Highly Efficient Deracemization of Racemic 2-Hydroxy Acids in a Three-Enzyme Co-Expression System Using a Novel Ketoacid Reductase

Article abstract of DOI:10.1007/s12010-018-2760-0

Enantiopure 2-hydroxy acids (2-HAs) are important intermediates for the synthesis of pharmaceuticals and fine chemicals. Deracemization of racemic 2-HAs into the corresponding single enantiomers represents an economical and highly efficient approach for synthesizing chiral 2-HAs in industry. In this work, a novel ketoacid reductase from Leuconostoc lactis (LlKAR) with higher activity and substrate tolerance towards aromatic α-ketoacids was discovered by genome mining, and then its enzymatic properties were characterized. Accordingly, an engineered Escherichia coli (HADH-LlKAR-GDH) co-expressing 2-hydroxyacid dehydrogenase, LlKAR, and glucose dehydrogenase was constructed for efficient deracemization of racemic 2-HAs. Most of the racemic 2-HAs were deracemized to their (R)-isomers at high yields and enantiomeric purity. In the case of racemic 2-chloromandelic acid, as much as 300 mM of substrate was completely transformed into the optically pure (R)-2-chloromandelic acid (> 99% enantiomeric excess) with a high productivity of 83.8 g L^?1 day^?1 without addition of exogenous cofactor, which make this novel whole-cell biocatalyst more promising and competitive in practical application.

Full text of DOI:10.1007/s12010-018-2760-0

Appl Biochem Biotechnol
https://doi.org/10.1007/s12010-018-2760-0
Highly Efficient Deracemization of Racemic 2-Hydroxy
Acids in a Three-Enzyme Co-Expression System Using
a Novel Ketoacid Reductase
1
,2
1,2
1,2
Ya-Ping Xue & Chuang Wang & Di-Chen Wang
&
1
,2
1,2
Zhi-Qiang Liu & Yu-Guo Zheng
Received: 2 February 2018 /Accepted: 10 April 2018
Springer Science+Business Media, LLC, part of Springer Nature 2018
#
Abstract Enantiopure 2-hydroxy acids (2-HAs) are important intermediates for the synthesis
of pharmaceuticals and fine chemicals. Deracemization of racemic 2-HAs into the correspond-
ing single enantiomers represents an economical and highly efficient approach for synthesizing
chiral 2-HAs in industry. In this work, a novel ketoacid reductase from Leuconostoc lactis
(LlKAR) with higher activity and substrate tolerance towards aromatic α-ketoacids was
discovered by genome mining, and then its enzymatic properties were characterized. Accord-
ingly, an engineered Escherichia coli (HADH-LlKAR-GDH) co-expressing 2-hydroxyacid
dehydrogenase, LlKAR, and glucose dehydrogenase was constructed for efficient
deracemization of racemic 2-HAs. Most of the racemic 2-HAs were deracemized to their
(R)-isomers at high yields and enantiomeric purity. In the case of racemic 2-chloromandelic
Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12010-018-
760-0) contains supplementary material, which is available to authorized users.
2
*
Yu-Guo Zheng
zhengyg@zjut.edu.cn
Ya-Ping Xue
xyp@zjut.edu.cn
Chuang Wang
9
47980211@qq.com
Di-Chen Wang
9
31213479@qq.com
Zhi-Qiang Liu
microliu@zjut.edu.cn
1
2
Key Laboratory of Bioorganic Synthesis of Zhejiang Province, College of Biotechnology and
Bioengineering, Zhejiang University of Technology, Hangzhou 310014, China
Engineering Research Center of Bioconversion and Biopurification of Ministry of Education,
Zhejiang University of Technology, Hangzhou 310014, China

Appl Biochem Biotechnol
acid, as much as 300 mM of substrate was completely transformed into the optically pure (R)-
−
1
2
-chloromandelic acid (> 99% enantiomeric excess) with a high productivity of 83.8 g L
−1
day without addition of exogenous cofactor, which make this novel whole-cell biocatalyst
more promising and competitive in practical application.
.
.
.
.
.
Keywords Biocatalysis 2-Hydroxyacid Deracemization Ketoacid reductase Co-expression
R)-2-Chloromandelic acid
(
Introduction
Enantiopure 2-hydroxy acids (2-HAs), a class of particularly important chiral building blocks,
are widely used in the synthesis of pharmaceuticals and fine chemicals [1–5]. For instance, (R)-
2
-chloromandelic acid is a valuable intermediate in the manufacture of anti-thrombotic agent,
(S)-clopidogrel [6–8]. (R)-2-Hydroxy-4-phenylbutyric acid plays an important role in the
production of angiotensin-converting enzyme (ACE) inhibitors such as cilazapril and
benazapril [9, 10].
Due to their importance, a series of approaches for the preparation of enantiopure 2-HAs
have been explored, and a great progress has been achieved in the past decades. Compared
with the traditional chemical approaches, several biocatalyst-mediated strategies are more
potential and preferable due to their high yields, strict enantioselectivity, mild reaction
conditions, and environmental friendliness [1, 11], mainly involving kinetic resolution of
racemic 2-HAs and derivatives with oxidases or hydrolases [12–14], asymmetric reduction
of prochiral 2-ketoacids with ketoacid reductases [15], and cascade deracemization of
racemic 2-HAs by coupling oxidation and reduction reactions with targeted oxidoreduc-
tases [16, 17]. Among these enzymatic routes, deracemization by coupling redox reactions
in one pot represents an economical and highly efficient fashion, which can efficiently
transform racemates of 2-HAs into the corresponding single enantiomers, thus
circumventing yield-reducing and time-consuming isolation of intermediates in multiple
reactions [18–20]. In recent years, multi-enzyme cascade reaction has become a very
important synthetic strategy in the field of biocatalysis and attracted a lot of attentions
[
21–25]. With the advance of genetic and metabolic engineering tools, redox cascade
reactions can be implemented by multi-gene co-expression in a target host cell, such as
Escherichia coli [26–29]. This method is more efficient, cost-saving, and easy to operate
compared with multi-strain cascade. In previous studies, the redox cascade reaction was
used for the synthesis of (R)-2-HAs including (R)-2-chloromandelic acid in one pot by
multi-strain cascade or a single recombinant E. coli co-expressing (S)-2-hydroxy acid
dehydrogenase (HADH), ketoacid reductase (KAR), and glucose dehydrogenase (GDH)
[
20, 30]. However, the substrate concentration was only 20 mM due to the poor activity of
KAR used in the reduction reaction.
In this work, we firstly discovered a novel ketoacid reductase from Leuconostoc lactis,
LlKAR, with higher activity and substrate tolerance by means of genome mining, and then
constructed a recombined E. coli co-expressing HADH, LlKAR, and GDH for highly efficient
deracemization of racemic 2-HAs (Fig. 1). (S)-2-HAs in the racemic mixtures were
enantioselective oxidized to the corresponding 2-ketoacids by HADH (FMN-dependence)
[31]. Then prochiral ketoacids were asymmetric reduced to (R)-2-HAs by LlKAR (NADH-
dependence), and NADH-regeneration was implemented by GDH. The substrate loading and

