Bioreductive production of enantiopure (S)-duloxetine intermediates catalyzed with ketoreductase ChKRED15

Article abstract of DOI:10.1016/j.molcatb.2015.01.008

The blockbuster antidepressant drug duloxetine contains one stereo-center derived from chiral alcohol intermediates. The stereoselective bioreductive production of five of such intermediates could be achieved using the recombinant ketoreductase ChKRED15, yielding the enantiopure (S)-alcohols with >99% ee. Sequence alignment indicated that ChKRED15 lacks the conserved G-rich motif, which was then amended by a single mutation of S12G. The resulting S12G mutant displayed significantly improved catalytic activity and protein stability. When coupled with a cofactor recycling system, the S12G mutant was able to catalyze the complete conversion of ethyl 3-oxo-3-(thiophen-2-yl)propanoate within 6 h and N-methyl-3-oxo-3-(thiophen-2-yl)propanamide within 24 h at substrate concentrations of 10 and 50 g/l, respectively, without the compromise of enantioselectivity.

Full text of DOI:10.1016/j.molcatb.2015.01.008

Journal of Molecular Catalysis B: Enzymatic 113 (2015) 76–81
Contents lists available at ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
Bioreductive production of enantiopure (S)-duloxetine intermediates
catalyzed with ketoreductase ChKRED15
Zhi-Qiang Rena,b,c, Yan Liua,b,c, Xiao-Qiong Peia,b,c, Hai-Bo Wanga,b, Zhong-Liu Wua,b,∗
a Key Laboratory of Environmental and Applied Microbiology, Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu 610041, China
b Environmental Microbiology Key Laboratory of Sichuan Province, Chengdu 610041, China
c University of the Chinese Academy of Sciences, Beijing 100049, China
a r t i c l e i n f o
a b s t r a c t
Article history:
The blockbuster antidepressant drug duloxetine contains one stereo-center derived from chiral alco-
hol intermediates. The stereoselective bioreductive production of five of such intermediates could be
achieved using the recombinant ketoreductase ChKRED15, yielding the enantiopure (S)-alcohols with
>99% ee. Sequence alignment indicated that ChKRED15 lacks the conserved G-rich motif, which was then
amended by a single mutation of S12G. The resulting S12G mutant displayed significantly improved
catalytic activity and protein stability. When coupled with a cofactor recycling system, the S12G mutant
was able to catalyze the complete conversion of ethyl 3-oxo-3-(thiophen-2-yl)propanoate within 6 h and
N-methyl-3-oxo-3-(thiophen-2-yl)propanamide within 24 h at substrate concentrations of 10 and 50 g/l,
respectively, without the compromise of enantioselectivity.
Received 4 December 2014
Received in revised form 9 January 2015
Accepted 16 January 2015
Available online 24 January 2015
Keywords:
Bioreduction
Chiral alcohol
Ketoreductase
© 2015 Elsevier B.V. All rights reserved.
Short-chain dehydrogenase/reductase
Duloxetine
1. Introduction
asymmetric reduction of the corresponding ketones (Fig. 1). How-
ever, catalytic systems developed by synthetic chemists only
Ketoreductases are an abundant group of oxidoreductases that
catalyzes the stereoselective reduction of ketones. The bioreduc-
tion serves as a powerful approach for the production of optically
active alcohols, which are widely used as fundamental interme-
diates for pharmaceuticals, agrochemicals and other products with
chiral centers [1–6]. With the increased availability of recombinant
ketoreductases, industrial-scale processes using cell-free enzymes
coupled with efficient recycling systems have been developed
rapidly in the past decade, which provides high volumetric pro-
ductivity, excellent stereoselectivity, absence of side reactions and
other economic advantages [7–12].
yield the product with moderate enantiomeric excesses (ee) of
80–88%, and additionally carry the risk of metal contamination
[13,14,17,18]. Therefore, the predominant approach to date is via
the resolution-racemization-recycle synthesis from racemic N,N-
dimethyl-3-hydroxy-3-(2-thienyl)-1-propanamine (DHTP, 2b) or
N-methyl-3-hydroxy-3-(2-thienyl)-1-propanamine (MHTP, 1b) to
form the corresponding (S)-enantiomer, which involves multiple
reaction steps and stoichiometric amount of (S)-mandelic acid as
the resolution reagent [19].
On the other hand, biocatalytic asymmetric reductions involv-
ing ketoreductases often lead to excellent stereoselectivity. It has
been reported that intermediates (S)-2b and (S)-6b can be achieved
with >99% ee using the whole cells of Candida tropicalis [20] and
Rhodotorula glutinis [16], respectively, and (S)-3b and (S)-5b with
98% ee using isolated ketoreductases from Exiguobacterium sp. [21]
and Thermoanaerobacter sp. [22], respectively. The use of isolated
ketoreductases avoids side reactions, which would greatly facili-
tate the scale-up with high volumetric productivity. However, the
isolated ketoreductases typically tolerate a low substrate concen-
tration of 5–10 mM [22]. For substrate 3a, a substrate concentration
of 50 mM (9.9 g/l) has been reported using the Exiguobacterium sp.
F42 ketoreductase expressed in Escherichia coli to achieve 80% prod-
uct yield (>98% ee) in 6 h [21].
