Towards microbiome-informed strategies for predicting and preventing pregnancy complications

Despite substantial and credible evidence demonstrating the absence of the microbiome in the uterine environment, several studies, which in most cases are devoid of appropriate controls, continue to assert the presence of a microbiome in these otherwise sterile sites ( de Goffau et al. 2019 , Sterpu et al. 2021, Granser & Foessleitner 2024 , Liu et al. 2024 , Olaniyi et al. 2024 , Uddipto et al. 2024 ). When analysing sites with minimal microbial biomass, bacterial DNA from environmental contaminants is amplified during PCR and detected in sequencing, resulting in misleading interpretations. Using existing methods, we and others have shown that the vast majority of signals are due to sample contamination from maternal microbiome during delivery or the tissue-processing reagents, including the DNA extraction and sequencing reagents, and there are no culturable microbiome in these sites ( de Goffau et al. 2019 , Sterpu et al. 2021 ). Despite this, most recent studies emphasising the presence of placenta microbiome were devoid of extraction reagent controls ( Li et al. 2024 b , Liu et al. 2024 , Olaniyi et al. 2024 , Saadaoui et al. 2024 ). Pathogens and pathobionts, such as group B Streptococcus (GBS), may traverse the utero–placental–fetal barrier in cases of chorioamnionitis ( Musilova et al. 2016 , Lukanović et al. 2023 , Banchi et al. 2025 ). However, these are typically associated with ascending infections and must not be confused for the presence of culturable microbes with a taxonomic function ( de Goffau et al. 2019 ). Moreover, microbial extracellular vesicles and nucleic acids can be detected within the utero–placental–fetal unit, but these do not constitute viable or culturable microbial communities. Thus, host–microbe interactions are likely mediated by molecules that enter the systemic circulation and influence maternal physiological processes.

This work was supported by a Loke Centre for Trophoblast Research Next Generation Fellowship to D-WPE. C-JDS received funding from NIHR Cambridge Biomedical Research Centre (NIHR203312). RGL was funded by the National Institutes of Health via the NIH Oxford–Cambridge Scholars Program. BJA was funded by the Queen’s College, University of Cambridge Alexander Crummell Scholarship. The views expressed are those of the authors and not necessarily those of the NHS, the NIH, the NIHR, or the Department of Health and Social Care.

Pregnancy involves systemic changes in energy metabolism, immune function, and endocrine activity to meet maternal–fetal needs and prevent fetal rejection. These adaptations are well reviewed elsewhere and will not be discussed here ( Chang & Streitman 2012 , Napso et al. 2018 ). Nevertheless, hormones produced by both the mother and the placenta are critical to these systemic changes ( Napso et al. 2018 ). During ovulation, progesterone produced by the corpus luteum prepares and maintains the endometrium in a receptive, secretory state to facilitate implantation and support early pregnancy ( Günther et al. 2023 ). Within two weeks of fertilisation, trophectoderm cells (the precursors of the placenta/trophoblast) begin secreting human chorionic gonadotropin (hCG), which peaks by week 12 and then declines ( Guibourdenche et al. 2010 , Korevaar et al. 2015 ). Optimal trophoblast development depends on the fine balance between the renewal of the cytotrophoblast stem cells and their differentiation into the invasive extravillous trophoblast (EVT) and the multinucleated syncytiotrophoblast (STB) epithelial cells. During early fertilisation and implantation, the uterine environment is pro-inflammatory, driven by decidual NK cells, macrophages, and Tregs that release angiogenic factors and cytokines ( Moffett & Shreeve 2023 ). These influence and regulate EVT invasion of the uterine mucosal lining ( Aplin et al. 2020 ). EVTs subsequently erode the vascular smooth muscles and elastic lamina, remodelling the uterine vasculature from high-resistance arterioles into low-resistance, high-flow vessels ( Crespo et al. 2016 , Burton et al. 2021 ). This transformation is critical for ensuring stable and low velocity blood flow to the uterus and placenta ( Burton et al. 2021 ). Thus, normal uterine spiral artery remodelling hinges on complex interactions between EVTs, immune cells, and decidual and uterine vascular cells ( Pollheimer et al. 2018 , Peng et al. 2023 ). The STB cell layer acts as an interface between the mother and the fetus, mediating nutrient and gaseous exchange, hormonal secretion, and metabolite production, eliminating toxins and pathogens ( Malassiné & Cronier 2002 ). An imbalance in nutrients, immune cell signalling, hormones, toxins, and pathogens can profoundly affect these developmental processes. Inadequate spiral artery remodelling leads to high maternal blood pressure in pregnancy (preeclampsia (PE)), a condition that can only be alleviated by the removal of the placenta ( Burton et al. 2019 ). Altered STB development can result in placental insufficiency, leading to miscarriage, stillbirth, fetal growth restriction (FGR), preterm birth (PTB), and increased risk of infections ( Megli & Coyne 2022 ). Infections also increase the risk of pregnancy complications and congenital anomalies, and poor fetal growth increases the risk of chronic diseases in adulthood ( Gluckman et al. 2016 , Megli & Coyne 2022 ). In addition, pregnancy complications predispose women to cardiometabolic diseases and mental disorders ( McNestry et al. 2023 ). The decline in hCG coincides with increased production of progesterone and oestrogens (oestradiol and oestrone) mainly from the corpus luteum, but also from the ovaries and adrenal cortex in early pregnancy and predominantly from the placenta in late pregnancy ( Schock et al. 2016 ). These hormones suppress the secretion of follicle-stimulating hormone and luteinising hormone by the pituitary gland and regulate the transformation of endometrial stromal cells into decidual cells. Decidualisation is key for the production of pregnancy-supporting molecules, such as prolactin, glycogen, lipids, and IGFB-1 ( Okada et al. 2018 ). Progesterone and related molecules create a uterine environment favourable for implantation by promoting anti-inflammatory cytokines (IL-4 and IL-10) and suppressing pro-inflammatory signals. This supports implantation and prevents contractions. Late in pregnancy, the uterine environment shifts to a pro-inflammatory state with cytokines (IL-1 and TNF-α), phospholipids, and prostaglandins, triggering labour and fetal expulsion ( Palomba & Daolio 2018 ). Timely changes in these hormones regulate maternal adaptations, including increased β-cell size, resulting in enhanced β-cell glucose uptake and insulin secretion. Consequently, late pregnancy is punctuated by increased pro-inflammatory cytokines, hyperinsulinaemia, and hyperglycaemia ( Kumar & Magon 2012 ). Alterations in the levels of pregnancy-related hormones, immune function, and glucose metabolism are associated with adverse pregnancy outcomes ( Williams et al. 2000 , Jarmund et al. 2021 ). Microbes proficiently integrate various signals to sense and adapt to environmental cues, such as nutrients, hormones, pH, and pathogens.

