Introduction

Alcohol dehydrogenases (E.C. 1.1.1.x; x = 1 or 2, ADHs) are a group of oxidoreductases that catalyze the interconversion between alcohols and aldehydes or ketones by transferring C4-hydride from NAD(P)H to the carbonyl carbon of an aldehyde or ketone substrate (Musa and Phillips 2011). During the last decade, ADHs have been widely studied and applied in the asymmetric synthesis of chiral alcohols, based on their high stereoselectivity under mild conditions (Musa and Phillips 2011; Zhang et al. 2015). Although ADHs have various catalytic advantages, the widespread application of ADHs to the large-scale production of target products is still restricted to their commercial availability and the low aqueous solubility of the substrates. More attention has been paid to the exploration of robust enzymes with high catalytic efficiency and the development of novel non-aqueous media to improve the solubility of substrates (Carvalho 2016; Wang et al. 2016; Zhang et al. 2015).

In the past few years, a considerable number of publications on ADHs and the corresponding asymmetric biosynthesis have been documented and reviewed from various aspects. Hall and Bommarius (2011) exhaustively discussed the major advances of reductase-catalyzed and oxidase-catalyzed redox reactions and covered reduction, oxidation, and corresponding cofactor regeneration. Musa and Phillips (2011) provided an overview of ADH-catalyzed asymmetric synthesis of hydrophobic alcohols. Biocatalytic asymmetric reduction to chiral alcohols was also summarized (Patel 2013). Nealon et al. (2015) discussed the factors that influence the substrate stereospecificity and specificity of ADHs. Recently, Zhang et al. (2015) reviewed the rational and/or irrational design of ADHs and the methods for screening the desired ADH variants.

This review focuses on the latest advances in biological applications of ADHs in the preparation of chiral pharmaceutical intermediates and fine chemicals, followed by future aspects of ADH-related research to enhance the industrial applications.

Biological applications of ADHs in the production of chiral pharmaceutical intermediates

Antidepressants

(S)-Duloxetine, (S)-(+)-N-methyl-3-(1-naphthalenyloxy)-3-(2-thienyl)propanamine, is not only used as a new antidepressant and a dual inhibitor of serotonin and norepinephrine reuptake but is also regarded as a potential drug for the treatment of stress urinary incontinence. The chemical synthesis of (S)-duloxetine was first described by Eli Lilly using a stereoselective reduction of prochiral ketones 1a–1f to the (S)-chiral alcohols 2a–2f as the key step (Fig. 1). The chiral metal ligand catalysts used in this step are costly, and the process resulted in trace metal contamination in the products. ADH-catalyzed processes for reduction of the ketones 1a–1f were developed to synthesize (S)-duloxetine to avoid the shortcomings of the chemical methods (Chen et al. 2016). Whole Candida tropicalis PBR 2 cells were first used for the asymmetric reduction of compound 1b to produce compound (S)-2b with 80% product yield and >99% enantiometric excess (ee). Furthermore, whole Candida viswanathii cells were used for the gram-scale reduction of compound 1b to compound (S)-2b at >81% conversion and >99% ee (Soni et al. 2008). Recently, the Lactobacillus kefir ADH was used as a template, and a series of ADH variants were obtained through protein engineering by Codexis (Redwood City, CA, USA) that reduced compound 1a to compound (S)-2a and compound 1b to compound (S)-2b. Using the isopropanol-based cofactor regeneration, the concentration of substrate was up to 150 g/L and the product yield and ee were 90 and >99%, respectively, which opened the door for the bioproduction of (S)-duloxetine (Savile et al. 2014).

Fig. 1
figure 1

Bioreduction process for synthesis of (S)-duloxetine intermediates (Tang et al. 2011)

Full size image

Befloxatone, (5R)-5-(methyloxymethyl)-3-[4-(3R)-4,4,4-trifluoro-3-hydroxybutoxy]phenyl-2-oxazalidinone, a novel oxazolidinone derivative, has been introduced into clinical practice as the first potent antidepressant drug which selectively and competitively inhibits monoamine oxidase (MAO)-A. (R)-Ethyl 3-hydroxy-4,4,4-trifluorobutanoate ((R)-EHTB) was regarded as the common intermediate in befloxatone synthesis. The chemical synthesis of (R)-EHTB from ethyl 3-oxo-4,4,4-trifluorobutanoate (EOTB) requires two steps involving NaBH4-dependent reduction and a subsequent kinetic resolution, which was constructed by Novozyme (Bagsvaerd, Denmark) (Tixidre et al. 1998). Due to the low yield in the chemical process, the biocatalyst-based asymmetric reduction of ketone EOTB to (R)-EHTB was developed using permeabilized Bacillus pumilus Phe-C3 cells containing an NADPH-dependent ADH and a G-6-PDH, with 67% product yield and 95% ee at a substrate concentration of 60 mM (Zhang et al. 2006). Whole Saccharomyces uvarum SW-58 cells were further used in an aqueous-organic solvent biphasic system for the asymmetric reduction of ketone EOTB to (R)-EHTB with 85.0% conversion and 85.2% ee (He et al. 2007). Whole Kluyveromyces marxianus cells also have the ability to catalyze EOTB to (R)-EHTB, with a relatively low yield (81%) and ee (24%) (Oliveira et al. 2013). Recently, Kara et al. (2014) established a highly productive biocatalytic reduction route for EOTB to (R)-EHTB using an isolated ADH in non-aqueous media, and a high conversion of 99.9% was obtained, with >99.9% ee. These studies make it possible for (R)-EHTB to be produced using the biocatalytically asymmetric reduction method.

