5-Enolpyruvylshikimate 3-Phosphate Synthase: Chemical Synthesis of the Tetrahedral Intermediate and Assignment of the Stereochemical Course of the Enzymatic Reaction

Article abstract of DOI:10.1021/ja036627+

A chemical synthesis of both diastereomers of the tetrahedral intermediate involved in 5-enolpyruvylshikimate 3-phosphate synthase (EPSPS) catalysis has been accomplished. Combination of methyl dibromopyruvate with a protected shikimic acid derivative, ph

Full text of DOI:10.1021/ja036627+

Published on Web 09/26/2003
5-Enolpyruvylshikimate 3-Phosphate Synthase: Chemical
Synthesis of the Tetrahedral Intermediate and Assignment of
the Stereochemical Course of the Enzymatic Reaction
Ming An, Uday Maitra,† Ulf Neidlein,‡ and Paul A. Bartlett*
Contribution from the Center for New Directions in Organic Synthesis,§
Department of Chemistry, UniVersity of California, Berkeley, California 94720-1460
Received June 11, 2003; E-mail: paul@fire.cchem.berkeley.edu
Abstract: A chemical synthesis of both diastereomers of the tetrahedral intermediate involved in
5-enolpyruvylshikimate 3-phosphate synthase (EPSPS) catalysis has been accomplished. Combination of
methyl dibromopyruvate with a protected shikimic acid derivative, phosphorylation, and lactonization afforded
the intermediates (S)-15 and (R)-15, whose configurations were assigned by NMR. After introduction of
the 3-phosphate group and deprotection, photoinitiated radical debromination of the dibromo analogues
(S)-5 and (R)-5 was accomplished with tributyltin hydride in mixed aqueous solvents in the presence of
surfactant to give the pyruvate ketal phosphates (R)-TI and (S)-TI, respectively. These compounds are
stable at high pH, but decompose at pH 7 with a half-life of ca. 10 min. (R)-TI proved to be inert to EPSPS,
while (S)-TI was converted by the enzyme to a mixture of 5-enolpyruvylshikimate 3-phosphate, shikimate
3-phosphate, and phosphoenolpyruvate. The demonstration that the enzymatic intermediate possesses
the S-configuration at the ketal center confirms the mechanism as an anti addition followed by a syn
elimination. Furthermore, it appears that the syn stereochemistry of the second step requires the phosphate
leaving group to serve as the base in catalyzing its own elimination.
Introduction
catalyzed by 5-enolpyruvylshikimate 3-phosphate synthase
(EPSPS).
Few biosynthetic sequences are as replete with unusual
chemical transformations as the shikimate-chorismate path-
way.1 The novelty of its reactions and its key position in primary
metabolism and as a herbicide target have stimulated great
interest in elucidating the chemical mechanism and steric course
for each of the enzymes involved.2 By 1990, all of the stereo-
chemical questions had been answered, except for one: the
mechanism of one of the most important reactions of all, that
Sprinson’s original suggestion3 that the transfer of a carboxy-
vinyl group from phosphoenolpyruvate (PEP) to shikimate
3-phosphate (S3P) to form 5-enolpyruvylshikimate 3-phosphate
(EPSP) proceeds via an addition-elimination mechanism was
confirmed by Anderson and co-workers at Monsanto by isolating
the tetrahedral intermediate (TI) from the enzyme (Figure 1).4
In an experiment employing equimolar amounts of enzyme and
S3P with excess PEP and Pi, equilibrium was established
between the enzyme complexes of S3P and PEP, TI, and EPSP
and Pi. Under these conditions, the TI constitutes up to 33% of
all enzyme-bound species, with the balance primarily the EPSP
and Pi complex. This mixture was denatured with neat triethyl-
† Current address: Department of Organic Chemistry, Indian Institute
of Science, Bangalore-560 012, India.
‡ Current address: BASF Aktiengesellschaft, D-67056 Ludwigshafen,
Germany.
§ The Center for New Directions in Organic Synthesis is supported by
Bristol-Myers Squibb as a Sponsoring Member and Novartis Pharma as a
Supporting Member.
1
amine, and the TI was isolated and characterized by H, 13C,
and 31P NMR.4,5 However, the configuration of the ketal
stereocenter could not be assigned. Anderson et al. noted the
instability of the TI at neutral pH, where it decomposes rapidly
to S3P, pyruvate, and Pi. The relevance of the isolated
intermediate was further demonstrated by reexposure to EPSPS,6
which catalyzed its decomposition to EPSP and Pi as major
products and S3P and PEP as minor products, consistent with
(1) Haslam, E. Shikimic Acid Metabolism and Metabolites; Wiley: New York,
1993.
(2) DHQ synthase: (a) Rotenberg, S. L.; Sprinson, D. B. Proc. Natl. Acad.
