PROJECT SUMMARY
The gut microbiome aids in the digestion of complex polysaccharides, the absorption of vitamins, and the
conversion of primary bile acids, drugs, and other bioactive compounds into metabolites that can be absorbed
by the host. Thus, the metabolic activity of commensal microbes is closely intertwined with human physiology
and the nutritional impact of our diet. However, there is limited understanding of how variation in the ecology of
our intestinal flora modulates the biological impact of diet on human health and nutrition. Recent work has shown
that differences in the composition of the gut microbiome can help explain person-to-person heterogeneity in
glycemic responses, blood lipid profiles, and weight loss. In this proposal, we present an innovative platform for
personalized metabolic modeling of the gut microbiome using metagenomic and dietary data as constraints. We
propose the integration of tissue-resolved metabolic models of relevant host tissues, including a cell-type-specific
metabolic reconstruction of the gut epithelium, into our existing microbial community model to improve estimates
of metabolic fluxes between the gut microbiota, the diet, and the host. We will call this host-diet-microbiome
metabolic model ‘CyberGut.’ Using existing multi-omic data from a cohort of >3,000 adults, we will constrain and
validate CyberGut with paired measurements of diet, host blood metabolomes, and gut microbiomes. In addition,
we will generate cross-sectional training and validation data consisting of paired blood and fecal metabolomes,
fecal microbiomes, and detailed 3-day dietary recall data from a new cohort of 100 healthy participants. Using
these data, we will refine and test two novel and independent diet-inference algorithms, which leverage stool
metagenomes and stool untargeted metabolomes, respectively. Furthermore, using samples taken from a subset
of this new cohort (N=40), we will perform ex vivo stool culturing experiments, designed to directly quantify
metabolic fluxes and bacterial growth rates in vitro. These fluxomic data will be used to directly test in silico
CyberGut flux predictions in response to a diverse panel of dietary and host metabolite interventions. In addition
to contributing to the refinement and testing of our CyberGut model, the paired diet, microbiome, and
metabolomic data, including replicate fluxomic assays, generated in this proposal will be an invaluable resource
to the precision nutrition and human microbiome research community. In summary, we will build, refine, and test
a novel platform for tracking dietary intake and predicting personalized nutritional responses to diet, which has
the potential to fundamentally alter how we design and test dietary interventions.