Antianxiety drugs

Buspirone (Buspar®), a potent serotonin 5HT1A receptor partial agonist, is used for the treatment of generalized anxiolytic disorder. It could bind to D3 and D4 receptors with high affinity and is applied to treat cocaine self-administration. Notably, 6-hydroxybuspirone is a major metabolite of buspirone and contributes to the effectiveness of buspirone, due to its high plasma concentration. There is no apparent distinction between (R)-6-hydroxybuspirone and (S)-6-hydroxybuspirone, showing that 6-hydroxybuspirone is effective and could be developed as a potential anxiolytic agent. The enantioselective microbial reduction of 6-oxo-8-[4-[4-(2-pyrimidinyl)-1-piperazinyl]butyl]-8-azaspiro[4,5]decane-7,9-dione (6-ketobuspirone) to either (R)-6-hydroxybuspirone or (S)-6-hydroxybuspirone has been developed. Patel et al. (2005) screened approximately 150 strains for the enantioselective reduction of 6-ketobuspirone, where whole Rhizopus stolonifer SC 13898 cells showed the capability of producing (S)-6-hydroxybuspirone, with >53% yield and >96.6% ee. Meanwhile, the strain Candida maltose SC 16112 exhibited the ability to produce (R)-6-hydroxybuspirone, with >57% yield and >97% ee. A NADPH-dependent ADH from Hansenula polymorpha SC 13845 was expressed together with a G-6-PDH from Saccharomyces cerevisiae and catalyzed the reduction of 6-ketobuspirone to (R)-6-hydroxybuspirone with a 99% yield and 99.9% ee at 50 g/L substrate (Goldberg et al. 2006). In addition, an ADH isolated from Pseudomonas putida SC 16269 was combined with formate dehydrogenase from Pichia pastoris and reduced 50 g/L 6-ketobuspirone to (S)-6-hydroxybuspirone with a >98% yield and >99.9% ee (Goldberg et al. 2006).

Antiasthmatics

Montelukast, (2S)-3-[[(1S)-3-(2-acetylphenyl)-1-[3-[(1E)-2-(7-chloro-2-quinolinyl)ethenyl]phenyl]propyl]thio]-2-methylpropanoic acid, was identified by Merck as a cysteinyl leukotriene 1 receptor antagonist and developed as a therapeutic agent for treating asthma. Merck described the chemical synthetic route for the production of montelukast using a stereospecific reduction of (E)-2-(3-(3-(2-(7-chloro-2-quinolinyl)ethenyl)phenyl)-3-oxopropyl)benzoic acid methyl ester (keto ester M) to (S,E)-methyl 2-(3-(3-(2-(7-chloroquinolin-2-yl)vinyl)phenyl)-3-hydroxopropyl)benzoic acid methyl ester ((S)-hydroxy ester) as the key step. Because there were several disadvantages of the chemical methods, enzyme-catalyzed processes for this reduction were explored. The first successful reaction used whole Microbacterium sp. MB 5614 cells containing a ketoreductase for the synthesis of (S)-hydroxy ester with >95% ee (Roberge et al. 1996). Codexis engineered L. kefir ADH through directed enzyme evolution for the asymmetric reduction of keto ester M to the corresponding (S)-hydroxy ester. After a series of mutations, one mutant was finally obtained exhibiting excellent productivity of (S)-hydroxy ester, which was improved by 2000-fold and had high enantioselectivity and stability. The final yield and ee reached 99.3 and 99.9%, respectively, at a substrate concentration of 100 g/L (Liang et al. 2012).

(R)-(−)-3-Quinuclidinol, which contains a double-ring structure with nitrogen, is a valuable chiral building block for pharmaceuticals such as talsaclidine and revatropate, which are used as bronchodilator agents. Several routes have been reported for the preparation of (R)-(−)-3-quinuclidinol, including the resolution of racemic (R)-(−)-3-quinuclidinol and asymmetric reduction of 3-quinuclidinone (Tsutsumi et al. 2009). More attention was paid to developing the direct reduction of 3-quinuclidinone to (R)-(−)-3-quinuclidinol by ADH, with a maximum theoretical yield of 100%, to break through the bottleneck of 50% theoretical yield in the racemic resolution approach. A carbonyl reductase from Rhodotorula rubra JCM3782 was coexpressed with glucose dehydrogenase (GDH) for cofactor regeneration and catalyzed the reduction of 3-quinuclidinone (618 mM) to (R)-(−)-3-quinuclidinol with 98% conversion and >99.9% ee (Uzura et al. 2009). A new strain, Nocardia sp. WY1202, was isolated for the stereospecific reduction of 3-quinuclidinone to produce (R)-(−)-3-quinuclidinol with a 93% yield and >99% ee (Wang et al. 2013a). In addition, two novel reductases (QNR and bacC) that were capable of reducing 3-quinuclidinone to (R)-(−)-3-quinuclidinol were cloned from Microbacterium luteolum JCM 9174 and coexpressed with an ADH from Leifsonia sp. (LSADH) for cofactor regeneration. As a result, 939 mM 3-quinuclidinone was converted to (R)-(−)-3-quinuclidinol, with a conversion of 100 and 99.9% ee, by immobilized Escherichia coli cells containing recombinant QNR (Isotani et al. 2012). Recently, Zhang et al. (2013b) described a new ketoreductase (ArQR) from Agrobacterium radiobacter ECU2556 that was capable of catalyzing the conversion of >200 g/L substrate 3-quinuclidinone to (R)-(−)-3-quinuclidinol with 99% ee and a high space-time yield of up to 916 g/L/day, which is competitive for biomanufacturing (R)-(−)-3-quinuclidinol on an industrial scale.

