Combining elementary mode analysis with a network embedded thermodynamic approach for analysis of microbial adipic acid production

Stoichiometric and thermodynamic constraints applied for non-obvious knock out strategies have unexpected significance for pathway choice

Adipic acid is a main precursor for the production of nylon-6,6 and polyurethanes. To date it is still derived from the precursor benzene in a process of oxidation of cyclohexane to cyclohexanol/cyclohexanone. While these precursors are all petro-chemistry based and thus non-sustainable, the further oxidation with nitric acid has an additional environmental impact as the greenhouse gas N2O is a large by-product in this process.
The bio-production of adipic acid offers a promising alternative to chemical synthesis. There
is currently a strong interest in the metabolic engineering community (both commercial and scientific) to develop a biological replacement process. There are mainly three types of approaches that can be distinguished: (i) Direct production of adipic acid, (ii) production of the unsaturated adipic acid precursor cis,cis-muconic acid and (iii) production of D-glucaric acid. Muconic and glucaric acid are subsequently reduced to adipic acid by chemical hydrogenation.
In the first part of this study, all possible sub-routes of the plenty of available pathways to adipic acid were identified. In total 16 different production routes based on dextrose or fatty acids as carbon-feed exist. These routes were compared under aerobic and anaerobic conditions in E. coli and S. cerevisiae, using a new approach combining elementary flux mode analysis with network embedded thermodynamic analysis, for feasibility and maximum yield.
One approach that is currently of particular attraction is to derive the precursor muconic acid from shikimate pathway. After early works by the Frost laboratory in the 1990's there has been a renaissance in interest in this topic as can be seen from recently published work by three independent laboratories (Boles, Alper and Yan). We noticed that all three approaches suffer from very low product titres (in the mg / L range) and decided to complement our study with an in silico comparison of the strain construction strategies.
Among the 16 different routes the theoretical mass yields ranged from 0.07 to 1.16 g g-1 on glucose and/or palmitate, highlighting the importance of pathway choice. Infeasibility patterns were determined for each pathways flux distribution, revealing that all but one pathway contained reactions that limit product formation through a thermodynamic equilibrium lying on the substrate side. Almost half of these appeared to have an equilibrium that deemed the whole pathway infeasible. This problem also affected recently patented routes. The pathway feasible over a large range of concentrations and pH values was the
production of muconic acid via the shikimate pathway derived tryptophan precursor
anthranilic acid (496 mgmuconic ggluco -1).
In the second part of this study we used in depth elementary mode analysis of shikimate pathway based routes to determine new knock-out strategies to enhance muconic acid production.
We found that the availability of sensible knock out targets depends not only on the pathway but also on the organism, allowing significantly different maximum yields. In addition it was possible to create a scenario where product formation is directly coupled to central metabolism, allowing the introduction of a minimum yield constrain of 45%. This is particularly applicable in S. cerevisiae.
While this is an in silico analysis, we believe it adds an important angle to the current efforts in creating bio-based muconic acid production and will help researchers to prioritize their research efforts.
The outcomes may also be applicable to other target compounds in different pathways where similar bioconversion steps are involved.