Aldehyde Reductase Activity of Carboxylic Acid Reductases

Carboxylic acid reductase enzymes (CARs) are well known for the reduction of a wide range of carboxylic acids to the respective aldehydes. One of the essential CAR domains ‐ the reductase domain (R‐domain) ‐ was recently shown to catalyze the standalone reduction of carbonyls, including aldehydes, which are typically considered to be the final product of carboxylic acid reduction by CAR. We discovered that the respective full‐length CARs were equally able to reduce aldehydes. Herein we aimed to shed light on the impact of this activity on aldehyde production and acid reduction in general. Our data explains previously inexplicable results and a new CAR from Mycolicibacterium wolinskyi is presented.

Carboxylic acid reductase enzymes (CARs, E.C. 1.2.1.30)are studied intensively for the direct reduction of a wide range of carboxylic acids to their respective aldehydes. [1,2]CARs are found in bacteria and fungi and are currently categorized into 5 distinguished types based on their origins: Type I (Actinomycota), Type II (Ascomycota and Basidiomycota) Type III (Ascomycota) to Type IV (Basidiomycota) and Type V (amino acid reducing CARs; Actinomycota, Ascomycota and Basidiomycota), respectively. [2]The current understanding is that CARs are 'product-selective', because they intrinsically circumvent the kinetic preference of aldehyde versus carboxylic acid reduction, due to their reaction mechanism: [2,3] the deprotonated acid 1) attacks ATP to form a reactive mixed anhydride (acyl-AMP) which is 2) attacked by the thiol of an enzyme-bound phosphopantetheine to give an enzyme-bound thioester (acyl-PPT).The acylated phosphopantetheinyl arm exits the adenylation domain (A-domain) and enters the active site of a reduction domain (R-domain), where 3) the enzyme-tethered thioester is reduced to aldehyde.The aldehyde is released to the bulk, while the liberated phosphopantetheine swings back to the A-domain to react with the next acyl-AMP molecule. [4,5]he simplified reaction is shown in Scheme 1.Although the aldehydes b are the primary product of CARs upon acid a reduction, a number of studies was focused on CAR mediated synthesis of alcohols c from carboxylic acids, [6][7][8][9][10] with b as transient intermediates.
We recently solved the structure of the R-domain of a fungal CAR from Neurospora crassa (NcCAR, PDB-code 8AEP) and found significant carbonyl reducing activity of the highly pure protein (Scheme 1, box): [11] Benzaldehyde -the natural product of CAR mediated benzoic acid reduction -was reduced to benzyl alcohol.Up to this point, detected alcohols in the context of CAR reductions were exclusively assigned to host-background reactivities.To minimize aldehyde depletion, knock-out strains [12] like the E.coli RARE [13] strain were tailored to serve as platform strains for aldehyde synthesis with CARs.Overreduction to c was still detected, [14] even with additional in-situ product removal using an organic second phase [15] and the phenomenon seemed to be strongly substrate dependent.While < 2 % of benzylalcohol were found in full length NcCAR mediated cell free reductions, [16] > 30 % of 3-nitro-benzylalcohol and 4-cyano-benzylalcohol accumulated when the respective acids were treated with NcCAR in the presence of in vitro cofactor recycling components under the very same conditions.Strikingly, 97 % of mono-methylterephthalate was converted to the respective alcohol, which had previously been inexplicable. [17]Not only the fungal NcCAR but also the broadly applied bacterial CAR from Mycobacterium marinum consumed NADPH in the presence of octanal or benzaldehyde in the spectrophotometric assay, corroborating GC results. [11]Similar observations have been made for R-domains of polyketide synthases (PKS) and non-ribosomal peptide synthetases (NRPS) with C-terminal R-domains.Like R-domains of CARs, the PKS and NRPS domains R GPL and R NPS , respectively, show significant homology to short-chain dehydrogenases and rely on the strictly conserved catalytic residues tyrosine and lysine. [18]Moreover, structural and biochemical studies showed their ability to catalyze 4 electron reductions in a non-processive manner.In other words, the thioester reduction is followed by the dissociation of aldehyde and NADP + involving a large con-formational change to an open form.Subsequently, the second reducing equivalent (NADPH) is bound, which triggers a decrease in radius of gyration and consequently the formation of a closed conformation of the R-domain.The previously released aldehyde is reassociated and further reduced to the alcohol.Evidently, CARs may also catalyze 4 e À -reductions under certain circumstances.To understand the impact of this 'side-activity' of CARs on the innate acid reduction potential, we investigated this activity in more depth.
The selective preparation of aldehydes from acids is one of the major assets of CARs. [19,20]In this light we were striving to understand more about the propensity of CARs to additionally catalyze carbonyl reduction.We herein investigated several CARs (type I, III and IV), and used benzoic acid  1). [15]The activity for the reduction of 1 b was approximately 20 % of the activity for 1 a reduction in case of bacterial CARs.In case of fungal CARs, the activity for both acid and aldehyde were in the same range.Aliphatic aldehydes (2 b, 3 b) were even reduced with initial rates that approached those of the corresponding acids (2 a, 3 a) with all examined CARs in the presence of ATP (Figure S1).These preliminary experiments were conducted with the CARs produced in E. coli RARE and purified by Niaffinity chromatography.Blank reactions were carried out for each reaction.Specifically, aldehyde and acid reductions were carried out both in the presence and absence of ATP to account for any NADPH consuming activities caused by protein impurities.
In course of this study, we included a new CAR from Mycolicibacterium wolinskyi (MwCAR).While this enzyme shows similar initial rate activity for acid reduction as MmCAR for several substrates (Fig S2 ), it has a preference for a pH of 6.5 (Figure S4 and S5) in contrast to the reported pH optimum of MmCAR. [21]hile initial rate activities are considered a useful measure for screening-type enzyme characterization, ultimately, biotransformation reactions deliver more important insight with respect to operational stability and application potential.We therefore subjected type I, III and IV to biotransformation reactions with aldehydes 1 b-3 b.In addition, 4-formyl-benzoic acid methyl ester 4 b (the product of terephthalic acid mono methyl ester reduction) as well as 3-and 4-nitro-benzaldehyde 5 b and 6 b were chosen as substrates, because the respective acids 4 a-5 a had delivered surprisingly high amounts of the respective alcohols in previous studies (Figure 1). [16,17]Reactions were carried out both in the presence and absence of ATP and a no-enzyme control was subtracted (for the full dataset see Supporting information, TableS1).The results in Table 1 corroborate a strong dependency of carbonyl reduction potential on both the CAR and the chemical structure of the substrate.In case of bacterial MwCAR, only the para-substituted aldehyde 4 b was reduced in significant amounts.The R-domain of Type III CAR delivered substantial amounts of alcohols only in the case of para substituted aldehydes 4 b and 6 b, respectively, but not for the unsubstituted and ortho substituted analogs 1 b and 5 b, respectively.A similar but much more pronounced trend can also be observed for type IV CAR from P. cinnabarinus.Remarkable is also the chain length dependency: More C-8 alcohol 3 c was detected as compared the C-6 counterpart 2 c.
Both the stereoelectronic properties of the substrate as well as the nature of the R-domain seem to determine the degree of 4-electron reduction in each substrate/enzyme combination.Selected IMAC purified CAR preparations were subjected to size exclusion chromatography (SEC, Figure S6 and S7) and the reactions were repeated.In case of MmCAR, detected amounts of c were in the range of 0-1 % for both preparations (Table S1).For PcCAR4, the reduction of aliphatic aldehydes 2 b and 3 b were reduced significantly, while the reduction of aromatic aldehydes 4 b-6 b still delivered high amounts of alcohol.Since both, MmCAR and PcCAR4 were prepared from the same vector and host strain using identical conditions for culturing and protein purification, we concluded that the aldehyde reducing activity for aromatic aldehydes is catalyzed by the R-domain of the PcCAR4 but that host background proteins may be a major contributor, depending on the chemical structure of the substrate.
The reductase domains of CARs are structurally similar to short chain dehydrogenases, and our results show that carbonyl reduction ability is a general feature of CARs across the phylogenetic tree.In many cases, the amounts of produced alcohols remain negligible, but both the substrate structure and the chosen CAR determine the intrinsic propensity of alcohol formation.Of course, the ability of CAR R-domains to perform 4 e À -reductions might deliberately be exploited for alcohol production without the need for additional carbonyl reductases.][24] On the other hand, our findings explain the inability to fully suppress alcohol formation despite thorough reaction optimization and the use of engineered strains that has been reported previously. [14,16,17]