Maternal physiological adaptations during pregnancy occur alongside large, metabolically and signalling-active commensal microbial communities that dynamically respond to environmental cues. The microbiome has an immense capacity to metabolise dietary components, such as complex carbohydrates, amino acids, l -carnitine, and choline, into bioactive metabolites, including SCFAs, indole derivatives, vitamins, and TMA or trimethylamine-N-oxide (TMAO) ( Yisireyili et al. 2017 , Negatu et al. 2020 , Roy et al. 2020 , Day-Walsh et al. 2021 , Qin et al. 2025 ). In response to nutrient availability, drugs, and host-secreted molecules (e.g. mucins, hormones, and cytokines), the microbiome also produces signalling molecules such as nucleotide second messengers: cyclic adenosine monophosphate (cAMP), cyclic guanosine monophosphate (cGMP), guanosine pentaphosphate ((p)ppGpp), cyclic di-GMP (c-di-GMP), and c-di-adenosine monophosphate (c-di-AMP) ( Burdette et al. 2011 ). Microbes detect changes in population size or invading species through quorum sensing (QS) molecules. As cell density rises, QS molecules accumulate until a threshold triggers coordinated expression of genes controlling processes such as dispersal and virulence ( Tao et al. 2021 , Mirtaleb et al. 2025 ). Through second messengers and QS molecules, microbiomes and their metabolites can regulate the production of substrates such as serotonin (SER) and glucose by the host ( Legan et al. 2022 , Oteng & Liu 2023 ). In addition, microbial metabolism of dietary substrates can redirect the production of microbial metabolites; for example, complex carbohydrates shift microbial metabolism towards beneficial indole derivatives indole-3-lactic acid (ILA) and indole-3-propionic acid (IPA) from indole production, which is subsequently metabolised by the host to the uraemic toxin indoxyl sulphate (IS) ( Sinha et al. 2024 ). Thus, the microbiome can directly or indirectly influence the host through several mechanisms, some of which include the following: Alterations in microbial composition and functional capacity that enhance nutrient utilisation may reduce the availability of essential nutrients required in pregnancy. Bioactive molecules derived from the microbiome can exert beneficial or harmful effects on the mother, placenta, and fetus. Microbial QS molecules can sense shifts in microbial population dynamics that impact biofilm formation, virulence, and epithelial barrier integrity, while altering functional redundancy and resistance to infection. This may impact the dynamic changes in the microbiome during pregnancy, microbial colonisation in infants, the risk of infection, and responses of the host to interventions such as probiotics ( Tao et al. 2021 ). The production of host-derived molecules or microbial-derived molecules that affect immune, metabolic, and neuroendocrine signalling can be regulated. Alterations in microbial composition and functional capacity that enhance nutrient utilisation may reduce the availability of essential nutrients required in pregnancy. Bioactive molecules derived from the microbiome can exert beneficial or harmful effects on the mother, placenta, and fetus. Microbial QS molecules can sense shifts in microbial population dynamics that impact biofilm formation, virulence, and epithelial barrier integrity, while altering functional redundancy and resistance to infection. This may impact the dynamic changes in the microbiome during pregnancy, microbial colonisation in infants, the risk of infection, and responses of the host to interventions such as probiotics ( Tao et al. 2021 ). The production of host-derived molecules or microbial-derived molecules that affect immune, metabolic, and neuroendocrine signalling can be regulated. Here, we examine SCFAs, tryptophan-derived metabolites, QS molecules, and second messengers, as well as choline, carnitine, and trimethylamines, alongside microbial-associated molecular patterns and extracellular vesicles, as potential mediators of host physiology during pregnancy. SCFAs are aliphatic carboxylic acids consisting of a carbon chain composed of less than six carbons ( den Besten et al. 2013 ). They are produced primarily by the fermentation of complex carbohydrates (including fibre and resistant starch) in the large intestine by the gut microbiome ( Fig. 3 ). However, they may also directly derive from dietary fermented and polyphenol-rich foods ( Silva et al. 2020 , Sankarganesh et al. 2025 ). Acetate, propionate, and butyrate are the most abundant SCFAs, constituting 90–95% of total SCFAs ( O’Riordan et al. 2022 ). However, other SCFAs, including lactate isomers, valerates, and branched-chain SCFAs, are present in smaller quantities ( Fusco et al. 2023 ). Homeostatic regulation of SCFA levels is essential, as they modulate gut microbiota composition, systemic metabolic flux, and immune signalling and are implicated in the pathogenesis of adverse pregnancy outcomes, including PE and GDM and impaired fetal programming ( Altemani et al. 2021 , Wang et al. 2022 , Husso et al. 2023 , Sankarganesh et al. 2025 ) ( Fig. 3 ). Pathways involved in microbial short-chain fatty acid production with key enzymes shown. Following hydrolysis of polysaccharides into monosaccharides, these substrates are converted to phosphoenolpyruvate (PEP) via glycolytic pathways (Embden–Meyerhof–Parnas, pentose phosphate, or Entner–Doudoroff pathways), which then undergoes further metabolism through NADH-dependent fermentation, hydrogen metabolism with interspecies transfer, or primitive electron transport chains to generate SCFAs as end products. Cross-feeding of intermediary metabolites between different bacterial species plays a critical role in SCFA production, with specific bacterial pathways determining the ratio of individual SCFAs produced. (A) Propanediol pathway: pduC/D/E, propanediol dehydratase large/medium/small subunits; pduP, propionaldehyde dehydrogenase; PCT, propionyl-CoA transferase. (B) Succinate pathway: SUC, succinyl-CoA synthetase; MUT, methylmalonyl-CoA mutase; MCEE, methylmalonyl-CoA/ethylmalonyl-CoA epimerase; MMDA, methylmalonyl-CoA decarboxylase α-subunit; PCT, propionyl-CoA transferase. (C) Acrylate pathway: key enzymes: PCT, propionyl-CoA transferase; LCD, lactyl-CoA dehydratase; ACR, acrylyl-CoA reductase. (D) Wood–Ljungdahl pathway: key enzymes: FDH, formate dehydrogenase; FTHFS, formyl-THF synthetase; FTC, formyl-THF cyclohydrolase; MTHFD, methylene-THF dehydrogenase; MTHFR, methylene-THF reductase; MTase, methyltransferase; ACS, acetyl-CoA synthase; PTA, phosphotransacetylase; ACK, acetate kinase. (E) Butyrate synthesis pathway (via crotonyl-CoA): key enzymes: ACAT, acetyl-CoA acetyltransferase; BHBD, β-hydroxybutyryl-CoA dehydrogenase; ECH, enoyl-CoA hydratase; BCD, butyryl-CoA dehydrogenase; PTB, phosphotransbutyrylase; BUT, butyryl-CoA:acetate CoA-transferase; BUK, butyrate kinase. Image created using BioRender.com . Acetate is predominantly produced via the Wood–Ljungdahl pathway involving several enzymes ( Fig. 3 ). Microbes involved in this pathway include; Bifidobacterium , Lactobacillus , Akkermansia muciniphila , Prevotella sp. , and Ruminococcus sp. ( Ziętek et al. 2021 ). Acetate serves as an important inter-species metabolic hub through cross-feeding by acting as the primary substrate for the butyryl-CoA:acetate CoA-transferase pathway ( Fig. 3 ). Dominant butyrate producers (e.g. F. prausnitzii and Roseburia spp .) depend on this pathway and grow poorly without acetate ( Duncan Sylvia et al. 2002 ). Thus, acetate-producing taxa effectively sustain the metabolic niche of butyrate producers ( Louis & Flint 2009 ). As the most abundant SCFA reaching the circulation, acetate serves as both a metabolic substrate and signalling molecule, providing 5–10% of total host energy, driving lipid and cholesterol synthesis, regulating protein acetylation and epigenetic modifications, and modulating metabolic homeostasis through G-protein-coupled receptor (GPCR) activation ( Fig. 4 ). ( Tang et al. 2019 , Moffett et al. 2020 a , b , Fu et al. 2024 ). Proposed mechanisms by which SCFAs influence maternal, placental, and fetal physiology during pregnancy. SCFAs interact with GPCR, histone deacetylase (HDAC), and free fatty acid receptor (FFAR) proteins or are transported into the cell to stimulate secretion of hormones, which regulate pancreatic β-cell proliferation, insulin secretion, and appetite during gestation. In the gut, SCFAs support energy metabolism and mucin production by intestinal epithelial and goblet cells, promoting gut barrier integrity and beneficial microbiota adhesion. Hepatic propionate utilisation for gluconeogenesis may contribute to gestational hyperglycaemia and insulinaemia. Placental uptake of SCFAs may occur via monocarboxylate transporters (MCTs), breast cancer resistance protein (BCRP), and free fatty acid receptors (FFARs). Within the cell, they may activate transcription factors such as peroxisome proliferator-activated receptor (PPAR) and forkhead box O (FoxO). These affect host physiology by modulating innate immunity, trophoblast differentiation, cytokine production, and neurodevelopment. Altered microbial profiles