Structural basis for inverting the enantioselectivity of arylmalonate decarboxylase revealed by the structural analysis of the Gly74Cys/Cysl88Ser mutant in the liganded form

Article abstract of DOI:10.1021/bi9015605

Arylmalonate decarboxylase catalyzes the enantioselective decarboxylation of α-aryl-α-methylmalonate to produce optically pure α-arylpropionate. The enzyme is comprised of two α/β domains and contains an active site situated between the two domains. The site is formed by Tyr48, Gly74-Thr75-Ser76, Tyr126, and Cys188-Gly189-Gly190 residues. Since it has been observed that the Gly74Cys/Cys188Ser mutation inverts the enantioselectivity of the enzyme, we determined the crystal structure of the Gly74Cys/ Cys188Ser mutant in the liganded form at a resolution of 1.45 A to understand the structural basis for this inversion. The overall structure of the enzyme overlapped well with that of the benzylphosphonate-associated wild-type enzyme, and the mutations had little effect on the structure of the active site. A ligand molecule bound to the active site in an unusual semiplanar conformation resembling the planar enediolate reaction intermediate could be assigned as phenyl acetate. The inversion in enantioselectivity by the paired mutation is explained by the mirror symmetry between Cys74 in the mutant and Cys188 of the wild type with respect to the carbon atom in the ligand to be protonated. Comparison of the wild-type and Gly74Cys mutant crystal structures suggested that ligand binding induces a positional shift of the Cys188-Gly189-Gly190 region toward the Gly74-Thr75 pair which provides two oxyanion holes necessary to stabilize the negatively charged enediolate reaction intermediate. The ligand binding also simultaneously induces the formation of a hydrophobic cluster over the active site cleft. Thus, AMDase is proposed to have "open" and "closed" conformations of the active site that are regulated by ligand binding. These results may provide an effective strategy for the rational design to invert the enantioselectivity of enzymes.

Full text of DOI:10.1021/bi9015605

Biochemistry 2010, 49, 1963–1969 1963
DOI: 10.1021/bi9015605
Structural Basis for Inverting the Enantioselectivity of Arylmalonate Decarboxylase
Revealed by the Structural Analysis of the Gly74Cys/Cys188Ser Mutant
†,‡
in the Liganded Form
§
,§,
Rika Obata and Masayoshi Nakasako*
§
Department of Physics, Faculty of Science and Technology, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama,
Kanagawa 223-8522, Japan, and The RIKEN Harima Institute/SPring-8, 1-1-1 Kouto, Sayo-cho, Sayo-gun,
Hyogo 679-5148, Japan
Received September 7, 2009; Revised Manuscript Received February 2, 2010
ABSTRACT: Arylmalonate decarboxylase catalyzes the enantioselective decarboxylation of R-aryl-R-methyl-
malonate to produce optically pure R-arylpropionate. The enzyme is comprised of two R/β domains and
contains an active site situated between the two domains. The site is formed by Tyr48, Gly74-Thr75-Ser76,
Tyr126, and Cys188-Gly189-Gly190 residues. Since it has been observed that the Gly74Cys/Cys188Ser
mutation inverts the enantioselectivity of the enzyme, we determined the crystal structure of the Gly74Cys/
˚
Cys188Ser mutant in the liganded form at a resolution of 1.45 A to understand the structural basis for this
inversion. The overall structure of the enzyme overlapped well with that of the benzylphosphonate-associated
wild-type enzyme, and the mutations had little effect on the structure of the active site. A ligand molecule
bound to the active site in an unusual semiplanar conformation resembling the planar enediolate reaction
intermediate could be assigned as phenyl acetate. The inversion in enantioselectivity by the paired mutation is
explained by the mirror symmetry between Cys74 in the mutant and Cys188 of the wild type with respect to the
carbon atom in the ligand to be protonated. Comparison of the wild-type and Gly74Cys mutant crystal
structures suggested that ligand binding induces a positional shift of the Cys188-Gly189-Gly190 region
toward the Gly74-Thr75 pair which provides two oxyanion holes necessary to stabilize the negatively charged
enediolate reaction intermediate. The ligand binding also simultaneously induces the formation of a
hydrophobic cluster over the active site cleft. Thus, AMDase is proposed to have “open” and “closed”
conformations of the active site that are regulated by ligand binding. These results may provide an effective
strategy for the rational design to invert the enantioselectivity of enzymes.
Enantiomerically pure chemicals are important building
blocks in organic chemistry and are in increasing demand in
chemical and pharmaceutical industries. Enantioselective organo-
catalysts and asymmetric metal complexes with asymmetric
architectures have been developed for the synthesis of chiral
products from prochiral substrates with high yields (1). Natural
evolution has provided a number of enzymes with high enantio-
selectivities and positional (regio) selectivities suitable for the
synthesis of enantiomerically pure compounds with few bypro-
ducts (2). Enzymes have been used in industrial chemical synth-
esis under mild conditions, and their use provides benefits for
green chemistry.
proaches, the direct evolution method (4, 5) and rational
design (6), have been applied to invert enantiopreference by
mutating essential residues in wild-type enzymes. Multiple muta-
tions to change substrate orientation have reversed the enantios-
electivity of several enzymes (6). One strategy is pairwise
mutagenesis to exchange simply catalytic residues in mirror
symmetry to the wild type (6). Pairwise mutagenesis of indust-
rially interest enzymes has begun, and the applications are
reported for oxidase (7), elastase (8), and decarboxylase (9).
