Structure-based rationalization of aldolase-catalyzed asymmetric synthesis

Article abstract of DOI:10.1139/v02-094

This paper describes a structure-based approach to elucidate the stereospecificity, including inversion of enantioselectivity, of the 2-deoxyribose-5-phosphate aldolase-catalyzed asymmetric aldol addition reaction using unnatural substrates designed for t

Full text of DOI:10.1139/v02-094

643
Structure-based rationalization of aldolase-
catalyzed asymmetric synthesis
Junjie Liu, Grace DeSantis, and Chi-Huey Wong
Abstract: This paper describes a structure-based approach to elucidate the stereospecificity, including inversion of
enantioselectivity, of the 2-deoxyribose-5-phosphate aldolase-catalyzed asymmetric aldol addition reaction using unnatu-
ral substrates designed for the total synthesis of epothilones. In addition, an aldolase variant with Ser-238 being altered
for Asp was found to be 2.5 times more effective than the wild type in accepting the unphosphorylated substrate
D-glyceraldehyde. A new H-bonding interaction between the Asp-238 carboxylate and the 3-hydroxyl of the substrate
was identified and was used to rationalize the rate enhancements.
Key words: aldol reaction, 2-deoxyribose-5-phosphate aldolase, mutagenesis, inversion of enantioselectivity.
Résumé : Cette communication décrit une approche basée sur la structure d’élucider la stéréospécificité, y compris
l’inversion d’énantiosélectivité, de la réaction d’addition aldolique asymétrique catalysée par l’aldolase de 2-
désoxyribose-5-phosphate; cette étude fait appel à des substrats qui ne sont pas naturels et qui ont été développés pour
la synthèse totale des épothilones. De plus, on a observé qu’une variante de l’aldolase, portant une Asp au lieu de la
Ser-238, est 2,5 fois plus efficace que la forme naturelle pour accepter le substrat d-glycéraldéhyde non phosphorylé.
On a identifié une nouvelle interaction par liaison hydrogène entre le carboxylate de l’Asp-238 et l’hydroxyle en posi-
tion 3 du substrat; elle permet de rationaliser les augmentations de vitesses de réaction.
Mots clés : réaction aldolique, aldolase de 2-désoxyribose-5-phosphate, mutagénèse, inversion d’énantiosélectivité.
[Traduit par la Rédaction] Liu et al. 645
Introduction
forms a Schiff base with acetaldehyde to generate an
enamine intermediate, which then attacks the Re-face car-
bonyl of the incoming acceptor to form a new C–C bond
with a new stereogenic center in the R-configuration. Both
the Schiff base and the carbinol amine intermediates have
been trapped and the structures have been determined (5).
According to the structure, the C-3 phosphate of the accep-
tor substrate has H-bonding interactions with a hydrophilic
site composed of Ser-238, Ser-239, and Gly-205, and the C-
2 hydroxyl group has H-bonding interactions (5) through a
water molecule with another hydrophilic site composed of
Thr-170 and Lys-172. The C-2 hydrogen is positioned in a
hydrophobic pocket (Fig. 1). The enzyme also accepts the
corresponding unphosphorylated substrate and related struc-
tures with a much weaker affinity and activity (4).
Over the past 20 years, enzymes have been exploited as
catalysts in large-scale organic synthesis. Recent advances in
molecular genetics, protein engineering, and site-specific
modification of enzymes have further expanded the scope of
enzyme catalysis with regard to its synthetic application (1).
Most of these new developments are based on the inspiring
work of a few pioneers in the field, particularly that of Pro-
fessor Bryan Jones (2). To date, many useful transformations
based on enzymes have been developed on large scales for
the chemical and pharmaceutical industries, and new devel-
opments based on biocatalysis continue to evolve (1, 3).
