Preparation of d-threonine by biocatalytic kinetic resolution

Article abstract of DOI:10.1016/j.molcatb.2015.09.011

D-Threonine is one of the important unnatural amino acids used as chiral building blocks in pharmaceutical drugs. Owing to the presence of two chiral centers, a synthetic protocol, either through chemocatalysis or biocatalysis, has not yet been available for one-step preparation of stereochemically pure d-threonine in terms of enantiomeric and diastereomeric excesses (i.e., both >99%). Here we demonstrate that facile production of d-threonine can be implemented using threonine deaminase (TD) via kinetic resolution of dl-threonine that can be readily prepared by conventional organic synthesis. TD catalyzes the dehydration/deamination of l-threonine, leading to generation of 2-oxobutyrate and ammonia. In contrast to mild substrate inhibition of the TD activity by l-threonine (i.e., apparent inhibition constant (KIapp) = 950 mM), d-threonine turned out to be a strong inhibitor (i.e., KIapp = 41 mM). In addition to the enzyme inhibitions by both enantiomers of threonine, cell lysis observed during small-scale kinetic resolutions of ≥1 M dl-threonine led us to carry out a preparative-scale reaction at 500 mM racemic substrate. The preparative-scale kinetic resolution in a 50 mL reaction mixture charged with 3 g dl-threonine and 3400 U whole cells was completed at 5 h with >99% ee of d-threonine. Product isolation by a cation-exchange chromatography led to white solid of d-threonine (1.36 g, 90.7% isolation yield). To explore whether our strategy could afford coproduction of another valuable unnatural amino acid, the pass-through solution from the cation-exchange column was further processed by a ω-transaminase (ω-TA) reaction where 2-oxobutyrate was converted to enantiopure homoalanine using isopropylamine as an amino donor. Addition of S- and R-selective ω-TA to the pass-through solution led to 93.2 and 90.9% reaction yield within 12 h with both >99% ee of the produced l- and d-homoalanine, respectively.

Full text of DOI:10.1016/j.molcatb.2015.09.011

Journal of Molecular Catalysis B: Enzymatic 122 (2015) 227–232
Contents lists available at ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
Preparation of d-threonine by biocatalytic kinetic resolution
Sang-Woo Han, Jong-Shik Shin∗
Department of Biotechnology, Yonsei University, Seoul 120-749, South Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 2 June 2015
Received in revised form 3 September 2015
Accepted 21 September 2015
Available online 26 September 2015
d-Threonine is one of the important unnatural amino acids used as chiral building blocks in pharmaceuti-
cal drugs. Owing to the presence of two chiral centers, a synthetic protocol, either through chemocatalysis
or biocatalysis, has not yet been available for one-step preparation of stereochemically pure d-threonine
in terms of enantiomeric and diastereomeric excesses (i.e., both >99%). Here we demonstrate that facile
production of d-threonine can be implemented using threonine deaminase (TD) via kinetic resolution of
dl-threonine that can be readily prepared by conventional organic synthesis. TD catalyzes the dehydra-
tion/deamination of l-threonine, leading to generation of 2-oxobutyrate and ammonia. In contrast to mild
substrate inhibition of the TD activity by l-threonine (i.e., apparent inhibition constant (KIapp) = 950 mM),
d-threonine turned out to be a strong inhibitor (i.e., KIapp = 41 mM). In addition to the enzyme inhibi-
tions by both enantiomers of threonine, cell lysis observed during small-scale kinetic resolutions of
≥1 M dl-threonine led us to carry out a preparative-scale reaction at 500 mM racemic substrate. The
preparative-scale kinetic resolution in a 50 mL reaction mixture charged with 3 g dl-threonine and 3400 U
whole cells was completed at 5 h with >99% ee of d-threonine. Product isolation by a cation-exchange
chromatography led to white solid of d-threonine (1.36 g, 90.7% isolation yield). To explore whether our
strategy could afford coproduction of another valuable unnatural amino acid, the pass-through solu-
tion from the cation-exchange column was further processed by a ␻-transaminase (␻-TA) reaction
where 2-oxobutyrate was converted to enantiopure homoalanine using isopropylamine as an amino
donor. Addition of S- and R-selective ␻-TA to the pass-through solution led to 93.2 and 90.9% reaction
yield within 12 h with both >99% ee of the produced l- and d-homoalanine, respectively.
Keywords:
d-Threonine
Kinetic resolution
Threonine deaminase
Transaminase
Homoalanine
© 2015 Published by Elsevier B.V.