Sci. U.S.A. 1970, 67, 1669-1672. (b) Turner, M. J.; Smith, B. W.; Haslam,
E. J. Chem. Soc., Perkin Trans. 1 1975, 52-55. (c) Widlanski, T.; Bender,
S. L.; Knowles, J. R. Biochemistry 1989, 28, 7572-7582. Dehydro-
quinase: See also ref 2c. (d) Shneier, A.; Harris, J.; Kleanthous, C.; Coggins,
J. R.; Hawkins, A. R.; Abell, C. Bioorg. Med. Chem. Lett. 1993, 3, 1399-
1402. EPSP synthase: (e) Grimshaw, C. E.; Sogo, S. G.; Copley, S. D.;
Knowles, J. R. J. Am. Chem. Soc. 1984, 106, 2699-2700. (f) Asano, Y.;
Lee, J. J.; Shieh, T. L.; Spreafico, F.; Kowal, C.; Floss, H. G. J. Am. Chem.
Soc. 1985, 107, 4314-4320. (g) Kim, D. H.; Tucker-Kellogg, G. W.; Lees,
W. J.; Walsh, C. T. Biochemistry 1996, 35, 5435-5440. Chorismate
synthase: (h) Hill, R. K.; Newkome, G. R. J. Am. Chem. Soc. 1969, 91,
5893-5894. (i) Onderka, D. K.; Floss, H. G. J. Am. Chem. Soc. 1969, 91,
5894-5896. (j) Floss, H. G.; Onderka, D. K.; Carroll, M. J. Biol. Chem.
1972, 247, 736-744. Chorismate mutase: (k) Sogo, S. E.; Widlanski, T.
S.; Hoare, J. H.; Grimshaw, C. E.; Berchtold, G. A.; Knowles, J. R. J. Am.
Chem. Soc. 1984, 106, 2701-2703. See also ref 2f.
(3) Bondinell, W. E.; Vnek, J.; Knowles, P. F.; Sprecher, M.; Sprinson, D. B.
J. Biol. Chem. 1971, 246, 6191-6196.
(4) Anderson, K. A.; Sikorski, J. A.; Benesi, A. J.; Johnson, K. A. J. Am.
Chem. Soc. 1988, 110, 6577-6579.
(5) Anderson, K. S.; Sammons, R. D.; Leo, G. C.; Sikorski, J. A. Biochemistry
1990, 29, 1460-1465.
(6) Anderson, K. A.; Johnson, K. A. J. Biol. Chem. 1990, 265, 5567-5572.
9
10.1021/ja036627+ CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 12759-12767
12759

A R T I C L E S
An et al.
Figure 1. Addition-elimination mechanism for the transfer of a carboxyvinyl group catalyzed by EPSP synthase.
Figure 2. Remaining stereochemical pathways for EPSP synthase involving si face protonation of PEP and overall retention of the double bond configuration.
the partition ratio predicted from the forward and reverse rate
constants determined previously.7
µg from an equilibrium mixture with substrate in the presence
of 1 g of enzyme), and it is highly unstable, with a reported
half-life of about 45 min at neutral pH.4,6 Thus, although
Anderson et al. could establish the identity of the isolated TI
by NMR and through its chemical and enzymatic transforma-
tions, they were unable to determine the configuration of the
ketal center.
Two paths may be followed for each of three different steps
in the formation and decomposition of the TI, resulting in eight
possibilities for the detailed stereochemical mechanism of the
overall transformation. In the first step, the proton may be added
to either the re or si face of the PEP double bond. In the second,
the 5-hydroxyl of S3P may add syn or anti to this hydrogen.
And in the third, the elimination of phosphate may also occur
in a syn or anti sense. Three independent pieces of information
are therefore required to define completely the stereochemical
course of the process.
The first piece of information came from the Knowles and
Floss groups, who used stereoisotopically labeled PEP (3H and
2H) to show that the double bond of EPSP is generated with
the same geometry as that in the PEP substrate.2e,f This finding
meant that the addition and elimination steps proceed with
opposite stereochemistries; i.e., one is syn and one is anti. The
Walsh group supplied the second piece of information by
studying the EPSPS-catalyzed addition of S3P to (E)- and (Z)-
fluoro-PEP in deuterated water.2g Trapping the fluorinated TI
and determining the configuration of the chiral fluoromethyl
group (CHDF) allowed them to conclude that the proton is added
to the si face of PEP in the first step of the EPSPS reaction.
These experiments left only two possible stereochemical
mechanisms to be distinguished (Figure 2): anti addition to form
an S-configured ketal phosphate, followed by syn elimination,
or the opposite, syn addition/anti elimination via an (R)-ketal
phosphate. To resolve this question required that the absolute
configuration of the ketal phosphate stereocenter in the TI be
determined.
The structure of a decomposition product, bicyclic ketal 1,
suggested the S-configuration for the TI, on the assumption that
cyclization to 1 occurs with inversion at the ketal center.5,8 Stable
analogues of the TI have provided mixed signals: the greater
affinity of the (R)-phosphonate 2 in comparison to the S-isomer
first led us to suggest that the TI also has the R-configuration.9
However, similar affinity for the trifluoro-TI diastereomers 3
and greater affinity for the opposite diastereomer of the difluoro
analogue 4 support the assignment as S (note that fluorine inverts
the Cahn-Ingold-Prelog prioritization among the ketal sub-
stituents).10 More recently, the crystal structure of the complex
2-
of EPSPS with S3P and glyphosate (-O2CCH2 NH2+CH2PO3
)
has been determined (Figure 3).11 With the assumption that the
active site orientation of PEP is the same as that of glyphosate,
Scho¨nbrunn et al. proposed that Glu-341 serves as the proton
donor in catalyzing anti addition; the consequence of such a
process would be formation of a TI with the S-configuration.
