Microbial–host molecular exchange and its functional consequences in early mammalian life

See allHide authors and affiliations

Science  08 May 2020:
Vol. 368, Issue 6491, pp. 604-607
DOI: 10.1126/science.aba0478


Molecules from symbiotic microorganisms pervasively infiltrate almost every organ system of a mammalian host, marking the initiation of microbial–host mutualism in utero, long before the newborn acquires its own microbiota. Starting from in utero development, when maternal microbial molecules can penetrate the placental barrier, we follow the different phases of adaptation through the life events of birth, lactation, and weaning, as the young mammal adapts to the microbes that colonize its body surfaces. The vulnerability of early-life mammals is mitigated by maternal detoxification and excretion mechanisms, the protective effects of maternal milk, and modulation of neonatal receptor systems. Host adaptations to microbial exposure during specific developmental windows are critical to ensure organ function for development, growth, and immunity.

Living organisms are constrained by the thermodynamic boundaries of obtaining and using sufficient energy within the available environment to construct their cellular and noncellular biomass during mammalian fetal and neonatal development. The consumption of plant, animal, and other xenobiotic material that is partly metabolized by microorganisms in the maternal and neonatal gastrointestinal tract improves energy harvesting for growth but exposes the developing mammal to a wide range of chemicals. Although it is possible for mammals to live in the absence of a microbiota, analytic techniques using isotopically labeled intestinal bacteria have revealed that normally even systemic organ systems and potentially the fetus are promiscuously bathed by molecules synthesized either by microbes on mucous membranes, or by dietary intake, but which cannot be generated by host metabolism itself (1, 2). Here, we examine the potential impact of exposure to microbial constituents and other xenobiotics in early life, from fetal development to the early postnatal period.

Maternal exposure to microbial metabolites and xenobiotics

Mice may be bred in aseptic “germ-free” conditions in isolators with normal fecundity, development, and life span provided that their food is fortified with micronutrients. Vitamin K, for example, is usually produced by bacteria and must be given as a dietary supplement to germ-free mice for normal blood clotting (3). The microbiota triggers extensive adaptations in every organ system (3), either through sensing of the live microbes at body surfaces or through signaling from microbial metabolites that reach host tissues. Colonized animals are less susceptible to opportunistic infections through the competitive protective effect of the microbiota. Maturation that drives the adaptive and innate immune systems allows scalable responses to later pathogen challenges (4). In addition to the functional adaptations that occur in mucous membranes exposed to the microbial biomass, there are also extensive long-range effects on the neural, cardiovascular, hepatic, endocrine, adipose, and skeletal systems.

Although many of these pervasive adaptations can be recovered by colonizing an adult germ-free animal, there are also precise developmental time windows that require microbial colonization for normal immune development and the assembly of a healthy microbiota (5). The question is whether such windows are exclusively postnatal, as the young mammal acquires its own microbiota, or whether the molecular diaspora from the mother’s microbiota is also important for antenatal development.

Maternal nutrition is clearly an important factor for development in early life. The energy requirements of the fetus and neonate benefit from the mother’s microbiota, which optimizes her overall nutritional state by providing vitamins and essential amino acids, detoxifying xenobiotics, and breaking down otherwise indigestible foods (6). Bacterial folate positively influences embryonic and fetal development, and short-chain fatty acid microbial metabolites are important to sustain maternal intestinal barrier integrity (6).

Effects of the microbiota are mainly studied by comparing germ-free and colonized mice. To understand the specific effects of a maternal microbiota alone, transitory colonization systems have been used, so the mother is only colonized for a short time in pregnancy but delivers germ-free pups (Fig. 1). This approach has been combined with nonradioactive isotope labeling of the transitory bacteria to trace the penetration of microbial molecules into the mother and her offspring. The model shows that maternal microbial metabolites drive the development of her pups’ innate immune system and maturation of the intestinal epithelium (7). One of the receptors for the chemical signals involved is the aryl hydrocarbon receptor (AhR), which is essential for normal immune function (8).

Fig. 1 Experimental model of reversible gestational colonization.