Anticholesterol drugs

Statins, also known as hydroxymethylglutaryl CoA (HMG CoA) reductase inhibitors, can reduce low-density lipoprotein cholesterol (LDL-C) by inhibiting liver cholesterol synthesis and are regarded as one of the top-selling drugs for hypercholesterolemia treatment. (R)-4-Cyano-3-hydroxybutyrate ((R)-HN) is a key intermediate for the production of atorvastatin and can be converted to the appropriate side chain upon Claisen condensation and reduction. An enzymatic process for the preparation of (R)-HN from ethyl 4-chloro-3-oxobutanoate (COBE) was developed by Codexis using a two-enzyme system with a ketoreductase and halohydrin dehalogenase (HHDL) (Davis et al. 2010). For this process, the stereospecific reduction of COBE to (S)-ethyl 4-chloro-3-hydroxybutanote ((S)-CHBE) was achieved by reductases from Geotrichum candidum, Pichia stipitis, Candida parapsilosis ATCC 7330, and other previously reported strains (Kaliaperumal et al. 2011; Liu et al. 2015; Patel et al. 1992; Ye et al. 2011; You et al. 2014). A substrate-coupled biocatalytic process driven by a NADPH-dependent sorbose reductase from Candida albicans was described to convert 2.5 M COBE to (S)-CHBE, with a 93.6% yield and 99.6% ee (Cai et al. 2012). In addition, highly efficient synthesis of (S)-CHBE was developed with whole E. coli BL21 cells containing a reductase gene cloned from Streptomyces coelicolor, and the reaction used 2-propanol for cofactor regeneration in a water/toluene biphasic system. As much as 3.6 M COBE was asymmetrically reduced within 22 h, resulting in a 93% yield and >99% ee (Wang et al. 2011a). Recently, novel reductases identified by genome mining were applied to produce (S)-CHBE with a high concentration of COBE (3 M), a high yield (>99.0%), and a high ee value (>99.9%), indicating their great potential in the industrial production of (S)-CHBE (You et al. 2014; Liu et al. 2015).

As shown in Fig. 2, compounds 6a–6d are the key chiral precursors for HMG-CoA reductase inhibitors that are synthesized by stereoselective reduction of compounds 3a–3d, which contain two carbonyl groups. Using a suspension of Acinetobacter calcoaceticus SC 13876 cells as the biocatalyst, both compounds 3a and 3b were stereospecifically reduced to the corresponding (3R)-5a–5b and (5S)-4a–4b compounds, followed by further reduction to the corresponding dihydroxy esters (3R,5S)-6a and (3R,5S)-6b. The desired (3R,5S)-6a compound was obtained with a 99% ee and 97% diastereomeric excess (de), whereas the (3R,5S)-6b compound was produced with 99% ee and 63% de (Guo et al. 2006). Subsequently, Goldberg et al. (2008) cloned and identified a desired ketoreductase III from A. calcoaceticus SC 13876 that catalyzed the stereoselective reduction of ketoesters 4b to (3R,5S)-6b with 99.3% yield, 100% ee, and 99.8% de. In addition, a purified diketoreductase (DKR) from Acinetobacter baylyi ATCC 33305 was described to stereoselectively reduce compound 4a to compound 6a, with both de and ee values greater than 99.5% (Wu et al. 2009). Recently, Wang et al. (2015) and Luo et al. (2015) cloned and expressed CaAKR from C. albicans XP1463 and KlAKR from Kluyveromyces lactis XP1461, respectively, to catalyze the stereoselective reduction of ketoester 4c to compound (3R,5S)-6c with both de and ee values greater than 90 and 99% for CaAKR and KlAKR, respectively. Furthermore, 11.25-fold higher catalytic efficiency (kcat/Km) of KIAKR towards compound 4c was achieved through two rounds of site saturation mutagenesis (Luo et al. 2016). Different cofactor regeneration systems and an aqueous-organic biphasic reaction system were evaluated for this biocatalytic process using purified recombinant DKR. An aqueous-hexane biphasic system was further scaled up to 0.5 L at a substrate concentration of 105 g/L, and the product was obtained with 83.5% yield, >95.5% de, and >99.5% ee (Wu et al. 2011). In addition, Codexis developed a series of mutants from a S. cerevisiae ketoreductase using a directed enzyme evolution method, and the final mutant was isolated to catalyze the reduction of both compounds 4c and 4d to compounds (3R,5R)-6c–6d or (3R,5S)-6c–6d with 99.9% ee and 99.99% de at a substrate concentration of 150 g/L (Sun et al. 2013). These biocatalytic approaches avoided most of the disadvantages that occurred in the chemical methods, which paved the foundation for the practical synthesis of statin drugs.

Fig. 2
figure 2

Bioreduction process for synthesis of HMG-CoA reductase inhibitor intermediates (Goldberg et al. 2008; Guo et al. 2006)

Full size image

Ezetimibe (Zetia® and Ezetrol®) is a novel cholesterol absorption inhibitor that decreases the LDL-C levels by binding and thereby inhibiting the function of Niemann-Pick C1-Like 1 (NPC1L1), the main cholesterol absorption transporter in the intestine. The key step of ezetimibe synthesis involves the asymmetric reduction of 1-(4-fluorophenyl)-5-(2-oxo-4-phenyl-oxazolidin-3-yl)-pentane-1,5-dione (FOP dione) to 3-[5-(4-fluorophenyl)-5(S)-hydroxypentanoyl]-4(S)-4-phenyl-1,3-oxazolidin-2-one ((S)-FOP alcohol). The microbial enantioselective reduction of FOP dione to (S)-FOP alcohol with >99% ee was initially performed by whole Schizosaccharomyces octosporus ATCC 2479 and Burkholderia cenocepacia MTCC 5427 cells with 54 and 34% yield, respectively (Homann and Previte 1997; Singh et al. 2009). Subsequently, Codexis developed a series of engineered ketoreductases from L. kefir, Lactobacillus brevis, and Lactobacillus minor for the reduction of FOP dione to (S)-FOP alcohol, and one mutant with improved properties was screened and employed to produce (S)-FOP alcohol at a preparative scale in the presence of 100 g/L substrate. The yield was approximately 99%, and the stereomeric purity was >99% (Mundorff and De Vries 2013).

