Attempt to simultaneously generate three chiral centers in 4-hydroxyisoleucine with microbial carbonyl reductases

Article abstract of DOI:10.1016/j.bmc.2017.06.044

A panel of microorganisms was screened for selective reduction ability towards a racemic mixture of prochiral 2-amino-3-methyl-4-ketopentanoate (rac-AMKP). Several of the microorganisms tested produced greater than 0.5 mM 4-hydroxyisoleucine (HIL) from rac-AMKP, and the stereoselectivity of HIL formation was found to depend on the taxonomic category to which the microorganism belonged. The enzymes responsible for the AMKP-reducing activity, ApAR and FsAR, were identified from two of these microorganisms, Aureobasidium pullulans NBRC 4466 and Fusarium solani TG-2, respectively. Three AMKP reducing enzymes, ApAR, FsAR, and the previously reported BtHILDH, were reacted with rac-AMKP, and each enzyme selectively produced a specific composition of HIL stereoisomers. The enzymes appeared to have different characteristics in recognition of the stereostructure of the substrate AMKP and in control of the 4-hydroxyl group configuration in the HIL product.

Full text of DOI:10.1016/j.bmc.2017.06.044

Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Attempt to simultaneously generate three chiral centers in 4-
hydroxyisoleucine with microbial carbonyl reductases
Makoto Hibi a,b, Koji Takahashi c, Junko Kako a, Yuuta Wakita c, Tomohiro Kodera d, Sakayu Shimizu c,
Kenzo Yokozeki a, Jun Ogawa c,
⇑
a Industrial Microbiology, Graduate School of Agriculture, Kyoto University, Kitashirakawa-oiwakecho, Sakyo-ku, Kyoto 606-8502, Japan
b Biotechnology Research Center and Department of Biotechnology, Toyama Prefectural University, 5180 Kurokawa, Imizu, Toyama 939-0398, Japan
c Division of Applied Life Sciences, Graduate School of Agriculture, Kyoto University, Kitashirakawa-oiwakecho, Sakyo-ku, Kyoto 606-8502, Japan
d Institute of Food Sciences & Technologies Flavor Innovation Group, Ajinomoto Co, Inc, 1-1, Suzuki-cho, Kawasaki-ku, Kawasaki 210-8681, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
A panel of microorganisms was screened for selective reduction ability towards a racemic mixture of
prochiral 2-amino-3-methyl-4-ketopentanoate (rac-AMKP). Several of the microorganisms tested pro-
duced greater than 0.5 mM 4-hydroxyisoleucine (HIL) from rac-AMKP, and the stereoselectivity of HIL
formation was found to depend on the taxonomic category to which the microorganism belonged. The
enzymes responsible for the AMKP-reducing activity, ApAR and FsAR, were identified from two of these
microorganisms, Aureobasidium pullulans NBRC 4466 and Fusarium solani TG-2, respectively. Three AMKP
reducing enzymes, ApAR, FsAR, and the previously reported BtHILDH, were reacted with rac-AMKP, and
each enzyme selectively produced a specific composition of HIL stereoisomers. The enzymes appeared
to have different characteristics in recognition of the stereostructure of the substrate AMKP and in control
of the 4-hydroxyl group configuration in the HIL product.
Received 6 February 2017
Revised 13 June 2017
Accepted 27 June 2017
Available online xxxx
Keywords:
4-Hydroxyisoleucine
Stereoisomer
Carbonyl reductase
Kinetic resolution
Asymmetric biosynthesis
Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction
4-Hydroxyisoleucine (HIL) is a hydroxy amino acid originally
found in the annual herbaceous plant, fenugreek (Trigonella foe-
The biocatalytic production of chiral compounds in the pharma-
ceutical, agrochemical, and fragrance industries has grown signifi-
cantly in recent years. Two types of enzymatic methods are
typically used for the generation of chiral compounds. The first
method is the kinetic resolution of racemic compounds using
enzymes able to recognize particular stereoisomers as their sub-
strates.1 The second method involves the asymmetric conversion
of prochiral compounds using enzymes catalyzing stereospecific
reactions.2,3 Although each method has been utilized individually,
less is known about the combination of the kinetic resolution and
the asymmetric conversion in the production of chiral compounds
from racemic prochiral substrates. If the combined methods were
applied in chiral alcohol production, multiple chiral carbon centers
could be simultaneously generated, and only the desired stereoiso-
mer would be obtained from a racemic mixture of the prochiral
ketone. In such a case, sophisticated carbonyl reductases with
stereoselective features both of substrate recognition and carbonyl
reduction would be required as the biocatalysts.