1
Arylmalonate decarboxylase (AMDase, EC 4.1.1.76) from
Alcaligenes bronchisepticus (10) is one of the enzymes to have
inverted enantioselectivity through the exchange of key catalytic
residues (9). Wild-type AMDase catalyzes the enantioselective
decarboxylation of prochiral R-aryl-R-methylmalonates to opti-
cally pure R-arylpropionates at high enantiometric excess and
yields (11) and has been shown to be dependent upon the Cys188
residue (12). The reaction proceeds through the enantioselective
decarboxylation of the substrate and the enantio-face selective
protonation of the enediolate intermediate (10) (Figure 1A).
The weakness of using enzymes stems from the lack of mirror-
image enzymes for the formation of either product enantiomer in
asymmetric synthesis (3). To overcome this, two type of ap-
†
This work was supported by grants from MEXT (15076210 and
20050030) and JSPS (1920402) to M.N. The biochemical assays using
PA were conducted under the approval of the Kanagawa Prefectural
Government.
‡
The atomic coordinates and the structure factors of the Gly74Cys/
Cys188Ser mutant were deposited in the Protein Data Bank as entry
1
3IXL.
Abbreviations: AMDase, arylmalonate decarboxylase; β-ME, β-
*
To whom correspondence should be addressed: Department of
mercaptoethanol; BnzPO, benzylphosphonate; BrPA, R-bromopheny-
lacetate; H-bond, hydrogen bond; HPLC, high-performance liquid
chromatography; MR, molecular replacement; PA, phenylacetate;
PDB, Protein Data Bank; PEG5000, polyethylene glycol 5000; PM,
phenylmalonate; rmsd, root-mean-square deviation.
Physics, Faculty of Science and Technology, Keio University, 3-14-1
Hiyoshi, Kohoku-ku, Yokohama, Kanagawa 223-8522, Japan. Tele-
phone: (81)-45-566-1713. Fax: (81)-45-566-1672. E-mail: nakasako@
phys.keio.ac.jp.
r 2010 American Chemical Society
Published on Web 02/05/2010
pubs.acs.org/Biochemistry

1
964 Biochemistry, Vol. 49, No. 9, 2010
Obata and Nakasako
F
IGURE 1: (A) Enzymatic reactions catalyzed by wild-type AMDase (10), the Gly74Cys/Cys188Ser double mutant (9), and the Gly74Cys
mutant(16). Prochiralsubstratesare enantioselectivelydecarboxylated bythe wild-type and doublemutantAMDases. The productsofthe double
mutant have an opposite chirality versus that of the wild-type enzyme. The dotted lines indicate the racemization of chiral R-substituted R-
arylacetates by the Gly74Cys mutant. The planar enediolate intermediate expected for both the decarboxylation and racemization reactions is
shown in brackets. (B) Crystal structure of the Gly74Cys/Cys188Ser mutant in the liganded form. The N-terminal (residues 1-103 and 211-230)
and C-terminal domains (residues 104-210 and 231-240) are colored pink and green, respectively. Helix RB (residues 45-65) (blue) displays a
large bend in the liganded state. Helix RC (yellow) contributes to the binding of the ligand through the residues of the N-terminal face. The colored
stick models represent the side chains of the key residues for ligand binding (Tyr48, Gly74, Cys74, Cys188, Ser188, and Tyr126). (C) Hydrophobic
cluster surrounding the active sites of the liganded Gly74Cys/Cys188Ser mutant (left) and SO -associated wild-type enzyme (right). The panels
4
illustrating the molecular structures were generated using PyMol (37).
AMDase consists of compact N- and C-terminal domains with
(BrPA), 1% (v/v) dioxane, and 70 mM HEPES (pH 7.6) was
used.
an R/β fold (13). The crystal structures of PO - and benzylpho-
4
sphonate (BnzPO)-associated wild-type enzymes demonstrate
that the active site is formed by the Cys188-Gly190 loop,
Gly74-Thr75-Ser76, and Tyr126 (14, 15). The Gly74Cys/Cys188-
Ser mutant enzyme produces an arylpropionate of the opposite
enantiomorph versus that produced by the wild type (9). In
addition, a Gly74Cys mutation results in an enzyme catalyzing
the racemization of R-arylpropionate (15).
X-ray Diffraction Data Collection, Structure Determi-
nation, and Refinement. X-ray diffraction data of Gly74Cys/
Cys188Ser mutant crystals were collected at BL26B1 (18) of
SPring-8 via the mail-in system (19). The X-ray wavelength was
˚
1.0000 A, and a MAR CCD225 detector was used. Crystals
harvested from hanging drops were set in micro dialysis cells
filled with precipitant solution. Then, they were dialyzed against
a cryo buffer containing 2.2 M ammonium sulfate, 22% (w/v)
glycerol, 5 mM BrPA, and 0.25% (w/v) ethanol (pH 6.5) for 12 h.
