Chemoenzymatic synthesis and enzymatic analysis of 8-modified cytidine monophosphate-sialic acid and sialyl lactose derivatives

Article abstract of DOI:10.1021/ja102644a

The sialic acids found on eukaryotic glycans have remarkably diverse core structures, with a range of modifications at C5, C7, C8 and C9. These carbohydrates have been found to play key roles in cell-cell interactions within eukaryotes and often serve as the initial site of attachment for viruses and bacteria. Consequently simple changes to the structures of the sialic acids can result in profoundly different and often opposing biological effects. Of particular importance are modifications at the 8-position. These include O-acetylation, carried out by an acetyl transferase, and particularly polysialylation, catalyzed by a polysialyltransferase. As part of a structural and mechanistic study of sialyltransferases and polysialyltransferases, access was needed to sialic acid-containing oligosaccharides that are modified at the 8-position of the sialic acid to render this center non-nucleophilic. The free 8-modified sialic acid analogues were synthesized using a concise, divergent chemical synthetic approach, and each was converted to its cytidine monophosphate (CMP) sugar donor form using a bacterial CMP-sialic acid synthetase. The transfer of each of the modified donors to lactose by each of two sialyltransferases was investigated, and kinetic parameters were determined. These yielded insights into the roles of interactions occurring at that position during enzymatic sialyl transfer. A transferase from Campylobacter jejuni was identified as the most suitable for the enzymatic coupling and utilized to synthesize the 8 -modifed sialyl lactose trisaccharides in multimilligram amounts.

Full text of DOI:10.1021/ja102644a

Published on Web 06/17/2010
Chemoenzymatic Synthesis and Enzymatic Analysis of
8-Modified Cytidine Monophosphate-Sialic Acid and Sialyl
Lactose Derivatives
Thomas J. Morley and Stephen G. Withers*
Department of Chemistry, UniVersity of British Columbia, 2036 Main Mall,
VancouVer V6T 1Z1, Canada
Received April 7, 2010; E-mail: withers@chem.ubc.ca
Abstract: The sialic acids found on eukaryotic glycans have remarkably diverse core structures, with a
range of modifications at C5, C7, C8 and C9. These carbohydrates have been found to play key roles in
cell-cell interactions within eukaryotes and often serve as the initial site of attachment for viruses and
bacteria. Consequently simple changes to the structures of the sialic acids can result in profoundly different
and often opposing biological effects. Of particular importance are modifications at the 8-position. These
include O-acetylation, carried out by an acetyl transferase, and particularly polysialylation, catalyzed by a
polysialyltransferase. As part of a structural and mechanistic study of sialyltransferases and polysialyl-
transferases, access was needed to sialic acid-containing oligosaccharides that are modified at the 8-position
of the sialic acid to render this center non-nucleophilic. The free 8-modified sialic acid analogues were
synthesized using a concise, divergent chemical synthetic approach, and each was converted to its cytidine
monophosphate (CMP) sugar donor form using a bacterial CMP-sialic acid synthetase. The transfer of
each of the modified donors to lactose by each of two sialyltransferases was investigated, and kinetic
parameters were determined. These yielded insights into the roles of interactions occurring at that position
during enzymatic sialyl transfer. A transferase from Campylobacter jejuni was identified as the most suitable
for the enzymatic coupling and utilized to synthesize the 8′′-modifed sialyl lactose trisaccharides in
multimilligram amounts.
Introduction
Sialic acid, or N-acetyl neuraminic acid, is a functionally
dense nine-carbon sugar that is normally found as the terminal
sugar of the glycan part of glycoproteins and glycolipids in the
extracellular matrix of mammalian cells.1 Currently over 50
variations of the core structure of sialic acid are known,2,3 and
it has been shown that even subtle changes can have drastic
biological effects. The 8-position is particularly important
because sialylation at this position4 gives rise to the R-2,8-linked
sialic acid polymers (PSA) that play key roles in brain
development5,6 and are found to be overexpressed in some
cancerous tumors,7 as well as being essential to the pathogenicity
of some bacteria.8,9 The 8-position has also been found to be
acetylated,10 methylated, and sulfated,11 with each such modi-
Figure 1. Biologically relevant modifications to the 8-position of sialic
acid containing glycans.
fication possessing its own biological activity (see Figure 1).
Convenient access to sialic acids bearing modifications at the
8-position would therefore be very useful in investigating the
role played by sialic acid in each of these systems.
Of particular interest to us is ready access to sialic acid
containing oligosaccharide derivatives in which the 8-position
of the terminal sialic acid has been rendered non-nucleophilic
by deoxygenation, deoxyfluorination, or O-methylation or by
complete removal through truncation of the glycerol side chain.
Such minimally modified analogues might act as useful “in-
competent” acceptor analogues for the study of enzymes that
transfer sialic acids or other substituents to the 8-position of
sialic acid. They should act as inhibitors of these enzymes,
allowing measurement of binding affinities through kinetic
analyses and permitting X-ray crystallographic structural analy-
(1) Schauer, R. Zoology 2004, 107, 49–64.
(2) Angata, T.; Varki, A. Chem. ReV. 2002, 102, 439–469.
(3) Du, J.; Meledeo, M. A.; Wang, Z.; Khanna, H. S.; Paruchuri, V. D. P.;
Yarema, K. J. Glycobiology 2009, 19, 1382–1401.
(4) Muhlenhoff, M.; Eckhardt, M.; Gerardy-Schahn, R. Curr. Opin. Struct.
Biol. 1998, 8, 558–564.
(5) Rockle, I.; Seidenfaden, R.; Weinhold, B.; Muhlenhoff, M.; Gerardy-
Schahn, R.; Hildebrandt, H. DeV. Neurobiol. 2008, 68, 1170–1184.
(6) Rutishauser, U.; Acheson, A.; Hall, A. K.; Mann, D. M.; Sunshine, J.
Science 1988, 240, 53–57.
(7) Cipolla, L.; Peri, F.; Airoldi, C. Curr. Med. Chem. Anti-Cancer Agents
2008, 8, 92–121.