Appl Biochem Biotechnol
productivity were significantly improved due to the high activity of LlKAR, indicating that this
synthetic method is more competitive and promising in the industrial application.
Materials and Methods
Materials
The gene of HADH was cloned from Pseudomonas aeruginosa (AGM49308.1). The gene of
GDH was cloned from Exiguobacterium sibiricum (WP_012369122.1). The kits for rapid
extraction of plasmid and purification of DNA were purchased from Axygen Biotechnology
Co., Ltd. (Hangzhou, China). ClonExpress® II (One Step Cloning Kit) was supplied by
Vazyme Biotech Co., Ltd. (Nanjing, China). Plasmids pET28b and pCDFDuet (Novagen,
Darmstadt, Germany) were used as expression vectors. Both E. coli DH5α and E. coli BL21
(DE3) were cultured in Luria-Bertani (LB) medium as cloning host and expression host,
respectively. All the chemicals and agents used in experiments were obtained commercially.
Mining and Screening of KARs
The genome mining strategy was adopted for searching novel KARs with high activity and
enantioselectivity. Two functionally known KARs (LeKAR and ScKAR) was used as tem-
plates for the NCBI-pBLAST search. All the potential genes were synthesized in vitro, cloned
into pET28b, and then transformed into E. coli BL21 (DE3).
The specific activity and enantioselectivity of KARs were tested using the standard enzyme
activity assay and enantiomeric excess (e.e.) determination protocol, respectively. The further
comparison of KARs from different sources was carried out in a reaction system (10 mL)
containing potassium phosphate buffer (PPB, 100 mM, pH 7.0), phenylglyoxylic acid (100
+
and 400 mM), glucose (2.0 equiv. vs. substrate), NAD (0.5 mM), KARs (800 U/mL), and
Fig. 1 Whole-cell of recombinant E. coli (HADH-LlKAR-GDH) used for highly efficient deracemization of
racemic 2-HAs

Appl Biochem Biotechnol
GDH (800 U/mL, for regeneration of cofactor). Reactions were performed for 3 h at 35 °C,
7
00 rpm and titrated automatically with 3.0 M NaOH for maintaining pH at 7.0. Samples were
centrifuged (12,000 ×g for 3 min), diluted, and filtered for chiral HPLC analysis to calculate
conversion and e.e. values.
Cultivation of Recombinant E. coli
All the engineered E. coli strains were cultured in LB medium containing a certain amount of
antibiotics (50 μg/mL kanamycin for pET28b, 50 μg/mL streptomycin for pCDFDuet, and
both antibiotics for both vectors existing simultaneously) at 37 °C. When the OD600 of the
cultures reached 0.6–0.8, IPTG was added into the medium to a final concentration of 0.1 mM.
The cultures were incubated continually at 28 °C, 150 rpm for another 12 h. The cells were
collected by centrifugation (9000 ×g, 4 °C, 10 min) and then washed twice with physiological
saline.
Purification and Kinetic Parameters Analysis of KARs
Resting cells were resuspended in 100 mM PPB (pH 7.5) and disrupted by sonication. Then
the crude enzyme solution was obtained via centrifugation and stored at 4 °C for further use.
Bio-Scale™ Mini Nuvia™ IMAC Ni-charged cartridges (Bio-Rad Laboratories, Inc.) were
used for the purification of KARs. Detailed operating procedures referred to the instruction
manual and recommended protocol in the quick start guide. The fractions containing target
protein were pooled and dialyzed against 20 mM PPB (pH 7.0) for desalting. The purified
protein samples were analyzed by SDS-PAGE.
The kinetic parameters of purified enzymes were investigated by determining the initial
velocities of the enzymatic reactions as described previously [32, 33]. The concentration of
phenylglyoxylic acid, as a mode substrate, was varied from 0.5 to 20 mM with a fixed NADH
concentration (5 mM).
Catalytic Properties Characterization of Purified LlKAR
The optimum pH of LlKAR was examined in different buffers (100 mM, pH 5.0–9.0): CH3
COOH-CH COONa (pH 5.0–6.0), KH PO -K HPO (pH 6.0–8.0), and Tris-HCl (pH 8.0–
3
2
4
2
4
9
.0). The pH stability was investigated by determining the residual activity after pre-
incubating the purified LlKAR in different buffers (pH 5.0–9.0) at 4 °C for 12 h. The
optimum temperature was examined at various temperatures (20–55 °C). Thermal stability
was investigated by determining the residual activity of purified LlKAR at different
temperatures (25–65 °C), and then the corresponding parameters, such as thermal inacti-
vation rate constant (k ) and half-lives (t1/2), were obtained according to the reported
d
methods [34]. In addition, the thermal stability was further studied by recording the CD
spectra at different temperatures (25–65 °C, pre-incubating for 30 min, respectively) and
variable wavelengths (190–350 nm), using a spectropolarimeter J-815 (Jasco Co., Tokyo,
Japan).
The effects of metal ions and chemical agents on LlKAR activity were evaluated. The
residual activity was determined using the standard enzyme activity assay after the interaction
of purified LlKAR and different compounds (1.0 mM) for 20 min at 35 °C. Relative activity
here represents a percentage of the enzyme activity in the absence of any tested compound.