Antihypertensive drugs

Ethyl (R)-(−)-2-hydroxy-4-phenylbutyrate ((R)-8) (Fig. 3) is a versatile intermediate for the production of angiotensin-converting enzyme (ACE) inhibitors, including benazepril, lisinopril, and enalapril, which are widely used to treat hypertension. Different approaches for (R)-8 synthesis were explored, including the kinetic resolution of racemic 8, asymmetric reduction of ethyl 2-oxo-4-phenylbutyrate (7), and multi-step synthesis (D’Arrigo et al. 2010). Biocatalytic asymmetric reduction was used as an attractive approach for the synthesis of compound (R)-8. A large number of microorganisms, including Candida boidinii CIOC21, Candida krusei SW2026, Rhodotorula mucilaginosa CCZU-G5, and Candida magnoliae CGMCC 2.1919, were shown to be the highly enantioselective biocatalysts in this reduction process (>99% ee) (Chen et al. 2008; Wang et al. 2013b; Xia et al. 2013; Zhang et al. 2009). Li et al. (2010) identified and characterized a highly enantioselective carbonyl reductase from C. krusei SW 2026, which was responsible for the reduction of compound 7. Furthermore, whole E. coli cells containing recombinant β-ketoacyl-ACP reductase (FabG) from Bacillus sp. ECU0013 were reported to reduce up to 620 g/L of compound 7 to compound (R)-8 with >99% ee (Ni et al. 2011), suggesting that FabG is a promising biocatalyst for the production of compound (R)-8.

Fig. 3
figure 3

Bioreduction process for synthesis of benazepril, lisinopril, and enalapril intermediate (Qian et al. 2014)

Full size image

Substrate feeding approaches and an aqueous/organic biphasic system were successfully constructed to improve the substrate tolerance and enhance the production of compound (R)-8. Ni et al. (2013) and Su et al. (2012) developed a 1 L aqueous/octanol biphasic reaction system for the synthesis of compound (R)-8 with a recombinant E. coli strain that coexpressed a novel carbonyl reductase and GDH, resulting in a space-time yield of 660 g/L/day and 99.5% ee at a substrate concentration of 330 g/L. Recently, a D-lactate dehydrogenase mutant from Lactobacillus bulgaricus ATCC 11842 was reported to efficiently catalyze the reduction of compound 7 to compound (R)-8 with high productivity (47.9 mM/h) and high enantioselectivity (>99% ee) when it was coexpressed with formate dehydrogenase in E. coli (Sheng et al. 2014). In addition, Qian et al. (2014) constructed a recombinant P. pastoris to express carbonyl reductase CgKR2 from Candida glabrata for the stereoselective reduction of compound 7 to compound (R)-8, achieving a complete conversion of 1.0 M compound 7 at 0.5 L scale with 77.9% yield and 97.3% ee. These studies cumulatively represent the industrial process for producing compound (R)-8 with a high conversion, ee, and yield.

Antithrombotic drugs

(S)-Clopidogrel (Plavix®), a platelet aggregation inhibitor, is widely used to treat heart attack or stroke caused by the formation of blood clots. To date, several synthetic routes, such as hydrolysis of a racemic nitrile compound, fractional crystallization from the racemic mixture, and asymmetric reduction of methyl o-chlorobenzoylformate (CBFM), have been developed for the preparation of methyl (R)-o-chloromandelate ((R)-CMM), the key chiral synthon for (S)-clopidogrel (Ma et al. 2012). The asymmetric reduction of CBFM to prepare (R)-CMM was considered as a more straightforward and attractive method, which could realize 100% theoretical yield. Whole S. cerevisiae cells were reported to reduce CBFM to (R)-CMM with 96.1% ee (Jeong et al. 2010). A purified ADH from Thermus thermophilus was identified and employed in the effective reduction of CBFM, resulting in a 95% yield and 92% ee using a GDH for cofactor regeneration (Pennacchio et al. 2011). Using genome mining, a novel carbonyl reductase, CgKR1, from C. glabrata was identified and used for the stereospecific reduction of CBFM to (R)-CMM. When combined with GDH for cofactor generation, up to 300 g/L CBFM was almost completely converted to (R)-CMM, with 98.7% ee (Ma et al. 2012). The CBFM concentration was further improved to 500 g/L using whole E. coli cells that coexpressed the aldo-ketoreductase from a Bacillus sp. and GDH, and excellent productivity as high as 812 g/L/day with 88% yield and 99% ee was obtained at 20 °C in a 50-g scale without the external addition of expensive cofactor (Ni et al. 2012). In addition, Xu et al. (2014) described that glycerol improved the thermal and operational stability of an NADPH-dependent aldo-ketoreductase (YtbE), by which the substrate conversion was increased from 62.9 to 98.7% and 70.5 to 96.6% at 0.1 and 1.0 M CBFM, respectively.

Thrombin, a trypsin-like protease enzyme, plays a critical role in the production of a thrombus and thrombin inhibitors and was developed for venous thromboembolism therapy. (R)-3,3-Dimethyl-2-hydroxybutyric acid ((R)-DHBA) is a key chiral intermediate of the thrombin inhibitor. An enzymatic process was constructed for the asymmetric reduction of ethyl 3,3-dimethyl-2-oxobutanoate to (R)-DHBA using commercially available ketoreductase KRED1001, together with a GDH, which was used for cofactor regeneration. (R)-DHBA was isolated and saponified to the corresponding enantiomerically pure hydroxyl acid with 82% yield and 99.5% ee (Nelson et al. 2004).

