In Vitro and in Vivo One-Pot Deracemization of Chiral Amines by Reaction Pathway Control of Enantiocomplementary ω-Transaminases

Article abstract of DOI:10.1021/acscatal.9b01546

Biocatalytic cascade conversion of racemic amines into optically pure ones using enantiocomplementary ω-transaminases (ω-TAs) has been developed by thermodynamic and kinetic control of reaction pathways where 12 competing reactions occur with pyruvate and isopropylamine used as cosubstrates. Thermodynamic control was achieved under reduced pressure for selective removal of a coproduct (i.e., acetone), leading to elimination of six undesirable reactions. Engineered orthogonality in substrate specificities of ω-TAs was exploited for kinetic control, enabling suppression of four additional reactions. Taken together, the net reaction pathway could be directed to two desired reactions (i.e., oxidative deamination of R-amine and reductive amination of the resulting ketone into antipode S-amine). This strategy afforded one-pot deracemization of various chiral amines with >99% ee_S and 85-99% reaction yields of the resulting S-amine products. The in vitro cascade reaction could be successfully implemented in a live microbe using glucose or l-threonine as a cheap amino acceptor precursor, demonstrating a synthetic metabolic pathway enabling deracemization of chiral amines which has never been observed in living organisms.

Full text of DOI:10.1021/acscatal.9b01546

Subscriber access provided by UNIV AUTONOMA DE COAHUILA UADEC
Article
In vitro and in vivo One-pot Deracemization of Chiral Amines by
Reaction Pathway Control of Enantiocomplementary #-Transaminases
Sang-Woo Han, YOUNGHO JANG, and Jong-Shik Shin
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.9b01546 • Publication Date (Web): 09 May 2019
Downloaded from http://pubs.acs.org on May 10, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
In vitro and in vivo One-pot Deracemization of
Chiral Amines by Reaction Pathway Control of
Enantiocomplementary -Transaminases
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
†
†
,†
Sang-Woo Han, Youngho Jang and Jong-Shik Shin*
†
Department of Biotechnology, Yonsei University, Yonsei-Ro 50, Seodaemun-Gu, Seoul 03722,
South Korea
1
ACS Paragon Plus Environment

ACS Catalysis
Page 2 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
Biocatalytic cascade conversion of racemic amines into optically pure ones using
enantiocomplementary -transaminases (-TAs) has been developed by thermodynamic and
kinetic control of reaction pathways where twelve competing reactions occur with pyruvate and
isopropylamine used as cosubstrates. Thermodynamic control was achieved under reduced
pressure for selective removal of a coproduct (i.e. acetone), leading to elimination of six
undesirable reactions. Engineered orthogonality in substrate specificities of -TAs was exploited
for kinetic control, enabling suppression of four additional reactions. Taken together, the net
reaction pathway could be directed to two desired reactions, i.e. oxidative deamination of R-amine
and reductive amination of the resulting ketone into antipode S-amine. This strategy afforded one-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
pot deracemization of various chiral amines with > 99 % ee and 85 - 99 % reaction yields of the
S
resulting S-amine products. The in vitro cascade reaction could be successfully implemented in a
live microbe using glucose or L-threonine as a cheap amino acceptor precursor, demonstrating a
synthetic metabolic pathway enabling deracemization of chiral amines which has never been
observed in living organisms.
KEYWORDS: deracemization, chiral amine, transaminase, protein engineering, metabolic
engineering
2
ACS Paragon Plus Environment

Page 3 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
INTRODUCTION
In vivo biochemical networks consist of a vast number of cascade reactions where reaction
intermediates are processed by specific enzymes dedicated for each reaction step.1-2 Coordination
of the complex reaction pathways into a well-defined hierarchy is elegantly achieved by an
exquisite combination of thermodynamic and kinetic control that keep the reaction fluxes under
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
-4
precise regulation. Thermodynamically unfavorable reactions are coupled with energetic drivers
such as ATP hydrolysis and NADH oxidation, so energy-consuming pathways are driven forward.5
Undesirable exergonic reactions are kinetically suppressed by exploiting orthogonal substrate
specificities of relevant enzymes with a minimal overlap that allows promiscuous utilization of
6
common intermediates. Construction of a synthetic metabolic pathway by mimicking the
biological design principles is promising for coming up with novel bioprocesses that living
organisms have not yet pursued.7-10
Although racemization is a naturally occurring reaction in living organisms to supply a minor
enantiomer from a major antipode reservoir,11-12 deracemization has never been observed in native
metabolism presumably owing to the lack of biological demand for this functionality. However,
deracemization is conceptually and commercially interesting in the chemical industry because
preparation of racemic mixtures is usually readily available relying on conventional organic
synthesis.13 In the case of chiral amines that are essential building blocks of a number of
1
4
pharmaceutical drugs, deracemization remains a huge challenge for chemocatalytic approaches
and instead employs biocatalytic methods which mostly rely on monoamine oxidase.15-18 In the
monoamine oxidase-based deracemization, racemic amines are stereoselectively oxidized by
monoamine oxidase to imines which are reduced back to the amines without stereoselectivity by
ammonia borane,13, 15, 18 palladium formate/hydrogen or artificial metalloenzyme. It is notable
16
17
3
ACS Paragon Plus Environment

ACS Catalysis
Page 4 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
that complete deracemization requires oxidation of actually two equivalents of an unwanted
enantiomer owing to the non-stereoselective reduction of the resulting imine. Recently,
stereoselective reduction of the imine intermediate has been demonstrated using imine reductase
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
19
for deracemization of 2-substituted piperidines and pyrrolidines. More recently, Aleku et al.
reported a new deracemization approach where monoamine oxidase was substituted by reductive
aminase.20
Besides the monoamine oxidase-based approaches, deracemization of chiral amines was
demonstrated using two enantiocomplementary -transaminases (-TAs) in a stepwise fashion,
i.e. serial processing of the two -TA reactions in the presence of auxiliary enzymes for
equilibrium shift.21-23 More recently, native orthogonality in substrate specificity afforded one-pot
deracemization where both -TA reactions occurred simultaneously.24 This method was
contingent on a non-canonical activity of an -TA from Polaromonas sp. for -ketoglutarate. Note
that -TAs known to date adopt pyruvate as a universal amino acceptor.25-26 Therefore,
applicability of this approach is limited because the enzyme from Polaromonas sp. is the only -
TA, known to date, to be active for -ketoglutarate.24 Moreover, both cosubstrates (i.e., -
ketoglutarate and D-alanine) are expensive and additional enzymes should be supplied to
overcome unfavorable equilibrium in the transamination between ketone and D-alanine. Setting
up orthogonal substrate specificities of the two enzymes has been also demonstrated by coupling
2
7
-TA and enantiocomplementary amine dehydrogenase. However, to allow industrial feasibility,
one should come up with a generally applicable strategy that employs cheap cosubstrates and facile
equilibrium shift without relying on auxiliary enzymes and external cofactors.
In this study, we aimed at developing such a -TA-based protocol for deracemization of various
chiral amines (1a-f) using typical cosubstrates such as pyruvate (3) and isopropylamine (5)
4
ACS Paragon Plus Environment

