Gene Cloning and Characterization of Two NADH-Dependent 3-Quinuclidinone Reductases from Microbacterium luteolum JCM 9174

Abstract

We used the resting-cell reaction to screen approximately 200 microorganisms for biocatalysts which reduce 3-quinuclidinone to optically pure (R)-(−)-3-quinuclidinol. Microbacterium luteolum JCM 9174 was selected as the most suitable organism. The genes encoding the protein products that reduced 3-quinuclidinone were isolated from M. luteolum JCM 9174. The bacC gene, which consists of 768 nucleotides corresponding to 255 amino acid residues and is a constituent of the bacilysin synthetic gene cluster, was amplified by PCR based on homology to known genes. The qnr gene consisted of 759 nucleotides corresponding to 252 amino acid residues. Both enzymes belong to the short-chain alcohol dehydrogenase/reductase (SDR) family. The genes were expressed in Escherichia coli as proteins which were His tagged at the N terminus, and the recombinant enzymes were purified and characterized. Both enzymes showed narrow substrate specificity and high stereoselectivity for the reduction of 3-quinuclidinone to (R)-(−)-3-quinuclidinol.

INTRODUCTION

Many methods have been used for the synthesis of optically pure isomers, including enantioselective synthesis, derivatization of natural compounds, and the optical resolution of racemic compounds. Enantioselective organic synthesis is the most efficient and attractive method for producing chiral starting materials for pharmaceuticals and agrochemicals. Chiral metal catalysts such as BINAP-Ru (1) and chiral Co(II) salen complex (2, 3) have been successfully used in producing chiral alcohols or chiral diols from various ketones or epoxides. However, these catalysts are costly and leave trace metal contamination in the products, and they are sometimes difficult to handle. The enzymatic reduction of ketones to optically pure alcohols is more environmentally sustainable and thus more attractive for pharmaceutical manufacturing (4). Several excellent methods using biocatalysts have been reported for producing chiral alcohols (5–10).

(R)-(−)-3-Quinuclidinol, which has a double-ring structure containing nitrogen, is a valuable intermediate for pharmaceuticals. It has been used as the chiral synthone for a cognition enhancer, a bronchodilator, and a urinary incontinence agent (11). The following enzymes have been reported to catalyze the reduction of 3-quinuclidinone to (R)-(−)-3-quinuclidinol: tropinone reductase, from the plant henbane (Solanaceae, Hyoscyamus niger) (12), and NADPH-dependent 3-quinuclidinone reductase (QNR), from the yeast Rhodotorula rubra (mucilaginosa) (13). Yamamoto et al. reported the production of (R)-(−)-3-quinuclidinol using recombinant formate dehydrogenase (FDH) and tropinone reductase (14). Uzura et al. reported a reaction system using recombinant glucose dehydrogenase and 3-quinuclidinone reductase from Rhodotorula rubra (13). The optical resolution of (±)-3-quinuclidinol esters by the hydrolysis reaction of Aspergillus melleus protease was reported by Nomoto et al. (15).

In this study, we report two novel 3-quinuclidinone reductase genes, qnr and bacC, from Microbacterium luteolum JCM 9174 and their cloning and heterologous expression in Escherichia coli. The recombinant enzymes were characterized and evaluated as biocatalysts for producing (R)-(−)-3-quinuclidinol. This is the first report of the enzymatic function of bacC in the bacilysin synthetic gene cluster.

MATERIALS AND METHODS

Chemicals.

3-Quinuclidinone hydrochloride, 4-acetylpyridine, tetrahydrothiopyran-4-one, and 7-oxabicyclo[4.1.0]heptan-2-one were purchased from Sigma-Aldrich, Missouri. 3-Quinuclidinol, 3-methylene-2-norbornanone, verbenone, 2-acetylpyridine, 4-hydroxy-1-cyclohexanecarboxylic acid δ-lactone, and 2-azabicyclo[2.2.1]hept-5-en-3-one were purchased from Tokyo Chemical Industry, Tokyo, Japan. (R)-(−)-3-Quinuclidinol was purchased from Kanto Chemical, Tokyo, Japan, and 3-acetylpyridine, 1-methyl-4-piperidone, and tropinone were purchased from Wako Pure Chemical Industries, Osaka, Japan. All other chemicals used in this study were of analytical grade and are commercially available.

Cultivation of various microorganisms.