Experimental Section
The gene sequence of Mycolicibacterium wolinskyi putative carboxylic acid reductase (NCBI accession No WP_085142124.1)was optimized for expression in E. coli, synthesized and cloned into the pETDuet1 vector harboring Mycobacterium marinum phosphopantetheinyl transferase (PPTase; NCBI accession No WP_094357779.1)by GenScript.Cultivation and purification by nickel affinity chromatography (IMAC) was carried out as described previously. [14]E.coli RARE [13] was used for the production of all CARs used in this study.DNA and protein sequence of MwCAR as well as a NUPA-Gel are given in the Supporting information.
Biotransformations were conducted in MES buffer (100 mM, pH 6.5 containing 20 mM of MgCl 2 ) in a total reaction volume of 500 μL.The typical enzyme concentration was 200� 50 μgmL À 1 of single step purified CAR (IMAC) or 170 � 30 μgmL À 1 of two step purified CAR (IMAC + SEC).Final concentrations were 2 mM of substrate, ATP 6 mM or no ATP, NADPH 10 mM and DTT 1 mM, respectively.Substrate was added from stock solutions in DMSO (final conc: 5% v/v).Reactions were conducted at 28°C and 600 rpm on an Eppendorf thermomixer for 3 h in Eppendorf tubes.They were terminated by addition of 50 μL of HCl (2-3 M) and extracted into 500 μL of ethyl acetate.Reactions were analyzed by GC/FID.Sample aliquots of 1 μL were injected in split mode (split ratio 1: 10) at 240°C injector temperature and 320°C detector temperature.The carrier gas was N 2 .The oven temperature was 70°C for 4 minutes, 130°C at 5°C min À 1 and a rise to 300°C at 45 °C min À 1 with a hold of 2 minutes.The total run time was 21.78 minutes.Calculation of conversions is based on area normalization.

Figure 1 .
Figure 1.Carboxylic acids (a) and aldehydes (b) used in this study

Table 1 .
Conversion of b (2 mM) to c catalyzed by CAR in MES buffer (pH 6.5) in the presence of 20 mM MgCl 2 , 1 mM DTT, 10 mM NADPH, 2 % v/ v DMSO, 3 h, 28 °C.Samples were analyzed by GC.