Structural determinants for the non-canonical substrate specificity of the ω-transaminase from paracoccus denitrificans

Article abstract of DOI:10.1002/adsc.201300786

Substrate binding pockets of ω-transaminase (ω-TA) consist of a large (L) pocket capable of dual recognition of hydrophobic and carboxyl substituents, and a small (S) pocket displaying a strict steric constraint that permits entry of a substituent no larger than an ethyl group. Despite the unique catalytic utility of ω-TA enabling asymmetric reductive amination of carbonyl compounds, the severe size exclusion occurring in the S pocket has limited synthetic applications of ω-TA to access structurally diverse chiral amines and amino acids. Here we report the first example of an ω-TA whose S pocket shows a non-canonical steric constraint and readily accommodates up to an n-butyl substituent. The relaxed substrate specificity of the (S)-selective ω-TA, cloned from Paracoccus denitrificans (PDTA), afforded efficient asymmetric syntheses of unnatural amino acids carrying long alkyl side chains such as lnorvaline and l-norleucine. Molecular modeling using the recently released X-ray structure of PDTA could pinpoint an exact location of the S pocket which had remained dubious. Entry of a hydrophobic substituent in the L pocket was found to have the S pocket accept up to an ethyl substituent, reminiscent of the canonical steric constraint. In contrast, binding of a carboxyl group to the L pocket induced a slight movement of V153 away from the small-pocketforming residues. The resulting structural change elicited excavation of the S pocket, leading to formation of a narrow tunnel-like structure allowing accommodation of linear alkyl groups of carboxylatebearing substrates. To verify the active site model, we introduced site-directed mutagenesis to six active site residues and examined whether the point mutations alleviated the steric constraint in the S pocket. Consistent with the molecular modeling results, the V153A variant assumed an elongated S pocket and accepted even an n-hexyl substituent. Our findings provide precise structural information on substrate binding to the active site of ω-TA, which is expected to benefit rational redesign of substrate specificity of ω-TA.

Full text of DOI:10.1002/adsc.201300786

FULL PAPERS
DOI: 10.1002/adsc.201300786
Structural Determinants for the Non-Canonical Substrate
Specificity of the w-Transaminase from Paracoccus denitrificans
+
a
+a
a
a
Eul-Soo Park, Sae-Rom Park, Sang-Woo Han, Joo-Young Dong,
a,
and Jong-Shik Shin *
a
Department of Biotechnology, Yonsei University, Shinchon-Dong 134, Seodaemun-Gu, Seoul 120-749, South Korea
Fax : (+82)-2-362-7265; phone: (+82)-2-2123-5884; e-mail: enzymo@yonsei.ac.kr
+
E.-S. Park and S.-R. Park equally contributed to the work.
Received: September 6, 2013; Published online: January 8, 2014
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201300786.
Abstract: Substrate binding pockets of w-transami- pocket accept up to an ethyl substituent, reminiscent
nase (w-TA) consist of a large (L) pocket capable of of the canonical steric constraint. In contrast, binding
dual recognition of hydrophobic and carboxyl sub- of a carboxyl group to the L pocket induced a slight
stituents, and a small (S) pocket displaying a strict movement of V153 away from the small-pocket-
steric constraint that permits entry of a substituent forming residues. The resulting structural change eli-
no larger than an ethyl group. Despite the unique cited excavation of the S pocket, leading to forma-
catalytic utility of w-TA enabling asymmetric reduc- tion of a narrow tunnel-like structure allowing ac-
tive amination of carbonyl compounds, the severe commodation of linear alkyl groups of carboxylate-
size exclusion occurring in the S pocket has limited bearing substrates. To verify the active site model,
synthetic applications of w-TA to access structurally we introduced site-directed mutagenesis to six active
diverse chiral amines and amino acids. Here we site residues and examined whether the point muta-
report the first example of an w-TA whose S pocket tions alleviated the steric constraint in the S pocket.
shows a non-canonical steric constraint and readily Consistent with the molecular modeling results, the
accommodates up to an n-butyl substituent. The re- V153A variant assumed an elongated S pocket and
laxed substrate specificity of the (S)-selective w-TA, accepted even an n-hexyl substituent. Our findings
cloned from Paracoccus denitrificans (PDTA), af- provide precise structural information on substrate
forded efficient asymmetric syntheses of unnatural binding to the active site of w-TA, which is expected
amino acids carrying long alkyl side chains such as l- to benefit rational redesign of substrate specificity of
norvaline and l-norleucine. Molecular modeling w-TA.
using the recently released X-ray structure of PDTA
could pinpoint an exact location of the S pocket Keywords: active site; asymmetric amination; site-di-
which had remained dubious. Entry of a hydrophobic rected mutagenesis; substrate specificity; w-transami-
substituent in the L pocket was found to have the S nase
Introduction
synthetic toolkit for asymmetric synthesis of enantio-
pure amines and amino acids from achiral carbonyl
[
1c–f,2]
w-Transaminase (w-TA) mediates the stereoselective precursors.
amino group transfer between an amino donor and an
Substrate specificities of more than forty w-TAs
acceptor using pyridoxal 5’-phosphate (PLP) as a pros- identified to date, including both (S)- and (R)-selec-
[1]
thetic group. The broad substrate specificity of the tive ones, turned out to share key features which
enzyme enables oxidative deamination of primary could be explained by an empirical two-binding-site
amines and amino acids as well as reductive amina- model consisting of a large (L) and a small (S) pocket
[
1c–f]
[1c,e,3]
tion of ketones, keto acids and aldehydes.