Appl Biochem Biotechnol
Several α-ketoacid substrates with aliphatic or aromatic substituents were chosen to
investigate the substrate specificity of LlKAR using the standard enzyme activity assay and
e.e. determination protocol. The activity towards phenylglyoxylic acid was regarded as 100%.
Analytic Methods
The standard enzyme activity assays of KARs were carried out in a reaction system (1.0 mL)
including 100 mM PPB (pH 7.0), 10 mM phenylglyoxylic acid, 5 mM NAD(P)H, and enzyme
at appropriate concentration. Reaction mixture and enzyme were preheated at 35 °C for 5 min,
respectively. Also, reactions were conducted at 35 °C and 700 rpm for 2 min and terminated by
6
.0 M HCl. Samples were centrifuged (12,000 ×g for 3 min), diluted, and filtered for chiral
HPLC analysis. One unit (U) of enzyme activity was defined as the amount of enzyme
required to catalyze the reduction of phenylglyoxylic acid for producing 1.0 μmol of mandelic
acid in 1.0 min under the standard enzyme activity assay conditions.
The e.e. values were determined via a reaction system (1.0 mL) consisted of 100 mM PPB
(pH 7.0), 10 mM phenylglyoxylic acid, 15 mM NAD(P)H, and enzyme at suitable concen-
tration. Reactions were performed at 35 °C and 700 rpm for 10 h and terminated with 6.0 M
HCl. Samples were centrifuged (12,000 ×g for 3 min), diluted, and filtered for chiral HPLC
analysis, which was performed according to the method reported previously [20].
Construction of Recombinant E. coli (HADH-LlKAR-GDH)
The gene of GDH from E. sibiricum and the gene of LlKAR from L. lactis were successively
cloned and linked to the expression plasmid pCDFDuet using seamless cloning technology
with ClonExpress® II (Vazyme Biotech Co., Ltd., Nanjing, China). Then the resulting
pCDFDuet-LlKAR-GDH was transformed into E. coli BL21 (DE3), and the target recombi-
nant transformants (E. coli (LlKAR-GDH)) were screened on the LB plates with 50-μg/mL
streptomycin.
The gene of HADH from P. aeruginosa was linked to the expression plasmid pET28b.
Then recombinant plasmids pET28b-HADH and pCDFDuet-LlKAR-GDH were extracted and
co-transformed into the expression host E. coli BL21 (DE3). Kanamycin (50 μg/mL) and
streptomycin (50 μg/mL) were added to the medium simultaneously for the selection of
positive transformants (E. coli (HADH-LlKAR-GDH)). The co-expression of three enzymes
(HADH, LlKAR, and GDH) was determined using SDS-PAGE.
Deracemization of Racemic 2-HAs with E. coli (HADH-LlKAR-GDH)
The cascade deracemization of racemic 2-HAs (1a-1s) was performed in a 10 mL reaction
system involving PPB (100 mM, pH 7.0), substrate (20 mM), glucose (20 mM), and whole-
cell catalyst (8 g DCW/L). Reactions were carried out at 30 °C, 700 rpm for 6 h. Samples were
taken once an hour, centrifuged (12,000 ×g for 3 min) for removing the catalyst, diluted, and
filtered to determine yields without isolation and e.e. values by chiral HPLC analysis.
Whole-Cell Biocatalytic Synthesis of (R)-1f
The synthesis of (R)-1f with whole-cell biocatalyst E. coli (HADH-LlKAR-GDH) was con-
ducted in a 10 mL reaction system including PPB (100 mM, pH 7.0), rac-1f, glucose (1.0