Antiepileptic drugs

Talampanel, a non-competitive antagonist of the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor, is a novel therapeutic drug for treating epilepsy. (S)-α-Methyl-1,3-benzodioxole-5-ethanol ((S)-MBE) is an important chiral building block for the synthesis of talampanel. The stereoselective bioreduction of 3,4-methylenedioxyphenylacetone (MDA) to (S)-MBE was regarded as the promising route for the scale-up production of (S)-MBE and was initially performed using whole Zygosaccharomyces rouxii and Debaryomyces hansenii cells, both of which had a >90% yield and >99% ee (Erdélyi et al. 2006). Furthermore, a series of ketoreductases were used to prepare (S)-MBE, where purified horse liver alcohol dehydrogenase (HLADH) from horse liver, CPADH from C. parapsilosis, and RS-1 ADH from Rhodococcus species-1 exhibited high activities and high enantioselectivities. HLADH showed a 99% conversion and 95% ee, while CPADH and RS-1 ADH showed 99% conversion and 99% ee (Broussy et al. 2008). Itoh et al. (2012) described an efficient method for producing (S)-MBE by the asymmetric reduction of MDA in an isopropanol-water medium with a recombinant ADH from Leifsonia sp. S749 (LSADH), resulting in a 73% yield and >99% ee. Recently, Simon et al. (2014) investigated various ADHs for the stereoselective bioreduction of MDA, where crude ADH-A from Rhodococcus ruber and purified ADH-T from Theromanaerobacter were proved to be the best enzymes for the preparation of (S)-MBE with high ee (>99%) and good conversion (>96%). Moreover, these biotransformations were performed on a preparative scale (75 mM), and the synthesis of (S)-MBE was achieved in optically pure form (>99%) and obtained with an approximately 95% yield.

β-Lactam antibiotics

Penems and carbapenems, next-generation β-lactam antibiotics, are considered as the “first-line drugs” for the treatment of severe nosocomial infections due to their broad antibacterial spectra, antibacterial activities, and stabilities to most β-lactamases. 4-Aceoxyazetidinone, [(2R,3R)-3-(R)-1-(tert-butyldimethylsilyloxy)ethyl)-4-oxoazetidin-2-ylacetate], is a critical chiral intermediate for the production of penems and carbapenems, where (R)-1,3-butanediol ((R)-10) is a key starting material for producing 4-aceoxyazetidinone derivatives (Fig. 4a). The biocatalytic process for the preparation of compound (R)-10 was achieved by two approaches, stereospecific oxidation of the undesired compound (S)-10 to the racemate and asymmetric reduction of 4-hydroxy-2-butanone (9). Whole C. parapsilosis IFO 1396 and K. lactis IFO 1267 cells were identified for the resolution of racemate 9 to produce (R)-10 and (S)-10, with an optical purity of 97% for C. parapsilosis IFO 1396 and 99% for K. lactis IFO 1267 (Matsuyama et al. 2001). A novel (S)-specific ADH (CpSADH) from C. parapsilosis was characterized. The activity of recombinant CpSADH expressed in E. coli was more than 2-fold higher than the wild-type whole C. parapsilosis cells, resulting in a higher production yield of compound (R)-10. Under the optimized conditions, the yield of compound (R)-10 reached 72.6 g/L, with a molar recovery yield of 48.4% from a 15% racemate reaction mixture and an optical purity of 95% ee (Yamamoto et al. 2002). Furthermore, the enzyme-catalyzed asymmetric reduction of compound 9 to compound (R)-10 was achieved by a newly isolated strain, C. krusei ZJB-09162, which exhibited a 83.9% conversion with a substrate concentration of 45.0 g/L and produced 38.7 g/L compound (R)-10 with 99.0% ee (Zheng et al. 2012).

Fig. 4
figure 4

Bioreduction process for synthesis of 4-aceoxyazetidinone intermediate (Gandolfi et al. 2009; Matsuyama et al. 2001; Rimoldi et al. 2011)

Full size image

(2S,3R)-2-Benzamidomethyl-3-hydroxybutyrate derivatives 11a–11c (Fig. 4b) are also key intermediates in the synthesis of 4-aceoxyazetidinone. The production of enantiomers (2S,3R)-12a–12c has previously been performed using ruthenium-catalyzed asymmetric reduction of the racemic compounds 11a–11c, which required expensive ruthenium catalysts and harsh reaction conditions (Noyori et al. 1989). Biocatalysis-based asymmetric reduction of 11a–11c has been developed to overcome these disadvantages. Cultured plant cells from Parthenocissus tricuspidata and Gossypium hirsutum were previously reported to convert compounds 11a and 11b to the corresponding (2R,3S) and (2S,3S) configurations with high stereoselectivities. The reduction by P. tricuspidata cells produced compound (2R,3S)-12a with a 97% yield, 100% de, and >99% ee. On the other hand, the reduction with G. hirsutum cells exhibited a (2S,3S)-selectivity to produce compound (2S,3S)-12a with a 98% yield, 84% de, and >99% ee (Shimoda et al. 2009). Furthermore, the yeast strains K. marxianus var. lactis CL69 and Pichia glucozyma CBS 5766 were investigated in the stereoselective reduction of compound 11b. Under the optimized conditions, whole K. marxianus var. lactis CL69 cells produced compound (2R,3S)-12b with >99% ee and 98% de, whereas whole P. glucozyma CBS 5766 cells produced compound (2S,3S)-12b with >99% ee and 86% de (Gandolfi et al. 2009). In addition, the reduction of compound 11c to compound 12c was achieved using whole P. glucozyma CBS 5766 and K. marxianus CBS 1553 cells, with high yields and stereoselectivity. The former generated compound (2R,3S)-12c with >99 ee and 70% de, whereas the latter furnished the enriched diastereoisomer (2R,3R)-12c with >99 ee and 88% de (Rimoldi et al. 2011). Notably, Codexis described a synthetic route for compound 11a using ADHs from L. brevis and Devosia riboflavin as catalysts, providing compound (2S,3R)-12a with >99% ee and >89% de (Yasohara et al. 2009). Chen et al. (2015) also discovered bgADH2 from the SDR-producing strain Burkholderia gladioli and showed that it could be used for the asymmetric synthesis of compound (2S,3R)-12a with 99% ee and 98.5% de.