The following culture media were used for screening: 1.0% (wt/vol) peptone, 0.5% yeast extract, 0.5% NaCl, and 0.3% glycerol (pH 7.0) for bacteria; 2.0% glucose, 0.2% yeast extract, 0.1% l-glutamate, and 0.5% malt extract (pH 5.6) for yeasts and fungi; and 1.0% glucose, 0.2% yeast extract, 0.1% beef extract, 0.2% N-Z-amine, and 0.2% malt extract (pH 7.2) for actinomyces. Each microorganism was aerobically cultured in a large test tube containing the appropriate medium (20 ml), with shaking (121 rpm), at 30°C for 2 to 7 days. After cultivation, the cells were collected by centrifugation (27,000 × g, 4°C, 20 min), washed once with 50 mM potassium phosphate buffer (KPB) (pH 7.0), and used for the resting-cell reaction.

Resting-cell reaction to convert 3-quinuclidinone to chiral 3-quinuclidinol.

The reaction mixture, with a total volume of 15 ml, consisted of microbial cells from 20 ml culture broth, 50 mM KPB (pH 7.0), 30 mM 3-quinuclidinone hydrochloride, 0.5 mM NADP⁺, 0.5 mM NAD⁺, and 1% (wt/vol) glucose in a 50-ml polypropylene tube. The reaction was shaken (170 rpm) for 48 h at 25°C in a water bath and was stopped by raising the pH to 12.0 with 6 N NaOH. The product was extracted with 1-butanol (15 ml), containing 5 mM 1-octanol as an internal standard, and was then dried with anhydrous Na₂SO₄. After centrifugation, the solvent was recovered and used for gas chromatography (GC).

Product analysis by GC.

The enantiomer of 3-quinuclidinol was analyzed by using a GC system (HP 6890; Hewlett Packard, California) equipped with a chiral capillary column (CP-cyclodextrin-β-236-N19, 0.25 mm by 25 m; Varian, California) with a flame ionization detector. The GC conditions were as follows: the column temperature program ramped from 70°C to 180°C at 10°C min⁻¹, the injection and detection temperatures were 250°C, and the He flow rate was 3.3 ml min⁻¹ with a linear velocity of 50 cm s⁻¹ and a split ratio of 50. The retention times were as follows: 8.17 min for the 1-octanol internal standard, 10.94 min for 3-quinuclidinone, 12.69 min for (S)-(+)-3-quinuclidinol, and 12.77 min for (R)-(−)-3-quinuclidinol.

Large-scale cultivation of M. luteolum.

M. luteolum JCM 9174 cells were grown aerobically in medium consisting of 1.5% (wt/vol) peptone, 0.5% yeast extract, 0.5% NaCl, 0.3% sodium glutamate, and 1% sucrose (pH 7.0). Precultivation was carried out in the medium (each 20 ml) in two large test tubes for 24 h at 30°C, with shaking (300 rpm). A portion of the culture medium (30 ml) was added to fresh medium (3 liters), which contained antifoam PE-H (final concentration of 0.1%) in a jar fermentor, and was cultured at 30°C for 17 h at 500 rpm, with an aeration rate of 0.75 liters min⁻¹.

Enzyme assay.

3-Quinuclidinone reductase activity was assayed spectrophotometrically by measuring the decrease in the absorbance of NADH at 340 nm (ε = 6.22 mM⁻¹ cm⁻¹). The assay was performed in a reaction mixture, with a total volume of 1.0 ml, which consisted of the substrate (3 μmol), NADH (0.3 μmol), KPB (50 μmol, pH 7.0), and enzyme solution (10 μl). One unit of enzyme was defined as the amount of enzyme that converted 1 μmol of NADH per min at 25°C.

Purification of 3-quinuclidinone reductase.

All purification procedures were performed at 0 to 4°C in 20 mM KPB (pH 7.0) containing 10% (vol/vol) glycerol, 1 mM MgCl₂, and 1 mM 2-mercaptoethanol, unless otherwise specified. M. luteolum was cultured as described above. The culture (3 liters) was centrifuged (10,000 × g, 10 min), and the precipitate (16.9 g [wet weight]) was washed once with the buffer. The cells were resuspended in the buffer (25.5 ml), which also contained 1 tablet of Complete Mini EDTA-free protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) per 10 ml of buffer. The cells were disrupted by sonication (201 M ultrasonic oscillator; Kubota, Osaka, Japan) for 8 min at 180 W and then centrifuged (9,000 × g, 20 min). The supernatant was applied to a column (20 mm by 210 mm, DEAE-Toyoperal 650M; Tosoh, Tokyo, Japan) equilibrated with the buffer. The enzyme was eluted with a linear 0 to 0.8 M NaCl gradient in the same buffer (flow rate, 1.0 ml min⁻¹, 4 ml/fraction). The fractions with high enzyme activity were collected. The solution was dialyzed in buffer containing 1.2 M ammonium sulfate. After centrifugation (20,000 × g, 30 min), the supernatant was applied to an high-performance liquid chromatography (HPLC) column (7.5 mm by 75 mm, Ether-5PW; Tosoh) equilibrated with the buffer containing 1.2 M ammonium sulfate. The enzyme was eluted with a linear 1.2 to 0 M ammonium sulfate gradient in the buffer (flow rate, 0.5 ml min⁻¹, 1.5 ml/fraction). The fractions with high enzyme activity were collected and dialyzed against the buffer (pH 7.0). After centrifugation (20,000 × g, 30 min), the supernatant was applied to an HPLC column (4.6 mm by 50 mm) (Bioassist Q; Tosoh) equilibrated with the buffer. The enzyme was eluted with a linear 0 to 0.8 M NaCl gradient in the buffer (flow rate, 0.35 ml min⁻¹, 1.0 ml/fraction). After the fractions were collected, the enzyme was dialyzed in pH 6.5 buffer and repurified using the same HPLC column. The enzyme was eluted with a linear 0 to 0.8 M NaCl gradient in pH 6.5 buffer (flow rate, 0.35 ml min⁻¹, 0.5 ml/fraction). The active fractions were collected and used as the enzyme solution.

