Nonenzymatic breakdown of the tetrahedral (α-carboxyketal phosphate) intermediates of MurA and AroA, two carboxyvinyl transferases. Protonation of different functional groups controls the rate and fate of breakdown

Article abstract of DOI:10.1021/ja0349655

The mechanisms of nonenzymatic breakdown of the tetrahedral intermediates (THIs) of the carboxyvinyl transferases MurA and AroA were examined in order to illuminate the interplay between the inherent reactivities of the THIs and the enzymatic strategies used to promote catalysis. THI degradation was through phosphate departure, with C-O bond cleavage. It was acid catalyzed and dependent on the protonation state of the carboxyl of the α-carboxyketal phosphate functionality, with ionizations at pK_a = 3.2 ± 0.1 and 4.3 ± 0.1 for MurA and AroA THIs, respectively. The solvent deuterium kinetic isotope effect for MurA THI at pL 2.0 was 1.3 ± 0.4, consistent with general acid catalysis. The pK_a's suggested intramolecular general acid catalysis through protonation of the bridging oxygen of the phosphate, though H₃O⁺ catalysis was also possible. The product distribution varied with pH. The dominant breakdown products were {pyruvate + phosphate + R-OH} (R-OH = UDP-GlcNAc or shikimate 3-phosphate) at all pH's, particularly low pH. At higher pH's, increasing proportions of ketal, arising from intramolecular substitution of phosphate by the adjacent hydroxyl and the enolpyruvyl products of phosphate elimination were observed. With MurA THI, the product distribution fitted to pK _a's 1.6 and 6.2, corresponding to the expected pK_a's of a phosphate monoester. C-O bond cleavage was demonstrated by the lack of monomethyl [³³P]phosphate formed upon degrading MurA [ ³³P]THI in 50% methanol. General acid catalysis through the bridging oxygen is consistent with the location of the previously proposed general acid catalyst for THI breakdown in AroA, Lys22.

Full text of DOI:10.1021/ja0349655

Published on Web 09/20/2003
Nonenzymatic Breakdown of the Tetrahedral (r-Carboxyketal
Phosphate) Intermediates of MurA and AroA, Two
Carboxyvinyl Transferases. Protonation of Different
Functional Groups Controls the Rate and Fate of Breakdown
Bartosz Byczynski,†,§ Shehadeh Mizyed,†,§ and Paul J. Berti*,†,‡,§
Contribution from the Department of Chemistry, Department of Biochemistry, and
Antimicrobial Research Centre, McMaster UniVersity, 1280 Main Street West,
Hamilton, ON, L8S 4M1, Canada
Received March 3, 2003; E-mail: berti@mcmaster.ca
Abstract: The mechanisms of nonenzymatic breakdown of the tetrahedral intermediates (THIs) of the
carboxyvinyl transferases MurA and AroA were examined in order to illuminate the interplay between the
inherent reactivities of the THIs and the enzymatic strategies used to promote catalysis. THI degradation
was through phosphate departure, with C-O bond cleavage. It was acid catalyzed and dependent on the
protonation state of the carboxyl of the R-carboxyketal phosphate functionality, with ionizations at pKa )
3.2 ( 0.1 and 4.3 ( 0.1 for MurA and AroA THIs, respectively. The solvent deuterium kinetic isotope effect
for MurA THI at pL 2.0 was 1.3 ( 0.4, consistent with general acid catalysis. The pKa’s suggested
intramolecular general acid catalysis through protonation of the bridging oxygen of the phosphate, though
H3O+ catalysis was also possible. The product distribution varied with pH. The dominant breakdown products
were {pyruvate + phosphate + R-OH} (R-OH ) UDP-GlcNAc or shikimate 3-phosphate) at all pH’s,
particularly low pH. At higher pH’s, increasing proportions of ketal, arising from intramolecular substitution
of phosphate by the adjacent hydroxyl and the enolpyruvyl products of phosphate elimination were observed.
With MurA THI, the product distribution fitted to pKa’s 1.6 and 6.2, corresponding to the expected pKa’s of
a phosphate monoester. C-O bond cleavage was demonstrated by the lack of monomethyl [33P]phosphate
formed upon degrading MurA [33P]THI in 50% methanol. General acid catalysis through the bridging oxygen
is consistent with the location of the previously proposed general acid catalyst for THI breakdown in AroA,
Lys22.
The mechanisms of enzymatic reactions reflect an interplay
between the inherent reactivities of the substrate(s) and whatever
catalytic strategies may be employed to lower activation
energies. To understand enzyme mechanisms, we must therefore
also understand the analogous nonenzymatic reactions to learn
which catalytic strategies will be effective and which will not.
The enzymes MurA1-6 and AroA (EPSP synthase)7-10 have
been extensively studied because of their novel catalytic
mechanisms and because they are targets for biocidal agents.11,12
They are the only two enzymes known to catalyze carboxyvinyl
transfer with phosphoenolpyruvate (PEP). Both enzymes cata-
lyze their reactions through fully reversible addition/elimination
mechanisms that pass through tetrahedral intermediates (THIs,
Figure 1).
Relatively little is known about the stability and reactivity
of the THIs; they appear to be the only examples in the chemical
literature of acyclic R-carboxyketal phosphates. Two compounds
with related functionalities are CMP-NeuAc (1) and CMP-
deoxyoctonate (2), which possess R-carboxyketal phosphodiester
groups.
† Department of Chemistry.
‡ Department of Biochemistry.
§ Antimicrobial Research Centre.
(1) Abbreviations: AroA, EPSP synthase; AroAH6, AroA with a C-terminal
extension terminating with His6; CMP, cytidine 5′-monophosphate; CMP-
NeuAc, cytidine 5′-monophosphate N-acetyl neuraminic acid; EPSP,
enolpyruvylshikimate 3-phosphate; EP-UDP-GlcNAc, enolpyruvyl uridine
diphospho N-acetyl-glucosamine; KIE, kinetic isotope effect; MurA, EP-
UDP-GlcNAc synthase; HMBC, heteronuclear multiple bond correlation
spectroscopy; NeuAc, N-acetyl neuraminic acid; PEP, phosphoenolpyruvate;
Pi, inorganic phosphate; PMSF, phenylmethylsulfonyl fluoride; S3P,
shikimate 3-phosphate; THI, tetrahedral intermediate; UDP-GlcNAc, uridine
diphospho N-acetylglucosamine.
(2) Gunetileke, K. G.; Anwar, R. A. J. Biol. Chem. 1968, 243, 5770.
(3) Marquardt, J. L.; Siegele, D. A.; Kolter, R.; Walsh, C. T. J. Bacteriol.
1992, 174, 5748.
(4) Marquardt, J. L.; Brown, E. D.; Walsh, C. T.; Anderson, K. S. J. Am. Chem.
Soc. 1993, 115, 10398.
THIs are unstable, especially at low pH. MurA THI was
reported to have half-lives of 38 s at pH 5 and 2 h at pH 8,13
while AroA THI half-lives were 45 min at pH 7 and >48 h at
(5) Lees, W. J.; Walsh, C. T. J. Am. Chem. Soc. 1995, 117, 7329.
(6) Schonbrunn, E.; Sack, S.; Eschenburg, S.; Perrakis, A.; Krekel, F.; Amrhein,
N.; Mandelkow, E. Structure 1996, 4, 1065.
(7) Anderson, K. S.; Johnson, K. A. Chem. ReV. 1990, 90, 1131.