Duloxetine is a blockbuster antidepressant drug that contains
one stereo-center derived from
a chiral alcohol intermedi-
ate. Only the (S)-enantiomer is pharmaceutically active for the
treatment of anxiety, diabetes-related pain, fibromyalgia and
chronic musculoskeletal pain, including lower back pain and
osteoarthritis [13–15]. According to the retrosynthetic strategy,
several chiral alcohol precursors act as the key intermedi-
ate for (S)-duloxetine [16], which could be achieved by direct
∗
Corresponding author at: Chengdu Institute of Biology, Chinese Academy of
Sciences, 9 South Renmin Road, 4th Section, Chengdu, Sichuan 610041, China.
Tel.: +86 28 82890434; fax: +86 28 82890434.
Here, we report a recombinant ketoreductase ChKRED15 iden-
tified from the genome of Chryseobacterium sp. CA49 [23,24] that
E-mail address: wuzhl@cib.ac.cn (Z.-L. Wu).
http://dx.doi.org/10.1016/j.molcatb.2015.01.008
1381-1177/© 2015 Elsevier B.V. All rights reserved.

Z.-Q. Ren et al. / Journal of Molecular Catalysis B: Enzymatic 113 (2015) 76–81
77
Fig. 1. Key intermediates of (S)-duloxetine derived from ketone reduction. MOTP, N-methyl-3-oxo-(2-thienyl)-1-propanamine (1a); MHTP, N-methyl-3-hydroxy-(2-thienyl)-
1-propanamine (1b); DOTP, N,N-dimethyl-3-oxo-(2-thienyl)-1-propanamine (2a); DHTP, N,N-dimethyl-3-hydroxy-(2-thienyl)-1-propanamine (2b); KEES, ethyl 3-oxo-3-
(thiophen-2-yl)propanoate (3a); HEES, (S)-ethyl 3-hydroxy-3-(thiophen-2-yl)propanoate (3b); KMES, methyl 3-oxo-3-(thiophen-2-yl)propanoate (4a); HMES, (S)-methyl
3-hydroxy-3-(thiophen-2-yl)propanoate (4b); CTP, 3-chloro-1-(thiophen-2-yl)propan-1-one (5a); CTPO, (S)-3-chloro-1-(thiophen-2-yl)propan-1-ol (5b); MOTPA, N-methyl-
3-oxo-3-(thiophen-2-yl)propanamide (6a); MHTPA, (S)-N-methyl-3-hydroxy-3-(2-thienyl)propionamide (6b); KTPN, 3-oxo-3-(thiophen-2-yl)propanenitrile (7a); HTPN,
(S)-3-hydroxy-3-(thiophen-2-yl)propanenitrile (7b).
can catalyze the bioreductive production of several chiral alcohol
precursors as the key intermediate for (S)-duloxetine with excel-
lent enantioselectivity (>99% ee). Up to 50 g/l substrate of 6a could
be reduced to completion within one day using the S12G mutant
of ChKRED15.
using the following primers: 5ꢀ-GAATTC ATG AAA ACA GTA TTA ATT
ACA GGC GCC-3ꢀ and 5ꢀ-AAGCTT CTA CCA CGG ACT GAT TCC GGT
TTC ATC-3ꢀ; and ligated into pMD19-Tvector with restriction sites
of EcoR I and Hind III. The resulting plasmids verified by sequenc-
ing were subcloned into pET28a (+) plasmid using the EcoR I and
Hind III restriction sites to achieve the expression plasmid pET28a-
chkred15.
2. Materials and methods
Site-directed mutagenesis was performed with the Quikchange
Mutagenesis Kit (Stratagene, La Jolla, CA, USA) following the
manufacturer’s protocol using the plasmid pET28a-chkred15 as
the template with the following primers: 5ꢀ-ACAGGCGCCAA-
CAGAGGTATTGGCCTTGAAACC-3ꢀ and 5ꢀ-GGTTTCAAGGCCAATAC-
CTCTGTTGGCGCCTGT-3ꢀ. The successful introduction of the desired
mutations was confirmed by DNA sequencing at Shanghai Invitro-
gen Life Technologies.
2.1. General
2-Acetylthiophene and substrate 7a was purchased from Alfa-
Aesar (Tianjin, China). Substrates 3a, 4a and 6a were synthesized
from 2-acetylthiophene following established methods [25,26].
Substrate 5a was synthesized from thiophene following estab-
lished methods [27]. All racemic alcohols were prepared following
literature methods [26,27]. NADH, NADPH, NADP+, isopropyl
␤-d-1-thiogalactopyranoside (IPTG), and glucose dehydrogenase
(GDH) were purchased from Sigma (St. Louis, USA). Restriction
enzymes and T4 DNA ligase were purchased from New England
Biolabs (Beverly, MA, USA). All other reagents were obtained
from general commercial suppliers and used without further
purification.
2.3. Expression and purification of the wild-type and mutant
ChKRED15
The recombinant plasmids were transformed into E. coli BL21
(DE3) cells for protein expression. Single colonies were grown
overnight at 37 ◦C in Luria-Bertani (LB) medium containing 50 ␮g
kanamycin/ml. Two milliliters of overnight culture was then inoc-
ulated into 200 ml of LB medium containing 50 ␮g kanamycin/ml.