Amino acid sequence analysis.

The enzyme was electrophoresed on a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel, transferred to a polyvinylidene difluoride membrane (Bio-Rad Laboratories, California) using a semidry electroblotting apparatus (NA-1512; Nippon Eido, Tokyo, Japan), and then stained with Coomassie brilliant blue G-250. The amino acid sequences of the N-terminal end and internal part of the enzyme on the polyvinylidene difluoride membrane were determined by APRO Science, Tokushima, Japan.

Cloning of the qnr reductase gene from M. luteolum.

Genomic DNA was extracted from M. luteolum grown in Luria-Bertani (LB) medium (1.0% peptone, 0.5% yeast extract, 1.0% NaCl [pH 7.0]) for 24 h at 30°C, with shaking. After centrifugation, the cells were resuspended in Tris-EDTA (TE) buffer and disrupted with an equal volume of glass beads by using a cell disruptor (Multi-Beads Shocker; Yasui Kikai, Osaka, Japan). Genomic DNA was obtained by phenol-chloroform-isoamyl alcohol (PCI) extraction and ethanol precipitation from the lysate. The target gene was amplified by PCR using genomic DNA as a template in combination with the degenerate primers (forward, 5′-ATGMGNYTNGARAAYAA-3′, and reverse, 5′-AANGCRTTNGTRTCYTG-3′) (Nippon EGT, Toyama, Japan), which were designed on the basis of the N-terminal amino acid sequence (MRLENKK) and two internal amino acid sequences (ALAIDHGPAGIR and QLAQDTNAFLAE; the underlined sequence was used for the design of the reverse primer) (see Fig. S1 in the supplemental material). The following conditions were used: 94°C, 2 min, followed by 94°C, 20 s; 55°C, 30 s; and 60°C, 1 min, for a total of 30 cycles, and then 72°C, 10 min, in accordance with the manufacturer's protocol for Ex Taq DNA polymerase (TaKaRa Bio, Otsu, Japan). The amplified DNA fragment (∼500 bp) was excised from agarose gel and purified. After TA cloning into pGEM-T (Promega, Wisconsin), the nucleotide sequence was determined using a genetic analyzer (ABI PRISM 310; Life Technologies, California). The entire qnr gene was cloned by cloning partially digested genomic DNA with Sau3AI into the BamHI site of pUC19. E. coli DH5α cells transformed with these plasmids were cultured on an LB agar plate with ampicillin (0.1 mg ml⁻¹). The gene library was constructed by collecting the E. coli cells by scraping the plate with TE buffer (1 ml) and then extracting the plasmids. Inverse PCR was conducted using this gene library with the following primers: 5′-CACCGGGATGCTGC-3′ and 5′-GGTCATGTTCACCAC-3′. The nucleotide sequence in the C-terminal part of the qnr gene was determined (see Fig. S1). To obtain the complete qnr gene, PCR was performed with the following primers: 5′-TTTCATATGCGGCTGGAGAATAAGAAGGC-3′ (the NdeI site is underlined) and 5′-TTTAAGCTTGACACCGGCGGACGCGCGAC-3′ (the HindIII site is underlined). KOD FX DNA polymerase (Toyobo, Osaka, Japan) was used with genomic DNA as the template. After the addition of A to the 3′ end of the PCR product by Taq DNA polymerase, the DNA fragment was cloned into the pGEM-T vector. The amplicon was digested with NdeI and HindIII and cloned into the same sites of the pET28-a(+) vector to generate pET28-QNR. E. coli BL21(DE3) (Merck KGaA, Darmstadt, Germany) was used for the expression of pET28-QNR, and E. coli BL21(pET28-QNR) was cultivated at 37°C for 24 h in LB medium containing 0.05 mg ml⁻¹ kanamycin sulfate and 0.4 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG).

