Resolving challenges in quantitative modeling of microbial community dynamics

Abstract

Microbial communities can perform biochemical activities that monocultures cannot. Controlling communities requires an understanding of community dynamics. Here, we mathematically predict the growth rate of an engineered community consisting of two S. cerevisiae strains, each releasing a metabolite required and consumed by the partner. Initial model parameters were based on strain phenotypes measured in batch mono-cultures with zero or excess metabolite, and failed to quantitatively predict experimental results. To resolve model-experiment discrepancy, we chemically identified the correct exchanged metabolites, but this did not improve model performance. We then re-measured strain phenotypes in chemostats mimicking the metabolite-limited community environments, while mitigating or incorporating effects of rapid evolution. Almost all phenotypes we measured varied significantly with the metabolite environment. Once we used parameters measured in community-like chemostat environments, prediction agreed with experimental results. In summary, using a simplified community, we uncovered, and devised means to resolve, modeling challenges that are likely general.