The auxotrophic Escherichia coli HA107 strain, which is deficient in the synthesis of two bacterial-specific amino acids, meso-diaminopimelic acid and d-alanine, can be used to study the effect of the maternal microbiota on offspring development in the absence of an endogenous microbiota in the offspring. This strain is unable to replicate within the murine germ-free intestine where meso-diaminopimelic acid and d-alanine are absent. If germ-free pregnant dams are treated with HA107, they are transiently colonized but return to germ-free status before giving birth and will thus deliver germ-free pups. This experimental setup has been used to study the effect of maternal microbiota only during pregnancy on developmental processes in the offspring (7). i.g., intragastric; c.f.u., colony-forming units; PBS, phosphate-buffered saline.


There are poorly understood differences in how different microbes metabolize dietary xenobiotics, drugs, or environmental contaminants (9), because microbial genes are mainly annotated by inference from sequence homologies, or their function is unknown. There are also fundamental differences between how eukaryotes and prokaryotes handle xenobiotic (bio)chemicals. The mammalian host generally detoxifies xenobiotic metabolites through the addition of polar functional groups (-OH, epoxide, -SH, or -NH2), followed by the addition of additional polar head groups (glucuronyl, acetyl, methyl, and sulfonyl), which make the resultant polar molecule susceptible to renal elimination. By contrast, microbial metabolism has evolved to break molecules for carbon sources and/or dispose of reducing equivalents using hydrolases, lyases, reductases, group transfer, or radical chemistry (9, 10). An important unmet need is to understand how far exposure to natural xenobiotics, whether sourced directly from the diet or taken in artificially high amounts as food supplements, may show idiosyncratic effects on the fetus or neonate after metabolism by the mother’s intestinal microbes. This is ethically difficult, as animal models do not precisely mirror human development mechanisms or the human microbiota composition.

The maternal–fetal interface

Placentation and antenatal hematopoiesis

The basis for antenatal exposure of the developing fetus to circulating microbial metabolites and xenobiotics is the placental interface between the maternal and fetal bloodstream (6). Mice and humans have a hemochorial placenta, in which the fetal trophoblast invades the uterine tissue and the endothelium of the maternal blood vessels. A vascular labyrinth starts to form as the fetal umbilical arteries invade the maternal decidua on embryonic day 12.5 (E12.5) in mice (within the first trimester in humans), progressing to an efficient interface between the maternal and fetal vascular systems for respiratory gas exchange, nutrient transfer, excretion, and some detoxification. This provides only a limited barrier to molecular transfer, and non-ionized molecules of relative molecular mass Mr < 500 reach the fetal circulation by passive diffusion. Fetal organ systems and immunity are developing in parallel. In mice, primitive hematopoiesis begins in the yolk sac at E7, and hematopoietic stem cells are found in the fetal liver and thymus starting from E10.5 (6).

Placental and maternal handling of xenobiotics

The potential risks of the limited placental barrier became apparent through the thalidomide disaster, when a synthesized xenobiotic taken in early pregnancy caused nonhereditary phocomelia (grossly underdeveloped or missing limbs) in babies (6). To regulate key compound classes, the placenta has a series of transport proteins, which help to coordinate maternofetal nutrient, excretory, and xenobiotic exchange (11). Multidrug resistance protein 1 (MDR1, P-glycoprotein, ABCB1) is an adenosine 5′-triphosphate (ATP)–dependent transporter of xenobiotics present on trophoblast cells, hepatocytes, intestinal epithelial cells, and renal tubular cells. Overall, this transporter helps to protect the fetus from xenobiotics as shown by the occurrence of cleft palate deformities in globally MDR1-deficient mouse fetuses where the dam was administered a polycyclic antihelminth analog (12). Two additional transporters (ABCG5 and ABCG8) expressed in the placenta, liver, and intestine are known to limit the uptake of plant-derived sterols, which are themselves subject to microbiota metabolism. Female mice that are globally Abcg5 deficient exhibit cardiomyopathy, thrombocytopenia, and infertility (13). How far selective placental expression of any of these transporters contributes to regulating fetal exposure to maternal microbial metabolites remains unclear.