This (Scheme 1).
The L pocket displays a dual recogni-
unique catalytic property of w-TA, executable without tion mode that allows accommodation of a hydropho-
the aid of an external cofactor, has prompted recent bic substituent (Scheme 1A) as well as a carboxylate
research efforts to harness the enzyme as a powerful (Scheme 1B), reminiscent of aspartate transaminase
212
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2014, 356, 212 – 220

FULL PAPERS
Structural Determinants for the Non-Canonical Substrate Specificity
Results and Discussion
Comparison of Amino Acceptor Reactivities
To scrutinize the difference in the steric constraints of
the S pockets of OATA and PDTA, we set out to
compare amino acceptor reactivities of a-keto acids
carrying structurally diverse alkyl substituents that
served as probes to assess the steric constraint
Scheme 1. The two-binding-site model of w-TA. A) Binding
of a-methylbenzylamine to a PLP form of the enzyme
(
Table 1). As expected, the S pocket of OATA
A( E-PLP). B) Binding of 2-oxobutyrate to a pyridoxamine
C
H
T
U
N
G
T
R
E
N
N
U
N
G
5
’-phosphate form of the enzyme (E-PMP).
showed the canonical steric constraint not allowing
entry of a substituent larger than an ethyl group.
Compared with glyoxylate (1a) and pyruvate (1b), 2-
oxobutyrate (1c) showed a drastic decrease in the
The S amino acceptor reactivity and eventually relative re-
[
1a,4]
and aromatic amino acid transaminase.
pocket is characterized by a strict steric constraint activities less than 1% were detected with a-keto
that allows entry of substituents only up to an ethyl acids carrying substituents bulkier than an ethyl
group (Scheme 1B). Indeed, no naturally occurring w- group (i.e., 1d–1i). In contrast, PDTA showed a mark-
TA has been reported to be able to readily accept edly relaxed steric constraint that enabled high reac-
a substituent bulkier than an ethyl group in the S tivities of a-keto acids carrying linear alkyl substitu-
[1c,e]
pocket.
The two-binding-site model has successful- ents even up to an n-butyl group [i.e., 2-oxohexanoate
ly explained observed substrate spectrum of w-TAs, (1f)]. To our surprise, 2-oxooctanoate (1i) carrying an
which underlies canonical substrate specificity of the n-hexyl substituent displayed even 3% reactivity rela-
enzyme.
tive to 1b. However, contrary to the a-keto acids car-
In the context of asymmetric synthesis of structural- rying linear alkyl substituents, PDTA did not show
[
5]
[6]
ly diverse chiral amines and amino acids which are any activity toward branched-chain a-keto acids (1e,
important chiral building blocks of various pharma- 1g–1h). Taken together, the S pocket of PDTA
ceuticals, the severe steric constraint in the S pocket showed a relaxed steric constraint toward linear alkyl
seriously limits the product range accessible by the w- substituents which is distinct from canonical w-TAs
[7]
TA approach. To cope with this limitation, one such as OATA.
could use either protein engineering to excavate the S
[
2b,8]
pocket of naturally occurring enzymes
or public
database searching to discover a new enzyme showing
a desirable substrate specificity. The former option
was successfully implemented to create a variant of
Table 1. Amino acceptor specificities of OATA and PDTA
toward a-keto acids carrying alkyl side chains.
[
9]
(
R)-selective w-TA from Arthrobacter sp. capable of
[
a]
R
Relative reactivity [%]
accepting bulky substituents of aryl alkyl ketones
used for asymmetric synthesis of chiral amines.
[
2b]
OATA
PDTA
In this report, we aim at addressing following ques-
tions. First, is there a naturally occurring w-TA that 1a
disobeys the canonical steric constraint in the S 1b
pocket? Second, where is the exact location of the S 1c
pocket and which residues serve as a steric barrier re-
sponsible for rejecting bulky substituents? In the pre-
vious study, we identified and cloned (S)-selective w-
TAs from Ochrobactrum anthropi (OATA) and Para-
coccus denitrificans (PDTA) by public database
-H
-CH3
-CH
-(CH ) CH
-CH
-(CH ) CH
-CH
-C(CH )
-(CH ) CH
3
129Æ7
100Æ4
14Æ2
103Æ6
100Æ3
82Æ7
105Æ3
n.r.
CH
2
3
[b]
1
1
1
1
1
1
d
e
f
g
h
i
n.r.
2
2
3
3
A
H
U
G
R
N
U
G
n.r.
n.r.
3
2
77Æ8
2
3
ACHTUNGTRENN(UG CH )CH CH n.r.
n.r.
3 2 3
A
T
N
T
E
N
N
n.r.
n.r.
n.r.
3Æ1
3
3
2
5
[2c,10]
searching.