Finally, there are strong similarities in structure and in chemical
mechanism between EPSPS and MurA, which catalyzes the
transfer of a carboxyvinyl group from PEP to an N-acetyl-
glucosamine hydroxyl group as a step in bacterial cell wall
biosynthesis. The demonstration by Walsh and co-workers that
the MurA reaction proceeds with anti addition to an S-tetrahedral
(8) Leo, G. C.; Sikorski, J. A.; Sammons, R. D. J. Am. Chem. Soc. 1990, 112,
1653-1654.
Assignment of the ketal phosphate configuration has been
confounded by the elusive nature of the TI. It can only be
isolated in small quantities from the enzymatic reaction (ca. 300
(9) Alberg, D. G.; Bartlett, P. A. J. Am. Chem. Soc. 1989, 111, 2337-2338.
(10) Alberg, D. G.; Lauhon, C. T.; Nyfeler, R.; Fa¨ssler, A.; Bartlett, P. A. J.
Am. Chem. Soc. 1992, 114, 3535-3546.
(11) Scho¨nbrunn, E.; Eschenburg, S.; Shuttleworth, W.; Schloss, J. V.; Amrhein,
N.; Evans, J. N. S.; Kabsch, W. Proc. Natl. Acad. Sci. U.S.A. 2001, 98,
1376.
(7) Anderson, K. S.; Sikorski, J. A.; Johnson, K. A. Biochemistry 1988, 27,
7395-7406.
9
12760 J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003

5-Enolpyruvylshikimate 3-Phosphate Synthase
A R T I C L E S
Scheme 1
Figure 3. Active site of EPSP synthase occupied by S3P and glyphosate.11
intermediate provided further circumstantial evidence for the
same reaction course for EPSPS.12
because the hemiketal or hemiacetal is favored in the intramo-
lecular equilibrium with the acyclic forms. Both problems are
mitigated by electron-withdrawing substituents in the ketone,
which stabilize the hemiketal adduct relative to the ketone and
disfavor mechanisms for decomposition of the ketal phosphate
that involve loss of phosphate and formation of an oxocarbonium
ion. We took advantage of these effects in synthesizing the
fluorine-substituted TI analogues 3 and 4, which are exception-
ally stable.10 Our strategy for the synthesis of the TI was
patterned on the synthesis of the stabilized analogues 3 and 4,
employing bromine in place of fluorine for stabilization of the
intermediates.10 We anticipated that the carbon-bromine bonds
could be reduced as the last step in the synthesis, which would
involve the diastereomeric dibromo-TI analogues 5 as the
immediate precursors to the TI and its stereoisomer.
Model System. To evaluate conditions that would effect
debromination of a highly charged species and allow purification
of an acid-labile product, we first explored a relatively simple
model substrate. The ketal phosphate 7 was prepared in a
straightforward manner from 2-phenethyl alcohol and methyl
dibromopyruvate as shown in Scheme 1. Methyl dibromo-
pyruvate was prepared by heating pyruvic acid with bromine
in refluxing chloroform, followed by Fischer esterification in
methanol and vacuum distillation from P2O5.13 In a one-pot
process, the hemiketal formed on combination of the alcohol
and ketone was treated successively with PCl3, 2-(p-nitrophen-
yl)ethanol, and m-chloroperbenzoic acid to generate the pro-
tected derivative 6. The phosphate esters were cleaved by
elimination of p-nitrostyrene, induced by DBU in the presence
of bis(trimethylsilyl)acetamide (BSA).10 As noted previously,10
inclusion of BSA silylates the anionic intermediates and ensures
that elimination of the second NPE ester from each phosphate
group is as rapid as that of the first. The intermediate
di(trimethylsilyl)phosphate esters are hydrolyzed along with the
lactone and carboxyl ester on subsequent exposure to 1 N
NaOH.
We have now removed any remaining ambiguity as to the
mechanism of EPSPS through the synthesis and stereochemical
assignment of both diastereomers of the TI itself. The observa-
tion that only one of them is processed by the enzyme defines
the stereochemical course of the EPSPS reaction and provides
the last piece in the fascinating puzzle of the shikimate-
chorismate pathway.
Synthetic Strategy. Acyclic hemiacetal and hemiketal phos-
phates were apparently unknown prior to the isolation of the
EPSPS TI. Direct synthesis by phosphorylation of a hemiketal
hydroxyl group poses two challenges: first, in the bimolecular
equilibrium with alcohol and ketone, the hemiketal is usually
disfavored, and second, the product is exceptionally prone to
decomposition if the phosphate group is esterified or protonated.
The first problem is overcome in the cyclic analogues (for
example, the anomeric phosphates of furanoses and pyranoses)
A variety of reductive conditions were assessed in aqueous
or alcoholic solutions. Dissolving metal reagents (e.g., with Na,
Li, Ca, Mg, and Zn), chromous ion, borohydrides, and catalytic
hydrogenation conditions were all ineffective. Radical reduction
by photolysis in the presence of AIBN and the water-soluble
(12) Skarzynski, T.; Kim, D. H.; Lees, W. J.; Walsh, C. T.; Duncan, K.