One subject of current interest to us is the use of enzyme-
structure information to design new protein catalysts or new
substrates for enzymatic reactions with altered stereo-
selectivity. The enzyme employed in this study is 2-
deoxyribose-5-phosphate aldolase (DERA, EC 4.1.2.4),
which catalyzes the aldol reaction between acetaldehyde and
D-glyceraldehyde 3-phosphate to form D-2-deoxyribose-5-
phosphate (4). The structure of this enzyme has been solved
at the atomic level (~1 Å resolution) and the reaction mecha-
nism has been studied in detail (Scheme 1) (5). The enzyme
Results and discussion
In an effort to find new substrates, we found that the en-
zyme catalyzes the aldol reaction with racemic 2-methyl-3-
hydroxypropanal 1a to give the pyranose 2a (Scheme 2) (6).
Characterization of the product by oxidative conversion to
the lactone with Br2 indicates that the enzyme exhibits a re-
Received 17 December 2001. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 29 May 2002.
Dedicated to Professor J. Bryan Jones on the occasion of his 65th birthday.
J. Liu, G. DeSantis, and C.-H. Wong.1 Department of Chemistry and the Skaggs Institute for Chemical Biology, The Scripps
Research Institute, 10550 N. Torrey Pines Road, La Jolla, CA 92037, USA.
1Corresponding author (e-mail: wong@scripps.edu).
Can. J. Chem. 80: 643–645 (2002)
DOI: 10.1139/V02-094
© 2002 NRC Canada

644
Can. J. Chem. Vol. 80, 2002
Scheme 1. Highlights of the mechanism of DERA catalyzed aldol addition reaction. Shown are those steps associated with C–H cleav-
age and C–C bond formation. Asp-102 and Lys-201 act as a proton relay system to assist the proton abstraction and carbonyl
protonation mediated by the conserved water molecule. The enzyme side chains interacting with the phosphate group and the C-4
hydroxyl (corresponding to the C-2 OH of acceptor) have been identified and designated each as a hydrophilic pocket. The C-4 hydro-
gen is located in a small hydrophobic pocket without interaction with side chains.
Lys-201
Lys-201
H
+
NH3
Asp-102
Asp-102
O
O
NH2
H
H
O
O
H
H
O
H
O
H
(Z)
O
(R)
=
N
CH3
OPO3
HN
CH2
OH
Lys-167
Lys-167
Ser-238,239
H
OH
(Z)
(R)
Gly-205
=
(S)
N
OPO3
H
OH
Hydrophilic
pocket
Thr-170
Lys-172
Hydro-
phobic
Lys-167
Scheme 2. Enantioselective aldol reaction catalyzed by DERA. A reversal of enantioselectivity was observed with acceptor substrates
with small hydrophobic substituents at C-2 as indicated in the conversions of 1a–c to 2a–c, respectively.
O
OH
H
OH
O
H
(S)
(S)
(S)
(S)
O
OH
H
OH
O
CH3
R
R
H
OH
R
1a. R = CH3
2a. R = CH3
1b. R = OCH3
1c. R = CH2CH3
2b. R = OCH3
2c. R = CH2CH3
O
OH
O
H
H
OH
(R)
(S)
(R)
R
(S)
H
OH
R
O
CH3
O
OH
OH
R
H
4a. R = N3
4b. R = OH
3a. R = N3
3b R = OH
versal of enantioselectivity in the aldol reaction, accepting
the L-aldehyde as an acceptor. Similarly, reverse enantio-
selectivity was observed with the 2-methoxy and 2-ethyl de-
rivatives 1b and 1c, respectively. Alternatively, when the
racemic 2-azido or 2-hydroxy derivative was used (3a and
3b, respectively), enantioselectivity that followed that of the
natural reaction was observed, i.e., the D-enantiomer was
converted to the pyranose product (4a and 4b). In all cases,
no change in facial selectivity was observed (6). The equi-
librium of the reaction is driven by the formation of the
pyranose ring in favor of condensation (Fig. 2). With the en-
zyme structure now available, the results can be rationalized
such that the 2-methyl, 2-methoxy, or 2-ethyl groups is
bound to the hydrophobic pocket and the C-2 hydrogen is in
the hydrophilic pocket. If the methyl, methoxy, or ethyl
group was in the hydrophilic pocket, it would have a close
contact with the water molecule and the carbonyl oxygen of
Thr-170, resulting in a less favorable binding. Since the
azide or the hydroxy group is considered to be hydrophilic,
it fits into the hydrophilic pocket (Scheme 3).