We fortuitously found that PDTA
[
[
a]
Relative reactivity represents the initial reaction rate
i.e., conversion <10%) normalized by that of 1b. Reac-
tion conditions to measure the initial rates were a-keto
could readily aminate 2-oxopentanoate as efficiently
as it did pyruvate. This finding led us to explore
whether PDTA is the first example of w-TA that devi-
ates from the canonical steric constraint.
(
acid (10 mM) and (S)-a-methylbenzylamine (20 mM) in
5
0 mM phosphate buffer (pH 7.0) at 378C.
n.r.: not reactive (i.e., relative reactivity <1%).
b]
Adv. Synth. Catal. 2014, 356, 212 – 220
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
213

Eul-Soo Park et al.
FULL PAPERS
Table 2. Amino donor specificities of OATA and PDTA
toward l-a-amino acids carrying alkyl side chains.
Table 3. Amino donor specificities of OATA and PDTA
toward (S)-a-methylbenzylamine and its analogues.
[a]
R
Relative reactivity [%]
[a]
R
Relative reactivity [%]
OATA
PDTA
2
a
-H
-CH3
-CH CH
-(CH ) CH
-CH
-(CH ) CH
3Æ1
2Æ1
100Æ6
37Æ3
81Æ2
n.r.
OATA
PDTA
l-2b
l-2c
l-2d
l-2e
l-2f
l-2g
l-2h
l-2i
100Æ5
7Æ1
(
S)-3a
-CH3
100Æ2
7Æ1
100Æ3
14Æ3
5Æ2
n.r.
2
3
[b]
(S)-3b
-CH CH
2
3
n.r.
2
2
3
3
[b]
(
(
(
S)-3c
S)-3d
S)-3e
-(CH ) CH
n.r.
2
2
3
A( CH )
C
H
T
U
N
G
T
R
E
N
N
U
N
G
n.r.
n.r.
3
2
[c]
[c]
-CH
A
H
U
G
R
N
U
G
n.r.
n.r.
3
2
23Æ4
2
3
-C(CH )
A
H
N
T
E
U
n.r.
3
3
-CH
-C(CH )
-(CH ) CH
3
ACHTUNGTRENNUNG( CH )CH CH n.r.
n.r.
3 2 3
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
n.r.
n.r.
n.r.
n.r.
[a]
3
3
Relative reactivity represents the initial reaction rate
normalized by that of (S)-3a. Reaction conditions were
(S)-amine (20 mM) and 1b (20 mM).
n.r.: not reactive (i.e., relative reactivity <1%).
Racemic amine (40 mM) was used in the initial rate mea-
surement.
2
5
[
a]
Relative reactivity represents the initial reaction rate
normalized by that of l-2b. Reaction conditions were
propanal (20 mM) and l-a-amino acid [20 mM; except l-
[
[
b]
c]
2
i (1 mm) owing to the low solubility (ꢀ1.6 mM)].
[
b]
n.r.: not reactive (i.e., relative reactivity <1%).
pocket (i.e., hydrophobic vs. carboxyl) seems to affect
spatial arrangement of the residues forming the S
pocket of PDTA.
Comparison of Amino Donor Reactivities
To confirm the relaxed steric constraint in the S
pocket of PDTA, we examined substrate specificity
toward l-a-amino acids (Table 2). Consistent with the Synthetic Utility of the Non-Canonical Substrate
reactivities of a-keto acids, OATA exhibited no activi- Specificity
ties to l-a-amino acids carrying side chains bulkier
than an ethyl group (i.e. l-2d–2i). In contrast, PDTA To demonstrate exploitation of the relaxed steric con-
showed substantial activities toward l-norvaline (l- straint of PDTA, we carried out asymmetric syntheses
2
d) and l-norleucine (l-2f) whose corresponding a- of unnatural amino acids from a-keto acids bearing
keto acids were reactive amino acceptors. Notably, a substituent bulkier than an ethyl group. To this end,
both w-TAs displayed very low reactivities of glycine isopropylamine (IPA) was chosen as an amino donor
(
2a) relative to l-alanine (l-2b) despite the higher re- owing to its cheap price and the high volatility of a re-
[2b,e,11]
activity of 1a than 1b.
sulting deamination product (i.e., acetone).
In
The L pocket is taken up by a carboxylate when the previous study, amino donor reactivity of IPA rel-
the l-a-amino acid forms a Michaelis complex with ative to (S)-3a was 7 and 43% measured with PDTA
[
12]
E-PLP. To examine whether entry of a hydrophobic and OATA, respectively. To illustrate the synthetic
substituent, instead of the carboxylate, into the L utility of PDTA over OATA, we carried out transami-
pocket might influence the steric constraint in the S nation between 1f and IPA using the two w-TAs (Fig-
pocket of PDTA, we measured amino donor reactivi- ure 1A). PDTA afforded conversion higher than 99%
ties of (S)-a-methylbenzylamine [(S)-3a] and its ana- at 4 h. Enantiomeric excess (ee) of the produced l-2f
logues in which the methyl group of (S)-3a was substi- was >99%. In contrast, very low activity of OATA
tuted by bulkier alkyl groups (Table 3). As expected, toward 1f (i.e., less than 1% of 1b) led to only 10%
OATA lacked activities to the structural analogues of conversion after 4 h despite the much higher relative
(
S)-3a where the alkyl substituents were larger than reactivity of IPA.
