Resolution of chiral phosphate, phosphonate, and phosphinate esters by an enantioselective enzyme library

Article abstract of DOI:10.1021/ja0658618

An array of 16 enantiomeric pairs of chiral phosphate, phosphonate, and phosphinate esters was used to establish the breadth of the stereoselective discrimination inherent within the bacterial phosphotriesterase and 15 mutant enzymes. For each substrate,

Full text of DOI:10.1021/ja0658618

Published on Web 11/17/2006
Resolution of Chiral Phosphate, Phosphonate, and
Phosphinate Esters by an Enantioselective Enzyme Library
Charity Nowlan,§ Yingchun Li,§ Johannes C. Hermann,‡ Timothy Evans,§
Joseph Carpenter,§ Eman Ghanem,§ Brian K. Shoichet,*,‡ and Frank M. Raushel*,§
Contribution from the Department of Chemistry, P.O. Box 30012, Texas A&M UniVersity,
College Station, Texas 77842-3012, and Department of Pharmaceutical Chemistry, UniVersity of
California, San Francisco, MC 2550, 1700 4th Street, San Francisco, California 94158-2230
Received August 11, 2006; E-mail: shoichet@cgl.ucsf.edu; raushel@tamu.edu
Abstract: An array of 16 enantiomeric pairs of chiral phosphate, phosphonate, and phosphinate esters
was used to establish the breadth of the stereoselective discrimination inherent within the bacterial
phosphotriesterase and 15 mutant enzymes. For each substrate, the leaving group was 4-hydroxyac-
etophenone while the other two groups attached to the phosphorus core consisted of an asymmetric mixture
of methyl, methoxy, ethyl, ethoxy, isopropoxy, phenyl, phenoxy, cyclohexyl, and cyclohexoxy substituents.
For the wild-type enzyme, the relative rates of hydrolysis for the two enantiomers ranged from 3 to 5.4 ×
105. Various combinations of site-specific mutations within the active site were used to create modified
enzymes with alterations in their enantioselective properties. For the single-site mutant enzyme, G60A,
the stereoselectivity is enhanced relative to that of the wild-type enzyme by 1-3 orders of magnitude.
Additional mutants were obtained where the stereoselectivity is inverted relative to the wild-type enzyme
for 13 of the 16 pairs of enantiomers tested for this investigation. The most dramatic example was obtained
for the hydrolysis of 4-acetylphenyl methyl phenyl phosphate. The G60A mutant preferentially hydrolyzes
the SP-enantiomer by a factor of 3.7 × 105. The I106G/F132G/H257Y mutant preferentially hydrolyzes the
RP-enantiomer by a factor of 9.7 × 102. This represents an enantioselective discrimination of 3.6 × 108
between these two mutants, with a total of only four amino acid changes. The rate differential between the
two enantiomers for any given mutant enzyme is postulated to be governed by the degree of nonproductive
binding within the enzyme active site and stabilization of the transition state. This hypothesis is supported
by computational docking of the high-energy, pentavalent form of the substrates to modeled structures of
the mutant enzyme; the energies of the docked transition-state analogues qualitatively capture the
enantiomeric preferences of the various mutants for the different substrates. These results demonstrate
that the catalytic properties of the wild-type phosphotriesterase can be exploited for the kinetic resolution
of a wide range of phosphate, phosphonate, and phosphinate esters and that the active site of this enzyme
is remarkably amenable to structural perturbations via amino acid substitution.
Introduction
possessed by enzymes are the extraordinary rate enhancements
and the high degree of selectivity for chiral centers. However,
Enzymes have become powerful tools in the synthesis of
value-added chemicals from cheaper and more readily available
starting materials.1 For example, the catalytic activity of xylose
isomerase is currently being exploited for the preparation of 3
million tons of high-fructose corn syrup per year via the
isomerization of D-glucose.2 D-Hydantoinases are employed in
the kinetic resolution of racemic mixtures of D-phenylglycine
and D-p-hydroxy-phenylglycine for the semisynthesis of penicil-
lins with annual production runs of several thousand tons.3
Relative to chemical catalysts, the most significant advantages
many wild-type enzymes exhibit a rather narrow substrate
specificity that can limit the range of useful and practical
applications. Nevertheless, recent advances in the construction
and assay of mutant protein libraries have demonstrated that in
vitro evolution of existing enzyme templates can result in the
enhancement and expansion of catalytic function for a desired
chemical transformation.4
The bacterial phosphotriesterase (PTE) catalyzes the hydroly-
sis of activated organophosphate triesters as shown in Scheme
1 for the detoxification of the insecticide paraoxon, with kinetic
§ Texas A&M University.
constants kcat ≈ 104 s-1 and kcat/Km ≈ 108 M-1 s-1 5
. The
‡ University of California, San Francisco.
substrate specificity is rather broad, and the wild-type enzyme
is able to distinguish between enantiomers that are chiral at the
phosphorus center.6,7 The substantial catalytic activity of phos-
(1) Liese, A. In Enzyme Catalysis in Organic Synthesis; Drauz, K., Waldmann,
H., Eds.; Wiley-VCH: Weinheim, Germany, 2002; Vol. III, pp 1419-1459.
(2) Esaki, N.; Kurihara, T.; Soda, K. In Enzyme Catalysis in Organic Synthesis;
Drauz, K., Waldmann, H., Eds.; Wiley-VCH: Weinheim, Germany, 2002;
Vol. III, pp 1281-1332.
(3) Syldatk, C.; Muller, R.; Siemann, M.; Wagner, F. In Biocatalytic Production
of Amino Acids and DeriVatiVes; Rozzell, D., Wagner, F., Eds.; Hanser
Publishers: Munchen, Germany, 1992; pp 75-128.
(4) Minshull, J.; Stemmer, W. Curr. Opin. Biotechnol. 1999, 35, 48-51.
(5) Aubert, S. D.; Li, Y.; Raushel, F. M. Biochemistry 2004, 43, 5707-5715.
9
15892
J. AM. CHEM. SOC. 2006, 128, 15892-15902
10.1021/ja0658618 CCC: $33.50 © 2006 American Chemical Society

Resolution of Chiral Phosphorus Esters
A R T I C L E S
Scheme 1
Chart 1
photriesterase is ideally suited for a variety of chemical
transformations. For example, PTE is one of the predominant
enzymes under current consideration as the primary catalytic
component for the detection and degradation of chemical threat
agents such as sarin, soman, and VX.8 The detoxification of
agricultural and household organophosphate insecticides is a
logical extension of this technology, and commercial insecticides
may be enzymatically resolved by PTE into separate enanti-
omers that are differentially toxic to specific insect targets.9 The
enantioselective properties of PTE may be applicable to the
resolution of chiral phosphorus containing pro-drugs and the
chemo-enzymatic syntheses of chiral phosphine ligands for
transition metal catalysts.10,11 Unlike chiral carbon centers, there
is no reservoir of naturally occurring chiral phosphorus centers
that can be chemically transformed to the target of choice.