In another effort to engineer the enzyme to accept
unphosphorylated neutral substrates more effectively, Ser-
238 was exchanged for Asp. It was anticipated that this
change would still retain the hydrophilic nature of the C-3
© 2002 NRC Canada

Liu et al.
645
Table 1. Kinetic parameters of wild-type DERA and the Ser238Asp variant in the retroaldol reaction.
Enzymea
WT
Substrateb
kcat (s–1)
Km (mM)
0.64 ± 0.01
57 ± 7
kcat/Km (s–1 mM–1)
106 ± 2
0.0020 ± 0.0003
DERP
DER
68 ± 1
0.107 ± 0.005
Ser238Asp
DERP
DER
0.58 ± 0.05
0.21 ± 0.01
61 ± 11
39 ± 6
0.01 ± 0.001
0.005 ± 0.0009
Note: Enzymes and kinetic parameters were determined according to the procedures previously described (5)
aWT, wild-type DERA.
bDERP, D-2-deoxyribose-5-phosphate; DER, D-2-deoxyribose.
Scheme 3. (a) Acceptor model of wild-type DERA; (b) acceptor
model of the Ser238Asp variant.
strate specificity of an enzyme by protein engineering or
substrate design. Methyl pyranose 2a has been used as a key
synthon in the total synthesis of the anti-cancer agents
epothilones (6, 7).2 Work is in progress to further utilize the
structural information to design new substrates for DERA-
catalyzed aldol addition reactions.
O
Ser-238,239
Gly-205
A)
OH
OH
H
R1 R2
Hydrophilic
Hydrophobic
Acknowledgment
We thank the National Institutes of Research for support
of this research.
O
H
B)
O
Asp-238
R1 R2
Reference
O
R
1 = CH3,OCH3, CH2CH3,
H
Hydrophilic
Hydrophobic
1. For a recent special issue on this subject, see: Nature (Lon-
don), 409, 226 (2001).
R2 = OH, N3, H
2. J.B. Jones. Tetrahedron, 42, 3351 (1986); G. DeSantis and J.B.
Jones. Curr. Opin. Biotechnol. 10, 324 (1999).
3. R.N. Patel. In Stereoselective biocatalysis. Dekker, New York.
2000, and refs. cited therein.
4. T.D. Machajewski and C.-H. Wong. Angew. Chem. Int. Ed. 39,
1352 (2000), and refs. cited therein.
5. A. Heine, G. DeSantis, J.G. Luz, M. Mitchell, C.-H. Wong,
and I.A. Wilson. Science (London), 294, 369 (2001).
6. For details, see: J. Liu and C.-H. Wong. Angew. Chem. Int.
Ed. 41, 1404 (2002).
7. K.M. Koeller and C.-H. Wong. Nature (London), 409, 232
(2001).
binding pocket, but substrates with a negative charge in the
C-3 position would be disfavored. Indeed, the Ser238Asp-
mutant enzyme is 2.5 times more active (based on the
kcat/Km value of the retroaldol reaction) than the wild-type to
D-glyceraldehyde as a substrate, as indicated in Table 1. Mo-
lecular modeling indicates that the C-3 hydroxy hydrogen
forms a 2.9–3.2 Å hydrogen bond with the Asp-238 car-
boxylate, accounting for the increase in reactivity (Fig. 3).