IPTG was added at a final concentration of 1 mM when OD600
of the culture reached 0.6, and the cultivation was continued at
30 ◦C for 5 h. Cells were harvested by centrifugation, washed twice
using 20 mM sodium phosphate buffer (pH 8.0), and resuspended
in buffer A (20 mM sodium phosphate buffer, 500 mM NaCl and
10 mM imidazole, pH 8.0).
1H NMR spectra were recorded on a Brucker-600 (600/150 MHz)
spectrometer in CDCl3. All signals are expressed as ppm down field
from tetramethylsilane. Optical rotations were measured with a
Perkin Elmer 341 polarimeter at temperatures that allowed better
comparison of the measured values with those reported in the liter-
atures. HPLC was conducted with a Shimadzu Prominence LC-20AD
system connected to a PDA-detector. The conversion rates were
determined on an Inertsil SIL-100A HPLC column (4.6 nm × 250 nm,
GL Sciences Inc., Japan), and the optical purities were determined
on Daicel OJ-H (6b and 7b), Chiralcel OD-H (3b and 5b) and Chiral-
cel AD-H (4b) columns (4.6 mm × 250 mm, Daicel, Japan) at 35 ◦C
with n-hexane/2-propanol as the mobile phase (90/10 for 3b and
6b, 97/3 for 4b and 5b, 80/20 for 7b, v/v) at a flow rate of 1 ml/min
(3b and 7b), or 0.8 ml/min (4b and 5b) or 0.5 ml/min (6b).
After disruption with
a high-pressure homogenizer (ATS-
AH100B, ATS Engineering Inc., Canada), the cell debris was removed
by centrifugation at 2 × 104 g for 25 min at 4 ◦C. The resulting super-
natant can be directly used as crude enzyme extracts or lyophilized
to provide a dry powder of the crude enzyme. For purification,
it was loaded onto Ni2+-nitrilotriacetic acid columns (Bio-Rad)
equilibrated with buffer A. The enzyme was eluted with buffer A
containing a gradient of imidazole ranging from 10 to 500 mM at
a flow rate of 1 ml/min. The fractions containing the target protein
were collected and dialyzed against 20 mM potassium phosphate
buffer (pH 7.0). Purified enzymes were analyzed by SDS-PAGE and
used for enzymatic assays. Protein estimations were carried out
with a commercial BCA Protein Assay kit with bovine serum albu-
min as a standard (Tiangen, Beijing, China). The purified enzyme
solution and lyophilized powder were stored at −80 ◦C.
2.2. Construction of plasmids encoding ChKRED15 and S12G
mutant
The strain Chryseobacterium sp. CA49 was deposited at the China
Center for Type Culture Collection (Wuhan, China) under the acqui-
sition number CCTCC M 2012484. The DNA fragment encoding
ketoreductase ChKRED15 (GenBank accession no. KC342015) was
amplified from the genomic DNA of Chryseobacterium sp. CA49 [23]

78
Z.-Q. Ren et al. / Journal of Molecular Catalysis B: Enzymatic 113 (2015) 76–81
2.4. Measurement of enzyme activity
CH3), 5.36 (t, 1H, J = 6 Hz, CHOH), 6.95 (m, 2H, Ar-H), 7.23 (m, 1H, Ar-
H). [␣]D20 = −41.4 (c = 1.0, CHCl3) for >99% ee (lit. [29] [␣]D20 = −25.0
(c = 0.9, CHCl3) for 93% ee, (S)); Retention times: tR (S) 39.0 min, tR
(R) 40.5 min.
All reactions were performed in triplicate using purified
enzymes. The reaction mixture containing 10 mM substrate 6a,
10 mM NADPH and 0.1 mg enzyme in 1 ml potassium phosphate
buffer (100 mM, pH 8.0) was incubated at 30 ◦C for 30 min. The
reaction was terminated by extraction with ethyl acetate, and the
conversion rate and optical purity were determined by HPLC. One
unit of enzyme was defined as the quantity of enzyme required to
catalyze the conversion of 1 ␮mol substrate in 1 min at 30 ◦C.
To determine the pH optimum, standard assay method was
applied except that the reaction time was reduced to 10 min, and
different buffers were used for different pH ranges, which included
potassium phosphate buffer (pH 6.0–8.0), Tris–HCl buffer (pH
7.5–9.0) and sodium carbonate buffer (pH 9.0–10.0). The optimum
temperature for ChKRED15 was determined at various temper-
atures ranging from 20 to 60 ◦C for 10 min. To investigate the
thermo stability of ChKRED15, the enzyme (1 mg/ml) was incu-
bated at 35, 40, or 45 ◦C, withdrawn at intervals, cooled in ice,
and the residual activity was assayed following the standard assay
method.
