Conversion of a PLP-dependent racemase into an aldolase by a single active site mutation

Article abstract of DOI:10.1021/ja036707d

Alanine racemase (Alr) [EC 5.1.1.1] from Geobacillus stearothermophilus is a pyridoxal 5′-phosphate-dependent enzyme that catalyzes the first committed step in bacterial cell wall biosynthesis. It is converted to an aldolase upon replacement of Tyr265, which normally serves as a catalytic base in the racemase reaction, with alanine. The Y265A mutation increases catalytic efficiency for cleavage of β-phenylserine to benzaldehyde and glycine by 2.3 × 10⁵ fold as compared to the wild-type racemase, while racemase activity is greatly decreased. Additional mutagenesis suggests that His166 may act as the base that initiates the retroaldol reaction. The Y265A mutant is highly stereoselective for (2R,3S)-phenylserine, a d-amino acid, and does not process its enantiomer. This preference is consistent with the expected binding mode of substrate in the modified active site and supports the proposal that naturally occurring d-threonine aldolases and alanine racemases derive from a common ancestor. Copyright

Full text of DOI:10.1021/ja036707d

Published on Web 08/05/2003
Conversion of a PLP-Dependent Racemase into an Aldolase by a Single
Active Site Mutation
Florian P. Seebeck and Donald Hilvert*
Laboratorium fu¨r Organische Chemie, Swiss Federal Institute of Technology, ETH Ho¨nggerberg,
CH-8093 Zu¨rich, Switzerland
Received June 16, 2003; E-mail: hilvert@org.chem.ethz.ch
Pyridoxal 5′-phosphate (PLP) serves as a cofactor for a wide
range of enzyme-catalyzed reactions, including racemizations,
transaminations, â/γ-eliminations, and aldol condensations. This
functional diversity can be attributed to the modulation and
enhancement of the cofactor’s intrinsic chemical properties at the
respective enzyme active site. Previous work has established that
subtle modifications of such an environment are often sufficient to
change either substrate1,2 or reaction3-6 specificity. In this report,
we present an example of a single amino acid substitution that
changes both substrate preference and reaction profile, turning an
Figure 1. Superposition of the PLP cofactor at the active sites of alanine
isomerase into a lyase. We find that alanine racemase (Alr) [EC
5.1.1.1] from Geobacillus stearothermophilus7 is provided with
structure shows PLP complexed as a Schiff base with the inhibitor (R)-1-
racemase7 (yellow) and L-threonine aldolase8 (gray). The alanine racemase
novel aldolase activity, and greatly decreased racemase activity,
when Tyr265 is mutated to alanine.
aminoethylphosphonic acid and residues His166 and Tyr265. In L-threonine
aldolase, PLP-L-allothreonine forms a hydrogen bond with His83.
Alanine racemase from G. stearothermophilus and L-threonine
aldolase from Thermatoga maritima8 are evolutionarily unrelated
and have completely different tertiary and quaternary structures,
substrate preferences, and reaction specificities. Nevertheless, both
initiate their catalytic cycles by forming aldimine intermediates
between PLP and their respective substrates, alanine or a â-hydroxy-
R-amino acid.9 The aldimine is subsequently converted to a quinoid
intermediate by abstraction of the CR-hydrogen in the case of alanine
racemase and by general base-promoted cleavage of the CR-Câ
bond in L-threonine aldolase. Protonation of the quinoid followed
by aminolysis liberates alanine or glycine and regenerates an internal
aldimine between PLP and an active site lysine.