Appl Biochem Biotechnol
equiv. vs. substrate), and whole-cell catalyst. A series of reactions were performed under
various conditions as listed in Table 5 at 30 °C and 700 rpm. The process of deracemization
was monitored by chiral HPLC analysis and stopped until no further conversion. Whole-cell
biocatalytic synthesis of (R)-1f with recombinant E. coli (HADH-LlKAR-GDH) by adding
rac-1f in a fed-batch mode was started in a 10 mL potassium phosphate buffer (100 mM, pH
7
7
.0) containing rac-1f (100 mM), glucose (100 mM), and catalyst (20 g DCW/L) at 30 °C and
00 rpm. Rac-1f (100 mM) and glucose (100 mM) were added per 4 h. All the reactions were
titrated automatically with 3.0 M NaOH for maintaining pH at 7.0.
Results and Discussion
Mining and Screening of KARs
In previous work, a HADH (AGM49308.1) with high activity has been used for the oxidation
of 2-HAs. In order to match the activity of HADH in the three-enzyme co-expression system,
it is necessary to search for a KAR with high activity towards 2-keto acids to avoid its
accumulation. The strategy of genome mining, as an effective method, has been widely used in
the discovery of novel biocatalysts and could largely shorten the mining period [35]. In view of
this, a mini-library of KARs was constructed by a NCBI-pBLAST search with two amino acid
sequences of LeKAR from Leuconostoc mesenteroides and ScKAR from Saccharomyces
cerevisiae as templates. Six predicted KARs owning 62–84% identities of amino acid se-
quence with probe proteins were selected, and its genes were synthesized and overexpressed in
E. coli BL21 (DE3).
For screening an ideal biocatalyst from the mini-library, the activity and enantioselectivity
of all these KARs were tested with phenylglyoxylic acid as a mode substrate (Table 1).
LeKAR and KARs from entry1–3 exhibited better specific activities and perfect selectivity,
and among them LlKAR was the best one, giving a specific activity of 3.7 kU/mg crude
enzyme and e.e. value of > 99%. The substrate loadings of these KARs were further compared
using 100 and 400 mM phenylglyoxylic acid, respectively (Fig. 2). We found only LlKAR
could transform phenylglyoxylic acid at two various concentrations to the corresponding (R)-
1
a with excellent conversion (> 99%) and enantioselectivity (> 99%) within 3 h.
Four recombinant KARs were purified from crude enzyme solution by IMAC Ni-charged
cartridges to determine their kinetic parameters (Table 2). SDS-PAGE analysis displayed that
the electrophoretic pure KARs were obtained, and a single band of about 32 kDa was
presented in the corresponding lane (Fig. S1). Results of kinetic parameters analysis revealed
−1
−1
that LlKAR owned the maximum Vmax (7.29 mmol min mg ) and the minimum Km
(
(
(
4.30 mM) towards phenylglyoxylic acid compared with other KARs. The catalytic efficiency
−1
−1
K /K ) of LlKAR reached 981.48 mM s , up to 3.7 times of the original enzyme
cat
m
LeKAR), which indicated the application of LlKAR in cascade deracemization system was
of great potential. Thus, LlKAR from L. lactis was used in the following experiments.
Catalytic Properties Characterization of Purified LlKAR
In order to better understand and utilize LlKAR with promising activity and substrate
tolerance, we did a series of experiments to characterize its catalytic properties. LlKAR has
been proved to be a NADH-dependent reductase in initial experiments, the same as a α-

Appl Biochem Biotechnol
Table 1 Evaluation of KARs using phenylglyoxylic acid as a mode substrate
Entry
Name
GenBank
accession
no.
Microbial
strain
Identity to probe Specific
e.e. (%)
enzyme
(%)
activity
(kU/mg)
Probe
en-
zyme
LeKAR ANN45946.1
Leuconostoc
mesenteroides
–
–
1.1138
0.7849
> 99 (R)
> 95 (R)
ScKAR AAS56366.1
Saccharomyces
cerevisiae
1
2
LlKAR WP_068852189.1 Leuconostoc lactis 84 to LeKAR
LmKAR WP_041775202.1 Leuconostoc 74 to LeKAR
mesenteroides
KoKAR WP_047720787.1 Klebsiella oxytoca 49 to LeKAR
3.7077
3.3692
> 99 (R)
> 99 (R)
3
4
2.8895
0.1689
> 99 (R)
> 99 (R)
SkKAR EJT41757.1
Saccharomyces
kudriavzevii
85 to ScKAR
69 to ScKAR
62 to ScKAR
5
6
KaKAR XP_003955441.1 Kazachstania
0.1317
Not
> 99 (R)
Not
africana
Candida glabrata
CgKAR KTB10995.1
detectable
detectable
ketoacid reductase from L. mesenteroides [15], but unlike most other reported reductases for
catalytic reduction of ketoacids/esters especially from eukaryotic yeasts [32, 36–38].
The effect of pH on activity of purified LlKAR was determined in different buffers (pH 6.0–
9
.0). As shown in Fig. S2a, the optimal activity was observed at pH 7.0 and over 80% enzyme
activity could be examined at pH 6.0–7.5. The pH stability of LlKAR was excellent in a
neutral environment, and over 90% activity was still retained after being incubated at pH 6.0–
8
.0 and 4 °C for 12 h (Fig. S2b). The effect of temperature on activity was assayed at 20–55 °C
Fig. 2 Conversion and enantioselectivity of LeKAR, LlKAR, LmKAR, and KoKAR-catalyzed asymmetric
reductions of phenylglyoxylic acid at 0.1 and 0.4 M. Reaction conditions (35 °C, 700 rpm, 3 h): phenylglyoxylic
+
acid (100 and 400 mM), glucose (2.0 equiv. vs. substrate), NAD (0.5 mM), KARs (800 U/mL), and GDH
(
800 U/mL) without purification, in a 10 mL potassium phosphate buffer (100 mM, pH 7.0). Reactions were
titrated automatically with 3.0 M NaOH for maintaining pH at 7.0