If the catalytic potential of PTE is to be exploited for the
kinetic resolution of chiral phosphorus centers, then the substrate
specificity and stereoselectivity must be expanded and enhanced.
Prior work with lipases and hydantoinases suggests that
manipulation of the amino acid sequence can result in the
production of enzyme variants with altered catalytic proper-
ties.12,13 For example, Reetz and colleagues have elegantly
shown that directed evolution of a lipase from Pseudomonas
aeruginosa can be utilized to enhance the stereoselectivity for
the hydrolysis of p-nitrophenyl 2-methyl decanoate from an
initial value of 1.1 to >51 via a combination of error-prone
mutagenesis, gene shuffling, and saturation mutagenesis.14 Only
six amino acid changes from the wild-type sequence were
required to achieve this result. Even more impressive, the same
strategy was used to identify mutant proteins with the opposite
stereoselectivity. A mutant with 11 amino acid changes was
isolated that favored the other enantiomer by a factor of 30.15
Many, but not all, of the amino acid substitutions were located
away from the active site, and a molecular explanation for the
changes in catalytic properties is not generally obvious.16
In this paper, we have probed the stereoselective properties
of the wild-type PTE and selected mutant enzymes. The breadth
of the substrate specificity has been addressed with a series of
racemic mixtures of fully esterified phosphate, phosphonate, and
phosphinate derivatives. The structural diversity for members
of this library of substrates is shown in Chart 1. We
demonstrate that the inherent enantioselectivity of the wild-type
PTE can be greater than 105 and can be further enhanced up to
3 orders of magnitude by a single amino acid change to the
active site. We also establish that a relatively small number of
amino acid changes within the active site can invert the
stereoselectivity of the wild-type enzyme and favor the opposite
enantiomer by up to 3 orders of magnitude. In the most dramatic
example, the differences in stereoselectivity for each enantiomer
can be as high as 100 million for two mutants that differ by
only four amino acids.
Materials and Methods
Materials. The bacterial phosphotriesterase and the site-directed
mutants were prepared and purified to homogeneity as previously
described.6,7,17 Methylphenylphosphinic chloride was obtained from
Acros Organics. The other chemicals involved in the synthesis of the
1
substrates for phosphotriesterase were obtained from Aldrich. H and
31P NMR spectra in CDCl3 were obtained on Varian Unity INOVA
spectrometers. The 31P NMR spectra were obtained with phosphoric
acid (85%) as the external reference. Exact mass measurements were
obtained on a PE Sciex APJ Qstar Pulsar by the Laboratory for
Biological Mass Spectrometry at Texas A&M University.
(6) Chen-Goodspeed, M.; Sogorb, M. A.; Wu, F.; Hong, S.-B.; Raushel, F.
M. Biochemistry 2001, 40, 1325-1331.
(7) Chen-Goodspeed, M.; Sogorb, M. A.; Wu, F.; Raushel, F. M. Biochemistry
2002, 40, 1332-1339.
(8) Raushel, F. M. Curr. Opin. Microbiol. 2002, 5, 288-295.
(9) Miyazaki, A.; Najamura, T.; Kawaradani, M.; Marumo, S. J. Agric. Food
Chem. 1988, 36, 835-837.
Synthesis of Racemic Substrates. The phosphate, phosphonate, and
phosphinate esters synthesized for this investigation are depicted in
Chart 1. Compounds 1-4, 6-8, and 10-16 were prepared according
to the general procedure described below. Triethylamine (1 equiv) was
added dropwise into a stirred mixture of an alkyl phosphonic dichloride
or alkyl dichlorophosphate (10-20 mmol) and the corresponding
alcohol or phenol (1 equiv) in ethyl ether or dichloromethane (∼100
mL) at 0 °C. After the reaction mixture was stirred for 3-8 h,
p-hydroxyacetophenone (1 equiv) was added, along with another portion
(10) Reddy, K. R.; Boyer, S. H.; Erion, M. D. Tetrahedron Lett. 2005, 46, 4321-
4324.
(11) Pietrusiewicz, K. M.; Zablocka, M. Chem. ReV. 1994, 94, 1375-1411.
(12) May, O.; Nyuyen, P. T.; Arnold, F. H. Nat. Biotechnol. 2000, 18, 317-
320.
(13) Reetz, M. T.; Zonta, A.; Schimossek, L.; Liebeton, K.; Jaeger, L.-E. Angew.
Chem. 1997, 109, 2961-2963.
(14) Reetz, M. T.; Wilensek, S.; Zha, D.; Jaeger, L.-E. Angew. Chem. 2001,
113, 3701-3703.
(15) Zha, D.; Wilensek, S.; Hermes, J.; Jaeger, K.-E.; Reetz, M. T. Chem.
Commun. 2001, 2664-2665.
(16) Bocola, M.; Otte, N.; Jaeger, K.-E.; Reetz, M. T.; Thiel, W. Chem. Biol.
Chem. 2004, 5, 214-223.
(17) Hill, C. M.; Li, W.-S.; Thoden, J. B.; Holden, H. M.; Raushel, F. M. J.
Am. Chem. Soc. 2003, 125, 8990-8991.
9
J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006 15893

A R T I C L E S
Nowlan et al.
of triethylamine (1 equiv). The reaction mixture was stirred at room
temperature for another 24-48 h, and then the triethylammonium
chloride was removed by filtration. The reaction mixture was condensed
to dryness under reduced pressure. The residue was applied to a column
of silica gel and eluted with a mixture of hexanes and ethyl acetate. A
general workup gave the desired compounds in yields ranging from
20 to 85%.
4-Acetylphenyl ethyl methylphosphonate (1) was prepared from
methyl phosphonic dichloride and ethanol. 1H NMR (ppm): 7.98 (2H,
d, J ) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 4.08-4.30 (2H, m), 2.60 (3
H, s), 1.66 (3H, d, J ) 17.8 Hz), 1.35 (3H, t, J ) 7.2 Hz). 31P NMR
(ppm): 33.0. Mass spectrometry (ESI positive ion mode for M + H):
calculated, 243.08; found, 243.08.
(phenyl)phosphinate (2.5 mmol) was mixed with oxalyl chloride (3.8
mmol) and stirred at room temperature under dry conditions for 40
min. After removal of the volatile components under reduced pressure,
the residue was dissolved in ethyl ether (15 mL), and then 4-hydroxy-
acetophenone (2.4 mmol) and triethylamine (2.4 mmol) were added.