The collected diffraction data were processed using the HKL2000
suite (20) (Table 1). In the crystal, one mutant molecule occupied
the crystallographic asymmetric unit.
The crystal structure of the mutant was determined by the
molecular replacement (MR) method with AMoRe (21) in
the CCP4 suite (22). We used the crystal structure of the wild-
type enzyme (Supporting Information, PDB entry 3DTV) as the
search model. Crystallographic structure refinement was per-
formed for the MR solution using Refmac5 (23). The model was
constructed using turboFRODO (BioGraphics) and difference
Fourier electron density maps calculated in every refinement
round. The statistics of the refined model are listed in Table 1.
The main chain dihedral angle of Ser188 was in a disallowed
region in the Ramachandran plot because of interactions with a
ligand molecule bound to the active site (see the Results).
Stereochemical restraints for the ligand were prepared using
the libcheck program in the CCP4 suite.
Here, we determined the crystal structure of a Gly74Cys/
Cys188Ser AMDase mutant in the liganded form at a resolution
˚
of 1.45 A to reveal the structural requirements for inversion in the
enantioselective reaction. The Cys74 side chain in the mutant is in
the mirror symmetry versus the Cys188 side chain in the wild-type
enzyme. Through a comparison to known AMDase structures,
we propose that ligand binding induces an open-close motion of
the active site cleft that is responsible for the enantio-face selective
protonation.
MATERIALS AND METHODS
Protein Purification and Crystallization. The Gly74Cys/
Cys188Ser mutant and the wild-type AMDase were overex-
pressed in Escherichia coli and purified to homogeneity according
to the reported procedures (16, 17). The purified enzymes were
concentrated to 20 mg/mL in a buffer containing 10 mM Tris-
HCl, 0.5 mM EDTA, and 5 mM β-mercaptotheanol (β-ME) (pH
8
.0). Crystallization of the Gly74Cys/Cys188Ser mutant enzyme
was performed by the hanging-drop vapor-diffusion method at
93 K. Crystals appeared as stacked plates when a precipitant
2
Measurements of Enzymatic Kinetics. The decarboxyla-
tion of phenylmalonate (PM) by wild-type AMDase was mea-
sured in the presence of phenylacetate (PA). Prior to the reaction,
solution containing 1.2 M ammonium sulfate, 15% (w/v) poly-
ethylene glycol 5000 (PEG5000), 10 mM bromophenylacetate

Article
Biochemistry, Vol. 49, No. 9, 2010 1965
RESULTS
Table 1: Data Collection and Refinement Statistics for the Gly74Cys/
Cys188Ser Mutant
Structure of the AMDase Gly74Cys/Cys188Ser Mutant.
The crystal structure of the Gly74Cys/Cys188Ser mutant with a
ligand molecule bound to the active site was refined at a
Data Collection
˚
space group
P2
1
2
1
2
resolution of 1.45 A (Figure 1B). The structure of the double
˚
unit cell parameters a, b, c (A)
38.39, 63.35, 82.41
100-1.45 (1.48-1.45)
370811/35026
mutant was very similar to those of the PO -associated (14) and
4
a
˚
resolution (A)
BnzPO-associated wild-type enzymes (15) as indicated by the
no. of observed/unique reflections
completeness (%)
˚
root-mean-square deviation (rmsd) of CA atoms of ca. 0.25 A,
a
95.8 (70.8)
a
despite the different molecular packing modes of their crystals.
However, the double mutant displayed conformational differ-
ences in both the active site and helix RB when compared to the
ÆI/σ
merge
I
æ b
28.83 (5.14)
a
R
0.054 (0.246)
Refinement
wild-type and Gly74Cys mutant enzymes in the SO -associated
4
˚
state (Supporting Information). Helix RB of the double mutant
bends at Ser54 and extends to the Glu46-Tyr48 region, which is
resolution (A)
20.00-1.45
33171
no. of reflections
c
R
R
work
d
free
0.172
in a loop conformation in SO -associated enzymes (Figure 1C).
4
0.194
˚
Accompanying the bend, the Tyr48 side chain travels ca. 18 A
no. of atoms
protein
from its position in the SO -associated form.
4
1703
21
Above the active site cleft, Val43 of the β2-RB loop contacts
Thr154 and Val156 of the β6-RG loop which induces the
formation of a hydrophobic cluster covering the active site.
ligand and ion
water
248
2
˚
average B factor (A )
protein
10.8
30.7
28.9
However, in the SO -associated wild-type and Gly74Cys mutant
4
ligand and ion
water
enzymes, these two loops are separated from each other and the
hydrophobic cluster is not observed (Figure 1C). Considering the
structural similarity between the liganded double mutant and the
BnzPO-associated wild-type enzyme, the conformational differ-
ences observed are likely induced by ligand binding rather than
artifacts of crystallization. Thus, AMDase possibly adopts the
e
rmsd from ideal geometry
˚
bond lengths (A)
0.007
1.319
bond angles (deg)
f
Ramachandran regions (%)
most favored
93.8
5.2
additionally allowed
generously allowed
disallowed
“
open” conformation in the unliganded or SO -associated forms
4
0.5
and the “closed” conformation when bound to ligand.
g
0.5 (Ser188)
Structure of the Mutant Active Site. The active site
structure of the Gly74Cys/Cys188Ser mutant was unambigu-
ously modeled from the clear electron density map (Figure 2A,B).