Cloning of the bacC reductase gene from M. luteolum.

The oligonucleotide primers of forward primer 1 (5′-ACNGCNYTNGTBACNGGYGGY-3′) and reverse primer 2 (5′-NCCNGTNAYRTARCTNGCNGCNGG-3′) were synthesized based on the conserved amino acid sequences of tropinone reductase I and II of Hyoscyamus niger (see Fig. 3) (12), which are known to reduce 3-quinuclidinone. PCR was carried out using the M. luteolum genomic DNA as the template with the primers under the following conditions: 94°C, 2 min, and then 98°C, 20 s; 55°C, 30 s; 68°C, 1 min, for a total of 30 cycles, in accordance with the manufacturer's protocol for KOD FX DNA polymerase. The gene fragment was then cloned into pCR2.1-TOPO (Life Technologies) using the TOPO TA cloning kit and E. coli TOPO 10 (Life Technologies) and was sequenced by a genetic analyzer. E. coli was cultured at 37°C in LB medium containing 0.1 mg ml⁻¹ ampicillin for gene cloning. Primers were synthesized based on the partial internal sequence, and the 5′- and 3′-terminal regions of the gene were amplified by the DNA walking SpeedUp kit (Seegene, Seoul, South Korea). The amplified fragments were analyzed to determine the DNA sequence of the gene. The entire gene was amplified using PCR with the M. luteolum genomic DNA as the template and the following primers: BACforEco (5′-GAATTCATGATCATGAACCTCACCGATAAAACCGT-3′) and BACrevHind (5′-AAGCTTCTATTGTGCGGTGTATCCTCCGTCTGCGG-3′). The gene, to which the restriction enzyme sites (EcoRI and HindIII, underlined) were added, was cloned into the pCR2.1-TOPO plasmid. The plasmid containing the gene was digested with the EcoRI and HindIII restriction enzymes and cloned into the same pET28-a (+) sites to generate the pET28-BacC expression vector. Gene induction was achieved by cultivating E. coli BL21(pET28-BacC) at 37°C for 24 h in LB medium containing 0.05 mg ml⁻¹ kanamycin sulfate and 0.4 mM IPTG.

Fig 3.

The amino acid sequence alignment of QNR and BacC from M. luteolum with SDR superfamily enzymes. 3-Quinuclidinone reductases from M. luteolum JCM 9174 (QNR, accession number AB733448; BacC, accession number NP_391652), Hyoscyamus niger (tropinone reductase I and II, D88156 and L20485), Streptosporangium roseum (YP_003343110), Stackebrandtia nassauensis (YP_003511498), Verminephrobacter eiseniae (YP_997220), and Rhodotorula rubra (mucilaginosa) (NADPH-dependent 3-quinuclidinone reductase, AB469142) are aligned. The gray box denotes the Rossmann fold for the NAD/P(H)-binding region. Alignment gaps are indicated by dashes. Identical amino acid residues are enclosed in black boxes. The amino acid sequences boxed with a dotted line are identical to those determined by protein sequencing. Asterisks denote the position of conserved amino acid sequences of tropinone reductase I and II of Hyoscyamus niger used for the design of degenerate primers.

Purification of the recombinant proteins.

E. coli BL21(DE3) cells harboring pET28-QNR or pET28-BacC grown in LB medium with kanamycin sulfate and IPTG were harvested by centrifugation from the culture broth (20 ml), suspended in 20 mM KPB (pH 7.0), and disrupted with an ultrasonic oscillator. After centrifugation (13,000 × g, 10 min), the resulting supernatant was filtered with a membrane filter (0.45 μm). The recombinant enzyme with the His₆ tag and thrombin recognition sequence at the N terminus was purified from the crude extract by nickel-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen, Venlo, The Netherlands) column chromatography (8 mm by 40 mm) according to the supplier's protocol. The fractions eluted with 300 mM imidazole solution which displayed high activity were collected and desalted by using a filter (Centriprep YM-30; Merck Millipore, Darmstadt, Germany) with a molecular weight cutoff of 30,000. The enzyme solution thus obtained was used as the purified enzyme for further characterization.

Protein assay.

The protein concentration was estimated by using the Bradford method (16) with a Bio-Rad protein assay kit (Bio-Rad Laboratories), calibrated with bovine serum albumin as the standard protein.

Molecular weight.

