Sialic acid biosynthesis: Stereochemistry and mechanism of the reaction catalyzed by the mammalian UDP-N-acetylglucosamine 2-epimerase

Article abstract of DOI:10.1021/ja021309g

The bifunctional enzyme, UDP-N-acetylglucosamine 2-epimerase/ManNAc kinase, catalyzes the first two steps in the biosynthesis of the sialic acids in mammals. The epimerase domain converts UDP-GlcNAc into ManNAc and UDP. This paper demonstrates that α-ManNAc is the first formed anomer and therefore the reaction proceeds with a net retention of configuration at C-1. Studies in deuterated buffer show that solvent-derived deuterium is quantitatively incorporated into the C-2 position of the product during catalysis, but it is not incorporated into the remaining pool of substrate. This indicates that the inversion of stereochemistry is ultimately brought about by the removal and replacement of a proton at C-2 and is consistent with a two-base mechanism. Studies with ¹⁸O-labeled UDP-GlcNAc show that the anomeric oxygen of the glycosyl phosphate bond departs with the UDP product and therefore the net hydrolysis reaction involves C-O bond cleavage. Incubation of the putative intermediate, 2-acetamidoglucal, with the enzyme resulted in a slow hydration reaction to give the product, ManNAc. Additional kinetic isotope effect and positional isotope exchange (PIX) experiments address the nature of the rate-determining step of the reaction and show that C-H bond cleavage is not rate limiting. Overall, these results support a reaction mechanism involving an anti-elimination of UDP to give 2-acetamidoglucal, followed by a synaddition of water.

Full text of DOI:10.1021/ja021309g

Published on Web 02/06/2003
Sialic Acid Biosynthesis: Stereochemistry and Mechanism of
the Reaction Catalyzed by the Mammalian
UDP-N-Acetylglucosamine 2-Epimerase
Wayne K. Chou,† Stephan Hinderlich,‡ Werner Reutter,‡ and Martin E. Tanner*,†
Contribution from the Department of Chemistry, UniVersity of British Columbia, VancouVer,
British Columbia, V6T 1Z1 Canada, and Institut fu¨r Molekularbiologie und Biochemie,
Fachbereich Humanmedizin, Freie UniVersita¨t Berlin, Arnimallee 22, D-14195 Berlin-Dahlem, Germany
Received October 30, 2002 ; E-mail: mtanner@chem.ubc.ca
Abstract: The bifunctional enzyme, UDP-N-acetylglucosamine 2-epimerase/ManNAc kinase, catalyzes the
first two steps in the biosynthesis of the sialic acids in mammals. The epimerase domain converts UDP-
GlcNAc into ManNAc and UDP. This paper demonstrates that R-ManNAc is the first formed anomer and
therefore the reaction proceeds with a net retention of configuration at C-1. Studies in deuterated buffer
show that solvent-derived deuterium is quantitatively incorporated into the C-2 position of the product during
catalysis, but it is not incorporated into the remaining pool of substrate. This indicates that the inversion of
stereochemistry is ultimately brought about by the removal and replacement of a proton at C-2 and is
consistent with a two-base mechanism. Studies with 18O-labeled UDP-GlcNAc show that the anomeric
oxygen of the glycosyl phosphate bond departs with the UDP product and therefore the net hydrolysis
reaction involves C-O bond cleavage. Incubation of the putative intermediate, 2-acetamidoglucal, with
the enzyme resulted in a slow hydration reaction to give the product, ManNAc. Additional kinetic isotope
effect and positional isotope exchange (PIX) experiments address the nature of the rate-determining step
of the reaction and show that C-H bond cleavage is not rate limiting. Overall, these results support a
reaction mechanism involving an anti-elimination of UDP to give 2-acetamidoglucal, followed by a syn-
addition of water.
Introduction
The sialic acids are nine-carbon polyhydroxylated R-keto
acids that are found on the distal end of glycan chains in
vertebrate glycoconjugates.1,2 Most of the sialic acids are
derivatives of neuraminic acid (Neu) and contain an amino
functionality at the C-5 position. The most common of these is
the biosynthetic precursor for the family N-acetylneuraminic
acid (Neu5Ac). The positioning of the sialic acids at the termini
of cell-surface oligosaccharides is responsible for many of their
crucial biochemical roles, such as serving as key recognition
elements and mediating cellular adhesion processes.3,4 In
addition, sialylation represents an important form of protein
glycosylation,5 and the metastatic potential of tumor cells has
been correlated to the extent of cell surface sialylation.6,7
The mammalian biosynthesis of the sialic acids begins with
conversion of UDP-N-acetylglucosamine (UDP-GlcNAc) into
N-acetylmannosamine (ManNAc) (Figure 1).1 The ManNAc is
Figure 1. The biosynthetic pathway for CMP-N-acetylneuraminic acid in
mammals. A ) UDP-N-acetylglucosamine 2-epimerase, B ) N-acetyl-
mannosamine kinase, C ) N-acetylneuraminic acid 9-phosphate synthase,
D ) N-acetylneuraminic acid 9-phosphate phosphatase, E ) CMP-N-
acetylneuraminic acid synthetase.
then phosphorylated at C-6 and condensed with phospho-
enolpyruvate to give Neu5Ac 9-phosphate. Dephosphorylation
and coupling with CTP generate the activated form of the sialic
acids, CMP-Neu5Ac. The first two steps in this pathway are
catalyzed by the bifunctional enzyme, UDP-GlcNAc 2-epime-
rase/ManNAc kinase.8-10 This enzyme has been found to
† University of British Columbia.
