Combinatorial Mutation Analysis of ω-Transaminase to Create an Engineered Variant Capable of Asymmetric Amination of Isobutyrophenone

Article abstract of DOI:10.1002/adsc.201900184

ω-Transaminase (ω-TA) is an important enzyme for asymmetric synthesis of chiral amines. Rapid creation of a desirable ω-TA variant, readily available for scalable process operation, is demanded and has attracted intense research efforts. In this study, we aimed to develop a quantitative mutational analysis (i. e., R-analysis) that enables prediction of combinatorial mutation outcomes and thereby provides reliable guidance of enzyme engineering through combination of already characterized mutations. To this end, we determined three mutatable active-site residues of ω-TA from Ochrobactrum anthropi (i. e., leucine 57, tryptophan 58 and valine 154) by examining activities of nine alanine-scanning mutants for seven substrate pairs. The R-analysis of the mutatable residues is based on assessment of changes in relative activities for a series of structurally analogous substrates. Using three sets of substrates (five α-keto acids, six arylalkylamines and three arylalkyl ketones), we found that combination of two point mutations display additive effects of each mutational outcome such as steric relaxation for bulky substrates or catalytic enhancement for amination of ketones. Consistent with the R-analysis-based prediction, the ω-TA variant harboring triple alanine mutations, i. e. L57A, W58A and V154A, showed high activity improvements for bulky substrates, e. g. a 3.2×10⁴-fold activity increase for 1-phenylbutylamine. The triple mutant even enabled asymmetric amination of isobutyrophenone, carrying a branched-chain alkyl substituent to be accepted in a small binding pocket that normally shows a steric limit up to an ethyl group, with >99% ee of a resulting (S)-amine. (Figure presented.).

Full text of DOI:10.1002/adsc.201900184

Title: Combinatorial Mutation Analysis of ω-Transaminase to Create
an Engineered Variant Capable of Asymmetric Amination of
Isobutyrophenone
Authors: Hong-Gon Kim, Sang-Woo Han, and Jong-Shik Shin
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Adv. Synth. Catal. 10.1002/adsc.201900184
Link to VoR: http://dx.doi.org/10.1002/adsc.201900184

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
FULL PAPER
DOI: 10.1002/adsc.201((will be filled in by the editorial staff))
Combinatorial Mutation Analysis of -Transaminase to Create
an Engineered Variant Capable of Asymmetric Amination of
Isobutyrophenone
a
a
a,
Hong-Gon Kim, Sang-Woo Han and Jong-Shik Shin *
a
Department of Biotechnology, Yonsei University, Yonsei-Ro 50, Seodaemun-Gu, Seoul 03722, South Korea
[
Phone: (+82)-2-2123-5884; Fax: (+82)-2-362-7265; e-mail: enzymo@yonsei.ac.kr]
Received: ((will be filled in by the editorial staff))
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201######.
Abstract. -Transaminase (-TA) is an important three arylalkyl ketones), we found that combination of two
enzyme for asymmetric synthesis of chiral amines. Rapid point mutations display additive effects of each mutational
creation of a desirable -TA variant, readily available for outcome such as steric relaxation for bulky substrates or
scalable process operation, is demanded and has attracted catalytic enhancement for amination of ketones. Consistent
intense research efforts. In this study, we aimed to with the R-analysis-based prediction, the -TA variant
develop a quantitative mutational analysis (i.e., R- harboring triple alanine mutations, i.e. L57A, W58A and
analysis) that enables prediction of combinatorial V154A, showed high activity improvements for bulky
4
mutation outcomes and thereby provides reliable guidance substrates, e.g. a 3.2  10 -fold activity increase for 1-
of enzyme engineering through combination of already phenylbutylamine. The triple mutant even enabled
characterized mutations. To this end, we determined three asymmetric amination of isobutyrophenone, carrying a
mutatable active-site residues of -TA from branched-chain alkyl substituent to be accepted in a small
Ochrobactrum anthropi (i.e., leucine 57, tryptophan 58 binding pocket that normally shows a steric limit up to an
and valine 154) by examining activities of nine alanine- ethyl group, with > 99 % ee of a resulting (S)-amine.
scanning mutants for seven substrate pairs. The R-analysis
of the mutatable residues is based on assessment of Keywords: -transaminase; chiral amines; asymmetric
changes in relative activities for a series of structurally synthesis; protein engineering; combinatorial mutation
analogous substrates. Using three sets of substrates (five
-keto acids, six arylalkylamines and
Introduction
amino donor, accompanied by conversion of an
enzyme-PLP form (E-PLP) to an enzyme form
harboring pyridoxamine 5-phosphate (PMP), and 2)
reductive amination of an amino acceptor, leading to
regeneration of E-PLP that resumes another catalytic
cycle. Compared with -transaminase responsible for
amino group transfer exclusively between -amino
acids and -keto acids, -TA displays a unique
catalytic property that enables deamination of amines
devoid of a carboxyl group and thereby amination of
the resulting ketones and aldehydes as listed in Fig.
Chiral amines are found in
a
number of
[1]
pharmaceutical drugs as an essential building block,
which attracts extensive research efforts to develop
chemocatalytic[ and biocatalytic
production of an enantiopure form. -Transaminase
-TA), more specifically amine transaminase or
2]
[3]
methods for
(
amine:pyruvate transaminase, is one of the competent
workhorses available for asymmetric synthesis of the
chiral amines starting with prochiral ketones and
cheap amino donors such as isopropylamine and
alanine.[ Successful development of a scalable
process using -TAs often depends on availability of
a suitable enzyme for amination of a target ketone.
This has invited huge research efforts aimed at
engineering of native enzymes to endow desired
properties such as altered substrate scope for bulky
substrates and enhanced catalytic performance for
amination of ketones.[
4]
5]

-TA exploits pyridoxal 5-phosphate (PLP) as an
.[5e,7] One of the important research goals in protein
1
internally bound cofactor to mediate amino group
transfer from an amino donor (D) to an acceptor
Scheme 1. A whole catalytic cycle of the -TA reaction.
E-PLP:D and E-PMP:A represent Michaelis complexes.
DP and AP stand for a deamination and an amination
6]
(
A).[ The whole catalytic cycle consists of two half
reactions (Scheme 1); 1) oxidative deamination of an
1
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
product, respectively.
are depicted in the active site of E-PLP and E-PMP,
respectively.
overcome for successful protein engineering in order
[5d,5e]
to improve an enzyme activity for a given ketone.
In the previous studies, we demonstrated that a
single point mutation introduced in the active site of
[
5e,9e,13]

-TA could greatly alter enzyme properties.
For example, a single point mutation of tryptophan 58
in a (S)-selective -TA from Ochrobactrum anthropi
(
OATA) was proven to dramatically improve
[5e]
catalytic turnover of ketones.
Moreover, OATA
could be engineered to accept even a n-hexyl group
of an -keto acid in the S pocket by a point mutation
[13b]
of leucine 57.
Nevertheless, we have not
succeeded in creating an OATA variant capable of
accommodating a branched-chain alkyl group in the S
pocket although protein engineering to accept
arylalkyl ketones carrying an isopropyl or a t-butyl
[8b,11b]
group was reported elsewhere.
The aim of this
study is the development of a predictive analysis tool
for combinatorial mutations and its application to
active site engineering of OATA that allows the
accommodation of an isopropyl group in the S pocket
and thus endows activities for (S)-D4 and A4. Instead
of mutant library generation and screening, we sought
to develop a knowledge-guided strategy that was
based on productive combination of single point
mutations. The decision-making on whether or not to
combine specific point mutations was guided by the
proposed analysis tool that enabled quantitative
interpretation of mutational effects and qualitative
prediction of combinatorial mutation outcomes. Our
strategy afforded rapid construction of an OATA
variant showing activities for (S)-D4 and A4 through
triple point mutations.
