promoting immune tolerance in early life.177 Antibodies in breast milk include IgA, IgG and IgM, the proportion of which varies in different lactation periods and populations.11 Among these antibodies, the most abundant is secretory immunoglobulin A (SIgA), which acts as a bridge between microbiome colonisation and immune regulation.120 180 Many studies have demonstrated that maternal SIgA plays a vital role in bacterial adherence, pathogen clearance and intestinal homeostasis to protect against viral infection and enhance microbial transfer from mothers to infants.180 183–185 Additionally, breast feeding is suggested to alter the maternal metabolic process, greatly decreasing the risk of hypertension, hyperlipidaemia and cardiovascular diseases.186 Although many studies have associated breast milk with a decreased risk of allergies in infants, some of them have indicated that the effectiveness of breast milk in allergy prevention is insufficient and is highly related to the timing and type of introduction of solid food.179 187–189 Such conflicting evidence remains to be further explored.

Infants fed exclusively with breast milk harbour increasing abundances of Bifidobacterium, Lactobacillus, Streptococcus and Staphylococcus. 190 Among these, strains of Bifidobacterium were the most prevalent, with B. breve, B. longum, B. dentium, B. infantis and B. pseudocatenulatum accounting for over 70% of the whole community.191–194 The strains of Bifidobacterium, especially B. bifidus and B. longum subsp. Infantis, have a high digestion capacity of HMOs, a group of glycans in human milk that provide a protective effect to intestinal mucosa and promote maturation of the immune system.195 196 In turn, HMOs also boost the increase in Bifidobacterium. 182 196

Breast milk was once considered a sterile fluid; however, increasing evidence has recently demonstrated that breast milk harbours a complex microbial community, most of which belongs to Firmicutes, Proteobacteria and Actinobacteria.182 197 Although several studies have suggested that breast milk exhibits great variability among different populations, Streptococcus and Staphylococcus are identified as the core genera in most lactating women, followed by some other bacteria, such as Lactobacillus, Bifidobacterium, Propionibacterium, Corynebacterium, Enterococcus and Rothia, depending on geographical location, diet, length of gestation and mode of delivery.182 197–199 Interestingly, Streptococcus and Staphylococcus are representative bacteria in the oral mucosa and on the skin surface.200 Large population analysis also confirmed that at the very beginning of life, these two genera, as well as Bifidobacterium, act as pioneer bacteria colonising the gut habitat of most newborns.194 201 Recently, some studies demonstrated that the consumption of breast milk with a reduced microbial richness in the first month was associated with allergy development, implying the great importance of vertical transmission of the microbiome and organised colonisation in early life.202 203 In contrast to direct exclusive breast feeding, some studies have indicated that feeding breast milk collected with breast pumps or formula milk leads to the depletion of some important bacteria, interrupting microbial transmission between mothers and infants.204–206 Specifically, one study reported that sucking on the infant’s pacifier before it was given to the infants was associated with a lower risk of eczema and asthma.207 This evidence implies putative microbiome communication between mothers and infants during lactation, highlighting the importance of direct breast feeding at early ages.

Contact and other sources

Apart from transmission via delivery and feeding, daily contact between neonates and family members, including parents, siblings and pets, also contributes to the postnatal transmission of the microbiome, helping to establish the bacterial communities of infants at an early age (figure 4C). Vertical transmission from mothers via physical contact is still the primary source in daily life. Bacteria such as Parabacteroides distasonis, Alistipes onderdonkii, Bacteroides faecis, Bacteroides caccae and Bacteroides salyersiae from the maternal oral cavity and skin are frequently transferred to infants through kissing, touching and hugging.208–210 One study focused on early microbial seeding and suggested that fathers also play an essential role in the establishment of the neonatal microbiome.8 Distinct from maternal seeding at the start of birth, fathers mainly provide novel strains that colonise infants later in life.8 In addition, infants who grow up with siblings or pets or live on farms exhibit higher diversity and more mature functionality associated with the gut microbiome, where frequent communication of the microbiome, known as horizontal transmission, is observed between cohabitants.5 16 211 212 The hygiene hypothesis213 214 or the alternative ‘old friends’ hypothesis215 216 explains why environmental exposure to a more abundant microbiome contributes to the maturation of the immune system of infants and thus decreases the risk of early diseases, such as asthma and allergies.16

Postnatal development in infants

The first 3 years after birth is the most critical window of early development.217 218 Recently, we conducted a longitudinal study of infants to explore microbial succession at the early stage.201 With the analysis of over ten thousand faecal samples of infants from 17 countries, we observed a considerable but predictable change in the gut microbiome from birth to 3 years of age.201 Such organised microbial assembly in global populations suggests that the development of the microbiome from infancy to adulthood follows a deterministic transition. In addition to microbial succession, the neonatal immune system also develops gradually in this critical period, exhibiting codevelopment between the microbiome and immune system.219 220