‡ Freie Universita¨t Berlin.
(1) Angata, T.; Varki, A. Chem. ReV. 2002, 102, 439-469.
(2) Traving, C.; Schauer, R. Cell. Mol. Life Sci. 1998, 54, 1330-1349.
(3) Varki, A. FASEB J. 1997, 11, 248-255.
(4) Rutishauser, U. Curr. Opin. Cell Biol. 1996, 8, 679-684.
(5) Lis, H.; Sharon, N. Eur. J. Biochem. 1993, 218, 1-27.
(6) Gorelik, E.; Galili, U.; Raz, A. Cancer Metastasis ReV. 2001, 20, 245-
277.
(8) Effertz, K.; Hinderlich, S.; Reutter, W. J. Biol. Chem. 1999, 274, 28771-
28778.
(9) Sta¨sche, R.; Hinderlich, S.; Weise, C.; Effertz, K.; Lucka, L.; Moorman,
P.; Reutter,W. J. Biol. Chem. 1997, 272, 24319-24324.
(7) Yogeeswaran, G.; Salk, P. L. Science 1981, 212, 1514-1516.
9
10.1021/ja021309g CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 2455-2461
2455

A R T I C L E S
Chou et al.
Figure 3. The mechanism of the reaction catalyzed by the bacterial UDP-
N-acetylglucosamine 2-epimerase.
enzyme is known to catalyze an anti-elimination of UDP from
UDP-GlcNAc to give 2-acetamidoglucal, followed by the syn-
addition of UDP to give the product, UDP-ManNAc (Figure
3). The epimerase domain in the rat epimerase/kinase shares a
22% amino acid identity with the bacterial enzyme, suggesting
the two have evolved from a common ancestor and likely
employ a similar mechanism.8,15
Figure 2. Three potential mechanisms for the reaction catalyzed by the
mammalian UDP-N-acetylglucosamine 2-epimerase. Path A ) epimerization
preceding hydrolysis, Path B ) hydrolysis preceding epimerization, Path
C ) elimination preceding hydration.
Previous isotope incorporation studies have shown that the
reaction in tritiated water produces ManNAc containing tritium
at the C-2 position.20 This indicates that the inversion of
stereochemistry is ultimately brought about by the removal and
replacement of a proton at C-2, and is consistent with all three
of the proposed mechanisms. Initial velocity kinetics and product
inhibition studies have been performed and were interpreted in
favor of an ordered mechanism in which UDP was the first
product released, followed by an irreversible formation of
ManNAc.21 This agrees most closely with the mechanism shown
in Path C. Further experiments involved incubating the enzyme
with the putative reaction intermediates and testing them for
catalytic competence. The incubation of the enzyme with UDP-
ManNAc did lead to the formation of ManNAc as might be
expected if Path A were operative.22 Further studies in tritiated
water, however, showed that the ManNAc produced contained
tritium at C-2 and the amount incorporated was about 87% of
that incorporated when UDP-GlcNAc was the substrate. In
addition, the recovered UDP-ManNAc from an incomplete
reaction bore no tritium label, and no trace of UDP-GlcNAc
could be detected. These results are inconsistent with either the
direct hydrolysis of UDP-ManNAc or the reversible epimer-
ization step shown in Path A. It was suggested that the UDP-
ManNAc is simply serving as an alternate substrate and is not
an intermediate in the normal reaction mechanism. To test for
the possibility that Path B was operative, free GlcNAc was
incubated with the enzyme; however, no epimerization was
observed.23 Perhaps the most convincing experiment was
performed when 2-acetamidoglucal was incubated with the
enzyme.21 This resulted in the production of a compound that
coeluted with ManNAc when analyzed by high voltage paper
electrophoresis. This observation supports the reaction mech-
anism outlined in Path C.
catalyze the rate-limiting steps in sialic acid biosynthesis and
therefore serves as the key regulator of cell surface sialylation
in humans.11 It has also been shown to be essential in the
embryonic development of mice,12 and point mutations result
in the human disease hereditary inclusion body myopathy.13
An examination of the UDP-GlcNAc 2-epimerase/ManNAc
kinase amino acid sequence suggests that the enzyme is
comprised of two domains. The N-terminal domain is respon-
sible for the epimerase activity and catalyzes both the inversion
of stereochemistry at C-2 of UDP-GlcNAc as well as hydrolysis
to produce free ManNAc and UDP (this reaction is essentially
irreversible and technically not a true epimerization). The
C-terminal domain is responsible for the kinase activity that
generates ManNAc 6-phosphate. These assignments have been
supported by both sequence homology studies and mutagenesis
studies.8
Potential mechanisms for the “epimerization” reaction include
the initial epimerization of UDP-GlcNAc to give UDP-ManNAc,
followed by a hydrolysis step that releases UDP and ManNAc
(Path A, Figure 2). The first step would presumably involve a
2-acetamidoglucal intermediate analogous to the one shown in
Path C (vide infra). Alternatively, hydrolysis may occur first,
and epimerization of the free sugar would generate ManNAc
(Path B, Figure 2). In this case, the epimerization would likely
proceed via the open chain form of the sugar and involve a
deprotonation/reprotonation of the relatively acidic proton at
C-2. Finally, the mechanism may proceed via an initial anti-
elimination of UDP from UDP-GlcNAc to give the intermediate
2-acetamidoglucal, followed by the hydration of the glycal to
give free ManNAc (Path C, Figure 2). Paths A and C are
indirectly supported by studies on the bacterial UDP-GlcNAc
2-epimerase that catalyzes the interconversion of UDP-GlcNAc
and UDP-ManNAc (and is therefore a true epimerase).14-19 This
In this work, the epimerase reaction is shown to proceed with
a retention of stereochemistry at C-1, and the loss of UDP is
found to occur via a C-O bond cleavage process. In addition,
(10) Hinderlich, S.; Sta¨sche, R.; Zeitler, R.; Reutter, W. J. Biol. Chem. 1997,
272, 24313-24318.
