Asymmetric synthesis of (S)-3-chloro-1-phenyl-1-propanol using Saccharomyces cerevisiae reductase with high enantioselectivity
- First Online:
- Cite this article as:
- Choi, Y.H., Choi, H.J., Kim, D. et al. Appl Microbiol Biotechnol (2010) 87: 185. doi:10.1007/s00253-010-2442-5
- 293 Views
3-Chloro-1-phenyl-1-propanol is used as a chiral intermediate in the synthesis of antidepressant drugs. Various microbial reductases were expressed in Escherichia coli, and their activities toward 3-chloro-1-phenyl-1-propanone were evaluated. The yeast reductase YOL151W (GenBank locus tag) exhibited the highest level of activity and exclusively generated the (S)-alcohol. Recombinant YOL151W was purified by Ni-nitrilotriacetic acid (Ni-NTA) and desalting column chromatography. It displayed an optimal temperature and pH of 40°C and 7.5–8.0, respectively. The glucose dehydrogenase coupling reaction was introduced as an NADPH regeneration system. NaOH solution was occasionally added to maintain the reaction solution pH within the range of 7.0–7.5. By using this reaction system, the substrate (30 mM) could be completely converted to the (S)-alcohol product with an enantiomeric excess value of 100%. A homology model of YOL151W was constructed based on the structure of Sporobolomyces salmonicolor carbonyl reductase (Protein Data Bank ID: 1Y1P). A docking model of YOL151W with NADPH and 3-chloro-1-phenyl-1-propanone was then constructed, which showed that the cofactor and substrate bound tightly to the active site of the enzyme in the lowest free energy state and explained how the (S)-alcohol was produced exclusively in the reduction process.
KeywordsAntidepressant drugsChiral intermediateDocking modelEnantioselectivityReductase
The asymmetric reduction of ketones to optically pure alcohols has attracted a great deal of attention due to their possible use as chemotherapeutic drugs and chiral building blocks. Many oxidoreductases (EC 1.1.1.-) from a variety of microorganisms may be capable of performing the reduction of carbonyl groups with chemo-, regio-, and stereoselectivity (Goldberg et al. 2007; Schroer et al. 2007). Enzymatic ketone reductions can be achieved using isolated enzymes (Goldberg et al. 2007; Inoue et al. 2005; Kaluzna et al. 2005) or whole-cell systems (Ema et al. 2006; Moore et al. 2007; Xu et al. 2006). Saccharomyces cerevisiae (baker's yeast) carbonyl reductases have been previously reported to exhibit catalytic activity, converting some ketone substrates into enantiomeric alcohols (Kayser et al. 2005; Kaluzna et al. 2005).
Biotransformation of ketone substrates with reductase enzymes requires the presence of the cofactor NAD(P)H. Cofactor regeneration can be conducted via an enzyme-coupled (Ema et al. 2006; Ernst et al. 2005; Moore et al. 2007) or a substrate-coupled (Inoue et al. 2005; Makino et al. 2005; Schroer et al. 2007) approach. The former approach is characterized by the use of a second enzyme that catalyzes the oxidation of a cosubstrate to regenerate the reduced cofactor, while the latter process employs only one enzyme for the production of the desired compound and the regeneration of the cofactor.
In this study, we generated ten different microbial reductases and screened these enzymes for their ability to produce a CPPO enantiomer. The properties of the carbonyl reductase thus selected, YOL151W, were characterized, and a reductase–glucose dehydrogenase coupling reaction was conducted in order to generate (S)-CPPO. Finally, a homology model of YOL151W and its docking model with cofactor and substrate were constructed to explain the detailed mechanism of the enantioselective reduction process.
Materials and methods
3-CPP, (R)- and (S)-CPPO, glucose dehydrogenase (GDH, Thermoplasma acidophilum), NAD(P)H, and NAD(P)+ were purchased from Sigma-Aldrich Co. (St. Louis, MO). All other chemicals were of analytical grade.
Cloning and expression of reductase gene
S. cerevisiae (KCTC 7904) chromosomal DNA was acquired from the Korea Research Institute of Bioscience and Biotechnology (KRIBB). The Leuconostoc citreum strain (KCTC 3721), which was supplied by the Biological Resource Center (Korea), was cultivated in lactobacilli formulations of deMan, Rogosa and Sharpe (MRS) broth at 30°C, and its genomic DNA was isolated by cell lysis and phenol/chloroform extraction. The Corynebacterium glutamicum (ATCC 13035) oxidoreductase gene (GenBank ID: NP-601323.1) was acquired from KRIBB.