Scheme 1 a
The crucial carbon-carbon bond cleavage step in threonine
aldolase is initiated by proton abstraction from the â-hydroxyl group
of the cofactor-bound substrate. A histidine, His83, which π-stacks
against the cofactor, has been assigned this role (Figure 1).8 Alanine
racemase also has an active site histidine, His166, which stacks
against the other face of the cofactor, but this residue does not
interact with the substrate directly (Figure 1). Instead, it forms a
hydrogen bond with the side chain of Tyr265 to lower the pKa of
the phenol.10 The latter has been postulated to be the base needed
to convert L- to D-alanine, and its replacement by alanine has been
shown to decrease racemase activity by more than 4 × 103-fold.11
We surmised that removal of this tyrosine would also allow an
appropriately substituted â-hydroxy-R-amino acid to bind in the
enlarged active site and form an aldimine with PLP; His166 might
then serve as a general base to initiate a retroaldol reaction (Scheme
1). To test this hypothesis, we prepared and purified wild-type
alanine racemase together with the Alr Y265A variant.12
In accord with our expectations, alanine racemase gains consid-
erable aldolase activity when Tyr265 is replaced by alanine. Table
1 contains a summary of the kinetic data obtained for several
variants of the racemase and a natural threonine aldolase. At 30
°C, Alr Y265A converts â-phenylserine to benzaldehyde and
glycine with multiple turnovers. The steady-state parameters for
this reaction are kcat ) 5.7 min-1 and KM ) 8.5 mM.13 Ten-fold
a PLP-dependent conversion of â-phenylserine to glycine and benzal-
dehyde. The general base B may be the imidazole side chain of His166
(see text).
higher activities are observed at 60 °C, as expected for an enzyme
of thermophilic origin. The apparent second-order rate constant (kcat/
Km) for the Alr Y265A-catalyzed retroaldol reaction is 2.3 × 105-
fold higher than that for wild-type alanine racemase. By this
criterion, the variant is still 103-fold less efficient than L-threonine
aldolase from Escherichia coli (eTA), but it has a rather large Km
value for substrate. Indeed, Alr Y265A has a kcat value that is only
50-times smaller than that of eTA. Furthermore, comparison of kcat
with the first-order rate constant for decomposition of PLP-â-
phenylserine to PLP-glycine in solution14 shows that the modified
active site enhances the rate of carbon-carbon bond cleavage by
a factor of at least 4 × 103.
To test whether His166 is required for the retroaldol reaction,
the double mutant Alr H166N/Y265A was prepared and character-
9
10158
J. AM. CHEM. SOC. 2003, 125, 10158-10159
10.1021/ja036707d CCC: $25.00 © 2003 American Chemical Society

C O MMU N I C A T I O N S
Table 1. Steady-State Parameters for the Conversion of
â-Phenylserine to Benzaldehyde and Glycinea
grateful to the Swiss National Foundation, the ETH Zu¨rich, and
Novartis Pharma for generous support of this work.
k
K
(mM)
kcat/K
m
cat
(min-1
m
Supporting Information Available: Experimental details for the
preparation and characterization of the proteins, kinetic measurements,
and product analysis (PDF). This material is available free of charge
via the Internet at http://pubs.acs.org.
enzyme
Alr-WT
Alr-Y265A
Alr-H166N/Y265A
eTAc
)
(M-1 min-1
0.0029b
670
11
)
5.7
0.34
278
8.5
30
0.38
7.3 × 105
References
a The proteins were assayed at 30 °C in 100 mM HEPES buffer (pH 8)
with â-phenylserine (i.e., a racemic mixture of (2S,3R)-phenylserine and
(2R,3S)-phenylserine). The Km values were calculated assuming that only
D-threo-â-phenylserine is a substrate for the enzyme (see text). b No
saturation was observed up to 200 mM â-phenylserine. The kcat/Km
parameter was therefore estimated from initial rate measurements using the
equation kcat/Km ) V0/[S][E]. c Data for E. coli L-threonine aldolase measured
under similar conditions with â-phenylserine are from Contestabile et al.24
The standard error on all kinetic parameters is e21%.
(1) Yano, T.; Oue, S.; Kagamiyama, H. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 5511-5515.
(2) Rothman, S. C.; Kirsch, J. F. J. Mol. Biol. 2003, 327, 593-608.
(3) Bertoldi, M.; Gonsalvi, M.; Contestabile, R.; Voltattorni, C. B. J. Biol.
Chem. 2002, 277, 36357-36362.