From this model, we observed that the Cys74 residue was free
from the β-ME modification found for Cys188 of the wild-type
enzyme crystallized in the presence of β-ME (13). In addition, the
side chains of Pro14, Tyr48, Cys74, and Leu77 formed a
hydrophobic pocket together with the N-terminal face of helix
RC, and the active site of the double mutant could be super-
a
b
I
Values in parentheses are for the highest-resolution shell.
R
merge
=
P
P
P
P
hkl
i
|I
i
(hkl) - ÆI(hkl)æ|/ hkl
i
I
R
i
(hkl), where I
i
(h) is the intensity of the
|Fobs(h) - Fcalc(h)|/
P
P
c
ith observation of reflection hkl.
h), where Fobs(h) and Fcalc(h) are the observed and calculated structure
factors, respectively, of reflection h. The Rfree factor was calculated for the
% of unique reflections that were not used in the structure refinement
e
throughout (35). Root-mean-square deviations from the ideal stereo-
chemical geometry. Distribution of Ramachandran angles according to
PROCHECK (36). The dihedral angle of the Ser188 main chain is
work
=
h
h obs
F -
(
d
5
f
g
distorted by the contacts between the side chain and the ligand.
imposed on those of the PO - and BnzPO-associated wild-type
4
a 480 μL solution of 0.15 μM AMDase in 10 mM Tris-HCl
enzymes (14, 15) (Figure 2C). The Ser188 side chain was in the
same conformation as that of Cys188, and Cys74 was just fit to
that in the Gly74Cys mutant (Figure 2D). In the closed con-
formation of the active site, the side chains of Cys188 and Ser188
face the crevice, while in the open conformation, the Cys188 side
chains are confined to the hydrophobic pocket (Figure 2D,E).
Ligand binding is thought to induce the rotation of the main
chain of Cys188 (or Ser188) and its movement toward the Gly74
(or Cys74) residue.
(
pH 8.5) was mixed with 60 μL of a 23, 45, 90, or 180 mM PA
solution and the mixture was incubated for 30 min. The
decarboxylation reaction was initiated by addition of 60 μL
of a 100 mM PM solution. After 10 min, the reaction was
quenched by addition of 10 μL of 1 M HCl. The amount of PM
and PA remaining in the quenched solution was quantitatively
analyzed by a Gold high-performance liquid chromatography
(
HPLC) system (Beckman) with an ULTRASPHERE ODS
column (Beckman). The system was operated at a flow rate of
The crystal structure analysis of the double mutant revealed an
unknown ligand molecule was bound to the active site. The
electron density of the molecule suggested that it was composed
of a phenyl group (atoms C1-C6), a carboxyl group (atoms O1,
C, and O2), and a carbon atom (CR) that connected the two
groups (Figure 2A). We also determined that the molecule had a
semiplanar conformation with a slight twist at the CR atom with
a C2-C1-CR-C dihedral angle of 29.0ꢀ and resembled the
enediolate form of PA (Figure 2B). The structure model suggests
1
2
mL/min, and the eluate was monitored by absorption at
54 nm. The eluant contained 10% (v/v) aqueous acetonitrile
and 0.05% (v/v) trifluoroacetate. Under the operating con-
ditions, the retention time of the PM substrate was 3.2 min,
while PA appeared 4.8 min after injection. The relationship
between reaction velocity and the amount of synthesized
PA was calculated by subtracting the initial amount (before
the reaction) from the amount remaining in the quenched
solution.
The absorption spectra of all enzymes, PA, PM, and their
mixtures were recorded at 298 K in quartz cells using a Hitachi
U-2001 UV-vis spectrophotometer.
2
3
a distorted sp hybridization rather than sp , and it should
be noted that the electron density was never approximated as
3
the mixture of electron density from PA in the sp hybridiza-
2
tion and the enediolate intermediate in the sp hybridization.

1
966 Biochemistry, Vol. 49, No. 9, 2010
Obata and Nakasako
F
IGURE 2: (A) Stereoplot illustrating the structure of the active site in the liganded Gly74Cys/Cys188Ser mutant. The blue fish net represents the
˚
omit-refined Fobs - Fcalc difference Fourier electron density map calculated using the reflections between the Bragg spacing of 20.0 and 1.45 A and
contoured at the 4.0σ level from the average. The ligand atoms are labeled as a reference for Table 2. (B) Stick model representing the active site of
the double mutant as depicted in panel A. The dotted red lines represent H-bonds formed around the carboxylate group of the ligand, while dotted
blue lines denote the contributions of the NH groups of the oxyanion holes toward neutralizing the carboxylate groups. The liganded active site
structure is compared with the active site structures of the BnzPO-associated wild-type enzyme (PDB entry 3IP8) (15) (C), the SO
4
-associated
Gly74Cysmutant(D), and the SO -associated wild-type enzyme (PDB entry 3DTV). The structuresare superimposedsothatthe residuesforming
4
the N-terminal face of helix RC optimally overlap.