Screening of reductases suitable for the reduction of 3-chloro-1-pheny-1-propanone to 3-chloro-1-phenyl-1-propanol
GenBank locus tag
Reductase activity (U/mL)
Alpha–keto amide reductase
Hydroxy acid reductase
Assay of reductase and glucose dehydrogenase activity
Reductase activity was assayed at 30°C by measuring the decrease in absorbance at 340 nm for 10 min using a spectrophotometer. The reaction mixture (1 mL) consisted of 1 mM 3-CPP (100 mM stock in DMSO), 0.2 mM NAD(P)H, 50 mM Tris–HCl buffer (pH 7.5), and 50 μL of cell-free extract. One unit of enzyme was defined as the quantity of enzyme required to catalyze the oxidation of 1 μmol NAD(P)H in 1 min at 30°C.
The glucose dehydrogenase oxidation reaction mixture (1 mL) consisted of 10 mM glucose, 2 mM NADP+, 50 mM Tris–HCl buffer (pH 7.5), and 0.05 μL of commercial glucose dehydrogenase. The reaction rate was monitored with a spectrophotometer on the basis of the increase in absorbance at 340 nm for 10 min at 30°C. One unit of enzyme was defined as the amount required to reduce 1 μmol of NADP+ in 1 min at 30°C.
Purification of reductase YOL151W
Reductase no. 8 (YOL151W) from baker's yeast was identified as an appropriate enzyme for the reduction of 3-CPP to CPPO in an enantioselective manner. The reductase protein in the cell extract was purified as follows. Firstly, the cell-free extract was loaded onto a Ni-NTA column (QIAGEN GmbH, Hilden, Germany) equilibrated with a 50-mM Tris–HCl buffer (pH 7.8) containing 50 mM imidazole and 300 mM NaCl. The recombinant reductase was then eluted from the column by the application of a 250-mM imidazole buffer. The active fractions were then collected and desalted with a PD-10 desalting column (GE Healthcare Bio-Sciences AB, UK).
Effects of temperature and pH
The effects of temperature and pH were evaluated using the purified YOL151W enzyme. Reactions were monitored spectrophotometrically by measuring the decrease in NADPH absorbance at 340 nm with 42.3 μg of enzyme. The reaction rate was measured at various temperatures (10–60°C). Meanwhile, in order to assess temperature stability, the enzyme was preincubated for 30 min at 10–50°C, then the remaining activity was assayed at 30°C. The following buffers (50 mM) were utilized to assess the effects of pH: pH 3–6, acetic acid/sodium acetate; pH 6–8, KH2PO4/K2HPO4; pH 7.5–10, Tris–HCl; pH 11–12, KCl–K2HPO4–K3PO4 buffers. In order to determine the pH stability, the enzyme was preincubated for 30 min in the pH buffers listed above at 0°C, then adjusted to pH 7.5, under which condition, the residual activity was determined.
Effect of buffer and DMSO concentrations
The effect of the concentration of Tris–HCl buffer (pH 7.5) on the reductase reaction rate was measured. The buffer was used at a concentration of 50, 100, 200, 250, or 300 mM and included 1 mM 3-CPP, 0.2 mM NADPH, and 10 μL of cell-free extract.
To evaluate the effect of DMSO, a reaction solution (100 mM Tris–HCl buffer, pH 7.5) containing 1 mM 3-CPP, 0.2 mM NADPH, 10 μL of cell-free extract, and DMSO at a final concentration of 1, 2, 3, 5, or 10% (v/v) was incubated, and the reaction rate was determined by observing the decrease in absorbance at 340 nm.
Enzymatic coupling reaction and reaction mixture analysis
The enantioselectivity of the enzymatic reduction of 3-CPP ketone was evaluated using an NADPH-recycling system. The general procedure was as follows: d-glucose (1.5-fold higher concentration than 3-CPP), 20 units of glucose dehydrogenase, 1 mM NADPH, various concentrations of 3-CPP in DMSO, and 10 units of reductase YOL151W were mixed in a total volume of 10 mL of 100 mM Tris–HCl buffer (pH 7.5), and the mixture was incubated at 30°C. The pH of the reaction mixture was monitored with a pH meter and maintained at 7.0–7.5 via the addition of 1 M NaOH. Four hundred microliter aliquots of the reaction mixture were sampled, mixed with 1.2 mL of ethyl acetate, and centrifuged at 10,000×g for 10 min. The upper phase containing ethyl acetate (1 mL) was obtained, combined with MgSO4, filtered, and then dried via vacuum pump centrifugation. The sample was resuspended in ethyl acetate (500 μL) and analyzed by a high-performance liquid chromatography (HPLC) system equipped with a CHIRALPAK IB column (Daicel Chemical Industries, Ltd., Tokyo, Japan). Separate peaks for the ketone substrate and (R)- and (S)-alcohols were obtained using n-hexane and 2-propanol (95:5, by volume) as the mobile phase at a flow rate of 0.8 mL/min. Detection was performed at 254 nm. Relative quantities were calculated on the basis of the peak area, which was suitably calibrated with standards of known concentration. Enantiomeric excess (ee) values were calculated from the alcohol products.