(8) Devi, S. J. N.; Schneerson, R.; Egan, W.; Vann, W. F.; Robbins, J. B.;
Shiloach, J. Infect. Immun. 1991, 59, 732–736.
(10) Tiralongo, J.; Schauer, R. Trends Glycosci. Glycotechnol. 2004, 16,
1–15.
(9) Jennings, H. J.; Roy, R.; Michon, F. J. Immunol. 1985, 134, 2651–
2657.
(11) Zanetta, J. P.; Pons, A.; Iwersen, M.; Mariller, C.; Leroy, Y.;
Timmerman, P.; Schauer, R. Glycobiology 2001, 11, 663–676.
9
9430 J. AM. CHEM. SOC. 2010, 132, 9430–9437
10.1021/ja102644a  2010 American Chemical Society

8-Modified CMP-Sialic Acid and Derivatives
A R T I C L E S
ses of ternary complexes in which both donor and acceptor
substrate (analogues) are bound.12 Sialic acids in extracellular
glycoconjugates are commonly found conjugated to lactose or
N-acetyl-lactosamine as shown in Figure 1.13 8′′-Modified sialyl
lactose derivatives were chosen as our synthetic targets because
the enzymes under study work well on lactose and aryl
lactosides are more readily prepared than the analogous N-acetyl
glucosamine-containing derivatives. The synthesis of such 8′′-
modified sialyl lactose derivatives could, in principle, be carried
out by chemical modification of sialyl lactose. However, the
complexity of the protecting-group chemistry required and the
presence of multiple functional groups would render such a route
challenging. A more promising alternative is a chemoenzymatic
approach in which a chemical modification of the 8-position
on the sialic acid monosaccharide is carried out first and then
enzymes are employed to transfer the modified sugar onto the
lactose moiety.
Despite the relevance of the 8-hydroxyl group of sialic acids
in biology, there have been few reports of chemical manipulation
of this position. Of these studies most have concentrated on
deoxygenation,14,15 sulfonation16 and simple alkylation,17,18 and
reports of halogenation at this position, with the exception of
iodination,19,20 are entirely absent. These syntheses utilize
lengthy reaction sequences to obtain a protected sialic acid
derivative with only OH8 free, which is further manipulated as
required. Lengthy reaction sequences have been necessary
because direct regioselective functionalization of OH8 is difficult
and thus multiple deprotection and reprotection steps have been
required to gain access to a sugar that has only OH8 free.21 We
present a relatively short reaction sequence, exploiting the
different chemical reactivities of each of the five hydroxyl
groups, to gain access to OH8 starting from a simple sialic acid
glycoside. This route allows the ready manipulation and
installation of multiple functionalities at this position.
relatively remote from the anomeric center, we were optimistic
that this approach could work. Similarly our experience with a
range of sialyltransferases gave us confidence that the CMP
derivatives so formed could act as substrates for at least one of
these enzymes. If successful, such an approach would not only
provide us access to our desired sialyl lactose analogues but
also provide further insights into the specificities of representa-
tives of the two classes of enzymes employed in their synthesis.
This manuscript describes the chemical synthesis of 8-deoxy,
8-deoxy-8-fluoro and 8-O-methyl analogues of sialic acid, as
well as a truncated heptose analogue of sialic acid in which C8
and C9 have been removed. It further describes the conversion
of each of these analogues to their activated CMP donor sugar
using a bacterial CMP-sialic acid synthetase and then use of
these derivatives as substrates for two different bacterial
sialyltransferases to prepare multimilligram quantities of each
modified sialyl lactose derivative. Insights into the specificities
of these two sialyltransferases, Cst-I from Campylobacter
jejuni25 and Pm0188h from Pasteurella multocida,26 were
provided by determination of kinetic parameters for transfer by
each enzyme.
Materials and Methods
All enzymes were obtained from the Sigma-Aldrich company,
with the exception of CMP-sialic acid synthetase,22 Cst-I27 and
Pm0118h,28 which were expressed and purified as previously
described.
The general method for synthesis of 8-modifed CMP-donor
sugars is as follows. Sialic acid analogue (1 equiv) and CTP
disodium salt (1.05 equiv) were dissolved in Tris buffer (100 mM,
pH 8.5) containing magnesium chloride (20 mM) and DTT (0.1
mM) to give a final sialic acid concentration of 15 mM. CMP-
sialic acid synthetase (1 U/µmol) and inorganic pyrophosphatase
(1 U/mmol) were added, and the mixture was tumbled at ambient
temperature. The pH of the solution was checked regularly, and
aqueous sodium hydroxide solution (1 M) was added as appropriate
to keep the pH constant. On completion of reaction, as observed
by TLC analysis (ethyl acetate/methanol/water/concd ammonia
solution in a 4:3:2:1 ratio mobile phase), the mixture was cooled
to -80 °C. Once it had thawed, it was filtered (0.44 µm) and
incubated with alkaline phosphatase (20 U/mmol) for 10 min. The
mixture was filtered (0.44 µm) again and loaded directly onto an
ion-exchange column, pre-equilibrated with ammonium formate (50
mM). After an initial wash, using the same buffer, a stepped gradient
(50 mM to 1 M) was performed, and products containing fractions
were identified by TLC, pooled, and lyophilized. This crude product
was dissolved in the minimum volume of buffer (20 mM am-
monium formate buffer, pH 8.1) and loaded onto a size exclusion
column. The column was run at 10 mL/h, and fractions were
collected every 15 min. Product-containing fractions were identified
by TLC analysis, pooled, and lyophilized.
Once the modified sialic acids are chemically synthesized,
the second step of this chemoenzymatic approach requires access
to two classes of enzymes that will, respectively, synthesize
the CMP-sialic acid derivatives (CMP-sialic acid synthetase)
and transfer the sialic acid (sialyltransferase). Because enzymes
are often rather specific for their substrates, this approach carries
the risk that the enzymes might well not function with such
modified species. However, previous studies have shown that
CMP-sialic acid synthetases can be remarkably tolerant to
substrate modification,22-24 and because the 8-position is
(12) Persson, K.; Ly, H. D.; Dieckelmann, M.; Wakarchuk, W. W.; Withers,
S. G.; Strynadka, N. C. J. Nat. Struct. Biol. 2001, 8, 166–175.
(13) Yu, R. K.; Bieberich, E.; Xia, T.; Zeng, G. C. J. Lipid Res. 2004, 45,
783–793.
