Aminocaproic

One-Pot Biocatalytic Transformation of Adipic Acid to 6‑Aminocaproic Acid and 1,6-Hexamethylenediamine Using Carboxylic Acid Reductases and Transaminases

INTRODUCTION

Polymers and their precursors constitute the largest fraction of all chemical products manufactured today. The global polymer market was valued at USD 522.7 billion in 2017 and is expected to grow at 4.0% annually over 2019−2025 (Grand View Research).

The fastest growing polymer markets include polyamides and polyesters, with compound annual growth (CAGR) up to 8%.1 In addition, worldwide demand for synthesis by biocatalysis or fermentation can be expected to be feasible.

The global biopolyamide market was valued at USD 110.5 million in 2016 and is projected to increase at a CAGR of 12.9% from 2017 to 2025 (Grand View Research, online report 2017).1,2 Nylon-6 and nylon-6,6 account for approx 90% of the total amount of nylon produced today (worldwide nearly 7 million tons annually).1

Nylon-6 is a homopolymer of 6-aminocaproic acid (6-ACA), whereas nylon-6,6 is a biobased polymers is quickly increasing, mostly driven by reduction of carbon footprints, switching to renewable feedstocks, and rising consumer awareness concerning sustainability issues.1−4

ACA are synthesized using chemical catalysis from benzene (via cyclohexane) and caprolactam, respectively, whereas HMD is predominantly produced by Ni-catalyzed hydro- cyanation of 1,3-butadiene.5−7

Currently, almost all polymer building block chemicals are produced using petroleum-based chemical processes, which are inherently nonsustainable and have detrimental impacts on the environment.1−4

Thus, there is an increasing global demand to replace petroleum-based chemical processes with biobased chemical production from renewable resources.

Biobased production of AA has been demonstrated using fatty acids, glucose, and glycerol as renewable feedstocks in Escherichia coli, yeast, and Thermobifida fusca, expressing engineered or natural (T. fusca) biosynthetic pathways.8−17

Depending on the pathway and host used, these processes produced from 3 mg/L to 68 g/L of AA.2,4 Biosynthesis of ACA was achieved using E. coli cells expressing the α-ketoacid decarboxylase KdcA from Lactococcus lactis and amino- transferase Vfl from Vibrio f luvialis, with the ACA titer up to 160 mg/L.2,18

Although several patents propose various non- natural metabolic pathways for HMD biosynthesis,2 biobased production of this commodity chemical from renewable feedstocks has not yet been realized. Thus, further efforts are required to establish and improve the biobased production of 6-ACA and HMD.

Recently, carboxylic acid reductases (CARs) have emerged as attractive biocatalysts for biotransformation of organic acids to various chemicals.19−22 In the presence of ATP and NADPH, CARs catalyze the reduction of carboxylic acids to an aldehyde.23

These large enzymes (over 1000 amino acids) have three domains: an N-terminal adenylation (A) domain fused via a PCP-domain (peptidyl carrier protein, a phosphopantetheine attachment site) to a C-terminal reduc- tase (R) domain (Figure S1).24,25

The recombinant CAR from Nocardia iowensis required post-translational activation (phos- phopantetheinylation) by a phosphopantetheinyl transferase.25

The CAR A-domain catalyzes the initial reaction between ATP and a substrate acid to form an acyl-AMP intermediate, which is then attacked by the phosphopantetheine thiol forming a covalently bound acyl-thioester with the release of AMP.25

The acyl-thioester phosphopantetheine then swings from the A- domain to the R-domain resulting in the reduction of the thioester by NADPH and producing the aldehyde prod- uct.23,25,26 Crystal structures of A-PCP and PCP-R didomains from N. iowensis and Segniliparus rugosus revealed large-scale domain motions during the catalysis with two different conformations of the active site region (the “on” and “off” states).27

Purified recombinant CARs have been shown to accept a broad range of substrates including aromatic and aliphatic carboxylic acids.28,29 Several successful applications of CARs have already been demonstrated for the production of alkanes, aromatic aldehydes, and chiral amines.19,20,29−33