9
10.1021/ja0349655 CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 12541-12550
12541

A R T I C L E S
Byczynski et al.
respectively) using pyruvate phosphate dikinase16 (a generous gift from
Prof. D. Dunaway-Mariano, University of New Mexico) as described
previously.17 [33P]PEP was prepared from [γ-33P]ATP essentially as
described previously.18 The AroA ketal product was prepared as
described previously19 and purified under the same conditions as AroA
THI (see below).
Enzymes. Recombinant E. coli AroA bearing a C-terminal His6 tag
to allow Ni2+-affinity purification, AroAH6, was prepared as described
previously.17
Recombinant E. coli MurA was PCR amplified from E. coli strain
DH5R and ligated into the pET41a plasmid (Novagen, Inc.), replacing
the NdeI-BamHI fragment. The insert was sequenced and found to
have the same amino acid sequence as that of Genbank entry P28909.
The MurA-containing plasmid was transformed into E. coli strain
BL21*(DE3) (Invitrogen).
LB broth containing 30 µg/mL kanamycin was inoculated with 1
vol % of overnight culture and grown at 37 °C until OD600 ) 0.6.
MurA expression was induced with 1mM isopropyl-â-D-thiogalacto-
pyranoside (IPTG) for 4 h at 37 °C. The cells were harvested by
centrifugation at 5000 × g for 10 min, and the pellet was resuspended
in 10 mL per L of culture of buffer A (50 mM Tris‚HCl, pH 7.5, 1
mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM
benzamidine) and frozen or lysed immediately.
Figure 1. Reactions catalyzed by the carboxyvinyl transferases MurA and
AroA. R-OH is uridine 5′-diphospho N-acetylglucosamine (UDP-GlcNAc)
for MurA and shikimate 3-phosphate (S3P) for AroA. The reactive hydroxyls
are underlined.
pH 12.14 At pH 4, the products were S3P, pyruvate, and
phosphate (Pi) (3). At pH 7 the products also included
enolpyruvyl S3P (EPSP, 5) and the ketal (4) formed by
substitution of the phosphate with S3P O4′H.
All subsequent steps were performed at 0-4 °C. DNase and RNase
(100µg/mL) were added to the resuspended cells immediately before
lysis in a French press. Cell debris was removed by centrifugation at
10 000 × g for 10 min. The protein was precipitated in 70% saturated
ammonium sulfate and pelleted by centrifugation at 20 000 × g for 10
min, followed by resuspension in buffer A + 1 M ammonium sulfate.
Insoluble material was removed by centrifugation at 20 000 × g for
10 min. MurA was purified by hydrophobic interaction chromatography
on a 40 mL Phenyl Sepharose (high sub, Amersham Biosciences)
column, with elution on a gradient of 1-0 M ammonium sulfate in
buffer A over five column volumes at a 5 mL/min flow rate with A280
detection. MurA eluted at near 0.5 M ammonium sulfate. The buffer
was exchanged by diafiltration over a YM-10 (Amicon) membrane to
buffer A. It was further purified by anion exchange chromatography
on a 40 mL Q-Sepharose (fast flow, Amersham Biosciences) column.
MurA was eluted on a gradient of 0-1 M KCl in buffer A. MurA was
the first of two peaks eluted, near 0.2 M KCl. It was concentrated over
a YM-10 membrane to a concentration of 1-3 mM (50-150 mg/mL).
AroA accelerates THI breakdown by >106-fold in order to
achieve the observed rates of catalytic turnover.14 The mecha-
nistic imperative of the enzymes is thus 2-fold, to promote THI
breakdown and at the same time to form the correct enolpyruvyl
products, not {R-OH15 + pyruvate + Pi}. The acid lability of
the THIs implies that acid catalysis promotes phosphate and/or
R-OH departure. However, important questions remain, includ-
ing the following: (i) what atom or functional group is
protonated to promote breakdown; (ii) how effective can acid
catalysis be; that is, how much can it accelerate the reaction;
(iii) how does this compare with the catalytic enhancement
attributed to the general acid catalyst in the AroA-catalyzed
reaction; and (iv) what controls the product distribution?
Answering these questions will help answer the question of how
AroA and MurA accelerate and control THI breakdown.
Studies on the THIs have been limited by the difficulty in
making them. They are obtained in substoichiometric amounts
by quenching enzymatic reactions in KOH, yielding 50-200
nmol of THI per synthesis. By producing gram amounts of E.
coli AroA and MurA and using them to synthesize THIs, we
have been able to characterize the kinetics and products of THI
breakdown over a pH range of 1-12.
The protein concentration was determined from the A280, using ꢀ280
)
13 250 M-1 cm-1, as determined by the method of Edelhoch.20 Aliquots
were flash frozen in liquid nitrogen and stored at -80 °C. Yields were
typically 40-75 mg/L of culture.
THI Synthesis: MurA. Reaction mixtures (up to 2000 µL)
containing 2 mM MurA, 2 mM UDP-GlcNAc, 200 µM PEP, and 50
mM potassium phosphate, pH 7.5, were prepared and quenched after
5 s with KOH to 200 mM. The quenched reaction was extracted
repeatedly with chloroform to precipitate and remove all protein. THI
was purified on a Mono-Q anion exchange column (5 × 50 mm2,
Amersham Biosciences) with a gradient of 100-500 mM KCl in 10
mM NH4Cl, pH 10 over 15 column volumes at a 0.5 mL/min flow rate
and A260 detection. MurA [14C]THI synthesis was as above but with
0.5 µCi [1-14C]PEP. In some reactions, MurA THI was purified by
anion exchange chromatography using ammonium bicarbonate, pH 10
Materials and Methods
General. [1-14C]PEP and [13C3]PEP were prepared from labeled
pyruvates (Amersham Biosciences and Cambridge Isotope Laboratories,
(14) Anderson, K. S.; Johnson, K. A. J. Biol. Chem. 1990, 265, 5567.
(8) Anderson, K. S.; Sammons, R. D.; Leo, G. C.; Sikorski, J. A.; Benesi, A.
J.; Johnson, K. A. Biochemistry 1990, 29, 1460.
(15) R-OH refers to both the O3′′H of UDP-GlcNAc and O5′H of S3P (Figure
1).
(9) Anderson, K. S.; Sikorski, J. A.; Johnson, K. A. Biochemistry 1988, 27,
1604.
(16) Wang, H. C.; Ciskanik, L.; Dunaway-Mariano, D.; von der Saal, W.;
Villafranca, J. J. Biochemistry 1988, 27, 625.
(10) Anderson, K. S.; Sikorski, J. A.; Johnson, K. A. Biochemistry 1988, 27,
7395.
(17) Mizyed, S.; Wright, J. E. I.; Byczynski, B.; Berti, P. J. Biochemistry 2003,
42, 6986.
(11) Hendlin, D.; Stapley, E. O.; Jackson, M.; Wallick, H.; Miller, A. K.; Wolf,
F. J.; Miller, T. W.; Chaiet, L.; Kahan, F. M.; Foltz, E. L.; Woodruff, H.
B.; Mata, J. M.; Hernandez, S.; Mochales, S. Science 1969, 166, 122.
(12) Sikorski, J. A.; Gruys, K. J. Acc. Chem. Res. 1997, 30, 2.
(13) Kim, D. H.; Lees, W. J.; Haley, T. M.; Walsh, C. T. J. Am. Chem. Soc.
1995, 117, 1494.
(18) Roossien, F. F.; Brink, J.; Robillard, G. T. Biochim. Biophys. Acta 1983,
760, 185.
(19) Leo, G. C.; Sikorski, J. A.; Sammons, R. D. J. Am. Chem. Soc. 1990, 112,
1653.