2.6.3. (S)-3-Chloro-1-(thiophen-2-yl)propan-1-ol (5b)
94% yield; 1H NMR (600 MHz, CDCl3): ı 2.15 (dd, 2H, J = 6.6 Hz,
J = 6.0 Hz, CH2), 3.56 (t, 2H, J = 6.6 Hz, CH2), 5.22 (t, 1H, J = 6 Hz,
CHOH), 6.95 (m, 2H, Ar-H), 7.23 (m, 1H, Ar-H). [␣]D20 = −43.1
(c = 1.0, CHCl3) for >99% ee (lit. [30] [˛]D20 = −32.3 (c = 1.0, CHCl3)
for 96% ee, (S)); Retention times: tR (S) 21.6 min, tR (R) 23.7 min.
2.6.4. (S)-3-Hydroxy-N-methyl-3-(thiophen-2-yl)propanamide
(6b)
94% yield; 1H NMR (600 MHz, CDCl3): ı 2.67 (d, 2H, J = 6 Hz, CH2),
2.80 (s, 3H, NCH3), 5.36 (t, 1H, J = 6 Hz, CHOH), 5.75 (br, 1H, NH), 6.95
(m, 2H, Ar-H), 7.23 (m, 1H, Ar-H). [␣]D20 = −41.0 (c = 1.0, CHCl3) for
>99% ee (lit. [16] [␣]D25 = −31.0 (c = 1.0, CHCl3); for 99% ee, (S));
retention times: tR (S) 22.8 min, tR (R) 24.2 min.
2.6.5. (S)-3-Hydroxy-3-(thiophen-2-yl)propanenitrile (7b)
94% yield; 1H NMR (600 MHz, CDCl3): ı 2.77 (dd, 1H, J = 7.8 Hz,
CH2), 3.01 (dd, 1H, J = 7.2 Hz, J = 6.6 Hz, CH2), 5.27 (t, 1H, J = 6 Hz,
CHOH), 6.95 (m, 2H, Ar-H), 7.23 (m, 1H, Ar-H). [␣]D30 = −33.8
(c = 1.0, CHCl3) for >99% ee (lit. [31] [␣]D30 = −34.5 (c = 1.0, CHCl3)
for 99% ee, (S)); Retention times: tR (S) 13.8 min, tR (R) 15.7 min.
The steady-state kinetic parameters toward various substrates
were investigated using standard assay method except that MOPTA
was replaced with other substrates at varied concentrations. The
kinetic parameters toward cofactors were investigated in the pres-
ence of 10 mM MOTPA at varied cofactor concentrations ranging
from 1 to 200 ␮M and 0.1 to 10 mM for NADPH and NADH, respec-
tively. Data were fitted to the Michaelis–Menten equation using
Graph-Pad Prism v5.0 (GraphPad Software, San Diego, CA, USA) to
generate estimates of Km and kcat values.
3. Results and discussion
3.1. Identification and sequence analysis of ChKRED15
2.5. Preparative bioreductions catalyzed with mutant S12G and
GDH
ChKRED15 is one of the six ketoreductases mined from the
genome of Chryseobacterium sp. CA49 that follows the Prelog’s rule
[23]. It catalyzed the bioreduction of five substrates, 3a–7a, to the
corresponding (S)-alcohols with excellent enantioselectivity (>99%
ee), and the yield reached >95% for substrate 6a at a concentration
of 10 g/l (Fig. 2). For 1a and 2a, none of the six enzymes could cat-
alyze the bioreduction, which was most probably due to the strong
electron-donating effect of the beta-amine group.
ChKRED15 belonged to the short-chain dehydroge-
nase/reductase family (SDR). It contained 245 amino acids, a
typical length for SDR enzymes, which usually contain around
250–300 amino acids. The sequence was compared with those
in GenBank by using the BLASTp program. Several hypothetical
proteins were found to share maximal identities of around 90%
with ChKRED15, such as the predicted short-chain dehydroge-
nases from Chryseobacterium hispalense (WP 029294226.1), and
The reaction was carried out at 30 ◦C in 50 ml potassium phos-
phate buffer (300 mM, pH 8.0) containing 0.2 mM NADP+, 5% (w/v)
glucose, GDH (4 U/ml), crude enzyme extract or lyophilized powder
of the crude enzyme of the S12G mutant (2 mg/ml, 0.96 U/mg) and
the substrate (10 g/l). For substrate 6a, higher substrate concentra-
tion of 20–50 g/l was applied, and the crude enzyme of mutant S12G
and GDH were used at 2–10 mg/ml (0.96 U/mg) and 4–20 U/ml,
respectively. The pH of the reaction mixture was monitored and
maintained at 7.0–8.0 by the addition of NaOH (1 M). To monitor
the time-course of the biotransformation, samples were taken at
intervals for analysis. The reaction was terminated by extraction
with ethyl acetate, then the combined organic extracts were dried
with anhydrous sodium sulfate, concentrated under reduced pres-
sure and purified using silica gel column chromatography eluted
with petroleum ether/ethyl acetate (10:1, v/v). The products were
identified by 1H NMR analysis. The absolute configuration of the
product was determined by comparing the optical rotation with
the literature data.