num-graecum Leguminosae).4 In HIL molecule, there are three chi-
ral centers generating eight stereoisomers consisting of four pairs
of enantiomers: (2S,3S,4S) and (2R,3R,4R)-HIL [SSS- and RRR-HIL];
(2S,3S,4R) and (2R,3R,4S)-HIL [SSR- and RRS-HIL]; (2S,3R,4R) and
(2R,3S,4S)-HIL [SRR- and RSS-HIL]; and (2S,3R,4S) and (2R,3S,4R)-
HIL [SRS- and RSR-HIL]. Of these, SRS-HIL is the major stereoisomer
naturally present in fenugreek.5 SRS-HIL was reported to have
potential for the treatment of non-insulin-dependent type II dia-
betes mellitus, where it stimulated insulin secretion depending
on plasma glucose level.6 Besides SRS-HIL, non-natural stereoiso-
mers were also reported to have antidiabetic activities.7,8 There-
fore, HIL stereoisomers are potential candidates for a novel class
of antidiabetic agent acting on the essential dysfunctions of type
II diabetes.
Recently, we reported that BtHILDH, an NAD-dependent alcohol
dehydrogenase derived from Bacillus thuringiensis 2e2, oxidized
SRS-HIL into (2S,3R)-2-amino-3-methyl-4-ketopentanoate [SR-
AMKP].9 BtHILDH also catalyzed the reverse carbonyl reducing
reaction, or asymmetric formation of SRS-HIL from SR-AMKP; how-
ever the reactivity of BtHILDH towards other AMKP stereoisomers,
⇑
Corresponding author.
E-mail address: ogawa@kais.kyoto-u.ac.jp (J. Ogawa).
http://dx.doi.org/10.1016/j.bmc.2017.06.044
0968-0896/Ó 2017 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Hibi M., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.044

2
M. Hibi et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
or its ability to produce other HIL stereoisomers, has not been
examined.
and HindIII endonucleases, and cloned into the expression vector
pQE80L (QIAGEN, CA, USA), which was digested with the same
endonucleases. The resultant plasmids, pQE-ApAR and pQE-FsAR
were transformed into E. coli JM109. A BtHILDH-expressing strain,
pET-BtHILDH E. coli Rosetta2 (DE3), was constructed as previously
reported.9
In present study, we tested various microorganisms for their
ability to react with a racemic mixture of prochiral AMKP (rac-
AMKP), and observed a tendency to form HIL isomers depending
on the microbial taxonomic category. Moreover, we successfully
purified and identified two carbonyl reductases acting on AMKP
from Aureobasidium pullulans NBRC 4466 and Fusarium solani TG-
2. These NADPH-dependent AMKP reductases, ApAR and FsAR,
were minutely estimated together with BtHILDH as biocatalysts
selectively producing specific HIL stereoisomers from rac-AMKP.
2.4. Expression and purification of recombinant AMKP-reducing
enzymes
Each of the E. coli transformants carrying pQE-ApAR, pQE-FsAR,
and pET-BtHILDH were cultured at 28 °C in LB medium comprised
of 1% (w/v) tryptone, 0.5% (w/v) yeast extract, and 1% (v/v) NaCl
with the addition of appropriate antibiotics. At an OD600 = 1.0, iso-
2. Materials and methods
propyl-b-D-thiogalactopyranoside (IPTG) was added to a final con-
2.1. Conditions for rapid amino acid analysis by achiral high-
performance liquid chromatography (achiral HPLC)
centration of 1 mM, and the cultures were incubated for 16 h at
28 °C with shaking at 300 rpm. The cell suspension (250 mL) was
centrifuged at 3000 rpm for 10 min and the cell pellet was sus-
pended in 10 ml of binding buffer containing 20 mM Tris-HCl buf-
fer (pH 7.4), 0.5 M NaCl, 20 mM imidazole, and 1 mM DTT, and
disrupted for 1 h by sonication with an Insonator 201 R (KUBOTA,
Osaka, Japan). The lysate was centrifuged at 12,000g for 30 min,
Amino acids were derivatized using the AccQ-Tag Derivatiza-
tion Kit (Waters, MA, USA) according to the manufacturer’s
instructions and analyzed with an Alliance 2695 HPLC system
(Waters). An XBridge C18 column (5
l
m, 2.1 Â 150 mm; Waters)
was used for separation at 40 °C. The mobile phases used were
10 mM ammonium acetate at pH 5.0 (eluent A) and methanol (elu-
ent B), and the flow rate of the eluent was 0.3 ml/min. The eluent
gradients were 0–1% (v/v) B for 0–0.5 min, 1–5% B for 0.5–
18 min, 5–9% B for 18–19 min, 9–17% B for 19–29.5 min, 17–60%
B for 29.5–40 min, and 60% B for 40–43 min. The AccQ-Tag deriva-
tives were detected with a fluorescence detector (Ex. 250 nm, Em.