of acetate producers and reduced maternal acetate are associated with PE and GDM, while higher cervicovaginal acetate correlates with PTB ( Jin et al. 2022 , Wang et al. 2022 ). In PE, gut dysbiosis is characterised by reduced abundance of Bifidobacterium , Akkermansia , and Coprococcus and correlates with increased abundance of opportunistic pathogens and decreased SCFA production ( Chen et al. 2020 ). Reduced maternal acetate precedes the clinical presentation of PE and is parallelled by reduced acetate concentrations in the fetal circulation ( Hu et al. 2019 ). A review of 11 case–control studies (1,318 participants; 352 with PE) found that five studies reported significantly lower acetate levels in women with PE, while one study observed higher acetate levels specifically ( Zhao et al. 2025 ). Low maternal acetate levels are associated with reduced thymic size and function in the offspring, which persist up to 4 years of age ( Hu et al. 2019 ). In mice, acetate supplementation rescues compromised fetal thymic CD4+ and regulatory T cell (Treg) development by inducing expression of the autoimmune regulator (AIRE). In GDM, decreased circulating SCFAs during the second and third trimesters is accompanied by reduced placental GPR41/43 expression, increased HDAC activity, enhanced inflammation, and activated glycolysis ( Wang et al. 2022 ). PTB, which is often accompanied by cervical-vaginosis, is characterised by reduced lactobacilli and increased anaerobes (including Gardnerella vaginalis ) and is associated with elevated acetate and increased PTB risk. Mechanistically, in HTR8/SVneo (first-trimester EVT) cells, acetate increases pro-inflammatory cytokines IL-6 and IL-8 and promotes EVT migration and invasion by activating the ERK1/2 signalling pathway ( Meng et al. 2024 ). This in turn regulates cell proliferation, differentiation, and migration and acts upstream of the SCFA receptor FFAR2 ( Fig. 4 ). In human trophoblast stem cells and primary cytotrophoblast, acetate replenishes the acetyl-CoA pool, thereby maintaining histone H3K9/18/27 and H4K16 acetylation, activating essential placental syncytialisation genes while suppressing pro-inflammatory trophoblast fates ( Yu et al. 2024 ). Butyrate is produced through two primary pathways: the butyryl-CoA:acetate CoA-transferase pathway (the dominant route) and the butyrate kinase pathway, both utilising acetyl-CoA as the initial substrate, with acetate serving as a critical co-substrate for the transferase pathway ( Fig. 3 ) ( Cui et al. 2023 ). Although butyrate is produced in much smaller amounts than acetate and propionate, it plays a significant role in maintaining energy metabolism and intestinal homeostasis ( Ziętek et al. 2021 ). Major butyrate-producing genera include Faecalibacterium , Roseburia , Eubacterium , Anaerostipes , Coprococcus , Subdoligranulum , and Anaerobutyricum ( Singh et al. 2022 ). Butyrate is largely consumed locally by the gut epithelium and serves as the primary energy source for colonocytes, the absorptive cells lining the large intestine [187]. In addition, it supports gut barrier integrity by fuelling energy metabolism, stimulating mucus production, reducing inflammation via NF-κB inhibition, and enhancing gap junction protein expression ( Yin et al. 2001 , Peng et al. 2009 ). At high concentrations, butyrate inhibits HDACs and activates GPCRs ( Fig. 4 ). Butyrate stabilises hypoxia-inducible factor (HIF) to maintain an anaerobic gut environment, regulates tight junction protein expression (claudin-1) and cytoskeletal proteins (synaptopodin) to support gut barrier integrity, and inhibits oncogenic signalling pathways (Akt/ERK, Wnt, TGF-β) ( Yin et al. 2001 , Peng et al. 2009 , Noah et al. 2011 , Wang et al. 2012 , Hodgkinson et al. 2023 ). In immune cells, butyrate promotes differentiation of regulatory T cells (Tregs) and regulatory B cells, suppresses NF-κB and MAPK activation, reduces pro-inflammatory cytokine production, and maintains immune homeostasis through epigenetic modifications ( Mann et al. 2024 ). Dysbiosis of the butyrate-producing gut microbiota and decreased circulating butyrate levels are associated with adverse pregnancy outcomes. In PE, systemic butyrate levels are consistently lower ( Zhao et al. 2025 ). Gut dysbiosis with reduced butyrate-producing Coprococcus species correlates with low serum butyrate and reduced abundance of genes encoding the terminal step in bacterial butyrate formation ( but gene) ( Altemani et al. 2021 , Cui et al. 2023 ). Notably, reduced butyrate precedes the onset of PE symptoms, particularly in obese pregnant women ( Hu et al. 2019 , Altemani et al. 2021 , Cui et al. 2023 ). Furthermore, oral butyrate supplementation in LPS-induced hypertensive pregnant rats significantly lowered blood pressure, highlighting its potential therapeutic role ( Chang et al. 2020 , Ziętek et al. 2021 ). GDM, consistently features reduced beneficial butyrate-producing taxa and increased pathogenic strains, mirroring patterns observed in type 2 diabetes mellitus ( Han & Lin 2014 , Ferrocino et al. 2018 , Hasain et al. 2020 , Medici Dualib et al. 2021 , Wang et al. 2022 , 2024 c , Sohrabi et al. 2025 ). High-fermentable dietary fibre protects against GDM in mice through the gut–placental axis ( Huang et al. 2023 ). It increases the abundances of Lachnospiraceae and butyrate, which enhances gut barrier function and limits the transfer of bacterial-derived LPS. This reduces placental inflammation, ultimately preventing high-fat-diet-induced insulin resistance. In a PTB mouse model, vancomycin-induced dysbiosis increased preterm birth rates by 43% ( Uchida et al. 2025 ). This was reversed by butyrate supplementation, which also restored Treg numbers. Reduced butyrate producers ( Lachnospiraceae , Ruminococcaceae ) correlate with a higher risk of PTB and shorter gestation ( Uchida et al. 2025 ). In primary decidual, amniotic mesenchymal, and amniotic epithelial cells, high butyrate and propionate concentrations inhibit TNF-α- and IL-1β-induced expression of pro-inflammatory cytokines, chemokines, the uterotonic prostaglandin PGF2α, and matrix-degrading enzymes ( Moylan et al. 2020 ). Maternal butyrate also leads to lasting anti-inflammatory effects, increased microbiome diversity, and protection against colitis in the offspring, suggesting a role in developmental programming ( Barbian et al. 2022 ). Propionate is produced via three different pathways, with the succinate pathway being the most dominant ( Fig. 4 ). Microbes encoding enzymes for this pathway include Bacteroidetes and Negativicutes , while those involved in the acrylate pathway include Lachnospiraceae and Negativicutes ( Reichardt et al. 2014 ). Ruminococcus obeum , Roseburia inulinivorans , and Eubacterium hallii encode enzymes for the propanediol pathway ( Fig. 4 ). A clear functional divergence exists between butyrate- and propionate-producing microbes, with very few taxa exhibiting pathways for the production of both ( Kircher et al. 2022 ). In the liver, propionate serves as a key substrate for gluconeogenesis and supports multiple physiological processes ( Tsukuda et al. 2021 ). Beyond its metabolic functions, propionate is the most potent endogenous agonist for GPR43, which is expressed mainly in immune cells, but also in adipocytes, enterocytes, and hepatocytes (similar to the other SCFAs) ( Fig. 4 ) ( Yoshida et al. 2019 , Cui et al. 2023 ). It can also influence whole-body glucose metabolism by inhibiting hepatic gluconeogenesis through GPR43/AMPK and HDAC signalling to downregulate glucose-6-phosphatase and phosphoenolpyruvate carboxykinase while simultaneously stimulating intestinal gluconeogenesis ( Pingitore et al. 2017 , Yoshida et al. 2019 ). This enhances insulin sensitivity by improving β-cell function and potentiating glucose-stimulated insulin secretion and inducing energy expenditure via browning in mesenteric adipose tissue. Propionate also plays a role in appetite regulation by stimulating the release of satiety hormones, including GLP-1 and PYY, from enteroendocrine cells by activating GPR41/GPR43, leading to reduced energy intake and potential weight loss ( Fig. 4 ) ( Arora et al. 2011 ). Alterations in propionate-producing species and systemic propionate levels are associated with PE, GDM, and PTB. In PE, maternal circulating propionate shows variable patterns. One study reported significantly higher propionate levels, while other two studies reported significantly lower levels ( Zhao et al. 2025 ). Notably, propionate supplementation significantly alleviates PE symptoms in rat models by promoting autophagy and M2 polarisation in the placenta. In addition, propionate significantly inhibits the expression of anti-angiogenic and inflammatory factors in HTR-8/SVneo cells and promotes trophoblast invasion ( Jin et al. 2022 ). In a hypoxia-induced FGR mouse model, maternal propionate treatment reverses reduced birth weight in male offspring, reduces hepatic insulin resistance, gluconeogenesis, lipogenesis, and lipid accumulation, and improves liver function ( Chen et al. 2022 ). This is mediated through peroxisome proliferator-activated receptor (PPAR) and forkhead box O (FoxO) transcription factor signalling pathways. Thus, through their interaction with G protein-coupled receptors (GPCRs) GPR42, GPR43, and GPR109a, SCFAs can stimulate the secretion of hormones including glucagon-like peptide-1 (GLP-1), peptide YY (PYY), and leptin ( Lin et al. 2012, Tang et al. 2022 ) ( Fig. 4 ). Both GLP-1 and PYY are key to pancreatic β-cell proliferation, expansion, and glucose-dependent insulin secretion thus important for gestation-associated changes in appetite regulation and glucose and insulin signalling ( Yeo et al. 2022 , Mangliar et al. 2024 ). Human placental monocarboxylate transporters (MCTs), breast cancer resistance protein (BCRP), and free fatty acid receptors (FFARs) may mediate the uptake and utilisation of SCFAs by the placenta and the fetus for energy metabolism ( Fig. 4 ) ( Qin et al. 2025 ). Amino acids are important for host protein accretion, energy metabolism, and cellular signalling ( McIntyre et al. 2020 ). They also serve as precursors for microbial energy and bioactive molecules that function as QS signals in both microbe–microbe and microbe–host interactions ( Darkoh et al. 2019 , Hu et al. 2024 ). Amino acids, including phenylalanine, methionine, and tryptophan (TRP), are metabolised by the microbiome to produce p-cresol derivatives, S-adenosyl methionine, and indole derivatives, respectively ( Lin et al. 2017 , Kang et al. 2019 ). TRP, an essential amino acid required for host protein synthesis during development, represents one of the most extensively studied in the context of host–microbial metabolism ( Karahoda et al. 2020 ). It is metabolised through three different pathways: the host kynurenine (KYN) and SER pathways account for 95% and 1–2% TRP metabolism, respectively, and the microbiota pathway accounts for 4–6%. Reduced maternal serum TRP is associated with GDM, FGR, PE, PTB, and spontaneous abortion ( van Zundert et al. 2022 ). The metabolism of TRP to KYN is mediated by the rate-limiting enzymes indoleamine 2,3-dioxygenases (IDO1 and IDO2). KYN and its downstream metabolites support maternal immune tolerance by modulating T-cell activity and promoting an anti-inflammatory environment for the conceptus, while also influencing vascular tone in both maternal and placental tissues ( Notarangelo & Pocivavsek 2017 , Siska et al. 2021 ). Altered KYN metabolism is associated with pregnancy complications, including GDM, FGR, and PTB ( van Zundert et al. 2022 ). Suppression of KYN production by inhibiting IDO leads to fetal loss in mice, while increased placental IDO methylation in humans is associated with miscarriage ( Munn et al. 1998 , Spinelli et al. 2019 ). SER is important for normal placental and fetal development and is critical to maternal neuroendocrine signalling ( Sanjuan et al. 2008 , Lugo-Candelas et al. 2018 , Wu et al. 2019 , Morris et al. 2025 ). Maternal inflammation increases SER production, preeclamptic mothers exhibit high circulating SER levels and altered SER metabolism, and signalling correlates with neuroendocrine dysregulation, FGR, PTB, and sudden infant death syndrome ( Suzuki et al. 2001 , Côté et al. 2003 , Goeden et al. 2016 , Ranzil et al. 2019 , Miyamoto et al. 2024 , Broekhuizen et al. 2025 , Kinney et al. 2025 , O’Reilly Sparks et al. 2025 ). In animal models, SER increases uterine arterial blood pressure and decreases uterine blood flow and aberrant SER signalling leads to cardiac dysfunction and heart failure ( Lang et al. 1993 ). During infections, microbes can also shift TRP metabolism from KYN and SER synthesis to indole production via QS signalling. The conversion of TRP to indole is mediated by pathobionts such as Shigella boydii , Vibrio cholerae , Proteus vulgaris, Escherichia fergusonii , and E. coli . In diabetic mice, indole improves glucose tolerance and insulin sensitivity by stimulating GLP-1 secretion and promoting L-cell differentiation, although in women with GDM, fetal indole concentrations are higher than in women without GDM ( Chimerel et al. 2014 , Trajkova et al. 2021 ). Together with p-cresol-sulphate, IS is a well-known uraemic toxin that promote renal fibrosis, vascular dysfunction, and inflammation via AHR and NF-κB signalling ( Niwa 2010 ). Renal dysfunction and vascular dysfunction are key features of PE ( Dines et al. 2023 ). Maternal plasma levels of IS do not change across gestation, although IS supplementation in mice induces renal injury and increases BBB permeability in the mother and fetal resorption ( Griffin et al. 2023 ). Elevated levels of indole-producing microbes are associated with PE in women with urinary tract infections and with GDM ( Kuang et al. 2017 , Kaduma et al. 2019 ). Pathogens (e.g. Vibrio cholerae ) colonising the host gut metabolises TRP to produce indole as a QS molecule to regulate their biofilm formation and virulence ( Fig. 5 ) ( Holoidovsky & Meijler 2020 ). As the population density of V. cholerae increases, QS molecules mediate the inhibition of bacterial TRP metabolism, diverting TRP to SER production, which interacts with 5-HT receptors to activate host immune system ( Jugder et al. 2022 ). Thus, infection-led QS signalling can impact TRP metabolism and pregnancy physiology. Regulation of host–microbial TRP metabolism and immune responses. Colonising pathogens metabolise TRP to produce indole derivatives and other QS autoinducers that regulate microbial growth, biofilm formation, and virulence. As microbial density increases, rising QS autoinducer levels suppress microbial TRP metabolism while promoting TRP production by the microbiome, diverting TRP towards SER synthesis. SER, together with indoles and QS autoinducers, activates host SER receptors and the AHR-modulating protein phosphorylation and gene transcription to elicit host physiological responses, including immune activation. Gut microbes associated with this pathway include Bifidobacterium , Lactobacillus , Lactiplantibacillus , and various Clostridium species ( Dodd et al. 2017 , Sakurai et al. 2019 , Wang et al. 2024 a ). Consistent with shared synthetic pathways, ILA increases IPA and indole-3-acetic acid (IAA) production via microbial cross-feeding ( Dodd et al. 2017 , Wang et al. 2024 a ). Bifidobacterium and Lactobacillus are used as probiotics to restore a healthy microbiome, including in preterm infants ( Binda et al. 2025 ). IPA and ILA are potent antioxidants with anti-inflammatory effects. They improve insulin sensitivity, strengthen epithelial barrier integrity, and protect against IS-induced damage, and ILA mitigates sFlt-1-induced PE-like effects via the activation of AHR ( Yisireyili et al. 2017 , Negatu et al. 2020 , Zhang et al. 2022 , Wei et al. 2025 ). Species involved in the production of IAA include Lactobacillus reuteri, Clostridium sporogenes , and Clostridium scindens . The effects of IAA on the host are context dependent ( Dou et al. 2015 , Ji et al. 2020 , Shang et al. 2023 , Nayak et al. 2024 , Qu et al. 2024 ). In a model of renal injury, IAA reversed renal dysfunction and enhanced intestinal barrier function by activating AHR signalling and alleviated colitis induce tumourigenesis through pregnane X receptor (PXR) signalling ( Cao et al. 2025 , Wang et al. 2025 ). Maternal supplementation with IAA or indole during pregnancy confers long-term benefits to the offspring, including reduced hepatic steatosis, improved insulin sensitivity, and prevention of liver fibrosis ( Mandala et al. 2026 ). However, together with IS and pCS, IAA is also classified as a uraemic toxin as it increases tissue factor production and inflammation in endothelial cells and positively correlates with renal dysfunction, risk of cardiovascular diseases, and all-cause mortality ( Gondouin et al. 2013 , Cernaro et al. 2022 ). IAA can be synthesised via multiple microbial pathways. The ecological context of its production warrants careful consideration, as cross-feeding interactions among microorganisms influence community structure and metabolic outputs. These interactions may lead to the generation of additional metabolites with effects that range from beneficial to deleterious for the host. Species that produce TAM include Ruminococcus gnavus, Enterocloster asparagiformis, Clostridium nexile, Blautia hansenii , and Clostridium sporogenes ( Bhattarai et al. 2018 , Otaru et al. 2024 ). The colonisation of germ-free mice with the top TAM producer R. gnavus or TAM supplementation in mice and monkeys leads to insulin insensitivity and contributes to metabolic syndrome in IBS through the activation of trace amine-associated receptor 1 (TAAR1) ( Zhai et al. 2023 ). However, in diet-induced insulin resistance, TAM reduces lipolysis and improves insulin sensitivity, suggesting that the effects of TAM may also be context dependent ( Lee et al. 2025 ). TAM reduces the abundance of microbes involved in the metabolism of complex carbohydrates to SCFAs, including the keystone species Faecalibacterium prausnitzii and others such as Faecalibacterium duncaniae, Blautia spp., Bacteroides uniformis, Bacteroides thetaiotaomicron , and Parabacteroides vulgatus ( Otaru et al. 2024 ). While the microbiome can produce TAM at millimolar concentrations, its concentration in