Glycosylation of N-acetylglucosamine: Imidate formation and unexpected conformation

Article abstract of DOI:10.1021/ol034669x

(Matrix presented) Rhamnosylation in mild conditions of a disaccharide containing N-acetylglucosamine afforded the imidate 6 while at higher temperature and concentration of promoter trisaccharide 7 was isolated. The kinetic imidate 6 was independently re

Full text of DOI:10.1021/ol034669x

ORGANIC
LETTERS
2003
Vol. 5, No. 15
2607-2610
Glycosylation of N-Acetylglucosamine:
Imidate Formation and Unexpected
Conformation
Liang Liao and France-Isabelle Auzanneau*
Department of Chemistry and Biochemistry, UniVersity of Guelph,
Guelph, Ontario N1G 2W1, Canada
auzanneau@chembio.uoguelph.ca
Received April 21, 2003
ABSTRACT
Rhamnosylation in mild conditions of a disaccharide containing N-acetylglucosamine afforded the imidate 6 while at higher temperature and
concentration of promoter trisaccharide 7 was isolated. The kinetic imidate 6 was independently rearranged in 50% yield to the thermodynamic
trisaccharide 7. Comparative NMR studies of 7 in CDCl3 and DMSO-d6 suggest the formation of a nonchair conformation in CDCl3. The structure
of 7 was confirmed through the independent synthesis of the N-acetylacetamido trisaccharide 11.
The overexpression of carbohydrate structures at the surface
of tumor cells has been extensively documented.1 Since
carbohydrates can be involved in specific immune responses
directed against the cells or bacteria that display them, a
number of these tumor-associated carbohydrate antigens are
currently being investigated as potential vaccines against
cancer.2 The Lewis blood group antigens (Figure 1) Lea (1)
and LeaLex (3) are known to be overexpressed on tumor
tissues and cells.1 However, since Lea is also largely
expressed on normal cells, anti-cancer vaccines based on
these antigens are likely to trigger an immune reaction that
will lead to the destruction of these normal cells.3 To avoid
such autoimmune reactions, we are studying analogues of
(1) Hakomori, S.-I. AdV. Cancer Res. 1989, 52, 257-331. Hakomori,
S.-I.; Y. Zhang, Y. Chem. Biol. 1997, 4, 97-104. Hakomori, S.-I. Acta
Anat. 1998, 161, 79-90.
(2) Danishefsky, S. J.; Allen, J. R. Angew. Chem., Int. Ed. 2000, 39,
837-863. Wang, Z. G.; Williams, L. J.; Zhang, X.-F.; Zatorski, A.;
Kudryashov, V.; Ragupathi, G.; Spassova, M.; Bornmann, W.; Slovin, S.
F.; Scher, H. I.; Livingston, P. O.; Lloyd, K. O.; Danishefsky, S. J. Proc.
Natl. Acad. Sci., U.S.A. 2000, 97, 2719-2724. Slovin, S. F.; Ragupathi,
G.; Alduri, S.; Ungers, G.; Terry, K.; Kim, S.; Spassova, M.; Bornmann,
W. G.; Fazzari, M.; Dantis, L.; Olkiewicz, K.; Lloyd, K. O.; Livingston, P.
O.; Danishefsky, S. J.; Scher, H. I. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
5710-5715.
Figure 1. Blood group antigens and analogue: Lea (1), Lea
analogue (2), and LeaLex (3).
10.1021/ol034669x CCC: $25.00 © 2003 American Chemical Society
Published on Web 06/26/2003