Figure 1. Substrates used in this study. (A) Amino donors
and (B) amino acceptors.
engineering of -TAs is to increase a reaction rate for
a target ketone, focusing on excavation of a narrow
active site to facilitate productive binding of bulky
ketones.[
8]
Substrate scope of -TAs can be defined by a
couple of general rules which were suggested by
experimental data of quantitative substrate-activity
relationship[
7a]
and then verified by structural
[9]
modeling using substrate docking. A two-binding-
site model, consisting of a large (L) and a small (S)
pocket, has successfully explained substrate
specificity and stereoselectivity of most -TAs (Fig.
[4a,7a,10]
2).
First of all, the S pocket shows a strict steric
limitation that denies entry of larger than an ethyl
substituent. This feature has frustrated use of bulky
substrates, attracting a great deal of protein
engineering research to relieve the steric
Results and Discussion
Development of a Knowledge-guided Strategy for
Productive Combinatorial Mutations
constraint.[8,11]
Second, the
L
pocket can
accommodate a hydrophobic as well as a carboxyl
group. This dual recognition is mediated by an active
site arginine that undergoes a gross movement
depending on the type of a substituent that enters the
Amino acid residues in the active site play a specific
role to catalyze an enzyme reaction, which might be
generally classified into two groups, i.e. the residues
responsible for a binding step and the ones for a
L pocket.[6a,12]
Third, ketones are very poor substrates
catalytic turnover step that are assessed by K and kcat
M
even though their cognate amines are good
substrates.[5d] For example, acetophenone (A1) shows
values, respectively. One of the difficult decisions to
be made during protein engineering is how to select
mutatable sites and how to predict productive
combination of the beneficial and/or adverse point
mutations that are identified in a previous step. As the
number of point mutations increases, docking
simulation results tend to be unreliable unless X-ray
structures of intermediate-stage mutants are available.
Therefore, those decisions often resort to cumulative
hands-on experience and require time-consuming
trial-and-error experimentation for library generation
and screening.
a very low reactivity while its amination product, i.e.

-methylbenzylamine (D1), is a typical amino donor
for most -TAs. Besides the steric issue in the S
The motivation of this study is to develop a
knowledge-guided protein engineering strategy that
exploits a predictive analysis tool for determining
pocket, this is another crucial problem to be
Figure 2. A two-binding-site model. -MBA and pyruvate
2
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
[
a]
productive combination of the mutatable residues. As
outlined in Scheme 2, the proposed strategy starts
with
Reaction conditions: 10 mM amino donor, 10 mM acceptor
and 2 M enzyme.
[
[
[
[
b]
c]
d]
e]
It represents an initial reaction rate per enzyme concentration.
It is based on v
n.r.: not reactive (i.e., < 0.5 % of v
Due to low activities for ketones, high enzyme concentration
10 M) was used for reliable measurement of the initial rate.
i
for (S)-D1/A6.
i
for (S)-D1/A6).
(
prediction of an enzyme activity of a native -TA for
a given substrate pair. We determined relative
activities of OATA for seven representative substrate
pairs using (S)-D1/A6 as a reference substrate pair
(
Table 1). Replacing A6, a typical amino acceptor for

-TAs, with A5 led to a similar activity, consistent
[9f]
with good activities of most -TAs for aldehyde.
This property is in line with the steric constraint in
the S pocket because the aldehydic hydrogen is the
last one that might undergo steric hindrance in the S
pocket. Moreover, the same steric consideration
explains the drastic activity loss for A8, A11 and (S)-
D3. Although the substrate structure would not
undergo rejection in the S pocket, A1 shows a very
low reactivity for most -TAs as also observed with
Scheme 2. Work flowchart of the proposed knowledge-
guided protein engineering.
identification of mutatable residues to rule out
detrimental mutations that are likely to cause a
significant loss of desirable properties. This step is
carried out by alanine scanning of the active site
residues and activity measurements of the resulting
mutants using representative substrate pairs. When
the mutatable residues are determined, quantitative
mutation analysis, i.e. R-analysis, is performed using
a set of structurally related substrates, with a
cosubstrate fixed, to probe whether the mutation site
responds positively or negatively to the structural
difference of a test substrate relative to a reference
substrate. This analysis is based on comparison of
relative reaction rates of the mutants to those of a
wild-type enzyme. The next step is assessment of
combinatorial mutation outcomes using the R values,
enabling prediction of productive combination of
point mutations. When the combinatorial mutation is
determined, the final step is to verify the variant by
activity measurement with a target substrate.
0
.07 % activity of OATA for A1 relative to that for
A6. D8 is regarded as an ideal amino donor for
[5a,14]
amination of ketones.
However, not many -TAs
[7a,15]
display a significant activity for D8.
The good
activity of OATA for D8 renders this enzyme
attractive for practical synthesis of chiral amines, as
[16]
mentioned elsewhere.
The two-binding-site model is visualized with a X-
ray structure of OATA, determined in a previous
[5d]
study, using docking models of (S)-D1 and A6 (Fig.
Native Activities for Representative Substrates
The two-binding-site model readily allows qualitative
Table 1. Native enzyme a]activities of OATA for
[
representative substrate pairs.
[
b]
Substrate
pair
Specific reaction rate (v
i
)
Relative
activity (%)
-
1
-1
[c]
(mM min mM-enzyme )
(
(
(
(
(
S)-D1/A6
S)-D1/A5
S)-D1/A8
S)-D1/A11
S)-D3/A6
120 ± 25
114 ± 17
0.9 ± 0.2
n.r.[d]
100
97
0.7
< 0.5
< 0.5
0.07
39
n.r.
0.08 ± 0.01[e]
L-D9/A1
D8/A6
47 ± 20
3
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
3
). These docking poses represent typical productive
mutations, W58A was strongly beneficial for L-
[5e]
D9/A1 as reported previously. In addition, W58A
was found to be the only alanine mutant showing a
beneficial activity change for (S)-D1/A5 whose
amino acceptor is an
Figure 3. Docking models of (A) (S)-D1 in E-PLP and (B)
A6 in E-PMP. Bound substrates are shown as thick sticks.
The green solid lines represent (A) N(S)-D1C4PLP and (B)
CA6NPMP. The green dotted lines represent hydrogen
bonds. The active site is visualized by a Connolly surface.
The L pocket is exposed to a solvent side and points
toward a viewer.
aldehyde.
F86A caused a drastic activity loss for (S)-D1/A6,
which led us to deter from measuring activities for
other substrate pairs. Likewise, Y20A, Y151A,
F323A and T324A turned out to be detrimental
mutations for most substrate pairs. Despite beneficial
activity changes for both (S)-D1/A5 and L-D9/A1,
I261A led to a drastic activity loss for D8/A6. Note
that D8 is one of the preferred amino donors for
amination of ketones. Taken together, these six
residues were crossed out from a list of mutatable
residues.
substrate binding in the active site of -TA. Side
chains of nine amino acid residues participate in the
active site where arginine 417 acts as
a
conformational switch that controls the dual
recognition mode. In contrast to arginine 417
involved in the substrate binding, lysine 287 play a
pivotal role in catalytic turnover as a nucleophile and
a Lewis acid/base. Note that OATA forms a
homodimeric structure where the active site is located
in the subunit interface and phenylalanine 86 and
phenylalanine 323 come from the other subunit.
