Mechanism and active site residues of GDP-fucose synthase

Article abstract of DOI:10.1021/ja807799k

L-Fucose, 6-deoxy-L-galactose, is a key component of many important glycoconjugates including the blood group antigens and the Lewis^x ligands. The biosynthesis of GDP-L-fucose begins with the action cof a dehydratase that converts GDP-D-mannose into GDP-4-keto-6-deoxy-mannose. The enzyme GDP-fucose synthase, GFS, (also known as GDP-4-keto-6-deoxy-D-mannose epimerase/reductase, GMER) then converts GDP-4-keto-6-deoxy-D-mannose into GDP-L-fucose. The GFS reaction involves epimerizations at both C-3 and C-5 followed by an NADPH-dependent reduction of the carbonyl at C-4. This manuscript describes studies that elucidate the order of the epimerization steps and the roles of the active site acid/base residues responsible for the epimerizations. An active site mutant, Cys109Ser, produces GDP-6-deoxy-D-altrose as its major product indicating that C-3 epimerization occurs first and premature reduction of the GDP-4-keto-6-deoxy-D-altrose intermediate becomes competitive with GDP-L-fucose production. The same mutation results in the appearance of a kinetic isotope effect when [3 - ²H]-GDP-6-deoxy-4-keto- mannose is used as a substrate. This indicates that Cys109 is the base responsible for the deprotonation of the substrate at C-3. The Cys109Ser mutant also catalyzes a rapid wash-in of solvent derived deuterium into the C-5 position of GDP-fucose in the presence of NADP⁺. This confirms the order of epimerizations and the role of Cys109. Finally, the inactive His179Gln mutant readily catalyzes the wash-out of deuterium from the C-3 position of [3 - ²H]-GDP-6- deoxy-4-keto-mannose. Together these results strongly implicate an ordered sequence of epimerizations (C-3 followed by C-5 ) and suggest that Cys109 acts as a base and His179 acts as an acid in both epimerization steps.

Full text of DOI:10.1021/ja807799k

Published on Web 12/03/2008
Mechanism and Active Site Residues of GDP-Fucose
Synthase
Stephen T. B. Lau and Martin E. Tanner*
Contribution from the Department of Chemistry, UniVersity of British Columbia, VancouVer,
British Columbia, V6T 1Z1 Canada
Received October 2, 2008; E-mail: mtanner@chem.ubc.ca
Abstract: L-Fucose, 6-deoxy-L-galactose, is a key component of many important glycoconjugates including
the blood group antigens and the LewisX ligands. The biosynthesis of GDP-L-fucose begins with the action
of a dehydratase that converts GDP-D-mannose into GDP-4-keto-6-deoxy-mannose. The enzyme GDP-
fucose synthase, GFS, (also known as GDP-4-keto-6-deoxy-D-mannose epimerase/reductase, GMER) then
converts GDP-4-keto-6-deoxy-D-mannose into GDP-L-fucose. The GFS reaction involves epimerizations
at both C-3′′ and C-5′′ followed by an NADPH-dependent reduction of the carbonyl at C-4. This manuscript
describes studies that elucidate the order of the epimerization steps and the roles of the active site acid/
base residues responsible for the epimerizations. An active site mutant, Cys109Ser, produces GDP-6-
deoxy-D-altrose as its major product indicating that C-3′′ epimerization occurs first and premature reduction
of the GDP-4-keto-6-deoxy-D-altrose intermediate becomes competitive with GDP-L-fucose production. The
same mutation results in the appearance of a kinetic isotope effect when [3′′-2H]-GDP-6-deoxy-4-keto-
mannose is used as a substrate. This indicates that Cys109 is the base responsible for the deprotonation
of the substrate at C-3′′. The Cys109Ser mutant also catalyzes a rapid wash-in of solvent derived deuterium
into the C-5′′ position of GDP-fucose in the presence of NADP+. This confirms the order of epimerizations
and the role of Cys109. Finally, the inactive His179Gln mutant readily catalyzes the wash-out of deuterium
from the C-3′′ position of [3′′-2H]-GDP-6-deoxy-4-keto-mannose. Together these results strongly implicate
an ordered sequence of epimerizations (C-3′′ followed by C-5′′) and suggest that Cys109 acts as a base
and His179 acts as an acid in both epimerization steps.
attractive targets for drug development.8 In many strains of
bacteria L-fucose is found as a component of capsular polysac-
Introduction
charides and lipopolysaccharides (LPS).9 Notably, the human
pathogen Helicobacter pylori (the causative agent of peptic and
duodenal ulcers), expresses the LewisX trisaccharide as the major
component of its LPS in an effort to evade the host’s immune
system via a strategy of molecular mimicry.10
L-Fucose, or 6-deoxy-L-galactose, is found as a key compo-
nent of many biologically important glycoconjugates in both
prokaryotes and eukaryotes.1,2 In mammals, L-fucose is a
terminal sugar in the glycans that comprise the human ABO
blood group antigens.3 It is also an essential component of the
cell surface ligands for the selectin proteins.4 The binding of
endothelial selectins to the fucose-containing LewisX and sialyl
LewisX structures on the surface of leukocytes is responsible
for initiating the inflammatory response. Humans unable to
fucosylate proteins suffer from the immune disorder known as
leukocyte adhesion deficiency type II.5 Conversely, O-fucosy-
lation of the Notch receptors is critical for the proper functioning
of the Notch signaling pathways and increased levels of
fucosylation has been linked to cancer.6,7 In human parasites
such as Trypanosoma brucei (the causative agent of African
sleeping sickness), fucose biosynthesis is essential for the life
cycle of the parasite and the biosynthetic enzymes represent
L-Fucose is biosynthesized as the sugar nucleotide, GDP-L-
fucose, in two enzymatic steps from the precursor GDP-D-mannose
(Figure 1).2,11,12 The first step is catalyzed by GDP-mannose 4,6-
dehydratase (GMD) and produces GDP-6-deoxy-R-D-lyxo-hexos-
4-ulose (GDP-6-deoxy-4-keto-mannose).13,14 This enzyme cata-
lyzes an overall redox neutral dehydration reaction involving
an initial NADP+-dependent oxidation at C-4′′, followed by an
elimination of water between C-5′′ and C-6′′, and a final
reduction at the C-6′′ carbon of the resulting enone intermediate.
(8) Turnock, D. C.; Izquierdo, L.; Ferguson, M. A. J. J. Biol. Chem. 2007,
282, 28853–28863.
(9) Maki, M.; Renkonen, R. Glycobiology 2004, 14, 1R–15R.
(10) Wirth, H. P.; Yang, M. Q.; Karita, M.; Blaser, M. J. Infect. Immun.
1996, 64, 4598–4605.
(1) Ma, B.; Simala-Grant, J. L.; Taylor, D. E. Glycobiology 2006, 16,
158R–184R.
(11) Wu, B. Y.; Zhang, Y. X.; Wang, P. G. Biochem. Biophys. Res.
Commun. 2001, 285, 364–371.
(2) Becker, D. J.; Lowe, J. B. Glycobiology 2003, 13, 41R–53R.
(3) Watkins, W. M. Transfus. Med. 2001, 11, 243–265.
(4) Lowe, J. B. Curr. Opin. Cell Biol. 2003, 15, 531–538.
(5) Yakubenia, S.; Wild, M. K. FEBS J. 2006, 273, 4390–4398.
(6) Miyoshi, E.; Moriwaki, K.; Nakagawa, T. J. Biochem. 2008, 143, 725–
729.
(12) Ginsburg, V. J. Biol. Chem. 1961, 236, 2389–2393.
(13) Mulichak, A. M.; Bonin, C. P.; Reiter, W.-D.; Garavito, R. M.
Biochemistry 2002, 41, 15578–15589.
(14) Somoza, J. R.; Menon, S.; Schmidt, H.; Joseph-McCarthy, D.; Dessen,
A.; Stahl, M. L.; Somers, W. S.; Sullivan, F. X. Structure 2000, 8,
123–135.
(7) Stanley, P. Curr. Opin. Struct. Biol. 2007, 17, 530–535.