LeaLex that no longer display the Lea antigen but might retain
common antigenic determinants with LeaLex. To assess its
cross reactivity with Lea, we have embarked on the prepara-
tion of analogue 2 in which the fucosyl residue is replaced
by a rhamnosyl unit. Thus, our synthetic scheme involved
the rhamnosylation of the known4 disaccharide 4 which,
along with other similar glycoside derivatives, were suc-
cessfully fucosylated to prepare Lea trisaccharide analogues.5-7
Studies with glycosyl donors less reactive than fucosyl
donors have clearly established8 the poor reactivity of the
4-OH group in N-acetylglucosamine derivatives when com-
pared to the 4-OH group in other acceptors. In fact,
construction of a glycosidic bond at this position is often
achieved after the introduction of an N-phathalimido9 or
azido10 substituent at C-2 of the glucosamine residue. Thus,
it is not surprising that glycosylation of 4 with a less reactive
rhamnosyl donor was more delicate than its fucosylation.
The low reactivity of the 4-OH group in N-acetylglucosamine
derivatives has long been believed to result from steric
hindrance at this position.8 Recently Crich et al. proposed
an alternative explanation based on the formation of a
hydrogen bond involving the glucosamine amide group and
lowering the reactivity of such acceptors toward glycosyla-
tion.11 In our attempt to synthesize analogue 2, we have found
that, under mild conditions, the NH group in acceptor 4 could
act as a competitive nucleophile in the glycosylation reaction
with donor 5 and lead to the formation of imidate 6 as the
major product (Scheme 1). Although the formation of such
6 was stirred in these glycosylation conditions it rearranged
to the trisaccharide in 50% yield. Therefore, our results
provide a third explanation for the low reactivity of the 4-OH
groups in N-acetylglucosamine glycosyl acceptors. We
propose that imidates form first as the kinetic products of
the glycosylation reactions and that depending on the
structure of the glycosyl donor and the reaction conditions
they do or do not rearrange to the thermodynamic glycosides.
NMR analysis of trisaccharide 7 in deuterated chloroform
showed that the glucosamine ring adopted an unusual
conformation that disappeared when it was dissolved and
analyzed in deuterated DMSO. To confirm the structure of
7, we also report here the unequivocal and efficient synthesis
of the N-acetylacetamido trisaccharide 11 (Scheme 2) via
Scheme 2
Scheme 1a
the glycosylation of acceptor 10 with rhamnosyl donor 5.
Deacetylation of both 7 and 11 led to the identical trisac-
charide 12.
Coupling of the known4 disaccharide glycosyl acceptor 4
with R-L-rhamnopyranosyl trichloroacetimidate glycosyl
donor13 5 was first attempted at low temperature and with
0.1 equiv of triethylsilyl trifluoromethanesulfonate (TESOTf)
as promoter (Scheme 1, a). In these mild conditions the
(4) Khare, D. P.; Hindsgaul, O.; Lemieux, R. U. Carbohydr. Res. 1985,
136, 285-308.
(5) Lemieux, R. U.; Driguez, H. J. Am. Chem. Soc. 1975, 97, 4063-
4068.
a Reagents and conditions: (a) TESOTf (0.1 equiv), CH2Cl2, -78
°C f rt. (b) TESOTf (0.5 equiv), CH2Cl2, rt.
(6) Lemieux, R. U.; Bundle, D. R.; Baker, D. A. J. Am. Chem. Soc.
1975, 97, 4076-4083.
(7) Jacquinet, J. C.; Sinay¨, P. J. Chem. Soc., Perkin Trans. 1 1979, 319-
322.
(8) Paulsen, H. Angew. Chem., Int. Ed. Engl. 1982, 21, 155-173.
(9) For example, see: Nilsson, M.; Norberg, T. Carbohydr. Res. 1988,
183, 71-82.
imidates has been postulated12 these had never been isolated.
When the glycosylation was conducted at higher temperature
and concentration of promoter, the O-glycosylated product
7 was obtained in 52% yield. In addition, when the imidate
(10) For example, see: Toepfer, A.; Schmidt, R. R. J. Carbohydr. Chem.
1993, 12, 809-822.
(11) Crich, D.; Dudkin, V. J. Am. Chem. Soc. 2001, 123, 6819-6825.
(12) Hindsgaul, O.; Norberg, T.; Le Pendu, J.; Lemieux, R. U. Carbohydr.
Res. 1982, 109, 109-142.
(13) van Steijn, A. M. P.; Kamerling, J. P.; Vliegenthart, J. F. G
Carbohydr. Res. 1991, 211, 261-277.
(3) Lemieux, R. U.; Baker, D. A.; Weinstein, W. M.; Switzer, C. M.
Biochemistry 1981, 20, 199-205.
2608
Org. Lett., Vol. 5, No. 15, 2003