Environmental toxins, such as dioxins and dioxin-like compounds, are metabolized through the cytochrome P450 superfamily of enzymes, such as Cyp1a1, which are induced through the AhR. High AhR expression is found in the placenta, the liver, and in mucous membranes. Although small amounts of AhR ligands derived from food and the microbiota are beneficial to fetal development and postnatal immune function, toxic amounts are restricted through overall AhR-dependent activation of Cyp1a1 in intestinal, hepatic, and placental tissues (14). Limiting fetal xenobiotic exposure is therefore generally a function of a triple layer consisting of maternal intestine, liver, and placenta. Failure to protect the fetus or neonate from exposure to maternal microbial metabolites, as well as other ingested chemicals, may have lasting developmental consequences and predispose these children to metabolic and immunological diseases.

The microbiota potentially affects the biochemical environment of the fetus and neonate in a variety of ways

Effects of microbiota metabolites can be categorized according to whether the chemicals are synthesized endogenously by the microbiota or whether they are secondary metabolites of compounds that are taken as food, pharmaceuticals, or as environmental contaminants. Each of these categories can alter the composition and biomass of the microbiota, as well as generate different portfolios of xenobiotic chemical exposures according to the metabolic capacity of various taxa present [reviewed in (15)]. The general effects of dysbiosis on adult (maternal) intestinal, metabolic, and immune functions have been considered in numerous reviews.

Endogenous microbial compounds

Many endogenous microbial compounds are recognized by innate pattern recognition receptors. Comparisons of adult germ-free and colonized mice indicate steady-state penetration of host tissues by Toll-like receptor (TLR) ligands [e.g., lipopolysaccharide (LPS) or flagellin] and NOD ligands, which have been linked to important maturation processes in the host immune system (16, 17). Although we believe that the developing fetus and the placenta are sterile (18) (box 1), stable isotope–labeling studies show that there is rather promiscuous penetration of most classes of endogenous microbial compounds into the adult host, especially from bacteria in the lower small intestine (2), and these are likely to reach the placenta.

Box 1

Is the developing fetus sterile?

The healthy fetus is enclosed in utero by the amniotic membrane and receives blood from the placenta. It has long been thought to be completely sterile, but this notion has been challenged by recent studies that reported a very small microbial biomass in placental tissue, cord blood, or meconium. However, methodological challenges, contradictory results, and our current immunological understanding cast doubts on the interpretation of these findings (18). The detection of very small amounts of bacterial DNA may be confounded by laboratory or reagent contaminants. Furthermore, viable bacteria are transiently found in the bloodstream of healthy neonates following minor trauma common during birth. Additionally, bacterial profiles described vary substantially, with taxa from the oral or the skin microbiota that are potential contaminants. Because placental tissue has no lumen, colonizing bacteria would be subject to immune elimination. Finally, cesarean section and sterile fostering of fully colonized experimental animals generate germ-free neonates. Pathogenic bacteria that successfully resist antimicrobial destruction are expected to provoke an inflammatory reaction predisposing to premature birth.

Central nervous system alterations resulting from antenatal LPS exposure have long-term behavioral consequences in rodent models. However, doses have been used that simulate intrauterine infections, leaving open the question of whether there are also effects from the steady-state penetration of LPS in the healthy pregnant female. Microglia are extensively branched with longer dendrites in adult germ-free mice, whereas microglial immune activation is greater in colonized animals and those treated with LPS during postnatal compared to fetal life (19). Therefore, the barriers of the intestine and the placenta coupled with early-life insensitivity to LPS signaling presumably provide protection against steady-state antenatal exposure. Damage to the intestinal barrier (through alcohol intake or parasitic infection) can increase LPS exposure at the maternal–fetal interface (20). LPS and glutamyl-diaminopimelic acid (binding TLR4 and NOD1, respectively) induce inflammation at the maternal–fetal interface and can be a risk factor for preterm birth (21).

