Engineering yeast shikimate pathway towards production of aromatics: rational design of a chassis cell using systems and synthetic biology

Abstract

Aromatic building blocks are amongst the most important bulk feedstocks in the chemical industry. As these compounds are commonly derived from petrochemistry, obtaining them is becoming more and more a matter of costs and sustainability. Biochemistry gives rise to a wealth of compounds that can potentially replace current petroleum-based chemicals or be used for novel materials. The aromatic compounds para-aminobenzoic acid (pABA) and para-hydroxybenzoic acid (pHBA) and the aromatics derived compound cis,cis-muconic acid (ccMA) can be precursors for, but are not limited to, terephthalic or/and adipic acid. These are essential feedstocks for the production of PET and nylon. The three compounds can be derived from the shikimate pathway, an anabolic pathway leading to the biosynthesis of aromatic amino acids, present in certain prokaryotes and eukaryotes, including fungi. By combination of metabolic modelling with genetic engineering, a microbial production system based on the yeast Saccharomyces cerevisiae can be designed, which effectively channels flux into the target compounds. In order to develop a competitive bio-based process, yields, titers and rates need to be maximized. While productivity or rates in a process can be altered using genetic engineering, carbon yield, and pathway feasibility are stoichiometrically and thermodynamically predetermined. Both limitations need to be considered when designing a microbial production system. For formation of adipic acid and precursors many bio-based routes exists. More rational than just picking one for in vivo studies rather all available biochemical pathways were examined in silico using metabolic modelling. To compare theoretical yields and reaction thermodynamics an interface that allowed network-embedded thermodynamic analysis of elementary flux modes was developed. This allowed distinguishing between thermodynamically feasible and infeasible flux distributions. Feasible maximum theoretical product carbon yields were substantially different in E. coli and S. cerevisiae metabolic models and ranged from 32% to 92%. Further, many pathways appeared to be restricted by a thermodynamic equilibrium lying on the substrate side, some even infeasible. The only routes that deliver significant product yields and were thermodynamically favoured were shikimate pathway based. Being currently of strong scientific interest, recent implementations of these pathways in E. coli and S. cerevisiae were evaluated and strain construction strategies optimized, using the concept of constrained minimal cut-sets. Especially in S. cerevisiae a single non-obvious knock-out target allowed coupling of growth to product formation; in particular, the deletion of the pyruvate kinase reaction resulted in a minimum yield constraint of 28%. Though unique to shikimate pathway, this strategy is transferrable to other products which are derived from chorismate and also involve the formation of pyruvate as a by-product. This applies to pHBA. With further optimizations, the strategy was applied in vivo for the production of this compound in S. cerevisiae. As the pyruvate kinase knock-out entails a growth defect on glucose, a synthetic circuit was used, which allowed conditional knock-down and activation of the determined genetic modifications and by this dynamic control of the phenotype. Thus, production could be separated from growth and it could be proven that the in silico determined genetic intervention strategy holds valid in vivo, resulting in a 1.1 mM final product titer. Further, production of pABA from shikimate pathway in S. cerevisiae was investigated and optimized: Several alleles from different yeast strains of the genes (ABZ1 and ABZ2) for pABA formation from chorismate were screened, using a strain genetically engineered to channel flux to chorismate. ABZ1 of AWRI1631 and ABZ2 of QA23 delivered the highest pABA production. To further increase production, the impact of carbon-source on product yield was investigated in silico using metabolic modelling. It was found that combined glycerol-ethanol was a superior carbon-source than glucose, glycerol or ethanol alone, especially when employed in molar ratios between 1:2 and 2:1. This could be confirmed in vivo with carbon yields reaching up to 2.64%. A fed-batch process on glycerol-ethanol delivered total aromatics titers as high as 1.73 mM. It could be shown that feasibility and viability of adipic acid production greatly depends on the pathway and the organism chosen for engineering. Weaknesses of existing strain construction strategies for ccMA production could be identified and a radical optimization strategy was determined. Inferences from metabolic modelling were proven experimentally for pHBA and pABA production. The obtained concentrations and yields are the highest in S. cerevisiae to date and among the highest in a microbial production system, underlining the potential of yeast as a cell factory for renewable aromatic feedstocks.