Appl Biochem Biotechnol
Table 2 Kinetic parameters analyses of KARs with phenylglyoxylic acid as the substrate
max (mmol min− mg−1
1
)
K
m
(mM)
K
cat (s−1
)
m
Kcat/K )
(mM−1 s−1
Enzyme
V
LeKAR
LlKAR
LmKAR
KoKAR
3.71
7.29
7.25
4.47
8.22
4.30
5.11
5.90
2168.07
4220.91
4254.29
2525.64
263.78
981.48
832.38
427.93
(Fig. S2c). The maximum activity of LlKAR was detected at 35 °C, and it retained over 90%
catalytic activity at a relatively broad temperature (30–40 °C). Thermostability of purified
LlKAR was studied at 25–65 °C (Fig. S2d). The enzyme activity decreased relatively slowly at
less than 45 °C and still maintained about 30% of initial activity even being incubated at 45 °C
for 24 h. The rates of inactivation at 55 and 65 °C markedly accelerated with half-lives of 0.42
and 0.22 h, respectively (Fig. S2e & Table S1). In addition, CD spectra of LlKAR at various
temperatures were also used for the thermostability analysis. Obviously, the shapes of charac-
teristic peaks changed greatly at 55 and 65 °C, which was very consistent with the inactivation
curve above (Fig. S2f).
The effects of metal ions and chemical agents on LlKAR activity were evaluated as
2
+
2+
2+
illustrated in Fig. S3. The presence of Mn , Fe , and Ca has relatively slight stimulative
+
2+
+
effects on the activity of LlKAR. But, it was significantly inhibited by Cu , Cu , and Ni ,
which maybe cause by the interaction with key residues in enzyme structure, such as His
[
33, 39]. All the chemical agents tested have the inhibitory effects on the enzyme activity to
varying degrees. Among them, water miscible DMSO, water immiscible n-heptane, and
EDTA exhibited relatively small impacts, meanwhile indicating that LlKAR is not a
metalloenzyme.
Table 3 Substrate specificity of purified LlKAR
Substrate
Relative activity (%) a Yield (%) b e.e. (%) b
4
5
9.7
6.4
65.2
71.8
>99
97.7
>99 (R)
>99 (R)
>99 (R)
>99 (R)
100.0
85.6
83.5
96.0
>99 (R)
a
Reactions were carried out at the standard enzyme activity assay conditions and the activity towards
phenylglyoxylic acid was regarded as 100%
b
Reaction conditions (35 °C, 700 rpm, 10 h): substrate (10 mM), NADH (15 mM) and purified LlKAR (80 U),
in a 1.0 mL potassium phosphate buffer (100 mM, pH 7.0)

Appl Biochem Biotechnol
The substrate specificity of LlKAR was researched using several various α-ketoacid
substrates with aliphatic or aromatic substituents. As described in Table 3, LlKAR exhibited
higher activity and conversion towards aromatic α-ketoacids than aliphatic α-ketoacids, the
same as a phenylpyruvate reductase from Lactobacillus sp. CGMCC 9967 [33], probably
because the presence of benzene ring in the substituent group could promote hydrogen transfer
process. In addition, the increased distance between the benzene ring and the carbonyl group
could decrease enzyme activity and catalytic efficiency to a certain extent. All tested substrates
were reduced with perfect e.e. values (> 99%) following the anti-Prelog’s rule.
Construction of Recombinant E. coli (HADH-LlKAR-GDH)
The efficient and economical regeneration of NAD(P)H, due to their high cost and large
usage, is always critical for industrial applications of a large number of reductases. So,
many related strategies have been developed at present [40, 41]. Herein, we adopted the co-
expression of GDH from E. sibiricum and LlKAR selected above with promising activity
and substrate tolerance in one E. coli cell for the regeneration of NADH in the asymmetric
reduction reaction. Therefore, the recombinant plasmid pCDFDuet-LlKAR-GDH was
constructed and transformed into E. coli BL21 (DE3), and then SDS-PAGE analysis
indicated the co-expression of LlKAR and GDH in one cell was successfully realized with
two clearly visible bands of 32 and 28 kDa (Fig. S4). Subsequently, to establish redox
cascade reaction system for highly efficient deracemization of racemic 2-HAs, the resulting
pCDFDuet-LlKAR-GDH and recombinant plasmid pET28b-HADH (the gene from
P. aeruginosa) with different antibiotic resistances were co-transformed into the expression
host E. coli BL21 (DE3). SDS-PAGE analysis confirmed that the co-expression of HAHD,
LlKAR, and GDH in one cell was successfully achieved with three clearly visible bands of
Table 4 Enantioselective cascade biocatalysis for deracemization of racemic 2-HAs with recombinant E. coli
a
(
HADH-LlKAR-GDH)
Entry
Substrate
Reaction time (h)
Yield of (R)-1 (%)
e.e. of (R)-1 (%)
1
2
3
4
5
6
7
8
9
rac-1a
rac-1b
rac-1c
rac-1d
rac-1e
rac-1f
rac-1g
rac-1h
rac-1i
rac-1j
rac-1k
rac-1l
rac-1m
rac-1n
rac-1o
rac-1p
rac-1q
rac-1r
rac-1s
2
2
2
2
2
2
2
2
2
2
2
2
2
4
4
4
4
4
4
98.16
98.19
98.37
98.60
99.52
98.36
98.75
99.40
98.19
98.99
99.27
98.96
98.98
85.27
81.98
90.53
86.64
96.70
95.80
> 99
> 99
> 99
> 99
> 99
> 99
> 99
> 99
> 99
> 99
> 99
> 99
> 99
95.1
94.3
> 99
> 99
93.4
91.6
10
1
1
12
13
14
15
16
17
18
19
a
Reaction conditions (30 °C, 700 rpm, 6 h): each substrate (1a-1s, 20 mM), glucose (20 mM), and whole-cell
catalyst (8 g DCW/L), in a 10 mL potassium phosphate buffer (100 mM, pH 7.0)