Page 5 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
Scheme 1). Note that 5 is one of the preferred amino donors for amination of ketones owing to a
low price and easy removal of its deamination product, i.e. acetone (6), under reduced pressure.28-29
As a proof-of-concept, we explored conversion of racemic amines into S-enantiomers via
stereoinversion of unwanted R-enantiomers using a R-selective -TA from Arthrobacter sp.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
0
31
(ARTA) and a S-selective one from Ochrobactrum anthropi (OATA). Because 3 and 5 are
common substrates for both ARTA and OATA, a number of competing reactions would occur in
the reaction mixture charged with racemic amine, 3 and 5 in the presence of the two -TAs.
Suppression of unwanted reactions was carried out using process and protein engineering
strategies, leading to reaction pathway control that enabled the competing reactions to be directed
to the two desired reactions consisting of an energetically down-hill amino group transfer between
R-1a-f and 3 and an up-hill one between 2a-f and 5. We demonstrated that the reaction pathway
control by combining thermodynamic and kinetic control afforded deracemization of chiral amines
both in vitro and in vivo.
EXPERIMENTAL SECTION
Materials
S-1e, R-1e, S-1f and R-1f were purchased from Alfa Aesar (Ward Hill, MA, USA). 5 and L-
threonine were obtained from Junsei Chemical Co. (Tokyo, Japan). Dimethyl sulfoxide (DMSO),
methanol, ethanol, acetonitrile, glucose and sodium acetate were obtained from Duksan Pure
Chemicals Co. (Ansan, South Korea). DL-4 was purchased from Samchun Pure Chemical Co.
(Seoul, South Korea). All other chemicals including racemic amines were purchased from Sigma-
Aldrich Co. (St. Louis, MO, USA). Materials used for preparation of culture media were obtained
from Difco (Spark, MD, USA).
5
ACS Paragon Plus Environment

ACS Catalysis
Page 6 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Site-directed Mutagenesis of AA
Construction of OATA carrying W58L and R417A substitutions (OATAW58L/R417A) was performed
using a QuikChange Lightning site-directed mutagenesis kit (Agilent Technologies Co.) according
to an instruction manual. Mutagenesis PCR was performed using pET28-OATAW58L, created
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
previously,32
as
a
template.
Forward
(5′-CGCGG
CGTCATTTCCGCGGCA
ATGGGCGATACG-3′) and reverse (5′-CGTATCGCCCATTGCCGCGGAAATGACG
CCGCG-3′) primers for the mutagenesis were designed by a primer design program
(https://www.genomics.agilent.com/ primerDesignProgram.jsp). The site-directed mutagenesis
was verified by DNA sequencing.
Preparation of Purified -TAs
Expression and purification of His -tagged -TAs were performed as described before with minor
6
modifications.33 Expression vectors used for the protein expression were listed in Table S1.
Escherichia coli BL21(DE3) cells were transformed with a pET28a(+) expression vector
harboring a -TA gene and the transformant cells were cultivated in 1 L of LB medium containing
5
0 g/mL kanamycin at 37 °C. Protein expression was induced by adding IPTG (final
concentration = 0.1 mM) at OD600  0.4. After further cultivation at 37 °C for 12 h, cells were spun
down by centrifugation (10,000×g, 10 min, 4 °C) and the resulting cell pellet was resuspended in
1
5 mL buffer (50 mM sodium chloride, 1 mM EDTA, 1 mM -mercaptoethanol, 0.1 mM PMSF
and 50 mM Tris, pH 8). Cells were disrupted by a sonicator, followed by centrifugation (10,000×g,
0 min, 4 °C) to remove cell debris. The resulting supernatant was used as cell-free extract for
protein purification.
6
6
ACS Paragon Plus Environment

Page 7 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
The cell-free extract was subjected to protein purification using a HisTrap HP column (GE
Healthcare) on an ÄKTAprime plus (GE Healthcare). The His -tagged -TA was eluted by a linear
6
gradient of imidazole (0.05-0.5 M imidazole, 0.15 M sodium chloride, 0.5 mM PLP and 20 mM
sodium phosphate, pH 7.4). Desalting was carried out on a HiTrap HP column (GE Healthcare)
using a buffer (0.15 M sodium chloride, 0.5 mM PLP and 50 mM sodium phosphate, pH 7). When
necessary, the enzyme solution was concentrated by an Amicon Ultra-15 30K (Millipore Co.).
Concentration of the purified -TA was determined by measuring UV absorbance at 280 nm with
a UV-1650PC spectrophotometer (Shimadzu Co.). Molar extinction coefficients of the
homodimeric -TAs were obtained by a protein extinction coefficient calculator (http://www.
biomol.net/en/tools/proteinextinction.htm) and are listed in Table S1.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Enzyme Activity Assay
All the enzyme assays were carried out at 37 °C and pH 7 (50 mM potassium phosphate buffer).
Initial reaction rates of -TAs for reactions using 1a/3, 5/2a and 5/3 as substrate pairs were
measured at < 20 % conversion by HPLC analyses of 2a, 1a and 4, respectively. Typical reaction
volume was 100 L and the reactions were stopped by adding 600 L of acetonitrile, 37.5 L of
16 % (v/v) HClO and 20 L of 5 N HCl for reactions using 1a/3, 5/2a and 5/3 as substrate pairs,
4
respectively.
Determination of Rate Constants for Mathematical Modeling
To enable mathematical modeling with a minimum set of kinetic parameters, we assumed that the
enzyme kinetics might be approximated to a second-order rate equation, i.e. first order with respect
to each substrate concentration, by ignoring variation in the distribution of enzyme species such
7
ACS Paragon Plus Environment

ACS Catalysis
Page 8 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
as E-PLP, E-PMP and Michaelis complexes. Based on the pseudo-second-order approximation,
reaction velocity (푣 ) with respect to reaction time can be expressed by a following equation
푡
푣푡 = 푘푎푝푝[A] [D]
(Eq. 1)
푡
푡
where 푘푎푝푝 represents an apparent rate constant at a given enzyme concentration ([E]). [A] and
푡
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
[
D] represent time-dependent concentrations of an amino acceptor (A) and a donor (D),
푡
respectively. Using Eq. 1, specific reaction rate (i.e., initial velocity per enzyme concentration; 푣 /[
푖
E]) can be expressed as follows.
푣푖/[E] = (푘푎푝푝/[E])[A] [D]
(Eq. 2)
푖
푖
Therefore, 푘푎푝푝/[E] values for ARTA, OATA and OATAW58L/R417A could be obtained by dividing
specific reaction rate with [A] and [D] . Table S2 lists the 푘푎푝푝/[E] values. In the case of
푖
푖
OATAW58L and OATAR417A, the 푘푎푝푝/[E] values were taken from those of OATA and
OATAW58L/R417A by assuming that the activity changes by each point mutation are completely
independent. Thus, it was assumed that the W58L substitution did not affect the reactions for 1a/3
and 5/3 substrate pairs but affected the reaction for 5/2a. In contrast, the R417A substitution was
assumed to affect the reactions for 1a/3, and 5/3 substrate pairs but not the one for 5/2a. The 푘푎푝푝/[
E] values assigned for OATAW58L and OATAR417A are listed in Table S3. Numerical solutions of
simultaneous differential equations for the competing reactions were obtained using the 푘푎푝푝/[E]
values by the software Mathematica (Wolfram Research Inc.).
Small-scale Deracemization of Chiral Amines
All the deracemization reactions were performed at 37 °C and 460 Torr. The reduced pressure was
achieved by a vacuum pump with a vacuum controller (EYELA Co.). Typical reaction volume
was 1 mL and the reaction conditions were 10 mM rac-1a-f, 6.5 mM 3, 100 mM 5, 10 M ARTA,
8
ACS Paragon Plus Environment