an ethyl group [i.e., (S)-3c–3e]. Intriguingly, unlike For the semi-preparative scale synthesis of unnatu-
the relaxed steric constraint observed with l-a-amino ral amino acids, we performed syntheses of l-2d and
acids, PDTA showed much lower relative reactivities l-2f from the corresponding a-keto acids (100 mM)
of (S)-a-ethylbenzylamine [(S)-3b] and (S)-1-phenyl- using PDTA in 10 mL reaction mixture (Figure 1B).
butylamine [(S)-3c] than l-2c and l-2d, respectively. Owing to the higher reactivity of 1d than 1f, asym-
This result suggests that binding of a hydrophobic metric synthesis of l-2d proceeded faster than that of
substituent to the L pocket restores the steric con- l-2f. After 100 min, conversions of both reactions
straint in the S pocket to a canonical state. Taken to- reached >99% and the enantiopurities of the result-
gether, the type of a substituent taking up the L ing l-amino acids were > 99% ee.
214
asc.wiley-vch.de
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2014, 356, 212 – 220

FULL PAPERS
Structural Determinants for the Non-Canonical Substrate Specificity
To provide a structural basis for the relaxed steric
constraint in the S pocket of PDTA, we performed
molecular modeling using the recently published X-
ray structure of homodimeric PDTA (PBD ID:
[
13e]
4GRX).
Although overall structures of the two
subunits were very similar (root-mean-square devia-
tion of backbone atoms=0.59 ꢁ), the active site R415
showed a significant conformational difference in the
two subunits (Figure 2A). The side chain of R415
points away from the active site in the subunit A
owing to the presence of a substrate analogue (i.e., 5-
aminopentanoate) bound to the active site. The out-
ward-facing conformation of the active site Arg was
also recently observed with other w-TAs (PDB ID:
[
9b]
3
HMU and 3FCR). In contrast, R415 in the subunit
B adopts an inward-facing conformation and conse-
quently restricts the L pocket.
We conjectured that the active sites of the subunit
A and B adopted structural arrangements competent
for binding to substrates carrying a hydrophobic and
a carboxyl group, respectively. Accordingly, we first
carried out docking of (S)-3a in the active site of the
subunit A using the Discovery Studio package (Fig-
ure 2B). The phenyl group of (S)-3a was tightly
packed in the L pocket where five aromatic amino
acids, that is, F19, W57, F85*, F86* and Y150, were
located proximally to the phenyl group and thereby
provided hydrophobic environments. Note that the as-
terisks in F85* and F86* mean that the residues come
from the other subunit. The methyl group of (S)-3a
Figure 1. Asymmetric synthesis of unnatural amino acids pointed toward a shallow pit which was surrounded
from a-keto acids and IPA. A) Conversion of 1f with l-2f by F19, F85*, Y150, V153, F321* and a phosphate
using PDTA and OATA. Reaction conditions were 1f
20 mM), IPA (50 mM) and w-TA (4 mM). B) Semi-prepara-
tive synthesis of l-2d and l-2f using PDTA. Reaction condi-
tions were a-keto acids (100 mM), IPA (200 mM) and
PDTA (40 mM) in 10 mL reaction mixture.
group of the internal aldimine, formed between PLP
and K285. Molecular modeling shows that alkyl sub-
stituents only up to an ethyl group fit in this pit [see
Figure S1 in the Supporting Information for a docking
model of (S)-3b], which is in agreement with the ex-
perimental results. Therefore, the modeling results
clearly indicate that the shallow pit serves as the S
pocket.
(
Molecular Modeling of the Substrate Binding
We previously proposed a two-binding-site model of
Substrate specificity studies performed with a-keto
w-TA based on active site mapping using structurally acids and l-a-amino acids indicated that the steric
[
3]
diverse substrates. The active site model has been in constraint in the S pocket was remarkably relaxed
accordance with available X-ray structures of w- when the L pocket was bound to a carboxylate.
[9b,13]
TAs.
However, in contrast to the L pocket readily Indeed, molecular modeling showed that the S pocket
noticeable in the active site, the exact location of the in the subunit B assumed a deep narrow pit capable
S pocket remained unclear. This was due to a failure of accepting up to an n-butyl group of l-2f (Fig-
in obtaining a reliable docking model of arylalkyl- ure 2C). Due to the extended structure, the S pocket
ACHTUNGTRENNUNGa mines, such as (S)-3a, within the active site where an forms additional contacts with the side chains of S118
active site Arg, responsible for recognition of a car- and F321*. The inward-facing R415 enables formation
boxylate, protrudes into the L pocket and thereby in- of multiple hydrogen bonds with the a-carboxylate of
[
10]
terferes with the phenyl group of (S)-3a.
The l-2f. The energetic gain coming from the H-bond for-
inward-facing conformation of the active site Arg was mation seems to drive the amino group of l-2f to be
exclusively observed in the X-ray structures of both positioned close to the C-4’ of the PLP moiety, which
subunits of the w-TAs from Pseudomonas putida is an essential spatial requirement for subsequent cat-
[
13a]
[13d]
(
PDB ID: 3A8U),
Vibrio fluvialis (3NUI)
and alytic conversion to an external aldimine. The same
[
13b]
Chromobacterium violaceum (4A6T).
spatial requirement of the amino group of (S)-3a, as
Adv. Synth. Catal. 2014, 356, 212 – 220
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
215

Eul-Soo Park et al.