The initial 3D structural model of YOL151W was constructed based on the crystal structure of Sporobolomyces salmonicolor carbonyl reductase (SSCR; Protein Data Bank code: 1Y1P) (Kamitori et al. 2005). The target structure was modeled in silico by means of a structural homology search using HHpred/HHsearch (Söding et al. 2005) and homology modeling using MODELLER (Sali and Blundell 1993), then optimized using FoldX (Schymkowitz et al. 2005). The YOL151W active site was examined by comparison with the corresponding X-ray template of NADPH-bound SSCR. The geometry of the substrate (CPP) and cofactor (NADPH) was optimized using the Hartree–Fock method with a 6-31G* basis set as implemented in the Gaussian 03 program (Frisch et al. 2004). The restrained electrostatic potential procedure of the antechamber module from the AMBER suite was used to generate input files with charges for docking programs (Case et al. 2005). The result was used as a valid input for the AutoDock ligand preparation procedure.
Docking was performed with AutoDock (version 4.0), using the implemented empirical free energy function and the Lamarckian genetic algorithm (Huey et al. 2007). The best docked conformation was the one found to have the lowest binding energy and the greatest number of members in the cluster, indicating good convergence. The best orientation was identified and optimized using the scoring function based on the AMBER force field FF99 and energy minimization according to the Nelder and Mead algorithm (Case et al. 2005) for induced-fit simulation. The parameters embedded in the AMBER package were used for energy minimization and molecular dynamics, (Wang et al. 2004) and then, molecular dynamics (MD) simulations were performed. In addition, the docking of CPP and NADPH to our target protein was initially made according to predicted topological binding sites by several algorithms (Huang and Schroeder 2008).
Screening for enantioselective reductases
Many microbial reductases have been isolated, and their genes are cloned, sequenced, and deposited in the GenBank database. In this study, we generated ten different microbial reductases in E. coli cells (Table 1). Eight of these originated from S. cerevisiae, and the remaining two enzymes were obtained from Leuconostoc and Corynebacterium species.
Although the expression levels of the recombinant enzymes were somewhat variable, the majority of target protein bands could be detected in soluble fractions on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE; see Supplementary data 1). The reductase activity in soluble fractions was determined via a spectrophotometric method according to the basic principle that NAD(P)H is oxidized when 3-CPP is reduced by the enzyme. Although the results differed according to which recombinant enzyme was tested, the majority evidenced measurable reductase activity (0.2–3.2 U/mL) toward 3-CPP (Table 1).
HPLC analysis was subsequently performed to detect the reaction product 3-CPPO. Among the enzymes tested, only reductase no. 8 produced a significant amount of (S)-CPPO.
Purification and characterization of reductase
Recombinant reductase no. 8 was purified in an effort to assess its reaction properties. As the enzyme harbored a His-tag at its C terminus, it could be readily purified via two sequential purification steps: Ni-NTA and PD-10 column chromatography. The purified enzyme had a specific activity of 0.99 U/mg toward the 3-CPP substrate.
In order to convert large quantities of substrate using a reductase, the reaction process should be coupled to a GDH enzyme reaction. The latter could regenerate NAD(P)H, a cofactor of the reductase. However, as this coupling reaction continues, glucose, a substrate of GDH, is oxidized into gluconic acid, and the solution pH is decreased, making the selection of buffer system critical. Therefore, the activity of reductase no. 8 was measured in an increasing molar concentration of Tris–HCl buffer. Reductase activity increased in concentrations of up to 100 mM, but decreased in concentrations over 200 mM (Fig. 1e).
In addition, since low substrate solubility presents another problem for the enzyme reaction, the substrate was dissolved in DMSO at a concentration of 1 M and then utilized for the reaction. Reductase activity was measured in increasing concentrations of DMSO in order to elucidate its effects (Fig. 1f). DMSO reduced enzyme activity only slightly up to a concentration of 3%, but its inhibitory effect increased significantly at concentrations over 5%.
Kinetic parameters of YOL151W reductase
kcat/Km (min−1 mM−1)
Production of (S)-CPPO by reductase–GDH coupling reaction system
In order for the ketone reduction reaction to continue, sufficient NADPH must be supplied. Since this cofactor is a relatively expensive material, an NADPH regeneration system is required. Toward this end, we employed a GDH coupling reaction.