(4) Bertoldi, M.; Castellani, S.; Voltattorni, C. B. Eur. J. Biochem. 2001,
268, 2975-2981.
(5) Jackson, L. K.; Brooks, H. B.; Osterman, A. L.; Goldsmith, E. J.; Phillips,
M. A. Biochemistry 2000, 39, 11247-11257.
(6) Graber, R.; Kasper, P.; Malashkevich, V. N.; Strop, P.; Gehring, H.;
Jansonius, J. N.; Christen, P. J. Biol. Chem. 1999, 274, 31203-31208.
(7) Stamper, C. G. F.; Morollo, A. A.; Ringe, D. Biochemistry 1998, 37,
10438-10445.
ized. The kinetic parameters for this protein (Table 1) show that
removal of histidine reduces catalytic efficiency by a factor of 60,
stemming mostly from a decrease in kcat (17-fold). This result is
consistent with a possible role for the imidazole side chain as a
general base in catalysis.15 Residual activity of the double mutant
presumably reflects other features of the alanine racemase active
site which have evolved to stabilize a dianionic transition state.16
In contrast to the reaction promoted by PLP in solution, the Alr
Y265A-catalyzed retroaldol reaction is stereoselective. The rate of
reaction decreases to background level after 50% of the â-phe-
nylserine sample has been transformed to product. Analysis of the
product mixture by circular dichroism spectroscopy revealed that
(2R,3S)-phenylserine, a D-amino acid, is the preferred substrate,
whereas its enantiomer is not processed by the modified enzyme.
This preference is consistent with a binding mode that places the
CR-Câ bond of the â-hydroxy-R-amino acid orthogonal to the PLP
plane, allowing the aromatic side chain to bind in the newly created
cavity. In the case of the wild-type racemase, the CR-H bond of
L-alanine occupies an analogous position (Figure 1). The specificity
of the reengineered enzyme thus supports the proposal17 that some
naturally occurring D-threonine aldolases, in contrast to L-threonine
aldolases, are evolutionarily related to alanine racemases.
Our data show that a single point mutation is able to convert an
alanine racemase into an aldolase, changing both substrate specific-
ity and reaction profile simultaneously. While the starting enzyme
abstracts the CR-hydrogen of either D- or L-alanine, the mutant
selectively cleaves the CR-Câ bond of a D-â-hydroxy-R-amino acid
with a bulky aromatic side chain. The 2.3 × 105-fold increase in
aldolase activity coupled with the 4 × 103-fold decrease in racemase
activity underscores the ease with which the intrinsic catalytic
promiscuity of the PLP cofactor can be redirected to access a new
reaction manifold, while suppressing another. As previous studies
on both natural18,19 and directed evolution1-6,20 have shown,
conservative redesign of an enzyme active site is far from
uncommon in nature. Rather, it represents a fundamental strategy
for creating a new function that can be readily mimicked in the
laboratory. Optimization of Alr Y265A through further mutagenesis
and selection may yield useful catalysts for the enantioselective
synthesis of â-hydroxy-R-amino acids21,22 and their R-alkylated
analogues.23
(8) Kielkopf, C. L.; Burley, S. K. Biochemistry 2002, 41, 11711-11720.
(9) Mehta, P. K.; Christen, P. AdV. Enzymol. Relat. Areas Mol. Biol. 2000,
74, 129-184.
(10) Sun, S. X.; Toney, M. D. Biochemistry 1999, 38, 4058-4065.
(11) Watanabe, A.; Yoshimura, T.; Mikami, B.; Esaki, N. J. Biochem. (Tokyo)
1999, 126, 781-786. In our hands, the Y265A variant exhibited 3 × 103-
fold lower racemase activity than wild-type alanine racemase.