(20) Pace, C. N.; Vajdos, F.; Fee, L.; Grimsley, G.; Gray, T. Protein Sci. 1995,
4, 2411.
9
12542 J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003

Rate and Fate of Tetrahedral Intermediate Breakdown
A R T I C L E S
in the gradient instead of KCl. The yield of MurA THI was 3-10%
relative to the protein concentration, with a specific activity of ∼1.2
mCi/mmol in [14C]THI.
THI Synthesis: AroA. AroA THI was prepared as described
previously.17 Briefly, the quenching method was similar to MurA
reactions, but KOH and KOH-saturated 2-propanol were needed to
denature the protein. Anion exchange chromatography used a gradient
of triethylammonium chloride (Et3NH‚Cl). Triethylammonium salts
were required to resolve THI from EPSP.14,17 The yield of AroA THI
was 2.5-5% relative protein concentration.
THI Breakdown. The rate and product distribution of THI
breakdown were quantitated using HPLC separation under similar
conditions to those above, with quantitation of products and residual
THI by A260 (MurA THI only) or 14C (AroA and MurA THI). All THI
breakdown experiments were conducted at 25 °C.
dimensional 1H, 13C, and two-dimensional HMBC spectra were
collected using a 5 mm inverse geometry probe at 600 MHz for H.
1
The HMBC spectrum was collected in 128 increments with 256 scans
per increment and a delay time of 1 s between scans. The data were
linear-predicted in F1 to 1024 points before Fourier transformation. The
spectra were collected without 13C decoupling, so correlations are seen
1
to the 13C satellites of the methyl signal in the H spectrum.
Data Analysis. pKa values for rates and product distributions were
fitted by nonlinear regression to equations for single or double
ionizations (eqs 1 and 2), with proportional weighting of data points,
using the program Grafit (Erithacus Software Ltd.).
Limit1 + Limit3 × 10(pH-pK )
j
y )
(1)
10(pH-pK ) + 1
j
Purified MurA THI contained ca. 300 mM KCl from the chroma-
tography buffer, with a minimal amount of NH4Cl (10 mM) to control
pH. For breakdown reactions, the purified THI (50 µL, ∼1.5 nmol)
was diluted with 450 µL of buffer (50 mM final concentration) at the
desired pH. The buffers used were as follows: pH < 2, HCl; pH 2.5-
3.5, glycine; pH 4.0-5.5, acetate; pH 6.0-9.5, bis(trispropane); pH
7.5-8.5, Tris; pH 9.5-0.5, glycine; pH 12, KOH. Chromatographic
conditions were as those for THI synthesis but with a flow rate of 1.0
mL/min. If necessary, reactions were stopped with KOH to pH 12 before
HPLC injection. For reactions at pH > 9, KOH traps were used to
prevent atmospheric CO2 from entering the solution and lowering the
pH. Without KOH traps, THIs appeared to be stable for several days
and then quickly degraded as the pH dropped.
The solvent deuterium kinetic isotope effect (KIE) was measured at
pL 2.0 using 50 µL of MurA THI prepared as described above and
450 µL of buffer containing >99% with D2O, with the pL adjusted
with concentrated HCl. The pH meter reading was adjusted by 0.4
units: pL ) pH(meter reading) + 0.4.21 The solvent deuterium molar
fraction was thus 0.90, and the observed reaction rate was extrapolated
to a deuterium molar fraction of 1.0.
Purified AroA THI contained ca. 700 mM Et3NH‚Cl, pH 9.0 from
the chromatography buffer. In a typical reaction, 100 µL of THI (ca.
1000 cpm 14C) was mixed with 750 µL of buffer (50 mM final
concentration), which gave ca. 100 mM of Et3NH‚Cl in the reaction
mixture. The added buffer had a higher or lower pH than the target pH
to compensate for the buffering capacity of Et3NH‚Cl. The pH was
determined using test mixtures containing the same buffer composition
as those the reaction mixtures. The reaction buffers used were as
follows: pH < 4, glycine; pH 4-6, citrate; pH 6-8, Tris; pH 8-9,
bis(trispropane); pH > 9.0, triethylamine. If necessary, reactions were
stopped with KOH to pH 12 before HPLC injection.
Limit1 + Limit2 × 10(pH-pK ) (Limit2 - Limit3) × 10(pH-pK )
j
k
y )
-
10(pH-pK ) + 1
10(pH-pK ) + 1
j
k
(2)
For k-versus-pH profiles, y ) ki, the rate constant at pH ) i; Limit1
to Limit3 were kmax, kmid, and kmin at low, mid, and high pH, respectively;
and pKj ) pKa2, pKk ) pKa4. For product-versus-pH profiles, y ) f(X)i,
the fraction of product X formed at pH ) i, Limit1 to Limit3 were
f(X)low, f(X)mid, and f(X)high at low, mid, and high pH, respectively,
and pKj ) pKa1, pKk ) pKa3. The fraction of products were calculated
as for example, f(3) ) [3]/([3] + [4] + [5]), and so forth.
For AroA THI, an alternate method of was used for fitting the
k-versus-pH data. Fitting directly to eq 1 gave a curve that was visibly
below most of the data points (data not shown).23 A better fit was
achieved by a linear fit of log(k)-versus-pH and setting kmax to the
average of k at the lowest pH values. The value of kmin was estimated
by then fitting to eq 1 with kmax and pKa fixed. For some fitted curves
for product distributions, the limits at extrema were fixed; that is, f(3)low
) 1.0, f(4)low ) 0, and f(5)low ) 0. Fixing these values in the fitted
curves gave greater consistency in pKa’s. When not fixed, the fitted
values of f(X)low were within 0.004 of the fixed values.
Results
The rate constants and product distribution of THI breakdown
were determined for both the MurA and AroA THIs (Table 1,
Figures 2 and 3). It was possible to follow MurA THI
breakdown by anion exchange HPLC using absorbance detection
of the uracil ring at 260 nm or detection of the labeled products
of [1-14C]THI breakdown. Detection of AroA THI breakdown
was by 14C only as it is a poor chromophore and the anion
exchange buffer, triethylammonium chloride,14,17 had significant
UV absorbance. MurA THI breakdown was shown by UV
absorbance detection to follow first-order kinetics at all pH’s
(data not shown). AroA THI breakdown was also therefore
assumed to follow first-order kinetics, making it possible to
determine rates at a given pH with one or a few time points.
Breakdown Rates. The pH dependence of the MurA THI
breakdown rate was most consistent with two ionizations, the
most important one with pKa2 ) 3.2 ( 0.1 and a second
ionization with pKa4 ) 8.8 ( 0.3. The AroA data were fitted to
MurA [33P]THI Breakdown in Methanol. MurA [33P]THI was
reacted in 50% (v/v) methanol to determine whether there was P-O
bond cleavage. Reaction conditions were as those above except for
the inclusion of methanol. The products were separated on a Mono-Q
column with a gradient of 20-500 mM KCl in 10 mM KOH, pH 12,
over 15 column volumes at a flow rate of 0.5 mL/min. Because the
products could not be detected by absorbance, reaction mixtures were
spiked with 5 µmol of monomethyl phosphate and 0.5 µmol of
phosphate immediately before chromatography. Fractions were col-
lected, and the Malachite Green assay was used to detect phosphate.22
Monomethyl phosphate did not react directly with Malachite Green
and was first treated with alkaline phosphatase (Sigma) to release
phosphate. This made it possible to distinguish between the two
products. Monomethyl phosphate and phosphate eluted at 4.5 and 8.0
min, respectively.