The semiplanar conformation of the ligand is unusual consider-
residues. Taken together, these findings suggested that the ligand
molecule acted as an inhibitor of AMDase. The ligand had
the chemical composition of PA and resembled the structure of
the planar enediolate intermediate expected in the decarboxyla-
tion of PM (Figures 1A and 3A), although PM was absent
from the crystallization buffer. However, BrPA degrades to
several components that likely include PA during the crystal-
lization process as determined by HPLC (data not shown).
While PA was not a substrate for either the wild type or the
double mutant, its inhibitory activity for the enzymes was
unknown.
In the presence of increasing concentrations of PA added prior
to the reaction, the apparent reaction velocity in the decarbox-
ylation of PM by wild-type AMDase was shown to decrease
(Figure 3B). The amount of PA produced from PM was very
small when the concentration of PA reached 20 mM against
0.15 mM AMDase. This indicates that PA possesses an inhibi-
tory activity on this enzyme.
ing that C -C -CR-C dihedral angles of 60-90ꢀ are usually
2
1
observed in PA molecules bound to other enzymes (24-27).
The unusual ligand conformation was stabilized by six hydro-
gen bonds (H-bonds) and 12 van der Waals contacts (Figure 2B
and Table 2). The four interacting pairs (O1-Thr75 NH,
O1-Gly189 NH, O2-Tyr126 OH, and O2-Ser76 OG) are far
from the ideal H-bond angle (28), suggesting that the Thr75 NH
and Gly189 NH pair in particular provides an oxyanion hole to
stabilize the enediolate intermediate as well as the Tyr126 OH and
Ser76 O pair, as supported by a previous report (14). The phenyl
ring of the ligand was shown to contact Pro14, Gly189, and
Gly190 (Figure 2B and Table 2).
Enzymatic Activity of AMDase in the Presence of
Phenylacetate. The strong interactions between the active site
residues and the ligand caused an unfavorable main chain
dihedral angle in Ser188, and the thermal factors of the
ligand atoms were comparable with those of the active site

Article
Biochemistry, Vol. 49, No. 9, 2010 1967
The absorption spectrum of the mixture containing the wild-
type enzyme and PM displayed an absorption maximum of
ca. 330 nm (Figure 3C). The absorption maximum of a π-electron
system shifts to longer wavelengths when the π-electron system
expands to a larger area. Thus, the observed spectra suggest that
the bound substrate PM or the re-bound PA has a π-electron
system larger than that in the phenyl rings of PM or PA in
solution. The mixture of the double mutant and PA displayed an
absorption maximum very similar to that of the mixture of the
wild type and PM (Figure 3D), suggesting the binding of PA to
the active site of the double mutant as deduced from the crystal
structure described above. If we assume the molar extinctions of
liganded PA are comparable between the experiments shown in
panels C and D of Figure 3, the absorption data imply that the
wild-type AMDase tends to form the PA-liganded state more
frequently (∼60 times) than the double mutant in the steady state,
taking the absorption at 330 nm and the concentrations of the
enzymes and ligands used.
No crystals appeared when crystallization buffer was used for
the wild-type AMDase enzyme. The results from the steady state
kinetics and crystallization suggested that the mutated active site
may be advantageous for capturing and holding PA versus the
wild-type active site and that the wild-type active site is suitable
for catching and releasing PA more frequently than the mutated
active site. In addition, when BrPA in the crystallization buffer
was replaced with PA, the double mutant failed to crystallize. It is
difficult to deny the possibility that the chemical components
from the degradation of BrPA induce or help the nucleation of
the liganded double mutant.
Table 2: Interactions in the Active Site Cleft of the Liganded Gly74Cys/
Cys188Ser Mutant
a
Ligand-Protein Interactions
˚
ligand atom protein atom distance (A)
angle (deg)
O1
O2
CR
Thr75 N
2.99
2.81
2.80
2.84
2.67
2.67
3.50
3.70
3.49
3.52
3.60
3.58
3.45
3.59
3.42
3.62
3.68
3.48
Thr75 N O1-C
101.9
Thr75 OG1 O1-C 124.5
3
3 3
Thr75 OG1
Gly189 N
Ser76 N
3
3 3
Gly189 N O1-C
124.4
122.8
127.8
3
3 3
Ser76 N O2-C
3
3 3
Ser76 OG
Tyr126 OH
Cys74 SG
Ser188 OG
Gly190 CA
Gly190 CA
Gly190 N
Pro14 CD
Pro14 CG
Gly190 N
Gly189 C
Pro14 CG
Gly189 C
Tyr126 OH
Ser76 OG O2-C
3
3 3
Tyr126 OH O2-C 120.1
3 3
3
C3
C4
C5
C6
C
a
Among Protein Atoms
˚
protein atom protein atom distance (A)
angle (deg)
Tyr48 OH
Ser76 OG
2.62
Tyr48 CZ-Tyr48
99.6
OH Ser76 OG
3
3 3
Tyr126 OH
Ser187 OG
Thr75 CB
3.58
2.81
Thr75 OG1
Ser187 CB-Ser187
139.4
OG Thr75 OG1
3
3 3
DISCUSSION
a
The upper limits to pick up H-bonds and van der Waals contacts are 3.4
The crystal structure of the Gly74Cys/Cys188Ser mutant was
˚
and 3.7 A, respectively.
