First and later pregnancies differ biologically, yet most studies of the maternal and infant microbiome treat them as interchangeable. My research asks how a mother’s pregnancy history and diet shape the gut and microbiota her child starts life with, and which of those early differences last into childhood and later life. The work combines human birth cohorts, mouse models of developmental programming, and methods built for low-biomass and longitudinal microbiome data.
Pregnancy history and the maternal–infant microbiome
The maternal gut microbiome adapts differently to a first pregnancy than to later ones, and birth order shapes neonatal colonization from the first days of life.
First-born children are smaller at birth, one sign that first and later pregnancies differ biologically. In a prospective cohort at Charité–Universitätsmedizin Berlin, I found that parity modifies how BMI and gestational weight gain relate to the maternal gut microbiota. Associations between excess adiposity and community composition were largely confined to first pregnancies, and multiparous participants followed different microbial trajectories. In a second pregnancy cohort (Bugs in BHIP), subcommunities of co-occurring taxa expanded and contracted differently in first and later pregnancies, and their postpartum recovery diverged. Pooling primiparous and multiparous participants can therefore obscure real biological signal.
The differences extend to the newborn. In neonates, delivery mode, birth order, and sex each independently shape early colonization, and birth order modifies how delivery mode and intrapartum antibiotic prophylaxis (IAP) affect colonization in the first days of life. In the Baby & Mi birth cohort (245 infants, sampled from 3 days to 3 years), I am asking whether these differences persist through early childhood and whether they track with growth and atopic outcomes.
Strain transmission at birth: the Good Bugs Study
Birth-order differences in the infant microbiome are present within the first day of life. Because the fetal gut has no detectable microbiota before birth, these differences likely arise from what the mother transmits during delivery. I designed and lead the Good Bugs Study, funded by a Farncombe Innovation Grant, which uses culture-informed metagenomics to track which maternal vaginal and gut strains colonize newborns in first and later pregnancies. The study banks isolates for functional testing, including strains that may limit mother-to-infant transmission of potential pathogens.
Maternal diet, the placenta, and the fetal gut
In mice, maternal high-fat diet disrupts the gut–placenta–fetus axis, and its effects on the offspring gut persist into adulthood.
My doctoral work used a mouse model of diet-induced obesity to trace how maternal diet affects the gut–placenta–fetus axis. Characterizing the maternal gut microbiome, maternal intestine, placenta, and fetal intestine in the same cohort revealed a chain running from diet to fetus: disrupted maternal intestinal adaptation to pregnancy, placental hypoxia, and altered barrier protein and immune marker profiles in the fetal gut. That chain is not visible in studies of a single compartment.
Follow-up work showed that the fetal intestinal inflammation depends on TNF signalling, and that the effects persist. In adulthood, offspring had altered gut microbiota composition, increased gut barrier permeability, and greater susceptibility to a postnatal dietary challenge.
Papers: Gohir, Kennedy et al. 2019, J Physiol · PhD thesis, McMaster University, 2021
Methods for low-biomass and longitudinal data
Meconium collected before birth has no detectable microbiota, so gut colonization begins at birth.
Reports of bacterial DNA in placenta, amniotic fluid, and meconium suggested that colonization begins in utero. Most of those studies lacked the contamination controls that low-biomass samples require. When a sample contains very little microbial DNA, reagent- and laboratory-derived sequences can dominate the result while still looking biologically plausible. Through a collaboration with Charité, I applied contamination-aware protocols to meconium collected immediately before caesarean delivery and found no detectable microbiota. I then re-analysed the published data behind a high-profile paper that I thought overstated the evidence for fetal microbial exposure, which became a peer-reviewed letter in Cell, and led a Nature perspective with 47 colleagues setting out principles for contamination-aware study design. That attention to technical artefacts goes back to my MSc work, which showed that PCR amplification introduces consistent, primer- and region-dependent biases in Illumina-based 16S rRNA gene profiling.
Longitudinal microbiome studies usually track individual taxa or diversity metrics over time. With McMaster’s Department of Mathematics and Statistics, I developed TALDA, which adapts latent Dirichlet allocation to model subcommunities of co-occurring taxa over time. In the Bugs in BHIP cohort it detected parity-dependent dynamics that conventional methods missed, and the approach applies to longitudinal microbiome data well beyond pregnancy.
From early life to healthspan
The question that runs through all of this work is which early-life differences last, and which can be changed. The next step is to follow them forward, from childhood growth and allergy to gut barrier function and health in aging, and to ask where along that path prevention could start.
Early-life programming is well established as a factor in who is susceptible to diet- and lifestyle-related disease. Whether it also shapes who benefits from interventions, such as changes in diet or exercise, is largely untested, and the answer could help match prevention to the people most likely to benefit.