(R)-3-Hydroxythiolane is a key intermediate in sulopenem, and its original production process was developed by Pfizer (Volkmann et al. 1992). Codexis successfully applied enzyme evolution technologies a KRED, resulting in unprecedented high enantioselectivity (>99% ee) for reduction of tetrahydrothiophene-3-one to (R)-3-hydroxythiolane. This process has been successfully scaled up to 100 kg level and applied to the industrial production (Liang et al. 2009).

Non-steroidal anti-inflammatory drugs

Profens are a subclass of non-steroidal anti-inflammatory drugs, which are non-selective inhibitors of the cyclooxygenase involved in prostaglandin synthesis via the arachidonic acid pathway. (2S)-2-Arylpropanols (14a–14f) can be oxidized to (2S)-2-arylpropionic acids, which are used as key intermediates in the synthesis of Profens. Compounds 14a and 14b (Fig. 5) were obtained with high yields and enantiomeric ratios using HLADH as the catalyst through dynamic kinetic resolution in organic solvents or in phosphate-buffered aqueous solution (Giacomini et al. 2007). Purified HLADH was further used to prepare compounds 14a–14f by the bioreduction of 2-arylpropionic aldehydes (13a–13f) under mild conditions (Galletti et al. 2010). High yields and high enantioselectivity towards the (S)-enantiomer were achieved in a dynamic kinetic resolution process by coupling the yeast ADH-catalyzed or HLADH-catalyzed reduction with a chemical base-catalyzed racemization of the unreacted aldehydes, which is a promising alternative process and can overcome the 50% yield limitation in classical kinetic resolution (Galletti et al. 2010). Recently, a histidine-tagged HLADH was successfully purified and immobilized for the improvement of enantioselectivity and reusability (Quaglia et al. 2013). Purified SsADH-10 from hyperthermophile Sulfolobus solfataricus was also employed in the dynamic kinetic resolution to produce compounds 14a–14f with high ee and a preference for the (S)-enantiomer (Friest et al. 2010).

Fig. 5
figure 5

Bioreduction process for synthesis of profen intermediates (Galletti et al. 2010; Friest et al. 2010; Quaglia et al. 2013)

Full size image

Anticancer drugs

Taxol (Paclitaxel®) is an anticancer drug with great potential that was approved by FDA to treat ovarian and breast cancers by blocking microtubule disassembly. (2R,3S)-N-Benzoyl-3-phenyl isoserine ethyl ester ((2R,3S)-16), a chiral intermediate of the C-13 side chain of taxol, was prepared by the stereoselective reduction of 2-keto-3-(N-benzoylamino)-3-phenyl propionic acid ethyl ester (15) (Fig. 6). The stereoselective reductions were performed by whole H. polymorpha 13865 and Hansenula fabianii 13894 cells, and product yield >80% and ee >98% were ultimately obtained (Patel et al. 1995). Yeast-catalyzed reduction of compound 15 to generate compound (2R,3S)-16 was also reported (Kearns and Kayser 1994). A new ADH cloned from Clostridium acetobutylicum was used for the asymmetric reduction of 3-phenyl-2-chloro-3-oxopropionic acid ethyl ester (17), providing a building block (2S,3R)-2-chloro-3-hydroxy-3-phenylpropionic acid ethyl ester ((2S,3R)-18) with 95% yield, 95% de, and 99% ee (Applegate et al. 2011).

Fig. 6
figure 6

Bioreduction process for synthesis of taxol side chain intermediates (Applegate et al. 2011; Kearns and Kayser 1994)

Full size image

Crizotinib (PF-2341066), a dual inhibitor of the mesenchymal-epithelial transition (MET) and anaplastic lymphoma kinase (ALK), was approved by the FDA for the treatment of advanced non-small cell lung cancer (NSCLC). (S)-1-(2,6-Dichloro-3-fluorophenyl)-ethanol ((S)-DFPE) is a key intermediate required for the preparation of crizotinib. (S)-DFPE can be synthesized by enzymatic resolution, chemical esterification, and stereoselective bioreduction of 2,6-dichloro-5-fluorophenacetone (DFPA). Codexis developed a series of engineered ketoreductases from L. kefir ADH (ADH-LK) and L. brevis ADH (ADH-LB) to convert DFPA to (S)-DFPE, with >99% ee (Liang et al. 2013). Furthermore, (S)-DFPE was produced on a preparative scale using 100 g/L of DFPA in a 0.5 L reaction system, resulting in a product yield of 90% and ee of 99.9% (Liang et al. 2013).

Brivanib, a novel dual inhibitor of the VEGF receptor (VEGFR) and FGF receptor (FGFR) signaling pathways, is currently under clinical evaluation for the treatment of hepatocellular carcinoma (HCC), which is the fifth most common primary neoplasm. (R)-1-[4-(4-Fluoro-2-methyl-1H-indol-5-yloxy)-5-methylpyrrolo[2,1-f][1,2,4]triazin-6-yloxy]-propan-2-ol ((R)-MPO) is an important precursor for the synthesis of brivanib. Codexis provided a series of engineered ketoreductases from ADH-LK and ADH-LB to reduce 1-[4-(4-fluoro-2-methyl-1H-indol-5-yloxy)-5-methyl-pyrrolo[2,1-f][1,2,4]triazin-6-yloxy]-propan-2-one to (R)-MPO with high enantioselectivity (Ching et al. 2013). Using glucose/glucose dehydrogenase or isopropanol for cofactor recycling, the product was obtained with >99% ee at 12 g/L substrate concentration (Ching et al. 2013).

