LAUSR

The gut microbiome is a complex microbial community that has a significant influence on the host. Microbial interactions in the gut are mediated by dietary substrates, especially complex polysaccharides. In this environment, breakdown products from larger carbohydrates and short chain fatty acids are commonly shared among gut microbes. Understanding the forces that guide microbiome development and composition is important to determine its role in health and in the intervention of the gut microbiome as a therapeutic tool. Recently, modeling approaches such as genome-scale models and time-series analyses have been useful to predict microbial interactions. In this study, a bottom-up approach was followed to develop a mathematical model based on microbial growth equations that incorporate metabolic sharing and inhibition. The model was developed using experimental in vitro data from a system comprising four microorganisms of the infant gut microbiome (Bifidobacterium longum subsp. infantis, Lactobacillus acidophilus, Escherichia coli, and Bacteroides vulgatus), one substrate (fructooligosaccharides, FOS), and evaluating two metabolic products (acetate and lactate). After parameter optimization, the model accurately predicted bacterial abundance in co-cultures from mono-culture data. In addition, a good correlation was observed between the experimental data with predicted FOS consumption and acid production. B. infantis and L. acidophilus were dominant under these conditions. Further model validation included cultures with the four-species in a bioreactor using FOS. The model was able to predict the predominance of the two aforementioned species, as well as depletion of acetate and lactate. Finally, the model was tested for parameter identifiability and sensitivity. These results suggest that variations in microbial abundance and activities in the infant gut were mainly explained by metabolic interactions, and could be properly modeled using Monod kinetics with metabolic interactions. The model could be scaled to include data from larger consortia, or be applied to microbial communities where sharing metabolic resources is important in shaping bacterial abundance. Moreover, the model could be useful in designing microbial consortia with desired properties such as higher acid production.