(14) Miyazaki, T.; Sato, H.; Sakakibara, T.; Kajihara, Y. J. Am. Chem.
Soc. 2000, 122, 5678–5694.
All Michaelis-Menten kinetic parameters were determined using
a CMP release assay.29 Cst-I kinetics were performed at 37 °C in
HEPES pH 7.5 buffer (20 mM) containing sodium chloride (50
mM), manganese chloride (10 mM), and magnesium chloride (10
mM) at substrate concentrations from 0.1 to 2 mM. Pm0118h
(15) Zbiral, E.; Schreiner, E.; Salunkhe, M. M.; Schulz, G.; Kleineidam,
R. G.; Schauer, R. Liebigs Ann. Chem. 1989, 519–526.
(16) Tanaka, M.; Kai, T.; Sun, X. L.; Takayanagi, H.; Furuhata, K. Chem.
Pharm. Bull. (Tokyo) 1995, 43, 2095–2098.
(17) Dufner, G.; Schworer, R.; Muller, B.; Schmidt, R. R. Eur. J. Org.
Chem. 2000, 1467–1482.
(18) Hanashima, S.; Ishikawa, D.; Akai, S.; Sato, K.-i., Carbohydr. Res.
2009, 344, 747–752.
(25) Gilbert, M.; Brisson, J. R.; Karwaski, M. F.; Michniewicz, J.;
Cunningham, A. M.; Wu, Y. Y.; Young, N. M.; Wakarchuk, W. W.
J. Biol. Chem. 2000, 275, 3896–3906.
(19) Brandstetter, H. H.; Zbiral, E. Liebigs Ann. Chem. 1983, 2055–2065.
(20) Hasegawa, A.; Adachi, K.; Yoshida, M.; Kiso, M. Carbohydr. Res.
1992, 230, 257–272.
(26) Ni, L. S.; Chokhawala, H. A.; Cao, H. Z.; Henning, R.; Ng, L.; Huang,
S. S.; Yu, H.; Chen, X.; Fisher, A. J. Biochemistry 2007, 46, 6288–
6298.
(21) Hemeon, I.; Bennet, A. J. Synthesis 2007, 1899–1926.
(22) Karwaski, M. F.; Wakarchuk, W. W.; Gilbert, M. Protein Expression
Purif. 2002, 25, 237–240.
(27) Chiu, C. P. C.; Lairson, L. L.; Gilbert, M.; Wakarchuk, W. W.; Withers,
S. G.; Strynadka, N. C. J. Biochemistry 2007, 46, 7196–7204.
(28) Yu, H.; Chokhawala, H.; Karpel, R.; Yu, H.; Wu, B. Y.; Zhang, J. B.;
Zhang, Y. X.; Jia, Q.; Chen, X. J. Am. Chem. Soc. 2005, 127, 17618–
17619.
(23) Nahalka, J.; Patoprsty, V. Org. Biomol. Chem 2009, 7, 1778–1780.
(24) Mizanur, R. M.; Pohl, N. L. Appl. Microbiol. Biotechnol. 2008, 80,
757–765.
9
J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010 9431

A R T I C L E S
Morley and Withers
Scheme 1. Synthesis of General Precursorsa
a Reagents and conditions: (i) anisaldehyde dimethyl acetal, TsOH (cat.) in MeCN, 54% of 2, 37% of 3. (ii) BzCl in pyr/CH2Cl2, -40 to 0 °C, 91% for
4, 44% for 5. (iii) NaBH3CN and HCl·Et2O in THF at 0 °C, 72% of 6. (iv) Et3SiCl and NEt3 in CH2Cl2 at 0 °C, then (v) DDQ in CH2Cl2 with Na2SO4 then
TBAF, 82% over 2 steps.
kinetics and data were performed at 37 °C in HEPES pH 8.5 buffer
(20 mM) containing sodium chloride (50 mM) and magnesium
chloride (20 mM) at substrate concentrations from 0.02 to 1 mM.
Data were analyzed using GraFit from Erithacus software.
The general method for synthesis of 8′′-modified sialyl lactoses
derivatives is as follows. ꢀ-Phenylsulfanyl lactoside30 (1.1 equiv)
was incubated with 8′-modified CMP-sialic acid (1.0 equiv) donor
in HEPES (50 mM, pH 7.5) buffered solution containing manganese
chloride (10 mM) in the presence of Cst-I (65 µM) and alkaline
phosphatase (300 U/mL) at room temperature for 2 h. The mixture
was filtered (0.44 µm) and loaded onto a pre-equilibrated Waters
SepPak C18 (2 g) column and washed with water (10 column
volumes), and products were eluted with 5% acetonitrile in water.
Product-containing fractions were identified by TLC (ethyl acetate/
methanol/water, in a 7:2:1 ratio mobile phase), pooled and
lyophilized. Further purification was performed using reverse-phase
HPLC as described in the Supporting Information.
the H7 and H9 protons indicates that H8 is in an axial position,
and the conformation of 3 shown in Figure 1 is the most likely.
Benzoylation of the 4-hydroxyl of 2 proceeded smoothly to give
compound 4, but benzoylation of 3 was less selective, giving a
mixture of 4- and 8-O-benzoyl compounds, favoring the desired
compound 5. This derivative (5) now contains OH8 as the only
free hydroxyl and was used for some of the modification steps
described below. Compound 5 could also be obtained from the
five-membered acetal 4 via the common intermediate, 9-O-PMB
derivative 6, which was formed by reductive ring opening of
the benzylidene acetal. Protection of the 8-hydroxyl group of
diol 6, using a triethylsilyl group, followed by oxidative ring
closure under anhydrous conditions,33 gave the six-membered
acetal 5 after TBAF treatment to remove the silyl protecting-
group. The 8-O-TES group is required to force the formation
of the six-membered acetal, since oxidative ring closure in its
absence results in preferential reformation of the five-membered
acetal 4. Importantly diol 6 is also a very useful intermediate
for modification of the 8-position, since the reactivity of OH7
is so low.
