Dramatically improved catalytic activity of an artificial (S)-selective arylmalonate decarboxylase by structure-guided directed evolution

Article abstract of DOI:10.1039/c1cc11953b

Using three rounds of structure-guided directed evolution, the catalytic activity of the (S)-selective arylmalonate decarboxylase variant G74C/C188S could be increased up to 920-fold. The best variant had a 220-fold improved activity in the production of

Full text of DOI:10.1039/c1cc11953b

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
ChemComm
Cite this: Chem. Commun., 2011, 47, 7503–7505
www.rsc.org/chemcomm
COMMUNICATION
Dramatically improved catalytic activity of an artificial (S)-selective
arylmalonate decarboxylase by structure-guided directed evolution
Yusuke Miyauchi,a Robert Kourist,ab Daisuke Uemuraa and Kenji Miyamoto*a
Received 7th April 2011, Accepted 26th April 2011
DOI: 10.1039/c1cc11953b
Using three rounds of structure-guided directed evolution, the
catalytic activity of the (S)-selective arylmalonate decarboxylase
variant G74C/C188S could be increased up to 920-fold. The best
variant had a 220-fold improved activity in the production of
(S)-naproxen with excellent enantioselectivity (499% ee).
Enantiomerically pure chemicals are important building
blocks in organic chemistry and are in increasing demand in
chemical and pharmaceutical industries. Enzymes with high
enantio- and regio-selectivities are suitable for the synthesis of
optically pure compounds with few by-products, and, as such,
can be used for industrial chemical synthesis under mild
conditions, which is more beneficial for green chemistry.
Fig. 1 Asymmetric decarboxylation catalysed by AMDase.
Arylmalonate decarboxylase (AMDase, EC.4.1.1.76) is a
with good enantioselectivity (94% ee).6 In the active site of the
unique enzyme that cleaves a-aryl-a-methylmalonates into
optically pure a-arylpropionates (profens).1 Profens are an
double mutant, Cys74 donates a proton toward the re-face of
the enolate intermediate, which leads to an inverse enantio-
face selectivity of the wild-type.7 Unfortunately, however, this
variant has only 0.005% catalytic activity compared to the
wild-type enzyme. Random mutagenesis experiments resulted
in variant S36N with a 10-fold increase in activity.8 Ser36 is
situated in the periphery of the protein and its effect on the
catalytic activity is difficult to elucidate. Despite this considerable
improvement, the activity remained too low for preparative
purposes. Therefore, we were interested in improving the
activity of an (S)-selective AMDase variant, but such a task
is a notoriously difficult problem in protein engineering.
Recently, the catalytic mechanism of AMDase has been
suggested based on three-dimensional structures (PDB code
3DG9, 3IP8, 3IXL).7,9,10 The AMDase active site is composed
of a so-called oxyanion hole (Thr75, Ser76, Tyr126, Gly189),
an aryl-binding pocket (Pro14, Pro15, Gly189, Gly190), a
hydrophobic pocket (Leu40, Val43, Tyr48, Val156, Met159),
and a proton donor (Cys188 or Cys74 in variant G74C).
Accordingly, the enantioselective decarboxylation reaction is
initiated by stabilization of the pro-S carboxylate by the
oxyanion hole and destabilization of the pro-R carboxylate
by the hydrophobic pocket.
important group of physiologically active compounds since
they are active as non-steroidal anti-inflammatory drugs
(NSAIDs).2 Because the biological activity of these drugs
resides exclusively in the (S)-enantiomer, considerable effort
has been invested in developing efficient routes for their
preparation; for instance, (S)-naproxen (2c) has been
prepared via recrystallization of diastereomeric mixtures.3 The
carboxylesterase-catalyzed kinetic resolution of (R/S)-naproxen
methyl ester achieves excellent optical purity of the
product (499% enantiomeric excess (ee)).4 Nevertheless, the
AMDase-catalyzed asymmetrization of prochiral a-aryl-a-
methylmalonates gives rise to a 100% theoretical yield of profens
which is a clear advantage over the 50% yield obtained from
kinetic resolutions. Unfortunately, wild-type AMDase produces
exclusively the undesirable (R)-enantiomers.