In summary, this study has demonstrated that with a well-
defined enzyme structure, one can effectively alter the sub-
2 General enzymatic synthesis: To a 100 mL buffer solution (0.1 M KH2PO4, pH 7.5) containing 0.1 M acceptor aldehyde and 0.3 M donor
acetaldehyde was added 3000 units of DERA. The resulting solution was stirred in the dark for 3–6 days under argon. The reaction was
quenched by addition of 2 volumes of acetone. The mixture was then stirred at 0°C for 1 h and centrifuged to remove the precipitated en-
zyme. The aqueous phase was concentrated in vacuo, and the residue was purified by flash chromatography (silica, 1:2 to 4:1 EtOAc–hex-
ane) to yield 31% of 2c. Characterized by its lactone form: [a]D = 30.2 (c 0.7, CHCl3). IR (film) (cm–1): 3424.6, 2970.0, 1719.2, 1190.4,
1114.0, 1055.2. 1H NMR (600 MHz, CDCl3) d: 4.43 (dd, J = 4.4, 11.4 Hz, 1H), 3.97 (dd, J = 7.9, 11.4 Hz, 1H), 3.96 (m, 1H), 2.85 (dd, J =
5.7, 17.1 Hz, 1H), 2.74 (d, J = 3.9 Hz, 1H), 2.55 (dd, J = 6.2, 17.6 Hz, 1H), 1.79 (m, 1H), 1.64 (m, 1H), 1.35 (m, 1H), 1.01 (t, J = 7.1 Hz,
3H). 13C NMR (150 MHz, CDCl3) d: 171.07, 69.21, 67.81, 42.32, 38.03, 21.63, 11.23. HRMS m/z calcd. for C7H12O3 (M+): 144.0786;
1
found: 167.0678 ([M + Na]+). D-aldehyde product; 3% yield. Characterized by its lactone form: [a]D = –1.0 (c 0.6, CHCl3). H NMR
(600 MHz, CDCl3) d: 4.37 (t, J = 10.9 Hz, 1H), 4.25 (dd, J = 4.9, 11.0 Hz, 1H), 4.22 (br s, 1H), 2.73 (dd of AB, J = 3.1, 18.0 Hz, 1H), 2.68
(dd of AB, J = 4.4, 18.4 Hz, 1H), 2.37 (br s, 1H), 1.86 (m, 1H), 1.48 (m, 1H), 1.35 (m, 1H), 1.00 (t, J = 7.4 Hz, 3H). 13C NMR (150 MHz,
1
CDCl3) d: 170.30, 68.91, 64.50, 39.28, 39.18, 19.60, 11.25. 4a; 47% yield. H NMR (500 MHz, CDCl3) d major isomer): 5.14 (dt, J = 2.6,
7.7 Hz, 1H), 4.22 (m, 1H), 4.13 (dd, J = 10.1, 11.9 Hz, 1H), 3.71 (dd, J = 4.8, 11.8 Hz, 1H), 3.58 (ddd, J = 2.9, 4.8, 9.9 Hz, 1H), 2.93 (d,
J = 5.2 Hz, 1H), 2.10 (ddd, J = 2.6, 10.2, 19.3Hz, 1H), 1.92 (dt, J = 3.3, 16.8 Hz); minor isomer: 5.32 (q, J = 3.3 Hz, 1H), 4.24 (m, 1H),
4.10 (dd, J = 2.6, 12.5 Hz, 1H), 3.84 (dd, J = 4.8, 12.1 Hz, 1H), 3.77 (q, J = 3.3 Hz, 1H), 1.98 (ddd, J = 3.0, 9.9, 12.8 Hz), 1.88 (dt, J = 4.0,
13.2 Hz, 1H). 13C NMR (125 MHz, CDCl3) d major isomer: 92.28, 67.12, 58.97, 56.83, 35.14; minor isomer: 91.74, 64.92, 61.07, 60.63, 35.33.
© 2002 NRC Canada