Please use this identifier to cite or link to this item: http://hdl.handle.net/1946/22558
Humans harbor a complex microbial ecosystem in their intestines. This human gut microbiota performs essential functions for host health and wellbeing. Disruption, or dysbiosis, in gut microbiota composition has been directly linked to the pathogenesis of complex diseases. Understanding the mechanisms behind these links between microbiota and host health and disease states requires predictive, mechanistic computational models.
Constraint-based metabolic modeling uses manually curated and validated metabolic reconstructions that constitute a genomically, genetically and biochemically structured knowledge base of a target organism. These reconstructions can be converted into mathematical models that allow the in silico prediction of an organism’s phenotype under given environmental constraints Constraint-based modeling has been successfully
applied to predict multispecies interactions but rarely in the context of host-microbe interactions. In this thesis, a framework for predicting metabolic interactions between a host and its gut symbionts was developed. As reconstructions of representative gut symbionts were lacking prior to this thesis, metabolic reconstructions targeting representatives of the two main phyla in the gut microbiota were manually constructed and validated. These reconstructions were first applied to predict the metabolic interactions between a model gut symbiont and its murine host, and, using a combined in silico/ in vitro approach, the metabolic potential of an abundant but poorly studied beneficial microbe. Subsequently, using nine additional published manually curated reconstructions, a model gut community was constructed in silico. Linking the model community with a global reconstruction of a human cell enabled the systematic prediction of the microbes’ potential to affect human metabolism. The in silico approach demonstrated the gut microbiota’s potential to act as an additional organ from the host perspective. Moreover, the 11 microbes were joint pairwise in all combinations and mutualistic, commensal, parasitic, and competitive interactions were investigated on varying nutrient environments. The potential to engage in mutualistic cross-feeding was species-specific and predicted to be highly dependent on anoxic conditions.
In summary, using constraint-based modeling, host-microbe and microbe-microbe metabolic interactions were predicted in a bottom-up, mechanistic manner. The multispecies modeling framework developed in the course of this thesis can readily incorporate any host and number of microbes and will have valuable applications for elucidating the gut ecosystem and its effects on the human host.