The reaction catalyzed by AMDase proceeds through
enantioselective decarboxylation of the substrate and enantioface-
selective protonation of the planar enolate intermediate by
Cys188 (Fig. 1).5 On the basis of the structural similarity of
AMDase to cofactor-independent racemases (around 30%
homology), we have designed an AMDase variant, G74C/
C188S, that gives the desired (S)-enantiomers of 1a and 1b
To examine this further, molecular docking was used to
position phenylmalonate (1d) in the active site of a structural
model of the G74C/C188S mutant (PDB code 3IXL, Fig. 2).7
Pro14, Leu40, Val43, Tyr48, Leu77, Val156, and Met159 were
positioned within a 5 A radius of the pro-R carboxylate
such that they were likely to form a hydrophobic pocket.
a Department of Biosciences and Informatics, Keio University,
3-14-1 Hiyoshi, Kohoku-ku, Yokohama, Kanagawa, 223-8522,
Japan. E-mail: kmiyamoto@bio.keio.ac.jp; Fax: +81-45-566-1783;
Tel: +81-45-566-1786
b Institute of Chemistry of Biogenic Resources, Technische Universitat
¨
Munchen, Schulgasse 16, D-94315 Straubing, Germany
¨
c
This journal is The Royal Society of Chemistry 2011
Chem. Commun., 2011, 47, 7503–7505 7503

View Article Online
based on structural information.14 First of all, the unfavorable
effect of Ser188 should be corrected by focusing the mutagenesis
on position 188. Earlier results show that modifications of the
hydrophobic pocket can strongly increase catalytic activity,10
which can be explained by an improved destabilization of
pro-R carboxylate. We speculated that by focusing on residues
of the hydrophobic pocket, we might increase the catalytic
activity of an (S)-selective variant accordingly.
We followed a strategy of three generations of mutagenesis
and screening. The first screening generation focused
on Ser188, while the second and third focused on residues
Leu40, Val43, Tyr48, Leu77, Val156 and Met159 in the
hydrophobic pocket.
Fig. 2 Residues of the hydrophobic pocket and catalytic machinery
of G74C/C188S with phenylmalonate (1d).
Directed evolution was performed as follows. The gene
encoding the Bordetella bronchiseptica AMDase carrying a
C-terminal poly-His-tag13 was subjected to saturation
mutagenesis of targeted amino acids. Site-directed mutagenesis
with primers bearing an NNK degenerating codon was used to
construct libraries containing all 20 possible amino acids in the
desired position.14 Variants were screened for phenylmalonate
(1d) decarboxylation activity. The resulting pH shift of the
solution was visualized using the pH indicator, bromothymol
blue.9,10 In each case, a set of 112 transformants were screened
to assure that each set would statistically contain full coverage
of all 20 amino acid exchanges.14 In total, 784 variants were
screened in the first round.
Interestingly, both Cys74 and Ser188 potentially form hydrogen
bonds with the oxygen atoms of the pro-S carboxylate.
The activity of the G74C and C188S mutants is much lower
than that of the native enzyme.11 In experiments using the
wild-type and the G74C/C188S mutant in deuterium oxide, no
kinetic isotope effect was observed (Data not shown). This
finding confirms that the rate-limiting step of the reaction
catalyzed by AMDase is decarboxylation, not protonation;
this is reflected in the G74C/C188S mutant activity decrease.
We speculated that Cys74 and Ser188 undergo unfavorable
interactions with the substrate, which may disturb the
stabilization of the pro-S carboxylate and the destabilization
of the pro-R carboxylate.
Fig. 3 and Table 1 summarize the progress of the evolution
of (S)-specific activity of AMDase variants for 1d. The first
screening showed that the G74C/C188G mutant has a 5.6-fold
increase in activity compared with the G74C/C188S mutant.