After the reaction mixture was stirred at room temperature for 24 h, it
was filtered and condensed to dryness under reduced pressure. The
residue was applied to a column of silica gel and eluted with a mixture
of hexanes and ethyl acetate (1:2, v/v). 1H NMR (ppm): 7.82-7.88 (4
H, m), 7.56-7.60 (1H, m), 7.24-7.26 (2H, d, J ) 9.0 Hz), 2.58 (3H,
s), 2.04-2.24 (2H, m), 1.22 (3H, dt, J ) 19.7 Hz, J ) 7.3 Hz). 31P
NMR (ppm): 53.9. Mass spectrometry (ESI positive ion mode for M
+ H): calculated, 289.10; found, 289.11.
4-Acetylphenyl ethyl phenylphosphonate (10) was prepared from
phenylphosphonic dichloride and ethanol. 1H NMR (ppm): 7.80-7.85
(4H, m), 7.52-7.60 (1H, m), 7.44-7.48 (2H, m), 7.22-7.30 (2H, d, J
) 9.0 Hz), 4.20-4.30 (2H, m), 2.58 (3H, s), 1.36-1.40 (3H, t, J )
7.2 Hz). 31P NMR (ppm): 18.5. Mass spectrometry (ESI positive ion
mode for M + H): calculated, 305.09; found, 305.10.
4-Acetylphenyl ethyl methyl phosphate (2) was prepared from methyl
dichlorophosphate and ethanol. 1H NMR (ppm): 7.98 (2H, d, J ) 9.0
Hz), 7.30 (2H, d, J ) 9.0 Hz), 4.22 (2H, m), 3.84 (3H, d, J ) 11.8
Hz), 2.60 (3H, s), 1.98 (3H, t, J ) 7.2 Hz). 31P NMR (ppm): -5.04.
Mass spectrometry (ESI positive ion mode for M + H): calculated,
259.07; found, 259.08.
4-Acetylphenyl phenyl ethylphosphonate (11) was prepared from
ethylphosphonic dichloride and phenol. 1H NMR (ppm): 7.94 (2H, d,
J ) 9.0 Hz), 7.37-7.15 (7H, m), 2.58 (3H, s), 2.19-2.09 (2H, m),
1.41-1.34 (3H, dt, JH-P ) 21.8 Hz, JH-H ) 7.8 Hz). 31P NMR (ppm):
28.4. Mass spectrometry (ESI positive ion mode for M + H):
calculated, 305.09; found, 305.10.
4-Acetylphenyl methyl ethylphosphonate (3) was prepared from ethyl
phosphonic dichloride and methanol. H NMR (ppm): 7.98 (2H, d, J
) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 3.84 (3H, d, J ) 11.8 Hz), 2.60
(3H, s), 1.84-2.00 (2H, m), 1.24 (3H, dt, J ) 19.7 Hz, J ) 7.3 Hz).
31P NMR (ppm): 37.0. Mass spectrometry (ESI positive ion mode for
M + H): calculated, 243.08; found, 243.08.
1
4-Acetylphenyl cyclohexyl ethyl phosphate (12) was prepared from
ethyl dichlorophosphate and cyclohexanol. 1H NMR (ppm): 7.98 (2H,
d, J ) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 4.50 (1H, m), 4.20 (2H, m),
2.58 (3H, s), 1.20-2.02 (13H, m). 31P NMR (ppm): -7.10. Mass
spectrometry (ESI positive ion mode for M + H): calculated, 327.14;
found, 327.14.
4-Acetylphenyl isopropyl ethylphosphonate (4) was prepared from
ethylphosphonic dichloride and isopropyl alcohol. 1H NMR (ppm): 7.98
(2H, d, J ) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 4.80 (1H, m), 2.60 (3H,
s), 1.82-1.98 (2H, m), 1.98-2.18 (9H, m). 31P NMR (ppm): 34.3.
Mass spectrometry (ESI positive ion mode for M + H): calculated,
271.11; found, 271.11.
4-Acetylphenyl cyclohexyl ethylphosphonate (13) was prepared from
ethylphosphonic dichloride and cyclohenxanol. 1H NMR (ppm): 7.98
(2H, d, J ) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 4.55 (1H, m), 2.60 (3H,
s), 1.20-2.02 (16H, m). 31P NMR (ppm): 34.0. Mass spectrometry
(ESI positive ion mode for M + H): calculated, 311.14; found, 311.14.
4-Acetylphenyl methyl cyclohexylphosphonate (14) was prepared
from cyclohexylphosphonic dichloride and methanol. 1H NMR (ppm):
7.98 (2H, d, J ) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 3.80 (3H, d, J )
11.8 Hz), 2.58 (3H, s), 1.20-2.10 (11H, m). 31P NMR (ppm): 35.0.
Mass spectrometry (ESI positive ion mode for M + H): calculated,
297.12; found, 297.13.
4-Acetylphenyl methyl(phenyl)phosphinate (5) was prepared upon
mixing a solution of methyl(phenyl)phosphinic chloride (20 mmol) in
ethyl ether (100 mL) with a saturated solution of 4-hydroxyacetophe-
none (20 mmol) in ethyl ether and stirred at 0 °C. Triethylamine (20
mmol) was added dropwise, and the reaction mixture was stirred at
room temperature for 42 h. After removal of the solvent under reduced
pressure, the residue was applied to a column of silica gel and eluted
with a mixture of hexanes and ethyl acetate (1:1, v/v). 1H NMR
(ppm): 7.78-7.88 (4H, m), 7.42-7.60 (3H, m), 7.20 (2H, d, J ) 9.0
Hz), 2.49 (3H, s), 1.90 (3H, d, J ) 17.8 Hz). 31P NMR (ppm): 50.6.
Mass spectrometry (ESI positive ion mode for M + H): calculated,
275.08; found, 275.08.
4-Acetylphenyl ethyl cyclohexylphosphonate (15) was prepared from
1
cyclohexylphosphonic dichloride and ethanol. H NMR (ppm): 7.98
4-Acetylphenyl methyl phenylphosphonate (6) was prepared from
(2H, d, J ) 9.0 Hz), 7.30 (2H, d, J ) 9.0 Hz), 4.04-4.24 (2H, m),
2.58 (3H, s), 1.20-2.10 (14 H, m). 31P NMR (ppm): 33.7. Mass
spectrometry: (ESI positive mode for M + H): calculated, 311.14;
found, 311.14.
4-Acetylphenyl cyclohexyl methyl phosphate (16) was prepared from
methyl dichlorophosphate and cyclohexanol. 1H NMR (ppm): 7.98 (2H,
d, J ) 9.0 Hz), 7.29 (2H, d, J ) 9.0 Hz), 4.3 (1H, m), 3.85 (3H, d, J
) 11.8 Hz), 2.60 (3H, s), 1.2-2.0 (10H, m). 31P NMR (ppm): -5.88.