For (S)-D1/A8 and (S)-D3/A6 pairs containing a
substrate harboring a n-propyl group that exceeds the
steric constraint of the S pocket, L57A elicited
dramatic activity improvements for both substrate
pairs owing to relocation of the S pocket as studied
[13]
Identification of Mutatable Residues by Alanine
Scanning
previously. In addition, V154A led to an activity
increase for (S)-D1/A8. Note that valine 154 closes
the narrow S pocket tunnel as shown in Fig. 3 and
thereby V154A leads to direct excavation of the S
pocket. In line with this result, V153A of the -TA
from Paracoccus denitrificans (PDTA) was observed
Using the active site model, we set out to determine
which active site residues are amenable to
engineering. This decision was made by eliminating
the residues whose mutations led to significant loss of
desirable activities, which was based on evaluation of
activity changes of alanine scanning mutants that
harbored a single alanine mutation of the nine active
site residues. The resulting alanine mutants were
subjected to activity measurements for the seven
representative substrate pairs (Table 2). Relative
activity less than 50 % of the wild-type enzyme was
regarded as detrimental, shown in italic, whereas that
over 150 % was considered as beneficial, shown in
bold. Activity improvement over 10-fold was
assigned as strongly beneficial and marked bold and
underlined.
[9e]
to relieve the steric constraint in the S pocket.
However, unlike PDTA, the steric relaxation was not
observed for (S)-D3/A6 with V154A of OATA. This
result suggests that V154A might not attain steric
relaxation strong enough to allow a detectable
activity increase for (S)-D3/A6. It is intriguing that
W58A showed a high activity increase for (S)-D3/A6
but caused an adverse effect for (S)-D1/A8. The
opposite mutational effects of W58A for A8 and (S)-
D3 seem to be relevant to the type of substituents to
be placed in the L pocket. Docking simulations
predict that, unlike L57A, W58A leads a carbonyl
oxygen of A8 to form a hydrogen bond with arginine
417 and thereby renders the carbonyl carbon
inaccessible to a nucleophilic attack by PMP (Fig.
S1). In contrast, docking models showed productive
binding of (S)-D3 to both L57A and W58A (Fig. S1).
In the case of (S)-D1/A11, no mutant showed a
Three residues, tryptophan 58, valine 154 and
isoleucine 261, showed beneficial activity changes
for (S)-D1/A6 by the alanine substitution. It is notable
that the same mutants displayed activity
improvements for L-D9/A1 that is a substrate pair for
the reverse reaction of (S)-D1/A6. Among the three
Table 2. Fold-changes in the activities of OATA caused by alanine mutation of the nine active site residues.[a]
Fold-change in specific reaction rate (r = vi, mutant/vi, wild-type)[b]
Mutation
(
S)-D1/A6
(S)-D1/A5
(S)-D1/A8
(S)-D1/A11
(S)-D3/A6
L-D9/A1[f]
D8/A6
Y20A
L57A
0.18 ± 0.14
1.07 ± 0.10
1.99 ± 0.31
0.05 ± 0.01
0.39 ± 0.09
1.86 ± 0.53
1.60 ± 0.33
0.42 ± 0.14
0.81 ± 0.16
0.15 ± 0.02
0.62 ± 0.06
1.80 ± 0.03
n.m.[e]
0.69 ± 0.27
31.3 ± 0.7
0.31 ± 0.14
n.m.
n.r.[c]
n.r.
n.r.
n.m.
n.r.
n.r.
n.r.
n.r.
n.r.
n.r.
0.43 ± 0.03
0.58 ± 0.05
34.5 ± 4.5
n.m.
0.40 ± 0.05
0.54 ± 0.21
0.57 ± 0.28
n.m.
> 17[d]
> 23[d]
n.m.
n.r.
W58A
F86A
Y151A
V154A
I261A
F323A
T324A
0.12 ± 0.06
1.23 ± 0.16
0.80 ± 0.14
0.46 ± 0.15
0.33 ± 0.26
0.58 ± 0.37
8.41 ± 0.61
0.25 ± 0.05
0.14 ± 0.05
0.10 ± 0.12
0.01 ± 0.001
1.65 ± 0.01
2.80 ± 0.01
0.53 ± 0.04
0.48 ± 0.01
0.07 ± 0.01
1.26 ± 0.16
0.05 ± 0.03
0.15 ± 0.07
0.07 ± 0.05
n.r.
n.r.
n.r.
n.r.
4
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
[
[
[
[
a]
b]
c]
d]
Font code for fold-changes: italic for r < 0.5, regular for 0.5 < r < 1.5, bold for r > 1.5 and bold and underlined for r > 10.
Reaction conditions: 10 mM amino donor and 10 mM acceptor.
n.r.: not reactive (i.e., < 0.5 % of vi, wild-type for (S)-D1/A6).
Owing to a non-detectable activity of the wild-type OATA under the reaction conditions, the minimum fold-change was estimated by
dividing vi, mutant for (S)-D3/A6 with 0.5 % of vi, wild-type for (S)-D1/A6.
n.m.: not measured.
[
[
e]
f]
[5e]
The fold-change data for L-D9/A1 were taken from a previous study.
Figure 4. Activity changes of OATA induced by either of and 3) V154A leads to mild steric relaxation in the S
L57A, W58A and V154A for three sets of substrates. (A) pocket.

-Keto acids with (S)-D1 as an amino donor (both 10 mM),
B) (S)-arylalkylamines with A6 as an acceptor (both 10 Activity Changes of the Three Alanine Mutants
(
mM) and (C) arylalkyl ketones (50 mM) with D8 (500
mM) as a donor in the presence of 15 % (v/v) DMSO.
To scrutinize precise mutational effects of the three
mutatable residues, we examined activity changes of
detectable activity seemingly due to a too bulky size the three alanine mutants with three sets of substrates,
of the benzyl group of A11 to enter the S pocket.
i.e. -keto acids (A6-A10), arylalkylamines (D1-D3
Taken all together, we chose leucine 57, tryptophan and D5-D7), and arylalkyl ketones (A1-A3) (Fig. 4).
8 and valine 154 as mutatable residues for further Note that A6-A10, D1-D3 and A1-A3 carry n-alkyl
5
study. The alanine scanning results could be side chains to be accommodated in the S pocket.
recapitulated as follows; 1) L57A elicits strong steric The wild-type OATA showed very low activities
relaxation in the S pocket irrespective of the for -keto acids carrying a side chain larger than an
substituent type in the L pocket but does not achieve ethyl group, i.e. A8-A10 (Fig. 4A). L57A showed
activity improvement for ketone, 2) W58A alleviates high activity improvements for such bulky -keto
the steric constraint of a hydrophobic substrate and acids. Besides, V154A led to such improvements
induces a dramatic activity improvement for ketone though in a lesser degree. In contrast, W58A did not
engender the activity gain for A8-A10.
5
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
assay, from the mutational outcome. For example, Fig.
4
A shows that V154A leads to an activity for A7 1.7-
fold higher than L57A does. However, a similar
activity increase for A6 makes it ambiguous to
conclude that V154A exerts a beneficial effect for A7
stronger than L57A does because a part of the
beneficial effect might come from an activity increase
for (S)-D1 used as an amino donor.