(23) This was an artifact arising from fitting relatively noisy data to a logarithm-
containing equation. With proportional weighting, deviation of the fitted
curve below the data points is less unfavorable than deviation above the
data points, and therefore the fitted curve is below most of the data points.
The effect can be observed by graphing simulated data on a log scale with
error bars equal to >50% of each point. For experimental data with small
errors, such as with MurA THI, this effect is negligible. The parameters
derived from fitting directly to eq 1 were pKa ) 4.0, kmax ) 0.66 s-1, kmin
) 3 × 10-7 s-1. The differences from the values in Table 1 were chemically
insignificant.
NMR of MurA Ketal (4). The ketal product (17 µg) of MurA [13C3]-
THI breakdown at pH 7.5 was repurified and identified by NMR. One-
(21) Salomaa, P.; Schaleger, L. L.; Long, F. A. J. Am. Chem. Soc. 1964, 86, 1.
(22) Lanzetta, P. A.; Alvarez, L. J.; Reinach, P. S.; Candia, O. A. Anal. Biochem.
1979, 100, 95.
9
J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003 12543

A R T I C L E S
Byczynski et al.
Table 1. pH Dependence of Rate Constants and Product
Distributions
MurA THI
AroA THI
rate constants
pKa2
pKa4
3.2 ( 0.1
4.3 ( 0.1
8.8 ( 0.3
k
k
k
max (s-1) 0.72 ( 0.12
0.79 ( 0.08
(3 ( 3) × 10-7 a
mid (s-1
min (s-1
)
)
(1.6 ( 0.8) × 10-5
(7 ( 3) × 10-8 a
product distribution
{R-OH + pyruvate + Pi} (3) pKa1
1.7 ( 0.3
pKa3
6.2 ( 0.1
4.7 ( 0.4
f(3)low
f(3)mid
f(3)high
pKa1
1b
0.986 ( 0.001
0.907 ( 0.002
1.5 ( 0.1
6.1 ( 0.1
0b
0.012 ( 0.001
0.062 ( 0.003
0.94 ( 0.03
0.66 ( 0.03
{ketal + Pi} (4)
pKa3
6.0 ( 0.6
f(4)low
f(4)mid
f(4)high
pKa1
0.02 ( 0.01
0.12 ( 0.02
{enolpyruvyl + Pi} (5)
pKa3
6.2 ( 0.2
4.5 ( 0.7
f(5)low
f(5)mid
f(5)high
0b
0.05 ( 0.04
0.23 ( 0.03
0.030 ( 0.006
a The numerical value of kmin was almost completely defined by single
ki at the highest pH measured. b Value fixed in fitted curves.
Figure 3. Product distributions for THI breakdown. For each compound,
the proportion of the total products at each pH is plotted for (left) MurA
THI and (right) AroA THI. (O) (UDP-GlcNAc or S3P) + pyruvate (3),
(0) ketal (4), (4) EP-UDP-GlcNAc or EPSP (5).
Figure 2. Rate constant profiles of THI breakdown for (top) MurA THI
and (bottom) AroA THI. Solid lines represent fitted curves.
Figure 4. HMBC spectrum of ketal (4) from MurA THI. The spectrum
was collected without 13C decoupling, so the 13C satellites of the methyl
1
a single ionization, pKa2 ) 4.3 ( 0.1. The solvent deuterium
KIE for MurA THI breakdown at pL 2.0 was 1.3 ( 0.4.
Breakdown Products: MurA THI. The MurA THI structure
was determined previously;4 in the present study, MurA [13C3]-
THI was synthesized and its identity confirmed by 13C NMR
(data not shown). Repurification and 13C NMR of 17 µg of the
putative ketal (4, Figure 4) and enolpyruvyl (5, data not shown)
products confirmed their structures. The ketal (4) has not been
characterized previously. The 1H NMR spectrum was not
informative because the sample contained some contaminating
UDP-GlcNAc. In an HMBC experiment, the H3 methyl protons
signal are observed in the H spectrum.
1
were coupled to 13C3 (δ ) 1.54 ppm, JC3-H3 ) 128.2 Hz).
Cross-peaks were observed to 13C2 (δ ) 107.8 ppm) and 13C3
(δ ) 22.0 ppm, 1JC2-C3 ) 43.6 Hz). These chemical shifts and
coupling constants compare well with the reported values for
the AroA ketal:19 H3 (δ ) 1.68 ppm, 1JC3-H3 ) 128.5 Hz); C2,
110.6 ppm; C3, 25.96 ppm. Very weak peaks indicating a
possible correlation to 13C1 were observed at 176.5 ppm but
were not clearly above the noise. Unidentified less intense peaks
were observed near 53.3 and 90.2 ppm.
9
12544 J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003

Rate and Fate of Tetrahedral Intermediate Breakdown
A R T I C L E S
The only products of nonenzymatic breakdown detected by
UV absorbance were UDP-GlcNAc, 4 and 5 (Figure 3).
Breakdown of MurA [1-14C]THI resulted in [1-14C]pyruvate in
the same proportion as UDP-GlcNAc, as well as 4 and 5.
The product distribution was determined as a function of pH.
At all pH’s, the dominant products were {R-OH + pyruvate
+ Pi} (3), with the proportions of the ketal (4) and enolpyruvyl
(5) products increasing with pH (Table 1, Figure 3). The
proportion of UDP-GlcNAc decreased from 1 to 0.907, with
pKa1 ) 1.7 and pKa3 ) 6.2. This agreed well with ketal (4)
formation, which increased in proportion from 0 to 0.062 with
pKa1 ) 1.5 and pKa3 ) 6.1. The proportion of EP-UDP-GlcNAc
(5) also increased with pH from 0 to 0.030. There was no
detectable 5 at pH < 3, but pKa3 ) 6.2 was fitted, in good
agreement with the pH dependence of 3 and 4 formation.
At pH 10, it is possible that a vanishingly small amount of
[1-14C]PEP was formed, ∼0.4% of products. The amount formed
was too small for the structure to be confirmed and could have
arisen from cross-contamination during chromatography.
Breakdown Products: C-O versus P-O bond Cleavage.
Pyruvate could arise from either C-O or P-O bond cleavage.
To detect whether P-O cleavage occurred, MurA [33P]THI was
degraded in 50% (v/v) methanol, and the labeled products were
separated by anion exchange chromatography. No monomethyl
[33P]phosphate was detected at pH 1, 8, 9, or 10. The limit of
detection for P-O bond cleavage was estimated at 3% based
on the molar fraction of methanol, 0.31, and amounts of [33P]-
phosphate detected.
The breakdown rate in 50% (v/v) methanol was 1.4 × 10-4
s-1 at pH 7.0 compared with 1.3 × 10-4 s-1 in buffer. The
apparent product distribution changed, with a ratio 3:4:5 of 0.73:
0.24:0.02. The apparent increase in proportion of ketal to 0.24
was presumably due to a methyl ketal product coeluting with
the intramolecular ketal product (4). The methyl ketal would
be formed by methanol attack on the cationic intermediate. The
proportion of enolpyruvyl product (5) was unchanged.