Microbiome colonisation and immune responses are closely related. Microbiome-mediated immune responses play a very important role in the maintenance of intestinal integrity. Before birth, cryptopatches and lymphoid tissues start to develop, helping the fetus prepare for exposure to the extrauterine world (figure 5A).221 222 Some studies have indicated that the prenatal microbiome is involved in this essential process.133 221 After delivery, a wide range of bacteria in the neonatal gut stimulates further development of the immune system (figure 5B).16 Paneth cells, a major source of antimicrobial substances in adult tissue, are immature in the neonatal mucosa, making the neonatal intestine very sensitive to external perturbation.221 Instead, cathelicidin-related antimicrobial peptides are expressed during this stage to help prevent bacterial infections (figure 5C).221

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Microbial and immune development during the first 3 years of life. (A) Before birth, the ontogenesis of the immune system begins with stimulation by maternal factors. A few recent studies have indicated that early colonisation occurs during this period. Nevertheless, considering the low biomass in placental or uterine samples, whether the detected signal is associated with contamination needs further exploration. (B) After delivery, the gut microbiome transitions from a facultative anaerobic community to an obligate anaerobic community. Correspondingly, the neonatal immune system switches from a tolerogenic response to an antimicrobial response. (C) Bifidobacterium becomes the most abundant taxon during lactation, especially in the guts of infants fed maternal breast milk. Maternal breast milk also stimulates the immune system to mature rapidly and protects the neonatal intestinal mucosa from colonisation by pathogens. (D) At the time of weaning, Bacteroidetes and Firmicutes dominate the gut ecosystem, and bacterial divergence between different individuals starts to decrease. Weaning reactions occur and induce variations in immune cells with the expansion of the intestinal microbiome. AMPs, antimicrobial peptides; CRAMP, cathelicidin-related antimicrobial peptide; PAMPs, pathogen-associated molecular patterns; Tregs, regulatory T cells.

Various postnatal factors determine immune development in early life. Caesarean delivery is associated with asthma and many immune-related disorders in childhood.223 Frequently accompanied by preterm birth and the use of antibiotics at birth, infants born via caesarean delivery not only harbour a disordered microbial community but also have a delayed immune system,16 224 with reduced microbial diversity and significantly lower levels of the Th1-associated chemokines CXCL10 and CXCL11 in blood.225 Especially for children with a microbial community that has not been restored to normal at the age of 1 year, the level of immune mediators, such as tumour necrosis factor alpha, interleukin 4 (IL-4), IL-3 or IL-1β, is much lower.226 227 Rodent models also suggested that caesarean delivery might lead to immune deficiency in progeny, which could be partially restored with prebiotics in the postnatal period.228 229

In the first few days after birth, facultative anaerobic bacteria, as pioneer colonisers, dominate the infant gut, where species of Enterobacteriaceae, Staphylococcus and Streptococcus exhibit the highest abundance (figure 5B).194 201 230 With the reduction of oxygen, facultative anaerobic microbes are replaced by a group of obligate anaerobic bacteria. Among these microbes, strains of Bifidobacterium, which are mainly involved in the metabolism of HMOs, dominate the ecological niche with the start of breast feeding (figure 5B).194 Infants fed breast milk have a decreased risk of necrotising enterocolitis, which is a devastating and the most common disease in preterm infants, causing high morbidity and mortality.11 197

In contrast, infants who are fed cow’s milk formula or soy formula harbour more bacteria belonging to Bacteroidetes and Firmicutes, such as Bacteroides, Clostridium and Ruminococcus, showing a higher alpha diversity and more mature microbial pattern compared with exclusively breastfed infants.197 206 As reduced microbial diversity is associated with the development of allergies and atopic eczema,231–233 the reason why this difference occurs remains unknown, and the protective effect of breast feeding against autoimmune diseases requires further research.187

Most infants were weaned between the sixth month and the eighteenth month of life, when Bifidobacterium significantly decreased and Bacteroides began to thrive (figure 5C,D).194 195 201 The period spanning birth to weaning is considered to be a ‘window of opportunity’, during which host-microbe crosstalk frequently occurs, leading to the development of a balanced immune system and preventing pathological imprinting later in life.226 234 235 At the time of weaning, the composition of the gut microbiome as well as the population of immune cells are greatly altered, inducing a vigorous immune response called the ‘weaning reaction’ and promoting the maturation of the immune system.234

As infants grow, the gut microbiome enters the transitional stage,194 characterised by a high abundance of Bacteroides, Lachnoclostridium, Faecalibacterium, Lachnospiraceae and Ruminococcus (figure 5D).230 Following this, around the 18th month of life, Prevotella, the dominant taxon in one of the most frequent enterotypes of adults, emerges and continues to increase.236 By 3 years of age, infants in distinct developmental stages harbour microbial communities dominated by different bacteria, leading to significantly stratified patterns in global populations.201 230 At this phase, the gut microbiome of infants transforms to an adult-like microbiome and remains stable. Along with maturation of the microbiome, functionalities associated with pathways of complex polysaccharide metabolism are significantly upregulated to better adapt to adult-like lifestyles.201 217