Replacement of the 8-hydroxyl by fluorine to give a product
of the same stereochemistry requires a double inversion of
stereochemistry to be performed at C8. This was achieved
relatively simply by taking advantage of the greater reactivity
of the 8-hydroxyl over the 7-hydroxyl in 6 in order to form the
mono-O-triflate at C8 by reaction with trifluoromethane sulfonic
anhydride (2 equiv in pyridine at -20 °C). Warming of this
intermediate to ambient temperature resulted in formation of
the L-glycero-D-galacto configured 7,8-epoxide 7, by intramo-
lecular nucleophilic displacement of the 8-O-triflate by the
7-hydroxyl.34 This one-pot reaction proceeds smoothly to yield
a 25:1 ratio of the desired compound 7 and the D-glycero-L-
altro configured 7,8-epoxide, which arises from displacement
of the 7-O-triflate by the 8-hydroxyl. All attempts to open
epoxide 7 with fluoride to install fluorine at the 8-position failed.
However, removal of the 9-O-PMB group, using ceric am-
monium nitrate, to give alcohol 8 greatly increased the reactivity
of the epoxide toward nucleophilic addition. The successful ring
opening was achieved using hydrogen fluoride-pyridine as the
fluoride source in the presence of a catalytic amount of boron
trifluoride to give the desired 8-fluoro derivative 9, along with
equal quantities of a diol byproduct 10, with undetermined
stereochemistry at C8 and C7. Monitoring of this transformation
Results and Discussion
Selective chemical modification at the 8-position of sialic acid
requires that suitably partially protected derivatives of sialic acid
be prepared in which the 8-hydroxyl is uniquely exposed, or at
least preferentially reactive. To achieve this it is necessary to
first protect the anomeric center, and in our case the simple
methyl glycoside derivative 1 was chosen.31 Since the relative
reactivities of the remaining hydroxyls in 1 generally fall in
the order OH9 > OH4 > OH8 > OH7,21 it is essential only to
protect OH9 and OH4 in order to carry out selective reactions
at the 8-position, but in some cases having only OH8 exposed
would be advantageous.
A convenient strategy is to protect both OH9 and OH7
simultaneously using an acetal, and such a method was described
by Magnusson32 on an analogous system that involved the use
of a benzylidene acetal to protect part of the glycerol side chain,
followed by benzoylation at C4. Thus, treatment of 1 with
anisaldehyde dimethyl acetal under acidic conditions gave a
(3:2) mixture of the 9,8- and 9,7-acetals (2 and 3, respectively),
with the five-membered ring product predominating (Scheme
1). This five-membered acetal 2 was isolated as a (4:3) mixture
of diastereoisomers, epimeric at the benzylidene center, whereas
the six-membered acetal 3 was isolated as a single isomer in
which all the substituents occupied equatorial positions. The
large (J ) 10 Hz) coupling constants between H8 and each of
(29) Gosselin, S.; Alhussaini, M.; Streiff, M. B.; Takabayashi, K.; Palcic,
M. M. Anal. Biochem. 1994, 220, 92–97.
(30) Fischer, E.; Delbruck, K. Chem. Ber. 1909, 42, 1476–1482.
(31) Ogura, H.; Furuhata, K.; Itoh, M.; Shitori, Y. Carbohydr. Res. 1986,
158, 37–51.
(33) Evans, D. A.; Cee, V. J.; Siska, S. J. J. Am. Chem. Soc. 2006, 128,
9433–9441.
(34) Brandstetter, H. H.; Zbiral, E.; Schulz, G. Liebigs Ann. Chem. 1982,
(32) Ercegovic, T.; Magnusson, G. J. Org. Chem. 1995, 60, 3378–3384.
1–13.
9
9432 J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010

8-Modified CMP-Sialic Acid and Derivatives
A R T I C L E S
Scheme 2. Synthesis of 8-Fluoro Sialic Acida
a Reagents and conditions: (i) Tf2O in pyridine from -20 °C to rt, 82%. (ii) CAN in MeCN/H2O, 96%. (iii) HF ·pyr in CH2Cl2 at 0 °C, then immediately
BF3 ·Et2O, 44% 9 and 42% 10. (iv) Ac2O in pyr, 80%. (v) Mg(OMe)2 in MeOH, then LiOH (1 M) in THF and water, followed by HCl (aq) (25 mM) at 60
°C, 66% over 3 steps. (vi) Anisaldehyde dimethyl acetal with TsOH (cat.) in MeCN, 40%.
Scheme 3. Synthesis of 8-Deoxy and 8-O-Methyl Sialic Acida
by TLC proved to be difficult because of coelution of starting
material and product in all solvent systems attempted. Similar
problems were encountered in purification, so after an initial
silica gel chromatography step the crude product 9 was
converted to the 9,7-diacetate 11, which could be purified to
homogeneity. Removal of the acetyl and benzoyl groups
presented some problems because basic conditions tended to
lead to degradation. These difficulties were overcome by using
magnesium methoxide in methanol, a much milder method of
acetate removal.35 Subsequent saponification of the methyl ester
was also carried out at low temperature to minimize any
degradation, and the 8-fluoro sialic acid product (12) was
purified by ion-exchange chromatography (Scheme 2).
a Reagents and conditions: (i) PhOC(S)Cl with DMAP (cat.) in pyr at 0
°C. (ii) Bu3SnH with AIBN (cat.) in toluene at reflux, 44% over 2 steps.
(iii) NaOMe in MeOH, then NaOH (0.1 M). (iv) HCl (aq.) (25 mM) at 60
°C, 62% for 15 and 60% for 17, over 2 steps. (v) MeI with NaH in DMF
at 0 °C, 72%.