the circulation is regulated by monoamine oxidases (MAO-A/B). MAO activity is thought to be important for the activation of AHR by mediating the production of an unknown TAM metabolite speculated to be IAAld ( Vikström Bergander et al. 2012 , Vikström Bergander et al. 2012 ). In mice and colonic cells, TAM increases fluid secretion by activating the SER or 5-hydroxytryptamine receptor 4 (5-HTR4) ( Bhattarai et al. 2018 ). Notably, TAM is among proposed markers of bacterial vaginosis ( Srinivasan et al. 2015 , Gajjar & Seshadri 2025 ). Thus, most indole metabolites directly or indirectly interact with host AHR, PXR, TAAR, and SER receptors with varying effects depending on the context ( Parker et al. 2020 ). Signalling through these receptors regulate various homeostatic components associated with pregnancy physiological adaptations, including beta-cell proliferation, glucose uptake and metabolism, insulin secretion, immune maturity, and cellular proliferation and differentiation ( Sayed et al. 2022 ). The activation of AHR also regulates blood pressure during inflammation and activates the production of prostaglandins, which are important for uterine contractions during labour ( Thornton & Ramphul 2023 ). Both AHR and PXR are implicated in altered placental vascularisation and the pathogenesis of FGR ( Jiang et al. 2010 ). Modulating the microbiome to favour ILA and IPA over indole and IS using complex carbohydrates offers a potential avenue for clinical intervention and pregnancy screening. IAA and the uraemic toxins p-cresol-sulphate and IS may also be useful as markers for renal and cardiovascular dysfunction associated with PE. QS molecules include the autoinducer-1 (AI-1) AHLs, autoinducer-2 (AI-2) molecules ((2S,4S)-2-methyl-2,3,3′,4-tetrahydroxytetrahydrofuran-borate (S-THMF-borate) and (2R,4S)-2-methyl-2,3,3′,4-tetrahydroxytetrahydrofuran (R-THMF)), and autoinducer-3 (AI-3). QS production is mediated by environmental cues, such as nutrients and hormones (oestrone, oestriol, and oestradiol) ( Beury-Cirou et al. 2013 , Paul et al. 2021 ). Microbes such as Klebsiella and E. coli can alter their virulence and biofilm formation by sensing AHL produced by Pseudomonas aeruginosa , while they themselves do not produce AHL. Similarly, E. coli can produce AI-2 signalling molecules, which promote the growth of Bacillota while concurrently suppressing the colonisation of Bacteroidota ( Thompson et al. 2015 ). As E. coli increases with advancing gestation, this may offer a mechanism through which the Bacillota-to-Bacteroidota ratio increases during pregnancy ( Ren et al. 2023 ). In addition, lactobacilli L. crispatus , which confers a healthier and stable vaginal microbiome, together with L. jensenii , produces more AHLs than L. gasseri , which is a poor coloniser and confers a less stable vaginal microbiome ( Sanchez et al. 2024 ). The ability to produce AHLs by these lactobacilli is associated with biofilm formation, indicating the importance of biofilms in vaginal health, particularly as vaginosis positively correlates with adverse pregnancy outcomes. QS molecules also play an important role in inter-kingdom signalling, thus influencing host–microbe interactions ( He et al. 2023 ). Although the direct role of QS during pregnancy is unclear, several QS molecules, including AHLs, produced by pathobionts such as Pseudomonas aeruginosa and E. coli can directly bind AHR to modulate host innate immunity ( Moura-Alves et al. 2019 ). AHLs regulate several aspects of cellular energy metabolism, and low AHL concentrations are associated with altered amino acid metabolism and piglet birth weight ( Tao et al. 2021 , Liu et al. 2022 ). AHLs also regulate cell proliferation, migration, and differentiation of human intestinal cells ( Karlsson et al. 2012 ). Similar to intestinal cells, the placenta and the fetus undergo dynamic cell proliferation, migration, and differentiation, processes that heavily rely on the dynamic changes in gene expression, DNA/RNA methylation and genetic stability ( Koukoura et al. 2012 ). Microbial production of AHL requires SAM as a substrate, and SAM is produced from the essential amino acid methionine ( Dong et al. 2017 ). SAM is a key component of one-carbon metabolism and DNA/RNA methylation. One-carbon metabolism and methylation processes are vital for the regulation of gene expression, genetic stability, and epigenetic modifications during development in the host. Considering that AHLs are produced by a wide range of bacteria, the partitioning of methionine and SAM between the host and the microbiome could create a competitive environment with possible detrimental effects on pregnancy outcomes. The microbiome produces second messengers, including c-AMP and c-GMP. In the host, these are intracellular molecules that mediate transcription of genes involved in developmental processes, such as cell migration, proliferation, and differentiation. However, their function is indistinguishable from the c-AMP/c-GMP produced by the host. The production of (p)ppGpp, c-di-GMP, and c-di-AMP is, however, conserved among prokaryotes including several members of the microbiome. The excretion of these second messengers into the environmental niche mediated by multi-drug efflux pumps (MDRs) is important for mediating microbe–microbe interactions ( Woodward et al. 2010 ). It is widely thought that host cellular responses to these molecules require the entrance of the microbe into the host cell where the second messenger is excreted into the cytosol. However, the direct incubation of host cells with these second messengers also elicits a similar response where the second messenger directly binds stimulator of interferon genes (STING) and stimulates type I IFN response ( Burdette et al. 2011 ). This points to the presence of receptors for microbial second messengers in the host, although their identity is yet to be elucidated. Both c-di-GMP and c-di-AMP primarily regulate host innate immune signalling via STING–TBK1–IRF3 to induce interferon type I (type-I IFN family) excretion ( Andrade et al. 2016 ) ( Fig. 6 ). The activation of STING also regulates NF-κB signalling to induce cytokine production. Both these pathways are crucial for innate immune maturity and cytokine production during pregnancy and are implicated in adverse pregnancy outcomes, including PE and spontaneous miscarriages ( Ariyakumar et al. 2021 ). Effects of microbial-derived second messengers on maternal–placental–fetal physiology. Maternal hormones and nutrients may induce the microbiome to produce signals such as metabolites and quorum sensors that have a direct impact on the production of second messengers (cAMP, cGMP, c-di-GMP, and c-di-AMP). These as soluble factors can interact with host proteins such as stimulator of interferon genes (STING), which induces phosphorylation of TANK-binding kinase 1 (TBK1) and inhibitor of nuclear factor kappa-B kinase (IKK) and subsequently interferon regulatory factor 3 (IRF3) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), respectively. These translocate to the host nucleus and induce transcription of genes involved in innate immunity, such as type I interferons (IFN), and inflammatory cytokines, such as tumour necrosis factors-alpha (TNF-α), with implications for innate immunity activation and maternal inflammatory status with advancing gestation. Maternal hormones and nutrients may also induce microbial production of the alarmones guanosine pentaphosphate ((p)ppGpp)) and guanosine tetraphosphate ((p)pGpp)), which may have direct effects on the host nutrient availability and metabolism with implications for maternal glycaemic and insulin status, trophoblast function and development, and fetal growth and development. In E. coli , (p)ppGpp responds to nutrient deprivation by inhibiting RNA synthesis and upregulating the transcription of genes involved in amino acid synthesis and transport. Microbial (p)ppGpp production is also regulated by fasting and fed states, particularly during glucose depletion ( Zhao et al. 2024 ) ( Fig. 6 ). Thus, (p)ppGpp may be produced in response to changes in host glucose metabolism and insulin signalling during pregnancy, which may in turn regulate shifts in microbial community dynamics ( Ontai-Brenning et al. 2023 , Zhao et al. 2024 ). The conversion of choline to TMA requiring the glycyl radical enzyme choline TMA-lyase (cutC) and its activating enzyme (cutD) is the most predominant pathway for TMA production from choline ( Fig. 7 ). CutC/D genes are predominantly found in species belonging to the Firmicutes and Proteobacteria phyla, including Clostridium aminobutyricum, Enterocloster citroniae, Lachnospiraceae bacterium, and Desulfovibrionales such as Desulfovibrio desulfuricans . The direct conversion of choline to betaine as an intermediate for TMA formation has been demonstrated in mice. However, using human faecal microbiome, we demonstrated that the conversion of choline to TMA does not require the formation of betaine as an intermediate ( Day-Walsh et al . 