˚
determined in a liganded form at a resolution of 1.45 A (Figures 1
F
IGURE 3: (A) Reaction scheme for the decarboxylation of PM. (B) Reaction velocities in the decarboxylation of PM by wild-type AMDasein the
presence of PA. The error bars were calculated from the data of three independent experiments. (C) Absorption spectra of the mixture of the wild-
type enzyme (1.5 μM) and PM (0.1 mM) in 10 mM Tris-HCl (pH 8.6). The blue dots show the spectrum measured when the enzyme and the
compound solutions were mixed, and red dots show the spectrum measured 10 min after the mixing. The green and black dots show the spectra of
the enzyme and the compound in Tris buffer at the same concentrations with the mixture. (D) Absorption spectra of the Gly74Cys/Cys188Ser
mutant (3.9 μM) and PA (3 mM) in 10 mM Tris-HCl (pH 8.6). The coloring scheme is the same as that in panel C.

1
968 Biochemistry, Vol. 49, No. 9, 2010
Obata and Nakasako
F
IGURE 4: (A) Comparison of the active site structures of the liganded Gly74Cys/Cys188Ser mutant AMDase and glutamate racemase (PDB
entry 1ZUW) (29), Cys82Ala mutant aspartate racemase (PDB entry 2DX7) (30), and diaminopimelate epimerase (PDB entry 2GKE) (31). The
dotted gray lines represent possible H-bonds between the ligand and the active site residues, while the dotted blue lines indicate the contribution
from the NHgroupsforming the N-terminalface of helices corresponding toRC inAMDase. (B) Schematic illustrationfor the proposedsubstrate
binding mode and enantioselective decarboxylation of 2-methyl-2-phenylmalonate by the wild-type AMDase.
and 2 and Table 1). The ligand was assigned as PA and possessed
a semiplanar conformation resembling the enediolate intermedi-
ate expected in the decarboxylation of PM (Figures 2 and 3).
These results allow us to speculate about how the double
mutation enables the inversion of enantioselectivity and propose
the ligand binding mode and reaction scheme.
First, it should be noted that the ligand binding mode in the
active site of the double mutant is similar to those reported in
representative enzymes of the enolase family (Figure
by Gly74, Thr75, Ser76, Tyr126, and Gly189 contribute to the
neutralization of the negative charge of the enediolate interme-
diate, as previously proposed (14).
In this binding mode, the enantio-face selective protonation,
i.e., the chirality of the final product, depends simply on the
location of the reactive SH group of Cys188 or Cys74 relative to
the substrate CR atom to be protonated. For example, in the
wild-type enzyme, Cys188 located on the si face protonates the
enediolate intermediate to give (R)-2-phenylpropionate. The
Gly74Cys mutant has two cysteine residues, Cys74 and
Cys188, on the si and re faces of the intermediate whose side
chains are expected to be in the free thiol state rather than
thiol-thiolate pairs. Therefore, these two residues are thought to
act as proton donors to produce racemic products. In the
Gly74Cys/Cys188Ser double mutant, the Cys74 side chain is in
mirror symmetry versus the Cys188 side chain of the wild-type
enzyme with regard to the CR atom of the enediolate intermedi-
ate. Thus, the thiol group of the Cys74 side chain protonates the
intermediate from the re face, while the hydroxyl group of Ser188
protonates the intermediate from the si face.
4A) (29-31). In this family, the carboxylate group of the ligand
forms H-bonds with either Asn or Thr residues on the N-terminal
face of a helix, which in AMDase corresponds to helix RC. The
carboxylate group also binds to an oxyanion hole formed
between the N-terminal face of the helix and the loop located
at the opposite side of the active site cleft.
Considering the similarities of ligand binding in the enolase
family, we propose a binding mode for 2-methyl-2-phenylmalo-
nate, a model substrate of AMDase, which is illustrated in
Figure 4B. In this model, the phenyl ring of the substrate is
sandwiched among the Gly189-Gly190 pair, Gly74, and Pro14,
and the dicarboxylate group contacts Tyr48, Gly74, Thr75,
Ser76, Tyr126, and Gly189. The binding mode and the arrange-
ment of the active site residues limit the size of the substrate and
explain why AMDase binds only ligands with CR substituents
smaller than ethyl and aryl groups smaller than naphthyl (10, 12).
In the proposed binding mode, the decarboxylation at the pro-
R side is driven by the stereoelectronic effect (32) and/or the
hydrophobic destabilization in the hydrophobic pocket formed
by the Gly74, Lue77, and Tyr48 side chains (14). The CO2
molecule produced by the decarboxylation possibly remains in
the pocket until the Tyr48 side chain moves away. Though an
enediolate intermediate is expected in the enantioselective dec-
arboxylation (10, 14), AMDase has no metal ion and/or electron
sink in the active site unlike enzymes producing enediolate
reaction intermediates (33, 34). Thus, the oxyanion holes formed
One approach to designing an enzyme with inverted enantios-
electivity is to move essential residues to the opposite face of the
active site that binds the substrate (6). Our crystal structure of the
AMDase double mutant demonstrates that this strategy allows
the rational design of enzymes if the active site conformation of
an enzyme is unchanged by mutations for the inversion.