Breakdown Products: AroA THI. The products of AroA
THI breakdown have been characterized previously.14,19 In the
present study, it appeared initially that there was no [1-14C]-
pyruvate formed at high pH, but extended incubation of
authentic [1-14C]pyruvate at high pH caused its HPLC retention
time to shift from 13.0 to 8.2 min, presumably due to a
condensation reaction.24 At pH > 11, with reaction times of up
to 3 weeks, the pyruvate peak was shifted completely to the
8.2 min peak. At pH 8-10, both peaks were detected, while, at
lower pH’s, only the pyruvate peak was detected. Taking this
peak into account during THI breakdown gave correct mass
balance for the reaction. As with the MurA THI, the dominant
products were {R-OH + pyruvate + Pi} (3), with the
proportions of the ketal (4) and EPSP (5) increasing with pH
(Figure 3). The cause of the less reproducible product distribu-
tion data for AroA THI breakdown is not clear, though the fact
that reactions had to be taken almost to completion in order to
get enough 14C counts in the small amounts of ketal (4) and
EPSP (5) contributed.
tion. For MurA THI, the ratio of products 3:4:5 was 0.91:0.06:
0.03 at pH > 7.5 in 50 mM bis(trispropane), Tris, or glycine.
Adding 100 mM Et3NH‚HCO3 had no effect, with a 3:4:5 ratio
of 0.90:0.07:0.03. However, with 60 mM NH4‚HCO3, the 3:4:5
ratio changed to 0.86:0.05:0.09 at pH 7.5, and 0.80:0.07:0.13
at pH 10.0. That is, the proportion of enolpyruvyl product
increased, becoming greater than the ketal, which remained
largely unchanged. The rate constant at pH 10.0 did not change,
k ) 1.0 × 10-6 s-1. Thus the breakdown products of MurA
THI were sensitive to the reaction conditions.
Discussion
Understanding which catalytic strategies an enzyme uses
necessarily requires understanding which strategies are capable
of working. Studying the nonenzymatic breakdown of the THIs
of MurA and AroA lends insight into their reactivity, as well
as the related compounds CMP-NeuAc (1) and CMP-deoxy-
octonate (2). Studies on the THIs have been limited by their
difficult syntheses and instability. By producing gram amounts
of E. coli AroA and MurA and using them to synthesize THIs,
we have been able to characterize the kinetics and products of
THI breakdown over a pH range of 1-12.
Previously, it was shown that the MurA13 and AroA14 THIs
were acid labile. The half-life of AroA THI was >48 h at pH
12. AroA THI yielded {S3P + pyruvate + Pi} (3) at low pH,
with {ketal + Pi} (4) and {EPSP + Pi} (5) appearing at pH
7.14 The dominant products of MurA THI degradation were
{UDP-GlcNAc + pyruvate + Pi} (3). Our results agree with
those findings. The rate constants of MurA and AroA THI
degradation were similar to each other, within 15-fold at all
pH’s. The half-lives ranged from 1 s at pH 1 to 89 and 24 days
at pH 12 for MurA and AroA THIs, respectively.
Mechanism of THI Breakdown. The proposed mechanism
is described briefly here, followed by detailed discussions of
specific parts. In the proposed mechanism, THI breakdown
occurs by phosphate departure through a rate-limiting C-O
bond cleavage that is intramolecular general acid catalyzed by
the R-carboxyl group (Figure 5). The immediate product of
C-O bond cleavage is proposed to be an ion pair complex25,26
between an oxocarbenium ion, and phosphate, 6-8. The
dominant reaction at all pH’s is addition of water, ultimately
leading to 3. As pH increases, the product distribution changes,
with larger amounts of 4 and 5. The product distribution is
affected by the phosphate protonation state and the reaction
conditions. The experimental results are discussed below in light
of this proposed mechanism.
Evidence for C-O Bond Cleavage. The products 4 and 5
could only arise through C-O bond cleavage. However, {R-
OH + pyruvate + Pi} (3) could arise through either C-O or
P-O bond cleavage. MurA THI was degraded in the presence
of 50% aqueous methanol at pH’s 1-10 to detect any P-O
bond cleavage. In the event of P-O bond cleavage, methanol
would trap the metaphosphate intermediate (PO3-), forming
monomethyl phosphate. The fraction of monomethyl phosphate
formed is directly proportional to the mole fraction of methanol
in the solvent when the reactant is monoanionic phosphate,27,28
Effect of Buffer on Product Distribution. The relatively
large differences in the product distributions for MurA and AroA
THIs (Figure 3) suggested that buffer conditions and/or other
functional groups in the molecules affected the product distribu-
(25) In the Winstein formalism,26 ion pairs may be intimate or solvent separated.
Our results do not allow us to make this distinction; therefore, we refer to
them simply as ion pairs.
(26) Carey, F. A.; Sundberg, R. J. AdVanced Organic Chemistry, 3rd ed.; Plenum
(24) Margolis, S. A.; Coxon, B. Anal. Chem. 1986, 58, 2504.
Press: New York, 1990.
9
J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003 12545

A R T I C L E S
Byczynski et al.
Figure 5. Reaction scheme for MurA THI breakdown. The proposed mechanism involves rate-limiting dissociation of phosphate (k1, k2, k3) which is acid
catalyzed by the R-carboxylic acid. The product distribution for MurA THI is shown. It depends on the fate of the initially formed oxocarbenium-phosphate
ion pair complex (6-8). Increasing amounts of 4 and 5 at higher pH are attributed to increased ion pair lifetimes, allowing increasing amounts of nucleophilic
addition by the adjacent hydroxyl to give 4 or deprotonation to give 5. The product distribution for AroA THI is different (see text).
or in higher proportion when the reactant is dianionic phos-
phate.27 There was no detectable monomethyl [33P]phosphate
formed from MurA [33P]THI under any conditions. Levels of
P-O bond cleavage at least as low as 3% would have been
detected.
acid and 3-phosphate from the shikimate ring) nearby, raising
pKa2 via through-space interactions.
Protonation of the THI phosphate had little or no effect on
the rate. It would have taken a ≈3-fold change in rate to be
observable in the pH-profile. The lack of effect on rate was
unexpected; protonating the nonbridging oxygens of phosphate
would have been expected to make it a better leaving group
and therefore increase the rate. On the other hand, the lack of
effect on protonation is consistent with the fact that the THIs
have similar stabilities to the phosphodiester-containing 1 and
2, as well as 9 and 10.32
pH Dependence of Breakdown Rate. THI breakdown was
acid catalyzed, with the pH dependence dominated by a
ionizations with pKa2 ) 3.2 (MurA) or 4.3 (AroA). Rates for
MurA and AroA THIs were within a factor of 15 of each other
at all pH’s. The value of pKa2 for MurA THI was consistent
with the combined electron withdrawing effects of a ketal
oxygen and phosphate adjacent to the carboxylic acid. In
comparison, the pKa of the carboxylic acid of 2-phosphoglycolic
acid is 3.6,29 and that of the 2-cyclohexyloxyacetic acid is 3.5.30
There was a weak dependence on an ionization with pKa4
)
8.8 in the MurA THI reaction, consistent with ionization of the
uridine ring.31 A negative charge on the uridine ring could
electrostatically stabilize the proposed cationic intermediate. The
reason for the higher apparent pKa2 ) 4.3 for AroA THI was
not clear but may have arisen from interaction with the
triethylammonium ions present in the reaction mixture and/or
the fact that there are more ionizable groups (the 1-carboxylic
THI breakdown was insensitive to other changes in reaction
conditions. The rate of MurA THI breakdown was unaffected
by buffer conditions that changed the product distribution or
by 50% methanol. The lack of effect with methanol was
consistent with the previous observation of only a 2-fold
(27) Kirby, A. J.; Varvoglis, G. A. J. Am. Chem. Soc. 1967, 89, 415.
(28) Allen, C. M., Jr.; Jamieson, J. J. Am. Chem. Soc. 1971, 93, 1434.
(29) Hartman, F. C.; LaMuraglia, G. M.; Tomozawa, Y.; Wolfenden, R.
Biochemistry 1975, 14, 5274.