395 nm).
and the supernatant was filtered with a 0.45-lm Millex Syringe-
driven Filter Unit (Millipore, MA, USA). The protein solution was
applied to a Ni-Sepharose column (His Trap HP 5 ml; GE Health-
care Bioscience) equilibrated with binding buffer. The column
was washed with binding buffer, and the proteins were eluted with
a linear gradient of 0.02–0.50 M imidazole in binding buffer. The
fractions containing the recombinant enzyme were pooled, con-
centrated by ultrafiltration and used as the purified enzyme. The
specific activities of these enzymes towards rac-AMKP as a sub-
2.2. Microbial screening for reduction activity of AMKP
strate were measured and expressed in lmol/min/mg (U/mg).
rac-AMKP was prepared according to the chemical synthesis
2.5. Simultaneous analysis of eight stereoisomers of HIL and four
stereoisomers of AMKP by chiral HPLC
previously reported.10 Approximately 10 mg of air-dried cells of
each strain preserved in our laboratory was added to 100 ll of
the reaction mixture in a test tube (10 Â 105 mm). The reaction
mixture contained 35 mM rac-AMKP, 1 mM NADH, 1 mM NADPH,
250 mM d-glucose, and 15 U/ml glucose dehydrogenase, in 50 mM
Tris-HCl buffer (pH 7.4). The reaction was carried out at 28 °C for
16 h with shaking (300 rpm). After 16 h, the reaction mixture
was centrifuged at 1200Âg for 15 min, and the supernatant was
analyzed by achiral HPLC.
Amino acid solutions (0.5 mg/ml amino acids and 4 mg/ml tri-
ethylamine in 50% acetonitrile) were mixed in equal amounts with
2 mg/ml 2,3,4,6-tetra-O-acetyl-b-D-glucopyranosyl isothiocyanate
(GITC) in acetonitrile, and were reacted for 30 min at RT. The amino
acid derivatives were analyzed using a LC-10A HPLC system (Shi-
madzu, Kyoto, Japan) equipped with a UV detector. The CAPCELL
PAK C18 column (5
l
m, 4.6 Â 250 mm; Shiseido, Tokyo, Japan)
was used for the separation at 45 °C. The mobile phases were
10 mM KH2PO4 (pH 2.8; eluent A) and acetonitrile (eluent B), and
the flow rate of the eluent was 0.3 ml/min. The eluent gradients
were 20–25% (v/v) B in 0–60 min and 25% (v/v) B in 60–70 min.
HIL and AMKP diastereomers were kind gifts from Ajinomoto Co.,
Inc (Kawasaki, Japan).11 The GITC-derivatives of stereoisomers
were detected spectrophotometrically at 250 nm.
2.3. Construction of expression strains for AMKP reductases, ApAR and
FsAR
Total RNA was prepared from A. pullulans NBRC 4466 and F.
solani TG-2. Approximately 0.1 g of wet cells of each strain was
added into a 2-ml stainless blender tube with a metal cone, pre-
cooled in liquid nitrogen, and then disrupted at 1700 rpm for
10 s with a Multi-Beads Shocker (Yasui Kikai, Osaka, Japan). The
total RNA fraction was extracted with ISOGEN (Nippon Gene,
Tokyo, Japan) according to the manufacturer’s instructions. The
cDNA was prepared from the total RNA using a PrimeScript RT-
PCR Kit (Takara Bio, Shiga, Japan). The ApAR gene was amplified
from cDNA obtained from A. pullulans NBRC 4466 cells by PCR
using the following primers: ACAGGATCCATGGACACCTCAAAAGC-
2.6. Reaction conditions for AMKP-reducing enzymes
The standard reaction conditions were as follows: 35 mM rac-
AMKP, 1 mM NADH or 1 mM NADPH, and 0.1 U/ml ApAR, FsAR,
or BtHILDH in 50 mM Tris-HCl buffer (pH 7.4) at 28 °C. The enzyme
activity was spectrophotometrically determined by measuring the
decrease in the absorbance at 340 nm resulting from the oxidation
of NAD(P)H. The amount of NAD(P)H was estimated using a molar
absorption coefficient of 6.22 Â 103 MÀ1ÁcmÀ1. The effect of pH on
the activity was examined by varying the reaction pH between 3.0
and 10.0. The effect of temperature on the activity was examined
by varying the reaction temperature between 10 °C and 70 °C.