1. Introduction
The stereochemical complexity of threonine renders stereos-
elective organic synthesis of d-threonine impracticable for an
Unnatural amino acids have gained a growing attention in
chemical industry owing to their versatile utilities as chiral
intermediates for a wide range of pharmaceutical drugs [1–4]. d-
Threonine is one of the important unnatural amino acids and is used
as a chiral building block of peptidomimetic drugs for treatment of
peripheral opioids side effects [5] and for analgesic applications
[6]. Threonine bears two chiral centers, leading to four stereo-
chemical isomers among which (2S, 3R) and (2R, 3S) configurations
correspond to l- and d-enantiomers of threonine, respectively. In
contrast to l-threonine which is mass-produced by microbial fer-
mentation for animal feed applications [7,8], such a fermentative
production is not yet available for d-threonine.
industrial scale-up due to low conversion and/or coproduction of
stereochemical impurities [9–11]. In contrast, organic synthesis of
dl-threonine with a negligible level of diastereomeric impurities
has been well established [12–14]. This has spurred development
of chiral resolution of dl-threonine by preferential crystallization,
which requires several successive crystallizations to achieve a high
enantiopurity [15–17].
In the case of biocatalytic approaches, enzymatic kinetic reso-
lution of dl-threonine derivatives were demonstrated using acid
phosphatase [18] and lipase [19]. These methods require an addi-
tional chemical step for the derivatization of racemic threonine.
Another biocatalytic approach employs asymmetric synthesis of
d-threonine from acetaldehyde and glycine using d-threonine
aldolase [20,21]. However, this biocatalytic synthetic method has
been suffered by substantial coproduction of unwanted stereo-
chemical impurities.
∗
Corresponding author at: Engineering Building
2 (Rm 507), Yonsei Uni-
In this study, we aimed at developing an industrially feasible
method for production of optically pure d-threonine by enzymatic
kinetic resolution of underivatized racemic threonine (Fig. 1). To
versity, Shinchon-Dong 134, Seodaemun-Gu, Seoul 120-749, South Korea.
Fax: +82 2 362 7265.
E-mail address: enzymo@yonsei.ac.kr (J.-S. Shin).
http://dx.doi.org/10.1016/j.molcatb.2015.09.011
1381-1177/© 2015 Published by Elsevier B.V.

228
S.-W. Han, J.-S. Shin / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 227–232
(GE Healthcare, Piscataway, USA) as described previously [27]. The
cell-free extract was loaded on a HisTrap HP column (GE Health-
care) and eluted by an elution buffer (20 mM sodium phosphate,
0.5 M sodium chloride, 0.5 mM PLP, pH 7.4) with a linear gradient of
imidazole (0.05–0.5 M). A HiTrap desalting column (GE Healthcare)
was used for removing imidazole using an exchange buffer (50 mM
sodium phosphate, 0.15 M sodium chloride and 0.1 mM PLP, pH 7).
2.3. Enzyme assay
Unless otherwise specified, enzyme assays were carried out
at 37 ◦C in 50 mM phosphate buffer (pH 7). One unit of TD is
defined as the enzyme amount catalyzing formation of 1 ␮mole
2-oxobutyrate in 1 min at 50 mM l-threonine. One unit of ␻-TA is
defined as the enzyme amount catalyzing formation of 1 ␮mole
acetophenone in 1 min at 10 mM pyruvate and 10 mM (S) or
(R)-␣-methylbenzylamine, depending on the stereoselectivity of
the ␻-TA. After 10 min reaction, the enzyme reaction (typically
100 ␮L reaction volume) was stopped by adding 600 ␮L acetoni-
trile. For the initial rate measurements (i.e., conversions <10%),
2-oxobutyrate and acetophenone for TD and ␻-TA, respectively,
were analyzed by HPLC.
Fig. 1. Schematics for the kinetic resolution of dl-threonine using threonine deam-
inase. d-Threonine is obtained from the reaction mixture when l-threonine is
completely converted to 2-oxobutyrate and ammonia.
this end, we choose threonine deaminase (TD, EC 4.2.1.16) which
is a pyridoxal phosphate (PLP)-dependent enzyme and performs
stereoselective conversion of l-threonine into 2-oxobutyrate and
ammonia [22,23]. Preparative-scale kinetic resolution and prod-
uct isolation were performed to demonstrate scalable production
of d-threonine. In addition, we demonstrated that 2-oxobutyrate
generated from the kinetic resolution could be converted to another
valuable unnatural amino acids using ␻-transaminase (␻-TA).