Appl Biochem Biotechnol
4
2, 32, and 28 kDa (Fig. S4). This strategy of one-cell three-enzyme co-expression could
avoid, to a certain extent, cumbersome processes of cell cultivation and mass transfer
difficulties in multi-cellular system.
Deracemization of Racemic 2-HAs with E. coli (HADH-LlKAR-GDH)
To investigate the applicability of recombinant E. coli coexpressing three enzymes, the
cascade deracemization of racemic 2-HAs (1a-1s) was conducted using the resting cells of
E. coli (HADH-LlKAR-GDH) as catalysts. The yields and e.e. values of (R)-1 without
isolation at different reaction times were determined by chiral HPLC as listed in Table 4.
The nature, position, and number of substituent on the benzene ring and the distance
between benzene ring and hydroxy group have different effects on the yields and e.e.
values of target products. For racemic 1a-1m, the corresponding (R)-2-HAs were obtained
with excellent yields (> 98%) and e.e. values (> 99%) at 2 h. Interestingly, all the yields
towards 2-HAs (1b-1m) with different substituents on benzene ring, either electron-
withdrawing or electron-donating groups, were slightly higher than that without substituent
(
1a). The yields for 2-HAs with o-, m-, and p-substituents are higher than 98% although the
steric effects (1f-1h, 1i-1k). For racemic 1n-1s, the yields and e.e. values (except for 1p-1q
with e.e. values of > 99%) of products were relatively lower than that of other substrates
even when the reaction time was extended to 4 h, as results of the (S)-isomer of 2-HAs
residues (1r-1s), or intermediate ketoacids accumulation (1p-1q), or both (1n-1o). In
general, the recombinant E. coli coexpressing HADH, LlKAR, and GDH could achieve
highly efficient deracemization of most racemic 2-HAs with satisfactory yields and e.e.
values. The results demonstrated that this newly constructed system has a great potential for
application.
Whole-Cell Biocatalytic Synthesis of (R)-1f
(R)-1f is a significant chiral synthon for preparation of clopidegrel, one of salable anti-
thrombotic agents [6, 7, 14]. In order to further confirm the application potential of this newly
built cascade deracemization system, the resting cells of recombinant E. coli (HADH-LlKAR-
GDH) coexpressing three enzymes were cultured and collected for the preparation of (R)-1f.
As shown in Table 5, when the catalyst dosage was 10 g DCW/L, 150 mM of rac-1f was
completely transformed into the corresponding (R)-1f with perfect yields (> 98%) and e.e.
a
Table 5 Whole-cell biocatalytic synthesis of (R)-1f with recombinant E. coli (HADH-LlKAR-GDH)
b
Entry
Catalyst (g DCW/L)
rac-1f (mM)
Time (h)
Yield (%)
e.e. (%)
1
2
3
4
5
6
10
10
10
10
20
20
50
8
> 99
> 99
> 99
95.7
> 99
94.8
> 99
> 99
> 99
> 99
> 99
> 99
100
150
200
200
300
14
18
20
16
36
a
Reaction conditions (30 °C, 700 rpm): rac-1f (various concentrations as listed above), glucose (1.0 equiv. vs.
substrate), catalyst (various dosages as listed above), in a 10 mL potassium phosphate buffer (100 mM, pH 7.0).
Reactions were performed until no further conversion. pH was kept at 7.0 by adding automatically with 3.0 M
NaOH in the process of reaction

Appl Biochem Biotechnol
Fig. 3 Whole-cell biocatalytic synthesis of (R)-1f with recombinant E. coli (HADH-LlKAR-GDH) by adding
rac-1f in a fed-batch mode. Reaction was started in a 10 mL potassium phosphate buffer (100 mM, pH 7.0)
containing rac-1f (100 mM), glucose (100 mM), and catalyst (20 g DCW/L) at 30 °C and 700 rpm. Also, rac-1f
(
100 mM) and glucose (100 mM) were added per 4 h. pH was kept at 7.0 by adding automatically with 3.0 M
NaOH in the process of reaction
values (> 99%). In the case of 200 mM rac-1f, (R)-1f could be produced with a yield of > 99%
and an e.e. value of > 99% at 16 h after increasing the catalyst dosage to 20 g DCW/L, and a
productivity of 55.9 g L− day was achieved. Subsequently, the substrate loading was further
increased up to 300 mM and only a yield of 94.8% was obtained until 36 h, and further
increasing cell dosage was not beneficial to the biotransformation. However, when the strategy
of adding substrate in a fed-batch mode was used in this experiment, 300 mM rac-1f could be
completely transformed into (R)-1f with a yield of > 99% and an e.e. value of > 99% within
1
−1
−1
−1
1
6 h as shown in Fig. 3, giving a higher productivity of 83.8 g L day . Compared with the
previous work for the deracemization of rac-1f (substrate concentration:20 mM, productivity
−1
−1
of 44.0 g L day ) [20], The substrate concentration and the productivity were increased 14
times and 90.6%, respectively. The result indicated that the newly constructed E. coli (HADH-
LlKAR-GDH) has a greater potential for practical application.
Conclusion
In summary, a novel ketoacid reductase (LlKAR) exhibiting higher activity and substrate
tolerance towards aromatic α-ketoacids has been successfully discovered and characterized.
The newly reconstructed recombinant E. coli co-expressing HADH, LlKAR, and GDH
exhibited higher catalytic efficiency for the deracemization of racemic 2-HAs. As much as
3
00 mM rac-1f could be completely transformed into (R)-1f with high yield and enantiopurity
without addition of exogenous cofactor, making this novel whole-cell biocatalyst more
promising and competitive in practical application.