To clarify the ambiguity in the mutational analysis
complicated by the contribution of a cosubstrate, we
decided to develop a quantitative analysis method
enabling elimination of the cosubstrate effect through
normalization of relative activity of a mutant with that
of a wild-type enzyme. For this analysis, we presume
that the two half reactions, shown in Scheme 1, are
independently affected by the active site mutation
with a mutational scaling factor (i.e., f) relative to a
wild-type enzyme, i.e. f
D
and f for deamination and
A
amination steps, respectively. As a result, mutational
effect on an overall reaction rate (vD,A) for a
donor/acceptor (D/A) pair could be approximated to a
product of the two f values as expressed in Eq. 1
where M and WT stand for a mutant and a wild-type,
respectively.
(
푣D,A)M = 푓 푓 (푣D,A)WT
(Eq. 1)
D A
We define relative activity (vrel) for a D/A pair as vD,A
divided by that for a reference substrate pair
consisting of D
R
and A .
R
푣rel = 푣D,A⁄푣DR,AR
(Eq. 2)
(Eq. 3)
Combining Eq. 1 and 2 leads to
푓 푓
D A
ꢀ
푣relꢁM
=
ꢀ푣relꢁWT
푓 푓
DR AR
Because the reaction rates for different substrates in
Fig. 4 were determined with a cosubstrate fixed, one
of the mutational scaling factors in Eq. 3 can be
cancelled out. For example, 푓 is the same as
D
The steric constraint in the S pocket was observed
푓 when using the activity data in Fig. 4A with (S)-
D
R
to be more striking with D1-D3 and the native activity
was almost abolished even for D2 carrying an ethyl
group (Fig. 4B). The steric relaxation caused by L57A,
as observed with -keto acids, allowed activity
improvements for D2 and D3 as well as for D5-D7.
Similarly, activity improvements for the bulky
arylalkylamines were observed with W58A. However,
V154A did not induce such a beneficial effect for D3
and D5.
D1/A6 chosen as a reference substrate pair. As a
result, Eq. 3 can be rearranged to give definition of
the ratio of f values for amino acceptors (R
A
).
푓
ꢀ푣 ꢁ
A
ꢂꢃꢄ ꢅ
푅A =
=
|
(Eq. 4)
푓AR
ꢀ
푣 ꢁ
ꢂꢃꢄ ꢆꢇ
fixed amino donor
Likewise, the activity data in Fig. 4B can be used to
determine the ratio of f values for amino donors (R )
D
as shown in Eq. 5.
In the case of arylalkyl ketones, W58A showed
activity increases for all the three ketones (Fig. 4C).
However, both L57A and V154A led to activity
decreases for A1 although a significant activity gain
was found with A2 and A3.
푓
ꢀ푣 ꢁ
ꢂꢃꢄ ꢅ
D
푅D =
=
|
(Eq. 5)
푓
D
ꢀ푣 ꢁ
ꢂꢃꢄ ꢆꢇ
R
fixed amino acceptor
The R value represents how much the mutation is
beneficial for a target substrate relative to a reference
substrate. Therefore, the R value can be interpreted as
whether the mutation responds positively or
negatively to the structural difference of a target
substrate compared to a reference. Consequently, the
R value over unity indicates a beneficial mutational
effect on the change in the substrate structure whereas
Theoretical Background of Quantitative Mutation
Analysis
Fig. 4 enables direct activity comparison and thereby
rapid determination of which mutation is the best for
a
given substrate pair. However,
a
crucial
0
< R < 1 indicates an adverse effect. As a result, the
shortcoming of this is the inability to rule out
contribution of a cosubstrate, used in the activity
R-analysis can be used to provide a quantitative
6
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
measure of the mutational effect on the structural placed in the L pocket and attains the mild steric
difference of substrates.
relaxation purely by excavation in the S pocket.
R-Analysis of the Single Point Mutations
R-Analysis of the Double Point Mutations
Using Eq. 4 or Eq. 5, the R-analysis was carried out The R-analysis enabled quantitative assessment of the
with specific reaction rate data in Fig. 4 (Table 3). steric relaxation induced by each point mutation
The R-analysis with D3, A2 and A3 was not possible depending on the substrate type. Although we could
because the activities of the wild-type OATA for not perform the R-analysis with ketones, W58A was
these substrates were lower than a detection limit found to elicit dramatic activity improvements for A1
under the assay conditions. The R value of L57A is and A2 (Fig. 4C). Keeping these results in mind, we
2
.3 for A7 but becomes much larger as the size of the decided to tackle an intriguing question: what does
side chain
combination of two point mutations lead to in terms
of an additive effect? To address this, we constructed
three combinatorial mutants and carried out activity
measurements with the three sets of substrates (Fig. 5).
Note that L57A/W58A is proximal mutations whereas
L57A/V154A and W58A/V154A are distal mutations.
For A6 and A7, the two double mutants harboring
W58A showed higher activities than the
L57A/V154A mutant did (Fig. 5A). However, an
opposite trend was observed for A8-A10. This result
is ascribable to the adverse effect of W58A for A8-
A10. It is notable that activity increases for A8-A10
induced by L57A/ V154A are larger than those
induced by each point mutation. For example, activity
of the L57A/V154A mutant for A10 is 3.5 and 8.2-
fold higher than those of L57A and V154A mutants,
respectively. Taken together, enzyme activities of the
Table 3. R-analysis of the three single point mutations.[a]
R
Substrate
L57A
W58A
1.2 ± 0.1
0.3 ± 0.1
0.0
V154A
A7
2.3 ± 0.1
29.1 ± 7.4
30.4 ± 4.6
11.5 ± 0.8
14.4 ± 1.5
12.4 ± 1.3
7.5 ± 1.2
17.6 ± 3.2
2.2 ± 0.2
4.5 ± 1.2
4.6 ± 0.7
5.2 ± 0.4
5.6 ± 0.5
0.9 ± 0.1
4.6 ± 0.8
4.7 ± 0.5
A8
A9
A10
0.0
(
(
(
(
S)-D2
S)-D5
S)-D6
S)-D7
25.8 ± 2.2
6.0 ± 0.5
15.7 ± 1.9
15.2 ± 3.8
[
a]
The reference substrate pair is (S)-D1/A6.
of -keto acid exceeds an ethyl group (A8-A10). The double mutants for bulky -keto acids showed
similar R values of L57A for A8 and A9 suggest that additive effects of each point mutation, i.e. strongly
n-propyl and n-butyl substituents undergo a similar beneficial by L57A,
degree of steric relaxation in the S pocket. A 2.5-fold Figure 5. Enzyme activities of OATA variants, carrying
decrease in the R value for A10, compared to those double mutations, for three sets of substrates. (A) -Keto
for A8 and A9, indicates that the steric relaxation acids, (B) (S)-arylalkylamines and (C) arylalkyl ketones.
induced by L57A becomes less favorable for a n-
hexyl group. The steric relaxation by L57A leads to weakly beneficial by V154A and strongly detrimental
high R values for arylalkylamines. It is notable that by W58A.
the R value for (S)-D5 is similar to that for (S)-D2
The additive effect was also observed with
although a structural difference between (S)-D5 and a arylalkylamines (Fig. 5B). However, as indicated by
reference amino donor ((S)-D1), unlike between (S)- the R-analysis in Table 3, W58A exhibited a strongly
D2 and (S)-D1, lies in the substituent that enters the L positive effect on the steric relaxation for the
pocket. This result is consistent with the previous arylalkylamines. As a result, the L57A/W58A mutant
investigation that the steric relaxation in the S pocket showed the highest activity for (S)-D2, (S)-D3 and
induced by L57A comes from relocation of the S (S)-D5 among the three double mutants. For (S)-D6
[
13a]
pocket caused by the expansion of the L pocket.