FULL PAPERS
Figure 2. Molecular modeling of the active site of PDTA. A) Structural alignment of the key active site residues in subunit
A and B, colored in green and magenta, respectively. R415 is shown in a ball-and-stick representation. B) Docking of (S)-3a
(
shown as a ball-and-stick model) in subunit A. C) Docking of l-2f (shown as a ball-and-stick model) in subunit B. Green
dotted lines represent hydrogen bonds between the carboxylate of l-2f and the guanidinium group of R415. D) Structural
alignment of the small-pocket-forming residues in subunit A (colored in green) onto the docking model of l-2f in subunit B
(
colored in magenta).
shown in Figure 2B, seems to be driven by the tight subtle difference in the bulkiness of substituents en-
packing of the phenyl group in the hydrophobic L tering the S pocket, depending on the type of a sub-
pocket, owing to the structural transition of R415 into stituent bound to the L pocket.
an outward-facing conformation.
To identify amino acid residues that act as a confor-
mational switch controlling the shape of the S pocket, Structure-Guided Engineering of the S Pocket
structural alignment of the five relevant residues and
the internal aldimine was performed (Figure 2D). We Based on the molecular modeling results, we could
found that V153 was the only residue that moved pinpoint the exact location of the S pocket and reveal
back from the S pocket in the subunit B. In contrast the structural basis for the non-canonical substrate
to the slight movement of V153 beneficial for the S specificity of PDTA. To verify these findings, we car-
pocket extension, other residues and the internal ald- ried out alanine scanning mutagenesis of six active
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
imine did not make such a contribution. Therefore, site residues (i.e., F19, L56, F85*, Y150, V153 and
V153 seems to play a key role in discriminating the L417) and examined whether the point mutations into
216
asc.wiley-vch.de
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2014, 356, 212 – 220

FULL PAPERS
Structural Determinants for the Non-Canonical Substrate Specificity
Figure 5. Substrate specificity of the V153A variant. A) Re-
activities of a-keto acids carrying linear alkyl side chains.
Reaction conditions: a-keto acids (10 mM), (S)-3a (10 mM)
and w-TA (0.1 mM). B) Reactivities of (S)-3a and its ana-
logues. Reaction conditions: 1b (10 mM), (S)-amine (10 mM)
Figure 3. Alanine scanning mutagenesis to identify a residue
involved in the steric constraint in the S pocket. Relative ac-
tivity represents the initial reaction rate of the alanine
mutant normalized by that of the wild-type PDTA. Reaction
conditions were 1i (10 mM), (S)-3a (10 mM) and w-TA
and w-TA (0.1 mM). Initial rate measured at 10 mM 1b and
À1
1
0 mM (S)-3a was 375 mMmin .
(
0.2 mM). Initial rate for the wild-type enzyme was
À1
20 mMmin .
a much smaller amino acid enhanced enzyme activity that l-2i formed productive binding to the active site
toward 1i carrying an n-hexyl substituent. Among the of subunit B (i.e., proximal orientations of the carbox-
six mutations, only V153A permitted an increase in yl and the amino group of l-2i to the guanidinium
the reactivity of 1i compared with the wild-type group of R415 and C-4’ of the PLP, respectively) with-
PDTA (Figure 3). The four-fold activity enhancement out a steric clash between the linear alkyl side chain
of the V153A mutant relative to the wild-type and the small-pocket-forming residues. In contrast,
enzyme is in agreement with the molecular modeling V153 present in the wild-type enzyme engendered
results indicating that V153 would be a conformational steric clashes with the terminus of the n-hexyl group.
switch to the structural alteration of the S pocket.
To examine whether the excavation of the S pocket
To visualize how the V153A mutation led to induced by the V153A mutation promotes accommo-
a higher activity toward the substrate carrying an n- dation of alkyl chains shorter than a n-hexyl group,
hexyl substituent, we performed docking simulations we measured activities of the V153A variant toward
of l-2i and the V153A variant (Figure 4). We found a-keto acids carrying linear alkyl substituents of vary-
ing lengths (Figure 5A). Taken together with Table 1,
the V153A mutation did not affect relative reactivities
of 1c, 1d and 1f, carrying side chains that were readily
accepted by the wild-type S pocket, but improved re-
activity of 1i. This result indicates that entry of alkyl
substituents, from ethyl to n-butyl groups, in the S
pocket does not undergo steric hindrance and thereby
excavation of the S pocket did not elicit reactivity en-
hancement.
Compared with the subunit B, the S pocket in the
subunit A is sterically more restricted by the slight
protrusion of V153 toward the S pocket. Therefore, it
was expected that the V153A variant might exhibit
enhanced reactivities of arylalkylamines carrying
alkyl substituents bulkier than an ethyl group. Indeed,
the V153A variant showed 34% relative activity
toward (S)-3c, compared with (S)-3a (Figure 5B).
