Fully biological production of adipic acid analogs from branched catechols

Microbial production of adipic acid from lignin-derived monomers, such as catechol, is a greener alternative to the petrochemical-based process. Here, we produced adipic acid from catechol using catechol 1,2-dioxygenase (CatA) and a muconic acid reductase (MAR) in Escherichia coli. As the reaction progressed, the pH of the media dropped from 7 to 4-5 and the muconic acid isomerized from the cis,cis (ccMA) to the cis,trans (ctMA) isomer. Feeding experiments suggested that cells preferentially uptook ctMA and that MAR efficiently reduced all muconic isomers to adipic acid. Intrigued by the substrate promiscuity of MAR, we probed its utility to produce branched chiral diacids. Using branched catechols likely found in pretreated lignin, we found that while MAR fully reduced 2-methyl-muconic acid to 2-methyl-adipic acid, MAR reduced only one double bond in 3-substituted muconic acids. In the future, MAR’s substrate promiscuity could be leveraged to produce chiral-branched adipic acid analogs to generate branched, nylon-like polymers with reduced crystallinity.

Scientific RepoRtS | (2020) 10:13367 | https://doi.org/10.1038/s41598-020-70158-z www.nature.com/scientificreports/ showed that switching feedstock from glucose to lignin monomers reduced adipic acid minimum selling price by 50% due to increased productivity and decreased feedstock cost 12 . Taken together, lignin is a more desirable feedstock than sugars for adipic acid production. Lignin depolymerisation results in a number of aromatic compounds, including catechol, a key intermediate in the production of adipic acid from both glucose and lignin-derived aromatics such as ferulic acid and p-coumaric acid 1 . Niu et al. achieved the fully biological production of adipic acid from lignin-derived aromatics in P. putida. Specifically, 4-hydroxybenzoic acid was converted to catechol and subsequently 3-ketoadipoyl-CoA via the β-ketodipate pathway. Using three heterologous enzymes and one endogenous thioesterase, 3-ketoadipoyl-CoA was converted to adipic acid with 17.4% molar yield 13 . Sun et al. developed a shorter pathway in E. coli, where glucose is metabolized to catechol via the shikimate pathway and catechol is converted to adipic acid using catechol-1,2-dioxygenase (CatA) and an enoate reductase previously shown to reduce muconic acid to adipic acid (i.e. muconic acid reductase, MAR) 14 . This pathway achieved 80.6 µg/L/hr of adipic acid 15 . Therefore, a better understanding of the CatA-MAR enzyme cascade can be directly applied to adipic acid production pathways that route through catechol.
Here, we optimized the CatA-MAR enzyme cascade in E. coli and discovered that the cascade is capable of converting branched catechols to adipic acid analogs (Fig. 1a). First, we optimized muconic acid production from catechol by screening and optimizing the expression of CatAs from five different organisms. Next, with the optimal CatA in hand, we maximized adipic acid production by optimizing the co-expression of two known MARs. Interestingly, during muconic acid production, the media acidified leading to the isomerization of ccMA to cis,trans-muconic acid (ctMA). To determine the extent to which MAR reduced ctMA, we fed ccMA and ctMA to cells expressing MAR and observed preference to reduce ctMA over ccMA. However, when we fed ccMA and ctMA to the cell lysate, MAR showed no preference between ccMA and ctMA. This data suggests that ctMA is preferentially transported into the cell over ccMA. Given the promiscuity of MAR, we explored its utility by feeding the CatA-MAR cascade branched catechols likely found in lignin 16 to produce chiral-branched diacids. We found that MAR reduced one double bond in muconic acid analogs with short alkyl chains at the 3-position and both double bonds of 2-methyl-muconic acid to produce 2-methyl-adipic acid. The promiscuity of MAR highlights the potential utility of the CatA-MAR enzyme cascade in lignin valorization, as lignin depolymerisation results in a heterogeneous mixture of aromatics.

Strains and plasmids.
Screening catA for muconic acid production from catechol. Overnight  production of adipic acid analogs from lignin-derived aromatic monomers.  (Fig. 1d), catechol and muconic acid were quantified using a multi-wavelength detector (MWD) (274 nm catechol, 260 nm muconic acid). Quantification was done using the area under the peak for the corresponding MWD signal. For MAR screening (Fig. 2d)