2.6. Spectral data for biotransformation products
2.6.1. (S)-Ethyl-3-hydroxy-3-(thiophen-2-yl)propanoate (3b)
92% yield; 1H NMR (600 MHz, CDCl3): ı 1.29 (t, 3H, J = 7.8 Hz,
CH3), 2.61 (dd, 1H, J = 6.0 Hz, J = 4.8 Hz, CH2), 2.86 (dd, 1H, J = 6.0 Hz,
J = 4.8 Hz, CH2), 4.12 (q, 2H, J = 6.0 Hz, CH2), 5.14 (t, 1H, J = 6.0 Hz,
CHOH), 6.95 (m, 2H, Ar-H), 7.23 (m, 1H, Ar-H). [␣]D25 = −17.1
(c = 1.0, CHCl3) for >99% ee (lit. [28] [␣]D25 = −17.2 (c = 1.0, CHCl3)
for 99% ee, (S)); Retention times: tR (S) 7.4 min, tR (R) 14.5 min.
Fig. 2. Bioreduction of different substrates catalyzed with the wild-type ChKRED15.
Shown are the conversion (white) and enantiomeric excesses (black). The reaction
mixture containing 10 mM substrate, 10 mM NADPH and 0.1 mg enzyme in 1 ml
potassium phosphate buffer (100 mM, pH 8.0) was incubated at 30 ◦C for 30 min.
The conversion and optical purity were determined by HPLC after extraction with
ethyl acetate.
2.6.2. (S)-Methyl 3-hydroxy-3-(thiophen-2-yl)propanoate (4b)
94% yield; 1H NMR (600 MHz, CDCl3): ı 2.61 (dd, 1H, J = 6.0 Hz,
J = 4.8 Hz, CH2), 2.86 (dd, 1H, J = 6.0 Hz, J = 4.8 Hz, CH2), 3.71 (s, 3H,

Z.-Q. Ren et al. / Journal of Molecular Catalysis B: Enzymatic 113 (2015) 76–81
79
Fig. 3. Multiple sequence alignment of ChKRED15 with several members of the SDR family. The alignment was performed with the program DNAMAN. ChKRED15, short
chain dehydrogenase/reductase in this work (NCBI accession No. AHC30853.1); GDH-TM, gluconate 5-dehydrogenase from Thermotoga maritima (PDB: 1VL8); SDR-Ex, short
chain dehydrogenase/reductase from Exiguobacterium sp. F42 (NCBI accession No. AB154409); ADH-LK, alcohol dehydrogenase from Lactobacillus kefir (NCBI accession No.
AAP94029); CR-ML, carbonyl reductase from mouse lung (PDB: 1CYD).
Chryseobacterium daeguense (WP 027378341.1), but the alignment
of ChKRED15 with known short-chain dehydrogenase indicated
a much lower identity. The enzyme was most closely related to
gluconate 5-dehydrogenase from Thermotoga maritima with 35%
identity (Fig. 3).
Conserved residues of the SDR family, such as the active-site
tetrad of N-S-Y-K [32], the structure association motif NNAG [33]
and activity association motif PGXXXT [32], were highly conserved
in the ChKRED15 sequence (Fig. 3). However, the possible coenzyme
binding site of ChKRED15 was GANRSIG (Fig. 3), which was slightly
different from the G-rich motif of the SDR family (GXXXGXG) [34]
by the substitution of Ser to Gly.
3.2. Construction of mutant S12G and protein expression
Based on the sequence analysis, ChKRED15 lacks the G-rich
motif at the N-terminal conserved sequence region. This motif is
known to play important role in enzyme activity and stability for
SDRs [32,34]. Therefore, site-directed mutagenesis was performed
to recover the Gly12, resulting in the mutant S12G with a conserved
G-rich motif.
Fig. 4. SDS-PAGE analysis of ChKRED15. SDS-PAGE was performed on a 12% gel
under reduced condition. Lane 1, molecular weight markers; Lane 2, total soluble
protein of the wild-type ChKRED15; Lane 3, purified wild-type ChKRED15; Lane 4,
total soluble protein of mutant S12G; Lane 5, purified mutant S12G.
SDS-PAGE analysis indicated that both the wild-type ChKRED15
and mutant S12G were highly expressed in E. coli BL21 (DE3) cells
and that the majority of the recombinant proteins produced was in
soluble form (Fig. 4). Expression in the pET28a(+) vector resulted
in proteins with an N-terminal hexa-histidine tag that facilitated
their purification. Single-step affinity chromatography using Ni-
NTA agarose yielded the homogenous protein (Fig. 4). The apparent
molecular weight was around 30 kDa, which was in accordance
with the calculated molecular weight of the enzyme (27 kDa).
efficiency (kcat/Km) compared with the wild-type (Table 1), sug-
gesting the positive effect of the amino acid substitution, which
was in accordance with the critical role of the conserved G-rich
region in catalytic activity.
The S12G mutant also displayed increased kcat values toward
all of the five substrates (3a–7a) than the wild-type, but with
slightly lower apparent binding affinity as judged by the higher
Km (Table 2). The kinetic parameters varied among the substrates,
and substrates 3a, 4a and 6a had relative higher kcat values than the
other two, which was consistent with the results shown in Fig. 2.
3.3. Cofactor preference and catalytic activity of the wild-type
and mutant ChKRED15
Since ketoreductases require NADH or NADPH as the hydride
donor, we investigated the kinetic parameters of the wild-type
ChKRED15 and mutant S12G toward these two cofactors at varied
concentrations in the presence of 10 mM substrate 6a (Table 1).
