Towards a platform organism for terpenoid production – in silico analysis of metabolic networks of potential hosts and in vivo validation

Abstract

Terpenoids are mostly plant derived compounds with industrial and medicinal applications. These compounds can be provided via biotechnological production as environmental friendly alternative to chemical synthesis or plant extraction. However, productivities from microbial hosts require improvement in order to be economically competitive. Therefore, microbial hosts are compared and new metabolic engineering targets aimed at increasing terpenoid production are investigated by stoichiometric metabolic network analysis and selected strategies are validated in vivo. The in silico analysis of the metabolic networks of Escherichia coli and Saccharomyces cerevisiae (yeast) led to several promising metabolic engineering targets with potential to increase the theoretical maximum terpenoid yield. Production of the sesquiterpenoid patchoulol in yeast was chosen for the in vivo validation study. A two-phase cultivation method for terpenoid capture with dodecane overlay was established and the produced spectrum of sesquiterpenoids was determined. Precursor metabolite flux towards the desired product was increased via overexpression of a truncated HMG-CoA reductase and fusion of FPP synthase with patchoulol synthase. Two in silico identified strategies were implemented: (i) disruption of α-ketoglutarate dehydrogenase gene redirected the metabolic flux as predicted, however, the intermediate acetate was produced in high amounts instead of the desired product; (ii) expression of ATP-citrate lyase from Arabidopsis did not increase terpenoid production due to insufficient in vivo activity. The findings of this thesis contribute to an increased knowledge about enhancing terpenoid production in both E. coli and S. cerevisiae, as well as metabolic behavior of yeast. The in silico stoichiometric metabolic network analysis can be used successfully as a metabolic prediction tool. This study highlights that kinetics, regulation and cultivation conditions may interfere with predictions, resulting in poor in vivo performance. These findings promote developments of metabolic modelling to increase the predictive power and accelerate microbial strain improvement.