2021 ). The conversion of l -carnitine to TMA involves both the direct aerobic pathway and the indirect dual-anaerobic pathway. The former involves the l -carnitine monooxygenase oxygenase subunit (cntA/YeaW) and the l -carnitine monooxygenase reductase subunit (cntB/YeaX) ( Fig. 7 ). The latter involves the cai -operon and the fix -operon, which convert l -carnitine to the obligate intermediate γ-BB. The γ-BB is further metabolised by the γ-BB utilisation ( bbu )-operon to TMA. YeaW and YeaX genes are found in Proteobacteria, including Citrobacter freundii, Klebsiella pneumoniae, and Acinetobacter baumannii . Microbes containing the cai -operon, such as E.coli, Salmonella enterica, and Proteus mirabilis, are more diverse compared with those containing the bbu gene cluster . Emergencia timonensis is one of the few strains known to contain the bbu gene cluster and to metabolise γ-BB to TMA. Microbes involved in TMA production are pathobionts often associated with PTB and PE. TMA is primarily oxidised to TMAO by the host hepatic flavin monooxygenases (FMOs) ( Koeth et al. 2014 ). Metabolic pathways involved in choline and l -carnitine metabolism. CutC/cutD are involved in the direct anaerobic conversion of choline to TMA, yeaW/yeaX in the direct aerobic conversion of l -carnitine to TMA, and the CaiTABCDEF and fixABCX operon in the complex, indirect, and anaerobic conversion of l -carnitine to y-BB, and the BbuABCDE operon in the anaerobic conversion of y-BB to TMA. This is another example of competitive nutrition partitioning between the maternal gut microbiota and the maternal–placental–fetal unit in which the microbiome consumes nutrients that are essential to the growing fetus causing deprivation of these nutrients and also produces potentially toxic compounds. Choline is important for many cellular functions, including the development of fetal membranes, folate metabolism, and neurodevelopment. Being key to one-carbon and folate metabolism as a methyl donor, choline deficiency can lead to altered methylation status and energy metabolism in placental and fetal tissues. This could subsequently lead to epigenetic alterations, which can have long-term health effects later in the offspring's adulthood ( Blusztajn & Mellott 2012 , Sowton et al. 2020 ). Alterations in choline metabolism and bioavailability are implicated in neural tube defects, non-alcoholic fatty liver diseases, and increased risk of CVD ( Zhu et al. 2003 , Shaw et al. 2009 , Roe et al. 2017 ). The microbiome also requires choline for the formation of its essential biofilm component teichoic acids ( Bärland et al. 2022 ). l -carnitine is critical for lipid metabolism, and its plasma levels decrease almost by 50% during pregnancy ( Keller et al. 2009 ). High plasma TMAO levels offer a better indicator of extreme secondary events after heart failure than traditional markers, such as N-terminal pro-B-type natriuretic peptide (NT-proBNP), and can therefore be used as a biomarker for screening patients who are at a high risk of death within 1–3 years ( Suzuki et al. 2016 ). In pregnancy, high plasma TMAO levels positively correlate with endothelial dysfunction, systemic inflammation, and an increased risk of PE ( Wen et al. 2019 , Wang et al. 2024 b ). FMT from PE patients into pregnant mice leads to PE symptoms, including hypertension and inflammation, and FGR and was shown to be blocked by co-supplementation of FMT with 3,3-dimethyl-1-butanol (DMB), an inhibitor of TMA formation ( Wang et al. 2024 b ). Beyond soluble metabolites, the gut microbiome may influence the maternal–fetal interface via microbe-associated molecular patterns (MAMPs) and bacterial extracellular vesicles (BEVs). MAMPs are highly conserved structural motifs that include lipopolysaccharide (LPS), peptidoglycan (PGN), and lipoteichoic acid (LTA), among others ( Furnari et al. 2025 ). These interact with pattern recognition receptors (PRRs), including Toll-like receptors (TLRs) and NOD-like receptors (NLRs), initiating host immune signalling and inflammation ( Manshouri et al. 2024 ). Both trophoblast and fetal membrane cells express TLRs and NLRs, which interact with MAMPs and pathogen-associated molecular patterns (PAMPs) during infection, leading to chorioamnionitis and preterm birth ( Koga & Mor 2010 , Hoang et al. 2014 ). However, little is known about how commensal-derived MAMPs interact with these tissues under normal physiological conditions ( Koga & Mor 2010 , Padron et al. 2020 ). Nevertheless, commensal-derived MAMPs can regulate gut barrier integrity via TLR signalling ( Burgueño & Abreu 2020 ). Recent evidence suggests that TLR2/TLR6-mediated microbial sensing triggers neuropilin-1 (NRP1 lysosomal degradation), a positive-feedback regulator of Hedgehog (Hh), thereby suppressing its signalling and weakening the intestinal epithelial barrier ( Pontarollo et al. 2023 ). Notably, intestinal permeability increases with advancing gestation, which may increase opportunities for microbial metabolites and MAMPs to enter maternal circulation ( Ma et al. 2025 , Ma et al. 2025 , Barrientos et al. 2026 ). Excessive barrier disruption increases translocation of MAMPs, such as LPS into circulation, driving metabolic endotoxemia and chronic TLR4-mediated inflammation ( Basu et al. 2011 , Mohammad & Thiemermann 2020 ). Circulating LPS activates IDO, shifting tryptophan metabolism towards the kynurenine pathway and reducing AHR-activating indole derivatives that support tight junction integrity and IL-22-mediated mucosal immune defence ( Laurans et al. 2018 ). Dysbiosis also reduces SCFA-producing bacteria and butyrate availability, further impairing barrier integrity and mucosal immune regulation ( Singh et al. 2022 ). Collectively, these processes create a cycle where dysbiosis-driven barrier disruption increases MAMP translocation and metabolic endotoxemia, further reducing indole derivatives and SCFAs that maintain barrier integrity ( Massier et al. 2021 ). This is associated with placental dysfunction, insulin resistance, GDM, and PE ( Xie et al. 2010 , Basu et al. 2011 , Yang et al. 2016 , Mokkala et al. 2017 ). BEVs provide an additional route for gut-derived microbial signals to reach the maternal–fetal interface by transporting nucleic acids, lipids, metabolites, and MAMPs within a protective lipid bilayer ( Taitz et al. 2023 ). This enables simultaneous activation of multiple pattern-recognition receptors and long-range delivery to gestational tissues ( Gilmore et al. 2022 , Barrientos et al. 2026 ). BEVs resembling those of the maternal gut microbiota have been detected in human amniotic fluid and cross biological barriers to reach the intra-amniotic space ( Kaisanlahti et al. 2023 ). BEVs are also taken up by decidual and cytotrophoblast cells and appear to be normal constituents of pregnancy at physiological concentrations ( Menon et al. 2023 ). Faecal BEVs from pregnant women shift T cells towards a pregnancy-supportive regulatory profile, increasing Th2 and reducing Th17 responses, suggesting a role in immune adaptation during pregnancy ( Dietz-Ziegler et al. 2025 ). Collectively, the translocation of both free MAMPs and BEVs from the maternal gut into the circulation represents an emerging route by which the maternal microbiome may influence immune regulation and placental function during pregnancy, although the precise contribution of each pathway under normal physiological conditions and in disease remains to be fully elucidated. The paternal gut microbiome also plays a crucial role as an interface between the paternal preconception environment and intergenerational health ( Argaw-Denboba et al. 2024 ) ( Fig. 1 ). For example, the induction of paternal gut dysbiosis in male mice using non-absorbable antibiotics (nABX), alternative antibiotic regimens, and osmotic laxatives results in reduced microbial diversity and richness. This is accompanied by significantly lower birth weights, heightened susceptibility to severe growth restriction, and increased mortality rates in their offspring. Transcriptomic analysis revealed distinct gene expression profiles in the brain and brown adipose tissue of affected offspring, suggesting that adverse effects were transmitted through sperm. Histological analysis further demonstrated smaller testes, reduced sperm count, and abnormalities in seminiferous tubules in dysbiotic males, linking microbiota disruption to reproductive system alterations. Importantly, these adverse effects were reversible following eight weeks of microbiota recovery, underscoring the transient nature of dysbiosis-induced outcomes ( Argaw-Denboba et al. 2024 ). While the precise molecular mechanisms remain to be elucidated, this highlights the potential importance of the paternal microbiome in intergenerational health and proposes a novel gut–germline axis influencing offspring development. Next-generation sequencing and culture studies have also identified a link between various paternal species and vaginal flora, supporting findings from other studies that suggest bacteria are shared between men and women and that sexual intercourse can influence the microbiome ( Gallo et al. 2011 , Mändar et al. 2015 ). Further research is needed to explore the sharing of taxa and its potential implications for infertility, pregnancy, and disease processes ( Farahani et al. 2021 ).

Well-powered, well-characterised studies suggest that the microbiome changes concomitantly with advancing gestation. For example, in a Finnish cohort of 90 pregnant women, Koren et al. observed that first-trimester β-diversity closely resembled non-pregnant controls but increased markedly by the third trimester ( Koren et al. 2012 ). In parallel, the abundance of Pseudomonadota and Actinomycetota increased significantly ( Koren et al. 2012 ). Notably, third-trimester microbial profiles persisted in both mothers and infants for at least one month postpartum, suggesting maternal-to-infant gut microbiota transfer ( Koren et al. 2012 , Soderborg et al. 2020 ). Streptococcus, commonly associated with inflammatory states also increased in late gestation along with a decrease in SCFA-producing genera, such as Faecalibacterium and Eubacterium ( Koren et al. 2012 ). These microbial changes parallelled with elevated maternal inflammatory markers and insulin resistance. A complementary longitudinal study involving 640 women from low-income countries (sub-Saharan Africa, South Asia, and Central America) reported a decline in Lachnospiraceae and Ruminococcaceae (major SCFA producers) alongside an increase in Prevotella ( Tang et al. 2022 ). In contrast, a study of 179 healthy women in China and Hong Kong reported increased butyrate-producing species ( Flavonifractor plautii and Lawsonibacter asaccharolyticus ) and taxa involved in hormone and bile acid metabolism ( Bilophila wadsworthia ) during the second and third trimesters compared to the first ( Wang et al. 2024 d ). While a systematic analysis of 1,479 health pregnant women in China concluded that ‘the influence of gestational age on women’s gut microbiota is limited, but considerable’, they did show significant increases in Collinsella, Megamonas , and unclassified- Erysipelotrichaceae and decreases in Ruminococcus, Dialister , and unclassified- Lachnospiraceae ( Yang et al. 2020 ). A further study in Israel also demonstrated enriched SCFA-producing bacteria, including Blautia and Bifidobacterium , and progesterone- and oestrogen-metabolising genera, including Bacteroides, Ruminococcus, Clostridium, Akkermansia, and Prevotella, while Dehalobacterium and Clostridium decreased by the third trimester ( Nuriel-Ohayon et al. 2019 ). Similar changes were observed in both lean and obese Finnish women, although the latter showed higher levels of Bacteroides and Staphylococcus ( Collado et al. 2008 ). In contrast, several studies involving fewer participants from the USA and Austria ( n ≤ 40), which included women with preterm births, pre-existing cardiometabolic conditions, and antibiotic use, reported no significant microbiome changes across body sites as gestation progressed ( DiGiulio et al. 2015 , Granser & Foessleitner 2024 ). The latter also only sampled a very short period in gestation (32–37 weeks). Collectively, robust studies indicate that dynamic shifts in maternal hormones and metabolism during pregnancy are accompanied by changes in microbial populations across gestation possibly to ensure metabolic and homeostatic synchrony ( Koren et al. 2012 ). However, baseline microbiome composition and host metabolic state may shape these shifts, with SCFA-producing microbes increasing, decreasing, or remaining unchanged across gestation. Mechanistic studies in mice support bidirectional interaction between the microbiome and the host during pregnancy. Mice receiving third trimester human FMT develop a pro-inflammatory status along with insulin resistance, but those receiving microbiome from the first trimester do not ( Nuriel-Ohayon et al. 2019 ). Escherichia coli (E. coli) increases with advancing gestation, and the colonisation of germ-free mice with E. coli induces inflammation and insulin resistance, further substantiating the influence of the microbiome on host pregnancy physiology ( Ju et al. 2023 ). Signals from the host also affect the microbiome as progesterone elevates the abundance of gut Bifidobacterium, which is higher in pregnant than in non-pregnant mice ( Nuriel-Ohayon et al. 2019 ). Colonisation of germ-free pregnant mice with Bifidobacterium or antibiotic supplementation alters placental structural integrity, vascularisation, immune response, endocrine function, and nutrient transport, as well as rates of fetal resorption/loss ( Benner et al. 2021 , Pronovost et al. 2023 , Lopez-Tello et al. 2025 ). In non-pregnant women, the vaginal microbiome is dominated by lactic acid-producing Lactobacillus species, primarily L. crispatus , L. gasseri , L. iners , and L. jensenii ( Bhattacharya et al. 2023 ). These define microbial community states (CST I–III and V, respectively) ( Fig. 2 ). CST IV is dominated by vaginosis- and high pH-associated pathobionts, such as Mycoplasma hominis, Atopobium vaginae, Ureaplasma, Megasphaera, Mobiluncus, Gardnerella vaginalis, Sneathia , and Prevotella ( Aagaard et al. 2012 , Ma & Li 2017 ) ( Fig. 2 ). Lactobacillus is crucial for vaginal health as it prevents predominant colonisation of the afore-mentioned pathobionts by maintaining low vaginal pH ( Paige et al. 1998 ). High abundance of L. crispatus rather than that of L. jensenii and L. gasseri corresponds with a more stable and healthier vaginal microbiome although a recent study suggests that a modular assessment of these community state types offers a better measure of health than individual species alone ( Verstraelen et al. 2009 , Aagaard et al. 2012 , Lebeer et al. 2023 ). With this assessment, microbial modules consisting of L. crispatus, L. jensenii , and Limosilactobacillus were associated with a healthy vaginal microbiome phenotype ( Lebeer et al. 2023 ). Across multiple longitudinal and cross-sectional studies involving participants from European, American, and South African cohorts, the vaginal microbiome consistently shifted during pregnancy towards relatively stable, low-diversity communities dominated by Lactobacillus , often enriched with L. crispatus or L. iners , parallelled by metabolomic changes (elevated lactate and amino acids and reduced BV-associated species) ( Aagaard et al. 2012 , Huang et al. 2015 , Nunn et al. 2021 ) ( Fig. 2 ). In contrast, the postpartum period is characterised by increased microbial diversity, a decline in Lactobacillus abundance, and colonisation by anaerobes such as Gardnerella , Prevotella , Anaerococcus , Peptoniphilus , and Streptococcus . Thus, vaginal microbial dynamics differ from the conventional concept of eubiosis, in which high microbial diversity is typically associated with a healthy state ( Fig. 2 ). Ethnicity also influences vaginal microbiome composition. Hispanic and Black women typically exhibit higher CST IV-associated microbes, with increased L. iners and reduced L. crispatus , while L. gasseri is often absent ( Ravel et al. 2011 , MacIntyre et al. 2015 ). In contrast, Asian and Caucasian women display greater abundance of L. jensenii . L. iners which frequently dominates during transitions from eubiotic to dysbiotic states, suggesting its role as a marker of instability ( Huang et al. 2015 ). Most vaginosis CST IV-associated microbes utilise iron and often produce siderophores (iron-chelating molecules) to acquire iron from the host environment, while Lactobacillus sp . have low iron requirements ( Imbert & Blondeau 1998 , Jarosik et al. 1998 , Leung & Folk 2002 , Marimuthu et al. 2025 ). Notably, bacterial vaginosis is more common in women with uterine fibroids and endometriosis, conditions highly associated with heavy menstrual bleeding and more prevalent in Hispanic and Black women ( Farland et al. 2022 , Nishimura et al. 2022 , Orellana et al. 2022 , Cheng et al. 2023 , Li et al. 2024 a ). This suggests that these diseases may influence the baseline microbiome, or vice versa, and should be accounted for in future evaluations of microbial shifts during pregnancy particularly as they pose a high risk for pregnancy complications. Vaginal microbiome composition across reproductive states. In non-pregnant women, the vaginal microbiome is typically dominated by lactic acid-producing Lactobacillus species defining community state types (CSTs), which can vary based on ethnicity. During pregnancy, the vaginal microbiome shifts towards low-diversity, stable communities dominated by Lactobacillus , accompanied by metabolomic changes. Postpartum, diversity increases with reduced Lactobacillus and colonisation by anaerobes.

Maternal gut and vaginal microbiomes undergo phylogenetic and taxonomic shifts that coincide with metabolic, endocrine, and immune adaptations during pregnancy. These microbial communities not only respond to maternal physiological cues but also actively regulate host functions through complex microbe–microbe and microbe–host interactions mediated by molecules produced by both the host and microbiome. Emerging evidence suggests that gut-derived metabolites, such as SCFAs, TRP-indole derivatives, and TMAs, influence maternal, placental, and fetal physiology. Microbes also produce QS molecules and second messengers (e.g. cAMP and di-GMP) that modulate microbial population dynamics and host immune responses, although their roles in human physiology and pregnancy progression remain poorly understood. Functional redundancy within microbial communities complicates the interpretation of taxonomic shifts, as most studies lack integrated functional analyses. Understanding how maternal physiological adaptation coincides with both taxonomic and functional changes will be crucial for unravelling mechanisms and identifying therapeutic targets, such as probiotics, prebiotics, and FMT against adverse pregnancy outcomes. This will require multidisciplinary approaches integrating publicly available bioinformatic databases and multi-omics to study community and functional dynamics. High-throughput culturomics coupled with metabolomic isotope tracing, as well as microbial and host transcriptomic, proteomic, and peptidomic analyses, will provide clarity on competing or symbiotic microbial communities within specific nutritional or hormonal contexts. Current research on maternal and infant microbiomes faces significant limitations. Many studies focus exclusively on taxonomic composition without incorporating functional or metabolic profiling, despite functional genes and metabolites driving community behaviours between health and disease states. This will be important in understanding key drivers of microbial functional redundancy and resistance during health and complicated pregnancies. The predominance of cross-sectional or semi-longitudinal designs limits the understanding of microbiome dynamics across pregnancy and postpartum. Comparisons across geographical regions further complicate interpretation due to differences in baseline microbiome composition, dietary patterns, and socioeconomic factors. Global representation remains uneven, with high-income countries dominating datasets and regions such as Africa and Asia underrepresented, despite evidence of distinct microbiome trajectories. Although several user-friendly databases provide metadata on the associations between disease states and microbial profiles, closer examination reveals significant limitations ( Jin et al. 2021 , Zhou et al. 2023 , Liu et al. 2025 ). For instance, datasets often juxtapose preterm infant microbiome changes with maternal profiles without harmonised sampling strategies or temporal alignment. Research disproportionately emphasises GDM, while conditions such as PE, stillbirth, and FGR remain severely underexplored. Addressing these challenges will require integrated, longitudinal, multi-omics studies with globally representative cohorts, standardised metadata, and harmonised sampling protocols. Such efforts are essentialfor elucidating microbiome-mediated mechanisms underlying adverse pregnancy outcomes and for developing targeted interventions for improving maternal and fetal health.