The second screening generation identified the triple mutant
G74C/M159L/C188G with a 210-fold increase in activity, and
Encouraged by this observation, we decided to improve
enzyme activity by applying iterative saturation mutagenesis
Y48F/G74C/C188G with
a 23-fold increase in activity
compared with the G74C/C188S mutant. We continued the
third screening generation from G74C/M159L/C188G by
individually saturating Leu40, Val43, Tyr48, Leu77 and
Val156. Consequently, the quadruple mutant Y48F/G74C/
M159L/C188G showed a 920-fold activity increase relative
to the G74C/C188S mutant.
Next, we analyzed the substrate specificity of these variants
(Table 2). G74C/M159L/C188G had the highest specific
activity toward a-aryl-a-methylmalonates(1a–c). This variant
showed a 220-fold higher specific activity toward 1c than that
of G74C/C188S, giving rise to (S)-naproxen (2c) of excellent
optical purity (Table 3).
Fig. 3 Evolution of (S)-specific AMDase activity for 1d. Red arrow:
1st screening; green arrows: 2nd screening; blue arrows: 3rd screening.
The improvement in AMDase activity depends on the substrate
structure. Variants were optimized for 1d.
The G74C/C188G mutant showed higher activity than the
G74C/C188S mutant, indicating that the hydroxyl-group of
Ser188 or steric hindrance of this residue decreases AMDase
Table 1 Activity of AMDase variants for 1d
Variant
k
cat/sÀ1
Km/mM
kcat/Km/sÀ1 mMÀ1
Relative activitya
G74C/C188S
G74C/C188G
Y48F/G74C/C188G
G74C/M159L/C188G
Y48F/G74C/M159L/C188G
Wild-type
0.020
0.050
0.44
3.3
11
260
7.5
3.3
7.1
5.9
4.6
5.1
0.0027
0.015
0.062
0.56
2.5
1.0
5.6
23
210
920
18 000
50
a
Relative activity was calculated using: (kcat/Km of variant)/(kcat/Km of G74C/C188S).
c
7504 Chem. Commun., 2011, 47, 7503–7505
This journal is The Royal Society of Chemistry 2011

View Article Online
Table 2 Activity of AMDase variantsa
Specific activity/unit mgÀ1
indicating their decisive role in providing the hydrophobic
environment for the decarboxylation. The considerable
increase depended to some degree on the substrate structure
and differed between the model compound of screenings (1d)
and sterically-hindered prochiral a-aryl-a-methylmalonates
(1a–c). Nevertheless, activity improvements of 210 fold
for the G74C/M159L/C188G mutant and 920 fold for the
Y48F/G74C/M159L/C188G mutant are extremely high and
show how efficient protein engineering can be applied for
the optimization of biocatalysts. We were thus able to
rescue a good part of the activity which had been lost to the
inversion of enantioselectivity. In addition, we were able to
considerably improve the stereoselectivity of (S)-selective
variants, giving rise to an attractive catalyst for the production
of profens with excellent enantiopurity and yield under
sustainable conditions.
1a
1b
1c
1d
Variant
G74C/C188S
G74C/C188G
Y48F/G74C/C188G
G74C/M159L/C188G
G74C/Y48F/M159L/C188G
Wild-type
0.0015
0.010
0.014
1.3
0.63
13
0.055
0.46
0.26
9.8
1.7
120
0.023
0.18
0.13
5.1
0.91
66
0.040
0.10
0.82
6.2
24
550
a
The substrate concentration is 20 mM.
Table 3 Enantioselectivity of AMDase variantsa
Enantiomeric excess (%)
2a 2b
98(S) 91(S)
2c
Variant
G74C/C188S
G74C/C188G
Y48F/G74C/C188G
G74C/M159L/C188G
G74C/Y48F/M159L/C188G
Wild-type
91(S)
499(S)
499(S)
499(S)
499(S)
499(R)
499(S)
499(S)
499(S)
499(S)
499(R)
499(S)
499(S)
499(S)
499(S)
499(R)
Notes and references
1 (a) K. Miyamoto and H. Ohta, J. Am. Chem. Soc., 1990, 112, 4077;
(b) K. Miyamoto and H. Ohta, Eur. J. Biochem., 1992, 210, 475.