The stereochemistry at C8 of compound 9 was confirmed by
conversion to the 9,7-PMB acetal 13, allowing a direct
comparison with the spectra of the 8-hydroxy parent compound
5. Again a large (J ) 10 Hz) coupling constant is observed for
H8 to H7 and H9a, indicating that H8 is in the desired axial
position. The 19F NMR spectrum of the 8-fluoro compound 13
shows a single resonance as a double triplet with one large (J
) 49 Hz) coupling to H8 and two small (J ) 5.5 Hz) couplings
to each of H9-axial and H7, indicating that the fluorine is in an
equatorial position. The stereochemistry at C7 and C8 of diol
byproduct 10 was not determined, though comparison with
spectra of a later compound, 18, confirmed that 10 did not have
the natural C7/C8 stereochemistry. This byproduct could be
either the L-glycero-L-altro derivative, from ring opening by
water at the C7 position, or the L-glycero-D-galacto derivative
from a Payne-rearrangement,36 followed by ring opening at the
8-position by water.
9,7-O-PMB acetal 5 provided the ideal departure point to
deoxygenate C8. This was achieved using a Barton-McCombie
protocol to give the 8-deoxy acetal 14 in a reasonable yield.
Most of the loss in yield was due to hydrolysis of the
thiocarbonate intermediate during workup, resulting in isolation
of the 8-O-phenyl-carbonate (not shown). Treatment of 14 with
sodium methoxide, followed by saponification, removed the
4-O-benzoyl group and the C1-ester. Subsequent acidic hy-
drolysis to remove both the benzylidene acetal and methyl
glycoside gave the desired 8-deoxy product (15), which was
isolated by ion-exchange chromatography and possessed spec-
troscopic data identical to those previously published,37 but in
fewer steps and higher overall yield (Scheme 3).
Synthesis of 8-O-methyl sialic acid (17) was also achieved
from the common intermediate 5 by simple methylation of the
8-hydroxyl to give ether 16, plus a small amount of the
overmethylated N-methyl-acetamide, which was isolated as a
minor byproduct. Deprotection of 16 was achieved in a manner
similar to that used for the 8-deoxy compound to give 17 after
ion-exchange chromatography, with identical characteristics to
that previously published, and again in a higher yield and fewer
steps.17,38
Oxidative cleavage of the vicinal-diols in 18 using sodium
periodate initially gave an aldehyde at C7, which was reduced
in situ to give alcohol 19 (Scheme 4). Removal of the esters
under basic conditions proceeded smoothly; however, under the
acidic conditions required to hydrolyze the methyl glycoside, a
deacetylated byproduct 21 was surprisingly also obtained.
Because N-deacylation of this kind is not seen in other
deprotections of this type, it seems probable that this byproduct
arises from acetate migration from the C5-nitrogen to the
primary alcohol at C7. The subsequent deacetylation of the 7-O-
acetate can then occur under the acidic conditions of the reaction.
The desired C7-truncated compound 20 was obtained as a minor
(35) Xu, Y.-C.; Bizuneh, A.; Walker, C. J. Org. Chem. 1996, 61, 9086–
(37) Miyazaki, T.; Sato, H.; Sakakibara, T.; Kajihara, Y. J. Am. Chem.
Soc. 2000, 122, 5678–5694.
9089.
(36) Payne, G. B. J. Org. Chem. 1962, 27, 3819.
(38) Khorlin, A. Y.; Privalov, I. M. Carbohydr. Res. 1970, 13, 373.
9
J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010 9433

A R T I C L E S
Morley and Withers
Scheme 4. Synthesis of a C7-Truncated Heptose Analogue of
Sialic Acid
ion-exchange chromatography on Dowex 1x2-200 resin. Prod-
ucts were further purified and desalted by size-exclusion (Bio-
Gel P2) chromatography to yield the desired, labile products in
excellent yields as shown in Table 1.
To provide insight into the importance of the 8-hydroxyl in
the enzyme-catalyzed sialyl transfer, as well as to identify the
optimal catalyst for the synthesis of sialyl lactose derivatives,
two different sialyltransferases from different GT families and
of different three-dimensional fold were studied. In each case
an initial assessment was performed by monitoring the enzy-
matic transfer reaction by TLC, using a fluorescent BODIPY-
lactose acceptor. This was followed, in each case, by measure-
ment of Michaelis-Menten kinetic parameters for the transfer
reaction, as well as for the enzyme-catalyzed hydrolysis in the
absence of the lactose acceptor, using a previously established
CMP-release assay.29
a Reagents and conditions: (i) TFA/H2O (3:1 ratio), 96%. (ii) NaIO4 in
MeOH, then (iii) NaBH4 in MeOH at 0 °C, 59% over 2 steps. (iv) NaOMe
in MeOH at 0 °C, then NaOH (1M) at rt, followed by HCl (aq.) (25 mM)
at 60 °C, 11% for 20 and 49% for 21.
product but could be readily generated from 21. Spectroscopic
data for 20 were consistent with those previously reported.39
Although the CMP-derivative of each sialic acid analogue
could be accessed by chemical synthesis,40 enzymatic syntheses
are far more attractive because they are more direct and less
laborious.23,24,41 The CMP-sialic acid synthetase from Neisseria
has proved to be useful in this regard, and the equilibrium can
be driven to completion by inclusion of inorganic pyrophos-
phatase to degrade the pyrophosphate generated during the
reaction.22 Indeed it is possible to combine the sialic acid
aldolase, CMP-sialic acid synthetase and sialyltransferase
enzymes in a single reaction mixture to generate sialic acid-
containing oligosaccharides from the corresponding N-acetyl
mannosamine sugars, although reports of this type have mainly
centered on sialylglycosides that are modified at C5 or C9.28,42-47
Our approach was to synthesize and isolate each of the
8-modified CMP-sialic acids, which could be subsequently
evaluated kinetically as substrates for the available sialyltrans-
ferases.
This approach required the isolation and purification of the
nucleotide phosphosugar. The enzymatic synthesis of the CMP-
derivatives was achieved by incubating each of the 8-modified
sialic acid analogues (up to a 50 mg scale) with CTP (1.05
equiv) in the presence of the CMP-sialic acid synthetase (1
U/µmol substrate) and inorganic pyrophosphatase (1 U/mmol
substrate) in a Tris buffer (100 mM, pH 8.5) containing
magnesium chloride (20 mM) and DTT (0.1 mM) (Scheme 5).