However, the reactivity enhancement was not ob-
served with (S)-3a analogues carrying branched alkyl
substituents [i.e., (S)-3d and 3e]. Taken together, the
Figure 4. Docking of l-2i in the subunit B of the V153A var-
iant. l-2i is shown as a CPK model where red and blue
hemispheres represent oxygen and nitrogen atoms, respec-
tively. Green dotted lines represent hydrogen bonds be-
mutational study corroborates the molecular model-
tween the carboxylate of l-2i and the R415. Thin yellow and
thick grey sticks represent the original V153 and its alanine ing results revealing that V153 controls accessibility
mutation, respectively.
of bulky substituents in the S pocket by a slight move-
Adv. Synth. Catal. 2014, 356, 212 – 220
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
217

Eul-Soo Park et al.
FULL PAPERS
ment in response to the type of a substituent bound the His
to the L pocket.
-tagged w-TA and eluted using an elution buffer
20 mM sodium phosphate, 0.5M sodium chloride, 0.5 mM
6
(
PLP, pH 7.4) with a linear gradient of imidazole. Buffer ex-
change of the purified enzyme solution was carried out by
a HiTrap desalting column (GE Healthcare) using an elu-
tion buffer (50 mM sodium phosphate, 0.15M sodium chlo-
ride and 0.5 mM PLP). When necessary, the purified
enzyme solution was concentrated by an ultrafiltration kit
Conclusions
In this study, we report the first example of a naturally
occurring w-TA that can readily accept substituents
(
Ultracel-30) purchased from Millipore Co. (Billerica,
bulkier than an ethyl group in the S pocket. The re- USA).
laxed steric constraint in the S pocket confers non-
canonical substrate specificity on PDTA, which af-
Site-Directed Mutagenesis
fords extension of a product pipeline accessible by w-
TAs. Molecular simulations enabled us to identify the Six variants of PDTA carrying a point mutation into alanine
were created by a QuikChange Lightning site-directed mu-
exact location of the S pocket and to reveal that V153
tagenesis kit (Stratagene) according to an instruction
of PDTA serves as a molecular switch controlling the
manual. Mutagenesis primers, shown in Table S1 in the Sup-
S pocket structure which is determined by a substitu-
porting Information, were designed using a primer design
ent type taking up the L pocket. This structural infor-
program (http://www.stratagene.com). The template used for
mation was successfully exploited for engineering of
the mutagenesis PCR was pET28-pdTA which was previous-
the S pocket to accept bulky substituents even up to
an n-hexyl group, otherwise up to an n-butyl group.
We expect that these findings may facilitate active
site engineering to create an w-TA variant that can
accommodate structurally diverse substrates.
[2c]
ly constructed.
Intended mutagenesis was confirmed by
DNA sequencing.
Enzyme Assay
Unless otherwise specified, enzyme assays were carried out
at 378C and pH 7 (50 mM potassium phosphate buffer).
Typical reaction volume was 200 mL and the reaction was
stopped by adding 75 mL of 16% (v/v) perchloric acid after
Experimental Section
1
0 min reaction. One unit of w-TA activity was defined as
Chemicals
the enzyme amount catalyzing the formation of 1 mmole of
acetophenone in 1 min at 10 mM (S)-3a and 10 mM 1b.
1b and 1h were obtained from Kanto Chemical Co. (Tokyo,
Japan) and Tokyo Chemical Industry Co. (Tokyo, Japan), re-
spectively. l-2b was purchased from Acros Organics (Geel,
Belgium). l-2e and l-2g were purchased from Samchun
Chemical Co. (Seoul, South Korea) and Daejung Chemical
Substrate Specificity
To examine substrate specificity, initial reaction rates were
measured. Conversions were less than 10% in the initial
rate measurements which were independently triplicated.
Reaction conditions to measure amino acceptor reactivities
were 10 mM a-keto acid and 20 mM (S)-3a in 50 mM phos-
phate buffer (pH 7.0) at 378C. For PDTA variants, concen-
tration of (S)-3a was 10 mM. Produced acetophenone was
analyzed by HPLC.
&
Metals Co. (Siheung, South Korea), respectively. (S)-3c
was purchased from Alfa Aesar (Ward Hill, USA). rac-3d
and rac-3e were purchased from Enamine Ltd. (Kiev,
Ukraine). Isopropylamine (IPA) was purchased from Junsei
Chemical Co. (Tokyo, Japan). All other chemicals were pur-
chased from Sigma Aldrich Co. (St. Louis, USA) and were
of the highest grade available.
To measure amino donor reactivities of l-a-amino acids,
amino donor (20 mM) and propanal (20 mM) were used as
substrates and the a-keto acids produced were analyzed by
HPLC. Final concentration of l-2i used in the reaction mix-
ture was 1 mM due to the low solubility. To measure amino
donor reactivities of (S)-amines, amino donor (20 mM) and
Preparation of Purified w-TA
Overexpression and purification of OATA, PDTA and var-
iants of PDTA were carried out as described previously with
[2c,10]
a few modifications.