9
10.1021/ja807799k CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 17593–17602 17593

A R T I C L E S
Lau and Tanner
4-keto-L-galactose). Finally, an NADPH-dependent reduction
converts the C-4′′ ketone into a hydroxyl group and generates
GDP-L-fucose. Studies that support this mechanism include the
observation that GDP-6-deoxy-4-keto-mannose bearing a tritium
label at C-3′′ loses its label to solvent during catalysis.20 This
indicates that the inversions of stereochemistry occur via
deprotonation/reprotonation events accompanied by solvent
isotope incorporation. This is also consistent with the observa-
tion that GFS can catalyze the epimerization steps in the absence
of NADPH.17 Further studies have shown that GFS catalyzes
the stereospecific transfer of the pro-S hydride from NADPH
to the C-4′′ position of the sugar nucleotide.17 When deuterated
NADPH was employed, the reaction was slowed by a kinetic
isotope effect (KIE) on Vmax of 1.4, indicating that reduction
was only partially rate-limiting during catalysis.
Figure 1. Biosynthesis of GDP-L-fucose from GDP-D-mannose.
The structure of GFS from Escherichia coli in complex with
NADP+ and NADPH has been solved by X-ray crystallography;
however,nostructureswithaboundGDP-sugarareavailable.16,18,19
The structures clearly show that GFS is a member of the short
chain dehydrogenase reductase (SDR) family of enzymes,21,22
and that its fold closely resembles that of UDP-galactose
4-epimerase,23 an enzyme that also catalyzes redox chemistry
at the C-4′′ position of a sugar nucleotide. GFS bears the
characteristic Ser-Tyr-Lys catalytic triad that is a hallmark of
the SDR family members and is responsible for promoting the
hydride transfer step.24 From both the structural analysis of GFS
and literature precedence with other sugar nucleotide-modifying
SDR enzymes,25-27 Tyr136 of this triad is strongly implicated
as the catalytic acid that protonates the C4 carbonyl during the
final reduction step (B3 in Figure 2). Two additional active site
residues, Cys109 and His179, which are not conserved through-
out the SDR family have been implicated as the acid/bases
involved in promoting the inversions of stereochemistry (B1 and
B2 in Figure 2). While the absence of bound substrate
complicates this assignment, modeling studies suggest these are
the only two residues that would be in an appropriate position
to play these roles.16,18,19 In addition, site directed mutations
of the Tyr136, Cys109, and His179 residues lead to enzymes
with dramatically lowered activities indicating that they are all
crucially important for efficient catalysis.16
Figure 2. Putative mechanism for the reaction catalyzed by GDP-L-fucose
synthase. Accurate conformations are not portrayed in this figure. B1, B2,
and B3 represent active site acid/base residues.
This step introduces the characteristic 6-methyl group of fucose
as well as the 4-keto functionality that is required for the
subsequent epimerizations. The second step in the biosynthesis
is catalyzed GDP-fucose synthase or GFS (also known as GDP-
4-keto-6-deoxy-D-mannose epimerase/reductase or GMER).15-20
GFS is remarkable in that it is able to catalyze three distinct
reactions within a single active site. These include inversions
of stereochemistry at both C-3′′ and C-5′′, as well as an
NADPH-dependent reduction of the ketone functionality at C-4.
A putative mechanism for the reaction catalyzed by GFS is
shown in Figure 2. An initial deprotonation at C-3′′ generates
an enolate intermediate that is reprotonated on the opposite face
to give the C-3′′ epimer, GDP-6-deoxy-R-D-arabino-hexos-4-
ulose (GDP-6-deoxy-4-keto-altrose). A similar deprotonation/
reprotonation sequence then inverts the stereochemistry at C-5′′
to give GDP-6-deoxy-R-L-xylo-hexos-4-ulose (GDP-6-deoxy-
Further insight into the GDP-fucose synthase reaction has
been gained from studies on the related enzyme GDP-mannose
3,5-epimerase involved in vitamin C biosynthesis in plants
(Figure 3).28-31 This enzyme can convert GDP-D-mannose into
(21) Kallberg, Y.; Oppermann, U.; Jo¨rnvall, H.; Persson, B. Protein Sci.
2002, 11, 636–641.
(22) Jo¨rnvall, H.; Persson, B.; Krook, M.; Atrian, S.; Gonza`les-Duarte, R.;
Jeffrey, J.; Ghosh, D. Biochemistry 1995, 34, 6003–6013.
(23) Thoden, J. B.; Frey, P. A.; Holden, H. M. Biochemistry 1996, 35,
5137–5144.
(24) Filling, C.; Berndt, K. D.; Benach, J.; Knapp, S.; Prozorovski, T.;
Nordling, E.; Ladenstein, R.; Jo¨rnvall, H.; Oppermann, U. J. Biol.
Chem. 2002, 277, 25677–25684.
(25) Gerratana, B.; Cleland, W. W.; Frey, P. A. Biochemistry 2001, 40,
9187–9195.
(26) Thoden, J. B.; Wohlers, T. M.; Fridovich-Keil, J. L.; Holden, H. M.
Biochemistry 2000, 39, 5691–5701.
(15) Rentmeister, A.; Hoh, C.; Weidner, S.; Dra¨ger, G.; Elling, L.; Liese,
A.; Wandrey, C. Biocatal. Biotransform. 2004, 22, 49–56.
(16) Rosano, C.; Bisso, A.; Izzo, G.; Tonetti, M.; Sturla, L.; De Flora, A.;
Bolognesi, M. J. Mol. Biol. 2000, 303, 77–91.
(27) Liu, Y.; Thoden, J. B.; Kim, J.; Berger, E.; Gulick, A. M.; Ruzicka,
F. J.; Holden, H. M.; Frey, P. A. Biochemistry 1997, 36, 10675–10684.
(28) Major, L. L.; Woluka, B. A.; Naismith, J. H. J. Am. Chem. Soc. 2005,
127, 18309–18320.
(17) Menon, S.; Stahl, M.; Kumar, R.; Xu, G.-Y.; Sullivan, F. J. Biol. Chem.
1999, 274, 26743–26750.
(29) Woluka, B. A.; Van Montagu, M. J. Biol. Chem. 2003, 278, 47483–
47490.
(18) Somers, W. S.; Stahl, M. L.; Sullivan, F. X. Structure 1998, 6, 1601–
1612.
(30) Woluka, B. A.; Persiau, G.; Van Doorsselaere, J.; Davey, M. W.;
Demol, H.; Vandekerckhove, J.; Van Montagu, M.; Zabeau, M.;
Boerjan, W. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 14843–14848.
(31) Barber, G. A. J. Biol. Chem. 1979, 254, 7600–7603.
(19) Rizzi, M.; Tonetti, M.; Vigevani, P.; Sturla, L.; Bisso, A.; De Flora,
A.; Bordo, D.; Bolognesi, M. Structure 1998, 6, 1453–1465.
(20) Chang, S.; Duerr, B.; Serif, G. J. Biol. Chem. 1988, 263, 1693–1697.
9
17594 J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008

Mechanism and Residues of GDP-Fucose Synthase
A R T I C L E S
anticipated that if Cys109 were acting as a catalytic acid/base
residue, the Cys109Ser mutant would retain sufficient residual
activity to allow for thorough kinetic analysis and product
studies. Literature precedence indicates that such a replacement
of a cysteine acid/base catalyst typically leads to a mutant that
retains 0.1-1.0% of the wild type activity.28,32,33 To study the
importance of His179, it was not possible to make such a
conservative mutation and the His179Gln mutant was generated.
The substrate for GFS, GDP-6-deoxy-R-D-lyxo-hexos-4-ulose
(GDP-6-deoxy-4-keto-mannose), was generated by treatment of
GDP-mannose with GMD at room temperature.15 Due to the
sensitive nature of this compound toward decomposition, only
minimal purification was employed, and the material was freshly
prepared before each use. The reaction was monitored by mass
spectral analysis until complete and the enzyme was removed
by ultrafiltration. The product was concentrated by lyophilization
and then used directly in experiments with GFS.