and (S)-D7, the two double mutants harboring W58A
In contrast to L57A, the R values of W58A for A8- displayed similar activities which were higher than
A10 were much lower than unity. This results clearly those of the L57A/V154A mutant.
indicate that W58A is detrimental to enzyme activity
For ketones, W58A exhibited a strongly beneficial
for -keto acid carrying a side chain larger than an effect as shown in Fig. 4. Consistent with this result,
ethyl group. However, the adverse mutational effect the two W58A-bearing mutants displayed activities
of W58A turns into
a
beneficial one for for A1 and A2 much higher than the L57A/V154A
arylalkylamines as good as that of L57A.
mutant did (Fig. 5C). It is notable that L57A/W58A
V154A leads to the R values higher than unity up led to an impressive activity improvement for A3
to 5.6, indicating mild steric relaxation compared to whereas W58A/V154A failed to induce such an
L57A and W58A. It is notable that the R value for increase, consistent with the steric relaxation in the S
(
S)-D5 is close to unity, indicating that V154A pocket by L57A stronger than that by V154A as
benefits the enzyme activities for (S)-D5 and (S)-D1 observed before.
by the same degree. This result suggests that V154A
does not tell a structural difference of the substituents
To quantify the combinatorial mutation effects, we
7
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
additional contribution to the steric relaxation
achieved by L57A for n-propyl and n-butyl groups.
However, L57A/V154A attains a synergic effect for
A10 carrying a bulkier side chain, leading to 2.7 and
6
-fold increases in the R values relative to those of
L57A and V154A, respectively. It is notable that the
synergic effect was found with the L57A/V154A
mutant for all the four arylalkylamines. Such a
synergic effect was also observed for (S)-D2 with
L57A/W58A.
R-based Prediction of Triple Mutation
Based on the R-analysis results so far, we examined
whether the R-analysis could be used to predict
mutational outcome of the variant harboring triple
mutation. For -keto acids, an adverse effect of
W58A was observed for bulky substrates in alone as
well as in combination with L57A or V154A.
Unfortunately, the R-analysis is not available to
predict what mutational outcome is caused by
addition of W58A to a L57A/V154A double mutant.
Instead, the R-analysis would allow quantitative
evaluation of the mutational effects of L57A and
V154A for bulky -keto acids on a W58A
background. To this end, we carried out a modified R-
analysis in which the W58A mutant served as a
parental enzyme (Table 5). As a result, R-analysis was
carried by normalizing relative activities of the
Table 5. A modified R-analysis of double mutants for -
[
a]
keto acids using the W58A mutant as a parental enzyme.
R
Substrate
L57A/W58A
2.7 ± 0.1
W58A/V154A
1.4 ± 0.1
A7
A8
Table 4. R-analysis of the variants carrying double
12.0 ± 3.3
8.1 ± 2.2
mutations.[
a]
[a]
The reference substrate pair is (S)-D1/A6.
R
Table 6. A modified R-analysis of double mutants for
Substrate
[
a]
L57A/W58A
3.1 ± 0.2
L57A/V154A
3.7 ± 0.5
W58A/V154A
1.6 ± 0.1
ketones using the W58A mutant as a parental enzyme.
A7
R
Substrate
A8
3.2 ± 0.8
28.8 ± 8.0
28.2 ± 5.3
31.3 ± 3.9
53.8 ± 5.0
21.1 ± 2.1
10.0 ± 1.9
25.1 ± 3.2
2.2 ± 0.6
L57A/W58A
2.8 ± 0.3
W58A/V154A
1.8 ± 0.1
A9
1.6 ± 0.3
2.1 ± 0.3
A2
A10
0.04 ± 0.01
49.4 ± 6.0
9.1 ± 0.9
0.9 ± 0.1
A3
76.0 ± 6.2
2.5 ± 0.2
(
(
(
(
S)-D2
S)-D5
S)-D6
S)-D7
25.5 ± 2.9
5.1 ± 0.6
[
a]
The reference substrate pair is D8/A1.
11.7 ± 1.4
11.2 ± 2.3
13.5 ± 2.2
15.2 ± 1.6
L57A/W58A and W58A/V154A mutants with that of
the W58A mutant. Owing to the non-detectable
activities of W58A for A9 and A10, R-analysis was
performed only with A7 and A8. The R-analysis
[
a]
The reference substrate pair is (S)-D1/A6.
carried out the R-analysis of the double mutants suggests that addition of L57A or V154A to W58A
Table 4). Owing to the adverse effect of W58A for would lead to a weak beneficial effect for A7 as
(
bulky -keto acids, the two W58A-bearing mutants observed with the R-analysis of L57A and V154A on
led to R values for A8-A10 lower than those of L57A a wild-type enzyme in Table 3. In contrast, both
or V154A single mutants. In contrast, L57A/V154A mutations were predicted to confer stronger beneficial
led to high R values for the bulky -keto acids. The R effects for A8 to the W58A mutant similarly to the
values of the L57A/V154A mutant for A8 and A9 wild-type enzyme shown in Table 3. Therefore, the
were not very different from those of the L57A modified R-analysis suggests that the beneficial
mutant, indicating that V154A does not make
8
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
A9
297 ± 49 (340)
28 ± 3 (58)
38.2, 6.2, 33.4
228.6, 1.0, 13.3
A10
Arylalkylamines
(
(
(
(
(
(
S)-D1
S)-D2
S)-D3
S)-D5
S)-D6
S)-D7
Ketones
709 ± 3 (6)
1.1,
1.8,
1.9,
3.4,
1.4,
1.8,
3.1, 1.2
4.9, 4.1
10.2, 11.1
4.3, 7.0
4.6, 1.4
2.4, 1.5
741 ± 32 (500)
777 ± 10 (32000)
314 ± 11 (170)
993 ± 31 (88)
308 ± 18 (110)
effects of L57A and V154A for bulky -keto acids
are conserved even after the adverse W58A mutation.
For arylalkylamines, we found that all the three
double mutants led to beneficial mutational effects
and showed high R values. Therefore, the triple
mutant is expected to display enzyme activities for the
bulky amines higher than all the double mutants.
3.41 ± 0.08 (5)
A1
A2
1.1,
3.1,
2.5,
5.3, 0.6
14.7, 2.2
88.6, 45.9
.86 ± 0.05 ( > 1100)[d]
.83 ± 0.02 ( > 790)[d]
3
2
A3
a]
[
[
Reaction conditions were the same as those in Fig. 4.
Fold-changes represent a reaction rate of the L57A/W58A/
V154A mutant relative to those of L57A/W58A, L57A/V154A
and W58A/V154A mutants.
The number in parenthesis represents a fold-change in the
specific reaction rate relative to that of the wild-type enzyme.
Due to non-detectable activities of the wild-type enzyme for A2
and A3, the minimum fold-change was based on the detection
limit (i.e., 0.5 % activity of the wild-type enzyme for A1).
b]
For ketones, we could not determine the R values of
the single point mutations because of too low
activities of the wild-type enzyme for A2 and A3.
However, Fig. 4C and 5C indicate that W58A is a
strong driver for enhancing activities for ketones
while L57A and V154A are hard to be evaluated. To
clarify the mutational effects of L57A and V154A for
ketones, we carried out again the modified R-analysis
using the W58A mutant as a parental enzyme (Table
[
[
c]
d]
notable that the triple mutation led to a 2.5-fold
activity improvement for A3 relative to L57A/W58A.