ACKNOWLEDGMENT
We are grateful to Prof. M. Yamamoto, Dr. N. Dohmae, and
Dr. H. Miyatake of RIKEN for their help in the X-ray diffraction
experiments and Dr. G. Newton for his careful reading of and
comments about the manuscript. We also thank Prof. H. Ohta
and Dr. K. Miyamoto of Keio University for their help in sample
preparation.

Article
Biochemistry, Vol. 49, No. 9, 2010 1969
SUPPORTING INFORMATION AVAILABLE
operation of the RIKEN structural genomics beamlines at SPring-8.
J. Synchrotron Radiat. 12, 380–384.
Crystal structures of the wild-type enzyme without β-ME
modification of the Cys188 residue and the Gly74Cys mutant.
This material is available free of charge via the Internet at http://
pubs.acs.org.
1
2
9. Ueno, G., Hirose, R., Ida, K., Kumasaka, T., and Yamamoto, M.
(2004) Sample management system for a vast amount of frozen
crystals at SPring-8. J. Appl. Crystallogr. 37, 867–873.
0. Otwinowski, Z., and Minor, W. (1997) Processing of X-ray
diffraction data collected in oscillation mode. Methods Enzymol.
276, 307–326.
REFERENCES
21. Navaza, J. (1994) AMoRe: An automated package for molecular
replacement. Acta Crystallogr. A50, 157–163.
1
. Gaunt, M. J., Johansson, C. C. C., McNally, A., and Vo, N. T. (2007)
Enantioselective organocatalysis. Drug Discovery Today 12, 8–27.
. Miguel, D., Toscano, T., Woycechowsky, K. J., and Hilvert, D. (2007)
Minimalist active-site redesign: Teaching old enzymes new tricks.
Angew. Chem., Int. Ed. 46, 3212–3236.
2
2
2
2. Collaborative Computational Project, Number 4 (1994) The CCP4
suite: Programs for protein crystallography. Acta Crystallogr. D50,
2
760–763.
3. Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997) Refine-
ment of macromolecular structures by the maximum-likelihood
method. Acta Crystallogr. D53, 240–255.
4. Alkema, W. B., Hensgens, C. M., Kroezinga, E. H., de Vries, E.,
Floris, R., van der Laan, J. M., Dijkstra, B. W., and Janssen, D. B.
3
. Gerlt, J. A., and Babbitt, P. (2009) Enzyme (re)design: Lessons from
natural evolution and computation. Curr. Opin. Chem. Biol. 13, 1–9.
. May, O., Nguyen, P. T., and Arnold, F. H. (2000) Inverting enantios-
electivity by directed evolution of hydantoinase for improved produc-
4
(2000) Characterization of the β-lactam binding site of penicillin
tion of L-methionine. Nat. Biotechnol. 18, 317–320.
acylase of Escherichia coli by structural and site-directed mutagenesis
studies. Protein Eng., Des. Sel. 13, 857–863.
5
. Cherry, J. R., and Fidantsef, A. L. (2003) Directed evolution of
industrial enzymes: An update. Curr. Opin. Biotechnol. 14, 438–443.
. Mugford, P. F., Wagner, U. G., Jiang, Y., Faber, K., and Kazlauskas,
R. J. (2008) Enantiocomplementary enzymes: Classification, molec-
ular basis for their enantiopreference, and prospects for mirror-image
biotransformations. Angew. Chem., Int. Ed. 47, 8782–8793.
2
5. Alkema, W. B. L., Hensgens, C. M. H., Snijder, H. J., Keizer, E.,
Dijkstra, B. W., and Janssen, D. B. (2004) Structural and kinetic
studies on ligand binding in wild-type and active-site mutants of
penicillin acylase. Protein Eng., Des. Sel. 17, 473–480.
6
2
2
2
2
6. Duggleby, H. J., Tolley, S. P., Hill, C. P., Dodson, E. J., Dodson, G.,
and Moody, P. C. (1995) Penicillin acylase has a single-amino-acid
catalytic centre. Nature 373, 264–268.
7. Brownlee, J. M., Carlson, E., Milne, A. C., Pape, E., and Harrison,
D. H. (2006) Structural and thermodynamic studies of simple aldose
reductase-inhibitor complexes. Bioorg. Chem. 34, 424–444.
8. Matsuoka, D., and Nakasako, M. (2009) Probability distributions of
hydration water molecules around polar protein atoms obtained by a
database analysis. J. Phys. Chem. B 113, 11274–11292.
9. Ruzheinikov, S. N., Taal, M. A., Sedelnikova, S. E., Baker, P. J., and
Rice, D. W. (2005) Substrate-induced conformational changes in
Bacillus subtilis glutamate racemase and their implications for drug
discovery. Structure 13, 1707–1713.