Page 9 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
1
00 M OATAW58L/R417A in 50 mM potassium phosphate buffer (pH 7) containing 15 % (v/v)
DMSO. At the predetermined reaction time, aliquots of the reaction mixture (typical 50 L) were
taken and mixed with 18.75 L of 16 % (v/v) HClO to stop the reaction. Amines were analyzed
4
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
by quantitative chiral HPLC, which were used for determination of reaction yield and enantiomeric
excess. When the reaction yield exceeded 90 %, 3 was added in the reaction mixture for chiral
polishing, and then running for 30 min (final concentration of 3 = 0.5 mM except 0.3 mM for rac-
1
d).
Preparative-scale Deracemization and Product Characterization of S-1d
Preparative-scale deracemization of rac-1d was carried out at 37 °C and 460 Torr in 100 mL
reaction mixture containing rac-1d (1.49 g, 10 mmol), 3 (0.88 g, 8 mmol), 5 (1.18 g, 20 mmol),
ARTA (10 M), OATAW58L/R417A (200 M), 15 % (v/v) DMSO and 50 mM Tris buffer (pH 7).
When the reaction yield exceeded 80 %, 3 was additionally added to the reaction mixture for chiral
polishing (final concentration = 1.5 mM) and the reaction was allowed to proceed for 1.5 h.
Product isolation was carried out using ion-exchange chromatography. For protein removal,
pH of the reaction mixture was adjusted to 1 by adding 5 N HCl and the resulting solution was
centrifuged (10,000×g, 30 min), followed by filtration through a glass-fritted filter funnel. The
filtrate solution was loaded on a glass column packed with Dowex 50WX8 cation-exchange resin
(40 g). Washing and elution were carried out by sequentially loading water (40 mL) and 10 %
(v/v) ammonia solution (800 mL). The eluates were collected and then S-1d was extracted three
times with n-hexane (3×300 mL). The resulting extractant pool was evaporated at 50 °C and 0.45
bar, resulting in clear liquid of S-1d.
9
ACS Paragon Plus Environment

ACS Catalysis
Page 10 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
13
1
The isolated S-1d was structurally characterized by H NMR, C NMR and LC/MS. H NMR
1
3
and C NMR spectra were recorded on an Avance II FT-NMR spectrometer (Bruker Co.) using
tetramethylsilane as a standard. Mass spectral data were obtained with a LTQ Orbitrap Mass
Spectrometer (electrospray ionization, positive ion mode, Thermo Fisher Scientific Inc.).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Construction of Expression Plasmids for in vivo Deracemization
Cloning was performed to construct three plasmids (i.e., pCDFDuet-ARTA, pCDFDuet-
ARTA/TD and pRSFDuet-OATAW58L/R417A) using pCDFDuet-1 and pRSFDuet-1 expression
plasmids (Novagen) as parental vectors. For construction of pCDFDuet-ARTA, the gene encoding
3
4
ARTA was amplified from pET28-ARTA, constructed in a previous study, using a forward (5′-
GATATACATATGGCATTCAGCGCCGATA-3′) and a reverse primer (5′-CTCGAGGGTACC
TCAATACTGAACCGGTGT-3′). The PCR product and the pCDFDuet-1 vector were digested
by NdeI/KpnI and then ligated using T4 DNA ligase after agarose gel purification. The pCDFDuet-
ARTA/TD plasmid was prepared by cloning the ilvA gene encoding threonine deaminase into
pCDFDuet-ARTA. Gene amplification was performed using a pET21 plasmid harboring ilvA,
3
5
created in a previous study, as a template with a forward (5′-ATAAGGAGATATACCATGGC
TGACTCGCAACCC-3′) and a reverse primer (5′-ATTATGCGGCCGCAAGCTTTTAACCCG
CCAAAAAGAA-3′). The PCR product was ligated with a linearized pCDFDuet-ARTA, digested
with NcoI/HindIII, using an In-Fusion HD Cloning Kit (Clontech Laboratories). For construction
of pRSFDuet-OATAW58L/R417A, the -ta gene encoding OATAW58L/R417A was amplified from
pET28-OATAW58L/R417, created in this study, using a forward (5′-GATATACATATGACTGCTC
AGCCAAAC-3′) and a reverse primer (5′-CTCGAGGGTACCTTACCTGGTGAGGCTTGC-3′).
The PCR product and a linearized pRSFDuet-1 vector, both cut by NdeI/KpnI, were ligated using
1
0
ACS Paragon Plus Environment

Page 11 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
T4 DNA ligase.
In vivo Deracemization Using Engineered E. coli
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Deracemization reactions were done at 37 °C and 460 Torr in 50 mM potassium phosphate buffer
(pH 7). Typical reaction volume was 1 mL and the reaction mixture was mixed by a magnetic
stirrer. For deracemization using glucose as an amino acceptor precursor, E. coli BL21(DE3) cells
harboring pCDFDuet-ARTA and pRSFDuet-OATAW58L/R417A were used. The reaction conditions
were 50 mM rac-1a, 200 mM 5, 7.5 - 50 mM glucose and 25.8 mg dcw/mL engineered E. coli
cells. The reaction was stopped at predetermined reaction times by adding aliquots (typically 10
L) of reaction mixture into 60 L of acetonitrile or 127.5 L of 4.7 % (v/v) HClO solution for
4
analyses of hydrophobic or hydrophilic compounds, respectively. After centrifugation (13,000 rpm,
1
0 min), the supernatant was subjected to HPLC analysis.
For deracemization using L-threonine as an amino acceptor precursor, E. coli BL21(DE3) cells
harboring pCDFDuet-ARTA/TD and pRSFDuet-OATAW58L/R417A were used. The reaction was
performed at 50 mM rac-1a, 500 mM 5 and 27.5 mM L-threonine using 17.2 mg dcw/mL
engineered E. coli cells. The procedures for HPLC sampling and analysis were the same as those
for the in vivo deracemization using glucose.
HPLC Analysis
HPLC analyses were performed using an Alliance system (Waters Co.), a 1260 Infinity II LC
System (Agilent Technologies Inc.) or a YL9300 HPLC system (YL Instrument Co., South Korea).
Chiral analyses of 1a-d were carried out using a Crownpak CR(-) or a Crownpak CR-I(-) column
(Daicel Co., Japan) under detection by an UV detector tuned at 200 nm. For quantitative chiral
1
1
ACS Paragon Plus Environment

