Coen Berns (University Of Amsterdam)

Summary of presentation:
Introduction:
In early life, the microbiome undergoes dramatic transitions driven by dietary changes and exposure to external microbes. Breastfed infants typically develop a Bifidobacteria-rich microbiome that metabolizes human milk oligosaccharides (HMOs). These carbohydrates can shape community composition and, thereby, short-chain fatty acid (SCFA) production. In our eight-species infant Bacterial Community (iBaCo), different HMO substrates led to shifts in species abundance and SCFA levels. Because it is difficult to test this mechanism in vivo and time-consuming to perform in vitro experiments, we use iBaCo as a benchmark to understand whether genome-scale metabolic models (GEMs) can explain the interaction mechanisms leading to the observed community-level SCFAs.
Methods:
We developed a reconstruction pipeline that generates GEMs from bacterial genomes and mono-culture growth data across HMO conditions. Curated models were used with flux balance analysis to predict SCFA secretion, and the eight GEMs were integrated into community models to screen for candidate cross-feeding metabolites. Selected species combinations were then simulated with dynamic flux balance analysis to capture time-dependent growth and metabolite exchange.
Results:
Under matched external conditions (yeast extract–casitone–fatty acid (YCFA) in silico medium), curated iBaCo GEMs reproduced experimental mono-culture SCFA secretion patterns more accurately than reference models (MCC 0.67 vs. 0.26). In a three-species community containing two Bifidobacterium species and the only butyrate producer Anaerostipes caccae, lactate was observed as a potential cross-feeding metabolite supporting butyrate production. Dynamic simulations further predicted that A. caccae could grow in silico on lacto-N-tetraose as the sole carbon source when paired with an HMO degrader, suggesting an interaction-dependent mechanism since A. caccae cannot degrade HMOs.
Conclusion:
Butyrate formation in early-life communities may rely on cross-feeding between HMO degraders and A. caccae. Using validated metabolic models in a defined synthetic community provides a framework to predict how specific microbes and substrates shape SCFA production.