(12) The Alr gene was cloned by PCR into a pET22b vector using genomic
DNA from G. stearothermophilus as template. Recombinant protein was
produced as a C-terminal (His)6 fusion protein in E. coli BL21 cells and
purified to homogeneity using Ni2+-ion affinity chromatography, followed
by anion-exchange on a monoQ column. Mutations were introduced using
standard overlap-extension PCR methods (Sambrook, J.; Fritsch, E. F.;
Maniatis, T. Molecular cloning: A laboratory manual, 3rd ed.; Cold
Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 2001). The
proteins were analyzed by electrospray ionization mass spectroscopy (ESI-
MS) and shown to have the expected mass (wild-type Alr, found 44 648
( 5 Da, expected 44 644 Da; Alr Y265A, found 44 561 ( 5 Da, expected
44 566 Da; Alr Y265A/H166N, found 44 539 ( 5 Da, expected 44 543
Da).
(13) Retroaldol activity was assayed at 30 °C in 100 mM HEPES (pH 8) using
â-phenylserine (i.e., a racemic mixture of (2S,3R)-phenylserine and
(2R,3S)-phenylserine). Formation of benzaldehyde was monitored at 279
nm using a molar extinction coefficient of 1.4 × 103 M-1 cm-1 23
.
Benzaldehyde production was confirmed by HPLC, 1H NMR, and
enzymatic reduction with NADH and horse liver alcohol dehydrogenase
(Liu, J. Q.; Odani, M.; Yasuoka, T.; Dairi, T.; Itoh, N.; Kataoka, M.;
Shimizu, S.; Yamada, H. Appl. Microbiol. Biotechnol. 2000, 54, 44-51).
(14) The first-order rate constant for decomposition of pyridoxal-â-phenylserine
(0.1 M) to pyridoxal-glycine and benzaldehyde was measured by 1H NMR
in H2O at pH 8, 30 °C, as previously described (Tatsumoto, K.; Martell,
A. E. J. Am. Chem. Soc. 1978, 100, 5549-5553). The value we obtained,
2.5 × 10-5 s-1, is twice the reported value. This discrepancy presumably
reflects a solvent kinetic isotope effect, because the earlier measurements
were performed in D2O rather than in H2O.
(15) This assignment is supported by the sigmoidal kcat versus pH profile for
the Y265A variant (pKa ) 7.1 ( 0.3) and the bell-shaped pH dependence
of kcat/Km (pKa1 ) 6.8 ( 0.3; pKa2 ) 10.0 ( 0.3) (see Supporting
Information). Reliable pH-rate data for the double mutant could not be
obtained because of its low activity and instability at the pH extremes.
(16) Badet, B.; Inagaki, K.; Soda, K.; Walsh, C. T. Biochemistry 1986, 25, 5,
3275-3282.
(17) Paiardini, A.; Contestabile, R.; D’Aguanno, S.; Pascarella, S.; Bossa, F.
Biochim. Biophys. Acta 2003, 1647, 214-219.
(18) Todd, A. E.; Orengo, C. A.; Thornton, J. M. J. Mol. Biol. 2001, 307,
1113-1143.
(19) Gerlt, J. A.; Babbitt, P. C. Annu. ReV. Biochem. 2001, 70, 209-246.
(20) Joerger, A. C.; Mayer, M.; Fersht, A. R. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 5694-5699.
(21) Kimura, T.; Vassilev, V. P.; Shen, G. J.; Wong, C. H. J. Am. Chem. Soc.
1997, 119, 11734-11742.
(22) Schoemaker, H. E.; Mink, D.; Wubbolts, M. G. Science 2003, 299, 1694-
1697.
(23) Our preliminary results indicate that Alr Y265A also promotes the
retroaldol reaction of D-R-methylthreonine which has not been reported
for naturally occurring threonine aldolases.
(24) Contestabile, R.; Paiardini, A.; Pascarella, S.; di Salvo, M. L.; D’Aguanno,
S.; Bossa, F. Eur. J. Biochem. 2001, 268, 6508-6525.
Acknowledgment. We thank Brigitte Brandenberg and Bern-
hard Jaun for assistance with the NMR measurements. We are
JA036707D
9
J. AM. CHEM. SOC. VOL. 125, NO. 34, 2003 10159