Biochemistry 1998, 37, 2572.
(13) Cooper, D. J.; Owen, L. N. J. Chem. Soc. C 1966, 533.
9
J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003 12761

A R T I C L E S
An et al.
Scheme 2
tris(3-(2-methoxyethoxy)propyl)tin hydride reagent introduced
by Breslow was also evaluated.14 The photoinitiated process
proved to be the most successful using the conventional
tributyltin hydride reagent in combination with cetyltrimethyl-
ammonium bromide (CTAB) as a solubilizing detergent. An
emulsion of the substrate, tin hydride, detergent, and AIBN in
a NaHCO3/Na2CO3 buffer was stirred under irradiation with a
mercury vapor lamp. At the end of the reaction, the detergent
and neutral reagents were adsorbed onto a cation-exchange resin
and readily separated from the anionic product and byproducts.
The desired pyruvate ketal phosphate 9 was detected by the
1
characteristic methyl singlet at δ 1.73 ppm in the H NMR
spectrum, which is replaced by the singlet of pyruvic acid at δ
2.43 ppm on acidification.
A major byproduct in this reduction, formed in varying
amounts, proved to be the cyclic ketal phosphate 10, arising
from cyclization of the monobromo intermediate 8 in competi-
tion with the second reductive step. Two modifications to the
cocktail used to solubilize the reagents and substrate were found
that reduced the amount of cyclic product. The first was the
inclusion of zinc ion, in the form of ZnCl2, which was identified
from a series of metal salts screened for their potential chelation
by the phosphate-carboxylate trianion. We reasoned that such
a chelate would reduce the nucleophilicity of the phosphate
moiety and slow cyclization. Indeed, when 10 equiv of ZnCl2
was included in the photolysis mixture, nearly quantitative
conversion to the desired ketal phosphate 9 was observed. A
more straightforward method for reducing competition from
cyclization was to carry out the reduction in methanol. In this
solvent, a homogeneous solution was obtained even in the
absence of a detergent (tetrabutylammonium iodide was sub-
stituted for CTAB), and the photochemically induced radical
debromination was more rapid. Unfortunately, it turned out that
neither of these modifications could be applied when it came
time to reduce the TI precursors 5 (see below), but recognizing
the need for a homogeneous process turned out to be significant.
Synthesis of Diastereomeric Dibromo Ketal Phosphates.
The ketal phosphate moiety of the EPSPS tetrahedral intermedi-
ate was assembled in the same one-pot process employed for
the model compound 7 (Scheme 2). A 2.2 M solution of methyl
shikimate 3,4-acetonide 1115 and 8.2 M methyl dibromopyru-
vate13 in CH2Cl2 was stirred with molecular sieves. Such high
concentrations of starting materials and the excess pyruvate were
required to favor formation of hemiketal 12. After successive
treatments with PCl3, 2-(p-nitrophenyl)ethanol, and mCPBA,
the phosphate 13 was isolated in 49% overall yield as a ∼1:1
mixture of R/S-diastereomers. The efficiency of this four-step,
one-pot sequence may be limited by the intrinsic equilibrium
between hemiketal 12 and the starting materials, because the
yield was reduced to 29% when only 1 equiv of the dibromo-
pyruvate was used.
lactonization of one of the product diastereomers on silica gel
chromatography further complicated the optimization process.
Fortunately, we eventually found that treatment of the mixture
of acetonides 13 in 9:1 AcOH/0.1 N HCl affords the diols 14
reproducibly in greater than 80% yield.
In the subsequent lactonization step, the markedly different
reactivity of the two ketal phosphate diastereomers became
apparent. One lactone diastereomer, eventually shown to be S,
not only forms more readily but also decomposes more rapidly.
If separation of the diol isomers 14 was attempted directly, much
of the S-isomer converted to the lactone 15 on chromatography.
However, if lactonization of the mixture of diols 14 was driven
to completion prior to separation, the proportion of (S)-15 in
the product was significantly reduced. These problems were
circumvented by adsorbing the diol mixture onto silica gel to
effect lactonization of the S-isomer; elution then afforded the
(S)-lactone (S)-15 in 25% yield and the (R)-diol (R)-14 in 26%
yield, based on acetonide 13. Subsequently, diol (R)-14 was
lactonized with K2CO3 and molecular sieves in acetonitrile to
afford lactone (R)-15 in 66% yield.
The next step, selective cleavage of the acetonide ketal in
the presence of the ketal phosphate, put our stabilization strategy
to the test and required considerable optimization. Some mild
conditions were ineffective (e.g., iodine in methanol), while
others resulted in decomposition or complex product mixtures
(e.g., methanolic solutions in the presence of sulfonic acid resins,
toluenesulfonic acid, 1 N HCl, or FeCl3 on silica). Ready
The remarkable dichotomy in chemical reactivity between
the stereoisomers is ascribed to the different anomeric effects
in the (S)- and (R)-lactones. The shikimate 5-O in (S)-lactone
(14) Light, J.; Breslow, R. Tetrahedron Lett. 1990, 31, 2957.