(15) Campbell, R. E.; Mosimann, S. C.; Tanner, M. E.; Strynadka, N. C. J.
Biochemistry 2000, 39, 14993-15001.
(11) Keppler, O. T.; Hinderlich, S.; Langner, J.; Schwartz-Albiez, R.; Reutter,
W.; Pawlita, M. Science 1999, 284, 1372-1376.
(16) Morgan, P. M.; Sala, R. F.; Tanner, M. E. J. Am. Chem. Soc. 1997, 119,
10269-10277.
(12) Schwarzkopf, M.; Knobeloch, K. P.; Rohde, E.; Hinderlich, S.; Wiechens,
N.; Lucka, L.; Horak, I.; Reutter, W.; Horstkorte, R. Proc. Natl. Acad. Sci.
U.S.A. 2002, 99, 5267-5270.
(17) Kawamura, T.; Ishimoto, N.; Ito, E. J. Biol. Chem. 1979, 254, 8457-8465.
(18) Kawamura, T.; Kimura, M.; Yamamori, S.; Ito, E. J. Biol. Chem. 1978,
253, 3595-3601.
(19) Salo, W. Biochim. Biophys. Acta 1976, 452, 625-628.
(20) Glaser, L. Biochim. Biophys. Acta 1960, 41, 534-536.
(21) Sommar, K. M.; Ellis, D. B. Biochim. Biophys. Acta 1972, 268, 590-595.
(22) Salo, W. L.; Fletcher, H. G. Biochemistry 1970, 9, 882-885.
(23) Comb, D. G.; Roseman, S. Biochim. Biophys. Acta 1958, 29, 653-654.
(13) Eisenberg, I.; Avidan, N.; Potikha, T.; Hochner, H.; Chen, M.; Olender,
T.; Barash, M.; Shemesh, M.; Sadeh, M.; Grabov-Nardini, G.; Shmilevich,
I.; Friedmann, A.; Karpati, G.; Bradley, W. G.; Baumbach, L.; Lancet, D.;
Ben Asher, E.; Beckmann, J. S.; Argov, Z.; Mitrani-Rosenbaum, S. Nat.
Genet. 2001, 29, 83-87.
(14) Tanner, M. E. Acc. Chem. Res. 2002, 35, 237-246.
9
2456 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Sialic Acid Biosynthesis
A R T I C L E S
tube, and Chelex-100 resin was added (20 mg of 200-400 mesh, Na+
form, previously rinsed with D2O). 1H and proton-decoupled 31P NMR
spectra were taken. The scrambling experiment was initiated by the
addition of UDP-GlcNAc 2-epimerase in deuterated buffer (200 µL,
200 µg of protein), and the resulting solution was incubated for 24 h
the product of the incubation with 2-acetamidoglucal is posi-
tively identified as ManNAc by H NMR spectroscopy. These
1
results support a mechanism involving an anti-elimination of
UDP followed by the syn-hydration of 2-acetamidoglucal.
Positional isotopic scrambling experiments and kinetic isotope
effect measurements are described that address the nature of
the rate-determining transition states in this reaction.
1
at 25 °C. H and proton-decoupled 31P NMR spectra were taken at
timed intervals. Proton-decoupled 31P NMR spectra were obtained using
a Bruker 300 MHz spectrometer operating at a frequency of 121 Hz.
Acquisition parameters were sweep width ) 2437 Hz, acquisition time
) 13.4 s, delay between pulses ) 2.0 s, and pulse width ) 10 µs.
The final solution was applied to a size exclusion column (Biogel
P-2, 200-400 mesh, 2.5 cm × 70 cm) and eluted with distilled water.
Fractions exhibiting absorbance at 254 nm (uridine chromophore) were
collected and lyophilized to dryness. 1H NMR spectra of relevant
fractions revealed that they contained UDP. Mass spectral analysis
confirmed the presence of the 18O-label in the UDP produced: -ESIMS
m/z 405 (M - H+, 18O, 100), 403 (M - H+, 16O, 46.5). 31P NMR
spectroscopy confirmed that the 18O-label was incorporated into the
â-phosphate: 31P NMR (D2O) δ-8.001 (d, Jp-p ) 21.0 Hz, 0.32P,
â-P-16O), -8.030 (d, Jp-p ) 21.0 Hz, 0.68P, â-P-18O), -9.645 (d, Jp-p
) 21.0 Hz, 1P, R-P).
Experimental Section
General Methods. UDP-N-acetylglucosamine, lactate dehydrogenase
(Type II from rabbit muscle), and pyruvate kinase (Type II from rabbit
muscle) were purchased from Sigma Chemical Co. 18O-enriched water
(95%) was purchased from Icon Stable Isotopes. Protein concentrations
were determined by the method of Bradford using bovine serum
albumin as the standard.24
Enzyme Expression and Purification. UDP-GlcNAc 2-epimerase/
ManNAc kinase from rat was overexpressed in Spodoptera frugiperda
cells (Sf9, Life Technologies, Inc.) using a baculovirus expression
system.8 Suspension cultures of Sf-9 cells (50 mL) were grown to a
density of 2 × 106 cells/mL at 120 rpm and 27 °C, and then infected
with the recombinant baculovirus containing the sequence coding for
the gene of interest, at a multiplicity of infection of 1. After an optimal
infection period of 60 h, the cells were pelleted by centrifugation (5000
rpm, 20 min), and the following purification procedures were carried
out at 4 °C.