Figure 3. Reactions catalyzed by the enzyme GDP-mannose 3,5-epimerase.
Kinetic analysis of the GFS reaction was performed by
directly monitoring the disappearance of NADPH chromophore
at 340 nm. The wild type enzyme followed Michaelis-Menten
kinetics with values of kcat ) 0.65 ( 0.02 s-1 and KM(GDP-Man)
) 2.2 ( 0.2 µM (saturating NADPH, see Supporting Informa-
tion). The Cys109Ser mutant also followed Michaelis-Menten
kinetics with values of kcat ) 0.0012 ( 0.0001 s-1 and KM(GDP-
Man) ) 0.9 ( 0.2 µM (saturating NADPH). The 540-fold
decrease in the value of kcat for this mutant is consistent with
the notion that the residue is involved in acid/base catalysis.
When the His179Gln mutant was subjected to kinetic analysis,
extremely low levels of activity could be detected, however, it
was not possible to accurately measure kinetic constants using
the direct assay. It was estimated that the value of kcat had been
reduced by at least 10 000-fold by this mutation. This indicates
that His179 is a key catalytic residue and is consistent with
past studies showing that the His179Asn mutant suffered from
a 1 000-fold decrease in the value of kcat.16
Product Analysis in the Wild Type and Cys109Ser Reac-
tions. In order to determine the specificity of the reactions
catalyzed by the wild type and Cys109Ser synthases, product
analyses were performed. Freshly prepared GDP-6-deoxy-4-
keto-mannose was incubated with each of the proteins in the
presence of excess NADPH. The reactions were run to comple-
tion as monitored by mass spectrometry, and the GDP-hexose
produced was isolated using ion-exchange chromatography. In
the case of the reaction catalyzed by wild type GFS, GDP-fucose
was identified as the sole product using 1H spectroscopy.34 This
assignment is clear from the presence of a signal due to the
anomeric proton of a sugar nucleotide (C-1′′) at a chemical shift
corresponding to that of GDP-fucose (doublet of doublets at
4.90 ppm in Figure 4b). The upfield shift of this signal and
large JH1,H2 value is expected for GDP-ꢀ-L-fucose that adopts a
1C4 conformation in solution. This finding highlights the high
specificity in the reduction step of the GFS reaction. The enzyme
will not reduce the carbonyl of the starting material or the singly
inverted intermediate at any significant rate and only delivers a
hydride once both inversions of stereochemistry have taken
place.
an equilibrium mixture of GDP-D-mannose (80%), GDP-L-
gulose (5%), and GDP-L-galactose (15%) via reversible epimer-
izations at both the C-3′′ and C-5′′ positions. This enzyme is
also a member of the SDR family of enzymes and shares 24%
sequence identity with GFS. The proposed mechanism for this
reaction involves an initial oxidation at C-4′′ by a tightly bound
NAD+ cofactor. Subsequent epimerizations take place first at
C-5′′ and then at C-3. Reduction of the C-4′′ carbonyl can occur
with any of the three ketone intermediates leading to the
equilibrium mixture of the three epimeric GDP-hexoses. The
ordering of the epimerization steps in this enzyme is supported
by the observation that the enzyme generates GDP-L-gulose as
a product (C-5′′ epimerization only), but does not produce GDP-
D-talose (C-3′′ epimerization only).28,29
Despite the current level of understanding of the mechanism
of the GDP-fucose synthase reaction, several key aspects remain
unknown. The order of epimerization steps (C-3′′ first or C-5′′
first) is not known and thus the nature of the true intermediate
formed during catalysis is unclear (C-3′′ first is shown in Figure
2). The identity and precise roles of the acid/base residues that
promote the inversions of stereochemistry is not fully under-
stood. While it is clear that the two residues, Cys109 and His179,
fulfill at least some of these duties, it is quite conceivable that
as many as four active site residues could be required to catalyze
two distinct epimerizations (B1′/B1′′ and B2′/B2′′ in Figure 2 could
all be unique residues). This manuscript presents evidence in
favor of a mechanism involving C-3′′ inversion followed by
C-5′′ inversion with GDP-6-deoxy-4-keto-altrose as an inter-
mediate (Figure 2). Cys109 is implicated as the basic residue
responsible for deprotonation at both C-3′′ and C-5′′ (B1′/B1′′
in Figure 2) and His179 is implicated as the acidic residue
responsible for protonation at both C-3′′ and C-5′′ (B2′/B2′′ in
Figure 2).
Results
Enzyme Preparation and Kinetic Analysis. Genes encoding
for the recombinant versions of the E. coli GDP-mannose 2,4-
dehydratase (GMD) and GDP-fucose synthase (GFS) bearing
N-terminal hexahistidine tags were expressed in E. coli and the
resulting enzymes were purified by metal affinity chromatog-
raphy. In past studies, the importance of Cys109 in the GFS
reaction was probed by generating the Cys109Ala mutant and
demonstrating that it suffered from a 7000-fold decrease in
enzymatic activity.16 In this study, a more conservative change
was desired and the Cys109Ser mutant was prepared. It was
In the case of the Cys109Ser mutant, a different outcome
was observed in that a mixture of two products were formed
(Figure 4a). Examination of the 1H spectrum (Figure 4c)
(32) Koo, C. W.; Sutherland, A.; Vederas, J. C.; Blanchard, J. S. J. Am.
Chem. Soc. 2000, 122, 6122–6123.
(33) Glavas, S.; Tanner, M. E. Biochemistry 1999, 38, 4106–4113.
(34) Timmons, S. C.; Jakeman, D. L. Org. Lett. 2007, 9, 1227–1230.
9
J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008 17595

A R T I C L E S
Lau and Tanner
Figure 5. (A) Preparation of [5′′-2H]-GDP-6-deoxy-4-keto-D-mannose. (B)
The preparation of [3′′-2H]-GDP-6-deoxy-4-keto-D-mannose.
reaction is accompanied by solvent isotope incorporation at C-5′′
(Figure 5a).35,36 Lyophilization and resuspension in buffer
prepared with H2O prepares the substrate for the wash-out
experiment. In order to prepare [3′′-2H]-GDP-6-deoxy-4-keto-
mannose, a sample of GDP-mannose was extensively incubated
with GDP-mannose 3,5-epimerase (Figure 3) in D2O to generate
a mixture of [3′′,5′′-2H]-GDP-D-mannose (80%), [3′′,5′′-2H]-
GDP-L-gulose (5%), and [3′′,5′′-2H]-GDP-L-galactose (15%)
(Figure 5b). This mixture was isolated by size-exclusion
chromatography and then treated with GMD in a buffer prepared
with H2O. GMD was found only to accept [3′′,5′′-2H]-GDP-D-
mannose as a substrate and therefore generated a sample of [3′′-
2H]-GDP-6-deoxy-4-keto-mannose (80%) contaminated with
[3′′,5′′-2H]-GDP-L-gulose (5%), and [3′′,5′′-2H]-GDP-L-galactose
(15%) that was used directly in the wash-out experiment. Both
labeled substrates were treated with the Cys109Ser mutant and
excess NADPH and the resulting products were analyzed by
mass spectrometry. In the case of [3′′-2H]-GDP-6-deoxy-4-keto-
mannose (m/z ) 587), no deuterium was detected in the
spectrum of the reduced products (m/z ) 588), whereas with
[5′′-2H]-GDP-6-deoxy-4-keto-mannose (m/z ) 587), 75% of the
resulting product retained one deuterium label (m/z ) 589).
Since solvent isotope incorporation accompanies an inversion
of stereochemistry,20 it is clear that the isomer formed by the
Cys109Ser mutant is GDP-6-deoxy-D-altrose (epimerization at
C-3′′ followed by reduction). This indicates GFS first inverts
the stereochemistry at C-3′′ and then inverts the stereochemistry
at C-5′′ during normal catalysis (Figure 2). With the Cys109Ser
mutant, premature reduction of the intermediate can occur before
the second epimerization takes place. The observation that
premature reduction is competitive with GDP-fucose formation
indicates that Cys109 plays a key role in the second epimer-
ization step.
Figure 4. Product analysis of the reaction catalyzed by the Cys109Ser
1
mutant. (A) Structures of the reaction products. (B) H spectrum of GDP-
L-fucose obtained from a reaction with the wild type enzyme. (C) 1H
spectrum of products obtained from a reaction with the Cys109Ser mutant.
1
(D) H spectrum of a GDP-6-deoxy-D-mannose standard.
indicated that only 25% of the product was GDP-fucose (H-1′′,
4.90 ppm) and that the remaining 75% was an isomer (H-1′′,
5.34 ppm). Reasonable possibilities for the structure of this
isomer include GDP-6-deoxy-D-mannose (reduction of starting
material), GDP-6-deoxy-D-altrose (epimerization at C-3′′ fol-
lowed by reduction) or GDP-6-deoxy-L-gulose (epimerization
at C-5′′ followed by reduction). We have found that it is possible
to prepare an authentic sample of GDP-6-deoxy-D-mannose by
treating GDP-6-deoxy-4-keto-mannose with GDP-mannose 4,6-
dehydratase and excess NADPH. Apparently the ′′tightly bound′′
NADP+ cofactor in this enzyme will slowly exchange with the
reduced cofactor in solution and then reduce the C-4′′ carbonyl
1
of its normal reaction product. H analysis of GDP-6-deoxy-
D-mannose prepared in this fashion showed an anomeric proton
shifted downfield from that of the unidentified isomer (H-1′′,
5.43 ppm, Figure 4d). The observed chemical shift is similar to
that of GDP-mannose and is expected for a sugar nucleotide in
a 4C1 conformation. This indicates that the unidentified isomer
did not result from the direct reduction of the starting material
and that a single epimerization had occurred prior to reduction.
The structural elucidation of the isomer by an analysis of
coupling constants or NOE measurements is complicated by
the fact that the species will have at least two substituents in
1
an axial orientation and will therefore adopt both the C4 and
4C1 conformations in solution. These conformations will rapidly
interconvert on the 1H time scale and a time-averaged spectrum
will be observed (this is consistent with the midrange chemical
shift at 5.34 ppm). Instead, the site of epimerization was deduced
using an isotopic wash-out experiment using either [5′′-2H]-
GDP-6-deoxy-4-keto-mannose or [3′′-2H]-GDP-6-deoxy-4-keto-
mannose. [5′′-2H]-GDP-6-deoxy-4-keto-mannose can easily be
prepared by carrying out the GMD reaction in buffer prepared
with D2O since it has been well-established that the dehydratase
Kinetic Isotope Effects in the Wild Type and Cys109Ser
Reactions. In order to probe the role of Cys109 as an acid/base
residue in the GFS reaction, kinetic isotope effects (KIEs) were
(35) Hegeman, A. D.; Gross, J. W.; Frey, P. A. Biochemistry 2002, 41,
2797–2804.
(36) Melo, A.; Elliott, W. H.; Glaser, L. J. Biol. Chem. 1968, 243, 1467–
1474.
9
17596 J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008

Mechanism and Residues of GDP-Fucose Synthase
A R T I C L E S
measured with deuterated versions of GDP-fucose. It was
anticipated that any steps requiring this residue would be slowed
in the Cys109Ser mutant and this may result in them becoming
rate-determining. [5′′-2H]-GDP-6-deoxy-4-keto-mannose and
[3′′-2H]-GDP-6-deoxy-4-keto-mannose (containing 5% [3′′,5′′-
2H]-GDP-L-gulose and 15% [3′′,5′′-2H]-GDP-L-galactose) were
used to probe whether deprotonation at either C-5′′ or C-3′′ are
rate-determining steps. The labeled compounds were prepared
as described previously (Figure 5), and the unlabeled counter-
parts were simultaneously prepared using identical conditions
and reagents, but with H2O instead of D2O. Control reactions
(monitored by mass spectrometry) showed that the deuterium
labels in the starting materials did not exchange with solvent at
any significant rate either in the absence or the presence of GFS/
NADPH. With the wild type GFS, the KIEs on the values of
kcat for [5′′-2H]-GDP-6-deoxy-4-keto-mannose and [3′′-2H]-
GDP-6-deoxy-4-keto-mannose were found to be kH/kD ) 1.0
( 0.1 and kH/kD ) 1.0 ( 0.1, respectively. The absence of a
primary KIE with either substrate indicates that the deprotona-
tions at C-3′′ and C-5′′ are not rate-determining steps in the
normal GFS reaction.
In the case of the Cys109Ser mutant, the KIE on the value
of kcat for [5′′-2H]-GDP-6-deoxy-4-keto-mannose was found to
be 1.0 ( 0.1. The absence of any significant KIE with this
substrate was expected since the major product with this mutant
is GDP-6-deoxy-D-altrose that is formed in a process that does
not involve deprotonation at C-5. However, when [3′′-2H]-GDP-
6-deoxy-4-keto-mannose was tested with Cys109Ser, the KIE
on the value of kcat was found to be 1.5 ( 0.1 (see Supporting
Information). This suggests that deprotonation at C-3′′ has
become a partially rate-determining step with this mutant and
a masked primary KIE is observed. This observation strongly
supports the notion that Cys109 serves as the basic residue
involved in the deprotonation at C-3′′ in the first step of the
reaction (B1′ in Figure 2).
Figure 6. 1H spectra showing the wash-in of solvent derived deuterium
into GDP-L-fucose. (A) GDP-L-fucose before treatment with enzyme. (B)
[3′′,5′′-2H]-GDP-L-fucose formed after treatment of GDP-L-fucose with wild
type GFS and NADP+ in D2O. (C) [5′′-2H]-GDP-L-fucose formed after
treatment of GDP-L-fucose with the Cys109Ser mutant and NADP+ in D2O.
His179Gln mutant was examined, no appreciable deuterium
incorporation into product could be detected under similar
conditions. These results are consistent with the notion that C-5′′
is epimerized after C-3′′ in the normal reaction direction, and
therefore the C-5′′ proton is the first to exchange with solvent
deuterium in the reverse direction. The base that deprotonates
the C-5′′ position of the GDP-6-deoxy-4-keto-L-galactose
intermediate is assigned to His179 that normally functions as
an acid in the forward reaction direction (B2′′ in Figure 2). Since
this residue is unperturbed in the Cys109Ser mutant, C-5′′
deprotonation and exchange may readily occur. This exchange
likely occurs during the lifetime of the enol(ate) intermediate
and does not involve epimerization at C-5′′ since Cys109 (B1′′
in Figure 2) has been mutated. In the case of the His179Gln
mutant, it is likely that the reversible oxidation of GDP-fucose
can still occur, but no deprotonation events take place because
the appropriate basic residue is missing.
Solvent Isotope Exchange into GDP-6-deoxy-4-keto-man-
nose with His179Gln. In all of the experiments discussed thus
far, the His179Gln mutant has been essentially inactive toward
catalysis. This could either be due to the mutation of a key
catalytic residue or a more dramatic structural perturbation that
effectively destroys the active site of the enzyme. The observa-
tion of partial reactions with this mutant would support the
former notion and therefore isotopic washout experiments were
performed. A sample of bis-labeled [3′′,5′′-2H]-GDP-6-deoxy-
4-keto-mannose was generated in a similar fashion as the 3′′-
labeled substrate (Figure 5b) with the exception that the GMD
reaction was also carried out in D2O. This material was
incubated with His179Gln and NADPH in a buffer prepared
with H2O and the reaction was monitored by mass spectroscopy.
Under these conditions the GFS activity of this mutant was so
low that no GDP-fucose formed during the incubations.