ACS Catalysis
Page 12 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
analyses of 1e, 1f and 4, chiral derivatization with 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide
3
6
(
Marfey’s reagent) was employed as described elsewhere. The reaction sample was analyzed
using a Symmetry C18 column (Waters Co.) at 340 nm. Detailed HPLC analysis conditions were
described in Table S4.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Analysis of 2a was carried out using the Symmetry C18 column with isocratic elution of 60/40
%
(v/v) MeOH/water eluent, both containing 0.1 % (v/v) trifluoroacetic acid, at 0.6 mL/min by an
UV detector tuned at 254 nm. Retention time of 2a was 6.5 min.
Key metabolites in the in vivo deracemization were analyzed using an Aminex HPX-87H
column (Bio-Rad Inc.). Elution was carried out with 5 mM H SO at 0.4 mL/min. Organic acids
2
4
were detected by an UV detector (210 nm). For detection of glucose and ethanol, a RI detector was
used under the temperature control at 35 °C. Retention times were 13.3 (glucose), 13.7 (3), 19.1
(lactate), 22.7 (acetate) and 30.3 min (ethanol).
RESULTS AND DISCUSSION
Identification of the Competing Reactions
To identify undesirable reactions to be suppressed, we carried out reaction pathway analysis of the
competing reactions. To this end, -methylbenzylamine (1a) was chosen as a model amine owing
3
4
to availability of equilibrium constant (K ) data. Note that - TA harbors pyridoxal 5-phosphate
eq
2
5
(PLP) as a prosthetic group. It shuttles between two enzyme forms, i.e. a PLP-form (E-PLP) and
a pyridoxamine 5-phosphate form (E-PMP), to mediate two half reactions consisting of oxidative
deamination of an amino donor and reductive amination of an acceptor (Scheme 2). Hence, five
half reactions should be considered in the reaction pathway analysis (Table 1). Owing to the
1
2
ACS Paragon Plus Environment

Page 13 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
R
D
S
L
opposite stereoselectivity of the -TAs, I and III are catalyzed by ARTA and so are I and III
by OATA. Note that absolute configurations of D and L-alanine (4) are R and S, respectively.
Because 5 is achiral, reaction II is carried out by both -TAs.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Combining the two half reactions listed in Table 1, one in a forward and the other in a reverse
direction, cancels out the enzyme terms and leads to construction of a whole catalytic cycle. For
R
example, an ARTA reaction between R-1a and 3 can be represented by adding I to the reverse of
D
R
D
III . We denote the resulting reaction as I /III . Likewise, an OATA reaction between 5 and
S
acetophenone (2a) can be denoted as II/I . Consideration of stereoselectivity and reversibility of
the -TA reactions leads to twelve possible reactions that occur in the reaction mixture (Table 2).
R
D
S
Among the twelve reactions, I /III and II/I are the ones constituting the desired deracemization
pathway in Scheme 1A.
Thermodynamic Control of the Competing Reactions
Successful deracemization depends on how effectively the unwanted reactions are suppressed
while the two desired reaction are unaffected or reinforced. The twelve competing reactions are
listed in descending order of equilibrium constants (Table 3). Note that the two reactions in the
same row are thermodynamically equivalent because they differ only in chirality. As a first step
for the reaction pathway control, we carried out thermodynamic analysis of the twelve reactions.
Because reactions I/III, II/III and I/II are exergonic (Greaction < 0), suppression of these reactions
is not thermodynamically feasible. In the case of endergonic reactions whose K values are < 1,
eq
maximum conversions when starting with the same molar concentrations of substrates were
calculated to be 20, 11 and 3 % for reactions II/I, III/II and III/I, respectively. Therefore, the
intrinsic thermodynamic constraint does not ensure sufficient suppression of II/I and III/II while
1
3
ACS Paragon Plus Environment

ACS Catalysis
Page 14 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the most unfavorable reaction III/I could be regarded negligible.
As a second step, we explored an artificial thermodynamic constraint which could be set up
under certain process conditions. We posited that removal of acetone (6) under reduced pressure
would suppress reactions I/II and III/II where 6 is one of the two substrates. Note that selective
removal of 6 is implementable without undesirable coevaporation of 2a, as demonstrated before,28,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2
o
because of a big difference in the boiling point (i.e., 56 vs 202 C at 1 atm). Therefore,
equilibrium positions of reactions I/III and III/I do not change under the reduced pressure
conditions while those of II/III and II/I are completely displaced to the product side. It is notable
S
that the energetically unfavorable II/I , i.e. one of the two desired reactions, could be driven to
completion owing to the equilibrium shift.
Kinetic Control of the Competing Reactions
As aforementioned, the thermodynamic filtering of undesirable reactions would allow suppression
D
R
L
S
R
S
D
L
of III /I and III /I by the intrinsic equilibrium and I /II, I /II, III /II and III /II by the process
S
L
D
L
R
engineering (Figure 1). However, four reactions (i.e., I /III , II/III , II/III and II/I ) still remain
to be suppressed. This is a challenging task to be tackled by thermodynamic control because the
two desired reactions should be allowed to go to completion while their energetic equivalents, i.e.
S
L
R
I /III and II/I , are to be suppressed. We reasoned that this might be accomplished by kinetic
control based on protein engineering that could provide desirable orthogonality in substrate
utilization of the two -TAs. For the kinetic control, we paid attention to the half reaction IIIL
S
L
L
because this is a part of the two undesired reactions, i.e. I /III and II/III . We postulated that
suppression of these two reactions could be attained if OATA lost its native activity for 3. To this
end, R417A substitution was decided to be introduced because R417 is an active site arginine
1
4
ACS Paragon Plus Environment

Page 15 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
36
responsible for recognition of a carboxylate group of an incoming substrate. We posited that the
resulting OATA mutant (OATAR417A) would lack the catalytic competence for amination of 3 as
well as deamination of L-4.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
D
R
For suppression of the remaining unwanted reactions II/III and II/I , we noticed that an activity
3
7
of ARTA for 5 was much lower than that for R-1a (i.e., 2 % relative activity). Note that the
3
7
D
activity of OATA for 5 is 43 % relative to that for S-1a. However, this does not ensure that II/III
R
and II/I would occur at negligible reaction rates during the cascade reaction. Hence, we devised
D
R
D
additional measures to render II/III and II/I kinetically insignificant. In the case of II/III , we
expected that precise control over a feeding amount of 3 would prevent unnecessary reaction
progress. Assuming the lack of enzyme activity of OATAR417A for 3, ARTA would consume most
R
D
D
37
R
D
of 3 via either I /III or II/III . Because of the 2 % reactivity of 5 relative to R-1a, I /III would
D
proceed 50-fold faster than II/III . Therefore, addition of 3 to the reaction mixture slightly over a
D
R
D
molar equivalent of R-1a would render II/III terminated upon depletion of R-1a via I /III
R
because 3 would not be available any more. In the case of II/I , equilibrium is shifted to the product
side under reduced pressure and thereby failure to effectively suppress the reaction leads to
R
significant chiral impurity of the resulting S-1a product. Therefore, the reaction rate of II/I should
S
32
be much lower than that of II/I . Based on the previous results, we reasoned that the requisite of
S
R
II/I much faster than II/I could be fulfilled by protein engineering of OATA to improve activities
for 5 and 2a. To this end, we decided to introduce an additional mutation to OATAR417A, i.e. W58L
that was found to elicit remarkable activity improvements for various ketones (e.g., 340-fold
32
increase in k /K for 2a) as well as 5 (i.e., 9-fold increase in k /K ) in a previous study.
cat
M
cat
M
Protein Engineering of OATA
1
5
ACS Paragon Plus Environment