The molecular weight of the enzyme was determined by analytical HPLC (column, 7.5 mm by 30 cm) (TSK-Gel G3000sw; Tosoh) at a flow rate of 0.5 ml min⁻¹ with 50 mM Tris-HCl (pH 7.0) containing 0.1 M NaCl. The molecular mass of the native enzyme was determined by comparing the retention times with standard proteins.

Measurement of the enantiomeric excess of (R)-(−)-quinuclidinol in the enzymatic reaction.

The 3-quinuclidinone conversion reaction with the coenzyme regenerating system was performed in a reaction mixture with a total volume of 1 ml consisting of 200 mM KPB (pH 7.0), 5% (wt/vol) of 3-quinuclidinone (313 mM), 1 mM NAD⁺, 1 M sodium formate, 0.1 units of the purified 3-quinuclidinone-reducing enzymes (QNR or BacC; see enzyme assay), and 0.2 units of formate dehydrogenase from Candida boidinii (Roche Diagnostics, Mannheim, Germany) in a 2-ml polypropylene tube. The reaction proceeded for 12 h at 25°C (Bioshaker MBR-022UP; Taitec, Saitama, Japan). A concentration of 6 N NaOH (0.2 ml) was added to the reaction mixture (1.0 ml), and the product was extracted twice with 1-butanol (0.5 ml) which contained 5 mM 1-octanol, and the 1-butanol layer was thoroughly dried with anhydrous Na₂SO₄. After centrifugation, the solution was analyzed by GC as described above.

Nucleotide sequence accession number.

The nucleotide sequence of qnr is shown in Fig. S1 in the supplemental material and has been deposited in the DDBJ, EMBL, and GenBank databases under the accession number AB733448.

RESULTS

Identification of 3-quinuclidinone-reducing microorganisms.

Stock cultures in our laboratory, including 68 strains of fungi, 28 strains of actinomycetes, 24 strains of yeasts, and 91 strains of bacteria, were screened for the conversion of 3-quinuclidinone. Twelve strains reduced 3-quinuclidinone in the resting-cell reaction (Table 1). The majority of the strains converted 3-quinuclidinone to (R)-(−)-3-quinuclidinol; only Micrococcus lutea produced (S)-(+)-3-quinuclidinol. The data suggested that the genera Microbacterium and Kluyveromyces may be suitable producers of (R)-(−)-3-quinuclidinol. Microbacterium luteolum JCM 9174 showed the highest production of (R)-(−)-3-quinuclidinol and an enantiomeric excess of >99%. Therefore, this strain was selected for further experiments.

Table 1.

Microorganisms producing chiral 3-quinuclidinols

Microorganism	3-Quinuclidinol concn (mM)	Absolute configuration, enantiomeric excess (%)
Tsukamurella paurometabola IFO12160	0.43	(R), 90
Micrococcus lutea IFO 12708	0.25	(S), >99
Alcaligenes sp. strain IFO 14130	0.5	(R), >99
Kurthia zopfii IFO 12083	1.2	(R), >99
Microbacterium luteolum JCM 9174	21.9	(R), >99
Microbacterium esteraromaticum IFO 3751	1.3	(R), 95
Microbacterium arabinogalactanolyticum JCM 9171	1.8	(R), 30
Geotrichum candidum IFO 4599	0.27	(R), 22
Kluyveromyces marxianus IAM12237	5	(R), 86
Kluyveromyces polysporus JCM 3705	4.1	(R), >99
Acremonium sp.	0.89	(R), 95
Mucor sp.	4.2	(R), 82

Purification of 3-quinuclidinone-reducing enzymes.

The purification of 3-quinuclidinone-reducing enzymes from M. luteolum JCM 9174 was carried out by anion-exchange chromatography. Four peaks showing enzymatic activity were observed (Fig. 1), and the main enzyme peak was purified (Fig. 1, indicated by an arrow).

Fig 1.

Anion-exchange chromatography of 3-quinuclidinone-reducing enzymes from Microbacterium luteolum JCM 9174. The absorbance at 280 nm (○) and the quinuclidinone-reducing activity (▲) of the fractions are shown. The gradient concentration of NaCl (0 to 0.8 M) is indicated by closed squares.

The QNR enzyme was purified by a combination of different types of anion-exchange chromatography using DEAE-Toyopearl, Bioassist Q, and Ether-5PW hydrophobic resin. The enzyme showed no affinity for dye-ligand matrices, such as Blue-Sepharose and Red-Sepharose. The enzyme could not be purified to homogeneity using SDS-PAGE with a sufficient yield; Fig. 2 shows that at least five protein bands were observed in the final preparation. The amino acid sequence of major C and D bands on the gel were analyzed. The N-terminal peptide sequence of band C was HDPVVQPVLGSG, indicating that it was similar to zinc metal protease or aminopeptidase I; therefore, it was disregarded. The N-terminal peptide sequence of band D was MRLENKK, and those of the internal amino acid sequence were ALAIDHGPAGIR and QLAQDTNAFLAE. These sequences were similar to previously reported putative short-chain oxidoreductases (SDR) (Fig. 3). Therefore, band D was identified as the target enzyme and the amino acid sequences were used for cloning the qnr gene from M. luteolum.