imidate 6 was isolated in 42% yield. The imidate structure
for 6 was first assigned on the basis of its characteristics
measured by NMR. While 1H NMR at 600 MHz showed no
signal corresponding to the 2-NH expected in the wanted
trisaccharide, an exchangeable signal coupling to H-4 of the
glucosamine unit was identified at 3.90 ppm supporting the
presence of the free 4-OH group. In accordance with the
few examples of N-methyl glycosyl acetimidates reported
by Sinay¨ et al.7 a signal assigned to the rhamnosyl H-1′′
was found at 6.08 ppm. In addition, an HMBC experiment
showed a long-range correlation between this signal and a
quaternary carbon at 162 ppm, which was assigned to the
imidate carbon CdN. In the 13C NMR spectrum the
rhamnosyl anomeric carbon gave a signal at lower chemical
shift (92 ppm) than that expected for an O-glycoside (∼100
that, depending on their stability and the reaction conditions,
may or may not rearrange to the thermodynamic glycosides
constitutes a third explanation to the low reactivity toward
glycosylation of the 4-OH group in N-acetylglucosamine
derivatives.
1
Careful study of the H NMR spectrum measured for 7
dissolved in CDCl3 showed unusual features. The coupling
3
1
constants JH-1,H-2 and JC-1,H-1 measured for the glu-
cosamine ring were respectively 4.2 and 168 Hz and did not
4
support the expected C1 conformation for this residue. As
observed for Fuc H-5′′ in protected10,15 and deprotected6 Lea-
containing oligosaccharides, we also expected that, providing
they had similar conformations, H-5′′ in the rhamnose
analogue 7 would be deshielded to ∼4.7 ppm due to its
proximity to GlcNAc O-3, Gal O-4′, and Gal O-5′.6 However,
this signal was found at 4.30 ppm, a chemical shift typically
measured for H-5 of rhamnosyl residues when they are not
submitted to any deshielding effect.16 Therefore, it appears
that trisaccharide 7 has a different conformational behavior
than Lea derivatives. On the basis of the Karplus equation17
the 3JH-1,H-2 coupling constant can result from 7 existing as
a major conformer in which the glucosamine residue is
flattened leading to a dihedral angle ΦH-1,H-2 ≈ 130-140°.
However, since coupling constants measure averaged spin
couplings, 7 could also exist as a mixture of two conform-
ers: the 4C1 chair (ΦH-1,H-2 ≈ 180°) and a 1S3 skew
(ΦH-1,H-2 ≈ 110°). Low-temperature NMR experiments
(down to -55 °C) did not allow a conformational freeze
out, therefore if 7 exists as a mixture of conformers in CDCl3
the energy barrier to interconversion between these conform-
ers must be small. So far NMR experiments did not allow a
complete and unambiguous analysis of the conformational
behavior of 7 in CDCl3.
1
ppm). The JC,H coupling constant (177 Hz) measured for
this anomeric carbon was in perfect agreement with that of
an R-rhamnosidic linkage.14 Finally, the N-acetimidate methyl
group gave a signal at 16 ppm while the N-acetyl methyl
group in acceptor 4 was found at 24 ppm. The structure of
6 was further supported by HRCI-MS giving an observed
molecular ion [M + H] at 928.3427 for a calculated value
of 928.3450. The formation of imidates during the glyco-
sylation of N-acetylglucosamine glycosyl acceptors has long
been hypothesized by Hindsgaul et al.12 However, because
they are highly susceptible to hydrolysis and did not survive
chromatography on silica gel, no structural evidence could
be obtained. Although we did see degradation during workup
and purification steps, the imidate 6 was stable enough to
undergo a quick column chromatography and be obtained
in sufficient amounts and purity to permit its full character-
ization. When the amount of TESOTf was increased to 0.5
equiv and the reaction conducted overnight at room tem-
perature, only traces of 6 were seen by TLC. A new major
product was formed and isolated pure in 52% yield (Scheme
1, b). The NMR characteristics in CDCl3 for this new
compound showed the presence of an R-rhamnosyl residue
(H-1′′ 4.87 ppm, 1JC-1′′,H-1′′ ) 172 Hz) and the disappearance
of the glucosamine 4-OH signal while the NH signal was
found at 5.87 ppm. These assignments together with the
HRCI-MS data obtained for the molecular ion (found [M +
H] 928.3513, calcd 928.3450) led us to identify this
compound as being the wanted trisaccharide 7. Thus it
appears that higher temperature and concentration of TESOTf
favored the formation of the trisaccharide while the imidate
was readily formed at low temperature and concentration of
TESOTf. An obvious explanation is that imidate 6 formed
initially as the kinetic product of the reaction and then
rearranged to the thermodynamically more stable trisaccha-
ride 7 when the temperature and catalyst concentration were
high enough. To investigate this hypothesis the imidate 6
was allowed to rearrange in the conditions used when 7 was
formed. Indeed, the trisaccharide 7 was then isolated in 50%
yield while some degradation of the imidate giving the
acceptor 4 and the rhamnose hemiacetal was also observed
by TLC. Thus the competitive formation of kinetic imidates
Upon Zemple´n deacetylation trisaccharide 7 gave a
compound the NMR and HRCI-MS characteristics of which
did not present any conformational or structural abnormalities
and supported its identification as the mono-benzylated
analogue 12 (Scheme 2). However, since to our knowledge
this is the first time that a conformational behavior such as
that observed in trisaccharide 7 is being reported, we engaged
in an alternative synthesis of trisaccharide 12. To prevent
imidate formation and thus increase the reactivity of the
glycosyl acceptor toward glycosylation at O-4 of the glu-
cosamine residue, we investigated the methodology devel-
oped by Crich et al.11 in which the amino group in
glucosamine glycosyl acceptors was bis-acetylated. The
known4 disaccharide 8 was N-acetylated in the presence of
Hu¨nig’s base to give 9 that was converted to the wanted
glycosyl acceptor 10 by treatment with NaBH3CN/HCl. As
expected, the glycosylation of 10 with donor 5 could be
conducted at low temperature, using only 0.15 equiv of
TESOTf, and gave the trisaccharide 11 in excellent yield
(91%). Zemple´n deacetylation of 7 and 11 gave in both cases
(15) Lay, L.; Manzoni, L.; Schmidt, R. R. Carbohydr. Res. 1998, 310,
157-171.
(16) Auzanneau, F.-I.; Bundle, D. R. Can. J. Chem. 1993, 71, 534-
548. Auzanneau, F.-I.: Forooghian, F.; Pinto, B. M. Carbohydr. Res. 1996,
291, 21-41.
(14) Bock, K.; Pedersen, C. J. J. Chem. Soc., Perkin Trans. 2 1974, 293-
297.
(17) Karplus, M. J. Am. Chem. Soc. 1963, 85, 2870-2871.
Org. Lett., Vol. 5, No. 15, 2003
2609