2.4. Substrate inhibition of TD
To examine substrate inhibition of TD by l-threonine, initial
reaction rates (vi) were measured at varying concentrations of
l-threonine (i.e., 10–400 mM). To determine the apparent inhibi-
tion constant (i.e., KIapp), curve fitting to a hyperbolic function (Eq.
(1)) was performed using the initial rate data where the activ-
ity decrease by the substrate inhibition was observed (i.e., above
100 mM l-threonine for cell-free extract).
2. Materials and methods
2.1. Chemicals and bacterial strains
v0
d-Threonine was purchased from Alfa Aesar (Ward Hill, USA).
l-Threonine and isopropylamine were purchased from Junsei
Chemical Co. (Tokyo, Japan). Materials used for preparation of
culture media including yeast extract, tryptone and agar were pur-
chased from Difco (Spark, USA). All other chemicals were purchased
from Sigma–Aldrich Co. (St. Louis, USA) and of the highest grade
available. Bacterial strains used for this study were Escherichia coli
BL21(DE3) cells transformed with the pET21a(+) expression vector
harboring a TD gene cloned from E. coli DH5␣ and the pET28a(+)
vectors harboring ␻-TA genes from Ochrobactrum anthropi and
Arthrobacter sp., that were constructed previously [24–26].
vi =
(1)
1 + L − L /Kapp
([ ] [ ]0)
I
[l]0 is the concentration of l-threonine at which the activity
decrease begins and v0 represents the initial rate at [l]0. The Kapp
value was calculated from the linear regression of 1/vi agaiInst
[l] − [l]0.
In the case of the enzyme inhibition by d-threonine, initial
rates were measured at varying concentrations of d-threonine (i.e.,
0–250 mM) with the concentration of l-threonine fixed at 50 mM.
To determine KIapp for d-threonine, the hyperbolic decay in the
enzyme activity was fitted to Eq. (2).
v0
vi =
(2)
2.2. Preparation of cell suspension, cell-free extract and purified
enzyme
1 + [D]/KIapp
[d] is the concentration of d-threonine and v0 represents a refer-
ence reaction rate where the enzyme inhibition by d-threonine is
absent (i.e., [d] = 0). The KIapp value was calculated from the linear
regression of 1/vi against [d].
To examine substrate inhibition by racemic threonine, initial
rates were measured with cell suspension at varying concentra-
tions of dl-threonine (i.e., 10–500 mM). 2-Oxobutyrate produced
was analyzed by HPLC for the initial rate measurements.
Overexpression of the enzymes was carried out as described
previously with minor modifications [27]. E. coli BL21(DE3) cells
carrying the expression vectors were cultivated in a LB medium
(typically 1 L) containing 50 ␮g/mL ampicillin or kanamycin at
37 ◦C. Protein expression was induced by 0.1 mM IPTG at 0.4 OD600
and the cells were allowed to grow for 10 h. Cells were harvested
by centrifugation, washed twice with potassium phosphate buffer
(50 mM, pH 7) and then resuspended in the same buffer (typically
10 mL). This cell suspension was used for the whole cell reactions.
To prepare cell-free extracts, the harvested cell pellet was resus-
pended in a 15 mL cell disruption buffer (50 mM NaCl, 1 mM EDTA,
1 mM ␤-mercaptoethanol, 0.1 mM PMSF, 0.02% sodium azide and
0.5 mM PLP in 50 mM Tris–HCl at pH 7). Cells were disrupted by
ultrasonication and then cell debris were removed by centrifuga-
tion (10,000 × g, 60 min, 4 ◦C). The resulting supernatant was used
as the cell-free extract for enzyme reactions.
2.5. Small-scale kinetic resolution of dl-threonine
Unless otherwise specified, small-scale kinetic resolutions were
carried out in 50 mM potassium phosphate buffer (pH 7) at 37 ◦C
under magnetic stirring. Racemic threonine was prepared by mix-
ing equal amount of each enantiomer. To examine the effect of
substrate feeding methods, kinetic resolution reactions were car-
ried out by adding cell suspension (final concentration = 13.5 U/mL)
to the reaction mixture (5 mL reaction volume). The batch reac-
tion starting with 300 mg dl-threonine (final concentration = 0.5 M)
In the case of ␻-TAs, purified proteins were used for enzyme
reactions. Purification of ␻-TAs was performed on ÄKTAprime plus

S.-W. Han, J.-S. Shin / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 227–232
229
was compared with a fed-batch reaction which started with 90 mg
dl-threonine (final concentration = 0.15 M) and was supplemented
with dl-threonine twice during the reaction (90 and then 120 mg)
when the ee value exceeded 90%.