Mass spectrometry (ESI positive ion mode for M + H): calculated,
313.12; found, 313.12.
1
phenyphosphonic dichloride and methanol. H NMR (ppm): 7.82-
7.87 (4H, m), 7.58-7.62 (1H, m), 7.44-7.54 (2H, m), 7.27 (2H, d, J
) 8 Hz), 3.90 (3H, d, J ) 11 Hz), 2.48 (3H, s). 31P NMR (ppm):
20.3. Mass spectrometry (ESI positive ion mode for M + H):
calculated, 291.08; found, 291.08.
4-Acetylphenyl phenyl methylphosphonate (7) was prepared from
1
methylphosphonic dichloride and phenol. H NMR (ppm): 7.80 (2H,
d, J ) 9.0 Hz), 7.20-7.40 (7H, m), 2.60 (3H, s), 1.86 (3H, d, J ) 17.8
Hz). 31P NMR (ppm): 29.8. Mass spectrometry (ESI positive ion mode
for M + H): calculated, 291.08; found 291.09.
Assay of Catalytic Properties. The time courses for the hydrolysis
of the racemic mixtures of substrates were monitored at 294 nm using
a SpectraMax360 UV-vis spectrophotometer. The extinction coefficient
used for 4-hydroxyacetophenone was 7.71 × 103 M-1 cm-1. Enzyme
was added to the substrate in a final volume of 3.0 mL, containing 50
mM CHES buffer, pH 9.0. The initial concentration of the racemic
substrates varied from 50 to 100 µM. The change in absorbance was
recorded, and the data were fit to either eq 1 or eq 2,
4-Acetylphenyl methyl phenyl phosphate (8) was prepared from
methyl dichlorophosphate and phenol. 1H NMR (ppm): 7.98 (2H, d, J
) 9.0 Hz), 7.15-7.5 (7H, m), 3.98 (3H, d, J ) 11.8 Hz), 2.58 (3H, s).
31P NMR (ppm): -10.7. Mass spectrometry (ESI positive ion mode
for M + H): calculated, 307.07; found, 307.08.
4-Acetylphenyl ethyl(phenyl)phosphinate (9) was prepared by mixing
diethyl phenylphosphonite (28 mmol) and ethyl iodide (3.2 mmol) in
a sealed tube and stirred at 95 °C for 48 h. Volatiles were removed
under reduced pressure. The reaction mixture was dissolved in ethyl
ether (200 mL) and washed with sodium sulfate (10% aqueous solution)
until the reaction solution was colorless. Removal of the ether gave
ethyl ethyl(phenyl)phosphinate (4.2 g) with a yield of 78%. Ethyl ethyl-
At ) Ao + A1(1 - e-k t
)
(1)
(2)
1
At ) Ao + A1(1 - e-k t) + A2(1 - e
)
-k2t
1
9
15894 J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006

Resolution of Chiral Phosphorus Esters
A R T I C L E S
where At is the absorbance at any time t, Ao is the initial absorbance,
A1 and A2 are net absorbance changes for the individual phases of the
reaction, and k1 and k2 are the first-order rate constants for each phase.
The Michaelis constants for a representative selection of substrates were
measured and found to be greater than 300 µM.
structures of the compounds used in this investigation are
presented in Chart 1.
Measurement of Stereoselectivity for Wild-Type PTE. The
relative rate of hydrolysis for each of the two enantiomers within
the 16 racemic mixtures was obtained by monitoring the full
time course (based on the end-point analysis for hydrolysis with
KOH) for each of the two isomers immediately after the addition
of an appropriate amount of enzyme. The initial concentration
of the ester substrate was established by hydrolysis with an
appropriate amount of hydroxide. For example, in Figure 1A is
shown the time course for the chemical hydrolysis of compound
6 with 67 mM NaOH. The hydrolysis reaction follows first-
order kinetics, and a fit of the data to eq 1 yields a rate constant
of 0.25 min-1. The time courses for the enzymatic hydrolysis
of 6 with 2.0 (gray line) and 200 nM (red line) wild-type PTE
are presented in Figure 1B. The time course for the lower
amount of enzyme clearly indicates that one of the two
enantiomers is hydrolyzed significantly faster than the other,
since only one-half of the total substrate is hydrolyzed in 10
min. At substrate concentrations lower than the Michaelis
constant, the time courses follow pseudo-first-order kinetics,
and the associated rate constants are directly proportional to
the amount of enzyme added.24 The rate constant from a fit of
the data to eq 1 is 0.47 min-1. Dividing the first-order rate
constant by the concentration of enzyme provides kcat/Km ≈ 3.9
× 106 M-1 s-1. At the higher level of enzyme, the fast phase is
essentially complete during the mixing time prior to data
acquisition. A fit of the data to eq 1 gives a rate constant of
0.24 min-1 and a value of kcat/Km for the slow enantiomer of
∼2.0 × 104 M-1 s-1 after correction for the amount of enzyme
added. The ratio of rates for the hydrolysis of the fast and slow
enantiomers of 6 is 200. Similar experiments were conducted
for the other 15 pairs of enantiomers, and the results are
presented in Table 1. In certain cases (such as with compounds
1-4), the times courses for both enantiomers could be obtained
with a single concentration of enzyme. In these instances, the
data were fit to eq 2 to provide the rate constants for both phases.
For compounds 1-16, the degree of discrimination between
the two enantiomers ranges from ∼3 for compounds 1 and 2 to
5.4 × 105 for compound 12.
Docking Calculations. Full details of the docking methods are given
in the accompanying paper.33 Briefly, structures for a number of the
PTE mutant enzymes were generated. These mutant enzymes differed
from the wild type by only a few specific residues, and these residues
were positioned such that their side chains projected into ligand open
regions of the active site. The substitutions either reduce or extend the
mutated side chain, but no major backbone movements are expected.
Consistent with this view, the actual X-ray structure of the H254G/
H257W/L303T mutant differs from that of the wild-type PTE by an
RMSD of only 0.27 Å.17-19 Therefore, we were conservative in our
model building, holding the structure of the overall enzyme constant,
and only modified the position of the substituting side chains
themselves. Beginning with the wild-type structure (PDB code 1HZY18),
substituted residues were fit into the structure by keeping the coordinates
of as many common atoms as possible. A dockable molecular database
was calculated that contained the pentavalent, bipyramidal high-energy
intermediate forms of 11 of the 16 experimentally tested enantiomeric
pairs. This flexibase was generated in analogy to the ZINC database;20
the primary innovation was the creation of pentavalent structures which
were obtained by modifications on the SMILES level followed by multi-
conformer structure calculations using Corina and Omega.21,22 These
structures resemble the compounds in the state after nucleophilic attack
of the catalytic hydroxide on the phosphorus and should represent a
state which the enzyme will preferentially recognize, enabling finer
distinctions.23 This database was docked into the active sites of PTE
and its mutants using DOCK3.5.54 and scored on the basis of van der
Waals and electrostatic complementarity, corrected for ligand desol-
vation energies. For each pair of enantiomers, the difference of their
docking energy scores was calculated and used as a proxy for the
binding affinity.