Highly efficient one-pot enzyme cascades for the biosynthesis of chiral piperidines and pyrrolidines or for biocatalytic N- alkylation of amines have been developed using CAR, transaminase (TA), imine reductase, and reductive aminase biocatalysts.34−36 TAs are pyridoxal-5′-phosphate (PLP)- dependent enzymes that belong to the PLP fold types I (S- selective) and IV (R-selective).37−39

These enzymes catalyze the transfer of an amino group between an amine donor (different amino acids and amines) and an amine acceptor (a ketone or aldehyde). Of particular interest are ω-transaminases (ωTAs) which do not require the presence of a carboxylic group in substrates and can accept a large variety of carbonyl substrates.39

Over the past decade, TAs have attracted considerable interest in biocatalysis both as individual biocatalysts and as part of multienzyme cascades for the synthesis of various amines.38−41

In our recent work, we identified several bacterial CARs with robust activity toward both aromatic and aliphatic substrates.42 Substrate screening of purified CARs against a panel of monocarboxylic acids revealed that these enzymes showed significant reductase activity against C3−C10 substrates with the maximal activity toward C5−C7 substrates.42

In addition, the subgroup II CARs, MAB4714 from Mycobacterium abscessus and MCH22995 from M. chelonae, exhibited significant activity against AA (C6), 7-aminoheptanoic acid (C7), and 8-aminooctanoic acid (C8), suggesting that some CARs can tolerate the presence of the second charged group in substrates.

Purified CARs supplemented with cofactor regenerating systems (for ATP and NADPH) and an aldo- keto reductase (AKR) catalyzed up to 76% conversion of AA to 1,6-hexanediol (Figure 1)42.

Here we explored the biotransformation of AA to 6-ACA (Cascade-1) and then to HMD (Cascade-2) using CARs and ωTAs (Figure 1). When supplemented with cofactor regenerating systems, the wild type CAR and ωTA demonstrated up to 95% substrate conversion in the in vitro transformation of AA to 6-ACA (Cascade-1) but showed low activity in the bioconversion of 6- ACA to HMD (Cascade-2).

To enhance CAR activity toward 6-ACA, the crystal structure of its substrate-binding domain was determined in complex with AMP and succinate and used for structure-based protein engineering.

Three mutant variants were found to exhibit enhanced activity against 6-ACA, as well as toward 7-aminoheptanoic acid and 8-aminooctanoic acid. In combination with ωTAs, the CAR L342E mutant protein showed up to 80% conversion of 6-ACA to HMD (Cascade-2).

With AA as substrate, the mixture of the CAR wild type and L342E proteins and two ωTAs demonstrated up to 30% conversion to HMD and 70% to 6-ACA.

MATERIALS AND METHODS

Gene Cloning and Protein Purification. The genes encoding CARs (MAB4714 and MCH22995), TAs, and other enzymes used in this work (Table S1) were amplified by PCR using corresponding genomic DNA and cloned into a modified p15TVLic vector (Novagen) containing an N-terminal 6His-tag as described previously.43

The E. coli inorganic pyrophosphatase PPA was expressed and purified using a clone from the E. coli ASKA collection.44 The phosphopantetheinyl transferase (PPT) BSU03570 (Sfp) from Bacillus subtilis was cloned into a pCDFDuet plasmid without a 6His-tag for coexpression with CARs.

Site-directed mutagenesis of the Mycobacterium abscessus CAR (MAB4714) and formate dehydrogenase (FDH) from Pseudomonas sp. strain 101 were performed using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer’s protocol, and the mutations were verified by DNA sequencing.

All plasmids were transformed into the E. coli BL21(DE3) Gold strain (Agilent). Cells were grown in 1 L TB cultures, induced with 0.4 mM IPTG overnight at 16 °C (or at 26 °C for CARs). Recombinant proteins were purified to near homogeneity (>95%, Figure S2) using Ni-chelate affinity chromatography on Ni-NTA Superflow resin (Qiagen) as described previously.45 The multiple sequence alignment was prepared using the MAFFT online tool and STRAP web server.46,47

Enzymatic Assays. Carboxylate reductase activity against differ- ent carboxylic acids was determined spectrophotometrically using an NADPH oxidation-based assay by following the decrease in absorbance at 340 nm.