Reactions were monitored by TLC and once complete were
terminated by cooling the reaction mixture to -80 °C, after
which they were thawed, filtered to remove protein precipitates,
and then treated with alkaline phosphatase to degrade any CMP
to neutral cytosine. Initial isolation of each of the crude 8′-
modified CMP-sialic acid conjugate products was achieved by
Cst-I is a member of CAZy48 family GT-42 and catalyzes
the formation of R-2,3 linkages to galactose residues using
CMP-sialic acid as the donor substrate. The three-dimensional
structures of this enzyme, both free and as its binary and ternary
complexes, have been published and show that the enzyme
adopts a GT-A fold.27 The TLC experiments (Table 2) show
that Cst-I transfers each of the modified sialic acid donors to
BODIPY-lactose, albeit with a reduced efficiency compared to
the natural donor substrate. The transfer of the natural donor
(lane b) was complete in less than 2 min, whereas transfer of
each of the modified sugars (lanes c to f) required 60 min for
completion under the same conditions.
Kinetic parameters determined using a CMP-release assay29
reveal that the modified donors bind to the enzyme with similar
affinities to the parent substrate, with the exception of the 8-O-
methyl derivative 23, which binds some 4-fold more weakly.
This is a surprisingly small effect given the potential for the
bulk of the methyl group to cause detrimental interactions within
the binding pocket of the enzyme and suggests that ground state
interactions with the substrate at this position are minimal. A
more significant effect is seen on kcat, with an approximate 10-
fold drop, implying that more important interactions at this
position are formed at the transition state. It is also interesting
to note that while the parent substrate undergoes enzyme-
catalyzed hydrolysis at around 10% of the rate of transfer in
the absence of lactose acceptor, no significant transfer to water
was observed for any of the modified sugar donors. This could
imply that interactions formed at this position are particularly
important to the hydrolysis reaction but that binding of acceptor
can partially overcome any deficiency caused by their absence.
The crystal structure of Cst-I, solved in the presence of CMP-
3-fluoro sialic acid (see Figure 2)27 shows that the donor sugar
binding site is not solvent-accessible and that there is only one
significant contact (Gln47) made between the 8- and 9-hydroxyls
of the donor sugar and the enzyme. This may explain the
tolerance of the enzyme toward the modifications at the
8-position. Even more surprising is the relatively high kcat/Km
value obtained for the truncated compound 25, whose kinetic
parameters are essentially identical to those of the 8-deoxy
analogue 22 despite the complete absence of the C8 and C9
hydroxymethyl groups. Presumably the ground state interactions
of OH8 and OH9 with Cst-I are weak enough that their absence
is not deleterious, but interactions at the transition state are
satisfied by the presence of only one hydroxyl of the glycerol
side chain. Indeed the crystal structure shows that OH7 interacts
(39) Hasegawa, A.; Ito, Y.; Ishida, H.; Kiso, M. J. Carbohydr. Chem. 1989,
8, 125–133.
(40) Wagner, G. K.; Pesnot, T.; Field, R. A. Nat. Prod. Rep. 2009, 26,
1172–1194.
(41) Rauvolfova, J.; Venot, A.; Boons, G. J. Carbohydr. Res. 2008, 343,
1605–1611.
(42) Wang, Y. H.; Ye, X. S.; Zhang, L. H. Org. Biomol. Chem 2007, 5,
2189–2200.
(43) Yu, H.; Chokhawala, H. A.; Huang, S. S.; Chen, X. Nat. Protoc. 2006,
1, 2485–2492.
(44) Yu, H.; Cheng, J.; Ding, L.; Khedri, Z.; Chen, Y.; Chin, S.; Lau, K.;
Tiwari, V. K.; Chen, X. J. Am. Chem. Soc. 2009, 131, 18467–18477.
(45) Chokhawala, H. A.; Huang, S. S.; Lau, K.; Yu, H.; Cheng, J. S.; Thon,
V.; Hurtado-Ziola, N.; Guerrero, J. A.; Varki, A.; Chen, X. ACS Chem.
Biol. 2008, 3, 567–576.
(46) Cao, H.; Muthana, S.; Li, Y.; Cheng, J.; Chen, X. Bioorg. Med. Chem.
Lett. 2009, 19, 5869–5871.
(47) Yu, H.; Huang, S. S.; Chokhawala, H.; Sun, M. C.; Zheng, H. J.;
Chen, X. Angew. Chem., Int. Ed. 2006, 45, 3938–3944.
(48) Cantarel, B. L.; Coutinho, P. M.; Rancurel, C.; Bernard, T.; Lombard,
V.; Henrissat, B. Nucleic Acids Res. 2009, 37, D233–D238.
9
9434 J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010

8-Modified CMP-Sialic Acid and Derivatives
A R T I C L E S
Scheme 5. Synthesis of CMP Donor Sugarsa
a Reagents and conditions: (i) TFA/H2O (3:1 ratio), 96%. (ii) NaIO4 in MeOH, then (iii) NaBH4 in MeOH at 0 °C, 59% over 2 steps. (iv) NaOMe in
MeOH at 0 °C, then NaOH (1 M) at rt, followed by HCl (aq.) (25 mM) at 60 °C, 11% for 20 and 49% for 21.
Table 1. Summary of Enzymatic Syntheses of
8-Modified-CMP-Sialic Acid Derivatives Using CMP-Sialic Acid
25 (about 1:3.2), meaning that a larger proportion of the donor
Synthetase and 8-Modified Sialic Acids
to transfer measured for 22 and 23 (about 1:1.8) versus that of
is being hydrolyzed by the enzyme. In fact prolonged incubation
product
no.
scale (mmol)
time (h)
yield (%)
times resulted in significant hydrolysis of the BODIPY-sialyl-
lactose transfer products back to BODIPY-lactose such that after
18 h all sialyl-lactoside conjugates, with the exception of the
truncated transfer product (lane f), had been degraded back to
their lactoside derivatives. Interestingly two product spots are
seen in lane c corresponding to the transfer of 8-deoxysialic
acid, suggesting the formation of two different products. The
faster running product has an Rf consistent with that of the
transfer product of Cst-I, the R-2,3-linked product, and the lower
running product is most likely the R-2,6-linked sialoside, since
the Pasteurella enzyme is known to have this alternative
transferase activity. Furthermore, incubation of the 8-deoxy-
sialyl lactose products at a lower pH (5.3) in the presence of
the Pasteurella enzyme resulted in exclusive degradation of the
faster running product. This is further evidence that this product
is the R-2,3-linked sialoside, since the enzyme is known to
cleave R-2,3-linked sialosides at low pH values but leaves R-2,6-
linked sialosides untouched.28
CMP-8-H-NeuNAc
CMP-8-O-Me-NeuNAc
CMP-8-F-NeuNAc
CMP-C7-NeuNAc
22
23
24
25
0.11
0.20
0.03
0.08
0.5
1
6
83
78
82
72
4
with Asn66, and these kinetic results could suggest a significant
contribution of this residue to stabilization of the transition state.
