Inferring Metabolic Mechanisms of Interaction within a Defined Gut Microbiota

Within the gastrointestinal tract of mammals, the diversity and number of species present enable a multitude of metabolic interactions. Methods for predicting these interactions are attractive because they may enable engineering of microbiome function. However, identifying the mechanism and consequences of metabolic interactions between even two species is incredibly challenging. In this work, we developed, applied, and experimentally tested a framework for identifying potential metabolic mechanisms associated with interspecies interactions.

We performed pairwise co-culture growth experiments using bacterial species a model mouse microbiota. After 72 hours of growth, we quantified the abundance of each strain and determined metabolite consumption and production using untargeted supernatant metabolomics. We applied our framework, which we call the Constant Yield Expectation (ConYE) model, to dissect emergent metabolic behaviors that occur in co-culture.

The ConYE model assumes the yield of a metabolite is constant in monoculture and co-culture. Using ConYE, we identified widespread indications of increased efficiency of biomass production in co-culture. We interrogated an amino acid crossfeeding interaction that is likely to confer a growth benefit to one ASF strain (Clostridium sp. ASF356) in coculture with another strain (Parabacteroides goldsteinii ASF519). We attempted in silico identification of growth-enhancing metabolites by constructing genome-scale metabolic network reconstructions for the ASF strains. Using the simulations together with the ConYE results, we designed media supplementation experiments and verified that the proposed interaction leads to a growth benefit for this strain.

Our results reveal the types and extent of emergent metabolic behavior in microbial communities and demonstrate how metabolomic data can be used to identify potential metabolic. Although we focus on growth-modulating interactions, the framework we develop can be applied to generate specific hypotheses about mechanisms of interspecies interaction involved in any phenotype of interest within a microbial community.