ACS Catalysis
Page 16 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
R
D
To implement the proposed kinetic filtering as shown in Figure 1, I /III should overwhelm the
S
L
L
D
S
three reactions competing for cosubstrate 3, i.e. I /III , II/III and II/III . Likewise, II/I should be
R
much faster than II/I to ensure high enantiopurity of the S-1a product. To verify feasibility of the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
kinetic control design via protein engineering of OATA, we prepared and purified the
OATAW58L/R417A variant (Figure S1). Reaction rates of OATAW58L/R417A for I/III, II/I and II/III were
measured and compared with those of ARTA (Table 4). Indeed, the reaction rate of ARTA for
R
D
D
I /III was much higher than that of ARTA for II/III as well as those of OATA
for
W58L/R417A
S
L
L
S
I /III and II/III . Besides, the reaction rate of OATA
for II/I was 117-fold higher than
W58L/R417A
R
that of ARTA for II/I . Taken together, OATA
turned out to exhibit desirable kinetic
W58L/R417A
properties to fulfill the kinetic filtering conditions.
In the protein engineering of OATA, we posited that the activity changes, each resulting from
R417A and W58L, would not interfere with each other because the two residues belong to different
subdomains and are not directly adjacent (Figure S2). Indeed, an activity improvement of
S
OATAW58L/R417A for II/I , compared with the parental enzyme, was 600-fold (Table 4) which is
even higher than a 165-fold activity increase caused by a single W58L mutation observed in a
previous study.32 In addition, the R417A substitution turned out to abolish an activity of
S
L
OATAW58L for 3. Activity of OATAW58L/R417A for I /III was only 0.16 % compared with that of
OATAW58L (Table S5). Owing to the activity loss for 3, OATAW58L/R417A showed only 0.4 %
L
activity for II/III compared with the wild-type enzyme (Table 4). In contrast, the enzyme activity
was not affected by R417A when the activity assay was carried out with an amino acceptor, devoid
of a carboxyl group, such as propanal (Table S6). It is notable that incorporation of both intended
S
S
L
L
properties into OATA (i.e., reinforcing II/I and suppressing I /III and II/III ) led to a complete
S
L
change in the amino acceptor bias from 3 to 2a. The II/I -to-II/III reaction rate ratio was
1
6
ACS Paragon Plus Environment

Page 17 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
-
4
dramatically increased from 4.9  10 for the parental enzyme to 72 for OATA
(i.e., 1.5
W58L/R417A
5
 10 -fold increase), in line with experimental results showing actual partitioning of the amino
group of 5 in the presence of both 2a and 3 (Table S7).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Computational Modeling of the Cascade Reaction

-TA is a bisubstrate enzyme following a ping-pong bi-bi mechanism, leading to complicated
3
8
formulation of enzyme kinetics. Therefore, reaction progresses were obtained under pseudo-
second-order approximations of enzyme kinetics. In the mathematical modeling of the
deracemization cascade, six reactions among twelve in total were discarded owing to the
R
D
S
L
D
L
R
thermodynamic filtering. The remaining six reactions, i.e. I /III , I /III , II/III , II/III , II/I and
S
II/I , were considered for the mathematical formulation of deracemization. We assumed that
removal of 6 under reduced pressure would be achieved efficiently enough to approximate that
concentration of 6 could be negligible throughout the reaction (i.e., [6]  0). Using Eq. 1, mass
푡
balance equations for the five chemical species were derived as follows.
d[R - 1a]푡
=
―푘
IR/IIID
[R - 1a] [ퟑ] + 푘 [ퟓ] [ퟐ퐚]
푡
II/IR
(Eq. 3)
푡
푡
푡
dt
d[S - 1a]푡
dt
= ―푘
IS/IIIL
[S - 1a] [ퟑ] + 푘 [ퟓ] [ퟐ퐚]
푡
II/IS
푡
푡
푡
(Eq. 4)
(Eq. 5)
d[ퟑ]푡
=
―푘
IR/IIID
[R - 1a] [ퟑ] ―푘
IS/IIIL
[S - 1a] [ퟑ] ―푘
II/IIID
[ퟓ] [ퟑ] ―푘
II/IIIL
푡 푡
[ퟓ] [ퟑ]
푡
푡
푡
푡
푡
푡
dt
d[ퟓ]푡
=
=
―푘II/I
R
[ퟓ] [ퟐ퐚] ― 푘 [ퟓ] [ퟐ퐚] ―푘
II/IS
II/IIID
[ퟓ] [ퟑ] ―푘
II/IIIL
[ퟓ] [ퟑ]
푡
(Eq. 6)
(Eq. 7)
푡
푡
푡
푡
푡
푡
푡
dt
d[ퟐ퐚]푡
―푘II/I
R
[ퟓ] [ퟐ퐚] ― 푘 [ퟓ] [ퟐ퐚] +푘
II/IS
IR/IIID
[R - 1a] [ퟑ] +푘
IS/IIIL
[S - 1a] [ퟑ]
푡
푡
푡
푡
푡
푡
푡
푡
dt
Computational simulations of the reaction progress were carried out using numerical solutions of
the five simultaneous differential equations (Eq. 3 - 7).
1
7
ACS Paragon Plus Environment

ACS Catalysis
Page 18 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
To predict whether the reaction pathway control permits intended deracemization of rac-1a into
S-1a, we carried out computational modeling of the cascade reaction starting with 10 mM rac-1a,
5.5 mM 3 and 100 mM 5 using four different enzyme pairs (Figure 2). Simulation results predicted
that use of wild-type -TAs would cause decreases in both enantiomers of 1a (Figure 2A),
consistent with experimental results (Figure S3). Neither of the single-point mutations in OATA
was predicted to allow the cascade reaction to result in the intended deracemization, i.e. reaction
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
yield > 90 % (i.e., [S-1a] > 9 mM) and ee > 95 % (Figure 2B and 2C). Simulation results showed
S
that such a desired reaction progress was obtainable only with a ARTA/OATAW58L/R417A pair
(Figure 2D). However, the model prediction suggested that R-1a was not sufficiently consumed to
R
D
allow ee > 99 % seemingly because of a shortage of 3 available for completion of I /III (Figure
S
S4). Increase in the concentration of 3 no less than 6.5 mM was predicted to allow ee > 99 % with
S
reaction yield > 99 % (Figure S5).
In vitro Deracemization
Based on the simulation results, we performed deracemization of 10 mM rac-1a with 6.5 mM 3
and 100 mM 5 using ARTA/OATAW58L/R417A (both 10 M) (Figure 3A). Indeed, the reaction
progress looked similar to the simulation result shown in Figure 2D. However, accumulation of S-
1
a was much slower than expected for such a fast depletion of R-1a. As a result, reaction yield of
S-1a was only 73 % after 7-h reaction. Another problem was a non-negligible build-up of R-1a
observed after 2 h, resulting in 95.1 % final ee . To boost up the slow accumulation of S-1a, we
S
R
D
increased OATAW58L/R417A level to 100 M. To attain ee > 99 %, we opted to reinitiate I /III at
S
the final stage of deracemization by adding 3 slightly more than the R-1a impurity (i.e., a “chiral
polishing” step). Indeed, the two modifications led to 95 % reaction yield with 99.1 % ee at 7 h
S
1
8
ACS Paragon Plus Environment