The results showed that NADPH was more efficient than NADH
as a cofactor for both the wild-type ChKRED15 and the mutant
S12G as the Km values of both enzymes toward NADH were three
orders of magnitude higher than those toward NADPH, indicating
a much higher binding affinity of NADPH. In addition, the mutant
S12G displayed increased turn-over frequency (kcat) and catalytic
3.4. Effect of pH and temperature on the enzymatic activity
The pH dependence of activities of the wild-type ChKRED15
and mutant S12G were measured using purified enzymes with
6a as the substrate (Fig. 5). Both the wild-type ChKRED15 and
S12G mutant were active at a wide range of pH, and displayed the
maximal activity at pH 8.0 in potassium phosphate buffer (Fig. 5).
The components of the buffer did not have remarkable impact
on the enzymatic activity as the activity at optimal pH (pH 8.0)

80
Z.-Q. Ren et al. / Journal of Molecular Catalysis B: Enzymatic 113 (2015) 76–81
Table 1
Kinetic parameters of the wild-type and mutant ChKRED15 toward different cofactors.
Enzyme
Cofactor
Km (mM)
kcat (min−1
)
kcat/Km (mM−1 min−1
)
NADH
NADPH
2.4 0.4
9.4 0.5
80.5 1.2
3.8 0.6
Wild-type
2.5 0.2 × 10−3
3.2 0.3 × 104
NADH
NADPH
1.6 0.06
36.3 0.4
22.7 0.6
S12G
2.6 0.3 × 10−3
114
2
4.4 0.5 × 104
Table 2
Steady-state kinetic parameters of the wild-type and mutant ChKRED15.
Substrate
Enzyme
Km (mM)
kcat (min−1
)
kcat/Km (mM−1 min−1
)
Wild-type
S12G
1.8 0.1
2.6 0.2
80.9 1.2
44.9 2.6
43.2 3.5
3a
117
2
Wild-type
S12G
1.8 0.1
2.8 0.2
73.2 1.4
40.6 2.4
39.2 2.9
4a
5a
6a
7a
110
2
1
Wild-type
S12G
4.9 0.4
6.2 0.2
11.8
2.4 0.3
5.6 0.2
34.7 0.4
Wild-type
S12G
2.0 0.2
3.0 0.2
69.0 1.7
34.5 3.6
35.4 2.5
106
2
Wild-type
S12G
3.2 0.3
5.4 0.4
32.4 1.2
56.3 1.3
10.1 1.0
10.4 0.8
Fig. 7. Thermal inactivation of the wild-type ChKRED15 (dotted line) and the S12G
Fig. 5. Effect of reaction pH on the activity of the wild-typeChKRED15 (dotted line)
and the S12G mutant (solid line). Four different buffers, including sodium citrate (᭹),
Potassium phosphate (ꢁ), Tris–HCl (ꢀ), and sodium carbonate (ꢁ) were applied.
mutant (solid line). The enzymes were incubated at 35 ◦C (ꢀ), 40 ◦C (᭹) and 45 ◦
(ꢁ) for varied times and cooled on ice before the assay.
C
of the maximal activity of the wild-type, indicating enhanced
thermal tolerance. Additional measurements on thermal stability
confirmed that the S12G mutant was more thermostable than the
wild-type (Fig. 7). After 1-h incubation at 45 ◦C, the wild-type lost
over 60% activity, while the S12G mutant retained 99% activity.
Both enzymes were relatively stable at 35 ◦C, retaining over 70%
activity after 10-h treatment (Fig. 7).
The results clearly demonstrated that the recovery of the G-rich
motif by S12G mutation remarkably increased both the enzyme
activity and thermal tolerance, which is in accordance with pre-
vious structural studies showing the side-chain and main-chain
interactions of this motif within the central ␤-sheet and strands
␤D and ␤E [35,36]. Those interactions are considered essential for
coenzyme binding and maintenance of central ␤-sheet that are
related to catalytic activity and protein stability [34].
Fig. 6. Effect of reaction temperature on the activity of the wild-type ChKRED15 (ꢀ)
and the S12G mutant (ꢂ).
3.5. Preparative-scale biotransformations
in potassium phosphate buffer was similar with that in Tris–HCl
buffer. The mutant S12G displayed higher specific activity than the
wild-type under all the tested pH conditions.
The temperature dependence of the wild-type ChKRED15 and
mutant S12G were measured from 20 to 60 ◦C. The wild-type
displayed the maximal activity at 45 ◦C, and retained 42% of the
maximal activity at 50 ◦C (Fig. 6), while the mutant S12G had an
optimal temperature of 55 ◦C, and its specific activity was 1.7-fold
The preparative-scale biotransformations were performed in
the presence of the GDH-catalyzed cofactor recycling system, using
the crude enzyme of mutant S12G prepared from the cell extract
of recombinant E. coli. Because the recycling system would result
in the accumulation of stoichiometric gluconic acid, the pH of the
reaction mixture was continuously monitored and maintained at
7.0–8.0 by the addition of 1 M NaOH. For substrates 5a and 7a,

Z.-Q. Ren et al. / Journal of Molecular Catalysis B: Enzymatic 113 (2015) 76–81
81
Acknowledgements
This work was supported by the National Natural Science Foun-
dation of China (21372216), and the Open Fund of Key Laboratory
of Environmental and Applied Microbiology (KLCAS-2013-06 &
KLCAS-2014-05) and the Key Research Program (KGZD-EW-606-
14) of the Chinese Academy of Sciences.