(15) Diels, O.; Fritsche, P. Chem. Ber. 1911, 44, 3018.
9
12762 J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003

5-Enolpyruvylshikimate 3-Phosphate Synthase
A R T I C L E S
Table 1. Chemical Shift Trends among Ketal Phosphate Lactones
in CDCl3 Solvent
(S)-15 shares its axial lone pair electrons with the σ* antibonding
orbital of the antiperiplanar pseudoaxial C-O ketal phosphate
bond. While anomeric stabilization favors formation of this
lactone in comparison to the R-isomer, it also lowers the barrier
to its decomposition. Thus, (S)-lactone (S)-15 is at once
thermodynamically more stable and kinetically less stable than
(R)-lactone (R)-15.16
The (S)- and (R)-lactones 15 were carried forward in parallel
to the TI diastereomers (Scheme 2). The 3-hydroxyl groups
were phosphitylated with di-2-(4-nitrophenyl)ethyl-N,N-diiso-
propylphosphoramidite catalyzed by tetrazole.10 The resulting
phosphites were oxidized in situ with mCPBA to give the (S)-
and (R)-bis(phosphates) 16 in 69% and 73% yields, respectively.
The bisphosphates 16 (of MW 1172!) include in protected form
all the components of the TI (MW 404). In the subsequent
deprotection steps, the anionic phosphates of the lactones 16
were unveiled first. The p-nitrophenethyl groups were eliminated
as described for the model compound 6 by stirring a solution
of the starting material in dry acetonitrile containing BSA and
DBU;10 subsequent exposure to 1 N NaOH then hydrolyzed
the phosphate silyl esters along with the lactone and carboxyl
ester. The fully deprotected dibromo hexaanions 5 were isolated
and purified via anion exchange chromatography, eluting with
a gradient of triethylammonium bicarbonate. A small amount
of 1 M aqueous NaOH was added to each fraction before
lyophilization to prevent acidification of the product caused by
removal of triethylamine. Although the hexasodium salts were
isolated in good overall yields (65-81%), they were contami-
nated with comparable amounts of Na2CO3 and NaHCO3 as a
consequence of this basification step.
S-isomer (δ, ppm)
R-isomer (δ, ppm)
compd
Y
CX
H4
H5
H4
H5
3
17a
18b
19b
20ad
20bd
15
PO3Bn2
PO3NPE2
CH3
CF3
4.45
4.65
4.44
4.45
4.56
4.65
4.75
4.88
5.14
5.02
5.1
4.46
4.95
4.92
4.92
5.05
3.99
4.46
4.68
4.70
4.98
c
c
H
H
H
H
CHF2
CHBrF
CHBrF
CHBr2
CHBr2
5.1
5.10
4.98-5.17e
4.92-5.25f
16
PO3NPE2
5.15e
a Reference 9; phosphonate (PO3Bn2) in place of the ketal phosphate
moiety; stereochemical designation (CIP) reversed. b Reference 10. c NPE
) 2-(p-nitrophenyl)ethyl. d 20a and 20b are diastereomers due to the CHBrF
stereocenter (configuration unknown): D. Alberg and U. Neidlein, unpub-
lished results. e H3 and H5 not clearly distinguished. f H4 and H5 not clearly
distinguished.
rapid, the monobromo intermediate undergoes intramolecular
displacement to form the cyclic phosphate (e.g., 10). The
challenge of solubilizing both the hydrophobic tributyltin
hydride reagent and the charged dibromo ketal phosphate model
substrate 7 could be solved with a cationic surfactant in aqueous
solution or by using methanol as the solvent. However, these
conditions were not effective for the hexaanionic substrates 5.
In the water-CTAB medium, a complex mixture was obtained
in which the major product appeared to be the cyclic phosphate
21; addition of zinc chloride to this mixture simply led to an
insoluble precipitate. Methanol could not be used directly as
solvent, since it does not dissolve the hexaanions either.
However, we were able to achieve a homogeneous solution by
dissolving the substrate in a small amount of water, diluting
with an equal volume of methanol, and then adding this solution
to a nearly saturated solution of CTAB in methanol. In this way,
a 2.4 mM solution of the dibromo ketal phosphate (S)-5 in 95:5
methanol/water could be prepared. After addition of tributyltin
hydride and AIBN, photolysis, and isolation by anion exchange,
the desired product, (R)-TI, was obtained in 60% yield. The
diastereomeric substrate, (R)-5, was typically isolated from the
column fractions with a larger amount of sodium carbonate and
bicarbonate salts; as a consequence, homogeneous solutions of
this material proved to be more difficult to achieve. However,
we ultimately succeeded with a 50:45:5 methanol/DMSO/water
combination. A 0.5 mM solution of (R)-5 in this mixture
afforded the (S)-TI product in 50% yield, contaminated with a
small amount of the cyclic phosphate (S)-21.17
Assignment of Ketal Configuration. The crucial stereo-
chemical assignment of the ketal phosphate configurations in
the bicyclic lactones 15 and 16 was accomplished in two
ways: by direct NOE analysis and by NMR comparison to
related structures whose configurations had been determined
unambiguously by proton-fluorine NOE experiments involv-
ing both isomers.9,10 In the (R)-hydroxylactone (R)-15, a weak
nuclear Overhauser effect is observed between the dibromo-
methyl hydrogen and the shikimate H5 hydrogen, reflecting their
cis relationship. No comparable NOE (e.g., to shikimate H4) is
observed for the other diastereomer, which we attribute to the
pseudoequatorial orientation of the dibromomethyl group in (S)-
15.