6). The two W58A-bearing double mutants show
similar R-values for A2, suggesting that both L57A
and V154A would lead to a weak beneficial effect on
steric relaxation for A2 relative to A1. The R-values
for A3 suggest a strongly and a weakly beneficial
effect of L57A and V154A, respectively, on enzyme
activity of the W58A mutant for A3 relative to A1. In
addition to the high R value of L57A/W58A for A3,
the R value of W58A/ V154A higher than unity led us
to posit that addition of both mutations to the W58A
mutant would be worth trying to improve activities
for bulky ketones.
Motivated by the R-based prediction for bulky
amines and ketones, we prepared the triple mutant
carrying L57A/W58A/V154A and measured the
enzyme activities for various substrates in comparison
with the three double mutants (Table 7). The additive
effect of the three point mutations was remarkable to
confer dramatic activity improvements for bulky
substrates (i.e., A8-A10, D2-D3, D5-D7 and A2-A3)
to a wild-type enzyme. For example, the triple mutant
achieved 550, 32000 and over 790-fold activity
increases for A8, D3 and A3, respectively, compared
to a wild-type enzyme. It is notable that the triple
mutation led to activities for A8 and A9 higher than
those of the L57A/V154A variant. Likewise, it is
Asymmetric Synthesis of (S)-D4
To examine how much the triple mutation allows
steric relaxation for bulky substrates, we tested (S)-D4
as a probe substrate. Single point mutations did not
engender a measurable activity for (S)-D4. The triple
mutant showed a specific reaction rate of 86 ± 2 mM
-1
-1
min mM-enzyme , at 20 mM rac-D4 and 10 mM A6,
which corresponds to 5, 23 and 41-fold activity gains
relative to those of L57A/W58A, L57A/V154A and
W58A/V154A mutants, respectively. The activity of
the triple mutant for (S)-D4 corresponds to even 72 %
activity of a wild-type enzyme for (S)-D1. Docking
simulation suggests that the expanded active site
induced by the triple mutation allows productive
binding of (S)-D4 in a way different from that of (S)-
D1 in the wild-type active site (Fig. S2). The high
activity for (S)-D4 encouraged us to test A4 as an
amino acceptor. Among all the OATA mutants used
in this study, the triple mutant was the only one that
allowed a measurable activity for A4, i.e. specific
-1
-1
reaction rate = 0.06 ± 0.01 mM min mM-enzyme at
5
0 mM A4 and 500 mM D8.
To demonstrate synthetic utility of the triple mutant,
we carried out analytical-scale synthesis of (S)-D4 at
15 % (v/v) DMSO under ambient pressure (Fig. 6).
Table 7. Activity improvement of the triple mutant.
[
a]
Figure 6. Asymmetric synthesis of (S)-D4 using the triple
mutant. Reaction conditions were 5 mM A4, 500 mM D8,
Specific reaction rate
(
Fold-change relative to
Substrate
-
1
-1
[b]
mM min mM-enzyme ) 57/58, 57/154, 58/154
1
00 M triple mutant and 15 % (v/v) DMSO at 1 atm.

A6
A7
A8
-Keto acids
Reaction yield is based on the produced (S)-D4 analyzed
709 ± 3 (6)[c]
1.1, 3.1, 1.2
1.1, 2.7, 2.5
30.8, 10.0, 52.7
by HPLC.
556 ± 38 (19)
487 ± 26 (550)
Note that DMSO is a popularly used cosolvent to
increase solubility of hydrophobic ketones used as a
9
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
substrate for -TA reactions. Reaction yield reached NMR, 13C NMR and LC/MS (see Supporting
4 % at 11.5 h and then leveled off seemingly information).
4
because the reaction approached a thermodynamic
limit. The enantiopurity of the resulting (S)-D4 was
over 99 % ee.
Conclusion
Encouraged by the analytical-scale result, we
moved on to preparative-scale asymmetric synthesis Rapid creation of a desirable -TA variant is crucial
of (S)-D4 using the triple mutant. To this end, we to successful process development for asymmetric
decided to increase the DMSO content owing to the amination of a target ketone. In this study, we
low solubility of A4 in water. Solubility of A4 was demonstrated that combinatorial mutation of active
measured at 0 - 50 % (v/v) DMSO content, showing site residues, aided by R-analysis for quantitative
gradual solubility increases from 5.5 mM up to 41.5 assessment of mutational effects, allowed such a rapid
mM at 50 % DMSO (Fig. S3). Considering that construction of a desirable variant capable of
cosolvent is usually detrimental to enzyme stability, accepting even an isopropyl group in the S pocket of
we examined how the varying DMSO contents OATA. We found that the mutatable residues of
affected amination of 100 mM A4 with 500 mM D8 OATA exhibit additive effects of enzyme properties,
at 10 - 50 % (v/v) DMSO (Fig. 7). Note that A4 affording prediction of combinatorial mutation
beyond the solubility forms emulsion under the outcomes for bulky amines and ketones. However, the
DMSO contents. Reaction progress became higher up R-analysis has a limitation in predicting mutational
to 30 % DMSO seemingly owing to the solubility outcomes when combining beneficial and detrimental
increases. However, the reaction progress was slowed mutations as illustrated with bulky -keto acids.
down as the DMSO content exceeded 30 % and Nevertheless, the practical utility of the proposed
eventually led to a drastic decease at 50 %. These approach is that it does not require time-consuming
results suggest that enzyme inactivation
library construction and screening but instead exploits
prior knowledge on mutational effects of the active
site residues. Considering a very low activity of a
wild-type -TA for a target ketone in most cases, the
R-analysis would better be performed with a cognate
amine and then test the combinatorial mutation with
the target ketone.
Experimental Section
Chemicals
(
S)-D3, A3 and A5 were purchased from Alfa Aesar (Ward
Hill, MA, USA). D4 was purchased from Enamine Ltd.
Kiev, Ukraine). D8 and L-D9 were obtained from Junsei
Chemical Co. (Tokyo, Japan) and Acros Organics Co.
Geel, Belgium), respectively. Methanol, acetonitrile,
(
(
DMSO and perchloric acid were obtained from Duksan
Pure Chemicals Co. (Ansan, South Korea). All other
chemicals were purchased from Sigma-Aldrich Co. (St.
Louis, MO, USA). Materials used for preparation of culture
media were obtained from Difco (Spark, MD, USA).
Figure 7. Effect of the DMSO content on the asymmetric
amination of A4. Reaction conditions were 100 mM A4,
5
00 mM D8, 10 M triple mutant and 10 - 50 % (v/v)
DMSO at 1 atm.
caused by DMSO would overwhelm the gain in the
reaction rate by the solubility increase as the DMSO
content becomes higher than 30 %.
Site-directed mutagenesis
Based on the reaction progress results above, we Mutagenesis was carried out using
a QuikChange
performed a 50-mL preparative-scale reaction at 50 Lightning site-directed mutagenesis kit (Agilent
mM A4, 500 mM D8, 100 M triple mutant and 30 % Technologies Co.) according to an instruction manual. All
(
v/v) DMSO under reduced pressure (i.e., 0.6 bar) for the nine alanine scanning mutants and the double mutant
equilibrium shift by removing volitile acetone. After carrying L57A/ W58A were constructed in our previous
[5d,13b]
4
6-h reaction, reaction yield reached 45.6 % and the studies.
