Genetic and environmental perturbation effects on metabolic networks and engineering objectives

Abstract

Phenotypic variation is produced through a complex web of interactions between genes and the environment and determines whether a trait has the ability to respond to natural or artificial selection. In metabolism, these concerted interactions define how energy is supplied and distributed throughout the cell. Since metabolic engineers seek to use cellular processes to improve the synthesis of valuable biochemical substances, there is great interest among scientists and engineers alike in determining the relative importance of genetic and environmental factors. To study the environmental and genetic effects on metabolic engineering applications, I use constraint-based metabolic modeling to develop a computational framework that systematically simulates and analyzes extracellular and intracellular perturbations on the metabolism of three microorganisms: Escherichia coli, Saccharomyces cerevisiae, and Shewanella oneidensis. Media compositions and gene- deletion strains are designed to optimize single or multiple engineering objectives, such as the maximization of production rate, yield and purity, or the minimization of the economic cost of raw materials. I use the framework to evaluate the production of several industrially important chemical commodities such as acetate, D-lactate, hydrogen, ethanol, formate, and succinate. By evaluating over 435 million simulated conditions and using 36 engineering metabolic traits, I classify the resultant phenotypes into 10-30 dominant meta-phenotypes for each organism. The meta-phenotypes illustrate global phenotypic variation and highlight organism-specific differences in biological and engineering capabilities. I show biological causality of high-performance engineering phenotypes and make available a web-based tool that was developed to permit public queries and visualization of optimal engineering designs and resultant metabolic pathway activities. Finally, I discuss relationships between engineering traits and phenotypes, trade-offs among multiple engineering objectives, and differences in phenotypic sensitivities that depend on perturbation type. Given the increasing number of sequenced genomes, model accuracy and available computing power, it is foreseeable that the developed framework can be extended to query a growing range of organisms, phenotypic variability, engineering applications and biological insights.