However, it was clear that the mutant was capable of catalyzing
the wash-out of deuterium from the bis-labeled material and
control reactions demonstrated that this process was not due to
a nonenzymatic background reaction. The loss of deuterium was
Solvent Deuterium Exchange into GDP-Fucose. Information
on the order of epimerizations and on the role of the active site
residues can also be obtained by monitoring solvent-derived
deuterium incorporation into the product, GDP-fucose. When
GDP-fucose and NADP+ are incubated with GFS, no apparent
reaction is observed since the equilibrium of the GFS reaction
strongly favors these products. Nevertheless, it is expected that
the reverse reactions are still kinetically accessible and that the
oxidation and epimerization steps may still take place. To
observe these processes, a sample containing GDP-fucose and
NADP+ was incubated with GFS in a buffer prepared from D2O
and the reaction was monitored by 1H spectroscopy. An
examination of the spectrum taken before the addition of enzyme
shows all four signals due to the H-2′′-H-5′′ protons in the
region between 3.5 and 3.8 ppm (Figure 6a). After treatment
with wild type GFS the signals due to H-3′′ and H-5′′ readily
disappear due to exchange of these protons with solvent-derived
deuterium (Figure 6b). The incorporation of two deuterium
atoms is also observed in purified samples analyzed by mass
spectrometry. Monitoring the time course of this exchange
shows that H-5′′ exchanges roughly twice as fast as H-3′′ (not
shown in Figure 6). The deuterium incorporation results show
that the reverse reaction is kinetically accessible despite the
thermodynamics favoring GDP-fucose production.
Quite a different result was obtained when the Cys109Ser
mutant was subjected to the same experiment. In this case, H-5′′
readily exchanged with solvent deuterium whereas H-3′′ did
not, even after extended incubations (Figure 6c). When the
9
J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008 17597

A R T I C L E S
Lau and Tanner
Figure 7. ESI mass spectra showing the His179Gln catalyzed wash-out of deuterium from labeled GDP-6-deoxy-4-keto-D-mannose. (A) The reaction of
[3′′-2H]-GDP-6-deoxy-4-keto-D-mannose in H2O monitored as a function of time. (B) The reaction of [5′′-2H]-GDP-6-deoxy-4-keto-D-mannose in H2O
monitored as a function of time.
biphasic and one deuterium was lost about 20 times faster than
the other. In order to determine which deuterium was washed-
out more readily, samples of [3′′-2H]-GDP-6-deoxy-4-keto-
mannose and [5′′-2H]-GDP-6-deoxy-4-keto-mannose were in-
dividually treated with the His179Gln mutant and NADPH. In
the case of [3′′-2H]-GDP-6-deoxy-4-keto-mannose, mass spectral
analysis indicated that ∼35% of the label had been lost after 5
min of incubation (Figure 7a). With [5′′-2H]-GDP-6-deoxy-4-
keto-mannose, a similar extent of isotopic wash-out required
90 min of incubation under identical conditions (Figure 7b).
The observation that the His179Gln mutant catalyzes isotope
exchange into starting material indicates that its active site has
not been dramatically perturbed and it is still able to bind
substrate and catalyze deprotonation events. Cys109 can still
act as a functional base (B1′ in Figure 2) and is able to generate
enolate intermediates, however, the absence of a functional acid
(His179 or B2′ in Figure 2) prevents epimerization and conver-
sion to product. The fact that label is lost from the C-3′′ position
much more rapidly than from the C-5′′ position is consistent
with the notion that C-3′′ is epimerized first in the normal
reaction mechanism.
Figure 8. Proposed mechanism for the reaction catalyzed by GDP-L-fucose
synthase.
Discussion
states. In the final step of the reaction, a reduction at C-4′′ is
The results of this study support a mechanism for the GDP-
fucose synthase reaction involving an ordered sequence of
epimerizations catalyzed by two active site acid/base residues
(Figure 8). In the first epimerization, Cys109 serves as a base
to deprotonate the substrate at C-3′′ and His179 serves as an
acid to protonate the opposite face. In the second epimerization,
both residues play analogous roles in inverting the stereochem-
istry at the C-5′′ position. For the same residues to play the
role of either base or acid in two consecutive catalytic steps,
there must be a rapid shuttling of protons to and from the active
site of the enzyme during the lifetime of the GDP-6-deoxy-4-
keto-altrose intermediate in order to regenerate the appropriate
protonation states. Alternatively, the residues may directly
exchange protons within the active site in order to achieve these
catalyzed by Tyr136 of the conserved catalytic triad found in
all SDR enzymes. With the wild type enzyme, the C-4′′ carbonyl
is only reduced once both epimerizations have occurred, and
GDP-fucose is the only product detected. Although the studies
in this manuscript do not directly address the reason behind
this selectivity, an attractive explanation is that the reduction
will only occur on a hexulose that preferentially adopts the 1C4
conformation as would be expected for the GDP-6-deoxy-4-
keto-L-galactose intermediate (Figure 8).
The fact that the Cys109Ser mutant generates GDP-6-deoxy-
altrose as the major product supports the notion that C-3′′ is
epimerized first and that Cys109 is involved in the second
epimerization step. According to the proposed mechanism, the
mutation will slow down both deprotonation steps in the forward
9
17598 J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008

Mechanism and Residues of GDP-Fucose Synthase
A R T I C L E S
reaction direction (k1, k-1 and k3, k-3 in Figure 8 will all
decrease). GDP-6-deoxy-4-keto-mannose will bind to the mutant
enzyme but is not reduced as it solely adopts the 4C1 conforma-
tion. Instead, a slow epimerization at C-3′′ (involving the Ser109
residue as a base) generates the GDP-6-deoxy-4-keto-altrose
intermediate. This intermediate may partition forward in the
second epimerization step and give GDP-fucose, or it may be
prematurely reduced to generate GDP-6-deoxy-altrose. Since
Cys109 is involved in the second epimerization step, reduction
of this intermediate becomes competitive with GDP-fucose
production. It is likely that the GDP-6-deoxy-4-keto-altrose
Studies with His179Gln showed that while this mutant is
essentially inactive toward GDP-fucose formation, it readily
catalyzes the exchange of the C-3′′ deuterium in [3′′-2H]-GDP-
6-deoxy-4-keto-mannose with a solvent-derived proton. This
indicates that the mutant is still properly folded and capable of
catalyzing partial steps of catalysis. Since Cys109 is still
functional, it may readily abstract the deuteron from C-3′′ and
generate an enol(ate) intermediate. No inversion of stereochem-
istry takes place since the acidic His179 residue has been
mutated, but exchange of the deuteron with a solvent-derived
proton can occur during the lifetime of the intermediate.
Delivery of this proton back to the C-3′′ position results in the
observed isotopic exchange. This observation provides further
support for the ordered epimerization in the GFS reaction and
the proposed role of His179 in acting as an acid in the first
epimerization step.
4
intermediate preferentially adopts a C1 conformation while
bound in the active site and is not reduced in reactions with the
wild type enzyme. However, in the case of the Cys109Ser
mutant the lifetime of this intermediate is sufficient for
1
premature reduction of a minor C4 conformation to become
competitive with turnover to GDP-fucose.
It is interesting to compare the mechanism of the reaction
catalyzed by GDP-fucose synthase to that of GDP-mannose 3,5-
epimerase (Figure 3). A notable difference between these two
enzymes is that the order of epimerizations are opposite. In the
case of GDP-mannose 3,5-epimerase, a product of the reaction
is GDP-L-gulose indicating that the C-5′′ position is inverted
before the C-3′′ position is inverted.28,29 Crystallographic studies
have provided structures of this enzyme complexed with GDP-
D-mannose, GDP-L-galactose, and a mixture of GDP-L-gulose
and GDP-4-keto-L-gulose.28 These structures provide excellent
insight into the conformational changes that accompany catalysis
and on the position of the active site residues relative to the
bound hexose moiety. The tyrosine of the conserved catalytic
triad is positioned appropriately to act as the acid/base catalyst
during the hydride transfer steps of the reaction. In all of the
structures the active site cysteine that is homologous to Cys109
of GFS is positioned appropriately to act as the base responsible
for proton abstraction at either the C-3′′ or C-5′′ positions (GDP-
D-mannose to GDP-L-galactose direction). Similarly, an active
site lysine that is in a homologous position to His179 in GFS,
is implicated as the acid that protonates the enolate intermedi-
ates. Thus the structural studies on GDP-mannose 3,5-epimerase
are in agreement with our conclusions regarding the roles of
active site residues in the GFS reaction. The structural studies
on GDP-mannose 3,5-epimerase also indicate that the bound
GDP-L-gulose adopts a 1C4 conformation in the active site and
thus has undergone a ring flip subsequent to (or during) the
first epimerization. It is difficult to speculate on how this relates
to the GFS mechanism since the conformational preferences of
the two intermediates will be different and may be controlled
by active site architecture. It is also important to note that GDP-
mannose 3,5-epimerase must be able to catalyze redox reactions
at the C-4′′ position of all three stereoisomers, and thus the
conformational requirements for hydride transfer are less
stringent with this enzyme.