the mono-benzylated trisaccharide 12 in quantitative yield,
thus confirming that the NMR characteristics measured for
7 in CDCl3 resulted from this trisaccharide adopting an
unusual conformation. Despite their similar structures, trisac-
charides 11 and 7 did not show the same conformational
behavior in CDCl3. Thus, we postulated that the conforma-
tional behavior of 7 resulted from the formation of an inter-
or intramolecular hydrogen bond involving the glucosamine
NH group rather than from steric hindrance around the
glycosidic bonds. To support this hypothesis we recorded
NMR data for trisaccharide 7 in the hydrogen bonding-
solvent DMSO-d6. Indeed, in this solvent the coupling
In summary, we have confirmed that imidate formation
could be the predominant reaction under glycosylation
conditions when N-acetylglucosamine derivatives are used
as glycosyl acceptors. Such an imidate was isolated for the
first time in amounts and purity that allowed its full
characterization by NMR and HRCI-MS. Subsequent gly-
cosylation at higher temperature and concentration of catalyst
led to the formation of the wanted trisaccharide in 52% yield.
When the kinetic imidate was stirred in these conditions, it
rearranged to the thermodynamically more stable trisaccha-
ride that was obtained in 50% yield. Thus, the formation of
kinetic imidates provides an additional explanation to the
exceptionally low reactivity of 4-OH groups toward glyco-
sylation in N-acetylglucosamine glycosyl acceptors. We have
also established that oligosaccharides containing N-acetyl-
glucosamine could adopt unusual conformations induced by
inter- or intramolecular hydrogen bonding when dissolved
in non-H-bonding solvent. To the best of our knowledge,
this is the first time that such conformational behavior is
reported for N-acetylglucosamine derivatives. Finally, to
confirm the chemical structure assigned to this unusual
conformation, the Lea trisaccharide analogue was synthesized
unequivocally and in excellent yield employing, as an
alternative synthetic strategy, a N,N-diacetyl protected glu-
cosamine disaccharide derivative as the glycosyl acceptor.
3
1
constants JH-1,H-2 and JC-1,H-1 measured for the glu-
cosamine ring were respectively, 8.0 and 161 Hz and thus
4
typical of a C1 conformation for this â-glycoside.14 In
addition, the H-5′′ signal of the rhamnosyl residue was then
downfield shifted to 4.75 ppm as one would expect in such
derivatives.6,10,15 Therefore, these results support the forma-
tion of an inter- or intramolecular H-bond in trisaccharide 7
when it is dissolved in non-hydrogen bonding solvents.
Further NMR experiments at various temperatures showed
a linear negative temperature dependence of the NH chemical
shift δNH for 7 dissolved in CDCl3. Linear regression led to
a ∆δNH/∆T value of -3.1 ppb/K, which suggests the
formation of an intramolecular hydrogen bond.18,19 However,
NMR spectra collected for various concentrations of 7 in
Acknowledgment. We thank the Research Corporation,
the National Science and Engineering Research Council of
Canada, the Canada Foundation for Innovation, the Ontario
Innovation Trust, and Aventis Pasteur Ltd for financial and
in kind support of this work.
CDCl3 showed a positive concentration dependence of δNH
,
which is usually indicative of an intermolecular H-bond.18,19
Therefore we conclude that the NH group in 7 can form
either an inter- or an intramolecular hydrogen bond leading
to the unusual conformational behavior that we observed by
NMR in CDCl3.
Supporting Information Available: Spectroscopic data
for compounds 6, 7, and 9-12. This material is available
free of charge via the Internet at http://pubs.acs.org.
(18) Vasella, A.; Witzig, C. HelV. Chim. Acta 1995, 78, 1971-1982.
(19) Flower, P.; Bernet, B.; Vasella, A. HelV. Chim. Acta 1996, 79, 269-
287.
OL034669X
2610
Org. Lett., Vol. 5, No. 15, 2003