The three mutants carrying L57A/V154A,
enatiopurity of the produced (S)-D4 was > 99 % ee. It W58A/V154A and L57A/W58A/V154A mutations were
is notable that enzyme stability was remarkably high generated by V154A substitution of L57A, W58A and
during the preparative-scale reaction and less than L57A/W58A mutants, respectively, as templates. The
1
0 % activity loss was detected at the end of the V154A substitution was carried out using a forward (5′-
reaction (data not shown). Product isolation was CGGCTATCACGGTGCG ACGATTGCCTCTG-3′) and a
primer (5′-CAGAG
carried out using cation-exchange chromatography, reverse
leading to 86 mg (S)-D4 (51 % recovery yield). The GCAATCGTCGCACCGTGATAGCCG-3′). Mutagenesis
1
isolated (S)-D4 was structurally characterized by H was confirmed by DNA sequencing.
1
0
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
Protein expression and purification
the cofactor and G118/S119/T324, were found to be
conserved after the structural replacement. Second,
coordinates of arginine 417 of 5GHF were replaced by
those of arginine 415 of 4GRX following the same
procedures. Structural models of OATA mutants were
prepared by alanine substitution of the mutation sites of the
wild-type structure. Docking simulations were performed
using a CDOCKER module under a default setting (i.e.,
Escherichia coli BL21(DE3) cells transformed with a
pET28a(+) expression vector harboring a TA gene were
used for protein expression. Cultivation of E. coli
BL21(DE3) cells, overexpression of the His -tagged -TAs
6
and protein purification were carried out as described
elsewhere.[ Molar concentrations of the purified -TAs
were determined by measuring UV absorbance at 280 nm.
Detailed information on the expression plasmids and molar
extinction coefficients is shown in Table S1.
5d]
2
000 steps at 700 K for a heating step, 5000 steps at 300 K
for a cooling step and 8 Å grid extension).
Asymmetric synthesis of (S)-D4
Enzyme assay
Asymmetric synthesis of (S)-D4 was performed in a 50
mM Tris buffer (pH 9.0) at 37 °C and ambient pressure.
The reaction conditions were at 5 mM A4, 500 mM D8, 15 %
All the enzyme assays were carried out at pH 7 (50 mM
potassium phosphate buffer) in a water bath set to 37 °C.
Enzyme activities for -keto acids (i.e., A6-A11) were
measured at 10 mM amino acceptor and 10 mM (S)-D1.
Typical reaction volume was 100 L. The reaction was
allowed for 10 min and then stopped by adding 600 L
acetonitrile. Protein precipitate was removed by
centrifugation (13,000 rpm, 10 min) and the supernatant
was subjected to HPLC analysis of the produced A1. The
(
v/v) DMSO and 100 M triple mutant. To monitor the
reaction progress, aliquot (typically 10 L) of the reaction
mixture was taken at predetermined reaction times and then
4
mixed with 190 L of 4.4 % (v/v) HClO solution. After
centrifugation, the resulting supernatant was subjected to
chiral HPLC analysis of the produced (S)-D4.
reaction conditions for measurement of enzyme activities Preparative-scale synthesis of (S)-D4
for amines (i.e., D1-D7) were 10 mM (S)-amine (or 20 mM
To determine the optimal amount of DMSO for
racemic amine for D4, D6 and D7) and 10 mM A6.
Produced ketones were analyzed by HPLC.
preparative-scale synthesis of (S)-D4, small-scale reactions
were carried out at 100 mM A4, 500 mM D8, 0.1 mM PLP
and 10 M triple mutant in the presence of 10-50 % (v/v)
DMSO. The reaction volume was 1 mL. All the enzyme
reactions were conducted at pH 9 (50 mM Tris buffer) and
To measure activities of -TAs for arylalkyl ketones
i.e., A1-A4), enzyme reactions were conducted at 50 mM
(
ketone and 500 mM D8 in the presence of 15 % (v/v)
DMSO. Typical reaction volume was 0.4 mL. To obtain the
initial rates, aliquots (typically 20 L) of the reaction
mixture were taken after 1 h and mixed with 380 L of 4.4 %
3
7 °C under ambient pressure.
Preparative-scale synthesis of (S)-D4 was performed in a
00 mL reactor charged with 50 mL reaction mixture
1
(
v/v) HClO
4
solution to stop the reaction. After
containing 50 mM A4, 500 mM D8, 0.1 mM PLP, 100 M
triple mutant, 30 % (v/v) DMSO and 50 mM Tris buffer
centrifugation, the resulting supernatant was subjected to
chiral HPLC for analyses of produced amines.
(
pH 9). The preparative-scale reaction was carried out at
Molecular modeling
37 °C and 0.6 bar under mild magnetic stirring. At
predetermined reaction times, aliquot of the reaction
mixture (typically 10 L) was taken and mixed with 190
Molecular modeling was carried out using a Discovery
Studio package (version 3.5.0, Accelrys) as described
previously with minor modifications.
4
L of 4.4 % (v/v) HClO solution. The reaction progress
[5d]
Docking
was monitored by HPLC analysis of produced (S)-D4. To
measure the residual activities of the triple mutant during
the reaction, activity assay was conducted by adding
aliquot of the reaction mixture to an assay solution
containing 10 mM (S)-D1, 10 mM A6 and 50 mM
potassium phosphate buffer (pH 7.0). Produced A1 was
analyzed by HPLC.
Product isolation was carried out using cation-exchange
chromatography. The pH of the reaction mixture was
adjusted to 1.0 by adding 5 N HCl and the protein
precipitate was removed by centrifugation (10000×g, 20
min), followed by filtration through a glass-fritted filter
funnel. The resulting filtrate was loaded on a glass column
packed with Dowex 50WX8 cation-exchange resin (40 g).
Washing and elution were carried out by sequential loading
of water (200 mL) and then 10 % (v/v) ammonia solution
simulations of -keto acid and arylalkylamine were
performed with an E-PMP structure harboring an inward
conformation of an active site arginine (E-PMPin) and an E-
PLP structure harboring an outward-pointing arginine (E-
PLPout), respectively. A crystal structure of OATA (PDB
ID: 5GHF)[ corresponds to E-PMPin and thus could be
directly used for docking simulation of -keto acid. For
docking simulation of arylalkylamines, the E-PLPout
structure of OATA was constructed by structural
replacement of arginine 417, lysine 287 and PMP of the E-
PMPin structure of OATA with arginine 415 and internal
aldimine of the E-PLPout structure of PDTA (PDB ID:
5d]
GRX)[ whose sequence identity is 41 % with OATA.
17]
[13b]
4
First, coordinates of lysine 287 and PMP of 5GHF were
substituted by those of the internal aldimine of 4GRX after
parallel translation of coordinates where C of the catalytic

lysine of 4GRX coincided with that of 5GHF. To render
the transported internal aldimine closer to a catalytically
competent structure, hybridization states of the C4 and N
(
840 mL). The eluates were collected and then (S)-D4 was
extracted with n-hexane (600 mL). The resulting extractant
pool was evaporated at 50 °C and 0.45 bar, yielding clear
liquid of (S)-D4. The isolated (S)-D4 was structurally
2
atoms were changed to sp . Hydrogen bonds in a phosphate
binding cup, i.e. formed between the phosphate group of
1
13
characterized by H NMR, C NMR and LC/MS. Details
of the structural characterization are described in the
1
1
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
Supporting Information.
[6] a) K. Hirotsu, M. Goto, A. Okamoto, I. Miyahara,
Chem. Rec. 2005, 5, 160-172; b) M. D. Toney, Arch.
Biochem. Biophys. 2014, 544, 119-127.
HPLC analysis
[
7] a) J. S. Shin, B. G. Kim, J. Org. Chem. 2002, 67,
848-2853; b) U. Kaulmann, K. Smithies, M. E. B.