Quite different results were obtained with the Pasteurella
enzyme, which is a member of the GT-80 family and has a
GT-B fold. This is a much more efficient enzyme than Cst-I,
with kcat/Km values approximately 250-fold higher. Earlier
studies28 have revealed a much broader specificity, with the
ability to form both R-2,3 and R-2,6-linkages to galactose along
with having R-2,3-sialidase and R-2,3-trans-sialidase activities.
The precise mechanisms of these latter activities, along with
their different pH-dependences, remain unclear.
The experiments described in this paper were performed at
pH 8.5, conditions described as being optimal for the R-2,3-
sialyltransferase activity. Both the TLC and kinetic data in Table
3 reveal that the Pasteurella enzyme will transfer all of the
modified sugars to lactose, although the TLC studies revealed
that, even in the presence of a 2-fold excess of donor sugar
over acceptor, reactions with 22, 23 and 24 did not proceed to
completion. This may be due to the higher ratio of hydrolysis
The kinetic parameters reveal essentially no effect at all of
8-position modification on kcat and only very minimal effects
upon Km. This would imply that no significant interactions are
formed between the enzyme and the 8-position at either the
ground state (Michaelis complex) or at the reaction transition
state. Further, and also in marked contrast to what was seen
with Cst-I, there was no deleterious effect upon the hydrolytic
Table 2. Kinetic Parameters for the Transfer of Modified CMP-Sialic Acid Derivatives to Lactose and to Water by Cst-Ia
a Also shown are the transfers of each of the modified donors to bodipy-lactose by Cst-I (15 µM), performed at ambient temperature, as monitored by
TLC (ethyl acetate/methanol/water/acetic acid mobile phase in a 4:2:1:0.1 ratio): lane a, bodipy lactose acceptor; b, CMP-sialic acid donor; c, donor 22;
d, donor 23; e, donor 24; f, donor 25.
9
J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010 9435

A R T I C L E S
Morley and Withers
Figure 2. Interaction of glycerol side chain with Cst-I (left) and Pm0188h (right) sialyltransferases.
Table 3. Kinetic Parameters for the Transfer of Modified CMP-Sialic Acid Derivatives to Lactose and to Water by PM0188ha
a Also shown are the transfers of each of the modified donors to bodipy-lactose by PM0118h (5 µM), performed at ambient temperature, as
monitored by TLC (ethyl acetate/methanol/water/acetic acid mobile phase in a 4:2:1:0.1 ratio): lane a, bodipy lactose acceptor; b, CMP-sialic acid
donor; c, donor 22; d, donor 23; e, donor 24; f, donor 25.
Scheme 6. Synthesis of 8′′-Modified Sialyl Lactose Analoguesa
Table 4. 1H NMR Chemical Shifts of H3 Protons of Sialyl Lactose
Conjugates 26-30 in Comparison with Previously Published
R-Sialosides
compound
H3ax
H3eq
26
27
28
29
30
1.66
1.64
1.62
1.63
1.62
1.59
1.69
1.68
2.62
2.54
2.54
2.56
2.57
2.70
2.69
2.63
R-OH-Neu5Ac50
R-O-Me-Neu5Ac51
R-O-Gal-Neu5Ac52
a Reagents and conditions: (i) Cst-I, 50 mM HEPES pH 7.5, 10 mM
MnCl2 and MgCl2, alkaline phosphatase, 4 mM ꢀ-thiophenyl lactoside and
5 mM CMP-8-modified donor sugar.
An important consequence of this behavior is that, although
the Pasteurella enzyme is an excellent catalyst for syntheses
with the natural as well as C5- and C9-modified CMP-sialic
acid donors, especially given its high efficiency, it is not as
useful for substrates modified at the 8-position. The shift from
a 5:1 ratio of transfer rate to lactose:hydrolysis rate constant
for the natural sugar to a ratio of under 2:1 with 22 and 23
renders the enzyme less useful for such syntheses, especially
given the need to very carefully monitor reaction progress
in order to minimize hydrolysis of the sialylated product.
Thus the transferase from Campylobacter (Cst-I) was chosen
reaction; indeed for two of the substrates the rate of sialyl
transfer to water was stimulated 2-fold. These findings are
entirely consistent with the published three-dimensional struc-
tures of the Pasteurella sialyltransferase in complex with
substrates,26 which showed that both the 8- and 9-hydroxyls
are essentially fully solvent-exposed, with the OH7 being the
only hydroxyl within the glycerol side chain to make significant
contacts with the enzyme through a hydrogen bond to Trp270
(see Figure 2).
9
9436 J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010

8-Modified CMP-Sialic Acid and Derivatives
A R T I C L E S
Table 5. 1H and 13C NMR Chemical Shifts of the Galactose Residue in Sialyl Lactose Conjugates 26-30 in Comparison with
ꢀ-PhS-Lactoside
compound
C2′
C3′
C4′
C6′
H2′
H3′
H4′
H6’a
H6’b
ꢀ-PhS-lactoside
70.4
68.8
68.9
68.8
68.9
69.0
72.0
74.9
74.9
75.2
75.0
74.9
68.0
66.9
67.2
67.2
67.0
66.9
60.5
59.5
59.6
59.5
59.6
59.6
3.40
3.43
3.43
3.43
3.43
3.42
3.52
3.97
3.96
3.94
3.97
3.97
3.78
3.81
3.85
3.84
3.84
3.85
3.61
3.69
3.69
3.69
3.69
3.68
3.65
3.83
3.84
3.85
3.83
3.82
26
27
28
29
30
as the optimal catalyst to synthesize 8′′-modifed sialyl lactose
analogues. It is also worth noting that Cst-I can also transfer
to GM1-glycan, T-antigen and lactosamine acceptors,49
indicating that further sialic acid containing oligosaccharides
could be synthesized using this methodology, which may not
be limited to lactose derivatives.