To compare reaction progresses of kinetic resolutions at differ-
ent concentrations of dl-threonine, the reactions were started by
adding cell suspension (final concentration = 68 U/mL) to the reac-
tion mixture (1 mL reaction volume) containing 60, 120 and 180 mg
dl-threonine (final concentration = 0.5, 1 and 1.5 M, respectively).
Because the solubility of threonine is 745 mM at room temperature
[28], solid threonine was added to the reaction mixtures instead
of using a concentrated stock solution. 50 ␮L aliquots of the reac-
tion mixture were taken at predetermined reaction times and were
mixed with 10 ␮L of 5 N HCl. To determine ee and reaction yield,
quantitative chiral analysis of dl-threonine was performed by HPLC
after derivatization with a Marfey’s reagent [29].
Fig. 2. Effect of the cultivation time on the specific cellular activity of TD. Specific
reaction rate (closed circle) and optical density of the cell culture (open circle) were
monitored for 15 h after IPTG addition to the culture broth. Specific reaction rate
was measured at 50 mM l-threonine in phosphate buffer (50 mM, pH 7).
2.6. Preparative-scale kinetic resolution
Preparative-scale kinetic resolution was carried out in a 50 mL
reaction mixture (50 mM phosphate buffer, pH 7) charged with 3 g
dl-threonine (500 mM) and 3400 U cell suspension (430 mg dcw) at
37 ◦C under magnetic stirring. When the ee of d-threonine exceeded
99%, the reaction mixture was processed for product isolation.
Cells were removed by centrifugation (10,000 × g, 20 min, 4 ◦C).
The pH of the supernatant was adjusted to 1.0 by adding 5 N HCl
and then the resulting protein precipitate was filtered off using a
glass-fritted filter funnel. The filtrate solution was loaded on a glass
column packed with Dowex 50WX8 cation-exchange resin (40 g),
followed by washing with 0.1 N HCl (50 mL) and water (50 mL).
Elution was done by 10% ammonia solution (150 mL). The elution
fraction was evaporated at 50 ◦C and 0.1 bar, resulting in white solid
of d-threonine.
dissolved in acetonitrile), 18 ␮L of sodium bicarbonate solution
(1 M) and 50 ␮L of DMSO were added to 22 ␮L reaction samples
(the molar ratio of Marfey’s reagent to amino acid was >1.4). The
reaction mixture was incubated at 40 ◦C for 8 h and then cooled at
room temperature. 20 ␮L of 1 N HCl solution was added to the mix-
ture to quench the derivatization reaction. UV detector was tuned
at 340 nm.
3. Results and discussion
3.1. Determination of optimal cultivation time for cell suspension
preparation
2.7. Asymmetric synthesis of homoalanine
Whole cells expressing intended enzymes are usually preferred
over purified enzymes for a cost-effective biocatalyst formulation.
Therefore, we undertook determination of an optimal cell culti-
vation time to achieve the highest enzyme activity required for
efficient whole cell reactions. Protein expression was induced by
0.1 mM IPTG at 0.4 OD600 and then the cellular activity of TD nor-
malized by a dry cell weight as well as the cell optical density was
monitored for 15 h (Fig. 2). Cell growth reached a stationary phase
at 6 h after the IPTG addition. However, specific cellular activity
of TD kept increasing until 10 h (i.e., 7.9 U/mg dcw) and then the
prolonged cell cultivation led to a decrease in the specific activity.
Therefore, all the experiments later were carried out using the TD-
expressing cells harvested at 10 h after the IPTG induction. Based
on the specific activity and OD600 data, total cellular activity of TD
obtainable from 1 L cell culture at 10 h was estimated to be 6300 U.
The pass-through solution from the cation-exchange column
was used as a stock solution of 2-oxobutyrate after pH adjust-
ment to 7.0 by adding 5 N NaOH. Asymmetric syntheses of l- and
d-homoalanine were carried out in 1 mL reaction mixture (50 mM
phosphate buffer, pH 7) containing 50 mM 2-oxobutyrate, 100 mM
isopropylamine and the purified ␻-TA (5.9 and 11.2 U/mL of ␻-TA
from O. anthropi and Arthrobacter sp., respectively) at 37 ◦C under
magnetic stirring. Reaction yield and ee were monitored by quan-
titative chiral analysis of homoalanine after derivatization with a
Marfey’s reagent [29].