As the coupling reaction continued, gluconic acid accumulated, and the solution pH consequently decreased. The optimal pH of reductase no. 8 was 7.5–8.0, and its activity rapidly decreased at pH 6.0. In this experiment, the solution pH was maintained between pH 7.0–7.5 via occasional additions of a 1 M NaOH solution.
Homology modeling and evaluation
By searching a similar structure for YOL151W using HHpred/HHsearch (Söding et al. 2005), we found an X-ray crystal structure, 1Y1P (Kamitori et al. 2005) as a potential homology template. The percent of sequence identity between the template and target was calculated to be 31.8% by Jotun–Hein method using DNAStar MegAlign program (see Supplementary data 2). We superimposed the active site of the target protein on the template (see Supplementary data 3).
The final model for YOL151W was assessed by Profile-3D (Luthy et al. 1992) and Procheck (Laskowski et al. 1993). Using Profile-3D, its compatibility score was 195 (the expected score for a protein of this size is 208). By employing Procheck, the reliability of the backbone torsion angles ψ–φ of the target protein was examined. The percentage of ψ–φ angles in the core Ramachandran region was 85.8%, which is comparable to that of template (89.7%). These data indicated that the 3D model (Fig. 4a) was reliable for further docking studies.
Docking of CPP to YOL151W
In order to explain the catalytic activity of YOL151W toward CPP in wet experiments, we performed a docking study (Fig. 4b). In the CPP–YOL151W complex, CPP is stabilized by hydrogen bonding and hydrophobic interactions in the center of the active site. Two hydrogen bonds are formed between them: the carbonyl oxygen of CPP binds to Ser127 and Tyr165. The complex has a favorable total interaction energy of −41.68 kcal/mol, in which the van der Waals and electrostatic energies are −33.09 and −8.59 kcal/mol, respectively. As seen in Fig. 4b, Phe132 and Met134 are positioned close to the benzyl ring of CPP, resulting in the second largest interaction energy, −3.35 kcal/mol (the strongest interactions come from the NADPH group), via π–π and sulfur–π interactions. Phe89 is also holding the ethyl chloride group of CPP via van der Waals interaction.
(R)- and (S)-CPPO are chiral intermediates of antidepressant drugs including fluoxetine as mentioned above. These compounds could be generated from a prochiral ketone via an enantioselective reduction reaction, as depicted in Scheme 1. For the efficient production of (R)- and (S)-CPPO, a suitable reductase enzyme should be employed.
In this study, the reductase YOL151W was selected from among ten different microbial reductases for the production of (S)-CPPO. The gene GRE2 encoding YOL151W has previously been reported (Kayser et al. 2005) as have the substrate specificity and enantioselectivity of the reductase toward several compounds (Ema et al. 2006). Ema et al. reported YOL151W could produce ethyl-(R)-3-hydroxy-3-phenylpropanoate, an intermediate of fluoxetine with ee value of 70%. However, there has not yet been a report addressing the reactivity of the enzyme toward 3-CPP, the chiral intermediate of fluoxetine.
As our experimental data shows, reductase YOL151W exhibited both regiospecificity and enantioselectivity toward the carbonyl group of the 3-CPP molecule. The enzyme catalyzed the enantioselective reduction of the substrate, producing (S)-CPPO exclusively.
The reductase activity as measured by the spectrophotometric method was different from that obtained via HPLC (Table 1). We could not explain the exact reason why the different activities were observed, however, it seemed to be caused by the different assay systems; spectrophotometric assay (continuous assay) and HPLC assay (stop–sample, discontinuous assay).
Some reductase reactions have occasionally been conducted using whole-cell systems. In this system, the reductase and glucose dehydrogenase enzymes were coexpressed in the same cell, thereby continuously recycling NAD(P)H (Ema et al. 2006; Moore et al. 2007). Additionally, various two-phase ionic liquid systems have been employed as reaction solvents to avoid substrate and/or product inhibition (Brautigam et al. 2007; Pfruender et al. 2006).
Collectively, the present research shows that, together with a glucose dehydrogenase coupling reaction, the recombinant yeast reductase YOL151W can be utilized to produce (S)-CPPO, and we might propose that it could be used to generate high concentrations of this product by adopting a whole-cell system or ionic liquid solvent system approach.
This work was supported by the 21C Frontier Microbial Genomics and Applications Center Program, Ministry of Education, Science and Technology, Republic of Korea and the 2009 Research Fund of the Catholic University of Korea.