Page 19 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
Figure 3B). In this reaction, we also monitored time-dependent changes of 2a and 3. After the
R
D
initial burst owing to I /III , concentration of 2a kept going down to 0.04 mM at 6 h. 3 was below
a detection limit from 0.5 h on.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To expand the synthetic applicability of the proposed method, we carried out deracemization of
five additional amines shown in Scheme 1B (Table 5). It took 4.2 - 8 h to achieve over 90 %
reaction yield and 99 % ee starting with 10 mM rac-1b-f (entry 1-5). In the case of 1d (entry 3),
S
initial concentration of 3 (i.e., 6.5 mM) turned out to be deficient to allow complete consumption
of R-1d. Thus, 3 was added at 2.1 h (1 mM) to remove the R-1d leftover (0.6 mM), followed by
second feeding (0.3 mM) at 3.7 h for chiral polishing (Figure S6). Besides the analytical scale
reactions, we carried out preparative scale reaction with 10-fold higher concentration of 1d (1.49
g) in a 100 mL reaction volume (entry 6). After 21.5 h, reaction yield reached 85 % with > 99 %
ee . Product isolation led to 1.01 g of S-1d (79 % recovery yield) of which structural
S
characterization was done by NMR and HRMS (Figure S7).
In vivo Deracemization
Encouraged by the successful implementation of the reaction pathway control, we moved on to
construction of a synthetic metabolic pathway enabling the deracemization cascade in a live
microbe where 3 is available from a much cheaper precursor by endogenous glycolysis (Figure 4).
Toward this end, Escherichia coli BL21(DE3) cells were cotransformed with a low and a high
copy number plasmid expressing ARTA and OATAW58L/R417A (pCDFDuet-ARTA and pRSFDuet-
OATAW58L/R417A, respectively; Figure S8), considering that reaction yield would benefit from a
high activity of OATAW58L/R417A as observed in Figure 3. The resulting cells exhibited 61 and 55 %
activities of ARTA and OATAW58L/R417A, respectively, relative to those measured with E. coli cells
1
9
ACS Paragon Plus Environment

ACS Catalysis
Page 20 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
expressing a single -TA (Table S8). The engineered cell harbors a deracemization pathway (Dp)
where 3 is supplied from glucose by the Embden–Meyerhof–Parnas (EMP) pathway. Note that the
engineered cell should undergo anaerobic metabolism due to the lack of oxygen under the reduced
pressure conditions. Therefore, NADH generated from the EMP pathway, i.e. 1 mol per mol of 3,
should be consumed in the fermentative pathway (Fp) to maintain redox balance, mandating
conversion of 3 to lactate, ethanol and acetate (Figure S9).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The metabolically engineered cells were incubated with 50 mM rac-1a in the presence of
glucose and 5 (0.25 and 4 molar equivalents, respectively, to rac-1a). Note that 2 mol of 3 is
produced per mol of glucose by the EMP pathway. Therefore, 0.25 molar eq. (i.e., 12.5 mM) is a
R
D
minimal amount of glucose required for completion of I /III if 3 is flowed exclusively into Dp.
It turned out that 0.25 molar eq. afforded complete removal of 25 mM R-1a at 18 h with significant
accumulation of S-1a, resulting in 99.5 % ee and 65 % reaction yield without a chiral polishing
S
step (Figure 5A). 2a kept decreasing after 3 h but, unlike Figure 3B, depletion of 2a was much
larger than generation of S-1a, suggesting that 2a might be hijacked by an unknown metabolic
reaction. In line with this result, we observed that an unknown peak grew bigger with respect to
reaction time in the HPLC analysis tuned for 2a whereas the peak did not appear in the in vitro
deracemization (Figure S10).
To perform flux analysis of 3, we measured concentrations of key metabolites at 18 h. Glucose
was completely consumed, indicating that 25 mM 3 was generated along with 25 mM NADH by
the EMP pathway. A substantial amount of 3 was detected (i.e., 0.1 mM), explaining why the
chiral polishing step was not needed to attain > 99 % ee . The outcomes of Fp were 4.6 (lactate),
S
4
.8 (ethanol) and 2.7 mM (acetate), indicating that 12.1 mM 3 was partitioned into Fp. Taken
together with the secretion amount, maximal flux of 3 into Dp is 12.8 mM that is not in accordance
2
0
ACS Paragon Plus Environment

Page 21 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
with the complete removal of 25 mM R-1a. A possible explanation for this disagreement is that 3
might be recycled from D-4 via L-4 by alanine racemase and L-alanine dehydrogenase (Figure
S11), which is supported by detection of L-4 in the reaction mixture where the concentration of
L-4 was 1.7-fold higher than that of D-4 at 18 h.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To examine how the change in glycolytic flux affects deracemization performance, the feeding
amount of glucose was varied from 0.15 to 1 molar eq. Increases in the glucose feeding led to
increasing glucose consumption and glucose leftover was observed from 0.5 eq. on (Table S9).
Compared with 0.25 eq., increasing glucose feeding caused gradual reduction in the reaction yield
although enantiopurity of the resulting S-1a was further increased (Figure 5B). In line with the
pyruvate recycling hypothesis, use of 0.15 eq. glucose (i.e., 7.5 mM) afforded reaction yield as
good as that obtained with 0.25 eq. However, complete removal of R-1a was not achieved and the
resulting enantiopurity was only 96.3 % ee . This result is seemingly ascribable to a shortage of 3
S
when starting with 0.15 eq. glucose, consistent with a non-detectable level of 3 in the reaction
mixture.
It is notable that lactate formation in the engineered E. coli was unusually high compared to that
observed in anaerobic cultivation of wild-type cells (i.e., < 0.2 mM) at similar glucose feeding.39-
4
0
The overflow of 3 to lactate seems to be relevant to NADH overload that caused undesirable
generation of toxic products in Fp. In the previous reports, we demonstrated that threonine
deaminase (TD) could readily convert L-threonine, a cheap amino acid, to 2-oxobutyrate that is a
good amino acceptor for most -TAs.35, 41 Therefore, we posited that Fp could be shut down in the
engineered E. coli by reinforcing expression of native TD and replacing glucose with L-threonine
(Figure S12). To this end, the ilvA gene encoding TD of E. coli was cloned in pCDFDuet-ARTA
and E. coli BL21(DE3) was cotransformed with the resulting plasmid (i.e., pCDFDuet-ARTA/TD;
2
1
ACS Paragon Plus Environment