The assignment of Cys109 as the base involved in the first
deprotonation step is strongly supported by the measurement
of kinetic isotope effects. With the wild type enzyme, no KIEs
were observed with either [5′′-2H]-GDP-6-deoxy-4-keto-man-
nose or [3′′-2H]-GDP-6-deoxy-4-keto-mannose indicating that
the deprotonations (k1 and k3 in Figure 8) are not rate-limiting
steps in catalysis. This is in agreement with previous studies
showing that reactions with deuterated NADPH are slowed by
a kinetic isotope effect and therefore reduction of the ketone is
a rate-limiting step.17 In the reaction catalyzed by the Cys109Ser
mutant, GDP-6-deoxy-D-altrose is the major product and GDP-
L-fucose is the minor product, but in forming either of them a
deprotonation at the C-3′′ position is required. The observation
of a kinetic isotope effect when [3′′-2H]-GDP-6-deoxy-4-keto-
mannose is used as a substrate with Cys109Ser indicates that
the mutation of this residue has raised the barrier to removal of
the C-3′′ proton and this step has become partially rate-limiting
(k1 in Figure 8 has decreased). Thus, Cys109 can be assigned
as the base responsible for the first deprotonation step. A similar
approach of using Cys-to-Ser mutants combined with kinetic
isotope effect measurements has previously been used to identify
the roles of cysteine residues in the reactions catalyzed by
glutamate racemase and diaminopimelate epimerase.32,33
The final experiment described with the Cys109Ser mutant
involved the wash-in of solvent-derived deuterium into GDP-
fucose in the presence of NADP+. This experiment supports
the ordering of the epimerization steps and the proposed role
of Cys109 as a base in the second epimerization step (forward
reaction direction) (Figure 8). With wild type enzyme the label
is readily incorporated into both the C-3′′ and C-5′′ positions
due to the freely reversible nature of the reaction. However,
with the Cys109Ser mutant deuterium is selectively incorporated
into the C-5′′ position. This is consistent with the proposed order
of the inversions since C-5′′ would be expected to be depro-
tonated first in the reverse reaction direction. The oxidation of
GDP-fucose to give the intermediate GDP-6-deoxy-4-keto-L-
galactose is facile since Cys109 is not involved in the redox
step of the reaction. His179 will readily deprotonate the C-5′′
position of this intermediate, however, an inversion of stereo-
chemistry does not occur since Cys109 (which would act as an
acid in the reverse reaction direction) has been mutated. Instead,
the histidinium moiety may exchange its proton with solvent-
derived deuterium during the lifetime of the enol(ate) intermedi-
ate, and subsequently transfer this deuterium to back to the C-5′′
position.
Conclusions
GDP-fucose synthase catalyzes a notably complex reaction
involving two epimerizations and a reduction. It does so with
remarkable fidelity and only generates a single isomeric product.
Our studies support a mechanism involving an initial epimer-
ization at C-3′′, followed by an epimerization at C-5′′, and then
a final reduction of the C-4′′ carbonyl. The roles of the basic
and acidic residues involved in both of the epimerization steps
have been assigned to Cys109 and His179, respectively. Further
9
J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008 17599

A R T I C L E S
Lau and Tanner
N2 and stored at -80 °C. The cells were resuspended in 15 mL of
lysis buffer (20 mM sodium phosphate, pH 8.0, containing 2 mM
ꢀ-mercaptoethanol, 1 µg/mL aprotinin, and 1 µg/mL pepstatin A)
and lysed by passage through a French Pressure cell at 12 000 psi
The lysate was clarified by spinning in a Sorvall SLA-1500 rotor
at 6500 rpm for 1 h followed by filtration (0.2 µm pore, 33 mm
syringe driven filter, Millipore). A column (20 mL) of Chelating
Sepharose Fast Flow resin (GE Healthcare) was packed and flushed
with distilled water. The column was charged with 2 column
volumes (CV) of 100 mM NiSO4, followed by washing with 2 CV
of distilled water and 3 CV of start buffer (20 mM sodium
phosphate, pH 8.0, containing 0.5 M NaCl, and 5 mM imidazole).
The clarified cell lysate was loaded at a rate of 2 mL/min, and
start buffer was passed through the column at a rate of 3 mL/min
until no more flow-through protein eluted as determined by
monitoring A280. To remove nonspecifically bound proteins, a
mixture of 80% start buffer and 20% elution buffer (20 mM sodium
phosphate, pH 8.0, containing 0.5 M NaCl, 10% v/v glycerol, and
500 mM imidazole) was applied until no more protein eluted.
Finally the histidine-tagged GMD was eluted with 100% elution
buffer. The protein solution was then concentrated using Amicon
Ultra centrifugal filter devices (Millipore). The concentrated enzyme
solution was then divided into aliquots and flash frozen in liquid
N2. Enzyme could be stored at -80 °C for at least 2 months without
significant loss in activity. Typically 50-60 mg of enzyme was
purified from 500 mL culture. GFS (from pT7GFS) and the
Cys109Ser and His179Gln mutants were generated and purified in
an identical manner with the exception that the cell cultures were
incubated for a total of 12 h at 25 °C following induction by IPTG.
GFS could be stored at -80 °C for at least 3 months without
significant loss in activity. Typical protein yields obtained from a
500 mL culture were 110 (GFS), 80 (Cys109Ser), and 60 mg
(His179Gln).
Enzymatic Synthesis of GDP-6-Deoxy-4-keto-D-mannose. To
a mixture of 20 mg GDP-D-mannose, 0.3 mg NADP+, and 1 µL
ꢀ-mercaptoethanol in 2 mL of sodium phosphate buffer (10 mM,
pH 7.0, containing 10 mM NaCl) was added 20 mg of GMD
(previously exchanged into 800 µL of the same buffer using
Amicon Ultra centrifugal filter devices (Millipore)). The reaction
mixture was incubated for 12 h at rt and monitored by negative
electrospray ionization mass spectrometry (-ESI MS, GDP-
mannose m/z ) 604, GDP-6-Deoxy-4-keto-D-mannose m/z )
586) to ensure the reaction reached completion. Protein was
removed using Amicon Ultra centrifugal filter devises (Milli-
pore), and the filtrate was lyophilized to yield GDP-6-deoxy-
4-keto-D-mannose in a mixture with phosphate buffer salt. This
material was used directly in all experiments.
Enzyme Kinetics. Kinetic constants for the GFS reaction were
obtained using a continuous spectroscopic assay that directly
monitors the conversion of NADPH to NADP+ by following
absorbance changes at 340 nm. Assay mixtures contained 50
mM sodium phosphate buffer, pH 8.0, 100 µM NADPH
(saturating), and 0.25-10 µM GDP-6-deoxy-4-keto-D-mannose
in a total final volume of 1.0 mL. The concentration of GDP-
6-deoxy-4-keto-D-mannose stock solutions was determined by
measuring A260 and using an ε260 of 11.8 mM-1 cm-1. Reactions
were initiated by addition of 10 µL of enzyme solution
(containing 0.1 µg of WT-GFS or 60 µg of C109S-GFS) and
the decrease in A340 was monitored at 25 °C. Initial velocities
were calculated using an ε340 of 6 220 M-1 cm-1. Kinetic
parameters were determined from fitting the initial velocities to
the Hill equation using the program GraFit.