7
. van den Heuvel, R. H. H., Fraaije, M. W., Ferrer, M., Mattevi, A.,
and van Berkel, W. J. H. (2000) Inversion of stereospecificity of
vanillyl-alcohol oxidase. Proc. Natl. Acad. Sci. U.S.A. 97, 9455–9460.
. Bartsch, S., Kourist, R., and Bornscheuer, U. T. (2008) Complete
inversion of enantioselectivity towards acetylated tertiary alcohols by
a double mutant of Bacillus subtilis Esterase. Angew. Chem., Int. Ed.
8
4
7, 1508–1511.
9
. Ijima, Y., Matoishi, K., Terao, Y., Doi, N., Yanagawa, H., and Ohta,
H. (2005) Inversion of enantioselectivity of asymmetric biocatalytic
decarboxylation by site-directed mutagenesis based on the reaction
mechanism. Chem. Commun. 239, 877–879.
1
1
1
0. Miyamoto, K., and Ohta, H. (1992) Purification and properties of a
novel arylmalonate decarboxylase from Alcaligenes bronchisepticus
KU1201. Eur. J. Biochem. 210, 475–481.
1. Miyamoto, K., and Ohta, H. (1990) Enzyme-mediated asymmetric
decarboxylation of disubstituted malonic acids. J. Am. Chem. Soc.
3
0. Ohtaki, A., Nakano, Y., Iizuka, R., Arakawa, T., Yamada, K.,
Odaka, M., and Yohda, M. (2008) Structure of aspartate racemases
complexed with a dual substrate analogue, citric acid, and implica-
tions for the reaction mechanism. Proteins: Struct., Funct., Bioinf. 70,
1
12, 4077–4078.
1167–1174.
2. Miyazaki, M., Kakidani, H., Hanzawa, S., and Ohta, H. (1997)
Cysteine 188 revealed as being critical for the enzyme activity of
arylmalonate decarboxylase by site-directed mutagenesis. Bull. Chem.
Soc. Jpn. 70, 2765–2769.
3
1. Pillai, B., Cherney, M. M., Diaper, C. M., Sutherland, A., Blanchard,
J. S., Vederas, J. C., and James, M. N. (2006) Structural insights into
stereochemical inversion by diaminopimelate epimerase: An antibac-
terial drug target. Proc. Natl. Acad. Sci. U.S.A. 103, 8668–8673.
1
3. K u€ ttner, E. B., Keim, A., Kircher, M., Rosmus, S., and Str €a ter, N.
(
2008) Active-site mobility revealed by the crystal structure of aryl-
malonate decarboxylase from Bordetella bronchiseptica. J. Mol. Biol.
77, 386–394.
32. Dunathan, H. C. (1966) Conformation and reaction specificity
in pyridoxal phosphate enzymes. Proc. Natl. Acad. Sci. U.S.A. 55,
712–716.
3
1
4. Okrasa, K., Levy, C., Hauser, B., Baudendistel, N., Leys, D., and
Micklefield, J. (2008) Structure and mechanism of an unusual malonate
decarboxylase and related racemases. Chem.;Eur. J. 14, 6609–6613.
5. Okrasa, K., Levy, C., Wilding, M., Goodall, M., Baudendistel, N.,
Hauer, B., Leys, D., and Micklefield, J. (2009) Structure-guided
directed evolution of alkenyl and arylmalonate decarboxylases.
Angew. Chem., Int. Ed. 48, 1–5.
33. Wise, E. L., Yew, W. S., Gerlt, J. A., and Rayment, I. (2003)
Structural evidence for a 1,2-enediolate intermediate in the reaction
catalyzed by 3-keto-L-gulonate 6-phosphate decarboxylase, a member
of the orotidine 5 -monophosphate decarboxylase suprafamily.
Biochemistry 42, 12133–12142.
34. Dreyer, M. K., and Schulz, G. E. (1996) Catalytic Mechanism of the
Metal-dependent Fuculose Aldolase from Escherichia coli as Derived
from the Structure. J. Mol. Biol. 259, 458.
0
1
1
6. Terao, Y., Miyamoto, K., and Ohta, H. (2006) Introduction of single
mutation changes arylmalonate decarboxylase to racemases. Chem.
Commun. 34, 3600–3602.
35. Br u€ nger, A. T. (1993) Assessment of phase accuracy by cross valida-
tion: The free R value. Methods and applications. Acta Crystallogr.
D49, 24–36.
17. Miyamoto, K., and Ohta, H. (1992) Cloning and heterogeneous
expression of
Alcaligenes bronchisepaticus KU1201. Appl. Microbiol. Biotechnol.
8, 234–238.
8. Ueno, G., Kanda, H., Kumasaka, T., and Yamamoto, M. (2005)
Beamline Scheduling Software: Administration software for automatic
a
novel arylmalonate decarboxylase gene from
36. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton,
J. M. (1993) PROCHECK: A program to check the stereochemical
quality of protein structure. J. Appl. Crystallogr. 26, 283–291.
37. DeLano, W. L. (2002) The PyMOL Molecular Graphics System,
DeLano Scientific, Palo Alto, CA.
3
1