For substrate specificity analysis, various carbonyl compounds
were used as the substrate in place of rac-AMKP. In order to pro-
CATCAACCCTTCG
and
AGTAAGCTTTTAGTATGTCGCGGGCAT-
CATGCTCAGGCG. The FsAR gene was amplified from cDNA
obtained from F. solani TG-2 cells by PCR using the following pri-
mers: ATAGGATCCATGGCTTCCGAAAACGCGAATAGCAGCGA and
TATAAGCTTTCAATTGGCCGCAGGGAAATTGCCCAGACG. The PCR
was carried out with Prime STAR polymerase (Takara Bio) under
the following conditions: 30 cycles of 10 s at 98 °C, 10 s at 58 °C,
and 1.5 min at 72 °C. The PCR product was digested with BamHI
duce HIL stereoisomers, 250 mM D-glucose and 15 U/ml glucose
Please cite this article in press as: Hibi M., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.044

M. Hibi et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
3
100%
80%
60%
40%
20%
0%
dehydrogenase were added to the standard reaction mixture.
Determination of HIL stereoisomers formed in the reaction mixture
was performed by the chiral HPLC analysis.
SRS/RSR
SRR/RSS
SSR/RRS
SSS/RRR
3. Results
3.1. Survey of the AMKP-reducing enzyme activity of microorganisms
A total of 1093 microbial strains from our culture collection
were analyzed for their ability to reduce AMKP into HIL. In 16 of
the strains, including 2 strains of bacteria, 5 strains of ascomyce-
tous yeasts, 7 strains of filamentous ascomycetes, and 2 strains
of basidiomycetes, HIL production of greater than 0.5 mM from
35 mM rac-AMKP was observed in the reaction mixture (Table 1).
These results indicated that the ability to reduce AMKP was widely
distributed across various genera of microorganisms. Of these, four
strains produced more than 5 mM of HIL (Bacillus thuringiensis
NBRC 3951, Sporobolomyces gracilis NBRC 1033, Aureobasidium pul-
lulans NBRC 4466, and Crinipellis stipitaria NBRC 30259), with S.
gracilis NBRC 1033 in particular showing the highest production
(11.6 mM).
Fig. 1. Selective production of HIL enantioisomers with various microorganisms.
HIL formed in the reaction mixture was analyzed with the achiral HPLC. Percentage
composition of enantioisomers produced from rac-AMKP is indicated.
These results suggest that microorganisms may possess several
AMKP-reducing enzymes with different characteristics.
Among the 16 strains, their reduction reaction products from
rac-AMKP differed in the composition of HIL enantioisomers
(Fig. 1). In most cases, the composition was dependent on the tax-
onomic category to which these microorganisms belong. Bacterial
strains such as NBRC 3951 and JCM 3205 produced only SRS/RSR-
and SSR/RSS-HIL isomers as their HIL products. All yeast strains
tested, NBRC 1033, NBRC 0385, NBRC 0715, NBRC 0879, and NBRC
4466, produced mainly SRS/RSR- and SSS/RRR-HIL isomers. Inter-
estingly, basidiomycetous strains such as NBRC 30259 and NBRC
30604 also produced SRS/RSR- and SSS/RRR-HIL isomers. Six strains
of filamentous ascomycetes, TG-2, NBRC 6605, NBRC 6813, NBRC
30538, 20-14, and NBRC 9435, showed low selectivity in the AMKP
reduction reaction and produced all enantioisomers of HIL. One fil-
amentous ascomycete, however, NBRC 30015, showed a similar
tendency to bacteria to produce SRS/RSR- and SSR/RSS-HIL isomers.
3.2. Purification and identification of microbial AMKP reductases
In order to purify AMKP reductase, we first tested for AMKP-
reducing activity in the cell-free extract of the above 16 strains.