2.8. Analytical methods
All the HPLC analysis were performed with a Waters HPLC sys-
tem (Milford, USA). Analysis of 2-oxobutyrate was carried out using
a Aminex HPX-87H column (Bio-Rad, Hercules, USA) with isocratic
elution of water (pH adjusted to 2.0 by sulfuric acid) at 0.5 mL/min.
UV detection was done at 210 nm and the column oven temperature
was maintained at 40 ◦C. Analysis of acetophenone was performed
using a Sunfire C18 column (Waters Co.) with isocratic elution of
60% methanol/40% water/0.1% trifluoroacetic acid at 1 mL/min. UV
detection was done at 254 nm.
Quantitative chiral analysis of threonine and homoalanine were
carried out using the Sunfire C18 column (Waters Co.) with iso-
cratic elution of aqueous MeOH solution, 35 and 45% v/v for analysis
of threonine and homoalanine, respectively, containing 0.1% tri-
fluoroacetic acid at 1 mL/min after derivatization with a Marfey’s
reagent (1-fluoro-2,4-dinitrophenyl-5-l-alanine amide) which is
widely used for chiral analysis of amino acids [29]. In a typical
derivatization procedure, 10 ␮L of Marfey’s reagent stock (14 mM
3.2. Substrate inhibition of TD
TD is involved in the metabolic pathway for synthesis of
isoleucine and is one of the typical allosteric enzymes. This enzyme
is known to undergo feedback inhibition by isoleucine (i.e., the end
product of the metabolic pathway) as well as feedback activation
by valine (i.e., the end product of the parallel pathway) [22,23]. In
addition, it is known that the active site of TD shows high bind-
ing affinity for several amino acids, including 2-aminobutyric acid,
alanine and d-threonine [22,23,30]. During the kinetic resolution
reaction, TD should be exposed to both enantiomers of threonine
much above a physiological concentration range. Therefore, we
examined how the TD activity responded to high concentrations
of l- and d-threonine. To this end, we measured TD activities of

230
S.-W. Han, J.-S. Shin / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 227–232
Fig. 3. Substrate inhibitions of TD by (A) l-threonine and (B) d-threonine. Specific reaction rate of whole cell (closed circle) and crude extract (open circle) represents initial
rate normalized by cell or protein concentration. Relative activity was measured at varying concentration of d-threonine under a fixed l-threonine concentration (50 mM).
both whole cells and cell-free extract at varying concentrations of
each threonine enantiomer.
activities at 50 mM d-threonine in the reactions with whole cells
and cell-free extract, respectively). As observed with the substrate
inhibition by l-threonine, the whole cells showed slightly lower
inhibition by d-threonine than the cell-free extract did. To deter-
mine KIapp of TD for d-threonine, we performed linear regression of
the inverse of the relative activity of the cell-free extract against the
concentration of d-threonine (r2 = 0.99). The KIapp value was calcu-
lated to be 41 4 mM, which is in a reasonable agreement with the
average dissociation constant (=19.8 mM) reported elsewhere [30].
Because of the enzyme inhibitions by both enantiomers of thre-
onine, use of high concentrations of dl-threonine would not be
desirable for an efficient kinetic resolution process. We examined
how the mixed enzyme inhibitions affected the reaction rate by
measuring the TD activity of whole cells at varying concentration
of racemic threonine (Fig. 4). As expected, whole cells exhibited
strong substrate inhibition by dl-threonine. The decrease in the
reaction rate begun even at 50 mM dl-threonine and the residual
activity at 500 mM was 58% relative to the maximal reaction rate.
We found that the TD activities in both enzyme formulations
showed mild substrate inhibitions by l-threonine higher than
100 and 150 mM for cell-free extract and whole cells, respec-
tively (Fig. 3A). The whole cells displayed slightly lower substrate
inhibition than the cell-free extract did, presumably because cel-
lular membrane acts as a diffusion barrier to l-threonine and
the resulting intracellular concentration would be lower than the
extracellular one. The inverse of the specific reaction rate measured
with cell-free extract was subjected to a linear regression against
the concentration of l-threonine (r2 = 0.98), leading to the apparent
inhibition constant (KIapp) = 950 210 mM. The substrate inhibition
by l-threonine has not yet been reported because all the kinetic
measurements reported to date were done at less than 100 mM l-
threonine [22,31,32]. To the best of our knowledge, this is the first
report on the substrate inhibition of TD.