Diet-derived xenobiotics that are metabolized by commensal microbiota

Extensive literature is available on various dietary components and how these can potentially be metabolized by the maternal microbiota possibly with effects on early-life development (table S1). These compounds are mainly plant-synthesized polycyclics (including flavones, isoflavones, and anthraquinones), terpenes, and polyols. Although there is little or no direct epidemiologic evidence either for or against relevant effects on a human fetus, some compounds are consumed in high amounts—for example, with the intention of influencing the gender of human babies.

In other cases, there is clearer evidence for clinically important effects. Retinoids in the maternal diet, whose availability in the intestine is directly regulated by microbial taxa, such as Clostridia (22), can affect the number of lymphoid tissue inducer cells in the fetus and thus development of secondary lymphoid organs in the offspring (23), as well as the induction of oral tolerance in models of allergy (24).

Microbes found in the large intestine, such as Bacteroides species, ferment dietary fibers to short-chain fatty acids (SCFAs). These contribute to host immune maturation, either by activating G protein–coupled receptors (GPCRs) on the surface of immune cells, or through inhibition of lysine deacetylases. In a murine model of asthma, feeding female pregnant mice a fiber-rich diet limited inflammatory airway responses and the development of asthma in their offspring through the production of SCFAs (25). Similarly, maternal dietary fiber fermentation during pregnancy and lactation can induce the differentiation of regulatory T cells in the offspring (26).

Bile acids have a distinctive position in the portfolio of maternal microbial metabolites that affect the fetus. Bile acid pools are generally increased through maternal hepatic synthesis in late pregnancy, and secondary bile acids are formed by microbial metabolism in the maternal gastrointestinal tract. In addition to lipid solubilization in the postnatal intestine, bile acids are known to exert metabolic and growth effects through the GPCR TGR5 and the farnesyl X nuclear receptor (FXR). The fetus is exposed to bile acids both from its own synthesis and from maternal transfer, although renal excretion depends on the mother. Solute transport proteins of the SLC21, SCL22, and ABCG2 classes are expressed in the placenta and may regulate fetal exposure to bile acid metabolites (27).

Nuclear receptors, including the AhR, FXR, pregnane X receptor, constitutive androstane receptor, and vitamin D receptor, can bind diet-derived ligands and are likely to detect different maternal microbial compound classes (28). They can direct transcriptional activity through epigenetic mechanisms (histone modifications or DNA methylation) and may be of special importance during in utero development, which constitutes the most active period for epigenetic DNA imprinting in a mammal’s lifetime.

Birth and postnatal effects

Postnatal colonization with microbes and lactation

The neonate’s body surfaces are colonized at birth, exposing the offspring to microbes and their molecular diaspora without maternal intestinal or placental barriers. Maternal microbial molecules still reach the neonate through breast milk, although direct microbial exposure now becomes far more important.

Breast milk shapes the unstable early-life intestinal microbiota through secretory antibodies (29), milk oligosaccharides (30), or milk proteins, including lactalbumin and lactoferrin. Because the antibody repertoire of maternal milk is shaped by the mother’s own microbiota and her previous exposure to pathogens, breast-feeding is an efficient way to transfer mucosal and systemic immune memory from mother to offspring. Once the offspring starts to consume solid food, these protective effects of milk disappear, leaving endogenous microbes to stimulate a “weaning reaction” in a critical developmental window in the young mammal (31).

Postnatal colonization and the innate immune system

As in adults, stromal and immune cells of the fetus and neonate express a series of innate immune receptors and antimicrobial effector molecules, allowing them to mount potent and protective immune responses upon infection. However, neonates are also susceptible to inappropriate inflammation upon encounter of microbial stimuli from commensal bacteria, as in the pathogenesis of necrotizing enterocolitis (NEC), a devastating immune-mediated disease in preterm neonates (32).