(30) Charton, M. J. Org. Chem. 1964, 29, 1222.
(31) Privat, E. J.; Sowers, L. C. Mutat. Res. 1996, 354, 151.
(32) Liang, P.-H.; Lewis, J.; Anderson, K. S.; Kohen, A.; D’Souza, F. W.;
Benenson, Y.; Baasov, T. Biochemistry 1998, 37, 16390.
9
12546 J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003

Rate and Fate of Tetrahedral Intermediate Breakdown
A R T I C L E S
decrease in the rate of CMP-NeuAc (1) breakdown in neutral
70% ethanol.33
Evidence for General Acid Catalysis. THI breakdown was
acid catalyzed. The solvent deuterium KIE at pL 2.0 was 1.3
( 0.4. This is typical of general acid catalysis previously
observed with various acetal hydrolyses, 1.0-1.5,29,34-36 and
was similar to that for hydrolysis of 9 and 10 (1.1-1.2).37 It
was inconsistent with the inverse solvent deuterium KIEs
observed for specific acid catalysis,38 including 0.45 observed
with CMP-NeuAc (1).39 This is less than the solvent deuterium
KIE for hydrolysis of the enol ether 11,40 ∼5, which presumably
reflects the larger intrinsic barrier to proton transfer involving
C-H bonds compared with O-H‚‚‚O in THI breakdown.
It is not possible to determine the exact magnitude of the
effect of R-carboxyl ionization on THI breakdown because the
pH profile was flat at the lowest pH’s measured. However, based
on the fact that the pH profile was flat for at least 1.5 (MurA)
or 2.0 (AroA) pH units, the effect of carboxyl ionization was
at least 30- to 100-fold. This is significantly larger than the
inductive effects assigned to R-carboxyl ionization in 14 and
15. On this basis, we propose that the general acid catalyst in
THI breakdown is the intramolecular carboxyl group, rather than
H3O+, though either alternative is possible. Regardless of the
acid catalyst, the most important observation is that THI
breakdown is catalyzed through protonation of the phosphate
bridging oxygen rather than the nonbridging oxygens.
The THIs meet the conditions for general acid catalysis of
acetals elucidated by Fife,45 namely a good leaving group and
a relatively stable oxocarbenium ion. They also meet the
condition38 that the pKa of the proton acceptor, the bridging
oxygen of the phosphate, changes dramatically between the
reactant (∼ -4) and phosphate product (1.8, 6.8, or 11.7),46 with
the pKa of general acid catalyst, 3.2 (MurA) or 4.3 (AroA),
between the two, except at pH < 1.6, where the product is
H3PO4.
Product Distribution. The product distribution was affected
by the identity of the THI (MurA versus AroA), pH, buffer and
solvent composition. This implies that once the rate-limiting
step had occurred, that is, cleavage of the C-O bond to form
an ion pair complex (6-8), three distinct breakdown pathways
were possible, all with relatively similar activation energies. This
made the product distribution sensitive to reaction conditions.
Water addition yields 3; C3 deprotonation gives an overall
phosphate elimination, yielding 5, while nucleophilic attack by
the adjacent hydroxyl yields 4.
If general acid catalysis is operative, the question becomes
whether it is intramolecular catalysis by the R-carboxyl or the
kinetically equivalent intermolecular catalysis by H3O+. With
intramolecular general acid catalysis, proton transfer would be
through a favorable five-membered ring structure (12). Kirby
and co-workers showed that general acid catalysis of acetal
hydrolysis with rate enhancements of almost 105 is possible even
with a less favored seven-membered ring during proton transfer
(13).41,42
For intermolecular general acid catalysis by H3O+, the pH
dependence corresponding to pKa2 would correspond to the
inductive effect of ionization of the R-carboxyl group. Car-
boxylate, COO-, being electron donating relative to carboxylic
acid, COOH, can better stabilize a cationic intermediate. This
will lead to a decrease in rate upon R-carboxyl protonation,
consistent with the rate profiles of the THIs. The most relevant
analogues to help estimate the strength of the inductive effect
would be the hydrolyses of 1443 and 15,34 where the cationic
intermediate formed would be very similar to the THIs. The
effect of carboxyl ionization was small in both cases, only 3-fold
for 14 and 5-fold for 15. In 15, the reaction was reported to be
general acid catalyzed by H3O+, so the R-carboxyl ionization
effect was an inductive effect. Another example of inductive
effects is 16,44 where ionization of the carboxyl group increased
hydrolysis 90-fold. This was attributed to inductive effects
because intramolecular general acid catalysis was not geo-
metrically possible. Intramolecular general acid catalysis was
proposed for hydrolyses of 1741 and 18,36 where 100- and 300-
fold effects were observed.
Without an ion pair complex, the oxocarbenium ion would
undergo rapid water addition to yield only pyruvate (3).
Oxocarbenium ions are highly reactive, with small intrinsic
barriers to nucleophilic attack.47 The oxocarbenium ions of
sugars such as glucose are too reactive to have a finite lifetime
in solution.48 One of the more stable oxocarbenium ions found
to date is from CMP-NeuAc (19), which had a lifetime in
solution of g3 × 10-11 s.39 Its relatively high stability was
attributed to the R-carboxyl group and the lack of an adjacent
hydroxyl, features present in 6-8.
(33) Ruano, M.-J.; Cabezas, J. A.; Hueso, P. Comp. Biochem. Physiol. 1999,
123B, 301.
(34) Ashwell, M.; Guo, X.; Sinnott, M. L. J. Am. Chem. Soc. 1992, 114, 10158.
(35) Barber, S. E.; Dean, K. E. S.; Kirby, A. J. Can. J. Chem. 1999, 77, 792.
(36) Brown, C. J.; Kirby, A. J. J. Chem. Soc., Perkin Trans. 2 1997, 1081.
(37) Baasov, T.; Kohen, A. J. Am. Chem. Soc. 1995, 117, 6165.
(38) Cordes, E. H.; Bull, H. G. Chem. ReV. 1974, 74, 581.
(39) Horenstein, B. A.; Bruner, M. J. Am. Chem. Soc. 1998, 120, 1357.
(40) Kresge, A. J.; Leibovitch, M.; Sikorski, J. A. J. Am. Chem. Soc. 1992,
114, 2618.
Stabilization of the oxocarbenium ion in an ion pair complex
makes formation of 4 and 5 possible. In the elimination reaction,
(41) Hartwell, E.; Hodgson, D. R. W.; Kirby, A. J. J. Am. Chem. Soc. 2000,
122, 9326.
(45) Fife, T. H. Acc. Chem. Res. 1972, 5, 264.
(42) Dean, K. E. S.; Kirby, A. J. J. Chem. Soc., Perkin Trans. 2 2002, 428.
(43) Chou, D. T. H.; Watson, J. N.; Scholte, A. A.; Borgford, T. J.; Bennet, A.
J. J. Am. Chem. Soc. 2000, 122, 8357.
(46) Saha, A.; Saha, N.; Ji, L.-n.; Zhao, J.; Gregan, F.; Sajadi, S. A. A.; Song,
B.; Sigel, H. J. Biol. Inorg. Chem. 1996, 1, 231.
(47) Richard, J. P.; Williams, K. B.; Amyes, T. L. J. Am. Chem. Soc. 1999,
121, 8403.
(44) Cherian, X. M.; Van Arman, S. A.; Czarnick, A. W. J. Am. Chem. Soc.
1990, 112, 4490.
(48) Banait, N. S.; Jencks, W. P. J. Am. Chem. Soc. 1991, 113, 7951.