In addition to the substrate inhibition by l-threonine, TD dis-
played strong inhibition by d-threonine although the d-enantiomer
is a catalytically inert substrate analog (Fig. 3B). d-Threonine is
known to bind tightly to the active site of TD in a competitive man-
ner [30]. Indeed, the nonproductive binding was strong enough to
inhibit deamination of l-threonine (e.g., 58 2 and 48 2% residual
3.3. Small-scale kinetic resolution
Based on the results in Fig. 4, initial concentration of dl-
threonine around 50 mM would be optimal to exploit high reaction

S.-W. Han, J.-S. Shin / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 227–232
231
Fig. 4. Substrate inhibition of TD by racemic threonine. Specific reaction rate rep-
resents initial rate normalized by cell concentration.
rates. However, this concentration level is far below an indus-
trial requirement when the volumetric productivity is taken into
account. As trial runs before a preparative-scale reaction, we
compared reaction progresses of small-scale kinetic resolutions
(reaction volume = 5 mL) using different substrate feeding meth-
ods, i.e., batch vs. fed-batch reactions (Fig. 5A). Total amount of
substrate doses was set to be identical (i.e., 300 mg dl-threonine)
in both reactions where the initial substrate concentrations were
0.5 and 0.15 M for the batch and fed-batch reactions, respectively.
The fed-batch reaction vial was additionally supplemented with
dl-threonine twice during the reaction (i.e., 0.15 and then 0.2 M)
when the ee value exceeded 90%. The rate of increase in the ee
value of dl-threonine was slowed down as the substrate doses
were added to the fed-batch reaction vial. This result is ascribable
to the strong enzyme inhibition by d-threonine, as concentration
of the inhibitory d-threonine overwhelmed that of the reactive
l-enantiomer along with the supplementation of the racemic
substrate. As expected, the batch reaction led to a lower ee value at
the end of the reaction compared to the fed-batch reaction.
To examine how the substrate concentrations affected the reac-
tion progress in a batch reaction mode, we performed small-scale
kinetic resolutions (reaction volume = 1 mL) at 0.5, 1 and 1.5 M dl-
threonine (Fig. 5B). Because the batch reaction shown in Fig. 5A
ended up with only 79% ee, we used a 5-fold higher cell concentra-
tion (8.5 mg dcw/mL). Because the solubility of threonine is 0.74 M
[28], solid dl-threonine was present at the beginning of the reaction
in the reaction mixture for 1 and 1.5 M substrate concentrations.
The ee value of d-threonine in the reaction starting with 0.5 M dl-
threonine was 99.3% at 3 h and reached >99.9% at 6 h with 50%
conversion. However, the reactions starting with 1 and 1.5 M dl-
threonine showed 95.1 and 61.7% ee values at 11 h with 48.6 and
36.9% conversions, respectively. Prolonged incubation of the reac-
tion mixtures until 24 h ended up with modest increases in the ee
values (i.e., 96.2 and 63.0% ee values for the reactions using 1 and
1.5 M dl-threonine, respectively). Besides the stronger inhibition at
the higher substrate concentration, the disappointingly slow reac-
tion progresses after 11 h seemed to result from progressive cell
aggregation and lysis that started to be observed at 6 h. In contrast,
no cell aggregates were observed throughout the reaction using
0.5 M dl-threonine. It is presumable that the apparent cell lysis at
≥1 M dl-threonine is ascribed to a substantial damage of cellular
membrane by solid threonine under the mixing conditions and/or
too high ionic strength detrimental to cellular integrity.
Fig. 5. Small-scale kinetic resolution of dl-threonine. (A) Effect of substrate feed-
ing methods on the reaction progress. The batch reaction (closed circle) and the
fed-batch reaction (closed triangle) were started with 0.5 and 0.15 M dl-threonine,
respectively. Additional substrate feeding was carried out during the fed-batch reac-
tion, which was indicated by arrows. Reaction conditions: TD (1.7 mg dcw/mL), total
dl-threonine added (0.5 M) and phosphate buffer (50 mM, pH 7) in 5 mL reaction
mixture. (B) Kinetic resolutions of different concentrations of dl-threonine. Reac-
tion conditions: TD (8.5 mg dcw/mL), dl-threonine (0.5, 1 and 1.5 M represented by
circle, triangle and square symbols, respectively) and phosphate buffer (50 mM, pH
7) in 1 mL reaction mixture.