Escherichia coli BL21 (DE3) cells
1
b (20 mM) were used and a ketone product was analyzed
transformed with the pET28a(+) expression vector harbor-
ing the w-TA gene were cultivated in LB medium contain-
by HPLC. In the case of (S)-3d and 3e, produced l-2b was
analyzed due to the commercial unavailability of the ketone
products. Racemic 3d and 3e (40 mM) were used in the re-
actions due to the commercial unavailability of the (S)-
enantiomers. For PDTA variants, 10 mM (S)-amine [except
À1
ing 50 mgmL kanamycin at 378C. Protein expression was
induced by IPTG around 0.6 OD600 and then the cells were
allowed to grow further for 10 h. The culture broth was cen-
trifuged and the resulting cell pellet was resuspended in re-
suspension buffer (50 mM Tris-HCl at pH 7, 50 mM NaCl,
(
S)-3d and 3e; 20 mM racemic amine) and 10 mM 1b were
used.
1
0
mM EDTA, 1 mM b-mercaptoethanol, 0.1 mM PMSF and
.5 mM PLP). The cells were disrupted by an ultrasonic dis-
ruptor and then centrifuged to remove cell debris.
Asymmetric Synthesis of Unnatural Amino Acids
Enzyme purification was carried out on an ꢂKTAprime
plus (GE Healthcare, USA). The cell-free extract was
loaded on a HisTrap HP column (GE Healthcare) to purify
To compare the reaction progresses of asymmetric amina-
tion of 1f using OATA and PDTA, reaction conditions were
218
asc.wiley-vch.de
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2014, 356, 212 – 220

FULL PAPERS
Structural Determinants for the Non-Canonical Substrate Specificity
2
0 mM 1f, 50 mM IPA, 0.5 mM PLP, 4 mM w-TA and 50 mM Acknowledgements
phosphate buffer (pH 7) in 0.5 mL reaction mixture.
For semi-preparative synthesis of l-2d and l-2f using
PDTA, reaction conditions were 100 mM a-keto acids (1d
or 1f), 200 mM IPA, 0.5 mM PLP, 40 mM PDTA and 50 mM
phosphate buffer (pH 7) in 10 mL reaction mixture.
Conversions were measured by analyzing residual a-keto
acid substrate by HPLC. Enantiomeric excess of the amino
acid product was measured by HPLC.
This work was supported by the Advanced Biomass R&D
Center (ABC-2011-0031358) through the National Research
Foundation of Korea funded by the Ministry of Education,
Science and Technology.
References
Analytical Methods
HPLC analyses were performed with a Waters HPLC
system (Waters, USA) or an YL9300 HPLC system (YL In-
strument Co., South Korea). Quantitative analyses of a-keto
acids and ketones using an Aminex HPX-87H column (Bio-
Rad, USA) and a Sunfire C18 HPLC column (Waters,
USA), respectively, were carried out as described elsewher-
[1] a) K. Hirotsu, M. Goto, A. Okamoto, I. Miyahara,
Chem. Rec. 2005, 5, 160–172; b) B. Y. Hwang, B. K.
Cho, H. Yun, K. Koteshwar, B. G. Kim, J. Mol. Catal.
B: Enzym. 2005, 37, 47–55; c) D. Koszelewski, K.
Tauber, K. Faber, W. Kroutil, Trends Biotechnol. 2010,
28, 324–332; d) M. Hçhne, U. T. Bornscheuer, Chem-
CatChem 2009, 1, 42–51; e) M. S. Malik, E. S. Park, J. S.
Shin, Appl. Microbiol. Biotechnol. 2012, 94, 1163–1171;
f) S. Mathew, H. Yun, ACS Catal. 2012, 2, 993–1001.
[2] a) D. Koszeleivski, I. Lavandera, D. Clay, G. M. Gue-
bitz, D. Rozzell, W. Kroutil, Angew. Chem. 2008, 120,
9477–9480; Angew. Chem. Int. Ed. 2008, 47, 9337–9340;
b) 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; c) E. Park, M.
Kim, J. S. Shin, Adv. Synth. Catal. 2010, 352, 3391–
[
2c]
e. Chiral analyses of 2d and 2f were performed by HPLC
after derivatization with Marfeyꢃs reagent as described else-
[14]
where. Determination of the analyte concentrations was
carried out using calibration curves obtained with authentic
samples. Details of the HPLC analyses are described in the
Supporting Information.
Molecular Modeling
All molecular modeling studies were performed with the
Discovery Studio package (version 3.5.0, Accelrys, USA)
using the CHARMM force field. Docking simulations of
3
398; d) M. Hçhne, S. Kꢄhl, K. Robins, U. T. Born-
(
S)-3a were carried out using the CDOCKER module
scheuer, ChemBioChem 2008, 9, 363–365; e) K. E. Cas-
simjee, C. Branneby, V. Abedi, A. Wells, P. Berglund,
Chem. Commun. 2010, 46, 5569–5571.
within the active site of the subunit A identified by the
Binding-Site module. The CDOCKER module was run at
a default setting (heating step: 700 K and 2,000 steps; cool-
ing step: 300 K and 5,000 steps; grid extension: 8 ꢁ).
[
[
[
3] J. S. Shin, B. G. Kim, J. Org. Chem. 2002, 67, 2848–
2
853.