Fig 2.

SDS-PAGE analysis of 3-quinuclidinone-reducing enzymes from M. luteolum. (I) The five bands of the purified enzyme preparation are labeled A to E. The enzyme was subjected to a 12.5% SDS-PAGE and stained with Coomassie blue after electrophoresis. Lane a contains molecular mass standards (XL ladder; APRO), and lane b contains the purified enzyme preparation. (II) Purified recombinant 3-quinuclidinone reductase (QNR). (III) Purified recombinant BacC.

Sequence analysis of 3-quinuclidinone-reducing enzymes.

The qnr gene was successfully isolated from the M. luteolum genome according to the procedures described in Materials and Methods. The gene contained an open reading frame (ORF), which consisted of 756 nucleotides corresponding to 252 amino acid residues with a calculated molecular mass of 25,739.1 Da. The qnr gene belongs to the SDR family and displayed moderate identities with the putative SDR of Streptosporangium roseum DSM43021 (68.4% amino acid identity; accession number YP_003343110), the putative SDR of Stackebrandtia nassauensis DSM44728 (59.6%; YP_003511498), and the putative SDR of Verminephrobacter eiseniae EF01-2 (54.7%; YP_997220). It showed a low identity with NADPH-dependent 3-quinuclidinone reductase from Rhodotorula rubra (mucilaginosa) (24.2%; AB469142), tropinone reductase I (27.4%; D88156) and II (29.8%; L20485) derived from Hyoscyamus niger, and BacC (35.7%; NP_391652) (Fig. 3).

A gene encoding a different 3-quinuclidinone-reducing enzyme was also isolated based on the assumption that such an enzyme may possess sequence similarity with known 3-quinuclidinone-reducing enzymes, such as tropinone reductases (12). This approach resulted in the cloning of a gene different from the qnr gene from M. luteolum. Sequence analysis of this gene indicated that it was identical to the putative oxidoreductase gene (bacC) in a bacilysin biosynthetic gene cluster derived from Bacillus subtilis (17). Bacilysin is a dipeptide antibiotic that is active against a wide range of bacteria and Candida albicans (18). bacC contained an ORF which consisted of 765 nucleotides corresponding to 255 amino acid residues with a calculated molecular mass of 27,324.3 Da. In order to verify the purity of the genomic DNA, it was confirmed that bacC could be amplified from 10 freshly prepared genomic DNA samples derived from 10 independently isolated single colonies (see Fig. S2 in the supplemental material). Moreover, we checked the 16S rRNA gene sequence of this strain and confirmed that it is specific for M. luteolum (see Fig. S3 and S4 in the supplemental material) (19).

Expression of qnr and bacC in E. coli and their purification.

Two expression vectors, pET28a-QNR and pET28a-bacC, were constructed, which encoded qnr and bacC, respectively. The ORFs for these genes were fused with the His₆ tag and thrombin recognition sequence at the N terminus under the control of the T7 promoter.

Each cell extract of recombinant E. coli cells was purified using Ni-NTA agarose column chromatography (Table 2). Homogeneous purification was confirmed by SDS-PAGE (Fig. 2), and recombinant 3-quinuclidinone reductase (QNR) and BacC in E. coli were estimated to account for 42 and 59% of the total amount of soluble protein in the cell extract, respectively.

Table 2.

Purification of recombinant enzymes from E. coli transformant cells

Enzyme and sample source	Total activity (U)	Total protein (mg)	Specific activity (U/mg)	Yield (%)	Fold
QNR
Crude extract	65	18.5	3.5	100	1
Ni-NTA agarose	38	4.6	8.4	59.2	2.4
BacC
Crude extract	4.8	16.0	0.3	100	1
Ni-NTA agarose	0.5	1.1	0.5	11.4	1.7

Molecular mass and subunit structure.

The molecular mass of recombinant QNR was estimated to be 94,000 Da with analytical HPLC on TSK-Gel G3000sw. SDS-PAGE revealed a single band, and the molecular mass was 30,600 Da (theoretical value, 27,902 Da). This suggests that QNR is probably composed of three identical subunits.