References
[1] K. Balke, M. Kadow, H. Mallin, S. Sass, U.T. Bornscheuer, Org. Biomol. Chem. 10
(2012) 6249–6265.
[2] P.P. Deshpande, V.B. Nanduri, A. Pullockaran, H. Christie, R.H. Mueller, R.N. Patel,
J. Ind. Microbiol. Biotechnol. 35 (2008) 901–906.
[3] F. Hollmann, I.W.C.E. Arends, D. Holtmann, Green Chem. 13 (2011) 2285–2314.
[4] J.C. Moore, D.J. Pollard, B. Kosjek, P.N. Devine, Acc. Chem. Res. 40 (2007)
1412–1419.
[5] Y. Ni, J.H. Xu, Biotechnol. Adv. 30 (2012) 1279–1288.
[6] U.T. Bornscheuer, G.W. Huisman, R.J. Kazlauskas, S. Lutz, J.C. Moore, K. Robins,
Nature 485 (2012) 185–194.
Fig. 8. Time courses of the bioreduction of KEES (3a, dotted line) and MOTPA (6a,
solid line). MOTPA was applied at concentrations of 20 g/l (♦), 30 g/l (ꢁ), or 50 g/l
(᭹), using crude enzyme of the S12G mutant at 2, 5 and 10 mg/ml, respectively.
KEES was applied at 10 g/l (ꢁ), 20 g/l (ꢃ) using crude enzyme of the S12G mutant
(2 mg/ml).
[7] O.W. Gooding, R. Voladri, A. Bautista, T. Hopkins, G. Huisman, S. Jenne, S. Ma,
E.C. Mundorff, M.M. Savile, Org. Process Res. Dev. 14 (2010) 119–126.
[8] B. Kosjek, J. Nti-Gyabaah, K. Telari, L. Dunne, J.C. Moore, Org. Process Res. Dev.
12 (2008) 584–588.
[9] J. Liang, J. Lalonde, B. Borup, V. Mitchell, E. Mundorff, N. Trinh, D.A. Kochrekar,
R. Nair Cherat, G.G. Pai, Org. Process Res. Dev. 14 (2010) 193–198.
[10] J. Liang, E. Mundorff, R. Voladri, S. Jenne, L. Gilson, A. Conway, A. Krebber, J.
Wong, G. Huisman, S. Truesdell, J. Lalonde, Org. Process Res. Dev. 14 (2010)
188–192.
complete conversion could only be achieved at lower substrate
concentration of 5 mM with isolated yields of 94%, while for sub-
strates 3a, 4a and 6a, at least 10 g/l substrate could be completely
reduced. Because the biotransformation time-courses of substrates
3a and 4a closely resembled each other, only those of 3a and 6a
were shown (Fig. 8).
When the biotransformation of 3a or 4a was performed with
a substrate concentration of 10 g/l (50 mM), the substrate conver-
sion reached >99% after 6 h with absolute stereoselectivity (>99%
ee) (Fig. 8). The products (S)-3b and (S)-4b were isolated and
purified with yields of 92% and 94%, respectively. However, at a
substrate concentration of 20 g/l, the highest conversion of either
substrate was only 70% after 12-h reaction, which did not increase
with prolonged reaction time, most probably due to product
inhibition.
[11] S.K. Ma, J. Gruber, C. Davis, L. Newman, D. Gray, A. Wang, J. Grate, G.W. Huisman,
R.A. Sheldon, Green Chem. 12 (2010) 81–86.
[12] D.M. Solano, P. Hoyos, M.J. Hernaiz, A.R. Alcantara, J.M. Sanchez-Montero, Biore-
sour. Technol. 115 (2012) 196–207.
[13] F.P. Bymaster, E.E. Beedle, J. Findlay, P.T. Gallagher, J.H. Krushinski, S. Mitchell,
D.W. Robertson, D.C. Thompson, L. Wallace, D.T. Wong, Bioorg. Med. Chem. Lett.
13 (2003) 4477–4480.
[14] J. Deeter, J. Frazier, G. Staten, M. Staszak, L. Weigel, Tetrahedron Lett. 31 (1990)
7101–7104.
[15] H. Liu, B.H. Hoff, T. Anthonsen, Chirality 12 (2000) 26–29.
[16] C.G. Tang, H. Lin, C. Zhang, Z.Q. Liu, T. Yang, Z.L. Wu, Biotechnol. Lett. 33 (2011)
1435–1440.
On the contrary, substrate 6a apparently did not result in inhibi-
tion at the tested substrate concentrations (100–250 mM). Despite
of the slightly lower catalytic efficiency compared with 3a and
4a at substrate concentration of 10 mM (Table 2), 6a was the
best accepted substrate in preparative-scale bioreduction with
elevated substrate concentrations. Complete conversion could be
achieved within 3 h at a concentration of 20 g/l with no by-product
observed. The isolated yield of the product (S)-6b was 93%, and
the enantiopurity was >99% ee. When subjected to higher sub-
strate concentration of up to 50 g/l, the performance remained
excellent, achieving enantiopure (S)-6b with complete conver-
sion within 1 day with increased enzyme usage (5-fold) (Fig. 8).