In the case of l-2b, the aforementioned docking proce-
dures failed to generate a substrate pose assuming the pro-
ductive binding in the active site of the subunit B. There-
fore, l-2b was initially positioned in the active site with the
productive binding mode fulfilled (i.e., proximal orienta-
tions of the carboxyl and the amino group of l-2b to the
guanidinium group of R415 and C-4’ of the PLP, respective-
ly). Energy minimization of the manually positioned l-2b
was performed with only the substrate allowed to move
4] V. N. Malashkevich, J. J. Onuffer, J. F. Kirsch, J. N. Jan-
sonius, Nature Struct. Biol. 1995, 2, 548–553.
5] a) T. C. Nugent, Chiral Amine Synthesis: Methods, De-
velopments and Applications, Wiley-VCH, Weinheim,
2
010; b) M. Breuer, K. Ditrich, T. Habicher, B. Hauer,
M. Kesseler, R. Sturmer, T. Zelinski, Angew. Chem.
2
004, 116, 806–843; Angew. Chem. Int. Ed. 2004, 43,
7
88–824.
[
6] a) J. S. Ma, Chimica Oggi 2003, 21, 65–68; b) M.
Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Keßeler,
R. Stꢄrmer, T. Zelinski, Angew. Chem. 2004, 116, 806–
À1
until the RMS gradient reached 0.1 kcalmol ꢁ. The
methyl group of the energy-minimized conformation of l-2b
pointed toward the same S pocket as the methyl group of
8
43; Angew. Chem. Int. Ed. 2004, 43, 788–824; c) A. S.
Bommarius, M. Schwarm, K. Drauz, Chimia 2001, 55,
0–59; d) S. Servi, D. Tessaro, G. Pedrocchi-Fantoni,
(
S)-3a did. To obtain the docking model of l-2f, the methyl
group of l-2b was modified to an n-butyl group and then
energy minimization of l-2f was carried out.
5
Coord. Chem. Rev. 2008, 252, 715–726.
The active site model of the V153A variant was created
by substituting V153 with alanine and then by energy-mini-
mizing the alanine moiety. To obtain the docking model of
l-2i, the docking model of l-2f obtained with the wild-type
PDTA was modified to l-2i and then energy minimization
of the resulting l-2i was performed within the active site of
the V153A variant.
[
[
7] E.-S. Park, J.-Y. Dong, J.-S. Shin, ChemCatChem
DOI:10.1002/cctc.201300571, in press.
8] U. T. Bornscheuer, G. W. Huisman, R. J. Kazlauskas, S.
Lutz, J. C. Moore, K. Robins, Nature 2012, 484, 185–
1
94.
[
9] a) M. Hçhne, S. Schꢅtzle, H. Jochens, K. Robins, U. T.
Bornscheuer, Nature Chem. Biol. 2010, 6, 807–813;
b) F. Steffen-Munsberg, C. Vickers, A. Thontowi, S.
Schꢅtzle, T. Tumlirsch, M. S. Humble, H. Land, P.
Berglund, U. T. Bornscheuer, M. Hçhne, ChemCatChem
2
013, 5, 150–153.
[
10] E.-S. Park, M. Kim, J.-S. Shin, Appl. Microbiol. Bio-
technol. 2012, 93, 2425–2435.
Adv. Synth. Catal. 2014, 356, 212 – 220
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
219

Eul-Soo Park et al.
FULL PAPERS
[
11] a) M. E. B. Smith, B. H. Chen, E. G. Hibbert, U. Kaul-
mann, K. Smithies, J. L. Galman, F. Baganz, P. A.
Dalby, H. C. Hailes, G. J. Lye, J. M. Ward, J. M. Wood-
ley, M. Micheletti, Org. Process Res. Dev. 2010, 14, 99–
K. E. Cassimjee, M. H¼kansson, Y. R. Kimbung, B.
Walse, V. Abedi, H. J. Federsel, P. Berglund, D. T.
Logan, FEBS J. 2012, 279, 779–792; c) C. Sayer, M. N.
Isupov, A. Westlake, J. A. Littlechild, Acta Crystallogr.
Sect. D. Biol. Crystallogr. 2013, 69, 564–576; d) T. H.
Jang, B. Kim, O. K. Park, J. Y. Bae, B. G. Kim, H. Yun,
H. H. Park, Acta Crystallogr. Sect. F: Struct. Biol. Cryst.
Commun. 2010, 66, 923–925; e) C. Rausch, A. Lerch-
ner, A. Schiefner, A. Skerra, Proteins: Struct., Funct.,
Bioinf. 2013, 81, 774–787.
1
07; b) F. G. Mutti, C. S. Fuchs, D. Pressnitz, J. H. Sat-
tler, W. Kroutil, Adv. Synth. Catal. 2011, 353, 3227–
233.
12] E.-S. Park, J.-Y. Dong, J.-S. Shin, Org. Biomol. Chem.
013, 11, 6929–6933.
3
[
[
2
13] a) N. Watanabe, K. Sakabe, N. Sakabe, T. Higashi, K.
Sasaki, S. Aibara, Y. Morita, K. Yonaha, S. Toyama, H.
Fukutani, J. Biochem. 1989, 105, 1–3; b) M. S. Humble,
[14] E.-S. Park, M. S. Malik, J.-Y. Dong, J.-S. Shin, Chem-
CatChem 2013, 5, 1734–1738.
220
asc.wiley-vch.de
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2014, 356, 212 – 220