Optimization based redesign of microbial production systems
Abstract (Summary)iii The primary objective of this research is to develop computational tools for guiding experimental strain engineering strategies. The key questions addressed in this work are (i) how can optimal gene deletions be selected so that biochemical production in a stoichiometric network is coupled to biomass formation?, (ii) alternatively, how can one select the recombination candidates from a database of biotransformations to confer the ability to produce a specific chemical from an optimal substrate in a host organism?, (iii) what are the best reaction candidates for modulation to enhance the yields of biochemicals in a microbial network, and (iv) finally, what are the smallest enzyme sets that can be modulated to achieve the maximum possible enhancement of a specific reaction flux if a detailed kinetic model is used to describe metabolism? This thesis begins with the description of a bilevel framework introduced to identify not only optimal reaction deletions but also the optimal transport rates of key metabolites so that complex compounds can be produced as an obligatory byproduct of growth. Case studies involve prediction of deletion strategies for different amino acids in Escherichia coli. Next, an integrated framework, OptStrain is presented to uncover and investigate different alternative pathways in conjunction with the examination of multiple organisms and substrates for selecting the best strategy for producing a target metabolite. These alternative pathways are identified from a database of reactions compiled from publicly available biopathway databases and stoichiometric models of metabolism. Results include manipulation strategies for overproducing hydrogen in networks of vastly different microbial organisms such as E. coli, Clostridium acetobutylicum and iv Methylobacterium extorquens, and for vanillin production in E. coli. The array of in silico genetic manipulations that can be predicted by using optimization frameworks is completed by the OptReg framework which can predict inhibition, up regulation and deletion of reactions for strain redesign. The applicability of this tool is demonstrated for ethanol overproduction in Escherichia coli. Finally, a kinetic model of the central metabolism of E. coli is examined for elucidating the enzyme sets that can favorably influence the flux towards serine synthesis and through the phosphotransferase uptake system in the network. The broad array of genetic manipulation strategies identifiable though the proposed frameworks highlights their utility as efficient strain design tools.
School Location:USA - Pennsylvania
Source Type:Master's Thesis
Date of Publication: