Constraint-Based Modeling Reveals the Distinct Metabolic Potential in Gut Microbiomes of Inflammatory Bowel Disease Patients with Dysbiosis

The human gut microbiome plays an important role in human health. It is well known that inflammatory bowel diseases (IBD) are associated with microbial dysbiosis and altered fecal metabolomics profiles, however, the etiology of this complex, multifactorial disease remains poorly understood.

Recently, we published AGORA, a resource of curated genome-scale metabolic reconstructions for 773 common gut microbial strains. Moreover, we have published a COBRA Toolbox extension that enables the creation of personalized microbiome models from metagenomics data, as well as a high-performance implementation of constraint-based modeling in Julia. Here, we systematically profile the metabolic capabilities of gut microbiomes in health and disease states.

We retrieved strain-level relative abundances from metagenomics data from a cohort of pediatric Crohn’s Disease patients with and without dysbiosis and healthy control children (108 samples in total) and constructed a personalized microbiome model for each sample using the corresponding AGORA reconstructions. Subsequently, we predicted for each microbiome model the quantitative biosynthesis potential for all secreted metabolites as well as the strain-level contributions to each metabolite in each individual microbiome.

In total, 182 metabolites from diverse subsystems could be secreted by at least one individual microbiome, of which 124 had statistically significantly different production potential in dysbiotic IBD patient microbiomes. The production potential for potentially detrimental metabolites such as lactate, sulfide, trimethylamine, and acetaldehyde was increased in dysbiotic IBD microbiomes. These metabolites were mainly produced by Proteobacteria representatives in this group but not in non-dysbiotic IBD or healthy microbiomes. In contrast, dysbiotic IBD microbiomes were depleted in production potential of butyrate, and B vitamins due to lower abundances of the contributing taxa.

In summary, we present an efficient computational approach to systematically interrogate individual microbiome models. This approach enables us to mechanistically link disease-relevant metabolites with gut microbial taxa known to play a role in IBD.