The reaction afforded much higher substrate tolerance than previ-
ously reported bioreduction of duloxetine precursors [16,21,37,38].
For the same substrate, previous study used high concentration
of whole cells of R. glutinis (42 g CDW/l), which yielded compro-
mised stereoselectivity when substrate concentration exceeded
30 g/l [16].
[17] G.S. Forman, T. Ohkuma, W.P. Hems, R. Noyori, Tetrahedron Lett. 41 (2000)
9471–9475.
[18] S. Duprat de Paule, S. Jeulin, V. Ratovelomanana-Vidal, J.-P. Genêt, N. Champion,
P. Dellis, Eur. J. Org. Chem. 2003 (2003) 1931–1941.
[19] Y. Fujima, M. Ikunaka, T. Inoue, J. Matsumoto, Org. Process Res. Dev. 10 (2006)
905–913.
[20] K. Inoue, Y. Makino, N. Itoh, Appl. Environ. Microbiol. 71 (2005) 3633–3641.
[21] M. Wada, A. Yoshizumi, Y. Furukawa, H. Kawabata, M. Ueda, H. Takagi, S.
Nakamori, Biosci. Biotechnol. Biochem. 68 (2004) 1481–1488.
[22] M. Breuer, US 7790436 B2, 2010.
[23] Y. Liu, T.X. Tang, X.Q. Pei, C. Zhang, Z.L. Wu, J. Mol. Catal. B: Enzym. 102 (2014)
1–8.
[24] T.X. Tang, Y. Liu, Z.L. Wu, J. Mol. Catal. B: Enzym. 105 (2014) 82–88.
[25] G.Q. Chen, X.Y. Tang, M. Shi, Chem. Commun. 48 (2012) 2340–2342.
[26] J. Takehara, J.P. Qu, K. Kanno, H. Kawabata, Y. Dekishima, M. Ueda, K. Endo, T.
Murakami, T. Sasaki, H. Uehara, Y. Matsumoto, S. Suzuki, EP 1486493-A1, 2004.
[27] S.L. Boulet, S.A. Filla, P.T. Gallagher, K.J. Hudziak, A.M. Johansson, R.E. Karan-
jawala, J.J. Masters, V. Matassa, B.M. Mathes, R.E. Rathmell, M.A. Whatton, C.N.
Wolfe, WO 2004043931-A1, 2004.
[28] J. Brem, A. Liljeblad, C. Paizs, M.I. Tosa, F.D. Irimie, L.T. Kanerva, Tetrahedron:
Asymmetry 22 (2011) 315–322.
[29] R.J. Kloetzing, T. Thaler, P. Knochel, Org. Lett. 8 (2006) 1125–1128.
[30] J.N. Zhou, Q. Fang, Y.H. Hu, L.Y. Yang, F.F. Wu, L.J. Xie, J. Wu, S. Li, Org. Biomol.
Chem. 12 (2014) 1009–1017.
[31] A. Kamal, G.B.R. Khanna, R. Ramu, T. Krishnaji, Tetrahedron Lett. 44 (2003)
4783–4787.
4. Conclusion
[32] W.L. Duax, R. Huether, V. Pletnev, T.C. Umland, M.W. Charles, Int. J. Bioinform.
Res. Appl. 5 (2009) 280–294.
[33] B. Keller, A. Volkmann, T. Wilckens, G. Moeller, J. Adamski, Mol. Cell. Endocrinol.
248 (2006) 56–60.
[34] U. Oppermann, C. Filling, M. Hult, N. Shafqat, X.Q. Wu, M. Lindh, J. Shafqat, E.
Nordling, Y. Kallberg, B. Persson, H. Jornvall, Chem. Biol. Interact. 143 (2003)
247–253.
[35] C. Filling, K.D. Berndt, J. Benach, S. Knapp, T. Prozorovski, E. Nordling, R. Laden-
stein, H. Jornvall, U. Oppermann, J. Biol. Chem. 277 (2002) 25677–25684.
[36] U.C. Oppermann, C. Filling, K.D. Berndt, B. Persson, J. Benach, R. Ladenstein, H.
Jornvall, Biochemistry 36 (1997) 34–40.
The ketone reductase ChKRED15 from the genome of Chry-
seobacterium sp. CA49 was expressed in E. coli, and successfully
applied to catalyze the bioreduction of five duloxetine interme-
diates to yield the enantiopure (S)-alcohols with >99% ee. A single
mutation, S12G, could recover the conservative G-rich motif, which
significantly improved the catalytic activity and stability of the
enzyme. Complete conversion was achieved for substrate 3a within
6 h at a concentration of 10 g/l, and for substrate 6a within 24 h at
concentration of 50 g/l.
[37] M. Breuer, US Patent 0319208-A1, 2008.
[38] P. Soni, U.C. Banerjee, Appl. Microbiol. Biotechnol. 67 (2005) 771–777.