ACS Catalysis
Page 22 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure S13) and pRSFDuet-OATAW58L/R417A. The resulting engineered cells expressing TD and
both -TAs were incubated with 50 mM rac-1a, 27.5 mM L-threonine and 500 mM 5 at 17.2 mg
o
dcw/mL, 37 C and 460 Torr. After 23.5 h, reaction yield reached 71 % and ee was 99.1 %.
S
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
CONCLUSION
We have demonstrated deracemization of chiral amines in vitro and in vivo using reaction pathway
control. Engineered orthogonality in the substrate range of enantiocomplementary -TAs enabled
specific partitioning of common cosubstrates to an intended reaction step. Our strategy illustrates
how to cope with unfavorable reaction equilibrium and promiscuous substrate specificity that often
remain a challenge in designing a scalable cascade reaction. Besides the deracemization to S-amine
studied here, deracemization to an antipode enantiomer would be implementable using the same
set of substrates if the -TA pair fulfills the desired substrate orthogonality. For example, an
ARTA mutant engineered for sitagliptin production (ARTA ) is known to display activity
mut
2
8
improvements for ketones and 5. So, combination of a S-selective -TA that lacks activity for 5
and an ARTAmut variant that loses activity for 3 would afford deracemization to R-amine. The
deracemization cascade was successfully incorporated in E. coli where native glycolysis or
reinforced deamination of L-threonine was exploited to supply -keto acid from a cheap precursor.
Further engineering of the synthetic microbe is demanded to improve reaction yield, which might
S
involve reinforcement of a crucial bottleneck step (seemingly II/I ), abolishment of undesirable
metabolic interference (hijacking of 2a) and establishment of metabolic removal of 6. To the best
of our knowledge, this is the first example of an engineered metabolic pathway for deracemization.
ASSOCIATED CONTENT
2
2
ACS Paragon Plus Environment

Page 23 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
Supporting Information: The Supporting Information is available free of charge on the ACS
Publications website. Thermodynamic calculation of K and G. Supporting Tables S1-S9;
eq
Supporting Figures S1-S13; HPLC chromatograms.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
AUTHOR INFORMATION
Corresponding Author: * E-mail for J.-S.S.: enzymo@yonsei.ac.kr
Notes: The authors declare no competing financial interest.
ACKNOWLEDGMENT
This work was funded by the National Research Foundation of Korea under the Basic Science
Research Program (2016R1A2B4008470). Dr. S.-W. Han was financially supported by Initiative
for Biological Function & Systems under the BK21 PLUS program of Korean Ministry of
Education. We are grateful to the technical work of Saerom Park and Joonhyun Song.
REFERENCES
(
(
(
(
1) Sandefur, C. I.; Mincheva, M.; Schnell, S. Network Representations and Methods for the
Analysis of Chemical and Biochemical Pathways. Mol. BioSyst. 2013, 9, 2189-2200.
2) Feist, A. M.; Herrgård, M. J.; Thiele, I.; Reed, J. L.; Palsson, B. Ø. Reconstruction of
Biochemical Networks in Microorganisms. Nat. Rev. Microbiol. 2008, 7, 129-143.
3) Rohwer, J. M.; Hofmeyr, J.-H. S. Kinetic and Thermodynamic Aspects of Enzyme Control
and Regulation. J. Phys. Chem. B 2010, 114, 16280-16289.
4) Chakrabarti, A.; Miskovic, L.; Soh, K. C.; Hatzimanikatis, V. Towards Kinetic Modeling of
Genome-Scale Metabolic Networks without Sacrificing Stoichiometric, Thermodynamic
and Physiological Constraints. Biotechnol. J. 2013, 8, 1043-1057.
(5) Bar-Even, A.; Flamholz, A.; Noor, E.; Milo, R. Rethinking Glycolysis: On the Biochemical
Logic of Metabolic Pathways. Nat. Chem. Biol. 2012, 8, 509-517.
2
3
ACS Paragon Plus Environment

ACS Catalysis
Page 24 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
6) Nobeli, I.; Favia, A. D.; Thornton, J. M. Protein Promiscuity and Its Implications for
Biotechnology. Nat. Biotechnol. 2009, 27, 157-167.
(
7) France, S. P.; Hepworth, L. J.; Turner, N. J.; Flitsch, S. L. Constructing Biocatalytic
Cascades: In vitro and in vivo Approaches to de Novo Multi-Enzyme Pathways. ACS Catal.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
017, 7, 710-724.
(8) Schrittwieser, J. H.; Velikogne, S.; Hall, M.; Kroutil, W. Artificial Biocatalytic Linear
Cascades for Preparation of Organic Molecules. Chem. Rev. 2018, 118, 270-348.
(9) Sperl, J. M.; Sieber, V. Multienzyme Cascade Reactions—Status and Recent Advances. ACS
Catal. 2018, 8, 2385-2396.
(10) Medema, M. H.; van Raaphorst, R.; Takano, E.; Breitling, R. Computational Tools for the
Synthetic Design of Biochemical Pathways. Nat. Rev. Microbiol. 2012, 10, 191-202.
(
(
(
11) Tanner, M. E. Understanding Nature's Strategies for Enzyme-Catalyzed Racemization and
Epimerization. Acc. Chem. Res. 2002, 35, 237-246.
12) Femmer, C.; Bechtold, M.; Roberts, T. M.; Panke, S. Exploiting Racemases. Appl.
Microbiol. Biotechnol. 2016, 100, 7423-7436.
13) Schrittwieser, J. H.; Groenendaal, B.; Resch, V.; Ghislieri, D.; Wallner, S.; Fischereder, E.
M.; Fuchs, E.; Grischek, B.; Sattler, J. H.; MacHeroux, P.; Turner, N. J.; Kroutil, W.
Deracemization by Simultaneous Bio-Oxidative Kinetic Resolution and Stereoinversion.
Angew. Chem. Int. Ed. 2014, 53, 3731-3734.
(
14) Ghislieri, D.; Turner, N. J. Biocatalytic Approaches to the Synthesis of Enantiomerically
Pure Chiral Amines. Top. Catal. 2014, 57, 284-300.
(
15) Carr, R.; Alexeeva, M.; Enright, A.; Eve, T. S. C.; Dawson, M. J.; Turner, N. J. Directed
Evolution of an Amine Oxidase Possessing both Broad Substrate Specificity and High
Enantioselectivity. Angew. Chem. Int. Ed. 2003, 42, 4807-4810.
(16) Foulkes, J. M.; Malone, K. J.; Coker, V. S.; Turner, N. J.; Lloyd, J. R. Engineering a
Biometallic Whole Cell Catalyst for Enantioselective Deracemization Reactions. ACS Catal.
2
011, 1, 1589-1594.
(17) Köhler, V.; Wilson, Y. M.; Dürrenberger, M.; Ghislieri, D.; Churakova, E.; Quinto, T.;
Knörr, L.; Häussinger, D.; Hollmann, F.; Turner, N. J.; Ward, T. R. Synthetic Cascades are
Enabled by Combining Biocatalysts with Artificial Metalloenzymes. Nat. Chem. 2013, 5,
93-99.
2
4
ACS Paragon Plus Environment