that C3′ shows the largest deviation in chemical shift (ap-
proximately 3 ppm downfield) from the parent lactoside (72.0
ppm), with C2′ and C4′ showing smaller (less than 1 ppm) but
significant differences in chemical shift, consistent with sialyl-
ation at the 3-position of galactose. Further evidence is provided
by comparison of the 1H NMR spectra whereby the H3′ proton
in each case has moved downfield, by approximately 0.4 ppm
to around 3.95 ppm, compared with the parent acceptor
compound (3.52 ppm). Yet further evidence is the strong
correlation observed in the HMBC spectra between H3′ and
C2′′ for each trisaccharide. These data confirm that the modified
sialic acid-containing donors do not affect or change the
specificity of Cst-I and that the 8′′-modified-sialyl lactose
products possess the natural R-2,3-glycosidic linkage.
Completion of the synthesis of the 8′′-modified sialyl
lactose derivatives was carried out using a thiophenyl
lactoside as the acceptor sugar, since the aryl glycoside can
be used to aid purification via reversed-phase chromatogra-
phy, and also to monitor the progress of the reaction. Simple
alkyl and aryl galactosides can also be used as acceptors. A
slight excess of acceptor was used to ensure that all of the
valuable donor was converted to product, and the concentra-
tions of donor and enzyme were optimized to minimize the
reaction time. On completion of the reaction, as monitored
by TLC, each of the sialyl lactose conjugates was isolated
using a C18-Sep-pak cartridge and further purified by
reversed-phase HPLC. The isolated yields obtained (Scheme
6) in each case are good to excellent, even for the C7-
truncated analogue 30.
To confirm the stereochemistry and position of the sialyl lactose
glycosidic bond of compounds 26 through 30, the 1H and 13C NMR
spectra of each were unambiguously assigned. To address the
stereochemistry of the glycosidic bond, the chemical shifts of both
of the H3′′ protons were compared to those of previously reported
R-sialosides. The chemical shifts of the H3 protons of sialic acid
are extremely sensitive to, and thus diagnostic of, the anomeric
configuration at C2. Most R-configured sialosides display a
significantly downfield-shifted H3-equatorial proton (2.9-2.5 ppm)
compared with the same proton of a ꢀ-sialoside (2.3-1.9 ppm).
Furthermore R-sialosides possess a larger difference in chemical
shift between H3-equatorial and H3-axial protons (1.1-0.9 ppm
difference), compared with ꢀ-sialosides (0.3-0.6 ppm difference).51
The chemical shifts of the H3-equatorial protons of all of the
synthesized sialyl lactose conjugates (26-30) are entirely consistent
with those of an R-sialoside (see Table 4), as is the difference in
chemical shift between the two H3 protons, confirming that Cst-I
transfers 8-modified sialic acids to form R-sialosides.
Conclusion
A series of 8-modified sialic acids has been synthesized
and each converted to their CMP-derivative using a bacterial
CMP-sialic acid synthetase. These modified donors were
evaluated as substrates for two sialyltransferases. Cst-I from
C. jejuni was found to transfer each of the modified sugars
to lactose with a 10-fold lower catalytic efficiency compared
with the natural donor substrate. By contrast Pm0188h from
P. multocida was found to transfer the modified sugars at
the same rates as the natural donor but catalyzed their
hydrolysis more efficiently than is the case with the natural
substrates. Of these two enzymes Cst-I, with its minimal
hydrolysis of the modified donors and good transfer rates,
has the most desirable characteristics for efficient synthesis
of a number of 8′′-modified sialyl lactoses and was used as
such, with yields of around 80% being obtained. The
conditions used are compatible with other biological systems
and may be used to incorporate such modified sialic acids
onto glycoproteins and other sensitive biomolecules, a clear
advantage over chemical glycosylation techniques. The
8-modified sialyl lactose derivatives prepared will be em-
ployed in kinetic and structural studies on polysialyltrans-
ferases to probe substrate binding modes and energetics.
Acknowledgment. We thank the Canadian Institutes of Health
Research (CIHR) for financial support of this work. We also
acknowledge valuable discussions with Drs. Warren Wakarchuk
and Francesco Rao. T.J.M. was supported by a Leverhulme Trust
fellowship. CMP-N-acetyl neuraminic acid was a kind gift from
Neose Technologies.
To address the position of attachment of the sialic acid, the
1H and 13C chemical shifts of the galactose residues in each of
the sialyl lactose conjugates (26-30) were compared with those
of the ꢀ-thiophenyl lactoside acceptor (Table 5). For clarity only
data for those positions on the galactose that bear a free hydroxyl
are shown for comparison. It is clear from the 13C NMR data
Supporting Information Available: Experimental details and
1H and 13C NMR data for the chemical synthesis of all modified
sialic acid derivatives, the chemoenzymatic synthesis of all
sialyl-lactosides, and the kinetic analysis of all modified sialic
acid donors. This material is available free of charge via the
Internet at http://pubs.acs.org.
(49) Gilbert, M.; Karwaski, M. F.; Bernatchez, S.; Young, N. M.; Taboada,
E.; Michniewicz, J.; Cunningham, A. M.; Wakarchuk, W. W. J. Biol.
Chem. 2002, 277, 327–337.
(50) Kao, Y. H.; Lerner, L.; Warner, T. G. Glycobiology 1997, 7, 559–
563.
(51) Dabrowski, U.; Friebolin, H.; Brossmer, R.; Supp, M. Tetrahedron
Lett. 1979, 20, 4637–4640.
(52) Turnbull, W. B.; Harrison, J. A.; Ravindranathan Kartha, K. P.;
Schenkman, S.; Field, R. A. Tetrahedron 2002, 58, 3207–3216.
JA102644A
9
J. AM. CHEM. SOC. VOL. 132, NO. 27, 2010 9437