Results and Discussion
Chiral Substrate Library. The breadth of the substrate
specificity exhibited by the bacterial phosphotriesterase was
established by measuring the relative kinetic constants for a
broad series of phosphate, phosphonate, and phosphinate esters.
The leaving group for the entire series of compounds was
4-hydroxyacetophenone because the enzymatic formation of this
compound in solution at pH 9 can be easily monitored at 294
nm without interference from the protein or the initial substrate
ester. The remaining two groups, attached directly to the
phosphorus center, included methyl, methoxy, ethyl, ethoxy,
isopropoxy, cyclohexyl, cyclohexoxy, phenyl, and phenoxy
substituents. These side chains provided a reasonable array of
steric bulk and the inclusion of hydrophobic and hydrophilic
atoms for direct molecular interactions within the active site of
PTE. The nine substituents were mixed to generate a small but
diverse chemical library of racemic mixtures that included two
phosphinates, four phosphates, and ten phosphonates. The
Stereoselectivity for the Mutant G60A. The enantioselec-
tivities for the G60A mutant of PTE were determined for the
library of phosphate, phosphonate, and phosphinate esters in
the same manner as described for the wild-type enzyme.
Presented in Figure 1C are the time courses for the hydrolysis
of compound 6 with 0.50 nM (gray line) and 20 µM (red line)
G60A. The mutant enzyme clearly hydrolyzes one enantiomer
at a significantly faster rate than the other enantiomer. A fit of
the time course for the lower level of enzyme to eq 1 provides
a first-order rate constant of 0.91 min-1. Correcting for the
amount of enzyme gives a value kcat/Km ≈ 3.0 × 107 M-1 s-1
for the faster of the two enantiomers. Similarly, the rate constant
from a fit of the time course for the more concentrated enzyme
gives a first-order rate constant of 0.43 min-1 and a value of
kcat/Km ≈ 360 M-1 s-1. The mutant G60A prefers one isomer
by a factor of 8.3 × 104. Similar experiments were conducted
for the rest of the racemic mixtures, and in every case (except
for compound 3) the G60A mutant is more enantioselective than
(18) Benning, M. M.; Shim, H.; Raushel, F. M.; Holden, H. M. Biochemistry
2001, 40, 2712-2722.
(19) Benning, M. M.; Hong, S.-B.; Raushel, F. M.; Holden, H. M. J. Biol. Chem.
2000, 275, 30556-30560.
(20) Irwin, J.J.; Shoichet, B. K. J. Chem. Inf. Model. 2005, 45, 177-182.
(21) Sadowski, J.; Schwab, C. H.; Gasteiger, J. 3D Structure Generation and
Conformational Searching. In Computational Medicinal Chemistry and
Drug DiscoVery; Bultinck, P., De Winter, H., Langenaeker, W., Tollenaere,
J. P., Eds.; Dekker Inc.: New York, 2003; pp 151-212.
(22) Bostrom, J.; Greenwood, J. R.; Gottfries, J. J. Mol. Graphics Modell. 2003,
21, 449-462.
(24) Cleland, W. W. In The Enzymes; Boyer, P. D., Ed.; Academic Press: New
York, 1970; Vol. II, pp 1-65.
(23) Pauling, L. Chem. Eng. News 1946, 24, 1375-1377.
9
J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006 15895

A R T I C L E S
Nowlan et al.
Relaxation of Enantioselectivity. A selected set of mutant
enzymes that were obtained by alteration of those residues that
interact directly with the substrate within the active site of PTE
were tested with the small library of phosphorus containing
esters to ascertain the degree of change in the enantioselective
properties. Shown in Figure 2A are the time courses for the
hydrolysis of compound 8 at two different levels of enzyme.
At a concentration of wild-type PTE of 1.9 nM, one-half of the
racemic mixture is hydrolyzed with a first-order rate constant
of 0.61 min-1 (gray line). This corresponds to a value of kcat/
Km ≈ 5.4 × 106 M-1 s-1. At a concentration of wild-type PTE
of 190 nM, the fast enantiomer is hydrolyzed during the mixing
of enzyme with substrate and the slow enantiomer is hydrolyzed
with a first-order rate constant of 0.50 min-1, obtained from a
fit of the data to eq 1 (red line in Figure 2A). This corresponds
to a value of kcat/Km ≈ 4.4 × 104 M-1 s-1. The stereoselective
preference of the wild-type enzyme for compound 8 is thus 120.
Similar experiments with the G60A mutant established a
stereoselective preference of 3.9 × 105. When compound 8 is
hydrolyzed by the mutant H257Y (180 nM), both enantiomers
are hydrolyzed at nearly identical rates, as shown for the time
course presented in Figure 2B. A fit to eq 2 for the double
exponential gives rate constants of 0.27 and 0.16 min-1. The
corresponding values of kcat/Km are ∼2.5 × 104 and ∼1.5 ×
104 M-1 s-1, respectively, with a rather low enantioselective
preference of ∼2.
Reversal of Stereoselectivity. For the mutant I106G/F132G/
H257Y, the addition of high and low concentrations of enzyme
demonstrated that the enzyme has a stereoselective preference
of 970 for one enantiomer over the other (data not shown). In
order to determine whether the stereoselectivity of this mutant
is the same as or opposite to that possessed by the wild-type
and G60A enzymes, a low level of the triple mutant (34 nM)
was added to compound 8 to selectively hydrolyze the more
rapidly hydrolyzed enantiomer. The time course for this
experiment is presented in Figure 2C (gray line). After the
hydrolysis of the more rapidly hydrolyzed enantiomer was
complete, a low level of the G60A mutant was added to the
reaction mixture to determine if the enantiomer preferred by
the triple mutant was the same as that preferred by the G60A
mutant. The time course for the addition of the G60A mutant
(2.6 nM) is presented in Figure 2C (red line). Clearly, the G60A
and I106G/F132G/H257Y mutants preferentially hydrolyze
different enantiomers of compound 8. Similar experiments were
utilized to establish the stereoselective preferences for 15
mutants in addition to the wild-type PTE, and the results are
presented in Table 1. Not every mutant was tested with every
enantiomeric pair of substrates.