The resulting cell pellet was lysed in a phosphate buffer (10 mM
NaH2PO4, pH 7.5, containing 1 mM dithiothreitol, 1 mM EDTA, and
1 mM phenylmethylsulfonyl fluoride) by drawing the cell pellet in and
out of a syringe-needle (PrecisionGlide Needle, 26 gauge, 5/8” length,
Becton Dickinson & Co.). The crude cell lysate was clarified by
centrifugation (5000 rpm, 60 min) and filtration (0.2 µm pore, 25 mm
Acrodisc syringe Filter, HT Tuffryn Membrane, Pall Corp.). The filtrate
was loaded onto an ion-exchange column (1 mL, HiTrap Q HP,
Amersham Pharmacia Biotech) preequilibrated with column buffer (10
mM NaH2PO4, pH 7.5, containing 1 mM DTT, and 1 mM EDTA).
The column was eluted with a linear gradient of 0-1 M NaCl in column
buffer and monitored at 280 nm. Fractions containing the enzyme were
used directly in experiments. Enzyme in deuterated buffer was prepared
as above, except that both the lysis and the column buffers were
prepared with D2O (pD 7.9).
Stereochemistry and Solvent Deuterium Isotope Incorporation
Studies. A sample of UDP-GlcNAc 2-epimerase in deuterated column
buffer (950 µL, 950 µg of protein) was prepared, and a 1H NMR
spectrum was taken. A sample of UDP-GlcNAc prepared in the same
buffer (50 µL, 100 mM) was added, and the resulting solution (5 mM
UDP-GlcNAc, 1 mL) was incubated for 10 h at 25 °C. 1H NMR spectra
(400 MHz) were taken at timed intervals. A control reaction of UDP-
GlcNAc in the same buffer (5 mM, 1 mL) was also monitored under
identical conditions.
(iii) Test for Solvent 18O-Isotope Incorporation. A sample of UDP-
GlcNAc 2-epimerase in column buffer (300 µL, 300 µg of protein)
was mixed with 300 µL of 95% 18O-enriched H2O. The reaction mixture
was incubated at 30 °C for 10 h and then applied to a size exclusion
column (Biogel P-2, 200-400 mesh, 2.5 cm × 70 cm) and eluted with
distilled water. Fractions exhibiting absorbance at 254 nm were collected
and lyophilized to dryness. 1H NMR spectra of relevant fractions
revealed that they contained UDP. Mass spectral analysis indicated the
absence of the 18O-label in the UDP produced: -LSIMS (thioglycerol)
m/z 425 (M (monosodium salt) - H+, 16O, 100).
Catalytic Competence of 2-Acetamidoglucal. 2-Acetamidoglucal
was synthesized by the procedure outlined by Pravdic et al.25 A solution
of UDP-GlcNAc 2-epimerase in deuterated buffer (950 µL, 950 µg of
protein) was prepared, and a 1H NMR spectrum was collected. A sample
of 2-acetamidoglucal prepared in the same buffer (50 µL, 100 mM)
was added, and the resulting solution (5 mM 2-acetamidoglucal, 1 mL)
was incubated for 29 h at 25 °C. 1H and 31P NMR spectra were collected
at timed intervals. Similar samples, either containing 5 mM UDP or
lacking enzyme, were run under identical conditions. The extent of
hydration was determined by integration of the H-1 peaks of both the
2-acetamidoglucal and the ManNAc anomers produced.
Kinetic Isotope Effect Studies. [2′′-2H]-UDP-GlcNAc was prepared
according to the procedure by Morgan et al.16 1H NMR and mass
spectroscopy confirmed the extent of deuterium incorporation to be
>95%: -ESIMS m/z 607 (M - H+, 2′′-2H, 100). Concentrations of
stock substrate solutions were calculated from A262 measurements with
ꢀ ) 9890 M-1 cm-1. Kinetics were measured using a continuous
coupled assay for UDP formation.26 Each assay contained 50 mM NaH2-
PO4 buffer (pH 7.5), 10 mM MgCl2, 2 mM PEP, 0.2 mM NADH, 188
units of pyruvate kinase, 250 units of lactate dehydrogenase, and 5-300
µM UDP-GlcNAc or [2′′-2H]-UDP-GlcNAc (800 µL total volume).
UDP-GlcNAc 2-epimerase (10 µg) was added to initiate the enzymatic
reaction. Reaction rates were measured at 37 °C by monitoring the
decrease in absorbance at 340 nm. Kinetic parameters were determined
from initial velocities fit to Michaelis-Menten kinetics using the
computer program GraFit 4.
Tests for C-O versus P-O Bond Cleavage and PIX Experiment.
(i) Preparation of 18O-Labeled UDP-GlcNAc. Uridine 5′-(2′′-aceta-
mido-2′′-deoxy-[ 1′′-18O]-R-D-glucopyranosyl diphosphate), 18O-labeled
UDP-GlcNAc, was synthesized according to the procedure by Morgan
et al.16 1H and 13C NMR spectra were identical to those reported. The
extent of 18O incorporation was determined to be 68% by mass spectral
analysis: -LSIMS (thioglycerol) m/z 608 (M - H+, 18O, 100), 606 (M
- H+, 16O, 46.5). The location of the 18O-label was confirmed by 31P
NMR spectroscopy: 31P NMR (D2O) δ -11.789 (d, Jp-p ) 21.0 Hz,
0.32P, â-P-16O), -11.801 (d, Jp-p ) 21.0 Hz, 0.68P, â-P-18O), -10.095
(d, Jp-p ) 21.0 Hz, 1P, R-P).