Reproducible HIL production was observed in one bacterium
(NBRC 3951), four yeasts (NBRC 0385, NBRC 0715, NBRC 0879,
and NBRC 4466), three filamentous actinomycetes (TG-2, NBRC
6813, and 20-14), and a basidiomycetes (NBRC 30259). These nine
strains were then subjected to the enzymatic purification proce-
dure, and at present two proteins from A. pullulans NBRC 4466
and Fusarium solani TG-2 have been successfully identified as
enzymes responsible for AMKP-reducing activity.
3.3. ApAR, an AMKP reductase from A. pullulans NBRC 4466
The enzyme possessing AMKP-reducing activity was purified
from the cell-free extract of A. pullulans NBRC 4466 through seven
successive separation steps. Details of the purification procedure
are described in the Supplemental materials (Table S1). Homo-
geneity of the purified enzyme, ApAR, in the final preparation
was confirmed by SDS-PAGE and the estimated molecular weight
was approximately 36 kDa. Although the N-terminus of ApAR
was blocked to Edman degradation, its internal amino acid
sequences were successfully determined, as shown in Table S3.
The full-length genomic region containing the apaR gene was
determined by degenerate PCR and inverse PCR, which are
described in detail in the Supplemental materials. The apaR gene
consisted of 1010 bp of nucleotides including two introns and
three exons, and the protein-encoding region of the gene obtained
from the cDNA was 909 bp. The deduced amino acid sequence of
ApAR showed 95% homology with a short-chain dehydrogenases/
reductases (SDR) family enzyme of Aureobasidium melanogenum
CBS 110374 (GenBank ID: KEQ58950.1) on a BLAST search.
Table 1
Microorganisms possessing AMKP reducing activity to produce HIL.
Taxonomic category
Strain
HIL production
(mM)
Bacteria
Bacillus thuringiensis NBRC
3951
Rhodococcus ruber JCM3205
5.2
1.7
Ascomycetous yeasts
Sporobolomyces gracilis NBRC
1033
Cryptococcus albidus NBRC
0385
Rhodotorula pallida NBRC 0715 2.3
Rhodotorula marina NBRC 0879 2.0
Aureobasidium pullulans NBRC
11.6
2.6
5.1
4466
Filamentous
ascomycetes
Fusarium solani TG-2
3.5
Gibberella fujikuroi NBRC 6605
Acremonium fusidioides NBRC
6813
1.0
3.6
Acremonium terricola NBRC
30538
Acremonium strictum 20-14
Verticillium albo-atrum NBRC
9435
3.9
3.4. FsAR, an AMKP reductase from F. solani TG-2
3.5
0.6
F. solani TG-2 was isolated as a fungal strain utilizing triacry-
lonitrile.12 The enzyme possessing AMKP-reducing activity was
purified from the cell-free extract of F. solani TG-2 through seven
successive separation steps. Details of the purification procedure
are described in the Supplemental materials (Table S2). Homo-
geneity of the purified enzyme, FsAR, in the final preparation was
confirmed by SDS-PAGE and the estimated molecular weight was
Pithomyces maydicum NBRC
30015
0.5
Basidiomycetes
Crinipellis stipitaria NBRC
30259
Tricholoma matsutake NBRC
30604
5.8
2.0
Please cite this article in press as: Hibi M., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.044

4
M. Hibi et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
approximately 30 kDa. Although the N-terminus of FsAR was
blocked to Edman degradation, its internal amino acid sequences
were successfully determined, as shown in Table S4. The full-
length genomic region containing the fsaR gene was determined
by degenerate PCR and inverse PCR, which are described in detail
in the Supplemental materials. The fsaR gene consisted of
1047 bp of nucleotides including three introns and four exons,
and the protein-encoding region of the gene obtained from the
cDNA was 888 bp. The deduced amino acid sequence of FsAR
showed 98% homology with an SDR family enzyme of Nectria
haematococca mpVI 77-13-4 (GenBank ID: XP_003043314.1) on a
BLAST search.
3.7. Substrate specificity of microbial AMKP reductases
The recombinant ApAR, FsAR, and BtHILDH were reacted with
various carbonyl compounds, and the activities were determined
by consumption of NAD(P)H measured spectrophotometrically at
340 nm (Table 2). Similar substrate specificities were observed
for both ApAR and FsAR. Besides rac-AMKP, both enzymes showed
reducing activities towards 1,3-dichloroacetone, 2,3-indolinedione,
and levulinic acid. Relative activities for these substrates compared
to rac-AMKP were 16%, 10%, and 9.3% for ApAR, and 12%, 20%, and
30% for FsAR, respectively. On the other hand, BtHILDH had a differ-
ent substrate spectrum, and reacted with ketopantolactone, 1,4-
naphthoquinone, and various phenolic aldehydes much more than
rac-AMKP.