Results and discussion
catechol to muconic acid conversion. CatAs from Pseudomonas putida, Pseudomonas aeruginosa, Acinetobacter calcoaceticus, Acinetobacter baylyi, and Acinetobacter baylyi ADP1 have been used to produce muconic acid in E. coli 6,[21][22][23] . We compared the E. coli performance of CatAs from P. putida, Acinetobacter baylyi ADP1 (both dimers with moderate specific activity-P. putida 22.4 µM/min/mg), Candida albicans (dimer with a higher specific activity, 63 µM/min/mg), Rhodococcus sp. AN22 (a monomer) and Rhodococcus opacus (structurally well studied) 8,24-28 ( Supplementary Figs. S2, S3). The CatAs were expressed from three promoters with different strengths in low and medium plasmid copy number (Fig. 1b-d, Fig. S4). All CatAs performed similarly independent of promoter strength and plasmid copy number, reaching almost 100% catechol to muconic acid conversion after 24 h (Fig. 1d). Of note, Acinetobacter baylyi ADP1 CatA showed < 10% conversion in half of the tested conditions despite similar protein expression ( Supplementary Fig. S5). Active site alignment of the ADP1 CatA and P. putida CatA revealed that at position 76 ADP1 CatA codes for a proline, while P. putida CatA codes for an alanine ( Supplementary Fig. S6). A proline to alanine mutation at this position has been shown to increase ADP1 CatA specific activity 10 fold 27 . Therefore, P76 in ADP1 CatA may lead to the underperformance observed when compared to the other CatA homologs. For all subsequent experiments, we used Rhodococcus sp. AN22 CatA as it performed well and it is a monomer, reducing the metabolic load of the system. E. coli expressing AN22 CatA converts catechol to ccMA as the major product. However, we also detected the thermodynamically stable isomer ctMA, which is likely produced from ccMA after media acidification after 24 h growth, which reaches a pH 4.3 after starting at pH ~ 7 (Fig. 1b,c). We measured the maximum ccMA to ctMA isomerization rate to be between pH 3-5 ( Supplementary Fig. S7), with previous literature confirming the maximum at pH 4 17 . No isomerization of ctMA to ttMA was observed, which is consistent with the literature 17 .
Muconic acid to adipic acid conversion. We compared the E. coli performance of Bacillus coagulans MAR (MAR-BC) and Clostridium acetobutylicum MAR (MAR-CA) 14 expressed from two promoters with different strengths in low, medium and high plasmid copy number (Fig. 2a-c). MAR-BC achieved a 6.1% conversion of 500 µM ccMA to adipic acid using the pTrc promoter from a medium-copy plasmid after a 24 h anaerobic growth. This is lower than Sun et al. who used batch fermentation to convert 2.8 mM of ccMA to adipic acid (18.0% total conversion) using MAR-CA 15 . We rationalize our lower percent conversion in comparison by the use of minimal M9 media supplemented only with 0.5% glucose rather than a modified M9 medium, which contains 0.5% yeast extract, 0.25% glucose and 1% glycerol that aids in higher protein expression and increased cell density.

Medium Copy Plasmid Low Copy Plasmid
High Copy Plasmid Adipic Acid Yield (mol%) Thus, fermentations were run as a two-stage batch process with a 2 h aerobic stage followed by a 22-h anaerobic stage 29 . The CatA-MAR cascade was tested in a two-and one-plasmid system. As Fig. 2d shows, the one-plasmid system resulted in 1.6 mg/L of adipic acid after 24 h, or a 0.241% molar yield from the fed 1 g/L catechol, an 18-fold improvement over the two-plasmid set up.
MAR muconic acid isomer preference. At pH < 7, ccMA isomerized to ctMA, and both isomers were present at a roughly equal molar ratio in the fermentation broth (Fig. 2b). Hypothesizing that MAR may not reduce ctMA as efficiently as ccMA, we fed ccMA, ctMA and, for completion, ttMA to cells expressing MAR and to a cell lysate expressing MAR. In the cell-based experiment, both MARs showed a preference to reduce ctMA over ccMA with MAR-CA showing an eightfold preference. MAR-BC also showed a higher adipic acid yield than MAR-CA; 11.5-fold in the case of ccMA, and twofold in the case of ctMA (Fig. 3a). In the cell lysate-based experiment, both MARs reduced ccMA and ctMA to the same extent. Interestingly, MAR-CA showed a higher adipic acid yield than MAR-BC (Fig. 3b). MAR-CA also showed a preference for ttMA, which is consistent with previous literature 14 . Of note, the overall yields in the cell lysate experiment were lower than the cell-based experiment, likely due to enzyme deactivation from cell lysis solution components and the reduced co-factor concentration. Taken together, we rationalize the in vivo MAR substrate preference for ctMA to be the result of increased membrane permeability for ctMA over ccMA. We rationalize the higher in vivo activity of MAR-BC over MAR-CA to the fact that once a muconic acid isoform enters the cell the MAR experiences a higher localized substrate concentration. Previously, it has been suggested that MAR-BC has a lower substrate affinity but higher catalytic activity than MAR-CA 14 . Such MAR-BC enzymatic characteristics would fit the observed results.

conclusions
The fully biological production of adipic acid from catechol, a pretreated lignin monomer, was achieved by screening CatA from different sources and optimizing its co-expression with MAR in E. coli to produce adipic acid at 1.6 mg/L or a 0.241% molar yield. The muconic acid yields presented in this work were lower than in previous studies, which we attribute to the use of batch fermentation rather than a biocatalysis set up and the use of minimal media rather than rich media, which makes it problematic to calculate yields. A closer study of oxygen sensitivity differences between MAR-BC and MAR-CA may help to engineer a more oxygen tolerant enzyme that will be useful in the production of adipic acid from lignin-derived monomers as oxygen is both a substrate (CatA) and an inhibitor (MAR) of the process. Engineering MAR for oxygen tolerance will help improve catechol to adipic acid yields in the future.
A key finding of this work is the use of MAR to produce branched adipic acid analogs. Thus, the CatA-MAR cascade could be used to convert lignin-derived monomers to chiral branched dicarboxylic acids that may give tuneable properties to nylon-6,6 like polymers. Application of the enzyme cascade to a variety of lignin-derived monomers demonstrates increased utility as a lignin valorization approach.