9
J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003 12547

A R T I C L E S
Byczynski et al.
3:4:5 of 1:0:0 at pH 4 and 0.76:0.16:0.08 at pH 7.14 In the
present study, the product distribution also varied with pH, with
ratios 3:4:5 of 0.90:0.02:0.09 at pH 4 and 0.66:0.11:0.23 at pH
7. That is, the elimination product EPSP (5) was formed in
higher proportion than ketal (4). The reason for the difference
in product distribution compared to both MurA THI and the
previous report with AroA THI is not clear, but may be at least
partly due to the reaction buffer. Reactions were conducted in
the presence of ca. 100 mM triethylammonium ions, present
from the anion exchange buffer. Triethylammonium ions interact
specifically with AroA THI and/or EPSP, as evidenced by a
differential shift in retention times, making separation possible,
whereas these compounds coelute in other buffers.14,17 The pKa’s
affecting the product distribution were poorly defined but were
∼4.5 for EPSP (5) and 6.0 for the ketal (4). Clearly, the factors
controlling the product distribution of AroA THI were different
than those for MurA THI. The pKa controlling 4 formation may
have reflected the THI phosphate group, but the pKa for 5
formation was inconsistent with a phosphate ionization. It was
in the range expected for a carboxyl, but if intramolecular
general acid catalysis by the R-carboxyl group is assumed, this
group would always be a carboxylate in the ion pair complex,
and there is no reason to expect an effect from the remote
carboxyl of AroA THI.
phosphate could act directly as a general base to deprotonate
the C3 methyl or merely extend the lifetime of the oxocarbenium
ion enough to allow some other group to act as a base. Although
the methyl of the cationic intermediate is very acidic, pKa <
0,40 there is a significant kinetic barrier to C-H bond dissocia-
tion.47 Elimination is slow compared to water addition with
uncomplexed oxocarbenium ions, as illustrated by low rates and
general base-catalyzed deprotonation of substituted 1-phenyl-
ethyl carbocations,49 and with the oxocarbenium ion of ac-
etophenone, where water addition was 19 000-fold faster than
elimination,47 though R-carbonyl substituents can have a large
effect on product partitioning.50,51
There is no obvious role for phosphate in ketal formation
beyond extending the lifetime of the oxocarbenium ion.
Phosphate would be located on the opposite side of the
electrophilic carbon from the nucleophilic hydroxyl, making
general base catalysis difficult or impossible. At the same time,
general base catalysis is likely unnecessary given the extremely
high reactivity of the oxocarbenium ion.
Hydroxyl Departure. These experiments have been inter-
preted in terms of phosphate departure through C-O bond
cleavage. The methods used in this study would not have directly
detected hydroxyl (i.e., R-OH) departure, but all evidence
points to it being a minor pathway. Hydroxyl departure would
yield either {R-OH + PEP} or {R-OH + pyruvate + Pi}
(3). PEP would be formed by deprotonating the C3 methyl of
the cationic intermediate (6-8), giving an overall alcohol
elimination. In MurA [1-14C]THI breakdown at pH 10, a very
small amount of [1-14C]PEP may have been detected, ∼0.4%
of the total products. This was very small compared to
enolpyruvyl (5) formation, which also occurs through depro-
tonation of an oxocarbenium ion intermediate. Alternatively, 3
could be formed through either water addition to the cationic
intermediate (20) or rapid P-O bond cleavage. The latter
possibility was eliminated by the experiment with [33P]THI.
MurA THI Product Distribution. With MurA THI, the
proportions of 4 and 5 increased with pH. Two pKa’s controlled
the product distribution, with pKa1 ) 1.6 and pKa3 ) 6.2. These
were in the range expected for phosphate monoesters, 0.7-1.4
and 5.7-6.8,29,46,52,53 though somewhat lower than observed with
9 and 10, with pKa3 ) 6.8 and 7.2, respectively.32 Presumably
the lifetime of the complex would increase as phosphate changed
from neutral (6) to monoanion (7) to dianion (8) forms because
of greater electrostatic interactions. The proportions of 4 and 5
would then show a pH dependency reflecting the pKa’s of the
phosphate, precisely what was observed with MurA THI. The
fact that these pKa’s did not affect breakdown rates supports
the suggestion that the product distribution was determined after
rate-limiting formation of 6-8. This is further supported by the
observation that (i) 60 mM NH4‚HCO3 in the MurA THI
reaction mixture increased the proportion of 5, without changing
the proportion of 4 or the overall rate,54 and (ii) reaction in
50% methanol caused no change in rate but increased the
apparent proportion of ketal (4) to 0.24, presumably due to a
methyl ketal product coeluting with 4. The methyl ketal would
be formed by methanol attack on the cationic intermediate.
The similarity of the pKa’s controlling 4 and 5 formation
suggests that the role of phosphate may be the same in both,
extending the lifetime of the ion pair complexes.
Based on their pKa’s, phosphate would be a better leaving
group than R-OH at all pH’s. This is consistent with the 7 ×
106-fold higher rates of acid-catalyzed R-glucose 1-phosphate
hydrolysis (extrapolated from ref 55) compared with R-methyl
glucoside.56 Our conclusions would be essentially unchanged
if hydroxyl departure was a significant route of THI breakdown.
Other r-Carboxyketal Phosphates and Phosphodiesters.
The R-carboxyketal phosphate functionality is rare. Cyclic
R-carboxyketal phosphates 9 and 10 were synthesized as
potential enzymatic intermediates of Kdo8P synthase.32 Despite
being acid labile and having lifetimes similar to the THIs, the
AroA THI Product Distribution. The pattern of 4 and 5
increasing with increasing pH also held with AroA THI, but
the product distribution was less straightforward to interpret.
The product distribution was reported previously with ratios
(49) Richard, J. P.; Jencks, W. P. J. Am. Chem. Soc. 1984, 106, 1373.
(50) Richard, J. P.; Lin, S. S.; Buccigross, J. M.; Amyes, T. L. J. Am. Chem.
Soc. 1996, 118, 12603.
(51) Richard, J. P.; Amyes, T. L.; Lin, S. S.; O’Donoghue, A. C.; Toteva, M.
M.; Tsuji, Y.; Williams, K. B. AdV. Phys. Org. Chem. 2000, 35, 67.
(52) O’Connor, J. V.; Barker, R. Carbohydr. Res. 1979, 73, 227.
(53) Massoud, S. S.; Sigel, H. Inorg. Chem. 1988, 27, 1447.
(54) Other buffers used in MurA THI breakdown could potentially have similarly
affected the product distribution, but the smooth transitions between
different buffers and the fact that different buffers at a given pH gave the
same product distribution argue against this.
(55) Bunton, C. A.; Llewellyn, D. R.; Oldham, K. G.; Vernon, C. A. J. Chem.
Soc. 1958, 3588.
(56) Wolfenden, R.; Lu, X.; Young, G. J. Am. Chem. Soc. 1998, 120, 6814.
9
12548 J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003

Rate and Fate of Tetrahedral Intermediate Breakdown
A R T I C L E S
mechanism of breakdown was proposed to be different, with
the pH dependence of degradation reported to be inconsistent
with general acid catalysis by the R-carboxyl group.