Results
Enzyme Purification. The UDP-GlcNAc 2-epimerase/Man-
NAc kinase from rat was overexpressed in insect cells (Sf-9)
using a baculovirus expression system8 and was purified by a
(ii) Positional Isotope Exchange (PIX) Experiment and Test for
C-O versus P-O Bond Cleavage. A solution of 18O-labeled UDP-
GlcNAc in deuterated buffer (450 µL, 17 mM) was placed in an NMR
(25) Pravdic, N.; Franjic-Mihalic, I.; Danilov, B. Carbohydr. Res. 1975, 45,
302-306.
(26) Gosselin, S.; Alhussaini, M.; Streiff, M. B.; Takabayashi, K.; Palcic, M.
M. Anal. Biochem. 1994, 220, 90-97.
(24) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2457

A R T I C L E S
Chou et al.
single ion exchange chromatographic step. The resulting protein
was estimated to be 80% pure as analyzed by SDS-PAGE and
was used without further purification because it is unstable to
freezing and has limited stability upon storage at 4 °C. The
enzymatic activity was also lost upon centrifugal concentration
using ultrafiltration membranes, making buffer exchange/
concentration difficult. To prepare sufficiently concentrated
enzyme in a deuterated buffer for NMR studies, it was necessary
to both lyse the insect cells and run the purification column in
deuterated buffers, and then use the column fractions directly.
Stereochemistry and Solvent Isotope Incorporation Stud-
ies. To elucidate the stereochemistry of the epimerase reaction,
it was necessary to determine whether the R-anomer or the
â-anomer of ManNAc is the true reaction product. This
experiment is complicated by the fact that the anomers inter-
convert rapidly in the absence enzyme. To determine the rate
of this mutarotation process, a sample of commercial ManNAc
(predominantly the â-anomer in the solid state, vide infra) was
prepared in a deuterated phosphate buffer and immediately
monitored using 1H NMR spectroscopy. By integrating the
signals corresponding to the anomeric protons, it was determined
that the anomers interconvert with a half-life of approximately
10 min at 25 °C. At equilibrium, the ratio of anomers was
approximately 1:1. A second problem lay in deducing which
1
of the H NMR signals correspond to a given anomer. The
values of JH1,H2 are very similar for the anomers of mannose
derivatives and cannot be used to distinguish between them.
Figure 4. (A) The enzymatic formation of R-[2-2H]-ManNAc in D2O
followed by nonenzymatic mutarotation. (B) 1H NMR spectra following
the time course of the reaction.
Instead, it was necessary to use the technique of Bock and
27
Pedersen that relies on measuring the values of JC1,H1
.
The
R-anomers of mannose derivatives are generally found to have
C1,H1 values that are 10 Hz greater than those of the â-anomers.
not incorporated into recovered starting material to any signifi-
cant extent.20 Inspection of the signal corresponding to the
anomeric proton of UDP-GlcNAc showed no change in the
coupling pattern even after 80% of the material had been
converted to product.
J
An equilibrated sample of the ManNAc anomers in deuterated
phosphate buffer was analyzed by 2-D heteronuclear NMR
spectroscopy (HMQC), and it was established that the C-1
proton signals at 5.02 and 4.92 ppm corresponded to the C-1
carbon signals at 94.50 and 94.38 ppm, respectively. A proton-
decoupled 13C NMR spectrum was then used to determine that
the values of JC1,H1 were 171 and 161 Hz, respectively. This
Test for C-O versus P-O Bond Cleavage and PIX
Experiment. To test whether the loss of UDP proceeds via C-O
bond cleavage or P-O bond cleavage, a sample of UDP-
GlcNAc in a phosphate buffer (pH 7.5) containing 50% H218O
was treated with the epimerase. The reaction was allowed to
proceed to completion, and the resulting UDP was isolated. Mass
spectral analysis confirmed the absence of any 18O-isotope in
the UDP. This observation suggests that the hydrolysis proceeds
via C-O bond cleavage and that the solvent-derived oxygen
resides in the ManNAc. Because this relies on a negative result,
and because the detection of the 18O-label in the anomeric
position of ManNAc may be complicated by nonenzymatic wash
in/out, the experiment was repeated with 18O-labeled UDP-
GlcNAc.
A sample of UDP-GlcNAc bearing an 18O-label at the
anomeric or “bridging” position was chemically synthesized as
described previously.16 31P NMR spectroscopy and mass spectral
analysis confirmed the location of the label and indicated that
the extent of label incorporation was 68%. By treating this
compound with the epimerase in a deuterated phosphate buffer
and monitoring the reaction using 31P NMR spectroscopy, it
was possible to track the fate of the 18O-label in the UDP
produced (Figure 5a). In addition, this experiment tests for the
possibility of positional isotope exchange (PIX) in the remaining
starting material.28 A spectrum of the starting material before
the addition of enzyme shows that the two signals corresponding
1
led to the assignments of the H NMR signal at 5.02 ppm to
the R-anomer and the signal at 4.92 ppm to the â-anomer.