3.4. Preparative-scale kinetic resolution
To demonstrate practical applicability of the kinetic resolu-
tion, we performed a preparative-scale reaction in a batch reaction
mode. The apparent cell lysis at ≥1 M dl-threonine led us to carry
out a preparative-scale kinetic resolution starting with 0.5 M dl-
threonine. The reaction mixture (total 50 mL) was charged with 3 g
dl-threonine, 3400 U cell suspension (430 mg dcw) and phosphate
buffer (final concentration = 50 mM, pH 7), and then incubated at
37 ◦C under magnetic stirring. The reaction was completed within
5 h, leading to >99% ee of d-threonine at 50% conversion.
To isolate the resulting enantiopure d-threonine, the reaction
mixture was subjected to cell removal by centrifugation, pH adjust-
ment to 1.0 and then removal of the protein precipitate using a
glass-fritted filter funnel. The filtrate solution was loaded on a
cation-exchange column packed with the Dowex 50WX8 resin.
Elution of the bound d-threonine by 10% ammonia solution fol-
lowed by evaporation of the elution fraction led to white solid of
d-threonine (1.36 g, >99% ee, 90.7% isolation yield).

232
S.-W. Han, J.-S. Shin / Journal of Molecular Catalysis B: Enzymatic 122 (2015) 227–232
problem at ≥1 M dl-threonine should be addressed. Besides, a pro-
cess engineering strategy such as a fed-batch operation should be
also fashioned to attenuate the enzyme inhibition at high substrate
concentrations. Finally, it is notable that the process economics of
the TD-catalyzed kinetic resolution can benefit by the subsequent
asymmetric conversion of the unwanted side product.
Acknowledgements
This work was supported by the Advanced Biomass R&D Center
(ABC-2011-0031358) through the National Research Foundation
of Korea and the R&D grant (S2173394) funded by the Small and
Medium Business Administration of Korea.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.molcatb.2015.09.
011.
Fig. 6. Asymmetric synthesis of homoalanine from 2-oxobutyrate obtained as a side
product of the TD reaction. Circle and tringle symbols represent reaction yields of
the synthesis of homoalanine by OATA and ARTA, respectively. Reaction conditions:
2-oxobutyrate (50 mM), isopropylamine (100 mM), ␻-TA (5.9 and 11.2 U/mL for
OATA and ARTA, respectively), and phosphate buffer (50 mM, pH 7) in 1 mL reaction
mixture at 37 ◦C.
References
[1] J.S. Ma, Chim. Oggi 21 (2003) 65–68.
[2] M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Keßeler, R. Stürmer, et al.,
Angew. Chem. Int. Ed. 43 (2004) 788–824.
[3] A.S. Bommarius, M. Schwarm, K. Drauz, Chimia 55 (2001) 50–59.
[4] S. Servi, D. Tessaro, G. Pedrocchi-Fantoni, Coord. Chem. Rev. 252 (2008)
715–726.
3.5. Asymmetric synthesis of homoalanine from a side product of
kinetic resolution
[5] A. Lipkowski (2014) U.S. Patent 303095A1.
[6] A.W. Lipkowski, D. Carr, A. Misicka-Kesik, P. Kosson, A. Klinowiecka (2010)
U.S. Patent 273715A1.
[7] W. Leuchtenberger, K. Huthmacher, K. Drauz, Appl. Microbiol. Biotechnol. 69
(2005) 1–8.
[8] M.T. Kidd, B.J. Kerr, J. Appl. Poult. Res. 5 (1996) 358–367.
[9] R.F.W. Jackson, N.J. Palmer, M.J. Wythes, Tetrahedron Lett. 35 (1994)
7433–7434.
[10] J.P. Genet, C. Pinel, S. Mallart, S. Juge, S. Thorimbert, J.A. Laffitte, Tetrahedron:
Asymmetry 2 (1991) 555–567.
[11] J.P. Genet, Pure Appl. Chem. 68 (1996) 593–596.
[12] H. Adkins, W. Reeve, J. Am. Chem. Soc. 60 (1938) 1328–1331.
[13] J. Attenburrow, D.F. Elliott, G.F. Penny, J. Chem. Soc. (1948) 310–318.
[14] K. Pfister, E.E. Howe, C.A. Robinson, A.C. Shabica, E.W. Pietrusza, M. Tishler, J.
Am. Chem. Soc. 71 (1949) 1096–1100.
[15] D. Polenske, H. Lorenz, A. Seidel-Morgenstern, Chirality 21 (2009) 728–737.
[16] V.M. Profir, M. Matsuoka, Colloids Surf. A. Physicochem. Eng. Asp. 164 (2000)
315–324.