3.5. Multiple alignment analysis of AMKP reducing enzymes
The amino acid sequences of ApAR and FsAR were used for the
multiple alignment analysis together with that of BtHILDH derived
from B. thuringiensis 2e2, which we previously found as a NAD-
dependent HIL dehydrogenase (Fig. 2). ApAR and FsAR possessed
amino acid motifs for the NAD(P)-binding region (TGXXXGXG)
and for the active site (YXXXK), which are conserved in classical
short-chain alcohol dehydrogenase family enzymes.13 BtHILDH
possessed the conserved active site motif but a slightly different
NAD(P)-binding motif (SGXXXGXG). The sequence identities were
35% between ApAR and FsAR, 26% between ApAR and BtHILDH,
and 28% between FsAR and BtHILDH.
3.8. Selective production of HIL stereoisomers with ApAR, FsAR, and
BtHILDH
The recombinant ApAR, FsAR, and BtHILDH enzymes were eval-
uated as biocatalysts for the selective production of HIL stereoiso-
mers (Fig. 3). In the reaction mixture of FsAR with rac-AMKP as the
substrate, SSS-, SRS-, RSS-, and RRS-HIL were equally produced with
the conversion of >99%. This result indicated that FsAR catalyzed
the stereoselective carbonyl reducing reaction to form the (S)-4-
hydroxy group of HIL but did not recognize the stereo-configura-
tion of 2-amino and 3-methyl groups. Although the same result
was observed in the (S)-selectivity of AMKP reduction with ApAR,
its steric recognition of the substrate was somewhat different.
Interestingly, ApAR preferred SR-AMKP and also reacted well with
SS- and RR-AMKP, but poorly utilized RS-AMKP, resulting in the
conversion of 57%. This steric recognition ability of ApAR corre-
sponded to that of BtHILDH, but BtHILDH showed diverse stereos-
electivity in AMKP reduction. BtHILDH converted 30% of substrate
and produced SRS-, RRS- and a small amount of RSS-HIL like ApAR,
however SSR-HIL was produced instead of SSS-HIL. Thus, there was
a relationship between the stereo-configuration of the substrate
and stereoselectivity of carbonyl reduction in the HIL-producing
reaction by BtHILDH.
3.6. Catalytic properties of microbial AMKP reductases
The three types of AMKP reducing enzymes, ApAR, FsAR, and
BtHILDH, were heterologously expressed in E. coli and purified to
determine the various catalytic parameters for the reduction of
AMKP. ApAR and FsAR strictly preferred NADPH as a coenzyme,
in contrast to BtHILDH, an NADH-dependent enzyme. The maxi-
mum activities of ApAR and FsAR were both observed at pH 8.5
and 37 °C, and BtHILDH showed maximum activity at pH 6.5 and
45 °C. Specific activities of ApAR, FsAR, and BtHILDH were deter-
mined to be 5.5, 6.9, and 353 U/mg for rac-AMKP, respectively.
Fig. 2. Multiple alignment analysis of amino acid sequences of ApAR, FsAR, and BtHILDH. ApAR of A. pullulans NBRC 4466, FsAR of F. solani TG-2, and BtHILDH of B. thuringiensis
2e2 were used for the analysis. The common amino acid residues of these enzymes are in boxes. The NAD(P)-binding motif (TGXXXGXG) and active site motif (YXXXK) are
underlined, in which the conserved amino acid residues are highlighted in black.
Please cite this article in press as: Hibi M., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.044

M. Hibi et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
5
Table 2
Table 3
Substrate specificities of microbial AMKP reductases, ApAR, FsAR, and BtHILDH
Summary of catalytic characteristics of AMKP-reducing enzymes.
towards carbonyl compounds.
Enzyme
Cofactor
Stereoselectivity
Substrate
Relative activity (%)
dependency
Substrate recognition
Carbonyl
reduction
ApAR
FsAR
BtHILDH
Ketopantolactone
o-Chlorobenzaldehyde
1,4-Naphthoquinone
o-Nitrobenzaldehyde
p-Nitrobenzaldehyde
Terephthalaldehyde
m-Chrolobenzaldehyde
m-Nitrobenzaldehyde
Cuminaldehyde
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
100
n.d.
n.d.
n.d.
n.d.
n.d.