1
The relative H NMR chemical shifts of the H4 and H5
hydrogens in the lactone derivatives are strikingly different,
depending upon the configuration of the ketal stereocenter. In
structures as divergent as the methyl phosphonates 17 and the
difluoromethyl, trifluoromethyl, and bromofluoromethyl phos-
phates 18-20, the H5 hydrogen resonates downfield of H4
when the phosphorus moiety is axial (see Table 1). Their rela-
tive positions are reversed when the methyl or halomethyl
substituent is axial. Assignment of the S-isomers in the
dibromomethyl series is clear-cut by this criterion, although
in the case of the R-isomers there is little separation between
these hydrogens.
Radical Debromination. As noted above, smooth reduction
of the dibromo ketal phosphate depends on a homogeneous
reaction medium. If the two reduction steps are not reasonably
The stability of the TI diastereomers is highly pH dependent,
as expected. At pH 10.5, (R)-TI has a half-life on the order of
days in aqueous solution at room temperature. When the pH is
(16) We observed similar differences in the rates of lactone formation in our
earlier work with the trifluoro and difluoro analogues.10
9
J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003 12763

A R T I C L E S
An et al.
Figure 4. 1H NMR spectrum of (R)-TI and difference spectra in the presence of EPSPS.
Table 2. 1H NMR Chemical Shifts in an Enzyme-Buffer Mixturea
adjusted to 7.0, (R)-TI decomposes to S3P and pyruvate with a
half-life of approximately 10 min.
Assignment of the Configuration of the EPSP Synthase
1
Tetrahedral Intermediate. Comparison of the published H
NMR spectrum of the TI isolated from the enzymatic reaction4
with those of the individual R- and S-diastereomers provided
the first suggestion that the natural isomer possesses the
S-configuration. However, a definitive assignment from a direct
correspondence in chemical shifts was not possible in view of
the different counterions and buffer solutions employed.
Conclusive identification of the enzymatic TI as the (S)-ketal
phosphate came from incubation of the synthetic diastereomers
with EPSPS to determine which one is susceptible to enzymatic
conversion. (R)-TI (4 µmol) and (S)-TI (3 µmol) were each
incubated with 0.16 unit of EPSPS in D2O at pD 9 in an NMR
tube, and the 1H NMR spectra were recorded at 30 min intervals
at room temperature. Reaction progress was conveniently
monitored by subtracting the first spectrum from each subse-
quent spectrum, thereby emphasizing the changes and minimiz-
ing the peaks due to solvent, buffer, and counterions. Although
cancellation of the ancillary signals is imperfect, it is apparent
that (R)-TI is stable in the presence of the enzyme; none of the
peaks for this compound disappear, and no new peaks appear
(Figure 4). In contrast, in the presence of the enzyme, the peaks
for (S)-TI disappear to be replaced by those of S3P, PEP, and
EPSP (Figure 5, Table 2). The peaks for EPSP grow in more
prominently than those for S3P and PEP because the enzyme-
TI complex partitions to EPSP in preference to S3P and PEP
(Figure 6). A 3:1 preference can be determined most readily
from the peaks for the enolpyruvyl hydrogens of PEP (δ 5.17
and 5.38) and EPSP (δ 4.70 and 5.20). Indeed, the observed
ratio of these products agrees well with the 5:1 ratio of forward
and reverse rate constants determined by Anderson and Johnson
and observed in their treatment of the natural TI with EPSPS.6,7
hydrogen
S3P
PEP
(S)-TI
EPSP
H2
H4
H5
H6R
H6â
CH3
Hz
6.47
(3.8)
4.0
2.7
2.20
6.46
3.93
4.25
2.69
2.23
1.76
6.52
4.06
4.45
2.90
2.23
5.17
5.38
4.70
5.20
He
a NaHCO3/Na2CO3 buffer, pD 9.
Integrated Stereochemical Mechanism of EPSP Synthase.
Assignment of the S-configuration to the ketal stereocenter in
the EPSPS TI completes the stereochemical definition of the
reaction catalyzed by this enzyme. The TI is formed by si face
addition of the proton to the PEP double bond coupled with
anti addition of the S3P hydroxyl oxygen; the TI then collapses
by syn elimination of the methyl proton and phosphate (see
Figure 2). In combination with the recently reported crystal
structure of the complex of EPSPS with S3P and glyphosate,11
this stereochemical insight helps to define the chemical mech-
anism of the addition-elimination reaction and the active site
residues that play a key role.
