High throughput genome scale modeling predicts microbial vitamin requirements contribute to gut microbiome community structure

Abstract

Human gut microbiome structure and emergent metabolic outputs impact health outcomes. However, what drives such community characteristics remains underexplored. Here, we rely on high throughput genomic reconstruction modeling, to infer the metabolic attributes and nutritional requirements of 816 gut strains, via a framework termed GEMNAST. This has been performed in terms of a group of human vitamins to examine the role vitamin exchanges have at different levels of community organization. We find that only 91 strains can satisfy their vitamin requirements (prototrophs) while the rest show various degrees of auxotrophy/specialization, highlighting their dependence on external sources, such as other members of the microbial community. Further, 79% of the strains in our sample were mapped to 11 distinct vitamin requirement profiles with low phylogenetic consistency. Yet, we find that human gut microbial community enterotype indicators display marked metabolic differences. Prevotella strains display a metabolic profile that can be complemented by strains from other genera often associated with the Prevotella enterotype and agrarian diets, while Bacteroides strains occupy a prototrophic profile. Finally, we identify pre-defined interaction modules (IMs) of gut species from human and mice predicted to be driven by, or highly independent of vitamin exchanges. Our analysis provides mechanistic grounding to gut microbiome stability and to co-abundance-based observations, a fundamental step toward understanding emergent processes that influence health outcomes. Further, our work opens a path to future explorations in the field through applications of GEMNAST to additional nutritional dimensions.

Keywords

• Gut microbiome
• genome scale modeling
• computational biology
• networks
• interactions
• cofactors