2 J. Caldwell, A. J. Hutt and S. Fournel-Gigleux, Biochem.
Pharmacol., 1988, 37, 105.
3 P. J. Harrington and E. Lodewijk, Org. Process Res. Dev., 1997, 1,
72.
a
The ee values of products were determined by chiral-phase HPLC.
activity. This variant had comparatively low enantioselectivity,
suggesting that Ser188 also interacts with the enolate inter-
mediate. The M159L mutation would cause a steric effect at
the a-position of the substrate, thus altering its binding mode.
Moreover, exchanging Tyr48 for Phe would lower the polarity
of the hydrophobic pocket. Both events can further destabilize
the pro-R carboxylate. Surprisingly, the Y48F mutation
increases the activity toward 1d but has a decreasing effect
on the activity of AMDase toward a-aryl-a-methylmalonates
(1a–c). The crystal structure (PDB code 3IXL) indicates that
Tyr48 might form hydrogen bonds with the hydroxyl group of
Ser76 in the oxyanion hole.7 The Y48F mutation conceivably
changes the orientation of Ser76, resulting in unfavorable
substrate binding for decarboxylation.
4 L. Steenkamp and D. Brady, Process Biochem., 2008, 43, 1419.
5 (a) K. Miyamoto, S. Tsuchiya and H. Ohta, J. Am. Chem. Soc.,
1992, 114, 6256; (b) K. Matoishi, M. Ueda, K. Miyamoto and
H. Ohta, J. Mol. Catal. B: Enzym., 2004, 27, 161.
6 (a) Y. Ijima, K. Matoishi, Y. Terao, N. Doi, H. Yanagawa and
H. Ohta, Chem. Commun., 2005, 877; (b) Y. Terao, Y. Ijima,
K. Miyamoto and H. Ohta, J. Mol. Catal. B: Enzym., 2007, 45, 15.
7 R. Obata and M. Nakasako, Biochemistry, 2010, 49, 1963.
8 Y. Terao, K. Miyamoto and H. Ohta, Appl. Microbiol. Biotechnol.,
2006, 73, 647.
9 K. Okrasa, C. Levy, B. Hauer, N. Baudendistel, D. Leys and
J. Micklefield, Chem.–Eur. J., 2008, 14, 6609.
10 K. Okrasa, C. Levy, M. Wilding, M. Goodall, N. Baudendistel,
B. Hauer, D. Leys and J. Micklefield, Angew. Chem., 2009, 121,
7827 (Angew. Chem., Int. Ed., 2009, 48, 7691).
11 (a) M. Miyazaki, H. Kakidani, S. Hanzawa and H. Ohta, Bull.
Chem. Soc. Jpn., 1997, 70, 2765; (b) R. Kourist, Y. Miyauchi,
D. Uemura and K. Miyamoto, Chem.–Eur. J., 2011, 17, 557.
12 M. T. Reetz, Angew. Chem., 2011, 123, 144 (Angew. Chem., Int.
Ed., 2011, 50, 138).
In conclusion, we have successfully improved the activity
of an artificial (S)-selective AMDase variant by structure-
guided directed evolution. Notably, beneficial amino acid
exchanges from hydrophobic residues Leu40, Val43, Tyr48,
Leu77, Val156 and Met159 bear hydrophobic substitutions,
13 H. Aygun, S. Wojczewski, M. Kircher and S. Rosmus, WO/2005/
¨
078111, 2005.
14 M. T. Reetz, D. Kahakeaw and R. Lohmer, ChemBioChem, 2008,
9, 1797.
c
This journal is The Royal Society of Chemistry 2011
Chem. Commun., 2011, 47, 7503–7505 7505