The type 1 insulin-like growth factor receptor (IGF-1R) plays an important role in the establishment and maintenance of the transformed phenotype. It has strong antiapoptotic ac-tivity and a significant influence on the control of cell and body size. Downregulation of IGF-1R leads to massive apo-ptosis of cancer cells, which is an attractive target for cancer therapy. (S)-2-chloro-1-(3-chlorophenyl)ethanol ((S)-CCPE) is an important intermediate in the synthesis of IGF-1R inhibitors. Hanson et al. (2005) investigated 119 microorganisms for the asymmetric reduction of 2-chloro-1-(3-chlorophenyl)ethanone (CCPEO) and found that whole H. polymorpha ATCC 58401 and Rhodococcus globerulus ATCC 21505 cells exhibited relatively high enantioselectivity for producing the desired (S)-CCPE, with 73.8 and 71.8% ee, respectively. Hanson et al. (2005) also found that whole E. coli cells containing a recombinant ketoreductase from H. polymorphato were able to catalyze the synthesis of (S)-CCPE with 100% ee. Furthermore, the coexpression of ketoreductase and glucose-6-phosphate dehydrogenase reduced CCPEO to the corresponding (S)-CCPE, with 89% yield and 100% ee (Hanson et al. 2005). In addition, whole G. candidum and Candida ontarioensis cells were applied to the asymmetric synthesis of (R)-CCPE, with >98% ee and >94% yield (Hamada et al. 2001).

Antiemetic drugs

Aprepitant (Emend®) is a neurokinin (NK1) receptor antagonist that is available to prevent chemotherapy-induced nausea and vomiting (CINV) and is used as a relatively new antiemetic drug. (R)-1-[3,5-Bis(trifluoromethyl)phenyl]ethanol ((R)-BTPE) is a pharmaceutically important alcohol intermediate for the synthesis of aprepitant. Asymmetric reduction of 3,5-bis(trifluoromethyl)acetophenone (BTAP) to (R)-BTPE using biocatalysts has been studied (Wang et al. 2014a). Various commercially available ADHs, including HLADH, YADH, BYADH, and L. kefir ADH, were tested for the asymmetric reduction of BTAP, of which only L. kefir ADH showed an activity with 100% ee (Gelo-Pujic et al. 2006). Furthermore, several microorganisms were screened to prepare (R)-BTPE, and whole Penicillium expansum EBK-9 cells were identified as the most productive strain, which is able to reduce 3.35 g/L BTAP to (R)-BTPE with a 76% yield and >99% ee (Kurbanoglu et al. 2009). Whole Leifsonia xyli HS0904 and Trichoderma asperellum ZJPH0810 cells were subsequently shown to produce (R)-BTPE with BTAP concentrations of 70 and 50 mM, respectively. Under the optimized conditions, L. xyli HS0904 exhibited a 62% yield and 99.4% ee, whereas T. asperellum ZJPH0810 exhibited a 93.4% yield and 98% ee (Li et al. 2013; Wang et al. 2011b). More recently, a novel strain Chryseobacterium sp. CA49 that was capable of reducing BTAP to (R)-BTPE was reported, and the substrate concentration could be loaded up to 50 g/L (Liu et al. 2014). After sequencing the Chryseobacterium sp. CA49 genome, a recombinant ChKRED20 was identified and shown to synthesize (R)-BTPE with a >99% conversion and >99.9% ee at a BTAP concentration of 200 g/L using 2-propanol for cofactor regeneration, indicating the great potential for the industrial-scale application of ChKRED20 and production of (R)-BTPE (Liu et al. 2014).

Anti-AIDS drugs

Atazanavir, an aza-dipeptide, is a potent human immunodeficiency virus type-1 (HIV-1) protease inhibitor approved for the clinical therapy of acquired immune deficiency syndrome (AIDs). (1S,2R)-[3-Chloro-2-hydroxy-1-(phenylmethyl)propyl]carbamic acid, 1,1-dimethylethyl ester ((1S,2R)-CADE) is a key chiral intermediate for atazanavir synthesis. The microbial diastereoselective reduction of (1S)-[3-chloro-2-oxo-1-(phenylmethyl)propyl]carbamic acid, 1,1-dimethyl-ethyl ester to (1S,2R)-CADE was performed using Rhodococcus, Brevibacterium, and Hansenula strains (Patel et al. 2003). When whole Rhodococcus erythropolis SC 13845 cells were used as the biocatalyst, (1S,2R)-CADE was obtained with 95% yield, 98.2% de, and 99.4% ee.

Adrenergic receptor agonists

(R)-Denopamine, [(R)-(−)-α-(3,4-dimethoxyphenethylaminomethyl)-4-hydroxybenzyl alcohol] and (R)-salmeterol, [(R)-1,3-benzenedimethanol, 4-hydroxy-α-[((6-(4-phenyl-butoxy)hexyl)amino)methyl], are important β-adrenoreceptor agonists that are approved to treat congestive heart failure. The asymmetric reduction of ketones 19a–19e is an efficient way to obtain the corresponding alcohols (R)-20a–20e, which are key intermediates in the production of (R)-denopamine and (R)-salmeterol (Fig. 7). The enzyme from Daucuscarota root was previously used as a biocatalyst to synthesize compounds (R)-20a and (R)-20b with high enantioselectivity (Yadav et al. 2002). Whole R. rubra cells were used for the reduction of compounds 19c and 19d to obtain compounds (R)-20c and (R)-20d with 95% ee and 78–80% yield (Goswami et al. 2002). Furthermore, a concise approach was developed for the synthesis of (R)-salmeterol by converting compounds 19d and 19e to compounds (R)-20d and (R)-20e using a novel Rhodococcus sp. 1-0130 strain, in which both the aldehyde and ketone groups are reduced (Zhang et al. 2013a).