Enzymatic Synthesis of GDP-L-Fucose. A sample containing
20 mg of the lyophilized GDP-6-deoxy-4-keto-D-mannose was
redissolved in 2.0 mL of distilled water, and 30 mg of NADPH
and 20 mg of GFS (in 1.0 mL of storage buffer) were added. The
reaction was incubated at 28 °C for 2 h and monitored by negative
electrospray ionization mass spectrometry (-ESI MS, GDP-fucose
m/z ) 588, GDP-6-deoxy-4-keto-D-mannose m/z ) 586) to ensure
structural studies on GFS complexed with sugar nucleotides will
serve to elucidate the conformational changes that accompany
catalysis.
Experimental Section
Materials and General Methods. GDP-D-Mannose, NADPH,
and alkaline phosphatase (from bovine calf intestine) were pur-
chased from Sigma-Aldrich. D2O (99.9%) was purchased from
Cambridge Isotope Laboratories, Inc. GDP-mannose 3,5-epimerase
was prepared by overexpression of the pEHISTEV-GME plasmid
in E. coli Rosetta (DE3) cells and purified as described previously.28
Protein concentrations were determined by the method of Bradford
using bovine serum albumin as the standard.37 1H NMR spectra
were obtained on a Bruker AV300 spectrometer at a field strength
of 300 MHz. Proton decoupled 31P spectra were recorded on the
same spectrometer at 121.5 MHz. Mass spectra were aquired at
the Mass Spectrometry Centre at the University of British Columbia
(UBC) by electrospray ionization (ESI-MS) using a Bruker
Esquire∼LC ion trap mass spectrometer. Oligonucleotides were
synthesized by the Nucleic Acid Protein Service (NAPS) facility
at UBC.
Cloning of the GMD and GFS Genes from E. coli. The GMD
gene was PCR amplified from E. coli K12 genomic DNA using
Platinum Taq polymerase (Invitrogen) and the following two
primers: 5′-GGTATTGAGGGTCGCATGTCAAAAGTC-3′ (for-
ward sequence) and 5′-AGAGGAGAGTTAGAGCCTTATGACTC-
CAGCG-3′ (reverse sequence). The amplified DNA was purified
using a SpinPrep PCR Clean-up kit (Novagen) according to the
manufacturer’s directions. The PCR product was cloned into a pET-
30 Xa/LIC vector (Novagen) according to the manufacturer’s
instructions to yield pT7GMD. The pT7GMD plasmid was ampli-
fied in NovaBlue GigaSingles competent E. coli cells (Novagen).
This plasmid encodes for the expression of GMD bearing an
N-terminal 43-residue peptide that terminates in a hexahistidine
tag. An identical procedure was used to clone the E. coli GFS gene
and produce pT7GFS. PCR amplification employed the following
two primers: 5′-GGTATTGAGGGTCGCATGAGTAAACAACG-
3′ (forward sequence) and 5′-AGAGGAGAGTTAGAGCCTTAC-
CCCCGAAAG-3′ (reverse sequence).
Site-Directed Mutagenesis. Plasmids encoding for the GFS
mutants, Cys109Ser and His179Gln, were generated using the
following two sets of oligonucleotides (nonmatching nucleotides
are underlined): for Cys109Ser, 5′-CTCGGATCGTCCAGCATC-
TACCCGAAAC-3′ (forward sequence) and 5′-GTTTCGGGTA-
GATGCTGGACGATCCGAG-3′ (reverse sequence); for His179Gln,
5′-CACCCGAGTAATTCGCAGGTGATCCCAGCATTGC-3′ (for-
ward sequence) and 5′-GCAATGATGGGATCACCTGCGAAT-
TACTCGGGTG-3′ (reverse sequence). Mutations were introduced
into the pT7GFS vector using the QuikChange Sited-Directed
Mutagenesis Kit (Stratagene) according to manufacturer’s instruc-
tions. The GFS genes in the resulting plasmids were fully sequenced
to ensure that only the desired mutations had been introduced during
PCR amplification.
Purification of Recombinant Proteins. To generate GMD, the
pT7GMD plasmid was transformed into chemically competent
BL21 (DE3) E. coli. Cells were grown on Luria-Bertani (LB) agar
plates containing kanamycin (30 mg/L) at 37 °C for 18 h. A single
colony was used to inoculate 10 mL of LB medium containing 30
mg/L kanamycin and this was incubated for 12 h at 37 °C with
shaking at 225 rpm. This culture was poured into 500 mL of LB
medium containing 30 mg/L kanamycin and grown with shaking
at 225 rpm and 37 °C until an OD600 between 6.0 and 7.0 was
reached. At that point, isopropyl-1-thio-ꢀ-D-galactopyranoside
(IPTG) was added to a final concentration of 0.3 mM and the cells
were grown for an additional 4 h before harvesting by centrifugation
(6000 rpm, 20 min at 4 °C). Cell pellets were flash frozen in liquid
(37) Bradford, M. M. Anal. Biochem. 1976, 72, 248–254.
9
17600 J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008

Mechanism and Residues of GDP-Fucose Synthase
A R T I C L E S
it reached completion. The protein was removed using Amicon Ultra
centrifugal filter devices (Millipore) and the filtrate was loaded onto
a Dowex DE52 anion exchange resin column (1.5 cm diameter ×
28 cm length). Product was eluted with a linear gradient of 0-0.3
M triethylammonium bicarbonate buffer, pH 7.5. Eluent fractions
were monitored at A260 and GDP-L-fucose eluted at 0.18 M buffer
as a mixture with NADP+. NADPH eluted immediately afterward.
All the NADPH-free fractions were pooled and lyophilized. This
mixture of NADP+ and GDP-L-fucose was redissolved in 10 mL
of 20 mM Trien-HCl buffer, pH 8.0, and 80 units of alkaline
phosphatase was added. The resulting solution was incubated for
12 h at 30 °C and then lyophilized. A second anion exchange
column was run using identical conditions, and the fractions
containing GDP-L-fucose (well separated from those containing
NAD+) were pooled and lyophilized. The residue was redissolved
in 15 mL of distilled water and lyophilized again to GDP-L-fucose
(typical overall yield was 18-20 mg, ∼75%) as a bis-triethylam-
monium salt. GDP-L-fucose prepared in this fashion showed
identical -ESI MS, 31P and 1H spectra to those previously reported
in the literature.34
sample of Trien-DCl buffer in D2O (20 mM, pD 7.4) was prepared
by repeated (3×) lyophilization of a Trien-HCl buffer (20 mM,
pH 7.0) and dissolution in an equal volume of D2O. To 2.0 mL of
this buffer was added 25 mg of GDP-mannose, 1.0 µL of
ꢀ-mercaptoethanol, and 25 mg of GDP-mannose 3,5-epimerase
(previously exchanged into 800 µL of the same buffer using Amicon
Ultra centrifugal filter devices (Millipore)). This reaction mixture
was incubated for 12 h at 37 °C and monitored by negative
electrospray ionization mass spectrometry to ensure that two atoms
of deuterium were incorporated into >95% of the material (m/z )
606). The protein was then removed using Amicon Ultra centrifugal
filter devices (Millipore). The filtrate was lyophilized, redissolved
in 1.5 mL of distilled water, and loaded onto a size exclusion
chromatography column (Biogel P-2, 200-400 mesh, 2.5 cm
diameter × 70 cm length). The mixture of dideuterated sugar
nucleotides was eluted with distilled water and fractions absorbing
at 260 nm were pooled and lyophilized. Nondeuterated control
samples for use in KIE determinations were also prepared in
buffered H2O under otherwise identical conditions.
In order to obtain [3′′-2H]-GDP-6-deoxy-4-keto-D-mannose, the
mixture of dideuterated epimeric GDP-hexoses was treated with
GMD in buffered H2O as described for the synthesis of GDP-6-
deoxy-4-keto-D-mannose. The unlabeled control sample for use in
KIE studies was prepared in a similar fashion. In order to obtain
[3,5′′-2H]-GDP-6-deoxy-4-keto-D-mannose, the mixture of dideu-
terated epimeric GDP-hexoses was treated with GMD in buffered
D2O as described for the synthesis of [5′′-2H]-GDP-6-deoxy-4-keto-
D-mannose. The reactions were monitored by -ESI-MS and observed
to stop after approximately 80% of the total material had been
consumed, indicating that only GDP-mannose served as a substrate
for the GMD reaction. Mass analysis of the mono- and dideuterated
products indicated that >90% of the material contained one and
two atoms of deuterium, respectively (-ESI MS, [3′′-2H]-GDP-6-
deoxy-4-keto-D-mannose m/z ) 587, [3′′,5′′-2H]-GDP-6-deoxy-4-
keto-D-mannose m/z ) 588).