Appl Biochem Biotechnol
Funding Information This work was funded by the National Natural Science Foundation of China (No.
21676254).
Compliance with Ethical Standards
Conflict of Interest The authors declare that there is no conflict of interest.
Ethical Statement The authors declare that there are no studies conducted with human participants or animals.
References
1
. Groger, H. (2001). Enzymatic routes to enantiomerically pure aromatic α-hydroxy carboxylic acids: a
further example for the diversity of biocatalysis. Advanced Synthesis & Catalysis, 343(6–7), 547–558.
. Ma, B. D., Yu, H. L., Pan, J., Liu, J. Y., Ju, X., & Xu, J. H. (2013). A thermostable and organic-solvent
tolerant esterase from Pseudomonas putida ECU1011: catalytic properties and performance in kinetic
resolution of α-hydroxy acids. Bioresource Technology, 133(2013), 354–360.
2
3
4
. Chen, X., Wu, Q., & Zhu, D. (2015). Enzymatic synthesis of chiral 2-hydroxy carboxylic acids. Process
Biochemistry, 50(5), 759–770.
. Sheng, B. B., Xu, J., Ge, Y. S., Zhang, S., Wang, D. Q., Gao, C., Ma, C. Q., & Xu, P. (2016). Enzymatic
resolution by a D-lactate oxidase catalyzed reaction for (S)-2-hydroxycarboxylic acids. ChemCatChem,
8
(16), 2630–2633.
5
. Zhang, Z. J., Pan, J., Ma, B. D., & Xu, J. H. (2016). Efficient biocatalytic synthesis of chiral chemicals.
Advances in Biochemical Engineering/Biotechnology, 155, 55–106.
. Zhang, C. S., Zhang, Z. J., Li, C. X., Yu, H. L., Zheng, G. W., & Xu, J. H. (2012). Efficient production of
6
(R)-o-chloromandelic acid by deracemization of o-chloromandelonitrile with a new nitrilase mined from
Labrenzia aggregata. Applied Microbiology and Biotechnology, 95(1), 91–99.
7
. Wang, H., Sun, H., Gao, W., & Wei, D. (2014). Efficient production of (R)-o-chloromandelic acid by
recombinant Escherichia coli cells harboring nitrilase from Burkholderia cenocepacia J2315. Organic
Process Research & Development, 18(6), 767–773.
8
9
. Hu, Y., Wu, C., Wu, X. Y., Li, S. L., Sun, X. X., & Tang, Z. B. (2015). Efficient preparation of (R)-2-
chloromandelic acid via a recycle process of resolution. Chirality, 27(3), 281–285.
.
Bai, Y., & Yang, S. T. (2005). Biotransformation of R-2-hydroxy-4-phenylbutyric acid by D-lactate
dehydrogenase and Candida boidinii cells containing formate dehydrogenase coimmobilized in a fibrous
bed bioreactor. Biotechnology and Bioengineering, 92(2), 137–146.
1
0. Sheng, B. B., Zheng, Z. J., Lv, M., Zhang, H. W., Qin, T., Gao, C., Ma, C. Q., & Xu, P. (2014). Efficient
production of (R)-2-hydroxy-4-phenylbutyric acid by using a coupled reconstructed D-lactate dehydroge-
nase and formate dehydrogenase system. PLoS One, 9(8), e104204.
11. Xue, Y. P., Cao, C. H., & Zheng, Y. G. (2018). Enzymatic asymmetric synthesis of chiral amino acids.
Chemical Society Reviews, 47(4), 1516–1561.
1
2. Dewanti, A. R., Xu, Y., & Mitra, B. (2004). Role of glycine 81 in (S)-mandelate dehydrogenase from
Pseudomonas putida in substrate specificity and oxidase activity. Biochemistry, 43(33), 10692–10700.
3. Ma, B. D., Kong, X. D., Yu, H. L., Zhang, Z. J., Dou, S., Xu, Y. P., Ni, Y., & Xu, J. H. (2014). Increased
catalyst productivity in α-hydroxy acids resolution by esterase mutation and substrate modification. ACS
Catalysis, 4(3), 1026–1031.
1
1
1
1
1
4. Xue, Y. P., Shi, C. C., Xu, Z., Jiao, B., Liu, Z. Q., Huang, J. F., Zheng, Y. G., & Shen, Y. C. (2015). Design
of nitrilases with superior activity and enantioselectivity towards sterically hindered nitrile by protein
engineering. Advanced Synthesis & Catalysis, 357(8), 1741–1750.
5. Norihiro, K., Hiroaki, Y. & C, D (2004). Alpha-keto acid reductase, method for producing the same, and
method for producing optically active alpha-hydroxy acids using the same. US Patent Application,
2
004086993 A1.
6. Tsuchiya, S., Miyamoto, K., & Ohta, H. (1992). Highly efficient conversion of (±)-mandelic acid to its (R)-
−)-enantiomer by combination of enzyme-mediated oxidation and reduction. Biotechnology Letters,
4(12), 1137–1142.
(
1
7. Adam, W., Lazarus, M., Saha-Moller, C. R., & Schreier, P. (1998). Quantitative transformation of racemic 2-
hydroxy acids into (R)-2-hydroxy acids by enantioselective oxidation with glycolate oxidase and subsequent
reduction of 2-keto acids with D-lactate dehydrogenase. Tetrahedron: Asymmetry, 9(2), 351–355.