Pregnancy complications lead to nearly 5 million maternal and infant mortality and morbidity annually and increase the risk of the offspring developing metabolic diseases in adulthood ( WHO 2023 , 2025 ). Maternal gut and vagina microbiomes are implicated in adverse pregnancy outcomes, such as PTB, PE, GDM, and stillbirth ( Vander et al. 2022 ). Bacterial vaginosis positively correlates with PTB and LBW ( Meng et al. 2024 , Paramel Jayaprakash et al. 2016 ). An increase in Megasphaera Corynebacterium and Gardnerella in the gut of preterm infants negatively correlate with Lactobacillus ( Shu et al. 2025 ). Other genera, including Enterococcus, Pseudomonas, Citrobacter , and Acinetobacter , are also increased in the gut and the vagina of mothers with preterm pre-labour rupture of membranes (PROMS) ( Ambalpady et al. 2022 ). These microbial changes are reflected in the microbiomes of premature (PREM) infants after birth, suggesting ascending infections ( Gibson et al. 2016 , Zhou et al. 2023 ). The abundance of gut Lactobacillus, Bifidobacterium , and Collinsella negatively correlate with PE, while plasma TMA, which is associated with atherogenic effects in both humans and mice, increases in PE between 31- and 32-week gestation ( Cui et al. 2024 , Wang et al. 2024 b ). The abundance of SCFA-producing species Bifidobacterium along with Faecalibacterium and members of the Lactobacillaceae family is significantly decreased in mothers with GDM. At species level and across gestation, SCFA-producing microbes, including Akkermansia muciniphila , Coprococcus eutactus , Enterorhabdus caecimuris , Dialister sp., are also decreased, while Blautia and pathobionts linked to inflammation, including Fusobacterium mortiferum , Ruminococcus gnavus , and Anaerotignum lactatifermentans , are enriched ( Ferrocino et al. 2018 , Ye et al. 2019 , Soderborg et al. 2020 ). Notably, while the abundance of the choline- and carnitine-utilising genera, which produce TMA (C ollinsella, Eggerthella, Lachnospiraceae, Enterobacteriaceae) , is also increased in GDM, TMA-utilising archaea, such as Methanobrevibacter smithii , decrease, suggesting reduced TMA utilisation ( Kuang et al. 2017 ). Low abundance of hormone-metabolising species ( Adlercreutzia equolifaciens and Asaccharobacter celatus ) and high abundance of the acetate producer Parabacteroides distasonis were also associated with increased GDM risk ( Wang et al. 2024 d ). Other genera elevated in GDM include Fusobacterium, Eubacterium hallii, Desulfovibrio, Rothia, Ruminococcus, Klebsiella variicola, Prevotella, and Blautia ( Kuang et al. 2017 , Ionescu et al. 2022 ). These data suggest that SCFA-producing species negatively correlate with PTB and PE, while in some cases, they either negatively or positively correlate with GDM. High TMA levels, on the other hand, positively correlate with both PE and GDM. These studies are limited as they focus primarily on taxonomic and less so on functional or metabolic shifts and rarely consider underlying metabolic or disease states, such as fibroids or endometriosis. Additionally, since the uterine environment is typically sterile, they provide no clear mechanisms.

In early life, the gut microbiome undergoes considerable compositional changes, gradually stabilising and forming the core microbiome by around three years of age, as the infant transitions from a milk-based diet to solid foods ( Bäckhed et al. 2015 ). This developmental window is therefore crucial for establishing a healthy microbiome that supports lifelong well-being. The core gut microbiome comprises six major phyla, dominated by Firmicutes (Bacillota), Bacteroidetes (Bacteroidota), Proteobacteria (Pseudomonadota), Actinobacteria (Actinomycetota), Verrucomicrobia, and Fusobacteria ( Turnbaugh et al. 2009 ). The core microbiome provides essential functional components key for the survival of that microbial community and its host. The microbiome includes absolute commensal microbes and opportunistic pathogens (pathobionts). A healthy (eubiotic) microbiome exhibits high functional redundancy, with diverse species performing similar roles to ensure resilience, recovery, and pathogen resistance. It also has a strong capacity to produce beneficial metabolites, such as short-chain fatty acids (SCFAs) ( Gu & Li 2016 , Fassarella et al. 2021 , Mei et al. 2022 , Singh et al. 2022 , Tsuda 2023 ). An unhealthy (dysbiotic) microbiome lacks beneficial traits, shows low diversity with a high abundance of pathobionts, and produces fewer beneficial metabolites while generating more harmful compounds, such as trimethylamine (TMA) and pro-inflammatory lipopolysaccharides (LPS) ( Rath et al. 2017 , Luo et al. 2025 ). However, these definitions can vary depending on the context, as will be explored in relation to the vaginal microbiome during pregnancy. Both mode of delivery and feeding practices influence the infant’s microbial composition ( Ma et al. 2023 ). Vaginally delivered infants typically acquire beneficial maternal microbes, including Bacteroides fragilis and Bacteroides thetaiotaomicron and keystone microbes such as Bifidobacterium sp., whereas those born via caesarean section harbour skin, oral, and hospital-associated pathobionts, such as Clostridium perfringens , Enterobacter cloacae , and Clostridioides difficile ( Rutayisire et al. 2016 , Vu et al. 2021 ). The gut microbiomes of breastfed infants are typically enriched by SCFA-producing taxa, including Bifidobacterium and Lactobacillus , which thrive in the presence of human milk oligosaccharides, although others have also shown the predominance of Ruminococcus gnavus and Collinsella species ( Kirmiz et al. 2018 , Oyedemi et al. 2022 ). In contrast, formula-fed infants often exhibit greater microbial diversity, with higher abundances of taxa from genera such as Clostridium and Enterobacteriaceae ( EbioMedicine 2021 ). Although generally stable, the core microbiome can shift under prolonged stress induced by antibiotics, poor diet, xenobiotics or chronic inflammation, allowing the colonisation of pathobionts ( Chandra et al. 2021 , Li et al. 2023 ). The inability for the microbiome to revert to the original state may allow pathobionts to reorganise community dynamics, reestablishing the core microbiome to an ‘alternative dysbiotic state’ ( Fassarella et al. 2021 ). This may result in loss of functional redundancy and the inability to physiologically respond to pregnancy signals and gestational demands. In ulcerative colitis, failure to respond to faecal microbiota transplantation (FMT) correlates with high Prevotellaceae , low Ruminococcaceae , and persistent inflammation, which hinder donor engraftment and recovery ( Pinto et al. 2024 ). This is crucial when interpreting microbiome shifts in pregnancy, especially amid the debate over whether the gut microbiota changes significantly as gestation progresses. Nevertheless, certain ‘keystone species’ are essential for maintaining community stability, supporting functional redundancy, and preserving resilience against disturbances ( Wu et al. 2022 ). Ruminococcaceae, Lachnospiraceae , and Bifidobacterium are among the keystone species essential for maintaining gut barrier integrity and an anti-inflammatory state through the production of SCFAs ( Jiang et al. 2025 ). Eubiosis can be restored by promoting the enrichment of keystone species ( Wu et al. 2022 ). Identifying keystone species in healthy pregnancy and tracking their shifts across gestation is crucial for developing microbiome-based strategies to improve pregnancy outcomes. Dietary interventions with complex carbohydrates, for example, may increase the abundance of SCFA-producing taxa.

Pregnancy involves profound systemic adaptations, including maternal energy metabolism and endocrine and immune function ( Chang & Streitman 2012 , Hunt et al. 2025 ). These dynamic adaptations enable maternal adjustment to pregnancy, preventing fetal rejection, thereby supporting successful implantation, adequate placentation, and pregnancy progression ( Armistead et al. 2020 , Sharma et al. 2022 ). Changes in hormones, primarily produced by the placenta, mediate fluctuations in maternal circulating nutrient concentrations and vaginal pH. Maternal gut and vaginal microbiomes concurrently change with advancing gestation although the magnitude of change may be shaped by baseline community functionality, which is influenced by environmental and nutritional factors ( Koren et al. 2012 , Severgnini et al. 2022 , Tang et al. 2022 , Baud et al. 2023 ). The microbiome constitutes 10–100 trillion microorganisms, including bacteria, archaea, viruses, protozoa, and fungi, along with their collective gene pool, metabolic products, and structural components. These occupy specific niches within the host, including the skin, mouth, gut, and urogenital tract ( Fig. 1 ) ( Turnbaugh et al. 2007 ). The microbiome responds to changes in nutrients and pH by producing signalling molecules that influence host immune, neuroendocrine, and metabolic pathways, potentially impacting pregnancy physiology ( Corrigan & Gründling 2013 , Cereija et al. 2021 ). The diverse microbial protein-coding gene pool, which is 100-fold higher than that of the human host, multiplies the metabolic and signalling capacity of the host ( Qin et al. 2010 ). Thus, identifying factors that shape microbial communities, their functions, and molecular outputs is crucial to improving pregnancy physiology and outcomes ( Zhu et al. 2025 ). Microbial community niches and components important in pregnancy physiology and outcomes (image created in BioRender).

The authors declare the following competing interests: C-JDS received research support from Roche Diagnostics Ltd for studies of diagnostics and screening for FGR and preeclampsia. C-JDS received non-financial support from Illumina and support from Pfizer and Sera Prognostics. Cambridge Enterprise (UK) has filed patents relating to the prediction of pregnancy outcome with C-JDS as one of the named inventors. The other authors have no competing interests to declare.

RGL contributed to writing the original draft of the manuscript and preparation of figures. BJA contributed to manuscript writing. C-JDS contributed to manuscript writing, concept development, and funding acquisition. D-WPE led the conceptual framework, contributed to manuscript writing and preparation of figures, and secured funding. All authors reviewed and approved the final version of the manuscript.