Asymmetric hydrogenations are important reactions in industrial synthesis, as up to two stereogenic centres can be generated.1 Given the current trend towards more “green” or sustainable chemistry, biocatalytic approaches are becoming more important in the production of fine chemicals, pharmaceuticals, and agrochemical products. There are at least four known classes of enzymes that have been explored for their biocatalytic applicability, collectively known as ene-reductases (ERs).2 Each enzyme requires NAD(P)H as the hydride donor, whereas the mechanism, substrate scope and product stereo- and/or enantiospecificity differs between biocatalyst classes.
The most widely investigated class is the old yellow enzyme (OYE; EC 184.108.40.206) family and many recent reviews have summarised the (potential) applications of these enzymes for industrial syntheses, for example, Refs. 1–3. These flavin mononucleotide (FMN)-containing enzymes catalyse the CC reduction of a wide variety of α,β-unsaturated aldehydes, ketones, nitroalkenes, maleimides, dicarboxylic acids and their esters, and nitriles.1–3 A second less well-characterised enzyme class catalysing biocatalytic reductions are enoate reductases (EC 220.127.116.11).4 These oxygen-sensitive enzymes contain both FAD and [4Fe-4S], and are members of the NADH:flavin oxidoreductase/NADH oxidase family. They have been shown to catalyse the reduction of a variety of α,β-unsaturated monoacids and esters.4, 5 A more recent class of enzymes investigated are the medium-chain dehydrogenase/reductase (MDR) family of oxidoreductases (1.3.1.-).6 For example, the flavin-independent double-bond reductase from Nicotiana tabacum (NtDBR) catalyses the reduction of a wide variety of α,β-unsaturated aldehydes, ketones and nitroalkenes.7 There are a few examples in the literature of potential applications of a fourth class of enzymes, namely the flavin-independent short-chain dehydrogenase/reductases (SDR) from plants. Examples include two menthol dehydrogenases, (−)-menthol dehydrogenase (EC 18.104.22.168) and (+)-neomenthol dehydrogenase (EC 22.214.171.124), which catalyse the reduction of specific menthone isomers to menthol.8
Unfortunately, the high cost of NAD(P)H nicotinamide coenzymes makes them uneconomical for industrial-scale syntheses. Therefore, alternative methodologies have been adopted to supply the necessary hydride equivalents for alkene reduction. One such method is to include a coenzyme recycling system, in which catalytic levels of NAD(P)+ are constantly regenerated to NAD(P)H. Several systems have been developed and routinely employed for ER biocatalytic reactions, such as glucose dehydrogenase, glucose-6-phosphate dehydrogenase (G6PDH/glucose-6-phosphate), formate dehydrogenase and phosphite dehydrogenase, and reviewed elsewhere.1, 9 In each case, only catalytic levels of NAD(P)+ are needed but stoichiometric levels of a co-substrate are required to drive the recycling enzyme. Faber and coworkers have performed comparative biotransformations by using multiple cofactor recycling systems, substrates and ERs, to determine the ideal hydride source.10 Product yield and/or enantioselectivity can vary considerably, depending on which coenzyme recycling system has been employed.10 Additionally, the inclusion of a second enzyme system into large-scale (bio)synthesis may present problems, for example in maintenance of the activity and stability of two enzymes and the high cost of some co-substrates. Electrochemical regeneration of nicotinamide coenzymes has also been investigated, for example in the direct cathodic reduction of NAD(P)+. Unfortunately, this simple regeneration method is hampered by low selectivity and the formation of undesired NAD(P) dimeric side products.9