Absolute Stereochemistry. For the wild-type and G60A
mutant enzymes, it has been demonstrated that the enantiomer
hydrolyzed preferentially can be determined on the basis of the
relative size and orientation of the substituents attached to the
central phosphorus core.25-29 For the target substrates con-
structed for this investigation, the pairs of enantiomers are
Figure 1. (A) Time course for the hydrolysis of compound 6 with 67 mM
KOH. (B) Time course for the enzymatic hydrolysis of compound 6 with
2.0 nM PTE (gray line) and 200 nM wild-type PTE (red line) at pH 9.0 in
50 mM CHES buffer, 30 °C. (C) Time course for the enzymatic hydrolysis
of compound 6 with 0.5 nM (gray line) and 20 µM (red line) concentrations
of the G60A mutant of PTE. Additional details are provided in the text.
(25) Hong, S.-B.; Raushel, F. M. Biochemistry 1999, 38, 1159-1165.
(26) Wu, F.; Li, W.-S.; Chen-Goodspeed, M.; Sogorb, M.; Raushel, F. M. J.
Am. Chem. Soc. 2000, 122, 10206-10207.
is the wild-type enzyme. The most extreme case is for compound
16, where the stereoselectivity is greater than 6 million to one.
The enantioselective ratios for the hydrolysis of racemic
mixtures of chiral phosphorus esters by the mutant G60A are
provided in Table 1.
(27) Li, W.; Li, Y.; Hill, C. M.; Lum, K. T.; Raushel, F. M. J. Am. Chem. Soc.
2002, 124, 3498-3499.
(28) Li, Y.; Aubert, S. D.; Raushel, F. M. J. Am. Chem. Soc. 2003, 125, 7526-
7527.
(29) Li, Y.; Aubert, S. D.; Maes, E. G.; Raushel, F. M. J. Am. Chem. Soc. 2004,
126, 8888-8889.
9
15896 J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006

Resolution of Chiral Phosphorus Esters
A R T I C L E S
1.
Table
9
J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006 15897

A R T I C L E S
Nowlan et al.
preferred enantiomer for the wild-type and G60A mutant is listed
in Table 1. The notation changes for the phosphate, phosphonate,
and phosphinate esters because of the nomenclature rules. A
Brønsted analysis of the effect of the pKa of the leaving group
on kcat and kcat/Km has shown that the âlg for the wild-type
enzyme is -1.8, and thus the substituted phenol (pKa ) 8.0) is
always hydrolyzed rather than phenol (pKa ) 10) or any of the
alcohols.30
Substrate Specificity. In this study, we have utilized nine
different substituents attached to a central phosphorus core to
create a small library of phosphate, phosphonate, and phosphi-
nate esters. The wild-type and G60A variants are able to
efficiently hydrolyze these compounds, and thus PTE has the
ability to recognize and hydrolyze a rather diverse set of
substrates while simultaneously exhibiting a rather pronounced
stereoselectivity. For the wild-type enzyme, a clear stereochem-
ical preference can be demonstrated for 14 of the 16 compounds
tested. A significant distinction could not be demonstrated for
methyl versus ethoxy or methoxy versus ethoxy as they exist
in compounds 1 and 2, where the stereochemical preference
between the two enantiomers is less than 5. However, the
stereoselectivity for the G60A mutant is significantly more
pronounced than for the wild-type enzyme, and a clear distinc-
tion can be identified for all pairs of substituents, except for
methoxy versus ethyl (compound 3). For the wild-type enzyme,
when the difference in size between the two substituents is one
or two carbon/oxygen atoms, the inherent stereoselectivity is
less than an order of magnitude (see compounds 1-4). When
a phenolic group is paired with a small alkyl or alkoxy
substituent (see compounds 7, 8, and 11), the preferred
enantiomer is hydrolyzed about 2 orders of magnitude faster
than the slower enantiomer. When the larger substituent is either
phenyl or cyclohexyl, the degree of discrimination between the
two enantiomers is greater than a factor of 103. In every case,
except for compound 3, the stereoselectivity is further enhanced
by the mutant G60A. This increase in the stereoselectivity varies
from 1 to 3 orders of magnitude by a single amino acid
substitution within the active site of PTE. With compounds 12
and 16, the rate differential between the two enantiomers is
greater than a million-fold!
Inversion and Relaxation of Stereoselectivity. Single amino
acid changes within the active site of PTE can relax the inherent
stereoselectivity of the wild-type enzyme. For example, the
single H257Y mutation relaxes the stereoselectivity for com-
pounds 6, 8, and 11 from approximately 100:1 to less than 11:
1. This single amino acid change can significantly diminish the
enhanced stereoselectivity exhibited by G60A by up to 5 orders
of magnitude (see, for example, compound 8). Combinations
of mutational changes can also be used to invert the stereose-
lectivity exhibited by the wild-type and G60A enzymes. The
specific set of residues depends, in part, on the specific
substituents attached to the central phosphorus core of the target
substrate. Thus, the mutant H257Y/L303T is the best example
for the inversion of compounds 5, 6, 14, and 15. With this set
of compounds, the large substituent is either a phenyl or
cyclohexyl group attached directly to the phosphorus core. A
stereoselectivity as large as 1800 is observed for compound 14.
The mutant I106G/F132G/H257Y is the best enzyme for the
inversion of stereoselectivity of compounds 7, 8, 11, 12, 13,
Figure 2. (A) Time course for the hydrolysis of compound 8 with 1.9 nM
(gray line) and 190 nM (red line) wild-type PTE at pH 8.0 in 50 mM HEPES
buffer, 25 °C. (B) Time course for the hydrolysis of compound 8 with a
180 nM concentration of the H257Y mutant of PTE. (C) Time course for
the hydrolysis of the preferred enantiomer of compound 8 with a 34 nM
concentration of the I106G/F132G/H257Y mutant of PTE (gray line). After
hydrolysis of the preferred enantiomer of compound 8 (15 min), a 2.6 nM
concentration of the mutant G60A was added to hydrolyze the remaining
material (red line). Additional details are provided in the text.
presented as shown in Chart 1. The more rapidly hydrolyzed
enantiomer is the one where the substituent oriented to the right
in a Fischer projection is physically larger than the substituent
orientated to the left. This enables the preferred enantiomer to
be identified for the wild-type and G60A mutant forms of PTE,
except for compound 3, where the relative size of the two
substituents is essentially the same. The stereochemically
(30) Hong, S.-B.; Raushel, F. M. Biochemistry 1996, 35, 10904-10912.