Samples of UDP-GlcNAc in a deuterated phosphate buffer
(pD 7.9) were treated with the epimerase, and spectra were
immediately collected (Figure 4). New peaks that were identical
to those of UDP and free ManNAc gradually replaced those of
UDP-GlcNAc, as expected. Inspection of the spectra at the
earliest time points (5 min) showed that the R-anomer was the
predominant product by a factor of at least 3:1. At later time
points, rapid nonenzymatic mutarotation allowed the anomers
to interconvert and resulted in a 1:1 equilibrium mixture. This
result shows that the reaction of the mammalian enzyme
proceeds with a net retention of configuration at C-1 and
produces the R-anomer of ManNAc.
The experiment described above also served to demonstrate
that the reaction proceeds with the incorporation of solvent-
derived deuterium at C-2 and that the level of incorporation
was >95%. Signals for the C-2 protons of ManNAc were absent
in the spectra (not shown in Figure 4), and the signals for the
C-1 protons appeared as singlets (Figure 4). In addition, the
experiment confirmed the report that solvent-derived isotope is
(27) Bock, K.; Pedersen, C. J. Chem. Soc., Perkin Trans. 2 1974, 293-297.
9
2458 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Sialic Acid Biosynthesis
A R T I C L E S
Figure 6. 1H NMR spectra following the incubation of 2-acetamidoglucal
(A) without enzyme, (B) with epimerase, and (C) with both epimerase and
UDP.
that is shifted upfield by 0.029 ppm from the unlabeled material
due to the higher P-O bond order. This was seen in PIX
experiments with the bacterial UDP-GlcNAc 2-epimerase.16
Instead, only the 1:2 ratio of signals present in the starting
mixture was observed. To ensure that it would be possible to
detect this signal, the experiment was repeated with the bacterial
enzyme, and the previously reported scrambling was clearly
observed.
Catalytic Competence of 2-Acetamidoglucal. To test whether
2-acetamidoglucal is catalytically competent to serve as a
substrate for the second half of the reaction, samples were
incubated with the enzyme in the presence and the absence of
added UDP. 2-Acetamidoglucal was synthesized using previ-
ously reported methods.25,31 When this material alone was
incubated in deuterated phosphate buffer (pD 7.9), and moni-
tored by 1H NMR spectroscopy, no nonenzymatic hydration was
observed (Figure 6a). In the presence of added epimerase,
however, the formation of [2-2H]-ManNAc was confirmed by
the appearance of two singlets at 5.02 and 4.92 ppm in a 1:1
ratio (Figure 6b). No signals corresponding to GlcNAc were
detected. The formation of ManNAc was quite slow, and after
29 h only 6% conversion was observed. Under identical
concentrations of UDP-GlcNAc and enzyme, the normal reac-
tion proceeded to completion within 2 h under these conditions.
When the incubation was repeated in the presence of 5 mM
UDP, a 12% production of ManNAc was observed (Figure 6c).
The increase in the amount of ManNAc produced was repro-
ducible and grew with increasing levels of added UDP. These
experiments demonstrate that 2-acetamidoglucal is a catalytically
competent substrate for the mammalian epimerase.
Figure 5. (A) The production of 18O-labeled UDP from 18O-labeled UDP-
GlcNAc. (B) 31P NMR spectrum of 18O-labeled UDP-GlcNAc before the
addition of enzyme. (C) 31P NMR spectrum after the reaction proceeded to
80% completion (a ) 18O-labeled material, b ) unlabeled material, * stands
for 3% impurity of UMP dimer formed during the morpholidate coupling
step of the synthesis).
to the 18O-labeled and unlabeled â-phosphorus nuclei appeared
as two doublets separated by 0.012 ppm (Figure 5b, the signals
appear as doublets due to coupling with the adjacent phospho-
rus).16 The isotopic substitution of 18O for 16O is known to cause
an upfield shift of this magnitude,29,30 and the 1:2 integration
of these signals confirms that the upfield doublet corresponds
to the labeled material. This material was incubated with the
epimerase, and a spectrum was taken after the reaction had
proceeded to approximately 80% completion (Figure 5c). An
examination of the signals due to the â-phosphorus of the UDP
product clearly shows two doublets present in a 1:2 ratio with
a separation of 0.029 ppm. Thus, the spectrum indicates that
the 18O-label resides in the UDP, a fact that was further
established using mass spectrometry. This proves that the
enzymatic reaction proceeds via a C-O bond cleavage process.
An examination of the signals due to the residual UDP-GlcNAc
indicates that no detectable scrambling of the 18O-isotopic label
has occurred (it would be possible to detect the presence of
10% of the scrambled product). Material in which the label has
scrambled into a “nonbridging” position would show a doublet
Kinetic Isotope Effect Study. The reaction rates of [2′′-2H]-
UDP-GlcNAc and unlabeled UDP-GlcNAc were compared to
determine if cleavage of the C-H bond is rate determining
during catalysis. The deuterium label was introduced into the
C-2′′ position of UDP-GlcNAc using the bacterial UDP-GlcNAc
2-epimerase in a deuterated buffer, as described previously.16
(28) Raushel, F. M.; Villafranca, J. J. CRC Crit. ReV. Biochem. 1988, 23, 1-26.
(29) Hester, L. S.; Raushel, F. M. Biochemistry 1987, 26, 6465-6471.
(30) Cohn, M.; Hu, A. Proc. Natl. Acad. Sci. U.S.A. 1978, 75, 200-203.
(31) Lai, E. C. K.; Withers, S. G. Biochemistry 1994, 33, 14743-14749.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2459

A R T I C L E S
Chou et al.
The previous work with tritiated water also demonstrated that
incorporation occurred,20 but because it is a tracer labeling
experiment and subject to discrimination from isotope effects,
the observed extent of incorporation is difficult to interpret. This
work showed that the extent of incorporation was essentially
quantitative and that there was no significant internal return of
the original C-2 proton of UDP-GlcNAc into the product.