Several techniques have been developed to bypass the need for nicotinamide coenzymes by using alternative hydride donors to reduce the flavin cofactor of some ERs. Reetz and coworkers described a method whereby the FMN cofactor of the OYE homologue YqjM from Bacillus subtilis was photoreduced, employing free flavin and a sacrificial electron donor.13 The YqjM bound oxidised FMN was reduced by the free photoreduced FMN, thereby generating the active enzyme. This NAD(P)H-free system was successful in converting ketoisophorone to (R)-levodione, albeit with lower product enantiopurity. Similarly, Hollmann and co-workers described the photoenzymatic flavin reduction of YqjM and NEMR (N-ethylmalemide reductase from E. coli) by using alternative sacrificial electron donors formate and phosphite.14 These methods required strict anaerobic conditions to prevent the rapid reoxidation of reduced FMN by molecular oxygen.14
Early attempts at NAD(P)H-independent alkene reduction focussed on the use of artificial mediators, such as N,N-dimethyl-4-4-bipyridinium [methyl viologen (MV)], as a means of reducing the flavin in clostridial enoate reductases.15 A variety of reduced mediator recycling systems were employed, including direct electron transfer from a cathode and enzymatic-based recycling systems. Successful enzyme-coupled mediator recycling was achieved used the following systems: 1) hydrogenase/H2; 2) formate dehydrogenase/formate and 3) carbon monoxide dehydrogenase/CO.15
Recently, continuous biphasic miniaturised bioreactors have been designed for biocatalytic reductions catalysed by pentaerythritol tetranitrate reductase (PETNR; from Enterobacter cloacae PB2), which can be readily adapted for use with related enzymes (Figure 1).12, 16 This technology combines the use of electrochemistry and mediator redox chemistry to drive ER biocatalysis in a miniaturised bioreactor (Figure 1 a), and introduces a novel scalable bioreactor design with integrated on-chip spectroscopic reaction monitoring (Figure 1 b). An earlier microreactor design that used coenzyme recycling (G6PDH/glucose-6-phosphate) rather than electrochemical reduction of the biocatalyst was based on a recirculating, biphasic flow with electrostatic attraction to force-charged droplets to merge with the aqueous phase.16 Biphasic reaction conditions can dramatically improve product yield, as potentially inhibitory high concentrations of substrate and/or product are sequestered in the organic phase and product removal is simplified. Based on the fact that catalysis is performed at the solvent interface, the mass transfer between the two phases was improved via droplet-based microfluidics. This ensured short diffusion distances and high surface-to-volume ratios. The aqueous phase was injected into the immiscible carrier fluid, generating non-spherical droplets (slugs) with improved mass transfer efficiency. Control biotransformations were performed with a variety of α,β-unsaturated ketones and nitroalkenes in the presence of a cofactor recycling system (G6PDH/glucose-6-phosphate). In each case, the substrate conversions were similar or higher than those obtained under conventional reaction conditions.16
Electrochemical reduction of ER in a miniaturised bioreactor has led to further design improvements. These new designs have incorporated an electrochemical-based mediator recycling system for NADPH-independent PETNR-catalysed biotransformations (Figure 1).12 This was based on the ability of MV2+ to be electrochemically reduced to its blue monocation MV+, a known reductant for OYEs. Owing to the toxicity of viologens, this design incorporates retention of the water-soluble mediators within the microreactor. The aqueous phase is pumped into an electrochemical cell to reduce the mediator prior to mixing with the organic (substrate-containing) phase (Figure 1 b). Comparative biotransformations of PETNR and the thermophilic old yellow enzyme from Thermoanaerober pseudethanolicus (TOYE) with five α,β-unsaturated aldehydes, ketones and maleimides showed successful substrate conversions and easy product recovery, although with a lower efficiency than reactions catalysed with NADPH present. For example, the TOYE and PETNR-catalysed reduction of 2-cyclohexenone 1 a generated cyclohexanone 1 b at a rate of 1.25 and 0.7 mM h−1, compared to 3.1 mM h−1 in the presence of a two-fold excess of NADPH. Additionally, enzyme and mediator stability and reusability were maintained over 12 h. Bioreactor conditions were kept strictly anaerobic to prevent the reoxidation of the MV+ cation by molecular oxygen. This iss the first example of a versatile microfluidic electrochemical reactor used to drive NAD(P)H-independent activated alkene reduction by OYEs.12