Enzymatic Synthesis of GDP-6-deoxy-D-mannose. To a mix-
ture of 25 mg GDP-D-mannose, 0.38 mg NADP+, and 1.3 µL
ꢀ-mercaptoethanol in 2.5 mL of sodium phosphate buffer (10 mM,
pH 7.0, containing 10 mM NaCl) was added 20 mg of GMD
(previously exchanged into 800 µL of the same buffer using Amicon
Ultra centrifugal filter devices (Millipore)). The reaction mixture
was incubated for 12 h at rt and monitored by negative electrospray
ionization mass spectrometry (-ESI MS, GDP-mannose m/z ) 604,
GDP-6-deoxy-4-keto-D-mannose m/z ) 586) to ensure the reaction
reached completion. To this mixture was added 35 mg of NADPH
and the reaction was incubated for a further 12 h at rt. The progress
of the reaction was judged to be complete by the consumption of
GDP-6-deoxy-4-keto-D-mannose (m/z of 586) and the production
of GDP-6-deoxy-D-mannose (m/z of 588) as monitored by -ESI
MS. The enzyme was removed and product was purified as
described for GDP-L-fucose synthesis. The final product was
isolated as a bis-triethylammonium salt with an overall yield of
60% (∼20 mg). 1H (400 MHz, D2O, as triethylammonium salt, pH
∼7.4): δ (ppm) 8.13 (s, 1H, 8-H), 5.95 (d, J1′,2′ ) 6 Hz, 1H, 1′-H),
Product Analysis with the Cys109Ser Mutant. Lyophilized
GDP-6-deoxy-4-keto-D-mannose (20 mg, prepared as described
above) and 30 mg of NADPH were dissolved in 2.0 mL of water.
To this solution was added 20 mg of Cys109Ser (in 1.0 mL of
sodium phosphate buffer) and the reaction was incubated at 28 °C
for 4 h. The progress of the reaction was monitored by -ESI MS
and judged to be complete by the disappearance of the GDP-6-
deoxy-4-keto-D-mannose signal (m/z ) 586). The GDP-hexose
products were purified as described for the enzymatic synthesis of
GDP-L-fucose. The final product mixture was isolated as bis-
triethylammonium salts with an overall yield of 65% (16 mg). The
reactions of deuterated substrates were carried out under analogous
conditions and monitored by -ESI-MS.
5.46 (br, d, J1′′,P ) 6.8 Hz, 1H, 1′′-H), 4.80 (dd, J2′,1′ ) 6 Hz, J2′,3′
)
4.8 Hz, 1H, 2′-H), 4.54 (dd, J3′,2′ ) 4.8 Hz, J3′,4′ ) 3.6 Hz, 1H,
3′H), 4.37 (m, 1H, 4′-H), 4.23 (m, 2H, 5a′-H, 5b′-H), 4.06 (br, m,
J2′′,3′′ ) 3.2 Hz, 1H, 2′′-H), 3.94 (dd, J3′′,2′′ ) 3.2 Hz, J3′′,4′′ ) 10
Hz, 1H, 3′′-H), 3.90 (dd, J5′′,4′′ ) 10 Hz, J5′′,6′′ ) 3.2 Hz, 1H, 5′′-
H), 3.31 (t, J4′′,3′′ ) 10 Hz, J4′′,5′′ ) 10 Hz, 1H, 4′′-H), 1.25 (d,
J6′′,5′′ ) 3.2 Hz, 3H, 6′′-H); 31P {1H} NMR (121.5 MHz, D2O, as
triethylammonium salt, pH ∼7.4): δ (ppm) -10.88 (d, Jp,p ) 20
Hz, PR), -13.31 (d, Jp,p ) 20 Hz, Pꢀ). MS (ESI-) for C16H25N5O15P2
(free acid, 589.3) ) m/z 588.2 [M - H].
Deuterium Wash-in Experiments with GDP-L-Fucose. Deu-
terium wash-in experiments were performed using GFS-WT,
Cys109Ser, and His179Gln. A sample of deuterated phosphate
buffer (10 mM, pD 7.4) was prepared by repeated (3×) lyophiliza-
tion of a phosphate buffer (10 mM, pH 7.0) and dissolution in an
equal volume of D2O. Three samples containing 2.0 mg GDP-
fucose, 0.1 mg NADP+, and enzyme (3 mg of GFS-WT, 7 mg of
Cys109Ser, or 10 mg of His179Gln, previously exchanged into
deuterated buffer using Amicon Ultra centrifugal filter devices
(Millipore)) were prepared in the deuterated buffer (final volume
800 µL). The samples were incubated at 28 °C and the progress of
deuterium wash-in was monitored using 1H spectroscopy. The mass
of the final product was confirmed by -ESI-MS upon completion
of the deuterium wash-in process.
Deuterium Wash-out Experiments with His179Gln. Samples
containing 5 mg of substrate ([3′′, 5′′-2H]-GDP-6-deoxy-4-keto-D-
mannose, [3′′-2H]-GDP-6-deoxy-4-keto-D-mannose, or [5′′-2H]-
GDP-6-deoxy-4-keto-D-mannose), 4.5 mg of His179Gln (previously
exchanged into the same buffer using Amicon Ultra centrifugal filter
devices (Millipore)), and 5 mg NADPH were prepared in 10 mM
sodium phosphate buffer (pH 7.0, 10 mM, 750 µL final volume).
Enzymatic Synthesis of [5′′-2H]-GDP-6-Deoxy-4-keto-D-man-
nose. The labeled substrate was prepared in an identical fashion
to the unlabeled compound except that the GMD reaction was
carried out in buffered D2O. A deuterated sodium phosphate
buffer ((10 mM, pD 7.4, containing 10 mM NaCl) was prepared
by lyophilizing a sample of nondeuterated buffer and redissolving
it in an equal volume of D2O. This was repeated three times to
ensure complete isotopic exchange. The GFS reaction was run
as described above (with previous exchange of GFS into the
same deuterated buffer) and negative electrospray ionization
mass spectrometry indicated that the reaction proceeded to
completion and that >97% of the product contained a single
deuterium (-ESI MS, GDP-mannose m/z ) 604, [5′′-2H]-GDP-
6-deoxy-4-keto-D-mannose m/z ) 587).
Enzymatic Synthesis of [3′′-2H]-GDP-6-Deoxy-4-keto-D-man-
nose and [3, 5′′-2H]-GDP-6-Deoxy-4-keto-D-mannose. Both the
[3′′-2H]- and the [3, 5′′-2H]-labeled substrates were prepared from
a mixture containing [3, 5′′-2H]-GDP-mannose (80%), [3, 5′′-2H]-
GDP-L-gulose (5%) and [3, 5′′-2H]-GDP-L-galactose (15%) that was
generated using GDP-mannose 3,5-epimerase in buffered D2O. A
9
J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008 17601

A R T I C L E S
Lau and Tanner
The samples were initiated by the addition of enzyme, incubated
at rt, and monitored by -ESI-MS in 15 min intervals.
Supporting Information Available: Kinetic traces for the
reaction of the wild type GFS and the KIE determination with
the Cys109Ser mutant. This material is available free of charge
via the Internet at http://pubs.acs.org/.
Acknowledgment. We thank Dr. James H. Naismith and Louise
L. Major for donation of the pEHISTEV-GME plasmid. We thank
the Natural Sciences and Engineering Research Council of Canada
(NSERC) for financial support.
JA807799K
9
17602 J. AM. CHEM. SOC. VOL. 130, NO. 51, 2008