HPLC analyses were carried out on an Alliance system
2
(
Waters Co.) or a 1260 Infinity system (Agilent
Smith, H. C. Hailes, J. M. Ward, Enzyme Microb.
Technol. 2007, 41, 628-637; c) B. M. Nestl, S. C.
Hammer, B. A. Nebel, B. Hauer, Angew. Chem. Int.
Ed. 2014, 53, 3070-3095.
Technologies). Quantitative analyses of ketones were
performed using a Symmetry C18 column (Waters Co.).
Chiral analyses of amines were carried out using a
Crownpak CR-I(+) column (Daicel Co.). Details of the
HPLC analysis are described in the Supporting Information.
[
8] a) M. Voss, D. Das, M. Genz, A. Kumar, N. Kulkarni,
J. Kustosz, P. Kumar, U. T. Bornscheuer, M. Höhne,
ACS Catal. 2018, 8, 11524-11533; b) M. S. Weiß, I.
V. Pavlidis, P. Spurr, S. P. Hanlon, B. Wirz, H. Iding,
U. T. Bornscheuer, ChemBioChem 2017, 18, 1022-
Acknowledgements
1
026; c) I. V. Pavlidis, M. S. Weiß, M. Genz, P. Spurr,
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
S. P. Hanlon, B. Wirz, H. Iding, U. T. Bornscheuer,
Nat. Chem. 2016, 8, 1076-1082.
[
9] a) K. E. Cassimjee, B. Manta, F. Himo, Org. Biomol.
Chem. 2015, 13, 8453-8464; b) F. Steffen-Munsberg,
C. Vickers, A. Thontowi, S. Schätzle, T. Tumlirsch,
M. S. Humble, H. Land, P. Berglund, U. T.
Bornscheuer, M. Höhne ChemCatChem 2013, 5, 150-
&
Systems under the BK21 PLUS program of Korean
Ministry of Education. We are grateful to the technical
work of Saerom Park.
1
53; c) M. Höhne, S. Schätzle, H. Jochens, K. Robins,
U. T. Bornscheuer, Nat. Chem. Biol. 2010, 6, 807-
13; d) E. S. Park, J. Y. Dong, J. S. Shin, Appl.
References
8
Microbiol. Biotechnol. 2014, 98, 651-660; e) E. S.
Park, S. R. Park, S. W. Han, J. Y. Dong, J. S. Shin,
Adv. Synth. Catal. 2014, 356, 212-220; f) E.-S. Park,
M. Kim, J.-S. Shin, Appl. Microbiol. Biotechnol. 2012,
[
1] a) T. C. Nugent, Chiral Amine Synthesis: Methods,
Developments and Applications, Wiley-VCH,
Weinheim, 2010; b) U. T. Bornscheuer, G. W.
Huisman, R. J. Kazlauskas, S. Lutz, J. C. Moore, K.
Robins, Nature 2012, 484, 185-194; c) M. Breuer, K.
Ditrich, T. Habicher, B. Hauer, M. Kesseler, R.
Sturmer, T. Zelinski, Angew. Chem. Int. Ed. 2004, 43,
9
3, 2425-2435.
[
[
10] D. Koszelewski, K. Tauber, K. Faber, W. Kroutil,
Trends Biotechnol. 2010, 28, 324-332.
11] a) A. Nobili, F. Steffen-Munsberg, H. Kohls, I.
Trentin, C. Schulzke, M. Höhne, U. T. Bornscheuer,
ChemCatChem 2015, 7, 757-760; b) M. Genz, O.
Melse, S. Schmidt, C. Vickers, M. Dörr, T. van den
Bergh, H.-J. Joosten, U. T. Bornscheuer,
ChemCatChem 2016, 8, 3199-3202.
12] V. N. Malashkevich, J. J. Onuffer, J. F. Kirsch, J. N.
Jansonius, Nat. Struct. Mol. Biol. 1995, 2, 548-553.
13] a) S. W. Han, E. S. Park, J. Y. Dong, J. S. Shin, Adv.
Synth. Catal. 2015, 357, 2712-2720; b) S. W. Han, E.
S. Park, J. Y. Dong, J. S. Shin, Appl. Environ.
Microbiol. 2015, 81, 6994-7002.
14] A. W. H. Dawood, M. S. Weiß, C. Schulz, I. V.
Pavlidis, H. Iding, R. O. M. A. de Souza, U. T.
Bornscheuer, ChemCatChem 2018, 10, 3943-3949.
15] A. Iwasaki, K. Matsumoto, J. Hasegawa, Y. Yasohara,
Appl. Microbiol. Biotechnol. 2012, 93, 1563-1573.
16] E. S. Park, J. Y. Dong, J. S. Shin, Org. Biomol. Chem.
7
88-824; d) V. Gotor-Fernández, V. Gotor, Curr.
Opin. Drug Discov. Devel. 2009, 12, 784-797.
[
[
[
2] T. C. Nugent, M. El-Shazly, Adv. Synth. Catal. 2010,
3
52, 753-819.
3] D. Ghislieri, N. J. Turner, Top. Catal. 2014, 57, 284-
00.
3
[
[
4] a) M. S. Malik, E. S. Park, J. S. Shin, Appl. Microbiol.
Biotechnol. 2012, 94, 1163-1171; b) R. C. Simon, N.
Richter, E. Busto, W. Kroutil, ACS Catal. 2014, 4,
1
1
29-143; c) F. Guo, P. Berglund, Green Chem. 2017,
9, 333-360; d) S. A. Kelly, S. Pohle, S. Wharry, S.
Mix, C. C. R. Allen, T. S. Moody, B. F. Gilmore,
Chem. Rev. 2018, 118, 349-367.
[
[
5] a) C. K. Savile, J. M. Janey, E. C. Mundorff, J. C.
Moore, S. Tam, W. R. Jarvis, J. C. Colbeck, A.
Krebber, F. J. Fleitz, J. Brands, P. N. Devine, G. W.
Huisman, G. J. Hughes, Science 2010, 329, 305-309;
b) K. S. Midelfort, R. Kumar, S. Han, M. J.
Karmilowicz, K. McConnell, D. K. Gehlhaar, A.
Mistry, J. S. Chang, M. Anderson, A. Villalobos, J.
Minshull, S. Govindarajan, J. W. Wong, Protein Eng.
Des. Sel. 2013, 26, 25-33; c) A. Nobili, F. Steffen-
Munsberg, H. Kohls, I. Trentin, C. Schulzke, M.
Höhne, U. T. Bornscheuer, ChemCatChem 2015, 7,
[
[
[
2
013, 11, 6929-6933.
17] C. Rausch, A. Lerchner, A. Schiefner, A. Skerra,
Proteins 2013, 81, 774-787.
7
57-760; d) S. W. Han, J. Kim, H. S. Cho, J. S. Shin,
ACS Catal. 2017, 7, 3752-3762; e) S. W. Han, E. S.
Park, J. Y. Dong, J. S. Shin, Adv. Synth. Catal. 2015,
3
57, 1732-1740; f) K. E. Cassimjee, M. S. Humble, H.
Land, V. Abedi, P. Berglund, Org. Biomol. Chem.
012, 10, 5466-5470.
2
1
2
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.201900184
FULL PAPER
Combinatorial Mutation Analysis of -
Transaminase to Create an Engineered Variant
Capable of Asymmetric Amination of
Isobutyrophenone
Adv. Synth. Catal. Year, Volume, Page – Page
Hong-Gon Kim, Sang-Woo Han and Jong-Shik
Shin*
1
3
This article is protected by copyright. All rights reserved.