Page 25 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
(
(
18) Yasukawa, K.; Nakano, S.; Asano, Y. Tailoring D-Amino Acid Oxidase from the Pig Kidney
to R-Stereoselective Amine Oxidase and Its Use in the Deracemization of -
Methylbenzylamine. Angew. Chem. Int. Ed. 2014, 53, 4428-4431.
19) Heath, R. S.; Pontini, M.; Hussain, S.; Turner, N. J. Combined Imine Reductase and Amine
Oxidase Catalyzed Deracemization of Nitrogen Heterocycles. ChemCatChem 2016, 8, 117-
120.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
20) Aleku, G. A.; Mangas-Sanchez, J.; Citoler, J.; France, S. P.; Montgomery, S. L.; Heath, R.
S.; Thompson, M. P.; Turner, N. J. Kinetic Resolution and Deracemization of Racemic
Amines Using a Reductive Aminase. ChemCatChem 2018, 10, 515-519.
(21) Koszelewski, D.; Pressnitz, D.; Clay, D.; Kroutil, W. Deracemization of Mexiletine
Biocatalyzed by -Transaminases. Org. Lett. 2009, 11, 4810-4812.
(
22) Koszelewski, D.; Clay, D.; Rozzell, D.; Kroutil, W. Deracemisation of -Chiral Primary
Amines by a One-Pot, Two-Step Cascade Reaction Catalysed by -Transaminases. Eur. J.
Org. Chem. 2009, 2289-2292.
(23) Koszelewski, D.; Müller, N.; Schrittwieser, J. H.; Faber, K.; Kroutil, W. Immobilization of
-Transaminases by Encapsulation in a Sol-Gel/Celite Matrix. J. Mol. Catal. B Enzym.
010, 63, 39-44.

2
(24) Shin, G.; Mathew, S.; Shon, M.; Kim, B.-G.; Yun, H. One-Pot One-Step Deracemization of
Amines Using -Transaminases. Chem. Commun. 2013, 49, 8629-8631.
(25) Malik, M. S.; Park, E. S.; Shin, J. S. Features and Technical Applications of -
Transaminases. Appl. Microbiol. Biotechnol. 2012, 94, 1163-1171.
(
26) Kelly, S. A.; Pohle, S.; Wharry, S.; Mix, S.; Allen, C. C. R.; Moody, T. S.; Gilmore, B. F.
Application of -Transaminases in the Pharmaceutical Industry. Chem. Rev. 2018, 118, 349-
367.
(
27) Yoon, S.; Patil, M. D.; Sarak, S.; Jeon, H.; Kim, G.-H.; Khobragade, T. P.; Sung, S.; Yun,
H., ChemCatChem 2019, 11, 1898-1902.
(28) Savile, C. K.; Janey, J. M.; Mundorff, E. C.; Moore, J. C.; Tam, S.; Jarvis, W. R.; Colbeck,
J. C.; Krebber, A.; Fleitz, F. J.; Brands, J.; Devine, P. N.; Huisman, G. W.; Hughes, G. J.
Biocatalytic Asymmetric Synthesis of Chiral Amines from Ketones Applied to Sitagliptin
Manufacture. Science 2010, 329, 305-309.
2
5
ACS Paragon Plus Environment

ACS Catalysis
Page 26 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
29) Dawood, A. W. H.; Weiß, M. S.; Schulz, C.; Pavlidis, I. V.; Iding, H.; de Souza, R. O. M.
A.; Bornscheuer, U. T. Isopropylamine as Amine Donor in Transaminase-Catalyzed
Reactions: Better Acceptance through Reaction and Enzyme Engineering. ChemCatChem
2
018, 10, 3943-3949.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
30) Iwasaki, A.; Yamada, Y.; Kizaki, N.; Ikenaka, Y.; Hasegawa, J. Microbial Synthesis of
Chiral Amines by (R)-Specific Transamination with Arthrobacter sp. KNK168. Appl.
Microbiol. Biotechnol. 2006, 69, 499-505.
(
31) Park, E.-S.; Kim, M.; Shin, J.-S. Molecular Determinants for Substrate Selectivity of -
Transaminases. Appl. Microbiol. Biotechnol. 2012, 93, 2425-2435.
(32) Han, S. W.; Park, E. S.; Dong, J. Y.; Shin, J. S. Mechanism-Guided Engineering of -
Transaminase to Accelerate Reductive Amination of Ketones. Adv. Synth. Catal. 2015, 357,
1
732-1740.
(
33) Park, E. S.; Shin, J. S. Deracemization of Amino Acids by Coupling Transaminases of
Opposite Stereoselectivity. Adv. Synth. Catal. 2014, 356, 3505-3509.
(
34) Park, E.-S.; Malik, M. S.; Dong, J.-Y.; Shin, J.-S. One-Pot Production of Enantiopure
Alkylamines and Arylalkylamines of Opposite Chirality Catalyzed by -Transaminase.
ChemCatChem 2013, 5, 1734-1738.
(35) Park, E.; Kim, M.; Shin, J.-S. One-Pot Conversion of L-Threonine into L-Homoalanine:
Biocatalytic Production of an Unnatural Amino Acid from a Natural One. Adv. Synth. Catal.
2
010, 352, 3391-3398.
(
(
(
36) Han, S.-W.; Kim, J.; Cho, H.-S.; Shin, J.-S. Active Site Engineering of -Transaminase
Guided by Docking Orientation Analysis and Virtual Activity Screening. ACS Catal. 2017,
7
, 3752-3762.
37) Park, E. S.; Dong, J. Y.; Shin, J. S. -Transaminase-Catalyzed Asymmetric Synthesis of
Unnatural Amino Acids using Isopropylamine as an Amino Donor. Org. Biomol. Chem.
2013, 11, 6929-6933.
38) Han, S.-W.; Shin, J.-S. A Facile Method to Determine Intrinsic Kinetic Parameters of -
Transaminase Displaying Substrate Inhibition. J. Mol. Catal. B Enzym. 2016, 133, S500-
S507.
(39) Trotter, E. W.; Rolfe, M. D.; Hounslow, A. M.; Craven, C. J.; Williamson, M. P.;
Sanguinetti, G.; Poole, R. K.; Green, J. Reprogramming of Escherichia coli K-12
2
6
ACS Paragon Plus Environment

Page 27 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
Metabolism during the Initial Phase of Transition from an Anaerobic to a Micro-Aerobic
Environment. PLoS ONE 2011, 6, e25501.
(
40) Murarka, A.; Clomburg, J. M.; Moran, S.; Shanks, J. V.; Gonzalez, R. Metabolic Analysis
of Wild-Type Escherichia coli and a Pyruvate Dehydrogenase Complex (PDHC)-Deficient
Derivative Reveals the Role of PDHC in the Fermentative Metabolism of Glucose. J. Biol.
Chem. 2010, 285, 31548-31558.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
41) Malik, M. S.; Park, E. S.; Shin, J. S. -Transaminase-Catalyzed Kinetic Resolution of Chiral
Amines using L-Threonine as an Amino Acceptor Precursor. Green Chem. 2012, 14, 2137-
2
140.
2
7
ACS Paragon Plus Environment

ACS Catalysis
Page 28 of 48
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. -TA-based deracemization of chiral amines using 3 and 5 as cosubstrates. (A)
Conversion of racemic amine into S-enantiomer enclosed by a box. (B) Chiral amines used in this
study.
2
8
ACS Paragon Plus Environment

Page 29 of 48
ACS Catalysis
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 2. Whole catalytic cycle of a transaminase reaction consisting of two half reactions.
2
9
ACS Paragon Plus Environment