| The engineering of biological systems for industrial applications is a complex process. Optimization of a cellular phenotype is complicated by the sheer number of genetic and environmental variables. With regards to heterologous natural product biosynthesis, this is further complicated by the foreign nature of the metabolic pathways and structurally complex products involved. To address this problem, heuristic and systematic approaches were employed for the engineering of two heterologous natural products (a polyketide and an isoprenoid) in Escherichia coli. The methods developed and applied herein are critical to advancing the field of heterologous natural product biosynthesis to the scale of competitive industrial bioprocesses.;Stoichiometric modeling was applied to survey heterologous hosts for supporting polyketide biosynthesis. Simulations under different host and environmental conditions revealed multiple gene knockouts that were capable of improving product titer. Work has shown that multiple pathways exist in nature for producing the two precursors necessary for polyketide production; however, E. coli does not possess these. These heterologous pathways were expressed, and with concurrent substrate feeding experiments, their effects were analyzed on polyketide production. Native gene over-expressions and deletions also improved polyketide titer.;vii Due to an inability to thoroughly search genomic space with the aforementioned computational method, a new algorithm was developed to identify knockout targets based on network topology and applied to isoprenoid production. By using a genetic algorithm, this method identified a four knockout strain capable of improved titer, while reducing computation time by several orders of magnitude. When constructed in the laboratory using an accelerated genome evolution method, isoprenoid yield improved nearly 3-fold in some cases. The aforementioned algorithm was reformulated in an attempt to identify over-expression targets for improving isoprenoid titer. This method identified four targets, three of which improved titer when implemented genetically, though failed to meet the predicted levels of improvement. Upon over-expression of the isoprenoid biosynthetic pathway genes, one gene improved titer to a higher extent than the predicted targets (almost 4-fold), showing that the rate-limiting step lies within the pathway itself. Applying heuristics for isoprenoid production, heterologous gene promoter strength, strain background, and process-related parameters were varied and allowed for a 240-fold improvement in titer. |