The (S)-TI can be readily modeled onto the structures of S3P
and glyphosate in their bound conformations. The position of
glyphosate in the active site (Figure 3) is indicative of the
orientation of the carboxyl and phosphate moieties of PEP
(Figure 7a) and, in turn, of the ketal phosphate side chain of
the TI (Figure 7b). As Scho¨nbrunn et al. have concluded from
their X-ray analysis of the EPSPS-S3P-glyphosate complex,
the roles of the active site residues, Asp-313 and Glu-341, are
apparent.11,28 Glu-341 is positioned above the si face of the
double bond of PEP, opposite the 5-OH group of S3P, and
serves as the proton donor in the anti addition step leading to
the TI. Asp-313 is located adjacent to the 5-OH group of S3P
(17) Although we worked hard to minimize formation of the cyclic ketal
phosphate (S)-21 as a byproduct in the synthesis of the (S)-TI, this stable
analogue of the TI may be of interest as a potent inhibitor of EPSPS.
Modeling suggests that this structure would mimic very closely the bound
conformation predicted for the (S)-TI (see Figure 7c). Although we did
not pursue this idea directly, we noted that enzymatic conversion of (S)-TI
to its products was slowed in the presence of (S)-21 as a contaminant.
9
12764 J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003

5-Enolpyruvylshikimate 3-Phosphate Synthase
A R T I C L E S
Figure 5. 1H NMR spectrum of (S)-TI and difference spectra in the presence of EPSPS.
and can thus serve as the proton acceptor as the carbon-oxygen
bond is formed.
but there are structural features in the intermediates of both DHQ
synthase and EPSPS which make these reactions possible.
Interestingly, quite opposite effects are at play in making these
substrates prone to elimination (Figure 8). In the case of DHQ
synthase, cleavage of the C-H bond likely precedes cleavage
of the C-O bond, with the adjacent carbonyl group stabilizing
the partial negative charge that builds up in the transition state.
In contrast, the ether oxygen in the EPSP TI stabilizes an
adjacent positive charge, so that C-O cleavage precedes C-H
cleavage in loss of phosphate from this intermediate. In the
active site orientation proposed for the TI (Figure 7b), an
electron lone pair on the shikimate 5-oxygen is antiperiplanar
to the carbon-phosphate bond, and thus in a position to provide
anchimeric assistance in its cleavage. The proposal that phos-
phate induces its own elimination not only follows from the
stereochemical constraints, but is also consistent with evidence
that the transition state for the elimination step is more ionic
than that for the addition step. While the fluoromethyl-TI can
Specific roles for enzymatic residues in the syn elimination
of phosphate from the TI are not evident. The most logical
enzymatic base is the original proton donor, Glu-341. However,
this carboxylate is oriented 90° to the C-O bond that must be
broken (Figure 7b); to bring the TI and Glu-341 into alignment
for syn elimination would require a drastic and probably
impossible reorientation of the enzyme-TI complex. No other
active site residues are in a position to effect syn elimination.
The base in the most advantageous site to remove the methyl
proton syn to the phosphate C-O bond is the phosphate itself;
we suggest that the elimination is thus intramolecular, without
direct action by any enzymatic group. This possibility has also
been raised by Walsh et al. from their analysis of the related
MurA mechanism.12
Self-elimination of phosphate has also been proposed as a
key step in the mechanism of dehydroquinate synthase.2c
Phosphate derivatives in general are not particularly unstable,
Figure 6. Difference spectrum from treatment of (S)-TI with EPSPS (bottom) and reference spectrum of an equimolar mixture of S3P, PEP, and EPSP
(top).
9
J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003 12765

A R T I C L E S
An et al.
Figure 8. Contrasting mechanisms for self-elimination of phosphate.
Figure 9. Acids and bases involved in EPSPS-catalyzed synthesis,
illustrated in the forward direction.
EPSP requires phosphate ion,20 in contrast to hydrogen exchange
on PEP, which the enzyme is able to catalyze in the presence
of 4,5-dideoxy-S3P.21 These results were first thought to be
inconsistent, with the latter observation offered in support of
an enzyme-bound PEP adduct and the former as evidence for
the noncovalently bound TI. In fact, these results are fully in
accord with expectation if an enzymatic acid/base is responsible
for the half-reaction interconverting S3P and PEP with the TI
and if phosphate is the acid/base responsible for its addition/
elimination from the TI (Figure 9).22
For the elimination step, it appears that the enzyme simply
serves as a template to complement the steric and electronic
characteristics of the transition state. This role mirrors that now
ascribed to other enzymes of the shikimate-chorismate pathway,
namely, DHQ synthase and chorismate mutase. For DHQ
synthase, direct involvement of the enzyme side chains and the
Figure 7. Model of the EPSPS active site with bound substrates: (a) S3P
and PEP; (b) (S)-TI; (c) EPSP and Pi.
be generated enzymatically from fluoro-PEP,18 it is not inter-
converted with fluoro-EPSP.19
(20) Wibbenmeyer, J.; Brundage, L.; Padgette, S. R.; Likos, J. J.; Kishore, G.
M. Biochem. Biophys. Res. Commun. 1988, 153, 760-766.
(21) Anton, D. L.; Hedstrom, L.; Fish, S. M.; Abeles, R. H. Biochemistry 1983,
22, 5903-5908.