Different mechanisms modulate the innate immune system to control the fetal–postnatal transition. Most of these perinatal changes appear to be developmentally regulated and largely independent of the microbiota, probably reflecting the unreliable presence of microbial stimuli early after birth (33). Age-dependent differences exist for the expression of individual receptors and the anatomical distribution of antimicrobial peptides in mice and humans, but mechanistic insight into the functional role of the changes remains limited (34, 35). For example, decreased prenatal expression of TLR4 by the human intestinal epithelium or repressed TLR4 signal transduction in neonatal murine epithelium may help to prevent inflammation upon early postnatal colonization (3638). By contrast, enhanced expression of the flagellin receptor TLR5 by the murine neonatal epithelium contributes to the selection of a beneficial gut microbiota (34). Similarly, human blood monocytes undergo postnatal reprogramming through stimulation with the endogenous TLR4 ligands S100A8 and S100A9 to avoid hyperinflammation and promote immune homeostasis (39). Finally, maternally derived factors in amniotic fluid and breast milk modulate innate immune recognition and mucosal translocation of microbial stimuli to restrict their proinflammatory activity during the immediate postnatal period (31, 40). For example, breast milk–derived secretory immunoglobulin A with affinity to enteric bacteria increases bacterial diversity and protects against the development of NEC in preterm neonates (41).

Thus, neonatal innate immunity is not simply less developed or immature. Rather, it is highly adapted and finely tuned to facilitate the fetal–postnatal transition of rapidly increasing microbial biomass and the development of long-term host microbial mutualism.

Postnatal colonization and adaptive immunity

The full maturation of the adaptive immune system occurs predominantly at weaning, when the young host is exposed to new antigens through higher intestinal microbial and food antigen loads. The intestinal mucosa acquires antigen-experienced T cells and activated plasma cells in a microbiota-dependent process (3). Antenatal B and T cell development in the fetal liver shifts to the bone marrow and thymus, respectively, and naïve B and T cells migrate into the secondary lymphoid tissues.

The extent to which the preweaning microbiota contributes to the trajectory of B and T cell repertoire development between early life and adulthood is not yet fully understood, although premature diversification appears to be a disadvantage for later immune responses that depend on natural antibodies and can potentially bias the adult B cell repertoire (42). Maternal milk immunoglobulins, themselves shaped by the composition of the maternal microbiota as well as neonatal T regulatory cells, delay the onset of secretory antibody production (43) and mucosal T helper cell maturation (44, 45) in the offspring.

The process of development of some lymphocytes requires the presence of microbiota during a critical time window prior to weaning. Mice that are germ free until weaning have increased serum IgE concentrations and excessive intestinal mucosal natural killer T cells (5). Mucosal associated innate T (MAIT) cells, absent in germ-free mice, only efficiently seed tissues in response to microbiota-derived riboflavins in the first few weeks of life (46). Mucosal regulatory T cells require microbiota before weaning, limiting later susceptibility to colitis or allergic airway inflammation (5, 47).


In this Review, we have considered the impact of the microbiota on the early-life mammal. In fetal life, this comes mainly from penetration of molecules synthesized by maternal intestinal microbes or microbial metabolism of food substances (Fig. 2). Our knowledge of these effects at physiological levels of metabolite penetration remains very limited, and the epigenetic and signaling mechanisms involved have primarily been studied thus far in the context of toxicology. After birth, adaptations occur mainly to contain the impact of the rapidly increasing endogenous neonatal microbial biomass and its molecular diaspora. It is clear that there are age-dependent modulations of signaling to innate receptor ligands and that maternal antibodies can shield the neonatal immune system from premature repertoire diversification and shape the composition of the early-life microbiota.

Fig. 2 Host barriers for exposure of the developing fetus and neonate to microbiota-derived metabolites.

(A) Schematic view of the extent of exposure of the developing fetus or offspring to metabolites originating from or metabolized by the maternal microbiota in comparison to the biomass of microbes colonizing the offspring itself after birth. (B) A triple barrier consisting of the intestinal epithelial lining, the detoxifying liver, and the placental barrier partially protects the developing fetus from exposure to maternal bacterial and dietary metabolites. After birth, exposure to those metabolites continues through breast-feeding, further shaping the offspring’s endogenous microbiota and immunity.


Supplementary Materials

References and Notes

Acknowledgments: Funding: This work was supported by funding from ERC (HHMM Neonates) to A.J.M., Collaborative Research Center CRC1382 to M.W.H., and Peter Hans Hofschneider Professorship to S.C.G.-V. Competing interests: The authors have no competing interests.

Stay Connected to Science

Navigate This Article