The molecular mass of BacC was estimated to be 72,000 Da. The subunit molecular mass was estimated to be 32,600 Da (theoretical value, 31,144 Da) with SDS-PAGE, showing that BacC is a dimeric protein composed of identical subunits. However, the mobilities of the purified recombinant enzymes on SDS-PAGE did not coincide with the theoretical values, probably because of the His tag in the recombinant proteins.

Effect of pH on enzyme activity.

The effect of pH on the enzyme activity at a final buffer concentration of 0.2 M was measured in the following buffers: citric acid-phosphate buffer (pH 4.5 to 6.0), KPB (pH 5.0 to 8.5), and Tris-HCl (pH 7.5 to 9.0). The maximum activity of QNR was at pH 7.0 to 8.0, whereas it was at pH 7.0 for BacC (Fig. 4).

Fig 4.

Activity of QNR and BacC as a function of pH during the reduction of 3-quinuclidinone. The activity was measured in the following 0.2 M buffers: citric acid-phosphate buffer (pH 4.5 to 6.0) (circles), KPB (pH 5.0 to 8.5) (squares), and Tris-HCl (pH 7.5 to 9.0) (triangles). The white symbols represent QNR, and the black symbols represent BacC.

Kinetic properties of QNR and BacC.

The K_m values of QNR for 3-quinuclidinone and NADH in the reductive reaction were calculated as 6.5 and 0.02 mM from the Hanes-Woolf plot, respectively. The K_m values of BacC for 3-quinuclidinone and NADH in the reductive reaction were calculated at 13.8 and 0.03 mM, respectively. The V_max values of QNR and BacC for 3-quinuclidinone were calculated from the same plots to be 71.9 μmol min⁻¹ mg⁻¹ and 2.1 μmol min⁻¹ mg⁻¹, respectively. Neither enzyme showed activity with NADPH.

Substrate specificity.

The activity of the purified enzymes was measured for various substrates. The activities relative to the activity against 3-quinuclidinone are displayed in Table 3. Interestingly, QNR was strictly specific for 3-quinuclidinone and showed no activity toward several ketones, including tropinone. In contrast, BacC catalyzed the reduction of 3-quinuclidinone, 7-oxabicyclo[4.1.0] heptan-2-one (27.8%), which is an analogue of the intermediate in bacilysin biosynthesis (20), and 2-acetylpyridine (16.8%). However, the substrate specificity of BacC was also narrow. Neither enzyme catalyzed the oxidative reaction of (R)-(−)-3-quinuclidinol. This is the first time the function of the BacC enzyme has been reported.

Table 3.

Substrate specificity of purified recombinant QNR and BacC enzymes for the reductive reaction of ketones^a

Substrate	Relative activity (%)
	QNR	BacC
3-Quinuclidinone	100	100
(R)-(−)-3-Quinuclidinol*	0	0
2-Acetylpyridine	0	16.8
7-Oxabicyclo[4.1.0]heptan-2-one	0	27.8

^aBoth enzymes were inert toward the following compounds: 2-propanol*, acetone, acetophenone, 2,2,2-trifluoroacetophenone, 3-methylene-2-norbornanone, ethyl pyruvate, ethyl acetoacetate, 4-hydroxy-1-cyclohexanecarboxylic acid delta-lactone, verbenone, tropinone, 6-hydroxytropinone, 2-azabicyclo[2,2,1]hept-5-en-3-one, 1-methyl-4-piperidone, 3-acetylpyridine, 4-acetylpyridine, cyclohexanone, 2-methylcyclohexanone, 3-methylcyclohexanone, and tetrahydrothiopyran-4-one. An oxidative reaction is indicated by an asterisk.

Stereoselectivity of QNR and BacC for 3-quinuclidinone.

The production of (R)-(−)-3-quinuclidinol was tested using an NADH regeneration system with formate dehydrogenase. In this system, 3-quinuclidinone (5%, wt/vol; 313 mM) was reduced to (R)-(−)-3-quinuclidinol with a molar conversion yield of 100% for QNR and 94% for BacC (see Fig. S5 in the supplemental material). The (S)-isomer of 3-quinuclidinol was not detected in the reaction products of QNR or BacC. Thus, the optical purity of the (R)-(−)-3-quinuclidinol produced by the QNR and BacC reactions was >99.9%.

DISCUSSION

Various stock microorganism cultures were screened for reduction of 3-quinuclidinone. Activity was observed in at least 12 of the 211 strains, suggesting that the resting-cell reaction with 1% glucose was a suitable screening method. The majority of the organisms stereoselectively produced the (R)-isomer from 3-quinuclidinone. M. luteolum JCM 9174 was selected and showed multiple 3-quinuclidinone-reducing activity in the cell extract (Fig. 1). Two genes (qnr and bacC) for enzymes capable of reducing the 3-quinuclidinone to (R)-(−)-3-quinuclidinol were obtained from M. luteolum.