9
15898 J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006

Resolution of Chiral Phosphorus Esters
A R T I C L E S
Figure 3. Enantioselectivity for the hydrolysis of compound 8 by the wild-
type PTE and seven mutant enzymes at pH 9.0. The designated enzymes
are as follows: A, G60A; B, S308G; C, wild-type PTE; D, I106G; E, I106G/
F132G; F, I106T/F132A/H254G/H257W; G, I106G/H257Y; and H, I106G/
F132G/H257Y.
and 16. This collection of compounds invariably contains a
phenolic or O-cyclohexyl group as the largest substituent
attached to phosphorus. With these compounds, the stereose-
lectivity ranged from 12 to 970. The mutant I106A/F123A/
H257W was the best enzyme for the inversion of stereoselec-
tivity for compounds 1, 2, and 4. With these compounds, the
two substituents are relatively small, and the substituted amino
acids have larger side chains in comparison to the mutant I106G/
F132G/H257Y. Therefore, to a first approximation, the specific
amino acids that are mutated in the active site of PTE can be
selected and changed to a variety of residues that can be
customized for a specified task.
The control of the enantioselectivity in this system is
significant. For the wild-type enzyme, the stereoselectivity
ranges from 1 to >105. For the single amino substitution, G60A,
the stereoselectivity ranges from 6 to >106. As few as two to
three amino acid changes within the active site of PTE can invert
the inherent stereoselectivity such that the opposite enantiomer
is the preferred substrate by a factor greater than 1000. Perhaps
the most impressive range of stereoselectivity is exhibited for
compound 8. The mutant G60A prefers the SP-enantiomer by a
factor of 3.7 × 105, whereas the I106G/F132G/H257Y mutant
prefers the RP-enantiomer by a factor of 970. Therefore, the
difference in stereoselectivity has been altered by a factor of
3.6 × 108 with only four amino acid changes! The range of
enantioselectivity for compound 8 with the wild-type PTE and
selected mutants is presented in Figure 3.
Structural Constraints on Stereoselectivity. What controls
the inherent enantioselectivity of the wild-type enzyme and
enables this catalytic property to be relaxed, enhanced, or
inVerted with relatively few amino acid changes within the active
site? Some of the residues within the active site of PTE that
coalesce to form the binding pockets for substrate recognition
are presented in Figure 4.19 Shown in this image is a non-
hydrolyzed substrate analogue, diisopropyl methylphosphonate,
which interacts through the phosphoryl oxygen to the â-metal
ion in the active site. The overall binding pocket is largely
hydrophobic, and there appears to be no strong electrostatic
interactions between the protein and the substrate (other than
Figure 4. Structure of the active site of the wild-type PTE. The two divalent
cations (MR and Mâ) are shown in salmon, and the bridging hydroxide is
shown in red. The side chains and R-carbon atoms for the residues mutated
in this investigation (G60, I106, F132, H254, H257, L303, and S308) are
labeled. The substrate analogue, diisopropyl methylphosphonate (DIMP),
is positioned adjacent to the binuclear metal center. The coordinates were
taken from the PDB file 1ez2 and rendered with the program WebLab
Viewer Pro.
Chart 2
with the divalent cation). In this complex, there are subsites
for each of the three substituents attached to the central
phosphorus core.19 In a productive enzyme-substrate complex,
we assume that the leaving group must be positioned in a
manner that will allow for a nucleophilic attack by the bridging
hydroxide onto the substrate, with the leaving group orientated
at an angle of approximately 180°.
Since the individual enantiomers have the same inherent
chemical reactivity, the rate differential may be explained on
the basis of differences in binding to the chiral active site. Either
the binding is thermodynamically weak or the orientation of
binding exhibited by the more poorly hydrolyzed enantiomer
is unproductive. For the substrates examined in this analysis,
there are three different substituents attached to the central
phosphorus core. Each of the three substituents must be oriented
in some manner within the active site. If one assumes that there
are separate binding subsites for each of these substituents (X,
Y, and Z), then there would be three possible orientations (or
binding modes) for any given substrate. Since the leaving group
must be aligned properly with the attacking nucleophile, only
one of these three orientations will lead to productive turnover.
These orientations are presented graphically in Chart 2. In this
scenario, the slower enantiomer would have a preponderance
of the bound substrate in one of the two possible nonproductive
orientations. A more subtle distinction between the two enan-
tiomers would position the slower enantiomer at a less optimal
attack angle relative to the hydroxide nucleophile or bound to
9
J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006 15899

A R T I C L E S
Nowlan et al.
Figure 5. Changes in enantiomeric preference resulting from steric changes to key active-site residues for two docked structures of compound RP-13, (A)
WT PTE and (B) mutant I106A/F132G/H257W structures. In the WT structure, a catalytically competent geometry is found. This is less true in the mutant
structure, presumably owing to the steric changes in site. Oxygen atoms are colored red, enzyme carbons in gray, ligand carbons in yellow, hydrogens in
white, and nitrogens in blue. The molecular surface of the enzyme is shown, as are two of the residues whose sterics change the most on substitution. The
gray spheres represent the two catalytic zinc ions. This figure was rendered using PyMOL.32
the active site such that the phosphoryl oxygen is unable to be
polarized via a direct interaction with one of the two divalent
cations. The acquisition of X-ray structures of PTE bound with
chiral inhibitors will help to establish the root cause for the
enantioselectivity of PTE and the site-directed mutants.
The failure of DOCK,31 the program used for docking, to find
a catalytically optimal pose for a particular enantiomer suggests
that a highly productive pose is unlikely to occur, making
hydrolysis unlikely even when the energy score might be
favorable.
Qualitatively, the enantiomeric selectivities predicted by
docking correspond closely to the experimental results. Of the
29 experimentally observed inversions, 28 were predicted by
docking either through a significant change of the energy
difference between the two enantiomers or by the failure to find
a catalytically correct pose for one of the two enantiomers (see
Table 2 and Supporting Information for detailed docking scores).
For five enantiomer-mutant pairs, an inversion was predicted
that was not observed experimentally; four of these were for
compound 6. Similarly, the docking energies also qualitatively
tracked those pairs in which the relative selectivities between
the isomers changed from the WT, though the overall isomeric
preference remained the same. For 13 of 20 mutant-enantiomer
pairs, the docking scores can predict this acceleration and
deceleration correctly. Here, too, 4 of the 7 wrong predictions
owe to compound 6 alone. In all cases where DOCK failed to
predict the change in the experimental ratio, this change was
by a factor of about 10 or less compared to the WT. In those
cases where the acceleration or deceleration was larger than 10-
fold, this change of the ratios was captured correctly by the
docking predictions.