Relevance to the Enzymatic Breakdown of THIs. Nonen-
zymatic breakdown is proposed to proceed through an ion pair
complex. In the enzymatic reactions, there is considerable
circumstantial evidence for either cationic intermediates or
cationic transition states in both formation and breakdown of
the THIs.8,10,19,64-66
The dominant route of nonenzymatic THI breakdown was
phosphate departure through C-O bond cleavage, the same
bond that is broken in the forward enzymatic reaction. The
results presented here show that protonating the bridging oxygen
will be an effective catalytic strategy to promote phosphate
departure. Presumably, protonating the ketal oxygen in the THI
would promote R-OH departure. We recently proposed that
the general acid catalyst in AroA for THI breakdown is Lys22,17
located between the two potential leaving groups (26). Accord-
ing to the results presented here, it is ideally located to promote
either phosphate or R-OH departure.
There are two known R-carboxyketal phosphodiesters, CMP-
NeuAc (1)33,39,57,58 and CMP-deoxyoctonate (2).59 These are acid
labile and unstable at neutral pH, with half-lives at neutral pH
of 12 and 0.6 h, respectively. The unusual reactivity of 2 was
attributed to general acid catalysis combined with the hydroxyls
at positions 4, 5, and 7 (or 8) enforcing ring conformations that
are favorable for hydrolysis.59 The weak pH dependence of
hydrolysis at pH 7-8.5 and the large solvent deuterium KIE,
2.7, imply important differences from the THIs.60
The rates and products of THI breakdown were similar to
those observed with CMP-NeuAc (1). Degradation was acid
catalyzed, with pKa ) 4.7 (from Figure 2 of ref 57), and the
product distribution changed with pH.33,58,61 At pH 4, NeuAc
(21, analogous to 3) and CMP were the only products. The
proportion of the elimination product (22, analogous to 5)
increased to 10% at pH 6,58 30% at pH 13, and 50% in
anhydrous triethylamine.61 A small amount of NeuAc 2-phos-
phate (23) was produced from concentrated aqueous base (30%
NH4OH or 50% triethylamine) or 100% triethylamine, indicating
P-O5′ bond cleavage.
In the present study, THI breakdown at low pH was 14 000-
fold (MurA) or 3000-fold (AroA) faster than that at pH 7.5.
However, the specific activity of the Lys22Ala mutant of AroA
decreased only 120-fold. Clearly, general acid catalysis is a
relatively minor component of the 106-fold rate enhancement
in enzymatic THI breakdown and much less than total catalytic
advantage that could be achieved. Even if AroA took full
advantage of the potential catalytic benefit of general acid
Role of r-Carboxyl Group. An R-carboxyl adjacent to the
ketal phosphate is important for reactivity, as demonstrated by
the fact that the acetal phosphate functionalities of sugar
1-phosphates and sugar nucleotide diphosphates (e.g., UDP-
glucose) are not especially unstable. The half-life of R-glucose
1-phosphate is roughly 2 years at 25 °C and pH 7.55
Aside from its potential role as a general acid catalyst, the
R-carboxyl would be expected to have a significant effect on
THI stability based simply on steric effects. Acetals and ketals
with secondary and tertiary substituents at R1 (24),62 and R2
(25),63 are more reactive and more prone to general acid catalysis
catalysis, the rate, ∼1 s, would be less than kcat, 56 s-1 67
.
R-OH cleavage was a minor pathway in nonenzymatic
breakdown, but in AroA-catalyzed THI breakdown there was
only a 3:1 preference for phosphate departure over R-OH.10,17
This implies that greater catalytic power is brought to bear on
promoting R-OH departure to overcome its poorer inherent
leaving group ability.
The products of the nonenzymatic reaction, predominantly
3, were different from the enzymatic reaction, 5 (in the forward
reaction). As shown in this study, the products of THI
breakdown were altered relatively easily by reaction conditions.
Thus, the main mechanistic imperative of AroA or MurA can
be regarded as catalyzing C-O bond cleavage, with the products
being determined by ensuring that the appropriate base is present
to deprotonate C3 of the cationic intermediate (or cationic
transition state) and shielding the cation from nucleophiles,
particularly water.
than those with less sterically demanding substituents. Although
there are no close chemical analogues in the literature, it is
reasonable to expect that part of the high reactivity of the THIs
comes from the sterically demanding substituents, viz., a
secondary alkoxy group (S3P or UDP-GlcNAc), a phosphate
(sterically similar to R2 ) t-butyl), a methyl, and a carboxyl.
Conclusions
Nonenzymatic breakdown of the THIs of the carboxyvinyl
transferases MurA and AroA was examined. THI breakdown
was through phosphate departure by C-O bond cleavage. The
reaction was general acid catalyzed by protonation of the
(57) Comb, D. G.; Watson, D. R.; Roseman, S. J. Biol. Chem. 1966, 241, 5637.
(58) Horenstein, B. A.; Bruner, M. J. Am. Chem. Soc. 1996, 118, 10371.
(59) Lin, C. H.; Murray, B. W.; Ollmann, I. R.; Wong, C. H. Biochemistry 1997,
36, 780.
(60) Withers, S. G.; MacLennan, D. J.; Street, I. P. Carbohydr. Res. 1986, 154,
127.
(61) Beau, J. M.; Schauer, R.; Haverkamp, J.; Kamerling, J. P.; Dorland, L.;
Vliegenthart, J. F. G. Eur. J. Biochem. 1984, 140, 203.
(62) Wiberg, K. B.; Squires, R. R. J. Am. Chem. Soc. 1981, 103, 4473.
(63) Belarmino, A. T. N.; Froehner, S.; Zanette, D.; Farah, J. P. S.; Bunton, C.
A.; Romsted, L. S. J. Org. Chem. 2003, 68, 706.
(64) Anton, D. L.; Hedstrom, L.; Fish, S.; Abeles, R. H. Biochemistry 1983,
22, 5903.
(65) Bondinell, W. E.; Vnek, J.; Knowles, P. F.; Sprecher, M.; Sprinson, D. B.
J. Biol. Chem. 1971, 246, 6191.
(66) Pansegrau, P. D.; Anderson, K. S.; Widlanski, T.; Ream, J. E.; Sammons,
R. D.; Sikorski, J. A.; Knowles, J. R. Tetrahedron Lett. 1991, 32, 2589.
(67) Gruys, K. J.; Walker, M. C.; Sikorski, J. A. Biochemistry 1992, 31, 5534.
9
J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003 12549

A R T I C L E S
Byczynski et al.
phosphate bridging oxygen, with rate constants controlled by
ionizations with pKa ) 3.2 ( 0.1 (MurA) or 4.3 ( 0.1 (AroA).
The general acid catalyst was proposed to be the R-carboxyl
group of the THI, although intermolecular general acid catalysis
by H3O+ is also possible. The protonation state of the THI
phosphate had no measurable effect on the rate, but it did affect
the product distribution, with increasing the amounts of ketal
and enolpyruvyl products at higher pH, with pKa’s of 1.6 and
6.2 for MurA THI.
These results are consistent with the observation in the AroA
structure of a potential general acid catalyst of THI breakdown,
Lys22, located such that it could protonate either the bridging
oxygen of phosphate or the ketal oxygen. Protonating nonbridg-
ing oxygens in the THI phosphate is unlikely to be an effective
catalytic strategy. The detailed chemical mechanisms of AroA
and MurA are currently under investigation.
Acknowledgment. We thank Prof. Debra Dunaway-Mariano
(University of New Mexico) for a generous gift of pyruvate
phosphate dikinase and Dr. Don Hughes and Prof. Alex Bain
(McMaster University) for invaluable assistance with the NMR
analyses. We also thank the anonymous referees of this work
for their valuable questions and comments. This work was
supported by the Canadian Institutes of Health Research, as well
as a graduate scholarship from the Natural Sciences and
Engineering Research Council of Canada (B.B.).
JA0349655
9
12550 J. AM. CHEM. SOC. VOL. 125, NO. 41, 2003