[17] A. Ito, M. Matsuoka, Cryst. Growth Des. 8 (2008) 2716–2720.
[18] T. van Herk, A.F. Hartoga, H.J. Ruijssenaars, R. Kerkman, H.E. Schoemaker, R.
Wever, Adv. Synth. Catal. 349 (2007) 1349–1352.
[19] J.Y. Houng, C.L. Hsieh (1996) U.S. Patent 5552318.
[20] T. Kimura, V.P. Vassilev, G.J. Shen, C.H. Wong, J. Am. Chem. Soc. 119 (1997)
11734–11742.
[21] J. Steinreiber, K. Fesko, C. Mayer, C. Reisinger, M. Schürmann, H. Griengl,
Tetrahedron 63 (2007) 8088–8093.
[22] D.T. Gallagher, D. Chinchilla, H. Lau, E. Eisenstein, Method. Enzymol. (2004)
85–106.
[23] D.T. Gallagher, G.L. Gilliland, G. Xiao, J. Zondlo, K.E. Fisher, D. Chinchilla, et al.,
Structure 6 (1998) 465–475.
[24] E.S. Park, M. Kim, J.S. Shin, Adv. Synth. Catal. 352 (2010) 3391–3398.
[25] E.S. Park, M. Kim, J.S. Shin, Appl. Microbiol. Biotechnol. 93 (2012) 2425–2435.
[26] E.S. Park, M.S. Malik, J.Y. Dong, J.S. Shin, ChemCatChem 5 (2013) 1734–1738.
[27] E.S. Park, J.S. Shin, Adv. Synth. Catal. 356 (2014) 3505–3509.
[28] L.A. Ferreira, S.P. Pinho, E.A. Macedo, Fluid Phase Equilib. 270 (2008) 1–9.
[29] R. Bhushan, H. Bruckner, Amino Acids 27 (2004) 231–247.
[30] E. Eisenstein, Arch. Biochem. Biophys. 316 (1995) 311–318.
[31] D.H. Calhoun, R.A. Rimerman, G.W. Hatfield, J. Biol. Chem. 248 (1973)
3511–3516.
[32] E. Eisenstein, J. Biol. Chem. 266 (1991) 5801–5807.
[33] E.S. Park, J.Y. Dong, J.S. Shin, Org. Biomol. Chem. 11 (2013) 6929–6933.
[34] M. Sasa, J. Pharmacol. Sci. 100 (2006) 487–494.
[35] W.A. Nugent, J.E. Feaster, Synth. Commun. 28 (1998) 1617–1623.
The pass-through solution from the cation-exchange column
contained a valuable side product, i.e., 2-oxobutyrate, which can be
converted to more commercially valuable unnatural amino acids,
i.e., l- and d-homoalanine, by S- and R-selective ␻-TA, respectively,
as we demonstrated previously [33]. For example, l-homoalanine is
a key intermediate for levetiracetam and brivaracetam (antiepilep-
tic drugs) [34], and ethambutol (an antituberculosis drug) [35].
Simultaneous reactions by TD and ␻-TA in one pot supplemented
with dl-threonine and isopropylamine (i.e., an amino donor for
the ␻-TA reaction) would lead to coproduction of enantiopure
homoalanine along with d-threonine. However, the one-pot reac-
tion renders isolation of both amino acids very difficult. Therefore,
we decided to go with a stepwise approach rather than the one-pot
reaction.
The pass-through solution was used as a stock solution of 2-
oxobutyrate after pH adjustment to 7.0 without any purification.
As a proof-of-concept, we performed small-scale asymmetric con-
versions of 50 mM 2-oxobutyrate into l- and d-homoalanine using
S-selective ␻-TA from O. anthropi (OATA) and R-selective ␻-TA
from Arthrobacter sp. (ARTA), respectively (Fig. 6). The reaction per-
formed with OATA showed a 93.2% reaction yield of l-homoalanine
with >99% ee at 9 h. Likewise, the reaction with ARTA led to enan-
tiopure d-homoalanine (i.e., >99% ee) with 90.9% reaction yield at
12 h.
4. Conclusions
In this study, we demonstrated preparation of optically pure
d-threonine as well as both enantiomers of homoalanine by
TD-catalyzed kinetic resolution of dl-threonine followed by ␻-TA-
catalyzed asymmetric amination of 2-oxobutyrate. Because of the
lack of a cost-effective method for preparation of enantiopure d-
threonine, the TD-based method developed in this study might be
applicable for an industrial purpose. To this end, the cell stability