If simple sterics play a large role in selecting between
productive and unproductive substrate geometries, as suggested
by the preceding discussion, then it should be possible to capture
by relatively simple calculations of molecular fit between
substrate and enzyme. We turned to molecular docking calcula-
tions of high-energy intermediate forms of the substrates into
model structures of 7 of the 15 mutant enzymes. The database
of high-energy bipyramidal species of the substrates was docked
into the active sites of the mutants, and for each pair of
enantiomers the difference between their energy scores was
calculated (see Supporting Information). This score is a proxy
for the binding affinity, i.e., the stabilization of a compound by
a PTE variant. The increase or decrease of the difference
between docking energy scores for a pair of enantiomers in a
mutant compared to the wild type (WT) should predict rate
acceleration, deceleration, or even if the enantiomeric preference
has been inverted between the WT and a given mutant enzyme.
In addition to the docking energy scores, we also considered
the docked poses. We assumed that a catalytically competent
pose (see Figure 5A) of an intermediate would place the former
hydroxide between the two zinc ions, with the leaving group
apical to that oxygen and oriented toward the exit of the active
site in the leaving pocket. The two nonreactive side chains of
the phosphorus core should be accommodated in the two subsites
of the active site. A catalytically less competent pose (see Figure
5B) typically leaves one of the subsites empty (the small pocket
in Figure 5B). As a secondary event to the less complementary
steric fit, the resulting poses usually have impaired electrostatic
interactions between the oxygens of the ligand and the zinc ions.
An example of a dramatic change of the binding quality and
energy due to a substitution is illustrated in Figure 5. The high-
energy intermediate of compound RP-13 can be stabilized and
bound well by the wild-type enzyme (Figure 5A), where the
(31) Meng, E. C.; Shoichet, B. K.; Kuntz, I. D. J. Comput. Chem. 1992, 13,
505-524.
(32) http://pymol.sourceforge.net/
(33) Hermann, J. C.; Ghanem, E.; Li, Y.; Raushel, F. M.; Irwin, J. J.; Soichet,
B. K. J. Am. Chem. Soc. 2006, 128, http://dx.doi.org/10.1021/ja065860f.
9
15900 J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006

Resolution of Chiral Phosphorus Esters
A R T I C L E S
Table 2. Stereoselectivities Predicted by Molecular Dockinga
a Results are given relative to the ratio of rates of the wild type (see Table 1 and Supporting Information for more details). The wild-type preferences for
an enantiomer are given in the “WT” column. In each case, the same enantiomer is favored by docking, as was experimentally verified. The arrows indicate
the experimentally observed qualitative changes in enantioselectivity. The colors indicate whether docking was able to predict that change qualitatively. If
an arrow is red, the docking prediction is incorrect; if it is black, docking predicted the experimental enantioselectivity correctly. Up arrows indicate an
increase of the ratio of rates, down arrows indicate a decrease of the ratio of rates, and curving arrows indicate an inversion of the enantioselectivity.
oxygen is placed equidistant between the two zinc ions and all
pockets are complemented by the side chains of the phosphorus
core. Conversely, in the I106A/F132G/H257W mutant enzyme,
no highly productive pose can be found for this compound
(Figure 5B). The former small pocketsclose to A106sis empty,
and the oxygen is mainly interacting with one zinc ion only.
For the enantiomer SP-13, which has the two nonreactive side
chains switched, the opposite can be observed: SP-13 fits well
in the I106A/F132G/H257W mutant but not in the WT (not
shown). Poses and energies suggest, for enantiomer pair 13,
that the RP-enantiomer is preferred by the WT, whereas the SP-
enantiomer is preferred by the I106A/F132G/H257W mutant,
which is consistent with the experimental results (see Table 1).
These results suggest that the changes in the ratio of rates
and enantioselectivities of mutant PTEs are captured by
relatively simple stereochemical and electrostatic effects. These
effects can be represented by the geometry of the high-energy
intermediates docked in these calculations. An important in-
novation in these calculations is docking not of ground-state
substrate structure but rather of a high-energy intermediate that
resembles a transition-state-like state (see accompanying paper
for more discussion of this point).33 As a caution, we note that
docking, as used here, represents a relatively simple level of
theory, and consequently this analysis cannot be pushed too far.
Indeed, inspection of the docking energies shows that, whereas
they qualitatively agree with the experimental changes in
preferences, their quantitative agreement is poor (see Supporting
Information). For instance, the ratio of rates for enantiomer pair
6 increases about 400-fold in the G60A mutant compared to
the WT, which corresponds to a difference in docking scores
of 0.5 kcal mol-1. The ratio of rates for enantiomer 16 increases
by a factor of 80 in the same mutant, which corresponds to an
increase of 11.3 kcal mol-1 in the docking scores. So, even
when docking predicts the correct direction of a ratio change
very reliably, the magnitude of that change cannot be correlated
with the difference in the docking scores for the two enanti-
omers.
Overall, these docking results support the hypothesis that the
stereospecificity and efficiency of the PTE mutants can be
directly related to the modification of the shape of the three
hydrophobic binding pockets (the leaving group pocket and two
subsites for the other side chains). Primarily, the mutations
improve or disrupt the hydrophobic interactions between a
compound and the enzyme. A secondary but nevertheless
important distinguishing element is the polar interaction between
the substrate oxygens and the catalytic zincs, which can change
dramatically as a result of an unfavorable pose and orientation
of a compound within the active site during catalysis.
Of course, to be compelling, it would be important to use
these docking calculations prospectively, to predict enantiose-
lectivities on previously untested substrates. We return to just
such a study in the accompanying paper,33 where the selectivities
of several enantiomers on the part of the mutant enzymes were
predicted before the results were known.
Summary. Wild-type and mutant forms of the bacterial
phosphotriesterase have been shown to catalyze the stereose-
lective hydrolysis of racemic mixtures of chiral phosphate,
phosphonate, and phosphinate esters. For the wild-type enzyme,
one enantiomer can be preferred by up to 5 orders of magnitude.
A single active-site mutant, G60A, enhances the inherent
stereoselectivity of the wild-type enzyme by an additional 1-2
orders of magnitude. Mutations within the active site that expand
and contract the size of the binding subsites can relax and invert
the enantioselectivity of the wild-type PTE for the initially
9
J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006 15901

A R T I C L E S
Nowlan et al.
slower enantiomer by up to 3 orders of magnitude. Molecular
docking calculations suggest that the reason for these kinetic
effects might be the impaired or improved stabilization of high-
energy structures, owing to steric complementarity. This enan-
tioselective enzyme library can be used to kinetically resolve
chiral mixtures of racemic phosphorus-based esters.
by the NSF REU program CHE-0243829. J.C.H. thanks the
“Deutschen Akademie der Naturforscher Leopoldina” for a
fellowship under contract no. BMBF-LPD 9901/8-115.
Supporting Information Available: Detailed docking scores.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Acknowledgment. This work was supported in part by the
NIH (GM68550 and GM71790). J.C. and T.E. were supported
JA0658618
9
15902 J. AM. CHEM. SOC. VOL. 128, NO. 49, 2006