Internal return could only be seen with a one-base mechanism
in which the same residue is responsible for both the deproto-
nation and the reprotonation events.34,35 Internal return will not
be seen with a two-base mechanism in which two different
residues play the roles of catalytic base and acid. The fact that
no solvent-derived isotope is observed in the remaining pool
of starting material after 80% conversion can be explained if
one assumes the deprotonation step is irreversible or is preceded
by an irreversible step. Alternatively, it can be explained by a
“two-base” mechanism involving monoprotic residues that do
not exchange protons with bulk solvent during the lifetime of
the intermediates. This is the case with the reversible reactions
catalyzed by glutamate and proline racemase in which two
cysteine thiol(ate)s act as the catalytic acid/base residues and
solvent isotope incorporation is observed in the product but not
the starting material.36-38 These observations are consistent with
all three mechanisms shown in Figure 2.
The experiment with 18O-labeled UDP-GlcNAc showed that
the label was retained in the released UDP and confirms that
the reaction proceeds with C-O bond cleavage. In Paths A and
B (Figure 2), the hydrolysis step could occur via either C-O
or P-O bond cleavage depending on whether the water attacks
at carbon or at phosphorus, respectively. This finding rules out
the latter possibilities and is also consistent with Path C that
follows a mandatory C-O bond cleavage process. The same
experiment showed that there was no isotopic scrambling (PIX)
in the pool of starting material even after 80% of the material
was converted to products. The absence of PIX suggests that
the C-O bond cleavage is irreversible (or a step prior to it is
irreversible) and that once UDP is formed it does not partition
back into the pool of free starting material. An alternate
explanation is that the terminal phosphate in the bound UDP is
prevented from rotation by interactions with active site residues
(no divalent metal cations have been implicated in catalysis39).
Thus, even if the bond cleavage were reversible, the same
oxygen would add back to the anomeric position of the GlcNAc
residue. While this is certainly a possibility, it is not the case
in the bacterial enzyme where PIX is clearly observed.16 Even
when the bacterial reaction is run under irreversibly coupled
conditions using UDP-ManNAc dehydrogenase, complete scram-
bling of the isotope is observed upon each turnover.40
Figure 7. Plots of initial velocity versus substrate concentration for
unlabeled UDP-GlcNAc (O) and for [2′′-2H]-UDP-GlcNAc (×). The line
was fitted to the data obtained with the unlabeled substrate.
The isotopic enrichment in the resulting [2′′-2H]-UDP-GlcNAc
was found to be >95% as judged by 1H NMR and mass spectral
analysis. A continuous spectrophotometric assay employing
pyruvate kinase and lactate dehydrogenase gave values of 0.33
( 0.01 s-1 for kcat and 26 ( 4 µM for KM with unlabeled UDP-
GlcNAc (Figure 7). The corresponding values obtained with
[2′′-2H]-UDP-GlcNAc were 0.32 ( 0.01 s-1 for kcat and 27 (
4 µM for KM. These values give a kcat isotope effect of 1.0 (
0.1 and a kcat/KM isotope effect of 1.0 ( 0.4 and demonstrate
that cleavage of the C-H bond is not a rate-determining step
in catalysis.
Discussion
Most known racemases and epimerases operate at chiral
centers that are adjacent to carbonyl functionalities and bear a
relatively acidic proton.32-34 In these cases, the enzymes employ
nonstereospecific deprotonation/reprotonation mechanisms to
catalyze an inversion of stereochemistry. The UDP-GlcNAc
2-epimerases belong to the small group of enzymes that operate
at chiral centers lacking an acidic proton and must therefore
use alternate catalytic strategies. Most of the work to date has
focused on the bacterial UDP-GlcNAc 2-epimerase that has been
shown to catalyze an initial anti-elimination of UDP to give
2-acetamidoglucal, followed by a syn-addition of UDP to give
UDP-ManNAc (Figure 3).14-19 These elimination reactions
presumably employ E2- or E1-like mechanisms and thus avoid
the generation of any carbanionic intermediates. The homolo-
gous mammalian UDP-GlcNAc 2-epimerase catalyzes the
irreversible formation of UDP and ManNAc and has only
recently become available in a recombinant form.8,9 In this
paper, we probe the stereochemistry and mechanism of the
mammalian enzyme and compare the results to those obtained
with the bacterial enzyme.
1
By monitoring the reaction with H NMR spectroscopy, it
The experiments described thus far are consistent with all of
the mechanisms shown in Figure 2, but do not distinguish
between them. A previous report briefly mentioned that 2-ac-
etamidoglucal was hydrated to give ManNAc, providing evi-
dence in support of Path C.21 Because the product had only
was possible to determine that the ManNAc was initially formed
as the R-anomer and therefore the reaction proceeds with a net
retention of stereochemistry at C-1. This is also the case with
the bacterial enzyme. This experiment also showed that the
reaction in D2O generates [2-2H]-ManNAc, implying that the
inversion of stereochemistry is ultimately brought about by the
removal of a proton from C-2, followed by the replacement of
a solvent-derived proton in the opposite stereochemical sense.
(35) Rose, I. A. Annu. ReV. Biochem. 1966, 35, 23-56.
(36) Gallo, K. A.; Tanner, M. E.; Knowles, J. R. Biochemistry 1993, 32, 3991-
3997.
(37) Choi, S.-Y.; Esaki, N.; Yoshimura, T.; Soda, K. J. Biochem. 1992, 112,
139-142.