Appl Biochem Biotechnol
1
1
2
2
2
2
8. Schrittwieser, J. H., Sattler, J., Resch, V., Mutti, F. G., & Kroutil, W. (2011). Recent biocatalytic oxidation-
reduction cascades. Current Opinion in Chemical Biology, 15(2), 249–256.
9. Schrittwieser, J. H., Velikogne, S., Hall, M., & Kroutil, W. (2017). Artificial biocatalytic linear cascades for
preparation of organic molecules. Chemical Reviews, 118(1), 270–348.
0. Xue, Y. P., Zeng, H., Jin, X. L., Liu, Z. Q., & Zheng, Y. G. (2016). Enantioselective cascade biocatalysis for
deracemization of 2-hydroxy acids using a three-enzyme system. Microbial Cell Factories, 15(1), 162.
1. Lopez-Gallego, F., & Schmidt-Dannert, C. (2010). Multi-enzymatic synthesis. Current Opinion in Chemical
Biology, 14(2), 174–183.
2. Xue, R., & Woodley, J. M. (2012). Process technology for multi-enzymatic reaction systems. Bioresource
Technology, 115(2012), 183–195.
3. Bayer, T., Milker, S., Wiesinger, T., Rudroff, F., & Mihovilovic, M. D. (2015). Designer microorganisms for
optimized redox cascade reactions—challenges and future perspectives. Advanced Synthesis & Catalysis,
357(8), 1587–1618.
24. Liu, Y., Xu, G., Han, R., Dong, J., & Ni, Y. (2017). Identification of D-carbamoylase for biocatalytic cascade
synthesis of D-tryptophan featuring high enantioselectivity. Bioresource Technology, 249(2018), 720–728.
25. France, S. P., Hepworth, L. J., Turner, N. J., & Flitsch, S. L. (2016). Constructing biocatalytic cascades:
in vitro and in vivo approaches to de novo multi-enzyme pathways. ACS Catalysis, 7(1), 710–724.
26. Ema, T., Yagasaki, H., Okita, N., Takeda, M., & Sakai, T. (2006). Asymmetric reduction of ketones using
recombinant E. coli cells that produce a versatile carbonyl reductase with high enantioselectivity and broad
substrate specificity. Tetrahedron, 62(26), 6143–6149.
2
7. Hu, X., Liu, L., Chen, D., Wang, Y., Zhang, J., & Shao, L. (2017). Co-expression of the recombined alcohol
dehydrogenase and glucose dehydrogenase and cross-linked enzyme aggregates stabilization. Bioresource
Technology, 224(2017), 531–535.
28. Jiang, W., & Fang, B. (2016). Construction of a tunable multi-enzyme-coordinate expression system for
biosynthesis of chiral drug intermediates. Scientific Reports, 6(1), 30462.
2
9. Wei, P., Gao, J. X., Zheng, G. W., Wu, H., Zong, M. H., & Lou, W. Y. (2016). Engineering of a novel
carbonyl reductase with coenzyme regeneration in E. coli for efficient biosynthesis of enantiopure chiral
alcohols. Journal of Biotechnology, 230(2016), 54–62.
3
3
3
3
3
0. Xue, Y. P., Zheng, Y. G., Zhang, Y. Q., Sun, J. L., Liu, Z. Q., & Shen, Y. C. (2013). One-pot, single-step
deracemization of 2-hydroxyacids by tandem biocatalytic oxidation and reduction. Chemical
Communications, 49(91), 10706–10708.
1. Kara, S., Schrittwieser, J. H., Hollmann, F., & Ansorge-Schumacher, M. B. (2014). Recent trends and novel
concepts in cofactor-dependent biotransformations. Applied Microbiology and Biotechnology, 98(4), 1517–
1529.
2. Shen, N. D., Ni, Y., Ma, H. M., Wang, L. J., Li, C. X., Zheng, G. W., Zhang, J., & Xu, J. H. (2012). Efficient
synthesis of a chiral precursor for angiotensin-converting enzyme (ACE) inhibitors in high space-time yield
by a new reductase without external cofactors. Organic Letters, 14(8), 1982–1985.
3. Xu, G. C., Zhang, L. L., & Ni, Y. (2016). Enzymatic preparation of D-phenyllactic acid at high space-time
yield with a novel phenylpyruvate reductase identified from Lactobacillus sp. CGMCC 9967. Journal of
Biotechnology, 222(2016), 29–37.
4. Xu, Z., Cai, T., Xiong, N., Zou, S. P., Xue, Y. P., & Zheng, Y. G. (2018). Engineering the residues on BA^
surface and C-terminal region to improve thermostability of nitrilase. Enzyme and Microbial Technology,
113, 52–58.
3
5. Behrens, G. A., Hummel, A., Padhi, S. K., Schätzle, S., & Bornscheuer, U. T. (2011). Discovery and protein
engineering of biocatalysts for organic synthesis. Advanced Synthesis & Catalysis, 353(13), 2191–2215.
6. Ema, T., Ide, S., Okita, N., & Sakai, T. (2008). Highly efficient chemoenzymatic synthesis of methyl (R)-o-
chloromandelate, a key intermediate for clopidogrel, via asymmetric reduction with recombinant
Escherichia coli. Advanced Synthesis & Catalysis, 350(13), 2039–2044.
3
37. Guo, J., Mu, X., Zheng, C., & Xu, Y. (2009). A highly stable whole-cell biocatalyst for the enantioselective
synthesis of optically active alpha-hydroxy acids. Journal of Chemical Technology and Biotechnology,
84(12), 1787–1792.
3
8. Chen, R., Deng, J., Lin, J., Yin, X., Xie, T., Yang, S., & Wei, D. (2016). Assessing the stereoselectivity of
carbonyl reductases toward the reduction of OPBE and docking analysis. Biotechnology and Applied
Biochemistry, 63(4), 465–470.
3
4
4
9. Lei, J., Zhou, Y. F., Li, L. F., & Su, X. D. (2009). Structural and biochemical analyses of YvgN and YtbE
from Bacillus subtilis. Protein Science, 18(8), 1792–1800.
0. Chenault, H. K., & Whitesides, G. M. (1987). Regeneration of nicotinamide cofactors for use in organic
synthesis. Applied Biochemistry and Biotechnology, 14(2), 147–197.
1. Wu, H., Tian, C. Y., Song, X. K., Liu, C., Yang, D., & Jiang, Z. Y. (2013). Methods for the regeneration of
nicotinamide coenzymes. Green Chemistry, 15(7), 1773–1789.