Fig. 7
figure 7

Bioreduction process for synthesis of (R)-denopamine and (R)-salmeterol intermediate (Goswami et al. 2002; Yadav et al. 2002; Zhang et al. 2013a)

Full size image

Other optically pure substituted α-halohydrins also play important roles in the synthesis of β-adrenergic receptor agonists. Whole S. cerevisiae CGMCC 2.396, Aspergillus sydowii Ce19, and Pichia minuta JCM 3622 cells showed high activity and enantioselectivity for the synthesis of (R)-α-halohydrins or (S)-α-halohydrins (>99% ee) (Lin et al. 2009; Rocha et al. 2010; Tokoshima et al. 2013). A substrate-tolerant carbonyl reductase from Kluyveromyces thermotolerans (KrtCR) was successfully applied to the asymmetrical reduction of different ketones to corresponding chiral alcohols with 99% ee (Xu et al. 2012). Using α-choloacetophenone as a representative substrate, as much as 154 g/L of the substrate was successfully reduced within 12 h, and a yield of 92%, an enantiopurity of >99% ee, and a turnover number of 5000 were observed, indicating that KrtCR has great potential in the preparation-scale synthesis of valuable α-halohydrins as versatile chiral synthons (Xu et al. 2012).

(R)-Phenylephrine ((R)-PE), a selective α1-adrenergic receptor agonist, is widely used as a substitute for pseudoephedrine and has been approved for the treatment of the common cold, without the side effects caused by other ephedrine adrenergic drugs. Codexis provided a series of engineered ketoreductases from L. kefir ADH for the stereoselective reduction of 1-(3-hydroxyphenyl)-2-(methylamino)ethanone (HPMAE) to (R)-PE, which exhibited a relative activity that was at least 10-fold higher than the wild-type enzyme and an improved stereoselectivity with >99% ee (Alvizo et al. 2012). Furthermore, an NADPH-dependent short-chain dehydrogenase/reductase from Serratia marcescens BCRC 10948 was applied to synthesize (R)-PE with >99% ee. The productivity and yield were 0.57 mmol/h and 51.06%, respectively, at substrate concentration of 10 mM (Peng et al. 2014). In addition, Peng et al. (2014) discovered another short-chain ADH from Serratia quinivorans BCRC 14811 for the reduction of HPMAE to generate (S)-PE with >99% of ee, 86.6% of conversion, and 20.2 mmol/h of productivity.

Biological applications of ADHs to the preparation of fine chemicals

γ-Valerolactone

γ-Valerolactone (GVL) has been identified as a potential intermediate for the production of fuels. Targeted GVL production can be produced from methyl levulinate using ADHs. Díaz-Rodríguez et al. (2013) employed different ADHs from R. ruber (ADH-A), L. brevis (LBADH), and Ralstonia sp. (RasADH) to synthesize a series of enantiopure hydroxyl esters and lactones, and enantioenriched GVL was obtained with >99% conversion and >99% ee using LBADH and ADH-A as biocatalysts. During this process, the cofactor was recycled by the addition of isopropanol (Díaz-Rodríguez et al. 2013).

(2S,3S)-2,3-butanediol

2,3-Butanediol is a promising bulk chemical with extensive industrial applications. ADHs show the capability to reduce diacetyl to acetoin and then to three isomers of 2,3-butanediol, including meso-2,3-butanediol, (2R,3R)-2,3-butanediol, and (2S,3S)-2,3-butanediol, where optically pure 2,3-butanediol with two vicinal stereogenic centers is used as an important building block for chiral compounds (Ji et al. 2009). Wang et al. (2014b) developed a novel R. erythropolis WZ010 strain that was capable of reducing diacetyl to optically pure (2S,3S)-2,3-butanediol, in which the purified ADH displayed absolute stereospecificity in the reduction of diacetyl to (2S,3S)-2,3-butanediol via (S)-acetoin with 98% yield and 100% ee. Coupled with the stereoselective oxidation of 1-phenylethanol in the process of preparing (2S,3S)-2,3-butanediol, (R)-1-phenylethanol was simultaneously produced with 49.4% yield and 99.9% ee (Wang et al. 2014c).

Phenylethanols

Optically active phenylethanols and their derivatives are important intermediates in the synthesis of enantiomerically pure pharmaceuticals and fine chemicals. Prochiral acetophenone is usually used as a model substrate for the isolation of microorganisms with ADH activity. Singh et al. (2012) employed whole Metschnikowia koreensis cells for the asymmetric reduction of various substituted acetophenones. With 25 mM acetophenone as substrate, 92% conversion and >99% ee were obtained. Furthermore, a non-conventional thermotolerant K. marxianus cell was shown to be a promising biocatalyst for the production of optically active phenylethanols, and it was used for the bioreduction of different acetophenone derivatives to (S)-alcohols (Vitale et al. 2013). In addition, an immobilized short-chain ADH from T. thermophilus HB27 was used for the asymmetric reduction of 2,2,2-trifluoroacetophenone to (S)-α-(trifluoromethyl)benzyl alcohol with 96% ee (Rochamartín et al. 2012), and a purified ADH from Thermococcus kodakarensis KOD1 (TkADH) could reduce 2,2,2-trifluoroacetophenone to (R)-2,2,2-trifluoro-1-phenyl ethanol with high enantioselectivity (>99.6% ee) (Wu et al. 2013).

Conclusion

ADHs are important enzymes that catalyze interconversion between alcohols and aldehydes or ketones under mild conditions. The fundamental research and wide applications of ADHs in the preparation of enantiomerically pure compounds have drawn more attention in recent years. With the development of screening approaches, such as metagenome technology and genome mining, novel ADHs are expected to be identified and used as promising biocatalysts. The combination of rational and irrational protein design technologies provides a promising strategy to improve the catalytic properties of ADHs for their industrial applications in the production of high-value chemicals and pharmaceuticals.