The conversion of S3P and PEP to EPSP and Pi as catalyzed
by EPSPS is fully reversible.7 If phosphate elimination is indeed
self-induced in the forward direction, the principle of micro-
scopic reversibility dictates that phosphate itself provides the
proton to promote its addition to EPSP to form the TI in the
reverse direction. Wibbenmeyer et al. demonstrated that enzyme-
catalyzed solvent exchange of the enolpyruvyl hydrogens on
(22) The protonation state of the phosphate moiety involved in the elimination
step is thus of interest. It is clear that the TI is quite stable in basic solution,
since the phosphate trianion is a poor leaving group. Viewed from the other
direction, HPO42-, with a pKa of 12.3, is probably not acidic enough to
protonate the EPSP double bond. Thus, a reasonable assumption is that
the ketal phosphate moiety of the TI carries a single proton and a single
-
negative charge, eliminating H2PO4 (pKa ) 7.2) on formation of EPSP.
However, the crystal structure of the EPSPS complex with S3P and
glyphosate shows that each of the phosphonate oxygens of glyphosate is
within 3 Å of a positively charged nitrogen on Lys-22, Lys-411, or Arg-
124.11 Thus, the charged residues in the active site may stabilize the ketal
phosphate anion of the TI sufficiently to enable PO43- to serve as the leaving
(18) Kim, D. H.; Tucker-Kellogg, G. W.; Lees, W. J.; Walsh, C. T. Biochemistry
1996, 35, 5435-5440.
(19) Seto, C. T.; Bartlett, P. A. J. Org. Chem. 1994, 59, 7130-7132.
2-
group in the forward direction and HPO4 as the acid in the reverse
direction.
9
12766 J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003

5-Enolpyruvylshikimate 3-Phosphate Synthase
A R T I C L E S
NAD cofactor appears to be required only for the steps involving
oxidation and reduction; the phosphate elimination and final
aldol cyclization may occur spontaneously in the active site.23,24
For the chorismate mutases, which accelerate the formal [3,3]-
sigmatropic rearrangement of chorismate to prephenate, catalysis
results simply from binding the substrate in the reactive, diaxial
conformation and from stabilization of the charge that builds
up briefly in the asymmetric transition state.25-27
substrate-induced. EPSPS thus joins two of its clever compan-
ions in the shikimate pathway that get their substrates to do
some of the work.
Acknowledgment. This work was supported by a grant from
the National Institutes of Health (GM28965), by a fellowship
from the Deutsche Forschungsgemeinschaft (to U.N.), and by
a Fulbright travel grant (to U.M.). We thank Murtaza Alibhai,
Douglas Sammons, and Claire CaJacob of Monsanto Corp. for
a gift of EPSP synthase.
Conclusion
Supporting Information Available: Experimental procedures
for the synthesis and characterization of the tetrahedral inter-
mediates (R)-TI and (S)-TI. This material is available free of
charge via the Internet at http://pubs.acs.org.
The chemical synthesis and stereochemical assignment of the
tetrahedral intermediate of EPSPS not only defines the mech-
anism of this enzyme, but completes the stereochemical assign-
ment of all of the enzymatic steps of the shikimate-chorismate
pathway. The roles proposed previously11 for the catalytic
residues Gly-341 and Asp-313 in the anti addition step are fully
consistent with the (S)-TI configuration.28 Moreover, the ap-
parent absence of any enzymatic residues that could serve as
proton acceptors for the syn elimination step suggest that it is
JA036627+
(28) After the preparation of our paper, Mizyed et al. reported the effects of
active site mutations on the partitioning of enzymatically synthesized TI
to PEP and EPSP.29 All 14 residues within 5 Å of a reactive atom from the
TI were mutated individually; while many mutations reduced the catalytic
activity significantly, none produced a large change in the partition ratio.
Mizyed et al. thus concluded that the active site residues play similar acid/
base roles in the breakdown of the TI, regardless of the direction. They
proposed Glu-341 and Lys-22 as the dual-role acid/base catalysts; that is,
Glu-341 would be the proton donor in formation of the TI from S3P and
PEP and the proton acceptor in elimination of the TI to EPSP and Pi. Lys-
22 would serve as the proton acceptor from S3P in the first step and the
general-acid catalyst for phosphate elimination in the second. This
interpretation contrasts with our mechanistic proposal and those offered
previously,11,12 on the basis of the orientation of the TI in the active site.
(29) Mizyed, S.; Wright, J. E. I.; Byczynski, B.; Berti, P. J. Biochemistry 2003,
42, 6986-95.
(23) Bartlett, P. A.; McLaren, K. L.; Marx, M. A. J. Org. Chem. 1994, 59,
2082-2085.
(24) Carpenter, E. P.; Hawkins, A. R.; Frost, J. W.; Brown, K. A. Nature 1998,
394, 299-302.
(25) Guilford, W. J.; Copley, S. D.; Knowles, J. R. J. Am. Chem. Soc. 1987,
109, 5013-5019.
(26) Lee, A. Y.; Zhang, S.; Kongsaeree, P.; Clardy, J.; Ganem, B.; Erickson, J.
W.; Xie, D. Biochemistry 1998, 37, 9052-9057.
(27) Gustin, D. J.; Mattei, P.; Kast, P.; Wiest, O.; Lee, L.; Cleland, W. W.;
Hilvert, D. J. Am. Chem. Soc. 1999, 121, 1756-1757.
9
J. AM. CHEM. SOC. VOL. 125, NO. 42, 2003 12767