CAR assays were performed in a reaction mixture (0.2 mL) containing HEPES-K (100 mM, pH 7.5), 1 mM NADPH, 2.5 mM ATP, 10 mM MgCl2, substrates (10 mM of aliphatic acids and 2 mM of cinnamic or benzoic acids), and 5−10 μg of purified CAR.

Transaminase activity was determined in coupled reaction with alanine dehydrogenase (Ala-DH) BSU31930 or glutamate dehydrogenase (Glu-DH) GDH1 spectrophotometrically using an NAD(P)H oxidation-based assay by following the decrease in absorbance at 340 nm (adapted from Wilding et al.48).

Transaminase screens were carried out in a reaction mixture (0.2 mL) containing 100 mM HEPES-K (pH 7.5), 0.5 mM NADPH or NADH, 1 mM adipaldehyde, 0.05 mM PLP, 50 mM NH4Cl, 2 mM Ala, 2 mM Glu, 5 μg of purified aminotransferases, 2.5 μg of Ala-DH, and 2.5 μg of Glu-DH. To determine kinetic parameters of enzymes (Km and kcat), microplate-based assays were performed (in triplicate) using a SpectraMax M2 plate-reader in a reaction mixture containing a range of variable substrates at 30 °C.

Kinetic parameters were determined by nonlinear curve fitting from Michaelis−Menten plots using GraphPad Prism (version 5.00 for Windows, GraphPad Software, San Diego, CA).

Identification of Reaction Products. The reaction products of bioconversion reactions catalyzed by CARs and TAs (6-ACA and HMD) were quantified on a Varian ProStar HPLC system equipped with a fluorescence detector and C18 column (Pursuit 5 150 × 3.9 mm; Agilent Technologies, USA) using a modified protocol described previously.49

Another potential bottleneck in the transformation of AA to HMD might be associated with the low activity of SAV2585 toward aminated intermediates (e.g., 6-aminohexanal). We proposed that putrescine transaminases might exhibit higher aminotransferase activity toward substrates containing terminal amines.

A recent report indicated that putrescine TAs represent appealing biocatalysts for biotransformations involv- ing diamines.68 The E. coli putrescine TA PatA belongs to aminotransferase class III (Pfam PF00202) and catalyzes the transfer of an amino group from terminal diamine donor molecules (putrescine, cadaverine) to keto acid acceptors (α- ketoglutarate).69

The Km values for putrescine and α- ketoglutarate were found to be 9.2 mM and 19 mM, respectively.69 In pyruvate amination reactions, E. coli PatA showed a preference for C4−C6 diaminoalkanes as amine donors.68

Recently, E. coli PatA and two other putrescine Tas have been shown to exhibit broad specificity for terminal aliphatic diamines, making them attractive biocatalysts.

It was also noted that ωTAs, which can accept amines or substrates with an amino group distant from carbonyl, are of special interest for biocatalytic applications.40

In our work, we observed that purified E. coli PatA was also active in HMD deamination with Km 7.7 mM and kcat 0.4 s−1 (kcat/Km 0.6 × 101). As shown in Table 1, PatA exhibited lower activity in the transamination reaction with adipaldehyde as acceptor (with L-glutamate as nitrogen donor), but its Km for adipaldehyde was almost 2 orders of magnitude lower (0.1 mM) compared to that for HMD (7.7 mM).

Because PatA is likely to be tolerant to the presence of amino groups in substrates, we tested this enzyme as an additional ωTA in AA transformation using the MAB4714+SAV2585 cascade. In- deed, PatA addition stimulated HMD production both from AA and 6-ACA by the wild type MAB4714 and SAV2585 (Figure 3).