16
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
100
n.d.
n.d.
n.d.
n.d.
n.d.
12
822
478
424
375
323
297
245
126
122
100
83
28
27
12
6
n.d.
n.d.
n.d.
ApAR
FsAR
NADPH
NADPH
non-selective
2S/3R > 2S/3S = 2R/
3R > 2R/3S
2S/3R > 2S/3S = 2R/
3R > 2R/3S
4S
4S
BtHILDH NADH
4S (2S/3R,2R/
3R,2R/3S)
4R (2S/3S)
AMKP. On the other hand, ApAR preferred SR-AMKP to SS- and
RR-AMKP, and accepted RR-AMKP only slightly as its substrate.
This may be due to some steric hindrance in the substrate binding
pocket affecting access to RR-AMKP. The case of BtHILDH was more
complicated; its substrate preference was the same as FsAR and it
behaved like a (S)-selective carbonyl reductase, but an (R)-form
alcohol, SSR-HIL, was unexpectedly produced from SS-AMKP.
Structure analysis will be required to elucidate the structural fea-
tures around the catalytic sites and substrate binding pockets of
ApAR, FsAR, and BtHILDH, that result in the generation of the vari-
ous HIL stereoisomers.
rac-AMKP
p-Chlorobenzaldehyde
2,3-Butanedione
2-Decanone
Salicylaldehyde
2-Nonanone
1,3-Dichloroacetone
2,3-Indolinedione
Levulinic acid
10
9.3
20
30
n.d., not detected.
100%
80%
60%
40%
20%
The composition of HIL isomers produced was different
between ApAR and FsAR (Fig. 3) and the corresponding original
microorganisms, A. pullulans NBRC 4466 and F. solani TG-2
(Fig. 1). The cause might be that other predominant AMKP reduc-
tases with different selectivity or metabolic pathways for specific
HIL-isomers are exist in the original microorganisms. The novel
reductases and metabolic enzymes could be discovered from these
microorganisms by reexamination of enzymatic purification
method.
HIL(SSR)
HIL(SRS)
HIL(RSS)
HIL(RRS)
HIL(SSS)
We have already developed two types of enzymatic production
processes for the naturally occurring form of the stereoisomer SRS-
HIL. In the first case, acetaldehyde,
a-ketobutyrate, and L-gluta-
0%
mate were used for substrates of the cascade reaction with two
chirality-generating enzymes, 4-hydroxy-3-methyl-2-keto-pen-
tanoate aldolase and branched-chain amino acid transaminase.14
ApAR
FsAR
BtHILDH
Fig. 3. Selective production of HIL stereoisomers using ApAR, FsAR, and BtHILDH .
HIL formed in the reaction mixture was analyzed with the chiral HPLC. Percentage
composition of stereoisomers produced from rac-AMKP is indicated.
In the second case, a chiral substrate,
L-isoleucine, was directly
hydroxylated by Fe(II)/ -ketoglutarate-dependent dioxyge-
a
nase.11,15 Compared to these processes, the carbonyl reductase-
based method reported in this study has the advantage of generat-
ing three chiral carbon centers simultaneously from an inexpen-
sive racemic substrate. However, selectivity of these AMKP-
reducing enzymes is not enough to produce only a single stereoiso-
mer. To realize this concept, advanced biocatalysts with selectivi-
ties both in substrate recognition and carbonyl reduction need to
be obtained by further microbial screening or enzyme engineering
approaches. For example, mutation introduction to amino acid
residues around the substrate binding pocket is a promising
approach to increase selectivity of substrate recognition, especially
in the case of ApAR and FsAR that possess high selectivity for car-
bonyl reduction.
4. Discussion
According to BLAST search results of ApAR and FsAR, each of the
AMKP reductases showed homology with function unknown SDR
family enzymes of various species belonging to filamentous fungi.
BtHILDH is also an SDR family enzyme, but it has similarity with
many bacterial SDR family enzymes. Thus, it is not surprising that
the homologies are very low between ApAR and BtHILDH, or FsAR
and BtHILDH. However, only 35% homology was observed between
ApAR and FsAR, both fungal AMKP reductases. While BtHILDH has
been shown to convert SRS-HIL into SR-AMKP in bacterial sec-
ondary metabolism9, the physiological roles of ApAR and FsAR
are still unknown.