Some OYEs can operate in the absence of reducing coenzymes, if a 2-enone or 1,4-dione co-substrate acts as the hydride source, generating the equivalent phenolic co-product (see Figure 2 a).22 However phenolic compounds are often potent inhibitors of OYEs,23 and may lead to a significant reduction in alkane product yields.22 A comprehensive study of the disproportionation of 1 a (alkene acts as the hydride donor and substrate) by eleven OYEs showed poor conversions (<15 %) in the majority of cases.17 Recently, Faber and co-workers have overcome this phenomenon by the inclusion of an in situ phenolic co-product scavenging system using the polymeric adsorbent macroporous triethylammonium methylpolystyrene carbonate (MP-carbonate).17 For example, the OYE homologue chromate reductase (CrS) from Thermus scotoductus SA-01 reduced dimethyl citraconate 2 a to dimethyl (R)-2-methylsuccinate 2 b with moderate conversions (48 %) and absolute enantioselectivity (Figure 2 a). Co-substrate 1,4-cyclohexanedione 3 a was aromatised to co-product 1,4-dihydroxybenzene 3 b, which was sequestered by MP-carbonate. Increasing the enzyme concentration and reaction time further optimised the reaction, resulting in a conversion of 92 % with an enantiopurity of (R)-2 b of >99 %. However, in some cases both the alkene substrate and its alkane product were sequestered by MP-carbonate, and may have caused undesired racemisation of chirally sensitive α-substituted ketones, so this technique is not applicable to all substrates.17 Further investigations identified additional cheap and commerically available co-substrates, resulting in conversions and enantioselectivities comparable to those obtained with traditional two-enzyme NAD(P)H-recycling systems.18
The coenzyme-independent methods described above are not applicable to flavin-independent ERs (MDRs and SDRs) as the former rely on flavin reduction preceding alkene substrate binding that is, the reaction proceeds in two half-reactions involving flavin reduction (by the coenzyme) and flavin oxidation (by a target substrate).23 A recent report by Hollmann and co-workers described the use of synthetic nicotinamide coenzyme mimics (mNADs) as substitutes for the more expensive NAD(P)H in ER-catalysed reductions of α,β-unsaturated ketones.19 A variety of mNADs were investigated (4 a–8 a; Figure 2 b), which were found to reduce the OYE YqjM from Bacillus subtilis and homologues from Thermus scotoductus (TsER) and Ralstonia metallidurans CH34 (RmER). Biotransformations of ketoisophorone with the three ERs and either mNAD 4 a or NAD(P)H gave near identical conversions and product enantiopurities.19 Unfortunately, other nicotinamide coenzyme-requiring enzymes, such as alcohol dehydrogenases and monooxygenases, exhibit poor activity towards mNADs. However, this makes them useful catalysts in biotransformations with poorly purified ERs because contaminating ketoreductase activity is thereby eliminated.19 These mNADs are very cost-effective to synthesise, rendering them an economical choice for industrial-scale biotransformations. These are exciting developments, but currently there is no in situ method of regenerating the reduced mNADs. At present, their use is less cost-effective than cofactor recycling systems in large-scale industrial applications but they offer great promise for future developments. Work is currently underway to determine whether they can be useful for other OYEs, and the non-flavin containing MDR and SDR enzymes.
The development of alternative mechanisms for reducing ene-reductases highlights the importance of these enzymes in the development of new and improved industrial-scale biocatalytic reductions. Although “green” and sustainable syntheses are preferable over traditional chemical techniques, this must be balanced against the economic cost of these processes. Further development in either more cost-effective cofactor recycling or nicotinamide coenzyme-independent hydride sources is, therefore, a priority for realising the value and potential applications of industrial-scale biocatalytic hydrogenation reactions.