A protein blast search with the QNR enzyme showed the highest identity was 68.4%, with a putative SDR in Streptosporangium roseum. The identities with known 3-quinuclidinone-reducing enzymes were very low: 24.4% with NADPH-dependent 3-quinuclidinone reductase of R. rubra (mucilaginosa) (13), 27.4% with tropinone reductase I, and 29.8% with tropinone reductase II (12). Therefore, QNR is a novel enzyme belonging to the SDR family. bacC isolated from M. luteolum showed 99.9% identity (100% amino acid identity) with the putative oxidoreductase gene bacC of the bacilysin biosynthetic gene cluster derived from Bacillus subtilis (17, 21), suggesting that bacC was horizontally transferred from Bacillus species. We have not yet determined whether or not M. luteolum produces bacilysin.

Each recombinant enzyme was overproduced under the control of the T7 promoter in E. coli BL21(DE3) cells and accounted for 42% of the soluble protein for QNR and 59% for BacC in E. coli. Recombinant QNR was highly specific for 3-quinuclidinone and showed no activity for cyclohexanone derivatives, such as 1-methyl-4-piperidone and tetrahydrothiopyran-4-one, and the bicyclic ketones, such as tropinone, 3-methylene-2-norbornanone, and verbenone (Table 3). The physiological function of this enzyme in M. luteolum has not been explored. We speculate that it participates in the synthesis of a nitrogen-containing secondary metabolite, similar to plant alkaloids. Although the substrate specificity of BacC was also narrow, it showed activity for 7-oxabicyclo[4.1.0]heptan-2-one and 2-acetylpyridine, with relative activities of 27.8% and 16.8%, respectively, in addition to 3-quinuclidinone. 7-Oxabicyclo[4.1.0]heptan-2-one is an analogue of anticapsin, a precursor of the bacilysin biosynthesis route (Fig. 5) (20, 21). Rajavel et al. have reported the role of the bacA and bacB genes in bacilysin biosynthesis (20). Thus, BacC may also play a role in bacilysin biosynthesis, especially in the oxidoreduction of the bacilysin precursor, with reduction of the carbonyl or double bond of the compound, although the accurate reaction has not yet been clarified.

Fig 5.

Structure of 7-oxabicyclo[4.1.0]heptan-2-one and the precursor of the bacilysin biosynthesis route (analogue of anticapsin).

Table 4 shows the enzymatic properties of QNR and BacC compared with a 3-quinuclidinone reductase from the yeast R. rubra (13). The K_m values of QNR (6.5 mM) and BacC (13.8 mM) for 3-quinuclidinone in the reductive reaction were much lower than that of NADPH-dependent 3-quinuclidinone reductase from R. rubra (145 mM). The coenzyme specificity and high specific enzyme activity of QNR mean that it is a more suitable biocatalyst for producing (R)-(−)-3-quinuclidinol than other enzymes in the bioreduction process.

Table 4.

Comparison of the properties of recombinant QNR and BacC with R. rubra 3-quinuclidinone reductase^a

Property	QNR	BacC	R. rubra reductaseb
Molecular mass (kDa)	94	72	93
Subunit structure (kDa)	27.9 (trimer)	31.1 (dimer)	30 (trimer/tetramer)
Coenzyme	NADH	NADH	NADPH
Km (mM)
3-Quinuclidinone	6.5	13.8	145
NADH/NADPH	0.02	0.03	0.19
Vmax (μmol min−1 mg−1)
3-Quinuclidinone	71.9	2.1	ND
Stereoselectivity (% enantiomeric excess)
(R)-3-Quinuclidinol	>99.9	>99.9	>99.9
Optimum pH	7.0–8.0	7.0	6.5–7.0

^aBoth enzymes contain the His₆ tag and thrombin recognition sequence at the N terminus.

^bData from reference 13. ND, not determined.

The production of (R)-(−)-3-quinuclidinol was confirmed using a reaction system coupled with an NADH regeneration system by FDH and formate (Fig. 6; see also Fig. S5 in the supplemental material). The conversion yield of QNR was 100% for 313 mM 3-quinuclidinone in 12 h. The enzyme showed high stereoselectivity; the enantiomeric excess of the (R)-(−)-3-quinuclidinol produced was >99.9%. The conversion yield for BacC was 94%, and the enantiomeric excess of (R)-(−)-3-quinuclidinol was >99.9%. These results demonstrate that our production system with QNR or BacC could be used for the practical production of (R)-(−)-3-quinuclidinol.

Fig 6.

Production of (R)-(−)-3-quinuclidinol coupled by FDH with an NADH regeneration system.