A. Supplementary data
The catalytic characteristics of these three AMKP-reducing
enzymes are summarized in Table 3. In spite of the low sequence
similarity, ApAR and FsAR were found to both be NADPH-depen-
dent enzymes catalyzing the stereoselective reduction of AMKP
to produce (S)-4-hydroxy-HIL, and both had similar catalytic char-
acteristics including optimum reaction temperature and pH. How-
ever, there was a slight difference in their steric preference for
substrate AMKP, resulting in different compositions of the HIL
stereoisomers produced, in particular RSS-HIL. FsAR did not recog-
nize the stereoconfiguration of 2-amino and 3-methyl groups of
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2017.06.044.
References
[1]. Ogawa J, Shimizu S. Diversity and versatility of microbial hydantoin-
transforming enzymes. J Mol Catal B Enzym. 1997;2:163–176.
[2]. Kataoka M, Kita K, Wada M, Yasohara Y, Hasegawa J, Shimizu S. Novel
bioreduction system for the production of chiral alcohols. Appl Microbiol
Biotechnol. 2003;62:437–445.
Please cite this article in press as: Hibi M., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.044

6
M. Hibi et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
[3]. Hibi M, Ogawa J. Characteristics and biotechnology applications of aliphatic
amino acid hydroxylases belonging to the Fe(II)/alpha-ketoglutarate-
dependent dioxygenase superfamily. Appl Microbiol Biotechnol.
2014;98:3869–3876.
[10]. Perlman D, Perlman KL, Bodanszky M, Bodanszky A, Foltz RL, Matthews HW.
Microbial production of vitamin B12 antimetabolites: II. 2-Amino-4-keto-3-
methylpentanoic acids from Bacillus cereus 439. Bioorg Chem.
1977;6:263–271.
[4]. Fowden L, Pratt HM, Smith A. 4-Hydroxyisoleucine from seed of Trigonella
foenum-graecum. Phytochemistry. 1973;12:1707–1711.
[5]. Alcock NW, Crout DHG, Gregorio MVM, Lee E, Pike G, Samuel CJ.
Stereochemistry of the 4-hydroxyisoleucine from Trigonella foenum-
graecum. Phytochemistry. 1989;28:1835–1841.
[6]. Broca C, Gross R, Petit P, et al. 4-hydroxyisoleucine: Experimental evidence of
its insulinotropic and antidiabetic properties. Am J Physiol Endocrinol Metab..
1999;277:40–44.
[11]. Kodera T, Smirnov SV, Samsonova NN, et al. A novel l-isoleucine
hydroxylating enzyme, l-isoleucine dioxygenase from Bacillus thuringiensis,
produces (2S,3R,4S)-4-hydroxyisoleucine. Biochem Biophys Res Commun.
2009;390:506–510.
[12]. Asano Y, Ando S, Tani Y, Yamada H, Ueno T. Fungal degradation of
triacrylonitrile. Agric Biol Chem. 1981;45:57–62.
[13]. Kavanagh KL, Jörnvall H, Persson B, Oppermann U. Medium- and short-chain
dehydrogenase/reductase gene and protein families. Cell Mol Life Sci.
2008;65:3895–3906.
[7]. Broca C, Manteghetti M, Gross R, et al. 4-Hydroxyisoleucine: effects of
synthetic and natural analogues on insulin secretion. Eur
2000;390:339–345.
[8]. Patent WO2006117696.
[9]. Ogawa J, Kodera T, Smirnov SV, et al. A novel L-isoleucine metabolism in
Bacillus thuringiensis generating (2S,3R,4S)-4-hydroxyisoleucine, a potential
insulinotropic and anti-obesity amino acid. Appl Microbiol Biotechnol.
2011;89:1929–1938.
J
Pharmacol.
[14]. Ogawa J, Yamanaka H, Mano J, et al. Synthesis of 4-hydroxyisoleucine by the
aldolase-transaminase coupling reaction and basic characterization of the
aldolase from Arthrobacter simplex AKU 626. Biosci Biotechnol Biochem.
2007;71:1607–1615.
[15]. Hibi M, Kawashima T, Kodera T, et al. Characterization of Bacillus thuringiensis
L-isoleucine dioxygenase for production of useful amino acids. Appl Environ
Microbiol. 2011;77:6926–6930.
Please cite this article in press as: Hibi M., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.044