(38) Cardinale, G. J.; Abeles, R. H. Biochemistry 1968, 7, 3970-3978.
(32) Tanner, M. E. Curr. Org. Chem. 2001, 5, 169-192.
(39) Hinderlich, S.; Sonnenschein, A.; Reutter, W. Biometals 1998, 11, 253-
(33) Samuel, J.; Tanner, M. E. Nat. Prod. Rep. 2002, 19, 261-277.
(34) Tanner, M. E.; Kenyon, G. L. In InVersion of Stereocenters; Sinnott, M.,
Ed.; Academic Press: San Diego, 1998; Vol. II, pp 7-41.
258.
(40) Sala, R. F. Ph.D. Thesis Dissertation, The University of British Columbia,
1999.
9
2460 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Sialic Acid Biosynthesis
A R T I C L E S
been identified by paper chromatography, this experiment was
repeated and monitored by 1H NMR spectroscopy. The experi-
ment clearly showed that ManNAc was produced; however, it
was formed at a rate much slower than the normal reaction.
Thus, 2-acetamidoglucal is catalytically competent, but not
kinetically competent, to serve as an intermediate in this
reaction. It is likely that the slow rate of reaction is due to the
fact that the free enzyme does not normally bind 2-acetamido-
glucal from solution (or release it) and only a small fraction of
the enzyme is in the correct conformation and protonation state
to do so.41 The reaction was also run in the presence of UDP,
and a modest, but reproducible, increase in the amount of
produced ManNAc was observed. This demonstrates that UDP
is not an inhibitor of the hydration step and suggests that the
hydration may occur while UDP is still bound in the active site
of the enzyme. Alternatively, the increase may be simply due
to a stabilizing effect that keeps a greater proportion of the
enzyme active during the course of the experiment. As we do
not have a sensitive assay for the formation of 2-acetamidoglucal
in this very slow reaction, we were not able to measure initial
rates and test this possibility. The possibility that the hydration
was catalyzed by impurities of glycosidases can be ruled out
by the observations that only ManNAc was formed (as opposed
to GlcNAc)31,42 and that the rate was increased by the presence
of UDP.
The hydration of the glycal strongly supports the mechanism
shown in Path C (Figure 2). This is also consistent with kinetic
studies that were interpreted in favor of an ordered mechanism
in which UDP was the first product released, followed by an
irreversible formation of ManNAc.21 Path B seems quite unlikely
as GlcNAc is not epimerized by the enzyme,23 and 2-acetami-
doglucal would not be expected to serve as an alternate substrate.
Path A also seems unlikely because the 2-acetamidoglucal (a
putative intermediate in the first step) should not be hydrated
in the absence of UDP. In addition, the earlier reports of solvent
isotope incorporation studies during the hydrolysis of UDP-
ManNAc argue against Path A.22 It is likely that UDP-ManNAc
simply serves as an alternate substrate and that the net hydrolysis
occurs via an elimination of UDP to give 2-acetamidoglucal,
followed by the normal hydration of this intermediate. Thus,
Path C remains the most reasonable description of the minimal
catalytic mechanism for the mammalian UDP-GlcNAc 2-epi-
merase. When one considers the overall stereochemistry of the
reaction, the enzyme catalyzes an anti-elimination of UDP
followed by the syn-addition of water. This ability to catalyze
two reactions of opposite stereospecificity is similar to the
bacterial enzyme that catalyzes the nonstereospecific addition
of UDP to a 2-acetamidoglucal intermediate.
The absence of any kinetic isotope effect in the reaction of
[2′′-2H]-UDP-GlcNAc shows that cleavage of the C-H bond
is not a rate-determining step in the reaction mechanism. This
is somewhat different than in the case of the bacterial enzyme
that shows a primary KIE of 1.8 on the value of kcat for the
epimerization of UDP-GlcNAc.16 When this is considered
together with the observation that PIX does not occur, several
possibilities emerge. One possibility is that the absence of PIX
indicates cleavage of the C-O bond is the rate-determining step
of catalysis. This would imply that the elimination of UDP
proceeds in a stepwise E1 mechanism with deprotonation of
the oxocarbenium intermediate occurring in a rapid second step
that forms 2-acetamidoglucal. The intermediate in this scenario
could also be a species that is covalently bound to an enzyme
nucleophile, or an oxazoline, as has been implicated in certain
glycosidase mechanisms.43,44 Another scenario is that binding,
or a conformational change preceding the first chemical step,
is rate determining and there is a large commitment to catalysis.
Finally, it is possible that PIX was not observed due to restricted
rotation of the terminal phosphate in the bound UDP. In this
case, the rate-determining steps may be hydration of the glycal
intermediate or product release. Further studies must be
performed to distinguish between these possibilities.
Acknowledgment. We thank Dr. Elena Polishchuk and Louis
Luk (UBC) for help with the growing of insect cell cultures.
We thank the Natural Sciences and Engineering Research
Council of Canada (NSERC), the Fonds der Chemischen
Industrie, Frankfurt/Main, Germany, and the Sonnenfeld-Stif-
tung, Berlin, Germany, for financial support.
JA021309G
(43) Zechel, D. L.; Withers, S. G. Acc. Chem. Res. 2000, 33, 11-18.
(44) Knapp, S.; Vocadlo, D.; Gao, Z.; Kirk, B.; Lou, J.; Withers, S. G. J. Am.
Chem. Soc. 1996, 118, 6804-6805.
(41) Cleland, W. W. Biochemistry 1990, 29, 3194-3197.
(42) Legler, G. AdV. Carbohydr. Chem. Biochem. 1990, 48, 349-357.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2461