Crystal Structure of the CAR A-Domain in Complex with AMP and Succinate. Recent biochemical and structural studies of CARs indicated that A-domains of these enzymes play a major role in substrate recognition.21,27,70

To obtain crystal structures of CAR A-domains for structure-based protein engineering studies, we cloned 16 A-domain fragments from 11 different CARs from our previous work42 and purified four A-domains for crystallization (including MAB4714 and MCH22995 from M.chelonae, 88% sequence identity, Figure S1).

The MCH22995 A-domain (1−641 aa) produced diffracting crystals, and its crystal structure was determined to a resolution of 1.97 Å using molecular replacement and the structure of the N. iowensis CAR A-domain (PDB code 5MSD) as the model (Table S2).

In vitro biotransformation of 6-ACA and AA using the MAB4714 L342E protein. On the basis of kinetic parameters (Table 2), the MAB4714 L342E mutant protein was selected for further studies on AA biotransformation using the CAR+TA cascades. With 6-ACA as substrate, purified L342E and SAV2585 showed up to 50% conversion to HMD compared to 1.5% observed for wild type MAB4714 (Figures 3 and 6). Furthermore, the addition of PatA increased the conversion of 6-ACA to HMD to 70% (Figure 6).

As expected, adding wild type MAB4714 to L342E produced no improve- ment in 6-ACA transformation for reactions containing one or two TAs (due to low activity of wild type protein against 6- ACA).

With AA as substrate, the L342E+SAV2585 cascade produced small amounts of both 6-ACA and HMD, and the addition of PatA had a minor positive effect, suggesting that transformation is limited by the low activity of L342E toward AA (Figure 6).

On the basis of these results, we hypothesized that the conversion of AA to HMD can be improved using a combination of the MAB4714 wild type and L342E proteins, because the wild type enzyme is more active against AA (CAR1 in Cascade-1), whereas the L342E protein is more efficient toward 6-ACA (CAR2 in Cascade-2).

Similarly, the inclusion of PatA might also be beneficial, because this enzyme appears to improve HMD formation from 6-ACA (TA2 Cascade-2) (Figures 3 and 6).

Accordingly, when both the wild type and L342E proteins were used in combination with one or two TAs for AA transformation, the reactions resulted in 30% conversion to HMD (and 60−70% to 6-ACA) representing a two times improvement compared to the wild type MAB4714 alone (Figures 3 and 6).

CONCLUSIONS

The biocatalytic transformation of terminal diacids (adipic acid) to terminal diamines (HMD) using CARs and ωTAs represents a formidable challenge due to the necessity to accommodate substrates with different terminal groups (carboxyl, carbonyl, amine) in substrate binding sites. Here we demonstrated that purified wild type MAB4714 and SAV2585 catalyze efficient transformation of AA to 6-ACA with up to 95% conversion but showed low activity in the second round of reduction/amination reactions (6-ACA → HMD).

Using structure-based protein engineering, we generated three mutant variants of MAB4714 (L342E, G418E, G426W) with enhanced activity toward 6-ACA, as well as against 7-aminoheptanoic acid and 8-aminooctanoic acid. In combination with SAV2585 and putrescine trans- aminase PatA, the L342E variant showed up to 75% conversion of 6-ACA to HMD but exhibited low activity against AA (the Cascade-1).

Therefore, for one-pot trans- formation of AA to HMD (via 6-ACA, Cascades-1 and -2, Figure 1), the wild type MAB4714 and SAV2585 were supplemented with the MAB4714 L342E variant and putrescine transaminase PatA, resulting in a complete conversion of AA to HMD (30%) and 6-ACA (70%).

We propose that HMD production from AA can be further improved by combining the identified beneficial mutations of MAB4714 (L342E, G418E, and G426W) and potentially via protein engineering of TAs. Thus, our study provided insights into the molecular mechanisms of substrate selectivity of CARs and illustrated the suitability of CARs and ωTAs for two rounds of substrate reduction/amination in one-pot cascade systems. Aminocaproic